The Archaeology of Australia's Deserts

more information - www.cambridge.org/9780521407458

THE ARCHAEOLOGY OF AUSTRALIA’S DESERTS

This is the ﬁrst book-length study of the archaeology of Australia’s deserts, one of the world’s major habitats and the largest block of drylands in the Southern Hemisphere. Over the last few decades, a wealth of new environmental and archaeological data about this fascinating region has become available. Drawing on a wide range of sources, The Archaeology of Australia’s Deserts explores the late Pleistocene settlement of Australia’s deserts, the formation of distinctive desert societies and the origins and development of the hunter-gatherer societies documented in the classic nineteenth-century ethnographies of Spencer and Gillen. Written by one of Australia’s leading desert archaeologists, the book interweaves a lively history of research with archaeological data in a masterly survey of the ﬁeld and a profoundly interdisciplinary study that forces archaeology into conversations with history and anthropology, economy and ecology, and geography and earth sciences.

Mike Smith is the senior archaeologist at the National Museum of Australia. For more than 30 years, he has worked extensively across the Australian arid zone, piecing together the archaeology of this immense continental region of dunefields, sandy rivers, salt lakes, and desert uplands. His previous appointments include field archaeologist at the Northern Territory Museum in Darwin and Alice Springs, research Fellow in the Research School of Pacific and Asian Studies at the Aus- tralian National University, and lecturer in archaeology for the Depart- ment of Archaeology and Anthropology at the Australian National Uni- versity. A Fellow of the Australian Academy of the Humanities and of the Society of Antiquaries (London), he was awarded the Rhys Jones medal by the Australian Archaeological Association in 2006 for ‘out- standing contributions to Australian Archaeology’. In 2010, he received the Verco medal from the Royal Society of South Australia for his research.

CAMBRIDGE WORLD ARCHAEOLOGY series editor NORMAN YOFFEE, University of Nevada, Las Vegas editorial board SUSAN ALCOCK, Brown University TOM DILLEHAY, Vanderbilt University TIM PAUKETAT, University of Illinois STEPHEN SHENNAN, University College London CARLA SINOPOLI, University of Michigan DAVID WENGROW, University College London

The Cambridge World Archaeology series is addressed to students and professional archaeologists, and to academics in related disciplines. Most volumes present a survey of the archaeology of a region of the world, providing an up-to-date account of research and integrating recent findings with new concerns of interpretation. While the focus is on a specific region, broader cultural trends are discussed and the implications of regional findings for cross-cultural interpretations considered. The authors also bring anthropological and historical expertise to bear on archaeological problems and show how both new data and changing intellectual trends in archaeology shape inferences about the past. More recently, the series has expanded to include thematic volumes. recent books in the series li liu and xingcan chen, The Archaeology of China stephen d. houston and takeshi inomata, The Classic Maya philip l. kohl, The Making of Bronze Age Eurasia lawrence barham and peter mitchell, The First Africans robin dennell, The Palaeolithic Settlement of Asia christopher pool, Olmec Archaeology and Early Mesoamerica samuel m. wilson, The Archaeology of the Caribbean richard bradley, The Prehistory of Britain ludmila koryakova and andrej epimakhov, The Urals and Western Siberia in the Bronze and Iron Ages david wengrow, The Archaeology of Early Egypt paul rainbird, The Archaeology of Micronesia peter m. m. g. akkermansa and glenn m. schwartz, The Archaeology of Syria timothy insoll, The Archaeology of Islam in Sub-Saharan Africa cambridge world archaeology

THE ARCHAEOLOGY OF AUSTRALIA’S DESERTS mike smith National Museum of Australia cambridge university press Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, Sao˜ Paulo, Delhi, Mexico City Cambridge University Press 32 Avenue of the Americas, New York, ny 10013-2473, usa www.cambridge.org Information on this title: www.cambridge.org/9780521407458

C Mike Smith 2013

This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press.

First published 2013

Printed in the United States of America

A catalog record for this publication is available from the British Library.

Library of Congress Cataloging in Publication Data Smith, M. A., author. The archaeology of Australia’s deserts / Mike Smith, National Museum of Australia. pages cm. – (Cambridge world archaeology) Includes bibliographical references and index. isbn 978-0-521-40745-8 (hardback) 1. Deserts – Australia. 2. Environmental archaeology – Australia. 3. Archaeology – Australia. 4. Human ecology – Australia. I. Title. gb618.89.s65 2013 994.01–dc23 2012031958 isbn 978-0-521-40745-8 Paperback

Cambridge University Press has no responsibility for the persistence or accuracy of urlsfor external or third-party Internet Web sites referred to in this publication and does not guarantee that any content on such Web sites is, or will remain, accurate or appropriate. For Ben, Lost in the Strzelecki Desert as a 10-year-old and tracked through the night. And for Moshu, born in Alice Springs with hair the colour of red desert sands.

CONTENTS

Figures and Tables page xi Preface xvii Acknowledgements xxi Note on Calibration of Radiocarbon Dates xxv

1 The Archaeology of Deserts: Australia in Context ...... 1 2 DesertsPast:AHistoryofIdeas...... 17 3 TheEmptyDesert:InlandEnvironmentsPriortoPeople...... 45 4 Foundations: Moving into the Deserts ...... 70 5 Islands in the Interior: Last Glacial Aridity and Its Aftermath . . . . 109 6 The ‘Desert Culture’ Revisited: Assembling a Cultural System . . . 157 7 Rock Art and Place: Evolution of an Inscribed Landscape ...... 212 8 The Chain of Connection: Trade and Exchange across the Interior...... 266 9 The Last Millennium: Archaeology and the Classic Ethnographies...... 302

Glossary of Technical Terms 343 References 349 Index 391

FIGURES AND TABLES

figures 1.1 The modern distribution of deserts. page 2 1.2 Australia is the driest of the six inhabited continents. 2 1.3 Global spread of modern humans into Australasia. 3 1.4 Map of the Australian arid zone. 5 1.5 Inﬂuence of the Australian monsoon on Australia’s deserts. 9 2.1 Baldwin Spencer with Arrernte elders, 1896. 19 2.2 ‘Native village in the Northern Interior’, 1845. 21 2.3 Herbert Basedow’s tracing of rock engravings, 1907. 25 2.4 Robert Keble’s 1947 map of climatic belts. 27 2.5 Jim Bowler’s 1971 map of the Willandra Lakes. 33 3.1 The 1893 digatLakeMulligan. 46 3.2 Distribution of Diprotodon optatum and Genyornis newtoni. 47 3.3 Climatic variability over the last 350 ka showing marine isotope stages (MIS). 51 3.4 Bowler’s ‘hydrological threshold’ for Australian lakes. 53 3.5 The Lake Eyre basin during the last interglacial. 54 3.6 Lake-level curve for Lake Eyre. 55 3.7 Kocurek’s model of aeolian system response to climate change. 59 3.8 The distribution and diversity of late Quaternary megafauna. 62 4.1 Map showing archaeological sites dating more than 30 ka. 73 4.2 Map of the Willandra Lakes. 81 4.3 The excavation at Mungo B in 1976. 82 4.4 The Cuddie Springs bone bed. 85 4.5 Puritjarra rockshelter. 90 4.6 Excavations at Puritjarra rockshelter, 1988. 91 4.7 The Mungo 3 (WLH3) burial. 93 4.8 Artefacts from the 35 ka palaeosurface at Puritjarra. 97 5.1 Map of the continent during the last glacial maximum. 111 5.2 Biotic and human responses during the last glacial maximum. 112 5.3 The location of glacial refugia. 113 5.4 The impact of peak glacial aridity upon land use. 123

xi xii Figures and Tables

5.5 Plot of radiocarbon dates, showing the impact of the last glacial maximum. 124 5.6 Different types of stratigraphic or occupation hiatus. 131 5.7 Map of western Central Australia. 134 5.8 Stratigraphic section, Kulpi Mara excavations. 135 5.9 Large ﬂake implements from the late Pleistocene unit at Puritjarra. 137 5.10 Map of the Pilbara. 139 5.11 Temporal distribution of stone artefacts at Djadjiling. 140 5.12 Temporal distribution of stone artefacts at Serpents Glen. 147 5.13 The sinkhole at Koonalda Cave. 151 5.14 Fossil human footprint dating to 23–19 ka, Willandra Lakes region. 155 6.1 Excavations at Puntutjarpa rockshelter in 1969–70. 159 6.2 Summed probability plot for all radiocarbon ages from archaeological sites in Australian drylands. 161 6.3 Summed probability plot showing radiocarbon dates on Terebralia and Anadara shell middens. 171 6.4 Excavations at the Skew Valley midden, 1975–76. 173 6.5 Stratigraphy of the Skew Valley midden. 173 6.6 Stratigraphic section for Allens Cave. 177 6.7 Small-tool phase artefacts from Puritjarra rockshelter. 186 6.8 The spatial and temporal distribution of geometric microliths. 187 6.9 The distribution of tula adzes, pirri points and millstones. 190 6.10 Unifacial pirri points from the Lake Eyre basin. 191 6.11 Seed-grinders from Central Australia. 199 6.12 The distribution of Australian language families. 203 7.1 Panaramitee-style rock engravings at Florina station, Olary region. 213 7.2 The distribution of graphic and religious systems across Australia. 215 7.3 Denis Ebaterinja drawing the honey ant ‘dreaming’. 218 7.4 Large striped totemic designs at Emily Gap, Central Australia. 221 7.5 Desert-style provinces for mid-Holocene rock engravings. 229 7.6 Panaramitee-style rock engravings at Puritjarra. 230 7.7 Wanga East rockshelter, showing engraved rock slabs and location of dated samples. 237 7.8 Excavation of engraved boulders at Puritjarra. 239 7.9 Early petroglyph assemblages in the Dampier–Burrup area. 242 7.10 The ‘climbing men’ motif. 243 7.11 Archaic face engravings. 245 7.12 The Burrup rock art sequence. 249 7.13 The sequences of changes in Central Australian rock art. 255 7.14 The long painted frieze at Puritjarra. 256 7.15 Hand stencils and hand prints. 257 7.16 State-and-transition model applied to Panaramitee-style rock art. 265 8.1 The ethnographic exchange system in relation to major pituri groves and quarries for red ochre. millstones and stone axes. 267 8.2 Map showing the southern sector of the Lake Eyre basin. 268 8.3 Map showing major quarries and mines mentioned in the text. 269 8.4 Pearl-shell pendant with interlocking key design. 275 Figures and Tables xiii

8.5 The Wilgie Mia red ochre mine in about 1910. 278 8.6 The structure of the Wilgie Mia mine. 279 8.7 Survey plan of Narcoonowie grindstone quarry. 285 8.8 Australian stone axe or hatchet, 1861. 288 8.9 The chronological distribution of ground-edge axes. 289 8.10 Hafted stone knife with Melaleuca ‘paperbark’ sheath. 295 8.11 Model showing number of exchange contacts available to hunter-gatherer groups with increasing population density. 297 8.12 Mound of kopi mourning caps on a grave, Eyre Creek, central Australia. 299 9.1 A fire drive to hunt maala wallabies, Musgrave Ranges, 1933. 303 9.2 Time-series distribution of radiocarbon dates over the last 5,000 years. 311 9.3 Map of Central Australia. 314 9.4 Schematic diagram illustrating different site histories. 315 9.5 Stratigraphic profile for Tjungkupu 1, Central Australia. 315 9.6 Stratigraphy in Trench 1, Intirtekwerle rockshelter, Central Australia. 317 9.7 Principal components analyses of site inventories in Central Australia. 319 9.8 Excavations at Glen Thirsty 1, Central Australia, in 2004. 321 9.9 Changing foraging patterns, comparing the early and late Holocene. 328 9.10 Geographic distribution of the Western Desert language (Wati). 335 9.11 Syd Coulthard at Glen Thirsty, with rock paintings described by E Giles in 1872 as ‘Roman numerals’. 337 table s 3.1 Lake Eyre, comparing the size of the last interglacial lake and the largest historical filling 55 3.2 List of taxa, Upper Katapiri fauna (MIS4–MIS6), Cooper Creek–Lake Eyre region 63 3.3 Comparison of middle Pleistocene herbivore dietary guilds across Australia 64 3.4 List of fossil fauna from Lake Callabonna 64 4.1 Archaeological sites in Australian deserts and drylands dating 30–50 ka (chronology and distribution) 79 4.2 Archaeological sites in Australian deserts and drylands dating 30–50 ka (assemblages and site inventories) 95 4.3 Archaeological sites in Australian deserts and drylands dating 30–50 ka (subsistence remains from levels >30 ka) 100 5.1 Relative importance of dryland ranges and gorge systems as biological refugia 118 5.2 Archaeological sites in Australian deserts and drylands dating 30–12 ka 125 xiv Figures and Tables

5.3 Key archaeological sites where there is substantive evidence for (a) a stratigraphic or occupational hiatus during the last glacial maximum, or (b) for continuing occupation in this period 132 5.4 Changes in intensity of occupation at Puritjarra 138 5.5 Comparative data on the size and diversity of selected assemblages, dated to the last glacial maximum 142 6.1 Archaeological sites in Australian deserts and drylands dating 12–4 ka 166 6.2 The Puntutjarpa sequence 180 6.3 Comparative data on the size and diversity of selected assemblages, contrasting mid-Holocene (8–6 ka) and late Holocene levels (<4 ka) 181 6.4 Radiocarbon dates for initial appearance of backed implements and adze flakes by region 188 6.5 Typological changes in flaked artefacts and ground-stone assemblages at Puritjarra rock shelter 192 6.6 Inventory of wooden artefacts used by ethnographic groups in Central Australia 194 6.7 Radiocarbon dates for initial spread of the dingo (Canis lupus dingo) across Australia 206 6.8 Radiocarbon dates for the last Thylacines (Thylacinus cynocephalus) on the Australian mainland 207 7.1 Recent production of traditional rock art in Central Australia and Western Desert 219 7.2 Stratigraphic and direct dates for rock engravings in the arid zone 223 7.3 Estimates of net rock art production in Central Australia, comparing different time periods 224 7.4 The composition of Panaramitee-style engraved assemblages 231 7.5 Relative stylistic sequence for Central Australia 232 7.6 Key characteristics of Panaramittee-style rock engravings as a graphic system 233 7.7 Radiocarbon ages bracketing the age of Panaramitee-style rock engravings, Central Australia 236 7.8 Composition of rock art assemblages in the Pilbara, comparing the Dampier–Burrup coast and inland Pilbara 241 7.9 Relative stylistic sequence for petroglyphs in the Burrup region 248 7.10 Variation in patination for selected motifs in Burrup rock art 248 7.11 Testing the style sequence for Burrup rock engravings 250 7.12 Comparison of older and younger assemblages of rock engravings, Central Australia 254 7.13 Matrix of similarities between rock art assemblages in Central Australia, using Euclidean distance coefficient 257 8.1 Radiocarbon dates for baler shell (Melo sp.) in the Great Sandy Desert 274 8.2 Production estimates for arid-zone grindstone quarries 283 8.3 Sites with axe fragments in dated contexts in the arid zone 291 Figures and Tables xv

9.1 List of excavated late Holocene archaeological sites in Central Australia 313 9.2 Excavated assemblages at Tjungkupu 1, comparing the major late Holocene occupation unit and underlying levels 316 9.3 Changes in number and distribution of rock art sites in Central Australia 319 9.4 Habitat mosaic created under different ﬁre regimes in the Western Desert, 2002 325 9.5 Types of plant food in Central Australia, showing the number of species promoted by ﬁre 325 9.6 Index of fragmentation values (IOF) for faunal bone assemblages from archaeological sites in the Western Desert and Central Australia 327 9.7 Changes in hunting patterns shown in faunal assemblages at Puntutjarpa rockshelter, Western Desert 328 9.8 Excavated grindstone assemblage from Intirtekwerle (James Range East), Central Australia 329 9.9 Diyari drought terminology 333

PREFACE

Whatever else it may be, a desert landscape is a historical document preserving a complex record of the interaction of past climates, geomorphic processes and cultural systems. I like to think of these landscapes as a palimpsest of different deserts. Stratified in time, stacked one above another, each has its own climate, physical landscapes and environments; each its own social landscapes and people, places of association and belonging, territories, resources and itineraries. Some features of earlier deserts project through these layers to become part of the fabric and cultural geography of later deserts. Structural features and processes are held in common: wind and water shape landforms; the basin and range topography provides the formwork of the landscape. No one desert is erased entirely by succeeding deserts – a fact that makes archaeology possible. This monograph – the first book-length archaeological study of Australia’s deserts – is an attempt to map out these histories. The past four decades of fieldwork in Australia’s deserts and drylands has been an intense period of discovery, and a synthetic work that brings together the results of this research is sorely needed. This period has seen the duration of human settlement extended from 10,000 years to more than 40 millennia. At the same time, there has been a quantum growth in our understanding of the Quaternary history of arid Australia, fuelled in part by the availability of new dating techniques – such as thermoluminescence (TL), optically stimulated luminescence (OSL), electron spin resonance (ESR), amino acid racemisation (AAR), and accelerator mass spectrometry radiocarbon dating methods (AMS 14C) often in conjunction with acid–base oxidation (ABOX) pretreatment. (For a description of the various dating techniques, and an explanation of terminology and abbreviations, see the Glossary.) There has also been an exponential increase in knowledge about the last 1–2 millennia of hunter- gatherer settlement and society; for the first time, this has provided a solid basis for examining the prehistory of the classic desert hunter-gatherer societies, well known in the nineteenth-century ethnographies of Baldwin Spencer and FJ Gillen. xvii xviii Preface

At the outset, I should explain that this book is primarily a work of historical review and synthesis; it revisits basic questions about the deep history of Australia’s deserts, their Ice Age settlement and the development of Aboriginal societies in these regions. For me, the ‘big picture’ still has an important role in archaeological narrative. Theory and methodology are intrinsic to this book but are not my main concern here. I prefer, as historian John Ferry once said, ‘to be nudged by theory rather than dominated by it’ (1999: 14). In any case, methodological issues are more properly part of the internal dialogue of the discipline, best fought out in the journals. I draw on rangelands ecology, hunter-gatherer theory and mathematical modelling where it helps me understand archaeological data and have taken into account arguments about the structure and biases in the archaeological record, without making these discussions ends in themselves. However, a broad review – such as this – necessarily conflates vast stretches of time, unavoidably glossing over some of the procedural difficulties presented by the uncertainties of radiocarbon calibration, questions of site formation processes and issues of taphonomy. Nor have I structured this book as a catalogue of old battles about specific models and hypotheses. I think it is important (occasionally) to step back and ask how the data might fit together as a natural and cultural prehistory of Australia’s deserts. I will be delighted if this book gives impetus to fresh research. Part of its job is to frame new questions about desert prehistory that can be tested in the field, to point to frustrating gaps in basic fieldwork, to identify targets for critical application of theory or methodology and to encourage others to complete the crucial excavation and site studies that make up the building blocks of our field. I hope this book does all of these things and (after a polite interval) is dismantled by new thinking and finer-grained data. Any review is limited by the extent of previous work. I have found that for some questions, the data simply do not exist (I would have liked, for example, to say more about the interactions between desert people and the societies surrounding the arid zone). For others, the data exist but are yet to be collated (most regions have abundant traces of late Holocene occupation, but fine-grained regional reviews are few, limiting my ability to approach the desert as a mosaic of different histories). Where it seemed that loose ends could be easily tidied up through new fieldwork or additional radiocarbon dates, I have gone and done this (for example, the potential harvest rates in the Mulligan River pituri groves and the additional radiocarbon dates for Kulpi Mara). Working through excavation reports, I have been conscious of the need to stay as close to the excavator’s interpretation of chronology, sedimentary history and stratigraphy as I dare. But where I found basic errors, I have run with my own interpretation. Some published reports do not answer basic questions about the character of an occupation, artefact assemblages and site inventories, or provide workable site chronologies. In these cases, I made my own assessment (here, I am grateful for the generosity of colleagues who Preface xix answered requests for unpublished field data). Many studies use only basic age-depth plots to establish a chronological framework for a site. I have stayed close to the published version, unless I felt the evidence contradicted it. I need to add a word on the structure of this book. The chapters represent a chronological series, but their coverage in space and time shrinks. Late Pleistocene deserts were larger than those today – and my review follows the contracting drylands. It will be no surprise to learn that we have a much finer-grained record of the last millennium (Chapters 7–9)thanwedoforthe immense period from 45 ka to 24 ka (Chapter 4), so later chapters necessarily deal with narrower time slices. Chapter 2 explores the various ways that the past of these desert landscapes and desert societies has been interpreted since colonial explorers and scientists first encountered them, and points to recurring themes that are picked up in later chapters. Individual chapters each explore a major issue in desert research: the changing nature of the desert environment, the timing of initial human settlement, the impact of the last glacial maximum on desert settlement, postglacial adaptations, rock art and religion, prehistoric trade and exchange and the social history of classic ethnographic societies. Each chapter opens with a key moment in desert research (such as the discovery of the Lake Callabonna Diprotodon fossils); collectively, these vignettes build to provide a pen portrait of the region and its people. Each chapter also begins with a review of issues and ideas, presented as a framework for exploring the archaeological evidence. Structure and agency are important concepts for me: throughout the book, I focus on how changes in palaeoenvironment affected the structure of the landscape in which these hunter-gatherer groups operated – and how human actions in turn modified these environments. I am drawn to the idea of landesque capital, proposed by economic historian HC Brookfield, because it forces us to think about long-lived economic infrastructure, in this case, the wells, soakages, plant transpositions, fire mosaics and also (in a way) rock art that improve the amenity of a hunter-gatherer landscape. One of my aims has been to explore as far as possible changes in the political and social structure of these societies – and this book goes some way towards describing a prehistory that contains more than just economy and environment. I am not blindly committed to the idea that population growth is the primary determinant of all social change, but I do find the evidence to be more consistent with the view that the structure of these hunter-gatherer societies and their capacity to mobilise resources and actors are emergent properties of their demography. In this regard, I see economy as architecture and social life as agency. Throughout the book, I explicitly draw on ethnography, linguistics and Aboriginal history wherever they complement an ‘archaeology of place’. Pre- historic societies were not cast in an ethnographic mould, but ‘tracking back’ is one method that can help establish when they took on something approaching their classic forms. The wealth of ethnographic data deserves wider critical xx Preface

attention by archaeologists, and I think this can be done without archaeology necessarily becoming ‘ethnography with a shovel’. I deliberately thread an Aboriginal ‘voice’ throughout the book. It is important, I think, to retain a feeling for the contemporary cultural landscape that swirls around these prehistoric sites. Finally, I must say that the idea for this book began on a bush trip with Clive Gamble to Puritjarra rockshelter in 2001. With his wonderful book, The Palaeolithic Societies of Europe (1999), rolled up in my swag, I wondered then whether something similar might yet be done in Australia.

Centre for Historical Research National Museum of Australia, Canberra February 2012 ACKNOWLEDGEMENTS

The National Museum of Australia is a remarkable institution, a gem of a place to work. I am grateful for the time and intellectual space to tackle this book and the unobtrusive support of past and present directors: Dawn Casey, Craddock Morton and Andrew Sayers. I am fortunate to have a lively base in the Museum’s Centre for Historical Research (CHR). I thank CHR Director Peter Stanley for his support and Libby Robin, who in many ways mentored this project. An affiliation with the Fenner School of Environment and Society (where I am adjunct professor) gave me crucial access to the e-library resources of the Australian National University. Staff at the Museum’s own library also tracked down many obscure journal articles – with characteristic elan´ and efficiency. The process of writing this book was an adventure in step with further field surveys, excavations, 4WD expeditions and some marvellous desert traverses with a string of pack camels – all of which helped plug gaps in individual chapters. Sitting in my study in the expectant hours before dawn, when ideas seem to flow best, I found it a revelation to re-read all the primary research papers, student theses and consultancy reports. But ‘full immersion’ made for slow writing, and this book took far longer than I had imagined. And so I thank Norman Yoffee and Beatrice Rehl at Cambridge University Press for their patience. Some sections of the text were initially written for other publications. I use them here (in a modified form) because they still best reflect my thinking on an issue. Synthetic books always rely on the goodwill of colleagues. I wish to thank my ‘republic of letters’, those who sent me unpublished data; copies of theses or hard-to-get publications; took me to sites; gave me directions; shared the dust, flies and fatigue of fieldwork; took aerial photographs for me; or simply answered queries and shared ideas as this book took shape: Kim Akerman, Max Aubert, Alex Baynes, Claire Bowern, Jim Bowler, David Brooks, Adam Brumm, Brian Codding, Tim Cohen, Syd Coulthard, Bruno David, Iain Davidson, Nick Evans, Pat Faulkner, Judith Field, Dick Gould, xxi xxii Acknowledgements

Richard (Dizzy) Gillespie, RG (ben) Gunn, Giles Hamm, Luise Hercus, Paul Hesse, Peter Hiscock, Tiger Morris Japaljarri, Diana James, Ken Johnson, Philip Jones, Luke Keogh, Dick Kimber, Roger Luebbers, John Magee, Ben Marwick, Isabel McBryde, Scott McConnell, Pat McConvell, Jo McDonald, John McEntee, Gifford Miller, Mark Moore, Kate Morse, Ken Mulvaney, Warren Murgatroyd, Gerald Nanson, Lewis Kyle Napton, David Nash, Colin Pardoe, Sheila van Holst Pellekaan, Brad Pillans, Gavin Prideaux, Richard Robins, Andre´ Rosenfeld, June Ross, Don Rowlands, Jen Silcock, Moya Smith, Mark Stafford Smith, David Taylor, Peter Thorley, Douglas Multa Tjakamarra, Sean Ulm, Peter Veth, Glenda Wardle, Alan Watchman, Esmee Webb, Steve Webb, Rod Wells, Bernhard Weninger, Allan Whiting, John Wis- chusen, Alan Williams, Elizabeth Williams, Martin Williams, Trevor Worthy, Boyd Wright and Andrew Harper and the cameleers and travellers with Aus- tralian Desert Expeditions. I especially thank Barry Cundy, June Ross, Alan Williams and Manik Datar for sharing the sheer adventure of the enterprise. The grey literature contains a wealth of information, and the following generously made consultancy reports or student theses available to me: Eliza- beth Bradshaw; ben Gunn; Fiona Hook; Phil Hughes; Ken Mulvaney; Rio Tinto Australia; Birgitta Stephenson; Lynda Strawbridge; the Department of Archaeology and Palaeoanthropology, University of New England; and the Department of Archaeology, University of Western Australia. The artefact distribution maps in Chapter 6 are more detailed than usual because they draw on field diaries from my 30 years as a field archaeologist in the arid zone – and on Kim Akerman’s unparalleled knowledge of these things. I appreciate the help of colleagues who read and commented on the manuscript: David Brooks, Sarah Dunlop, Guy Fitzhardinge, Marg Friedel, Tom Griffiths, Paul Hesse, Dick Kimber, John Kinahan, Isabel McBryde, Steve Morton, John Mulvaney, Jim O’Connell, and Libby Robin and my feisty, iconoclastic Manik – her love of the craft of writing augmented my own interest in words. It takes a surprising amount of work to turn a manuscript into a book. In Canberra, Robert Nichols expertly edited the manuscript and Denise Sutherland created the index. Kay Dancey (ANU Cartography) drew the major base maps, and Alan Williams produced the time-series radiocarbon plots. Most other figures are from my own hand. At the National Museum, Anne Faris and Almaz Berhe helped clear copyright in the photographs. For permission to reproduce photographs in this book, I thank Peter Eve (Monsoon Studio), Museum Victoria (Spencer and Gillen collection), the South Australian Museum (EC Stirling collection), Wilfred Shawcross, Robert Edwards, Richard Robins, the Australian Institute of Aboriginal and Torres Strait Islander Studies (BJ Wright collection), Michel Lorblanchet, June Ross, the Western Australian Museum (Wilgie Mia collection) and the Northern Acknowledgements xxiii

Territory Archives – Alice Springs (HH Finlayson collection). Darwin pho- tographer, Peter Eve, graciously allowed me to use one of his striking desert images on the cover of this book. Finally, I must thank Mike and Jan Sexton for their ‘philanthropy without strings’; their generosity allowed me to chase ideas rather than money.

NOTE ON CALIBRATION OF RADIOCARBON DATES

Almost all archaeological research in Australia’s deserts has relied on an uncorrected radiocarbon chronology. However, calibration of the radiocarbon timescale is now unavoidable, if archaeological records are to be fully integrated with palaeoenvironmental and climate data, or with global marine isotope stages (MIS). Few archaeologists today would be unaware that ﬂuxes in atmospheric 14C have distorted the radiocarbon timescale, so that radiocarbon years are not calendar ages. However, calibration programs to convert radiocarbon ages (years BP) to sidereal years (years cal. BP) over the full span of the last 50,000 years have only recently become available. In this book, I have attempted to put everything on a common timescale by working primarily with calibrated ages (quoted as ‘ka’ or ‘years cal. BP’). Ages were calibrated using OxCal 4.1 and the INTCAL09 dataset. Marine09 has been used for shell dates, applying a marine reservoir correction of 450 ± 35 years, plus a regional offset delta-R correction 70 ± 70 for samples younger than 8 ka. For samples older than 8–7 ka, shifts in sea levels and currents may have affected rates of mixing, so the delta-R correction is effectively unknown. For these samples, I rely on the base correction of 450 ± 35 years, widely used in reporting Pleistocene shell dates from archaeological sites. The SHCal04 data set and calibration curve allows for a Southern Hemisphere offset of 55–58 years back to 11,000 BP. A full listing of the original radiocarbon ages for Australian desert sites is available in the AustArch1 database, at http:// palaeoworks.anu.edu.au/databases.html. For most archaeological use, a centre-point age estimate is more convenient than the unwieldy, non-normalised, 2SD age ranges routinely generated by calibration programs (even if these are technically more correct). Following Telford et al. (2004), I use the median of the calibrated age-distribution (median IntCal09 age ± 1SD), which is more robust than mode or intercept methods and not as sensitive to small changes in the calibration curve. Throughout this book, calendar ages are quoted as ‘ka’ (thousands of years ago) and rounded up, to avoid implying more chronological precision than the context allows. xxv

CHAPTER 1

THE ARCHAEOLOGY OF DESERTS: AUSTRALIA IN CONTEXT

On and on, stepping it out Across the wide country, Across the plain. On and on, stepping it out Across the wide country In the heat of the day. ‘Travel Song’ by Wajurrpirdi, original in Ngarluma (von Brandenstein and Thomas 1974: 44)

Of the six inhabited continents, Australia is by far the driest, with large internal areas of desert that make up nearly 70 per cent of the landmass. This immense region of duneﬁelds, ephemeral rivers, saltlakes and desert uplands forms one of the largest desert biomes in the world (United Nations Envi- ronment Programme [UNEP] 2006), comparable in size to Arabia or Central Asia; only the Sahara is larger (Figures 1.1 and 1.2). In this book, I have set out to review what we know of the archaeology of Australia’s deserts. At one level, this is a record of prehistoric settlement and adaptation in one of the world’s major desert regions. On another level, it is also a history of some distinctive Aboriginal societies. Unlike the deserts of Africa, Asia and the Amer- icas, the precontact history of Australia’s deserts is exclusively a history of the autonomous development of hunter-gatherer communities descended from the original late Pleistocene settlers of the continent.

POSITIONING THIS RESEARCH Archaeological research in these drylands has global as well as regional dimensions, and it is worth setting these out at the outset. The ﬁrst point to make is that the peopling of Australia’s deserts represents one end of an arc of dispersal involving the diaspora of early modern humans from Africa (Balme et al. 2009). Exploration and colonisation of deserts and drylands appears to have been part and parcel of this movement. McBrearty and Brooks (2000) 1 2 The Archaeology of Australia’s Deserts

Figure 1.1. The modern distribution of deserts. Based on Global Deserts Outlook (UNEP 2006). Australia’s deserts form one of the world’s major desert regions, as well as the largest desert biome in the Southern Hemisphere.

Figure 1.2. Australia is the driest of the six inhabited continents. Just how dry can be seen from the ﬁgures for rainfall and river discharge for each continent. Annual precipitation is averaged over the continental area. River discharge is the average annual discharge converted to equivalent rainfall (mm). Australia’s deserts represent the extreme end of a largely dry continent but do not include hyper-arid areas comparable to the Sahara, Namib or Atacama deserts. (Source: Australian Bureau of Meteorology 2008: 45) The Archaeology of Deserts: Australia in Context 3

Figure 1.3. The southern route for spread of modern humans out of Africa and into the Australasian region, showing the relationship to world deserts. Map courtesy of S. Oppenheimer (2012:fig.3). suggest that Aterian hunter-gatherers settled in the North African deserts from 110 ka down to 60 ka, as one prong of the dispersal of behaviourally and anatomically modern humans (see also Garcea 2004; Barton et al. 2009). The spread of modern humans along the southern margins of Eurasia 80– 60 ka brought people to the northern approaches of a prehistoric continent made up of the modern landmasses of Australia, New Guinea and Tasmania (Figure 1.3). The settlement of this continent, around 50 ka, involved two remarkable events: first, the crossings of the Timor Sea and the straits east of the Moluccas to establish separate landfalls on the Australian landmass; and second, the move south from the northern savannas into the deserts and drylands that form the greater part of the continent and represent one of its most distinctive biomes (Byrne et al. 2008). Some 40 years ago, Golson (1971) assembled botanical evidence showing that early colonists from Southeast Asia would have found a range of familiar plant food species in the swamps and woodlands of northern Australia. However, any move inland into the deserts and drylands involved significant adjustments to a new biota. There are questions about how rapidly this move occurred and whether there was a significant time lag between entering the continent and moving into the interior. 4 The Archaeology of Australia’s Deserts

Another theme is the history of the drylands themselves. Australia’s deserts have not always taken their modern form: aridity has intensiﬁed or relaxed, the deserts have expanded and contracted along their margins and the rivers and lakes of what is now the ‘dead heart’ have been periodically reactivated. The Quaternary environmental history of these regions is part of a global story of climate change across the last glacial–interglacial cycle. These climatic shifts inevitably affected the ecological baseline for hunter-gatherer groups in the region. One question is whether the interior of Australia was initially settled when the lakes and rivers that ring the heart of the continent were more active than they are today. Another is whether intense aridity during the last glacial maximum led to abandonment of large parts of the continental interior and fragmented the social and economic world of inland groups. The quantum growth in understanding of the Quaternary history of arid Australia – fuelled in part by the availability of new dating techniques such as optically stimulated luminescence (OSL), uranium series (U/Th) and acid– base oxidation radiocarbon dating (ABOX 14C) – now allows a fresh assessment of the environmental basis for prehistoric settlement in this region to be made. Any archaeology of Australia’s deserts is necessarily also a cultural history. Archaeologists working in Central Australia operate in the shadow of the classic ethnographies by Spencer and Gillen. The Native Tribes of Central Aus- tralia (1899)andThe Northern Tribes of Central Australia (1904) were founding works of social anthropology and provided prehistorians in Europe with ethnographic analogues for Middle or Upper Palaeolithic societies in West- ern Europe (Reinach 1903; Sollas 1911; Lubbock 1912). For instance, Sollas famously dubbed the Australians ‘the Mousterians of the Antipodes’ (1911: 170). Although few social anthropologists after Spencer or Sollas accepted the argument that Aboriginal society preserved elementary or primordial social traits, contemporary foraging groups have often been used as economic analogues for early hunter-gatherer societies on the basis, as Testart puts it, that they preserve ‘the most archaic way of life known to humanity’ (1988: 1). Archaeological research now complicates this picture by showing that these desert societies have their own histories, and that late Pleistocene societies were not cast in the ethnographic mould. When did life in the desert change to something resembling the ethnographic pattern? One of the concerns of this book is to reconstruct how these societies were assembled over time.

AUSTRALIA’S DESERTS There are large bands of deserts and drylands along the tropics in both northern and southern hemispheres (Middleton and Thomas 1997). Australia’s deserts form the major block of arid lands in the southern hemisphere; taken as a whole, they constitute a classic mid-latitude desert. The region’s aridity is a The Archaeology of Deserts: Australia in Context 5

Figure 1.4. The Australian arid zone, showing major geographic features: (1)Pilbara; (2) Western Desert; (3) Central Australia; (4) Arid rivers region; (5) Nullarbor Plain; (6) Darling–Paroo river system. The boundary of the region follows Morton et al. (2011). product of large-scale subsidence of dry, stable air along the tropics, enhanced by the size of the continent. The geographic boundaries for this block of Australian deserts and drylands are poorly deﬁned – and it is well to remember that modern climate indices are likely to be a poor guide to the extent of these deserts in the past. I use the boundary adopted by Morton et al. (2011). Referring to the standard UNEP aridity index P/PET (a ratio of precipitation to potential evapotranspiration), Morton et al. take a P/PET value of less than 0.4 as the boundary of arid Australia (see also Stafford Smith and Cribb 2009)(Figure 1.4). This includes areas that, on climatic criteria, would be classiﬁed as arid and semi-arid (UNEP aridity index: semi-arid P/PET 0.20– <0.50;arid0.05–<0.20), but aligns better with eco-regions where biota have clear adaptations to arid conditions (see Thackway and Cresswell 1995). The 6 The Archaeology of Australia’s Deserts

coverage of this book mostly deals with an arid core, that is, the 3 million km2 area that today would be formally classiﬁed as arid on UNEP criteria. However, the wider boundary used by Morton et al. suits my purposes because it also represents the maximum extent of the arid zone during the climatic conditions that prevailed during the late Pleistocene. Throughout this book, I use the terms ‘desert’ and ‘arid zone’ interchangeably to refer in general terms to the (shifting) arid core, and ‘desert margins’ to refer to the semi-arid or dry-subhumid fringes of the desert; and I refer to ‘Australia’s deserts’ whenever I need to remind myself that the ‘desert’ is not a uniform region. Diversity is an important theme here. ‘Deserts’, observed Isaiah Bowman, Director of the American Geographical Society, ‘are no more alike than mountains or plains’ (1924: 60). This is true in the Australian context, where patterns of rainfall, soil fertility and physical geography vary across the arid zone and frame a range of different opportunities for people. The zone of greatest aridity centres on Lake Eyre – a huge saltlake of 9,690 km2 – where annual rainfall averages about 100 mm. Out from this, average annual rainfall increases from 250 mm in the south to 500 mm in the north. The greater part of the interior forms an arid core of 3 million km2, where P/PET is less than 0.20 (meaning that rainfall makes up less than 20 per cent of the potential loss of moisture through evaporation). This is surrounded by a wide belt of semi-arid country, adding another 2–2.5 million km2. The named deserts tend to be major duneﬁelds or stony (‘gibber’) plains, but the arid zone encompasses a wider range of regions. We can list the more important of these, moving from west to east across Figure 1.4:

1. The Pilbara in northwest Australia is a zone of uplands and ranges, between the Fortescue and Ashburton Rivers. Importantly, this region has both an inland and a coastal component and connects the desert interior with an arid west coast that extends from the Dampier region, to Northwest Cape, and on to Shark Bay in the south. 2. In the western half of the arid zone, three sandy deserts (Great Sandy Desert, Little Sandy Desert and Great Victoria Desert) are grouped with the Gibson Desert (an expanse of lateritic gravel plains) and collectively called the ‘Western Desert’ (Gould 1977: 8). These deserts are part of a continental swirl of duneﬁelds that ring the heart of the continent (Hesse 2010). Most of the dunes here are linear sandridges less than 15 m high and are often stable and vegetated. More than 1.6 million km2 of the Australian arid zone is covered by aeolian sand, either sandplain or duneﬁelds. All of the sandy deserts also contain small range systems, with rockholes, gorges and run-on areas that provide focal points for human use of the region. 3. The Central Australian ranges form a large block of desert uplands in the heart of the continent – with the MacDonnell Ranges in the north The Archaeology of Deserts: Australia in Context 7

and the Mann-Musgrave Ranges south of Lake Amadeus – dubbed ‘the Red Centre’ by HH Finlayson (1936) for its crimson, deeply oxidised rock faces and red sands. Geographically, this is a plateau 500 mabove sea level, broken by the central ranges, which strike east–west across the region in an almost unbroken belt more than 400 km long and up to 160 km wide, separated by narrow sandplains or alluvial valleys. The northern ridges of the MacDonnell Ranges are rugged uplands of crystalline and metamorphic rocks. Local relief commonly exceeds 300 m, and Mt Liebig (1,267 m), Mt Zeil (1,531 m) and Mt Sonder (1,380 m) are the highest points in Central Australia. The major river systems of Central Australia – such as the Finke River – have their headwaters in these ranges, and this is a starkly beautiful landscape with dark red gorges and gum-lined river channels of white sand and gravel. The Central Australian ranges also contain the greatest concentration of rare, relict and endemic plant species in the arid zone, underlining their importance as the major biogeographic refugium in the desert. 4. The eastern desert – the ‘arid rivers’ region – is a large ephemeral riverine system that includes the catchments of the Mulligan and Georgina Rivers, and the Diamantina–Warburton and Cooper Creek–Barcoo drainages (Robin, Dickman and Martin 2010). These arid rivers intermittently transfer large flows of slow-moving floodwaters from the desert margins towards Lake Eyre, which is the terminus for this vast inter- naldrainagesystem(Kotwicki1986). Unlike the sand-bed channels of Central Australia, this is a muddy, in places anastomosing, channel system. During the 2009–10 La Nina,˜ shallow floodwaters reactivated a network of channels, swamps and floodplains, gradually inundating an area larger than Germany. In the lower part of the catchment, these intermittent flows create a sharp contrast between the transient richness of the floodplains and some of the most arid sandridge country in Australia, giving the eastern desert quite a different character to the Western Desert and Central Australia. Towards Lake Eyre, increasing aridity and desolation – as well as the prominence of apparently fossil lakes and palaeochannels – led JW Gregory to label this region ‘the dead heart of Australia’ (1906). 5. Among this set of deserts, the Nullarbor Plain on the southern margin of the continent has a unique character. This is a vast karst plain, forming a treeless chenopod steppe. Surface water is absent, although solution tubes form natural cisterns in places. From a human perspective, the most habitable part of the region is a narrow strip of arid woodland sandwiched between the open plain and sea cliffs up to 100 m high; it is along this southern margin that massive limestone caves and dolines preserve some of the deepest and best stratified archaeological sites in Australia. 8 The Archaeology of Australia’s Deserts

THE ECOLOGICAL BACKGROUND Plant and animal communities in inland Australia have been ‘shuffled and reassembled through filters of increasing aridity’ throughout the Quaternary (Morton et al. 2011: 315) and have adapted to a landscape poor in basic nutrients such as nitrogen and phosphorus. However, on shorter ecological timescales, Stafford Smith and Cribb (2009) point out that deserts are characterised not just by scarcity but also by transient richness. Today, Australia’s deserts are only moderately arid and mostly well vegetated, with large areas of spinifex (Triodia) hummock grassland and arid woodlands. But they experience extreme variation in rainfall and biological production. In some respects, the desert is an unstable extension of the savanna, shrublands and grasslands surrounding the region. Part of the reason for the unexpectedly high perennial biomass is that rain-bearing systems occasionally extend deep into the interior. Rainfall in Australian drylands is influenced by several weather systems, all of which attenuate towards the centre of the continent: the Australian summer monsoon in the north, tropical cyclones in the Pilbara and northwest, and the winter westerlies along the southern margins of the desert (Sturman and Tap- per 2006). The monsoon is the most important of these – and the Quaternary record of the desert is largely a history of shifts in the reach of monsoon rainfall into the interior and changes in the strength of this circulation system (Figure 1.5). The net effect of these circulation patterns is infrequent and unpredictable large rainfall events separated by long dry spells that sometimes last for several years. Although variability in rainfall is a generic characteristic of most deserts, Van Etten (2009) showed that the summer rainfall zone of Australia’s deserts (north of 27◦S) has extreme interannual fluctuations in rainfall, greater than most other desert regions. For instance, Alice Springs, with a median annual rainfall of 259 mm, received only 54 mm of rain in 1985 but a staggering 903 mm in 1974. The region also experiences extremes in temperature: at 69.3◦C, an area near Winton was the hottest area on Earth in 2003 (Mildrexler, Zhang and Running 2006). In the hummock grasslands of Central Australia and the Western Desert, wildfire is an important circuit-breaker in these ecosystems. Fires follow the buildup of fuel loads in good years, recycling scarce nutrients and removing climax vegetation. However, in the arid rivers region east of Lake Eyre, wildfires are less frequent (Morton et al. 2011: fig. 4) and here, episodic flooding is probably at least as important in redistributing nutrients and triggering biological activity. In general terms, the spinifex ecosystems of the western half of the desert are rejuvenated by fire, and those in the eastern part of the desert are rejuvenated by water (in the form of extensive floods). Despite the influence of erratic and intermittent rainfall on the ecology of these deserts, research has also shown that it is too simplistic to characterise The Archaeology of Deserts: Australia in Context 9

Figure 1.5. Inﬂuence of the Australian monsoon on Australia’s deserts. Bold line shows the southern limit of the summer rainfall zone (where rainfall in October–March is >3:1 rainfall in April–September). Shaded area shows zone of highest variability in rainfall (1.25–2.0: calculated as 9th decile–1st decile/median). Isohyets show mean annual rainfall. Variability is greatest at the limits of the monsoon. (After Morton et al. 2011:ﬁg.3)

biotic activity in these deserts as entirely a pulse-and-reserve system, whereby pulses of plant growth are separated by long dormant periods. This is because the high standing biomass and low turnover of plant matter also provide the basis for food webs between booms. Because infertile soils promote longer leaf lifespans and tough unpalatable vegetation, the major herbivores in Australia’s deserts are termites and ants consuming plant detritus and exudates. These invertebrate communities are comparatively stable and provide the basis for persistent guilds of consumers: reptiles, birds, insectivorous rodents and marsupials. These in turn provide an array of persistent resources for Aboriginal people, especially in the form of lizards and grubs. In the global context, the density and diversity of reptiles in these environments is unusual: Nosepeg Tjupurrula memorably described the abundance of varanid lizards in Pintupi country as making it like a ‘poultry run’ (Kimber 1988: 48). 10 The Archaeology of Australia’s Deserts

THE DESERT’S PEOPLE Historically, the Australian desert had some of the lowest population densities on record for human populations (as low as one person per 100–200 km2). At the time of European contact, somewhere between 60,000 and 100,000 people occupied Australia’s deserts. Berndt (1959) estimated that the precontact population of the Western Desert was between 10,000 and 18,000 people. For Central Australia, a figure of between 10,000 and 12,000 people seems probable, given population estimates for various areas. The highest population densities were in the ranges and river corridors – ranging from about 1 person per 13 km2 in the best parts of the Central Australian ranges (Strehlow 1965, 1970), through 1 person per 50 km2 in other parts of the ranges (Yengoyan 1968; Layton 1986), to 1 person per 90–200 km2 in the sandy deserts (Meggitt 1962; Long 1971; O’Connell, Latz and Barnett 1983). Reviewing cultural divisions across the continent, Peterson (1976) found that although arid-zone groups were distinct from those in other areas, the desert was not a single-culture area. The Western Desert groups were the most socially inclusive and fluid societies, including people we know today as the Martu, Ngaatjatjara and Pintupi (Sutton 2003). Some culturally related groups, such as the Pitjantjatjara and Yankunytjatjara, also occupied range country in the Mann-Musgrave Ranges. There was otherwise a sharp cultural divide, with Arandic-speakers occupying the Central Australian ranges, including the Arrernte, Alyawarr and Anmatyerre. Generally, these societies were marked by richer cultural paraphernalia; denser, more deeply etched mythologies and longer ceremonial song cycles. To the east, groups such as the Diyari and Yan- druwantha along Cooper Creek, and the Wangkangurru occupying the Simpson dunefield, were different again. Ranged along a demographic gradient, there were systemic changes in patterns of territoriality and land tenure, with territoriality apparently coupled to population density. Corporate descent groups had primary responsibility for land in Central Australia and probably also in the eastern part of the arid zone (Peterson 1986; Sutton 2003; Keen 2004)but were absent in the Western Desert. In the latter area, ‘sociality is stretched to its uttermost and the emphasis is on inclusion’: descent is reduced to ‘a metaphor manifested in such phrases as, “my father’s country” as an explanation or claim’ (Peterson 1986: 151–2), although ritual property is still jealously held.

HUMAN ECOLOGY Deserts are difficult environments for hunter-gatherers, not just because scarcity of water is a limiting factor but also because they are inherently patchy and highly variable on a range of spatial and time scales. Small parts of the landscape – springs, groundwater discharge zones, run-on areas and oases – are The Archaeology of Deserts: Australia in Context 11 often the key to people’s use of these zones, especially in Australia’s deserts, where plant and animal resources are thinly distributed throughout the region. Here, it is the small surface waters, seepages, wells and springs that provide access to the desert hinterland and a means of stepping through the country. Because the sandy mantle is mostly less than 10 m thick, these deserts have a network of small wells, rockholes and soakages. Although these yield only small amounts of water, it is often enough to provide people with a base for exploiting the resources of the surrounding desert. In the better watered ranges, most people lived in small groups of about twenty people, less than a day’s walk from their neighbours and within easy sight of their hunting fires (Peterson 1986: 44–6). Little theoretical attention has been given to the comparative cultural ecology of desert societies, but sources suggest several potential strategies (Barker and Gilbertson 2000; Veth, Smith and Hiscock 2005): 1. One response is to invest heavily in technology and infrastructure; to harvest or move water through deep wells, qanats or channels; or to store water and food to buffer aridity and scarcity, effectively stabilising returns from the land. 2. Another is ephemeral use of arid regions, exploiting seasonal changes in humidity and pasture to support long-distance transhumance – as some pastoral groups do – effectively evading aridity in time. 3. Some systems focus on major oases, quebrada or wadi systems, or other refuge areas within a desert. This is only viable where these pockets are also zones of very high productivity and where use is coupled with horticulture, as it is in oases in the Atacama Desert of northern Chile and in North Africa. In effect, this means evading aridity in space. 4. Where there are market economies on the desert margins, or demand for exotic or prestige goods, desert groups are sometimes able to exploit their geographic position for raiding or as nodes on important trade routes. Some desert communities, such as San Pedro de Atacama in the Atacama Desert, have grown rich trading with outside groups. 5. For most desert hunter-gatherers, the main strategy was to be highly mobile and opportunistic, as resources ebb and flow over time and space, using pulses of rainfall to disperse across foraging territories and falling back on small wells and waterholes as the country dries out. The last of these strategies was the one adopted by Aboriginal groups in Australia’s deserts. Here, there are few sizable pockets with levels of productivity comparable to the oases in the Atacama in Chile or the deserts of North Africa, and the scale of the Australian arid zone largely rules out transhumance from desert margins. 12 The Archaeology of Australia’s Deserts

For hunter-gatherers, mobility is a first-order response to local depletion of bush foods: in desert environments, this is accentuated by low primary productivity, and the spatiotemporal distribution of water determines how you step through the country. Ethnographically, the general pattern for these Aboriginal groups was to take advantage of rain to visit outlying parts of a foraging territory and to exploit the bush foods in these areas while they were accessible (Thomson 1964; Gould 1969a; Allen 1972; Tindale 1972; O’Connell 1977; Tonkinson 1978; W Jones 1979;Cane1984;Kimber1984). The temporary waters that supported these visits included claypans, soakages, rockholes and shallow wells. These were short-lived and often unpredictable. As they dried up, people fell back to camps near the main waters. The bush foods around these camps were left in reserve until foraging was restricted during the dry season to areas around the main waters: ‘the Aborigines,’ as Gould put it, ‘eat their way into a camp by first exploiting all the food resources whenever possible before settling at the main waterhole’ (1969a: 267). The lives of these desert people were deeply affected by uncertain rainfall and irregular production. People tailored their land use to take advantage of rain to access fresh foraging patches; they co-opted large pulses of biological production as opportunities for larger gatherings and ceremonies; and, during protracted droughts, the population would fission into small groups, dissolving as individuals activated their social networks to take up residence in neighbouring territory. People attempted to stabilise returns by stockpiling some foods to meet seasonal shortfalls and to stabilise access to country by maintaining wells and capping rockholes. Tonkinson commented on the ‘continuing dialectic between the ecological constraints that push people apart and the cultural pressures that draw them together’ (1978: 30). One of the challenges was to maintain a territorial and land tenure system that could operate in the face of a high degree of fluidity in residence patterns, group composition and the distribution of people on the land. Stanner (1965) draws a key distinction between an ‘estate’ (the religious core owned by a descent group) and the ‘range’ (the wider foraging territory). Aboriginal people attached themselves to foraging groups wherever they could establish some links and rights of access. This created residential groups loosely made up of people related through marriage, direct kin relations, place of conception or birth, ritual knowledge or perhaps the death of a parent or grandparent in the area. Underlying this, however, was a formal landscape of clan estates – usually an area of 900–1,500 km2 containing at least one reliable water source – in which a group of kin claimed primary tenure.

THE ARCHAEOLOGY OF DESERTS The archaeology of deserts is rarely treated as a discrete field of inquiry, but a stronger comparative framework would clearly benefit archaeological research The Archaeology of Deserts: Australia in Context 13 across the world’s arid regions (Barker and Gilbertson 2000; Smith and Hesse 2005;Vethetal.2005). Arid landscapes often bear the imprint of past climates in dry lakes, abandoned river channels or fossil dune systems. Climate change can reactivate a desert landscape or leave it stranded and lifeless. Small shifts in rainfall or groundwater can have dramatic consequences for desert communities. ‘Places die, just as men do,’ says Joseph Joubert (in Depardon et al. 2000: 102). It is not surprising, therefore, to find that desert research hinges on similar questions: whether prehistoric settlement in areas now deserts was restricted to humid episodes, or whether increasing aridity led to the abandonment of entire regions or fragmented existing patterns of settlement, creating significant gaps in occupation sequences. In practical terms, the sparse ground cover in deserts means that archaeological sites, rock art and stone arrangements are highly visible. For this reason, work on open sites is a feature of research in these areas. Although this is true in Australia’s deserts (e.g., Smith 1988; McConnochie 1996; Holdaway, Fanning and Shiner 2005; Hughes et al. 2001), the poor stone tool systematics means that, here, phasing of sites relies on direct dating of stratified deposits or remnant hearths to a greater extent than in other desert regions. Tapho- nomic problems are common to many regions and include the cycling of lags of artefacts in mobile dunes with repeated episodes of deflation and reburial (Hughes et al. 2011), and reworking of stone tool scatters on spatially extensive deflation surfaces (Fanning et al. 2009). Archaeological remains in stratified deposits are generally well preserved, but Australia’s deserts lack the preservation of wood, bone, fibre or plant material commonly found in archaeological sites in hyper-arid regions. The archaeological record for arid Australia is striking for its austerity: the archaeology of these hunter-gatherer systems relies as much on context as on material remains. Most sites have few artefacts and few categories of remains (usually only chipped stone artefacts, grindstone fragments, red ochre and charcoal). Many archaeological deposits are shallow, often no more than a metre deep, and poorly (or cryptically) stratified. Yet the history of research over the last 30 years suggests the regional records are remarkably robust. Most areas show a repeated pattern of discovery of sites with broadly similar age ranges, sequences and site assemblages. Australian research widely employs an approach that focuses on strategic points in the desert landscape – usually near critical waters, often in small gorge or range systems – and on reconstructing site histories. This is basically an ‘archaeology of place’ in the sense Binford (1982) describes. Characteristically, fieldwork has involved small rockshelter or open-site excavations, in locations where the excavated sites are part of complexes that include clusters of occupied rockshelters, open sites, stone arrangements, engravings and rock paintings (e.g., Thorley 1998b; Smith and Ross 2008; Veth, McDonald and White 2008). Many rockshelter deposits are extensions of occupation on adjacent sand 14 The Archaeology of Australia’s Deserts

mantles, blurring any straightforward dichotomy between open and rockshelter sites. In the depositional landscapes of Central Australia, stratified open sites are common (Smith 1988). However, in the southeastern part of the arid zone, extensive scatters of stone artefacts on deflation surfaces are more common – and these represent major geomorphic challenges to unpack (Holdaway et al. 2005). Most burials and skeletal material have also come from the southeastern margins of the arid zone, where calcareous soils preserve bone – in contrast to the rest of the arid zone, where bone is rarely preserved beyond a few hundred years. There are some obvious caveats in interpreting site data. Excavations are often limited to small exploratory trenches, not all of which are reported in detail, and few sites have been subject to reinvestigation. Few sites have seen detailed analysis of their sedimentary or geomorphic history. Samples of stone tool assemblages are invariably small and, for larger sites, usually represent only one of the potential discard zones within a rockshelter – with consequences for the type of assemblage recovered. Few excavators explicitly attempt to isolate and characterise separate phases of occupation in any detail. For some late Pleistocene sites, it is not clear whether the archaeological remains represent discrete, widely spaced pulses of occupation within a slowly accumulating sediment matrix, or whether they reflect the slow, continuous accumulation of a few artefacts per century. There are also some cautions in how we interpret site records in behavioural terms. Most excavation units probably represent palimpsests of different sorts of occupation, with a mix of short-term and extended-stay visits. These problems are accentuated by the coarse resolution of chronologies for late Pleistocene sites (where luminescence and 14C dates often have uncertainties of several thousand and several hundred years, respectively) and the lack of fine-grained sampling for radiocarbon chronologies for Holocene sites. Larger excavations and detailed analyses of individual site records potentially offer both greater precision and a richer picture of late Pleistocene and mid- Holocene societies – as work at Puritjarra, Lake Tandou TN36, Puntutjarpa and Skew Valley shows (see Chapters 4–6). The most important challenge for archaeologists in the Australian arid zone is to more fully exploit the intrinsic potential of late Pleistocene sites as archives of the cultural and economic landscapes in which they are embedded.

THE POLITICS OF PRACTICE Other issues concern the contemporary political and economic situation in Australia’s deserts. Biologists, ecologists, archaeologists and anthropologists all benefitted from the four-wheel drive (4WD) vehicle revolution that swept remote Australia from the 1960s onwards. But the same forces that opened up The Archaeology of Deserts: Australia in Context 15 the desert for archaeological research also placed more constraints on where, when and how this research could be done. Since the passage of the Aboriginal Land Rights (Northern Territory) Act in 1976,theAnangu Pitjantjatjara Yankunyt- jatjara Land Rights Act in 1981,andtheNative Title Act in 1993, archaeologists have required community approval to work on most lands in the region. Most state and federal jurisdictions manage archaeological sites as a form of cultural property and require appropriate consultation at the community level before issuing permits for field research. This is no bad thing in itself but imposes long lead times for field projects, mostly beyond the term of a PhD project or a 3-to5-year research grant. Although many veteran archaeologists have longstanding working relationships with particular communities – often extending over several generations – it has become much more difficult for students or new researchers to enter the field. The current cohort of senior researchers represents the last to have been involved in primary archaeological exploration of the continent and the last generation to have the opportunity to travel and work with desert people who have spent part of their lives in the bush without significant European contact. There has been little direct impact on research agendas or the types of research questions posed by archaeological projects. The main effect has been a shift away from regional surveys by solo researchers towards larger ramified multidisciplinary projects, often with bush communities as formal partners. These sometimes embed archaeological research within a wider project of cultural mapping, heritage management or community art. Archaeologists also find themselves working for land councils as part of native title claims, although my own experience is that bush people sometimes see the task of archaeology as providing tangible evidence of the ‘dreaming’ rather than as a dispassionate examination of evidence for precontact occupation of an area. Over the last 30 years, archaeology has become increasingly part of the fabric of contemporary life in the desert and a point of articulation among bush communities, government agencies, industry and wider Australian society. For instance, in support of the successful Spinifex (Pila Nguru) Native Title Claim in 2000, Chief Justice Michael Black noted that ‘the archaeological evidence suggests the emergence of a rudimentary nomadic society in the Western Desert at least 20,000 years ago’ (Hickman 2000: 7). Another factor shaping archaeological research in Australia’s deserts has been the boom in iron ore and coal mining and in oil and gas exploration, especially in northwestern Australia. Since the 1980s, this has generated archaeological survey and salvage projects in the Pilbara, but the last decade, in particular, has seen an explosive growth in this type of work, to the extent that salvage and survey contracts now employ the majority of archaeological graduates. Much of this fieldwork is of very uneven quality, aimed at compliance with legislation rather than at answering research questions. Few sites are reported 16 The Archaeology of Australia’s Deserts

in detail, fewer are published, and it is rare to find that sufficient work has been done to characterise the nature of occupation at sites or provide a detailed site chronology. Nonetheless, this large grey literature is a boon for researchers, and there have been attempts to put work in the Pilbara on a firmer research footing (Morse 2009). This Pilbara research has also produced a large excavated dataset and a remarkable suite of late Pleistocene sites. CHAPTER 2

DESERTS PAST: A HISTORY OF IDEAS

In the summer of 1896, more than a hundred Arrernte tribesmen gathered at an isolated repeater station on the Overland Telegraph Line to stage a great ingkwere cycle of ceremonies (Spencer and Gillen 1897). Lasting for more than four months, with ceremonies day and night, an ingkwere (‘Engwura’) was normally held at the totemic site associated with the principal ceremonies (Strehlow 1947: 100–12). However, on the colonial frontier, Alice Springs telegraph station promised a measure of security, away from police, missionaries and cattlemen. The stationmaster, FJ Gillen, was the subprotector of Aborig- ines and local magistrate. Only a few years earlier, he had arrested Mounted Constable WH Willshire for murdering two Aboriginal men (Vallee 2004–6). For Aboriginal elders, the deciding factor was that Gillen – a keen ethnog- rapher – was prepared to finance the ceremonies by providing rations for the participants. Joining Gillen in a makeshift shelter on the ceremonial ground in the stifling summer heat was Oxford biologist Baldwin Spencer (Figure 2.1). The two men would later write The Native Tribes of Central Australia (1899), a detailed ethnography that revealed the richly elaborate religious life of these desert societies (Mulvaney and Calaby 1985; Mulvaney, Morphy and Petch 1997). It became one of the most influential books of its day (Kuklick 2006). But it was not contemporary Arrernte society that excited most international interest. Under the influence of Spencer’s mentor, JG Frazer, the Arrernte entered European imagination as ‘humanity in the chrysalis stage’ (Frazer 1899: 648,inKuklick2006), a relic of an early stage in the evolution of religious and social institutions (Lang 1903; Durkheim 1912; Mulvaney 1964;PJBowler 1992;Hiatt1996;Kuklick2006). The Arrernte also provided prehistorians with analogues for Palaeolithic societies in western Europe (Reinach 1903; Sollas 1911; Lubbock 1912). For instance, the notion that Stone Age hunters in the Australian desert could create profound artistic and religious works led Salomon Reinach to suggest a magico-religious function for Palaeolithic cave art in France. 17 18 The Archaeology of Australia’s Deserts

Ostensibly, this glossed over the fact that Central Australian societies might also have a history, that contemporary Arrernte society might be more complex than its ancestral form. But, in fact, ‘evolutionism’ did little to dampen speculation about the deep history of these desert societies. Ideas about the nature of the desert itself also changed. And shifting environments of lakes, rivers and duneﬁelds implicitly framed a changing geography against which ideas about desert people were set. This chapter follows this stream of ideas, examining how the past history of desert landscapes and desert societies has been interpreted since European explorers and scientists ﬁrst encountered them. It interweaves two narratives: one explores emerging ideas about the environmental history of Australia’s deserts; the other looks at ideas about the prehistory of Aboriginal groups in this region. I want to chart the research that created the conceptual space within which we now work. I also want to highlight the recurrent themes that frame research into the natural and cultural prehistory of Australia’s deserts because these will provide the rationale and structure for later chapters.

THE DEAD HEART OF AUSTRALIA Almost as soon as colonial exploring parties entered the desert, they encountered the large saltpans that dominate this landscape. In 1840, just 4 years after the founding of Adelaide, EJ Eyre led an expedition north, following the Flinders Ranges, until he found his route blocked to the west, north and east by a chain of large saltlakes. Eyre thought these formed a single large ‘horse- shoe lake’. He suggested it was an old arm of the sea, describing the northern end of the ranges as ‘the termination of the island of South Australia, for such I imagine it once to have been’ (1845:I,128). JLewis(1875) mapped the boundaries of Lake Eyre in 1874–5. Other reports soon established the presence of a fossil fauna that included large fish, crocodiles, chelids, ‘an emu-like bird’ (probably Genyornis)andDiprotodon optatum (Tate 1886: 54). By 1885, enough was known to suggest there had been major environmental changes in the interior. Ralph Tate was able to sum up: ‘A vastly increased rainfall over what is now the arid region of Australia in former times is demanded by the extinct rivers and lakes and the former existence of large herbivores’, adding that ‘a glacial period and a pluvial period mean the same to me’ (Tate 1886: 53; see also Tate 1879: lxvii). The most striking evidence of these environmental changes was the discovery in 1892 of the fossil remains of hundreds of extinct giant marsupials and large flightless birds on the now-dry bed of Lake Callabonna (Stirling and Zietz 1899–1913). The following year, a museum field party excavated ‘a veritable necropolis of gigantic extinct marsupials and birds which have apparently died where they lie, literally in hundreds’ mired in lake muds (Stirling 1900:i–xv; see also Douglas 2006). In the title of his book, The Dead Heart of Australia Deserts Past: A History of Ideas 19

Figure 2.1. Baldwin Spencer with Arrernte elders at the 1896 ingkwerre ceremonies, Alice Springs, Central Australia. The publication of The Native Tribes of Central Australia in 1899 gave the Arrernte a prominent role in the public imagination, one not surpassed until the Western Desert Papunya Tula Art movement in the 1970s. (Photograph courtesy Museum Victoria – XP14279)

(1906), JW Gregory captured the consensus that a Pleistocene pluvial period had been followed by continental desiccation. In the 1940s, TW Edgeworth David drew together earlier ideas about an enlarged Pleistocene lake (Tate 1879: lxvii, 1886;Gregory1906;Howchin 1909: 102;David1932), labelling it ‘Lake Dieri’ (Browne 1945: vii). This ancestral mega-lake had been 10 timesthesizeofLakeEyreand20 m deep (David and Browne 1950: 616–17). Dating was uncertain. Although David and Browne suggested a middle Pleistocene age (180 ka), some writers took the view that pluvial conditions had persisted into the early Holocene (Crocker and Wood 1947). Later work was to show that ‘Lake Dieri’ conflated sediments and landforms of widely differing ages (Wasson 1983; Stevens 1991), although Loffler¨ and Sullivan (1979) briefly resurrected the notion of an immense ancestral lake. The great flooding of Lake Eyre in 1949–50 saw a resurgence of scientific interest in the playa (Mawson 1950; Peake-Jones 1955). This produced the first direct stratigraphic evidence of palaeo-shorelines around the lake (King 1956: 100) and the first use of radiocarbon (14C) dating in arid Australia (Baas- Becking and Kaplan 1956). But it was another set of palaeolakes, those in the 20 The Archaeology of Australia’s Deserts

Willandra, that rapidly became the focus of research in the 1970s and prompted a radical rethinking of the notion that the glacial climate of the inland had been ‘pluvial’.

DESERT SOCIETIES In 1699, the English buccaneer William Dampier explored the arid west coast from Shark Bay to Lagrange Bay but did not see Aboriginal people until he reached the Kimberley coast. French mariners were the first to encounter desert people. In 1803, at Shark Bay on the arid west coast, F Peron´ was confronted by a group of armed Malkana men (1824, II: 144–6). A generation later, PP King met family groups of Yaburara people on the Pilbara coast in 1818 (King 1827,I:38–47). The Central Australian Expedition (1844–6)was the first to travel deep into the arid interior, following the ephemeral rivers feeding Lake Eyre to reach 25◦S on the eastern side of the Simpson Desert in September 1845 (Davis 2002). This party, under Charles Sturt, was in the field for 17 months and encountered Aboriginal groups almost every day, providing a remarkable picture of a hunter-gatherer landscape peopled by family groups spaced a day’s travel apart (Figure 2.2). The tempo of exploration and contact quickened in mid-century, when rival expeditions by J McDouall Stuart (1858–62) and Burke and Wills (1860– 1) set out to cross the continent. Aboriginal groups along Cooper Creek came to public attention when they assisted John King, the only survivor of Burke’s advance party. In April 1860, McDouall Stuart became the first explorer to encounter the Arrernte people when he reached the MacDonnell Ranges in Central Australia. After his route north was blocked by armed Warrumungu men, the South Australian Parliament debated whether he should shoot his way through, justifying this on the basis that ‘the tribes on the northern coast were of Malay origin, and were not the original lords of the soil which they now occupied’ (see Exploration of the Interior [Parliamentary Debates, 9 October 1860]: 1008). The construction of the Overland Telegraph Line from 1870 to 1872 brought Aboriginal groups in Central Australia into sustained contact with Europeans and provided a base for further exploration west into the Western Desert. The Elder Scientific Exploration Expedition (1891–2) and the Horn Scientific Expedition to Central Australia (1894) (Spencer 1896) were the first explicitly scientific expeditions to Australian deserts; the latter was the first to focus on the detail of Aboriginal life. But, by 1900, much of Central Australia had been occupied by pastoralists. In this context, the 1896 ingkwere represented a remarkable attempt by senior Arrernte men to draw colonial authorities into their governing ideology and a virtuoso demonstration of their ritual property and knowledge. Deserts Past: A History of Ideas 21

Figure 2.2. ‘Native village in the Northern Interior’. The ‘Central Australian Expe- dition’ (1844–6) was the ﬁrst to travel deep into the arid interior of Australia. Charles Sturt recorded these huts, west of Tibooburra on the margin of the Strzelecki Desert, in late January 1845.(Source:Sturt1849:I,facingp.254)

Most writers accepted that Australian Aboriginal people shared a common origin – ‘all have originally sprung from the same stock’ (Eyre 1845: II, 151)– but differences in language, material culture and social organisation raised questions about the dispersal of people across the continent and their subsequent differentiation. In 1787, a year before the first British settlement in Australia, W Eden had argued that the uniformity of people and material culture evident on the northern and southern coasts of ‘New Holland’ implied not only a common origin but also rapid settlement, without time for differentiation of these groups to occur. Given the size of the continent, the only way this could have been achieved was if the spread of Aboriginal groups had been limited to ‘a narrow tract of sea coast’ and ‘continued circuitously along the coast, leaving the interior entirely desolate, or at least but partially inhabited’ (1787: 233). With better information on cultural variability but little more on the geography of the interior, Eyre suggested three separate routes of migration radiating out from the northwest coast, with one directly crossing the centre of the continent (1845: II, 393, 405–7): ‘each offset appearing to retain fewer or more oftheoriginalhabits...inproportiontothedistancetraversed,oritsisolated position . . . modified also, perhaps, in some degree, by the local circumstances of the country through which it may have spread’ (1845: II, 405–6). The idea of a community of custom and organisation, subsequently modified by distance, isolation and adaptation to a varied ecology, was echoed by many later 22 The Archaeology of Australia’s Deserts

writers (Favenc 1893; Spencer and Gillen 1899: 36, 117; 1904: 15, 17; Spencer 1921). Better knowledge of the interior drove changes in ideas about likely migration routes. After 1845, many writers saw the Cooper, Diamantina and Darling River systems as key routes for initial human dispersal into the interior (Smyth 1878: lxvi; Mathew 1889; Spencer and Gillen 1899: 596; 1904: 17). To account for the ‘remarkable homogeneity’ in core beliefs and customs shown in their ethnographic work, Spencer and Gillen (1904: 18–19) argued for initial settlement during the late Pleistocene under a pluvial climate, with increasing regional diversity as ‘the gradual desiccation of the central area’ took place. However, Gregory (1906) had not found any trace of people in the fossil beds along Cooper Creek and Warburton River, although this was an explicit objective of his work. This suggested that people were not coeval with either the extinct giant marsupials or contemporary with pluvial conditions in the interior. Walter Howchin, the indefatigable geologist who laid the foundations of South Australian stratigraphy, later found crude stone implements on the desert plateau near Lake Eyre that led him to argue for a desert ‘Eolithic’ phase of ‘considerable antiquity’ (1921). However, these objects were quickly dismissed as the products of thermal fracture and other natural agencies (Jones and Campbell 1925; Tindale 1932). A short chronology, however, sat uncomfortably with the notion of Stone Age survivals (Lubbock 1875: 427–9, 438–50; 1912; Spencer 1901: 12). R Brough Smyth (1878, I: lv–lvii) argued that without a Pleistocene presence, Aboriginal society could hardly be a Palaeolithic survival. In Ancient Hunters and Their Modern Representatives (1911), W Sollas offered a solution. Calling the Arrernte and other Aboriginal groups ‘the Mousterians of the Antipodes’, he suggested ‘that the inferior tribes of the Neandertal race were driven by stress of competition out of Europe, and wandered till they reached the Australian region’ (1911: 208). Time was not the only issue here: the extent of autonomous cultural development also posed a challenge to the notion that Central Australian society was a sociological fossil. Theories of global social evolution set much of the agenda for Spencer and Gillen’s research, but their field-based anthropology led them towards functionalist or sociological explanations of Arrernte life (Morphy 1997: 30–7). Recognising that these societies must also have a history of local development, they asserted that ‘it is impossible to believe that the far-away ancestors entered the continent provided with a highly complex set of customs and beliefs such as are now common to all the tribes. These have undoubtedly been elaborated in course of the long ages’ (Spencer and Gillen 1904: 15). Their interest in the historical dimensions of tradition led them to suggest a sequence of changes in Arrernte social institutions. The four stages of the altyerrenge (‘alcheringa’ or ‘dreamtime’) (Spencer and Gillen 1899: chapter 10) Deserts Past: A History of Ideas 23 began with the creation of men and women, followed by introduction of the four-class section system, circumcision and the use of stone circumcision knives, then introduction of subincision rites by travelling atyelpe (‘achilpa’ or native cat) ancestors, and finally the present system of marriage and social segmentation involving eight subsections. Although this was crude historicism at best, modern linguistic research on subsection systems (McConvell 1996) and archaeological work on sites associated with atyelpe ceremonies (Smith 1988: chapter 9;David2002) both offer a measure of support for this schema. The implications concerned A Lang (1904, 1910). He pointed out that because the Arrernte could not be the most primitive of Aboriginal groups, any conclusions about the evolution of totemism and early social organisation were invalid. Sollas also noted that the presence of more advanced implements, such as ground-stone axes, indicated that these groups ‘though still remaining in the Palaeolithic stage have made a considerable advance on the culture of Neanderthal man’ (1911: 179, 207). DS Davidson took the argument for local social evolution a step further, positing Central Australia as a centre of cultural innovation. The geographic distribution of cultural traits – such as subincision and class and subsection systems – ‘proved that Central Australia is the centre of cultural intensification and development of this trait [subincision] in addition to many others, regardless of the point of origin of each’ (1928: 139). By the 1920s, a picture of desert prehistory had emerged: a Palaeolithic people had migrated into the interior by following the inland river systems, and there had been some differentiation of desert groups and local evolution of social institutions and material culture. The limited evidence available suggested this had taken place in the Holocene, after the extinction of the megafauna and after the Pleistocene pluvial lakes had disappeared. However, towards the end of his career, Spencer restated the case for a longer chronology: to account for the development of a common core of beliefs and customs across the interior, initial settlement must have taken place in ‘early Pleistocene times’ (1921: 26), when there were fewer environmental barriers to communication. Revising his ethnography in The Arunta: A Study of a Stone Age People (1927), Spencer noted that in the earliest traditions, the ‘country was covered with salt water’, coinciding with geological evidence for ‘the existence of a great inland sea’ (Spencer and Gillen 1927:I,307,fn3).

ANCIENT PETROGLYPHS Rock engravings (petroglyphs) are the most tangible evidence of human antiquity in Australian deserts. Herbert Basedow (1907, 1914)wasthefirsttorecog- nise the significance of this ‘track and circle’ rock art, drawing on fieldwork in the Flinders Ranges between 1906 and 1911. Living Aborigines knew nothing of the engravings, he claimed. Some designs seemed to represent the tracks 24 The Archaeology of Australia’s Deserts

of extinct animals, including Genyornis and Diprotodon (Figure 2.3). In most cases, the engravings were on fractured and broken rock surfaces and were covered with a ‘dark rust-coloured “patina” or glazed surface film’ – all of which indicated a ‘very considerable antiquity’. Anticipating modern research on these mineral films, Basedow obtained samples from Egyptian quarries dating to ‘about 3000 years B.C.’ and compared the two, noting that those on Australian samples were thicker (Basedow 1914: 199–200). By implication, the rock carvings were older than 5,000 years (an age remarkably close to some modern estimates; see Smith, Watchman and Ross 2009). Basedow’s work led to a surge of research on rock engravings (Hale and Tindale 1925; Campbell 1925; Pulleine 1925; Mountford 1929a). More finds of extinct fauna followed – with Mountford famously identifying an engraving of an estuarine ‘crocodile’ at Panaramitee (1929b; Mountford and Edwards 1962) and Tindale identifying the tracks of Genyornis in engravings at Eucolo Creek (Hall, McGowan and Guleksen 1951; Tindale 1951). Mountford correlated his ‘crocodile’ with the ancestral Lake Eyre described by Gregory (1906)– although Berndt (1987) later showed it was probably a Ngadjuri hair-string ‘yarida’ sorcery figure. By 1960, the uniformity of this ‘extinct art’ across 2.5 million km2 of the desert interior was striking (Mountford 1960). Quantitative analyses reinforced the picture of an ancient continental rock art tradition (Edwards 1963, 1965, 1966, 1968; Mountford and Edwards 1964;Layton1992; Franklin, 2004), labelled the ‘Panaramitee style’ (Maynard 1979), that extended to Tasmania prior to the postglacial rise in sea level. The consistency of this tradition implied a high degree of cultural homogeneity in Australian deserts during the late Pleistocene (Edwards 1966, 1971;Maynard1979). This was taken as evidence that Pleistocene groups maintained open, extensive social networks, important for small mobile populations (Morwood, 1984). But this continental tradition had fragmented as populations grew in the postglacial period, resulting in greater regional diversity in rock art and an increase in territoriality.

THE ‘GREAT AUSTRALIAN ARID PERIOD’ The key to understanding postglacial environmental changes initially lay with desert dunes. Sturt’s 1844–5 expedition had been the first British party to enter the immense dunefields that ring the centre of the continent. When the expedition was finally blocked by the sandridges of the Simpson Desert, Sturt wrote to his wife:

Sunday September 7th 1845 ...AscendingoneofthesandridgesI saw a numberless succession of these terriﬁc objects rising above each other to the east and west of me. Northwards they ran away before me for morethanﬁfteenmiles...Howmuchfarthertheywentwiththesame undeviating regularity God only knows . . . The scene was awfully fearful, dearCharlotte...Itlooked like the entrance into Hell. (Sturt 1984: 72–4) Deserts Past: A History of Ideas 25

Figure 2.3. Herbert Basedow’s 1907 tracing of rock engravings at Wilkindinna in the northern Flinders Ranges. Basedow suggested that No. 3 represented the track of Diprotodon, an extinct giant marsupial known from the region. (Source: Basedow 1907: 715)

Sturt doubted that ‘winds had formed these remarkable accumulations of sand’astheywere‘asstraightasanarrow...acrosssixdegreesoflatitude’and suggested instead that they were scour features created when the interior had been uplifted and an ancient sea had violently drained away (1849: 380–1). ‘They exhibit a regularity that water alone could have given,’ wrote Sturt, ‘and to water, I believe, they plainly owe their first existence’ (1849: 381). There is an echo here of the ‘diluvialist’ ideas current in Britain when Sturt left for Australia in 1827. Steady-state geomorphic processes could not explain raised shell beds, glacial erratics, U-shaped valleys and glacial moraines. Most geologists in Europe accepted the reality of a geological deluge – even if a correlation with the biblical flood was doubted – and that parts of the continental landmasses had once been submerged. By 1840, however, the glacial origin of most of the anomalous landforms was clear, and diluvial ideas had morphed into ideas about a Pleistocene Ice Age (Rudwick 2008). In this context, Sturt’s ideas did not get much traction. Prior to 1870, geographical knowledge about the interior was simply too scanty to permit much of a debate about the origins of arid Australia. Most observers saw the sandridges as aeolian landforms reflecting contemporary arid conditions in the interior of Australia (e.g., Jutson 1934: 330). The fact that dunes had built over dry lake beds and appeared to be blown out of older fluvial sediments indicated that they postdated Pleistocene pluvial conditions. However, by 1940, regional geomorphic mapping in western Queensland (Whitehouse 1940: 67–8) and soil erosion studies in the Cooper 26 The Archaeology of Australia’s Deserts

basin (Ratcliffe 1936, 1938) indicated that the dunefields themselves were fossil landforms. This implicitly raised questions of chronology. Whitehouse thought the field evidence indicated that the dunefields had formed during the closing phases of the Pleistocene (1940: fig. 14). However, the most influential assessment was by RL Crocker, then a soil scientist at the University of Adelaide. In a series of papers, Crocker argued that ‘the Great Australian arid period’ was postglacial, probably dating to between 6,000 and 4,000 years ago (Crocker 1946; Crocker and Cotton 1946; Crocker and Wood 1947 – see also Browne 1945: xix). Impressed by exceptionally thick sheets of calcrete (locally termed kunkar) in soils in South Australia, Crocker argued that the likely source was windblown calcium carbonate (‘loessial lime’) deflating from shelly beaches formed during glacial low sea level. Later wind erosion of these soils had provided the sediment for dunes in this region. Given the time required for this chain of events, the arid period probably coincided with the thermal maximum in the first half of the Holocene (Crocker 1946). ‘A catastrophic decline in rainfall which initiated the aridity placed such a stress on the pre-arid flora,’ argued Crocker and Wood, so that ‘it was almost completely wiped out’. Under these conditions, both wind and water erosion were more aggressive ‘and carried considerable quantities of sand and silt into thedrainagebasins’(1947: 103–4). The idea of a short, sharp period of extreme aridity 6,000–4,000 years ago built on a decade of detailed regional mapping of soils, alluvial sediments and laterites and gained wide acceptance (Browne 1945;Gill1965). But the only fixed point in this chronology was the last marine transgression, and this is where it began to unravel. The key evidence was the identification of submerged linear dunes and lunettes on shallow continental shelves, drowned by the postglacial rise in sea level (Fairbridge 1961;Sprigg1964, 1979; Jennings 1975). RC Sprigg was quick to see the implications, concluding that ‘the extreme “desert” phases correlate with glacial eustatically low-sea levels, and in turn with recurring Pleistocene glaciation’ (Sprigg 1964). The dunefields were therefore significantly older than Crocker and Wood had proposed.

Shifts in Climatic Belts One of the perennial questions the duneﬁelds raised was whether Quaternary climate changes simply involved latitudinal shifts in weather systems. T Grifﬁth Taylor (1919) popularised the idea that ‘an increase in the temperature of the middle latitudes . . . pushes the arid belt and the polar rain belts away from the equator’ (1919: 297). During the Pleistocene, the arid belt would have been centred on Darwin (a shift of 1,300 km). Taylor’s model of a ‘drift’ in the position of the arid belt was widely accepted (Spencer 1921;Keble1947;Birdsell 1957: 60–1;Keast1959; Tindale 1959: 50) and seemed to provide a plausible mechanism for environmental shifts in the Lake Eyre basin. Keble (1947: 30) Deserts Past: A History of Ideas 27

Figure 2.4. Robert Keble’s 1947 map of climatic belts in the interior 15,000 years ago. In this version, the arid zone is centred further north than today, and winter rainfall reactivates rivers in the eastern part of the zone, creating a large palaeolake ancestral to modern Lake Eyre. (Source:Keble1947: 72) supported this view, arguing that ‘with the advance of the belts, a region may have passed successively from desert into savannah to tropical rainforest’ (1947: 30)(Figure 2.4). Keast (1959) argued that the biology of the arid zone indicated ‘that desert was never eliminated from the continent’, and so it made more sense to think of shifts in the arid belt rather than a Quaternary succession of wet and dry climates (1959: 31). However, Gentilli (1961) was scornful of this notion, arguing that the Pleistocene ‘pluvial’ climate reflected the simul- taneous encroachment of wetter climates from the equatorial and poleward margins of the continent, compression of the mid-latitude weather systems, and a major contraction of the desert. Aerial mapping undertaken by Cecil Madigan in 1929 and Donald Mackay in the 1930s produced the first continental map of the dunefields (Madigan 1936). Using newly available national air photograph coverage, Jennings (1968) revised this – and this provided a basis for testing ideas about shifts in the prevailing wind systems (Brookfield 1970; Wasson 1983, 1986; Hesse 1994). During the last glacial, the dunefields registered a small latitudinal shift (no 28 The Archaeology of Australia’s Deserts

more than 3◦, equivalent to 300 km) in pressure belts and expanded both to the north and south (Hesse 1994). The desert, it seemed, had neither swung as much as Grifﬁth Taylor (1919) had suggested nor contracted as severely as Gentilli (1961) had argued. What the data showed was a more modest expansion and contraction of the desert and some readjustment of circulation on its southern margin.

CULTURE HISTORIES A more detailed human history took shape in the 1940sand1950s, as physical anthropologists, archaeologists and linguists attempted to account for variation, across the continent, in the physical characteristics of Aboriginal people, in stone artefact assemblages and in language.

Physical Anthropology One outcome of a joint anthropological expedition by Harvard University and the University of Adelaide (1938–9) was the ﬁrst detailed map of Aboriginal tribes across the continent, which was compiled by NB Tindale (1940). JB Birdsell, Tindale’s partner in this research, argued that the new data on physical variation in Aboriginal populations across the continent were best explained by successive waves of migration from the north (1941, 1957, 1967). His ‘trihybrid theory’ echoed earlier ideas, especially those of Howitt (1904: 31), who had suggested that Tasmanians were a relict pocket of an earlier population displaced on the mainland by Australian Aborigines (see also Mathew 1889; Flower and Lydekker 1891; Spencer 1921). None of these schemes had much impact on ideas about the population history of Australia’s deserts. And, except for Howchin’s ‘desert eolithic’, there was never any evidence to suggest a ‘Tasmanoid’ phase of occupation there. What Birdsell’s data did show was that apart from a latitudinal cline in skin colour, body proportions and body hair (Lindsell 2001; Gilligan and Bulbeck 2007), phenotypical variation in desert groups was so high that no cohesive patterns could be isolated (Birdsell 1993).

Stone Tools NB Tindale also saw the desert in dynamic terms, using variability in stone artefact assemblages to argue for a succession of cultures across the continent and adopting the ‘culture-historical approach’ then current in North America and Europe. The task, as he saw it, was ‘an assessment of the ways of life, activities and effectiveness of peoples of a series of palaeolithic hunting tribal communities’ that had been ‘drawn off from the whole seething cauldron of Asia at various intervals of time’. Geographic distance and adaptation to Deserts Past: A History of Ideas 29 local ecological conditions had ‘rapidly introduced a diversity in modes of exploitation and spread’ (1959: 38–9). Tindale’s 1957 scheme relied on identification of successive cultural phases, each characterised by a small number of distinctive stone implements (‘type fossils’) and differences in flaking techniques (Tindale 1957;Smith2000). The first widespread phase of occupation in Australian deserts was during the Tartangan, which spanned the period between the last glacial maximum (LGM) and the mid-Holocene, and was contemporary with the megafauna and the rock engravings. From 4.7 ka onwards, people with Pirrian stone implements – including unifacial ‘pirri’ points – expanded across the desert interior, from the Lake Eyre basin westwards into Central Australia and the Great Victoria Desert. This spread was assisted by favourable environmental conditions and a new technology for ‘wet grinding of grass seeds to prepare grass seed meal’ that allowed Pirrian groups to exploit the expanding grasslands created by millennia of Aboriginal firing of the vegetation. Around 3,000 years ago, the Mudukian – with a microlithic tool kit – dispersed across the southern half of the continent and into Central Australia and the Western Desert. The later Murundian were an adze-using desert people, ‘pressing southward and eastward out of the desert areas into South Eastern Australia’ during the last millennium. This latest replacement, argued Tindale, reflected the impact of increasing aridity, prompting a shift to greater mobility and a more streamlined toolkit. In a remarkable unpublished manuscript, ‘Archaeology of S-W Centralian Desert’ (1958), JE Johnson, a geologist working in the Mann–Musgrave Ranges, tried to apply Tindale’s classification to artefact sites in Central Aus- tralia. This was the first independent test of the stone tool seriation, and problems quickly became apparent. Many type artefacts, especially adzes and microliths, occurred together in the same assemblages. Some Tartangan sites were clearly modern. Others were coeval with Mudukian occupation. Tindale was quick to recognise the potential of radiocarbon dating, and, by 1957,he was able to cite five 14C dates in support of his pan-continental sequence. However, by 1963, empirical problems with the cultural-succession model had become acute. Later work did confirm a chronological shift from large flake-and-core assemblages to small-tool assemblages with adzes, points and geometric microliths, but no discrete groups of artefact assemblages corresponding to Tindale’s ‘cultures’ were found.

Historical Linguistics Linguists also proposed a dynamic history of migration and population replacement. Although translation of the New Testament into Diyari – a desert language – was accomplished as early as 1897 (Reuther and Strehlow, 1897), there was little comparative data on Aboriginal languages until 1940. Schmidt 30 The Archaeology of Australia’s Deserts

(1919) had drawn attention to a major division between the languages of northern and southern Australia. In a series of later studies, Arthur Capell (1937, 1940, 1956, 1962) also showed there was a division between prefixing languages in tropical northern Australia and the suffixing languages found elsewhere, including those in the arid interior. Languages in the north represented ‘the remains of a very early language group, swamped by later comers’ (1940: 432). Speakers of an ancestral language he called ‘Common Australian’ had spread across much of the continent and evolved in situ to produce a large group of interrelated languages belonging to a single phylum. In the early 1960s, K Hale coined the term ‘Pama-Nyungan’ for this language family. Linguists proposed that the centre of dispersal was west of the Gulf of Carpentaria (Hale 1964) or in ‘the northern part of the Arandic speech- area’, or in areas immediately north or east of this (O’Grady 1979: 114–15). Because language spread involves the transfer of a complex semiotic system – in its entirety – linguists saw the expansion of Pama-Nyungan as a demographic event that effectively wiped out most of the continent’s existing languages and their speakers. The linguistic diversity within the Pama-Nyungan languages seemed to require 4,000–5,000 years to develop. S Wurm (1972) attempted to correlate these linguistic changes with new archaeological data on the spread of microliths and stone axes. Most archaeologists, however, thought that population replacement on such a scale was implausible (Mulvaney and Kamminga 1999: 72–5) – unless the linguistic patterns were remnants of much earlier events. RMW Dixon (2002), a prominent linguist, argued that rather than a phylogenetic group, Pama-Nyungan was the product of a long period of reticulate linguistic evolution in the open terrain of the desert, where millennia of borrowing, diffusion and typological drift had operated to create a Sprachbund or linguistic area. Whatever the interpretation, comparative linguistics seemed to point to a pattern of ‘explosive and massive sociolinguistic’ events originating on the margins of the desert, each followed by periods of stability and in-situ efflorescence (Sutton and Koch 2008: 498). The cultural and linguist homogeneity of the Western Desert suggested that speakers of Wati represented a recent expansion of people across the western half of the arid zone (McConvell 1996;Veth 2000). In contrast, the Arandic languages in Central Australia formed a more ‘deeply etched’ linguistic landscape (Capell 1962: 12–13;Wurm1972: 166), an ancient Sprachbund created early in the initial breakup of Pama-Nyungan, and now displaying considerable internal variability in language and dialect.

THE AUSTRALIAN DESERT CULTURE The ﬁrst, long stratigraphic record from an archaeological site in the desert – supported by nine 14C dates – gave quite a different picture of desert prehistory. Between 1967 and 1970, RA Gould – an American archaeologist – excavated the Puntutjarpa rockshelter near Warburton in the Western Desert (1968, Deserts Past: A History of Ideas 31

1971a, 1977). The archaeological record here showed a long period of cultural and economic stability beginning about 10,000 years BP. Gould proposed the term ‘Australian Desert Culture’ (1971a; 1977: 168–82), arguing that this represented a distinctive way of life, one highly adapted to desert conditions. Its hallmarks were high mobility, a ‘risk-minimizing opportunism’, the harvesting of wild seeds as grain and a toolkit that included microliths and hafted adzes. Puntutjarpa rockshelter provided ‘evidence for the success of this ancient desert culture adaptation throughout the post-Pleistocene of the Western Desert, culminating in the ethnographic desert culture of that region, in what must surely stand as one of the most dramatic cases of cultural conservatism on record’ (1977: 182). The underlying reason for this long-term stability was that the desert ‘limits the optional responses by anyone with a hunting-gathering mode of subsistence who tries to live there’ (1977: 182). In effect, this dismissed the possibility of internal pressures for socioeconomic change. Gould was writing at a time when archaeologists were exploring equilibrium models for hunter-gatherer populations and when social anthropologists also stressed the deep conservatism of Central Australian and Western Desert societies (Meggitt 1962; Strehlow 1947, 1971; Tonkinson 1978). But the immediate inspiration was the notion of an American ‘desert culture’ ( Jennings 1957)–a long stable ‘lifeway’, adapted to the aridity of the desert Southwest, and characterised by high mobility, broad-spectrum diets, a sparse but efﬁcient material culture and the harvesting of small seeds as grain. In Australia, the concept of a ‘desert culture’ never gained wide acceptance. Australian archaeologists soon pointed out inconsistencies in identifying microliths and in the dating (Glover and Lampert 1969) and showed that the earlier deposits did not contain either adzes (Hiscock and Veth 1991)or seed-grinding implements (Smith 1986a). The novel feature of Gould’s work, however, was that in proposing a long period of cultural stability in the interior, he portrayed desert groups as an ancient endemic society. It was also a distinctively ‘desert’ society rather than one derived from the desert margins, with an economic focus on the use of small seeds. This way of life had developed as a response to Holocene aridity, to ‘the onset and persistence of the rigorous environmental conditions of the last 10,000 years’ (Gould 1971: 175). The desert had been settled during the more equable conditions of a Pleis- tocene pluvial climate, but Holocene desiccation of the interior meant that, in a sense, the desert had formed around the people. This argument hinged on a particular view of the environmental history of Australia’s deserts – and, by 1970, the notion of Holocene aridity was already seriously out of date.

LAKE MUNGO AND THE WILLANDRA LAKES The prevailing view that higher rainfall, and active rivers and lakes, had marked the late glacial climate changed so abruptly in the mid-1960sthatby1975, 32 The Archaeology of Australia’s Deserts

little trace of it remained. The crucial demonstration of glacial aridity was made by RW Galloway (1965), who derived numerical estimates of rainfall and evaporation during the last glacial by comparing the limits of periglacial action and snowlines in the Snowy Mountains with palaeo-shorelines at nearby Lake George. This showed that with large reductions in temperature and evaporation, enlarged lakes did not necessarily imply greater rainfall. The terms ‘wetter’ and ‘drier’ became increasingly ambiguous without ‘separate thermal and effective precipitation histories’ (Bowler et al. 1976: 361). In arid and semi-arid Australia, much of the new evidence for glacial aridity was from lunettes: crescentic clay-rich dunes, found on the downwind side of palaeolakes. Hundreds of lunettes and small palaeolakes are dotted across southern Australia. They are clearly relict landforms, although the process of their formation was not understood for some time (Hills 1939, 1940; Stephens and Crocker 1946). Observations of modern clay dune formation in hypersaline lagoons on the Gulf Coast of Texas eventually resolved the issue: the dominant process was the deflation of sand-sized clay pellets, formed by wetting and drying and increasing salinity on exposed lake margins and blown by strong winds into clay-rich dunes (Bowler 1968). Lunettes quickly became an important indicator of seasonal drought and salinity (Bowler 1971). The Willandra Lakes, which include Lake Mungo, make up a chain of small palaeolakes, each bordered by a lunette, in western New South Wales (Figure 2.5). Fed by flood discharges from the southern highlands, the now-dry lakes were active systems during the late Pleistocene. In 1968–9, the first series of 14C dates for lunettes confirmed they were glacial-age landforms (Bowler 1971), but it was their potential as environmental and archaeological archives that caught most international attention (Bowler et al. 1970; Barbetti and Allen 1972; Bowler, Thorne and Polach 1972). In a now-classic paper, ‘Pleistocene salinities and climatic change’ (1971), Jim Bowler showed that lunette sediments recorded facies change from quartz sand dunes (freshwater environments) to clay-rich dunes (saline environments). Each cycle began with quartz beach sands blown into dunes during high lake stands. As lake levels fell and salinity increased, clay pellets were blown into the lunettes, carrying with them salts recording the chemical ‘ageing’ and increasing salinity of the lake water: first carbonates, then chlorides and finally sulphates were precipitated. As the water table fell, the lake dried sufficiently to switch off the formation of pelletal clay, the lunette stabilised and a soil formed. ‘These ancient lakes preserve in considerable detail the imprint of past changes in the physical environment to which they responded with particular sensitivity,’ wrote Bowler (1971: 47). These lakes also preserved a rich archaeological record. Interstratified within the lunette sediments at Lake Mungo were numerous small archaeological sites – including burials, a cremation, hearths, grindstones and shell middens – mostly dating between 20,000 and 30,000 years BP (Bowler et al. 1970; Deserts Past: A History of Ideas 33

Figure 2.5. Jim Bowler’s 1971 map of the geomorphology of the Willandra Lakes, a chain of Pleistocene palaeolakes in western New South Wales, each bordered by a lunette. This was the ﬁrst map of the Pleistocene lake system. (Source:Bowler1971: ﬁg. 5.4) 34 The Archaeology of Australia’s Deserts

Barbetti and Allen 1972; Bowler, Thorne and Polach 1972). Taken collectively, these gave a detailed picture of a Pleistocene economy based on ﬁsh, crayﬁsh, waterfowl, freshwater mussels, small mammals and grass seeds. It was, HR Allen argued, indistinguishable from the riverine economy practiced by the Darling River Bagundji during the last century (1974: 315–16). One effect of the new data was to push the notion of a phase of mega-lakes, followed by continental desiccation, back into the Pleistocene (Bowler 1976). But although the chronology had shifted, the drying of the lakes – and the destruction of the late Pleistocene economy they once supported – reinforced Gould’s picture of a society forced to adapt to increasing aridity.

INITIAL COLONISATION OF THE DESERT The timing of initial human moves into the desert had not been at issue (Birdsell 1957; Tindale 1959; Mulvaney 1961). As John Mulvaney, writing in 1961,said:

Any migration southwards from the Gulf could easily move across to the inland drainage basin, which in a pluvial period would provide a passage down the Diamantina River or Cooper’s Creek systems towards central Australia, or down the Darling River and tributaries to the south and south-east of the continent. (1961: 62)

But the translocation of Crocker and Wood’s ‘great arid period’ into the late Pleistocene created problems for this scenario and raised new questions about the pattern and timing of initial human settlement in the interior. Was the arid interior of Australia settled as part of the initial dispersal of people across the continent, or was the aridity of the region a barrier to settlement by these early human groups? In the ‘coastal colonisation hypothesis’, S Bowdler (1977) argued that early settlement of the continent was focused on coasts and littoral resources (cf. Eden 1787; Tindale 1959: 38). Any late Pleistocene sites in the interior, she argued, represented either a ‘transliterated coastal economy’ centred on the rivers and lakes of the last glacial (for example, at Lake Mungo) or merely the inland extension of territories otherwise anchored in coastal regions (in the case of then-known Pleistocene sites on the Nullarbor or in the Pilbara). Effective cultural and economic adaptation to aridity – such as the harvesting of grass seeds and the hunting of large macropods – was a consequence of late glacial drying of palaeolakes in the Willandra, with effective settlement of the desert core only after 12 ka (Bowdler 1977: 229–30, 236). This echoed ideas popular early last century, including the ‘oasis theory’ (Pumpelly 1908), in which increasing aridity forced local hunter-gatherers to develop new ways of living (in Trigger 1989: 248), and Childe’s idea that desiccation associated with Deserts Past: A History of Ideas 35 postglacial temperature rise led to the domestication of plants and animals in the Middle East (1928). Horton (1981) took more explicit account of the evidence for active rivers and lakes in the interior prior to the LGM. Using the distribution of fossil megafauna as a proxy for the distribution of potable surface waters, he suggested that much of the continent could have been settled before 30,000 BP, with the exception of the central part of the interior. From 25,000 to 12,000 BP, the human population was restricted to coastal regions by the harsh conditions that prevailed across inland Australia. Then, 12,000 years ago, people ‘moved back, first into some of the areas previously occupied where sufficient water was now available again, then, with improving knowledge and technology, into the arid core itself, which had never previously been occupied’ (1981: 26). In its various forms, the idea that ‘the Mungo Lacustrine Phase would have given major access to the reactivated river and lake systems which ringed the arid heart’ ( Jones 1979: 453) gained some currency (White and O’Connell 1982; Ross, Donnelly and Wasson 1992; Thorley 1998a; Hiscock and Wallis 2005: 43). In effect, the desert margins were seen as an adaptive filter, allowing ‘gradual conditioning by Aboriginal populations to increasing aridity . . . in western New South Wales, northern Australia and elsewhere’ (Ross et al. 1992: 101), shaping the way of life of Aboriginal groups on the periphery of the desert in ways that positioned them to colonise the arid core of the continent. The use of wild grass seeds as grain was widely seen as the key economic adaptation to the arid interior (Tindale 1959, 1977; Gould 1977;Bowdler 1977) and instrumental in initial settlement of the region.

We propose that the onset of arid conditions 17,000–18,000 BP led to critical reductions in the abundance of high-ranked foods and favoured the adoption of more expensive items, including seeds. Once the technology for processing seeds was available, it was possible for Aborigines to move to previously uninhabited or sparsely inhabited parts of the continent. (O’Connell and Hawkes 1981: 115)

It was some time before archaeological fieldwork identified late Pleistocene occupation sites in the arid heart of the continent – and the new data shattered this neat picture of desert prehistory. The decisive evidence came from Purit- jarra, a large rockshelter west of the MacDonnell Ranges in Central Australia ( Jones 1987;Smith1987). Initial excavations showed this site had been occupied 22,000 BP and later work extended this back to 35,000 years ago (Smith 1987, 2006; Smith, Prescott and Head 1997). Other finds confirmed late Pleis- tocene occupation in Central Australia (at Kulpi Mara) (Thorley 1998a)andin the Western Desert (at Serpents Glen) (O’Connor, Veth and Campbell 1998). These finds also broke the nexus between seed use and early occupation: none of the early sites contained seed-grinding implements (Smith 1986a, 2004). 36 The Archaeology of Australia’s Deserts

As the earlier dates gained acceptance, interest shifted from questions of chronology to questions about the impact of intensiﬁed aridity during the LGM.

BEYOND THE WILLANDRA Perspectives from the Willandra dominated the environmental history and archaeology of the arid zone into the 1980s. The Mungo lacustral phase was extrapolated as a major period of expanded lakes across Australia between 50 ka and 30 ka (Bowler 1982: fig. 8). Bowler acknowledged its limitations, but ‘while the actual age of the expanded lacustral phase in northern Australia remains speculative, the evidence of greatly expanded water bodies is beyond doubt’ (1982: 42). The problem was that the history of the desert was being written from rivers and lakes on its southeast margin. In 1981, Australian National University (ANU) researchers (on Bowler’s initiative) launched an ambitious program to rectify this by exploiting the potential of palaeolakes across the continent as environmental archives, looking at climate change, O and C cycles, groundwater interactions, geo- and biochemical evolution of the lakes and vegetation history (Chivas and Bowler 1986). Called SLEADS (‘saltlakes, evaporates and aeolian deposits’), this was initially a four-year drilling program to extract cores from inland lakes. By the mid-1990s, when the project concluded, cores were available for half a dozen inland lakes (Lakes Eyre, Frome, Callabonna, Woods, Lewis and Amadeus), although not all produced detailed late Quaternary records. Dating was a problem in the early stages of SLEADS (especially at Lake Eyre; see Gillespie et al. 1991), but the wider availability of first TL and later OSL dating techniques resolved this. SLEADS was followed by John Magee’s detailed study of the palaeohydrology of Lake Eyre from 1989 to 1997 (Magee et al. 1995; Magee and Miller 1998), which used a combination of the new Quater- nary dating methods to provide a solid chronology (14C, AMS 14C, U/Th, AAR, TL and OSL). This provided the first fine-grained, well-dated sequence of environmental change spanning the last 100 ka in central Australia and shifted the focus back to Lake Eyre as the centrepiece of a suite of palaeolake studies. The picture that emerged from SLEADS, and the projects that flowed from it, unpacked the idea that there had been a single ‘lacustral phase’. The new evidence not only provided a longer time frame but also a stronger picture of variability in landscape response across the continent. The northern lakes (Lake Woods and Lake Gregory) showed a series of ‘lacustral’ phases over the last 300 ka, each centred on an interglacial (Hesse, Magee and van der Kaars 2004). Elsewhere in the Australian desert, most dates for enhanced fluvial activity (such as along channels feeding Lake Eyre) or high lake levels (for example, Lake Eyre and Lake Lewis) centre on 100 ka, late in the last interglacial. Deserts Past: A History of Ideas 37

The last deep-water phase of Lake Eyre in Central Australia ended 60,000 years ago (Magee and Miller 1998), and this also marked the decline of the Cooper and Warburton River systems feeding the lake (Maroulis et al. 2007). Along with SLEADS, the first detailed 14C chronologies for the continental dunefields came in the 1980s, with RJ Wasson’s work on the Quaternary history of the Strzelecki and Simpson dunefields (1983). Most dune-building occurred ‘between 25,000 and 14,000 years B.P., with a peak between 20,000 and 16,000 years B.P.’focussed on the extreme conditions of the LGM (Wasson 1986: 59–60). There were limitations, however, to what could be achieved with radiocarbon dating. Organics are rare in dune sequences, and Wasson found he needed to rely on the interdigitation of aeolian and fluvial records to build a reliable chronology. This tied him to the floodplains of Cooper Creek and the Warburton River. Although he experimented in the mid-1980swithTL dating of dunes (Gardner et al. 1987; Wasson 1986: 59), luminescence dating was not widely deployed for another decade. A suite of studies using TL and OSL now provides fine-grained chronologies for dune activity in the western and northern Simpson Desert (Nanson, Chen and Price 1995; Hollands et al. 2006; Twidaleet al. 2001) and in the Strzelecki and Tirari Deserts (Fitzsimmons et al. 2007). These show both a longer and more nuanced history of dune- building, as well as contrasting records of dune activity in the arid core and desert margins. The first long vegetation record from the interior was from Lake Frome. Gurdip Singh had begun work here in 1975, building on his success in extracting pollen records from dry lakes and saltpans in Rajasthan (Singh 1981). A second core was drilled as part of the SLEADS program and, by 1991, Singh could claim that apart from his work in Rajasthan, this provided the only pollen record from deserts around the world that extended back to the LGM (Singh and Luly 1991). The Lake Frome work neatly confirmed the emerging picture of late Pleistocene aridity and Holocene amelioration. Singh argued that the Lake Frome record reflected changes in the extent to which the summer monsoon had reached into the interior. The monsoon had been significantly weaker during the late Pleistocene. The accumulation of evidence for intense aridity during the LGM raised questions about how human populations had coped.

DESERT REFUGIA ‘A mountain,’ said Charles Darwin, ‘is an island on the land’ (1859: 371). It is not surprising, therefore, that one of the aims of the 1894 Horn Scien- tiﬁc Exploring Expedition to Central Australia was the discovery of ancient endemic plant species that had survived the desiccation of the continent in montane refugia. Ralph Tate, the expedition’s botanist, commented: ‘I had 38 The Archaeology of Australia’s Deserts

pictured a vast mountain system capable of preserving some remnants of that pristine flora which had existed on this continent in Palaeocene times – probably a beech, possibly an oak, elm or sycamore’ (Tate 1896: 118). This, said Tate, indicated ‘the desirability of a systematic exploration of the oasis of the McDonnell Range’ (1896: 118). But the low relief of these ranges was a disappointment. The ‘awakening came suddenly and rudely’ because ‘there exist no such features as could possibly maintain . . . a fauna and flora of a sub-alpine or cold temperate clime’ (1896: 119). Although the expedition recorded relict palms (Livistona) and cycads (Macrozamia), its major contribution to biogeography was the idea that the arid zone constituted a single biogeographic zone, which botanists, following Tate (1896), called the ‘Eremian’ or ‘Eremean’ zone, and zoologists, following Spencer (1896), called the ‘Eyrean’ zone. It was not until the work of Crocker and Wood, in 1947,whichdrew on a century of work on plant distributions, that refuges and aridity became the dominant paradigm (see also Crocker 1959). Their 1947 paper not only put a ‘Great Australian arid period’ on centre stage, it also provided a major explanatory model for plant distributions in the interior. Aridity in the early Holocene had destroyed the existing vegetation of the interior so abruptly that it was impossible even for xeric species to colonise the region quickly enough to stabilise bare ground. ‘The present-day plant communities are the result of re-colonisation of vast, virtually bare, areas,’ argued Crocker and Wood, who compared the effects of aridity to ‘the Pleistocene ice-ages in the northern hemisphere’ (1947: 129–30). Their vocabulary of refuges, barriers, migration and recolonisation has framed biogeographic thinking about arid-zone biota ever since. Burbidge (1960) argued that expansions and contraction of the arid zone had driven plant speciation but that the region now had its own ‘autochthonous’ flora. Jacobs (1982) suggested that speciation in spinifex (Triodia) had been driven by a series of expansions and contractions from refuges. Gentilli (1949) suggested that the avifauna of the desert derived from a series of colonisations from the desert margins, whereas Smith-White (1982) proposed that the desert margins had acted as an ‘adaptive filter’ for species dispersing into the desert. Much of this constituted a running debate over the extent to which the biota of the arid zone was structured by dispersal or vicariance – paralleling arguments in historical linguistics. Cracraft (1991) argued for a ‘historically cohesive’ biota in the desert – an ancient regional fauna (including birds, mammals, snakes, frogs and lizards) with links to the north and northeast that pre-dated the Pleistocene. Crisp, Linder and Weston (1995) put the contrary view: the biota of the arid zone is related to adjacent coastal areas and does not form a coherent grouping. Part of the problem, suggested Stafford Smith, West and Thiele (1996), is that intermittent isolation and small populations lead to ‘reticulate evolution’ and ‘noisy taxa’. Deserts Past: A History of Ideas 39

For arid-zone refugia, the analogue of northern hemisphere ice sheets proved to be misleading (Byrne 2008). In Australian deserts, refugia tend not to be discrete regions. Inland ranges may have been refugia for birds and ﬁsh and some mesic plant species (Heatwole 1987), but the wider picture is of a mosaic of localised ‘species-speciﬁc, idiosyncratic’ refugia (Byrne et al. 2008: 4411).

Islands in the Interior Evidence for a significant intensification of aridity during the LGM (20–18 ka) suggested that refugia might also have been an important factor in the human history of the region. Many archaeologists thought that inland ranges – with their gorge systems and waterholes – represented one of the few possibilities for the survival of late Pleistocene hunter-gatherer groups in the interior (Lampert and Hughes 1987; Hiscock 1988a;Smith1988; 1989b;Veth1989b; 1993; Thorley 1998a), with abandonment of other parts of the arid zone during the LGM likely. Mabbutt (1971a) had been the first to put this forward, writing well before the discovery of late Pleistocene human occupation in the arid zone:

The favoured environments of upland and piedmont and riverine tracts would have persisted throughout, although with ﬂuctuating limits and changes in relative importance which would lead to associated changes in the range and frequency of human migration into adjoining desert lowlands. (Mabbutt 1971a: 78)

The most systematic exploration of these ideas was by Peter Veth,in Islands in the Interior (1993). As aridity increased, human groups had retreated to glacial- age refuges in the major ranges – the Pilbara, Kimberley, Flinders Ranges and Central Australian Ranges. These ‘islands’ formed important nodes of adaptation to aridity, fostering development of social, technological and behavioural adaptations to arid lands. Although this was widely accepted in principle, there were doubts about the specifics. The persistence of occupation at Puritjarra in spinifex and sandhill country west of the main MacDonnell Ranges warned against too easy a correlation of refugia with inland ranges. Some refugia appeared to have been dispersed and cryptic: human groups at Puritjarra may have survived in sandhill country because a network of small waters provided access to a block of country large enough to support a viable population (Smith 1989a). Veth (1989a, 1989b) also drew attention to the lack of evidence for occupation of the major continental dunefields prior to 5,000 years ago and argued that these regions would have been first colonised in the mid-Holocene. At issue was whether access to the sandy deserts was determined by climatic fluctuations – and their impact on wells and soakages, and plant and animal resources – or whether there were other more fundamental barriers to 40 The Archaeology of Australia’s Deserts

settlement of these regions (Smith 1993;Veth1995b). Veth argued that a combination of demographic factors, new technology and new modes of social organisation allowed their settlement and that these had developed during a period of adaptation in the glacial refuges and in the long process of slowly recolonising the more accessible desert lowlands. Later finds of late Pleistocene or early Holocene sites in the sandridge deserts (O’Connor, Vethand Campbell 1998; Veth, McDonald and White 2008) undermined this scenario, showing that Aboriginal people were using the sandy deserts at least 25,000 years ago. These biogeographic and archaeological models were important for the way they emphasised the diversity of Australian deserts. However, the idea of well- defined glacial refugia – each a centre of dispersal for a coherent biota – was a northern European perspective that foundered on the more complicated situation in Australia’s deserts. A sandy desert was not the equivalent of an ice sheet. Some Australian species show evidence for expansion and migration from mesic habitats outside the arid zone, whereas others show persistence in multiple localised refugia over several glacial maxima (Byrne et al. 2008). For human prehistory, these models suggested a long uninterrupted occupation of the major inland ranges and river systems, contrasting with an ebb and flow of population in desert lowlands – with recolonisation of abandoned areas from centres within the arid zone. The cultural and genetic implications of rapid range expansion from a small subset of preglacial populations have barely been explored, but they point to the likely importance of bottlenecks for small desert populations repeatedly cut back by environmental change (Bennet and Provan 2008).

‘THE AUSTRALIAN ABORIGINAL AS AN ECOLOGICAL AGENT’ Not all changes involved humans responding to climatic shifts. Tindale saw ‘the Australian aboriginal as an ecological agent’, arguing that ‘much of the grassland of Australia could have been brought into being as a result of his exploitation’ and firing of the vegetation (1959: 42–3). Millennia of human fires not only allowed arid tropical grasslands to expand but also provided the basis for an economy based on the harvesting and wet milling of grass seeds (Tindale 1957, 1959: 49) – later labelled the ‘Panara or grass seed culture’ (Tindale 1977). Rhys Jones (1969) also emphasised Aboriginal use of fire to manage the land, calling it ‘firestick farming’, and speculated on its long-term impact on the face of Australia.

A Land Transformed? One issue was whether the initial entry of people into Australia around 50,000 years ago had led to a massive increase in fires that transformed the continent’s vegetation and denuded its soils (Latz 1995, 2007). Working with isotopic Deserts Past: A History of Ideas 41 signatures in fossil emu and Genyornis eggshell, G Miller argued that the modern desert landscape was a human artefact, the historical legacy of a landscape irrevocably altered by the initial impact of human settlement (Miller, Fogel, Magee et al. 2005). The decline of C4 plants in the diet of emus, and the extinction of Genyornis around 50 ka (Miller et al. 1999), signalled the destruction of a distinctive mosaic of palatable trees, shrubs and tropical grasses. Furthermore, the removal of woodlands in northern Australia had diminished moisture transfer into the Lake Eyre catchment, altering the hydrology of the desert core (Miller, Mangan, Pollard et al. 2005b). This neatly stitched together ideas about fire and megafauna extinction (Flannery 1994) and the catastrophic fire histories of Peter Latz (1995, 2007; Megirian et al. 2002), and it stressed the influence of people not just on landscape but also on climate. However, none of the long records of palaeovegetation provided much support for these ideas (Kershaw, van der Kaars and Moss 2003; Kershaw et al. 2006;Smith2009a). Rather, the evidence suggested that weakening of the summer monsoon circulation was the predominant factor at play ( Johnson et al. 1999), with a train of regional responses to increasing aridity in the Lake Eyre–Cooper Creek region. Archaeologists also had problems with Miller’s thesis (O’Connell and Allen 2012). Aboriginal population densities in the arid interior appear to have been low throughout much of the late Pleistocene. This means that human fires would have been localised and patchy and are unlikely to have had a major impact on desert vegetation (Bowman 1998;Birdetal.2008).

Landesque Capital Tindale’s work had also drawn attention to the more subtle and incremental effects of Aboriginal fires on local ecology. Arid Australia was one of the last places where the effects of a traditional burning regime could be observed at first hand (Gould 1971b;Birdetal.2008). In the 1990s, Flannery’s ideas stimulated renewed interest in the issue of fire ecology (Bowman 1998;Burrows et al. 2006; Bird et al. 2008). These studies showed that repeated burning of vegetation during foraging trips led to the formation of small-scale successional mosaics, which in turn increased the productivity of the country in respect of plant foods and small game. The main effect was on the grain of the country, with greatest impact on habitat structure at the local rather than regional level. The effect of this patch-burning was cumulative and concentrated in localities were people spent the most time. ‘Aboriginal foragers,’ said Bird et al., ‘thus construct their own ecosystem’ (2008: 14799). Some archaeologists saw this as a form of ‘landesque capital’ (improvements to a productive estate), associated with the growth of population and the expansion of settlement that was increasingly evident in the last 1–2 millennia (Smith 1988;Veth1993; Smith and Ross 2008a). 42 The Archaeology of Australia’s Deserts

SOCIAL INTENSIFICATION Fire was not the only area where people were seen as having some agency. Large intergroup ceremonial gatherings, such as the Arrernte ingkwere,werea powerful demonstration of the way that religion housed the ruling ideology in these hunter-gatherer societies, as well as the way in which wealth and prestige were deﬁned in terms of ritual property and knowledge. This raised the question of whether incentives to increase the size and scale of ceremonies could have constituted an autonomous driver of social and economic change, independently of any pressure from environmental or demographic shifts. H Lourandos (1985, 1997) suggested that from about 5,000 BP,an escalation of competitive changes in social relations in the arid zone (including an increase in communal gatherings, territoriality and the scale of exchange and alliance networks) would have put pressure on production systems, leading to a reliance on grass and acacia seeds as staple foods, and that this in turn triggered growth in regional population and settlement of marginal areas, such as the sandy deserts. As a heuristic framework, these ideas were readily adopted by archaeologists interested in exploring a richer social history than ecological and environmental perspectives alone could provide (Veth 1989a; Ross, Donnelly and Wasson 1992;David2002; Ross and Davidson 2006). However, they proved difﬁcult to test empirically. There were problems in explaining why these changes were triggered in the late Holocene and not earlier. And the available archaeological evidence suggested a different sequence of events: changes in ceremonial structure appeared to lag behind, rather than lead, changes in population and economy (Smith 1988;David2002; Smith and Ross 2008a). These ideas were important, however, because they represented a renewed interest in the political and social dynamics of desert groups and in the development of their key institutions.

WRITING THE HISTORY OF THE DESERT Looking back over this history, several aspects stand out. First, ideas about the history of the land and its people have been intertwined almost from the moment colonial exploring parties first encountered the desert. This is seen most clearly in the impact the new Quaternary dating methods had in the 1960sand1990s, when ideas about land and people shifted rapidly to accommodate the new chronologies. Colonial scientists were able to provide an outline of the Quaternary history of the Lake Eyre region by the 1880s, less than 40 years after the first British exploring parties entered the interior. By 1886, a broad outline was in place that read the field evidence as indicating that the climate of the interior had been much wetter in glacial times and that this had been followed by the onset Deserts Past: A History of Ideas 43 of aridity and the creation of the immense dunefields that ring the centre of the continent. The fundamental problem was that there was no way to establish a timescale for these events: the order of events could be worked out tolerably well, but their chronology was utterly uncertain. This framework collapsed in the 1960s, as radiocarbon dating began to provide an independent chronology, and it changed again in the 1990s, as luminescence dating provided the means to unpack the environmental history of the interior into a series of arid and fluvial phases, and as the marine isotope chronology (Martinson et al. 1987) provided a new global framework for Australian data. Another striking aspect is the way in which the history of the desert has been written largely from its southeastern corner. The sense of a fossil, ‘dead’ landscape is strongest in the Lake Eyre region, with its saltlakes and stranded beach ridges, abandoned or buried river channels and rich fossil fauna. It is not surprising that into the 1950s, this area drove most thinking about the environmental history of the desert. The same is true of early research into desert rock engravings, work that between 1900 and 1950 was concentrated in the Flinders Ranges, the Olary upland and the Broken Hill region, those parts of the arid zone most accessible to researchers in Adelaide, Sydney or Melbourne. Research in the Willandra continued this trend, shifting the focus of field research to the desert margins, where the rich palaeoenvironmental and archaeological records from Lake Mungo dominated thinking about the prehistory of arid Australia throughout the 1970sand1980s. The major dune chronologies also relied on sequences worked out in the Strzelecki Desert and Cooper basin, because even with radiocarbon dating, aeolian sequences were difficult to date directly, except where they were intercalated with fluvial or lacustrine sediments. In contrast, the classic desert ethnographies follow the frontier: beginning with JG Reuther’s work at Killalpaninna on Cooper Creek between 1888 and 1906, followed in 1899 by Spencer and Gillen’s accounts of the Arrernte in Central Australia and WR Roth’s research in northwestern Queensland between 1897 and 1910, and shifting to the Western Desert after 1950 as the last remaining desert hunter-gatherers were drawn into missions and government settlements at Warburton, Jigalong, Balgo, Yuendumu and Papunya. After a tentative start at the Koonamore research station in the Olary region in 1926, arid-zone ecology has also centred on research in the spinifex and sandhill country of northern and central Australia. Given the diversity of Australia’s deserts, there is a danger here that while the history of the arid zone is written from its southeastern margin, the processes that govern its people and ecology are framed by work in central and northern Australia. A third aspect is the way in which recurring themes in different disciplines collectively provide an outline of persistent historical structures in Australia’s desert. The desert margins – the zone most affected by expansion and contraction of the arid zone – are widely seen as an area where species are conditioned 44 The Archaeology of Australia’s Deserts

to aridity, allowing them to disperse widely across the interior. Desert uplands, in contrast, are more stable nodes within the desert, centres of endemism and local dispersal, as well as more deeply etched human landscapes. The desert lowlands experience rapid spread of species, languages or cultural traits, periodically integrating large areas of the desert, followed by local differentiation until the next event. Finally, there is a surprisingly rich ﬁeld of ideas about the social development of desert societies – their languages, economies and key institutions – that runs counter to the notion that these were somehow a static relic of an earlier time. For desert people themselves, notions of historical change are subordinated to the ruling ideology, the ‘dreaming’. As Tonkinson says, ‘nowhere does their ideology admit structural change as a possibility’ (1978: 112). However, a century after the Alice Springs ingkwere, most Western scholars see these societies as the product of a long period of indigenous development, with their roots in the late Pleistocene but having been transformed by structural changes in the late Holocene and cultural efﬂorescence during the last millennium. CHAPTER 3

THE EMPTY DESERT: INLAND ENVIRONMENTS PRIOR TO PEOPLE

At the close of the nineteenth century, fossil finds at Lake Callabonna altered perceptions of the inland, showing that even the driest and bleakest parts of the Australian interior once teemed with life (Editorial, The South Australian Register, 24 May 1893). In 1893, members of a museum expedition to Lake Mulligan (as it was then known – a corruption of malakanha, the Adnyamathanha name for ‘a type of string bag’) found the saltlake covered with bones of extinct giant marsupials, ‘which have apparently died where they lie, literally in hundreds’ (Stirling 1900: iii) (Figure 3.1). The most common remains were of Diprotodon optatum, a large herbivore, which at 2.8 tonnes was the largest marsupial ever to have lived (Wroe et al. 2004)(Figure 3.2a). Hundreds of these animals had been trapped as they broke through the salt crust into the saturated clays below while attempting to cross the lake bed to nearby springs. The museum dig did not get off to a good start. When Edward Stirling, the director of the South Australian Museum, visited the site, he found ‘utter carelessness, slovenliness, and want of system’. It would, he said, ‘be difficult to imagine the camp of a scientific expedition in a more disgraceful condition’ (EC Stirling, Report to Board of Governors, 28 August 1893). Stirling promptly dismissed HC Hurst, the geologist initially contracted to extract the fossils, and left ACH Zietz, a museum officer, to assume direct control of the work. For three months, Zietz and his team worked at Callabonna, under appalling conditions: ‘Innumerable flies were, in the day-time, a constant and maddening source of annoyance’ (Stirling 1894: 210). Heavy rain, frequent sandstorms, daytime temperatures of 120◦F(49◦C) and bad water led to ‘ophthalmia and gastro-intestinal complaints’ (1894: 211). ‘Every stick of firewood’ and every drop of drinking water had to be carried in by camels. Meat sometimes ‘went bad before the day was out’ (1894: 187). Many of the fossil skeletons were exposed on the surface, outlined by patches of fragmented bone and pieces of tooth enamel. Zietz located buried skeletons by probing through the lake mud with steel rods. Rabbits swarmed over the diggings, 45 46 The Archaeology of Australia’s Deserts

Figure 3.1. The 1893 dig at Lake Mulligan (Lake Callabonna), showing excavation of a Diprotodon skeleton on the ﬂoor of the saltlake. (Photograph AA309,Stirling collection, courtesy South Australian Museum Archives)

destroying many bones in a frantic attempt to get at the salty groundwater seeping into hollows around Diprotodon skeletons. Bones disintegrated as sodium sulphate salts crystallised in fracture lines and cavities. By the close of the dig, the museum had recovered the remains of about 100 animals (Stirling 1894: 209), including the first complete skeletons of Diprotodon and partial skeletons of Genyornis (a large flightless bird), Phascolonus (an extinct giant wombat) and Sthenurus (an extinct short-faced kangaroo) (Stirling 1900;Pledge1994; Wells and Tedford 1995). In his book The Dead Heart of Australia, JW Gregory summed up the new picture. The Lake Eyre basin had formerly been ‘a vast inland sea’ but with declining rainfall, the lake had dried: ‘hot winds swept across the dusty plains, and the once fertile basin of Lake Eyre was blasted into desert’ (1906: 150–1). Later research has shown that these arid landscapes preserve a palimpsest of ‘lacustral’ or mega-lake phases of varying age and declining amplitude – beginning with the perennial lakes and palaeodrainage systems of the late Tertiary landscape (about 24 Ma), with active but more limited rivers and palaeolakes during marine isotope stages (MIS) 5 and 7 (132–71 ka; 236–186 ka), and a modest reactivation of semi-arid lake and river systems early in MIS3 (60–45 ka). This raises an obvious question. What was the desert like on the eve of human settlement? Did the first people entering this region encounter deserts and drylands, or did they find a landscape of rivers and lakes? In this The Empty Desert: Inland Environments Prior to People 47

Figure 3.2. Distribution of wide-ranging megafauna species: (a) Diprotodon optatum (Reconstruction from Murray 1984:ﬁg.27.10). Shoulder height is approximately 2 m. (Distribution map is based on Horton 1984:ﬁg.29.8). (b) Genyornis newtoni, reconstructed as a robust gooselike anseriform bird (courtesy Peter Murray). The height of Genyornis is estimated to have been 2.15 m. The distribution of this species is based on fossil bone (Horton 1984: 29.8) and eggshell. (Miller et al. 1999; personal communication) chapter, I review recent work on the age and origin of Australian deserts and their biota and look in more detail at Quaternary environments in MIS5 prior to human settlement of the region.

THE ‘DESERT TRANSFORMATION’ CONCEPT One possibility, widely canvassed, is that at the time of colonisation, aridity may have been only modest and that the ﬁrst colonists would have found a richer landscape, one with active rivers and large permanent lakes. In 1961, DJ Mulvaney wrote that ‘reliable rainfall extended across the heart of the continent. Perennial rivers drained into inland seas . . . there was a spread 48 The Archaeology of Australia’s Deserts

of rain-forest flora over the centre of the continent ...[and] a diversified and predominantly herbivorous land fauna’ (Mulvaney 1961: 63). Later, after Bowler’s research on the Willandra Lakes, the idea that ‘the Mungo Lacus- trine Phase would have given major access to the reactivated river and lake systems which ringed the arid heart’ ( Jones 1979: 453) gained currency among many researchers (White and O’Connell 1982; Hiscock 1988a; Ross, Don- nelly and Wasson 1992; Thorley 1998a; Hiscock and Wallis 2005: 43). In her ‘coastal colonisation hypothesis’, Bowdler (1977) argued that early sites in the Murray–Darling basin represented a ‘transliterated coastal economy’ centred on rivers and lakes. Thorley (1998a) extended this idea to Central Australia, suggesting that Lake Amadeus and the Finke–Palmer river system would have provided more regular and abundant resources at the time of colonisation than today. Hiscock and Wallis brought these ideas together in their ‘desert transformation’ model, arguing that the presence of large permanent water bodies prior to 45–40 ka ‘facilitated exploration and exploitation of these unique interior landscapes’ and that ‘the modern deserts of Australia came to inland dwelling people, rather than the reverse’ (2005: 41–3). The alternative view – which I endorse in this book – is that the first people to move into the interior found an arid landscape and xeric biota. This was a desert whose vital statis- tics were somewhat different from the desert today but an arid environment nonetheless. Chronology is critical to this question. Luminescence and palaeomagnetic dating have recently provided the means to unpick the series of fluvial and lacustrine events in the interior. As a more detailed environmental history has emerged, these archaeological ideas risk creating an amalgam of landscapes of widely different age and potential, conflating evidence from the last interglacial (MIS5.5, 132–115 ka) with conditions in early MIS3 (60–45 ka).

AGE AND ORIGIN OF AUSTRALIA’S DESERTS Australia’s deserts date to the early Pleistocene (0.9–2.0 Ma), when arid landforms such as dunefields, saltlakes and stony deserts first appear in the Cainozoic record. However, the development of aridity in Australia appears to have pro- gressed in a stepwise fashion (McLaren and Wallace 2010), and many elements of drylands biota appear much earlier, in the Pliocene (5–2.6 Ma) (Martin 2006). To understand the landscapes of arid Australia, we need to look beyond the Pleistocene to the late Tertiary. During the last 24 million years, Australia has been gradually transformed from a humid, forested land to an arid continent. A transition to drier conditions began in the mid-Miocene (15 Ma), as continental drift brought the continent into the mid-latitudes, and as strengthening of the Antarctic circumpolar current drove changes in global circulation (Bowler 1982). During the early part of this period, large perennial lakes dominated The Empty Desert: Inland Environments Prior to People 49 the central Australian landscape. These have left distinctive sedimentary units of lacustrine clays and dolomites across much of the region. The best known of these is the Etadunna formation in the Lake Eyre basin. Palaeomagnetic dating provides a late Oligocene age (26–24 Ma) for the Etadunna formation (Woodburne et al. 1993), but biostratigraphy suggests that the upper parts of the formation extended into the late Miocene. Whatever the case, Etadunna clays and dolomites represent the basal stratum underlying most Quaternary sequences in the Lake Eyre region. The Etadunna formation records an extensive shallow alkaline lake – or series of lakes – fluctuating in depth and extent and surrounded by alluvial plain and fan sediments (Wasson 1982; Alley 1998). These lakes supported an abundant aquatic fauna, including catfish and lung- fish, crocodiles, chelid turtles and river dolphins, as well as a diverse assemblage of waterfowl, including waders and flamingos (Miller 1963). Equivalent late Miocene units in Central Australia show that large perennial freshwater lakes were also present at Lake Lewis (Anmatyerre clay, English et al. 2001)and Lake Amadeus (Uluru clay, Chen and Barton 1991; Chen, Bowler and Magee 1993). Like many saltlakes, both Lewis and Amadeus are part of chains of playas following ancient palaeodrainage networks that last saw significant flows prior to 15 Ma (Van de Graaff et al. 1977;Magee2009). Palaeomagnetic dating of the laterites that form extensive hardpans across western and central Australia also shows that the main period of laterite formation was in the late Oligocene– early Miocene (Alley 1998), consistent with warm and wet climatic conditions throughout the Miocene. Plant fossils show that the Lake Eyre basin was covered by forest until about 24 Ma, with rainforest and swamp vegetation in valley bottoms and seasonally dry sclerophyll forest on slopes and ridges (Truswell and Harris 1982;Hill 1994;Martin2006). There is little evidence, however, for grasslands at this time (Martin 1990). Until recently, the distinctive Livistona palms in the Finke Gorge, Central Australia, were thought to be a relict of this Tertiary climate (Latz 1975; Wischusen, Fifield and Cresswell 2004; Crisp et al. 2010), but genetic research now shows that these palms were introduced during the late Pleistocene (Kondo et al. 2012). Evidence from the Pilbara shows that dry forests – mainly eucalypts and casuarina – with small patches of rainforest extended across the interior until the mid-Miocene (Macphail et al. 1994: 229). Increasing continental desiccation after 15–12 Ma is reflected in the end of the Miocene lakes – although these basins continued to structure later (less extensive) lacustrine phases. In Central Australia, Lake Amadeus switched from shallow freshwater conditions to a groundwater-controlled playa at 1.6–0.91 Ma, with the formation of gypsum shoreline dunes by 0.98–0.73 Ma marking the onset of pronounced seasonal aridity in the heart of the continent (Chen et al. 1993). Lake Lewis made the transition to saline conditions somewhat later, at about 350 ka (English et al. 2001). 50 The Archaeology of Australia’s Deserts

Drier conditions during the Pliocene are also reflected in the formation of widespread silcrete duricrusts that cap the Etadunna formation and create the distinctive central Australian landscape of tabletop hills and breakaways (Mabbutt 1971b; Alley 1998). Cosmogenic exposure ages show that by 4 Ma, the stripping of soil mantles from these silcrete tablelands may have begun, leading to the formation of Australian stony deserts between 4 and 2 Ma (Fujioka et al. 2005). Exposure ages on dunes in the western Simpson Desert also show that this enormous dunefield had begun to form by 1 Ma (Fujioka et al. 2009). Plant fossils show corresponding changes in palaeovegetation. The Pliocene saw a shift to dry eucalypt and casuarina woodlands, chenopod shrubland, and a major expansion of grasslands. Martin (2006) points out that many typically arid-zone taxa (for example, eucalypts, acacia, chenopods) were in Central Australia long before it became arid, occupying other niches (such as sandy or saline lakeshores). Continental desiccation simply filtered out taxa that could thrive in new circumstances. These environmental changes set the stage for a rapid evolutionary diversification of a range of dryland animal taxa. Most extant Australian marsupials owe their existence to a proliferation of species adapted to drylands – the marsupial ‘radiation’ – which took place in response to increasing continental aridity and the loss of forests and rainforests after 15 Ma (Long et al. 2002; Dawson and Dawson 2006; Dickman and Ganf 2007; Byrne et al. 2008). The earliest grazing kangaroos (Macropodines) and wombats appear in the early Pliocene, and these lineages diversified rapidly to exploit the expanding grasslands. Another group that was common in the late Tertiary fauna are the dromornithids: a family of large terrestrial birds that includes Genyornis and Dromornis. The Alcoota site, northeast of Alice Springs, provides one of the few fossil assemblages from Central Australia dating to the late Miocene (at about 8 Ma) (Megirian, Murray and Wells 1996; Murray and Vickers-Rich 2004). It represents waterhole-tethered mortality during a drought and is dominated by dromornithids and diprotodontid herbivores – but also includes thylacines (marsupial wolf ), crocodile, large varanid lizards (Megalania), ducks and flamin- goes. This diversification was accompanied by an evolutionary trend towards larger body mass that reached its most extreme expression in the 2.8 t Diprotodon optatum (see Figure 3.2a). In this sense, the Australian megafauna were a product of the development of Pliocene and Quaternary drylands. Larger herbivores coped better with a low-quality fibrous diet, but this advantage was offset by longer generation times, smaller populations and a greater reliance on free water (Archer and Hand 2006). The corollary of this is that although these animals were adapted to drylands, they were unable to cope with open desert or high environmental stochasticity. The Empty Desert: Inland Environments Prior to People 51

Figure 3.3. Climatic variability over the last 350 ka, showing the numbering and amplitude of marine isotope stages (MIS) for the late Quaternary (data from Bassinot et al. 1994). The last interglacial is MIS5.5. The last glacial maximum is MIS2.2.

Collectively, these data suggest that much of the desert landscape we see today had taken shape by the early Pleistocene, at a time when global cooling initiated the Quaternary ice ages. Open dryland habitats, such as grasslands and stony desert, developed earlier, in the Pliocene, and were accompanied by an expansion of dryland biota. But the diverse components of the modern desert biome were not assembled until the early Pleistocene.

THE LAST INTERGLACIAL IN AUSTRALIAN DESERTS

Quaternary Context The Pleistocene saw a much less stable climate, as global circulation adjusted to the periodic growth and decline of high-latitude ice sheets and glaciers. This period was marked by a repeated cycle of rapid global warming followed by more gradual cooling. These oscillations form a distinctive sawtooth pattern, with the warm peaks roughly 100,000 years apart (Figure 3.3). The cornerstone of this Quaternary sequence is the deep-sea oxygen isotope record from marine foraminifera, which define a series of marine isotope stages (Martinson et al. 1987; Bassinot et al. 1994). By convention, the cold, dry glacial stages are given even numbers. Warm interglacial stages have odd numbers. In Australian drylands, these climatic oscillations are mainly reflected in cyclic fluctuations in available moisture: interglacials reactivated lakes and rivers across the interior; the colder drier glacial periods are marked by active dunefields and windblown dust. 52 The Archaeology of Australia’s Deserts

The most important of these climatic oscillations for understanding the natural prehistory of Australia’s deserts is MIS5, which encompasses the last interglacial (substage 5.5 or 5e). This period represents the last signiﬁcant imprint of widespread ﬂuvial and lacustrine activity in the arid zone. It also provides a key benchmark for comparison because we can use conditions during the last interglacial as a best-case scenario for conditions likely to have been encountered by people moving into the desert 45,000 years ago (see Chapter 4). In this context, it represents the upper limit of the range of possibilities for these inland landscapes. Globally, the last interglacial (MIS5.5) was the last time when global climates were similar to those today (Kukla et al. 2002). It is the warmest interglacial in the 800 ka EPICA Dome C ice-core record (Masson-Delmotte et al. 2010) and may have been a little warmer and wetter than present. It is conventionally dated to 132–115 ka (Shackleton et al. 2002), but new thermal ionisation mass spectrometry (TIMS) uranium/thorium (U/Th) ages, coupled with a δ18O record of precipitation from speleothems at Dongge Cave, China, indicate a narrower span of 129–120 ka (Yuan et al. 2004). Most records then show a shift to progressively colder and drier conditions from 116–115 ka, with rapid swings in climate (Kukla et al. 2002). MIS5 ended about 71 ka as the world entered another dry, windy glacial phase. This is also marked by a peak in windborne dust at 70–60 ka in the Vostok ice core in Antarctica (Petit et al. 1999). Palaeoenvironmental records in Australia’s deserts register these shifts in various ways: ‘reduced monsoon rain in global cold stages caused lakes and rivers to dry, vegetation to become more sparse, sand dunes to become active, and dust advection to increase’ (Hesse, Magee and van der Kaars 2004: 118).

Lakes and Saltlakes One of the paradoxes in Quaternary research is that the history of deserts is commonly a history of lakes and rivers. In Australia, the sedimentary record of lakes and rivers in the arid and semi-arid zones has been the cornerstone of late Quaternary research in these regions. The best-developed records are those from Lake Eyre, Lake Frome and the Willandra Lakes, but important data have also come from Lake Amadeus and Lake Lewis, in Central Australia, and from Lake Gregory and Lake Woods on the northern desert margin. Bowler’s ‘hydrological threshold’ (1981) provides the key to understanding variability in these lake basins and their differential response to climatic shifts. Lakes in closed basins are controlled by catchment size (relative to lake area) and the hydrologic balance between inﬂow and evaporation (Figure 3.4). Most of the large saltlakes, dotted across Central and Western Australia, have small catchments relative to their surface area and are unlikely to have ever functioned as palaeolakes during the late Quaternary. In most cases, these have remained as playas, controlled by groundwater processes, for some 700 ka. In contrast, The Empty Desert: Inland Environments Prior to People 53

Figure 3.4. Bowler’s ‘hydrological threshold’ for Australian lakes. This shows the relationship between catchment and lake areas (Ac/Al) and climatic parameters. Some ephemeral lakes with large catchments will become permanent water bodies with only a modest climate change. Other lakes, which are playas under modern conditions, require large shifts in climate to change state. The climatic function (Fc) is (0.8E – P/Pf ) +1, using a pan evaporation (E) factor of 0.8 for the lake surface, precipitation (P) and a runoff coefficient (f ) of 0.1.(AfterBowler1981:fig.30.4; English et al. 2001:fig.3) lakes that have the ‘amplification’ of large catchments are particularly sensitive to changes in climate; with comparatively small changes in precipitation or evaporation, they can switch from dry lake basins to ephemeral or permanent palaeolakes. Amplifier lakes are rare in the desert. The main examples are Lake Eyre and Lake Lewis and lakes on the desert margins fed by catchments outside the arid zone, such as the Willandra Lakes, Lake Gregory, and Lake Woods. lake eyre: ‘a continental rain gauge’ Lake Eyre, today an ephemeral playa lake (9,690 km2), forms the centre of Australia’s largest internal drainage basin, covering 1,215,000 km2 (Figure 3.5). It is fed by summer monsoon rainfall in the north and receives water through the Warburton–Diamantina–Georgina and Cooper–Thomson–Barcoo river systems from the northeast, and the Macumba and Neales Rivers from the west. In the past, Central Australian rivers, such as the Finke and Todd, which are now cut off by the Simpson Desert dunefield, also reached the lake. Because Lake Eyre is sensitive to changes in precipitation over such a large proportion 54 The Archaeology of Australia’s Deserts

Figure 3.5. The Lake Eyre basin during the last interglacial (MIS5.5), showing the location of Lake Callabonna, the extent of large palaeolakes (after DeVogel et al. 2004) and major ﬁnd spots for Katapiri formation megafauna assemblages (after Webb 2006: map 5.1).

of Australia, geomorphologist John Magee has called it ‘a continental rain gauge’. In this context, the single most important contribution to the Quaternary history of the desert has been J Magee’s reconstruction of the lake’s palaeohydrology for the last 130 ka (Magee et al. 1995;Magee1997; Magee and Miller 1998)(Figure 3.6). His work on the stratigraphy of shoreline features and sedimentology of lacustrine sequences, and the dating and cross-correlation of these using a range of Quaternary dating techniques – including optically stimulated luminescence (OSL), U/Th, amino acid racemisation (AAR), and radiocarbon (14C) – shows there was a major lake expansion during MIS5 that The Empty Desert: Inland Environments Prior to People 55

Figure 3.6. Lake-level curve for Lake Eyre (after Magee et al. 2004:fig.2, updated from Magee 1997:fig.6.4). Dates shown are pooled ages (TL/OSL) for each lacustral phase. Lake levels are plotted relative to Australian Height Datum (AHD). created a deep permanent water body between 130 and 90 ka. This immense interglacial lake was up to 25 m deep and extended over more than 25,000 km2 (see Figure 3.5). By comparison, the 1974 flooding of Lake Eyre (Bonython and Fraser 1989) – the largest historical filling on record – created an ephemeral lake only 6 per cent of the size of the MIS5 palaeolake (Table 3.1). Although the interglacial palaeolake was permanent, it was mostly saline, with marked salinity stratification and anoxic bottom conditions. This ‘inland sea’ was not rich in resources compared with shallow freshwater lakes such as the Pleistocene Willandra Lakes. Shells and shell fragments are abundant in beach sediments. These are mainly Corbiculina bivalves and the lake-bottom gastropods Coxiella and Coxiellada (Magee 1997), none of which is a prime shellfish species for hunter-gatherers. The most common eggshell in these beach deposits is pelican. Numerous small fish bones (vertebrae, head plates and spines) are present, including some dense, thin, horizontally table 3.1. Lake Eyre, comparing the size of the last interglacial lake and the largest historical filling

Lake area Water depth Lake volume Lake Eyre km2 mkm3

Modern playa 9,690 0 0 Historic 1974 ﬂooding 8,020 6 21 Final deepwater lake 65 ka 9,920 19 74 Last interglacial MIS 5.5a 25,260 25 332 Last interglacial MIS 5.5b 34,250 25 430 a Excludes Frome-Gregory basin. b Includes Frome-Gregory basin. Source: DeVogel et al. 2004 56 The Archaeology of Australia’s Deserts

continuous accumulations of fish bone. These suggest a cycle of boom and bust associated with flood events. Today, ephemeral floods are notable for the high levels of nutrients flushed into the lake, leading to blooms of invertebrates, followed by rapid growth of fish and waterfowl populations, and the collapse of this system as the halite crust dissolves. Because any disruption to the salinity stratification of lake waters has catastrophic consequences for invertebrates, fish and waterfowl populations, the ecology of the interglacial lake is likely to have been characterised by high levels of environmental stochasticity. After 90 ka, the lake entered a long unstable phase, during which it frequently switched state – occasionally drying out, or oscillating among shallow ephemeral conditions, a deep saline lake, and brackish or fresh interludes. The final pulse of the lake was at 65–60 ka, when water returned in a still-stand that created a beach rich in tiny Coxiellada shells (a ‘coquina’): ‘This represents the last deep-water perennial lake in the basin’ (Magee et al. 2004: 886). From 65 ka to 60 ka, the lake switched to a groundwater-dominated playa (Magee and Miller 1998) – apart from minor lacustrine events around 50–40 ka and between 12 ka and 4 ka that each created a saline lake 5 m deep, smaller than the 1974 filling.

other inland lakes The palaeohydrology of Lake Eyre reflects a stronger, more active monsoon during MIS5, and it is not surprising, therefore, that other inland lakes, especially in northern Australia, share the imprint of this. On the northern margins of the desert, Lake Gregory and Lake Woods have palaeo-shorelines that indicate both were much larger perennial lakes about 100–96 ka (Bowler et al. 1998; Bowler, Wyrwoll and Lu 2001). Both remain active ephemeral lakes today, representing terminal lakes fed by inflows from the northern subtropics. These are likely to have been freshwater systems: at Lake Gregory, this is confirmed by the remains of beds of Velesunio mussels. The northern lakes are matched on the southeastern margin of the desert by the Willandra Lakes, which are fed by runoff from the southeast Australian highlands. These also appear to have had high lake levels during MIS5 (represented by the Golgol unit at Lake Mungo), but this is only poorly dated (>126–98 ka) or defined (Bowler and Price 1998). All of these lakes appear to have been reactivated to varying extent early in MIS3 (50–45 ka), making them an obvious focus for hunter-gatherer groups moving into the desert (for example, Bowler et al. 1998;Vethetal.2009). In Central Australia, Lake Amadeus was a hypersaline playa during the last interglacial, although the regional water table was higher than it is today. The deposition of shoreline gypsum before 82 ka and from 60 ka to 45 ka shows that the lake waters were mainly saturated brines, subject to seasonal drought, with deflation of sand-sized CaSO4 from the near-shore seepage zone to form local dunes (Chen et al. 1993). Nearby Lake Lewis shows a similar pattern The Empty Desert: Inland Environments Prior to People 57

(English et al. 2001). By 80–70 ka, this was a groundwater-controlled basin, with shallow lake waters, high levels of salinity and precipitation of carbonates and sulphates (Chen, Chappell and Murray 1995; English et al. 2001). To the southeast of Lake Eyre, Lake Frome appears to have been a perennial standing body of brackish water during the last interglacial (this mega-lake is termed Lake Millyera by Callen, 1984). This dried towards the close of MIS5. Callen saw this as recording ‘the demise of permanent brackish lakes in the northeast of South Australia’ (1984: 172). However, recent research shows Lake Frome refilled again around 70–60 ka and 48–45 ka (Cohen et al. 2012), with fluctuating salinities and water levels since then (De Deckker, Magee and Shelley 2011). From 30 ka, the lake became increasingly saline and ephemeral, with occasional flushes of fresh water and, after 20–15 ka,itwasmostlya saline, groundwater-controlled playa. This shows that fluvial and lake systems on the southeastern margins of the desert responded to shifts in winter rainfall and reduced evaporation and that they may have different histories from lakes further north. Lake Callabonna is one of a chain of saltlakes linking Lake Eyre and Lake Frome. An OSL age of 75 ± 9 ka for sediments inside one of the excavated Diprotodon crania (Roberts et al. 2001) suggests this system last carried water towards the end of MIS5. Both the presence of surface skeletons and the preservation of Diprotodon trackways on the lake floor (Tedford 1984) show that there cannot have been significant reactivation of this lake since that time. the arid rivers The river systems feeding Lake Eyre indicate stronger stream flows during the last interglacial, with waning activity after 85 ka. The Diamantina River and Cooper Creek are Australia’s best known arid river systems. Two characteristics warrant comment here. First, these are river systems on an immense scale. The catchments of the Georgina–Diamantina– Warburton and the Thomson–Barcoo–Cooper drainages begin in northern Australia and extend for more than a thousand kilometres into central Australia, feeding monsoonal floodwaters into Lake Eyre. Second, the dynamics of the slow movement of floodwater into the heart of the continent – through a variety of sumps and storages – adds another layer of environmental variability over and above that due to fluctuations in precipitation. The fluvial history of the Cooper and the Diamantina is now known in some detail. The major phase of fluvial activity on the middle and lower reaches of these rivers peaked around 110–109 ka and left a series of large palaeochannels, now buried by desert dunes and modern floodplains (Wells and Callen 1986; Nanson, Price and Short 1992;Magee1997; Nanson et al. 2008). These ancestral rivers created a broad meandering channel system up to four times the width of the modern floodplain. The palaeochannels contain a wide range of fossil fauna (which attracted the attention of JW Gregory in 1901–2)and 58 The Archaeology of Australia’s Deserts

are filled with fine cross-bedded white sand, indicating higher discharges than the present mud-dominated channels. Collectively, these channel deposits are known as the Katapiri formation, usually divided into the upper Katapiri last interglacial and the lower Katapiri (260–220 ka). OSL ages for the upper Katapiri show that these deposits range in age from 120 ka to 60 ka. Higher in the catchment – in the Channel Country – sand-dominated bed-loads indicate higher discharges around 120 ka (Nanson et al. 1988), but these were replaced from 85 ka by the network of low-energy anastomosing muddy streams that today give this region its name. Stream flows along Cooper Creek appear to have become strongly seasonal towards the end of MIS5.By85 ka, it formed large, sandy, meandering channels, with seasonally dry beds that allowed the prevailing winds to blow channel sands into source-bordering dunes. OSL dates for these source-bordering dunes form several age clusters, ranging from 120 to 100 ka and from 85 ka to 40 ka (Maroulis et al. 2007; Cohen et al. 2010). After 40 ka, a switch to increasingly low-energy flows inundated the floodplains and channels with mud. For Cooper Creek, ‘increasing aridity had greatly diminished this major Australian inland river by about 40–35 ka’ (Nanson et al. 2008: 126). The fluvial histories of other arid zone rivers are not known in any detail. In Central Australia, major alluviation along the Finke River took place around 100–90 ka, with no evidence of later activity until the Holocene (Nanson, Chen and Price 1995). The Todd River was active 75–21 ka, but little is known about its flow regime during the last interglacial (Hollands et al. 2006). The Sandover River may have been a more energetic fluvial system during MIS5, but the limitations of thermoluminescence (TL) and U/Th dating of palaeochannels and alluvial terraces in this part of Central Australia leave us without a solid Quaternary chronology for this system (Tooth 1997). The balance of the evidence suggests that by 85 ka, even the largest of these inland river systems had switched to strongly seasonal flows and that by 60 ka (the close of the Katapiri), they were operating primarily as arid fluvial systems. In Central Australia and along Cooper Creek, short-lived high-magnitude flows occurred throughout the late Pleistocene (Bourke 1998; English et al. 2001; Nanson et al. 2008), but these probably reflect the interdecadal variability that characterises these arid rivers today.

Desert Dunes and Dust The history of desert dunes provides a counterpoint to the records of palaeohydrology from rivers and lakes. The widespread availability of luminescence dating techniques has allowed a chronology of dune activity to be reconstructed in increasing detail and at scales ranging from the vertical accretion and longitudinal growth of individual dunes through to the history of entire The Empty Desert: Inland Environments Prior to People 59

Figure 3.7. Kocurek’s model of aeolian system response to climate change. Sediment accumulated during earlier conditions is mobilised under an arid climate. Dune construction is initially limited by ground cover (availability-limited influx) and later by wind speed (transport-limited availability). Dunes are reworked or destroyed as the supply of sand is exhausted. (Simplified from Kocurek 1998:fig.4) dunefields. In Australian deserts, just as in deserts in Southern Africa, this has complicated the conventional view that dune systems have a straightforward correlation with periods of aridity (Lancaster 2008). Recent work suggests that interactions between sediment supply and availability and wind speed govern the response of dunes to climate change and define a series of states of dune systems (Kocurek 1998: fig. 4)(Figure 3.7). Kocurek’s model of aeolian response shows that sand starvation will cause an aeolian system to flip from dune construction to dune reworking. One implication is that some apparent gaps in dune chronologies may be processual, not simply climatic. Where sediment is freely available, winds play a major role in determining periods of dune activity (a transport-limited system). Where sediment is not freely available, variations in rainfall and vegetation cover play a dominant role in development of dunes (an availability-limited system). Veg- etation cover mostly limits dune activity in the Australian arid zone. However, Ash and Wasson (1983) have argued that wind speed determines dune mobility in the arid core, where the sparse vegetation cover is not sufficient to stop sand movement today. In contrast to linear desert dunes, source-bordering dunes are dependent on stream flow to deliver sediment and on seasonally dry conditions to allow it to be blown into transverse dunes. 60 The Archaeology of Australia’s Deserts

Dune chronologies across the Australian arid zone show that a widespread phase of construction of linear desert dunes began at 70–60 ka, at the close of MIS5. Around Lake Frome, OSL ages show phases of dune activity at 66–57 ka and 22–11 ka (Fitzsimmons, Bowler, Rhodes et al. 2007). In the Strzelecki and Tirari dunefields, the onset of major dune activity is dated to about 70 ka (Lomax et al. 2003; Fitzsimmons, Rhodes, Magee et al. 2007), with multiple periods of dune reactivation at 73–45 ka, 35–28 ka, 22–18 ka and 14–10 ka, punctuated by episodes of environmental stability. On the western margins of the Simpson Desert dunefield, phases of linear dune activity date to 75–65 ka and 28 ka (Hollands et al. 2006). At Lake Gregory, on the northern margin of the desert, linear desert dunes built over the old lake floor at 70 ka (Bowler et al. 2001). There is a broad consistency among these records, showing the onset of widespread dune activity after 70 ka, with a peak in activity during the last glacial maximum (30–18 ka). The Tasman Sea dust record also registers an increase in dust concentration from 74 ka, with distinct peaks at 61 ka and 52 ka (Hesse 1994). Both dune and dust records reflect increasing disturbance to vegetation and ground cover across the interior as the climate became increasingly arid and more seasonal and as the northern monsoon weakened.

Inland Vegetation during the Last Interglacial Direct data on inland vegetation during the last interglacial are scarce. The only long pollen sequence is from a marine core, Fr10/95 GC17, near Exmouth, in northwestern Australia. This provides a record of vegetation, fire and climate for the last 100 ka (van der Kaars and De Deckker 2002; van der Kaars, De Deckker and Gingele 2006). Although taken 60 km offshore, wind patterns mean that Fr10/95 GC17 provides a strategic record of vegetation across adjacent parts of northern Australia: there were grass-rich eucalypt woodlands in the northern part of the arid zone between 100 ka and 81.5 ka, with a shift after this time to open arid woodland with chenopods and the native dryland conifer Callitris. Palaeoclimatic reconstructions (van der Kaars, De Deckker and Gingele 2006) suggest that regional rainfall after 81.5 ka was about 440 mm per annum (p.a.), bringing the region within the arid-zone range (500 mm p.a. is taken as the northern bound of the modern arid zone). The record from Fr10/95 GC17 is corroborated by carbon isotope records from OSL-dated sediment cores at Lake Gregory (Pack et al. 2003), which also show a long-term trend towards a more arid climate. In Central Australia, the phytolith sequence from Puritjarra rockshelter (Bowdery 1998;Smith2009b) shows that local grasslands were well established during the last interglacial (MIS5), with grass values at about 96 ka approaching modern levels. The evidence suggests an interglacial landscape with open, The Empty Desert: Inland Environments Prior to People 61 grass-rich vegetation and at least seasonally active dunes, analogous to conditions around the rockshelter today. Attempts to retrieve a long, continuous pollen record from Lake Eyre have been frustrated by deflation of the lake floor and periodic dissolution of the salt crust. The LE82/2 core from Madigan Gulf is poorly dated and provides only a broken record of vegetation change, but it does show that Callitris woodlands were an important component of the regional vegetation before 30 ka (Luly 2001). Pollen has also been recovered from upper Katapiri sediments at Punkrakadarinna waterhole on Warburton River (Hesse, Luly and Magee 2005). This shows that the interglacial environment included significant areas of open vegetation made up of Poaceae and Asteracea (grass and daisies), similar to the modern vegetation of the area. Carbon isotopes in fossil emu eggshell (Dromaius novaehollandiae)provide an alternative record of palaeovegetation in the Lake Eyre region over the last 65 ka ( Johnson et al. 1999). These show that C4 grasses – associated with summer precipitation – were more abundant in the early part of this record, between 65,000 and 45,000 years ago, and then declined, which implies that the Australian summer monsoon was more effective then than now. The overall picture is of open arid or semi-arid vegetation across much of the interior during the last interglacial, with a shift towards more dryland taxa, including Callitris and chenopods, towards the end of MIS5.However,itis clear that even after MIS5, summer monsoon rainfall was more effective in the interior than today and did not collapse until 45 ka.

Last of the Dryland Megafauna By the last interglacial, signiﬁcant guilds of dryland megafauna were restricted to the southern part of the Lake Eyre basin, mainly in the Cooper Creek area. This regional fauna – associated with the upper Katapiri formation – collapsed around 65 ka and contracted to the margins of the desert, where some megafauna species persisted until 50–45 ka. A recent review of the continental distribution of these species shows that the late Quaternary megafauna had a predominantly southeastern distribution, with greatest species diversity in the woodlands, plains and forests of southeastern Australia (Webb 2008)(Figure 3.8). Except for the Cooper Creek–Lake Eyre area, the arid zone appears to have been largely empty of megafauna by 132–115 ka. Isolated ﬁnds of Diprotodon optatum across the western interior and along the west coast all appear to be much older than the last interglacial but are undated. In this context, the rich fauna found in upper Katapiri sediments is an extension of the southeastern megafauna, from the desert margins along major inland river systems into the Lake Eyre region. 62 The Archaeology of Australia’s Deserts

Figure 3.8. The distribution and diversity of late Quaternary megafauna in Australia. Numbers and size of squares show number of megafauna species present in each region, based on 32 key species. The arid zone is empty of megafauna except for an extension of southeastern Australian species along inland rivers into the Lake Eyre region. (Source:Webb2008 and 2009)

the katapiri fauna At the regional level, the Katapiri fauna provides a striking picture of a rich interglacial savanna environment, with a mosaic of habitats, including riparian forest and woodland, seasonally inundated floodplains, deep permanent waterholes and back-swamps, as well as open drylands with chenopods, grasslands and Callitris pine woodlands. JW Gregory may have been thinking of the Serengeti when he commented, ‘when I noticed the richness of the soil, I could not but think what a paradise this country would be if only it had an East African rainfall’ (1906: 112). The most comprehensive listing of fauna from upper Katapiri sediments is by Webb (2009; see also Tedford and Wells 1990)(Table 3.2). The major macropods present are the short-faced kangaroos, Procoptodon and Sthenurus. Other large herbivores include Diprotodon optatum and the giant wombat Phas- colonus gigas. Arboreal species, such as possums (Trichosurus vulpecula) and koalas (Phascolarctos), are also present, as well as smaller taxa characteristic of arid-zone faunas such as bandicoots (Isoodon obesulus, Macrotis lagotis), rat kangaroos (Bet- tongia lesueur) and hare wallabies (Lagorchestes leporides). Large flightless birds include the emu (Dromaius) and the much larger Genyornis, estimated to weigh as much as 275 kg. This fauna supported a range of carnivores and scavengers, including a large varanid lizard (Megalania prisca), the marsupial lion (Thylacoleo carnifex), thylacines and the freshwater crocodile Palimnarchus. The latter is represented in these fossil deposits by numerous cranial fragments, scutes, teeth and vertebrae. Crocodile teeth marks on Diprotodon bones record its effectiveness as a predator in these inland waterways. These channels and waterholes also supported a range of fish, freshwater mussels and chelid turtles. The large fish vertebrae in these assemblages indicate that some individual fish weighed as much as 36–50 kg (Webb 2009). The Empty Desert: Inland Environments Prior to People 63 table 3.2. List of taxa, Upper Katapiri fauna (MIS4–MIS5), Cooper Creek–Lake Eyre region

Group Megafauna >50 kg Other species <50 kg

Marsupial carnivores Thylacoleo carnifex Sarcophilus laniarius Thylacinus cynocephalus Bandicoots and bilbies Isoodon obesulus Macrotis lagotis Koalas Phascolarctos stirtoni Wombats Phascolonus gigas Lasiorhinus latifrons Possums Trichosurus vulpecula Rat-kangaroos Bettongia lesueur Short-faced kangaroos Procoptodon browneorum Procoptodon gilli Procoptodon goliah Procoptodon pusio Procoptodon rapha Simosthenurus orientalis Simosthenurus pales Simosthenurus occidentalis Sthenurus atlas Sthenurus andersoni Sthenurus stirlingi Sthenurus tindalei Other kangaroos and wallabies Protemnodon anak Macropus fuliginosus Protemnodon brehus Lagorchestes leporides Diprotodonts Diprotodon optatum Troposodon minor Zygomaturus trilobus Rodents Leporillus conditor Rattus tunneyi Large ground birds Genyornis newtoni Dromaius novaehollandiae Large varanid lizards Megalania prisca Crocodiles Pallimnarchus pollens Fish and turtles Maccullochella peelii peelii Macquaria ambigua Chelid turtles Crustaceans Cherax sp. Shellﬁsh Unio sp. Coxiella sp.

Tedford and Wells (1990) list a range of other species but do not distinguish between upper and lower Katapiri fauna. Source:Webb2008

There are few dryland faunas that allow comparison with the upper Katapiri fauna. The spectacular pitfall assemblages from the Nullarbor caves are middle Pleistocene and date from at least 780 ka through to 200 ka (Prideaux et al. 2007). These represent an older, arid-adapted fauna occupying a mosaic of 64 The Archaeology of Australia’s Deserts

table 3.3. Comparison of middle Pleistocene herbivore dietary guilds across Australia

Forested Arid Guild Leewin/Naturaliste Naracoorte Nullarbor Lake Eyre Basin

Arboreal 19 24 12 9 Fungivores 24 16 8 5 Browser 38 42 36 50 Mixed feeders 5 3 16 18 Grazers 14 16 28 18 Species N 21 38 25 22

Data are number of species. Source: Prideaux et al. (2007) – supplementary table 5

woodland and shrubland. Comparisons across a number of middle Pleistocene assemblages (Table 3.3) show that browsers dominate all of these, but the dryland faunas from the Nullarbor and Lake Eyre basin characteristically have more grazers and mixed feeders than do woodland and forest faunas, and fewer arboreal species. It seems likely that this would have also been the case for the late Pleistocene Katapiri fauna.

lake callabonna The fauna from Lake Callabonna is more limited than that in the Cooper basin (Table 3.4), presumably because the Callabonna area is a more marginal environment. The Callabonna fossils are OSL-dated to 75 ± 9 ka (Roberts et al. 2001). Gillespie et al. (2008) also report a minimum radiocarbon age of 53,400 BP for the gut contents of a Diprotodon (see next section). At the time of the

table 3.4. List of fossil fauna from Lake Callabonna

Species

Dromaius novaehollandiae Genyornis newtoni Diprotodon optatuma Phascolonus gigas Macropus cf. titan Macropus sp. Protemnodon cf. brehus Sthenurus andersoni Sthenurus stirlingi Sthenurus tindalei

a Following Price (2008), D. australis and D. minor are merged into D. optatum. Source: Wells and Tedford (1995) The Empty Desert: Inland Environments Prior to People 65

Callabonna diprotodons, the area was a shallow, seasonally dry saline lake, surrounded by saltbush and Cypress pine, with eucalypts along watercourses drain- ing into the lake. Trackways and footprints preserved as carbonate-cemented imprints on the bed of the lake show there was well-established traffic across the lake (Tedford 1984). The Lake Callabonna diprotodons died attempting to cross mud flats to nearby mound springs when the lake level was low. population ecology Information about the ecology of these animals has come from a range of sources. Distribution patterns within the Lake Eyre basin (Webb 2009)show that most species favoured the main river channels and floodplains (11–17 species), with the highest species diversity along Cooper Creek. Only seven species are regularly found in arid terrestrial habitats away from rivers; these include Diprotodon and Genyornis, which are known to have been wide-ranging species. The distribution of Procoptodon fossils suggests that this was another species able to forage in the arid backcountry. Analyses of carbon isotopes in tooth enamel for upper Katapiri fauna confirm that Diprotodon was a mixed feeder, consuming C3 and C4 plants (δ13C −18.8 ± 3.0)(Grocke¨ 1997). In contrast, Sthenurus was a browser, consuming adietofC3 shrubs (δ13C −24.3 ± 0.9) and was probably more restricted to woodlands and riparian habitats. Other isotopic work has shown that Pro- coptodon was a chenopod-browse specialist (Prideaux et al. 2009)andthat Genyornis was a mixed feeder that selectively ate C4 plants (Miller et al. 1999; Miller, Fogel, Magee et al. 2005). This must have included saltbush because this dietary signature is also observed in its eggshell in temperate areas well beyond the distribution of C4 grasses (Smith 2009a). One of the remarkable finds at Lake Callabonna was the recovery of gut contents of Diprotodon; this provided a direct record of the diet of this giant marsupial. Stirling records that ‘associated with the skeletons of Diprotodon,in a relative position which corresponded with that of the abdominal cavity, were occasionally found loosely aggregated globular masses of what were judged to be the leaves, stalks, and smaller twigs of some herbaceous or arboreal plants’ (1900: xii). At the time, these were tentatively identified as ‘Salsolaceae’ or ‘the allied orders Amarantaceae or Nyctagineae’, indicating that these are most likely a chenopod. This has been confirmed by recent analyses (using x-ray diffraction and stable isotope measurements) showing that this material is Atriplex (saltbush), a C4 halophyte shrub (Gillespie et al. 2008). collapse of the katapiri fauna The end of this distinctive dryland megafauna coincides with the collapse of the regional ecosystem, as rivers, lakes and swamps failed in the shift to the drier conditions of MIS4. Diprotodon remains from Punkrakadarinna on Warburton River are TL dated to 64.9 ± 5 ka (Webb 2009) and mark the last evidence 66 The Archaeology of Australia’s Deserts

of significant populations of megafauna in this region. A couple of factors are likely to underlie local extinction of this fauna. First, many macropods have a marked dependence on shelter to avoid thermal stress, so an increasingly open vegetation would have created problems for larger species. Even today, medium to large kangaroo species are rare in the heart of the Lake Eyre basin. Second, larger body size also brought a need for free water, as well as late maturity and slower reproductive rates. Most megafauna species, therefore, would not have tolerated the increasing environmental stochasticity and the fragmentation of riparian habitats (Main and Bakker 1981; Horton 1984). The disappearance of this dryland megafauna represents one of the clearest examples of climate-driven extinction at the regional level. Although other studies have shown the resilience of megafauna communities in the face of Quaternary climatic fluctuations (Prideaux et al. 2007), the collapse of the Katapiri fauna demonstrates the critical importance of biogeographic factors. The interglacial palaeohydrology of the Cooper Creek–Lake Eyre region effectively made this area part of the desert margin. The switch to drier and more variable conditions appears to have disrupted migration corridors between the main body of megafauna in southeastern Australia (a source) and animal populations in the Lake Eyre basin (a sink) (Webb 2009). Without these connections, the dryland megafauna could not recover from environmental crashes. Some species, however, survived into the period 50–45 ka. An isolated find of a Diprotodon cranium near Lake Eyre, associated with an OSL date of 46.6 ± 3 ka (ANU-OD 1251), suggests that these animals persisted in the better- watered upper reaches of these drainage systems, dispersing downstream after exceptionally heavy rain (Webb 2009 and personal communication). Similarly, at Lake Menindee, the regional megafauna persisted into the period 66–51 ka (Cupper and Duncan 2006). Genyornis newtoni, a large flightless dromornithid bird, is the only megafauna species likely to have been widely encountered by the first people to move into the desert.

genyornis Genyornis newtoni was widely distributed across arid and semi-arid parts of southeastern Australia until 50 ± 5 ka (Miller et al. 1999) (see Figure 3.2b). The presence of nasal salt glands indicates that this bird could tolerate saline waters (Murray and Vickers-Rich 2004). However, the distribution of its distinctive eggshell suggests that Genyornis was probably tethered to riparian corridors for nesting, perhaps moving up and down channel systems and ﬂoodplains as seasonal conditions allowed. Genyornis had a wide range in those parts of the arid interior that are interlaced with river channels – in western New South Wales, the Lake Eyre basin, the Murray–Darling basin, and in southwestern Queens- land. But bones of the bird are not present in the Nullarbor fossil assemblages and, despite several ﬁeld surveys, its eggshell has not been found in the Great The Empty Desert: Inland Environments Prior to People 67

Victoria Desert or around Lakes Gregory or Mackay. There appear to have been isolated populations of Genyornis at Lake MacLeod on the arid west coast and at Lake Lewis in Central Australia. The latter almost certainly represents a limited extension of range following the corridor of the Finke River during good years. In the Lake Eyre basin, the eventual extinction of Genyornis coincided with a signiﬁcant decline in summer monsoon rainfall ( Johnson et al. 1999; Murphy, Williamson and Bowman 2012), which potentially disrupted both its summer breeding pattern and its preferred diet of C4 plants. It may also coincide with the arrival of people in these landscapes, especially because as any predation on eggs and young birds would have put additional pressure on these birds.

OVERVIEW: THE DESERT PRIOR TO PEOPLE

Interglacial Landscapes What was the last interglacial like in the interior of Australia? A bird’s-eye view across these interglacial landscapes would reveal open, arid woodlands and grasslands across much of the inland. Looking across the interior, we would see a landscape with significant fluvial and lacustrine systems inset into arid and semi-arid landscapes on the desert margins and in the Lake Eyre basin. Outside of these areas, the interior was a landscape of saltlakes, stabilised desert dunes and xeric vegetation. There were active freshwater lakes on the desert margins – at Lake Gregory and Lake Woods in the north and the Willandra Lakes in the southeast. These are terminal lakes fed by stream flows originating outside the desert, and all are likely to have been rich in fish, shellfish or crustaceans. In the heart of the continent, Lake Eyre was a large, deep, saline lake during the interglacial, a greatly expanded lake that incorporated Lake Callabonna and Lake Frome. This was also a terminal lake, on an immense scale, fed by river systems rising in the subtropics. But the lake itself was too saline to support significant resources for people, except for ephemeral resource blooms, when floods flushed nutrients and fish into the lake from connecting river systems. It is the lower and middle reaches of the Cooper–Diamantina channels that represent the richest resource zone associated with the lake. Here, the rivers and waterholes supported a rich biota that included fish, crocodiles, turtles, freshwater mussels and waterfowl. A regional mosaic of swamps, floodplains and riparian woodlands supported a diverse array of large herbivores. This megafauna was tethered to the major channels and floodplains. In open desert away from the floodplains, these animals were rare: the greater part of the interior did not support comparable communities of megafauna. Continuing our bird’s-eye view would show that the Cooper–Diamantina area was not typical of the desert. The western half of the continental interior 68 The Archaeology of Australia’s Deserts

does not have the long coordinated river systems of the Lake Eyre basin. There were no active Quaternary lakes (saline or otherwise) in the western half of the desert: here, the old Miocene lake basins were now saltlakes. And in Central Australia, Lake Amadeus and Lake Lewis had become hypersaline playas by the late interglacial.

The Landscapes of Colonisation The first human movements into the interior took place 45 ± 5 ka (see Chap- ter 4). Most Quaternary records indicate that the interior was significantly more arid at this time than at any stage during the last interglacial (Hesse et al. 2004). People moving into the interior early in MIS3 would have encountered extensive deserts and drylands, with a xeric vegetation of grasslands, chenopod shrublands and open woodlands often dominated by the dryland conifer Callitris (van der Kaars and De Deckker 2002; Hesse et al. 2004;Smith2009b). The vegetation around Puritjarra, in western Central Australia, at 45 ka, was an arid, open shrubland or herbland, with isolated trees and little grass in a landscape that was already registering the impact of intensified aridity (Smith 2009b). Continental dunefields had been widely reactivated by 70–60 ka, and greater dust fluxes at this time register increasing land surface instability. Most large palaeolakes ended well before the first human movements into the desert. The last deepwater phase of Lake Eyre ended 65,000 to 60,000 years ago (Magee and Miller 1998), coinciding with declining fluvial activity in the associated river systems and with the collapse of the Katapiri fauna. The flightless Genyornis was the only megafauna species still widely extant across the interior when people entered the desert. The exceptions to this pattern were the freshwater lakes on the desert margins. In the southeastern section of the arid zone, rivers and lakes in the Darling basin and Willandra region were active during MIS3 and MIS2 (especially between 55 ka and 15 ka; Bowler 1998), feeding runoff from the southeastern Australian highlands into a chain of terminal lakes, the most famous of which is Lake Mungo. Elsewhere, MIS3 saw some reactivation of rivers and lakes in the Lake Eyre basin – especially at Lake Frome – but appears to have had little impact in Central Australia or the Western Desert. There were episodic strong flows along Cooper Creek and presumably also along other channels in the Georgina–Diamantina system (Nanson et al. 2008: 119). However, given that regional dunefields in the Tirari and Strzelecki Deserts were also active at this time (Fitzsimmons, Rhodes, Magee et al. 2007), the picture during MIS3 is of an arid environment with high submillennial variability and strong seasonality. This suggests that the channels feeding Lake Eyre (Warburton River and Cooper and Kallakoopah Creeks) functioned as classic arid rivers: mostly dry channels with saline groundwater outcropping in places but periodically reactivated during ‘boom’ years, when they funnel The Empty Desert: Inland Environments Prior to People 69 large flows of floodwaters into Lake Eyre. Lake Eyre filled again during MIS3, creating a saline water body no larger than that created by modern floods. There is little evidence to support the proposition that early occupation of Australia’s deserts selectively relied on lacustrine or riverine resources or that it took place during a ‘lacustral phase’ (Hiscock and Wallis 2005). Late Pleistocene human groups in the Darling and Willandra regions had access to active river and palaeolake systems. But further inland, along the channels and back-swamps of Cooper Creek and the Warburton River, riverine resources are likely to have been less predictable and increasingly seasonal. People moving further into the interior would have been faced with an open, arid landscape of dunes and saltlakes, one that lacked coordinated river systems. By the time people entered the interior, this was an arid ecosystem that had been shaped by tens of millennia of selective pressure. However, its ecological structure and controls may have been subtly different from those of Australia’s deserts today. It is significant that although the Australian monsoon weakened substantially after MIS5, it remained stronger than it is today until 45 ka. Potable water, in claypans, small soakages, waterholes and springs, is likely to have been more widely available due to lower evaporation and a more active summer monsoon. These small waters would have given people greater flexibility in their annual and seasonal subsistence movements than today, allowing more efficient access to the plant and animal resources of these drylands. Mobility is crucial for desert hunter-gatherers, and these small water points would have provided an effective way of ‘stepping through’ these unique landscapes. Moving into the ‘dead heart’ of the continent, the first people would have found the old floodplains and dry lake beds littered with the bones of megafauna – as can still be seen today in a few areas, such as Lake Callabonna. In the Lake Eyre region, Thirrari, Diyari and Wangkangurru people incorporated the fossil bones into their cosmology, associating them with Kardimarkara (the Rainbow Serpent), a creative being of immense size closely associated with rain and waterholes. In the 1970s, Mick McLean Irinjili described how the old people had found big bones at a waterhole and had covered them over, ‘out of respect and pity’. CHAPTER 4

FOUNDATIONS: MOVING INTO THE DESERTS

Excavations in 1986 at a remote site in western Central Australia opened up the late Pleistocene archaeology of the desert (Smith 1987). The site – known by its Pintupi/Luritja name, Puritjarra – was not the first late Pleistocene site to be discovered. There had been earlier finds in the Pilbara (Maynard 1980)and on the Nullarbor Plain (Marun 1972), but Puritjarra was the first that could not easily be explained away as ‘opportunistic exploitation of a temporarily ameliorating arid zone fringe’ (Bowdler 1990a: 339). As Rhys Jones put it, ‘some believed that the Australian desert was not occupied until the end of the Ice Age, about 10–12 kyr ago’, but the Puritjarra find ‘at a stroke, more than doubles the known antiquity of the region’ ( Jones 1987: 666). Ideas about the pattern and process of colonising the arid zone are inevitably entangled with evidence about its timing – as the Puritjarra finds showed. In this chapter, I take stock of the picture that has emerged from archaeological research over the past 30 years. I also review what we know about the begin- nings of Aboriginal society in Australia’s deserts in the period 45–30 ka. We now know that people moved into Australia’s deserts and drylands sometime before 45,000 years ago, stepping out across land that until then no human had ever visited. By 30,000 years ago, small groups of highly mobile hunter- gatherers were using pockets of country across the interior of the continent, from Central Australia to the Pilbara, and from Lake Mungo to the southern Kimberley.

THE CONTINENTAL SETTING Settlement of the desert was part of a larger process of settling the continent. People appear to have ﬁrst reached Australia, by sea, shortly before 50,000 years ago. Sea passages of a few days would have been enough to reach the Australian and New Guinea landmass from adjacent parts of the Indo- Malaysian archipelago, suggesting that the process is just as likely to have been

70 Foundations: Moving into the Deserts 71 a steady trickle of people across the frontier as a discrete migration. In this context, I see the establishment of a viable population on the shores of Sahul as principally a statistical problem: once human groups with watercraft are established on the shores of Sunda, even accidental drift voyages could lead to more than 100 people on Australian shores within a generation, and the experiment could be repeated again and again until the right combination of sex, age, locality and number of people allowed a self-sustaining population to take root. Organised voyaging and sophisticated boat-building skills would accelerate this process (O’Connell and Allen 2012). But the immensely long time over which this experiment in biological invasion could be repeated means that even low probabilities of success would see establishment of human groups in Sahul within a few thousand years of their arrival in the eastern Indonesian archipelago. As Webb points out, ‘a slow, almost constant, trickle of people’ may have been ‘all that was needed to begin Australia’s human story’ (2006: 111). In Australia, the earliest sites are Malakunanja II (52 ka), Nauwalabila 1 (>53.4 ka) and Nawarla Gabarnmang (45.2 ka) in Arnhem Land (Roberts, Jones and Smith 1990; Roberts et al. 1994; David et al. 2011). Devils Lair cave, at the southwestern extremity of the continent, was occupied by 48 ka, and there are scattered artefacts in levels dating to 50 ka (Turney, Bird, Fifield et al. 2001). This limestone cave has a deep (∼6 m), finely stratified sedimentary sequence, cross-dated using 14C, OSL, U series, AAR and ESR methods. The first evidence for occupation within the cave is a nested series of ashy hearths with radiocarbon ages from 41,460 ± 1,400 (ANUA-11709)to 45,470 ± 1,420 (ANUA-11502) (Turney, Bird, Fifield et al. 2001), equivalent to calibrated ages of 45–48 ka. People were also present at Lake Mungo and Lake Menindee, in the southeast of the continent, at 45 ka and probably by 50 ka ( Johnston, Clark and White 1998; Cupper and Duncan 2006). To the south, access to the temperate mountainous island of Tasmania was contin- gent upon sea level dropping sufficiently to expose the Bassian Rise, at about 43 ka. Occupation followed soon afterwards. In New Guinea, which formed part of the Pleistocene landmass, people were present on the Huon Peninsula sometime before 44 ka (Groube et al. 1986; O’Connell and Allen 2004)and were seasonally harvesting pandanus fruit and small game in the New Guinea Highlands at elevations of 2,000 m soon afterwards – perhaps even as early as 49 ka (Summerhayes et al. 2010). Although a critical review of these early sites argued that there was little firm evidence for settlement of Australia before 45 ka (O’Connell and Allen 2004), the evidence has become increasingly difficult to accommodate within such a narrow timeframe. The balance of evidence indicates that people were established in temperate and tropical regions surrounding the arid zone by 50–45 ka, early in marine isotope stage 3 (MIS3: 60–24 ka) (Figure 4.1). 72 The Archaeology of Australia’s Deserts

A MODICUM OF IDEAS Archaeological debate about the process of colonising the desert has hinged on issues of chronology and adaptation. There is a long history of thought framing deserts as extreme environments at the limits of human settlement, requiring special adaptations to live in successfully. Jones put it well:

Having mastered the capacity to cross water, the true barrier to the colonisation of the Australian continent lay not in the occupation of the tropical north but in the ability to establish and maintain human settlement in the arid country to the south. (1987: 666, my emphasis)

Some archaeologists have argued that we could expect a significant time lag (×104 ka) between early settlement of the continent and the first movements into the desert. These ideas usually imply a stepwise settlement of the region. Occupation may have been restricted initially to a few favourable habitats on the margins of the desert, such as the arid coast and its hinterland, or the river and lake systems in the Darling basin and Willandra (Bowdler 1977), or the semi-arid woodlands on the eastern and southeastern fringes of the desert (Horton 1981). Or, it may have involved ‘opportunistic exploitation of a temporarily ameliorating arid zone fringe’ (Bowdler 1990a: 339). It is only later that there was sustained use of the arid core. Another view is that early settlement may have been widespread across the interior but concentrated in focal areas, such as ranges, uplands and gorge systems, with little use of the major sandridge deserts until much later (Veth 1989b, 1993, 1995b;see Smith 1993, for a contrary view). Others argue that the interior was widely settled, but only under more favourable climatic conditions, before current levels of aridity were established (Ross, Donnelly and Wasson, 1992; Thorley 1998a; Hiscock and Wallis 2005). A final group of models postulate archaeologically instantaneous settlement of the entire continent, including the arid interior, within a few thousand years of landfall (Birdsell 1957; O’Connell and Allen 2012). In the most recent of these, O’Connell and Allen (2012) argue that ‘serial depletion of high-ranked prey’ – accentuated by use of hunting fires – would have led to rapid occupation of the continent. This would have created a wavefront of colonisation, fuelled by the theoretical capacity of hunter-gatherer populations to double in size every 25–75 years. Unlike Birdsell (1957), who argued that population growth would have filled the continent at ethnographic population densities within 2,200 years, O’Connell and Allen suggest that recurrent, high-amplitude environmental changes during the late Pleistocene kept populations low until the climate stabilised in the Holocene. Various limiting factors have been posited for arid-zone settlement: the distribution of potable surface water (Horton 1981), the timing of the switch from a subsistence economy based on fish and shellfish to one exploiting Foundations: Moving into the Deserts 73

Figure 4.1. Late Pleistocene Australia showing the location of arid-zone archaeological sites dating more than 30 ka. The modern boundary of the arid zone is shown for general reference, as well as the extent of continental duneﬁelds and the Pleistocene coast at -65 m (its probable position at 45 ka). Black triangles show desert archaeological sites discussed in the text. Open squares show early sites dated greater than 42 ka: (1)DevilsLair;(2) Silver Dollar site; (3) Northwest Cape sites ( Jansz, Mandu Mandu, C99 and Pilgonaman); (4) Noala; (5) Juukan 1;(6) Djadjiling and Newman P0255; (7) Serpents Glen; (8)CarpentersGap;(9) Riwi; (10) Lake Gregory (Parnkupirti); (11) Puritjarra rockshelter; (12) Kulpi Mara; (13) Allens Cave; (14) Malakunanja II; (15) Nauwalabila 1;(16) Nawarla Gabarnmang; (17) Colless Creek Cave; (18)GRE8; (19) Lake Yantara; (20) Lake Menindee; (21) Lake Tandou; (22) Lake Mungo and Willandra Lakes sites; (23) Dempseys Lake; (24) Cuddie Springs; (25) Huon Peninsula sites; (26) Ivane Valley sites. 74 The Archaeology of Australia’s Deserts

macropods and grass and acacia seeds (Bowdler 1977), or the need to develop new technologies and new modes of social organisation to cope with the vast arid regions of the interior (Veth 1993). Increasing aridity may also have driven local adaptation, gradually shaping the way of life of Aboriginal groups in the interior and on its margins, and equipping them to extend settlement into the arid core of the continent (Bowdler 1977; Gould 1977; O’Connell and Hawkes 1981: 115). Many of these hypotheses have been overtaken by new ﬁnds and shifts in chronology. But with recalibration for longer timeframes and better palaeoenvironmental data, they collectively provide a portfolio of ideas about the pattern and process of colonising the arid zone.

Invasion Biology There also exists a body of ecological theory about the likely pattern of dispersal into new environments (e.g., Shigesada and Kawasaki 1997; Clobert et al. 2001; Winterhalder 2001; Rockman and Steele 2003;Davis2009). Shigesada and Kawasaki (1997) draw a distinction among establishment, expansion and saturation phases, indicating that we can expect successful settlement of a new region (‘colonisation’) to be preceded by an exploratory phase (‘dispersal’). In archaeological terms, Beaton (1991) distinguishes between ‘transient explorers’ and ‘estate settlers’, while Gamble (1993) lists ‘pioneers’ and ‘settlers’. Rindos and Webb (1992) argue that it is advantageous for colonising groups to rapidly disperse across new territory, even if this does not involve optimal population density or adaptation to local conditions. Once a human population was established on the periphery of the arid zone, pressure to move into the region would depend on several factors. ‘Invasion pressure’ is a product of (a) the number of dispersal events, (b) the number of individuals in each dispersal and (c) increases in resource availability in the new habitat (see Davis 2009: 88–96). Dispersal events can be triggered by declining returns in high-ranked resources due to population increases or as a consequence of short-term climatic fluctuations (as also predicted by the ‘ideal free distribution’ in human behavioural ecology – Winterhalder 2001). It follows that invasion pressure would grow as population numbers on the periphery of the arid zone increased, even if only a small proportion of these populations were involved in moves into new country. O’Connell and Allen (2012) argue that the underlying driver for human foragers would have been economic: as local resources were depleted, foraging returns declined, and groups would move their residential bases to areas where the high-ranked game species, fruits and tubers were still available, creating a pattern of rapid serial occupation of new territory. Pressure would also grow whenever climatic variability created a flush of resources in the desert. In fact, spatiotemporal variability in the availability Foundations: Moving into the Deserts 75 of suitable habitat is a key determinant of dispersal rate. This has particular relevance in the Australian context, where interannual variability is extreme. During ‘boom’ years, the normal constraints on movement are relaxed and the biotic pulse generated would provide strong incentives for opportunistic movement. Temporary amelioration of arid conditions during boom years would have allowed people to move deep into the interior almost as soon as they reached its borders. The problem was not dispersal across the desert so much as maintaining a presence in the face of continuing climatic variability. Species capable of long-distance dispersal have a characteristic pattern of expansion in which migrants generate new satellite colonies rather than simply expanding the occupied area from its periphery (Shigesada and Kawasaki 1997). This process creates a metapopulation (Forman 1995: 373–86), a group of spatially separated populations occupying a nexus of favourable patches within the region. During the dispersal phase, demographic and environmental stochasticity repeatedly eliminates small populations in some patches, but these are recolonised by immigrants from other nodes. Studies of exotic animals give an indication of how rapidly dispersal into the desert may have taken place. Rabbits spread across Australian drylands at rates of 100 km/yr, compared to 10–15 km/yr in forest and coastal formations (Stodart and Parer 1988), completing their dispersal across the arid zone in less than a century. Birdsell (1957) made a similar point, arguing that the reduced carrying capacity of arid environments would necessitate larger territories and increase the rate of dispersal of people in the interior. During the dispersal phase, rabbits changed their social structure: dispersing rapidly as individuals rather than as ‘warren’ groups. If this was the case for human groups (as seems likely), then the rate of dispersal need not be constrained by the need to replicate group structure (the ‘budding-off’ point). The scale of the demographic challenge is worth briefly considering. Some ethnographic desert groups lived at population densities around 1 person/200 km2. If we take this as the minimum required for a viable hunter-gatherer society – allowing for intergenerational transfer of knowledge, corporate descent groups, replication of social institutions, and networks for finding marriage partners – then the colonisation of the arid interior need only have required 25,000 people. With fertility rates that allow population growth of up to 3 per cent per annum (Pennington 2001), a hunter-gatherer population could double every 25 years, so a donor population of 500 people may only need 150 years to populate the arid zone. All of this suggests that initial dispersal of human groups across the arid zone could have proceeded very rapidly, within a few centuries of the establishment of a human population on the periphery of the desert. Even allowing for multiple unsuccessful dispersals, we could expect to see people occupying a nexus of favourable patches across the arid interior within a millennium. Not 76 The Archaeology of Australia’s Deserts

all of these would necessarily have led to successful colonisation. The early prehistory of the desert is likely to have involved multiple short-lived occupations, with the early establishment and maintenance of human populations occurring only in focal areas.

GEOGRAPHIC BACKGROUND TO COLONISATION OF THE DESERT At the time people first moved into the desert, sea level was lower (−65 m for much of the critical time between 60 and 45 ka) and the continent was somewhat larger (∼11.1 million km2) than the three large islands (New Guinea, Australia and Tasmania – totalling 8.6 million km2) that make up the region today. Deserts and drylands probably made up about 5 million km2 of this Pleistocene landmass. The boundaries of the arid zone early in MIS3 are not known in any detail. Vegetation records show increasing aridity from 45 ka, increasing again after 35 ka, but major changes in the boundaries of the arid zone are not clearly evident until the last glacial maximum, when active dune- building extended to the southern and northern margins of the continent. A more modest expansion of the arid zone, relative to its modern boundaries, is likely to have been in place around 50–35 ka, given the evidence for arid spinifex grassland in the southern Kimberley at 45 ka (Wallis 2001) and for chenopod shrublands on the eastern edge of the semi-arid zone at 36–34 ka (Dodson and Wright 1989; Field, Dodson and Prosser 2002). Many areas presently on the margins of the desert would have been incorporated into the arid zone during this period. Most Quaternary records also indicate that the interior was significantly arid during MIS3 (60–24 ka) and MIS4 (70–60 ka) and probably more arid than at present (Hesse et al. 2004). People moving into the interior early in MIS3 encountered extensive deserts and drylands, with xeric vegetation analogous to modern formations (van der Kaars and De Deckker 2002; Hesse et al. 2004;Smith2009b). However, lower temperatures and reduced CO2 suggest that seasonal and evaporative stresses – as well as physiological stresses on plants – would have been different to those applying today. Potable water in the form of soakages, waterholes and springs may have been more widely available, giving some logistical flexibility. However, as Chapter 3 showed, by 45 ka, most of the large palaeolakes were either no longer active or were saline water bodies. Late Pleistocene human groups in the Darling and Willandra regions had access to active river and palaeolake systems. But further inland, along the channels and back-swamps of Cooper Creek and the Warburton River, riverine resources are likely to have been less predictable and increasingly seasonal. People moving further into the interior would have found an open, arid landscape, one whose ecological structure and controls may Foundations: Moving into the Deserts 77 have been subtly different from Australia’s deserts today but an arid ecosystem nonetheless.

Routes The geography of the continent suggests some potential routes for movement into the interior. Both Lake Gregory and Lake Woods represent terminal lakes of catchments that today feed monsoonal runoff from the northern savannas into the desert. To the east of these, the catchments of the Georgina, Diamantina and Darling Rivers begin in northern Australia and extend for more than a thousand kilometres into central and southeastern Australia. All of these are natural corridors for human dispersal into the interior. The flush of biological productivity that today follows floods along the inland river systems would provide strong incentive to groups to follow a flood peak downstream along these corridors. The effect of this would have been to channel human groups into several focal areas: the northern fringes of the arid zone, the Central Australian ranges, the lower and middle reaches of Cooper Creek, and the Murray–Darling basin. People moving into the western half of the continent faced a rather different set of circumstances because there are no clear corridors for movement deep into the interior. High interannual and interdecadal variability in rainfall is likely to have been crucial in determining initial access to the arid west of the continent. Pulses of extreme rainfall would have allowed human groups to disperse widely across the interior, coalescing in focal areas, such as the Pilbara, as the country dried out. Aridity on the west coast of the continent is zonal (controlled by large-scale subsidence of stable air in the anticyclonic belt) and is reinforced by a modest oceanic effect. So, there is little possibility that lower sea levels during MIS3 would have created a fertile coastal strip. Entry into the Pilbara would have required a jump across several hundred kilometres of arid terrain. Similarly, the temperate southwest of the continent could not have been reached without crossing a thousand kilometres of open desert. Here, the presence of people at Devils Lair by 48 ka represents an important benchmark, showing that people had successfully dispersed across the arid interior by this date. It seems unlikely that Devils Lair was reached by using watercraft along the west coast, bypassing the desert. South of Wallacea and New Guinea, there is little evidence that watercraft remained the primary means of long-distance dispersal: evidence for pelagic fishing is rare or absent in Australian sites before the terminal Pleistocene (O’Connor, Ono and Clarkson 2011); populations in New Guinea and Australia had little genetic contact after landfall (Friedlaender et al. 2007; Hudjashov et al. 2007), indicating that topographical barriers such as mountains and swamps had more impact than shared coastlines; and the time lag in 78 The Archaeology of Australia’s Deserts

colonising Tasmania indicates that use of watercraft was limited in southern waters.

EARLY SITES: CHRONOLOGY AND DISTRIBUTION Table 4.1 and Figure 4.1 show the distribution of archaeological sites, dating between 50 ka and 30 ka, in arid Australia. With nearly forty known sites in this time range, this is one of the best-documented records of early human occupation of a major world desert. The evidence suggests a rapid initial dispersal of human groups across the interior, reaching the northern and southeastern margins of the desert by 50–45 ka, and the southwestern extremity of the continent by 48 ka. The fact that the dates for Devils Lair are so much earlier than the oldest known sites in the Pilbara or Central Australia also suggests that much of the ‘pioneer’ dispersal phase in the interior remains to be found. At most sites listed in Table 4.1, the earliest visits initiated a pattern of repeated site use, which is more indicative of local ‘establishment’ of human groups than of intermittent dispersal events. The earliest sites on the desert margins represent use of focal areas in an arid landscape, including limestone gorges in the southern Kimberly and Carpen- taria areas, or palaeolakes in the Willandra and Darling regions. The pattern in the desert interior is different. Most sites there are a little younger (40–35 ka) and not as tightly grouped in focal areas. Although further ﬁeldwork and dating may well change this picture, the chronology currently suggests a staged pattern of colonisation, with early use of focal areas on the desert margins 50–45 ka, followed by occupation of arid coasts and arid uplands 40 ka, with only later moves into the sandy deserts at around 30 ka.

Northern Desert Fringe The limestone gorge systems of the southern Kimberley – today only 100 km from the edge of the Great Sandy Desert – would have been an obvious focal area for early settlement. Even with reduced precipitation and surrounded by arid grasslands, these limestone ranges had substantial groundwater reserves sufﬁcient to maintain local wetlands, seepages and waterholes (Wallis 2001). The key sites are Riwi and Carpenters Gap 1. Riwi is a shallow limestone cave with a Pleistocene unit of brick-red sediment, a deeply oxidised quartzose silt, interlaced with intact hearths and lenses of charcoal dating to 45–34 ka. The presence of intact hearths, numerous stone artefacts, ochre and bone indicates that this represents a substantial occupation deposit, but it is yet to be investigated in detail (Balme 2000). Carpenters Gap 1 is a large limestone rockshelter in the Napier Range near Windjana Gorge (O’Connor 1995; McConnell and O’Connor 1997; Foundations: Moving into the Deserts 79 table 4.1. Archaeological sites in Australian deserts and drylands dating 30–50 ka (chronology and distribution)

Occupation Site or locale Region Setting unit Dating method Age ka Sources

N desert margins Carpenters Gap 1 Kimberley Rockshelter Phase 1 (Spits ABOX AMS 44–48 18, 19, 39–51) 14C, OSL 23, 34 Riwi Kimberley Cave Spits 5–16 14C, ABOX 34–45 2 AMS 14C GRE8 Gulf Plains Rockshelter Phase A 14C, AMS 14C 31–39 27 Colless Creek Gulf Plains Cave Unit B Geomorphic 30 35 estimate SE desert margins Willandra Lakes Murray-Darling Lunettes Lower Mungo 14C, AMS 14C, 40–50 16 basin sands and TL, OSL, Mungo ESR, AAR, soil (or U series equivalent) Lake Menindee Murray-Darling Lunette – AMS 14C, OSL 45 9 basin Cuddie Springs Murray-Darling Ephemeral Stratum 6a– 14C, ABOX 35–39 12, 13, basin lake AL2 AMS 14C, 14, 15 OSL, ESR Lake Tandou Murray-Darling Lunette Base of 14C 31–41 1, 3 basin Bootingee Dempseys Lake Flinders Ranges Ephemeral Hearth PACD 14C, OSL 44 36 lake H1 Arid interior Allens Cave Nullarbor Doline Basal 14C, AMS 14C, 40 7, 26 occupation OSL CB10-93 Pilbara (inland) Rockshelter Spit 5 (mid 14C >41 37 occupation) Djadjiling Pilbara (inland) Rockshelter Basal 14C 30–41 17 occupation Juukan 1 Pilbara (inland) Rockshelter Unit C (spits 14C 37 28 12–14) Puritjarra Central Rockshelter Unit 2c 14C, ABOX 35 29 Australia AMS 14C, TL, OSL Kulpi Mara Central Rockshelter Layer 3 14C, AMS 14C 29–34 30,31 Australia Lake Yantara Simpson Ephemeral – 14C 31 10 Strzelecki lake Newman P2055 Pilbara (inland) Rockshelter Spits 16–19 14C 31 6, 8 Lake Gregory Great Sandy Lake Unit C OSL >37 33 Desert shoreline (45–50?) Serpents Glen Western Desert Rockshelter Stratum 8 AMS 14C 28 24 (continued) 80 The Archaeology of Australia’s Deserts table 4.1 (continued)

Occupation Site or locale Region Setting unit Dating method Age ka Sources

Arid west coast Jansz North West Rockshelter Layer 4 14C, AMS 14C 36–40 25 Cape Mandu Mandu North West Rockshelter Unit 2 14C 24–39 20, 21, Cape 22 C99 North West Rockshelter Layer 4 14C, AMS 14C 25–38 25 Cape Pilgonaman North West Rockshelter Lower red unit 14C, AMS 14C 36 21 Cape Silver Dollar Shark Bay Sandridge Pleistocene unit 14C 22–35 4, 5, 11 Noala Montebello Rockshelter Layer 3 14C 31 32 Islands Sources: 1,Balme1995; 2,Balme2000; 3, Balme and Hope 1990; 4,Bowdler1990b; 5,Bowdler1999; 6,Brown1987; 7, Cane 1995; 8, Comtesse 2003; 9, Cupper and Duncan 2006; 10, Dury and Langford-Smith 1970; 11, Dyason 2007; 12, Field and Dodson 1999; 13, Fillios, Field and Charles 2010; 14, Fullagar and Field 1997; 15,Furby1995; 16, Johnston, Clark and White 1998; 17, Law, Cropper and Petchey 2010; 18, McConnell 1998; 19, McConnell and O’Connor 1997; 20,Morse1988; 21, Morse 1993a; 22, Morse 1993b; 23, O’Connor 1995; 24, O’Connor, Veth and Campbell 1998; 25, Przywolnik 2002; 26, Roberts et al. 1996; 27,Slack2007; 28, Slack, Fillios and Fullagar 2009; 29,Smith2006; 30, Thorley 1998a; 31, Thorley 1998b; 32,Vethetal.2007; 33,Vethetal.2009; 34, Wallis 2001; 35, Magee and Hughes 1982; 36,Walshe2012; 37, Hook, Sinclair and Wright 2012.

McConnell 1998; Wallis 2001). Like Riwi, the high alkalinity of the deposits there has led to exceptional preservation of organic materials throughout the sedimentary sequence. The basal occupation (Phase 1), dating to 48–44 ka, is represented by a small peak in occupation debris, including wood, seeds and grass stems, animal bone, ochre and stone artefacts. Further east, a pilot excavation at GRE8, a small rockshelter in Gregory River gorge, shows that people were also using the limestone gorge systems in the Carpentaria region by 37 ka (Slack 2007). The first phase of occupation at GRE8 is represented by eighteen small flakes of local chert, some pieces of local ochre and a large quartzite core. These need not represent much more than fleeting visits by people making greater use of nearby riverbank or floodplain locations. The presence of discrete lenses of Alathyria mussel shell suggests episodic deposition of occupation debris within a slowly aggrading silt and carbonate matrix.

The Willandra Lakes and Lower Darling River The Willandra Lakes, including Lake Mungo, is one of the most famous archaeological regions in Australia (Bowler et al. 1970; Barbetti and Allen 1972; Allen 1990; Johnston et al. 1998). Today, the lakes are dry, but between 50 ka and 15 ka, they formed a chain of active glacial-age lakes, fed by higher Foundations: Moving into the Deserts 81

Figure 4.2. The Willandra Lakes showing the distribution of late Pleistocene archaeological exposures. Inset shows the relationship of these lakes to the southeastern highlands. (After Johnston, Clark and White 1998: 114) discharges from the southeastern Australian highlands (Bowler 1971, 1998) (Figure 4.2). A series of archaeological sites provide evidence for occupation before 32 ka. These include the ochre-covered burial of a man, a cremation burial of a woman, hearths and ﬁreplaces (N = 8) and small shell middens 82 The Archaeology of Australia’s Deserts

Figure 4.3. The excavation at Mungo B in 1976, showing the Mungo palaeosol (upper dark band). (Photograph courtesy of FW Shawcross)

(N = 8–14). Altogether, about twenty sites are known from the period 40– 32 ka. Most sites represent short-lived lakeshore camps. The archaeological remains range from thin bands of shell to complexes of thin shell middens, hearths and scatters of fish otoliths, crustacean gastroliths and the bones of small land mammals, lizards and birds. The small middens are made up of freshwater mussels (Velesunio) and tend to be small scatters of shell (about 3.5 m2) no more than 2–3 shells thick. At Lake Mungo, two large excavation trenches have probed the base of the lunette sequence – Mungo A (22 m2) and Mungo B (48 m2)–down to 5 m. The Mungo B excavation (Figure 4.3) recovered 775 stone artefacts, mainly large silcrete flakes and cores, concentrated in levels dating to 43–45 ka (Shawcross and Kaye 1980). The lowest artefacts (consisting of eleven silcrete flakes from blocky silcrete platform cores) are bracketed by OSL ages 50.1 ± 2.4 and 45.7 ± 2.3 ka (Bowler et al. 2003); they show that people began using these palaeolakes around 50 ka, almost as soon as they were formed by the return of freshwater to the system (Bowler 1998). This pattern of finds is repeated around palaeolakes associated with the Darling River to the northwest, including Lake Tandou and Lake Menindee (Hope, Dare-Edwards and McIntyre 1983; Balme and Hope 1990; Cupper and Duncan 2006). At Lake Menindee, the earliest trace of people is a fireplace, dated by OSL and 14C methods to 45.1 ka, associated with burnt emu eggshell and a burnt piece of megafauna bone (a macropod tibia). This is also the case at Dempseys Lake, on the western margin of the Flinders Ranges, where the Foundations: Moving into the Deserts 83 earliest evidence is a hearth (PACD H1), dated to 44.4 ka, possibly associated with Genyornis eggshell and two silcrete flakes (Walshe 2012).

Cuddie Springs Cuddie Springs is an ephemeral lake, today on the margin of the semi-arid zone (Field and Dodson 1999; Field et al. 2002;Furby1995). A large-area excavation (15×15 m2) over several seasons has produced the largest archaeological assemblage available for arid Australia older than 30 ka (Field, Fullagar and Lord 2001). Most attention, however, has centred on the presence at this site of anomalously late megafauna and anomalously early seed-grinding implements (Gillespie and Brook 2006). The late Pleistocene unit is a layer of lacustrine muds, 70 cm thick and capped by a stony deflation pavement. This is overlain by 1 m of Holocene lake clays that have been extensively trampled by cattle using a well on the lake floor. The excavators recognise two distinct phases of use within the late Pleistocene unit. The upper part (layer 6a, archaeological levels 2 and 3) represents an open campsite on the bed of an ephemeral lake, concentrated around one of the few reliable wells in the region. A series of radiocarbon samples shows that this dates to around 35–39 ka (Beta-81378 32,900 ± 510 BP, Beta-81379 33,660 ± 530 BP, Beta-44375 30,280 ± 450 BP). No hearths or other features are preserved, but the archaeological assemblage includes a wide range of material: large pieces of charcoal, ochre, silcrete and quartzite flakes; cores and retouched implements; bones with cut marks; several grindstones (including a millstone) and broken and highly fragmented bones of both extinct and extant species. The lower part (layer 6b, archaeological level 1) represents a bone bed formed in a shallow marshy lake, with intact bones of extinct species, including Diprotodon, Genyornis, Sthenurus and Macropus titan, in a matrix of cracking clays (Figure 4.4). ESR and U-series dates on the bones and teeth of megafauna species show that the bone bed dates to around 50 ka (Grun¨ et al. 2010). This is a classic waterhole death assemblage and the excavators discount any significant human involvement in its formation (Fillios, Field and Charles 2010). Although stone artefacts are intermixed within the bone bed down to about 55 cm, the excavators argue that these represent independent use of the lake margins by people, with both people and megafauna drawn to a key water source. At a site like Cuddie Springs, any model of site formation has to assess the extent of mixing of assemblages due to swelling and contraction of the lake muds, periodic inundation of the lake floor and the effects of trampling by animals around a focal water point. Added to this is the likelihood of disturbance by people digging wells. Although the excavators forcefully defend 84 The Archaeology of Australia’s Deserts

their interpretations, their data indicate a series of palimpsests, with archaeological occupation (around 35–39 ka) overprinting an older bone bed (around 50 ka), and with Holocene artefacts intruding into the upper part of the late Pleistocene sequence. Within the bone bed (6b), the artefacts appear to represent intermixing of younger material from the dense occupation layer immediately above it. The number of artefacts declines sharply in 6b(downto176 artefacts/m3 compared with 1,878/m3 in the overlying occupation layer – Furby 1995: table 3.3), and there is some size sorting with depth (mean weight of cores declines from 78–96 gto23–29 g). There is also evidence for some mixing of sediments (OSL samples show multi-age grain populations – Roberts et al. 2001; Fillios et al. 2010), as well as displacement of charcoal (there is much less charcoal in 6b, and the radiocarbon dates are indistinguishable from those in 6a). Much of the fossil bone in the main occupation horizon (layer 6a – archaeological unit 2) may have been brought up by people digging pits for wells and ovens – especially as the density of occupation debris and artefacts suggests comparatively intensive use of the site at this time. All the megafauna bone in this level is highly fragmented, and some pieces are identified by the excavators as having been reworked from much lower levels (e.g., Palorch- estes and Pallimnarchus). Because cut marks are only found on the bones of extant species, these provide no evidence of primary association between megafauna and people. Trueman et al. (2005) used rare earth element (REE) signatures, with some success, to test whether the megafauna bone is contemporaneous with the other remains, but there are doubts about whether the bone is a closed system with respect to groundwater chemistry (Grun¨ et al. 2010). The pavement (layer 5) that caps the late Pleistocene layers was formed by aeolian deflation of about 1–1.5 m of overlying sediments. It is almost entirely a lag of stone artefacts (800–1,000 artefacts/m2), formed over a period of perhaps 10,000 years. Although the excavators claim the assemblage dates from the last glacial maximum (LGM), the age of this pavement is effectively unknown. Because any autochthonous charcoal will have been removed with the fines, the reported radiocarbon dates give only the age of the surface to which deflation has cut down (33 ka). Extensive disturbance to the muds that overlie the pavement means that OSL and 14C dates for these levels do not reliably bracket the age of layer 5. Wind abrasion and varnishing on the surface of the pavement show that it was a land surface for a considerable period of time. There must be some doubt, therefore, about the age of the artefacts in layer 5. The movement of modern cattle and horse skeletons down through the overlying clays to rest directly on the pavement also raises the question of whether it continued to incorporate younger artefacts after it formed – as Foundations: Moving into the Deserts 85

Figure 4.4. Cuddie Springs. Plan of the megafauna bone bed for squares E10–E12 in layer 6b. Skeletal elements with cross-hatching are limb bones of Genyornis.Gridsquares are 1 m2. (After Fillios, Field and Charles 2010:ﬁg.9a)

does the presence within it of late Holocene artefact types such as thumbnail scrapers (Field and Dodson 1999: 290) and possibly also tula adzes (Habgood and Franklin 2008: 206). Finally, although the deflation pavement (layer 5) now seals the late Pleis- tocene units from disturbance, this was not the case in the past. The 1997 excavation uncovered a series of cut-and-fill features sloping down towards the old well depression. These are truncated by the deflation pavement and so predate its formation. They represent erosion gullies (or perhaps pits or burrows) cut from a surface now removed by deflation. Excavation of one feature showed they sometimes contain artefacts, including at least one grindstone, but the fills were not routinely isolated during excavation. In this context, claims for late Pleistocene seed-grinding implements at Cuddie Springs (Fullagar and Field 1997) are tenuous: of twenty-one published late Pleistocene grindstones, fifteen are from the deflation pavement. Of the six grindstones recovered below this, only one is unequivocally a seed-grinding 86 The Archaeology of Australia’s Deserts

implement (CS6034), and its position with respect to the cut-and-ﬁll features is not reported. The main importance of Cuddie Springs is neither its late megafauna nor early seed-grinding implements – both of which are doubtful – but rather the rich record of early occupation that it provides for the upper reaches of the Darling basin.

The Arid West Coast Northwest Cape is one of the few points on the west coast of the continent where the coastal shelf is so narrow that the modern coast is within 5–10 km of the late Pleistocene coastline. Several phases of archaeological research here (Morse 1988, 1993a, 1993b, 1993c; Przywolnik 2002, 2005)showa remarkably consistent picture of occupation as early as 35–40 ka by people using the resources of the late Pleistocene littoral and making transitory use of rockshelters within a day’s walk inland of the coast. The limestone systems of the Cape Range are known to have significant reservoirs of subterranean water, and it is likely that these waters were critical in sustaining occupation along an arid coast. Most sites are small rockshelters eroded beneath a cap of coastal limestone. Mandu Mandu typifies the pattern. Structurally, it is a small cave, with a floor area of 80 m2. A series of excavation pits (3 m2 in total) revealed a deposit just over 90 cm deep with two stratigraphic units. The lower unit consists of deeply oxidised red carbonate-rich Pleistocene sands with a basal age of 34,200 ± 1,050 BP (Wk1513)(∼38.8 ka). This is overlain by greyish late Holocene sands, containing most of the occupation material. Between the two units, there is a stratigraphic (and possibly occupational) hiatus from 24 ka to 6 ka. When first occupied, Mandu Mandu would have overlooked a gently sloping coastal plain. The archaeological evidence suggests that people moved between the coast and the arid hinterland, seasonally exploiting macropods and emu eggs in the hinterland, but with only transitory use of sites such as Mandu Mandu until the late Holocene. Use of the Cape Range sites declined after 25 ka (as sea level fell) and does not appear to have resumed until the early to mid-Holocene, when the last marine transgression again brought the coast within the catchment of these sites. At Jansz rockshelter, a basal age of 35,230 ± 450 BP (Wk8919)(40.3 ka) on baler shell represents the earliest evidence for people on the northwest coast of Australia (Przywolnik 2002), comparable to the dates from Djadjiling and site CB10-93 in the inland Pilbara. Jansz appears to have been abandoned shortly after initial use (36–40 ka) and was not reoccupied until 10,730 ± 60 BP (Wk8920). C99 wasfirstusedat33,930 ± 320 BP (Wk8925)(38 ka) – when it would have been less than 2 km from the coast – and then abandoned between 21,110 ± 100 BP (Wk8933)and8,010 ± 280 BP (Wk8297)(Przywolnik2002). From Pilgonaman, there is a date of 31,770 ± 390 BP (R16098/1)(36 ka) on Foundations: Moving into the Deserts 87 a baler shell artefact, showing early use of this site, but problems with date inversions complicate any interpretation of site history (Morse 1993a). The Northwest Cape finds are corroborated by excavations on the Mon- tebello Islands, now 120 km off the northwest coast of Australia. There was intermittent use of Noala Cave at 27,220 ± 640 BP (32 ka) (Veth et al. 2007). At this time, the Montebello Islands were part of the mainland, about 8 km from the Pleistocene coast. The shell and stone artefacts show that people were moving between the Pilbara and the Pleistocene coast across a broad coastal plain. Some 400 km south of Northwest Cape, further evidence for use of the coastal hinterland comes from the Silver Dollar site (Bowdler 1990b, 1999). Presently on Peron Peninsula in Shark Bay, this is an open site capped by a Holocene shell midden. When first occupied, it would have been a campsite on a sandy ridge about 50 km from the Pleistocene coast. Excavations show that below the shell midden there is a late Pleistocene unit (70 cm thick) of unconsolidated sands stabilised by carbonate accretions with a variety of occupation debris, including stone artefacts, Melo shell, macropod teeth and abundant emu eggshell, spanning the period from 35 ka to 22 ka. With the Pleistocene coast beyond the range of daily foraging activities, the Silver Dollar site provides a record of people living on typically desert resources: hunting hare wallabies and gathering emu eggs.

Pilbara The Pilbara saw an early discovery of Pleistocene sites, with excavation of theNewmansite(P0187)in1976 (Maynard 1980). Since then, another nine sites with occupation dating before 15 ka have been located. Two recently discovered sites, Djadjiling and Juukan 1, confirm occupation in the time period 40–30 ka – and excavations at a third, CB10-93 in the Chichester Range, show that use of this site began before 41.2 ka. Djadjiling (HD07–1A-04) has a deep (2.2 m) deposit, representing decay and weathering of local iron-rich bedrock with variable inputs of charcoal and ash, stone artefacts and traces of ashy hearths (Law, Cropper and Petchey 2010). Four radiocarbon dates provide a mutually consistent series to 35,159 ± 537 BP (Wk22787), with stone artefacts present below this point. Like most Pilbara rockshelters, use of Djadjiling appears to have been fleeting. Visits to the shelter began sometime before 40 ka and continued until about 29 ka (24,522 ± 129 BP, Wk22786), where a well-preserved hearth marks a point after which visits to the shelter declined. Juukan 1 shows a similar pattern of fleeting early use, beginning around 32,920 ± 70 BP (Beta 249759) (Slack, Fillios and Fullagar 2009). A serious gap in research is the lack of investigation of the Millstream- Fortescue wetlands, which may well have been the focus of late Pleistocene 88 The Archaeology of Australia’s Deserts

settlement in the Pilbara. Recent ﬁnds at CB10-93 go some way towards addressing this. This site is a small rockshelter in the Chichester Range, within a day’s walk of the wetlands. The lowest radiocarbon sample (Wk33656 36,039 ± 272 BP) is from the mid-point of the occupation unit (Hook, Sinclair and Wright 2012). Because small numbers of stone artefacts are found beneath this for another 50–60 cm, initial use of the shelter may have begun by 45 ka or perhaps earlier. Of the other inland Pilbara sites, at least one may have basal occupation deposits of a comparable age. At Newman rockshelter (P2055), the lower levels of a small excavation produced evidence of occupation dated to 26,300 ± 500 BP (SUA1510)(Brown1987), but the full depth of the deposit was not probed and older cultural remains are probable. All of these sites are located near the upper reaches of the Fortescue River, an intermittent river that drains into the Indian Ocean. It is possible that late Pleistocene groups ranged from the coast at Northwest Cape 500 km inland along the Fortescue River, although there is at present no direct evidence for this.

Nullarbor Plain The Nullarbor Plain is another region where recent work has deepened earlier records of late Pleistocene occupation. This vast karst plain, now covered with chenopod steppe, attracted early archaeological interest because of its caves and sinkholes (Gallus 1968;Wright1971;Marun1972;Martin1973; Milham and Thompson 1976). At Koonalda Cave, a radiocarbon age of 31,000 ± 1,650 BP (V-82) for the lower hearth complex appears to be anomalous: other dates for the same hearth and for the stratigraphic column show that these levels date to no more than 28–26 ka (Wright 1971). However, further excavation and luminescence dating at Allens Cave (N145) shows that occupation at this site extends back to 40 ka (Cane 1995; Roberts et al. 1996; Turney, Bird and Roberts 2001). Allens Cave is a rockshelter formed beneath the edge of a collapsed doline. About 5.5 m of horizontally bedded sand and silt, interbedded with finely stratified hearths and lenses of gravel and limestone debris, have accumulated in the doline, making this the deepest and most finely stratified sedimentary sequence at any desert site. Sediments began accumulating in the doline around 65 ka (Roberts et al. 1996). The earliest traces of human visits are represented by a small ashy hearth (40 × 25 cm, 2 cm thick), a small flake of Nullarbor flint and three artefacts of flaked limestone, dated by OSL to 39,850 ± 3,100 years ago (OXODAC390). A pattern of intermittent use of the doline continued throughout the late Pleistocene. When Allens Cave was first used, it was approximately 100 km from the Pleistocene coastline and surrounded by open chenopod steppe (Martin 1973). Foundations: Moving into the Deserts 89

On this treeless plain, aridity is accentuated by the karst landscape. Permanent occupation would not be expected. The early use of Allens Cave probably registers seasonal hunting parties exploiting the abundant macropod fauna of these chenopod shrublands. The focal area for hunter-gatherer groups, however, is likely to have been the strip of arid woodland on the coastal plain to the south, where water was available in wells sunk in coastal sands, or in seepages against the base of the Wilson Bluff escarpment. A break in the 50 mhigh escarpment at Merdayerrah sandpatch, near Allens Cave, makes this a strategic point on any journey between the coastal woodland and the treeless plain.

Central Australia In Central Australia, the crucial evidence comes from Puritjarra rockshelter (Smith 1987; Smith, Prescott and Head 1997), where radiocarbon dates back to 31,140 ± 470 BP (ANUA10013)(35 ka) show that late Pleistocene occupation was not limited to the desert periphery but rather extended into the centre of the continent. Complementary studies of stone artefact assemblages; 14C, TL and OSL chronology; environmental evidence (charcoals and phytoliths); sedimentary history; grindstones and plant use; rock art; ochres and Aboriginal history have made this one of the best documented archaeological sites in the Australian desert. The western part of Central Australia contains a diversity of country, including sandplain, dunefields, stony desert, saltlakes and rock outcrops. Scattered across the dominant sandhill and spinifex country is an archipelago of small rocky ranges, often only a few kilometres long, and visible one from another on the horizon. Puritjarra is situated in the Cleland Hills, one of these small range systems, 60 km west of the MacDonnell Ranges (Figure 4.5). The area is a transitional zone between ranges and desert lowlands: biogeographically, it is part of the Great Sandy Desert, but it is within a few days’ walk of the main central Australian range system. The rockshelter is situated at the foot of a small sandstone escarpment near Murantji rockhole, a large semipermanent (possibly spring-fed) water source 2 km north of the rockshelter. Three seasons of archaeological excavations at the site between 1986 and 1990 opened up a series of trenches (14 m2 in total) to explore the spatial variability of occupation across the rockshelter (Figure 4.6). Twenty-six radiocarbon dates are available for the Puritjarra sequence, and they show three main phases of occupation. There are hints of fleeting, early visits to the rockshelter approximately 45 ka. These traces include several flakes, a core and a piece of red ochre, but the status of this material is uncertain (Smith 2006). Systematic but intermittent use of the rockshelter appears to have begun about 35 ka. This pattern of use continued until 8.3 ka, when use of the shelter became more frequent. The third phase of use coincides with the last millennium and is marked by much more intensive occupation of the site. 90 The Archaeology of Australia’s Deserts

Figure 4.5. Puritjarra rockshelter, a late Pleistocene site in the Cleland Hills. The latter are typical of many small sandstone range systems set in spinifex and sandhill country in western Central Australia. (Photograph: MA Smith)

Analysis of the Puritjarra assemblages (Smith 2006) supports the notion that late Pleistocene groups in Central Australia were small and highly mobile and ranged over large territories. A pattern of repeated site use was established about 35 ka, when growing aridity in the region (Smith 2009b) may have led to a more water-tethered pattern of land use. The first indications of these visits are small discrete scatters of silcrete and chalcedony flakes, charcoal and pellets of red ochre in the central part of the shelter floor. The excavations suggest a pattern of small shifting scatters of occupation debris at the northern end of the rockshelter – an area that constituted the major rock-free floor at this time. Many artefacts appear to have been left where they were produced or used. The paucity of broken tools or cores and the absence of evidence for clearing debris from occupation surfaces suggest that use of the rockshelter was transitory. The high proportion of artefacts made of exotic silcrete and chalcedony and the relatively low use of local sandstone also indicate greater residential mobility than at any other time in the history of the rockshelter. People visiting the rockshelter around 35 ka were moving over a wide territory (across an estimated 10,000 km2) west of the main MacDonnell Ranges: they acquired red ochre from the Karrku mine, 125 km northwest of Puritjarra (Smith, Fankhauser and Jercher 1998) and brought in flakes of fine-grained silcrete, probably from the western end of the main range system, as well as pieces of white chalcedony, probably from a small quarry called Puli Tjulkura, 60 km to the northeast. Foundations: Moving into the Deserts 91

Figure 4.6. Excavations at Puritjarra rockshelter, 1988. (Photograph: MA Smith)

A complementary record of late Pleistocene occupation has come from excavations at Kulpi Mara in the main ranges (Thorley 1998a, 1998b; Thorley, Faulkner and Smith 2011). Kulpi Mara, a long narrow rockshelter on a broad sandstone terrace perched above the surrounding country, has a strategic position in the catchment of the Palmer River, one of the main river systems of Central Australia. This excavation, therefore, provides a small window on what is undoubtedly a key focal area for hunter-gatherer occupation in Central Aus- tralia. The excavations (3 m2) showed that the first phase of use of the shelter began around 34 ka and continued until about 29 ka (bracketed by radiocarbon ages of 29,510 ± 230 BP Beta-98034 and 24,250 ± 620 Wk4583). The Pleis- tocene assemblage is made up of small well-made flakes of silcrete or quartzite, several retouched implements and small pieces of red ochre. Although none of the stone used for artefacts indicates movement over significant distances, it remains possible that Kulpi Mara and Puritjarra represent parts of one system of land use, with seasonal movement between the main ranges and the more marginal spinifex and sandhill country to the west.

Western Desert The longest record of occupation in the Western Desert is from Serpents Glen, a large sandstone rockshelter near the mouth of a valley in the Carnarvon Range, on the edge of the Little Sandy Desert (O’Connor, Veth and Campbell 1998). The setting of the site, near rockholes and soakages, seems to have attracted repeated visits from 30–28 ka but only light use of the shelter itself. 92 The Archaeology of Australia’s Deserts

A pilot excavation showed a deep (1.6 m) sequence of aeolian and sandplain sediments, including a Pleistocene unit of indurated reddish sediments containing stone artefacts, animal bone, ochre and charcoal. A single radiocarbon date of 23,550 ± 140 BP (ANSTO OZB582)(∼28 ka) provides a minimum age for the upper part of this unit. Use of Serpents Glen began somewhat earlier, possibly as early as 30 ka. The site was abandoned shortly after 28 ka (this is one of the few sites where a gap in the occupation sequence is marked by a substantial sterile layer), and use did not recommence until 4,710 ± 180 BP (ANU10026).

DESERT PEOPLE There are two potential sources of information about the identity of these early desert people: genetic data (mainly mtDNA) and skeletal remains. Genetic research in Australia has focused on phylogenetic rather than population history, partly because of the small samples available. Data are available for fewer than 200 individuals drawn from only two areas – Yuendumu in Central Australia and communities in the Murray–Darling basin (van Holst Pellekaan and Harding 2006). These data show that desert groups share the primary founding mtDNA lineages identiﬁed in Australian populations (haplogroups NS1aandM42), as well as some younger lineages (haplogroup NO1) (van Holst Pellekaan 2008), although all of these have divergence age estimates of at least 50 ka. There has not been enough research done to determine whether any haplogroups are restricted to desert populations, nor whether this pattern reﬂects an initial settlement of the desert at about 50 ka or simply later migration by people bearing common Australian haplogroups. The skeletal remains of 132 individuals have been recorded in the Willandra Lakes area (Webb 1989), but only three (WLH1,WLH3 and WLH135)are knowntodatetoatleast30 ka. All of these are from a 500-m stretch of the lunette at the southern end of Lake Mungo and represent the oldest human remains from Australia. WLH1 and -3 have been well studied, but WLH135 (a child burial in a crouched position) has been left in situ for nearly 20 years because of the opposition of contemporary Aboriginal groups to further work on these remains.

WLH1 (Mungo 1) WLH1 (Mungo 1) – the ﬁrst to be discovered (Bowler et al. 1970; Bowler, Thorne and Polach 1972) – is a cremation burial of a slightly built young woman, 16–25 years of age (Curnoe and Thorne 2006). The evidence suggests a staged mortuary ritual (Bowler et al. 1970). Her body may have been dried prior to cremation because some bones had minimal protection from soft tissue Foundations: Moving into the Deserts 93

Figure 4.7. The Mungo 3 (WLH3) burial, showing the outline of the grave (line), the distribution of powdered red ochre (light grey) and the denser concentrations of ochre on the head, chest and groin (dark grey). (After Bowler and Thorne 1976:ﬁg.2)

when her corpse was burnt. The burnt bones were collected, thoroughly smashed and buried in a shallow conical hole (16–20 cm deep, 75 cm in diameter) while the remains of the pyre were still hot enough to differentially calcine some pieces.

WLH3 (Mungo 3) WLH3 (Mungo 3) is a man buried in a shallow grave (80–100 cm deep) in beach sands, about 150 m from the edge of the lake (Bowler and Thorne 1976; Webb 1989: 52–7;Brown2000; Thorne and Curnoe 2000)(Figure 4.7). He would have been about 50 years old when he died. His corpse was covered with about 1 kg of powdered red ochre, thought to be high-grade vein haematite brought from the Broken Hill–Olary district, 200 km to the northwest. He died towards the end of the ﬁrst freshwater phase of Lake Mungo, just as regional aridity began to mobilise pelletal clays on the lake margins, cloaking 94 The Archaeology of Australia’s Deserts

the lunette with aeolian dust (wustenquarz¨ ) (Bowler and Magee 2000). An attempt to date the bone directly using ESR and U-series (Thorne et al. 1999) produced ages (61 ± 6 ka) that were inconsistent with the chronology of the Mungo lunette (Bowler and Magee 2000). Further OSL dating of the lower Mungo sediments indicated that both WLH1 and WLH3 dated to 40 ± 2 ka (Bowler et al. 2003), a result confirmed when OSL ages of 40.9 ± 4.5 ka were obtained directly on the grave fill of Mungo 3 (Olley et al. 2006). Other analyses give us a glimpse of the man and his life. He was a moderately tall, lightly built man, about 170 cm in height (Bowler and Thorne 1976; Thorne and Curnoe 2000 –butseeBrown2000 for a contrary assessment of the sex of WLH3). Extreme osteoarthritis had partially immobilised his right elbow some years before his death: repetitive stress from spear throwing appears to have precipitated a severe infection of his upper arm (Webb 1989: 52–7). The synchronous premortem loss of both lower canines suggests a form of tooth avulsion. There was also a distinctive pattern of heavy tooth wear that suggests he prepared string by drawing siliceous plant fibres across and down through his teeth. Significantly, he had none of the uniform heavy attrition that typifies groups relying on ground seed meal (Webb 1989: 67). MtDNA extracted from his skeletal remains (Adcock et al. 2001) also suggests that WLH3 may be haplogroup M42, one of the deepest Australian mtDNA lineages (van Holst Pellekaan and Harding 2006). Most analysts, however, discount the possibility that WLH3 represents a relict mtDNA lineage outside the modern range (see Trueman 2001;Groves2001;Colgan2001). WLH1 and -3 reinforce the picture of well-established societies on the fringe of the desert by 40 ka. By this time, mtDNA data suggest that the dynamics of genetic drift operating on small, widely dispersed hunter-gatherer populations (including founder effects, isolation and local extinction) had already led to significant divergence from the original founding populations in Sahul. We could expect the desert to be a strong vicariant agent, although it remains to be established how much of the early genetic differentiation actually took place in the arid zone.

ASSEMBLAGES AND SITE INVENTORIES On the basis of their initial work at Lake Mungo, Bowler et al. suggested that Pleistocene Australian stone artefact assemblages belonged to a widespread industry they called the ‘Australian core tool and scraper tradition’, albeit a tradition with ‘many variants in time and space’ (1970: 52). The accumulation of data since then has undermined this notion. Although few sites dating from 30–45 ka have produced large assemblages of stone artefacts (Table 4.2), most have generalised small ﬂake assemblages, except where the availability of large blocks of silcrete or silcrete cobbles has led to larger thicker ﬂakes and Foundations: Moving into the Deserts 95 table 4.2. Archaeological sites in Australian deserts and drylands dating 30–50 ka (assemblages and site inventories)

No. stone Assemblage Other finds and Site or locale Site m2 Excavated m2 artefacts composition features N desert margins Carpenters Gap 1505 ? Small flakes, broken Ochre, ochre-stained flakes and flaking rock slab debris Riwi ∼110 1 ∼500 No data Ochre, Dentalium shell beads, intact hearths GRE810119Small flakes and core Ochre SE desert margins Willandra Lakes –c 70 775b Large flakes, horsehoof Ochre-covered burial cores, steep-edged (male), cremation and convex scrapers burial (woman), intact hearths and fireplaces, small shell middens Lake Menindee – – – – Intact hearth Cuddie Springs ∼990 225 666a Flakes, cores (flaked Ochre, hammerstones, cobbles), horsehoof manuports, cores, retouched amorphous flakes and amorphous grindstones, retouched pieces seed-grinding implements (?intrusive) Lake Tandou – –g ? Flakes and cores Hearths, small shell middens, burnt faunal remains Arid interior Allens Cave 180 3d 4 Small flake and flaked Intact hearth pieces Djadjiling 80 4.5 <30 Small flakes and cores Intact hearths Juukan 1 200 1 2–3 No data Puritjarra 400 14 42f Small flakes, trimming Red ochre, grindstone flakes (convex scrapers), flaking debitage, platform and bifacial cores Kulpi Mara 600 3 92 Small flakes and Red ochre, diffuse retouched hearths implements Lake Yantara – – – – Isolated hearth Newman P2055 40 2 41 Small flakes, flake Ash lenses fragments and retouched implements (continued) 96 The Archaeology of Australia’s Deserts table 4.2 (continued)

No. stone Assemblage Other finds and Site or locale Site m2 Excavated m2 artefacts composition features Lake Gregory – 4+ 1 Platform core – Serpents Glen 160 1 32 Small flakes, flake Ochre fragments and multiplatform cores Arid west coast Jansz 40 1.25 299 Flakes, debitage and Melo shell artefacts, red cores or yellow ochre Mandu Mandu 80 3 ∼100e Large flakes, cores, Melo shell artefacts, debitage and Pinctada fragments steep-edged (nacreous retouched flakes ornaments?), Conus shell beads C99 ∼300 2.25 42 Flakes, core and debitage Melo shell artefacts Pilgonaman 105 2 – – Reuse of old (30 ka) Melo shell in early Holocene context Silver Dollar ? ? ? Stone artefacts Melo shell artefacts Noala 30 2.25 2 Calcarenite flakes Geloina clam shell a G10 and F10 totalling 2 m2 b Mungo B excavation c Twenty sites at ∼4 m2 each d 1989 excavations only (another 7 m2 were excavated in 1972) e Spits 16–19. f Only 6.5 m2 of unit 2c were uncovered. g About thirty small sites at the base of the Bootingee unit on the Tandou lunette, but only four dated to >30 ka, all are small sites 1–2 m2.

‘horsehoof’ cores. More common are assemblages dominated by small squarish flakes (20–30 mm long) from only lightly prepared platform cores or bifacial cores (Figure 4.8). In general, there is greater variability among sites than across time periods (i.e., intersite variability is greater than temporal variability). This suggests that local circumstances (proximity to stone sources, type of raw material, extent of reduction, and the logistics of stone transport) underlie much of the variability in these assemblages – even in the Willandra (Hiscock and Allen 2000). There has been little success in identifying industries with well-defined functional, temporal or geographic signatures, even allowing for the fact that site assemblages may only be facies of an industry. The overall impression is that these assemblages represent expedient flake production for maintenance rather than for extractive tasks. The failure to generate a workable lithic systematics despite four decades of research into Australian Pleistocene assemblages suggests that stone flaking was an unreflexive daily practice, peripheral to the material culture and wider technology of these groups. Foundations: Moving into the Deserts 97

Figure 4.8. Artefacts from the 35 ka palaeosurface at Puritjarra: (1)N13/24–1 is a large sandstone flake typical of the larger component of the flake assemblage; (2)M11/27–2 is a chalcedony flake with a short length of retouch or edge damage; (3)N13/25–3 shows a sandstone flake detached from a rotated core. Bottom two rows (4)–(10)show small finely made silcrete flakes: (4)M11/27–4 is a trimming flake detached from the retouched edge of a chalcedony implement; (5)N12/26–1;(9)M11/27–6 and (10) M11/ 27–5 are made on exotic silcrete. (Source:Smith2006:fig.13)

Few assemblages contain retouched implements: most implements are ﬂakes or pieces with a single retouched edge. Steep-edged or notched implements occur at a number of sites (Lake Mungo, Cuddie Springs, Puritjarra and Mandu Mandu). People visiting Puritjarra before 30 ka used ﬁnely retouched convex scrapers made of grey silcrete from sources in the Cleland Hills (but these are 98 The Archaeology of Australia’s Deserts

only represented by the trimming flakes removed from their edges when they were resharpened) (Smith 2006). Grindstones are rare or absent before 30 ka, suggesting that intensive processing of plant foods or bone meal was not a frequent activity: For these late Pleistocene groups, any depletion of high-ranked foods appears to have been offset by high residential mobility. Although grindstones are present amongst Pleistocene assemblages in northern Australia, the only arid-zone examples in dated contexts before 30 ka are from Puritjarra (Smith 2004) and Cuddie Springs (Fullagar and Field 1997). A distinction between expedient grindstones and specialised seed-grinding implements needs to be made: the latter are only known from late Holocene contexts (see Chapter 6), with the possible exception of Cuddie Springs (where they may be intrusive from Holocene levels). With nearly forty excavated sites in this time period, it is clear that the first human movements into the arid zone were not underwritten by the harvesting and milling of grass and acacia seeds. The widespread use of baler shell (Melo sp.) as containers, ornaments and implements is shown by site assemblages near the Pleistocene coast at North- west Cape, and at Shark Bay about 50 km inland. The most remarkable finds, however, are twenty-two small Conus shell beads found at Mandu Mandu, at the base of the deposit in a tight scatter (spread over no more than 50 cm) (Morse 1993b). The six complete shells retain perforations for stringing the beads. Two have worn notches, and the full string of beads may have been 180 mm long. This is paralleled by the recovery of ten shell beads at Riwi (Balme and Morse 2006). Made from the anterior ends of tusk shells (Dentali- idae), these range from 5.2 mm to 17.55 mm long. Like the Mandu Mandu beads, traces of wear, fibre and pigment show that the Riwi beads were mounted as a string of beads and coated with red ochre. Red ochre is ubiquitous in the archaeological record (see Table 4.2). Most often, this is found as small fragments within archaeological deposits, suggesting that it was used as body paint or as a preservative rather than in the production of rock paintings. At Carpenters Gap, however, the presence of a slab of roof fall with red ochre is strongly suggestive of some form of parietal art as early as 38–47 ka (O’Connor and Fankhauser 2001). This is significant because rock art is a strong correlate of culturally constructed and maintained landscapes. Geochemical sourcing of red ochre has also helped determine whether the archaeological evidence reflects an established hunter-gatherer system rather than intermittent use of a region. At Puritjarra, evidence for extraction of red ochre from the Karrku mine (Smith et al. 1998) points to the establishment of a fully operating regional system by 35 ka. Despite this, none of the deeply weathered ‘archaic’ rock engravings that are found widely across the arid interior can be shown to date as early as 30 ka (see Chapter 7). These site assemblages also provide an indication of the range of late Pleis- tocene groups in the arid zone. Ochre was moved over distances of 100–200 Foundations: Moving into the Deserts 99 km at Puritjarra and Lake Mungo, and possibly 300 km at Mandu Mandu (although local sources of ochre in the Cape Range are now known – Przy- wolnik 2002). Baler shell was moved 50 km inland to reach the Silver Dollar site, comparable to the maximum distance that most flaked stone was transported. At Noala Cave in the Montebello Islands, the presence of stone from sources in the Pilbara shows that stone was sometimes moved over 150 km (White 2007) when local sources were inadequate. All of these examples are within the range of annual movements by some ethnographic desert groups and point to similar levels of mobility in the late Pleistocene. It is worth remembering that even this scale of travel involved a degree of negotiated access because it brought people into areas where sources of stone or red ochre were controlled by other descent groups. Whether this was also the case in the late Pleistocene, we cannot tell. Some material, however, was moved over much longer distances. The Dentaliidae shell beads at Riwi must have come from the Pleistocene coast, 350–400 km away. Whether this indicates a form of long-distance exchange or more extensive territories at this time remains to be determined.

SUBSISTENCE AND ECONOMY These early groups were generalised foragers, exploiting reptiles, small macropods and emu eggs, supplemented with marine or lacustrine resources where available (Table 4.3). Whatever the economic orientation of the first colonists to reach the shores of Sahul, by the time people reached the borders of Aus- tralia’s deserts and drylands, they were well adapted to the inland savanna (O’Connor and Veth 2000). The most common prey species were small and medium macropods, such as Macropus agilis, M. eugeni/greyi and the desert hare wallaby Lagorchestes, as well as varanid lizards and the eggs of Dromaius novaehollandiae. For instance, at the Silver Dollar site, the faunal remains are dominated by Lagorchestes hirsutus (Dyason 2007) and emu eggs. In the Willandra, the Pleistocene shell middens tend to be small single-event accumulations, limited to thin scatters of shells extending over 3.5 m2.Ina review of the evidence, Johnston (1993) concluded that this reflected small- scale use of lake resources, supplemented with terrestrial foods, rather than a specialised lacustrine economy. This is supported by analysis of the faunal remains at Mungo B (Walshe 1998) that shows these represented year-round hunting by people opportunistically tapping into the resources of the lakes, arid grasslands and open woodlands around them. Most game was in the size range 40–1,500 g; it included native mice, dasyurids, bettongs, hare wallabies, small macropods (M. eugeni/greyi) and shingle-back lizards, as well as emu eggs, golden perch and freshwater mussels. In the limestone gorge systems on the desert fringe, a similar pattern may have been in place. The large size of Alathyria mussels at GRE8 indicates that early harvesting of these shellfish was 100 The Archaeology of Australia’s Deserts table 4.3. Archaeological sites in Australian deserts and drylands dating 30–50 ka (subsistence remains from levels >30 ka)

Site or locale Plant remains Terrestrial fauna Marine or aquatic fauna

N desert margins Carpenters Gap 1 Tr iodia and Small macropods and – Cyperaceae stems, reptiles wood shavings, chenopod seeds Riwi Tr iodia stems, seeds Small macropods and Dentalium shell beads and paperbark reptiles GRE8 – Medium-sized Alathyria mussels macropods and/or Trichosurus, varanid lizard SE desert margins Willandra Lakes – Native mice, dasyurids, Fish otoliths and fish bettongs, hare bone (golden perch), wallabies, small Velesunio shell macropods (M. middens, crustacean eugeni/greyi), gastroliths (yabby) single-back lizards and emu eggshell Lake Menindee – Emu eggshell – Cuddie Springs Starchy or silicious Large macropods – plants (including red kangaroo), potoroos, varanid lizards, and emu. Extinct megafauna (?not contemporaneous) Lake Tandou – Native mice and rodents, Fish otoliths and fish bettongs, hare bone (golden perch wallabies, bettongs, and Murrray cod), bandicoots, wombats, Velesunio and Alathyria small lizards and emu shell middens, eggshell crustacean gastroliths (yabby) Arid interior Juukan 1 – Large macropods – Kulpi Mara – Minor bone – – unidentified Serpents Glen – Minor bone – – unidentified mammalia Arid west coast Jansz – Macropods, emu eggshell Melo shell artefacts, marine turtle, fish, chiton, crustaceans Foundations: Moving into the Deserts 101

Site or locale Plant remains Terrestrial fauna Marine or aquatic fauna

Mandu Mandua – Medium-sized macropods Sandy shore mollusc (M. agilis), thylacine and including chiton, bandicoot (Isoodon) Dentalium and Pinctada, Melo shell artefacts, Conus shell beads C99 – Unidentifiable bone Melo shell artefact Pilgonaman – – Melo shell artefacts Silver Dollar – Lagorchestes hirsutus,emu Melo shell artefacts eggshell Noala – Medium-large macropods, Geloina clam shell, fish Lagorchestes,bilby bone, five small (Macrotis), bettongs, fragments of Terebralia bandicoots (Perameles), shell varanid and other lizards a Spit 16–19 only.

not sufficient to affect the size structure of the mussel population until after 30 ka, when the mussels declined in size (Slack 2007: 249). There were exceptions to the general pattern, however, when large numbers of people gathered to take advantage of locally abundant resources. People may have used gill nets to catch large numbers of fish at Lake Tandou (at two sites: TNL20 at 27 ka and TNL36 at 29 ka) (Balme 1983, 1995). This may also have been the case at Lake Mungo (Kefous 1977). So far, all the evidence for nets and large gatherings is younger than 30 ka, although WLH3 may well have been involved in the manufacture of cordage as early as 40 ka. The scale of harvesting is illustrated at TNL36. This site represents a single catch of 357 golden perch (Macquaria ambigua) – mostly small fish about 25 cm (adult perch can reach 50–76 cm long) – and three Murray cod (Maccullochella sp.), sufficient to feed 50–60 people. If gill nets were involved, then some 7,500–9,000 mof cord would have been required, which would have taken between 97 and 112 days to make (Balme 1995: 17). All of this would be a significant corporate and cooperative investment in planning, labour and consumption. However, one caution should be sounded: although the use of nets is inferred from the narrow size range of fish otoliths at these sites, the ecology of these lakes would lead to fish populations made up exclusively of sexually immature fish at certain times (Walshe 1998). TNL36 could therefore represent the harvesting of juvenile fish stranded by seasonal drying of Lake Tandou. Whether or not nets were used, these sites reflect significant gatherings of people to take advantage of local resource gluts. The Northwest Cape sites show that people were also foraging along the Pleistocene coast, with most of the evidence indicating use of sandy beaches 102 The Archaeology of Australia’s Deserts

near rocky peninsulas (Veth 1999). People were taking sea turtles and crabs from a shallow sandy or muddy littoral, as well as chitons from rock platforms. None of the evidence points to a dependence on coastal resources or a major investment in their harvesting (O’Connor and Veth 2000). Fish and shellfish are rare in assemblages before 30 ka, and marine shell is limited to pieces likely to have been used as containers, implements or ornaments. The lack of evidence for substantive harvesting of shellfish is significant because some sites lie within 2–5 km of the Pleistocene coast, and all have good preservation of shell and bone. These assemblages also lack fishhooks, barbs, or prongs for fishing spears or lures, sinkers, or other specialised fishing gear. Although we might have expected an initial focus on the coast with only later adaptation to an arid hinterland, this does not appear to be the case. In fact, at Mandu Mandu, use of coastal resources increased after 30 ka, which suggests that desert people were compelled to make greater use of the coast as aridity increased – despite the fact that falling sea levels meant the coastline was now farther away. Botanical remains are rarely preserved in sites, except in the limestone sites of the southern Kimberley. Both Riwi and Carpenters Gap show use of Triodia (an arid hummock grass species), possibly for resin. At Carpenters Gap, the excavations also recovered Cyperaceae stems (possibly for edible corms) and burnt chenopod seeds. The overall picture is of broad-based foraging with a diversity of subsistence patterns across Australian deserts and drylands by 30 ka. These people were using dryland resources, as well as exploiting the palaeolakes, limestone gorges and coasts on the margins of the desert. Wherever available, marine, lacustrine and riverine resources formed part of a mixed economy, but there is nothing to indicate that occupation of the arid zone early in MIS3 was conditional on the ability to use these resources.

ECOLOGICAL IMPACTS The arrival of any new species in an ecosystem will lead to a period of adjustment before a new equilibrium is reached, especially if the new arrival is a large predator. So, we could expect a priori that the entry of humans into arid Australia would have a measurable impact on arid ecosystems. The question is whether we can see this in the palaeoenvironmental record and, if so, whether it provides an independent marker for colonisation of these regions. There are two main lines of evidence: the impact of human fires on vegetation, and extinction of the megafauna. Research into Aboriginal fire ecology indicates that the human signature is a mosaic of small patches of vegetation in different successional stages (Bird et al. 2008). These reflect the distribution of people, and the extent of modification of local vegetation relates to the intensity of foraging and hunting activity. Foundations: Moving into the Deserts 103

Anthropogenic fires have little effect on the composition of vegetation at the regional level, but they do modify the grain of the landscape at the local level. Although late Pleistocene groups may have fired the vegetation, these studies indicate that the effects are likely to have been localised, mirroring the patchy distribution of this population. A very different view is put by Miller et al. (2005), who argue that the initial entry of people led to a massive increase in fires, and that this had transformed the vegetation of the continent by 50 ka. In the arid interior, this involved the complete loss of a distinctive mosaic of endemic deciduous woodland, vine thicket and C4 tropical grasses. The evidence for these vegetation changes is a shift in isotopic signatures of Dromaius eggshell from the Lake Eyre basin, reflecting changes in the diet of emu populations. Palynology, however, provides no direct evidence for the presence of this type of woodland in the interior at 50–100 ka (van der Kaars and De Deckker 2002;Smith2009b) (see Chapter 3). The Exmouth (Fr10/95 GC17)core shows that open eucalypt woodlands in the northern part of the arid zone responded to a drying trend, initiated well before the earliest presence of people; this reflected a decline in summer monsoon rainfall over the region and a corresponding shift to an open shrubland by 46–40 ka (Kershaw, van der Kaars and Moss 2003). Van der Kaars and De Deckker (2002) explicitly ruled out human involvement in these vegetation changes. In general, more frequent and more intense fires tend to be associated with climatic transitions and drier climates (Hope et al. 2004; Kershaw et al. 2006), suggesting that increasing fire frequency is unlikely to be an unambiguous indication of the first entry of people into a region. There is another reason for doubting that anthropogenic fires would have abruptly transformed Australian deserts: these are not fire-starved landscapes where a novel source of ignition (anthropogenic fires) is likely to breach an environmental threshold. Finally, the eggshell data themselves have been interpreted as a climatic rather than anthropogenic signal: Johnson et al. (1999) argued that the significant narrowing of the diet of emu at 45 ka registered a weakening of summer monsoon circulation, and this was supported by reanalysis of the eggshell data by Murphy et al. (2012). All of this suggests that the eggshell isotopic data are more simply explained as a record of a climate shift towards drier conditions in the Lake Eyre basin. The second area in which a human impact has been suggested is in the extinction of Australian megafauna. Across the continent, the last traces of many species occur at 46 ± 5 ka, coinciding with an expansion of human settlement across the continent (Roberts et al. 2001). In the arid zone, however, most megafauna were already gone by then. The Katapiri fauna in the Lake Eyre basin had collapsed with increasing aridity by 60–65 ka, and most species were contracting to the temperate parts of the continent (Webb 2009,see also Chapter 3). People would not have encountered significant guilds of megafaunal species until they reached the temperate woodlands in southeastern 104 The Archaeology of Australia’s Deserts

Australia. When people entered the arid zone, the only widely distributed species was Genyornis – a large flightless bird weighing about 270 kg – with a range mostly limited to the southern Lake Eyre basin and Murray–Darling region (Rich 1979, 1985). A large dated series of Genyornis eggshell (Miller et al. 1999)showsthat this bird had disappeared across much of its range by 50 ± 5 ka. Miller et al. (2005) argued that anthropogenic burning of the vegetation was responsible: Genyornis, a highly specialised feeder, could not cope with the destruction of the C4 vegetation on which it depended, whereas the emu, an opportunist, simply ate whatever was available. Other data contradict this picture: some Genyornis ate primarily C3 browse, indicating the bird was a selective but not an obligate C4 feeder (Miller et al. 1999). The presence of dietary C4 signatures in Genyornis eggshell in temperate regions beyond the distribution of C4 grasses also suggests this bird was feeding on C4 chenopods, not tropical grasses (Smith 2009a). None of this rules out the possibility that the extinction of Genyornis across its remaining range was due to the entry of hunter-gatherer groups into the Murray–Darling region and along the ephemeral river systems east of Lake Eyre. But the link between anthropogenic fires, habitat disruption and extinction is not well established. None of the early sites in the arid zone shows hunting or scavenging of megafaunal species (except perhaps Cuddie Springs). Isolated finds of extinct species in archaeological contexts at Lake Menindee (Cupper and Duncan 2006) and in the Willandra (McIntyre and Hope 1978) are thought to reflect bone reworked from earlier sediments. Many sites have abundant emu eggshell, but none preserves the distinctive and more robust eggshell of Genyornis. Burnt eggshell eroding from dunes in the Northwest Cape area may record human exploitation of Genyornis eggs, but this is yet to be confirmed. Currently, this leaves us with no direct data on the nature of the interaction between the expanding human population and the remaining Genyornis populations. The ecology of this bird suggests it was vulnerable to declining monsoon rainfall and habitat fragmentation across much of its range. Given this, even low levels of human predation and egg collecting may have been sufficient to destroy the long-term viability of these birds ( Johnson 2006;Smith2009a). Colonisation of new regions often involves translocation of useful plant and animal species (Grayson 2001). Interisland transport of phalangers is known from the New Guinea archipelago during the late Pleistocene (Allen, Gosden and White 1989), but there are few known examples from Australian prehistory apart from the dingo (see Chapter 6). Recent work by Kondo et al. (2012)suggests that Livistona mariae – an endemic palm that forms an isolated population in Central Australia – is a late Pleistocene introduction. They estimated that genetic divergence between L. mariae and its closest relative, L. rigida, found 1,000 km to the north along the Roper River in northern Australia, took place Foundations: Moving into the Deserts 105 around 15.4 ka or 32.9 ka (depending on estimated mutation rates). Livistona seeds are 12–20 mm in diameter and are normally dispersed either by fluvial action or when they are consumed and excreted by fruit-eating birds and bats. Because dispersal across 1,000 km by birds and bats is unlikely, people may have been the vector. This is not as straightforward as it first appears: Livistona is an unlikely candidate for trade or translocation because the palms are of limited economic importance (Latz 1995), and the seeds are not decorative. As Kondo et al. point out, ‘although it remains unclear which vectors introduced L. mariae to arid central Australia over such a huge distance and across a hostile landscape, it is as likely that it is a legacy of Aboriginal dispersal as it is that it was carried there by animals’ (2012: 8). If so, it is possible that the palm was introduced as part of the first human moves into the region, or perhaps later during the terminal Pleistocene as part of the expansion of human groups after the LGM. However, the wide age interval associated with the divergence estimates (ranging from 6,600 to 83,000 years) means the archaeological significance of this palm is presently unclear.

DISCUSSION: MOVING INTO THE DESERTS

A Global Perspective The ﬁrst human movements into Australia’s deserts and drylands formed part of the initial dispersal of hunter-gatherer groups across a new continent. From a global perspective, late Pleistocene settlement of these regions is one end of an arc of dispersal, one that takes modern humans from Africa to Sahul. This involved not only the geographic spread of people into new continents but also movements into other world deserts and dryland regions. People appear to have spread across arid North Africa by 100–70 ka (Garcea 2004), about the same time that traces of anatomically and behaviourally modern humans ﬁrst appear in the archaeological record (McBrearty and Brooks 2000). By the close of MIS5, people were using palaeolakes and playas in the eastern Sahara (Wendorf, Schild and Close 1993) and in the Kalahari (Robbins et al. 2000). Aterian hunters were using arid uplands in the central Sahara by 69–90 ka (Di Lernia 1999) as well as sites along the Moroccan littoral (Barton et al. 2009). In the ranges of the southern Namib Desert, a series of rockshelter sites shows that there was a resident Middle Stone Age hunter-gatherer population in this area sometime before 34 ka (beyond the limit of available dating at these sites) (Wendt 1972; Thackeray 1992). In Australia and the Americas, the movement into the deserts took place as part of dispersals across new continents, by people who had already demonstrated the capacity to undertake successful sea voyages (Australia), cross extreme high-latitude cold environments (North America) or occupy high-altitude environments (South America). 106 The Archaeology of Australia’s Deserts

The emerging picture is of adaptable hunter-gatherers occupying parts of the Australian arid zone almost as soon as they reached its borders 45,000– 50,000 years ago. MtDNA and other generic evidence shows that these were anatomically modern humans. And many of the traits present in these early desert societies – red ochre, mortuary rites and ornaments – are hallmarks of the new patterns of human behaviour represented by this global diaspora. As Veth points out,

Occupation of marginal landscapes such as deserts appears to be dated as early as other modern behaviours, such as the development of artistic expression, and the ability to make ocean crossings. This creates the fascinating scenario in which modern humans appear to have explored and colonized the world’s deserts at the same time that they developed other ‘modern’ attributes underpinning a successful global diaspora. (2005b: 299)

In this context, the speed of dispersal across Australian deserts is not unusual. Macaulay et al. (2005) estimated that the primary dispersal process from India into Australasia involved rates of about 4.0 km/yr. This implies these hunter- gatherers already had a degree of ﬂexibility in social arrangements, an ability to operate autonomously in small, dispersed groups and a history of moving rapidly into new terrain.

Dispersal and Colonisation Most modelling suggests that movement across the interior would have been archaeologically instantaneous. The exact date of initial dispersal of people across the interior (the ‘pioneer’ phase) is uncertain but must have been rapid as people had crossed the arid zone to reach the southwest extremity of the continent by 48 ka. Isolated finds of artefacts at Puritjarra at 46 ka and at Lake Gregory at 50–45 ka may reflect short-lived visits associated with this dispersal phase, but it is clear that we still lack a significant part of the archaeological record before 35 ka. It may have taken some time to establish viable human populations across the entire region, and this may be part of the explanation for the poor visibility of early sites. The cline in dates from desert fringe to desert uplands to sandy deserts may yet turn out to reflect real differences in the probability of successfully establishing and maintaining a resident population in these areas. Current data show early occupation of inland river systems, palaeolakes and limestone gorge systems on the desert fringe by 50–45 ka, followed by settlement of the arid west coast and the major desert uplands by about 40–35 ka, with the sandy deserts occupied later, around 30 ka. This probably involved a pattern of small, spatially separated and intermittently linked populations across the arid Foundations: Moving into the Deserts 107 zone (the ‘metapopulation’ concept). The initial stages of settlement are likely to have been demographically and socially unstable. Some focal areas would not have supported a stable population in the face of ongoing environmental variability. Some would have a history of repeated episodes of extinction and recolonisation.

Desert Societies 45–30 Ka What can we say about the nature of these early desert societies? The small size of the living areas on most sites – and the small artefactual and fauna assemblages – suggests use by small groups, no larger than a family. Their range appears to have been similar to some ethnographic desert groups, moving over 100–200 km annually. If population nodes were widely dispersed (as I suggest), any depletion of high-ranked resources could have been offset by mobility rather than by processing labour-intensive plant foods. The rarity of grindstones supports this: there appears to have been little intensive processing of geophytes or grass and acacia seeds. The paucity of retouched, resharpened or recycled ﬂaked stone artefacts in these assemblages is consistent with this picture of mobile groups making only ephemeral use of sites. Except perhaps for Carpenters Gap 1, none of these early sites is inscribed with rock art. This suggests that individual and corporate relationships to place were looser than those in the ethnographic period. Following Gamble (1999), we might even think of these sites as ‘locales’ of activity rather than inscribed ‘places’. And yet, there are limits to how far we can push this picture of small, highly mobile groups. There is little evidence for free-range movement of people over long distances (>200 km): most indicators point to repeated use of the same territory, indicating that people ranged over circumscribed areas. Several of the Lake Tandou sites and Cuddie Springs provide evidence for large (possibly seasonal) gatherings. In any small dispersed population, these have a crucial role in negotiating relationships among groups, facilitating various sorts of social exchanges and in building the basis for economic reciprocity. The use of red ochre and elaborate mortuary rites hints at levels of sociality we cannot yet trace, even in broad outline. Where we have data, they suggest that red ochre was used as body decoration – for example, at Puritjarra (Smith, Fankhauser and Jercher 1998; Rosenfeld and Smith 2002) and on the corpse of WLH3 (Bowler and Thorne 1976) – rather than for parietal art. People may have used it to signal individual or group identity or as a cosmetic to protect the skin. Finally, although this chapter has looked at the foundations of Aboriginal society in these drylands, we should not lose sight of the fact that a vast stretch of time – perhaps 200–400 generations – separates the ﬁrst use of Puritjarra or 108 The Archaeology of Australia’s Deserts

the burial of WLH3 at Lake Mungo from the earliest moves into Australia’s deserts and drylands. This is time enough for people to have forged a distinctive way of life in the desert prior to the ﬁrst traces at Puritjarra at 35 ka. These populations would have had to cope with increasing aridity from 45 ka, lending support to the idea that there would have been signiﬁcant economic, social and behavioural adaptation in situ and that ‘in some important ways the modern deserts of Australia came to inland dwelling people, rather than the reverse’ (Hiscock and Wallis 2005: 42). CHAPTER 5

ISLANDS IN THE INTERIOR: LAST GLACIAL ARIDITY AND ITS AFTERMATH

Early in October, with the cold of a desert winter past, Pinta Pinta Tjapananga and his family set up camp near a new government handpump at Winbarku (Mt Webb). In 1984, this area – around Lake Mackay – was still one of the most remote parts of Australia. It had been depopulated between 1923 and 1965,as the Pintupi and their neighbours left the Great Sandy Desert for missions and government ration depots in Central Australia. The last phase of this exodus had drawn people out of the core of the desert – the ‘inside Pintupi’ – leaving a vast swathe of unoccupied dunefield a thousand kilometres wide. By 1984, this tide had reversed. Changes in government policy encouraged a homelands movement, and Pintupi people moved back to establish ‘outstations’ west of the Kintore range, reoccupying country they had left in the 1960s. Pinta Pinta and his family settled in for the night, confident that they had the area to themselves. But then, just beyond the firelight, an unfamiliar voice softly framed a request: kapi pu: u – ‘water’, ‘give me water’. Spinifex was quickly thrown on the fire, revealing a shadowy stranger, impossibly tall in the brief flare of light. One of Pinta Pinta’s sons fired a shotgun in the air. Both parties fled, Pinta Pinta driving 60 kilometres on a flat tyre. The ‘stranger’ was Piyiti Tjapaltjarri. Hiding in the bush beyond the firelight was his brother, Warlimpirrnga. The rest of their family – two older women, two boys and three young women or girls – were safely in a camp two days’ walk to the north. Dubbed the ‘Pintupi Nine’ by news media, they were among the last autonomous hunting and gathering group in the Western Desert – with no previous contact with Europeans. They had been stranded as the desert emptied of people and had been isolated from other Pintupi people for more than 20 years (Myers 1988;Kimber2006). After the death of Piyiti’s father, they had walked south, having seen ‘smokes’, to look for relatives. The ‘Pintupi Nine’ were isolated by a diaspora set in motion by European contact and accentuated by the severe droughts of 1924–9, 1944–5 and 1957– 66. Their remarkable story – just one of many accounts of small groups left stranded in the vastness of the desert (e.g., Davenport, Johnson and Yuwali 109 110 The Archaeology of Australia’s Deserts

2005) – puts a human face to one of the major historical dynamics in Australian drylands: the process of periodically abandoning and reoccupying marginal parts of the desert. This operates at a range of temporal and geographic scales, from interdecadal responses to severe drought to the longer rhythms of Qua- ternary climate change. And it makes the more marginal parts of Australia’s deserts a broad ‘contact zone’ – ‘the space in which people geographically and historically separated come into contact with each other and establish ongoing relations’ (Pratt 1992: 6). This chapter looks at an extreme example of this dynamic: the ﬂux of prehistoric settlement in Australian deserts between 30 ka and 12 ka. This spans a period of intensiﬁed aridity around the last glacial maximum (LGM) (30–19 ka), when large parts of the desert may have been abandoned, and the climatic recovery beginning around 18 ka, which saw the expansion of settlement from glacial refugia within the desert.

IDEAS ABOUT REFUGIA: ARCHAEOLOGICAL FRAMEWORKS The impact of the LGM has dominated thinking about the late Pleistocene archaeology of the desert since the 1980s, when the severity of the full glacial climate in inland Australia first became apparent (Bowler and Wasson 1984). In the northern hemisphere, this period saw ice sheets reach their maximum extent. In the mid-latitudes, however, it was intense aridity rather than ice that marked the peak of the last glacial period. In the interior of Australia, conditions were substantially colder and more arid than today, creating the most difficult set of environmental circumstances that people in this region have ever faced. The arid zone was also much larger than its modern counterpart, reaching almost to the margins of the continent (Figure 5.1). The consequences for Aboriginal groups were first outlined by geomor- phologists (Mabbutt 1971a;Bowler1976). Bowler stressed that the severe aridity accompanying the LGM ‘would have imposed considerable stress on almost every element in the landscape including Man’, sufficient ‘to induce significant and, hopefully, detectable changes in the distribution and adaptation of human populations’ (1976: 72–5). These issues – changes in distribution and adaptation, and ways of detecting these changes in the archaeological record – have been recurrent themes in archaeological research over the past 30 years.

The Contraction of Settlement In the late 1980s, four independent archaeological projects framed the problem in more detail (Lampert and Hughes 1987; Hiscock 1988a;Smith1988, 1989b; Veth 1987, 1989a, 1989b), arriving at much the same conclusions: the LGM was a punctuation event involving a major contraction in settlement and population in the interior. Islands in the Interior: Last Glacial Aridity and Its Aftermath 111

Figure 5.1. The geography of the continent during the last glacial maximum, showing inferred distribution of xerophytic vegetation (arid and semi-arid grassland and woodland) (Hope et al. 2004; Kershaw and van der Kaars 2007), continental dune- ﬁelds ( Jennings 1968; Hesse 2010), glaciers and the Pleistocene coastline (-135 m). Black triangles show archaeological sites discussed in the text: (1) Silver Dollar site, (2) Millys Cave, (3)Malea,(4) Juukan 2,(5)Yirra,(6) Carpenters Gap, (7) Bush Turkey 3, (8) Serpents Glen, (9) Puntutjarpa rockshelter, (10) Puritjarra, (11) Allens Cave, (12) Koonalda Cave, (13) Colless Creek Cave, (14) Louie Creek, (15) Lake Tandou and Darling River sites, (16) Lake Mungo and Willandra Lakes sites, (17) Cuddie Springs. 112 The Archaeology of Australia’s Deserts

Figure 5.2. Diagrammatic representation of biotic and human responses during the last glacial maximum. (a)–(c) The human ecology of refugia, showing the interrelationship of carrying capacity (dark grey = high; light grey = low) and the distribution of watering points (white). (a) An ‘oasis’, in which a single focal water point provides access to a small area of high productivity. (b) Spatially clustered springs or waterholes. Although there is a high density of watering points, their spatial distribution does not allow foraging range to expand to compensate for declining carrying capacity. (c) Network of small wells providing access to a large area, offsetting a decline in carrying capacity per unit area. (d)–(e) Types of glacial refugium (after Bennett and Provan 2008:ﬁg.1). (d) A ‘classical’ refuge (black) involving contraction of a biome from much of its former range (grey). (e) A ‘cryptic’ refugium, in which a species survives in scattered pockets of microhabitat (black) across much of its former range. (f )–(g) Patterns of land use. (f ) ‘Point-to- point’: people move from well to well, exploiting only a small area around each watering point. (g) Ethnographic home-base model. The annual subsistence round centres on a focal water, with movements out to use ephemeral waters, conserving resources around the main water until people are tethered to it during dry months. Lampert and Hughes (1987) suggested that the Flinders Ranges were ‘a Pleistocene outpost in the arid zone’. In a wide-ranging ﬁeld survey of the southeastern sector of the arid zone, they found that late Pleistocene sites were restricted to the Flinders Ranges (Hughes and Lampert 1980). The ‘harsh, des- iccating conditions of the last glacial maximum’ had, they said, ‘caused depopulation, people perhaps falling back on the better-watered ranges’ (Lampert and Hughes 1987: 32). Once the stressful conditions of the LGM ameliorated, there were occasional forays into the surrounding plain. By the early to mid-Holocene, people were making visits of longer duration to the shores of Lake Frome. Most desert lowlands, however, were not occupied until the late Holocene. My own review (Smith 1988) combined the palaeoenvironmental data with ethnographic information about human ecology in Australian drylands. The thinning out of watering points during the LGM would have left many parts of the arid zone inaccessible to human groups. I suggested that ‘human Islands in the Interior: Last Glacial Aridity and Its Aftermath 113

Figure 5.3. The location of glacial refugia (after Veth 1989b). In this conceptual model, the major uplands (the Central Australian ranges, Pilbara uplands, Flinders Ranges and the Kimberley) are refugia, the sandy deserts are biogeographic barriers, and the intervening areas represent corridors. populations would have largely been restricted to the major montane and piedmont regions or major riverine tracts – areas with the highest density of reliable watering points. The desert lowlands, in particular the major areas of sandy desert or stony plains, would have been abandoned’ (1988: 303). It was the interaction of three factors that determined whether an area was abandoned: the distribution and spacing of permanent water, the carrying capacity of the land at these points and the total contiguous area likely to have been accessible to people in any particular region (Figure 5.2a–c). The interplay of these factors meant that human groups might survive even in resource-poor desert lowlands if a network of small wells allowed access to a sufﬁciently large tract of land – a picture that ﬁnds at Puritjarra rockshelter reinforced (Smith 1989b). On the northern margins of the desert, Hiscock focused on sites in limestone gorges, showing that people in this region responded to LGM aridity ‘by reducing their territory and focussing on the abundant and reliable resources available in the gorge systems’ (1988a: 260), a pattern he described as ‘permanent occupation of an oasis’ (1988a: 258). Veth (1989a, 1989b, 1993)drewthe various strands together in a continental model (‘islands in the interior’) using biogeographic terms to contrast areas where human populations are likely to have persisted during the LGM (‘refugia’) with areas unlikely to have been settled until much later (‘barriers’) and with a broad zone where the distribution of settlement was sensitive to climatic shifts (‘corridors’) (Figure 5.3). 114 The Archaeology of Australia’s Deserts

The contraction to glacial refugia must have involved a gradual succession of local abandonments as people retreated from marginal areas. The responses of ethnographic groups illustrate the type of individual events that may underlie the longer trends. Protracted drought led to temporary abandonment of some areas, ﬁssioning of local groups and retreat to regions with better water resources (Peterson 1986). Population densities in stricken areas dropped, both through migration, as individuals activated their social networks to take up residence in neighbouring territory, and through mortality. Kimber (1990: 161–2) estimates that 10–25 per cent of the hunter-gatherer population in Central Australia may have died during severe droughts. Ideas about the scale of these impacts tended to hinge on assumptions about hunter-gatherer adaptation to aridity.If the interior was initially settled during a wetter climate, or by foragers targeting riverine or lacustrine resources or well- watered gorge systems, these societies might have been unable to cope with extreme aridity during the LGM (Veth 1993: 114; Thorley 1998a, 1998b: 42; Hiscock 1988a: 260). As Thorley put it, ‘the glacial maximum provides the ﬁrst and most crucial test of human adjustment to a truly arid environment’ (1998b: 42). Veth highlighted the role of the LGM in driving adaptation, arguing that groups in the interior would only have adapted ‘to truly arid conditions within the refuges of the glacial maximum, the islands of the interior’ (1993: 114). Later work showed that successful dryland hunter-gather societies were already in place by 30 ka, tailored to a range of regional circumstances and with a long history of living in open, arid landscapes (see Chapter 4).

Life in Glacial Refugia Some researchers argue that there would have been a concomitant increase in population density in refugia as desert populations concentrated in these areas (O’Connor, Veth and Hubbard 1993). This is unlikely if carrying capacity in glacial refugia had declined in line with increasing aridity. However, higher population densities may have been possible if economic intensification took place in the refuges – either in the form of more intensive harvesting of fish and shellfish or a shift to more plant-food processing involving harvesting of native cereals (Allen 1990; Edwards and O’Connell 1995; Fullagar and Field 1997; McConnell 1998) – but there is little direct evidence for this (see Chapter 6). Even if desert lowlands were abandoned, interannual environmental variability would have allowed opportunistic use of country (‘corridors’) surrounding the glacial refugia (Lampert and Hughes 1987;Veth1989b: 81; Hiscock and Wallis 2005: 44). The intermittent contacts this allowed would have been critically important for the demographic viability of desert people, connecting groups in scattered refugia. Some archaeologists see the low regional diversity of Pleistocene rock engravings as evidence for ‘the “open” social networks required by low density populations to maintain social and biological viability Islands in the Interior: Last Glacial Aridity and Its Aftermath 115 in harsh environments’ (Morwood 1988, cited in C Smith 1992: 37). Oth- ers argue that the LGM disrupted established patterns of trade and exchange (Morse 1993c;Veth1995a: 744), reflecting increasing fragmentation of social networks in the desert.

Reoccupation of Desert Lowlands In the aftermath of the LGM, Aboriginal settlement expanded into the desert lowlands in a more sustained fashion. Just how over rapidly this took place – and what factors governed it – have been the focus of some debate (Smith 1993;Veth1995b; Edwards and O’Connell 1995). In most areas, water would have been a critical limiting factor. In 1988, I argued that reoccupation would have rapidly followed climatic amelioration, with most areas reoccupied by 16–14 ka (Smith 1988). Others argued for a staged process (Lampert and Hughes 1987;Veth1989a, 1989b). In the most detailed of these arguments – the ‘clinal model’ – Veth argued that the large sandridge deserts ‘were the last of the desert habitats to have been occupied permanently by prehistoric hunter-gatherers’ because there were biogeographic barriers to their settlement (1989b: 83). These were only breached in the mid-Holocene by people with a ‘uniquely desert economy’ and the capacity to cope with high levels of environmental stochasticity (1989b: 85–90). The net effect was that the desert lowlands were recolonised gradually, beginning with the ‘corridors’ and finishing with the ‘barrier deserts’, in a long process between 15 ka and 5 ka, driven by ‘a combination of demographic pressure, technological shifts and changes in social structure’ (1989b: 83). However, not everyone accepted the premise that the continental dunefields constituted biogeographic barriers in the way specified in the model (Smith 1993). Another problem was that although some key Holocene adaptations (desert wells, extended social networks) seemed just as likely to be features of late Pleistocene societies, there was little evidence that others (seed-grinders, tula adzes) were in place until well after 5,000 BP. The most important aspect of the ‘clinal model’ was that it presented a much more sophisticated account of the process of colonising the sandy deserts. It put the view that population expansion in the aftermath of the LGM was not simply a response to climatic amelioration but rather a gradual process that involved the dynamic interplay of changes in demography, technology and social networks. The move out of glacial refugia involved not just geographic expansion but also a trajectory that created the distinctive desert societies known from Australian ethnography. Subsequent fieldwork in the sandy deserts, however, undermined the ‘clinal model’ by showing that these regions were settled before the LGM – by 28.5 ka at Serpents Glen (O’Connor et al. 1998) – and were reoccupied prior to 5,000 BP – sometime between 12 ka and 8 ka at Bush Turkey 3 (Veth, Macdonald and White 2008). 116 The Archaeology of Australia’s Deserts

WHERE ARE THE REFUGIA? BIOGEOGRAPHIC PERSPECTIVES Where are the major glacial refugia? Most archaeologists agree that the Central Australian ranges, the Pilbara uplands and Flinders Ranges are likely to have had persistent human populations (Lampert and Hughes 1987;Smith1988;Veth 1989a). But the evidence does not support a focus on ranges alone. Veth(1989b) originally identified seven biogeographic refugia. These included the major uplands (see Figure 5.3) and the river systems of the Murray–Darling basin and left open the possibility of small local refugia outside these areas. Other areas variously identified as glacial refugia include the Pleistocene coastline (Morse 1993c: 163;Veth1995a: 734, 1999: 69), extant lake systems (Hiscock and Wallis 2005: 44), the gorge systems in the Carpentaria region (Hiscock 1988a), and areas where networks of small wells allowed people to exploit large tracts of resource-poor land (Smith 1989b). Biogeography identifies several types of refugia (Bennett and Provan 2008). ‘Classical refugia’ are remnants of habitat that persist despite a shift in climate and involve contraction of a particular biome (Figure 5.2d). The best example of this is the retreat of forests to glacial refugia in southern Europe, as documented in Quaternary pollen records (e.g., Tzedakis et al. 2002). A second line of work, on species distribution, rests on the idea that nodes of high species diversity are centres of origin and dispersal. This was formalised in the ‘refuge theory’ (Haffer 1969, 1982). Haffer argued that fragmentation of habitat in the Amazon basin during glacial periods drove speciation in lowland forests. Many biologists have followed this, using species diversity or genetic bottlenecks to identify ‘phylogeographic refugia’. Molecular taxonomy has suggested a third type: ‘cryptic refugia’ (Bennett and Provan 2008: 2451;Byrne2008). In a cryptic refugium, the relationship between range and abundance is decoupled. A species may survive across much of its former range but as scattered occurrences and at low densities (Figure 5.2e). This involves a thinning out of a population, which then survives in pockets of microhabitat widely distributed across its former range. Some archaeologists have made similar distinctions for glacial-age Europe. For instance, Gamble et al. (2005: 200) distinguish between ‘G’ refugia (the minimum geographic distribution of the human population, irrespective of population size) and ‘M’ refugia (the minimum size of the metapopulation, irrespective of geographic extent). In Australia, most biological research on refugia has been driven by studies of species diversity and distribution. Crocker and Wood (1947) focused on arid-zone plant species (mainly eucalypts, acacias and Atriplex)–and,in a remarkably early application of the refuge concept, identified the Central Australian ranges and the Flinders Ranges as refugia during their proposed ‘great arid period’ (1947: fig. 5). Other research used the concept to explain speciation in Australian avifauna (Gentilli 1949), reptiles (Pianka 1981; Cogger Islands in the Interior: Last Glacial Aridity and Its Aftermath 117 and Heatwole 1981), butterflies (Kitching 1981)andTriodia hummock grass species ( Jacobs 1982). Australian research was synthesised by Heatwole (1987), who argued that ‘the cyclic fluctuation of Australian climate in the late Cain- ozoic . . . provided conditions under which refugia played an important role in the distribution and speciation of animals’ (Heatwole 1987: 111). His conceptual framework of ‘refugia’, ‘barriers’ and ‘corridors’ of dispersal influenced archaeological ideas about glacial refugia (Veth 1989a). However, like most Australian biogeographic researchers, Heatwole focused on phylogeography and did not identify specific areas that might have functioned as classic refugia. Although none of the Australian desert uplands is high enough to represent montane refugia like the ‘sky islands’ in the desert southwest of North Amer- ica, some studies suggest they do have significantly higher biodiversity than surrounding regions. Morton, Short and Barker (1995) list the major biological refugia in Australian drylands. They use high frequencies of endemic or relict species to identify seventy-six refugia, which they rank in order of importance (Table 5.1). Not all of these are relevant for prehistoric human ecology, but their study does confirm the status of major desert uplands as refugia. Unques- tionably, the most important of these in size and biodiversity are the Central Australian ranges and the Pilbara. Reconstructions of palaeovegetation are too few to test the idea that mesic and riparian biomes contracted into these areas during the LGM. One of the few studies available, based on charcoals and phytoliths at Puritjarra (Smith, Vellen and Pask 1995;Smith2009b), suggests changes in abundance rather than in the distribution of common large tree species during the LGM – a pattern more consistent with a cryptic refugium than with the contraction of a biome. Byrne (2008) came to similar conclusions. She used molecular taxonomy to investigate patterns of divergence in arid-zone plant and animal species. Most species showed signs of a major genetic bottleneck in the mid-Pleistocene, but there was little evidence of major contraction of range during the LGM (2008: 2582). The pattern of high diversity within lineages, as well as diversity across geographic area, indicates a pattern of multiple small, localised refugia allowing individuals to rapidly colonise suitable habitat in a patchwise fashion as conditions ameliorated. Collectively, these studies provide a more complicated picture of glacial refugia than archaeological studies have allowed. The phylogeographic studies suggest that the main impact of refugia on speciation was in the mid-Pleistocene, not during the LGM. Much of the pattern of evolutionary biodiversity that underpins the identification of dryland refugia by Morton et al. (1995)may also originate before the last glacial cycle. Some major desert uplands clearly represent durable nodes of biodiversity and are likely to have remained as focal areas for human settlement during the LGM. But it is also clear that it is too simplistic to equate refugia with uplands because not all uplands have high levels of persistent biodiversity. These biological studies also suggest that there 118 The Archaeology of Australia’s Deserts

table 5.1. Relative importance of dryland ranges and gorge systems as biological refugia

Name Type of refugea Aread Rankd

Southern Kimberley Bungle Bungles Gorges S 6 Oscar and Napier Caves Cave system M 6 Edgar Ranges Mountain range M 4 Geikie Gorge Gorge VS 3 Windjana Gorge Gorge VS 2 Tunnel Creek Gorge VS 2 Carpentaria Carpentarian sandstone Mountain ranges M 7 Lawn Hill Gorge Gorge VS 5 Pilbara Hamersley and Chichester Rangesd Mountain ranges L 8 Central Australia Western MacDonnell Ranges Mountain ranges L 9 Eastern MacDonnell Ranges Mountain ranges M 8 George Gill Range Mountain ranges S 7 Davenport and Murchison Ranges Mountain ranges M 6 Dulcie Ranges Mountain ranges M 4 South Australian Ranges Northern Flinders Ranges Mountain ranges M 7 Gawler Ranges Mountain ranges M 6 Lake Eyre Basin Selwyn Range Mountain range M 5 Northern Grey Range Mountain range M 5 Other Major Refugiae Cape Range Cave system S 7 Queensland Channel Country Ecological refuge L 7 Nullarbor Cave system L 6 Coongie Lakes Wetlands M 6 Uluru and Kata Tjuta Ecological refuge S 6 Dalhousie Springs Mound springs VS 6 Lake Eyre mound springs Mound springs VS 6

a ‘Ecological refuges’ are seasonal fallback areas. b Area: VS <100 km2,S<1,000 km2,M<10,000 km2,L>10,000 km2. c Each refuge is ranked according to the number of relict or endemic species, or species listed by the Australian and New Zealand Environment Conservation Coun- cil (ANZECC) as endangered or vulnerable. 9 = most significant. d This area also includes wetlands and marshland along the middle and upper reaches of the Fortescue River, which contributes significantly to the biodiversity and rank of this refugium. e Not shown are refuges from modern land clearing or from European introduction of exotic plant and animal species; islands; refugia for specialised saltlake biota or low-ranked wetlands. Source: Morton, Short and Barker (1995) Islands in the Interior: Last Glacial Aridity and Its Aftermath 119 may be areas outside of desert uplands that functioned as ‘cryptic refugia’ for scattered human groups. Given the difficulty of identifying refugia independently of archaeological data, ‘islands in the interior’ is now best regarded as a conceptual rather than strictly predictive model.

INLAND ENVIRONMENTS DURING THE LAST GLACIAL MAXIMUM The LGM has taken on much of the explanatory role that Crocker and Wood’s ‘great arid period’ once had (1947). But the sharp punctuation event at 21 ka (∼18 kyr BP) assumed by most archaeological reconstructions has been complicated by recent palaeoclimatic research. Was the LGM a brief extreme event? An extended period of arid but stable climate? Or an extended arid period with significant fluctuations in local climate? Current research suggests the latter, and this has implications for how the archaeological record is interpreted. Some definitions are necessary to clarify what follows. Marine isotope stage 2 (MIS2)(24–12 ka) (Martinson et al. 1987) includes a sequence of abrupt and extreme climate changes forming two global climatic episodes: (a) The LGM (26.5–19 ka) is formally defined as the period when global glaciers were at their greatest extent (Clark et al. 2009). World climate then was dramatically different from that of today: forests shrank, deserts and ice sheets expanded and sea levels fell to their lowest recorded levels, 135 m below modern sea level. The LGM was the driest of all glacial cycles but (on strict definition) refers only to the maximum extent of global ice. It does not directly refer to peak aridity or to precipitation and temperature minima in regional records. (b) The last termination (18–12 ka) is the period when global temperatures were rising rapidly and regional climates were in transition. Glacial– interglacial warming occurred in two steps, interrupted between 14.5 ka and 12.5 ka by a colder, drier, dusty period called the Antarctic cold reversal (ACR). Although the LGM is conventionally dated to 24–18 ka (Mix, Bard and Schneider 2001 – LGM chronozone level 2), evidence has steadily mounted that it began several millennia earlier, between 31.4 ka and 26.5 ka (Lambeck, Yokoyama and Purcell 2002; Alloway et al. 2007; Newnham et al. 2007; Pether- ick, McGowan and Moss 2008; Clark et al. 2009). Using sea levels as a measure of integrated global ice volume, Lambeck et al. (2002) showed that glaciation had reached its maximum extent by 30 ka and remained nearly constant until 19 ka. My approach here is to adopt the extended LGM (30–19 ka) suggested by the Southern Hemisphere studies. Not only was the LGM longer, but recent work also suggests that it was not a single phenomenon – at least, in terms of its impact on inland environments. 120 The Archaeology of Australia’s Deserts

Some evidence indicates stadial-type climatic instability, with abrupt episodes of greater aridity at climatic transitions (e.g., Petherick et al. 2008). Other evidence shows that there was considerable interannual variability in LGM climates. Short-lived but high-magnitude ﬂows along stream channels in Central Australia (Bourke 1998; English et al. 2001) and along Cooper Creek (Nanson et al. 2008) record occasional extreme rainfall events even during the period of full glacial aridity.

The Impact on Australian Drylands The impact of the LGM on Australian drylands appears to have been a two-step process, with a rapid enlargement of deserts and drylands at 30 ka, followed by a further intensification of aridity 24–20 ka. The first stage occurred about 30 ka, when sea level fell rapidly from −85 m to a maximum low stand at −135 m, falling 50 minlessthan1,000 years (Lambeck et al. 2002). This increased the size of the Pleistocene continent, exposing wide continental shelves in the north (200–300 km wide) and south (100–200 km wide). There were two direct consequences for the arid zone. First, the increased continentality of the Pleistocene landmass reduced the influence of maritime air masses in the interior. Second, these changes disrupted the monsoon that brings summer rainfall to the northern half of Australia ( Johnson et al. 1999; Hesse, Magee and van der Kaars 2004; Griffiths et al. 2009). The result was an abrupt increase in aridity about 30 ka (Hesse et al. 2004). As a consequence, the arid zone expanded to its greatest extent, covering an estimated 6–7 million km2. Inland vegetation responded rapidly: arid and semi-arid woodlands, hummock grasslands, and chenopod steppe expanded on the northern and eastern margins of the desert (Dodson and Wright 1989; Wallis 2001; Hope et al. 2004; van der Kaars, De Deckker and Gingele 2006; Frawley and O’Connor 2010). Pollen records from the Fr10/95-GC17 core register intensified aridity from 32 ka with virtually no summer rainfall (van der Kaars et al. 2006). Other records, from the Lake Eyre basin, show retraction of tropical C4 grasses from 28 ka to 15 ka ( Johnson et al. 1999) and fragmentation of Callitris woodlands after 30 ka (Luly 2001). The reduction in monsoon activity also dramatically reduced flows along the extensive network of rivers and channels feeding Lake Eyre in central Australia: as early as 35 ka, the lake had dried completely and its bed had been cut down by wind deflation (Magee and Miller 1998). Reduced vegetation cover triggered a major phase of dune-building in the Lake Eyre basin between 35 ka and 28 ka (Fitzsimmons, Rhodes, Magee et al. 2007; Fitzsimmons, Bowler, Rhodes et al. 2007) and in Central Aus- tralia around 30 ka (Nanson, Price and Short 1992; Jacobson, 1996; Hollands et al. 2006). Dunefields were reactivated across more than 40 per cent of the Islands in the Interior: Last Glacial Aridity and Its Aftermath 121 continent, including northwest Australia, where desert dunes dating to 32 ka now lie submerged in the Fitzroy estuary, King Sound ( Jennings 1975; Goudie et al. 1993). Destabilisation of land surfaces also led to frequent dust storms, recorded downwind in greatly increased dust fluxes in lakes, swamps and valley fills in southeastern Australia between 35 ka and 29 ka (Petherick et al. 2008; Haberlah et al. 2010) and in ocean cores from the Tasman Sea from 27 ka (Hesse and McTainsh 2003) – much of this sourced to floodplains in the Lake Eyre basin or Murray–Darling catchment. The second stage began about 24 ka and is marked by exceptionally cold and dry conditions, especially in the southern third of the continent, marking a further stepwise adjustment of local atmospheric and ocean circulation patterns. In the mountains of southeastern Australia and Tasmania, valley and cirque glaciers reached their maximum extent between 20 ka and 17 ka (Bar- rows et al. 2002), and most periglacial activity centres on 21.9 ka (Barrows, Stone and Fifield 2004). Sea surface temperatures in the oceans surrounding Australia also indicate that the coldest period was around 20.5 ka (Barrows and Juggins 2005). By 24 ka, landscapes across the arid zone had already begun to register the effects of the intensely dry conditions this produced. In northwest Australia, the coldest, driest period occurred 25–23 ka (van der Kaars et al. 2006: 888) with an open vegetation of daisies, arid grasslands and chenopods, and with drought-tolerant Callitris replacing eucalypt woodlands (van der Kaars and De Deckker 2002). In Central Australia, the most intense aridity occurred at 24 ka, when the regional vegetation became an open, arid shrubland following the collapse of grassland formations, including the otherwise hardy Triodia hummock grassland (Smith 2009b). Dust storms intensified, with dust fallout peaking in most records (De Deckker 2001; Hesse et al. 2004; Petherick et al. 2008), and there was another widespread phase of dune activity (Chen, Chappell and Murray 1995; Nanson et al. 1992, 1995;Jacobson1996; Hollands et al. 2006; Fitzsimmons, Rhodes, Magee et al. 2007). Hydrological systems became more efficient as runoff increased under conditions of reduced ground cover. The period around 20 ka saw a resumption of floods and groundwater discharge in the Lake Lewis basin in Central Australia (English et al. 2001). Late Pleistocene palaeochannels of the Sandover River, in Central Australia, also reflect a regime of episodic higher discharges prior to 15 ka despite intense regional aridity (Tooth 1997). In Brachina Gorge, Flinders Ranges, the LGM was marked by an extensive wetland set within an otherwise arid landscape, as fine sediments redistributed from dust mantles choked the valley (Haberlah et al. 2010; Glasby, O’Flaherty and Williams 2010). Lake Frome was dry or ephemeral during the LGM but filled to 6.8 m shortly afterwards (between 17.6 ka and 15.8 ka) as depressed temperatures and shifts in winter rainfall altered the hydrological balance (Cohen et al. 2012). Peaks in dust emissions from the lake show this was punctuated by regional 122 The Archaeology of Australia’s Deserts

drought or seasonal dry periods, and it became an ephemeral hypersaline playa at 15 ka (De Deckker, Magee and Shelley 2011). The end of the LGM is reﬂected in a slow transition to a warmer, more humid climate in inland Australia, with summer monsoon incursions into the interior reestablished by 16–14 ka (Singh and Luly 1991; Johnson et al. 1999; Wyrwoll and Miller 2001; Spooner et al. 2005). The Holocene transition is discussed in more detail in Chapter 6.

Implications for Human Ecology in the Interior We can model the likely effects of the LGM climate on human ecology in the interior if we consider some key climatic parameters. First, effective rainfall across northern Australia and the interior was 30–50 per cent less than today (Bowler and Wasson 1984; Dodson and Wright 1989; Hesse et al. 2004), with the strongest reduction in summer rainfall (van der Kaars et al. 2006). Some areas, especially in the Lake Eyre basin, would have become hyperarid under these conditions. Second, the average ambient temperature in the interior was at least 6◦C, and perhaps as much as 9◦C, lower than today (Bowler and Wasson 1984; Miller, Magee and Jull 1997), probably with greater seasonal and diurnal extremes. Whether this was made up of much colder winters combined with hot, dry summers, or a shift to colder conditions across all seasons (Hesse et al. 2004: 95), the implication is that many plant species would have faced a shorter growing season than today. Third, evaporation rates were about 20 per cent higher during the LGM, owing to a combination of low humidity, high wind speeds and high summer radiant energy (Bowler and Wasson 1984). These effects were sufﬁcient to offset any reduction in evaporation due to lower temperatures. Evidence for intense evaporative events in lake basins (Bowler 1998; Magee and Miller 1998) and the reactivation of marginal dune ﬁelds (Ash and Wasson 1983; Hesse and McTainsh 1999) points to stronger winds throughout the year – although dust transport patterns do not show this (Hesse and McTainsh 1999). Regional water tables were lower across much of the interior, to the extent that groundwater discharge into some saltlakes ceased and aeolian sediments buried the halite crust (e.g., Lake Amadeus – Chen, Bowler and Magee 1993). Finally, lower carbon dioxide (CO2)levels also imposed a degree of physiological aridity on plants, quite apart from the actual rainfall received. The net effect of this climate was to lower the primary productivity of desert ecosystems, affecting plant food species as well as invertebrates and small- and medium-sized vertebrates. There would have been fewer permanent springs, soakages or rock holes because lower rainfall meant that these were replenished less often and water tables were lower. Just as critical would have been the reduction in ephemeral or seasonal waters, such as claypans, small rock holes and seepages. High evaporation rates combined with low rainfall would Islands in the Interior: Last Glacial Aridity and Its Aftermath 123

Figure 5.4. Schematic diagram showing the impact of peak glacial aridity upon land use (from Smith 1989b:ﬁg.6). The ethnohistoric pattern shows that foraging territories focused upon permanent water with an annual subsistence round involving cyclic dispersal and aggregation around these waters. During the last glacial maximum, the few permanent waters are concentrated in the ranges. The annual subsistence cycle involves movement between these points, with each foraging territory incorporating a suite of waters. have reduced the number and use-life of these waters, critically affecting the accessibility of country beyond the reach of the main waters. It seems clear that climatic conditions during the LGM could not have supported a ‘home-base’ pattern of cyclic annual dispersal and aggregation centred on the main waters (Figure 5.2g). There were two reasons for this: access to country away from the main waters would have been infrequent and unpredictable, and foraging returns from country surrounding the main waters would also have been greatly reduced. Both factors would combine to force a pattern of frequent moves between main waters – a point-to-point pattern of subsistence (Figure 5.2f )– rather than seasonal dispersal across the backcountry. In effect, patterns of residential mobility would be tethered to a series of restricted localities where water was available year round. Under such conditions, a hunter-gatherer group would need an annual foraging range with a grid of reliable watering points, implying that territories would also need to be commensurately larger (Figure 5.4). Many of the changes evident in the archaeological site records (see the following discussion) can be read in this way: sites near focal waters are visited more frequently, and visits to outlying sites cease altogether as foraging range contracts. 124 The Archaeology of Australia’s Deserts

Figure 5.5. The impact of the last glacial maximum in the arid zone. This is a summed probability plot of radiocarbon (14C) dates from archaeological sites across the arid zone. Data are calibrated 14C ages between 8 ka and 35 ka (N = 155 dates), with a 1 ka moving average and taphonomic correction (following methods used by Surovell et al. 2009). Samples from the Murray–Darling basin are excluded. This plot also suggests that the Antarctic cold reversal (ACR) (14.5–12.5 ka) had an impact, although this is harder to see in individual site records. (Source: AustArch1 database, courtesy A. Williams)

THE ARCHAEOLOGICAL RECORD 30–12 KA Since the debates of the 1980s, the number of late Pleistocene sites known for this period from Australian deserts and drylands has grown from just a handful to more than ﬁfty (Table 5.2). This allows a more detailed assessment of the impact of the LGM than was possible 30 years ago. The impact of the LGM is clear in time-series records (Figure 5.5), but there is little direct evidence for the abandonment of entire regions, except perhaps along the Pleistocene littoral or in the sandy deserts. The emerging picture is of resilient, low-density, hunter-gatherer populations that made a series of adjustments in land use, territory and mobility to conditions during the LGM. There was local reorganisation of settlement in the uplands: some sites were abandoned, others were occupied for the ﬁrst time (Marwick 2002a). Sites associated with rivers and lakes show abandonment of parts of the drainage system and a shift to channels that remained hydrologically active (Hope 1993). Sites in well- watered gorges show a reduction in site catchments (Hiscock 1988a;Marwick 2002a), indicating more focal use of local ‘oases’. Sites in sandhill country on the periphery of major ranges register changes towards greater residential mobility (Veth 2005a;Smith2006), with small groups moving around a network of small waters in a point-to-point rather than a home-base pattern of land use (Figure 5.2f–g) (Smith 1989b; Smith, Fankhauser and Jercher 1998). Islands in the Interior: Last Glacial Aridity and Its Aftermath 125 table 5.2. Archaeological sites in Australian deserts and drylands dating 30–12 ka

Site and No. region Setting datesa Occupation sequence Source

N desert margins Carpenters Gap 1 Rockshelter 10+ Occupation from 48 ka–present. O’Connor Declining use after 31.5 ka 1995; and only transient use of site McConnell 24–19 ka. Main occupation is 1998; Wallis <12 ka. 2001 Carpenters Gap 3 Rockshelter 10+ Unpublished but preliminary O’Connor and work suggests peak in Veth 2006 occupation around 22–21 ka. Riwi Cave 6 Basal occupation unit 44.8–33.9 Balme 2000 ka, disconformably overlain by Holocene occupation unit c. 6.8 ka. Colless Creek Cave 12 Undated pre-LGM occupation Hiscock 1988a; unit. Main use 20.7–16.6 ka Magee and but occupation continues to Hughes 1982 <3.4 ka. Louie Creek Cave 3 Occupied 22.2 ka–<1.0 ka. Hiscock 1988a Main period of use 21–18 ka. OLH midden Shell 3 Single phase site at 12.5 ka. Slack 2007 midden GRE8 Rockshelter 4 Basal occupation unit 38.9–30.6 Slack 2007 ka separated from upper occupation unit 16.9–1.5 ka by stratigraphic disconformity. Arid west coast Jansz Rockshelter 8 Basal occupation layer 40–36 ka. Pryzwolnik Occupation unit 12.1–10.8 2002 ka. Late Holocene occupation unit 1.8 ka–present. Mandu Mandu Rockshelter 9 Basal occupation layer 39–23.5 Morse 1993a ka. Increasing use at 5.8 ka associated with mangrove molluscs. Late Holocene occupation unit 2.5 ka–present. C99 Rockshelter 14 Basal occupation layer dating Pryzwolnik 38–32 to 24.7 ka. Early 2002 Holocene occupation unit 7.6–8.4 ka. Late Holocene occupation 1.9 ka–present. (continued) 126 The Archaeology of Australia’s Deserts table 5.2 (continued)

Site and No. region Setting datesa Occupation sequence Source

Pilgonaman Rockshelter 9 Pieces of transported marine Morse 1993a shell (or reworked older shell) at 36 ka and 20.3 ka. Main occupation 13.9–11.0 ka. Late Holocene occupation 0.5 ka–present. Yardie Well Rockshelter 3 Occupation 11.6 ka–late Morse 1993a Holocene. Main occupation 5.8 ka–present. Noala Rockshelter 4 Isolated visit at 31.4 ka. Veth et al. 2007 Sparse occupation 13.9–9.3 ka. Gum Tree Valley Manuport 1 Transported Syrinx aruanus Lorblanchet 1992 rock engravings shell. Reworked shell (?) or isolated visit at 22.1 ka. Silver Dollar Sandridge 14 Pleistocene occupation unit Bowdler 1999 34.5–21.9 ka. Overlain by shell midden 7.7–7.4 ka. Pilbara (inland) Djadjiling Rockshelter 14 Discrete basal occupation Law, Cropper and 40.6–40.1 ka. Occupation Petchey 2010 30.4–<2.9 ka. Major decline in artefact density between 24–18 ka, which does not recover until 7 ka. Juukan 1 Rockshelter 3 Fleeting basal occupation Slack, Fillios and 37.4–31.3 ka. Main phase Fullagar 2009 of use in (?) late Holocene. Juukan 2 Rockshelter 3 Basal occupation 24.0–19.3 Slack, Fillios and ka. A late Pleistocene Fullagar 2009 deposit with a veneer of late Holocene occupation. Cleft Rockshelter Rockshelter 5 Intermittent visits from Salmon 1998; >15.2–1.0 ka. Marwick 2002a Mesa J J24 Rockshelter 3 Occupation layers at 28.4 Hughes and and 4.4 ka interleaved Quartermaine with sparser occupation. 1992; Hughes, Quartermain and Harris 2011 Islands in the Interior: Last Glacial Aridity and Its Aftermath 127

Site and No. region Setting datesa Occupation sequence Source

Marillana A Rockshelter 3 Infrequent visits >16.0–4.0 ka Marwick 2005 with main occupation 10–4 ka. Basal unit >16 ka but undated. Millys Cave Rockshelter 4 Sealed LGM occupation Marwick 2002b 22.5–17.4 ka. Upper occupation unit 0.7 ka–present. Yirra Rockshelter 9 Initial occupation during Veitch, Hook and LGM, continuing 23.0 Bradshaw 2005 ka–present. Peak artefact and charcoal densities 23.0–20.2 ka and 3.5–present. Manganese Gorge 2 Rockshelter 6 Sterile basal layer dated Quartermaine 21.5–7.4 ka. Most 1994 occupation 7.4–4.7 ka. Malea Rockshelter 5 Occupation from 24.3 ka to Edwards and present, with in-situ Murphy 2001, artefacts 24.3–18.4 ka and 2003 7.5 ka. Most intensive phase of use < 3.0 ka. Newman P0187 Rockshelter 4 Individual ashy hearths at c. Maynard 1980; 24.8, 11.4, 6.0 and 3.2 ka. Brown 1987; Decline in use 24.8–11.4. Comtesse 2003 Highest concentration of artefacts is 6.0–3.2 ka. Newman P2055 Rockshelter 10 Individual ashy lenses at 31.1, Troilett 1982; 17.1, 10.1 and 9.9 ka. Brown 1987; Chronological hiatus Comtesse 2003 between 31.1–17.1 ka. Main use of site is 7.1–3.8 ka. Recent use at 1.3 ka and during contact period. Western desert or Murchison region Walga Rock Rockshelter 5 Basal occupation sparse but Bordes et al. 1983 undated. Occupation unit at 11.5–7.9 ka separated by disconformity and eroded deposits from late Holocene occupation unit. Main use of site 1.0 ka. (continued) 128 The Archaeology of Australia’s Deserts table 5.2 (continued)

Site and No. region Setting datesa Occupation sequence Source

Bush Turkey 3 Rockshelter 8 Basal occupation c. 8.4 ka Veth et al. 2008 possibly 12–8 ka. Occupation unit 4.0–3.0 ka.Mainperiodofsiteuse c. 1 ka–present. Serpents Glen Rockshelter 4 Basal occupation >28.5 ka. O’Connor et al. (Katjarra) Sterile LGM layer. Upper 1998 occupation layer 5.4–present. Main use of site is <1 ka. Katampul Rockshelter 5 Possible brief use at 25.3 ka. O’Connor and Main use 5.2 ka–present. Veth 1996 Puntutjarpa Rockshelter 9 First use of site 11.9 ka. Gould 1977 Major occupation 7.6 ka. Upper occupation unit 0.4 ka–present. Central Australia Puritjarra Rockshelter 31+ Ongoing site use from 35 Smith 2006, ka–present with 2009b well-deﬁned phases of occupation at 35 ka, 24–8.3 ka and during the last millennium. Kulpi Mara Rockshelter 11 Pulses of sedimentation with Thorley 1998a, occupation units at 1998b; 34.2–29.1 ka, 15.3–10.9 Faulkner and ka and 5.5 ka–present. Thorley 2009; Thorley, Faulkner and Smith 2011 Watarrka NEP22 Sandplain 1 Use of locale at 13.3 ka. Ross and Smith Stone artefacts in 2007 sandsheet. Lake Eyre basin Cuckadoo 1 Rockshelter 14 Possibly occupied at 18.4 ka Davidson, Sutton but unconﬁrmed. Discrete and Gale 1993 occupation units at 14.7–14.3 ka, 6.2–3.3 ka and 1.7 ka–present. Kaponyee Springs 1 Hardpan 1 Artefacts in hardpan dated to Robins 1993 11.7 ka. Islands in the Interior: Last Glacial Aridity and Its Aftermath 129

Site and No. region Setting datesa Occupation sequence Source

Youlain Springs Hardpan 8+ Regular use of springs from 16.9 Robins 1998 to 4.9 ka. Lake Yantara Hearth 130.9 ka. Dury and Langford-Smith 1970 Madigan Gulf 4 Hearths and 1 Single phase site 22.1 ka with Magee and Miller open site hearths, stone artefacts and 1998;Magee emu eggshell. personal communication White Crossing Hearths 2 Two hearths in dune core, 13.9 Veth et al. 1990 and 13.7 ka. JSN Hearths and 5 Multiple use of open site in Smith, Williams open site duneﬁeld with dated hearths and Wasson at 17.5, 17.0, 12.3 and 2.5 ka. 1991 Flinders ranges Hookina Creek 1 Te r r a c e 3 Occupation at 26–27 ka, 16 ka G. Hamm and during late Holocene. personal communication Hawker Lagoon Lunette 10 Occupation units 18.0–>9.3 ka Lampert and and 5.8 ka–present. Main use Hughes 1988; of site is 18–9 ka and last 1 ka. Walshe 2005 Nullarbor Norina Rockshelter 3 Occupation from 10.7 Marun 1972 ka–present with main use 7.0–4.5 ka. Koonalda Cave Cave 12 Quarrying activity 26.3–16.6 ka. Wright 1971; Dated torches in art chamber Sharpe and at 25.5, 23.9 and 21.9 ka. Sharpe 1976 Allens Cave Doline 15+ Discrete occupation events Cane 1995 associated with hearths at 39.9 ka and 22.2 ka, but only sporadic use until before 15 ka. Occupation peak at 10.8 ka with declining levels of use after 3 ka. SE desert margins Willandra Lakes Lunettes 107+ Regional decline in number of Bowler 1998; sites 28–24 ka but shell Hope 1993; middens, hearths and fossil Johnston, Clark footprints dated to 24–19 ka. and White 1998 (continued) 130 The Archaeology of Australia’s Deserts table 5.2 (continued)

Site and No. region Setting datesa Occupation sequence Source

Lower Darling Lunettes 81+ Widespread use of lakes and Balme and Hope rivers 30–26 ka and from 1990; Balme 19 ka–present. Few sites 1995 26–22 ka. Cuddie Springs Ephemeral 18+ Main occupation unit 36–33 Field and Dodson lake ka, overlain by deﬂation 1999; Fillios, pavement of artefacts Field and possibly dating to 33–19 Charles 2010 ka. a Number of radiocarbon (14C), thermoluminescence (TL) and optically stimulated luminescence (OSL) ages available for site.

Before looking at the sequence of changes in each region, we need to consider the nature of the available evidence.

Interpreting Site Histories and Stratigraphy Much of the evidence for regional abandonment relies on how breaks in sedimentation or occupation at individual sites are interpreted (Figure 5.6). Table 5.3 lists the main sites, ranked according to quality of the data. Archaeological debates highlight several issues that need to be taken into account. First, site abandonment need not imply regional abandonment. Where it is possible to compare multiple site sequences within one region, apparent gaps in occupation are often shown to be local adjustments in settlement rather than regional abandonment. Second, does a stratigraphic break necessarily indicate site abandonment? Aus- tralian archaeologists generally assume that sedimentation at rockshelter sites is coupled with occupation so that a break in deposition is also a gap in occupation (O’Connor, Veth and Barham 1999). This may be so where a rockshelter is small and confined – maximising the impact of anthropogenic disturbance to walls and the effects of humidity or fires – or where the bedrock is very friable (Hughes 1977). But it should not be extrapolated as a general premise. Sediment fluxes in rockshelters are usually more complex than this (Farrand 2001; Ward and Larcombe 2003). The few detailed studies of sedimentary history at arid-zone caves and rockshelters (Frank 1971; Magee and Hughes 1982;Morse1993a;Cane1995;Marwick2002a, 2005;Smith2009b) all show significant external input of aeolian, colluvial or fluvial sediments. These components of the sediment flux are largely independent of occupation and, Islands in the Interior: Last Glacial Aridity and Its Aftermath 131

Figure 5.6. Different types of stratigraphic or occupation hiatus (after O’Connor et al. 1999,ﬁgs.2, 3 and 5). (a) The case for abandonment of a site is strongest where a sterile layer is present, showing that although a sedimentary record of the relevant time interval exists, people had ceased using the site. (b) Some sites have stratigraphic disconformities, either as a result of the removal of sediments by erosion or minimal net accumulation of sediment over time. The argument that this also represents a break in occupation depends on whether or not archaeological material has been detected at the layer interface. (c) For other sites, the break is simply a chronological gap, in which radiocarbon (14C) dating has shown that poorly differentiated sediments conceal a major time break in deposition. (d) Shows the effect of changes in rate of sedimentation on partitioning of archaeological assemblages. (Adapted from Ward and Larcombe 2003:ﬁg.5) as the presence of a sterile layer at Serpents Glen shows, sedimentation at rockshelter sites can continue despite abandonment (O’Conner, Veth and Campbell 1998). A stratigraphic hiatus need represent nothing more than a period of unrecorded time. If we decouple sedimentation and site use, the layer interface becomes the focus of attention – and we need to determine whether any artefacts 132 The Archaeology of Australia’s Deserts table 5.3. Key archaeological sites where there is substantive evidence for (a) a stratigraphic or occupational hiatus during the last glacial maximum (LGM), or (b) for continuing occupation in this period. Ranked by quality of data (see text for explanation).

Site Type of hiatus Duration ka

Sites with an LGM hiatus Manganese Gorge 2 Sterile layer 21.5–10.1 Serpents Glen Sterile layer 28.5–5.4 Juukan 1 Gap in chronostratigraphy 31.3–? Newman P2055 Gap in chronostratigraphy 31.1–17.1 Jansz Stratigraphic disconformity 35.0–12.1 Riwi Stratigraphic disconformity 33.9–?6.8 Mandu Mandu Stratigraphic disconformity 23.5–5.8 GRE8 Stratigraphic disconformity 30.6–16.9 Kulpi Mara Stratigraphic disconformity 29.1–15.3 Sites with LGM occupation Willandra Lakes Middens, hearths and footprints 24.0–19.0 Lower Darling Middens and hearths 22.5–19.0 Koonalda Cave Hearths, torches, wall markings and stone arrangements 26.3–16.6 Allens Cave Hearths and peak in artefact density 22.2 Millys Cave Sealed occupation unit 22.5–17.4 Silver Dollar Discrete occupation unit 34.5–21.9 Louie Creek Peak in artefact, bone and shell density 22.2–18.0 Colless Creek Peak in artefact, bone and shell density 20.6–16.6 Puritjarra Peak in artefact density 24.0 Yirra Peak in artefact density 23.0–20.2 Juukan 2 Peak in artefact density 24.0–19.3 Malea Peak in artefact density 24.3–18.4

accumulated during the period represented by the ‘break’. This could take several forms: (a) a lag of artefacts; (b) a palimpsest of different occupations at a layer interface; or (c) a zone of enrichment immediately below the interface, where artefacts have been mixed or trodden into the top of a palaeosurface (Figure 5.6d). However, few sites have been excavated with sufﬁcient precision to isolate material relating to interfaces between major stratigraphic units. On the opposite side of the balance sheet, evidence for occupation during the LGM is strongest where hearths, shell middens or other features have been directly dated, or where there are sealed occupation units or discrete layers dated to the LGM. More problematic are sites where artefacts are distributed within a continuous sedimentary column, dated by interpolation from radiocarbon (14C) ages. Here, the best evidence is from sites with a large series of radiocarbon samples that provide the basis for a high-resolution age- depth curve and internal cross-checks on consistency in dating. The weakest Islands in the Interior: Last Glacial Aridity and Its Aftermath 133 evidence is from sites with only a handful of 14C ages and where artefact discard rates are very low. All of these issues can be resolved with larger samples of dated sites and by comparing multiple site sequences within a region.

The Desert Uplands central australia Although the Central Australian ranges form the largest and richest of the biological refugia in the Australian arid zone, only three late Pleistocene sites are known from this vast region – Kulpi Mara, Puritjarra and Watar- rka NEP22 (Figure 5.7). All are from the same area in the southwest of this region. At present, nothing is known of the Pleistocene archaeology of the Petermann–Mann–Musgrave range complex (the Anangu–Pitjantjatjara– Yankunytjatjara Lands), the eastern MacDonnell Ranges, or the Dulcie Range and Davenport–Murchison complex. Kulpi Mara is the most strategically located of the three sites, situated on the headwaters of the largest river system associated with these uplands. The rockshelter sits at the foot of a clifﬂine, on a high sandstone terrace overlooking a tributary of the Palmer River (Figure 5.8). Excavation revealed three discrete layers, each with traces of occupation (Thorley 1998a). The earliest unit (layer 3)datesto34.2–29.1 ka and is overlain by deposit dating from 15.3 ka to 10.9 ka, showing that the LGM is marked by a stratigraphic disconformity at this site. Because Kulpi Mara is part of a chain of narrow rockshelters formed by cavernous weathering at the base of a cliff, net sediment accumulation is determined by moisture regime (affecting rates of rock weathering and basal sapping) and by structural failure of the sandstone (creating a zone of block fall that retains sediments within the shelter) (Mabbutt 1977). It is unlikely, therefore, that net sedimentation rates and occupation are closely linked – or that the break in sedimentation directly registers a break in occupation. However, careful excavation did not show any concentration of artefacts on the surface of layer 3, so the site seems to have been genuinely abandoned during the LGM. To see why this may have been, we need to look more closely at the location of Kulpi Mara. The rockshelter is on the outer edge of a complex of ethnographic occupation and rock art sites associated with a large rain-fed rock hole. Local groups made use of such ephemeral waters as focal points for occupation whenever conditions allowed (Thorley 1998b). But with sharp declines in rainfall, these are the type of waters most likely to fail, leading to readjustment of settlement towards springs and waterholes lower in the catchment. The site chronology suggests that pulses of sedimentation and 134 The Archaeology of Australia’s Deserts

Figure 5.7. Western Central Australia showing the location of archaeological sites (black triangles), red ochre quarries or chalcedony quarries (open triangles), and the regional topography of longitudinal dunes, sandstone ridges and saltlakes. Open squares show the network of key springs and wells likely to have been available during the last glacial maximum.

phases of occupation coincide because both responded to changes in local moisture regime. A different sort of local reorganisation of settlement is shown at Purit- jarra, out in sandhill country on the margins of the central ranges, where a hunter-gatherer population persisted throughout the LGM (Smith 1989b). Islands in the Interior: Last Glacial Aridity and Its Aftermath 135

Figure 5.8. Interpretative stratigraphic section, west face of the Kulpi Mara excavations. This shows a veneer of late Holocene deposits made up of hearths (dark grey), pits and charcoal-rich sediments (grey), capping late Pleistocene units dating to 34–29 ka and 15–11 ka. Layer III contains a series of diffuse hearths (light grey). Inset:the profile of the cliffline a few metres on either side of the excavations, showing cavernous weathering at the foot of an escarpment. (Source: Thorley et al. 2011:fig.1)

This is the best-documented excavation in the interior, combining a robust age model (Smith, Prescott and Head 1997; Smith et al. 2001; Prescott, Williams and Hunt 2007) with a comparatively large excavation, sediment and thin- section analyses, plotting of individual artefacts in the late Pleistocene levels (Smith 2006) and reconstruction of local palaeoecology (Smith 2009b). The sequence shows a series of phased adjustments to conditions over the period from 35 ka to 15 ka. In the first phase, the pattern initiated at 35 ka of small shifting camps by highly mobile groups continued until about 24 ka. These people brought with them red ochre from the Karrku mine, situated 125 km northwest of Puritjarra across a formidable dunefield, white chert from the Puli Tjulkura quarry (see Figure 5.7) and used notched implements, steep-edged scrapers and finely made retouched convex scrapers. The next phase, centred on 24 ka (17980 ± 160 BP ANU 6918, redated using ABOX AMS14Cas20,240 ± 280 BP ANUA10010 – Smith et al. 2001), 136 The Archaeology of Australia’s Deserts

coincides with phytolith evidence for the most intense period of aridity in this area (Bowdery 1998;Smith2009b). This shows a discrete peak in number of artefacts and greater reliance on local sandstone for making tools. People still used a wide territory but now combined this with extended visits to the rockshelter. The final phase, from 24–22 ka to 8.3 ka, continues the trend towards more extended site visits and larger encampments. As arid conditions ameliorated, there was a shift in assemblage facies towards greater emphasis on large flake tools (mainly steep-edged scrapers, notched implements and amorphous retouched tools) (Figure 5.9), probably recording more focus on repair and replacement of wooden implements at this site. By 16–15 ka (13,000–12,000 BP), camps at Puritjarra were larger, and occupation debris extended across most parts of the shelter floor, leaving a band of large implements that can be traced across the site, as well as pits, hearths and grindstones. A concentration of ochre near the rockshelter wall suggests that rock paintings were being produced at this time, although none of this pigment art survives today. Geochemical sourcing of the ochre shows greater use of both local sources and ochre from Ulpunyali, the closest of the major ochre quarries (50 km south east of Puritjarra) (Smith, Fankhauser and Jercher 1998). This suggests a switch to a more geographically circumscribed population using local range and dune country. Visits must still have been relatively short-lived because there was no attempt to clear large debris from occupation surfaces and heavily trafficked areas, but the groups using the shelter were larger and appear to have been moving over a more limited territory. The data suggest that Puritjarra was a repeatedly visited locale within a late Pleistocene territory centred in sandhill country west of the main ranges. Its position on the western edge of a network of small semipermanent springs and wells, spaced about 25–50 km apart, means that late Pleistocene groups would have had the ability to range over a substantial area, even during the LGM. The presence of exotic ochre and stone confirms that the people using Puritjarra were visiting some of these areas. The overall picture is of a much dispersed pattern of land use from 35 ka to 24 ka, with small mobile groups and focal use of sites near strategic waters, very much as we would expect in a point-to-point strategy. This pattern is exaggerated during the intense aridity at 24 ka, with more frequent cycling through sites near the remaining waters. The relaxation of arid conditions saw a shift to a more territorially based pattern of land use, with rescheduling of maintenance activities and ‘gearing up’ at sites near permanent water, such as nearby Murantji rock hole. Despite these changes, levels of site use remained much lower than anything recorded ethnographically – and residential mobility appears to have been much higher. Hayden, who studied stone tool use among Pintupi people in the area immediately west of the Cleland Hills in 1971, calculated that a family would use around 150 ± 50 retouched artefacts per year and utilise about the Islands in the Interior: Last Glacial Aridity and Its Aftermath 137

Figure 5.9. Large flake implements from late Pleistocene levels of Puritjarra rockshelter, showing steep-edged scrapers (1–4), notched implements (6 and 10), dentate implements or saws (9 and 11), and amorphous retouched artefacts (5, 7 and 8). All are from unit 2a, except N13/20–1 (6) (unit 2b). (Source:Smith2006:fig.16) same number of unmodified flakes as tools (Hayden 1977a: 182; 1979: 165–6). On these estimates, the assemblage from unit 1a (late Holocene) at Puritjarra represents the long-term equivalent of use of the rockshelter by a nuclear family for about 3–6 weeks each year, whereas earlier levels, dating between 35 ka and 138 The Archaeology of Australia’s Deserts table 5.4. Changes in intensity of occupation at Puritjarra

A: Quantitative indicators Emu Pits and Analytical Lithics Cores Grindstones Ochre Charcoal eggshell Bone hearths unit no./10 kga no. no. no. g/10 kga wt. g wt. g no.

1a 48.63449335.21 12.662.610 1b 27.01411130.28 0.950.93 1c 14.0118 290.01 – 2.03 2a 2.2394 630.11 ––4 2b–d <0.131 58<0.01 ––– a Relative to weight of excavated sediment. B: Estimated artefact discard rates Analytical Estimated period Artefactsb Area excavatedc Estimated discard ratesd unit kyr cal BP no./100 yr % no./100 yr

1a 0–0.8 903 3.5 25800 1b 0.8–4.0 201 3.3 6090 1c 4.0–8.3593.3 1790 2a 8.3–24 17 3.0 570 2b–d 24–36 2 4.545 b Excavated trenches only. c Percentage of ﬂoor area available during each period. d Interpolated for entire deposit. C: Other indices Reduction Breakage rates Retouched Cores and Flakes Flakes Tool inventory Exotic materials Analytical ﬂakes tools (L)f (T)g unit (GIRe) % % % Richnessh Diversityi Stonej Ochresk

1a 0.42 25 9 14 5.63 1.95 0.64 0.70 1b 0.50 31 8 22 4.80 1.74 0.67 0.62 1c 0.39 27 6 17 3.69 1.04 0.26 0.41 2a 0.32 18 6 2 4.19 1.32 0.26 0.59 2b–d – 22 9 1 1.10 1.39 0.23 0.93 e Geometric index of reduction (GIR). f Longitudinal breaks. g Transverse snaps or breaks. h Following Odum (1971:144), richness is number of tool types/log (sample size). i Diversity is the Shannon-Weaver diversity index (H). j Ratio by number of pieces of exotic to local stone. Exotic stone is stone from sources 50–100 km from Puritjarra. k Ratio by number of exotic ochre (from sources 50–100 km from Puritjarra) to other ochres. A wide range of indicators is commonly used to characterise the nature of occupation at a site: quantitative measures of occupation intensity including artefact discard rates; patterns of artefact breakage and recycling; indices of assemblage richness and diversity; flake reduction indices including GIR, and flake/core or flake/ tool ratios; and various proxy measures of residential mobility. Units 1a–1c are Holocene. Unit 2ais24–8 ka. Units 2b–2dare35–24 ka. Source: Adapted from Smith 2006:table17 Islands in the Interior: Last Glacial Aridity and Its Aftermath 139

Figure 5.10. The Pilbara, showing the Hamersley plateau (dark grey indicates land above 500 m asl), major river systems, and the Burrup-Dampier coast. Archaeological sites mentioned in the text: (1) Noala, (2) Skew Valley midden, (3)MesaJJ24,(4)the Juukan sites, (5) Manganese Gorge, (6)Yirra,(7) Bastion shelter (Packsaddle Ridge), (8) Millys Cave, (9) Cleft rockshelter, (10) Marillana, (11) Djadjiling, (12)Malea,(13) Newman P0187,(14)NewmanP2055.

8.3 ka (units 2a–d), represent even more ephemeral use of the shelter. Indices of assemblage richness and diversity show a similar pattern (Table 5.4). Values for unit 1a at Puritjarra are comparable to those for residential base camps near springs in the Simpson Desert (richness 5.44;diversity1.58 – Barton 2003: table 5). Site inventories in units 2b–d are conspicuously poorer, implying even greater residential mobility than ethnographic hunter-gatherer groups in one of the most marginal parts of the arid zone today. the inland pilbara The only inland desert region with a large number of sites spanning the period 30–12 ka is the Pilbara. This region is dominated by the Hamersley plateau, a relatively small, arid upland (about 45,000 km2), lying between the Fortescue and Ashburton Rivers, both of which drain towards the Pleistocene coast some 500 km away (Figure 5.10). The northern side of the Hamersley plateau forms a high escarpment, broken by deep gorges that drain the plateau. 140 The Archaeology of Australia’s Deserts

Figure 5.11. Temporaldistribution of stone artefacts at Djadjiling rockshelter, Hamer- sley plateau: (a) age-depth plot, (b) relative artefact density (estimated discard rate) shown as no./100 yr, (c) no. of artefacts per spit. (Source: Law, Cropper and Petchey 2010)

The deeper gorges are permanently shaded, with groundwater-fed springs and pools and relict ferns. Both rivers flow only intermittently, but the Fortescue has several permanent pools and the relict palm Livistona alfredi. Typically, the Pleistocene sites are small rockshelters, with mostly small excavated assemblages, few retouched implements and no grindstones, although they often do contain red ochre. Because the Hamersley is a major iron-ore province, much of this work has been driven by the imperatives of mitigation or salvage archaeology, so few sites have been studied in detail. Sites in the catchments of the major main gorge systems show a pattern of repeated – and sometimes increased – use during the LGM. A critical threshold in pattern of site use occurs around 24–22 ka. Djadjiling, the best-dated sequence in the Pilbara, shows that occupation continued strongly until 24 ka. Artefact discard rates then declined sharply, although use of the rockshelter continued into the Holocene (Figure 5.11). Other sites show the first traces of use during the latter part of the LGM. Millys Cave (22.5–17.4 ka), Yirra (23.0–20.2 ka) and Malea (24.3–18.4 ka) all have significant LGM occupation units, beginning around 24–23 ka (Edwards and Murphy 2001, 2003;Marwick 2002b; Veitch, Hook and Bradshaw 2005). Islands in the Interior: Last Glacial Aridity and Its Aftermath 141

Millys Cave is a small rockshelter (34 m2) near a spring on a side branch of the Weeli Wolli catchment. The late Pleistocene deposit, dated 22.5–17.4 ka, is sealed beneath a layer of rock fall. Detailed analysis of the excavated assemblage (N = 837) (Marwick 2002a, 2002b) shows that the highest artefact concentrations are in the LGM unit. But there is no evidence for intensive reduction of flaked stone or differences in the working between transported and local cherts. Few artefacts have been retouched or resharpened. Despite the unusually high discard rate at this site (8.2 artefacts/100 years/m2), these levels appear to reflect frequent visits by small mobile groups, with no evidence for extended stays or the range and diversity of tool classes that might indicate a residential base camp (Table 5.5). Marwick concludes that although these data confirm that the Hamersley plateau was a glacial refugium, Millys Cave itself was not ‘a site of refuge for a regional population’ (2002b: 31). The types of flaked stone found there provide some indication of the foraging territory of these groups. Marwick draws a contrast between local banded ironstone formation (BIF) cherts, available in the rockshelter, and cherts transported to the site. The high proportion of BIF chert (27.4 per cent of all chert artefacts) in the LGM unit (compared to the Holocene levels of the site) indicates a smaller site catchment with less use of the marginal Pilbara lowlands, although the retraction in territory cannot have been marked because nonlocal cherts still make up more than 70 per cent of chert artefacts during the LGM. Away from the springs and main gorge systems, most sites were not heavily used until the Holocene. On the central plateau, use of the large Juukan 1 rockshelter was very fleeting between 37.4 ka and 31.3 ka and is marked by isolated artefacts and some burnt bone. Nearby Juukan 2 was intermittently used 24.0–19.3 ka, with intact hearths, unmodified ironstone flakes and bone from large- and medium-sized macropods found in these levels (Slack, Fillios and Fullagar 2009). Slack et al. concluded that the Hamersley population at this time was ‘more residentially mobile’ than refuge models suggested. ‘People were not just retreating into gorges,’ they said, ‘a more complex use of the landscape, perhaps following local weather patterns and allowing access to the less drained areas occurred’ (2009: 38). Similarly, the two rockshelters excavated near Mt Newman (P0187 and P2055) on the inland margin of the plateau show only intermittent, light use (Maynard 1980;Troilett1982;Brown1987) – as one might expect if they served as an occasional hunting station or wet weather bivouac. This low level of use is emphasised by the preservation of intact hearths at both sites, despite the low rates of sediment accumulation that prevailed. P2055 has dated hearths at 31.1 and 17.1 ka, with little evidence of use between these events. P0187 has a cluster of dated hearths centred on 24 ka (20,740 ± 345 BP SUA 1041) with a sharp decline in artefact numbers in overlying levels. Reanalysis of the excavated assemblage by Comtesse (2003) reinforces the picture of short-lived, intermittent site visits around 25–24 ka, showing that these levels are dominated 142 The Archaeology of Australia’s Deserts table 5.5. Comparative data on the size and diversity of selected assemblages, dated to the last glacial maximum

Sample Site Tool Age Lithicsa areab area Discard classesd Site Layer/unit (ka) (n) (m2) (m2) ratec (n) Ochre Richnesse

Colless P46/9–10 21–18 723 0.25 84 96.43 N 1.4 Creek Louie BJ29/9–11 21–18 434 0.25 1400 62.01 N 0.4 Creek Millys Unit 323–17 837 2.0348.20 3 N 1.0 Cave Malea Unit d 24–18 264 2.0282.24 ––– Mandu Unit 232–20 691 3.0801.92 6 Y 2.5 Mandu Newman Spits 8–10 24 75 1.0201.00f 3 Y 2.1 P0187 Djadjiling Layer 324–18 81 4.5340.30 –N– Puritjarra Unit 2b–d 35–24 303 6.5 145g 0.42g 6 Y 2.8 Kulpi Layer 334–28 124 3.0650.70 4 Y 2.4 Mara Serpents Spits 34–43 29 32 1.0 160 0.49h 2 Y 2.0 Glen

Most desert sites have comparable levels of site use and assemblage richness. The few sites with much higher discard rates also have low assemblage diversity – indicating that these represent frequently visited task-specific sites. a Assemblage size is limited to artefacts from the 5–6 mm sieve fraction, where this is given. b Sample area is area analysed, which may not be the entire area excavated. c Discard rate is no. lithics/100yr/m2, to standardise this index for differences in sample area. d The number of tool classes gives an indication of the range of activities carried out at a site. Only a few categories are widely reported in site reports: cores, notched implements, steep-edged scrapers, amorphous retouched artefacts, pebble manuports or hammerstones, amorphous grindstones and unmodified flakes. e Richness is a formal index of assemblage diversity (no. tool types/log(sample size)) that reduces the effect of sample size. In this case, the sum includes the tool classes listed in d plus ochre. To retain a degree of comparability, I have focused on nonperishable materials (stone and ochre) and excluded artefacts on wood, shell or bone found at some sites. f The full age span of this unit is unknown, but because sediment accumulation at this site averages 250 yr/cm, I have calculated discard rates on the basis that spits 8–10 represent about 7.5 kyr, spanning the period from 28.5–21.5 ka. g Site area is 400 m2 but the rock-free floor area in this period was limited to 145 m2. Extrapolating a last glacial maximum (LGM) site total for this site gives 45 artefacts/100 yr. h The full age span of this unit is unknown. I have used the excavator’s assessment of its likely age span (from Veth 2005a: table 6.3).

by unmodified flakes with only a few lightly retouched implements. Artefact discard rates then remained low until 11.4 ka. The archaeological record shows that a regional population continued to use the Hamersley plateau throughout the LGM. The evidence suggests a spatial reorganisation of land use, beginning around 30 ka but most marked from 24 Islands in the Interior: Last Glacial Aridity and Its Aftermath 143 ka. From this time, rockshelter sites in the more marginal parts of the plateau, or on its periphery, were used more intermittently or abandoned – often with long gaps between episodes of use. Sites in focal areas for occupation, around the major gorge systems, continued to be used, but with people cycling more frequently between sites. Some new shelters in these areas also came into service at 24–23 ka. None of these appears to have been occupied intensively. Instead, the archaeological record reveals a pattern of high residential mobility, increasingly tethered to a series of restricted localities where water was available year round, with only intermittent use of the backcountry. This is more in accord with a point-to-point pattern of land use (see Figures 5.2c and f ) than of intensive occupation of resource-rich oases (see Figures 5.2aandg). The ameliorating climate after 18 ka seems to have relaxed the constraints on land use. There is evidence for first use of several rockshelters (e.g., Marilana Aat16 ka; Cleft rockshelter at 15.2 ka), as well as increased visits to sites on the margins of the plateau (P2055 at 17.1 ka). With the end of the LGM, the use of focal sites such as Millys Cave and Yirra also declined, suggesting a further relaxation in the spatial footprint of these hunter-gatherer systems. And, although the response to the end of the LGM climate was rapid, consolidation of the regional population may have been delayed because many sites do not show significant occupation until 8.0–6.0 ka. other desert uplands The record in other desert uplands is patchy. Cuckadoo rockshelter, in the Selwyn Range on the northern rim of the Lake Eyre basin, may have been used at 18.4 ka, although most evidence is later, with discrete occupation units at 14.7–14.3 ka and 6.2–3.3 ka (Davidson, Sutton and Gale 1993). In the northern Flinders Ranges, artefacts occur in thick silt terraces along Hookina Creek and at Arkaba, near strategic springs, but there was little use of these localities between 27 ka and 16 ka (G. Hamm, personal communication). Although there were extensive Pleistocene wetlands in these valleys – associated with the valley fills (Glasby et al. 2010) – the combination of intense cold, slope instability and high water tables may have made these localities unattractive to foragers. In a more sheltered location, Hawker Lagoon, an open site on a lunette near an ephemeral lake, was not occupied until 18 ka (14,770 ± 270 BP SUA 2131) (Lampert and Hughes 1988). In the southern Kimberley, there was only transient use of Carpenters Gap 1 between 24 ka and 19 ka (Wallis 2001), although the number of burnt chenopod seeds in these levels may reflect some form of dryland plant processing (discussed in Chapter 6) (McConnell 1998). This site has rich occupation deposits – with stone artefacts, bone, eggshell and plant remains – dating to 31.5 ka, but use of the rockshelter declined after this time and did not recover until about 11 ka. Another site, Carpenters Gap 3 in Windjana Gorge, is closer to permanent freshwater and shows a peak in radiocarbon ages and occupation debris at 22–21 ka (O’Connor and 144 The Archaeology of Australia’s Deserts

Veth 2006: 35). This again illustrates some of the differences between regional histories and individual site records, showing shifts in the spatial structure of Pleistocene land use within refuges rather than a straightforward dichotomy between continuity and abandonment.

The Arid Core The most marginal part of the desert for people is made up of two quite different regions. The central part of the Lake Eyre basin forms the most arid part of the continent today. With a median annual rainfall of 100 mm or less, it is close to being hyperarid. This is a region of saltlakes and pans, stony desert and dunefields, interlaced with ephemeral river channels. The second region, in the western part of the continent, is made up of the major sandy deserts. The crucial difference between the two is the presence in the former of a vast network of ephemeral rivers and channels, making up a coordinated inland drainage system centred on Lake Eyre. Whereas the Western Desert is dependent on local rainfall, the Lake Eyre basin receives floodwaters from northern and eastern Australia, rejuvenating and reactivating riparian zones along the channels. At present, the Pleistocene archaeology of both regions is poorly understood: both are remote and difficult to access, rockshelters are uncommon, and many of the open sites are in geomorphically or taphonomically complex settings.

thelakeeyrebasin In the Lake Eyre basin, the prevalence of active dunes and floodplains limits the extent of Pleistocene exposures despite considerable archaeological reconnaissance (Hughes and Lampert 1980; Veth, Hamm and Lampert 1990;Robins 1998; Williams 1998; Barton 2003). Most of the evidence for Pleistocene occupation is from isolated hearths exposed in dune cores or on pans, or from artefacts cemented into hardpans. Although there are extensive, unstratified scatters of stone artefacts, artefact typology provides little guide to the age or cultural phase of this material. Detailed investigation of the complexes of mound springs – fed by the Great Artesian Basin – on the southwestern side of Lake Eyre has not identified any Pleistocene sites associated with these spring vents (Florek 1993). Most dated sites reflect use of the region after 18 ka, with a cluster of dated hearths between 17.6 ka and 12.3 ka, and with use of Youlain and Kaponyee Springs from 16.9 ka to 11.7 ka (see Table 5.2). Along the western edge of Lake Frome, several open sites were occupied between 10 ka and 5 ka (Lampert and Hughes 1987). The principal sequence is from Balcoracana Creek, where the main occupation horizon is marked by cores and flakes eroding from carbonate-rich sands TL dated to 9,900 ± 2,200 years ago (Gardner et al. 1987). Islands in the Interior: Last Glacial Aridity and Its Aftermath 145

Two sites warrant special mention. The first, MG4 (Madigan Gulf Site 4), shows that even at the height of the LGM, people were able to move into the most extreme part of the arid zone. This site was discovered on the northeast shore of Madigan Gulf during geomorphic mapping of Lake Eyre. Although yet to be fully described, MG4 includes two small, discrete fireplaces – including one dating to 22.1 ka (18,450 ± 165 BP, AA12615)– stone artefacts and emu eggshell, associated with sediments laid down when Lake Eyre was a saline, groundwater-controlled playa (Magee and Miller 1998: 312). The site is close to the point where Cooper Creek debouches into Lake Eyre, suggesting that it may represent a temporary downstream extension of foraging range during a short-lived flood event. Its real significance is that it indicates that some people continued to occupy the middle reaches of these river systems during the LGM – possibly upstream of Innamincka, 300 km away, where the bedrock-confined section of the Cooper has cut deep permanent waterholes (Hughes and Lampert 1980). The second site ( JSN) is an open site in the Strzelecki dunefield, 40 km west of Strzelecki Creek. The history of JSN indicates that a pattern of seasonal use of the dunefield was established almost as soon as the climate began to ameliorate after 18 ka. The site consists of a cluster of at least nine hearths on an interdune pan, associated with a scatter of baked-clay heat retainers, silcrete artefacts, grindstones and some fragments of freshwater mussel shell (Velesunio sp.). Radiocarbon dating shows that it was used on multiple occasions (at 17.5, 17.0, 12.3 and 2.5 ka) (Wasson 1983: 102–4; Smith, Williams and Wasson 1991). The oldest of the Pleistocene hearths is a large earth oven ( JSN/W3, 14,400 ± 400 BP ANU7196) that was reused on more than one occasion, possibly to cook Dromaius (emu). Another Pleistocene hearth, dated at 17 ka, contained fragments of Velesunio shell. People appear to have moved west from the Innamincka area along the Cooper floodplains, then south along interdune corridors to reach JSN, bringing mussel-shell implements, silcrete and other stone from the Innamincka area 115 km away. Large pieces of charcoal from the W3 hearth have been identified as acacia and Eucalyptus terminalis (now Corymbia terminalis), confirming that sandhill trees and shrubs were locally abundant by 17.5 ka. Both MG4 and JSN illustrate the importance of the Cooper Creek channel and flood-out zones in underwriting late Pleistocene use of this region, and the ability of local hunters to rapidly disperse across adjacent dunefields whenever conditions allowed. the problem of the sandy deserts The major continental dunefields are the most archaeologically intractable landscapes within the arid zone. They are also the critical regions for testing whether the LGM led to abandonment of marginal habitats. The key sequences are from Serpents Glen (O’Connor et al. 1998)andBushTurkey3 in the Little Sandy Desert (Veth et al. 2008). At present, nothing is known of the 146 The Archaeology of Australia’s Deserts

Pleistocene archaeology of the Simpson Desert, the Tanami or the Great Victoria Desert. With only one sequence extending back to 30 ka,itisdifficulttogaugethe impact of the LGM on settlement in the sandy deserts. Serpents Glen, on the southern edge of the Little Sandy Desert, has a sterile layer dating 28.5–5.4 ka, clearly marking abandonment of the site at the beginning of the LGM (Figure 5.12). Does this indicate that the sandy deserts were abandoned? O’Connor and Veth argue that it does: ‘the onset of the LGM appears to have effectively brought a stop to the exploitation of desert lowlands’ (2006: 43). But how strong is the evidence for this? On the one hand, Serpents Glen is in a strategic location and likely to record shifts in regional occupation. The site is a large rockshelter, near the mouth of one of the drainage lines that dissect the western side of the Carnarvon Ranges, one of many small sandstone or quartzite ranges that dot the sandy deserts in the western half of the continent. Local rock holes made it a focal area for occupation, and the presence of more than 100 basal grindstones on sandy flats nearby shows that large ethnographic gatherings of people took place here (Veth and O’Connor 1996). On the other hand, Serpents Glen was not reoccupied until 5.4 ka, which means that the occupation hiatus extends into a time period where there is good evidence for use of other sites in this region. For instance, Bush Turkey 3 in the Calvert Ranges, in the Little Sandy Desert, was in use by 8.4 ka and possibly by 12 ka (Veth et al. 2008). Looking across the arid core as a whole, we can see that hunter-gatherer use of sandridge deserts continued throughout the LGM wherever these regions adjoined extensive range systems (as at Puritjarra) or major river systems (in the case of MG4). In other areas, the evidence is less clear cut. Widespread abandonment of the immense continental dunefields in the western half of the interior remains a possibility, especially because there is still no direct evidence of their use during the LGM. But until there is comparative data from other sites, Serpents Glen may represent nothing more than the pattern of retraction and reorganisation seen in the Hamersley region. Resettlement of these regions after the LGM poses similar problems. Some sites in these desert lowlands were reoccupied by 12–8 ka (e.g., Walga Rock, 11.5 ka; Bush Turkey 3, 8.4 ka; Puntutjarpa, 11.9 ka), but use of both Serpents Glen and Katampul was delayed until about 5 ka.

The Shifting Margins We could expect LGM aridity to have been most dislocating on the margins of the desert – in those areas affected by a rapidly expanding arid zone and where environmental gradients are steep, or in areas where falling sea level disrupted coastal economies. The following case studies show a diversity of Islands in the Interior: Last Glacial Aridity and Its Aftermath 147

Figure 5.12. Temporal distribution of stone artefacts at Serpents Glen rockshelter, Little Sandy Desert, showing the break in site use after 28.5 ka. (Data from O’Connor et al. 1998) social and economic responses to increasing aridity, with the most dramatic changes occurring late in the LGM. the carpentarian gorge systems The coastal plains, surrounding the Gulf of Carpentaria, are rimmed by a dissected plateau that ends abruptly in cliffs rising 70–120 mabovetheland to the north. This escarpment is broken by a series of entrenched gorge systems containing perennial pools and springs. During the LGM, the area was incorporated into an expanding arid zone, while the Gulf itself was transformed by the fall in sea level into a vast lake (Lake Carpentaria). Today, these gorges are rich oasis-type ecosystems with vine thicket, Livistona palms, Nymphaea waterlilies, ﬁsh, turtle, Velesunio mussels, freshwater crustaceans and crocodiles. This also appears to have been the case throughout the late Pleistocene. An important biological refugium is formed by sandstone gorges along the western part of this escarpment (see Table 5.1). Most archaeological work, however, has focused on limestone and dolomite caves in the east at Colless Creek Cave, Louie Creek Cave and GRE8. The best studied of these sites, Colless Creek, is a medium-sized cave (84 m2) formed in dolomite in Lawn Hill gorge. Fine sediments from erosion of the soil mantle on the plateau have entered the cave via ﬁssures in the dolomite. Within 148 The Archaeology of Australia’s Deserts

the cave, these form two main stratigraphic units (A and B) separated by an erosional disconformity and a lag of gravel and patinated artefacts (Magee and Hughes 1982). Unit B is undated but may date to more than 30 ka, given the deep weathering and in situ pedogenic modification of these sediments. The overlying layer, Unit A, has a basal age of 20.7 ka (17,290 ± 470,ANU2331), which means that this site records events only at the close of the LGM. In fact, none of the excavated sites preserves a record showing the onset of LGM aridity around 30–24 ka. Although Hiscock (1988a)excavated3.5 m2, most analysis has focused on Square P46, which provides the most complete sequence. His analysis shows that the cave was more frequently used at the close of the LGM than during later periods and that local groups predominantly used the gorge systems, avoiding the northern plains. The most marked changes occur in levels dating to 18.0–16.7 ka. These have the highest concentrations of stone artefacts, burnt bone and shell, and most evidence for incidental burning of stone artefacts and for trampling damage to artefacts. The absence of stone from sources on the plain, 30 km north of the cave, and the decline in use of chert from quarries on the plateau, suggest a geographic contraction in territory used by these groups. By 18 ka, people visiting Colless Creek Cave were limiting their foraging to the gorge systems and had largely abandoned regular use of the plains and plateau. The faunal remains record the resulting pressure on local resources: there is a shift at 18 ka from freshwater turtle, snakes and macropods to more fish and shellfish, reflecting overexploitation of gorge biota. The picture is of more intensive use of these gorge systems, by groups more or less restricted to these oasis-type environments. However, none of the excavated sites appears to be a base camp (in the sense that Binford [1980] uses the term: a residential base and hub of subsistence activities, where most processing, manufacturing and repair activity takes place). The narrow site inventories and limited evidence for retooling or resharpening of implements suggest that these were task-specific sites (see Table 5.5) that were visited more frequently as use of the gorge systems intensified. Although the Colless Creek study (Hiscock 1988a) has been influential, it was based on analysis of 0.25 m2 of floor area in a site where the crucial late Pleistocene deposits are only 25–30 cm thick. The major trends are replicated at a second site – Louie Creek Cave – where the most intensive period of use is dated at 20.7–17.9 ka (Hiscock 1988a). Independent work at a third site – GRE8 (Slack 2007) – shows a sedimentary sequence very similar to Colless Creek, with a lower unit of oxidised red-brown sediments, dating from 38– 9–30.6 ka, overlain by a younger layer that began to accumulate at 16.9 ka. Although Slack (2007) claims that this is a continuous sedimentary sequence, the difference in weathering of the two units makes this unlikely. His data indicate that there is a major disconformity between 30.6 and 16.7 ka and that LGM levels are missing from GRE8. Islands in the Interior: Last Glacial Aridity and Its Aftermath 149

Although local groups responded to environmental stress 21–18 ka by reducing their use of the plains and focusing almost exclusively on the riparian resources of the gorges, individual gorge systems are too small to be self- supporting in demographic terms. The Lawn Hill Gorge comprises only 17.5 km2, sufficient to support a local population of 20–30 people (Hiscock 1988a: 246). A viable regional population (∼300–500 persons) would clearly need a network of such oases. The geography of the Carpentarian region shows that there are more than ten gorge systems along a 350-km stretch of the escarpment, from Robinson River in the west to Gunpowder Creek in the east. The Lawn Hill group would have been embedded in an extensive regional system that linked local groups, scattered in gorges over a 400-km range. Although these groups were economically tethered to productive riparian zones and gallery forest in gorges, they are unlikely to have had restricted social networks. the arid west coast Neither the fall in sea level nor the expansion of the margins of the desert greatly affected the Northwest Cape area. The continental shelf is so steep there that the advance of the coastline was only modest. Even at the maximum low stand, most sites were within 4–12 km of the Pleistocene coast. Instead, it was increasing continental aridity around 24–18 ka that posed most difficulty for local hunter-gatherer groups. This is reflected in the regional archaeology, where a mixed terrestrial and marine subsistence economy successfully continued throughout the early part of the LGM, until it collapsed around 24–23 ka (Morse 1993a; Przywolnik 2002;Vethetal.2007). Apart from Noala Cave, situated on the wide continental shelf to the north, the only site to be abandoned at the onset of the LGM was Jansz rockshelter. People continued to occupy the region and exploit marine and littoral resources until 24.7 ka at C99 and 23.5 ka at Mandu Mandu. Up to this time, the main changes were economic. Steep emergent coasts involve a shift from sandy to rocky-shore ecosystems. In this context, from about 26 ka, Mandu Mandu shows increased harvesting of reef or rocky- shore molluscs, fish, crab and sea urchin, together with an intensification of occupation and use of marine resources in the period just before the site was abandoned. Isolated finds of transported shell artefacts at 22.1 ka (Gum Tree Valley) and 20.3 ka (Pilgonaman) suggest that some human presence continued on this coast late into the LGM, but because these items may have been scavenged or reworked from older deposits, their significance is unclear. No site has occupation that continues uninterrupted throughout the period 30–18 ka. This pattern of declining use of the region is more consistent with increasing aridity than with changes in position of the Pleistocene coastline. The direct cause may have been failure of local springs or waterholes and the salinisation of groundwater. This also explains why the Silver Dollar site – then some 100 km 150 The Archaeology of Australia’s Deserts

inland – was also abandoned at 21.9 ka. O’Connor (1999a: 122) made a similar point about sites on the Kimberley coast, which were abandoned close to the period of maximum aridity, several thousand years after shifts in the Pleistocene coastline had left them well inland. The peak in occupation at Mandu Mandu just before the site was abandoned at 23.5 ka (Morse 1993a: 184–90) warrants comment. Could this be a zone of enrichment created by Holocene treadage on a long-exposed Pleistocene surface? This seems unlikely because it continues a trend that began at 26 ka. Rather, it suggests that land use in the Mandu Mandu area was undergoing a transition just before the regional economy collapsed – with more frequent use of a focal rockshelter (the closest to the Pleistocene coast), greater use of marine resources to offset increasing aridity and the ﬁrst appearance of exotic ochre (possibly from inland sources), indicating shifts in regional networks. The major gap in regional occupation occurs between 20.3 ka and 13.9 ka, straddling the end of the LGM. Following this collapse, it seems to have taken some time to reconstitute the regional economy. Resettlement of the area followed reestablishment of summer monsoon rainfall about 14 ka (Wyrwoll and Miller 2001) and the creation, by 12.7 ka, of productive mangrove habitats along the transgressive coastline (Przywolnik 2002: 301). By 13.9–11.6 ka, a number of sites were in use: Noala Cave, Yardie Well, Pilgonaman shelter and Jansz. But substantial occupation of the west coast was delayed until 8.0–7.0 ka, when an increase in coastal shell middens coincides with a major expansion of mangroves along coasts in northern Australia.

the nullarbor On the Nullarbor, the rapid fall in sea level at 30 ka also had little effect on regional geography. Today, the Nullarbor coast is dominated by limestone sea cliffs 40–100 m high extending for 200 km around the Great Australian Bight. To the south of these is an immense submarine plain, up to 200 km wide. By 30 ka, most of this was already exposed as a broad coastal plain, and the fall in sea level from −85 m to the last glacial low at −135 m added little additional territory. Increasing aridity is reflected in regional pollen records, with a retraction of mallee eucalypt woodland and arid scrub to the southern part of the exposed coastal plain by 22 ka (Martin 1973). At this time, most of the area north of the cliffline formed a vast, treeless, karst plain dominated by chenopod steppe. The key excavated sites are Koonalda Cave (Wright 1971) and Allens Cave (Marun 1972;Cane1995). Koonalda Cave is a spectacular limestone sinkhole on the treeless plain, about 160 km inland from the Pleistocene coast (Figure 5.13). From the base of a collapsed doline, a deep cave leads down to a series of subterranean lakes more than 70 m below the plain and to chambers with exposed nodules of high-grade flint. Excavations at the limit of daylight show that a phase of flint quarrying within the cave dates to 26.3–16.6 ka (Wright 1971). Further into the Islands in the Interior: Last Glacial Aridity and Its Aftermath 151

Figure 5.13. The sinkhole at Koonalda Cave, surrounded by treeless chenopod steppe on the Nullarbor Plain. (Photograph courtesy of Robert Edwards) cave, beyond the reach of daylight, the soft limestone walls are decorated with finger-markings and incised lines (Maynard and Edwards 1971)–nowpartly buried by a massive rock fall – and dated by the remains of several wooden torches to 25.5, 23.9 and 21.9 ka (Wright 1971: 28; Sharpe and Sharpe 1976). The cave also contains several possible stone arrangements (Gallus 1971;Sharpe and Sharpe 1976). Closer to the coast, Allens Cave registers intermittent visits from hunting parties between 39.8 ka and 10.8 ka, without any apparent change associated with the onset of the LGM (Cane 1995). A discrete band of artefacts at 22.2 ka is associated with several small hearths and coincides with maximum aridity in this area. In earlier syntheses, the archaeological and palaeoenvironmental sequence on the Nullarbor seemed to exemplify the impact of eustatic changes: rainfall isohyets, vegetation zones and human occupation all responded to falling sea level and exposure of the coastal plain (Martin 1973). New data have broken this nexus. The arid maximum at 22–21 ka in the pollen record at Allens Cave must reflect increasing continental aridity because much of the coastal plain would already have been exposed by 30 ka. Similarly, an apparent cultural hiatus between 17,000 BP and 15,000 BP,identified by Marun (1972), has been redated to the early part of the LGM, around 35–25 ka (Cane 1995; Roberts et al. 1996). By 24–18 ka, both sites were on the treeless plain in an area where rainfall is estimated to have been 160–180 mm per annum (p.a.) (Martin 1973). This arid karst landscape would have represented one of the most difficult environments 152 The Archaeology of Australia’s Deserts

for hunter-gatherers on the continent. Both sites, however, indicate the survival of a regional population during the height of the LGM. The key to this is probably the belt of arid woodland along the Pleistocene littoral, where local rainfall would have been around 250 mm p.a. Koonalda Cave suggests something more than opportunistic forays onto the treeless plain or expeditions to obtain flint. The combination of flint quarrying, stone arrangements and rock art at this site shows that the southern margin of the treeless plain had been incorporated into a cultural and economic landscape. Flint is unlikely to have been the primary attraction because it would have been widely available along the old sea cliffs. The fact that use of this deep cave coincides with last glacial aridity points to the importance of the subterranean water at Koonalda. The water in Koonalda’s lakes is brackish (3,100 mg/L TDS) (Wright 1971) and at the outer limit of tolerance for human consumption. However, some Central Asian communities today use drinking water with 3,500 mg/L TDS (Lioubimtseva and Henebry 2009), so use of the cave water cannot be ruled out. The different role of these two sites in a hunter-gatherer landscape is illustrated by changes after 17 ka. Climatic amelioration brought use of Koonalda Cave to an end, presumably as better water sources became available. But Allens Cave entered a period of increasing use as the last marine transgression brought the coast closer and as mallee woodland colonised the surrounding area. After 15 ka, there is evidence for transport of materials from the coast and increasing contacts across the treeless plain (Cane 1995, 2001). For example, shortly after 14.7 ka, pieces of abalone shell (Haliotis laevigata) were brought from the coast 65 km away, possibly for use as a pendant. A peak in occupation around 10.8 ka is associated with a large hearth, a high concentration of artefacts and the first appearance of desert silcrete, brought from sources on the inland margin of the treeless plain 200–300 km away. Use of other sites on the Nullarbor, such as Madura and Norina, increased between 7.0 ka and 4.5 ka as the climate improved (Martin 1973; Milham and Thompson 1976), and the early Holocene saw a consolidation of settlement in this region.

the darling river and willandra lakes On the southeastern margins of the desert, the Willandra Lakes show rapid oscillations between freshwater and brackish conditions from 30–24 ka, with frequent dry intervals leading to eventual failure of the system around 24 ka (Bowler 1998). At 22 ka, water brieﬂy returned to the Willandra Lakes before the system dried up and desert dunes began to encroach onto the lakeshores (Bowler 1998). From 19–17.5 ka, water returned to the northern part of the Willandra Lakes in a short-lived event that reactivated a freshwater system supporting abundant mussels and ﬁsh (Bowler 1998). In contrast, the lower Darling River, and the lakes associated with it, remained hydrologically active throughout the late Pleistocene (Balme and Hope 1990). Islands in the Interior: Last Glacial Aridity and Its Aftermath 153

The archaeology of this region tracks the hydrological history of these lakes and rivers, showing a persistent regional population but also a shift from the large Pleistocene lakes in the Willandra chain to the smaller lakes and channels fed by the Darling River (Hope 1993). In the Willandra, there is a decline in the number of sites dated 28–24 ka, with a complete absence of shell middens between 24 ka and 20 ka (although several dated fireplaces fall in this gap) ( Johnston, Clark and White 1998). The last phase of the large lakes at 19 ka saw local reinitiation of a lacustrine economy and its subsequent abandonment over 1–2 millennia. On the lower Darling, Lake Tandou has a large cluster of shoreline sites relating to the period 30–28 ka, with an apparent gap in occupation 28–24 ka (Hope 1993; Balme 1995). Occupation resumed again 21–18 ka, with more than fifty sites associated with the Pleistocene river channel (Tandou Creek and Talyawalka Creek) or small lakes such as Kangaroo Lake. The last phase of the Willandra Lakes may have seen incursions from riverine populations to the south. Although the majority of the human fossils recovered from the Willandra sites are undated, they are thought to coincide with the final phases of lake activity around 24–17 ka (Webb 1989;Pardoe1993). Most of this skeletal material appears to have eroded from burials in lunettes. Ini- tially, these were grouped into two different populations – ‘robust and gracile’ – on the basis of cranial robusticity (Thorne 1977). These were interpreted in evolutionary terms, with robust traits inherited from archaic human populations in Java. A number of lines of evidence have now shown this cannot be the case (Curnoe 2009): some of the shared features (e.g., a prominent supraorbital torus, marked cranial thickness and large frontal mastoid air cavities) are constructed differently in the Australian fossils (Webb 1989: 28–9), and some traits (e.g., receding frontal bones) have been accentuated by artificial cranial deformation through head binding (Brown 1981). Other studies question the distinction between the two groups: mtDNA analyses do not distinguish between robust and gracile groups (Adcock et al. 2001), whereas the strong correlation between sex and robusticity suggests that this may be a single population with marked sexual dimorphism in build (Pardoe 2006). But sexual dimorphism alone does not fully account for the difference. Par- doe (1993, 1994, 2006) suggests that the variation among these Pleistocene Australians also matches the difference between people with the tall, lean, long-limbed build of desert people and the stocky, heavily built individuals more characteristic of large, dense, riverine populations along the well-watered Murray Valley to the south. If this is the case, then towards the end of the LGM, the southeastern margins of the desert had become a broad contact zone between peoples with very different demographic histories and patterns of gene flow. A rare snapshot of human life during the LGM is provided by the extraordinary discovery in 2003 of a large series of fossil human footprints preserved 154 The Archaeology of Australia’s Deserts

on a cemented surface near Lake Garnpung (Webb, Cupper and Robins 2006; Webb 2007). Cleaning of the surface revealed 563 human footprints (Figure 5.14), organised in twenty-three trails or trackways, making this the largest single assemblage of Pleistocene human footprints in the world. The tracks occur on the edge of a small groundwater-fed pan; they extend over an exposed area of 850 m2 of carbonate hardpan and run under shoreline lunette sands for another 70 m. Excavation showed that they are sandwiched between sediments OSL dated to 23.0 ± 1.2 ka (below) and 19.4 ± 1.1 and 19.2 ± 1.9 ka (above), indicating that they relate to drying of the lake system following the short freshwater phase at 22 ka. People appear to have been regularly visiting a small pan for water and moving across nearby surfaces while the clays were damp. Using anthropometric data for modern Aboriginal populations, Webb (2007) identified twenty-three individuals – twelve adult men, four women and seven children (ranging in estimated age from 4 to 9 years). Although the number of individuals is suggestive of a local band of twenty to thirty people, the disproportionately low number of adult females is striking and may reflect some partitioning of activities in the landscape. Webb (2007) used Pintupi trackers to assist in interpretation of the tracks. One cluster of tracks (T1–T5 and T8)showssix men running in a flanking movement, as one might see in a cooperative hunt. Remarkably, this group included a one-legged man using a crutch (shown by circular indentations 20–30 mm in diameter) to keep pace with the fastest members of the group. It is relevant that DF Thomson recorded an agile one- legged man using a rough crutch, ‘like a vaulting pole,’ among some desert Pintupi in 1957 (1977: 68 and plate 19). Another cluster of trackways (T9–T13) records a man and a group of four children walking at a more leisurely pace west to east across the pan. These trackways provide a rare example of the collective and cooperative dimension of life in these hunter-gatherer societies at the height of the LGM.

DISCUSSION: THE LAST GLACIAL MAXIMUM REVISITED Thirty years of research have complicated the archaeology of the last glacial period in Australia’s deserts. Palaeoclimatic research shows that the LGM was a longer and more internally complex series of events than first thought, beginning around 30 ka and terminating around 19–18 ka. Archaeological research also shows a more complicated pattern of responses, with marked shifts in the spatial structure of hunter-gatherer land use but little direct evidence for abandonment of entire regions. Despite falling sea levels, expansion of the arid zone on its margins and retraction of the summer monsoon, most regional settlement systems survived the first phase of the LGM. The major impact was around 24–20 ka (as aridity intensified) or immediately after the LGM at 17–16 ka (as rising temperatures Islands in the Interior: Last Glacial Aridity and Its Aftermath 155

Figure 5.14. Fossil human footprint dating to 23–19 ka, impressed in carbonate-rich clays near Lake Garnpung, Willandra Lakes. (Photograph courtesy of R Robins) placed additional stresses on arid ecosystems). In some areas, this led to the collapse of established patterns of settlement. In most, it led to a spatial reorganisation of land use, accentuating a pattern of small population isolates across the interior. The recovery phase was just as complicated. At the regional level, archaeological records show that people responded rapidly to reestablishment of summer monsoon rainfall after 16–14 ka. But many individual sites do not register signiﬁcant use until 8–5 ka, when there appears to have been a consolidation of settlement across the interior. The slow recovery in the aftermath of the LGM (see Figure 5.5) suggests that the rate and amplitude of the climatic oscillations after 18 ka were as dislocating for hunter-gatherer groups as was the LGM itself. We are limited in what we can we say about desert societies during the last glacial maximum, but a few points stand out. First, the archaeological evidence indicates the presence of a small, highly mobile and very dispersed hunter-gatherer population throughout this period. In almost all regions, there was local contraction of settlement, with comparatively low levels of use of the remaining sites (see Table 5.5, Figure 5.5). Patterns of site use are much more attenuated than anything recorded ethnographically in these deserts – with no evidence for higher population densities, even in focal areas. 156 The Archaeology of Australia’s Deserts

Second, there is nothing in the archaeology to suggest systematic long- distance movement of people. People were able to disperse widely following extreme rain events (as the MG4 site suggests), but there is no evidence for systematic forays into the interior from the desert margins. The data are more consistent with resident regional populations. Land use appears to have been tethered to a series of restricted localities where water was available year round – a point-to-point pattern of subsistence – with only limited seasonal dispersal across the backcountry. This was a contraction of living space rather than territory: some groups still moved over extensive areas. Third, if we take the desert as a whole, the archaeological evidence is more consistent with a pattern of widespread ‘cryptic’ refugia than with a geographic division into refuges, corridors and barriers. People appear to have survived across much of the desert, but as scattered occurrences and at low densities – in effect, in pockets of microhabitat. Some regions may have been abandoned, including some areas of sandy desert and parts of the Pleistocene coast, but direct evidence for abandonment of large parts of the interior is more limited than once thought. ‘Each desert has its own barriers, corridors and refuges,’ says Cane, ‘and one should look to this inner variability in order to understand the true nature of desert colonisation and settlement’ (1995: 49). One implication of this picture is that these hunter-gatherer groups inhabited a much smaller social world than contemporary desert people. Open, inclusive social systems would have been essential for social reproduction and demographic viability. Even so, this fragmentation of the desert world suggests there was potential for significant regionalisation in art and material culture. We cannot test this at present because no desert rock art is securely dated to 30–19 ka (see Chapter 7). If we accept that environmental and demographic factors create boundary conditions for Australian social systems, we can also expect these desert societies to have had fluid social groupings, with high levels of personal autonomy, and a land tenure system focused on individual connections to place. Corporate descent groups are likely to have been weak or absent, and there would have been little incentive for strongly territorial systems. The LGM was a period when inland groups had to make a range of structural adjustments to intense aridity and environmental stochasticity. As Thorley puts it, ‘the glacial maximum provides the first and most crucial test of human adjustment to a truly arid environment ...thereisnothingthatwouldhave prepared them for the hyper-arid conditions of the last glacial maximum’ (1998b: 42). Taking a long historical view, the LGM also represents a disjunc- tion between the world established by the colonisation of the desert prior to 30 ka and the expanded social world after 8 ka, when larger populations provided the machinery for a more elaborate political and ritual economy. CHAPTER 6

THE ‘DESERT CULTURE’ REVISITED: ASSEMBLING A CULTURAL SYSTEM

The first archaeological excavation in the desert was one of the most auda- cious episodes in arid-zone research. When American archaeologist RA Gould began excavating Puntutjarpa rockshelter in 1967 – deep in the Western Desert – most Australian research was limited to the temperate or tropical margins of the continent. The pivot for Gould’s research was the handful of Aboriginal people still living as hunter-gatherers, north of Warburton mission. In 1965, the newly established Australian Institute of Aboriginal Studies had sponsored an expedition to film their way of life (beautifully documented by Ian Dunlop in the Desert People film series). Gould arrived at Warburton the following year to make an ethnographic study of these same groups, riding on international interest in hunter-gatherers surrounding the 1966 Man the Hunter symposium (Lee and DeVore 1968). Puntutjarpa is a small rockshelter in the Brown Range, a quartzite ridge set among spinifex and sandhills, not far from the mission. In desert mythology, the contours of the range evoke the muscular body of the ngirntaka ancestor (ngirntaka or ngintaka n. perentie – the largest of the desert varanid lizards). Gould’s excavations at this remote spot in 1967 and 1969–70 represent the largest archaeological dig ever undertaken at a desert rockshelter (altogether about 46 m2 wasexcavated)(Figure 6.1). Even today, it remains the richest mid-Holocene occupation deposit ever discovered in the arid zone, one of the most remote sites and one of the best reported (Gould 1977). The long stratigraphic record at Puntutjarpa appeared to show an extended period of cultural and economic stability spanning the Holocene. Gould proposed the term ‘Australian desert culture’ for this (1971a, 1977: 168– 82), arguing that it recorded a distinctive way of life, highly adapted to desert conditions. This had developed as a response to Holocene aridity, to ‘the onset and persistence of the rigorous environmental conditions of the last 10,000 years’ (Gould 1971a: 175). Its hallmarks were high mobility, a ‘risk-minimizing opportunism’, the harvesting of wild seeds as grain and a mobile toolkit that included microliths and hafted adzes. Puntutjarpa provided 157 158 The Archaeology of Australia’s Deserts

‘evidence for the success of this ancient desert culture adaptation throughout the post-Pleistocene of the Western Desert, culminating in the ethnographic desert culture of that region, in what must surely stand as one of the most dramatic cases of cultural conservatism on record’ (Gould 1977: 182). This was a bold statement for a pioneering exploratory dig, and the concept of an Australian ‘desert culture’ never gained support. Looking back, however, we can see that the Puntutjarpa work framed important questions for desert prehistory. How does aridity shape a hunter-gatherer way of life? Were there distinctive desert adaptations? Could these be traced in stone-tool assemblages? Puntutjarpa also emphasised early and mid-Holocene developments in the arid zone. This was a crucial formative time in the desert and, to see why, we need to turn to post-Pleistocene changes in the environment.

CHANGES IN THE PHYSICAL ENVIRONMENT Australia’s deserts, as we know them today, took shape during the Holocene – a period that began at 11.7 ka with the resumption of rapid global warming after the Younger Dryas stadial (INQUA Subcommission on Quaternary Stratigra- phy – Pillans 2007). Although the Holocene was a comparatively stable period, it was not uniform and can be divided into three distinct phases: r The early Holocene (11.7–8.0 ka) was a period when vegetation and geomorphic systems were still adjusting to rapid postglacial changes in temperature, sea levels and rainfall. r The mid-Holocene (8–4 ka) was a comparatively stable period when woodlands and grasslands reached their maximum extent in the interior, the summer monsoon was strong and regional water tables were high. r The late Holocene saw the onset of a drier, more variable climate in the interior at 4.0–3.7 ka, one dominated by El Nino–Southern˜ Oscillation (ENSO) climate disturbances.

Global Warming The deglacial period saw a protracted climatic transition that transformed the geography of the continent. The most fundamental change was global warming: temperatures rose about 6–9◦Cover7 millennia in the most rapid and prolonged climate change people would have experienced since settlement of the continent. Antarctic ice cores show that global temperatures had begun to rise by 18 ka (Watanabe et al. 2003; Jouzel and Masson-Delmotte 2007) (Figure 6.2) but abruptly fell again between 14 ka and 13 ka, in a stadial known as the Antarctic cold reversal (ACR). In the Southern Hemisphere, rapid warming resumed at 13 ka, peaking in a climatic optimum from 11.5 ka The ‘Desert Culture’ Revisited: Assembling a Cultural System 159

Figure 6.1. Excavations at Puntutjarpa rockshelter in 1969–70 showing the rich, dark archaeological soil exposed in trench 2. (Photograph courtesy of AIATSIS, Bruce Wright collection, image no. 140727) to 9.0 ka, with temperatures about 1.5◦C higher than today (Masson et al. 2000).

Rising Seas In mid latitudes, the most obvious effect of these changes was a rise in sea level as global ice sheets melted. Sea levels in northern Australia rose rapidly after 19 ka – with maximum rates of 15 m/kyr between 16 ka and 12.5 ka (Lambeck, Yokoyama and Purcell 2002). By 13 ka, this had removed all of the extra land created during the last glacial maximum (LGM). By 11.5 ka, most of the remaining parts of the exposed continental shelf had been ﬂooded (Murray-Wallace 2007). By 7.9–7.7 ka, sea level had reached its modern level and continued to rise to an Australian high stand of +1–+1.5 mby7.4 ka (Sloss, Murray-Wallace and Jones 2007). The regional impact was dramatic. In northern Australia, a huge area of lowlands (about 1.6 million km2), formed by the Sahul and Northwest Shelf, was submerged. By 12.2 ka, marine waters began to encroach on Lake Car- pentaria and, by 10.5 ka, this huge freshwater lake (190,000 km2) had become a marine embayment (Reeves et al. 2008). Along the arid northwest margin of the continent, coastal retreat removed a broad plain north of the Burrup Peninsula by 9 ka, isolating the Montebello and Barrow rises as islands and 160 The Archaeology of Australia’s Deserts

facilitating the widespread establishment of mangrove vegetation 8–6 ka, in what is known as the ‘big swamp phase’ (Crowley 1996). In southern Aus- tralia, along the Nullarbor coast, the steepness of the shelf initially limited the extent of marine transgression but, from 14.8 ka, the sea began to remove the coastal plain very rapidly, ﬂooding back to the base of the present sea cliffs by 11.5 ka.

The Summer Monsoon The most important consequence for arid Australia was that flooding of shallow continental shelves drove an intensification of monsoon rainfall (Griffiths et al. 2009). A range of evidence – from ocean cores, palaeovegetation, stable isotopes in emu eggshell and lake records – shows that summer monsoon incursions into the interior switched on around 15–14 ka, then intensified around 11–10.5 ka (Singh and Luly 1991; Johnson et al. 1999; Wyrwoll and Miller 2001; Spooner et al. 2005; Griffiths et al. 2009). One effect was to modestly revitalise rivers, lakes and groundwater in the interior. By 12 ka, shallow perennial lacustrine conditions had returned to Lake Eyre, creating a saline lake (Magee et al. 1995; Magee and Miller 1998; Magee 2007). Nearby Lake Frome shows a similar response: from 12 ka to 11.6 ka, it was a shallow brackish lake (De Deckker, Magee and Shelley 2011). In Central Australia, groundwater discharge in the Amadeus basin appears to have been reestablished by about 10 ka, with evidence for shallow brines at Lake Amadeus (Chen, Bowler and Magee 1993). These conditions continued into the mid-Holocene, with increasing evidence for fluvial activity along major rivers in Central Australia between 12.5 ka and 5 ka (Nanson, Chen and Price 1995; Bourke 1998; English et al., 2001). In the Great Sandy Desert, Lake Gregory was a semiperennial freshwater lake from 14 ka to 6 ka (Wyrwoll and Miller 2001), and regional groundwater tables had risen sufficiently by 7.2 ka to activate organic sedimentation at Dragon Tree Soak (Wyrwoll, Hopwood and McKenzie 1992).

Contraction of the Deserts During the Holocene transition, global deserts contracted from around 29.5 × 106 km2 during the LGM to 17.2 × 106 km2 in the Holocene (Adams et al. 1990: table 1). Australia’s deserts and drylands followed this trend, halv- ing in size to reach their modern extent some time in the early Holocene, as woodlands took over areas that had been desert margins during the late Pleistocene. In the southern Kimberley, charcoals, phytoliths and macrobotanical remains from Carpenters Gap show that tropical grass-rich woodland with Te r mi- nalia andboab(Adansonia gregorii) was in place by 10.9 ka (Wallis 2001; The ‘Desert Culture’ Revisited: Assembling a Cultural System 161

Figure 6.2. Summed probability plot for all radiocarbon (14C) ages from archaeological sites in Australia drylands (N = 908), plotted against changes in Greenland and Antarctic ice-core palaeoclimate records, NGRIP and EPICA (EDML), using δ18O as a proxy for global temperature (higher values correspond to warmer conditions). Numbered peaks (bold) mark inferred population events. ACR indicates position of Antarctic cold reversal (14.5–12.5 ka). (Source:Smithetal.2008:ﬁg.3)

Frawley and O’Connor 2010). On the eastern margins of the arid zone, eucalypt forest replaced chenopod shrubland around Ulungra Spring at 12.4 ka (Dodson and Wright 1989). In the southeast, woodland and tall shrubland reached their maximum extent within duneﬁelds in the lower Darling area and Little Desert between 9 ka and 4 ka, with dense mallee eucalypt woodlands and Callitris both present (Thomas, Enright and Kenyon 2001; Cupper 2005). On the Nullarbor, arid mallee shrubland followed the retreating coastline and, by 162 The Archaeology of Australia’s Deserts

10.1–5.8 ka, covered the southern edge of the formerly treeless plain (Martin 1973).

Vegetation Response within the Deserts In Central Australia, local vegetation recovered rapidly after 15 ka. By the early Holocene, the regional pattern appears to have been a mosaic of mulga (Acacia aneura) woodland and spinifex (Triodia) hummock grassland (Smith 2009b), supporting a range of animal species adapted to both shrublands and xeric grasslands, including the hairy-nosed wombat (Lasiorhinus)andTasmanian devil (Sarcophilus), both locally extinct today (Megirian et al. 2002). By 8.3 ka, revegetation of the catchment around Puritjarra had choked off sediment supply to the rockshelter, and the character of the sediments changed abruptly. A similar effect is also evident at Kings Canyon (60 km southeast of Puritjarra), where runoff into the joint lines and ﬁssures that feed the Palaeozoic sandstone aquifers on the rim of the Amadeus basin ceased around 8.1 ka as the recharge zone revegetated ( Jacobson et al. 1989). The phytolith record at Puritjarra rockshelter shows that Poaceae values peaked at 5.8 ka and then declined. In the arid northwest, the FR10/95-GC17 core shows that the period from 14 to 3 ka was wetter than today – with an annual rainfall between 300 and 400 mm, mainly in summer – supporting an open grass-rich vegetation with peak values for Poaceae around 9.8 ka (van der Kaars and De Deckker 2002). In the Lake Frome area, there was a sudden decline in Callitris woodland after 12 ka (possibly as a response to a more active ﬁre regime) and a switch to grasses and other summer-rainfall taxa (Singh and Luly 1991; De Deckker et al. 2011). In the northern Flinders Ranges, pollen and plant macrofossils from Leporillus rat middens show that a terminal Pleistocene vegetation dominated by halophytes was replaced by woodlands with a grassy understory around 10–8 ka (McCarthy, Head and Quade 1996; McCarthy and Head 2001). Taken as a set, these records suggest that woodlands and grasslands reached their maximum extent in much of the arid interior between 8 ka and 6 ka.

The Onset of the El Nino–Southern˜ Oscillation The favourable conditions of the mid-Holocene were truncated by a reorganisation of circulation around 5 ka. This saw a dramatic weakening in summer monsoon rainfall around 3.7 ka and a shift to an ENSO-dominated climate marked by high interannual and interdecadal variability in rainfall (Shulmeister 1999; Cosford et al. 2008; Moros et al. 2009; Quigley et al. 2010). Most arid- zone lakes switched to more ephemeral conditions (Lake Gregory) or became ephemerally flooded playas (Lakes Eyre, Frome and Lewis). At Lake Eyre, the modern saltpan was established at 4 ka (Magee 2007). Vegetation across the interior registers drier, more variable conditions from 4 ka (van der Kaars The ‘Desert Culture’ Revisited: Assembling a Cultural System 163 and De Deckker 2002). In southeastern Australia, there was a contraction of woodland and tall shrubland cover on the dunefields and a regional shift to more open vegetation with more chenopods, around 4 ka (Thomas et al. 2001; Cupper 2005). Summer grass taxa in the Lake Frome area were replaced by chenopods, acacia and myrtaceous shrubs (Singh and Luly 1991). In Central Australia, disruption of regional vegetation is marked by a sharp decline in Poaceae around 3.8 ka (Smith 2009b). The rapid increase in dust emissions from the Lake Eyre basin at 4.8 ka also signals a shift to a drier and more variable climate, with pulses of fluvial activity (controlling sediment supply) alternating with seasonal or interannual drought (allowing the sediment to be entrained as windborne dust) (Marx, McGowan and Kamber 2009).

A MODICUM OF IDEAS: HUNTER-GATHERER SOCIETIES IN THE HOLOCENE Against this background, the period from 12 ka to 4 ka stands out as probably the most under-researched and under-theorised period of desert archaeology. In the ‘desert culture’ model, a distinctive desert way of life formed at the beginning of the Holocene as a response to increasing aridity. This was a socially fluid, highly mobile hunter-gatherer society with a sparse but efficient material culture attuned to opportunistic use of short-lived flushes of desert resources and able to cope with high levels of environmental stochasticity. Here, Gould drew freely on the North American concept of a ‘desert culture’ ( Jennings and Norbeck 1955; Jennings 1957) – a proposal that faced much the same raft of objections as its Australian counterpart before being formally withdrawn in 1973 ( Jennings 1973; see O’Connell and Madsen [1982]for a summary of the objections). In both cases, there were problems with the radiocarbon dating and stratigraphy of the key site and in establishing continuity in artefact inventories and material culture. And the concept downplayed the impact of environmental change, as well as the difficulty of directly correlating archaeological evidence with ethnographic observations. In Australia’s Western Desert, the underlying reason for long-term cultural stability was that the subsistence economy was tightly constrained by the poverty and unpredictability of the desert environment (Gould 1977: 168–9, 179–80). The desert, Gould said, was ‘a great leveller’ of ambition: ‘Anyone attempting . . . to adapt in any other way would undoubtedly perish’ (1977: 182). Of course, the same might have been said about earlier societies in the desert (see Chapter 4), but in 1977, there was no evidence for late Pleistocene occupation in the Western Desert or Central Australia. With better data on the environmental history of the arid zone, the period from 12 ka to 4 ka now looks more likely to have been an expansive phase for both people and desert biota. The desert got wetter, not drier. An increase in primary production from 12 ka was likely coupled with finer-grained accessibility to country as the distribution 164 The Archaeology of Australia’s Deserts

of small water sources (springs, wells, soakages, seasonal claypan waters and rockholes) improved. This would have provided conditions for an early to mid-Holocene expansion of population in the desert and higher regional population densities (Smith 1988: 49–55) – although researchers disagree about the timing, speed and extent of this process (as debate over the history of the sandy deserts shows). Other work has shifted the ‘desert culture’ concept forward to the late Holocene. Many of the ‘specific desert economic and social traits’ that characterise ethnographic groups now appear to date to the late Holocene (Hiscock and Wallis 2005; see also Veth 1989b, 1995b). The most distinctive technological elements of ethnographic desert groups – adzes, seed-grinding implements and composite tools – appear around 4 ka, at a time when desert groups were faced with a shift to a more variable ENSO-dominated climate (Smith 1986a; Hiscock and Veth 1991; Veth et al. 2011). Adzes and microliths, the stone components of new lightweight composite tools made of wood, sinew and mastic, have been described as a technological response to ‘risk’, one tailored to the mobility that increased environmental variability imposed (Hiscock 1994; Veth, Hiscock and Williams 2011). The appearance of seed-grinders at 4 ka also hinges on shifts in mobility, but in a different way. Population growth through the first half of the Holocene would have increased the cost of relocating to another foraging patch when bush foods were locally depleted. With the added pressure of a deteriorating environment around 4–3 ka, one response was to increase diet-breadth to include more processing of seeds and grain and thus defer the necessity to move (Smith 1986a, 1988). On the coast, the widespread appearance at 4.2 ka of shell middens and shell mounds dominated by a single molluscan species, Anadara granosa, represents another response to a changing environment. Veitch (1999) draws a parallel between exploitation of Anadara bivalves and grass seeds, arguing that both are r-selected taxa (small- bodied opportunistic species with high intrinsic rates of increase), whose ready availability underpins settlement of marginal habitats with high environmental stochasticity. Each, in their own way, treats the beginning of the late Holocene (around 4 ka) as a nexus of economic and technological change, with ramifications for prehistoric demography and the sociopolitical structures of desert societies.

SITES AND TRENDS 12–4 KA An overview of archaeological trends in Australia’s deserts and drylands is given by Smith et al. (2008), using summed probability radiocarbon plots from the AUSTARCH database as a proxy for long-term population trends (see Figure 6.2). There is a broad correlation between population levels in Australian drylands and the stabilisation of global climate. In the late glacial period, during the ACR (14.5–12.5 ka), population growth appears to have The ‘Desert Culture’ Revisited: Assembling a Cultural System 165 stalled, and it did not resume until the early Holocene. Dryland populations reached their early Holocene maximum around 8 ka, followed by a widespread collapse of desert populations around 3 ka as ENSO intensiﬁed. To unpack these trends, we need to separate the effects of the last marine transgression on coastal groups from changes affecting societies in the desert interior. For the former, we can contrast two quite different regions: the arid west coast, from Broome to Shark Bay, where most archaeological evidence is from small shell middens; and the Nullarbor coast, where the key evidence is from deep excavations at limestone dolines and rockshelters. Table 6.1 provides a list of sites dated between 12 ka and 4 ka.

The Arid West Coast On the arid west coast, the last marine transgression left a distinctive imprint in the archaeological record. Extensive forests of Rhizophora mangroves formed along drowned coastlines in northern and northwestern Australia (Crowley 1996). This was not simply a matter of mangroves following a retreating coastline – because growth of mangroves is impeded by rapid rise in sea level – but rather it constituted a distinctive phase in the adjustment of coastal habitats. As the rate of sea level rise slowed, pollen counts from offshore core Fr10/95- GC17 record a rapid increase in the extent and diversity of mangrove habitats from 12 ka, peaking around 8 ka and then gradually declining until around 4.5 ka (van der Kaars and De Deckker 2002). The expansion of mangroves created new opportunities for coastal foragers. Archaeological evidence shows widespread use of mid-Holocene mangrove forests, as part of a local foraging economy that also included a range of other coastal and inland habitats. The mangrove gastropod Terebralia palustra first appears in archaeological sites at Jansz at 12.1 ka (Przywolnik 2002), at Wadjuru rockpool (now on Rosemary Island) at 9 ka (Bradshaw 1995)and Warroora shell midden at 8.2 ka (Kendrick and Morse 1982). Between 8 ka and 6 ka, small middens of Terebralia mangrove shells are the dominant site type, extending along the coast from Shark Bay in the south to Port Hedland in the north (Figure 6.3). Most of these sites are smaller and more diffuse than later shell middens and record use of rocky shore, reef and sandy intertidal areas, as well as mangroves. One of the best documented of these middens is the Silver Dollar site, dated to 7.7–7.4 ka (Bowdler 1999). Here, the Terebralia are associated with a range of other shellfish species, emu eggshell and fish otoliths – including inshore and reef species such as whiting (Sillago schomburgkii) and snapper (Chrysophrys auratus) – as well as a range of arid-zone marsupials including large macropods (Macropus fuliginosus and M. robustus), bettongs (Bettongia leseuer)and small wallabies (Lagorchestes hirsutus, Lagostrophus fasciatus and Onchogalea lunata) (Dyason 2007). This broad picture is corroborated by rockshelter deposits, 166 The Archaeology of Australia’s Deserts table 6.1. Archaeological sites in Australian deserts and drylands dating to 12–4 ka

Site and region Site type Holocene record Source

Arid west coast Noala Rockshelter Sparse occupation 13.9–9.3 ka Veth et al. 2007 Pilgonaman Rockshelter Main occupation 13.9–11.0 ka. Late Morse 1993a Holocene occupation 0.5 ka–present. Jansz Rockshelter Occupation unit 12.1–10.8 ka. Late Pryzwolnik Holocene occupation unit 1.8 2002 ka–present. Yardie Well Rockshelter Occupation 11.6 ka–late Holocene. Morse 1993a Main occupation 5.8 ka–present. Wadjur u Shell midden Open shell midden with mangrove Bradshaw 1995 rockpool molluscs 9.0–6.9 ka. Haynes Cave Rockshelter Sparse occupation 8.6–7.8 ka Veth et al. 2007 followed by abandonment of Montebello islands as sea level reached modern levels. C99 Rockshelter Early Holocene occupation unit Pryzwolnik 8.4–7.6 ka. Late Holocene 2002 occupation 1.9 ka–present. Warroora Shell midden Open shell midden with mangrove Kendrick and molluscs 8.6 ka. This is the earliest Morse 1982 Terebralia midden on the mainland. Silver Dollar Shell midden Open shell midden with mangrove Bowdler 1990b, molluscs 7.7–7.4 ka. 1999 Clarke’s Cave Cave Sparse occupation throughout Bradshaw 1995 Holocene. Pre-marine levels overlain by mangrove molluscs from 7.6 ka and Anadara by 0.9 ka. Skew Valley Shell midden Small shell midden with mangrove Lorblanchet P0406 molluscs 7.4–6.6 ka, overlain by and Jones much larger midden dominated by 1979 Anadara bivalves from 4.1–2.4 ka, with intermittent use during last two millennia. Anadara shelter Rockshelter Basal shell midden containing Bradshaw 1995 mangrove molluscs from 6.8 ka, overlain by Anadara-dominated midden from 4.2 ka. Not-so-secret Rockshelter Basal occupation at 6.2 ka associated Bradshaw 1995 rockshelter with mangrove molluscs, overlain by younger units with non–mangrove shell species. Mandu Mandu Rockshelter Use at 5.8 ka associated with Morse 1993a mangrove molluscs. Main late Holocene occupation unit is 2.5 ka–present. The ‘Desert Culture’ Revisited: Assembling a Cultural System 167

Site and region Site type Holocene record Source

Wobiri Rockshelter Occupation 5.5 ka–present with most Pryzwolnik intensive use 5.5–2.0 ka. 2002 West coast Shell middens Numerous small shell middens Bradshaw 1995 middens (including thirty middens with 14C Veth 1995a dates). Sites dominated by Pryzwolnik mangrove molluscs are mainly 8–5 2002 ka; sites dominated by other shell species are mostly <4.5 ka. Pilbara (inland) Yirra Rockshelter Initial occupation during LGM, Veitch et al. continuing 23.0 ka–present. Peak 2005 artefact and charcoal densities 23.0–20.2 ka and 3.5–present. Cleft Rockshelter Intermittent visits from >15.2–1.0 ka. Salmon 1998 Rockshelter Marwick 2002a Newman P0187 Rockshelter Individual ashy hearths at c.24.8, 11.4, Maynard 1980 6.0 and 3.2 ka. Highest Comtesse concentration of artefacts is 6.0–3.2 2003 ka. Newman P2055 Rockshelter Individual ashy lenses at 31.1, 17.1, Brown 1987 10.1 and 9.9 ka. Main use of site is Comtesse 7.1–3.8 ka. 2003 Marillana A Rockshelter Main occupation 10–4ka. Marwick 2005 West Angelas Rockshelter Basal occupation 9.0 ka overlain by Dias et al. 2006 DFSH4 late Holocene occupation 1.0–0.9 ka. Bastion Shelter Rockshelter Discrete occupation unit 9.0–2.5 ka Brown and P5315 interstratiﬁed within sterile Mulvaney sediments. 1983;Brown 1987 Manganese Rockshelter Occupation 8.0–4.6 ka and at 0.4 ka. Morse 2009 Gorge 8 Malea Rockshelter Occupation from 24.3 ka to present, Edwards and with in-situ artefacts 24.3–18.4 ka Murphy and 7.5 ka. Most intensive phase of 2001, 2003 use <3.0 ka. Manganese Rockshelter Occupation 7.4–4.7 ka followed by Quartermaine Gorge 2 collapse of rockshelter entrance. 1994 Newman P0959 Rockshelter Intermittent occupation throughout Comtesse 2003 Holocene with concentration of stone artefacts, ochre and hearths between 7.2 and 6.2 ka. Hope Downs Rockshelter Intermittent occupation from 7.2 ka Morse 2009 SH/N-2 to last millennium.

(continued) 168 The Archaeology of Australia’s Deserts table 6.1 (continued)

Site and region Site type Holocene record Source

Djadjiling Rockshelter Major decline in artefact density Law, Cropper 24–18 ka, which does not recover and Petchey until 7 ka. 2010 West Angelas Rockshelter Basal occupation 4.9–3.8 ka Dias et al. 2006 RR8 extending into the last millennium. West Angelas Rockshelter A veneer of occupation deposit Dias et al. 2006 DFSH3 4.6–4.4 ka on bedrock. Mesa J J24 Rockshelter Occupation layers at 28.4 and 4.4 ka Hughes et al. interleaved with sparser 2011 occupation. Western Desert Puntutjarpa Rockshelter Initial use of site 11.8 ka. Major Gould 1977 occupation unit 7.6–7.5 ka. Intermittent use 4.2 ka. Upper occupation unit 0.4 ka–present. Bush Turkey 3 Rockshelter Basal occupation c. 8.4 ka possibly Veth et al. 2008 12–8 ka. Occupation unit 4.0–3.0 ka. Main period of site use c. 1 ka–present. Jalpiyari Rockshelter Sparse occupation from 5.8 ka with Veth 1993 main concentration during last millennium. Serpents Glen Rockshelter Upper occupation layer 5.4–present. O’Connor et (Katjarra) Main use <1 ka. al. 1998 Murchison region Walga Rock Rockshelter Occupation unit at 11.5–7.9 ka Bordes separated by disconformity and et al. 1983 eroded deposits from late Holocene occupation unit. Main use of site 1.0 ka. Billibilong Open site Alluvial ﬂat with occupation from 6.1 Bordes et al. Spring ka, with microlithic assemblage 1983 from 4.1 ka. Katampul Rockshelter Main use 5.2 ka–present. O’Connor and Veth 1996 Gidgee Rockshelter Occupation 5.1–2.2 ka. E. Webb unpublished Mount East Rockshelter Occupation 5.1–3.4 ka. E. Webb rockshelter unpublished Uramurdah Open site Cluster of hearths on open site. One Bindon 1986 claypan dated to 4.6 ka. Lake Barlee 8 Rockshelter Basal occupation 4.4 ka. O’Connor and Veth 1996 The ‘Desert Culture’ Revisited: Assembling a Cultural System 169

Site and region Site type Holocene record Source

Central Australia Puritjarra Rockshelter Ongoing site use from 35 ka to Smith 2006, present with well-defined phases of 2009b occupation at 35 ka, 24–8.3 ka and during the last millennium. Kulpi Mara Rockshelter Pulses of sedimentation with Thorley 1998a occupation units at 15.4–10.9 ka, and 1998b; 5.4–2.6 ka and 0.6 ka–present. Thorley, Faulkner and Smith 2011 Wanga East W04 Rockshelter Discrete occupation layers at 9.4–5.8 Smith, ka and 0.4 ka–present. Watchman and Ross 2009 Lake Eyre Basin Youlain Springs Hardpan Regular use of springs from 16.9 to Robins 1998 4.9 ka. Kaponyee Springs Hardpan Artefacts in hardpan dated to 11.7 ka. Robins 1993 1 Cuckadoo 1 Rockshelter Possibly occupied at 18.4 ka but Davidson, unconfirmed. Discrete occupation Sutton and units at 14.7–14.3 ka; 6.2–3.3 ka Gale 1993 and 1.7 ka–present. JSN Hearths and Multiple use of open site in dunefield Smith, open site with dated hearths at 17.6, 17.0, Williams and 12.3 and 2.4 ka. Wasson 1991 Chidlee Well Hearths and Dated hearths at 4.9 ka. R. Callen open site unpublished Flinders ranges Hawker Lagoon Lunette Occupation units 18.0– >9.3 ka and Lampert and 5.8 ka–present. Main use of site is Hughes 1988 18–9 ka and last 1 ka. Balcoracana Transverse Major use 9.9–5.5 ka marked by Lampert and Creek dune carbonate-encrusted cores and Hughes blocky implements. Only 1988; intermittent use during last 5 ka. Gardner et al. 1987 Arkaroo Rock Rockshelter Basal occupation unit 7.1–6.7 ka Draper 1989 overlain by major occupation layer with stone-lined hearths, ochre and peak artefact densities 5.5 ka–present. Cons Bore Hearth Oldest of cluster of buried hearths, Nobbs 1983 Hearth 3 4.4 ka. Ethiadina Hearth Buried hearth 4.9 ka. Nobbs 1983 Hearth 1 (continued) 170 The Archaeology of Australia’s Deserts table 6.1 (continued)

Site and region Site type Holocene record Source

Nullarbor and South Coast Allens Cave Doline Only sporadic use until 15 ka. Cane 1995 Occupation peak at 10.8 ka with declining levels of use after 3 ka. Norina Rockshelter Occupation from 10.7 ka–present Marun 1972; with main use 7.0–4.5 ka. Martin 1973 Madura Cave Rockshelter Basal occupation 8.8 ka continuing Milham and throughout Holocene, with most Thompson intensive use from 4.8 ka, declining 1976 after ∼2.7 ka. Dingo appears 3.7 ka. Clare Bay Footprints Fossil footprints of humans, emus and Belperio and footprints macropods in margin of saline Fotheringham coastal lake ∼5–6 ka. 1990 Streaky Bay Shell middens Several shallow shell middens Nicholson 1994 middens dominated by Nerita and Austrocochlea periwinkles. Acraman Creek Ridge site: 7.2 ka; Granites site: 7.4 ka.

which include a variety of shellfish, chiton, crabs, barnacles, baler shell fragments, and fish bone and turtle remains, as well as small macropods and emu eggshell (Przywolnik 2002). The proliferation of Terebralia provided a small but constant source of protein, potentially available throughout the year and easily collected. The availability of this shellfish may have been a factor in stabilising settlement along these arid coastlines, underpinning at least part of the subsistence round. The frequency of Terebralia in archaeological sites at this time reflects the likely abundance of this whelk (up to 100 molluscs/m2 – Wells and Lalli 2003), which is often the dominant species in mangroves. However, Terebralia has low flesh yield compared to bivalve species (average meat weight per shellfish: Terebralia palustra 4.5 g compared to Anadara granosa 7.6 g). Regular harvesting of this comparatively low-ranked species may therefore signal a growing population and seasonal shortfalls in subsistence. Other archaeological evidence reinforces the picture of mid-Holocene expansion and consolidation of hunter-gatherer societies on these coastlines. Rockshelters from this period show more intensive use than in the late Pleis- tocene, with assemblages containing grindstones, as well as chipped stone artefacts and ochre. At least one fish trap related to the high stand of sea The ‘Desert Culture’ Revisited: Assembling a Cultural System 171

Figure 6.3. Summed probability plot showing radiocarbon (14C) dates on Terebralia and Anadara shell from archaeological sites. This illustrates the expansion and decline of mangrove habitats on the arid west coast around 8–6 ka. (Source: AustArch1 database)

level at 7.4 ka has been identified (Przywolnik 2002: 315–23). Mid-Holocene occupation of the Dampier–Burrup coast is also associated with a major phase of rock engraving involving the production of hundreds of motifs, depicting fish, dugong, turtles and human figures (see Chapter 7 for details). This is paralleled at Northwest Cape by the appearance in site C99 of dry crayons of white pigment at about 8.7 ka (Przywolnik 2002: 261). Mangrove habitats contracted after 6 ka and, by 4.5 ka, were reduced to their modern fragmented distribution on these coasts. The main drivers of this were morphodynamic adjustments of the coastline (including changes in sediment budget and progressive sedimentary infilling of estuaries and embayments) (Crowley 1996). But a decline in sea-surface temperatures (Stott et al. 2004) and in the northwest monsoon (O’Connor 1999b: 46) might also have been contributing factors. The decline of mangroves was associated with marked changes in the organisation of the foraging economy. And there is a distinct spatial pattern to human responses: south of 21.5oS (from Northwest Cape to Shark Bay), use of the coast declined sharply and became more intermittent and seasonal after 6 ka. Only a few sites have archaeological evidence dating to the period 6.0–2.5 ka. These record an economic shift to chiton and some species of reef fish (O’Conner 1999b; Przywolnik 2002), although wherever people had access to small local patches of mangroves, the use of mangrove shellfish continued until 4.0–3.8 ka (e.g., Eagle Bluff midden at Shark Bay). North of 21.5◦S, along the Pilbara coast, the coastal economy expanded, with 172 The Archaeology of Australia’s Deserts

a rapid switch to intensive harvesting of the bivalve Anadara granosa as the structure of coastal environments changed (see Figure 6.3).

The Anadara Middens The formation of large areas of open sandy and muddy intertidal flats in northern Australia created ideal conditions for the proliferation of Anadara granosa, a high-yielding and fast-growing shellfish. The decline of Terebralia- dominated mangroves probably also broadened the niche available to Anadara granosa, an r-selected shellfish able to fill vacant habitats quickly. On this part of the coast, the latest Terebralia middens date to 4.7–4.3 ka (e.g., Tubridgi and FMGP04–14). By 4.2 ka, these had been replaced by larger, single-species Anadara middens, often forming shell mounds or mounded earth and shell middens. This is part of a wider shift across northern Australia to intensive harvesting of Anadara during the late Holocene (Veitch 1999; Faulkner 2009). Although Harrison (2009) argues for a slightly earlier date on the Pilbara coast, his key radiocarbon sample (4,592 ± 133 BP, Wk-19850) in midden FMGP04–14 only provides a terminus post quem for the change. Sites on the Pilbara coast range from shallow, extensive bands of shell through to stratified shell mounds 20–40 m long and 1–4 m deep, typically made up of 90–98 per cent of Anadara (by shell weight) (Clune and Harrison 2009). Few sites were occupied for more than 600–800 years (probably reflecting the lability of local shell beds). The only large controlled excavation is Lorblanchet’s work at Skew Valley, in the Dampier–Burrup region, excavated in 1975–6 (Lorblanchet 1977, 1992; Lorblanchet and Jones 1979). This is a low, stratified mound of earth and shell, 20 m long and 1 m deep, at the foot of one of the hills of loose granophyre boulders that dominate the Burrup peninsula (Figure 6.4). By 1975, the midden had been partly destroyed by a road cutting. This provided a long section through the deposit (Figure 6.5) that revealed a small midden of Terebralia gastropods (∼14 m3) overlain by a much larger Anadara midden (∼118 m3). Excavation confirmed this picture of two major stratigraphic layers. The basal layer, dating to 7.4–6.6 ka (from 6,960 ± 100 BP ANU-1836 to 6,280 ± 90 BP ANU-1835B), is dominated by Terebralia, with an assemblage of cores and heavily retouched implements made from local granophyre. The upper Anadara-dominated layer has a more complicated stratigraphy of lenses of shell and charcoal, dating from 4.1 ka (4,150 ± 80 BP ANU-1834)to2.4 ka (2,770 ± 80 BP ANU-1838). Lorblanchet’s excavation was large enough (13 m2) to show the outline of structures marked by post holes and areas of finely crushed shell cleared of rocks, as well as more recent ovens dug into the surface of the midden. A thin basal unit of the Anadara midden contained an assemblage (mainly of quartz) that predates the small-tool phase. This was overlain by a deposit containing geometric microliths; stone artefacts made The ‘Desert Culture’ Revisited: Assembling a Cultural System 173

Figure 6.4. Excavations at the Skew Valley midden, 1975–6, Burrup–Dampier region. Part of the midden has been cut away by a road (top left). The main excavation (centre) focuses on part of the midden at the foot of a hill with engraved boulders. (Photograph courtesy M Lorblanchet) of chert, chalcedony and quartzite; grindstones or pestles; and fragments of engraved boulders from the nearby hill. Excavated bone shows that people using the midden were taking crabs, catﬁsh, barramundi, wrasse and rock wallabies, in addition to bivalves such as Anadara granosa. The data suggest much more intensive occupation from 4.1 ka: artefact density is signiﬁcantly

Figure 6.5. Stratigraphy of the Skew Valley midden, exposed in road cutting. (Source: Lorblanchet 1977:ﬁg.1) 174 The Archaeology of Australia’s Deserts

greater (∼110 lithics/m3 compared to 34 lithics/m3 in the Terebralia midden); the presence of grindstones on the margin of the midden, as well as in the excavated assemblage, suggests the processing of plant or animal foods; the presence of structures points to more sustained use of the site; and this also appears to have been the major production phase for the rock engravings on the nearby hill. Given the large number of sites and the volume of shell involved, archaeologists have generally interpreted the late Holocene Anadara middens as signalling a significant structural shift in the foraging economy, involving changes in residential mobility, seasonality and increasing population density on the coast. The contrary view is that it was simply a switch from one shellfish species to another as species abundance changed (O’Connor 1999b). But it is clear that these changes also represent an intensification of the earlier economic pattern based on Terebralia, with a shift towards significantly greater harvesting of shellfish and a focus on a single high-yield bivalve species. In northern Aus- tralia, Faulkner (2009) has shown that even though this fast-growing species is very resilient under harvesting pressure, the exploitation of Anadara granosa was sufficiently intensive and constant to cause a decline in the size of shellfish and to depress the age–size structure of the mollusc population. Although there are no comparable data for the Pilbara middens, the volume of shell suggests that intensive use of Anadara formed a regular part of the subsistence round. Clune and Harrison (2009) argue that this most likely represents a seasonal, semisedentary concentration of population during the late wet season (a time when local water sources allow gatherings) as one extreme of an annual round involving aggregation and dispersal of the regional population. Use of coastal resources may also have become more critical after 4 ka as an ENSO-dominated climate brought greater interannual variability to terrestrial ecosystems. Veitch (1999) notes that seasonal reliance on Anadara potentially reinforces ‘the weakest link in an economic and scheduling chain’ and that ‘an economy thus strengthened may allow larger populations to exist and some areas to be occupied for the first time’ (1999: 57). Some archaeologists draw a further link between the need for widely ramified social networks to buffer increasing environmental risk after 4 ka, and the potential function of mounds as locales for ceremonial gatherings to service these networks (Clune and Harrison 2009). There seems little doubt that coastal groups would have co-opted the abundant shellfish to finance ceremonial gatherings. The efflorescence of rock engraving and the numerous standing stones and thalu (‘increase’) cairns point to considerable investment in maintaining ritual and mythic traditions during this period. However, there are no hard data to substantiate a direct link between midden sites and ceremonies. Most Anadara middens represent shell-harvest rates within the range recorded for residential foraging groups in northern Australia (Meehan 1982). Extrapolating from site data in Harrison (2009), the total volume of shell in The ‘Desert Culture’ Revisited: Assembling a Cultural System 175 even the largest middens lies within the range that a foraging group of twenty- five women could be expected to gather over a few years (Meehan 1982). These shell middens also tend to be too small to allow the segregation of participants and partitioning of activities that major ceremonies require. None is associated with the sort of remains we might expect from major ceremonial gatherings, such as substantial quantities of ochre or exotic stone, or indicators of significant local pressure on resources (large assemblages of grindstones or anvils, evidence of logistical foraging, or increases in diet breadth). Rock engravings and stone arrangements are clustered in coastal areas, but this may reflect the distribution of population rather than imply a functional association between the middens and these features.

The Nullarbor Coast and Hinterland The last marine transgression also triggered an expansion of settlement along the Nullarbor coast. A narrow band of mallee eucalypt woodland followed the retreating coastline, replacing chenopod shrub-steppe (Martin 1973). By 11.5–10 ka, this had created a timbered strip 30–90 km wide. Most of the archaeological evidence associated with this derives from pioneering excavations in the 1960s at a series of limestone dolines and rockshelters, including Norina, Madura Cave and Allens Cave (N145)(Marun1972;Martin1973). These are remarkably deep, well-stratified deposits with numerous hearths and ash lenses. Marun’s excavations provide only a rudimentary chronology and typology, and these sites would clearly repay more detailed investigation. But, so far, only Allens Cave has seen a second phase of investigation (Walshe 1994; Cane 1995; Roberts et al. 1996; Turney et al. 2001a). On the available data, all three excavated sites record an expansion of settlement in the early Holocene: there was a marked increase in use of Allens Cave at 10.1–10 ka, Norina was first occupied at 10.6 ka, and occupation of Madura Cave began at 8.8 ka. The first indication of the early Holocene changes at Allens Cave is a large multiple-use domestic hearth (1.45 m long) of compacted white ash and charcoal, burnt bone and burnt flint artefacts, dated to 10 ka (9,530 ± 190 BP ANU6849; 9,270 ± 140 BP ANU6850, 10.1 ± 0.6 ka OxODAC150)(Figure 6.6). This episode initiated a phase of comparatively intensive site use from 10 ka to 3 ka, marked by a black, organic-rich occupation deposit with frequent lenses of ash and charcoal, burnt and calcined animal bone and many stone artefacts (Cane 1995). The bone assemblage illustrates the astonishing diversity of fauna in these mid-Holocene woodlands – as rich as anywhere in the arid zone – including possums (Trichosurus and Pseudocheirus); arid-zone wallabies (Onychogalea lunata, Lagostrophus fasciatus and Lagorchestes hirsutus); bandicoots (Perameles, Isoodon and Chaeropus); bettongs (Bettongia penicillata); and sticknest rats (Leporillus); as well as numbat (Myrmecobius fasciatus), bilby (Macrotis lagotis) and large kangaroos such as Macropus fuliginosus (Walshe 1994). Only about 2.2 176 The Archaeology of Australia’s Deserts

per cent of the archaeological bone appears to have been directly deposited as human-food remains. Most is due to owls and small carnivores sharing the cave. But enough evidence survives to indicate that the local subsistence economy focused on the abundant small- and medium-sized mammals with occasional hunting of large macropods. From 10 ka, the stone-tool assemblage is dominated by use of high-quality local flint, by then exposed in the sea cliffs, removing the necessity to mine flint in deep caves such as Koonalda. The appearance of other more exotic material suggests an expansion of foraging range and wider links with neighbouring areas. People collected tektites (small glassy meteorites) from ‘strewn fields’ on the Nullarbor Plain. The presence of small amounts of desert silcrete from sources 200 km to the north also suggests that people were now able to travel across the treeless plain. Cane speculates that ‘the presence of these exotic materials at Allens Cave may even mark the establishment of the first mythical pathways across the plain and associated systems of exchange and social interaction that were recorded at Wilson Bluff in the historic period’ (1995: 27). Although the early to mid-Holocene saw greater access to the interior, the rising ocean created a line of sea cliffs 60–120 mhighthatrestrictedaccess to the sublittoral zone. Despite living along the coast, people living on the edge of the Nullarbor could only reach the sea in a few areas. Fragments of a Katelysia bivalve at Allens Cave at 10 ka are probably the remains of a single shell implement. Apart from this, the Allens Cave assemblage contains just two fish vertebrae to indicate the presence of the Southern Ocean only 8 km away (Walshe 1994). Further east along the coast, on the margin of the Great Australian Bight, where marine resources were more accessible, the largest of the regional shell middens date to 7.4–7.2 ka, coinciding with stabilisation of sea level (Nicholson 1994). Near Fowlers Bay, dolomitic muds around a saline lake created during the marine transgression and then rapidly in-filled have also preserved human footprints (including those of an adult and a child) dating to 6 ka – as well as tracks of emus, kangaroos and wallabies (Belperio and Fotheringham 1990). The switch to a drier ENSO-dominated late Holocene climate affected these coasts just as it did those in the west. On the Nullarbor, woodland contracted on the landward margin of the timbered strip after 4.5 ka. Martin (1973) originally suggested this was environmental degradation due to human burning, but it now seems likely that it reflects the onset of a more arid climate in the late Holocene. All of the doline excavations record a fall-off in levels of site use around 4–3 ka. Use of Allens Cave declined sharply after 3 ka (Cane 1995). The major period of occupation at Madura Cave was from 4.8 ka to about 3 ka (age–depth interpolation suggests that the top of this main occupation unit may date to 2.7 ka) (Marun 1972). The major use of Norina was between 7.0 ka and 4.5 ka, after which it declined (Marun 1972: fig. 22). The ‘Desert Culture’ Revisited: Assembling a Cultural System 177

Figure 6.6. Stratigraphic section for Allens Cave, showing dated hearths (in black) and OSL samples (open circles). Column at left shows stone artefact densities (no. of lithics/kg sediment in squares D2,D3 and E4). The sharp stratigraphic break at 1.5 m is an erosional unconformity, created when storm waters stripped sediments from the doline and redeposited them at the back of the cave. (Source: Adapted from Cane 1995:ﬁgs4.and11)

Further east around the coast, there appears to have been only minimal use of marine resources after 7 ka. After this time, most middens are small unstratified clusters of shell and stone artefacts, the largest containing only 1,700 shells – with a focus on Nerita and Austrocochlea periwinkles and occasional use of large individual abalone, limpets, turbo or whelks. There is little evidence of fish or marine mammals. Nicholson (1994) argues that scarcity of potable water along this coast kept population low and mobile (ethnographic population density is estimated as 1 person/40 km2), well below the threshold at which it would have become necessary to supplement a diet of rats and small marsupials with shellfish. Nicholson comments that these were ‘desert people living on the coast; with their backs to the sea’ (Nicholson 1994: 131).

The Desert Interior Away from the coast, hunter-gatherer societies responded to strengthening monsoon rainfall across the interior. A modest expansion of settlement is 178 The Archaeology of Australia’s Deserts

evident across the arid zone as early as 15–14 ka, with new patterns of site use at Puritjarra, Kulpi Mara, Cuckadoo 1 and JSN (see Table 6.1). Unlike settlement on the coast, this was a patchier, more gradual process of expansion, which appears to have begun around 15–14 ka and gained momentum around 12–11 ka, with the ﬁrst use of Puntutjarpa and Walga Rock. This was followed by a further increase in site visibility and occupation in the period 8–6 ka as inland climate stabilised. This early and mid-Holocene expansion of settlement is reﬂected in various ways:

r reoccupation of sites and locales abandoned during the LGM, r an increase in the number of occupied sites, r more intensive use of existing sites, r an extension of land use into peripheral or locally marginal areas, r more evidence for rock art (both paintings and petroglyphs), r and the earliest occupation deposits with artefact densities matching those in the late Holocene.

inland pilbara and western desert The Pilbara typifies this process. Field surveys in advance of iron-ore development in the Pilbara have identified a large series of sites with mid-Holocene occupation (see Table 6.1), but few have been investigated in sufficient detail to answer basic questions about the character of occupation, artefact assemblages and site inventories, or to provide detailed site chronologies. Most sites occupied during the LGM remained in use (e.g., Yirra, Cleft rockshelter, P0187, P2055, Malea), but sites abandoned during the LGM were not reoccupied until well into the Holocene. Use of the desert uplands and gorge systems increased around 10–8 ka, gradually incorporating a range of new sites, several with significant early to mid-Holocene occupation deposits (e.g., Marillana A, P5315). The most intensively studied site is Marillana A, where Marwick (2005) compared a range of sediment analyses (phosphorus, organic carbon, magnetic susceptibility, and soluble salt concentrations) with a study of stone artefact reduction and artefact densities. In combination, these showed that the rockshelter was used most intensively from 10.4 ka (9,200 ± 200 BP Wk-2682)to 3.9 ka (3,630 ± 70 BP Wk-2683), with more frequent visits to the site rather than a shift to a more sedentary pattern of occupation. At Bastion shelter (P53515), careful stratigraphic excavation uncovered a posthole associated with an occupation layer dated to 9 ka – suggesting that a brush windbreak had been constructed across the front of the rockshelter (Brown and Mulvaney 1983). Further south, in the Murchison and Wiluna areas, this general pattern is repeated. The first use of the rockshelter at Walga Rock (Bordes et al. 1983) may record an extension of settlement into a marginal area on the The ‘Desert Culture’ Revisited: Assembling a Cultural System 179 periphery of the ranges. Here, the earliest occupation horizon dates from 11.5 ka to 7.9 ka. Most other excavated sites in this region – Katampul, Gidgee, Mount East rockshelter – have occupation peaking at 5–4 ka, just prior to the appearance of small-tool phase assemblages. This pattern is also repeated in the Western Desert. An early Holocene expansion of land use is represented by basal occupation at Puntutjarpa (11.8–7.5 ka) and Bush Turkey 3 (12–8 ka), extending to include other sites around 5 ka ( Jalpiyari, Serpents Glen). puntutjarpa Puntutjarpa is the only large-scale excavation in the Western Desert. Warapuju soak (10–12 km from the site) is a major fall-back water (Tindale 1974: 70), but without a network of these wells, it is unlikely the area was habitable during the LGM. The first use of Puntutjarpa, at 11.8 ka (10,170 ± 230 BP I-5319), probably records an extension of settlement from the main block of the Central Australian Ranges into marginal country on their periphery. Our picture of events at Puntutjarpa is complicated by problems with radiocarbon dating of the site. Several alternative interpretations of the chronology are possible (Glover and Lampert 1969; Gould 1977; Johnson 1979: 131–4; Smith 1988). It is unlikely that these can be resolved without further excavation, so, in Table 6.2, I have simply given my interpretation of the evidence. The rockshelter deposit is divided by a layer of rock fall into an upper unit (zone A) and a lower unit (zones B and C). Initial use of the site (zone C) is represented by a thin lens of occupation and a single hearth, dated to 11.8 ka, sealed between a large rock slab and bedrock. The trend of the 14Cdates indicates that the main part of the deposit (zone B) built up rapidly around 7.6–7.5 ka, with deposition of a rich, dark archaeological soil containing a high density of stone artefacts (∼165 lithics/100yr/m3), red ochre, numerous intact hearths and animal bone. Shortly after 7.5 ka, part of the shelter collapsed, leaving the overhang structurally unstable; this resulted in a series of further (but smaller) falls over the next few millennia. Rock slabs within the rock fall had red or yellow ochre on their undersurfaces, suggesting the collapse of the shelter had brought down an older (mid-Holocene) frieze of rock paintings. This is supported by the find of a rock slab with four patches of yellow ochre – presumably also from this painted frieze – sealed in levels dating to 7.5 ka. Some field evidence suggested that this rock fall rested unconformably on older deposits (Gould 1977: 63, 171, 176), but Gould dismissed this possibility on the grounds that it was not supported by analysis of the stone artefacts. But, as Table 6.2b shows, the stratigraphic distribution of backed blades, tula adzes and seed-grinders (Smith 1986a, 1988: table 10.1; Hiscock and Veth 1991) supports the idea that there is a sharp break in the cultural sequence between zone A and the underlying deposits. People dug out a small seepage well in 180 The Archaeology of Australia’s Deserts

table 6.2. The Puntutjarpa sequence A. Radiocarbon chronology for Puntutjarpa rock shelter. Anomalous dates are shown in italics in a separate column. The excavator rejects I-3389 as contaminated. I-5476 is clearly out of sequence. Source: Gould (1977)

Depth Lab Zone cm 14CageBP code

A 13 <185 –I-3386 A 46 435 ± 90 –I-5320

B 74 6,740 ± 120 –I-3387 B 81 6,710 ± 125 –I-5475 B 108 – 4,010 ± 105 I-5476 B 108 6,590 ± 140 –I-3388

C 119 10,170 ± 230 –I-5319 C 155 – 3,810 ± 160 I-3389

B. The stratigraphic distribution of backed blades, tula adzes, ochre (Trench 2), and seed-grinders (Trenches 2 and 4). Based on reexamination of catalogued specimens held in the Western Australian Museum (see Smith 1986a, 1988 for details). Data are number of artefacts.

Geometric Tula Seed- Other Zone microliths adzes grinders grindstones Ochre

A 35 89 5 19 39 B 3a 3a – 146 C– – – 46

a These are all within a 3-inch spit at the base of A and may reﬂect artefacts introduced at this level during excavation and removal of the rock-fall layer.

an adjacent rockshelter (the south cave) around 4.2 ka so were still visiting the area, but the main Puntutjarpa site does not preserve signiﬁcant occupation deposits dating between 7.5 ka and 0.5 ka. Major use of the shelter resumed once sandy sediments buried the rock fall sufﬁciently to create a level living surface. A radiocarbon date of 435 ± 90 BP (I-5320) from a hearth covered by the last episode of rock fall indicates that zone A dates within the last 500 years. This phase of use of the shelter was more intensive than anything previously, with a richer, more diverse artefact assemblage containing seed-grinding implements, red ochre, adzes and geometric microliths and with concentrations of artefacts more than ten times that of the mid-Holocene levels (>2,000 lithics/100yr/m3)(Table 6.3). The ‘Desert Culture’ Revisited: Assembling a Cultural System 181 table 6.3. Comparative data on the size and diversity of selected assemblages, contrasting mid-Holocene (8–6 ka) and late Holocene levels (<4ka)

Sample Site Tool Lithicsa areab area Discard classesd Site Unit Age (ka) (n) (m2) (m2) ratec (n) Ochre Richnesse

Mid-Holocene Walga Rock Layer II 11.5–7.9 713 2 1,200 10 5 Y 2.1 Marillana A Unit 2 10.0–4.0 903 1 124 15 4 N 1.4 Sq 9b Wanga East Layer II 9.4–5.8 124 2 60 2 4 N 1.9 Puritjarra Unit 1c 8.3–3.81,656 12.5 400 3 5 Y 1.9 Puntutjarpa Tr 2 7.6–7.541,346 ∼25 ∼200 165f 6 Y 1.5 zone B Puntutjarpa Tr 2 7.6–4.541,346 ∼25 ∼200 53g 6 Y 1.5 zone B Skew Valley Layer II 7.4–6.6 489 1.548276 N 2.2 Sq F Late Holocene Walga Rock Layers ∼2.0–03,838 2 1,200 96 7 Y 2.8 1–5 Puritjarra Unit 1b 3.8–0.73,794 12.5 400 10 9 Y 2.8 Puritjarra Unit 1a 0.7–07,227 12.5 400 826 10 Y 2.9 Puntutjarpa Tr 2 0.5–025,576 ∼25 ∼200 2,046 10 Y 2.5 zone A Skew Valley Layer I 4.1–2.412,956 13 168 59 9 N 2.2

There is significant variability in site usage in the mid-Holocene, with some sites – such as Puntutjarpa – approaching late Holocene rates of artefact use and discard. In most, desert assemblage richness increases markedly after 4 ka with the addition of new tool types such as geometric microliths, adzes, ground-edge axes and seed-grinders. a Assemblage size is limited to artefacts from the 5–6 mm sieve fraction, where this is given. b Sample area is area analysed, which may not be the entire area excavated. c Discard rate is no. lithics/100yr/m2, to standardise this index for differences in sample area. d The number of tool classes gives an indication of the range of activities carried out at a site. Only a few categories are widely reported in site reports: cores, notched implements, steep-edged scrapers, amorphous retouched artefacts, pebble manuports or hammerstones, amorphous grindstones and unmodified flakes – with the addition of geometric microliths, adzes, ground-edge axes and seed- grinders in late Holocene assemblages. e Richness is a formal index of assemblage diversity (no. of tool types/log(sample size)) that reduces the effect of sample size. In this case, the sum includes the tool classes listed in the Tool classes column, plus ochre. To retain a degree of comparability, I have focused on nonperishable materials (stone and ochre) and excluded artefacts on wood, shell or bone found at some sites. f Discard rate is worked out on the basis that zone B spans a millennium. g Discard rate worked out using age estimates for zone B in Gould (1977). central australia and lake eyre basin In the Central Australian Ranges, the pattern of site use at Puritjarra changed sharply at 8.3 ka. Wanga East (W04) has a dark, charcoal-rich occupation layer dated to 9.4–5.8 ka, and Kulpi Mara was reoccupied from 5.4 ka. At 182 The Archaeology of Australia’s Deserts

both Puritjarra and Wanga East, the mid-Holocene occupation is associated with production of rock engravings (discussed in Chapter 7), suggesting not only more intensive site use but also structural changes in the way people constructed the cultural landscape. In the Lake Eyre basin, the only well-dated mid-Holocene sequence is from Cuckadoo 1 rockshelter, strategically situated in the upper reaches of the Georgina River catchment (Davidson, Sutton and Gale 1993). This has a major mid-Holocene occupation unit dated to 6.2–3.3 ka, with intact hearths, pieces of red and yellow ochre and heat-treatment pits for ochre and chert. Elsewhere, there are indications that people were using the sandy deserts and river systems surrounding Lake Eyre between 14 ka and 12 ka. This is from dated hearths along Cooper Creek (at White Crossing; Veth, Hamm and Lampert 1990) and in the Strzelecki dunefield (at the JSN site; Smith et al. 1991). But there is surprisingly little direct evidence for use of this region in the early to mid-Holocene. In the heart of the Simpson Desert, the earliest trace of occupation dates to 2.8 ka and is from a hearth associated with Marapadi well, one of a series of small mikiri wells that underpinned historic Wangkangurru occupation of this dunefield (Smith and Clark 1993). As the only site with a long continuous sequence spanning this period, Puritjarra has been analysed in some detail (Smith 2006, 2010). From 8.3 ka, people began to use the rockshelter more intensively; this coincided with the onset of conditions sufficiently humid to alter patterns of sedimentation and weathering in the overhang. There is a significant increase in artefact discard rates, more broken tools and broken cores, and both implements and cores are more heavily worked. Larger debris was cleared from the floor of the rockshelter, reflecting extended episodes of occupation. The production of the archaic rock engravings, found on boulders in the shelter, appears to date to the beginning of this period (see Chapter 7 for details). Grindstones become more common, indicating greater use of plant processing gear. There was more use of locally available ochres and a greater reliance on local sandstone for artefacts. Except for white chalcedony, the use of nonlocal stone sources declines or remains low: exotic silcrete virtually disappears from the sequence. All this points to more local provisioning and visits to the rockshelter that, on average, were more sustained, not simply more frequent. The pattern of site use changed again at 3.8 ka. After this time, the people visiting Puritjarra were using a more diverse tool kit, including hafted adzes and scrapers and other composite artefacts of wood, resin and stone, as well as specialised seed- grinding implements for processing grass and acacia seeds. The overall picture is of a further increase in use of this rockshelter after 3.8 ka, with a more complex mix of activities and site inventories.

flinders ranges and lake frome The most important excavated site in this region is Arkaroo Rock, which has a discrete mid-Holocene occupation unit dated to 7.7–6.7 ka (Draper 1989). The ‘Desert Culture’ Revisited: Assembling a Cultural System 183

Here, the low artefact densities and the absence of ochre and exotic stone suggest that the site was only used episodically for short visits prior to 6.7 ka (5,910 ± 90 C5739). This unit is overlain by a dense occupation deposit from 5.5 ka, which includes stone-lined hearths, ochre pellets, ochre-stained slabs, burnt animal bone and large quartzite block and flake implements (including steep-edged scrapers and notched implements). Sometime between 5.5 ka and 1.1 ka, the assemblage is supplemented by the appearance of flake adzes, geometric microliths and unifacial points, as well as by a high density of small (<10 mm) trimming flakes. Away from the ranges, a series of sites along the western edge of Lake Frome show an expansion of settlement into an area with highly ephemeral surface water. The main phase of occupation occurred during the early Holocene, with typological and stratigraphic evidence indicating only sporadic visits after about 5.5 ka (Lampert and Hughes 1988). The principal sequence is from Balcoracana Creek, where the main occupation horizon is marked by cores and flakes eroding from carbonate-rich sands, TL dated to 9.9 ka (Gardner et al. 1987). Other sites can be assigned to the same period on typological and geomorphic grounds – in particular, one at Big John Creek – but Balcoracana is the only site to have been investigated in any detail.

The Southeast Margins With contraction of desert margins, much of the Murray–Darling basin now falls outside the arid zone (and outside the scope of this book). However, some discussion is warranted before we leave it entirely. Rivers in the Darling basin show abundant signs of occupation associated with a return to high river discharge in the early Holocene. Radiocarbon dates on freshwater mussels indicate initial flooding of Teryawynia Creek (an anabranch of the Darling River) at 9.0 ka, with a progressive drying of the system over the succeeding 2,000 years. This reactivation is associated with 150 sites around the Teryawynia lakes, clustering in age between 9 ka and 7 ka. These are mostly shell middens and open campsites, many associated with expedient grindstones (63.9 per cent of sites have grindstones) (Balme 1991; Hope 1993). The Lake Nitchie burial also belongs to this phase of occupation. A well- defined grave pit contained the skeleton of a large, muscular man (187.7 cm tall, aged 37 ± 4 years), dated on bone collagen to 7.2 ka (6,820 ± 200 BP NZ1225) (Macintosh, Smith and Bailey 1970;Macintosh1971). As part of the burial rites, a small fire had been lit in the grave fill. The removal of the man’s central upper two incisors suggests ritual tooth avulsion. Remarkably, this man was buried with a large single-strand necklace of 178 pierced canine teeth from the marsupial carnivore Sarcophilus harrisii (Tasmanian devil), an animal extinct on the mainland since 3 ka. These represent teeth from at least forty-seven devils and were added to the string over the span of more than a 184 The Archaeology of Australia’s Deserts

century to judge from their condition and the different techniques used to drill the teeth. Macintosh (1971) argued that the necklace represented ceremonial regalia, noting that it was too fragile for general use but was carefully curated, with replacements and additions over several generations. It pointed, he said, to a lost corpus of mythology and ritual associated with Sarcophilus. Further to the southeast, Ross (1981) showed that populations occupying the Murray valley extended their range into the northern half of the Victorian mallee between 9 ka and 8 ka. A radiocarbon date of 8.5 ka (7,650 ± 110 yrs BP SUA 766) is available for one site on Raak plains (Ross 1981: 148). Other sites are assigned to this period on typological grounds and on their association with shallow lakes that were full in the early Holocene but dry after 7 ka.

FLAKED STONE ASSEMBLAGES Despite the rich record of early to mid-Holocene occupation, few site assemblages have been described in detail. Australian researchers have not been able to identify discrete temporal or geographical stone industries, irrespective of whether this is approached using typological studies, analyses of reduction sequences and flake production, size change in artefacts or selectivity in tool-stone (Holdaway 1995). In reviewing this period, Holdaway noted that ‘without some intermediate unit by which to group assemblages for analysis, Australian stone artefact studies have concentrated on intra-site change, frequently on variation in a single attribute state’ (1995: 793). This currently leaves us with a fragmented view of early to mid-Holocene lithic assemblages – their typology, taphonomy, composition and variability – with few attempts made to adequately characterise flaked stone material in discrete chronological or stratigraphic units. The lack of significant change in stone artefact assemblages for much of the continent’s history led Flenniken and White (1985) to suggest that Australian stone artefacts represented a single technological tradition, with a long history of expedient use and production. Much of the stylistic, technological and functional variability may have been vested in the organic component of the material culture, leaving stone artefacts as a largely unreflexive everyday practice.

Early to Mid-Holocene Assemblages The main arid-zone sites with large, well-dated ﬂaked stone assemblages for this period are Skew Valley (Lorblanchet and Jones 1979), Puntutjarpa (Gould 1977) and Puritjarra (Smith 2006) – with promising studies of smaller assemblages at Kulpi Mara (Thorley 1998a; Faulkner and Thorley 2009) and Marillana A (Marwick 2002a). Most desert sites show production of small, squarish ﬂakes (20–30 mm long) from lightly prepared platform cores or bifacial cores. The ‘Desert Culture’ Revisited: Assembling a Cultural System 185

Retouched artefacts include steep-edged scrapers, notched implements, saws and a range of amorphous retouched or utilised artefacts. Many of these forms are likely to represent stages in the resharpening of an implement (Hayden 1977a; see also Hiscock 1998 for a similar assessment). Table 6.3 shows that discard rates are an order of magnitude greater than those in late Pleistocene sites. The overall impression is that these early to mid-Holocene assemblages represent expedient ﬂake production for general-purpose maintenance tasks such as woodworking and cutting, but there is much still to be done to understand the structure and behavioural correlates of these assemblages.

Late Holocene Assemblages Late Holocene assemblages have attracted more analytical attention, largely because this period saw a continent-wide shift to a suite of smaller, finely worked hafted tools at 4.5–3.5 ka: with bifacial projectile points proliferating in assemblages in northern Australia, geometric microliths and other backed artefacts across the southern two-thirds of the continent, and hafted adzes and unifacial points in the arid interior. Late Holocene assemblages in the desert are a regional variant of the Aus- tralian ‘small-tool phase’ (Mulvaney and Kamminga 1999, chapter 14). This term is another legacy of Gould’s work at Puntutjarpa: he coined the term ‘Australian Small-Tool Tradition’ to defuse criticism of his artefact typology (1969b: 233), although evidence now indicates that ‘phase’ would be a better term than ‘tradition’. These desert assemblages contain a functionally hetero- geneous suite of new tools – spear armatures, projectile points, adzes and hafted scrapers – linked by a preference for fine-grained stone and their function as elements of composite tools (as stone bits or barbs in multipart implements made of wood, sinew and resin). Table 6.3 shows the increase in assemblage richness associated with the small-tool phase. Individual histories of the key artefact types (see the following discussion) show that these have different origins, some originating in the desert, others adopted from the different parts of the desert margin. geometric microliths Geometric microliths are small, backed artefacts, usually crescentic or triangu- lar flake segments, with abrupt blunting retouch along part or all of one side (Figure 6.7) (Holdaway and Stern 2004: 262–4). In the woodlands and along the seaboard of southeastern Australia, they are sometimes associated with blade industries (hence the name ‘backed blade’), although in the arid zone they were usually made on flakes rather than blades. These microliths served as prefabricated elements in a range of composite implements. Some studies show their use as barbs or armatures on spears (McBryde 1977; Kamminga 1980). The most dramatic evidence for this was the discovery near Sydney of the body of a man killed about 3.7 ka with three stone-barbed spears (McDonald et al. 186 The Archaeology of Australia’s Deserts

Figure 6.7. Small-tool phase artefacts from the late Holocene levels of Puritjarra rockshelter: 1–8, geometric microliths; 9–13, thumbnail scrapers; 14–15,endscrapers; 16–18, tula adze slugs; 19, wooden spearhead (with scarf joint at base).

2007). Other studies, mainly of use-wear and residues, show the use of some microliths as teeth in hafted knives (Robertson, Attenbrow and Hiscock 2009), similar to the ethnographic taap saw-knife (Hayden 1973). Backed artefacts occur sporadically in late Pleistocene contexts in northern Australia and Tasmania but are more common in early Holocene assemblages in southeastern Australia between 9 ka and 5.5 ka (Hiscock and Attenbrow 1998). By 4.6 ka, they had spread along the inland river systems (Figure 6.8) and ﬁrst appear in artefact assemblages across the arid zone between 4.1 ka and 3.6 ka (Table 6.4a). Hiscock (2002) coined the term ‘proliferation event’ to draw a useful distinction between the ﬁrst appearance of a type and its subsequent proliferation in assemblages. In the desert, the appearance of these artefacts coincides with their proliferation in southeastern Australia, and the distribution data indicate a point of dispersal in this area (much as Pearce 1974 originally suggested).

tula adze flakes and slugs Hafted adzes were the primary woodworking tool of desert people and were used for making boomerangs, spear-throwers, deep wooden bowls and a range of other wooden artefacts. Tula adze ﬂakes formed the cutting element in these composite tools, with the stone ﬂake mounted in a resin haft on a curved The ‘Desert Culture’ Revisited: Assembling a Cultural System 187

Figure 6.8. The spatial and temporal distribution of geometric microliths and other backed artefacts. Density contours (shaded) show (a) a zone with a large number of sites with microliths, often with 100–500 microliths/site; (b) a zone with multiple sites in each region, often with 10–50 microliths/site; and (c) a zone with isolated sites containing <10 microliths each, sometimes with atypical forms. Figures show the earliest dates in each region (in ka): and sites with microliths <4.5 ka (open triangle); sites with microliths 4.5–5.5 ka (small black triangle); and sites with microliths >5.5 ka (large black triangle). Open squares show sites with rare occurrences of backed artefacts in late Pleistocene levels. From GRE8, Slack et al. (2004) claim two backed artefacts dating to 15 ka. This is based on simple linear interpolation between radiocarbon ages of 4,071 ± 50 BP (Wk-1228)and16,249 ± 120 BP (Wk-12229), ignoring stratigraphic evidence for a hiatus between Holocene and Pleistocene units. Additional data (Slack 2007) suggest that the backed artefacts lie at the interface of these units and date shortly before 4.6 ka. The two microliths from OLH are more securely dated to 15.2 ka but are from a small test pit and require corroboration by other finds. The backed artefacts from Walkunder Arch Cave at 16.4 ka (Campbell 1984) are identified by the excavator as atypical. The Tasmanian examples (McNiven 2000) represent a separate anvil-based technology. (Sources: Base map updated from Mulvaney 1985 with density data updated from Pearce 1974, both with additional data from Smith 1977;Dortch1977; Johnson 1979; Hiscock and Hughes 1980; Campbell 1982, 1984; Horsfall 1982; Morwood and Godwin 1982; Smith and Cundy 1985; Hiscock and Hall 1988;McNiven1988;David 1990;Pickering1990;Veth1993; Hiscock and Attenbrow 1998, 2004; Slack et al. 2004) 188 The Archaeology of Australia’s Deserts table 6.4. Radiocarbon dates for initial appearance of backed implements and adze flakes by region A. Geometric microliths and backed artefacts

Region Site Age kaa Source

Coastal Pilbara Skew Valley midden 3.6b Lorblanchet and Jones 1979 Yardie Well ∼3.2 Morse 1993a Inland Pilbara Newman P2055 4.1c Brown 1987 Marillana A 4.0 Marwick 2002a Newman P0187 ∼3.2 Comtesse 2003 Murchison Walga Rock 1.0–4.2d Bordes et al. 1983 Billibilong Spring 4.1 Bordes et al. 1983 Western Desert Puntutjarpa ∼1.0e Gould 1977 Nullarbor Allens Cave 2.6–4.7d Cane 1995 Madura 3.7–4.8d,f Marun 1972 Central Australia Kulpi Mara ∼4.0 Thorley 1998b Puritjarra 3.8b Smith 2006 Ilarari 17 3.5 Smith 1988 Lake Eyre basin Cuckadoo 14.8c Davidson et al. 1993 a Excludes several outliers: 6.3 ka (Puritjarra) and ∼7.0 ka ( Juukan 2). The former may be intrusive. The latter relies on uncritical interpolation of age–depth plots and is not directly dated. b First appearance within a continuous stratigraphic sequence from a large excavation with good chronological resolution. c Microliths from dated hearth. d Age range for layer: lowest microliths need not fall at maximum age. e The published age estimate is ∼4.0 ka. Reassessment of site chronology suggests that the earliest geometric microliths appear somewhere between bracketing ages of 0.4–8.4 ka, probably at ∼1.0 ka. f There is also some uncertainty about typological identiﬁcations in this early work.

wooden handle (Roth 1904; Spencer and Gillen 1904: 637–9; Horne and Aiston 1924). This implement was associated with an elaborate chain of production. Tula adze flakes are semidiscoidal artefacts with a pronounced bulb on the ventral surface and a convex cutting edge profile (Holdaway and Stern 2004: 253– 6). Their production involved a specialised knapping process (Moore 2004) and a narrow range of tool-stone (mainly chert or chalcedony). Experimental studies show that the specific morphology of these flakes improves the function of the adze and prolongs the life of the hafting cement by reducing impact stresses (Sheridan 1979). During use, adze flakes were reduced through wear and resharpening to a distinctive ‘slug’ form (see Figure 6.7), which is the most common tool type in late Holocene assemblages across the arid zone. Large amounts of hafting cement were also required to keep an adze in good repair. This was widely available in several species of Triodia (‘spinifex’), the dominant hummock grass taxa in the desert. Spinifex had to be collected and beaten to collect resin dust, which was then cleaned by winnowing before The ‘Desert Culture’ Revisited: Assembling a Cultural System 189

B. Tula adzes or adze-slugs

Region Site Age ka Source

Victoria River District Ingaladdi 3.0 Clarkson 2007 Jagoliya <2.9 Clarkson 2007 Garnawala 22.5–3.1b Clarkson 2007 Coastal Pilbara Skew Valley midden ∼2.5a Lorblanchet and Jones 1979 Yardie Well ∼3.2 Morse 1993 Mandu Mandu 2.5 Morse 1993a Inland Pilbara Newman P2055 <4.1 Brown 1987 Murchison Walga Rock 1.0–4.2b Bordes et al. 1983 Billibilong Spring 2.0–4.1b Bordes et al. 1983;Dortch1984 Western Desert Kaalpi 2.2 Veth, Smith and Haley 2001 Puntutjarpa ∼1.0 f Gould 1977 Nullarbor Norina <1.0 Marun 1972 Madura <3.7 Marun 1972 Allens Cave <4.2 Marun 1972 Central Australia Kulpi Mara <2.6 Thorley 1998b Puritjarra ∼2.7a,c Smith 2006 Kwerlpe 2.8–4.0b, d Smith 1988 Lake Eyre basin Mucklandama Creek <1.0 Hiscock 1988b Native Well 1 ∼3.8 Morwood 1981 Lower Murray valley Devon Downs 3.7a,e Smith 1977 a First appearance within a continuous stratigraphic sequence from a large excavation with good chronological resolution. b Age range for layer: lowest tula adzes need not fall at maximum age. c Lowest tula associated with hearth dated 2,640 ± 70BP (ANU-6602). d Lowest tulas in 3-inch spit immediately above 14Cdateof3,635 ± 90 BP (I-7602). e Layer 7 containing ten tulas is dated by 3,460 ± 100 BP (Gak-1022), with a further twenty-eight tulas in layers 4–6. f The published age estimate is ∼4.0 ka. Reassessment of site chronology suggests that the earliest tulas appear somewhere between bracketing ages of 0.4–8.4 ka, probably at ∼1.0 ka. being heated and pressed into a usable parcel of thermoplastic cement. Up to 8 m3 of spinifex was required to produce 600 g of resin (Sheridan 1979). In areas where parcels of spinifex resin could only be obtained through trade, other plant resins often had to be substituted. These were mainly Grevillea striata or Lechenaultia divaricata, although sometimes Xanthorrhea, known to be poor and rather brittle cement, was used. The combination of Triodia spinifex resin (as a hafting cement) with a specialised adze-morphology (the tula adze ﬂake) is an adaptation for working eucalypt and Acacia hardwoods (Sheridan 1979). Unfortunately, the history of development of these implements is still sketchy. The earliest known tula adzes date to 3.8–3.7 ka, at both ends of the inland river system (Figure 6.9a). They appear in arid-zone assemblages around 3.0–2.5 ka, somewhat later than microliths, and in the Victoria River district in northern Australia 190 The Archaeology of Australia’s Deserts

Figure 6.9. Arid-zone adaptations. (a) The distribution of tula adzes (modiﬁed from Clarkson 2007: 4, and Mulvaney and Kam- minga 1999:map12a), showing earliest dates in each region in bold. Tula adzes are rare in sites along the arid west coast but do occur in archaeological deposits in the Northwest Cape area and at Shark Bay. Open squares show a site with atypical adzes, initially reported as tulas (McNiven 1993). (b) The distribution of unifacial pirri points. This shows the major distribution of these points (based on Campbell 1960, updated from Mulvaney 1969:ﬁg.24; Mul- vaney and Kamminga 1999:map12c; and Smith and Cundy 1985). Isolated points also occur in Central Australia, the Mann– Musgrave Ranges, and in north Queens- land, where they may have been obtained by exchange. Dated sites (in bold)show that these points appear about 5 ka at both ends of the inland river system – at Kenniff Cave (Mulvaney and Joyce 1965), Native Well 1 (Morwood 1981), Devon Downs (Smith 1977) and Fromms Landing 2 (Mul- vaney 1960) – and disappear by about 3.5 ka. (c) The distribution of large millstones (adapted from Davidson and McCarthy 1957). Dark grey area shows the distribution of ‘Panara’ grass-seed exploitation as suggested by Tindale (1974, 1977). Early dated sites (in bold) show that seed-grinding implements appear at 3.8 ka in Central Aus- tralia (Smith 2004) and on the margins of the inland river system by around 3.2 ka (Morwood 1981;Smith1986a). Also shown is Cuddie Springs (open square), where late Pleistocene seed-grinders are claimed (Ful- lagar and Field 1997).

at 3 ka (Table 6.4b). In both areas, the main phase of proliferation of these implements is after 2.5 ka. Their distribution (shown in Figure 6.9a) suggests that tulas most likely originated on the desert margin, on the headwaters of the Cooper–Barcoo–Diamantina system at the northeastern limit of resin-bearing Triodia species. Their spread south along the river systems must have involved The ‘Desert Culture’ Revisited: Assembling a Cultural System 191

Figure 6.10. Unifacial pirri points. Mulka, Lake Eyre basin. (After Mitchell 1949: ﬁg. 26)

substitution of other resins. Their rapid adoption across the desert interior at 3 ka, however, would have been facilitated by the widespread availability of high-quality Triodia resin. unifacial pirri points Unifacial pirri points are pressure-flaked unifacial points having a distinctive leaf shape with a round base, with invasive retouch extending to a central keel along the dorsal face (Figure 6.10) (Campbell 1960; Holdaway and Stern 2004: 266–7). The platform is usually trimmed away or thinned, perhaps to facilitate hafting. These implements, presumed to be projectile points (Akerman 1978), are mainly limited to the southern part of the Lake Eyre basin (Campbell 1960), with smaller forms found both to the south and east in areas surrounding the arid zone (see Figure 6.9b). Although these are one of the most distinctive desert tool types, their prehistory is largely unknown. Pirri points first appear in stratified contexts in the northeastern part of the basin between 4.9 ka and 5.3 ka (Mulvaney and Joyce 1965; Morwood 1981) and in lower Murray valley sites around 5.0–5.6 ka (Mulvaney 1960;Smith1977). In both areas, they form a narrow stratum in these sites and decline or disappear by 3.5–3.0 ka. Their distribution suggests an origin within the desert, probably around Lake Eyre and Lake Torrens. 192 The Archaeology of Australia’s Deserts

table 6.5. Typological changes in ﬂaked artefacts and ground-stone assemblages at Puritjarra rock shelter

Analytical unit Artefact type 1a–b 1c 2a 2b–d

Flaked stone artefacts Geometric microliths 46 1a –– Tula adze ﬂakes and slugs 33 ––– Burren adzes 3 ––– Thumbnail scrapers 29 ––– Endscrapers 14 1b –– Steep-edged scrapers 39 5 14 1 Notched implements 18 – 61 Saws 424– Other formal implements – – 1 – Amorphous retouched implements 181 28 37 1 Retouched fragments and trimming ﬂakes 78 12 13 2 Total number lithics (all classes) 11,021 1,656 1,594 303 Grindstones Seed-grinders 18 ––– Amorphous grindstones 16 3 2 – Grindstone fragments 26 5 2 1 Edge-ground tools Flakes from ground-edge axes 2 ––– Hammerstones ––2 –

Late Holocene (units 1a–b, 0–3.8 ka); mid-Holocene (unit 1c, 3.8–8.3 ka); early Holocene/terminal Pleistocene (unit 2a, 8.3–21.9 ka); late Pleistocene (units 2b–d, 21.9– 35 ka). a N6/15–1. b M10/11–1, from the uppermost spit of 1c. Source:Smith2004, 2006

changing assemblages at puritjarra Gould had intended that the notion of a small-tool tradition would shift the focus from a few tool types to a more rounded consideration of stone artefact assemblages. So what do we know about these changes at the site level? The Puritjarra sequence exemplifies the pattern: an assemblage containing geometric microliths, tula adze slugs, endscrapers, thumbnail scrapers and seed-grinders is limited to levels dating from 3.8 ka and is grafted onto an older suite of implements that include steep-edged scrapers, notched tools and saws (dentate implements) (Table 6.5). It is too early to gauge the impact on site use and regional settlement patterns. At Puritjarra, the small-tool phase coincides with increasing use of the rockshelter, with relatively high levels of breakage of cores and tools, of treadage and trampling and of artefact reduction. Most mobility indicators The ‘Desert Culture’ Revisited: Assembling a Cultural System 193 point to longer, more sustained use of the rockshelter (Smith 2006: table 19). However, the large increase in exotic stone and ochre signals a significant broadening of site catchment, reflecting a shift to logistical procurement at this time. We could expect new hafting methods to have a wide impact on a stone artefact assemblage (Keeley 1982), and this is indeed what we see at Puritjarra. Here, the thumbnail scrapers also appear to have been hafted – as small scrapers or gouges. These implements are small discoidal scrapers (see Figure 6.7) (Holdaway and Stern 2004: 234–5). At Puritjarra, many are too small to have functioned without some form of haft, and the pattern of wear suggests woodworking or the scraping of fibrous materials. The impact of hafting is also evident in changes in the use and breakage of steep-edged scrapers from 3.8 ka. Some are so small (<10 g) that they must have been hafted to have sustained such heavy step-flaking on their working edges, and most are made on chert or chalcedony after 3.8 ka. However, not all of the new tool types were hafted. Endscrapers are the best example of this. These are long flakes with a convex retouched edge on the distal end, often with finely invasive fluting (see Figure 6.7). Some have well-developed use-polish, with striations running transverse to the edge. This suggests they may be smaller versions of ethnographic endscrapers known as yilugwa or alywek in Alyawarr, which were used as spoons and scrapers for consuming fibrous roots and tubers (O’Connell 1974). Like seed-grinding implements (discussed herein), they suggest a broadening of the resource base around 3.8 ka. Although the small-tool phase is the major typological change evident in desert stone-tool sequences, this was not an abrupt replacement of one industry by another. At Puritjarra, the underlying flake and core technology remained unchanged. None of the new implements involved fundamental changes in flaking techniques. For instance, none was associated with the production of blades or other specialised blanks (except perhaps tula adzes, which were not manufactured at the rockshelter). Nor was there a sharp change in selection of tool-stone at 3.8 ka. At Puritjarra, a shift from silcrete in the late Pleistocene and early Holocene to chert and chalcedony in the late Holocene is the product of a long-term trend beginning around 8.3 ka rather than a stadial change associated with the small-tool phase. Finally, not all of the new tool types appear at the same time: tula adzes appear later in the sequence than the other new types. Mulvaney’s suggestion, made nearly 50 years ago, that these are best characterised as hafted assemblages (Mulvaney and Joyce 1965)isclosertothe mark than most of the alternatives. The evidence suggests that increased use of thermoplastics as hafting cement brought lithics into closer articulation with the wooden tool kit (as stone components or bits in wooden composite tools). This is the proximate cause of much of the functional variability seen in the 194 The Archaeology of Australia’s Deserts

table 6.6. Inventory of wooden artefacts used by ethnographic groups in Central Australia

Subsistence Throwing sticks Digging stick Wooden bowls, dishes, and scoops Winnowing trays (for grain) Wooden spatula (for Solanum fruit) Fluted hunting boomerang Three-piece composite hunting spear Spearthrower (dished type or flat ovate type) Hafted stone axe Maintenance tools Hafted adzes Small hafted engravers and gouges Simple spindle (for hair-string) Weapons Large fighting clubs Hooked (No. 7) fighting boomerang Stone knife with wooden handle Hand-held one-piece fighting spear Hafted fighting pick Softwood parrying shield Sacra Incised wooden tjuringa boards Clapsticks (music and ceremonies)

Sources: Stirling (1896), Spencer and Gillen (1899, 1904), Brokensha (1975), Latz (1995)

small-tool phase. It suggests that the small-tool phase is not so much a signiﬁ- cant change in stone-tool technology as a fundamental change in the suite of wooden implements.

The Wooden Tool Kit Ethnographic groups in the desert used a range of wooden implements, including composite three-piece hunting spears, spear-throwers, digging sticks, hafted adzes, fluted boomerangs and several types of deep wooden dishes and winnowing trays (Table 6.6). In general, this wooden tool kit was more diverse in areas with higher population densities (Central Australia, the Lake Eyre basin and northern Central Australia) and more streamlined and portable in the Western Desert and Mann–Musgrave Ranges. Many of these implements were purpose-built: hafted adzes for woodworking, deep dishes for carrying water, winnowing trays for cleaning grain and boomerangs for hunting. Others were designed for efficiency in construction and repair. The major composite implements in this suite were hafted adzes and three-piece hunting spears (with a detachable tail shaft and a separate hard- wood mulga blade). The design of composite hunting spears was constrained The ‘Desert Culture’ Revisited: Assembling a Cultural System 195 by available lengths of the preferred spearwood (Pandorea vine) and by the need for a modular construction allowing easy repair (Cundy 1989: 95–104). The dry desert climate caused longitudinal cracking in spear shafts, which had to be replaced every 3 weeks (i.e., seventeen shafts per year; Gould 1978b). Tail shafts frequently broke under the strain imposed by a spear-thrower. Typically, an adze flake might need to be replaced more than twenty times over the course of a year (Gould 1978b). There is little direct evidence on the history of this wooden tool kit. Mulga spearblades are found on archaeological sites across Central Australia (see Figure 6.7) but rarely survive in deposits older than a few hundred years. Hafted stone axes are known to be a recent addition to this tool kit (see Chapter 8), whereas spear-throwers may have a history prior to 4 ka because they are shown in early to mid-Holocene rock paintings in northern Australia (Lewis 1988). However, a case can be made that the small-tool phase is, at least in general terms, the archaeological correlate of this wooden assemblage: r All of the new types (except pirri points) continued in use into recent precontact times. r Spinifex resin was intrinsic to the construction and repair of these wooden implements, both as hafting cement and as an adhesive, and is often found in late Holocene contexts (e.g., Thorley 1999; Veitch, Hook and Bradshaw 2005; Dias, Veitchand Hook 2006). The earliest direct evidence is from cement on a geometric microlith in the Pilbara, dating to 4.1 ka (Brown 1987: 27). r The appearance of a specialised adzing tool (the tula) marks a significant increase in the turnover of wooden implements. Spears, mulga spear blades, spear-throwers and wooden dishes were all major consumers of lithic material. By 3.0 ka, the scale of woodworking must have exceeded what Hayden termed a ‘threshold of lithic economy’ (1977b: 89): the point at which the high consumption of tool-stone makes it economical to adopt a specialised woodworking tool. r Small stone points (such as pirri points) are intrinsically linked to use of a spear-thrower because they are too light to provide balance and stability in hand-thrown spears (Akerman 1978; Cundy 1989). r Wooden winnowing trays and dishes were central to the harvesting and cleaning of grain, an aspect of the subsistence economy that becomes archaeologically visible with the appearance of seed-grinders at 3.8 ka (see the next section). All of this suggests that the small-tool phase marks a fundamental broadening of the wooden tool kit towards a more diverse assemblage, with greater emphasis placed on curated rather than impromptu implements, and often with a modular design organised around the use of thermoplastic cements and adhesives. 196 The Archaeology of Australia’s Deserts

A Technological Response to Risk? Hiscock (1994) argued that the small-tool phase is best understood as a technological response to foraging risk in the late Holocene. The proliferation of backed artefacts, adzes and points after 4 ka represents a portable, standardised and multifunctional tool kit that was adopted to offset ENSO-driven environmental variability in the late Holocene (Hiscock 1994; Attenbrow, Robertson and Hiscock 2009; Veth et al. 2011). In studies of technology, ‘foraging risk’ is usually defined as the probability of not meeting dietary needs plus the cost of that failure (Torrence 2001). For hunter-gatherers, there are three main types of response. First, risk can be mitigated through mobility. Mobility, however, is not a single variable: changes in residential mobility (shifting camp to relocate to another foraging patch) need to be distinguished from changes in logistical mobility (wider- ranging or more frequent hunting trips, or task-specific trips to acquire wood, tool-stone or resin). Second, risk can also be offset by storing food. In the arid zone, small-scale stockpiling of bush foods was effective in dealing with predictable seasonal shortfalls, but not with environmental stochasticity at the decadal or interannual scale (where shortfalls are unpredictable and may last several years). A third strategy is to invest in tools and technology designed to reduce the probability of failure. This might take one of three forms: more efficient, special-purpose tools; tools that support an increase in diet-breadth; or tools that are more reliable and more readily repaired. The small-tool phase conforms to this third strategy, but there are some important riders to the ‘risk’ model. First, historical context is important. Population growth is a crucial corollary of late Holocene ‘risk’ models, without which they make little sense. The shift to an ENSO-dominated climate in the late Holocene was only a modest environmental change compared with the abrupt, high-amplitude climate shifts at the end of the Pleistocene. Nor are there grounds for thinking that late Pleistocene groups were less mobile, or faced less environmental uncertainty or confronted fewer challenges in colonising new territory, than did late Holocene desert populations. The problems posed by an ENSO-dominated climate were accentuated precisely because these changes took place at the end of a long period of expansion of settlement in the interior. Around 5–4 ka, desert populations were at their highest level since the last glacial. Search costs for game are likely to have been increasing, and the costs of simply relocating to another foraging patch or hunting ground must also have been higher than in earlier periods. In these circumstances, adoption of a standardised, easily maintained, and more efficient tool kit is worthwhile even if the new tools require a more protracted and complicated process of production. Second, to understand the technological response, we need to look at the entire tool kit, not simply the stone components. Material culture inventories The ‘Desert Culture’ Revisited: Assembling a Cultural System 197 tend to increase as settlement mobility declines. Table 6.6 shows a diverse, curated tool kit, with implements purpose-built for specific functions, more suited to logistical than residential mobility. As an assemblage, it is not especially portable: explorers’ journals note that many wooden implements were left on-site when people moved camp. The array of wooden dishes and bowls for carrying water and collecting grain, grubs or fruit indicate greater investment in logistical mobility, field processing and transporting resources back to camp. The demands of the new tool kit also suggest frequent task-specific trips to acquire spearwood, high-quality tool-stone or spinifex resin. Some individual items, such as composite hunting spears and spear-throwers, are comparatively light and modular in construction, making them portable and easy to repair. Their adoption may well indicate increased mobility but, in the context of the entire tool kit, this is likely to have been logistical mobility (more frequent and longer hunting trips) rather than residential mobility. This is in line with computer simulations that demonstrate that logistical mobility reduces subsistence risk in hunting economies (Grove 2010).

SEED HARVESTING AND THE GROUND-STONE ASSEMBLAGES The harvesting of wild grain from acacias, grasses and forbs was one of the most distinctive patterns of resource use in Australia’s deserts (Gould 1969a; Tindale 1977; O’Connell, Latz and Barnett 1983;Cane1987). Current archaeological evidence indicates that this was another late Holocene development. In ecological terms, granivory is a common strategy in world deserts because seeds are diverse and abundant (Brown, Reichman and Davidson 1979). Ephemeral forbs and grasses rely on rapid growth and seeding to take advantage of unpredictable rainfall (Sowell calls it ‘a quick and dirty approach to plant development’ [2001: 51]), and these plants use delayed germination to bridge long dry spells. Perennial plants also invest heavily in seeds as a strategy for competitively occupying bare ground between established vegetation, and they use synchronised seeding to offset losses to granivorous species (mainly birds and ants in Australia’s deserts). For desert hunter-gatherers, therefore, grass and acacia seeds provide an abundant carbohydrate-rich form of grain (typically 50–70 per cent carbohydrate and 10–20 per cent protein). Of the 140 food-plant species used by ethnographic groups in Central Australia, 67 (47.9 per cent) were harvested for their seeds (Latz 1995). Grass seeds were usually stripped by hand into wooden dishes, husked by rubbing or by treading in pits and cleaned using winnowing trays. Other seeds were beaten or shaken from the plants (Brokensha 1975; Tindale 1977; O’Connell et al. 1983;Cane1984;Latz1995). Most seeds required grinding into ﬂour or paste before consumption. Some, like mulga (Acacia aneura), were parched in hot soil before grinding into a paste with water. Grass and portulaca seeds were directly ground with water into a paste and cooked as seed-cakes. 198 The Archaeology of Australia’s Deserts

Despite their abundance, seeds are relatively high-cost resources, requiring on average 6–6.5 hours/kg handling time (collecting, cleaning and processing). The bulk of this is taken up with grinding the seeds: it takes 2–6 hours to produce a kilogram of flour from most seeds (Brokensha 1975: 25; O’Connell and Hawkes 1981: 124–5; O’Connell et al. 1983: 92;Cane1987). Although grass-seed harvesting particularly caught the imagination of scholars, it is important to note that neither wet-milling nor intensive exploitation of native grain was limited to grasses. Only five of sixteen high-ranked seed- food species listed by Latz (1995) are grasses. Of the five seed-food staples in Central Australia, Acacia aneura (mulga) is a tall perennial shrub, Portulaca oleracea (munyeru) is an annual succulent, Tecticornia verrucosa is a chenopod, and only Eragrostis eriopoda and Panicum decompositum are grasses. In the Great Sandy Desert, one of the most important seed plants is a sedge, Fimbristylis oxystachya (Cane 1987). Tindale’s notion of a ‘Panara or grass seed culture’ (1974, 1977) limited to an arc of semi-arid grasslands on the desert margin (see Figure 6.9c) is often cited but is misleading. Modern vegetation maps do not show a grassland belt that corresponds to his ‘panara’ culture (Atlas of Australian Resources – Vegetation 1990). And harvesting of grass and acacia seeds was a major component of ethnographic subsistence patterns across a much wider area: in the Pilbara and Central Australia and across the Western Desert.

Seed-Grinding Implements Current knowledge of the archaeology of seed-based economies rests on the temporal distribution of the main types of seed-grinding implements – mortars, millstones and mullers (Figure 6.11) – whose function, morphology and distinctive use-wear are outlined in various studies (Davidson and McCarthy 1957; O’Connell 1977;Smith1985, 1986a;Cane1989):

r Millstones were used in the wet milling of seeds. These base-plates are large ﬂat-surfaced slabs with one or more long shallow grooves worn into the surface. The portable slabs also have a counterpart in the bedrock milling grooves and grinding patches associated with desert sites. r Mullers are the hand-sized topstones used with millstones. Their form varies depending on whether the blank was a cobble or a tabular piece of sandstone, but they always have a distinctive ground facet (Figure 6.11). Both millstones and mullers are progressively reduced throughout their use-life, before reaching the heavily worn forms eventually discarded (Smith 1985;Cane1989). However, diagnostic wear patterns develop quickly, and seed-grinders can be distinguished from other types of grindstone at various stages in their use-life, not simply as worn end products. The ‘Desert Culture’ Revisited: Assembling a Cultural System 199

Figure 6.11. Seed-grinders from Central Australia. (1) This multigrooved millstone shows residual rejuvenation pitting around the milling surfaces (arrowed) and has a ﬂaked edge. (2)–(4) Mullers (facetted handstones), showing some of the range of variation of these implements.

r Mortars are less common in excavated assemblages. These are sandstone blocks with a shallow oval or circular basin ground into one or both faces. These implements were used with pestles to pound very hard- coated seeds (mainly acacias) into a coarse meal, as a preliminary stage of processing before wet milling. r Expedient grindstones are also common in arid-zone assemblages. These are amorphous slabs with uneven wear and abrasion, often limited to small areas of the surface.

Despite a major increase in archaeological work in the arid zone since the chronology of seed-grinding was initially reviewed (Smith 1986a, 1989a; Cane 1989), grindstones with the distinctive patterns of wear associated with the wet milling of seeds have only been found in late Holocene contexts. Expedient grindstones occur in Pleistocene assemblages and become more 200 The Archaeology of Australia’s Deserts

common in early to mid-Holocene assemblages. Specialised seed-grinding implements appear around 3.8 ka in Central Australian assemblages (Smith 2004) and in the headwaters of the arid river systems by about 3.2 ka (Morwood 1981;Smith1986a;seeFigure 6.9c); and their main phase of proliferation in desert assemblages is within the last 1,500 years.

Problems and Prospects On the southeastern margins of the arid zone, some fieldwork challenges this late Holocene chronology. Balme (1991) reviewed artefact assemblages eroding from open sites around rivers and lakes in the Darling basin, suggesting that seed-grinding implements may have been introduced in this region in the early Holocene (8–7 ka). However, none of the grindstones were found in stratified deposits, and most sites were palimpsests of occupation of different ages. Other evidence has come from Cuddie Springs (discussed in Chapter 4), where the excavators argue that a seed-based economy was present in this region by 35– 30 ka (Fullagar and Field 1997). The evidence here is not persuasive because as most of the grindstones are from a deflation pavement that forms an erosional interface with highly disturbed Holocene sediments. If further work confirms the early development of seed-based economies in the Murray–Darling basin, it will also show that this region has a history different from the desert interior. Only Puntutjarpa and Puritjarra have been systematically investigated for carbonised plant remains (Gould 1977: 165;Smith2004). The results are promising, but direct seed remains are rare and so far have not added to our picture of seed-grinding economies. Elsewhere, excavations at Carpenters Gap 1 on the northern margin of the arid zone show a concentration of burnt chenopod seeds in occupation levels dating 26–18 ka. Taphonomic assessment suggests that the seeds are associated with human use of the rockshelter, but no grindstones are present in these levels (McConnell 1998; McConnell and O’Connor 1997). People at Carpenters Gap may have been harvesting chenopod seeds but, if so, the processing methods appear to have been different (chenopod seeds can be popped) and may not have involved either grinding or ground-stone implements. Have earlier seed-grinders simply been missed or misidentified? Although sampling issues have been highlighted by a number of commentators (Edwards and O’Connell 1995; Gorecki et al. 1997), the large number of excavated early to mid-Holocene and late Pleistocene sites makes it unlikely that earlier seed-grinding implements have been missed. Nor is there evidence that the predominance of rockshelter excavations has skewed this chronology. Large excavations at rockshelter sites have generally yielded large assemblages of grindstones: at Puritjarra (N = 90), Puntutjarpa (N = 29) and Intirtekwerle ( James Range East) (N = 104). The ‘Desert Culture’ Revisited: Assembling a Cultural System 201

Several studies have highlighted problems associated with applying typology to highly fragmented archaeological specimens (where it is not uncommon to find that 40 per cent of grindstones are too fragmented to classify). In such circumstances, analysis of residues (especially starch and phytoliths) and microwear has considerable potential to maximise classification of excavated material (Fullagar and Field 1997; Gorecki et al. 1997; Veth, Fullagar and Gould 1997). Balme et al. (2001) reexamined the grindstones from Puntutjarpa and identified starch and use-polish on most implements – including several from the mid-Holocene levels of this site – but could not identify the starch or demonstrate it was use-related. The major problems are methodological, both in identifying residue pathways and in scaling up analysis to look at a complete ground-stone assemblage. Starch is common in arid-zone soils and the presence of starch grains does not make a seed-grinder, and a few grindstones with starch do not make a seed-based economy. A reflective polish is often taken to indicate wet milling of seeds, but experimental work shows that this polish can be produced in other ways (e.g., by working bone or processing nuts), as Dubreuil (2004) found in his study of Natufian grindstones. There is clearly a need to establish what residues and microwear occur on ethnographic millstones as a baseline for interpreting fragmentary archaeological artefacts. One of the few studies to directly examine residues on ethnographic Australian millstones – by Stephenson (2011) – found that even these have complicated sets of residues: some related to the primary function of the implement; others reflecting secondary or more impromptu usage; and some residues (e.g., bat hair) that are unlikely to be use-related.

The Puritjarra Grindstone Assemblage In excavated assemblages, millstones are mostly represented by fragments (including sections of the ground groove, the rim or the median keel between grooves). At Puritjarra, the ninety excavated grindstones include twenty-one artefacts identified as pieces of seed-grinders. All the seed-grinders show signs of heavy use with a high rate of breakage. The earliest seed-grinders (parts of mullers) appear at 3.8 ka (Smith 2004). Many other ground-stone implements – and broken and undiagnostic pieces of grindstone – were also recovered from the late Holocene levels of this site (see Table 6.5). The latter include sandstone trimming-flakes and pieces of broken grindstone (some of which may be heavily recycled seed-grinders). This mirrors the pattern of other late Holocene sites: seed-grinding assemblages typically contain large quantities of grinding material, broken grindstones, trimming flakes and amorphous grindstones, as well as identifiable seed-grinders. Analysis of the temporal distribution of grindstones at Puritjarra shows a number of trends (Smith 2004). First, grindstones (irrespective of type) are not 202 The Archaeology of Australia’s Deserts

common at this site until after 8.3 ka. Second, there is a signiﬁcant increase in the relative frequency of grindstones (of all types) in implement inventories in the late Holocene levels. Third, within the grindstone assemblage, there is a shift from amorphous (i.e., expedient) grindstones to seed-grinders. These data indicate a suite of interrelated changes at Puritjarra, involving more intensive use of the rockshelter over time, greater overall use of plant processing gear such as grindstones per se and changes in the type of grindstones, with seed-grinders appearing from 3.8 ka.

Development of Seed-Grinding Economies The appearance of seed-grinders and the increasing importance of grain in hunter-gatherer subsistence economies is likely to reﬂect the interplay of several factors, including population growth, environmental stress and changes in residential mobility. Numerical modelling in behavioural ecology (Winter- halder and Kennett 2006) suggests that extended use of key sites (increasing patch residence time) becomes more probable as population growth increases the costs of relocating. A corollary of this is that use of acacia and grass seeds (which have high processing costs in comparison with other bush foods) will increase as higher ranked bush foods are locally depleted (O’Connell and Hawkes 1981). These pressures were sharpened by the switch to an ENSO- dominated climate after 4 ka, particularly in circumstances in which residential mobility could not be the sole response to subsistence shortfalls. In this context, the harvesting and processing of wild seeds represents a broadening of diet breadth to include labour-intensive bush foods in a well-peopled landscape (where hunter-gatherer groups were only a day’s walk apart). Although there are no theoretical grounds for ruling out a shift to seeds during earlier periods of environmental stress in the late Pleistocene (Edwards and O’Connell 1995), the paucity of grass and the fragmentation of acacia shrublands during the LGM would have signiﬁcantly increased the collection costs of seeds and made grain-based economies unviable. Excavated ground- stone assemblages also indicate that these seed-based economies appear too late to be simply a response to new resource opportunities, such as a post- Pleistocene increase in grasslands (Tindale 1959: 49–50).

OTHER MID-HOLOCENE DEVELOPMENTS The desert interior was an open system, often affected by distant events outside the arid zone. The mid-Holocene saw several other developments that, each in its own way, transformed life in Australia’s deserts. These include the proliferation of Panaramitee-style rock engravings (discussed in Chapter 7), the replacement of desert languages by languages of the Pama-Nyungan family and the dispersal of the dingo across the continent. The ‘Desert Culture’ Revisited: Assembling a Cultural System 203

Figure 6.12. The distribution of Australian language families, showing the Pama- Nyungan (PN) languages (grey area) and the location of Karrwan and Tangkic. Tree diagram shows relationship between PN language family and the northern languages. (From Evans 2005)

The Adoption of Pama-Nyungan Languages The distribution of Australian languages shows marked asymmetry. The north and northwest of the continent is a diverse, deeply etched linguistic mosaic made up of twenty-seven unrelated language families. In sharp contrast, a single language family – the Pama-Nyungan (PN) – with numerous closely related languages, blankets the rest of the continent, including the whole of the desert interior (Figure 6.12). Most Australian linguists see PN as a clade, derived from non-PN languages on the southern margin of the Gulf of Carpentaria and spreading ﬁrst into Central Australia and the Lake Eyre basin in the early to mid-Holocene and then into the western half of the continent (where the Nyungic languages are all younger) (McConvell 1996; Evans and Jones 1997; Bowern 2010). An alternative view is that PN is not a genetic grouping but rather a remnant of an older linguistic landscape overprinted and homogenised by intense borrowing and linguistic diffusion (Dixon 2002). Recent research on comparative linguistics, however, has tested and conﬁrmed the proposition that PN languages form a discrete historical grouping (Bowern and Koch 2004; 204 The Archaeology of Australia’s Deserts

Evans 2005). The north and northwest of the continent also have the greatest deep-level diversity (at family level) consistent with a centre of dispersal. Crucially, Evans and Jones (1997) have also shown that PN languages can be derived from one of the northern non-PN languages. This process of differentiation also gave rise to some non-PN languages, such as Karrwan and Tangkic, whose speakers now live around the southern part of the Gulf of Carpentaria (see Figure 6.12). An argument by Clendon (2006) that PN represents an east–west recolonisation of the continent, following depopulation during the LGM, takes an extreme view of events not consistent with either the archaeological or the palaeoenvironmental evidence (see Chapter 5). Setting aside the counterarguments, the evidence suggests that some time during the early Holocene, an existing linguistic mosaic in the desert was erased as PN languages spread across the interior. The timing of these events is only loosely constrained to between 15 ka and 4 ka. McConvell (1996)givesdatesof 6–4 ka for the initial break-up of PN, based on guesstimates of the minimum time required to accommodate the sequence of linguistic changes. Standard glottochronology suggests 8,000 years of separation between the youngest and oldest PN languages (based on 6 per cent of cognates shared between Pintupi and Yanyuwa) (O’Grady 1996). Linguists have assumed that the real figure might be half of this (4,000 years) because of high rates of lexical change in ethnographic groups (driven by word avoidance following deaths). But it may not be valid to project this practice into the past, as some recent work suggests much slower rates of lexical change. Some archaeological evidence supports the idea that the initial break-up of PN took place in the early Holocene and that McConvell’s estimates should be regarded as minimum ages. For instance, the mosaic of small, unrelated language families in northern Australia is contiguous with a part of the coast with a uniquely large continental shelf. Here, the last marine transgression removed nearly 1.6 million km2 of land, most of which had been lost by 11.5 ka. From 12.2 ka to 10.5 ka, rising sea level transformed Lake Carpentaria into a marine gulf, flooding the lowlands surrounding the freshwater lake. The correlation between the distribution of non-PN language families and the geography of the continental shelves suggest that the linguistic diversity in northern Australia reflects the displacement of populations on the Sahul and Northwest shelf. If so, the differentiation of the Karrwan-Tangkic-PN group was probably a vicariant process driven by the final stages of the flooding of the Gulf of Carpentaria around 10 ka. At the other end of the timescale, a terminus ante quem is suggested by the role of the PN/non-PN divide as a social boundary. Neither geometric microlithics (to the south) nor bifacial points (to the north) significantly crossed this boundary, indicating it must have been established before 4.5 ka (when these artefacts proliferated in tool assemblages). Some linguists have attempted to directly link the spread of small-tool phase The ‘Desert Culture’ Revisited: Assembling a Cultural System 205 artefacts and PN languages, but this is unlikely, as the distribution maps in Figures 6.8 and 6.9 show. If PN languages spread across the desert interior at about 8 ka, what drove this expansion? It is very unlikely that it was simply a case of large-scale migration and population replacement because this would require the replacement of existing hunter-gatherer groups across two-thirds of the continent, including those in the densely occupied Murray–Darling basin. The alternative is that it involved a language switch, in which existing desert groups adopted PN languages from their neighbours. Language shift begins with multilingualism, which is a boundary phenomenon. If the last marine transgression was the punctuation event that initiated the explosive expansion of one small language family, this may initially have been a consequence of accelerated social interaction generated when displaced groups were forced into close contact. The subsequent spread of PN languages occurred at a time when population levels in the interior were beginning to stabilise after a protracted period of rapid climatic and environmental change during the Pleistocene–Holocene transition. It also shows a broad pattern of transfer from high-density populations to less densely populated regions – from the Carpentarian lowlands into the Lake Eyre basin and Central Australia, and then into the western half of the continent (McConvell 1996) – thus supporting the notion that language may have formed part of the ‘ritual engine’ of major ceremonial transfers (Evans and Jones 1997; Gibbs and Veth 2002). A final point to be made concerns the likely identity of these populations. Evidence on the genetic distance between various Aboriginal populations does not support the idea of large-scale population replacement in the desert. Research on blood-group antigens, serum protein systems and enzyme groups shows that populations in the Western desert, Central Australia and Port Hed- land are genetically distinct from populations in the Kimberley region, Arn- hem Land and north Queensland (Kirk 1971; Balakrishnan, Sanghvi and Kirk 1975). The genetic distance is sufficient for Kirk to conclude that it ‘implies a long period of isolation without the benefit of fresh additions from outside. Indeed, the evidence from the gamma globulin group system indicates the complete absence of gene flow from the coast in toward the desert regions of the interior’ (1971: 335).

The Dingo The dingo (Canis lupus dingo) is widespread across Australia’s drylands, both as a wild population and as a semidomesticated member of hunter-gatherer groups across this region (Corbett 1995). On the fossil remains, dingoes had spread across the continent by 3.1 ka and possibly as early as 3.7 ka, with the earliest evidence coming from Madura Cave on the Nullarbor Plain (Table 6.7). 206 The Archaeology of Australia’s Deserts

table 6.7. Radiocarbon dates for initial spread of the dingo (Canis lupus dingo) across Australia

Locality Canid remains Age ka Source

Kimberley Ngurini Skeletal elements 3.0 (1) Nullarbor Madura Cave Skeletal elements 3.7 (2) Thylacine Hole Desiccated animal 2.2 (3) East coast Wombah midden Skeletal elements 3.5 (4) Balmoral Beach 2 Skeletal elements 3.0 (7) Blue Mountains Capertee 3 Skeletal elements 3.0 (5) Murray-Darling basin Fromms Landing 6 Articulated skeleton 3.4–3.1 (6)

Sources:(1)Veitch1996, 2,820 ± 90 BP (Wk-1615); (2) Milham and Thompson 1976, 3,450 ± 95 BP (ANU-807); (3) Lowry and Merrilees 1969, 2,200 ± 96 BP (NSW- 30); (4)McBryde1982, 3,230 ± 100 BP (Gak-568); (5) Johnson 1979, cited in Gollan 1984, 2,865 ± 57 (V33); (6) Mulvaney, Lawton and Twidale 1964, bracketed by ages 2,950 ± 91 (NPL-28)and3,170 ± 94 (NPL-29); (7) Attenbrow 2002: 167, <3,310 ± 50 (Beta 63204).

Dingo remains are generally absent from bone deposits older than 5 ka. The fossils, however, are too few and widely dispersed to provide more than a loose terminus ante quem for the introduction of this species to Australia. A recent study of mtDNA indicates separation of dingoes from domestic dogs in east and southeast Asia somewhat earlier, between 5.4 ka and 4.6 ka, and that the dingo represents a ‘unique isolate of early undifferentiated dogs’ (Savolainen et al. 2004: 12390). These mtDNA age estimates may be too old if the founder group was genetically isolated in southeast Asia prior to arrival in Australia, or if they had greater genetic diversity than has been estimated. Combining the fossil and mtDNA evidence would suggest a working estimate of about 4.6 ka for dispersal of the dingo across the continent. The genetic evidence points to a single founding event, involving a small group of individuals at the end of a long chain of migration and population bottlenecks. This supports a case for the introduction of the dingo on the coast of northern Australia during the maritime expansion of Austronesian agricultural groups south through the Indo-Malaysian archipelago around 5.0– 4.5 ka (Higham, Kijngam and Manly 1980; Spriggs 1989) – perhaps as another example of the deliberate animal translocation practised by these groups. The The ‘Desert Culture’ Revisited: Assembling a Cultural System 207 table 6.8. Radiocarbon dates for the last Thylacines (Thylacinus cynocephalus)onthe Australian mainland

Locality Thylacine remains Age ka Source

Nullarbor Thylacine Hole Desiccated animal 5.2 (1) Madura Cave Skeletal elements 3.7 (2) Murra-el-elevyn Desiccated animal 3.5 (3) Kimberley Widgingarri 1 Skeletal elements ∼5.0 (4) Tunnel Cave Skeletal elements modern? (5) Northwest Cape MonajeeCaveCR21 Skeletal elements (?4.0)(8) Southwestern WA Murray Cave Skeletal elements 3.3 (5) Murray–Darling basin Fromms Landing 2 Skeletal elements 4.3–4.1 (6) Devon Downs Skeletal elements 3.8–3.2 (7)

Sources:(1) Lowry and Merrilees 1969; Merrilees 1970, 4,650 ± 104 (NSW-28a), 4,550 ± 112 (NSW-28b), 4,650 ± 153 (NSW-28c); (2) Milham and Thompson 1976, 3,450 ± 95 BP (ANU-807); (3) Partridge 1967, 3,280 ± 90 (Gak-693); (4) O’Connor 1999a: 90, bones are in mid-Holocene shell horizon but not speciﬁcally dated; (5)Archer1974, Tunnel Cave, 0 ± 80 (Gak-3890) on bone in limestone tunnel – the thylacine bones were not directly dated and their condition indicates that they may be older. Murray Cave, 3,090 ± 90 (ANU-716) bracketed by thylacine bone on surface and in underlying layer; (6) Mulvaney et al. 1964, bracketed by ages 3,750 ± 85 (P308)and3,870 ± 85 (P309); (7)Smith1977, >2,980 ± 90 (Gak-1029) with thylacine elements in underlying layers, including units dated 3,468 ± 100 (Gak-1022)and4,360 ± 110 (Gak-1023); (8)Kendrick and Porter 1973, dingo and thylacine bones are found together in lithiﬁed cave earth at cave CR21 in the Northwest Cape region, implying a mid-Holocene age for this deposit (if the remains are coeval). spread of the dog inland is likely to have been rapid and to have been assisted by its commensalism, or ability to insinuate itself into human societies as a semiautonomous fringe dweller. Dingo pups may also have been exchanged or traded, although we lack evidence of this at present. However, the diversity of linguistic terms implies that the dingo spread more rapidly than did the name for the new animal. The dingo took over much of the ecological niche formally occupied by the marsupial carnivores – the thylacine (Thylacinus)and devil (Sarcophilus) – and may have contributed to their extinction, although it is unlikely to have been the sole factor in the mid-Holocene extinction of these species because thylacines were much larger animals and survived until 3.5 ka (Table 6.8) ( Johnson and Wroe 2003). 208 The Archaeology of Australia’s Deserts

In ethnographic desert societies, dogs had several roles. They assisted in ﬁnding and running down game (Hayden 1975), although their utility in this regard declined sharply once people moved to missions and settlements. Dogs were also camp companions (Hamilton 1972) and sentinels who warned against real and supernatural dangers. Desert people regard the dog as very close family (tjarntu) and see their dogs as a fundamental part of the basic social unit. By the ethnographic period, dogs had also assumed a privileged position in mythology and ritual life as an intermediary between the human and supernatural worlds (Kolig 1978); they often feature in myths as pursuing or killing other ancestral beings. Their integration into desert societies in the mid-Holocene represents a socioeconomic development as important, in its own way, as the adoption of the new tool kits at 4 ka.

OVERVIEW: THE ASSEMBLY OF A DESERT CULTURAL SYSTEM Many of the key elements of Australian desert societies were assembled during the mid to late Holocene. Their development was shaped by a dynamic interplay of changes in demography, technology and social networks, in a long historical trajectory beginning around 12 ka. Increasing population density was a key factor in this process and is structurally coupled with the behavioural ecology of hunter-gatherer groups (affecting technology, mobility and subsistence economy). This is illustrated in a range of ways by the history of these desert groups.

Growth and Expansion of Desert Populations Australia’s deserts took their modern form in the post-Pleistocene as inland climates responded to global warming and the stabilisation of sea level. From 8 to 6 ka, Holocene environments in the interior arguably provided the best conditions for a hunter-gatherer population for 30 millennia, with stronger monsoon rainfall and an extended growing season that boosted biological productivity in the Western Desert, the Central Australian ranges and the inland rivers feeding Lake Eyre. This period coincides with a widespread consolidation of hunter-gatherer settlement across the interior, shown by an increase in site numbers; the presence of grindstones in site inventories; the increased visibility of rock art, facilities and structures; and the ﬁrst appearance of rich, dense occupation deposits. In some localities, people responded rapidly to the reestablishment of summer monsoon rainfall after 16–14 ka, but many individual sites do not show signiﬁcant use until 8–5 ka. The only areas to register rapid, abrupt increases in intensity of occupation were the arid west and south coasts, where the last marine transgression directly affected subsistence economies. In the interior, demographic growth may have been initially absorbed by reoccupation of The ‘Desert Culture’ Revisited: Assembling a Cultural System 209 marginal land abandoned during the last glacial maximum. By 8–7 ka, focal parts of the sandy deserts had been reoccupied, showing that this was not solely a late Holocene phenomenon.

Technological and Economic Changes The period from 4 to 3 ka was an important nexus of technological and economic change in the desert, as it was in most parts of the continent. The most distinctive elements of ethnographic desert tool kits – adzes, seed-grinding implements and composite tools – all date to this time. These economic changes coincided with the onset of an ENSO-dominated climate and higher levels of environmental stochasticity than anything seen in the mid-Holocene (8–4 ka). Growing populations during previous millennia had made the desert a well-peopled landscape. Desert groups could no longer rely, to the same extent, on relocation and residential mobility as their major response to shortfalls in local bush foods. The result was widespread adoption of a more efficient tool kit, including hafted composite implements, a more diverse suite of wooden implements (many geared to logistical foraging) and new ground-stone assemblages associated with the processing of grass, acacia and chenopod seeds as grain. These changes also involved a significant increase in economic capacity in the desert, with more elaborate plant processing assemblages, increases in diet-breadth, a more effective wooden tool kit and the potential contribution of hunting dogs in detecting and pursuing game. Seed harvesting and seed- grinding underpinned changes towards more extended occupation of some sites (as shown at Puritjarra at 3.8 ka), suggesting that changes in population density, residential mobility, site inventories, assemblage size and economy are likely to be interrelated variables. At present, however, we lack a detailed picture of associated changes in settlement pattern or site use across the desert at 4–3 ka or the behavioural correlates of flaked stone assemblages in the crucial period 8–4 ka that led up to these changes.

The Desert as a ‘Spread Zone’ One implication of the post-Pleistocene archaeology is that these hunter- gatherer groups inhabited a much larger social world than desert people before 8 ka. Given that the desert was surrounded by more densely populated regions to the north and east, accelerated social interaction made the desert an open system for rapid spread of new traits from the desert margins into the interior. Historical geography shows many examples of this, including the spread of PN languages, Panaramitee-style rock engravings, geometric microliths and the dingo – and, in more recent times, the diffusion of subsection systems and the long mythological pathways such as the tingarri, wati kutjara (two men), 210 The Archaeology of Australia’s Deserts

kungkarankalpa (seven sisters) and tjilpa (native cat) traditions, which enter the heart of the continent from its margins. In this respect, the desert is a classic ‘spread zone’, a region periodically uniﬁed and homogenised by the spread of cultural, technological and linguistic traits from the more densely occupied margins. The dominant temporal trend, therefore, is one of integration rather than regional differentiation. Rather than fragmentation of an old pan- continental cultural system, we need to think of this as the Holocene assembly of a desert system out of small population and cultural isolates.

Social History All of the new traits would have had social consequences, but these are difficult to reconstruct using current archaeological evidence. For instance, a language is a complex semiotic system, and language differences tend to rest on ancestry, religious identity and land relationships. What, then, are we to make of the adoption of PN languages across the greater part of a hunter-gatherer continent? At the very least, it must indicate a political and ritual economy able to generate some form of social hegemony. Similarly, a shift towards intensive use of seeds is not simply an economic change but also can involve a suite of associated social changes in the relations of production (particularly in the sexual division of labour), residential patterns, the matrilineal ownership of grindstones, regional exchange patterns (trade is discussed in Chapter 8)orthe capacity to ‘finance’ large ceremonies. In her classic study of ‘dual technologies’, Hamilton (1980) drew attention to the ethnographic differentiation of desert tool kits into specialised men’s equipment (hafted adzes and composite spears) and women’s (seed-grinding implements and digging sticks). Although a sexual division of labour is likely to be an ancient feature of hunter-gatherer economic life, the changes in flaked and ground-stone assemblages around 4 kamakethisvisibleinthe archaeological record for the first time. It is also possible that the material and symbolic aspects of the new technology sharpened gender roles. The extent of the cultural changes around 4 ka is mirrored in the politico- ritual economy of desert groups. Although they are relatively recent additions, both the dingo and the use of seed-foods are thoroughly naturalised in the mythology and totemic geography of the desert. Many ancestral totemic figures grind seeds or prepare seed cakes. Seed-food species are intrinsic to particular songlines or dreaming tracks (such as the ngirntaka mythology). There is a corpus of songs used in increase ceremonies for grass, acacia and portulaca seeds, and some stone arrangements and natural landforms represent seed cakes. Myths are notoriously fluid, able to respond to new political and social contexts or accommodate new elements within an old paradigm. However, even with these caveats, it is clear that much of the fabric of desert tradition canbenoolderthan4 ka. The ‘Desert Culture’ Revisited: Assembling a Cultural System 211

The post-Pleistocene history of the desert illustrates the complexity of reconstructing the development of hunter-gatherer societies in Australia’s deserts. In his pioneering archaeological work at Puntutjarpa rockshelter, RA Gould famously saw Western Desert society as an adaptive response to the poverty of the desert environment, creating a narrowly prescribed way of life that was stable for 10–12 millennia. A new desert society did, in fact, form in the Holocene – but at 4 ka, not 12 ka. This represented a distinctive adaptation to arid conditions (much as Gould proposed), but archaeology now shows that this was only the last of a series of adaptations to an arid land marked by unusually high levels of environmental stochasticity. CHAPTER 7

ROCK ART AND PLACE: EVOLUTION OF AN INSCRIBED LANDSCAPE

Rock engravings are one of the most durable traces of prehistoric occupation in Australia’s deserts. Much of the arid zone shares an austere suite of ancient pecked designs of emu and macropod tracks, circles and arcs (Fig- ure 7.1). But before 1900, few travellers seem to have noticed them. And when they did, early observers often mistook the engravings for fossil tracks. When HYL Brown, the notoriously taciturn South Australian government geologist, reported in 1883 on a site in the northern Flinders Ranges, he uncharacteristically went to some length to address this:

On some of the hard blocks of quartzite and quartzose sandstone near this place there are marks somewhat resembling the impressions of the feet of human beings, kangaroo, birds, etc., which are considered to be fossil tracks. On examination they appear to be merely rough imitations of such, the smooth surface of the rock having been removed by some hard instrument to a slight depth, the pitted marks of such action being plainly visible. (1883: 5)

Even museum authorities sometimes got it wrong. When, in 1902, Mounted Constable EG Waterhousesent ‘part of a native tattoe [sic] of a kangaroo’s track’ from the now famous Mannahill (Karolta) engraving site to the South Aus- tralian Museum, he was told that the markings ‘were due to the action of certain algae and lichens’ (Royal Society of South Australia 1901–2: 326; Hale and Tindale 1925: 52–3). It did not help that when asked, Aboriginal people invariably claimed the engravings were not made by people but were natural features related to totemic mythological beings. It was not until 1906 that the archaeological potential of the rock engravings was recognised, when Herbert Basedow (1907, 1914) suggested that they represented an extinct art of ‘very considerable antiquity’ (1914: 200). Later work emphasised the uniformity of this rock art across 2.5 million km2 of the interior, creating the picture of an ancient, pan-continental art tradition. Today, it is a common working assumption that

212 Rock Art and Place: Evolution of an Inscribed Landscape 213

Figure 7.1. Deeply patinated Panaramitee-style rock engravings at Florina station, Olary region, South Australia. This panel on dolomitic siltstone shows a range of common motifs, including bird and macropod tracks, circles and sinuous lines. Scale: Small circles are approximately 15 cm diam. (Photograph courtesy of MA Smith) these Panaramitee style engravings date to the late Pleistocene, although the few direct dates indicate ages in the range of 6–5 ka. One legacy of Basedow’s work is that rock art studies form a major stream of archaeological research in the desert, one that precedes development of regional stratigraphic sequences in most parts of the arid zone. Because rock art is so durable, it provides an opportunity to chart the historical development of a distinctive graphic system, a record of shifting spheres of interaction and a rare window on the development of the political economy of the desert. Before turning to the archaeology, however, we need to look at the geography and social context of desert art.

ROCK ART REGIONS Rock art across the continent falls into several broad art provinces, as shown in Figure 7.2a. These were ﬁrst described by Maynard (1976) and later tested using multivariate analyses by Layton (1992) and Franklin (2004). Australia’s deserts have a distinctive rock art, covering an area that loosely corresponds to Davidson’s ‘central design area’ for decorative art (Davidson 1937: 134–5). The rock art of this region is made up of a corpus of bird and macropod tracks, circles and arcs, with only a limited range of ﬁgures (mainly 214 The Archaeology of Australia’s Deserts

solid silhouette figures of lizards and anthropomorphs). The apparent uniformity of this art across the interior reflects the use of a shared graphic vocabulary and common stylistic conventions over an immense area. In contrast, the temperate and tropical margins of the continent have a more regionally diversified rock art, with paintings or engravings dominated by figures of animals and people, ranging from elaborately naturalistic figures to simple stylised figures of people and animals. The Burrup–Pilbara region in northwestern Australia falls into the ‘figurative’ group, giving it a separate history to other parts of the desert (Wright 1968; Mulvaney 2010). Figure 7.2a glosses over two important considerations. First, large parts of the continent, including much of the sandy deserts and the Cooper basin, have no rock art at all. In this context, the striking uniformity of desert rock art indicates the efficiency by which other media, such as body art, stone arrangements, ground paintings and decorations on ceremonial paraphernalia, were able to maintain a shared graphic system across thousands of kilometres of open desert. Second, Figure 7.2a is an amalgam of different time periods. Some boundaries appear to be longstanding features of the graphic landscape, especially the boundary between figurative art and desert art, which persisted throughout the Holocene. Within the desert, the overall continuity in the art suggests gradual evolution of a single graphic tradition. On the eastern margin of the arid zone, however, the situation is more complicated. Here, an older phase of ‘track and circle’ engravings was replaced by figurative and geometric art during the late Holocene.

TOTEMIC GEOGRAPHY AND POLITICAL ECONOMY Rock art needs to be approached as part of the religious system and political economy of ethnographic groups. Desert rock paintings and engravings are grounded in a totemic mythology that articulates ritual relationships between people and place. And it is these relationships that form the foundation of traditional land tenure across Central Australia and the Western Desert and provide a mechanism that maintains the territorial spacing of desert groups (Peterson 1972). It is nearly a century since Reinach (1903) focused attention on the religious significance of Central Australian Aboriginal art and used it as an analogue for interpreting Palaeolithic European cave paintings (see also Palacio-Perez 2010). Since then, our understanding of the significance of rock art and other aspects of hunter-gatherer graphic systems has developed well beyond the concepts of totemism, hunting magic and ‘increase’ ceremonies outlined by Spencer and Gillen (1899). Durkheim (1912) was closer to the mark with his idea that designs representing a totemic ancestor were also emblems of a social group claiming allegiance to that ancestor. But these early theorists missed the extent to which totemic ancestors – and through them, people – are tied to specific Rock Art and Place: Evolution of an Inscribed Landscape 215

Figure 7.2. Graphic and religious systems across Australia: (a) rock art regions (adapted from Maynard 1976:ﬁg.6.23;Layton1992; Morwood 2002:ﬁg.2.1; and Franklin 2004); (b) Australian religious systems. (Adapted from Berndt 1974:pt.4, 23) 216 The Archaeology of Australia’s Deserts

places. We now know that people’s rights to land vary according to the links they have with the ancestral beings that created or traversed the landscape and were transformed into its physical features (Strehlow 1970, 1978; Sutton 2003). Rock paintings, stencils and engravings not only provide a tangible link to this totemic geography but also are seen by Aboriginal people today as natural features that validate the ‘dreaming’. Two concepts that coordinate the relationship between people and place are a belief in the ‘dreaming’ and the importance of totemic identity. The ‘dreaming’ refers to an ancestral world, variously called altyerrenge (‘alcheringa’) or tjukurrpa (Stanner 2009). This was shaped by mythological ancestral beings who travelled the land and, along the way, left something of themselves behind in rock art and in topographic features – such as hills, rock outcrops, trees, sandridges and ochre deposits. Their travels are represented by myth sequences (colloquially termed ‘songlines’), which form paths linking sites over a wide area and are expressed in song cycles and ceremonial reenactments. The other key canon of this belief system is use of a totemic identity. In Australian totemism, the natural world is personified in ancestral beings, each connected to a plant or animal species and able to shift shape between animal and human form. People are linked to nature through their identification with these totemic beings. TGH Strehlow (1970) dubbed this ‘geographically based totemism’ because the ancestors (or aspects of them) are more or less anchored to specific sites in the landscape. As Taylor put it, ‘in Aboriginal thought, place is a primary mnemonic for ancestral beings and events and also a means of linking social groups and individuals to the ancestral world’ (1979: 86). The ‘dreaming,’ then, is an ideological system that connects people to place in a fundamental way and connects series of places along the same songline. This ideology is articulated in a large repertoire of land-based mythology. Throughout Aboriginal Australia, access to this knowledge was regulated through ceremonies, the most important of which centred on (a) male initiation and the series of postinitiation revelatory stages that gave ‘access to an expansive and expanding area of knowledge’ relating to the totemic geography (Berndt 1974: 4); and (b) ‘increase’ ceremonies intended to maintain or revitalise the species and natural phenomena personified in the totemic ancestors. Although the basic elements of Aboriginal religious life were widely shared across the continent, Berndt (1974) identifies four different religious systems (Figure 7.2b). The northern and eastern parts of the continent have coordinated religious systems in which overarching traditions are held in common across descent groups and tribal or linguistic units, often giving predominance to one or two ‘super-totemic’ ancestral beings, such as the Rainbow Serpent, the Wagilag sisters, Namarrgon the lightning man, the Wandjinas, and sky gods such as Baiame and Daramulan. Among the rivers, swamps and savannas of Arnhem Land and the eastern Kimberley, the emphasis is on large-scale fertility Rock Art and Place: Evolution of an Inscribed Landscape 217 cults, centred on kunapipi ceremonies or their local variants. In the northeast, spirits of the dead are transformed into ancestral ‘cult heroes’. In eastern and southeastern Australia, two symbolic worlds, earth and sky connected by a pathway, are represented in the bora earth or stone circles found in this region. Local belief gives importance to supernatural beings – such as the ‘sky gods’ Baiami and Daramulun – who are set apart from people. In contrast, most desert people have a decentralised religious system, in which mythological and ritual knowledge is geographically compartmen- talised, with local descent groups (clans) each responsible for a section of the total body of knowledge, usually a segment of a ‘dreaming’ track or songline. The patri-clans responsible for different segments of the same mythological track sometimes coordinate their activities to stage major ceremonies, but ritual knowledge is always tightly held property. The territorial aspects of this arrangement structures other aspects of the ritual-political system: ritual life is strongly male-dominated; place-based religious paraphernalia is more elaborate than in other areas (in particular, the sacred boards and engraved stone plaques known as tjuringa, which are only found in the desert); living totemic ancestors intrinsically reside in people; initiation ceremonies are more elaborate, with an emphasis on formal instruction of initiates, on long song cycles and on the number of mythological reenactments; and secret sacred knowledge is well developed. Although the desert is often seen as an open system in terms of its demographic arrangements and foraging territories, this is clearly not the case in the religious sphere, where ritual property is strongly bounded. And it is this area that frames rights to land (in terms of clan ‘estates’) and is most strongly reflected in iconography and rock art, where ‘painting the land’ is restricted to those with core rights in country. Within the desert, however, there are some important variations in this pattern. Among Warlpiri and Arandic groups, corporate descent groups – represented by genealogically shallow patri-lineages – hold estates. Native title research indicates that these have mostly been stable in their links with core sites and ‘dreamings’ for a century or more (Sutton 2003:161–2). Western Desert groups are more fluid in composition of the land-holding group: local descent groups are absent and the region lacks the corporate character of land-holding groups in the rest of the desert. The classic desert segmentary totemic system is present, but ritual responsibility for core sites rests with a looser group of senior men recruited variously by long-term residence, conception affiliation or bilateral descent through either of the parents or grandparents. Western Desert groups tend to place more stress on inclusivity, and they emphasis the social connections that ‘dreaming’ tracks create. As Tonkinson (1978: 103) points out, ‘the very same totemic geography that binds a group to its estate and fosters pride and ethnocentrism also, through the many ancestral tracks, links them to all their neighbours and to other individuals and groups in distant areas.’ The social fluidity of Western Desert groups is usually regarded 218 The Archaeology of Australia’s Deserts

Figure 7.3. Denis Ebaterinja drawing designs representing the honey ant ‘dreaming’, Ellery Creek, Central Australia. (Pho- tograph courtesy of June Ross)

as a response to the poverty and variability of the desert environment, but Hamilton (1980) suggests that it may reﬂect their recent migration into this part of the desert and the fact that their institutions were still in transition at European contact. Comparison of Figures 7.2a and b shows a general correlation between the geographic totemism of the interior and the distribution of desert rock art. This suggests potential avenues for direct archaeological investigation of how the classic ethnographic pattern developed: by focusing on the differentiation of individual art sites within a common graphic system and by establishing an age for large ceremonies at key totemic centres.

The Graphic System and Rock Art Munn (1986) showed that the graphic elements used in sand drawings in narrative storytelling were recombined in the more elaborate designs used in body painting (for ceremonies), ground paintings and other ceremonial paraphernalia (such as shields and tjuringa), as well as in rock paintings and rock engravings (Figure 7.3). Links with the totemic geography were fundamental. Both Taylor (1979) (who looked at kulpidj or tjuringa)andDickins(1996)(who looked at ceremonial shields) showed that despite possessing a simple graphic vocabulary, the designs painted and inscribed on objects served to differentiate place, region and totemic identity. Rock Art and Place: Evolution of an Inscribed Landscape 219 table 7.1. Recent production of traditional rock art in Central Australia and Western Desert

Place Region Year Source

Retouching of painted panel Undiarra Finke River 1897 (1) Paintings Walinya Musgrave Range ∼1915 (2) Pounded designs Granites Tanami 1937 (3)(4) Pounded designs – Mann Range 1940 (3)(4) Pounded designs – Musgrave Range 1940 (3) Pounded cupules – Musgrave Range 1940 (4) Repainting of painted frieze Uluru Petermann Range 1940 (5) Pecked engravings – Tomkinson Range ∼1940 (6) Repainting of art frieze Ngama Tanami 1957–1965 (7) Paintings Wintjara Rawlinson Range 1966 (8) Paintings – Warburton ∼1966 (8) Paintings Puntutjarpa Warburton ∼1968 (9) Repainting of major art Gunadjari Tanami 1967 (10) frieze Paintings Yaru Yaru Kintore Range 1972 (11) Repainting of art frieze Walinya Musgrave Range 1974 (12)

Sources:(1) Spencer and Gillen (1899: 201); (2) Tindale (1972: 240); (3) Mountford (1955); (4) Mountford (1976: 76–8); (5) Mountford (1965); (6) Johnson (1964: 162); (7) Mountford (1968: 69); (8) Gould (1969: 146–8, 160); (9) Gould (1977: 51,figs.92–4); (10) Sandall (1967) (film); (11) Sandall (1972)(film);(12) Wallace and Wallace (1977).

What role did rock art play within this graphic system? Some general observations can be made, as production of rock art continued, in parts of the desert, into the period between 1940 and 1970 (Mountford 1965, 1968, 1976; Gould 1969c, 1977)(Table 7.1). First, most rock art is conceptually linked in some fashion to the place where it occurs and most relates to the ‘dreaming’. Painting of totemic designs in rockshelters appears to have been exclusively the work of senior initiated men, and the right to create such durable marks was restricted to those with core rights in country. However, the presence of rock art is unrelated to the importance of a site: there are many key totemic centres, and focal occupation sites, without rock art. And, unlike body designs or designs on ceremonial objects, rock art was not a central feature of ritual and ceremonial life in the desert (although repainting of major totemic sites such as Ngama, Rugari and Undiarra clearly did involve signiﬁcant ritual, blood and sacra). The absence of rock art from large parts of Australia’s deserts did not affect the character of ritual practice or the transmission of desert ideology. In some areas, stone arrangements represent an analogous link to the ‘dreaming’ – sometimes directly representing the body of totemic beings – but geoglyphs like those in the Atacama desert (which, in the strict sense, are graphics projected onto earth and stone) are not present in Australia’s deserts. 220 The Archaeology of Australia’s Deserts

Second, rock art mirrors the basic division seen in Central Australian art between secular and sacred art (Spencer and Gillen 1899; Berndt 1974, Munn 1986). Secular art is mainly composed of tracks, track-lines or circles, with a loose narrative syntax that parallels the structure of sand drawings. This may refer to the totemic geography or events in the associated mythology, but it does so at a general level. Sacred art consists of totemic designs and emblems (arlkenye – Spencer and Gillen’s ‘ilkinia’; kuruwari – Munn’s ‘guruwari’), symbolically differentiating place or totemic ancestor. The large striped emblematic designs at some Central Australian sites, such as Emily Gap, are good examples of these totemic designs (Figure 7.4), but sacred art also includes designs exclusively composed of clusters of bird or macropod tracks (such as those at Rugari; McCarthy 1958: 62), showing that a simple typological separation of sacred and secular art is not possible. Third, there is a further division between the graphic art of paintings and engravings on the one hand, and individual or informal marks resulting from interaction with a rock surface, such as hand stencils, prints, pounded pits, incised grooves and abraded patches on the other hand (Rosenfeld 1997). Munn (1986: 31) commented on the importance of sensory contact with totemic designs. Men visiting a site often formally trace the outline of key rock paintings, as Mountford observed at the Ngama site in 1951, when a senior man took the hand of an initiate and guided it to key designs (Mountford 1968: 69–70). Kimber (1985) records that hand stencils were made as part of an introduction to a site, as an afﬁrmation and greeting to the totemic ancestor – ‘hullo old man’, as a Pintupi man phrased it (1985, 1989: 97). Other activities in this category include the pounding and abrading of rock surfaces to activate the essence of a place, sometimes as part of ‘increase’ rituals (Spencer and Gillen 1899: 200; Mountford 1976: 231). Incising lines and ﬂaking the edges of boulders probably represent other ways of activating a site. Ross usefully groups abraded grooves, pits, abraded areas and battered edges as ‘associated rock art traditions’ (2003: 150).

Aboriginal Ontology In contemporary Aboriginal ontology, rock paintings and engravings are wholly natural phenomena that provide tangible evidence of the totemic ancestors and the tjukurpa (Berndt 1974: 18;Layton1992: 12). This view is encapsulated in the Warlpiri term for totemic designs, kuruwari (lit. ‘signiﬁ- cant mark’), which applies to any visible mark left by a totemic ancestor and which does not conceptually distinguish rock art from natural marks on rock surfaces (cf. Mountford and Edwards 1964: 849–50). For example, Edwards noted that his Aboriginal informants were adamant that rock engravings were ‘part of the “dreaming” site’ and ‘were made when the site was “created”’. Any suggestion that people had ‘made the engravings was met with surprise and Rock Art and Place: Evolution of an Inscribed Landscape 221

Figure 7.4. Large striped totemic designs, painted in red-and-white ochre, in the gorge at Emily Gap (Anthwerrke), Central Australia. These emblematic motifs were described by Spencer and Gillen (1899: chapter 9 and fig. opp. 632) and are associated with utnerrengatye caterpillar ancestors. The ‘eyes’ represent caterpillar eggs. (Source: Spencer 1928:fig.161; photograph courtesy of Museum Victoria – XP9776) incredulity’ (1966: 36). People see themselves as curating or restoring ancestral marks when they retouch rock paintings, in the same way that a stone arrangement might be cleared of vegetation and rebuilt. Even when a specific term for rock paintings is used – such as the Western Desert term, walka (lit. ‘pattern’) – this is often an implied comparison with ceremonial body art associated with totemic ancestors. Archaeologists have often misunderstood this worldview, assuming that Aboriginal denial of a human agency for rock engravings implies that this is an ‘extinct art’ (McCarthy 1958: 24–5; Mountford 1960: 145; Edwards 1966: 36), whereas their informants are simply asserting that these marks validate the religious system. Rock art is only a small part of a symbolic landscape that primarily relies on natural features and that exists irrespective of the presence or absence of rock art.

ARCHAEOLOGICAL APPROACHES Australian work is narrowly anchored in international writing about style and its relation to social identity, boundary maintenance, alliances, information exchange and population aggregation (Wobst 1977; Conkey 1980;Gamble 222 The Archaeology of Australia’s Deserts

1982; Wiessner 1984). Gamble (1982) drew on Australian ethnography to argue that stylistic similarities in Upper Palaeolithic figurines across Western Europe were evidence of an open interaction network among these hunter- gatherer groups. Morwood (1984) and C Smith (1992)reappliedthisto Australian rock art, arguing that the dichotomy between bounded (closed) social networks in the productive woodland and coastal environments and linking (open) social systems in more marginal desert environments was reflected in their respective graphic systems. The open system in the desert underwrote the uniformity of Central Australian art. However, almost all studies since then have reported more regional or structural variability in desert rock art than previously allowed (Forbes 1982; Gunn 1995; Galt-Smith 1997;Ross2003; Franklin 2004; Mulvaney 2010; McDonald and Veth 2011). McDonald and Veth comment: ‘against expectations, we have repeatedly found surprisingly high levels of stylistic heterogeneity across the arid zone, particularly in the art of the recent past’ (2011: 222). This should not be a surprise because the social anthropology of the region shows that it is important to distinguish between two kinds of relationships between people and land: ritual and economic (Hiatt 1962). The totemic geography articulates a more or less ‘bounded’ system of land tenure, rights in land and ritual property, whereas economic access to land is more open – based on a broader set of kin and interpersonal relationships – and varies with demographic circumstances across the desert. In these desert systems, rock art is a counterpoise, reinforcing a system of inalienable land tenure in societies marked by high mobility, seasonal dispersal and aggregation of the population and extreme fluidity in residential groupings. These ideas were extended to Panaramitee-style rock engravings, which were initially thought to be an ancient pan-continental art tradition (Maynard 1979). The uniformity in this art suggested that open social networks and long- distance exchange were the norm across the interior of the continent in the late Pleistocene. These gave way to more bounded systems with regionally distinct bodies of rock art in the late Holocene, as increasing population led to closure of social networks and greater territoriality. However, increasing evidence for regionalisation in the older desert rock art (Franklin 2004; Mulvaney 2010) has complicated the picture. As Rosenfeld pointed out 30 years ago, ‘the great unity of style and motif of ancient art styles in Australia may have been overstated in attempts to synthesise the evidence for the antiquity of rock art in the continent’ (Rosenfeld, Horton and Winter 1981: 89). The question of whether the rock engravings are simply an extinct art reincorporated into recent mythology (Maynard 1979: 93) has been resolved by evidence showing that production of rock engravings continued into the recent prehistoric period (Tables 7.1 and 7.2)andthatnet production rates for rock engravings did not fall significantly over time (Table 7.3)(1,330 motifs/ka for the last millennium, compared to a net rate of 1,696 motifs//ka over the previous 7 ka). table 7.2. Stratigraphic and direct dates for rock engravings in the arid zone

Site Age ka Method Type Status Source

Burrup-Pilbara complex Yule River ∼26.8–0.4 Microerosion Rock weathering ages for engravings (f) Bednarik 2002 Gum Tree Valley 21.7 14C Transported shell near art panel (e) Lorblanchet 1992 Bush Turkey 3 ∼8.4–3.1 14C AMS Engraved slab in roof fall (a) Veth et al. 2008 Skew Valley 3.6–2.4 14C Fragments of engravings (b) Lorblanchet 1992 Gum Tree Valley 3.5–1.0 14C Adjacent shell midden (e) Lorblanchet 1992 Kaalpi 2.2 14C Fragments of engravings (c) (g) Veth et al. 2001 MV1 (Calvert Ra.) 1.0 14C Engraved panel buried by deposit (c) McDonald 2005 Central Australian art region Karolta 147.3–1.4 14C AMS Desert varnish over engravings (d) (h) Nobbs and Dorn 1993 Karolta 131.6–1.4 Cation-ratio Desert varnish over engravings (f) Dorn et al. 1988 Sturts Meadows 12.3 14C AMS Carbonate over off-art desert varnish (e) Dragovich 1986 Puritjarra ∼8.3–4.0 14C AMS Engraved slab in roof fall (a) Smith et al. 2009 Wanga East 6.0–5.0 14C AMS Engraved slab in roof fall (a) Smith et al. 2009 Kurutiti ∼4.0 Typology Engravings in millstone quarry (b) Mulvaney and Gunn 1995 Intirtekwerle 0.8–0 14C Engraved slab in roof fall (a) Gould 1978a; Smith 1988 Keringke 0.8–0 14C Fragments of engravings (c) Stockton 1971;Smith1988 Therreyererte 0.6–0 14C Engraved slabs on surface of deposit (b) Smith 1988 Finger markings Koonalda Cave 25.5–21.9 14C Dated torches in cave with ﬁnger markings on walls (b) Wright 1971;SharpeandSharpe1976

Status: (a) Age constrained by stratigraphic or radiometric bracketing ages, (b) associated with dated occupation deposit or structure, (c) minimum age only, (d) interpolated age estimate or indirect date, (e) uncertain association of dates and rock art, (f) dating method not veriﬁed or inadequately calibrated, (g) motifs not identiﬁable, (h) dates retracted by author. 223 224 The Archaeology of Australia’s Deserts

table 7.3. Estimates of net rock art production in Central Australia, comparing different time periods

Phase Motifs/ka

From Gunn (2000) Recent (unpatinated) engravings 1,371 Older engravings (∼1–8 ka) 1,588 From Ross (2003) Recent (unpatinated) engravings 1,330 Paintings 3,697 Stencils 1,893 Total for recent art 5,920 Older engravings (∼1–8 ka) 1,696

In another key paper, Wiessner (1984) argued that stylistic behaviour is one means by which people negotiate personal and social identity. This has been very influential in Australian research, with a range of studies that treat rock art – especially the more recent pigment art – as a semiotic system that mediates social and territorial relations (McDonald and Veth 2006). We can expect a priori that variability will scale geographically: ranging from (a) differences between separate graphic and visual systems, (b) clinal differences in regional identity within a single graphic system, (c) stylistic differences marking intergroup boundaries within regions, and (d) the differentiation of site assemblages (as in segmentary totemic systems). Few studies, however, attempt to quantify variability in art assemblages or to distinguish different levels of diversity. Seasonal or ceremonial gatherings are a crucial institution for maintaining social integration in dispersed hunter-gatherer societies. Conkey (1980) argued that the history of these aggregations could be traced archaeologically, using high diversity in parietal art repertoires to identify sites where disparate groups from widely separated areas coalesced. In Australia’s deserts, many art complexes are situated in or around focal points in the desert landscape, where a diversity of archaeological remains can be expected (Gunn 1997). However, identification of art complexes specifically as ‘aggregation’ sites has had mixed success in desert research. Galt-Smith (1997) showed that different types of ethnographic sites were lumped under the term ‘aggregation’ site and that secular and sacred rock art had different patterns of production. McDonald and Veth (2011) argued that diversity in pigment art in the Western Desert is tiered, reflecting long-distance interactions, as well as signalling local identity. All of this raises doubts about whether parietal rock art (rather than art mobilier, exotic stone or imported ochre) is the right tool to identify sites of large ceremonial gatherings. One problem for this view is that desert rock art is place-based art – that is, designs are mediated by the totemic associations of a Rock Art and Place: Evolution of an Inscribed Landscape 225 place – and, in most cases, only people with core rights are entitled to produce the art. This undermines the premise that diversity of people is necessarily coupled with diversity of rock art. Another is that many large ceremonial or seasonal gatherings, such as those witnessed by McKinley in 1861–2 around the Coongie Lakes (Williams 1998), did not involve production of rock art. If rock art had a socially mediating role for large groups of people, it was clearly not fundamental to such gatherings, even in those parts of the arid zone known for long-distance trade and exchange networks (see Chapter 8). In Australia’s deserts, most of these analytical approaches would benefit from stronger explanatory frameworks and a more rigorously quantitative approach to diversity and style. The variability in religious patterns across the continent also suggests some general points that can be made about the structure of associated rock art and graphic systems. In regions with coordinated overarching traditions, we can expect a standardised iconography with the same motif or group of motifs, repeated at sites across a wide region – as is arguably the case in northern and eastern Australia. However, in a segmentary system, such as that in the desert, we should expect the art in each locale to be more indi- vidualised, albeit within a shared graphic vocabulary and conventions, with a strong trend towards repetition of motifs within a site.

ESTABLISHING A CHRONOLOGY The major impediment to integrating rock art with other archaeological records is the difficulty of dating the art. Despite some heroic efforts, only small gains have been made over the last 40 years (see Table 7.2). Desert sites provide few opportunities for direct stratigraphic dating. It is rare to find panels of engravings buried by archaeological deposits. Some success has come from radiocarbon dating of sediment surfaces beneath fallen rock slabs, providing a terminus post quem for engravings on the upper surfaces of the rock fall (David, Chant and Flood 1992; Smith, Watchman and Ross 2009). Pieces of engraving panels are occasionally found in archaeological deposits. The few examples date to the mid to late Holocene, but most are so fragmentary that stylistic assignment is uncertain (Stockton 1971; Veth, Smith and Haley 2001). Technically, dates for a layer containing fragments of engravings provide only minimum ages for the original art frieze, but a case can often be made that the rock art is associated with an occupation layer – especially as production of a major panel of paintings or engravings is itself likely to generate significant site use. Fragments of ochre in archaeological deposits often serve to identify phases of rock painting, and AMS 14C dating of organics in pigments has shown good correlation between the production of rock paintings and concentrations of ochre in archaeological deposits at the same site (McDonald and Veth 2011). For rock engravings, most work has focused on dating various mineral skins that cover both rock surfaces and rock art: carbonate layers, silica skins, 226 The Archaeology of Australia’s Deserts

oxalate crusts and desert varnish (more correctly termed rock varnish). Ini- tial attempts to date engraved rock surfaces directly using 14Cdatingof localised calcium carbonate deposits, interleaved with rock varnish, did not resolve questions about the relationship between varnish layers and the rock engravings (Dragovich 1984, 1986) and were invalidated when other work showed that pedogenic carbonates do not give reliable radiocarbon ages (Williams and Polach 1969; Callen, Wasson and Gilespie 1983; Amundson et al. 1994). One of the most characteristic mineral skins in arid areas is desert varnish, a thin, finely laminated film (<1 mm) of manganese and iron oxides, clays, and organic debris cemented to the rock surface. Cation-ratio dating, which assumes time-dependent leaching of mobile elements (Ca and K) relative to Ti, has been applied to varnishes covering archaic desert engravings at Karolta (Dorn, Nobbs and Cahill 1988) but foundered, as later work showed serious geochemical problems with the technique and its assumptions. These included the difficulty of calibrating cation-ratios against known-age samples, observed variability in cation-ratios across individual varnish layers and evidence that Ca and K were not preferentially leached from rock varnishes; other research showed that the initial ratio of Ca + K to Ti was not constant (Reneau, Oberlander and Harrington 1991; Watchman 2000). AMS 14Cdatingofthe organics in desert varnish has also been attempted (Nobbs and Dorn 1993), but it is now evident that varnishes are often remobilised and reworked and may contain carbon of widely different ages (Dragovich 2000; Watchman 2000). At present, the most promising methods appear to be uranium series dating applied to specific oxide phases within the varnish (Aubert et al. 2009)or compound-specific radiocarbon dating. A separate approach involves AMS 14C dating of oxalate skins on rock surfaces (Smith et al. 2009). Although oxalate minerals appear to be uncommon in arid environments, they provide a closed system for carbon and, once formed, are stable and insoluble (Watchman 2001). This research has shown that some common assumptions about desert varnish and oxalate crusts as climatic markers are incorrect. Desert varnish is sometimes assumed to be a fossil directeur for a period of intensified aridity. However, because its formation is controlled not only by supply of aerosols in desert dust but also by the biological processes (mainly bacteria) that fix iron and manganese on rock surfaces, it can form under a variety of climatic conditions. In Australia’s deserts, varnish appears to have formed episodically (because it is often laminated) under widely different conditions: during the last glacial maximum, during the early Holocene (Dragovich 1986) and over historical graffiti within the last century (Dorn et al. 1988). In contrast, oxalate crusts are sometimes seen as markers of more humid climatic conditions. In Australia, they are mainly found in humid to subhumid tropical regions where mean annual rainfall is greater than 700 mm (Watchman 1991). It is only Rock Art and Place: Evolution of an Inscribed Landscape 227 recently that rock art surfaces with oxalate skins have been identified in arid Central Australia, where they formed in sheltered locations throughout the Holocene (Smith et al. 2009). The trend in this research has been to isolate and date specific components within mineral skins. This is also true of recent work to radiocarbon date rock paintings, where a plasma-oxidation AMS 14C technique has been used to selectively date amorphous organics and fine carbon in the pigment, avoiding contamination from oxalates or carbonates on the rock surface (Watchman 1993; McDonald and Veth 2008). This gives good results for pigment art within the last 500–1,000 years but is open to contamination from subaerial biofilms on rock surfaces (Gorbushina 2007), and problems may well emerge as we attempt to date older rock art. Most rock art research in the desert has relied on relative dating to establish a stylistic sequence, using both superimposition of motifs and degree of patination as indicators of relative age. Neither is straightforward. It is notoriously difficult to determine the order of superimposition when two engraved designs overlap and, in the arid zone, cases of superimposition form only a small proportion (about 6–10 per cent) of the rock art assemblage. Patina- tion is a catch-all term that refers to several different processes: (a) chemical redox alteration of rock surfaces, (b) build-up of mineral crusts and varnishes, and (c) discolouration by various biofilms produced by lichens, cyanobacteria, microcolonial fungi and bacteria. The most common form is the subaerial weathering and oxidation of rock surfaces, producing an iron-rich patina and creating the rich red colours of Australia’s deserts. Development of a patina takes time, and the degree of repatination of engravings (where bright, freshly exposed rock has weathered to match an older rock surface) has been widely used as an indicator of relative age. One limitation is that redox patination rates are determined not just by exposure time but also by aspect and location and by rock type, and these can vary across a single surface. Another problem is that most studies do not systematically distinguish oxidation (weathering), varnishing (accretion of a manganese or iron-rich film) or biofilm discolouration (through water seepage or biological processes). The distinction is important because these processes operate on different timescales (with different rates of patination), and some involve episodic rather than continuous formation of a patina. Rock varnishes have also been shown to be removed by fire or salt spray (Dragovich 1994). Patination is also hard to quantify: most studies use a simple three-stage scale (fresh, moderately patinated, fully patinated), but Mulvaney (2010)relies on a highly subjective five-stage scale. Other work, using microerosion of engravings as a semiquantitative measure of age (Bednarik 2002), is promising but not widely used, and it involves extended extrapolation from a few calibration points bunched in the historical period. 228 The Archaeology of Australia’s Deserts

EARLY TO MID-HOLOCENE ROCK ART TRADITIONS The ubiquitous presence of red ochre in archaeological deposits leaves little doubt that late Pleistocene groups across the continent had some form of graphic art, although most of this is no longer visible. In northern Australia, both rock paintings and engravings were produced during the terminal Pleis- tocene (from 16.1 ka to 4.6 ka at Early Man rockshelter – Rosenfeld et al. 1981), as well as finger fluting in the soft limestone walls of Koonalda Cave on the Nullarbor plain (Wright 1971). However, most dates for rock engravings cluster in the early to mid-Holocene, particularly in the period 8–4 ka (Franklin 2004: table 1.1). This appears to reflect widespread investment in the corporate marking of country as Aboriginal populations increased in the postglacial period. Within the desert itself, there is no substantial evidence to support a Pleis- tocene age for the surviving rock art. Most indicators of antiquity are circumstantial and can be accommodated within an early to mid-Holocene time frame – and this is where the few secure dates also point (see Table 7.2). By the early to mid-Holocene, there were two largely contemporaneous graphic systems operating in the desert (Figure 7.5), one based on Panaramitee-style ‘track and circle’ art, the other consisting of the more figurative and complex geometric art of the Pilbara.

Panaramitee-Style Rock Engravings In the eastern half of the arid zone, the earliest identiﬁed rock art consists of Panaramitee-style rock engravings (Figures 7.1 and 7.6), which represent a widespread mid-Holocene graphic tradition. The geographic distribution of these engravings is shown in Figure 7.5. They occur throughout the Lake Eyre basin and Central Australia, extending west to the borders of the Western Desert. But it is not clear whether Panaramitee-style engravings extend across the Western Desert. What Figure 7.5 shows is that these engravings are not a pan-continental tradition but rather a graphic tradition largely restricted to the eastern half of the arid zone and adjacent drainage systems. At the extreme eastern end of this distribution, in the headwaters of the Barcoo River, Panaramitee-style engravings are limited to small assemblages (<200 motifs) at six sites (Morwood 1979), compared with an estimated 30,000 motifs at more than sixty Panaramitee sites in central Australia, ranging from sites with less than 200 motifs to those with 2,000–3,000 engravings (Table 7.4).

deﬁning the style These engravings were ﬁrst recognised as a continental art tradition by Edwards (1966), and the concept was formalised by Maynard (1979), who gave this art its name (from one of the key sites, Panaramitee, in the Olary region). The basic Rock Art and Place: Evolution of an Inscribed Landscape 229

Figure 7.5. Desert-style provinces for mid-Holocene rock engravings. Black squares show sites with archaic face engravings: (1) Burrup-Dampier sites; (2) Woodstock Abydos engraving; (3) Calvert Range and Durba Hills; (4) Cleland Hills; (5)Gap Hole, Toomba Range; (6) Carbine Creek; (7) Jalijbang. deﬁnition, as set out by Edwards and Maynard, was of an art with a limited range of recurring motifs (mainly tracks and circles) and a relatively uniform composition (in motif frequencies) across the arid zone (Table 7.4). Later studies showed that there is greater regional variability than previously believed (Forbes 1982; Nobbs 1984;Clegg1984;Franklin2004). A critical assessment of the status of Panaramitee by Rosenfeld (1991) – entitled ‘Panaramitee: Dead or Alive?’ – argued that over the intervening decade, the term had been too broadly applied to any ancient rock engravings and too narrowly restricted to engravings without considering other media. Not all ancient patinated engravings are necessarily Panaramitee art, nor is all ‘track-and-circle’ rock art. Another unresolved issue was the relationship between the early engravings and more recent rock art paintings with a similar range of graphic elements. Was there a ‘painted Panaramitee?’ asked Nobbs (1984: 111). So, what is ‘Panaramitee’? A graphic system cannot be deﬁned by motif frequencies alone. Most researchers accept that we also need to consider the stylistic conventions used in depicting common motifs, and the structure and syntax of the art. Rosenfeld (1991) showed that these excluded two bodies of rock engravings often lumped with the Panaramitee: those in tropical north Queensland and those in Tasmania. Similar considerations exclude the rock engravings at Ingaladdi (Mulvaney 1975;LewisandRose1995) – a rockshelter 230 The Archaeology of Australia’s Deserts

Figure 7.6. Panaramitee-style rock engravings at Puritjarra, Central Australia: (a) cluster of pecked bird tracks; (b) cluster of pecked pits (eggs) with nearby bird track; (c) plain pecked circles with incised grooves on same block. Scale is 150 mm diameter for largest circle. (Source:Smith2010)

with a dense vertical frieze dominated by pecked and abraded human footprints and abraded grooves – and also small isolated panels of recent ‘track and circle’ engravings found in southwestern Australia (Clarke 1983) and near Darwin (Bourke and Mulvaney 2003). Because bird and macropod tracks, circles and arcs are widespread elements in different graphic systems across the continent, they are best seen as a generalised homoplastic trait that might co-occur in unrelated regional art styles. Rock Art and Place: Evolution of an Inscribed Landscape 231 table 7.4. The composition of Panaramitee-style engraved assemblages

Percentages Site N Tracks Circles Other designs Source

Florina 985 48 40 12 (1) Panaramitee 1,003 58 27 16 (1) Tiverton 2,736 72 17 11 (1) Winnininnie 3,236 71 17 12 (1) Tukulnga 3,186 71 24 4 (1) Cleland Hills 387 50 33 17 (1) Carbine Creek 707 17 41 42 (2) Bull Hole 195 60 15 25 (6) Buckland Creek 104 62 <137 (6) Kurutiti 2,034 50 11 40 (4) Wanga East 200 75 10 15 (10) Roma Gorge E 1,251 42 30 28 (3) NDhala Gorge 4,469 37 22 41 (5) Kwerlpe 901 44 11 45 (9) Mootwingee 1,076 41 3 56 (7) Sturts Meadows 13,901 60 2 38 (2) Karolta 11,826 28 35 37 (8)

As originally proposed by Edwards (1971: table 24–1, shaded rows), compared with later additions. Note that these are whole-site ﬁgures because few studies isolate the older engraved component. Sources:(1) Edwards 1971;(2) Franklin 2004;(3) Worrall 1994;(4) Mulvaney and Gunn 1995;(5)Forbes1982;(6) Morwood 1979;(7)Maynard1979;(8) Nobbs and Dorn 1988; (9) Gunn 1994;(10) Rosenfeld 1990

Maynard (1979) suggested an evolutionary schema for Australian rock art, with tracks and nonfigurative art as the earliest stratum, developing into later figurative art traditions. However, there is no theoretical reason why we should not reverse this and see ‘track and circle’ art as a specialised form, derived from a simple figurative art tradition in which tracks made up a small percentage of motifs (as they do in most simple figurative rock art). This may have occurred a number of times, so not all ‘track and circle’ art need be part of one historical tradition. Characterising the Panaramitee-style has also meant distinguishing it from more recent engravings. The most detailed study of relative sequence and superimposition is by Ross (2003) in Central Australia, using data on 13,205 engravings (Table 7.5). Of these, 13 per cent were deeply patinated, 10 per cent were fresh in appearance and the majority of the assemblage (77 per cent) fell between these extremes. Clear superimposition of one engraving over another was limited to 867 cases (7 per cent of the engravings). But Ross was able to combine this with (a) data on weathering and patination, (b) analysis of 232 The Archaeology of Australia’s Deserts

table 7.5. Relative stylistic sequence for Central Australia

Phase Motif range Context

Petroglyphs 1 Group I: Deeply pecked circles (often as Middle gorge areas near repeated motif), concentric circles, circle and waterorinrock pit motifs, macropod and bird tracks (usually shelters. as individual tracks or track pairs). Group II: Simple figurative intaglio or linear motifs, often silhouette or outline figures of birds, macropods or lizards. 2 Group III: Core motifs in Group 1 are repeated Extensive engraved or reincorporated into later art. An extended complexes along gorge engraved repertoire includes complex poles, systems and onto fans spoked circles, radiating fans, radiating circles, – with motifs that ferns, arcs, meandering lines and trails of differentiate sites. macropod and bird tracks. Group V:Rareoruniqueabradedmotifs. Petroglyphs and pigment art 3 Group IV: Stylistically uniform suite of motifs Large art complexes with shared between engravings and paintings – numerous satellite sites including core motifs from earlier periods and with visually with addition of snake motifs, dominant site-specific anthropomorphs with headdresses or motif at central site – weapons, and roughly made bird and oftenonverticalfaces. macropod tracks in long, meandering trails. Engravings are shallow and irregular. A wider range of techniques used to produce motifs – including pecking, painting, incising, abrading, pounding and bruising. Group VI: Large bichrome paintings including striped designs. Group VII: Extensive use of hand stencils and hand prints. 4 Group VIII: Contact-period drawings including Within existing art concentric circles, grids, snakes and barred complexes. ovals as well as postcontact motifs of horses, camels and people with clothes and guns.

Source:Ross(2003)

assemblages on rock surfaces of different geomorphic age, and (c) the sequence of overlays at one or two sites with particularly dense panels of petroglyphs, where differences in patination on the one surface separated the petroglyphs into discrete phases of production (e.g., ASID 292). Her study showed that the earliest engravings are predominantly small (<150 mm) plain circles with broad, fully pecked outlines, circle and pit motifs, irregular concentric circles Rock Art and Place: Evolution of an Inscribed Landscape 233 table 7.6. Key characteristics of Panaramittee-style rock engravings as a graphic system

Content Narrow graphic vocabulary. Dominant motifs: Bird and macropod tracks, circles, arcs, concentric circles and sinuous lines. Other motifs: Solid silhouettes of lizards and birds, human footprints. Rare: Complex geometric designs (nets, rayed designs, spoked circles) or anthropomorphs. Form Small (<10–20 cm), ﬁnely pecked motifs, either solid or using pecked bands 10–20 mm wide. Tracks are quasi-naturalistic, showing sufﬁcient anatomical detail to identify broad divisions between species. Composition Single motifs or track pairs. Short track series, clusters of tracks, clusters of pecked dots (eggs). Tail and track combinations. Frequent repetition of the same motifs at a site. Placement Focal points in landscape. Horizontal or subhorizontal rock surfaces.

and pecked bird and macropod tracks showing anatomical details of heel and toe arranged in single tracks or short series of tracks. There are also some simple figurative motifs on vertical panels, notably intaglio pecked in-fill silhouettes of birds. Later engravings have a wider range of motifs, often have long track meanders and are more schematised, with only narrow pecked lines. This sequence confirmed trends loosely identified in other studies (Wilson 1982;Tacon1994; Rosenfeld and Mumford 1996; Gunn 2000), although Gunn (1995, 2000: 122; 2004) argues that large, thin-line pecked circles (>50 cm) represent an earlier phase of pre-Panaramitee-style engravings (this awaits publication of supporting data). ‘Panaramitee’ remains a useful label for a body of ancient rock engravings in the desert interior, but we need a better formal definition of this art. This should (a) treat Panaramitee art as a visual system, and (b) specifically focus on the early (fully patinated) engravings rather than lump all engravings together. Not all the issues can be addressed here, but I have tried to summarise the current picture in the following discussion (and in Table 7.6). Content. One of the key characteristics of this art is a narrow graphic vocabulary, initially limited to bird and macropod tracks, circles, arcs, concentric circles, sinuous lines and some solid silhouettes of lizards and birds, but later including rare and often site-specific motifs such as complex geometric designs (nets, rayed designs, spoked circles, spirals, mazes, geometric designs incorporating concentric circles) or anthropomorphs (small human figures with headdresses) (Gunn 2000: table 6;Ross2003). Bird and macropod tracks and circles usually dominate the repertoire, but motif frequencies vary among sites and among regions (Franklin 2004). 234 The Archaeology of Australia’s Deserts

Form and composition. The motifs are usually small (<200 mm), finely made pecked designs, either solid or using pecked bands 10–20 mm wide, usually regular in form and deeply engraved (Ross 2003: table 6.2). Bird and macropod track motifs tend to be quasi-naturalistic, showing sufficient anatomical detail to allow broad divisions between species (Forbes 1982: 71; McDonald 1993). Many motifs are used singly or in track pairs, and compositions are limited to short track series, clusters of tracks, clusters of pecked dots (eggs) or tail and track combinations, although dense overlays of engravings occur on many panels. A striking characteristic is frequent repetition of the same motifs at a site. Placement. Where early Panaramitee-style engravings occur, they are clustered in the landscape, usually around ephemeral or permanent water sources, often in the middle reaches of gorges on horizontal pavements or subhorizontal rock surfaces, or in rockshelters (Edwards 1971;Ross2003; Gunn 2004). Later Panaramitee sites often form extensive complexes of engravings along gorge systems. There was undoubtedly also a painted component to Panaramitee-style art, but little of this has survived. One rare find is the remains of a collapsed frieze of red and yellow ochre paintings, dating to 7.5 ka, at Puntutjarpa rockshelter (Gould 1977), although no motifs can be distinguished now. At Puritjarra, the concentration of ochre near the base of the rockshelter wall suggests that production of pigment art on the walls began sometime between 16 ka and 8.2 ka (Smith et al. 1998).

chronology The most vexed question is the chronology of these engravings. Although they are widely thought to date from the late Pleistocene (Edwards 1971; Maynard 1979; Morwood 2002: 40, 57–8;Franklin2004: 141), most evidence points to a mid-Holocene age, with the major period of production probably around 8–4 ka (see Table 7.2; see also Rosenfeld 1993; Bednarik 2010). Direct dating of Panaramitee-style engravings is limited to three case studies: at Sturts Meadows (Eight Mile Creek) in the Barrier Ranges, at the Karolta site in the Olary region, and in Central Australia at Wanga East and Puritjarra.

Carbonate Dating of Varnish at Sturts Meadows. At Sturts Meadows (Eight Mile Creek), an estimated 18,000 rock engravings occur on whaleback rock outcrops in arid shrubland near an ephemeral channel (Clegg 1987). Most are covered by desert varnish. In a widely quoted study, Dragovich dated soil carbonate (3 mm thick) covering manganese-rich varnish on a rock fragment, adjacent to the main engraved pavement (1984, 1986). This showed that varnish was forming prior to 12.3 ka (7,090 ± 310 BP SUA-2011; 10,250 ± 170 BP Beta-13803; 10,440 ± 170 BP Beta-13804). Further carbonate samples gave ages around 3.9–2.8 ka (2,680 ± 380 BP SUA2099; 3,530 ± 210 BP SUA2100) (Clegg 1987). Rock Art and Place: Evolution of an Inscribed Landscape 235

Both the age of the carbonate and its association with the rock art are uncertain. First, the carbonate layer appears to have recrystallised at some stage, so 12.3 ka is a minimum age for varnish formation. Second, microstratigraphic examination showed varnish was also present on the surface of the carbonate and within the carbonate layer itself, indicating that varnish continued to form after approximately12.3 ka, probably into the mid-Holocene. Third, this was an off-art sample: if the engravings were pecked into a surface already covered with desert varnish (as seems likely), then c.12.3 ka may be a maximum age for the petroglyphs. My interpretation is that the Sturts Meadows engravings were pecked into an existing varnished surface during the early to mid-Holocene, at a time when varnish was still actively forming.

Cation-Ratio Dating of the Karolta Engravings. Later work focused on direct dating of varnish within engravings at Karolta 1 – the site that had caught the eye of Mounted Constable Waterhouse in 1902 – near Mannahill in the Olary region (Dorn et al. 1988; Nobbs and Dorn 1988). For this site, cation- ratio dating provided twenty-four age estimates in a more or less continuous series from 31.7 ka to 1.4 ka. These were the first dates that appeared to confirm a Pleistocene age for Panaramitee engravings but were invalidated as the geochemical problems with cation-ratio dating became evident. There were also specific local problems with this dating method: cation ratios at Karolta were calibrated against three AMS 14C dates on organics in the varnish, and neither the ages nor the derived cation-leaching curve could be replicated in an independent study using the same calibration points (Watchman 2000: 266). Other work in the same region showed no significant difference in cation ratios between old and young rock surfaces (Dragovich 1998) and showed that the concentration of cations within a varnish layer was not uniform (Clarke 1989). Because the cation-ratio technique was a poor indicator of either absolute or relative age, later work at Karolta relied on AMS 14C dating of carbon within the varnish, which produced a large series of dates back to 47.3 ka (43,140 ± 3,000 AA6898)(NobbsandDorn1993). These results supported the initial cation-ratio study but were formally retracted (Dorn 1997) after radiocarbon laboratories reported foreign contaminants in these samples (Beck et al. 1998) and as better understanding of the dynamics of rock varnish showed that organics or particulate carbon did not necessarily provide an accurate date for formation of a varnish (Reneau et al. 1991; Dragovich 2000; Watchman 2000). So, what can we make of the Karolta dates? The best we can say is that these results provide a record of variability in varnish chemistry across a rock surface rather than either relative or absolute dates for the engravings. An age of approximately 30 ka is unlikely for other reasons: the motifs, which are on dolomitic siltstone, retain sharp, well-defined edges (a point also made by Bednarik 2010), and measurements of U/Th ratios in manganese oxides in the 236 The Archaeology of Australia’s Deserts table 7.7. Radiocarbon ages bracketing the age of Panaramitee-style rock engravings, Central Australia

Sample Lab code Age BP Association Material

Wanga East M4/2–1 Wk20451 320 ± 35 Layer I sediments covering Charcoal base of Slabs 1 and 2 W1 and W2 To p N Z A 27886 1,443 ± 25 Final phase of oxalate Oxalate formationinengravings W1 and W2 W1 Base NZA27392 3,820 ± 30 Base of oxalate in engraved Oxalate circle W1 W2 Base NZA27393 4,362 ± 30 Base of oxalate in engraved Oxalate circle W2 M5/7 Wk19122 5,021 ± 44 LayerIIsedimentscovering Charcoal base of Slabs 1 and 2 W3 Base NZA27388 7,271 ± 40 Initiation of oxalate on rock Oxalate surface M4/5–1 Wk19121 8,385 ± 52 Sedimentary surface beneath Charcoal Slab 1 Puritjarra P1 and P2 To p N Z A 27555 542 ± 30 Final phase of oxalate Oxalate formationinengravingsP1 and P2 P1 Base NZA27391 2,258 ± 30 Base of oxalate in engraved Oxalate circle P1 P2 Base NZA27387 2,613 ± 30 Base of oxalate in engraved Oxalate circle P2 ST5/10–3 ANU7460 13,570 ± 100 Sedimentary surface beneath Charcoal Block D

See Figure 7.7 for sample points. Source: Smith et al. (2009)

same varnishes show such low Th-230 levels that the engravings are unlikely to be of Pleistocene age (Aubert et al. 2009).

Dating of Oxalate Skins at Wanga East and Puritjarra. Currently, the best- constrained ages for Panaramitee-style engravings are from two sites in Central Australia: Wanga East and Puritjarra rockshelter (Smith et al. 2009). At both sites, engravings occur on the upper surfaces of rock slabs that have fallen from the ceilings of the shelters, overlying archaeological deposits. This allows 14C dating of charcoal preserved on sedimentary surfaces beneath the slabs to provide a terminus post quem for both the rock fall and the engravings. A sequence of AMS 14C ages for laminated calcium oxalate skins covering the engravings provides an upper age bracket for the engravings (Table 7.7). Both sites were chosen because extensive ﬁeld surveys indicated that these were amongst Rock Art and Place: Evolution of an Inscribed Landscape 237

Figure 7.7. (To p ) Plan of Wanga East rockshelter, showing the layout of excavation trenches (M4–5,L5) and engraved rock slabs 1 and 2;(bottom) rock slabs 1 and 2, showing engravings and location of dated oxalate samples (W1–W3). the oldest engravings in the region (based on their condition) and because they are representative of the earliest stylistic phase of rock engraving identiﬁed in relative art sequences in Central Australia (Rosenfeld 1990, 2002; Rosenfeld and Smith 2002;Ross2003). At Wanga East, the engravings cover a series of rock slabs, created by roof fall at the rear of a rockshelter (Figure 7.7). These represent a single phase of pecked engravings with small plain circles, concentric circles, lines, pecked pits, bird tracks and track series. Although no engravings extend below modern ground level, they could not have been made before the rock fall because this created the surfaces used for engraving. Excavation of ﬂoor deposits beneath 238 The Archaeology of Australia’s Deserts

the western end of the rock fall (slabs 1 and 2 in Figure 7.7) showed that these rocks fell onto a sediment surface dated to 9.3 ka (8,385 ± 52 BP Wk-19121). Relatively large fragments of wood charcoal (∼15 mm) are preserved directly underneath these slabs, indicating that they fell on what was then a young surface, protecting friable material from disintegration. Oxalate had started to form on these new rock surfaces by 8.1 ka (‘off art’ at W3, 7,271 ± 40 BP NZA27388). Allowing for time averaging of oxalate samples due to low rates of skin formation (0.25 mm/kyr), the oxalate started to form shortly after these surfaces were created. The build up of sediments buried the lower part of these slabs by 5.7 ka (5,021 ± 44 BP Wk19122). AMS 14Cdatingofthebase of oxalate skins within the engravings shows that these were made before 4.9 ka (4,362 ± 30 BP NZA27393)andwerecutintoasurfacewithanexisting oxalate skin, perhaps to highlight the engraved motifs. Oxalate was actively forming at this time and would have slowly begun to form in engravings shortly after they were made. Allowing for time averaging of basal oxalate samples, the Wanga East engravings date to approximately 6–5 ka and are associated with the basal occupation unit at this site dated 9–5 ka. They cannot be older than 9.3 ka because the rock fall provides a terminus post quem for the rock art. At Puritjarra, which has a long record of late Pleistocene and Holocene occupation, a cluster of large boulders, detached from the rear wall or roof of the shelter, is now partly buried by ﬂoor deposits. These boulders have oxalate-encrusted engravings on their upper surfaces and sides: all are small plain circles, representing a single phase of engraving (Rosenfeld and Smith 2002). Excavations showed that the engraved boulders rest in the upper part of layer II, in levels dating to the terminal Pleistocene (Figure 7.8). Large pieces of Callitris charcoal beneath these blocks indicate that they fell around 16.1 ka (13,570 ± 100 BP ANU-7460). 14C dates for the base of the oxalate in several pecked circles show that these were made before 2.8 ka (2,613 ± 30 BP NZA27387). But because the oxalate skin at this site is thinner (< 0.5 mm) than at Wanga East, time-averaging effects are greater, so the actual basal age of the oxalate may be 5.0 ka. The missing piece of evidence is any direct indication of when oxalate began to form on these rock surfaces. This is unlikely to be more than 8.3 ka because changes in rockshelter sediments indicate that the modern weathering regime was established around this time (Smith 2009b). Oxalate is still forming today at Puritjarra. If we assume that it formed episodically throughout the mid- to late Holocene (as seems likely), then it would have covered the engravings shortly after they were made. If so, the Puritjarra engravings were made around 5 ka, and they cut through an existing skin that had begun to form at 8.3 ka, in a sequence very similar to that at Wanga East.

subsequent history of the tradition Over the subsequent life of the engraving tradition, there appear to have been additions to the graphic assemblage (Rosenfeld and Mumford 1996;Ross 2003; Gunn 2004), with repetition of existing motifs at individual sites and Rock Art and Place: Evolution of an Inscribed Landscape 239

Figure 7.8. Excavation to establish the level of engraved boulders at Puritjarra rockshelter, Central Australia. Pecked circles can be seen at top left. The engravings on this boulder as also shown in Figure 7.6c. (Photograph courtesy of MA Smith) gradual addition of a range of new designs to the overall graphic vocabulary (Table 7.8). Sometime during the late Holocene, Panaramitee-style engravings were replaced by the more stylised forms used in the recent engravings. A date of around 4–5 ka for this change is suggested by several observations. First, at Kurutiti millstone quarry (see Chapter 8), the latest Panaramitee engravings overlap with operation of the quarry and occur within quarry pits, where they must date to less than 4 ka (Mulvaney and Gunn 1995). Second, although dingo tracks are rarely shown in Panaramitee-style engravings, they do occur at Kurutiti (Mulvaney and Gunn 1995), Red Gorge in the Flinders Range (Mountford and Edwards 1964), Thomas Reservoir (Alalya)andRainbow Valley in Central Australia (Ross 2003: 197). This indicates that production of these engravings continued to around 4.6 ka. Third, in the central Queensland highlands, there are a number of open engraving sites with small assemblages of Panaramitee-style engravings. Around 4.8 ka, these were replaced by a regionally distinctive style of rock art in the form of engravings of stone axes and vulvas (Morwood 1979). the social context of panaramitee art Field recording provides an indication of the scale of production of these petroglyphs: in Central Australia, there are an estimated 30,000 motifs at 240 The Archaeology of Australia’s Deserts

approximately sixty sites in an area of 700,000 km2 (Gunn 2000;Ross2003), and in the Olary region, there are more than 12,300 engravings at approximately twenty sites (Edwards 1971; Nobbs 1984), with another 18,000 engravings at Sturts Meadows alone (Clegg 1987). Estimates of the time required to produce a pecked motif range from 13 to 15 minutes (Nobbs 1984;Clegg 1987), suggesting that even dense assemblages, such as those at Sturts Mead- ows, could be produced within a few decades. (Clegg 1987 estimated that the 16,000–18,000 engravings at Sturts Meadow would take 491 days to produce.) Sites often have clear differences in patination of engravings on the same panel, suggesting that there were punctuated pulses of art production. This impression of episodic production is also the case at Puritjarra and Wanga East, where in each case the petroglyphs represent a single phase of engraving. It is possible that the greater part of the engraved assemblage was produced during the initial stages of creating an ‘inscribed landscape’, with only a smaller number of later additions as engraved panels were refreshed and curated. Not all focal points in the landscape were inscribed. However, once a location was marked, people often added to the assemblage: rock art accumulated in folds in the country, and art tended to attract more art. Other studies have looked more explicitly at the social context of production. Panaramitee-style engravings meet many of the criteria for ritual rather than secular production (Ross 2003; Ross and Davidson 2006): an invariant form, repetition, standardised symbols and evidence for performance and par- ticipation. The pitting, incising, flaking and abrading of engravings and nearby rock surfaces recalls practices used by ethnographic groups to ritually reactivate significant natural features. The repetition of existing engraved motifs, often the designs most characteristic of a specific site, is also a trait consistent with the properties of desert totemic systems. But the presence of engravings in highly visible locations, on pavements and panels around waterholes – or in occupied rockshelters – indicates that they do not share the secret, restricted and revelatory aspects of some ethnographic rock art. Nor is the differentiation between sites as marked as in recent rock paintings and engravings – and individual designs are less graphically complex than more recent rock paintings (Frederick 1997). All of this suggests that Panaramitee-style engravings represent the initial phase of marking out a totemic landscape, but one that had not yet developed the segmentation that characterises the classic ethnographic form of geographic totemism.

The Pilbara Art Province Northwest Australia forms a distinctive early art province, stretching from the Dampier coast (including the Burrup peninsula), inland through the ranges and uplands of the Pilbara, with outliers in the interior in small desert ranges such as the Durba Hills and Calvert Range (see Figure 7.5). This area presents Rock Art and Place: Evolution of an Inscribed Landscape 241 table 7.8. Composition of rock art assemblages in the Pilbara, comparing the Dampier–Burrup coast and inland Pilbara

Motif class % Major motifs

Burrup Geometric motifs 40.1 Arcs, ovals, lines, meandering lines Anthropomorphs 25.1 Simple human figures Birds 3.8 – Terrestrial fauna 11.6 Macropods Marine fauna 11.2 Turtle and fish Tracks 8.2 Emu tracks Inland Pilbara Geometric motifs 25.6 Arcs, circles, ovals Anthropomorphs 38.2 Stylised ‘Kurangura’ figures Birds 1.2 – Terrestrial fauna 6.9 Macropods, snakes Tracks 26.7 Macropod and bird tracks (16.7%)

Data from Wright (1968) and Mulvaney (2010). a graphic system structurally different from central Australian art. And it appears to have been a distinctive region since at least the early Holocene, showing that regionalisation in desert art is not exclusively a late Holocene phenomenon. The rock art of this region is known from field surveys on the Pilbara coast by McCarthy (1961) and Ride et al. (1964) on Depuch Island; Petri and Schulz (1951) and McCarthy (1962) on limestone ridges at Port Hedland; Dix and Virili (1977), Lorblanchet (1992), Donaldson (2009) and Mulvaney (2010) on the Burrup Peninsula; in the inland Pilbara by Wright (1968, 1972)and McNickle (1984); and in the Calvert Range by McDonald and Veth (2006, 2011). The Pilbara has the largest concentration of rock engravings in the desert, as well as the richest and most diverse assemblages. The sheer number of petroglyphs is impressive: in a 260 km2 area on the Burrup peninsula and adjacent islands, there are an estimated 270,000–435,000 engravings (of which 30,000 have been recorded); there are another 5,000 on Depuch Island, 15,000 on coastal limestone ridges at Port Hedland, 4,845 in the inland Pilbara and a few thousand in the Calvert Range. This is the largest concentration of petroglyphs in any of the major Southern Hemisphere deserts. Unlike the austere ‘track and circle’ art of central Australia, rock art assemblages in the Pilbara region are dominated by human figures and naturalistic figures of animals (see Table 7.8). The older patinated engravings in this region probably date to the Holocene transition (14–8 ka). The early assemblage includes large, complex geometric designs, concentric circles and arcs, large pecked and abraded outlines of macropods and emu, silhouettes of birds and small dot-headed anthropomorphs, large human figures with elaborate internal 242 The Archaeology of Australia’s Deserts

Figure 7.9. Examples of motifs in early petroglyph assemblages in the Dampier– Burrup area: (a) complex geometric design with bird tracks at top RHS; (b) large decorative in-ﬁll anthropomorph; (c) thylacine; (d) pecked and abraded outline macropod – surrounded by fully pecked turtles, macropod tracks and a pecked meander (LHS). (Source: a and d, Lorblanchet 1992:ﬁgs.29 and 32; b and c traced from photographs)

decoration (often >1 m long) and engravings of the striped dog-like thylacine (Figure 7.9). There are also speciﬁc repeated compositions, such as the ‘climbing men’ design (Figure 7.10), and iconic motifs, such as ‘archaic faces’ (Fig- ure 7.11). A core group of these motifs (pecked and abraded outline macropods and emu, archaic faces, large decorated in-ﬁll anthropomorphs, complex geometric panels and engravings of thylacines and the ‘climbing men’ design) is shared among the Burrup, inland Pilbara and Calvert Range. The internal variability of this large region has not been established but, as McDonald and Veth note, ‘clear chains of stylistic connection can be demonstrated from the Rock Art and Place: Evolution of an Inscribed Landscape 243

Figure 7.10. The ‘climbing men’ motif, found in early petroglyphs on the Burrup peninsula and in the Calvert Range. (Traced from photograph)

Pilbara coast to the desert interior with distinct and stylistically unique rock-art bodies’ (2008: 4). chronology The chronology of the Dampier engravings remains uncertain, although work in progress may soon provide better age estimates for rock weathering and varnish and carbonate formation in this region. In the interim, rock art researchers have generally argued for a late Pleistocene age for the earliest engravings, based on the deep weathering of some engravings, the lack of marine motifs and the assumption that a long chronology is necessary to accommodate the sequence of styles (Lorblanchet 1992; Mulvaney 2010). Mulvaney (2010: 215)boldly posits a date of 30 ka for some of the Dampier rock art (arguing that the Burrup was an oasis during the last glacial maximum), whereas, at the other extreme, Bednarik (2010) argues none is earlier than 8 ka (on the basis that the art is associated with the modern coastline). One line of evidence concerns the age of the desert varnish on these rock surfaces (Pillans and Fifield 2010). Wherever engravings are associated with varnish, they cut through it and must be younger than the varnish. In a few cases, however, small flecks of varnish have formed within engravings (including motifs of anthropomorphs with archaic faces), and this suggests that the earliest engravings were made while the varnish was still forming, albeit weakly. Two studies have tested the idea that the Dampier varnish is late Pleistocene in age (Dragovich 2000; Pillans and Fifield 2010). In the first, AMS 14C of carbon inclusions showed that these varnishes had either formed or had been extensively remobilised within the last 2.8 ka (Dragovich 2000). In the second, dating of soil carbonate overlying varnish gave 14Cages between 40 ka and 20 ka, with the oldest samples dating to 29.7 ka and 37.7 ka 244 The Archaeology of Australia’s Deserts

(Mulvaney 2010: 95). Because formation of these laminated carbonates ceased around 20 ka, they are not well placed to identify more recent episodes of varnish formation. However, if the major period of varnish formation (in this area) was controlled by aridity during the late Pleistocene, then it follows that primary varnish formation is likely to have ceased when the summer monsoon circulation switched on around 15–14 ka and intensified 11– 10.5 ka. The implication is that the oldest rock engravings date somewhere within the period 15–10.5 ka and that most of the art assemblage is Holocene in age. A second line of argument relies on evidence for a wetter climate and higher regional water tables in the Burrup area during the late Pleistocene – an argument that Mulvaney (2010) advances to reinforce the notion that the petroglyphs date to 30–25 ka. These palaeoenvironmental reconstructions hinge on which factors controlled the deposition of carbonates on Burrup Peninsula. These deposits range from soil carbonates on perched terraces to massive travertine deposits 1–2 m thick along some of the stream channels. There is little carbonate in the local rock: in this area, carbonate deposition is supply-driven rather than groundwater-controlled. The likely source is aeolian, sand-sized carbonate blown from the exposed continental shelf. Carbonate sands form extensive coastal dunes and linear sandridges along the Ningaloo coast to the south. Prior to the marine transgression, these formations probably extended along the Dampier coast. Aeolian carbonate is readily reworked by seasonal rainfall and would carry in solution through the open block piles that form the ranges to precipitate in the stream channels and saturate soils on the terraces: there is no need to invoke wetter conditions to account for this process. An aeolian source is consistent with radiocarbon dates for carbonates late in MIS3, before rising seas drowned the coastal dune systems, and with regional pollen records from the Fr10/95-GC17 core that show an open, arid vegetation and intensified aridity from 32 ka with virtually no summer rainfall by 25–23 ka (van der Kaars et al. 2006). During the late Pleistocene, the Burrup area is likely to have been visited by hunter-gatherer groups using seasonal seepages and rockholes, but there is no evidence that the area was an oasis, and it is unlikely to have been densely populated until the early Holocene. A third line of argument is whether the early engravings predate the rise in sea level, which began to flood this area around 9.5 ka. This was tested by underwater survey, searching areas where rocky slopes continued below sea level (Dortch 2002). Despite the abundance of engravings on nearby slopes, none was found on submerged slopes drowned by the rising sea. The oldest engravings are also rare on outlying islands in the archipelago, such as Rosemary Island and Depuch Island (Ride et al. 1964; Donaldson 2009; Mulvaney 2010). This is hardly conclusive, but it is consistent with other indications of an early Holocene rather than late Pleistocene age for the art. Rock Art and Place: Evolution of an Inscribed Landscape 245

Figure 7.11. Archaic face engravings: (a) one of the engraved blocks at Alalya (Thomas Reservoir), Cleland Hills, showing the face as part of a complex geometric design with afﬁnities to similar motifs on Burrup peninsula; (b) and (c) Burrup peninsula; (d) Cle- land Hills; and (e) Gap Hole, Simpson Desert. There are stylistic continuities between the heart-shaped face in (a) and (c), and between (d) and (e). [Source: Photograph in (a) is courtesy Robert Edwards; (b–d) are traced from photographs; (e), Ross and Smith 2009] direct dating of the engravings Direct dating of rock engravings is limited to a few ﬁnds in the Skew Valley (Dampier), the Calvert Range and the upper Yule River area (inland Pilbara) (Table 7.2). These show a graphic system that is increasingly visible in 246 The Archaeology of Australia’s Deserts

archaeological contexts in the Holocene, but there are no secure ages for the older, fully patinated petroglyphs.

The Skew Valley Midden. The Skew Valley midden lies against the base of a block slope with dense concentrations of engravings, with an estimated 5,000– 6,000 petroglyphs within 100 m of the midden. Archaeological excavations recovered four engraved slabs from the Anadara midden from levels dating to 3.6–2.7 ka (Lorblanchet 1992). Two of these have groups of human ﬁgures in a style and composition similar to those in Mulvaney’s phase 4, which suggests they were produced as part of the occupation of the late Holocene Anadara middens rather than being much older fragments incorporated into younger shell middens. Engravings also cluster around another Anadara midden dated 3.5–1 ka at Gum Tree Valley. A Syrinx shell at Gum Tree Valley dated to 21.7 ka is widely taken to show late Pleistocene use of the Burrup. However, this is a surface ﬁnd and probably represents reuse of an older shell as a water container during occupation of the late Holocene midden nearby: it has no demonstrable association with the rock art.

Inland Pilbara. On the upper Yule River, Bednarik (2002) used microerosion rates to estimate the age of several petroglyphs at Woodstock. The oldest motifs were pecked circles (or circle compositions) dating to 26.8 ka and 19.4 ka, but Bednarik warned that the technique does not give secure or precise ages at present.

Calvert Range. The Calvert Range is an area where the Burrup chronology can be tested because it potentially has the entire history of art production, uncomplicated by the last marine transgression. However, excavations so far have only produced equivocal ages for the earlier petroglyphs. An engraved panel at site MV1 contained an archaic face partly buried by sediments dating to 1 ka (1,050 ± 110 BP Wk-8807; McDonald 2005: 127). Archaeological excavations at Kaalpi rockshelter recovered an engraved fragment – with two pecked bands – from levels dating to 2.2 ka (Veth et al. 2001). Excavations at a third site, Bush Turkey 3, showed that rock fall containing panels of engravings (with tracks, circles and the ﬁgures of a human and some bush turkeys) dates between 8.4 ka and 3.2 ka (Veth, McDonald and White 2008). This work points to the potential for recovering older buried engravings in the Calvert Range, but there is clearly more to be done.

stylistic sequence on burrup peninsula The most detailed quantitative and stylistic study of the rock art is by Mulvaney (2010) in the Burrup region, using a sample of 5,650 motifs in seventeen art complexes across the Dampier archipelago. This area is a drowned coast, with partly inundated hills and ridges forming an archipelago of forty-two islands, Rock Art and Place: Evolution of an Inscribed Landscape 247 of which Burrup ‘peninsula’ is the largest. In this unusual landscape, the hills and ridges are barren block piles of deep red gabbro and granophyre boulders separated by low valleys following jointing in the base-rock. Local rainfall (261 mm per annum [p.a.]), most of which falls seasonally in summer, collects in the open rock piles to recharge seepages and rockholes in the channels, but self- replenishing springs do not appear to be present. Since the 1960s, the Burrup area has been the focus of industrial development, with the establishment of a deepwater port for exporting iron ore from the Pilbara and onshore facilities for handling natural gas from the North West Shelf gas fields. What is now Burrup peninsula was formerly Dampier Island until a causeway was built across the coastal saltflats that connect it to the mainland. Mulvaney (2010) uses differences in style, patination, superimposition and content to outline a relative sequence of five art phases (Table 7.9). In broad terms, this sequence documents a shift from patinated to fresher engravings, a shift from terrestrial to marine subjects and a trend to more stylised and schematic motifs. He suggests that the long sequence of styles is evidence in itself that the rock art spans the last 30,000 years. So, how well founded is this sequence? Two key issues are whether there are intrinsic reasons for grouping the art in each phase, and whether or not phases are clearly segregated in field data. Coherence in style. The data make it clear that these are polythetic style phases, grouped on patination. The marked diversity of styles within each phase suggests an art that is internally differentiated in context, production and visual conventions. There are stylistic links between some motif classes in a phase. However, each phase also includes groups of motifs that use quite different stylistic conventions, and there are traits that cut across phases (e.g., in use of deep pecked lines or links between decorative in-fill anthropomorphs and other outline anthropomorphs). The difficulty of grouping the early art is reflected in other studies that assign a different relative order to key early motifs (Lorblanchet 1992). Although there is a clear separation between the more stylised recent marine art and the older naturalistic art, style alone provides little intrinsic reason for separating phases 1 and 2. Patination. Phasing of the Burrup art relies on accurate identification of patination state. Mulvaney (2010) relies on a subjective assessment of patination state (in this case, rock weathering), coding ‘contrast state’ by eye on a five- point scale. It is doubtful whether this yields reproducible results. But a more serious problem is that irrespective of the reliability of the method, it does not provide clear separation of art phases (Figure 7.12 and Table 7.10). Superimposition: Superimposition provides another way of sequencing the rock art, although not one that is fully independent of patination state. Unfor- tunately, there are few cases of superimposition in the Burrup assemblage (557 superimposition pairs, or only 6.9 per cent of the assemblage). Seriation of this dataset supports the general trend of the style sequence but not the detail 248 The Archaeology of Australia’s Deserts

table 7.9. Relative stylistic sequence for petroglyphs in the Burrup region

Phase Motif range Context

1 Archaic faces; complex geometric designs; Main ‘valley’ systems concentric circles, ovals and arcs near potable water. Anthropomorphs with decorative in-fill Prominent vertical ‘Climbing man’ motif (and other ‘dot-headed’ faces. human figures) Simple birds and quadrupeds including thylacines 2 Large macropod and bird tracks Main ‘valley’ systems Large outline emu, macropods and near potable water. anthropomorphs Prominent vertical Thylacines and marine mammals faces. 3 Marine fauna Littoral. Stylised macropod and bird tracks Stylised outline and solid images of birds, macropods, reptiles and anthropomorphs (including ‘rainbow man’ motif) 4 Outline and internal-design fish and turtles, pit Littoral. clusters (turtle eggs), marine mammals Simple macropod tracks, aquatic bird tracks Dynamic grouped human figures Human figures with headdress designs and objects Shallow stylised bird and macropod tracks 5 Human figures with exaggerated hands, feet and Littoral. genitals Wide range of marine species including stingray, crab, turtles, fish and crayfish; bi-lobed ‘sting-ray liver’ motif Stylised birds Simple anthropomorphic stick figures

Source: Mulvaney (2010)

table 7.10. Variation in patination for selected motifs in Burrup rock art. Data are number of motifs in each category

Contrast state (patination) Motif 1234 5

Tur tle 6127212984 Fish 4 1 49 68 64 Thylacine 239 61 Decorative in-ﬁll anthropomorph 637 30 Archaic face 29 12 19 2 0 Complex geometric 4989102

Source: Mulvaney (2010) Rock Art and Place: Evolution of an Inscribed Landscape 249

Figure 7.12. The Burrup sequence showing motif groups arranged by phase and patination (‘contrast state’). (Source: Mulvaney 2010:ﬁg.11.2)

(Table 7.11), showing marine motifs as the most recent, with arcs, thylacines and decorative in-fill anthropomorphs in the oldest art. Content: Some subjects depicted in the rock engravings provide an intrinsic guide to chronology, including extinct species such as thylacines and fat-tailed macropods, and marine motifs such as fish, turtles, stingrays and sea mammals. Of these, marine motifs are the most critical for the art sequence. Marine subjects are not common until phases 3, 4 and 5. The crucial question is whether this change registers the formation of Dampier archipelago, beginning 9.5 ka, or whether it reflects the formation of marine-mangrove economies 250 The Archaeology of Australia’s Deserts

table 7.11. Testing the style sequence for Burrup rock engravings

Overlain %Heavily Seriation order n times patinated

Anthropomorph ‘stick’ figures 03 Marine fauna 46 1 Anthropomorphs with exaggerated hands, feet or genitals 21 4 Circle motifs (including concentric circles) 24 41 Simple geometric motifs 40 16 Lines and meandering lines 35 14 Simple human figures 24 8 Complex geometric designs 23 70 Zoomorphs (nondiagnostic animal or bird) 22 52 Ovals 29 14 Outline or solid birds 32 14 Outline or solid macropods 101 30 Anthropomorph compositions (rows, pairs, clusters) 20 2 Thylacines and other quadrupeds 14 10 Arcs (including concentric arcs) 20 24 Anthropomorphs with decorative in-fill 10 70

The first column shows seriation order, derived from the first component of a correspondence analysis using MULTIV, based on superimposition data in Mulvaney (2010: table AII.2). Motifs in bold are those assigned to the earliest phase of the Burrup sequence. Rows 1–3 include motifs assigned to the most recent phases of engraving. Seriation supports the general trend shown in Table 7.9 but not the detail. Column 1 is compared with other proxies for relative age in columns 2 and 3. Note that there is little agreement between number of overlays and extent of patination (correlation coefficient r = 0.12).

on this coast around 8–6 ka. The presence of some marine subjects (mainly turtles) in the earlier art suggests that proximity to a coast was already a factor during phases 1 and 2, and that phase 3 is correlated with the establishment of a marine-mangrove economy represented by the Terebralia middens 8–6 ka (see Chapter 6). The presence of thylacine depictions in the Burrup petrogylphs into phase 4 (see Table 7.10) shows this overlapped with the last thylacines in the region around 3.5–3.3 ka (see Table 6.8). If so, phase 4 must correlate with the Anadara middens dating to 4–3 ka (see Chapter 6). Further work may resolve some of the difficulties associated with building a robust relative art sequence on the Burrup peninsula. The safest interpretation at present is that we have a pre-marine art assemblage (conflating Mulvaney’s phases 1 and 2)probablydatingto14–8 ka, with a shift to an early marine art assemblage (phase 3) associated with archaeological evidence for well- established mangrove economies on this coast 8–6 ka and a later marine art assemblage associated with late Holocene Anadara middens (phases 4 and 5) (Lorblanchet 1992; Mulvaney 2010). Beyond the Burrup, this general sequence is supported by finds in the inland Pilbara and the Calvert Range. Deeply Rock Art and Place: Evolution of an Inscribed Landscape 251 pecked and abraded outlines of macropods and of emu, thylacines and large decorated anthropomorphs are among the earliest petroglyphs in the inland Pilbara (Wright 1968; McNickle 1984), whereas in the Calvert Range, the earliest engravings are intaglio bird and macropod tracks, circles and dots, thylacines, large decorative in-fill anthropomorphs, archaic faces, simple pecked silhouettes of animals and large abraded outline macropods and emu (McDonald 2005; McDonald and Veth 2006). social context of the pilbara engravings Few studies look explicitly at the social context of production of the early Pilbara petroglyphs. The large-scale production of petroglyphs on the Burrup, where about 30 per cent of the assemblage falls into the pre-marine group (Mulvaney 2010: appendix AI.4), suggests a marked focus on the Pilbara coast, with lower levels of rock art produced in the inland Pilbara. The emerging picture is of well-established hunter-gatherer societies on the Pilbara coast by 14–8 ka, at a time when settlement in the interior was still recovering from rapid climatic changes during the Holocene transition. Early Holocene groups in the Pilbara were using their rock art in a radically different way from those in the rest of the arid zone. The region appears to have more in common with tropical northern Australia than with the desert. In contrast to the desert interior, Burrup rock art is zonal rather than place-based: there is little differentiation of sites and few dense panels of engravings, and the rock art is broadly distributed across the landscape on isolated boulders, across hillsides and in gorges, rather than in discrete focal localities. Unlike the desert, there is more evidence for continuous (rather than punctuated) production of rock art. And production of the larger figures must represent a significantly greater investment of time, especially in the hard granophyre rock of the Burrup peninsula. A key characteristic of this art is the recurrent use of iconic motifs (such as the ‘climbing men’ design, found 600 km apart in the Burrup and in the Calvert Range, or the archaic faces, which occur at multiple locations on Burrup peninsula alone) and emblematic designs (such as complex geometric motifs and large decorated anthropomorphs), often placed on prominent vertical rock faces. This pattern is more characteristic of a coordinated system with an overarching ideology than a segmented totemic religion. It is relevant that although much of the rock art is sited to be visible, in full view of communities using the Burrup and the Calvert Range, the diversity of styles within the early petroglyphs suggests an art that is also internally differentiated in context, production and visual conventions. With the shift to a marine-mangrove economy on the Pilbara coast in the mid-Holocene, many of the structural features of the early rock art persisted. Although it remains widely distributed in the landscape, the later rock art is 252 The Archaeology of Australia’s Deserts

more strongly oriented towards the littoral zone and shell middens. Marine subjects were added to the repertoire, and the strong naturalistic and ﬁgurative character of the rock art shows a continuing concern with the productivity of the landscape. Repecking and abrading of engravings was more common in this phase (Lorblanchet 1992), which shows that revitalisation of existing designs was important to these groups. However, the development of the marine economy also appears to have weakened the ideological underpinnings of the existing society, and many of the most iconic or prominent motifs of the earlier art were no longer produced. This is also reﬂected in changes in the regional expression of the rock art. The early art is widespread, linking the Burrup, inland Pilbara and Calvert Range, but later art shows a diversity of styles within the Pilbara (Wright 1968; McDonald and Veth 2011), and the Calvert Range may have been absorbed into the desert graphic system by this time.

The Problem of the ‘Archaic Faces’ The archaic faces, one of the most enigmatic groups of engravings in Australia’s deserts, form a link between the two early graphic traditions. These engravings were initially reported in 1961 by prospector Michael Terry, who controversially claimed they were of Egyptian or Assyrian origin. The key site, Thomas Reservoir (Alalya) in the Cleland Hills, 16–18 km north of Puritjarra rockshelter, was investigated by Edwards (1968, 1971), who showed that the sixteen ‘faces’ were a small part of an assemblage of Panaramitee-style rock engravings. They remained very much an anomaly, however, until later work recorded similar pecked engraved faces in the Durba Hills and Dampier Archipelago (Dix 1977; Mulvaney 2010); in the Calvert Ranges (McDonald 2005) and inland Pilbara (Brady and Carson 2012); at Jalijbang on the northern margins of the desert (David et al. 1992); and on the northeast margin of the Simpson dunefield (Ross and Smith 2009) (see Fig- ure 7.5). This shows a pattern of occurrence at isolated and widely separated sites across the northern half of the desert, from the Pilbara coast inland to the eastern part of the Lake Eyre basin. The engraved faces are linked stylistically through the use of a bas-relief technique to form the design (a technique rarely used in desert petroglyphs), a full-frontal perspective and with eyes emphasised by the use of multiple graphic elements (often pit eyes enclosed by an outline) (Dix 1977; David et al. 1992; McDonald 2005). However, these faces tend to be highly individual (see Fig- ure 7.11): they are not standardised motifs, nor are they a single recurrent iconic design. Some faces are highly schematised, with eyes set within a geometric design (see Figure 7.11a); sometimes the eyes are emphasised by concentric circles. Some examples in both the Burrup region and the Cleland Hills have heart-shaped heads, often incorporating a nose (compare Figures 7.11a and c). In others, the face is U-shaped with eyes on the ends and an oval-shaped mouth Rock Art and Place: Evolution of an Inscribed Landscape 253 at the base (compare Figures 7.11d and e). Some have headdresses of rayed lines or antennae-like head appendages. They can form disarticulated faces, have bodies forming a decorative in-fill anthropomorph or borrow elements from complex graphic designs (McDonald and Veth 2006) (see Figure 7.11a). The most distinctive variations are repeated in widely separated sites, and recent trait analyses do not show clear clustering by geographic region (McDonald 2005; Mulvaney 2010). By far the largest concentration of archaic faces and the greatest diversity of face motifs occur on Burrup peninsula (n = 64), where Mulvaney (2010) distinguishes four types of faces on differences in graphic complexity. Because the more complex face motifs (such as Figure 7.11b) are not found outside the Dampier region, he argues that other ‘archaic faces’ represent convergence in design rather than a shared tradition. However, when compared to the Panaramitee-style rock engravings of the desert interior, the engraved faces stand out as a group distinguished by subject, technique, stylistic conventions and composition. There is a case, therefore, for seeing the Burrup region as the centre of dispersal of these designs, with loss of the more complex designs as the tradition spread from the coastal Pilbara across the arid interior. If so, they should be among the earliest petroglyphs in Central Australia. So far, all attempts to date these engravings have been unsuccessful. In one of the best stratigraphic studies, David et al. (1992) suggest the Jalijbang faces are likely to have been produced within the last 4.5 ka. Here, the archaic faces are engraved on the upper faces of rock fall within a gorge. Radiocarbon dating of hearths in silty sediments dammed by this rock fall shows that it was in place before 0.9 ka (1,010 ± 60 BP Wk-1550). However, fluvial gravels below this date suggest that water flow may have removed earlier sediments: this leaves the age of the rock fall unresolved. An AMS 14Cdateof4.5 ka on organics within the weathered cortex of the rock fall can be discounted: organics are likely to be mobile within rock cortex and need not date initial exposure of these surfaces. Without a more secure fix on the age of the rock fall, the antiquity of Jalijbang engraved faces will remain unknown. If archaic faces originated in the Burrup area, this would show that cultural influences from the vibrant early Holocene hunter-gatherer societies on the Pilbara coast reached deep into the interior. What social mechanism could have created such widespread distribution of these motifs? One possibility is that the archaic faces are an analogue of ethnographic travelling cults, like the Tingarri traditions that traversed the northern part of the desert. Contemporary Tingarri traditions are specific song-myth cycles, involving travelling groups of men and novitiates, and underpin the ritual education of men. Originating in the Arnhem Land Kunapipi ceremony, they were transferred to the desert via the Kimberley and now represent a super-totemic tradition that overlies the totemic geography. Although specifics will certainly differ, travelling cults provide a mechanism for dispersal of cultural traits over long distances and across cultural boundaries. 254 The Archaeology of Australia’s Deserts

table 7.12. Comparison of older and younger assemblages of rock engravings, Central Australia

Older Younger (patinated) (unpatinated) engravings engravings Motif class % %

Tracks 35 37 Circles and concentrics 18 14 Pecked pits 610 Other 11 – Geometric elements 18 32 Simple designs 83 Complex designs 11 Anthropomorphs 33 Contact period motifs – 1 To t a l n 11,118 1,371

‘Simple’ and ‘complex’ designs refer to motifs constructed from multiple geometric graphic elements. ‘Other’ includes outline or silhouette depictions of birds, lizards and macropods. Source: Gunn (2000)

ROCK ART OF THE LAST MILLENNIUM

Central Australia during the Last Millennium In Central Australia, most of the surviving paintings and hand stencils – as well as the unpatinated petroglyphs – probably date within the last millennium. There are no direct radiocarbon dates for the pigment art yet, but many of the art surfaces consist of friable, actively eroding sandstone able to carry only relatively recent rock art. A last millennium age is suggested by the concentration of fragments of ochre in archaeological deposits dating to less than 1 ka (Smith 1988; Thorley 1998b; Smith and Ross 2008a), the association between types of excavated pigments and adjacent wall art (Rosenfeld and Smith 2002; Smith and Ross 2008b) and the correlation between single-phase art and occupation sites (e.g., Napton and Greathouse 1985). There are strong continuities between earlier Panaramitee-style engravings and the recent rock art. This is most evident in a common graphic vocabulary and in the syntax of simple compositions (Table 7.12). Ross (2002: 86) found that 78 per cent of motif classes in Central Australia were shared by both painted and engraved art (as Nobbs [1984] also found in the Olary region). In this context, Panaramitee art is ancestral to more recent Central Australian decorative art and forms a long graphic tradition with a persistent core of motifs. Today, this art is familiar to anyone admiring the stunning Rock Art and Place: Evolution of an Inscribed Landscape 255

Figure 7.13. Diagram showing the sequences of changes in Central Australian rock art. (Source:Ross2005:fig.16.5) contemporary acrylic paintings of the Papunya Tula art movement (Bardon 1979, Sutton 1998). Stylistic differences in common motifs distinguish recent engravings from earlier Panaramitee engravings. In the former, the tracks are more stylised (bird tracks are basic linear forms or T-bar motifs, and macropod tracks are often shown as paired dashes), circles are often larger and less regular, and pecked lines are thin and shallow. The rough form of recent engravings suggests that it was the act of producing the motifs, rather than the finished form, that was important. The most striking change during the last millennium was the addition of a large corpus of painted art and stencils, amounting to a five-fold increase in the quantity of rock art (see Table 7.3)(5,920 motifs/ka, up from 1,696 motifs/ka), as well as the use of a more diverse range of techniques including drawing, stencilling, printing, pecking, painting, pounding and abrading (Gunn 2004). Some of the apparent increase is due to taphonomic factors, such as the poor survival of older pigment art. But the association with other lines of archaeological evidence, such as an increase in pieces of ochre in archaeological deposits of this age, suggests that there was a material expansion of pigment art production during the last 500 to 1,000 years. Both the pigment art and the engravings show that the graphic repertoire expanded during this period (Figure 7.13), with the appearance of fundamentally new designs and more complex motifs and compositions (Gunn 1995, 2004;Frederick1997; Galt-Smith 1997; Rosenfeld and Smith 2002;Ross 2003). The graphic vocabulary broadened to include motifs with no equivalent in the earliest engravings: for example, anthropomorphs with stylised headdresses, long meanders of macropod tracks (Figure 7.14), hand stencils 256 The Archaeology of Australia’s Deserts

Figure 7.14. Part of the long painted frieze at Puritjarra rockshelter, Central Australia. This shows a section of the long macropod track meanders that dominate the southern section of the rockshelter. Numbers along the base are distance in metres from south to north along the base of the shelter wall. (Source: Rosenfeld and Smith 2002:ﬁg.5)

and handprints (Figure 7.15), and large striped or bichrome paintings, the most famous of which are those at Emily Gap (Anthwerrke) (see Figure 7.4). This broader vocabulary underpinned greater intersite differentiation of art assemblages during the last millennium (Table 7.13). In Central Australia, hand stencils dominate rock art assemblages wherever the regional geology is dissected sandstone, often occurring in hundreds at individual sites. The long art frieze at Puritjarra rockshelter includes 279 hand stencils and 47 handprints, both of children and adults (Rosenfeld and Smith 2002). The recent art is also more internally partitioned, with evidence for production in a range of social contexts. Hand stencils and handprints reflect a highly individual form of expression. The long track meanders and track clusters appear to be secular art. The larger painted and engraved designs are totemic designs. The latter are large (60–98 cm) motifs; often on prominent vertical panels, these are usually site-specific and have been repeatedly repainted or retouched (Gunn 2003). By the last millennium, the character of the sites had also changed, forming larger, diversified complexes of art and occupation sites, with satellite art sites clustering around a major focal art site (Ross 2003). These complexes show a clearer segregation of secular and sacred motifs and more localised ritual practices within art complexes (Ross 2003). Rock Art and Place: Evolution of an Inscribed Landscape 257 table 7.13. Matrix of similarities between rock art assemblages in Central Australia, using Euclidean distance coefficient. There is greater inter-site differentiation amongst (a) recent (petroglyph and painting) assemblages than amongst (b) archaic (petroglyph only) assemblages. Mean Euclidean distance coefficient in (a) is 0.2209 compared to (b) 0.1225. Source: standardised site data from Ross (2003).

(a) Recent rock art assemblages NDhala Wallace Roma Joker Gorge RH Gorge Gorge N 344 61 1027 50 NDahla Gorge – Wallace RH 0.2003 – Roma Gorge 0.1983 0.2255 – Joker Gorge 0.2359 0.2179 0.2474 –

(b) Archaic petroglyphs NDhala Wallace Roma Gorge RH Gorge N 2311 739 3902 NDhala Gorge – Wallace RH 0.1321 – Roma Gorge 0.1202 0.1182 –

Figure 7.15. Hand stencils and hand prints are dominant features of recent art assemblages in Central Australia. This shows a panel from a site on the Watarrka plateau. (Photograph courtesy of MA Smith) 258 The Archaeology of Australia’s Deserts

Finally, the last millennium saw increased differentiation among regions, as well as among sites. Across much of the Central Australian Ranges, emblemic totemic art is represented by large bichrome paintings and striped designs, but in the Mann–Musgrave Ranges, large figurative motifs take their place (Gunn 2004. This period, therefore, is marked by increasing differentiation of rock art on a range of scales: the context of production, the differentiation of specific sites and locales, and increasing regional identity within areas that otherwise share the same graphic system. All of this suggests that by the last millennium, the segmentary religious system had taken shape, with geographic totemism in its classic ethnographic form. This was an elaboration of an older form of totemism represented by Panaramitee-style engravings towards one that was more strongly corporate in character (Layton 1992: 236; Galt-Smith 1997; Rosenfeld and Smith 2002; Gunn 2004). The last millennium itself is unlikely to have been a uniform period but, at present, it is difficult to distinguish its internal dynamics.

large ceremonies Excavations at large open sites in Central Australia show that ceremonies associated with this segmentary system also began within the last millennium. The best known of these sites is Therreyererte, where large ceremonial gatherings associated with the atylepe (native cat, Dasyurus geoffroii) traditions drew in several hundred people (see Kimber and Smith [1987] for an ethnographic reconstruction of one of the last large ceremonies at this site). Such large gatherings were often dependent on the harvesting and stockpiling of grass seed and other grain. At Therreyererte, the archaeological remains make up an extensive open site on a fan at the entrance to a gorge system on the northwestern edge of the Simpson Desert. The scale of the ethnographic gatherings at Therreyererte is corroborated by dense concentrations of stone artefacts (100/m2) and finely fragmented animal bone, the impressive number of large grinding slabs (620 grindstones, including 102 millstones and 152 mortars) and rock engravings present on small loose boulders across the surface of the site and on the grindstones themselves. Excavation showed that this mantle of archaeological debris forms a discrete layer, with dense concentrations of stone artefacts and flaking debitage (10,765/m2/ka); fragmented, burnt and calcined macropod bone; charcoal; red ochre; fragments of grindstones and seed-grinding implements (90 fragments/m2/ka); ground-edge axes; and features such as earth ovens (Smith 1988). Two radiocarbon dates bracket the base of this unit (400 ± 50 BP SUA-2520; 1,830 ± 110 BP SUA-2519), showing that it began to accumulate around 0.60 ka. The presence of a large assemblage of older Panaramitee-style rock engravings within the gorge suggests that Therreyererte was an important totemic centre before this time, but the large ceremonies that are an integral part of reenacting the classic atyelpe traditions are comparatively recent and appear to have begun about 600 years ago. Large Rock Art and Place: Evolution of an Inscribed Landscape 259 ceremonial sites like Therreyererte confirm the picture of structural changes in the totemic system in Central Australia during the last millennium that is emerging from rock art studies.

The Periphery of the Central Australian Graphic System the eastern margins On the desert margins, Panaramitee-style engravings were replaced by different graphic traditions in the late Holocene throughout eastern Australia. In the Mt Isa region, Ross (1997) records a distinctive, recurring painted anthropomorphic figure (usually bichrome and with a ‘fern’ headdress) that appears in the rock art during the last millennium, with slightly different but standardised forms in adjacent geographic areas. This region was the source area for stone axes used in long-distance exchange in the Lake Eyre basin, and Ross argues that use of this recurring motif signals a more bounded group identity as a response to increased interaction between groups involved in this exchange network. Because this painted motif also matches ceremonial body decorations recorded for local Kalkadoon people, it may represent not just increasing regionalisation but also the development of an overarching super- totemic ideology along the lines described in the ethnography (Berndt 1974). In the central Queensland highlands, in the upper catchments of the inland river systems, Panaramitee-style engravings were replaced around 4.8 ka by a highly distinctive regional art made up of large painted geometric motifs (grids, zigzags, lines and dots), stencils of objects and hands, some spectacular full-body stencils, abraded motifs of human and animals tracks and painted and abraded vulvas (Morwood 1984). Much of this is associated with mortuary evidence, and Morwood argues that it reflects a basic restructuring of institutions concerned with production of rock art and increasing social differentiation (1979, 1984). In the southeastern sector of the arid zone, Panaramitee-style engravings in the Barrier Range also appear to have been replaced by a recent painted and engraved art assemblage dominated by figurative motifs of humans and animals and by hand stencils (McCarthy and Macintosh 1962;Maynard1976). This assemblage has closer affinities with the simple figurative art on the Cobar pediplain, where it dates to within the last 500 years (McCarthy 1976b). the margins of the western desert The margins of the Western Desert are the most stylistically complex parts of the desert, open to significant influences from adjacent parts of the Pilbara, southern Kimberley and Gascoyne/Ashburton art areas (as Strehlow [1964] also noted). The best-studied area is the Little Sandy Desert, known mainly through research by McDonald and Veth (McDonald 2005; McDonald and Veth 2008, 2011;Vethetal.2008), although the rock art is yet to be reported 260 The Archaeology of Australia’s Deserts

in detail. Here, the rock art is concentrated in small focal range systems, each with springs and gorges, separated by extensive dunefields. The recent painted art is very diverse, and each range system has its own character (McCaskill 1968; McDonald and Veth 2011). Classic desert art occurs east of Lake Disappointment – for example, in the Runton Range, where the paintings are dominated by concentric circles, arcs, bird tracks in short track series and simple stylised outline or solid anthropomorphs and animals (mainly lizards) with some more elaborate bichrome red-and-white paintings. To the west of Lake Disappointment, the painted art of the Calverts is more strongly figurative with less influence from Central Australian graphics. There are some classic desert designs – such as concentric circles linked by paths, lines flanked by arcs and some track meanders – but these make up only a small component of the art. The paintings are dominated by bichrome snake motifs in red-and- white, small animal silhouettes and in-fill anthropomorphs with ceremonial headdresses (rays, concentric circles or ferns), often holding objects such as boomerangs, clap sticks or sacred boards. Radiocarbon dates on organics in the pigments show that most of the surviving paintings date within the last millennium (McDonald and Veth 2008, 2011). The pigment dates correlate well with concentrations of ochre in archaeological excavations at the same sites (O’Connor, Veth and Campbell 1998;Vethetal.2001, 2008), but there are not sufficient dates yet to isolate stylistic phases within the last millennium. Most studies emphasise local differences between rock art in different range systems. Differences in the style of headdresses on human figures appear to be an ‘extremely sensitive indicator of local group signifying behaviour’ (McDonald and Veth 2011: 230), and some of these forms occur in Central Australian rock art at Uluru (Mountford 1965) and Walinya (Cave Hill) (Mountford 1976). Much of this variability appears to be local variations on iconographic themes within a widely shared mythology. Some anthropomorphic figures recur widely across the region, with only small differences: these include figures associated with travelling cults such as the Wati Kutjarra tradition. The recent art of this region, therefore, shows some of the intersite differentiation we can expect of the desert segmentary system, overlain by the overarching traditions associated with travelling cults that interlaced parts of the Western Desert in recent ethnography, such as Wati Kutjarra and Tingarri.

Late Holocene Art in the Burrup–Pilbara Area The recent prehistoric rock art of the Pilbara region is dominated by petroglyphs: pigment art is rare. On the Dampier coast, there is a change to shallow pecked or pounded designs showing a wide range of marine species and simple stylised anthropomorphs. Groups of human figures are more common than in the earlier art, and the anthropomorphs are often shown holding objects, or with headdresses, and often have exaggerated hands and feet. Field surveys Rock Art and Place: Evolution of an Inscribed Landscape 261 show that the recent petroglyphs are concentrated along the modern coast, that the highest density of petroglyphs occurs around late Holocene shell middens and that there is little differentiation of motifs at individual sites (Lorblanchet 1992; Mulvaney 2010). The focus on living sites, and the exu- berant panoply of people and marine fauna, points to an art more tied to social and economic concerns than to a totemic geography. The implications of these changes remain to be explored, but it is clear that both the social context of production and the investment in producing individual motifs changed in the late Holocene; this art appears to be more secular and less formally structured than the early Holocene engravings in this area. The late Holocene art in Burrup phases 4 and 5 also has parallels elsewhere – in the engravings on Depuch Island, on the limestone ridges at Port Hedland and in the kurangara and other stylised human figures with exaggerated hands, feet and genitals in the inland Pilbara (Wright 1968). Across this large region, the late Holocene art retained a strongly figurative character and an emphasis on anthropomorphs in contrast to the ‘track and circle’ art of the interior. But it also became much more locally diversified than the earlier art, with distinct style provinces within the Pilbara, and with different conventions for depicting anthropomorphs (Wright 1968). It is not clear whether this registers increasing divergence of graphic systems (indicating smaller areas of social interaction) or reflects differentiation of style areas within a single graphic system (as we might expect with increased badging of group identity). We urgently need a better understanding of the structure of the rock art in the inland Pilbara and its articulation with other aspects of the archaeology of this region during the last millennium.

PREHISTORY OF DESERT ROCK ART: OVERVIEW AND SPECULATION Rock art adds to the evidence for the assembly of a distinctive desert world during the Holocene (see Chapter 6). Panaramitee-style ‘track and circle’ engravings are part of a widespread desert graphic system used throughout central Australia. This becomes archaeologically visible between 8 ka and 6 ka and is ancestral to contemporary Central Australian decorative art. But Panaramitee was not the only early graphic system in Australia’s deserts – as the rich ﬁgurative rock art of the Pilbara shows. I want to ﬁnish this chapter by reviewing four issues that are key areas for further research: chronology, regionalisation, the creation of an inscribed landscape and shifts in the political economy of the desert.

Chronology The age of much desert rock art remains uncertain, and our understanding of its chronology may well change sharply as dating methods improve. At present, 262 The Archaeology of Australia’s Deserts

there is no substantial evidence to support a Pleistocene age for the surviving rock art in either tradition. Neither is associated with initial peopling of Australia’s deserts: both are more usefully seen as later developments that reﬂect the dynamics of established desert societies. In central Australia, the major period of production of Panaramitee-style engravings appears to have been 8–4 ka, coinciding with expansion and consolidation of regional populations. Coastal populations responded more quickly to postglacial climatic amelioration than did those in the interior, and well-established hunter-gatherer societies on the Pilbara coast were producing an elaborate suite of engraved designs by the early Holocene. Both graphic systems probably have deeper roots but only become visible with the proliferation of rock engravings in the early to mid-Holocene.

Regionalisation and Interaction By the early to mid-Holocene, there were two largely contemporaneous graphic systems operating in the desert. It is significant that even the earliest identified rock art is ‘regionalised’ rather than pan-continental in character. In the northwest, the striking figurative art of the Burrup is a complex and internally differentiated corpus of petroglyphs, which has structural similarities to rock art in northern Australia and may represent an extension of these systems into the Pilbara. In the interior, the more austere ‘track and circle’ art has clear links to the graphics and syntax used in sand drawings and is likely to have developed out of these. The overall continuity in central Australian art suggests the gradual evolution of a single graphic tradition. Both graphic traditions identify ‘interaction spheres’: areas where information flow and social interactions were sufficient to maintain a shared visual language. But rock art was used quite differently in the two provinces. In the northwest, the Dampier coast and inland Pilbara shared not just a common graphic system but also specific iconic designs. These societies used rock art as a part of a coordinated ideological system, with the same canonical motifs produced at widely separated sites. In this system, rock art has many of the usual roles often claimed for it: social integration, reinforcement of group identity and a territorial role emphasised by use of visually prominent motifs. This bloc fragmented into a series of discrete stylistic areas in the later Holocene. These record changes in the character of social interaction and exchange across this region. What is not known is whether these represent separate graphics systems or merely stylistic differentiation within the one graphic system. In contrast, Panaramitee-style engravings in the interior represent a place- based art. The apparent uniformity of this rock art reflects use of a shared graphic vocabulary and common stylistic conventions across an immense area. Although there are clinal differences with geographic distance, much of the Rock Art and Place: Evolution of an Inscribed Landscape 263 variability in this art is employed to differentiate sites – places rather than regions. This is a large ‘interaction area’, maintained by the open character of desert social networks. By 8–4 ka, the eastern half of the desert was a cultural bloc with a distinct identity. And this appears to have persisted throughout much of the subsequent history of the desert. A mid-Holocene age for the first Panaramitee art suggests a link with the expansion of Pama-Nyungan (PN) languages across the interior at this time. The rock engravings are clearly not a direct correlate of this linguistic bloc, because the languages but not the graphics spread into the Pilbara – and into parts of eastern Australia, where Panaramitee engravings do not occur. However, a shared graphic system may be one outcome of the spread of PN languages, as linguistic expansion created a larger social world across the desert, and related languages would have facilitated interactions between groups. But what triggered the transfer of shared graphics to rock surfaces to create an inscribed landscape?

Establishment of an Inscribed Landscape Rock art first appears in both the northwest and the desert interior as hunter- gatherer populations in these areas expanded and consolidated. The pattern of change is time-transgressive: rock art first appears in the better-watered regions of northern Australia, then on the Pilbara coast and later in central Australia. This suggests that incremental changes in the distribution and density of desert populations reached a threshold that triggered rapid structural changes in these societies. State-and-transition models – widely used in rangelands ecology (Westoby, Walker and Noy-Meir 1989; Friedel 1991) – offer a heuristic framework for modelling the process. In state-and-transition models, disturbances such as fire, drought or extreme rainfall push an ecosystem towards a threshold at which there is a rapid nonlinear transition from one stable state to another. Rock art may have been initially produced as part of the transition to a new social geography (Figure 7.16). In the northwest, the interplay of rapid changes in physical geography during the last marine transgression, the reestablishment of the summer monsoon, and population growth triggered a shift to a system in which rock art was used to reinforce group cohesion and territoriality. Rock art production was an ongoing part of these coastal societies – and there were successive transitions as the morphodynamics of an evolving coastline altered the economic base of these groups, affecting the character of the art. In the desert interior, the punctuated pattern of rock art production suggests that rock art was produced mainly during transitions (see Figure 7.16), as a way of mapping out or reinforcing a social landscape. Desert groups responded to postglacial climatic amelioration from 15 ka, but the key threshold was not reached until 8–4 ka – a time that saw production of discrete assemblages 264 The Archaeology of Australia’s Deserts

of engravings at sites across central Australia. This ‘establishment’ phase was followed by a stable ‘maintenance’ phase, where fewer petroglyphs were produced and where the emphasis was on the curation, renewal or reactivation of existing panels of rock art. In state-and-transition models, the dynamics within stable communities may also build towards a tipping point that leads to structural and functional changes in these communities. These usually involve a set of interacting components, rather than discrete boundaries in space or time, which helps to explain other aspects of the central Australian sequence. For example, there were localised pulses of production as sites in marginal areas were colonised and brought into more regular use (as shown at Glen Thirsty at 1.5 ka; see Chapter 9) (Smith and Ross 2008b). In other cases, pulses of production may have been triggered by challenges to the social geography, when the identity of sites needed to be reasserted or when new ideological elements needed to be grounded within an existing paradigm (as is likely to have been the case with the efﬂorescence of pigment art and petroglyphs during the last millennium). Panaramitee is a land-based art, but the early engravings are not overly territorial in character. Unlike the Burrup rock art, early Panaramitee engravings use a very limited range of motifs: there is little graphic complexity in designs, not all key places are marked and the art is placed on horizontal panels where it does not have a prominent role in visual signalling. This suggests that Panarami- tee rock art was an adjunct to a totemic landscape and not its central feature. One scenario is that changes at 8 ka triggered the establishment of a totemic landscape, which provided a new framework within which people could assert or negotiate rights to land. Rock engravings helped to differentiate sites in this land-based mythology but were not central to articulating property rights or ritual. Indirectly, the rock art records a new way of codifying relationships between people and land in the face of a much larger social world created by increasing regional populations and by the spread of PN languages.

Changes in Political Economy As an ideological system that connects people to place, the ‘dreaming’ has its roots in the mid-Holocene, becoming visible with the ﬁrst desert rock engravings 8–4 ka. But it was not a static system: the rock art suggests that it became more elaborate over time. The last millennium is marked by increasing differentiation of rock art at a range of scales: the context of production, the differentiation of speciﬁc sites and increasing regional identity within areas that share the same graphic system. This suggests that by 1.5–1 ka, the segmentary religious system had taken shape, with geographic totemism assuming much of its classic ethnographic form (as David also argued in his bold book, Landscapes, Rock-art and the Dreaming, 2002). This was an elaboration of an older form of totemism, represented by Panaramitee-style engravings, towards one that Rock Art and Place: Evolution of an Inscribed Landscape 265

Figure 7.16. State-and-transition model applied to Panaramitee-style rock art. In State A, internal dynamics and external forcing factors build to a tipping point, triggering a rapid transition to State B. Trigger factors include incremental structural changes in the density and distribution of hunter-gatherer groups, as well as climate-driven environmental shifts. Most rock art is produced during the transition, in an establishment phase that reflects adoption of a new way of mapping onto country. On one side of the threshold (State A), hunter-gatherer groups use graphics but not land-based graphics. On the other side of the threshold (State B), hunter-gatherer groups maintain and revitalise rock art as part of a systemic practice, codifying relationships between people and land. was more strongly corporate in character, with local descent groups each responsible for a segment of a ‘dreaming’ track or songline. ‘Art,’ says Morphy (1998: 5), ‘provides a sacred charter to the land’. By the last millennium, the graphic system reinforced a totemic geography that articulated a more or less ‘bounded’ system of land tenure, rights in land and ritual property. The closure of social networks in the desert interior is paralleled by the break-up of the Pilbara art province into more sharply defined style areas, although the timing of this is not yet known in any detail. In Central Australia, rock art acts as a counterpoise, reinforcing a system of inalienable land tenure in societies marked by high mobility, seasonal dispersal and aggregation of the population, and extreme fluidity in residential groupings. It was also a social mechanism for maintaining the ecological spacing of desert groups in a now well-peopled landscape (Peterson 1972). The geographic totemism of Australia’s deserts exemplifies Schama’s dictum that ‘landscapes are culture before they are nature; constructs of the imagination projected onto wood and water and rock’ (1996: 61). It is a distinctive system, quite different from that in southern African or South American deserts, and, in its formal properties, there is little scope for rock art as hunting magic, as sorcery or as part of shamanic practice. CHAPTER 8

THE CHAIN OF CONNECTION: TRADE AND EXCHANGE ACROSS THE INTERIOR

The ﬁrst desert song to be written in English was linked with expeditions to bring red ochre from the famous Pukardu mine (Gason 1874;Hill2002: 1–3; Jones 2007). It was sung during the making of bags for carrying the ochre and on the long journey from Pukardu, near Parachilna in the Flinders Ranges, to Cooper Creek.

Put colours in the bags, Close it all around, And make the netted bag All the colours of the rainbow.

The original Diyari was translated by Mounted Constable Samuel Gason some time between 1865 and 1871. More than a century later, a Wangkangurru man from the Simpson Desert – Jimmy Russell Wangamirri – still knew the verses (Hercus and Koch 1997). ‘They sing the Pukardu song,’ he said, ‘for setting out and returning. They sing the song of Pukardu Hill, they sing it for a long time’ (Russell, in McBryde 2000: 160). In the late nineteenth century, the Diyari on the lower reaches of Cooper Creek were at the centre of a long trade route that connected the Flinders Ranges in the south with the Mt Isa uplands 1,200 km to the north (Fig- ure 8.1). At the southern end of this exchange system was the red ochre mine at Pukardu and a complex of grindstone quarries at Tooths Nob. At the northern end, the extensive stone-axe quarries in the Mt Isa–Cloncurry meta-basalts, and the groves of the desert narcotic pituri west of the Mulligan River, provided a counterweight. Constable Gason was uniquely placed to see this system in operation. As a police trooper stationed at Lake Hope on Cooper Creek between 1865 and 1871, just as pastoralists and missionaries were moving into this district, he witnessed at ﬁrst hand the departure of the red ochre expeditions as they headed south and the return of expeditions bringing pituri from the north. Reviewing this ‘chain of connection’, DJ Mulvaney famously commented that ‘it was possible for a man who had brought pituri 266 The Chain of Connection: Trade and Exchange Across the Interior 267

Figure 8.1. The ethnographic exchange system on the eastern margin of the arid zone in relation to quarries – red ochre (open triangle), millstone slabs (black square) and stone axes (open square)–and the major pituri (Dubosia)groves(black diamond). (Simpliﬁed from McBryde 1987: 260)

from the Mulligan River and ochre from Parachilna to own a Cloncurry axe, a Boulia boomerang and wear shell pendants from Carpentaria and Kimberley’ (1976: 80). Despite its prominence in the ethnography, however, the long-distance trade system observed by Gason was not typical of exchange in arid Australia. The more usual pattern was one of individual gift exchange, either between kin or between exchange partners at ceremonial gatherings. A myriad of small individual transactions was sufﬁcient to rapidly transfer objects deep into the desert, often thousands of kilometres from their source (Mulvaney 1976). The shift from this relatively undifferentiated gift economy to a more complex mix of exchange transactions, with long-distance expeditions by groups of men (as witnessed by Gason) coupled with large-scale production of prestige goods, appears to have taken place within the last millennium. It was limited to the eastern margin of the arid zone, where the inland river systems channelled trade into a corridor, creating a critical mass of intergroup exchange that was unique in the desert. This chapter reviews what is known about the development of 268 The Archaeology of Australia’s Deserts

Figure 8.2. The southern sector of the Lake Eyre basin, showing the location of the major quarries for red ochre (open triangles) and for millstones (squares). Archaeological sites mentioned in the text are numbered: (1) Marapadi, (2)MG4,(3) White Crossing hearths, (4)JSN,(5)HawkerSwamp,(6) Arkaroo Rock, (7) Balcoracana Creek, (8) Panaramitee, and (9) Karolta engraving site. The Chain of Connection: Trade and Exchange Across the Interior 269

Figure 8.3. Major quarries and mines mentioned in the text. Red ochre mines: (1) Wilgie Mia, (2) Karrku, (3) Pukardu. Stone-knife quarries: (4)Kanti,(5) Kankiritja. Grindstones: (6) Kurutiti (Helen Springs), (7) Anna Creek (Palthirri-pirdi), (8) Tooths Nob (Wadla wadlyu), (9) the Innamincka grindstone quarries, (10) Yambacoona Hill. Stone-axe quarries: (11) the Mt Isa axe quarries. these exchange systems, their differentiation and antiquity, and the history of the major mines and quarries (Figures 8.2 and 8.3). It is often said that to understand hunter-gatherer societies, we need to broaden out from place-based or regional studies to examine how they articulated within larger spatial and social entities (Wobst 1978). Nowhere is this more true than in Australia’s deserts. One of the distinctive features of desert societies was their large geographical scale and their permeability to the movement of goods, ceremonies and social phenomena. During the recent past, social institutions such as subsections (McConvell 1985), practices such as subincision (Tindale 1974) and ceremonies such as the ‘Molonga’ (Mudlungga) (Mulvaney 1976; Hercus 1980) diffused widely across the interior. Earlier cultural spreads that linked up large areas of the desert are registered by the diffusion of stone tool types, such as geometric microliths or Panaramitee-style rock art and the 270 The Archaeology of Australia’s Deserts

spread of Pama-Nyungan (PN) languages across an already well-peopled interior (see Chapters 6 and 7). Exchanges of this kind countered the tendency for small population isolates to diverge culturally and linguistically. They also show how much the cultural history of the desert was encapsulated by developments on the desert margins, or outside the arid zone. Whereas Chapters 6 and 7 looked at the development and consolidation of regional societies, the focus of this chapter is on how these were interconnected through trade and exchange – and how this changed over time.

THE SYSTEMATICS OF TRADE AND EXCHANGE In Australia, ethnographic exchange systems were scaled in terms of organisation, quantity of goods and distance (Mulvaney 1976; Keen 2004). Regional studies are available for groups in tropical northern Australia (Stanner 1933–4; Love 1936; Thomson 1949; Berndt 1951; Warner 1958; Falkenberg 1962)and for the arid interior (Roth 1897;Howitt1904). There were named exchange cycles in many regions, such as the Warlpiri warntarri ‘when people exchange giftswitheachother...Allsortsofthings.Theones that people give away to each other in turn’ (Laughren 1994). At one end of the spectrum was gift exchange between individuals. This was usually intragroup and involved kin or named exchange partners. These interpersonal exchanges were underlain by networks of obligation that formed the infrastructure of social relations in hunter-gatherer societies. Reciprocity was fundamental. Desert Pintupi expected exchange relations to be reciprocal or ngaparrku (level) (Keen 2004: 358). Ceremonial gatherings often provided the occasion for conducting a range of individual transactions on the side or, as Stanner put it, ‘in the shadow of more impressive events’ (1933–4: 160). The net effect of numerous small transactions through multiple connecting links was the wide diffusion of durable goods, ceremonies and new social forms across the interior of the continent. At the next level were formal intergroup exchanges between neighbouring language groups. These sometimes involved specific trade ceremonies, as well as preparation to receive visitors and stockpiling of food and exchange goods – as Berndt (1951) describes for Arnhem Land, and Roth (1897) and Aiston (1937) describe for the Lake Eyre basin. Mudburra and Jingili people on the northern margin of the desert initiated such exchanges because ‘they felt that a formal exchange of goods would communicate that the local community was able to assert its place within the wider area’ (Paton 1994: 180). In this context, exchange was as much about political economy as sociality. Thomson’s classic study (1949) of the ceremonial exchange cycle in Arnhem Land also shows how sensitive this was to external factors. In Arnhem Land, the turnover of goods accelerated as people sought to obtain exotic items brought by Macassan traders to the north coast. They did this by manipulating The Chain of Connection: Trade and Exchange Across the Interior 271 the exchange cycle, using the production of goods for exchange to trigger obligations for reciprocity. The acquisition of goods was fundamental to the exchange gatherings described by Berndt (1951) and Roth (1897), and these are located firmly towards the commodity end of the gift–commodity dichotomy. Outside of residential groups, food was rarely exchanged. Most of the goods changing hands derived their value from their properties as exotic or prestige goods – although this value was refracted through the ritual and totemic associations inherent in the object or its place of origin. In this context, local sources of red ochre or grindstones were sometimes bypassed in favour of material from more distant sources. In contrast to these ‘down-the-line’ exchange systems, the Lake Eyre basin saw the direct ‘long transfer’ of goods over 400–500 km via organised expeditions mounted specifically to acquire red ochre, pituri, grinding slabs and stone axe heads. These involved exchange on a corporate rather than an individual level, and the direct transfer of goods over long distances rather than a chain of small transactions. They were also distinctive in their large scale. For example, each red ochre expedition occupied fifty to eighty men on a two- month 800 km round trip that brought in about 2 t of ochre per expedition. In many respects, these ‘long-transfer’ systems represent an elaboration of the intergroup trade ceremonies described for northern Australia. Trade goods obtained through ‘long transfer’ also rapidly entered local reciprocal exchange networks, creating a secondary pattern of diffusion of goods.

ARCHAEOLOGICAL APPROACHES Archaeological research on trade and exchange has followed two complementary lines of approach: distribution studies looking at the spatial spread of exchange items, often using petrology or geochemistry to identify material from speciﬁc sources; and studies of production focusing on the major quarries. DS Davidson and FD McCarthy provide important early studies of distribution patterns (Davidson 1938, 1952; Davidson and McCarthy 1957; McCarthy 1938–40). The key to much modern research, however, is the 1973 review by DJ Mulvaney (1976), who argued that anthropological studies of trade and exchange needed to be more strongly grounded in the material evidence. McBryde’s work on the distribution of ground-edge axes in southeastern Aus- tralia (Binns and McBryde 1972; McBryde and Watchman 1976; McBryde 1978) provided the model for a range of other archaeological studies: on grindstones (McBryde 1987, 1997), on pearl shell and baler shell objects (Akerman and Stanton 1994) and on red ochre (Sagona 1994;Smith,Fankhauserand Jercher 1998). McBryde later extended her study of stone axes to include petrological examination of museum specimens from the Lake Eyre basin, but results are not yet available. She also initiated a similar project on grindstones 272 The Archaeology of Australia’s Deserts

in the Lake Eyre region, but this did not proceed beyond a pilot study. Geo- chemical sourcing of red ochres has been the subject of several studies (David et al. 1993; David, Watchman, Goodall et al. 1995; Ford, MacLeod and Haydock 1994; Goodall, David and Bartley 1996; Jercher et al. 1998;Smithand Fankhauser 2009), but its application to museum and archaeological collections is still in the early stages. Much of what is known about the distribution of axes, grindstones and red ochre still relies heavily on the ethnography. And these ethnographic distribution patterns are yet to be tested against archaeological data (except for rather limited analyses by Tibbett 2002 and Smith and Fankhauser 2009). Even where there is fine-grained distribution data, there may be problems distinguishing (a) passage of a gift along a chain of exchanges from (b) direct long-distance transport or from (c) direct acquisition of ochre or grindstones during the normal cycle of seasonal movements (‘embedded procurement’). Multiple individual exchanges can create the appearance of a long-distance exchange system, without requiring a network of fixed, formalised point-to- point trade routes – as Lawrence (1994) cautioned for customary exchange in the Torres Strait. The usual means of discriminating between these processes is to assume that well-developed distance-decay patterns result from ‘down-the- line’ exchange. In contrast, direct long-distance transport may be implicated in cases in which the comparative lack of wear (or reduction) on transported axes or grinding slabs indicates that they have been transferred without an extended time in circulation. One question rarely posed is whether these exchange systems only become archaeologically visible when particular durable goods (grindstones, axes, red ochre, shell pendants) are added to existing networks. A second line of research has focused on the production centres. Davidson et al. (2005) and Hiscock (2005) provide the best accounts of the Mt Isa axe quarries, although much of their research on this large complex is yet to be published. Of the grindstone quarries, only three have been described in detail: ToothsNob (Reap Hook Hill) (McBryde 1997), Kurutiti (Mulvaney 1997)and Narcoonowie (Smith, McBryde and Ross 2010). Information on the major red ochre quarries in the arid zone is compiled in Smith and Fankhauser (2009), and studies of individual ochre quarries are available for Wilgie Mia (Wood- ward 1914), Karrku (Peterson and Lampert 1985) and Pukardu (‘Bookartoo’) (Jones 2007). Silcrete quarries associated with the production of large blades (150 mm long) for hafted stone knives are known throughout Central Aus- tralia. The only detailed studies of these are by Murgatroyd (1991, Kankiritja), Paton (1994, Kankiritja), Graham and Thorley (1996, Kanti) and Binford and O’Connell (1984) – although Doelman (2008) provides details of general flake and blade production at other silcrete quarries. Quantitative estimates of quarry production rates are rare. Determining the chronology of the major quarries has also been difficult: excavation of quarry pits and mine-floor deposits has The Chain of Connection: Trade and Exchange Across the Interior 273 rarely recovered dateable material. Where fragments of grindstones or stone axes have been excavated from occupation sites, it is often uncertain whether these are made on quarried blanks or on stone derived from natural outcrops and exposures of the same material. Australian archaeologists have generally stressed the intangible value of exchange goods: their ritual properties, cosmological or totemic associations or their inherent symbolic values (McBryde 2000; Jones 2007; Brumm 2010). However, this does not account for why certain places and not others became foci of exchange and production, given that the entire landscape was interlaced with powerful mythological associations. Others have argued that the social networks and reciprocity created by exchange are crucial for buffering ecological risk in marginal environments. This may have been true for intragroup and regional exchange, but there is no evidence that ‘long-transfer’ systems functioned in this way. In any case, ‘ecological reciprocity’ was probably more dependent on the widely ramified kin networks created by exchange of women as wives than on reciprocal gift exchange (Yengoyan 1968; Peterson 1986: 50). The converse argument is that long-distance exchange systems were dependent for their operation on extended chains of social obligation, and that these in turn required a prerequisite degree of social solidarity and cultural integration at interregional level. One extension of the ‘risk and reciprocity’ argument is the claim that more effective social exchange networks – and the extended systems of reciprocity they generated – underpinned late Holocene settlement in the sandy deserts (Lourandos 1983: 91, 1985;Veth1989b, 1995b, 2006: table 1). The proposition originally put forward by Lourandos (1983, 1985) was that increasing intergroup and interpersonal competition in these hunter-gatherer societies led to changes in ceremonial exchange, providing the crucial incentive for economic expansion and resource intensification, and leading to population growth. This created a spiral of change towards greater social complexity and the abrupt late Holocene ‘appearance of intensive and possibly ceremonially based occupation of marginal zones, for example wetlands, rainforests, highlands...andaridzones’ (Lourandos 1983: 91). Archaeological fieldwork in the desert since the 1980s has weakened much of the basis for this argument but leaves us with important questions about the role of expanded social networks (‘alliances’) in the growth of desert societies and about whether an escalating demand for goods for ceremonial prestation may have driven an expansion of trade and exchange and the restructuring of social relations between groups. Other studies show that the turnover of these exchange systems was sensitive to external supply of exotic goods (demand-driven) and to opportunities for interpersonal contact (density-dependent) – both likely to increase with demographic growth in the desert (Thomson 1949; Akerman and Stanton 1994). 274 The Archaeology of Australia’s Deserts

table 8.1. Radiocarbon dates for baler shell (Melo sp.) in the Great Sandy Desert

Laboratory Radiocarbon Age with oceanic Years Site code age BP reservoir correction cal BP

Kiriwirri Wk5638 500 ± 70 50 104 Kiriwirri Wk5639 660 ± 150 210 233 Kurtararra Well Wk5640 2,710 ± 280 2,260 2,300 Kurtararra Well Wk5641 1,220 ± 90 770 734 Yurlpul Wk5642 2,270 ± 60 1,820 1,747

The oceanic reservoir correction is 450 ± 35 years for northwestern Australia. Source: Smith and Veth (2004)

The following sections present a series of case studies that look at the history of some major exchange commodities: marine shells, red ochre, grinding slabs, stone axes, pituri and stone knives.

EXCHANGE OF COASTAL BALER AND PEARL SHELL INTO THE INTERIOR Marine shells were traded deep into the interior of the continent and provide the best archaeological evidence for ‘down-the-line’ exchange systems. The most prominent examples are baler shell (Melo amphora) and pearl shell (Pinctada maximus or P. albina albina). Melo amphora is a large volute widely used as a container and for a range of objects with other functions, including scrapers, adzes and spoons (Schall 1985). Ethnographic accounts show that much of the shell exchanged into the interior of the continent was used for pendants: these brilliant, white, polished oval pendants, about 100 mm long and pierced for suspension, were important in rain-making ceremonies (Akerman 1973a). Although M. amphora is widely found in coastal waters around northern Australia, most ethnographic baler-shell artefacts in the interior came from either the northwest coast of the continent or the Gulf of Carpentaria (Akerman 1973a; Mulvaney 1976). In the Western Desert, baler shell was transferred through ‘down-the-line’ exchange between individuals across 1,700 km of desert, reaching western Central Australia and the southern margin of the Great Victoria Desert. Archaeology provides some evidence of the antiquity of these exchange networks. Baler shell is common in archaeological assemblages from Pleistocene sites along the arid west and Kimberley coastlines (see Chapter 4) but appears to be limited to within 100 km of the coast. This pattern changed dramatically during the last few millennia. Radiocarbon dates on small pieces of baler shell at campsites near Aboriginal wells in the Great Sandy Desert (Table 8.1)show that it was being exchanged into the arid interior by 2.3 ka (Smith and Veth 2004). The Chain of Connection: Trade and Exchange Across the Interior 275

Although the exchange of pearl shells mirrors many aspects of the exchange of baler shell, it differs in being largely a postcontact phenomenon. Pearl shells – either Pinctada maxima or the smaller P.albina albina – were made into large ovoid pendants up to 20 cm long for breastplates or pubic ornaments (Akerman and Stanton 1994). For desert people, the nacreous iridescent shell shimmered with the brilliance widely associated with ancestral power in Aboriginal Australia (Morphy 1989); more specifically, it was symbolically linked to water, rain and the rainbow serpent. Like baler shell, pearl shell originated on the Pilbara and Kimberley coasts and along the shores of Cape York Peninsula. There is little evidence for its trade into the arid interior until the pearling industry provided a ready Figure 8.4. Pearl-shell pendant with source of shells from the 1870s onwards. By the interlocking key design, from North- final decades of the century, plain pearl-shell pen- westernAustralia.Noscalegiven dants were widely distributed across the continent – but probably 12–15 cm long. (After although Stirling (1896: 109) records that in Central Davidson 1949a:fig.1d) Australia in 1894, baler shell was still more common than pearl shell. Akerman and Stanton (1994) show that there was a rapid increase in the production and exchange of pearl-shell pendants around 1900 and a shift towards decorated pendants with engraved designs vividly inlaid with red ochre and fat or with powdered charcoal. And, from the 1920s, these designs were dominated by a distinctive interlocking key motif that originated on the northwest coast (Davidson 1949a)(Figure 8.4). The diffusion of pearl shell illustrates some important properties of exchange systems in the desert. First, there is the speed and efficiency of transfer. The continental distribution of pearl-shell objects is the result of less than a century of exchange. Akerman and Stanton (1994) examined 800 engraved pearl-shell pendants and showed that some were moved over 4,000 km in just 20 years. An ethic of hand-to-hand exchange historically made the desert an open system for the dispersal of prestige objects and practices. Second, the rate and intensity of exchange is very sensitive to changes in external factors. The efflorescence in the exchange of pearl shell was fuelled by greater access to the shell through the pearling industry; an acceleration of social interaction and ceremonial activity triggered by the concentration of the Aboriginal population in missions, ration depots and on cattle stations; and by greater opportunities to travel using motor vehicles, railways or camels. These provided ‘oxygen’ for the ‘ritual engines’ of desert life, removing some of the constraints on precontact ceremonial exchange systems. 276 The Archaeology of Australia’s Deserts

THE RED OCHRE MINES: PUKARDU, WILGIE MIA AND KARRKU Fragments of red ochre recovered from dated occupation deposits show that exploitation of some ochre sources began in the late Pleistocene, but evidence from the mines and quarries also suggests a more recent escalation of production. Ochres are natural earthy red or yellow pigments, containing iron oxides or hydrated iron oxides, sometimes as crystalline haematite. Aboriginal people used red ochre for a variety of purposes: as body paint during ceremonies, as a covering to cool the skin, as a preservative on balls of dried fruit, as decoration on wooden implements and as pigment in rock paintings. Ochres were ground into a dry powder and rubbed onto the body either directly or after rubbing the body with grease. Both men and women used red ochre as a ritual cleanser during ceremonies. Mirroring European use of the term ‘haematite’, many red ochre deposits are equated mythologically with the blood of totemic beings. As Jones puts it, red ochre was ‘a medium or agent of transcendence, from sickness to health, death to renewal, ritual uncleanness to cleanness, the secular to the sacred, the present reality to the Dreaming’ (2007: 349). The most desirable ochres were those with a deep red colour, good covering qualities and a distinctive metallic lustre (produced by small, aligned ﬂakes of crystalline haematite). The diagenesis of ochre is complex: earthy haematite, limonite or goethite often originate in ancient marine sedimentary formations, and may occur as pockets of reworked and deeply weathered sediment in joint- lines or cavities in bedrock; ochre is also found in Tertiary-age laterites, whereas crystalline haematite mostly derives from gossans and hydrothermal lodes or veins (Smith and Fankhauser 2009). Across Australia’s deserts, ochre was widely obtained from outcrops of laterite or ferruginised sandstone. In some locations, small pits or tunnels were dug into earthy iron-rich sediments to extract the ochre. The best-known ochre mines – Pukardu, Wilgie Mia and Karrku (see Figure 8.3) – are discrete deposits where Aboriginal people extensively used open-cut pits or tunnels to follow an ochre seam. The high-grade red ochre from these sources was exchanged over hundreds of kilometres, sometimes involving organised expeditions. Pukardu (also known as ‘Bookartoo’) provides the best-documented ethnographic example of these red ochre expeditions (reviewed in detail by Jones 2007). The name comes from ‘the term for the heart’s blood of the ﬁerce dog-like animals, which was spilt at the site’ ( Jones 2007: 352). Gason (1874) witnessed red ochre expeditions leaving Cooper Creek on the 400-km journey to Pukardu in the Flinders Ranges. These annual expeditions usually took place in winter. Gason noted that

the party travels about twenty miles a day, and on arrival at the mine each member of it digs out his own ochre, mixes it with water, making it into The Chain of Connection: Trade and Exchange Across the Interior 277

loaves of about 20 lbs. weight, which are dried. Each man carries an average weight of 70 lbs. of ochre, invariably on the head, and has to procure his own food; the party seldom resting a day while on the journey, which lasts usually from six to eight weeks (1879: 281).

In most cases, these appear to have been parties of Diyari men, but some accounts mention men from ‘the source of the Diamantina and Herbert Rivers’ (1,200 km to the north) (Masey 1882). Howitt notes that the Yandruwantha also mounted annual expeditions to Pukardu (1904: 713), and there are first- hand accounts of Wangkangurru men taking part in some expeditions (Hercus and Koch 1997; Jones 2007). By 1860, the Pukardu mine lay well within the bounds of European settlement. Because large parties of armed men entering the pastoral districts posed a problem for colonial authorities, police and sheep station managers, there are a number of independent accounts of red ochre expeditions between 1864 and 1890. These corroborate Gason’s observations of the scale of these expeditions: reports of group size range from fifty to eighty men, with estimates of the average load of ochre ranging from 23 to 45 kg per man (Masey 1882; Howitt 1891, 1904: 712; Jones 2007). In one case, a party of fifty-six men was intercepted carrying between them ‘30 cwt’ (1.5 t) of red ochre (The South Australian Register, 26 July 1864). These accounts also make it clear that senior men of the local group regulated access to the ochre, and at least one party was ambushed and killed at the mine for breaching this protocol (Jones 2007). Despite its fame, Pukardu is a relatively small ochre mine. It is located on a spur high in the northern Flinders Ranges. The deposit is an earthy haematite, in-filling joints and cavities in dolomite (Keeling 1984). The remains of six open-cut pits (2–6 m in diameter) record the last major phase of quarrying at this site. Today, these are choked with silt and boulders and only 0.5 m deep. Local subsidence shows there are subterranean cavities nearby, suggesting that the pits were once deep enough to allow adits from their sides. Surrounding the pits, a raised rim of spoil skirts the spur – enclosing about 240 m2 of worked ground (all of the available area of the spur). Assuming the pits were originally 2 m deep, the visible workings account for about 40 m3 of excavated ochre. If we extrapolate to the entire area of the spur (and reconstruct the line of the ridge, allowing for 1.5 m of ochre quarried away, and assume that 25 per cent of this was dolomite spoil), then a maximum of 600 t of ochre may have been quarried at Pukardu. On these figures, Pukardu could not have sustained the scale of extraction recorded ethnographically (2.1 t per year) for much more than 300 years. Pukardu ochre is chemically and physically distinctive, ranging from friable dark pink ochre to a heavy greyish-red haematite (Jercher et al. 1998;Smith and Fankhauser 2009). Pieces of this ochre, identified by their geochemistry, have been found on surface sites along the lower Cooper (SAM A66497 and 278 The Archaeology of Australia’s Deserts

Figure 8.5. The Wilgie Mia red ochre mine, photographed in about 1910 by WH Kretchmar. This shows the base of the initial open cut. A large subterranean gallery, running off to the RHS, slopes down to a depth of 30 m below surface. (Photo- graph courtesy of Department of Anthropology and Archaeology, Western Australian Museum)

A66498, Smith and Fankhauser 2009), but so far it has not been found in dated archaeological contexts. The nearest excavated site, Arkaroo Rock, has a sequence spanning the last 7,000 years, with a large assemblage of ochre pieces dated to within the last 5.5 ka (Draper 1989). However, none of these ochres has the geochemistry or petrology characteristic of Pukardu ochre (Smith and Fankhauser 2009). The present picture, therefore, is of a mine mainly used within the last few hundred years. The scale of production at Pukardu is dwarfed by that at Wilgie Mia, the largest of the ethnographic ochre mines (Figures 8.5 and 8.6) (Woodward 1914; Kretchmar 1936;Davidson1952; Baynes 1984). Here, the workings form a large open-cut quarry, which leads into a massive subterranean gallery excavated under a cap-rock of ironstone on the crest of a ridge in the Weld Range. Davidson provides a description of the quarry in 1939: The Chain of Connection: Trade and Exchange Across the Interior 279

Figure 8.6. The structure of Wilgie Mia: (a) stratigraphic section from Crawford’s 1963 excavation (from Crawford, 1963, unpublished ﬁeldnotes; and Baynes 1984); (b) Woodward’s cross-section of the mine, showing the approximate location of Crawford’s excavation (Woodward 1914:ﬁg.37).

From the summit on the north side a great open cut varying between fifty and one hundred feet in width and possibly sixty-five feet in depth has been laboriously excavated. On the sides around the bottom are deeper chambers, while underneath them numerous tunnels follow the seams of red and yellow ochre, often for several yards . . . The excavation is so stupendous that it is difficult to attribute such an accomplishment to a people with a simple technology. (1952: 82) From the main chamber, a series of smaller galleries and crawl spaces – up to 60 m long – follow steeply dipping pockets of earthy haematite into the hill. Wooden scaffolds were used to provide access to seams of ochre, and hammer-stones and fire-hardened wooden digging sticks were used to wedge and pry out the blocks of ochre. Woodward (1914) provided the most detailed report on the mine. He estimated that about 15,000 cubic yards (11,468 m3) of rock and ochre had been quarried at Wilgie Mia, equivalent to about 24,000 t. Because the earthy ochre and haematite is in thin bands, interbedded with shale and jaspilite, much of this may have been mining waste. However, even a conservative estimate of yield (assuming 75 per cent mining waste) indicates that some 6,000 t of ochre may have been extracted. Commercial exploitation of the deposit began in 1945 and continued until 1978, extracting a further 9,131 t of ochre (Elias 1982: 20). In 1963,Crawford 280 The Archaeology of Australia’s Deserts

carried out an archaeological excavation in part of the main chamber left undisturbed by commercial mining (1963) (see Figure 8.6). This revealed more than 6 m of floor deposits – made up of fine layers of ash, quartzite rubble and haematite dust, with a concentration of mining implements in the upper 4 m. Other finds included a tula adze, a geometric microlith and a bone point. Crawford obtained two 14C dates – on wood in layer 12 (560 ± 70 BP Gak 1769)andinlayer23 (at 6 mdepth)(1,100 ± 90 BP Gak 1770)–showing that use of this mine spanned the last millennium. These dates also imply that significant excavation of ochre had already occurred by the time the floor deposits in this part of the mine began to accumulate 1,000 years ago. Based on Aboriginal informants in 1939,Davidson(1952) concluded that Wilgie Mia quarry provided most of the red ochre used in the southwestern sector of the arid zone, up to 450–600 km from the mine. For such a large mine, however, it is surprising that there is so little direct archaeological or ethnographic evidence for the trade and exchange of this ochre. Wilgie Mia ochre was used in rock paintings at nearby Walga Rock (Clarke 1976), but it has not been found in dated excavated contexts at this site (Bordes et al. 1983) or ‘down-the-line’ in Western Desert sites such as Kaalpi, Serpents Glen or Puntutjarpa (Smith and Fankhauser 2009). Karrku is a regional red ochre mine in the Campbell Range in the western part of Central Australia. Like Wilgie Mia, ochre mining at Karrku has formed a substantial subterranean chamber beneath the cap-rock of a flat-topped hill, with tunnels tracing a seam of ochre into the hillside for about 35 m. Peterson and Lampert (1985:2)estimatethat240 m3 of ochre has been dug up at Karrku, equivalent to 500 t (on my calculation that the density of Karrku ochre is 2.1g/cm3, equivalent to 2.1 t/m3). Today, Warlpiri people still use the mine. Before about 1940, Karrku ochre was traded locally to the north and south of the Campbell Range, reaching the Ehrenberg Ranges, 65 km south of Karrku. As with other ochre mines, access to Karrku was controlled by a group of closely related kin – although, in this case, this also involved a competitive assertion of rights by senior men belonging to different clan estates (Peterson and Lampert 1985). The remains of wood brought into the mine as firesticks are present among mining debris at the base of the modern work face. Radiocarbon dating of torches from the surface of the south chamber gives ages ranging from modern to 230 years (211 ± 37 BP Wk-21266; 223 ± 35 BP Wk-21267; 183 ± 34 Wk-21268; modern Wk-21269 –MASmith, unpublished field data). Earlier evidence of use is provided by the identification of Karrku ochre in the long archaeological sequence from Puritjarra rockshelter, 125 km to the south (Smith et al. 1998). Karrku ochre is a distinctive, dark-red micaceous haematite with a high proportion of lanthanide elements, characteristic of deep weathering (Smith and Fankhauser 2009). The archaeological ochres at Puritjarra superficially resembled Karrku material but were subject to a nested The Chain of Connection: Trade and Exchange Across the Interior 281 series of analyses to test this: looking at major and minor oxide composition, petrology, mineralogy, quartz provenance identity using stable isotope ratios, and trace-element analysis. The results show that people visiting the rockshelter around 35,000 years ago brought with them red ochre from Karrku, indicating that exploitation of this deposit had begun by this time. The slightly higher proportion of kaolinite to muscovite in the archaeological specimens indicates that these samples have undergone greater weathering than contemporary samples from the mine. This suggests that the late Pleistocene ochres were obtained from parts of the lode more exposed to surficial weathering, perhaps indicating an early stage in the development of the mine (viz. a time before the subterranean chambers were established). Whatever the case, archaeological evidence suggests that Karrku lay within the catchment of people using Puritjarra during the late Pleistocene, and that the ochre was obtained directly (‘embedded procurement’) rather than by long-distance exchange. At Puritjara, Karrku ochre is the most common type found in late Pleistocene levels, but its use declined after 15 ka, coinciding with other evidence for shifts in site territory. During the last millennium, it once again became the dominant ochre at Puritjarra. Smith et al. (1998) suggested that this reflected new exchange links with the Karrku area, but increased production at the mine might also have been a contributing factor. Age estimates are available for two other desert ochre quarries. The first, Ulpunyali, is a regional quarry in western Central Australia, where local Abo- riginal people dug a series of open pits to extract coarse-grained, greasy, pur- ple or dark-red ochre. Red ochre from Ulpunyali first appears in significant amounts in the Puritjarra sequence from 15 ka (Smith et al. 1998). Because this site is only 50 km from Puritjarra, the ochre is likely to have been obtained directly by the people using the rockshelter. The second example is the Anvil Creek ochre quarry (Ridges 2003: 63). Here, a shallow subsurface ochre-rich layer with charcoal is dated to 250 ± 70 BP (Beta 40523). Lichen on the surface of the ochre outcrop is dated 140 ± 100 BP (NZA 5730), suggesting that use of this quarry ceased before 1800 AD. Red ochre from this source (or a related source) was found in a heat treatment pit (hearth 5) excavated at Cuckadoo 1 rockshelter, dating to 4.8 ka (4,270 ± 70 BP Beta 38562) (Davidson, Sutton and Gale 1993), showing that roasting was used to improve the adherence of the ochre on rock surfaces. These ochre quarries have contrasting histories. Regional quarries with limited local distribution clearly have some antiquity, as shown in the case of Karrku, Ulpunyali and Anvil Creek. Pukardu – a ‘long-transfer’ quarry with large-scale production of ochre – seems to have operated for only a few hundred years. On present data, we cannot say where Wilgie Mia fits into this scheme. The scale of ochre mining at this site suggests that it must have fed into a ‘long-transfer’ network involving red ochre expeditions organised on a similar scale to those along the inland rivers on the eastern side of the desert. 282 The Archaeology of Australia’s Deserts

However, there is no direct ethnographic or archaeological evidence for this. If ochre was extracted at the rates recorded ethnographically for Pukardu (about 2.1 t per year), Wilgie Mia may have had an operating life of 3,000 years. On the other hand, Crawford’s radiocarbon dates indicate production rates within the last millennium at Wilgie Mia that were well in excess of those at Pukardu, suggesting the ethnographic workings need not have an antiquity much beyond 2 ka.

GRINDSTONE QUARRIES AND THE TRADE IN MILLSTONES The importance of grass and acacia seeds as grain led to substantial demand by desert groups for replacement grinding slabs. Tindale notes that in the Western Desert, ‘there is a woman’s song that laments the inertia of her husband who would not make the journey to fetch new tjungguri stones [topstones] for her. Her hand stone is worn so thin the tips of her fingers are bleeding as she makes his bread’ (1974: 100). In Central Australia and parts of the Western Desert, these replacement slabs were locally obtained as natural tabular blanks in the sandstone and quartzite ranges dotted across the desert. In sandy regions of the Western Desert, however, they were a scarce resource. ‘Men travelled to the known sources of stone, utilizing kinship ties with people in these areas,’ notes Hamilton. ‘The further diffusion of the stones depended on men replacing the grindstones of their wives, then appropriating the older ones and passing them on in trade to men of the west’ (1980: 8). In other areas, this demand was met by specialised grindstone quarries, which supplied sandstone slabs for local needs or fed finely finished millstones into long-distance exchange networks (McBryde 1987). These grindstone quarries are typically positioned on the edge of large tracts of sand plain, dunefield or stony ‘gibber’ desert – areas where suitable stone is scarce (see Figure 8.3). The best-known examples are the quarries at Helen Springs (Kurutiti), on the edge of the Barkly Tableland (Mulvaney 1997); Anna Creek (Palthirri- pirdi), near Lake Eyre (Hercus 2005); Tooths Nob (Wadla wadlyu), north of Reaphook Hill in the Flinders Ranges (McBryde 1997); Narcoonowie, in the Strzelecki dunefield (Smith et al. 2010); and the quarry complexes north and south of Cooper Creek at Innamincka (including Wild Dog Hill and Mount McLeod) (Hiscock and Mitchell 1993;McBryde1987). Access to the major quarries and local knowledge about the production of grinding slabs appears to have been carefully regulated. It was usually in the hands of senior men, although ethnographic accounts are sketchy. For instance, at the Walaya quarry in the Toko Ranges, Roth noted ‘the exact locality and process of removal from out of the natural rock is said to be a secret among the old men: that the slabs are in some way detached by means of fire and then split with bone pegs’ (1904: 25). The Chain of Connection: Trade and Exchange Across the Interior 283 table 8.2. Production estimates for arid-zone grindstone quarries

Quarry pits Excavated Estimated number Quarry (number) volumeb m3 of millstonesc Source

Regional Yambacoona Hill 370 3,700 (2,775) 68,300d Field et al. 2003 Kurutiti 200 2,000 (1,500) 87,200e Mulvaney 1997 Narcoonowie 39 196 (150) 2,050f Smith et al. 2010 Long-transfer Anna Creek – 1,032 (774)h 71,000g Worsnop 1897 Innamincka – – – – Tooths Nob 1,100a 19,800 (14,850)i 731,500d McBryde 1997 a McBryde 1997,figs.6 and 7,shows370 quarry pits covering 30 per cent of the quarry complex. By extrapolation, there are approximately 1,100 pits in total. b Figures in brackets assume 25 per cent wastage in the form of quarry debris. These figures are used for the net production estimates. Excavated volume is calculated assuming that an average quarry pit is 6 m in diam and 1 m deep = 10 m3. c Net millstone production assumes rough-outs are 50 per cent greater volume than finished millstones and uses the following estimates of millstone dimensions and volume: d Here I use figures for Innamincka millstones, 52 × 41 × 9.5 cm = 0.0203 m3. e Kurutiti, 55× 26 × 6 cm = 0.0086 m3. f Narcoonowie, 65 × 40 × 25 cm = 0.073 m3. This is an overall figure for rough-outs and is relatively high because the sandstone is poor and without good tabular cleavage. g Anna Creek, 46 × 30 × 8 cm = 0.0110 m3. h Worsnop’s estimate includes only a proportion of this extensive quarry. Actual production rates are likely to be three to four times greater, perhaps 200,000–280,000 millstones. i Quarry pits at Tooths Nob and Innamincka are larger on average than 10 m3. The figure used for these quarries is 18 m3 per pit – a figure midway between 10 m3 and 26 m3 (the larger pits are assumed to average 10 m in diameter and to be 1 m deep = 26 m3).

A distinction between regional and ‘long-transfer’ quarries is important here. The great quarry complexes at Anna Creek, Tooths Nob and Innam- incka provided sandstone slabs that were traded hundreds of kilometres north into the immense Simpson dunefield and along the arid inland rivers on its eastern margin. Supplementing the major grindstone quarries was a second tier of quarries based on smaller outcrops, often exploited less intensively, and serving the needs of communities within their local area or region. McBryde’s unpublished research shows that this difference is documented in the distribution of surface finds and museum collections of grindstones across the Lake Eyre region. Finally, as Table 8.2 shows, there is a significant scaling up of production from regional quarries to those identified as ‘long-transfer’ quarries.

The Lake Eyre Basin Quarries The most striking concentration of grindstone quarries in the arid zone lies in the southern part of the Lake Eyre basin (see Figure 8.2), where there are at 284 The Archaeology of Australia’s Deserts

least five known quarries in addition to the major complexes at Anna Creek, Tooths Nob and Innamincka. West of Lake Eyre, the Anna Creek quarry is known to Arabana- Wankangurru people as Palthirri-pirdi (lit. millstone [palthirri]quarry[pirdi]). Mythologically, this long narrow outcrop of fine grey-white sandstone, 200 m long, is said to represent a fish bone. Field surveys and oral accounts point to a wide distribution of Anna Creek grindstones across the southern part of the Simpson Desert, especially in the Kallakoopah area. The full extent of trade in this stone is yet to be established, but the quarrying activity at Anna Creek was clearly intensive: numerous quarry pits crater the outcrop. Worsnop calculated that 1,354 t of stone (1,032 m3) had been excavated from the main part of the quarry, which he estimated represents the production of about 71,000 millstones (he allowed for 25 per cent wastage and used field observations that showed the largest millstones measuring 46 × 30 × 8 cm) (1897: 98). Because he dealt with only part of the quarry, and because the quarry pits extend along the entire length of the outcrop, the actual production rate of millstones may have been three to four times this figure, perhaps around 200,000 millstones. The Wadla wadlyu quarry complex at Tooths Nob, on the eastern slopes of the Flinders Range, represents an even larger scale of production. Quarry pits extend for 1 km along sandstone ridges, with more than 370 quarry pits mapped (McBryde 1997; this represents a net production of between 500,000 and 1.5 million millstones. The quarrying was standardised both in product and process: tabular sandstone was split using stone wedges and, like most grindstone quarries, was excavated in oval pits up 10–20 m in diameter and up to 2 m deep. Of all the quarry complexes, ToothsNob is the site most explicitly linked to long-distance trade along the inland rivers. Millstones from Boulia, 1,000 km to the north, have been petrologically sourced to this quarry. Howitt links acquisition of the grindstones to ethnographic red ochre expeditions to the Pukardu ochre quarry, noting ‘the flagstones used for grinding seeds were obtained not far from the red ochre mine. Each man carried back either a slab of stone or lump of red ochre on his head’ (1904: 713); with ‘some kind of permit from the intervening tribes’ (1878: II, 305). At Innamincka ( Jidniminka), Cooper Creek debouches from the bedrock- controlled section of its channel and spreads out into a maze of floodplains among the dunefields to the west. At this point, the sandstone hills have been extensively quarried for grinding slabs. Reuther reports ‘these millstones are brought from Jidniminka for purposes of trade. The Jandrowonta and Jauru- worka people are the owners of this stone pit which is of great importance to them for the bartering trade’ (1981 [1914]: II, 238, 1081). In fact, there are a number of discrete grindstone quarries at Innamincka, extending along hillslopes and escarpments for 10 km on both sides of Cooper Creek. The distinctive, gritty brown sandstone was extracted in quarry pits, 2–20 min diameter and up to 2 m deep, with rock slabs piled around the rim, forming a maze of intersecting craters on some hills. Grinding slabs appear to have The Chain of Connection: Trade and Exchange Across the Interior 285

Figure 8.7. Survey plan of Narcoonowie grindstone quarry, showing the maze of circular quarry pits. Contour interval is 0.2 m and shows relative fall only. Individual pits are numbered: bold numbers indicate active (recent) pits; exhausted (older) pits are numbered in normal type. (Source:Smithetal.2010) been roughed out on the spot: flaked to shape, bifacially thinned and then hammer-dressed to produce ovoid millstones with a biconvex cross-section. Production appears to have been on a large scale, with a standardised approach to the reduction and dressing of grindstone slabs. However, the scale of these quarries presents difficulties for study and, at present, there is no detailed survey plan of the Innamincka quarries or any formal analysis of the production and distribution of the millstones. Narcoonowie provides a contrast to the large quarries. This is a small satellite quarry in the Strzelecki dunefield, 85 kilometres south of the main Innam- incka quarry complex (Smith et al. 2010). The Narcoonowie quarry is made up of a series of shallow circular pits (Figure 8.7) – covering a total area of 286 The Archaeology of Australia’s Deserts

4,000 m2– excavated into a low subhorizontal bed of low-grade sandstone exposed in a dune swale. The superimposition of pits shows that quarrying initially exploited a low bench of sandstone exposed in the swale and progressively extended to the northwest over the life of the site. Quarrying initially produced irregular, subangular to tabular slabs that were worked into standardised grindstone blanks in workshop areas adjacent to the quarry, in a process similar to those at the Innamincka quarries. Nearly 345 t of stone has been removed from the quarry area (assuming 1 m3 sandstone weighs 2.3 t), which translates to about 2,052 dressed millstone blanks. Many of the grindstones produced here were used locally within the Strzelecki Desert: implements made from Narcoonowie stone are mainly found within 65 kilometres of the quarry. Smith et al. (2010) calculated that the use-life of a millstone was 1,600 hours, effectively about 9 years using ethnographic consumption rates for native cereals (assuming an average daily preparation of 0.5 kilograms of ﬂour per adult woman for 90 days each year, a total of 180 hours of grinding time per woman per year). The Yandruwantha (in whose country the Narcoonowie quarry lies) would have needed to replace 1,400 millstones per century. These ﬁgures suggest that small quarries such as Narcoonowie could not have met local demand for more than 150–300 years.

Kurutiti (Helen Springs) Kurutiti – in the Ashburton Range, northern Central Australia – is an example of a major regional quarry outside the Lake Eyre basin (Mulvaney and Gunn 1995; Mulvaney 1997). It supplied grinding slabs to an area of about 135,000 km2, extending from the eastern part of the Tanami Desert to the Barkly Tableland. The workings extend over 5.5 ha, with more than 200 pits ranging from 1 mto20 m in diameter and up to 2 m deep, surrounded by sandstone quarry debris up to 5 m deep in the main part of the site. Remarkably, the workings are associated with large numbers of petroglyphs: Mulvaney and Gunn (1995) record 2,249 engraved motifs, mostly in the classic desert Panaramitee style. Like all of the quarries discussed here, production methods at Kurutiti were standardised – although the thick, uniform beds of sandstone demanded a distinctive approach. Small pits were pecked into exposed sandstone surfaces, 5–15 cm apart, to split the sandstone along the resulting stress lines, using both fire and hammer-stones. This method produced slabs 5–6 cm thick that were then flaked to shape, bifacially trimmed along the margins and then hammer- dressed to produce finished millstones. The grindstones appear to have been manufactured on demand because there is no evidence for stockpiling of finished slabs. The Chain of Connection: Trade and Exchange Across the Interior 287

The Chronology of Grindstone Quarries The development of seed-gathering economies around 3,000–4,000 years ago, and their expansion within the last millennium, would have required the provisioning of large areas of sandy desert with grinding slabs, an ongoing challenge for desert groups in these areas. The appearance of seed-grinding implements in archaeological deposits across the arid zone between 1 and 3.8 ka provides an upper limit for the age of the grindstone quarries. The large-scale production of grinding slabs in the Lake Eyre basin, however, goes well beyond the immediate provisioning needs of these landscapes: the major quarry complexes at Anna Creek, ToothsNob and Innamincka alone represent production of perhaps 1–2 million millstones over the life of these quarries. One scenario is that grindstone quarries initially developed to meet local and regional economic needs around 3–4 ka, and that later, as the quarries in the Lake Eyre basin were incorporated into long-distance exchange networks, production had to be scaled up to meet the new demands. Current data allow few opportunities to test this at present. Grindstones made on quarried slabs have not been distinguished in analyses of excavated assemblages. The chronology of the quarries themselves is uncertain. The northern part of the Innamincka complex appears to have been operating by at least 2,000 BP. This is suggested by a 14Cdateof1,700 ± 140 BP (Beta-30777), on charcoal (associated with the use of ﬁre to break up the sandstone) from a sealed context at the base of a quarry pit on Mt McLeod (Smith et al. 2010). On its own, however, this does not help to discriminate between production for local use and the hyper-production associated with ‘long-transfer’ quarries. At Narcoonowie, a range of circumstantial evidence suggests that this quarry is no older than 600 years and that it represents a late expansion of grindstone production at a marginal site. At Kurutiti, the presence of a late Holocene artefact assemblage that includes ground-edge axes (N = 15, all locally made on dolerite pebbles) (Mulvaney 1997)suggeststhatuseofthisquarrydatesto no more than 1,000 BP.

PRODUCTION OF STONE AXES FOR TRADE AND EXCHANGE The accelerated production of stone axe heads for trade and exchange provides one of the best examples of the transformations underway in the scale of these exchange systems during the last millennium. Stone axe heads formed part of a composite implement, more properly called a hatchet than an axe (Figure 8.8). Australian stone hatchets usually had a dolerite, diorite or basalt head (500–700 g), with a ground edge, bound with an adhesive (either resin or beeswax loaded with a cellulose ﬁller) into 288 The Archaeology of Australia’s Deserts

a wrap-around split-wood handle (Dickson 1981). In woodlands and savanna, they were an indispensable tool (Sharp 1952) – widely used to extract honey or possums from tree hollows, cut footholds in tree trunks, remove bark for shelters or canoes or cut and dress blanks for wooden implements. In Australian drylands, they were used in the northern and eastern parts of the arid zone, but historically they were absent from the south and west (Figure 8.9). This pattern is not simply due to more open desert terrain or an absence of suitable stone because pockets of acacia and eucalypt woodland, as well as outcrops of dolerite, occur across much of this region (e.g., Wingate, Pirajno and Morris 2004). One explanation, suggested by Hayden (1977b), is that the costs associated with making and maintaining a stone axe are only justiﬁed in areas where woodworking is extensive (and where the associated turnover of stone implements is relatively high). This may well be the case, but clearly it does not account for the archaeological distribution of stone axes because early Holocene woodlands were more extensive than today’s woodlands, and the wooden tool kit appears to have diversi- ﬁed around 3–4 ka, well before the earliest axes appear in the interior (see Chapter 6). An alternative view is that the ethnographic distribution Figure 8.8. Australian stone axe or of stone axes in the arid zone is an historical hatchet – with wrap-around handle legacy of shifts in patterns of exchange. bound with string, and stone head seated Stone axes entered desert societies by several in resin haft – collected by the Burke and different paths. On the margins of the desert, Wills Expedition in the Bulloo River axes were traded into the region from the south- area, 250 km southeast of Innamincka, ern Kimberley in the north and from temperate in 1861. The ground-edge head matches those from the Mt Isa quarries. The regions in the southeast (Binns and McBryde axe was acquired by E Giglioli, in Mel- 1972;McBryde1978). For instance, stone axes bourne, in 1867, from a member of from the Mt William quarry in central Victo- the committee that organised the state ria were exchanged along the Murray–Darling funeral for Burke and Wills. The axe is River system, reaching the Barrier Ranges in now in the Pigorini Museum, Rome. the southeastern sector of the arid zone, more (Drawn from original photograph, cour- than 700 km from their source. Within the tesy of P Jones) desert, there were also local axe quarries in Cen- tral Australia: in the Stuart Pass dolerite in the The Chain of Connection: Trade and Exchange Across the Interior 289

Figure 8.9. The chronological distribution of ground-edge axes in Australia. The earliest axes are from late Pleistocene sites in northern Australia (open squares). Axes are common in archaeological assemblages in eastern and northern Australia during the last 4,000–5,000 years (dark grey shading). They spread into the eastern part of the desert during the last millennium (light grey shading) (black triangles show excavated desert sites with stone axes dating less than 1 ka) but are rare or absent in the western half (unshaded), where they only occur in recent prehistoric or ethnographic contexts (open triangles), often as items traded from the north. (Updated from Davidson 1938; Davidson and McCarthy 1957;Dickson1981:fig.1; and Morwood and Hobbs 1995: fig. 1). Hatched area shows the distribution of axes from the Mt Isa quarries (black square). (Adapted from Davidson et al. 2005:fig.13.3)

MacDonnell Ranges, in diorites near Yuendumu in the Treuer Range, and near Kurundi in the Davenport Range (Spencer and Gillen 1899: 590). The largest arid-zone axe quarries – as well as an extraordinary density of quarries – occur in the meta-basalts around Mt Isa (Davidson et al. 2005) (see Figure 8.3). These quarries were pivotal in interregional trade because of the systematic large-scale mining and production that took place here, and their strategic position near the headwaters of the Georgina–Mulligan River system. Some forty discrete axe quarries are known in the Eastern Creek Volcanics (Davidson et al. 2005). The major quarries concentrated on ridges of meta- basalt (with a distinctive green schist facies), with less extensive quarrying of local rhyolite outcrops, maﬁc dykes and meta-andesites. Hiscock (2005) provides details of Lake Moondarra 1 (LM1), the largest of the meta-basalt quarries in the Mt Isa area. This quarry extends over an area of 2.4 km2 – with large volumes of debris, stockpiles of ﬁnished goods and clearly 290 The Archaeology of Australia’s Deserts

differentiated activity areas – and is comparable in scale to the better-known Mt William quarry in Victoria (McBryde and Watchman 1976). Quarrying activity focused on ridgetops, where circular pits, 3–10 m diameter and more than 0.6 m deep, were dug to extract the stone. Hiscock (2005) estimated that on the ridge crests, there is one quarry pit per 100 m2. Between these pits are dense piles of flaking debris up to 1 m deep and also piles of axe rough-outs, retouched to shape and then stacked. Further bifacial trimming of axe blanks took place on the flats between the ridges; some flaked blanks were ground to form finished axes, which also appear to have been stockpiled. The scale of production reflected in the Mt Isa quarries was immense. At LM1, Hiscock (2005) estimated that 800,000 axe rough-outs were stockpiled across the site. If we assume that this reflects only 75 per cent of production, and that use of LM1 spans no more than a millennium (see the following discussion), then this quarry alone represents production of about 1,000 axes per year. And if we scale up to include the other quarries in the Mt Isa district, it is clear that axe production may have been on the order of 10,000–15,000 axes per year, reflecting a considerable surplus above normal replacement needs for these implements within the region. But it is not just the scale that suggests large-scale production for exchange: finished implements were also highly standardised in size and shape, reflecting routine application of a set sequence of production. The ground-edge axes from this region ‘up to 23 cm in diameter and 2 kg in weight, smoothly flaked all over both sides, and often ground all around the edge, are ranked among the finest axes in Australia’ (McCarthy 1976a: 50). Ethnographic accounts of control and access to the Mt Isa quarries are sketchy. The Mt William and Central Australian axe quarries were controlled by men of the local patri-clans (Spencer and Gillen 1899: 590) – and this might alsohavebeenthecaseatMtIsa.WERoth(1897: 151) noted that it took a day to rough out an axe blank and another day to grind the finished axe, the latter ‘a tedious process, sometimes said to be the work of the women’ (1904: 19). Such an investment of time is corroborated by Spencer (1928: 498), although some experimental work indicates that flaking and grinding a basalt axe head need only take four hours (Dickson 1981: 41). On these figures, the production of axes at the Mt Isa quarries must have involved about 20,000– 25,000 man/hours per year, representing a significant investment of time by local hunter-gatherer groups outside normal subsistence activities. The Mt William quarries were worked in response to demand, whereas the stockpiling of axes and axe blanks at LM1 suggests the Mt Isa quarries were worked in anticipation of demand. Axes from these quarries have a wide distribution along river systems north and south of the Mt Isa district (see Figure 8.9). Their distribution extended 1,000 km south to the lower Cooper and Reap Hook Hill area in the Flinders Ranges, where axes petrologically sourced to these quarries have been found The Chain of Connection: Trade and Exchange Across the Interior 291 table 8.3. Sites with axe fragments in dated contexts in the arid zone

Levels with axe Site fragments Region Source

Upper Leichardt River <1,100 cal BP Mt Isa (6) Anvil Creek 1 <850 cal BP Selwyn Ra (1) Cuckadoo 1 <1,067 cal BP Selwyn Ra (1), (2) James Range East <800 cal BP Central Australia (3) Thereyererte <590 cal BP Central Australia (3) Puritjarra <800 cal BP Central Australia (4) Tjungkupu 1 <974 cal BP Central Australia (3) Wanmarra <1,000 cal BP Central Australia (3) Vlaming Head Midden 2 Surface ﬁnd NorthWest Cape (5)

Sources: (1) Ridges 2003: 63;(2) Davidson et al. 1993, 2005;(3)Smith1988;(4)Smith 2006;(5) Przywolnik 2002;(6) Hiscock personal communication. Fragments of dolerite with traces of grinding are found throughout Holocene and late Pleistocene levels of Mesa JJ24 in the Pilbara but are not identiﬁed as axe fragments (Hughes and Quartermaine 1992).

on archaeological sites (Veth, Hamm and Lampert 1990: 55;McBryde1997: 604–5). Trade along these inland river systems is known from the ethnography (Horne and Aiston 1924: 34) and from trade axes found cached along this route (Winnecke 1884: 12; Kelly 1968; Davidson et al. 2005). The uniformity of the axes, traded over hundreds of kilometres, is striking: the median length of axe heads from the Cooper Creek area (129 mm) is little different from that at the Mt Isa quarries (121 mm) (Tibbett 2002), consistent with rapid, long-distance transfer of these commodities. However, the evidence also indicates that ‘long-transfer’ quarries such as LM1 co-existed with small-scale production from local sources and secondary exchange systems (reciprocal ‘down-the-line’ exchange) that distributed axes away from the inland river systems (Tibbett 2002;Davidsonetal.2005). For instance, Mt Isa axes were used in the Selwyn Ranges, not just traded through this region. Although there is little direct evidence on the age of the Mt Isa quarries, the chronology of ground-edge axes provides some broad constraints on their antiquity. Ground-edge axes are found in late Pleistocene assemblages in tropical Northern Australia but only appear within the last 4,000–5,000 years in the woodlands of eastern and southeastern Australia (Morwood and Hobbs 1995) (see Figure 8.9). In Central Australia, they occur somewhat later and have not been found in contexts earlier than 800–1,000 years ago (see Table 8.3). Stone axes are not found on archaeological sites in the Great Sandy Desert but were traded into the region (Cane 1984: 167). Further aﬁeld, in arid northwestern Australia, there is little evidence that the presence of axes has any antiquity: despite the large number of archaeological excavations, axes have only been 292 The Archaeology of Australia’s Deserts

found in undated surface contexts in the Pilbara and Northwest Cape regions. These finds may reflect ‘down-the-line’ exchange from northern Australia into the desert within the last 200 years. It is relevant that, ethnographically, their use in the Pilbara was limited to rain-making ceremonies (axes are mythologically associated with lightning and the ‘lightning man’). Excavations in rockshelters near the Mt Isa quarries confirm this broad pattern: flakes and fragments of edge-ground axes only occur in levels dating less than 1.1 ka, although it is not definitely established that these are from quarried axes rather than meta-basalt river cobbles. At Cuckadoo 1, two meta- basalt axe fragments in levels dating less than 1,170 ± 50 BP (Beta 25057)have been sourced to the Group III (greenschist) meta-basalts at Mt Isa. On this evidence, the large-scale trade in stone axes along these inland river systems cannot be much earlier than the last millennium. The east-to-west chronology for the appearance of axes in the desert also suggests that the trade in Mt Isa axes in the eastern part of the arid zone may have stimulated interest in the value of axes as prestige and prestation goods in adjacent parts of Central Australia.

THE DESERT NARCOTIC PITURI The trade in pituri provides another example of long-distance exchange that was organised on a corporate and logistic level rather than on ‘down-the-line’ exchange between individuals. Colonial interest in the pharmacological properties of pituri was intense (Bancroft 1879;Howitt1904: 710–12; Johnston and Cleland 1933–4;Watson 1983). The drug – technically both a stimulant and an analgesic – was reported ‘to produce a voluptuous dreamy sensation’ (Bancroft 1879: 7). Europeans first encountered it on Cooper Creek in 1861, when members of the Burke and Wills exploring expedition were given ‘bedgery or pedgery’, which was mixed with acacia ash to form a quid for chewing (Wills 1863: 283). Abandoned and starving, the sole survivor of that expedition, John King, noted that ‘after chewing it for a few minutes I felt quite happy and perfectly indifferent about my position’ (in Moorehead 1985: 118). It took some time for Europeans to identify a sandhill shrub (1–2 mhigh), Duboisia hopwoodii, as the source of pituri. D. hopwoodii is widespread across the arid zone, but herbarium records show that it has a markedly disjunct distribution, with a small isolated population near the upper reaches of the Mulligan River, where there are groves of these shrubs (Silcock, Tischler and Smith 2012). Mulligan River Duboisia also has a distinctive chemistry, which suggests that taxonomic revision may eventually show it to be a separate variety, even though a recent analysis of plant DNA rules out the possibility that this is a clonal population with low genetic variability (French, James and Walsh 2012). The active agent is the alkaloid nicotine, present in Mulligan River The Chain of Connection: Trade and Exchange Across the Interior 293 pituri at levels ranging from 2.4 to 5 per cent (more than twice the content of commercial cigarettes or of the Nicotiana gossei bush tobacco used elsewhere in the desert). Chemical analysis shows that in Central and western Australia, D. hopwoodii is toxic because of high concentrations of D-nor-nicotine, a much more powerful alkaloid (Watson 1983; Watson, Luanratana and Griffin 1983). This explains why only Mulligan River Duboisia was exploited as a narcotic and widely traded. To prepare the drug, Duboisia shrubs were burnt to promote regrowth. Leaves and small stems were harvested in early March and artificially cured in heated sand in pits (Hodgkinson 1877;Gason1882;Aiston1937); this drying process took about two hours and arrested the enzyme action that normally degraded the nicotine level. The dried leaves and stems were then broken up and packed into small semicircular net bags, each containing about 1.4 kg (Gason 1882) and often painted with bands of red and yellow ochre. The scale of the trade was impressive. Aboriginal groups in the Cooper Creek area – including the Diyari, Wangkangurru and Yandruwantha – mounted annual expeditions involving a round trip of 800–1,200 km to obtain supplies of pituri (Gason 1882;HowittinSmyth1878;Howitt1904; Horne and Aiston 1924;Aiston1937). Pituri was also traded to groups outside the desert: 300–400 km northwards from the Mulligan along the Flinders and Gregory Rivers. And the word pijiri, originally a Pitta Pitta word from the upper Mulligan, appears to have followed the trade downstream. Gason, who witnessed the last of the traditional pituri expeditions, estimated that each man returned fully loaded with 70 lb (32 kg) of dried pituri – and this is loosely corroborated by other sources. If expeditions consisted of combined parties of about thirty to fifty Diyari, Yandruwantha or Wangkangurru men, we can estimate that about 900–1,600 kg of dried pituri was transported annually. If this figure is right, then 500–900 Duboisia shrubs would have to be harvested to sustain the Cooper Creek trade (the dry weight of foliage averaged 2 kg per shrub; Silcock et al. 2012). And if we factor in local demand as well as the trade to the north, the actual harvest may have involved 2,000 plants. Field data on the distribution and density of Mulligan River Duboisia suggests this was well within the productive capacity of these pituri groves (where there are an estimated 24,000 plants today), even if it would have been logistically difficult because the main pituri groves are in a dunefield 20 km from the nearest reliable water. What is important here is that this was direct, corporate, long-distance transfer of a commodity rather than dissemination of prestige goods through a diffuse network of individual contacts. The latter was undoubtedly important but as part of a secondary distribution process that dispersed the pituri obtained by these expeditions. Gason (1882) noted that within a few months of a pituri expedition, the Diyari had exhausted their supplies, presumably through demand-sharing and exchange as well as use. In this fashion, pituri 294 The Archaeology of Australia’s Deserts

reached the Flinders Ranges and Barrier Range a further 200–400 km to the south. Direct access to groves of pituri, as well as the use of the drug, appears to have been restricted to older men and was governed by totemic affilia- tions (Bancroft 1879;Howitt1904;Aiston1937). Some accounts suggest that particular descent groups (‘clans’, ‘pitcheri moora’ or mura) were associated with the pituri expeditions (Elkin 1931;Aiston1937). The contact period saw rapid changes in the character and scale of this exchange system. After 1880, most observers noted pituri being used freely by women; most pituri after this time was sun-dried rather than heat-cured; and much larger net bags from this period (up to 80–90 cm wide) appear in museum collections (e.g., Duncan-Kemp 1968: 185). Both Roth (1897: 100) and Aiston (1937) recorded large exchange centres, involving hundreds of people, operating as markets for exchange – apparently without the social prescriptions that applied earlier. European contact appears to have provided conditions for a rapid rise in production and more general consumption of pituri as a commodity. The archaeology of the pituri trade is unknown at present. We could expect to find evidence of processing pits associated with stands of Duboisia hopwoodii, as well as large, ephemeral open sites clustered on the periphery of the major groves. In the pituri heartland, along the Mulligan River, two major archaeological projects have been undertaken, by Barton (2001) and by Davidson (Davidson 1983; Davidson, Tarrago´ and Sullivan 2004;Ridges2003). Neither provides any direct indication of the antiquity of this trade, although Bar- ton notes that the surface archaeology in the Ethabuka area (including a site located within one of the major pituri groves) is exclusively late Holocene. Other evidence for activity along this trade network comes from a trade cache of tula adzes – artefacts that Roth (1904: 20) described as common trade items in this region. At Mucklandama Creek, near Boulia, Hiscock (1988b) excavated a pit containing thirty-three neatly stacked adzes, as well as seventeen flakes representing adze preforms. This cut through a charcoal lens dated 1,190 ± 120 BP (Beta-21518), indicating that the cache itself probably dates to within the last few hundred years.

STONE KNIVES AND LEILIRA BLADES The production of large stone blades for men’s knives shows that the restructuring of trade and exchange that occurred during the last few hundred years was not limited to the Lake Eyre basin. The 1894 Horn Scientific Expedi- tion provided the first account of these knives. They were large ‘lanceolate blades’ about 15 cm long, mounted with spinifex resin in a short wooden handle, together with a sheath of Melaleuca bark, bound with fur-string and ‘ornamented by a small plume of feathers’, to preserve the keenness of The Chain of Connection: Trade and Exchange Across the Interior 295 the sharp edge (Stirling 1896: 96–7)(Figure 8.10). Spencer and Gillen (1899, 1904) coined the term lalira or leilira blades (from the Arrernte alyweke, or stone knife). Ethno- graphically, these were men’s fighting knives and were also mythologically and symbolically linked with subincision (although never used for this purpose). As trade items, they were widely exchanged across northern and central Australia. In Arnhem Land, the quarry at Ngillipidji was a key production centre for parcels of blades circulated by the exchange cycle described by Thomson (1949). Across northern and central Aus- tralia, there are a series of large blade quarries where these knives were produced, usually exploiting outcrops of silcrete or (less often) chalcedony. In the arid zone, the best-known examples are Kanti, near Yuendumu (Graham and Thorley 1996), Kankiritja and Renner Springs in the Ashburton Range (Murgatroyd 1991;Paton1994)and several quarries in the James Range (Smith 1988)and near Macdonald Downs in Central Australia (Binford and O’Connell 1984). Typically, these sites consist of mantles of flaking debris, blades and cores (in places up to 30– 50 cm deep), surrounding a quarried outcrop or quarry pits and extending over 1–6 ha. Flaking aimed at the pro- Figure 8.10. Hafted stone duction of massive trigonal blades 15–20 cm long, struck knife with Melaleuca ‘paper- from prepared cores along facetted ridges (e.g., Binford bark’ sheath, Central Aus- and O’Connell 1984;Binford1986; Graham and Thorley tralia, 1894. Scale: the hafted 1996). blade on the right is 25 cm long. (From Stirling 1896,pt In northern Australia, dates from archaeological deposits 4,plate6,figs.1aandb) with large blades or fragments of these blades show that their production was limited to the last millennium. Despite the number of quarries, and the size and prominence of the stone knives, these artefacts are rarely found in archaeological deposits or on surface occupation sites in Central Australia (Graham and Thorley 1996; Allen 1997). One of the few reports is from Intirtekwerle (James Range East), where they are found in contexts dating sometime within the last 800 years (Gould 1978a: 109). Allen (1997) suggests their production was limited to the last millennium, whereas Graham and Thorley (1996) argue that it can be more tightly defined as a contact-period phenomenon: an archaeological manifestation of the acceleration of the northern exchange cycle described by Thomson (1949). The production of these stone knives, they argue, represents the expanded production of a local valuable for exchange, driven by the desire for the exotic goods circulating in northern Australia. If this is correct, it provides another example 296 The Archaeology of Australia’s Deserts

of the ‘encapsulation’ of desert trade and exchange by developments outside the arid zone.

REGIONAL AND TEMPORAL VARIABILITY IN DESERT EXCHANGE SYSTEMS Direct archaeological evidence for trade and exchange is still quite poor, and the best evidence comes from quarries, mines and production centres associated with ethnographic exchange systems. It is useful at this stage to group the available evidence in terms of (a) embedded procurement, (b) ‘down-the-line’ exchange and (c) long-distance transfer.

Embedded Procurement In Australia’s deserts, the archaeological record documents widespread transport of red ochre during the late Pleistocene. In most cases, the distances involved are less than 200 km and so probably represents ‘embedded procurement’, the acquisition of material during the course of seasonal or annual movements of residential groups. Evidence for long-distance exchange of goods – whether this is indirect transfer through multiple small reciprocal exchange transactions or by direct long-distance transfer – is limited to the late Holocene, mostly within the last 2 millennia.

‘Down-the-Line’ Exchange Reciprocal gift exchange or ‘down-the-line’ exchange is likely to have a long history in these desert societies. The earliest archaeological evidence, in this case for the exchange of coastal baler shell into the interior, dates to 2,300 years ago. But we can expect that much earlier evidence will eventually be found because sites on the northern margin of the desert show late Pleistocene exchange of coastal Dentalium shells 350–400 km inland (Balme and Morse 2006). Is there evidence for an expansion of exchange networks in Australia’s desert around 3–4 ka? The evidence is too poor to rule this out, but there is little to suggest an abrupt acceleration of ‘down-the-line’ exchange at this time. It is simpler to see the increasing visibility of exchange in the late Holocene record as one consequence of the trend towards higher regional populations in the desert. Birdsell (1958) has shown mathematically that higher population density increases opportunities for intergroup contacts (Figure 8.11). And ethnographic studies indicate that the turnover of exchange goods increased as opportunities for interpersonal contact grew. From this perspective, increased exchange activity is a likely corollary of growing population density. In turn, The Chain of Connection: Trade and Exchange Across the Interior 297

Figure 8.11. Geometric model showing number of contacts available to hunter- gatherer groups in Australia. Birdsell originally used this to show that the proportion of intertribal vs. intratribal contacts declined with an increase in the number of local groups. It is also shows that the total number of contacts available to local groups increases with increasing population density. (Adapted from Birdsell 1958: ﬁg. 1)

any acceleration of gift exchange would increase demand for exotic goods. This may explain the initial diffusion of ground-edge axes into the desert around 1.5 ka and the pressure to initiate local production of stone axes on the margins of the desert in the Mt Isa area. Some grindstone quarries, whose original purpose would have been to ‘provision’ parts of the desert with replacement grinding slabs, were co-opted by these exchange cycles. Intensive mining of red ochre at Wilgie Mia may have a similar history, although the exchange network that supported it seems not have persisted into the ethnographic period. The relationship of exchange turnover to population density also helps explain why the exchange of some commodities (e.g., stone axes, stone knives) and diffusion of certain social traits (e.g., subsections) began in higher density regions outside the desert and later spread into the arid interior, creating a repeated pattern of cultural ‘spreads’ from desert margins to desert core.

‘Long-Transfer’ Systems In the Lake Eyre basin, a more complex ‘long-transfer’ exchange system developed out of this undifferentiated gift economy around 1,500–1,000 years ago. In its early stages, this involved trade in stone axes and grinding slabs but later grew to include red ochre and probably pituri, within the last few hundred years. This system was limited to the arid rivers along the eastern margin of the desert and grew to take advantage of a particular set of geographic circumstances in which scarce resources were clustered in uplands at either 298 The Archaeology of Australia’s Deserts

end of a long riverine corridor. It also involved a signiﬁcant expansion of production compared to ‘down-the-line’ exchange. Estimates for the total production of goods over the operation of this exchange network include 1–2 million grinding slabs, more than 800,000 stone axes (possibly several million), 600 t of red ochre and an annual harvest of 900–1,600 kg of dried pituri. We can speculate what drove the development of this ‘long-transfer’ exchange system. One possibility is that once population growth provided initial conditions for a high turnover in gift exchange, the interplay of a range of local circumstances created a further escalation of demand and production. Several factors stand out:

r The geography of this part of the desert acted to funnel exchange into a long riverine corridor, which created a critical mass of intergroup exchange. The arid rivers also form a chain of ecological, social and historical connections across 10 degrees of latitude. This provided the prerequisite framework for long-distance expeditions to acquire Pukardu ochre or Mulligan River pituri. r There was a high level of interpersonal and intergroup competition along these arid river systems (W Jones 1979). Aboriginal groups along Cooper Creek, at the centre of this exchange network, had some of the highest population densities recorded for the arid zone (up to 1 person per 8 km2 – W Jones 1979: 138). Most early exploring parties met large groups of people, and ceremonial gatherings could number up to 1,000 people (Gason 1895: 174). This population was sustained by the extraordinary productivity of a network of channels, floodplains and ephemeral swamps. But although Cooper Creek has the highest flow rates of any dryland river in Australia, periods of high primary productivity (‘booms’) are punctuated by long dry periods (‘busts’) (Kingsford, Curtin and Porter 1999), accentuating environmental stresses on these groups. By the late nineteenth century, these stresses were evident in a number of ways: these groups had an extensive famine vocabulary; hunting had depressed the local availability of large macropods; there was extensive use of the seed-like sporocarps of ngardu (Marsilea drummondii), although this is nutri- tionally poor grain; and there was active intergroup conflict (summarised in W Jones 1979), as well as formalised duelling with murrawirrie (a large boomerang up to 180 cm long, which a fighter used like a broadsword, holding it with both hands over his head – Horne and Aiston 1924: 72–3, fig. 10). In this context, gift exchange between individuals, and ceremonial prestation between groups, assumed more than usual importance as components of the political economy: creating alliances and mutuality, reinforcing group solidarity and resolving feuds. The Chain of Connection: Trade and Exchange Across the Interior 299

Figure 8.12. Mound of kopi mourning caps on a grave, Eyre Creek, central Australia. These skull caps, made of gypsum plaster, were worn by women in the southeastern sector of the desert and the Murray–Darling basin as part of mourning rituals following the death of a relative and later placed on the grave (Davidson 1949b). In this example, the large number of mourning caps (forty to sixty individual caps) suggests the death of a senior ﬁgure. This site has not been radiocarbon dated, but an age of around 500 years would be consistent with the good condition of these fragile caps and the absence of European items, such as metal objects, or net, string or hessian liners. Associated artefacts include freshwater mussel shells (m), chipped stone implements (triangles)and a grindstone for ngardu (Ng). Goddard (1936) reported a similar grave – in this case of a woman – with seventy caps from the upper Darling region. (Source: Unpublished ﬁeld plan by MA Smith and J. Bertini) 300 The Archaeology of Australia’s Deserts

r Gerontocratic privilege (a form of social hierarchy in which older men monopolise women and resources) was marked among these groups, together with a well-developed system of community and ritual leadership (Howitt 1904; Horne and Aiston 1924)(Figure 8.12). In this politico- ritual economy, charismatic individuals with forceful personalities were able to assume the role of ‘bigmen’, redistributing exchange goods to reinforce their personal standing and create webs of social obligation. These structures also provided the capacity to organise long-distance expeditions, while the expeditions themselves provided a theatre for the exercise of leadership. The interplay of these factors may well have created a spiralling demand for exchange goods, especially exotic or prestige goods, putting strong production pressure on stone axe, grindstone and red ochre quarries and pituri groves on the geographic margins of this system. If so, the development of the long-distance exchange system was driven by the political economy of Aboriginal groups in the Cooper basin rather than by interannual shortfalls in the subsistence economy or by the need for ecological buffering (something that, in any case, was more effectively addressed within a region).

Comparisons Reciprocal gift exchange in Australia’s deserts has parallels in other desert societies. For example, in the Kalahari Desert, the hxaro exchange system created a network of reciprocity among families of San hunter-gatherers, mostly with exchange partners within 50 km (Wiessner 2002). However, the geographic scale of Australian gift-exchange systems is generally greater than those recorded for other comparable parts of the world (for southern Africa, see Mitchell 1996), reﬂecting patterns of residential movement. There are also examples of expanded production at mines and quarries in other deserts. In the northern Kalahari, there was intensive mining of specularite (a crystalline form of haematite) and native copper by hunter-gatherers (Robbins et al. 1998). But these supplied commodities to neighbouring Iron-Age pastoralists and illustrate how production among hunter-gatherer groups can become ‘encapsulated’ within wider economic systems (Gordon 1984; Wilmsen 1989; Sadr 2005). In contrast, the elaborate long-distance exchange system in the Lake Eyre basin appears to be unique among desert hunter-gatherers: an autonomous system of trade and exchange extending across half a continent, from the southern margins of the desert to its northern edge. Writing about Arnhem Land, Don- ald Thomson commented that objects acquired by ceremonial exchange have ‘a history, a background, a personality which ...givesthemagreatlyenhanced The Chain of Connection: Trade and Exchange Across the Interior 301 value in the eyes of the people who receive them. Here then, is the nucleus of a great system of exchange’ (1949: 80). Similar forces were at work in the Lake Eyre basin during the last millennium and increasingly differentiated the eastern part of the desert, with its great dryland river and channel systems, from Central Australia and the Western Desert. CHAPTER 9

THE LAST MILLENNIUM: ARCHAEOLOGY AND THE CLASSIC ETHNOGRAPHIES

Within a few months of completion of the Overland Telegraph Line through Central Australia in 1872, exploring parties began to strike out into country to the east and west of the line. Without maps or native guides, exploration consisted of a series of probes into unknown country, seeking the next water, often followed by a desperate forced march to get back to the last known water before the horses or camels died. A small, privately funded reconnaissance party led by Ernest Giles was the first of several colonial exploring parties to enter western Central Australia (Giles 1889). Giles was the quintessential Australian desert explorer, a man with an eye for the beauty and romance of these arid landscapes, who hired a novelist to ensure that his journals, published as Australia Twice Traversed, are some of the most readable and engaging accounts of desert exploration. Mounted on horses, Giles and his two companions pushed deep into spinifex and sandhill country. Small, ephemeral soakages and rockholes provide the key to this landscape. Two months into the expedition, Giles reached the painted rockshelters and waterhole at Tjungkupu (Giles’s ‘Tarn of Auber’). Rain had allowed Abo- riginal people to disperse out into the sandhill country from their fallback waters – and they were busy firing the surrounding country. ‘The natives were about,’ Giles noted, ‘burning, burning, ever burning; one would think they were of the fabled salamander race, and lived on fire instead of water’ (1889: I, 81)(Figure 9.1). Further south, in October 1872, Giles reached a small desert well on the northern rim of Lake Amadeus. Forced to use this as a staging point, he named the well Glen Thirsty, ‘for whenever we returned to it ourselves and horses were choking for water’ (1889:I,107). In nearby rock overhangs, Giles found numerous rock paintings. This was an ‘aboriginal National Gallery of paintings and hieroglyphics’, he wrote in his journal (1889: I, 101). One motif in particular caught his eye: it consisted of ‘two Roman numerals ...representingthefigureVI...paintedwithredochre’(1889:I, 95) (see Figure 9.11). Unable to travel further west, and blocked by the boggy

302 The Last Millennium: Archaeology and the Classic Ethnographies 303

Figure 9.1. A fire drive to hunt maala (Lagorchestes hirsutus) – a small spinifex wallaby – similar to hunting fires seen by Giles in 1872. This was photographed by the naturalist HH Finlayson on 9 January 1933 in the Musgraves Ranges, Central Australia. His description is worth quoting: ‘matters are so arranged that the areas where maala tracks are thickest are within the lines of flame, and upon this space attention is focused . . . The world seems full of flame and smoke ...theyswing their weight from foot to foot, twirling their throwing-sticks in their palms, and as they scan the advancing flames their great eyes glow and sparkle as the climax of the day draws near. It is their sport, their spectacle, and their meat getting, all in one’ (Finlayson 1936: 65–6). (Photograph from the HH Finlayson collection, courtesy of Northern Territory Archives) salt flats of Lake Amadeus to the south, the expedition retreated to the main ranges. Although the 1872 expedition never made direct contact with the local people, Giles’s vivid account describes a well-occupied region: this was a landscape shaped by anthropogenic fire, with a patchwork of foraging territories, and inscribed with rock art. Archaeological fieldwork in this region, at sites such as Puritjarra, Tjungkupu and Glen Thirsty, shows it was largely the product of economic and social changes among these desert societies during the last millennium, where growth of regional populations had created a more closely settled landscape. This final chapter looks at the immediate roots of the classic Arrernte and Luritja societies and their neighbours, internationally known through the hunter-gatherer ethnographies of Spencer and Gillen and others. Half a century after Giles, Spencer described these groups as relicts of an early stage in social evolution, providing a window on the origins of religion and social 304 The Archaeology of Australia’s Deserts

institutions elsewhere in the world. In a well-known passage in The Arunta,he wrote:

Australia is the present home and refuge of creatures, often crude and quaint, that have elsewhere passed away and given place to higher forms. This applies equally to the aboriginal as to the platypus and kangaroo . . . The aim of this work is to give as complete an account as possible . . . of the organization, customs, beliefs and general culture of a people that affords as much insight as we are now ever likely to gain into the manner of life of men and women who have long since disappeared in other parts of the world, and are now known to us only through their stone implements. (Spencer and Gillen 1927: vii)

Archaeological research challenges this view, providing a richer, more nuanced historical account of the development of these societies during the centuries immediately prior to European contact. Previous chapters in this book have looked at changes during this period in trade and exchange systems, in rock art and political economy, and in material culture. This chapter looks at development of the foraging economy during the last thousand years – before summarising what we now know about the social history of these desert societies.

THE CLASSIC ETHNOGRAPHIES There is a long series of ethnographies dealing with Australian desert societies, following the expansion of European inﬂuence deep into the core of the desert. The earliest accounts are of societies on the eastern and southeastern fringes of the arid zone (Roth 1897;Howitt1904; Reuther 1981 [1906]) or in parts of Central Australia opened up by the Overland Telegraph Line (Spencer and Gillen 1899, 1904, 1927; C Strehlow 1907–20); later ethnographies deal with Western Desert groups or their neighbours (Meggitt 1962; Tindale 1974; Thomson 1977; Mountford 1976; Tonkinson 1978;Myers1986). All of these are small-scale societies, having some of the lowest population densities on record (1 person per 200 km2, ranging up to 1 person per 13 km2 in the better areas), high levels of personal mobility and basic foraging economies that provide a striking contrast with the complexity of their religious life and the mythologically structured landscape. Few social anthropologists after Spencer accepted the argument that Aborig- inal society preserved elementary or primordial social traits, but the structural- functionalist approach promoted by AR Radcliffe-Brown in the 1930sdis- couraged direct research into the social history of Aboriginal groups, seeing this as crude historicism. Research in the 1980s, however, showed that hunter- gatherer societies were scaled in terms of population size, social hierarchy, extent of food processing and storage, mobility and territoriality (Price and The Last Millennium: Archaeology and the Classic Ethnographies 305

Brown 1985; Kelly 1995; Lourandos 1997). These comparative studies rejuvenated interest in the historical trajectories of hunter-gatherer societies and the factors that promote socioeconomic growth rather than stability. They also made it clear that if we want to reconstruct the social history of Australian desert societies, it will be important not just to establish when life in the desert changed to something resembling the ethnographic pattern but also to reconstruct the order in which key elements were added over time. So, what are the key social institutions of these societies? Social anthropologists identify several traits that distinguish Australian desert societies from hunter-gatherers elsewhere in the world (see especially Sutton 2003). These include the following: r Social segmentation by age and gender, combined with a classiﬁcatory kinship system based on generational moieties, sections and subsections. r A complex cultural and symbolic landscape based on geographic totemism, the ‘dreaming’ and totemic identities. Almost all accounts emphasise the extraordinary richness of ritual and ceremony among these groups and the complex cultural structuring of the landscape. r A distinctive graphic system based on tracks, circles and nonﬁgurative motifs. r Ritual property with many of the attributes of economic property, and which forms the basis of land rights. The importance of ritual property in the political economy was demonstrated in 1976, when the Aboriginal Land Rights (Northern Territory) Act recognised ‘primary spiritual responsibility’ for a site and the surrounding land as one of the criteria for ownership. Few of these traits can be traced directly using archaeological evidence, but most have correlates in the structure of economic landscapes and in the pattern of use of central places.

THE DYNAMICS OF HUNTER-GATHERER SOCIETIES Also at issue here are ideas about the dynamics of hunter-gatherer societies. Substantial social and economic change is now known to have occurred throughout Australia during the late Holocene but without general agreement on causal mechanisms. In the desert interior, the major changes took place around 1.5–1.0 ka (Smith and Ross 2008a) – somewhat later than elsewhere, where they occurred around 4 ka. Nevertheless, the region has featured prominently in debates about socioeconomic intensiﬁcation (Lourandos 1997). Different archaeologists have emphasised different aspects of these trends: economic intensiﬁcation associated with demographic packing (Smith 1988); shifts in settlement pattern (Thorley 1998b); changes in social relations, identity and relations to place (David 2002); and a decline in residential mobility, increased social interaction, and accelerated ritual and ceremonial activity (Veth 2006). 306 The Archaeology of Australia’s Deserts

Most would agree with the sentiments expressed by Veth: ‘an appreciation of nuanced social and political actions, set against a transforming set of ecological circumstances’ will provide ‘a more persuasive and satisfactory view of Indigenous history within the Western Desert of Australia’ (2006: 253). One question is the extent to which foraging societies are intrinsically limited by their economic mode. Ethnographic studies show that the structure of hunter-gatherer life is profoundly affected by the degree of residential mobility imposed by subsistence needs (Kelly 1995; Winterhalder 2001). Obligate residential mobility, combined with demand-sharing, limits the accumulation of surpluses of food or prestige goods and tends to promote personal autonomy rather than social hierarchy. Mobility requires portability and so constrains the richness of material culture. High residential mobility also contributes to the high infant mortality and wide birth-spacing observed in these groups, both of which depress demographic growth rates. All of these dampen capacity for population growth, social differentiation or development of highly elaborate material culture. This is pronounced in deserts, where the limited productivity of these ecologies imposes relatively high residential mobility and places strict constraints on population size (Cane 1987; Gould 1991; Keen 2004). However, over longer timeframes (archaeological rather than ethnographic), both environment and demography present key points of historical instability for hunter-gatherer societies. Even small net rates of demographic increase, likely to evade the usual social mechanisms of population control, potentially lead to substantial population growth over a few millennia (Pennington 2001) and create points of structural instability. Climatic shifts inevitably change the environmental baseline, enlarging or contracting the living space for desert groups. Previous chapters of this book have emphasised the sheer scale of the changes in paleoecology and environment faced by desert groups since late Pleistocene colonisation of the region. Where climatic shifts affect primary production or where demographic growth constrains residential mobility, we can expect flow-on effects in terms of structural change in these societies. These need not have always involved increases in social or economic complexity but can act in reverse: falling populations fundamentally alter the social capacity of hunter-gatherer groups, as must have occurred in many parts of the arid zone between 30 and 15 ka. All of this suggests that – on an archaeological timescale – some degree of structural change in these desert societies is inevitable (Rowley-Conwy 2001). A related question is whether the internal social dynamics of these societies create autonomous pressures for social or economic change: whether increasing social obligations, social competition and increasing demand for food and goods to finance ceremonies place pressure on production systems, independently of other factors. The classic statement is by Bender, who argued that in ‘ceremonial life lie the seeds of increased demands, more food for feast- ing, more goods for exchange, more intensive relations between elder and The Last Millennium: Archaeology and the Classic Ethnographies 307 initiate, and between elder and wives’ (1981: 154). Lourandos (1983, 1985, 1997) applied a version of this hypothesis to Australian desert societies, arguing that intergroup and interpersonal competition in these hunter-gatherer societies led to changes in ceremonial exchange, providing the crucial incentive for economic expansion and resource intensification and leading, in turn, to population growth. This created a spiral of change towards greater social complexity and the abrupt late Holocene ‘appearance of intensive and possibly ceremonially based occupation of marginal zones, for example wetlands, rainforests, highlands . . . and arid zones’ (1983: 91), as well as an economic reliance on grass and acacia seeds. Testing this sequence of events against archaeological data has been difficult: many studies simply assume that evidence of economic expansion is intrinsic evidence for social intensification. We will be on stronger ground if we can analytically decouple social and economic intensification and trace these separately in the archaeological record. The fact that the classic ethnographies describe societies that combine an extraordinarily complex ritual and ceremonial life, social segmentation and social differentiation by age and gender with a basic foraging economy should caution against the premise that social intensification inevitably drives structural changes in these economies. In some hunter-gatherer societies, increasing social competition or an escalation of social obligations may well have become an independent driver of economic change. But the latter has always seemed unlikely in Australian deserts, where there is little ethnographic evidence (a) that production for ceremonies exerted pressure for long-term structural changes in desert subsistence economies, (b) that significant quantities of food were regularly diverted towards ceremonial display, or (c) that the large ceremonies escaped strict environmental constraints on their scheduling and scale. In these societies, social intensification is more likely to be a structural adjustment to wider changes in economy and demography. The record of changes in political economy and trade and exchange discussed in previous chapters leaves little doubt that there were historical changes in the social relations of these societies. But the evidence is more consistent with the view that the social complexity of these hunter-gatherer societies is an emergent property of their demography: I see economy as architecture and social life as agency. What does come out of these debates is the role that population density plays as a key point of articulation between social and economic changes in hunter-gatherer societies. Almost all models accept that demographic growth will create a cascade of structural changes in foraging patterns, mobility and site use, trade and exchange, social relations and political economy. Binford (2001) provides a theoretical framework for understanding the articulation of the different variables. If population density rises to a point at which it begins to constrain mobility and reduce options for relocating foraging activities, a critical ‘packing’ threshold is reached. Once this threshold is crossed, 308 The Archaeology of Australia’s Deserts

hunter-gatherer groups have strong incentives to invest in strategies that enhance the yield from available habitat, to put more reliance on plant foods, to adopt more costly processing techniques and to use storage to offset shortfalls in the availability of bush food. Demographic packing also radically changes the scale of kinship articulations and promotes spatial in-ﬁlling by daughter groups. And the partitioning of groups into a greater number of competitive units leads to institutional stewardship of critical locations for resources. Any signiﬁcant reduction in residential mobility will affect birth spacing and infant mortality, amplifying the initial population increase. How did these factors play out in Australia’s deserts during the centuries before European contact? We can expect the region to have a mosaic of different histories based on the dynamic interplay of socioeconomic factors, the political economy and local environmental conditions. The challenge for archaeology has been to unpack the history of these groups in any detail.

ENVIRONMENTAL CHANGE 1,500–1,000 YEARS AGO A range of evidence suggests that during the last millennium, hunter-gatherer groups in Australia’s deserts inhabited a landscape where net productivity was higher than previously. Although the late Holocene saw the onset of a drier, more variable El Nino–˜ Southern Oscillation (ENSO)-dominated climate in the interior at 4.0–3.7 ka, these conditions appear to have relaxed around 1.0 ka in both northern and central Australia, with a strengthening of the summer monsoon and a weakening of the ENSO circulation (Shulmeister and Lees 1995; Marx, McGowan and Kamber 2009;Smith2009b). These changes result from a reorganisation of atmospheric circulation, driven by changes in insolation, and are reflected in other records of rapid environmental change from both high and low latitudes, including the Greenland and Antarctic ice cores, dust records and deep ocean foraminifera (Mayewski et al. 2004; Wanner et al. 2008;Marxet al. 2009; Moros et al. 2009). The regional impact of these varied, and not all sites responded synchronously or equally, but many records show modest global warming around 1.0 ka (represented in Europe by the Medieval Cli- mate Anomaly), followed by global cooling from 0.7 ka (Bertler, Mayewski and Carter 2011) (Europe’s ‘Little Ice Age’). In Australia’s deserts, fine-grained data on regional palaeoenvironments between 3.0 ka and 1.0 ka are still frustratingly sparse. In Central Australia, local grasslands had recovered by 1.5 ka, when the modern vegetation appears to have become established around Puritjarra (Smith 2009b). Phytolith spectra indicate an open shrubland with abundant grass in the understorey, approxi- mating modern sandhill and spinifex communities in the region today – and with the highest Poaceae values since the last interglacial. By 800 years ago, Leporillus rat middens in the Central Australian ranges show that vegetation The Last Millennium: Archaeology and the Classic Ethnographies 309 on rocky slopes was also responding to changes in regional rainfall; in this case, with an increase in perennial shrub cover (Nelson, Webb and Long 1990; Berry 1991; Webeck and Pearson 2005). The last 1,500 years was also marked by an increase in the frequency of high-magnitude palaeofloods along the Finke, Todd and Ross Rivers (Pickup, Allan and Baker 1988; Patton, Pickup and Price 1993; Bourke 1998), as well as by renewed dune activity in Central Australia (Nanson, Chen and Price 1995; Twidale et al. 2001; Hollands et al. 2006) and further south in the Strzelecki and Tirari Deserts (Fitzsimmons, Rhodes, Magee et al. 2007). Taken as a set, these records suggest an expansion of summer rainfall grassland in Central Australia during the last 1,500 years, with extreme wet–dry events reflected in episodic dune mobilisation and major floods on the Central Australian river systems. Summer precipitation in Central Australia is influenced by the interaction of several climate systems, including the northwest monsoon, tropical cyclones and ENSO, all of which attenuate towards the centre of the continent (Gentilli 1972). Collectively, these contribute to the high annual variability in rainfall in the interior, which is greater than in most other desert regions (Van Etten 2009). The ENSO, in particular, is associated with cyclic floods (La Nina)˜ and drought (El Nino)˜ in central Australia. The fine-grained ENSO records from South America show that the frequency of ENSO events (in this context, Australian droughts) peaked around 1.2 ka, then declined towards present levels (Moy et al. 2002; Riedinger et al. 2002; Wanner et al. 2008; Conroy et al. 2008). A decline in continental dust emissions, sourced mainly from the riverine systems east of Lake Eyre, records a similar decrease in climatic variability and ENSO activity from 0.9 ka (Marx et al. 2009). These changes clearly did not eliminate the importance of ENSO in the ‘boom and bust’ ecology of Australia’s deserts, but mark a reduction in the frequency and severity of these events. These palaeoecological and geomorphic records necessarily provide a time- averaged picture of climate during the last millennium. Contemporary climate records, however, give a more detailed picture of the sort of interannual and interdecadal variability that may have prevailed during this period. Holmgren et al. (2006) summarise the ecological effects of ENSO in drylands. Frequent severe ENSO events destabilise local ecology, but at lower frequencies they are an important driver of environmental productivity in arid and semi-arid systems, where productivity is strongly limited by precipitation. The extraordinary increases in primary production during La Nina˜ ‘boom’ years percolate up through entire food webs and are crucial for plant recruitment, productivity and diversity in these ecosystems. The effects of an ENSO event persist for years afterwards and drive the ‘pulse and reserve’ dynamics characteristic of these arid systems (Letnic and Dickman 2006). No one who witnessed the immense volume of floodwater moving through the desert’s river systems during the 2010–12 La Nina,˜ or the fields of chest-high grass, waves of insects, 310 The Archaeology of Australia’s Deserts

birds and native rats, and the extensive wildﬁres across Central Australia, could doubt the role of these events in revitalising the desert. Like most foraging economies, Aboriginal societies were intimately connected to their environment: critical resources such as the distribution and persistence of wells, rockholes and soakages constrained access to country and governed which parts of it could be exploited, and the productivity of the country within foraging range of a water point determined how long a group could stay in an area. Both of these factors – access and range, and productivity – appear to have improved during the last millennium. But environmental variability remained high on world scales. ‘Boom’ years removed constraints on expansion and growth, but protracted drought periodically put considerable pressure on desert societies, amplifying population-driven resource stress. Major ethnographic droughts, such as those around 1900 and in the late 1920s, exceeded the capacity of Aboriginal groups to cope in situ, prompting major out-migration and fallback to core areas, ﬁssioning of the remaining population into small mobile groups and high regional mortality (estimated at 10–25 per cent of the population; Kimber 1990: 162).

THE LATE HOLOCENE ARCHAEOLOGICAL RECORD The last 1,000–1,500 years saw signiﬁcant changes in the archaeological record in the interior of the Australian arid zone, with shifts in settlement pattern and site use, tool inventories and rock art. The overall trend in the interior is summarised in the time-series plot of radiocarbon dates shown in Figure 9.2. There are marked increases in counts of radiocarbon dates, numbers of dated sites in each time period and net accumulation rates of chipped stone artefacts (no. artefacts/m2/1,000 yr) at this time. Much of the evidence points to regional population growth, beginning 1.5–1.0 ka. As part of this, hunter- gatherer groups appear to have increased their use of marginal or outlying areas as these became seasonally accessible, with more extended occupation of existing sites in the better-watered ranges, more processing of acacia and grass seeds and an increase in territoriality (shown in greater differentiation of rock art complexes during the last few hundred years). Reviewing the evidence across the arid zone, two other trends are apparent. First, population growth in the core of the arid zone seems to have lagged behind that on the arid west coast, where comparable levels may have been reached by 4 ka (Smith et al. 2008a). Binford argues that the ‘packing threshold’ should be reached earlier in more stable habitats, where population growth is faster (2001: 439). If so, the broad-spectrum transition it triggers may have occurred earlier on the west coast than in the desert core. There is some evidence for this – as Veitch (1999) argued for the switch to Anadara middens after 4 ka (see Chapter 6) – but more work is needed. The Last Millennium: Archaeology and the Classic Ethnographies 311

Figure 9.2. Time-series distribution of radiocarbon (14C ages) from archaeological sites in the arid interior (inland Pilbara, Western Desert, Central Australia and Lake Eyre Basin) over the last 5,000 years. N = 263 dates. (a) Standard summed probability plot, showing an increase in occupation during the last millennium. (b) The same dataset, applying a theoretical correction for exponential decay (following Williams 2012). (c) The corrected plot with a 500-yr moving average. Neither (b) nor (c) removes the peak during the last 1 ka. (Source: AustArch1 database)

Riverine habitats on the temperate margins of the arid zone in southeastern Australia may also have seen an earlier transition, but at present we lack the data to test this. The second point is that even within the last millennium, there is evidence for changes in site use and settlement on a range of geographic scales and timescales, particularly within the last 500–200 years (Thorley 1998b; Holdaway et al. 2010). But, at present, the data are not suf- ﬁciently resolved to separate the dynamics of local systems from short-term historical trends. A number of studies ask whether the proliferation of sites within the last millennium is a real trend or wholly an artefact of differential preservation and loss of older sites (Robins 1993; Holdaway, Fanning and Rhodes 2008; Marwick 2009). For time-series radiocarbon data, Williams (2012) tested this, imposing a theoretical correction for exponential time-decay on the radiocarbon plot. This does not remove the peak in late Holocene activity, although 312 The Archaeology of Australia’s Deserts

it does reduce its amplitude (see Figure 9.2b). Using a computer simula- tion, Johnson and Brook (2011) also found that differences in site destruction could not account for the apparent increase in late Holocene occupation. Both studies give conﬁdence that the time-series trends record real changes in archaeological settlement during the late Holocene. However, the strongest evidence for late Holocene expansion relies on a multiproxy approach, in which counts of radiocarbon dates and numbers of dated sites in each time period coincide with changes in tool inventories, net accumulation rates of chipped stone artefacts (number artefacts/m2/1,000 yr) and patterns of site use. Many rockshelters have occupation debris intercalated within sedimentary sequences that are more or less continuous over the last 3 ka. These are well placed to record systemic changes towards more extended or more frequent episodes of occupation.

CENTRAL AUSTRALIA: SITES AND ASSEMBLAGES The best regional dataset is that from Central Australia (Thorley 1998b;Ross 2003; Smith and Ross 2008a), where there are now forty-two excavated sites, although only thirty-one of these have radiocarbon chronologies (Table 9.1). Most research has focused on areas south of the main MacDonnell Ranges or in the middle reaches of the major river catchments (Figure 9.3). Although the regional spread of sites is circumscribed, the sample includes sites in a range of localities: plateau and sandplain sites, sites in the middle reaches of major drainages, sites near springs, sites in the duneﬁelds and sites both on the rim and in the lower part of the Amadeus basin. The archaeological record for the late Holocene shows an expansion of settlement beginning around 1.5 ka and taking a millennium or so to play out. This involved more sites, more intensive use of existing sites and more use of peripheral localities within the main ranges and of marginal habitats in sandhill and spinifex country beyond the ranges.

Site Histories Excavated sites show several patterns of use (Figure 9.4). Single-phase sites have a discrete, relatively dense occupation unit, often with large amounts of charcoal in addition to high densities of chipped or ground artefacts (A on Figure 9.4). In most cases (n = 17), this is a late Holocene occupation unit, dating to within the last 1,500 years (but with considerable variability in basal age), either resting on bedrock or overlying sediments containing few artefacts. The pattern is one of more extended occupation of sites previously used in a more expedient or intermittent way. Comparison of artefact densities (calculated as no. of lithics/10 kg deposit) across this change reveals a mean increase table 9.1. List of excavated late Holocene archaeological sites in Central Australia (updated from Smith and Ross 2008a)

Site Late Holocene unitb Site name Type historya (basal age ka) Source Arnngu Rockshelter A 1.29 (14) Glen Thirsty 1 Rockshelter A 1.40 (7) Glen Thirsty 3 Rockshelter A 0.29 (7) Intirtekwerle (James Range E) Rockshelter A 0.78 (2,9) James Range NW Rockshelter A 1.57 (15) Kweyunpe 1 (106) Rockshelter A 0.58 (3,4) Kweyunpe 2 (107) Rockshelter A 0.37 (3,4) Kweyunpe 6 Rockshelter A 1.36 (8,10) Tjungkupu 1 Rockshelter A 0.90 (10) Watar rka NEP23 Rockshelter A 1.36 (18) Atnetye Open site A 0.19 (19) Keringke Open site A 0.73 (12) Kulpi Mara open site Open site A 0.75 (14) Simpson Desert 275 Open site ?A ?1.13 (5) Therreyererte Open site A 0.58 (10) Tjungkupu 2 Open site A 0.81 (10) Urre Open site A 1.15 (10) Wanmara Open site A 1.44 (10) Ilarari 17 Rockshelter B – (8,10) James Range S Rockshelter B 0.40 (2) Rrewurlpmurlpme kweke Rockshelter B – (10) Watar rka NEP22 Rockshelter B – (6) Jesse Gap Sand Quarry Hearthd B– (1) Mosquito Bore Hearthd B– (1) Simpson Desert 226 Hearthd B– (5) Ilarari 19 Rockshelter C 2.85c (14) Kwerlpe (66) Rockshelter ?C 3.90c (15) Irtikiri Rockshelter D 0.35 (14) Kulpi Mara Rockshelter D 3.95c (13) Puritjarra Rockshelter D 0.70 (11) Wanga East W04 Rockshelter D 0.74 (18) Camel Hump (17) Rockshelter ? ? (16) Camel Hump (23) Rockshelter ? ? (16) Deep Well Cave (63) Rockshelter ? ? (16) Kwerlpe (70) Rockshelter ? ? (16) Kwerlpe (71) Rockshelter ? ? (16) Madigan Cave (111) Rockshelter ? ? (16) Ooraminna rockshelter Rockshelter ? ? (17) Ooraminna (ORH16) Open site ? ? (8) Simpson Desert 162 Open site ? ? (5) Simpson Desert 86A Open site ? ? (5) Simpson Desert 86b Open site ? ? (5) Most excavated rockshelter sites also contain rock art assemblages, and many of the open sites have rock art assemblages on nearby rock faces or in adjacent rockshelters. a For definitions, see Figure 9.4. Not all sites have been reported in sufficient detail to establish the pattern of occupation. b Interpolated ages (cal BP) for the base of the main late Holocene occupation unit at each site. Not all sites have been dated or have significant late Holocene occupation. c These sites represent an earlier phase of occupation c. 2.5–3 ka. All appear to have a veneer of more recent occupation with well-preserved organics c. 0.20 ka. d Isolated hearths. Sources:(1) Bourke 1997;(2) Gould 1978a;(3) Greathouse 1985;(4) Napton and Greathouse 1985;(5)Naptonand Greathouse 1996;(6) Ross and Smith 2007;(7) Smith and Ross 2008b;(8) Smith 1983;(9) Smith 1986b;(10) Smith 1988;(11) Smith 2006;(12) Stockton 1971;(13) Thorley 1998a;(14) Thorley 1998b. Unpublished field notes:(15)RA Gould 1973;(16) LK Napton and A Albee 1973;(17) LK Napton 1980;(18) MA Smith and J Ross 2006;(19)MA Smith 2010. 314 The Archaeology of Australia’s Deserts

Figure 9.3. Central Australia. Land more than 700 mabovesealevelisshownindark grey and marks the major range complex (including the MacDonnell Ranges, James and Krichauff Ranges, and the George Gill Range). Triangles show late Holocene sites mentioned in the text: (1) Glen Thirsty, (2) Puritjarra, (3) Tjungkupu – Tarn of Auber, (4) Watarrka NEP22 and 23,(5) Wanga East W04,(6) Wanmara, (7) Irtikiri, (8) the Kulpi Mara complex, (9) the Illarari sites, (10) Arnngu, (11) Rwerlpmwerlpme kweke, (12) the Kweyunpe sites, (13) James Range Northwest, (14) Urre (Rainbow Valley), (15) Intirtekwerle (James Range East), (16)Kwerlpe,(17)Keringke(Santa Teresa) and (18) Therreyererte.

of 6:1 from the underlying levels to the major occupation unit. Tjungkupu – the locality visited by Giles in 1872 – is an example of this. Here, excavations in the largest rockshelter show a rich occupation unit with a basal age of 0.9 ka, capping a layer of aeolian sands and sandstone rubble (Figure 9.5). This marks the point at which the input of anthropogenic sediments became the dominant element in an otherwise natural accumulation of aeolian sands and sandstone rubble (Table 9.2). Another example is the Intirtekwerle (James Range East) site in Central Australia (Gould 1978a;Smith1986b, 1988). Here, the rockshelter sediments are an extension of the sand sheet that forms the valley ﬂoor The Last Millennium: Archaeology and the Classic Ethnographies 315

Figure 9.4. Schematic diagram illustrating different site histories (after Smith and Ross 2008a). (a) Sites with a major occupation unit <1.5 ka, either overlying a pattern of ephemeral usage or resting on bedrock (N = 17). (b) Sites with ephemeral usage throughout their history (N = 7). (c) Sites with a single major occupation unit >1.5 ka, with a subsequent decline in usage (N = 2). (d) Sites with multiple rich occupation units (N = 4). Calibrated ages (ka) are shown. Sites shown include James Range Northwest (JR NW), Intirtekwerle (IDK), Kweyunpe 6 (KYB6), Arnngu (ARN), Illarari 17 (IL17), Rwerlpmwerlpme kweke (RMK), Illarari 19 (IL19), Irtikiri (IRT) and Kulpi Mara rockshelter (KM).

Figure 9.5. Stratigraphic proﬁle for Tjungkupu 1, showing a dense occupation unit (layer I) with a basal age of 0.9 ka, overlying a layer of aeolian sand and sandstone rubble with a basal age of ∼3.0 ka. (Source:Smith1988:ﬁg.5.3) 316 The Archaeology of Australia’s Deserts

table 9.2. Excavated assemblages at Tjungkupu 1, comparing the major late Holocene occupation unit (layer I) (0–0.9 ka) and underlying levels (layer II, 0.9–3.0 ka)

Layer I II

Charcoal g/m3 5,971 12 Bone g/m3 828 – Eggshell g/m3 5 – Ochre g/m3 38 – Lithics no./m2/100 yr 1,974 4 Grindstones no./m2/100 yr 13 <1 Lithics (total no.) 5,624 14 Geometric microliths 14 – Tula adze slugs 20 – Burren adzes 2 – Amorphous retouched tools 68 1 Cores 16 1 Redirecting ﬂakes 30 Grindstones (total no.) 37 1 Seed-grinders 6 – Amorphous grindstones 31 Grindstone fragments 28 – Flakes from ground-edge axes 1 –

Source:Smith(1988)

in front of the shelter (Figure 9.6). Occupation debris extends out across this for 30 m beyond the dripline. Excavations show that stone artefacts occur in low numbers (75/m2/kyr) throughout an essentially continuous sedimentary sequence beginning around 3 ka (the base of layer III). Use of the site was more intensive after 0.8 ka (layer I), creating a dense charcoal-rich deposit with large quantities of burnt bone and grindstones and a high density of chipped stone artefacts (1241/m2/kyr), all of which reﬂects a fundamental shift in the way this site was used. This is a pattern repeated throughout Central Australia, at both rockshelters and open sites. Less common is a pattern in which a dense, discrete occupation unit occurs prior to 1,500 years (see C on Figure 9.4). There are only two examples of this: Ilarari 19, where a rich occupation unit dating 3.0–2.1 ka is overlain by a veneer of more ephemeral recent use (probably postcontact) (Thorley 1998b); and possibly Kwerlpe 66, where the major occupation unit dates to about 3.9 ka and is overlain by a more recent but undated layer. Essentially, these represent sites where extended occupation ceased prior to the last 1,500 years. A small number of sites have multiple occupation units (n = 4) (see D on Figure 9.4), all with signiﬁcant densities of occupation debris and artefacts, and sometimes with disconformities between the units. These sites include The Last Millennium: Archaeology and the Classic Ethnographies 317

Figure 9.6. Stratigraphy in Trench 1, Intirtekwerle (James Range East) rockshelter. This shows the interdigitation of sandsheet sediments and sandstone rock fall (layers II and III) and the major late Holocene occupation unit (layer I). (Source:Smith1988: ﬁg. 7.2)

Irtikiri, where a basal unit dating to 2.7 ka is overlain by a rich late Holocene unit dating within the last 350 years; Wanga East (W04), where the major use of the rockshelter dates to between about 9.0 ka and 5.0 ka, overlain by a late Holocene unit dating within the last 500 years; and two sites with long Pleistocene records of occupation: Puritjarra (Smith 2006)andKulpi Mara (Thorley 1998a). Irtikiri, Puritjarra and Wanga East have discrete late Holocene units, equivalent to those in Figure 9.4a. Kulpi Mara has a veneer of occupation debris dating to 0.6 ka overlying an occupation unit dating to 2.6 ka (Thorley, Faulkner and Smith 2011). There is also dense, late Holocene occupation on the open sandplain nearby (the Kulpi Mara open site) and also rich assemblages of late Holocene stencil art in nearby rockshelters. This is probably the clearest evidence in the sample for a local shift in the focus of occupation from rockshelters to open sites. Finally, there are a series of sites that register only intermittent use throughout their history (see B on Figure 9.4). These include Watarrka NEP 22, Ilarari 17, Rwerlpmwerlpme kweke and James Range South (and also the three isolated hearths listed in Table 9.1). Overall, the pattern is one of more sites with extended occupation during the last 500–1,500 years, superimposed on an earlier pattern of expedient use of the same localities. There are several sites with longer records of occupation and several which fall out of use after 5.0–2.5 ka. Although there may be a geomorphic as well as an occupation signal in these patterns – rates of aeolian deposition on sandplain sites and in some rockshelters are high, typically 500–1,000 mm/1,000 years (Smith 1988; Thorley 1998b: 244) – there has 318 The Archaeology of Australia’s Deserts

been considerable testing of deposits where there is no obvious indication of late-phase occupation. This has included exploratory augering, test excavation and the prospecting of erosion gullies cut into sediments in focal localities. In addition, at some sites – Intirtekwerle, for example – excavations exposed several metres of sandplain sediments underlying the latest occupation unit, without ﬁnding comparable earlier occupation units. Rich ‘buried’ occupation units are genuinely rare in this landscape. At sites with signiﬁcant late Holocene occupation, the basal age of this unit ranges from 0.3 ka to 1.6 ka, with a mean of 0.90 ± 0.4 ka. For most sites, the switch to more intensive use took place between 1,500 and 500 years ago. The spread of ages suggests that at the regional level, this is not an ‘event horizon’ but rather the outcome of a process that progressively affected sites over the course of a millennium.

Site Inventories Comparison of site inventories is difﬁcult because of differences in reporting of excavated assemblages and because of the limited number of artefact classes in these assemblages. Smith and Ross (2008a) were able to assemble data for thirty- seven assemblages, spanning the last 4,000 years, using the most commonly reported classes of artefact. Bone is rarely present in these assemblages and not routinely quantiﬁed. Red ochre is common but usually reported on a presence/absence basis. Smith and Ross (2008a) used multivariate analysis of these excavated assemblages to prospect for groups of sites and to examine whether there were differences between assemblages dating to less than 1.5 ka and those dating to 4.0–1.5 ka. The results show that there is a shift at 1.5 ka towards dense grindstone-rich deposits and larger, more diverse lithic assemblages (this is apparent even when the size of excavation is held constant; Figure 9.7). This is consistent with the idea that most plant processing gear will be associated with sites with extended occupation. Detailed analysis of individual site sequences – as at Puritjarra (Smith 2006)– corroborates this overall pattern: the major late Holocene unit, in this case dating within the last 800 years, is richer (richness: number of tool types/log (sample size) = 5.63 compared with 4.80) and more diverse (Shannon-Weaver diversity index, H = 1.95 compared with 1.74), with less residential mobility indicated than in the immediately underlying levels.

Rock Art A separate line of evidence is provided by rock art: both pigment art (painting and stencils) and rock engravings (see Chapter 7 for more detail). The main temporal trends (loosely dated to 1.5 ka) include a signiﬁcant increase in The Last Millennium: Archaeology and the Classic Ethnographies 319

Figure 9.7. Principal components analyses of site inventories in Central Australia. This shows variability in excavated assemblages (n = 37) and contrasts assemblages of less than 1.5 ka (black circles) with those of more than 1.5 ka (grey circles). The frequency of grindstones and ground-edge implements is more strongly correlated with the density of remains than with assemblage size. (Source: Smith and Ross 2008a,ﬁg.4) the number of art complexes, with a change from discrete focal art sites to larger complexes, each with an array of satellite sites containing smaller assemblages (Table 9.3). Overall, there is also greater differentiation among art complexes and greater production of rock art. Motif counts also illustrate the scale of differentiation between sites in the later period: 61 per cent of art complexes have fewer than 200 motifs, whereas 16 per cent (n = 8)have more than 1,000 motifs. Ross (2003: 160) also shows a statistically signiﬁcant association of hand stencils and handprints at sites with grindstones, paralleling table 9.3. Changes in number and distribution of rock art sites in Central Australia

Perioda No. art complexesb No. satellite sitesc

Art assemblages <1.5 ka 49 247 Art assemblages >1.5 ka 21 – a Estimate based on relative sequence and archaeological associations. b N = 51. These are clusters of art sites in a locality, containing from one to thirty-two individual sites. Some complexes have multiple phases of art production. c Satellite sites are outlying art sites associated with a complex. Most date within the last 0.5 ka. Source: Smith and Ross (2008a) 320 The Archaeology of Australia’s Deserts

the high frequency of grindstones in excavated late Holocene assemblages. At present, the lack of precise dating for rock art means that we cannot determine whether the changes in rock art lead or lag the changes in economic landscapes.

Site Location and Settlement Pattern Although this period involved some restructuring of land use within the main drainage systems and a broadening of resource zones, there does not appear to have been a major redistribution of population away from the middle and lower reaches of river valleys to sites on sandplains on the margins of the ranges or towards the lower reaches of these river systems. Analysis of site location shows that spinifex hummock grassland communities dominate site catchments throughout the last 3 ka (Smith and Ross 2008a). This is unsurprising because with more than 32 per cent of important or staple food species (36 of 112 species), spinifex-dominated sandplain communities have the richest and most diverse suite of plant food species (Latz 1995). The main temporal trend is for exploitation of a wider range of vegetation communities during the last millennium. One change that is evident in the data but is hard to quantify is greater use after 1.5 ka of peripheral locations. Within the main ranges, this involved establishment of new occupation sites and rock art sites away from the focal river channels. The establishment of sites in outlying areas near ephemeral waters includes sites visited by Giles in 1872 – Tjungkupu and Glen Thirsty 1 (Figure 9.8). Both are complexes of sites in an area where there is a long record of late Pleistocene and early Holocene occupation (at Puritjarra rockshelter), but neither locality was used in an extended fashion prior to 1.5 ka. Both are dependent on ephemeral rain-fed soakages and waterholes focused on isolated sandstone strike ridges, and use of both sites is likely to be sensitive to shifts in rainfall and evaporation. In both cases, local rock paintings and engravings appear to relate to the late Holocene occupation. Within the Glen Thirsty complex, the rock art and occupation histories of the sites provide intrinsic evidence for the establishment and consolidation of a new cultural landscape; they document a gradual expansion of use of this locality, moving outwards from initial focal areas, and a shift towards the painting of large site-speciﬁc iconic motifs at these sites (Smith and Ross 2008b). These data suggest some restructuring of land use and settlement in the last millennium or so. This requires better deﬁnition, but the overall pattern is one of local expansion of settlement, into peripheral areas within the core drainage systems and ranges, and into more marginal areas in sandhill and spinifex country dependent on very ephemeral waters. The Last Millennium: Archaeology and the Classic Ethnographies 321

Figure 9.8. Excavations at Glen Thirsty 1, August 2004. (Photograph courtesy of MA Smith)

Large Ceremonial Aggregation Sites A common pattern is for large open sites, often associated with major ceremonial gatherings, on sand mantles at the mouth of gorges with focal rock holes. Only two have been excavated in Central Australia – Therreyererte and Atnetye – but at these, the major occupation unit dates within the last 500 years (see Chapter 7). Although limited, the evidence indicates that the intensiﬁcation of ceremonies at these major totemic sites postdated changes in economic landscapes. Nevertheless, the large number of seed-grinders on these sites highlights the importance of grass, acacia and chenopod seeds in underpinning these ceremonial gatherings.

THE INLAND PILBARA AND WESTERN DESERT In the Pilbara, there is an increase in radiocarbon-dated sites during the late Holocene, especially after 1.2 ka, although regional data alone are currently too weak to substantiate these trends (Marwick 2009;RyanandMorse2009). The picture is stronger in the Western Desert. Here, almost all excavated sites show more dense accumulations of artefacts and charcoal during the last millennium – including Serpents Glen (O’Connor, Veth and Campbell 1998), Puntutjarpa (Gould 1977)andBushTurkey3 (Veth et al. 2008). At Kaalpi, there is a switch to a grindstone-rich assemblage around 1.2 ka (although the concentration of ﬂaked stone artefacts declines after 0.8 ka) (Veth, Smith 322 The Archaeology of Australia’s Deserts

and Haley 2001). In the north of the region, excavated sites in the Rudall River area show the same pattern with ‘a consistent trend in increasing rates of artefact discard, accelerating during the period 1400 to 800 BP’ (Veth 1993: 66). On the southwestern margin of the Western Desert, Walga rock records a two- or threefold increase in artefact densities around 0.9 ka (1,040 ± 180 BP LY-2098) (Bordes et al. 1983).

THE SOUTHEASTERN SECTOR OF THE ARID ZONE Archaeologists working in the southeastern part of the arid zone have also commented on the abundant traces of late Holocene occupation in the region (see Hughes and Lampert 1980; Lampert and Hughes 1987; Williams 1988, 1998; Veth, Hamm and Lampert 1990 for the Strzelecki dunefield and Cooper Creek area; Florek 1993 for campsites associated with mound springs along the western side of Lake Eyre; Holdaway et al. 2002 for the Barrier Ranges region; Slaytor 2005; Hughes et al. 2011 for the Lake Torrens area; and Ross 1981 for the Victorian mallee). Many of these are surface scatters of stone artefacts: although typological criteria only narrow their age to within the last 4 ka, wherever radiocarbon dates on hearths and shell are available, the trend is for most to date within the last millennium. Two regional studies illustrate these trends in more detail. The Coongie Lakes were studied by Williams (1988, 1998). These shallow, ephemeral and semipermanent freshwater lakes form an important arid-zone wetland at the terminus of the northwest channel of Cooper Creek. In 1862, explorer John McKinlay recorded unexpectedly large numbers of Aboriginal people in this area, often in groups of several hundred. Ethnographic accounts for the Cooper basin describe people living on fish and freshwater mussels, the seeds of portulaca and panicum, waterfowl and small marsupials, and rats, which Howitt describes being ‘roasted in the coals’ and ‘eaten seriatum, very much like biting a sausage’ (1862:Diary,10 July). Systematic field surveys in the Cooper basin complement the ethnography, confirming that the Coongie Lakes were a focal point for occupation (Veth et al. 1990; McConnochie 1996; Hutchins 1998). In this area, archaeological sites often extend over 2,000– 10,000 m2 and include scatters of heavily reduced stone artefacts (tula adze flakes and adze slugs, cores, small scrapers and unmodified flakes), grindstones, burnt animal bone, shell fragments, burials, burnt clay, intact hearths with termite-mound heat-retainers and the remains of hut frames (Williams 1988, 1998). Radiocarbon dates on hearths, fireplaces, burials and shell artefacts all cluster between 1.0 ka and 0.4 ka (Pretty 1968; Williams 1998). There are few sites older than this: Williams (1998) located a single hearth in dune sediments dating to 3.2 ka. The second study deals with spatially extensive stone artefact scatters on deflated surfaces along drainage lines in the region between the Cooper and Darling River systems. The large numbers of stone artefacts on this kind The Last Millennium: Archaeology and the Classic Ethnographies 323 of arid-zone site have attracted repeated attention over the years. But the practical difficulties of untangling both a complex palimpsest of material of different ages and different types of occupation and the effects of geomorphic reworking of these scatters have defeated most researchers. In one of the most ambitious attempts to tackle this problem, Holdaway et al. (2002, 2005, 2008, 2010 – but see also Shiner 2008) used geographical information system (GIS) techniques to map artefact scatters in detail across more than 10,000 m2 in four separate study areas and used a large series of radiocarbon dates (n = 96) on heat-retainer hearths, preserved as residuals on the eroded land surfaces to provide a chronology. These small alluvial systems are some of the most geomorphically complicated parts of the arid landscape: with discontinuous accumulation of sediments, plugs of sediments working downstream and phases of erosion and surface stripping. Fanning (1999; Fanning et al. 2009) used OSL dating of sediments and 14C dating of carbonates to determine the age of land surfaces and to factor out the effects of destruction of older archaeological remains. Even when allowance was made for geomorphic visibility, the results showed a temporal concentration of hearths between 1.0 ka and 0.4 ka and many fewer dated hearths between 3.5 ka and 2.0 ka (Holdaway et al. 2008, 2010). Unlike the Coongie Lakes sites, few of these artefact scatters represent central-place sites, semipermanent camps or large aggregations of people: rather they appear to reflect repeated but low-intensity use of resource zones along drainage lines within these foraging territories. In this area, Burkes Cave, located in Scropes Range about 30 km from the Darling River, also shows a dense, charcoal-rich occupation deposit with much higher densities of retouched artefacts and unmodified flakes after 1.7 ka (1,850 ± 240 BP ANU-704) (Allen 1972: 138–218). Shiner et al. (2007) show that this unit represents a palimpsest of different sorts of occupation, but they did not attempt to characterise changes in site use over time. Holdaway’s study illustrates some of the problems of separating fine-grained historical trends from the normal perturbations within a regional ecology. In this example, short gaps in the dated hearth series within the last 1,500 years may correlate with climatic shifts – as Holdaway et al. (2002) suggest. Alterna- tively, these may reflect vegetation dynamics within a single climatic regime. Arid rangelands typically have multiple stable states under a given climate, and flood and fire events can trigger a local flip between semi-arid grasslands and woodlands (Westoby, Walker and Noy-Meir 1989; Friedel 1991). This would have the effect of laterally shifting the foci of foraging within these river corridors, changing encounter rates for large kangaroos and local availability of seed grasses, and presumably affecting local rates of hearth construction.

FIRE AND PATCH BURNING The shorter timescales represented by the archaeology of the last millennium allow us to look at the interplay of human modiﬁcation of the landscape 324 The Archaeology of Australia’s Deserts

with climate change. Fire is the major vehicle for this, as Giles saw in 1872. A patchwork landscape made up of vegetation in early and late successional stages formed a key part of the landesque capital of these foraging economies. More than 30 years ago, Latz and Griffin (1978) outlined the importance of fire in the ecology of spinifex grasslands: it thins out climax vegetation (such as mature spinifex), unlocks nutrients stored in perennial vegetation, promotes the short-term productivity of herbaceous plants, increases local vegetation diversity and creates a more stable ecosystem, in which small frequent fires reduce fuel loads and the risk of extensive, intense, destructive fires (see also Morton et al. 2011). In Australia, much of the debate about the impact of anthropogenic fires has been dominated by questions about whether the initial entry of people into Australia around 50,000 years ago led to a massive increase in fires that transformed the continent’s vegetation, denuded its soils and led to extinction of some genera of megafauna (Tindale 1957, 1959;Latz1995, 2007; Miller et al. 2005). Recent research, however, suggests that fire has played a more interesting and more critical role during the last millennium in structuring the economic landscapes seen in the classic ethnographies (Bowman 1998; Haydon, Friar and Pianka 2000;Birdetal.2005, 2008; Burrows et al. 2006). Aboriginal groups in Australia’s deserts used fire for a variety of purposes: to flush out small marsupials, to hunt burrowing animals and reptiles, to ‘clean up country’, to encourage new growth to attract kangaroos and bustards, to burn fire breaks to protect religious sites or stands of fire-sensitive trees, to assert ownership of land and for signalling when travelling (Gould 1971b; Kimber 1983;Birdetal.2008). Work over the last decade has used satellite imaging to map anthropogenic fire mosaics in the Western Desert, including landscapes used by contemporary foragers (Bird et al. 2008) and landscapes recently abandoned by foragers (including that by the ‘Pintupi Nine’ in 1984) (Burrows et al. 2006). This work shows that anthropogenic fires are localised and patchy, modifying the structure of the landscape locally but with little effect on the composition of vegetation at the regional level. The characteristic signature of human fires is a mosaic of small patches of vegetation in different successional stages, closely mapping the density and distribution of the hunter- gatherer population, and produced as an epiphenomenon or incidental effect of fires lit for hunting small marsupials and lizards and for signalling (Table 9.4). As Bird et al. point out, ‘Martu burning does not increase the absolute amount of fire, it rescales its temporal and spatial impact’ (2008: 14800). Patch burning had clear economic benefits for foragers in Australia’s deserts, especially in the spinifex grasslands of Central Australia and the Western Desert, where it increased yields from staple plant-food species, medium-sized marsupials and reptiles. Most of these benefits accrue to women’s foraging rather than men’s pursuit of larger mobile game (Bird et al. 2005). Many plant-food species are promoted by fire (Table 9.5), including staple seed-food plants, most of the important Solanum (bush tomato) species, and key bulb or tuber The Last Millennium: Archaeology and the Classic Ethnographies 325 table 9.4. Habitat mosaic created under different fire regimes in the Western Desert, 2002. Areas regularly burnt by Martu foragers have more diverse vegetation and numerous small patches in a range of successional stages

Human ﬁres Wildﬁres

No. of patches 64 22 Patch size (ha.) 58 189 Diversity (SDI)a 1.15 0.63 a Shannon diversity index. Source: Bird et al. (2008) species such as Cyperus bulbosus (onion grass yalka)andVigna lanceolata (pencil yam). The resilience of fire-sensitive seed species, such as Brachychiton gregorii and the various species of acacia, and fruit trees such as Ficus platypoda (rock fig) and Santalum acuminatum (quandong), also depends on reducing the extent of destructive wildfires. These species are destroyed by intense wildfires but are given a measure of protection in a landscape made up of a mosaic of small successional patches. Bird et al. (2008: 14800) argue that efficient exploitation of seed resources may be critically linked to the creation of the small-scale successional mosaics created by anthropogenic burning. Of the sixteen most important seed-plant species listed in ethnobotanical research (Latz 1995), at least seven are most table 9.5. Types of plant food in Central Australia, showing the number of species promoted by fire

No. plant food species No. ﬁre-promoted

Acacia seeds 40 4 Grass seeds 14 8c Other seedsa 25 15c Fruits 30 11d Tubers, bulbs and roots 96e Grubs, lerp and galls 19 4 Honey 11 0 Otherb 12 3 a Other tree seeds, forbs and chenopods. b Edible gums, mushrooms and greens. c Includes several staple seed species: Eragrostis eriopoda, Panicum decompositum, Brachiaria subquadripara, Fimbristylis oxystachya, Por- tulaca oleracea, Dysphania kalpari and Yakirra australiense. d These are all species of Solanum including several staple species. e Includes several staple species: Boerhavia diffusa, Cyprus bulbosus and Vigna lanceolata. Data from Latz (1995: table 3). 326 The Archaeology of Australia’s Deserts

abundant in early successional sandplain habitats; these include annual grasses (Brachiaria subquadripara), perennial grasses (Eragrostis eriopoda, Panicum decompositum and Yakirra australiense), chenopods (Dysphania kalpari), annual sedges (Fimbristylis oxystachya) and succulent herbs (Portulaca oleracea). In contrast, the most important acacia seed species, such as Acacia aneura, are limited to late successional stages. For grass and acacia seeds, therefore, net yields over several consecutive years are likely to be highest in areas with a habitat mosaic that includes a range of successional stages. Similar benefits appear to accrue with respect to small and medium-sized game (Bolton and Latz 1978; Burbidge and McKenzie 1989). A mosaic of successional stages provides both food and cover for small- and medium- sized marsupials, such as hare wallaby, Lagorchestes hirsutus (Bolton and Latz 1978), and reptiles, and new growth attracts larger mobile species, such as kangaroo, emu and bustard. Some work suggests that patch-burning regimes do not directly benefit rat- and mouse-sized mammals but rather increase the resilience of species otherwise affected by widespread destruction of cover during wildfires (Letnic et al. 2004). Perhaps the most important effect here is that small-scale habitat mosaics allow for rapid recolonisation of suitable habitat by small- and medium-sized mammals, providing a more stable game resource than the ‘boom-and-bust’ pattern produced by spatially extensive wildfires. There are few opportunities to trace the development of fine-grained anthropogenic fire mosaics directly, but some associated changes predicted by behavioural ecology can be recognised in the archaeological record. Codding (2012) shows that these fire mosaics are an epiphenomenon of foraging economies based on small marsupials and reptiles. Bird et al. (2008: 14800) argue that proliferation of seed-grinding gear should closely correspond with establishment of anthropogenic fire mosaics. Both of these lines of evidence are explored in the following sections.

FAUNAL ASSEMBLAGES AND FORAGING PATTERNS In Central Australia and the Western Desert, faunal assemblages from occupation units dating within the last millennium typically contain a wide range of small game: rodents; varanid lizards and other reptiles; small marsupials such as bandicoots, bettongs (rat kangaroos) and brush-tailed possums; small- to medium-sized macropods such as hare wallabies (Lagorchestes sp.), rock wallabies (Petrogale sp.) and nailtail wallabies (Onychogalea sp.); large macropods such as the red kangaroo (Macropus rufus) and euro or hill kangaroo (M. robustus); as well as emu eggshell (Gould 1977; Webster 1982;Smith1988, 2009b, 2010; Smith and Ross 2008b;Veth1993; O’Connor et al. 1998). These reﬂect a foraging economy in which the meat component largely focused on small game but was supplemented by large macropods. The Last Millennium: Archaeology and the Classic Ethnographies 327

To date, however, faunal assemblages table 9.6. Index of fragmentation values have not played a major role in reconstruct- (IOF) for faunal bone assemblages from ing changes in prehistoric economy in any archaeological sites in the Western Desert and detail. Most assemblages are too small to Central Australia be analytically useful. The major limitation, Fragmentation of course, is preservation: bone is usually Site (IOF) highly fragmented in desert sites, often with few diagnostic elements remaining, and the Western Desert acidic nature of desert sand means that Puntutjarpa Trench 20.53 bone is generally only well preserved in lev- Serpents Glen 0.06 Kaalpi 0.10 els dating within the last millennium. The Central Australia only sites with significant excavated faunal Intirtekwerle Trench 10.50 assemblages are Puntutjarpa in the Western Puritjarra 0.04 Desert (n = >17,167 pieces; Gould 1977) and Intirtekwerle (James Range East) rock- Data from Gould (1996), Webster (1982), Veth (2005a), Veth et al. (2001), and Smith (2010). shelter in the Central Australian ranges (n = >6,514 pieces; Gould 1978a). Some work uses an index of fragmentation (IOF – total bone weight/total number of fragments) as a proxy for dietary stress, arguing that prehistoric resource stress was particularly marked in the Western Desert (Gould 1980, 1996; O’Connor et al. 1998;Veth2005a). This assumes that extreme fragmentation of archaeological bone directly reflects the pulverising of bone meal for consumption. Since this was first proposed, other work has shown that anthropogenic patterns of fragmentation are obscured by post-discard and post-depositional taphonomic factors, such as scavenging of bone by animals, post-discard burning of bone, post-depositional fragmentation within the deposit, fragmentation during excavation and handling during analysis (Webster 1982; Gould 1996; Walshe 2000). These analyses have also shown that there is no systematic difference in IOF values over time or between sites in the Western Desert and Central Australia (Webster 1982; Gould 1996) (Table 9.6). A more productive line of analysis looks at changes in the relative contribution of large macropods in these foraging economies (Codding 2012). Based on ethnoarchaeological research with contemporary Martu foragers in the Western Desert, Codding et al. (2010) show that variation in the proportion of large macropods to small game in foraging economies is a function of the success in hunting larger mobile prey (where returns are unpredictable) versus the time spent foraging for smaller game (with more predictable returns). Extended use of foraging territories quickly reduces encounter rates for larger macropods, increasing travel time for hunters and leading to greater reliance on women’s foraging for small game and plant foods (see Figure 9.9). Codding (2012) tested this with an analysis of excavated faunal material from Puntut- jarpa, which showed a greater reliance on small marsupials and reptiles during 328 The Archaeology of Australia’s Deserts

Figure 9.9. Changing foraging patterns, comparing (a) the early Holocene and (b) late Holocene. Smaller foraging territories and declining residential mobility reduce encounter rates and hunting returns for large macropods. Human behavioural ecology predicts a widespread shift to small marsupials and reptiles, in association with a fine- grained anthropogenic fire mosaic as a corollary of more extended occupation of sites. (Source: Codding 2012:fig.6.14)

the last millennium (Table 9.7). Similar trends are evident at other desert sites, wherever faunal remains are preserved beyond the last thousand years, including Puritjarra, Intirtekwerle and Serpents Glen (Codding 2012 – see also Webster 1982; O’Connor et al. 1998;Smith2010), and the Nullarbor sites, where faunal assemblages are increasingly underpinned by reptiles and small marsupials (Walshe 1994;Cane1995). At Puntutjarpa, these are not primarily taphonomic trends: Codding (2012: 237–46) was able to show that the shift recorded in Table 9.7 is not explained by sampling bias, differences in fragmentation, selective removal of small bone or differential preservation of the more robust elements of large macropods. Human behavioural ecology posits a widespread shift to small marsupials and reptiles as a corollary of more extended occupation of sites. If better samples and more detailed analyses of faunal assemblage conﬁrm this archaeological trend, there are some interesting implications for foraging technology and for

table 9.7. Changes in hunting patterns shown in faunal assemblages at Puntutjarpa rock shelter, Western Desert

Stratigraphic unit Zone A Zone B Zone C Last millennium Mid-Holocene Early Holocene

Small marsupials and reptilesa 59 47 30 Large macropods 41 53 70

a Excludes murids and introduced rabbits. Data are percentage minimum number of individuals (MNI ) from Codding (2012). The Last Millennium: Archaeology and the Classic Ethnographies 329 table 9.8. Excavated grindstone assemblage from Intirtekwerle (James Range East), Central Australia

Expedient Undiagnostic Layer Age (ka) Seed-grinders grindstones fragments Total (n)a

I 0–0.8 33 8 72 113 II/III 0.8–3.81 1 – 2 a An additional twenty-two grindstones from layer I are listed in the excavation catalogue but cannot be located in the museum collections. Data from Smith (1988). the social relations of production. First, the shift to more logistical foraging associated with innovations in hafted tool kits around 4 ka (described in Chap- ter 6) may represent an attempt to maintain hunting success at a time when the natural abundance of large macropods in sandridge deserts was declining as environmental stochasticity increased. Second, greater reliance on women’s foraging sharpened the gender division of labour – as shown in the ‘dual technology’ of late Holocene tool kits (see Chapter 6) – in which men adopted a high-risk/high-return strategy in hunting large mobile game and women foraged for reliable staples such as small marsupials, lizards and plant foods. This also created the circumstances in which men increasingly sought to control women’s labour as a form of capital, through polygyny and the reciprocal bestowal of daughters as the marital partners of older men (Gale 1978). As Codding puts it,

If hunters are more likely to be successful when encounter rates are higher, and this leads to variability in the gender-division of labor, including effects that resound into social and political arenas, then the distribution and abundance of kangaroo likely inﬂuences not only hunting success and foraging outcomes, but larger aspects of human social and political organisation as well. (2012: 132–4)

SEED-BASED ECONOMIES IN THE LAST MILLENNIUM The increasing proliferation of grindstones and seed-grinding implements in desert assemblages from 1.5 ka (Table 9.8) – as well as ethnographic evidence showing continuing pressure to expand or stabilise the supply of wild grain – shows that exploitation of seed-foods continued to intensify throughout the late Holocene. By the time the classic ethnographies were compiled, it was clear that historic population levels in the arid zone were underpinned by the use of seed-foods. When European explorers ﬁrst entered the desert, they noted that Aboriginal people there relied on wild seeds as food, harvested from a wide variety of plants, including forbs, grasses and acacias. On Cooper Creek, in 1845, Sturt commented that the women were ‘employed beating the seed for cakes, between two stones, and the noise they made was exactly 330 The Archaeology of Australia’s Deserts

like the working of a loom factory’ (Sturt 1849: II, 79). In the Diamantina area, Hodgkinson made a similar observation in 1876, recording ‘an incessant pounding of nardoo goes on, resembling nothing so much as the chipping of a stone-cutter’s yard’ (Hodgkinson 1877: 15). While on the upper reaches of Cooper Creek during his 1858 expedition, AC Gregory famously noted that

the natives reap a Panicum grass. Fields of 1,000 acres are there met with growing this cereal. The natives cut it down by means of stone knives, cutting down the stalk half way, leaving the straw which is often met with in large heaps; they winnow by tossing seed and husk in the air, the wind carrying way the husks. (Gregory 1887: 132)

For most ethnographic groups, the period of greatest stress during the annual cycle was late in the dry season, when the population had been tethered to country within reach of the permanent waters for some time. High-ranking bush foods such as fruit, tubers, bustards and large and medium-sized marsupials were quickly depleted. Seeds were particularly important at this time and were often stockpiled to meet expected shortfalls (Thomson 1964: 402;Cane1984: 76). The pressure on seed resources was evident in foraging economies across the arid zone. For instance, there was extensive use of ngardu sporocarps (Marsilea quadrifolia) in the Cooper basin, and occasional consumption of Triodia seeds in parts of the Western Desert, although both are poor foods and have high handling costs. Other features suggest that there was pressure to expand seed resources during the last few hundred years: for example, the construction of earthen dams to ﬂood stands of seed grasses, patch-burning to promote early to mid-succession stands of grass or to protect stands of ﬁre-sensitive mulga, the storage of seed in pits in the Cooper basin and the introduction of seed species into new areas (W Jones 1979: 140–3;Kimber1984;Smith1986a; Gerritsen 2008). Some of these features warrant further comment.

The Use of Nardoo and Spinifex ‘Nardoo’ (ngardu) is one of best known native ‘grains’, largely because of its association with the iconic 1861 Burke and Wills expedition. The plant, Marsilea quadrifolia, is a clover-leaf fern that grows in shallow water, and its lentil-sized sporocarps are abundant on the flood plains and dried swamps in the arid river systems east of Lake Eyre. The sporocarps are very hard and require quite laborious pounding before being mixed into a thin paste and cooked into cakes. Because the spores contain comparatively little starch within a gelatinous spore wall (Schneider and Pryer 2002), ngardu makes a filling but poor ‘farinaceous’ food. In the last entry in his journal, Wills – one of the expedition members who died of malnutrition and beriberi at Cooper Creek (Earl and McCleary 1994) – wrote that ‘starvation on nardoo is by no means very unpleasant’ (Wills 1863: 302). Actively growing plants The Last Millennium: Archaeology and the Classic Ethnographies 331 also contain high levels of thiaminase, an enzyme that destroys vitamin B1 and promotes beriberi. Major seed-foods – such as Portulaca oleracea, Panicum decompositum and acacia seeds – are prolific in the Cooper basin after rain or flooding. In this context, the use of such a high-cost, low-return food as ngardu suggests a subsistence system with very strong seasonal pressures on resources, and it is significant that ngardu was only used extensively in the Cooper basin, being almost entirely ignored in the Murray–Darling basin and in Central Australia. Another example of seeds with high handling costs and poor returns is from the Western Desert, where there are occasional references to the consumption of Triodia (spinifex) seeds (Clement 1904;Cane1987: table 1). Some archaeologists have taken this to mean that the hummock grasslands were a major seed resource in their own right. This is unlikely. Triodia is known to be a poor seeder: it produces abundant florets but very little seed, except after exceptionally good rain, a time when higher-ranked plant foods are usually widely available – including other seed grasses. The only circumstances in which the necessity to use spinifex seeds would coincide with the availability of this seed is where good rainfall allowed people to move into a large area of mature spinifex that had not been burnt for some time.

Reaping Knives and Hayricks Gregory’s (1887) reference has been very widely quoted but, in fact, it is the sole reference to the use of stone reaping knives in the Australian literature. All of the major grass seeds – including Panicum – can be harvested by hand- stripping the heads off the plants into dishes. The only circumstance in which it might be cost effective to cut the grass – and pile it into ‘hayricks’ – is where there is pressure to accelerate ripening of the seed and to synchronise seed fall. In this context, Gregory’s account provides another example of periodic pressure on seed resources in the Cooper basin.

Seed Storage Various methods of storing or stockpiling grain were used – pits, wooden dishes on platforms, skin bags, parcels covered with mud and grass – but, in most cases, this was on a small scale and involved short-term storage. Howitt’s 1862 observation is typical of many ﬁrsthand accounts: in an Aboriginal camp on Cooper Creek, he found a mud-and-grass parcel containing about 30– 50 kg of portulaca (‘one and a half bushels’; quoted in Allen 2011: 264). The largest recorded cache of seeds was described by Ashwin (1932), a member of a droving party supplying Overland Telegraph Line construction teams in 1871. At Newcastle Waters, on the northern margin of the desert, Ashwin noted a store of about 1 t of grain (possibly Australian wild rice Oryza australiense)in wooden dishes. Given that he recorded groups of 200 men assembled for a 332 The Archaeology of Australia’s Deserts

ceremony in this area, even this large store of grain represents only 10–14 days’ supply. The main problems in scaling up the storage of grain appear to have been logistical (to be effective, large granaries would need to be co-located with major water storage) and technical (the moisture content of native grain is too high for long-term storage of large quantities of seed, unless parching is employed – W Jones 1979). None of the ethnographic accounts describes ‘granaries’ or seasonal and massive storage of staples. However, the repeated references to use of seed-storage pits (paua kudu ‘a seed-grain pit’) (Reuther 1981 [1914]: II 329–982(4), II 155–903(3), III 119–1700(11), IV 60–3013(3), VII 278–1841) represent another example of the distinctiveness of the Cooper basin: this region had a large Aboriginal population, episodic hyperproduction during ﬂood cycles, with marked subsistence stress between resource ‘booms’. By the late nineteenth century, these stresses were clearly evident: although this was one of the most productive parts of the desert, these groups had an extensive famine vocabulary (Table 9.9).

Proliferation of Seed-Grinders in Archaeological Assemblages Although specialised seed-grinding implements appear in desert assemblages around 3.8–3.2 ka (see Chapter 6), their main phase of proliferation there occurred within the last 1,500 years (see Table 9.8). When these implements first appear, they are not common in archaeological assemblages, and seed- grinding may have been a supplement to seasonal shortfalls in other bush foods. However, from 1.5 ka, there is a shift towards more extended use of sites, with a corresponding increase in the use of grindstones (about 30–50 per cent of which are seed-grinding implements). Of the 378 grindstones recovered from radiocarbon-dated late Holocene contexts in Central Australia, 352 (93 per cent) date to less than 1.5 ka (Smith and Ross 2008a). This is also the period when the few dated mortars occur in sites in the Central Australian ranges. Their presence is an archaeological example of pressure to broaden seed use during the last millennium, in this case to include hard seeds – mainly varieties of acacia seeds that need processing in a mortar, prior to wet milling (see Smith 1985 for an account of these mortars). All of this shows that there is both ethnographic and archaeological evidence for pressure on these foraging economies during the last millennium. ‘Economic intensification’ is usually defined as the substitution of capital, labour and skills in a situation in which additional land cannot be brought into production. For hunter-gatherer groups, mobility is the main means of offsetting declining yields. However, the last millennium saw periodic shortfalls in subsistence despite modest climatic amelioration. This suggests that the more closely settled landscape, indicated in the archaeology, constrained mobility to the extent that it became worthwhile to invest in plant foods with high handling and processing costs, and which required specialised gear. The The Last Millennium: Archaeology and the Classic Ethnographies 333 table 9.9. Diyari drought terminology

Diyari term Reuther’s explanation Ms reference pitaru Drought. ‘Constantly to pound’ – a IV/19/2767 reference to use of seeds that require pounding, used because of lack of alternative foods. From pidana (‘to pound’). ngura pitaru A deserted camp, which was previously IV/20/2767 (8) inhabited by people. pitaru kana People who come in from a IV/20/2767 (10) drought-stricken area and settle down as guests. pitaru kana nari For desert dwellers to become IV/20/2767 (11) emaciated, starving, [as] dying, dead. pitaru terkana For a dry season to persist (‘stand’) and IV/20/2767 (15) not recede. pitarala Name by which those people are known IV/20/2768 who have perished during a drought. pitarurina To become an increasing desert [for a IV/20/2769 drought to intensify], for a dry season to get worse. kanta pitarurina For [edible] plants to become scarce. IV/21/2769 (3) pita pitarurina For trees to become more withered. IV/21/2769 (5) pariwilpa pitarurina For the sky to become more and more IV/21/2769 (6) cloudless. ngapa pitarurina For water to become increasingly scarce. IV/21/2769 (7) paua kudu A seed-grain pit. II/329/982(4)

Between 1888 and 1906, Lutheran missionary JG Reuther compiled a thirteen-volume, 5,000-page manuscript, The Diari, detailing the language, beliefs and customs of the Diyari people of the Cooper Creek area. His glosses on the Diyari word for drought (‘pitaru’) provide a vivid word-picture of famine – in a region otherwise known for its richness and high population density. Source: Reuther (1981) proliferation of grindstones in desert assemblages is one consequence of this shift; ﬁne-grained anthropogenic ﬁre mosaics are likely to be another.

EXPANSION OF THE WATI (WESTERN DESERT) LANGUAGE GROUP These economic changes were not the only consequences of the growth of desert populations during the last millennium. The expansion of the Wati (Western Desert) language group is another example of the restructuring of the cultural world of the desert at this time (McConvell 1996;Smith2005). Most Aboriginal people in the Western Desert today speak one or more of a series of closely related languages (e.g., Pitjantjatjara or Pintupi), known 334 The Archaeology of Australia’s Deserts

collectively as Wati. The recent expansion of this language group to cover the entire western half of the Australian arid zone is one of the few cases that unequivocally involved migration of people (unlike the earlier spread of Pama-Nyungan languages – see Chapter 6). This movement created a common language and culture and remarkable genetic homogeneity over a huge area spanning 2,000,000 km2. Linguistic and genetic data suggest that this spread took place at some time early in the last millennium, but the timing is not known with any certainty. Western Desert groups were making incursive movements east and southeast during the late nineteenth century. Major movements of Pintupi, Pitjantjatjara and Yankunytjatjara people took place in the first few decades of the twenti- eth century as people migrated eastwards, partly in response to severe drought conditions in their homelands. These groups were the most fluid of desert societies, lacking corporate descent groups and with a land tenure system focused more on individual connections than were other desert societies (Sutton 2003: 140). Hamilton (1980) argued that this was a society still in transition following recent migration into the area. A range of traits – linguistic, ceremonial, mythic and genetic – shows a marked cleavage between Western Desert groups and the Arandic speakers of Central Australia and points to a historical discontinuity between these populations (Sutton 2003: 96–7)(Figure 9.10). Genetic data emphasise the distinctiveness of Western Desert groups (Bala- krishnan, Sanghvi and Kirk 1975;Birdsell1993;White1997). This is particularly clear in serological data (in which the highest frequencies of blood group A1 are in the Western Desert) and in the phenotype for tawny hair (which is characteristic of this population) (Birdsell 1993: figs. B-1 and D-4). The genetic evidence indicates a rapid expansion of a small founding population drawn from existing desert groups. At the southeastern margin of their spread, Birdsell (1993) estimated there could have been no more than 500–1,200 years of genetic exchange between Western Desert and Arandic populations. McConvell (1996) outlined the linguistic prehistory of the region. He suggested that the break-up of Northern Nyungic languages around 3 ka involved movement of groups into the Tanami Desert, the Pilbara, the Gascoyne– Murchison region and the western edge of the Western Desert. Over the next 2,000 years, Warnman and Wati languages differentiated, and speakers of the latter moved south and east to occupy the desert, reaching the western edge of the Central Australian ranges by about 1 ka. This chronology is necessarily speculative: historical linguistics provides only a relative sequence of language differentiation, with poorly constrained estimates of the rate of change. But, by 1870, Western Desert groups occupied the Mann–Musgrave Ranges in central Australia, as well as the western and southwestern margins of the main Central Australian range complex. The implication is that some well-occupied archaeological landscapes in western Central Australia saw a recent replacement of population. At present, we cannot detect this in excavated sequences or in The Last Millennium: Archaeology and the Classic Ethnographies 335

Figure 9.10. Geographic distribution of the Western Desert language (Wati ), shown as percentage of words shared with Pintupi (PIN) (using a 100-word lexicostatistical test list). Abrupt boundaries to the south and east, and gradual boundaries to the north and west, suggest that Wati speakers have migrated to the southeast, corresponding with genetic patterns. (After O’Grady 1996)

rock art with any confidence, but more detailed analysis of changes around 0.5 ka might change this. Linguistic and genetic data suggest the following scenario. First, increasing environmental stochasticity, associated with the shift from 4 ka to an ENSO- dominated climate, disrupted settlement in such a way as to accelerate linguistic divergence among existing groups and create a genetic bottleneck. When climatic variability declined after 1.5–1.2 ka, Western Desert groups were able to expand rapidly and occupy new territory. During historic droughts, these groups were observed to fission and disperse into small groups, with the bulk of the population falling back on better-watered country. It is not difficult, therefore, to see this process acting as a sort of ‘cultural pump’, drawing people into the desert during good seasons and forcing them out towards the margins during drought periods. This process requires demographic growth. It is relevant that the best archaeological evidence for population growth – on both sides of the Western Desert – is dated 1.5–1.0 ka (Smith 1988;Veth1993; Smith and Ross 2008a). 336 The Archaeology of Australia’s Deserts

THE SPREAD OF SUBSECTION SYSTEMS Another example of changes associated with growing populations during the last millennium is the spread of subsections. Aboriginal groups in the northern half of the arid zone have a distinctive segmentary social system based on subsections – colloquially known as ‘skins’. This is a classiﬁcatory kin system that divides people into eight divisions; these govern behaviour according to quasi- kin relations between the subsections (e.g., potential ‘spouse’ or ‘classiﬁcatory father’). This system provides a framework for governing social interaction between people who are not direct kin, especially during initial contact with strangers. Yengoyan (1968) showed that population size is a critical factor in the operation of subsection systems: communities of less than 500 people contain too few eligible spouses to sustain the preferred marriage rules. Sub- sections, therefore, are a way of integrating larger numbers of people in these hunter-gatherer societies and cannot operate effectively in smaller dispersed communities. In the 1890s, subsections were actively spreading south across Central Australia (Spencer and Gillen 1899). Linguistic comparison of the subsection terms used by different groups reveals that the system originated in northern Australia and spread south into the desert and north into Arnhem Land within the last 500 years (McConvell 1985, 1996). This provides another example of the dynamic social effects of population growth in the recent past.

OVERVIEW: ARCHAEOLOGY AND THE CLASSIC ETHNOGRAPHIES Both the foraging landscapes Giles travelled through in 1872 (Figure 9.11)and the elaborate ritual and ceremonial life recorded a generation later by Spencer and Gillen, in 1896, appear to be products of historical changes within the last millennium. Some elements of the desert cultural system were in place in the late Pleistocene – as the various lines of evidence discussed in this book show – but ethnographic societies were largely the product of a rolling series of transformations in technology, foraging economy and political economy in the recent past.

Foraging Economies during the Last Millennium During the last millennium, foraging economies in Central Australia and the Western Desert appear to have undergone a ‘broad spectrum’ transition (Bird et al. 2008). Increasing population density was associated with lower residential mobility and extended site use, and a shift towards greater logistical foraging. This was a landscape where anthropogenic ﬁres had created a ﬁne-grained successional mosaic of vegetation. The subsistence pattern increasingly focused The Last Millennium: Archaeology and the Classic Ethnographies 337

Figure 9.11. The late Syd Coulthard (senior traditional owner) at Glen Thirsty, August 2003, showing the paintings described by Giles in October 1872 as ‘Roman numerals’. These are more typical of totemic designs used in body paintings. (Photograph courtesy of MA Smith) on small game, plant foods and resources that need elaborate processing, such as grass and acacia seeds. There is scope for more research on these matters, but current evidence suggests that a dynamic interplay of factors underlies archaeological trends during the last millennium. A modest amelioration of climate and environmental stochasticity around 1.5–1.0 ka provided the basis for more extended use of key sites and closer settlement overall. The reduction in residential mobility made it cost-effective to shift to more labour-intensive processing of plant foods, especially the winnowing, parching and grinding of grass and acacia seeds. It also led to creation of a small-scale successional mosaic as an epiphenomenon of closer settlement, as foragers hunting burrowing marsupials and reptiles regularly burnt the same areas around focal points in the landscape. In turn, this mosaic represented a form of landesque capital (i.e., improvements to a productive estate) generating higher and more stable net yields of plant foods and small game. ‘Aboriginal foragers,’ said Bird et al., ‘thus construct their own ecosystem’ (2008: 14799). Although the evidence suggests a productive mosaic would be established within a single generation, this is not simply a consequence of closer settlement but probably also a necessary corollary of more intensive economic use of these landscapes. The feedback this entails ampliﬁed the effects of modest climatic amelioration on hunter- gatherer settlement at the beginning of the last millennium, explaining the 338 The Archaeology of Australia’s Deserts

disproportionate response evident in regional archaeological records. It also ampliﬁed the stress in the system – when increasingly large regional populations faced intermittent and protracted drought – as shown in the mortality rates reported for historic droughts. These trends parallel broad-spectrum transitions among hunter-gatherer groups in other parts of the world in the late Pleistocene or early Holocene, but they occur much later in Australia (Edwards and O’Connell 1995). The low productivity of these regions, combined with high interannual and decadal environmental variability in Australia’s deserts, placed persistent constraints on population growth in this region and periodically destabilised hunter-gatherer settlement in the interior (Davidson 1990; Smith et al. 2008).

The Social History of Desert Societies What can we say about the social history of desert societies? Archaeologists working in Central Australia inevitably operate in the shadow of Spencer and Gillen, who were explicitly interested in the evolution of social institutions and religious life among these societies. The changes in economic landscapes at the beginning of the last millennium suggest that much of the social apparatus of these societies – their social segmentation, their political economy and the complexity of the ritual-mythological landscape – either took shape within the last millennium or altered their form within the new economic mode. This includes many of the traits considered to be most characteristic of Australian hunter-gatherer groups. However, to properly review the social history of desert groups, we need to range over the earlier history of these societies.

demographic trends The desert during the last millennium was a much larger social world than before. The late Pleistocene appears to have been characterised by isolated small groups of closely related people attached to particular locales. Regional populations seem to have been significantly lower than their ethnographic counterparts (possibly <1 person per 80–200 km2); territories appear to have been significantly larger, and foraging and residential mobility may have been extremely high. It seems likely (but hard to test) that the corollary of lower population density was an open inclusive social system, a high degree of personal autonomy and ceremonial gatherings that were small in scale and more concerned with social integration than ritual property. A smaller, unevenly distributed population also means that not all nodes in a social network can interact equally, so these societies would have been marked by a degree of localisation. Whatever form of a desert society was established prior to 30 ka, late Pleistocene groups would have faced protracted adjustments to an increasingly sparsely occupied social world, as last glacial maximum aridity intensified and The Last Millennium: Archaeology and the Classic Ethnographies 339 fragmented foraging territories. In the desert interior, societies must have been marked by extreme social fluidity between 24 and 15 ka. During the last glacial maximum, societies probably differed from the founding groups (45–35 ka) as much as they differed from ethnographic desert groups. Regional populations grew during the early to mid-Holocene as large- amplitude climatic variability decreased, creating a more reliable resource base. The flow-on effects on patterns of residence and mobility, and on networks of interaction, transformed these societies. Increasing population density in the early to mid-Holocene coincides with the appearance of rock art because living space was increasingly embedded in a process of negotiation of territory and identity. Further growth of regional populations in the last millennium created a more closely settled landscape and expansion into more marginal habitats. Lower residential mobility reduced birth interval and postpartum infant mortality and so may have amplified the effects of moderate climatic amelioration. Irrespective of the actual trigger, archaeological evidence suggests that much of the desert appears to have gone through a period of rapid demographic growth during the last millennium. long-term changes in foraging economy and technology Late Pleistocene societies in the desert appear to have been based on generalised small-game foraging economies: there is no evidence for targeted exploitation of particular species, either on the coast or in the interior. Foraging ranges are likely to have been large, but there is no direct evidence for long-distance transhumance. The role of large macropods appears to have declined as foraging territories became smaller in the late Holocene. Archaeological evidence suggests a switch to greater reliance on grain and other plant foods, small mammals and reptiles, coupled with more use of fire to manage these landscapes, more logistical mobility and small-scale storage or stockpiling of food to cover seasonal shortfalls. In Central Australia and the Western Desert, foraging economies developed their ethnographic structure after 1 ka, but on the arid west coast it is possible that similar shifts occurred earlier (around 4 ka). The most distinctive items of the ethnographic tool kit – seed-grinders, and hafted composite tools such as adzes and composite spears – first appear around 4–3 ka and relate to economic and technological restructuring in the late Holocene. These represent a nexus of technological additions that, along with dingo as hunting dogs, gave desert people greater economic capacity than ever before. shifts in social relations Social segmentation by sections and subsections is clearly recent in these societies because both were observed to be spreading south into the desert in the late nineteenth century. But some other aspects can be given a deeper historical context. The technological changes at 4–3 ka indicate a sharper gender 340 The Archaeology of Australia’s Deserts

division in material culture, and the economic shifts around 1 ka may have further accentuated the sexual division of labour in these foraging economies. Some of the dialectic between ideology and practice observed in ethnographic societies probably originates in the economic changes taking place at the beginning of the last millennium: men sought to control women’s labour as a form of capital; women’s ownership of large millstones and the cooperative nature of women’s work groups imposed a degree of matri-local residence on men. The limited ecological productivity of Australia’s deserts, and the levels of personal autonomy that these foraging economies allowed, imposed strict constraints on the extent of social hierarchy, mostly limiting this to differentiation by age and gender.

changing regional networks Desert societies were connected not only to their environment but also to each other. The nature of these linkages appears to have changed over time. Growing population saw an acceleration of social interaction and exchange across the desert (the ‘ritual engines’ of Gibbs and Veth 2002). By the mid-Holocene, the Pilbara and Central Australia represented separate systems of interaction. And, by the last millennium, the arid riverine area east of Lake Eyre had begun to diverge from the rest of the desert with the development of distinctive long- distance exchange systems for red ochre, pituri and millstones, and evidence for signiﬁcant pressure to expand production of exotic goods. Finally, although site records like those from Puritjarra provide evidence of long continuity in occupation, historical linguistics also reminds us that desert people spoke fundamentally different languages prior to the spread of Pama-Nyungan (PN) languages in the early Holocene and the spread of the Wati-Western Desert language around 1.5 ka: the implications for ideology, graphic systems and social interaction have barely been explored.

rituality and religion The ‘dreaming’ encodes proprietary relationships to land. For ethnographic groups, it provided both a mechanism for maintaining the ecological spacing of deserts groups, as well as a means of reinforcing a system of land tenure in the face of environmental variability and high residential mobility. As Myers explains, ‘holding country’ primarily consists ‘in control over the stories, objects, and rituals associated with the mythological ancestors of the Dreaming at a particular place’ (1993: 45). If the totemic landscape provides a framework for the negotiation of rights in country, rock art suggests that this system began to take shape in the early to mid-Holocene with the first appearance of Panaramitee-style rock engravings using the distinctive graphic system familiar from ethnographic art. Totemism and a belief in ancestral figures that shape the land is widespread across Aboriginal Australia – and so is probably much earlier than the mid-Holocene – but the inscription of landscapes The Last Millennium: Archaeology and the Classic Ethnographies 341 suggests the development of a more place-based system in the desert in the first half of the Holocene. The differentiation of rock-art sites and radiocarbon dates for sites associated with large ceremonies suggest that geographic totemism assumed a more corporate character during the last millennium. The complex symbolic landscape, the extent and character of ceremonial apparatus, and the link between ritual property and corporate rights in land all appear to be developments of the last millennium. The limited field evidence suggests that these were structural adjustments to shifts in economy and demography rather than autonomous drivers of change.

EPILOGUE This book began with Spencer and Gillen and the 1896 ingkwere ceremony. This was a watershed event in anthropological literature, a profound intellectual exchange between elite members of two very different societies. The chapters in this book attempt to historically situate the desert groups recorded in the classic ethnographies. In a sense, I have tried to approach the 1896 ingkwere from ‘the other side’, reconstructing the long history that shaped the world of the Arrernte elders sitting across the ceremonial ground from Oxford biologist Baldwin Spencer and Alice Springs postmaster Frank Gillen. We can now see that the classic ethnographic societies encountered by anthropologists on the desert frontier were the historical product of a long period of autonomous cultural development, shuffled and reassembled through a range of filters as aridity waxed and waned throughout the Quaternary. On the one hand, this process amplified and elaborated the social and economic traits of an ancient hunter-gather culture. On the other hand, archaeology shows that the classic ethnographies portray societies that fundamentally took shape in the centuries immediately prior to colonial contact. The last millennium is one of the most interesting periods of desert prehistory, a time when archaeological investigations are beginning to give a more nuanced historical picture of changes in the foraging economy, in trade and exchange, in rock art and in the political structure of these groups. As Rhys Jones once reminded us, ‘contemporary hunters are not static relicts from a frozen past; they also have their own past that can be investigated archaeologically’ (1984: 59). As I write this, there is a lively grassroots movement by remote desert communities – from Fregon and Blackstone to Punmu and Port Hedland – to bring their art, traditions and dance to the national stage. I hope my book adds a historical dimension to this celebration of the desert as a cultural landscape.

GLOSSARY OF TECHNICAL TERMS

TIME PERIODS Quaternary A period beginning 2.6 million years ago that includes the Pleistocene and Holocene epochs. The beginning of the Quaternary saw global cooling, glaciation of both poles and the appearance of the genus Homo. Pleistocene Beginning about 2.6 million years ago, the Pleistocene was a time marked by the periodic growth and decline of high-latitude ice sheets and glaciers. Late Pleistocene The late Pleistocene is made up of the last complete glacial–interglacial cycle, beginning about 126 ka with the onset of the last interglacial. LGM The last glacial maximum (26.5–19 ka). Formally deﬁned as the latest period when global glaciers reached their maximum extent. The LGM was the driest of all glacials: forests shrank, deserts and ice sheets expanded, and sea levels fell to their lowest recorded levels, 135 mbelow modern sea level. Holocene The Holocene began at 11.7 ka with the resumption of rapid global warming after the last glacial episode. Although this was a comparatively stable period, it was not uniform and can be divided into three distinct phases: the early Holocene (11.7–8.0 ka), when vegetation and geomorphic systems were still adjusting to rapid postglacial changes in temperature, sea levels and rainfall; the mid-Holocene (8.0–4.0 ka); and the late Holocene (from 4.0 ka), which saw the onset of a drier, more variable climate dominated by the El Nino-Southern˜ Oscillation (ENSO).

343 344 Glossary of Technical Terms

MIS Marine isotope stages. The cornerstone of the Quaternary climate sequence is the deep-sea oxygen isotope record from marine foraminifera, which define a series of marine isotope stages (also called oxygen isotope stages). By convention, cold, dry glacial stages are given even numbers, and warm interglacial stages have odd numbers. In Australian drylands, these climatic oscillations are mainly reflected in cyclical fluctuations in available moisture: interglacials reactivated lakes and rivers across the interior; the colder drier glacial periods are marked by active dunefields and windblown dust. MIS5 The last interglacial is marine isotope stage 5 (or, more precisely, substage 5.5). This represents the last significant imprint of widespread river and lake activity in the arid zone.

DATING METHODS 14C Radiocarbon dating is the major means of dating archaeological sites and organic materials in Australia’s deserts. Mainly applied to charcoal, wood, shell and some minerals such as carbonates, the technique relies on the decay of 14C – a common radioisotope of carbon – to provide a ‘clock’ that counts down. ABOX 14C Radiocarbon dating paired with a pretreatment process (acid–base oxidation) designed to remove otherwise intractable contaminants in charcoal. The technique is used with stepwise combustion at progressively higher temperatures to selectively remove the outer layers of a sample and separately date the resulting CO2. AMS 14C High-precision radiocarbon dating using an accelerator mass spectrometer to directly measure the 14C content of a sample (standard radiocarbon dating relies on radioactive emissions as an indirect measure of 14C content). AMS 14C is widely used where only small samples (as low as 1–2 mg) are available or where the necessity of extensive pretreatment reduces sample size. TL and OSL Luminescence methods are widely used to date sediments in the arid zone. Mainly applied to quartz sediments, the technique measures the charge accumulated in mineral grains after burial (or when sunlight is otherwise excluded). The ‘clock’ is zeroed Glossary of Technical Terms 345

by exposure to light and counts up after burial. Thermoluminescence (TL) uses heat to release the accumulated charge (emitted as photons), whereas optically stimulated luminescence (OSL) uses a laser. OSL is the technique of choice for high-precision work, for single-grain dating or for samples where mixed-aged sediments are suspected or where bleaching (i.e., zeroing) may have been incomplete (as in some fluvial or cave sediments). ESR Electron spin resonance is an alternative approach to measuring the dosage of nuclear radiation received by an archaeological sample. ESR is mainly used to date tooth enamel and speleothems, materials that are unsuitable for luminescence techniques. For these materials, ‘time-zero’ is formation of the crystals in these materials: hydroxyapatite in tooth enamel; calcite in speleothems. Cation-ratio dating A technique formerly used for dating mineral varnishes on rock surfaces but now shown to be unreliable. The method assumes that the leaching of mobile elements (Ca and K) in these varnishes is time-dependent. AAR Amino acid racemisation. In eggshell and marine shell, the progressive alteration of amino acids is a function of time and temperature. Provided one of these can be controlled, AAR can provide either an indication of age or a record of palaeotemperature. In Australia’s deserts, AAR dating has been applied to eggshell from emu and Genyornis, often calibrated against 14Cdating of the eggshell, or TL/OSL ages on associated sediments. U/Th and U-series One branch of uranium-series dating looks at the radioactive decay chain 238U–234U–230Th. U/Th dating specifically looks at the build-up of thorium-230 in a sample, assuming that the original sample was free of detrital thorium and there has been no subsequent uptake or leaching of uranium. In the arid zone, this dating method has been applied to tooth enamel or calcite in speleothems (cave flowstone and stalagmites) and less often to bone, carbonates or rock varnish. TIMS U/Th The use of thermal ionisation mass spectrometry (TIMS) has broadened the archaeological application of U/Th dating by allowing very small samples to be measured. 346 Glossary of Technical Terms

TIMESCALES BP Radiocarbon ages are given as years ‘before present’. By international convention, this is ‘years before 1950 AD’, the point at which burning of fossil fuels diluted the concentration of radiocarbon in the atmosphere. cal. BP A calibrated radiocarbon age. Even in the deep past, ﬂuxes in atmospheric 14C have distorted the radiocarbon timescale, so that radiocarbon years are not calendar ages. Calibration programs such as OxCal, CALIB or CALPAL are used to convert radiocarbon ages (years BP) to sidereal years (years cal. BP or ka). ka Throughout this book, calendar or sidereal ages are quoted as ‘ka’ in the sense of ‘thousands of years ago’.

OTHER C3 and C4 plants Different photosynthetic pathways in plants result in minor fractionation of carbon isotopes, giving them different isotopic signatures. C4 grasses are more resistant to water stress and are characteristic of hot climates or summer rainfall regions. C3 plants are mainly shrubs or temperate zone grasses. Summed probability Summed probability plots of radiocarbon ages plots provide a more sophisticated picture of archaeological trends than do simple frequency plots. Because radioactive decay is a random process, there are statistical counting uncertainties associated with each radiocarbon date. This is reﬂected in the way radiocarbon ages are reported: as mean age ± 1 standard deviation. For calibrated ages, the irregularity of the calibration curve introduces other uncertainties: a single radiocarbon date might have multiple age intercepts on the calibration curve. In effect, each calibrated radiocarbon age has a spread of probabilities. For a series of ages, summed probability plots combine these probabilities into a single plot. Ready access to radiocarbon calibration programs has made these plots the mainstay of time-series analysis of archaeological trends. ENSO The El Nino-Southern˜ Oscillation (ENSO) is a major driver of interannual and decadal climatic variability on both sides of the Paciﬁc. Differences in sea surface Glossary of Technical Terms 347

temperature (SST) determine the strength of airflow across the tropical South Pacific, contributing to droughts and floods in Australia. This pattern of atmospheric thermal circulation is inherently unstable, and extremes tend to recur on a 3-to8-year cycle.

REFERENCES

Adams JM, Faure H, Faure-Denard L, McGlade Allen H (1997) The distribution of large blades: JM and Woodward FI (1990) Increases in ter- Evidence for recent changes in Aborigi- restrial carbon storage from the last glacial nal ceremonial exchange networks. In P maximum to the present. Nature 348: 711–14. McConvell and N Evans (eds) Archaeology and Adcock GJ, Dennis ES, Easteal S, Huttley GA, Linguistics: Aboriginal Australia in Global Per- Jermiin LS, Peacock WJ and Thorne A (2001) spective,pp.357–76. Melbourne: Oxford Uni- Mitochondrial DNA sequences in ancient versity Press. Australians: Implications for modern human Allen H (2011) The space between: Aboriginal origins. Proceedings of the National Academy of people, the Victorian Exploring Expedition Science 98: 537–42. and the relief parties. In EB Joyce and DA Aiston G (1937) The Aboriginal narcotic pit- McCann (eds) Burke and Wills: The scientific cheri. Oceania 7: 372–4. legacy of the Victorian Exploring Expedition,pp. Akerman K (1973a) Aboriginal baler shell objects 245–74. Melbourne: CSIRO. in Western Australia. Mankind 9: 124–5. Allen J, Gosden C and White JP (1989) Human Akerman K (1973b) Further evidence of the Pleistocene adaptations in the tropical island manufacture and use of ground-edge axes in Pacific: Recent evidence from New Ireland, southwestern Australia. The Western Australian a greater Australian outlier. Antiquity 63: 548– Naturalist 21: 107–11. 61. Akerman K (1978) Notes on the Kimberley Alley NF (1998) Cainozoic stratigraphy, palaeo- stone-tipped spear focusing on the point haft- environments and geological evolution of the ing mechanism. Mankind 11: 486–9. Lake Eyre Basin. Palaeogeography, Palaeoclima- Akerman K and Stanton J (1994) Riji and Jakuli: tology, Palaeoecology 144: 239–63. Kimberley pearl shell in Aboriginal Australia. Alloway BV, Lowe DJ, Barrell DJA, Newnham Monograph Series No. 4. Darwin: Northern RM, Almond PC, Augustinus PC et al. and Territory Museum of Arts and Sciences. NZ-INTIMATE members (2007)Towardsa Allen H (1972) Where the crow flies backwards: Man climate event stratigraphy for New Zealand and land in the Darling Basin. Unpublished PhD over the past 30,000 years (NZ-INTIMATE thesis, Australian National University. project). Journal of Quaternary Science 22: 9–35. Allen H (1974) The Bagundji of the Darling Amundson R, Wang Y, Chadwick O, Trumbore Basin: Cereal gatherers in an uncertain envi- S, McFadden L, McDonald E et al. (1994) ronment. World Archaeology 5: 309–22. Factors and processes governing the 14C con- Allen H (1990) Environmental history in south- tent of carbonate in desert soils. Earth and Plan- western New South Wales during the late etary Science Letters 125: 385–405. Pleistocene. In C Gamble and O Soffer (eds) Archer M (1974) New information about The World at 18,000 BP.Volume 2: Low latitudes, the Quaternary distribution of the thy- pp. 298–321. London: Unwin Hyman. lacine (Marsupialia, Thylacinidae) in Australia.

349 350 References

Journal of the Royal Society of Western Australia Balme J (1995) 30,000 years of fishery in western 57: 43–50. New South Wales. Archaeology in Oceania 30: Archer M and Hand SJ (2006) The Aus- 1–21. tralian marsupial radiation. In JR Merrick, M Balme J (2000) Excavations revealing 40,000 Archer, GM Hickey and MSY Lee (eds) Evo- years of occupation at Mimbi Caves, south lution and biogeography of Australasian vertebrates, central Kimberley, Western Australia. Aus- pp. 575–646. Oatlands: Australian Scientific tralian Archaeology 51: 1–5. Publishing. Balme J, Davidson I, McDonald J, Stern N and Ash JE and Wasson RJ (1983) Vegetation and Veth P (2009) Symbolic behaviour and the sand mobility in the Australian desert dune- peopling of the southern arc route to Australia. field. Zeitschrift fur¨ Geomorphologie N.F. 45: 7– Quaternary International 202: 59–68. 25. Balme J, Garbin G and Gould RA (2001) Ashwin AC (1932) From South Australia to Port Residue analysis and palaeodiet in arid Aus- Darwin with sheep and horses in 1870–71. tralia. Australian Archaeology 53: 1–6. Royal Geographical Society of Australasia, (South Balme J and Hope J (1990) Radiocarbon dates Australian Branch) 32: 47–93. from midden sites in the lower Darling River Attenbrow VJ (2002) Sydney’s Aboriginal past: area of western New South Wales. Archaeology Investigating the archaeological and historical in Oceania 25: 85–101. records. Sydney: UNSW Press. Balme J and Morse K (2006) Shell beads Attenbrow V, Robertson G and Hiscock P and social behaviour in Pleistocene Australia. (2009) The changing abundance of backed Antiquity 80: 799–811. artefacts in south-eastern Australia: A response Bancroft J (1879) Pituri and tobacco. Transactions to Holocene climate change? Journal of Archae- of the Philosophical Society of Queensland 3: 1–16. ological Science 36: 2765–70. Barbetti M and Allen H (1972) Prehistoric man Aubert M, Richer-LaFleche` M, Watchman A, at Lake Mungo, Australia, by 32,000 years BP. McCulloch M and Mortimer G (2009) Investi- Nature 240: 46–8. gating rock varnish formation with sequential extrac- Bardon G (1979) Aboriginal art of the Western tion. Unpublished paper, Australian Archae- Desert. Adelaide: Rigby. ological Association Annual Conference, 11– Barker G and Gilbertson D (eds) (2000) The 14 December 2009, Flinders University, South archaeology of drylands: Living at the margin. Australia. London: Routledge. Australian Bureau of Meteorology (2008) Cli- BarrowsTTandJugginsS(2005) Sea-surface mate of Australia. Sydney: Commonwealth of temperatures around the Australian margin Australia. and Indian Ocean during the Last Glacial Australian Surveying and Land Information Maximum. Quaternary Science Reviews 20: Group (1990) Atlas of Australian Resources – 1017–47. Vegetation. Third series, Vol. 6. Canberra. Barrows TT, Stone JO and Fifield LK (2004) Baas-Becking LG and Kaplan IR (1956) The Exposure ages for Pleistocene periglacial microbiological origin of the sulphur nodules deposits in Australia. Quaternary Science of Lake Eyre. Transactions of the Royal Society of Reviews 23: 697–708. South Australia 79: 52–65. Barrows TT, Stone JO, Fifield LK and Cresswell Balakrishnan V, Sanghvi LD and Kirk RL (1975) RG (2002) The timing of the last glacial max- Genetic diversity among Australian Aborigines. imum in Australia. Quaternary Science Reviews Canberra: Australian Institute of Aboriginal 21:159–73. Studies. Barton HJ (2001) Mobilising lithic studies: An appli- Balme J (1983) Prehistoric fishing in the lower cation of evolutionary ecology to understanding pre- Darling, western New South Wales. In C historic patterns of human behaviour in the Simp- Grigson and J Clutton-Brock (eds) Animals son Desert, far western Queensland. Unpublished and archaeology: 2. Shell middens, fishes and birds, PhD thesis, University of Sydney. pp. 19–32. Oxford: BAR International Series BartonH(2003) The thin film of human action: 183. Interpretations of arid zone archaeology. Aus- Balme J (1991) The antiquity of grinding stones tralian Archaeology 57: 32–41. in semi-arid western New South Wales. Aus- Barton RNE, Bouzouggar A, Collcutt SN, tralian Archaeology 32: 3–9. Schwenninger J-L and Clark-Balzand L (2009) References 351

OSL dating of the Aterian levels at Dar es- Berndt RM (1959) The concept of “the tribe” Soltan I (Rabat, Morocco) and implications in the Western Desert of Australia. Oceania 30: for the dispersal of modern Homo sapiens. Qua- 81–107. ternary Science Reviews 28: 1914–31. Berndt, RM (1972) The Walmadjeri and Basedow H (1907) Felsgravierungen hohen Gugadja. In MG Bicchieri (ed) Hunters and Alters in Zentral-Australien. [Ancient rock Gatherers Today,pp.177–216.NewYork: engravings in central Australia] Zeitschrift fur¨ Holt, Rinehart and Winston. Ethnologie 1907: 707–17. Berndt RM (1974) Australian Aboriginal religion. Basedow H (1914) Aboriginal rock carvings of Iconography of Religions series. Leiden: EJ great antiquity in South Australia. The Jour- Brill. nal of the Royal Anthropological Institute of Great Berndt RM (1987) Panaramittee magic. Records Britain and Ireland 44: 195–211. of the South Australian Museum 20: 15–28. Bassinot FC, Labeyrie LD, Vincent E, Berry SL (1991) The potential of fossil mammal Quidelleur X, Shackleton NJ and Lancelot middens as indicators of vegetation history in Y(1994) The astronomical theory of climate Central Australia. Australian Journal of Botany and the age of the Brunhes-Matuyama mag- 39: 305–13. netic reversal. Earth and Planetary Science Letters Bertler NAN, Mayewski PA and Carter L (2011) 126: 91–108. Cold conditions in Antarctica during the Lit- Baynes A (1984) Native mammal remains from tle Ice Age: Implications for abrupt climate Wilgie Mia Aboriginal ochre mine: Evidence change mechanisms. Earth and Planetary Sci- on the pre-European fauna of the western arid ence Letters 308: 41–51. zone. Records of the Western Australian Museum Bindon PR (1986) Old campsites, new tenants: 11: 297–310. A study of recent and prehistoric fireplaces near Beaton JM (1991) Colonizing continents: Some Wiluna, Western Australia. Unpublished MA problems from Australia and the Americas. In thesis, University of Western Australia. TD Dillehay and DJ Meltzer (eds) The First Binford LR (1980) Willow smoke and dogs’ Americans: Search and Research,pp.209–30. tails: Hunter-gatherer settlement systems and Boca Raton: CRC. archaeological site formation. American Antiq- Beck W, Donahue DJ, Jull AJT and Burr G uity 45: 4–20. (1998) Ambiguities in direct dating of rock Binford LR (1982) The archaeology of place. surfaces using radiocarbon measurements. Sci- Journal of Anthropological Archaeology 1: 5–31. ence 280: 2132–5. Binford LR (1986) An Alyawara day: Making Bednarik RG (2002) First dating of Pilbara men’s knives and beyond. American Antiquity petroglyphs. Records of the Western Australian 51: 547–62. Museum 20: 415–29. Binford LR (2001) Constructing frames of refer- Bednarik RG (2010) Australian rock art of ence: An analytical method for archaeological the- the Pleistocene. Rock Art Research 27: 95– ory building using hunter-gatherer and environmen- 120. tal data sets. Berkeley: University of California Belperio AP and Fotheringham DG (1990)Geo- Press. logical setting of two Quaternary footprint Binford LR and O’Connell JF (1984)An sites, western South Australia. Australian Jour- Alyawara day: The stone quarry. Journal of nal of Earth Sciences 37: 37–42. Anthropological Research 40: 406–32. Bender B (1981) Gatherer-hunter intensifica- Binns RA and McBryde I (1972) A petrologi- tion. In A Sheridan and GN Bailey (eds) Eco- cal analysis of ground-edge artefacts from north- nomic Archaeology: Towards an Integration of eco- ern New South Wales. Prehistory and Material logical and social approaches,pp.149–57.British Culture Series 10. Canberra: Australian Insti- Archaeological Reports, International Series tute of Aboriginal Studies. 96. Oxford: Archaeopress. Bird DW, Bird R Bliege and Parker CH Bennett KD and Provan J (2008)Whatdowe (2005) Aboriginal burning regimes and hunt- mean by ‘refugia’? Quaternary Science Reviews ing strategies in Australia’s Western Desert. 27: 2449–55. Human Ecology 33: 443–64. Berndt RM (1951) Ceremonial exchange in Bird R Bliege, Bird DW, Codding BF, Parker Western Arnhem Land. Southwestern Journal CH and Jones JH (2008) The ‘fire stick of Anthropology 7:156–76. farming’ hypothesis: Australian Aboriginal 352 References

foraging strategies, biodiversity, and anthro- Bowdler S (1990a) Peopling Australasia: The pogenic fire mosaics. Proceedings of the National ‘coastal colonisation’ hypothesis reconsidered. Academy of Sciences 105: 14796–801. In P Mellars (ed), The emergence of modern Birdsell JB (1941) The tribrid origin of the Australian humans: An archaeological perspective,pp.327– Aborigines. Unpublished PhD thesis, Harvard 43. Edinburgh: Edinburgh University Press. University. Bowdler S (1990b) The Silver Dollar site, Shark Birdsell JB (1957) Some population problems Bay: An interim report. Australian Aboriginal involving Pleistocene man. Cold Spring Harbor Studies 1990/2: 60–3. Symposium on Quantitative Biology 22: 47–69. Bowdler S (1999) Research at Shark Bay, WA, Birdsell JB (1958) On population structure in and the nature of coastal adaptations in Aus- generalized hunting and collecting popula- tralia. In J Hall and IJ McNiven (eds), Aus- tions. Evolution 12: 189–205. tralian coastal archaeology,pp.79–84. Research Birdsell JB (1967) Preliminary data on the tri- Papers in Archaeology and Natural History hybrid origin of the Australian Aborigines. No. 31. Canberra: ANH Publications, The Archaeology and Physical Anthropology in Ocea- Australian National University. nia 2: 100–55. Bowern C (2010) Historical linguistics in Aus- Birdsell JB (1993) Microevolutionary patterns in tralia: Trees, networks and their implications. Aboriginal Australia: A gradient analysis of clines. Philosophical Transactions of the Royal Society B. Oxford: Oxford University Press. 365: 3845–54. Bolton BL and Latz PK (1978) The West- Bowern C and Koch H (2004) Australian lan- ern Hare-wallaby, Lagorchestes hirsutus (Gould) guages: Classification and the comparative method. (Macropodidae), in the Tanami Desert. Aus- Amsterdam: John Benjamins. tralian Wildlife Research 5: 285–93. Bowler JM (1968) Australian landform example Bonython CW and Fraser AS (eds) (1989) The no. 11: Lunette. The Australian Geographer 10: great filling of Lake Eyre in 1974. Adelaide: Royal 402–4. Geographical Society of Australasia (South Bowler JM (1971) Pleistocene salinities and cli- Australian Branch). matic change. In DJ Mulvaney and J Gol- Bordes F, Dortch C, Thibault C, Raynal JP son (eds) Aboriginal Man and Environment and Bindon P (1983) Walga Rock and Bil- in Australia,pp.47–65. Canberra: Australian libong Spring: Two archaeological sequences National University Press. from the Murchison Basin, Western Australia. Bowler JM (1976) Aridity in Australia: Age, ori- Australian Archaeology 17: 1–26. gins and expression in aeolian landforms and Bourke M (1997) The age of two human occu- sediments. Earth-Science Reviews 12: 179–310. pation sites in the eastern MacDonnell ranges, Bowler JM (1981) Australian salt lakes: A palaeo- Central Australia. Australian Archaeology 45: hydrologic approach. Hydrobiologia 82: 431–44. 55–56. Bowler JM (1982) Aridity in the late Tertiary Bourke MCA (1998) Fluvial geomorphology and and Quaternary of Australia. In WR Barker paleofloods in arid Central Australia. Unpub- and PJM Greenslade (eds) Evolution of the flora lished PhD thesis, Australian National Uni- and fauna of arid Australia,pp.35–45. Aus- versity. tralian Systematic Botany Society. Frewville, Bourke PM and Mulvaney K (2003)Archaeol- SA: Peacock Publications. ogy of the first recorded petroglyphs for the Bowler JM (1998) Willandra Lakes revisited: Darwin region. The Beagle: Records of the Muse- Environmental framework for human occu- ums and Art Galleries of the Northern Territory 19: pation. Archaeology in Oceania 33: 120–55. 1–6. Bowler JM, Duller GAT, Perret N, Prescott JR Bowdery DE (1998) Phytolith analysis applied and Wyrwoll K-H (1998) Hydrologic changes to Pleistocene-Holocene archaeological sites in the in monsoonal climates of the last glacial cycle: Australian arid zone. British Archaeological Stratigraphy and luminescence dating of Lake Reports Series 695. Oxford: Archaeopress. Woods, N.T. Australia. Palaeoclimates 3 (1–3): Bowdler S (1977) The coastal colonisation of 179–207. Australia. In J Allen, J Golson and R Jones Bowler JM, Hope GS, Jennings JN, Singh G (eds), Sunda and Sahul: Prehistoric studies in and Walker D (1976) Late Quaternary cli- Southeast Asia, Melanesia and Australia,pp.205– mates of Australia and New Guinea. Quater- 46. London: Academic Press. nary Research 6: 359–94. References 353

Bowler JM, Johnston H, Olley JM, Prescott JR, Reserve, northwestern Western Australia. Roberts RG, Shawcross W and Spooner NA Australian Archaeology 74: 98–102. (2003) New ages for human occupation and Brokensha P (1975) The Pitjantjatjara and their climatic change at Lake Mungo, Australia. crafts. Sydney: Aboriginal Arts Board. [Second Nature 421: 837–40. printing 1978]. Bowler JM, Jones R, Allen H and Thorne AG Brookfield M (1970) Dune trends and wind (1970) Pleistocene human remains from Aus- regime in Central Australia. Zeitschrift fur¨ Geo- tralia: A living site and human cremation from morphologie 10: 121–38. Lake Mungo, Western New South Wales. Brown HYL (1883) Report of government geol- World Archaeology 2: 39–60. ogist. South Australian Parliamentary Papers 4 Bowler JM and Magee JW (2000) Redating Aus- (146): 1–5. tralia’s oldest human remains: A sceptic’s view. Brown JH, Reichman OJ and Davidson DW Journal of Human Evolution 38: 719–26. (1979) Granivory in desert ecosystems. Annual Bowler JM and Price DM (1998) Lumines- Review of Ecology and Systematics 10: 201–27. cence dates and stratigraphic analyses at Brown P (1981) Artificial cranial deformation: Lake Mungo: Review and new perspectives. A component in the variation in Pleistocene Archaeology in Oceania 33: 156–68. Australian Aboriginal crania. Archaeology in Bowler JM, Thorne AG and Polach HA (1972) Oceania 16: 156–67. Prehistoric man in Australia: Age and signif- Brown P (2000) Australian Pleistocene variation icance of the Mungo skeleton. Nature 240: and the sex of Lake Mungo 3. Journal of Human 48–50. Evolution 38: 743–49. Bowler JM and Thorne AG (1976) Human Brown S (1987) Toward a prehistory of the Hamer- remains from Lake Mungo: Discovery and sley Plateau, northwest Australia. Occasional excavation of Lake Mungo III. In RL Kirk and Papers in Prehistory No. 6. Canberra: Depart- AG Thorne (eds) TheOriginoftheAustralians, ment of Prehistory, The Australian National pp. 127–38. Canberra: Australian Institute of University. Aboriginal Studies. BrownSHandMulvaneyKJ(1983) Test pit Bowler JM and Wasson RJ (1984) Glacial age excavations at Aboriginal sites P4623, P4627, environments of inland Australia. In JC Vogel P5315 and P5316, Perth-Darwin National High- (ed) Late Cainozoic palaeoclimates of the South- way, Newman-Port Hedland section. Unpub- ern Hemisphere,pp.183–208. Rotterdam: AA. lished report for Main Roads Department, Balkema. Perth, Western Australia. Bowler JM, Wyrwoll K-H and Lu Y (2001) Browne WR (1945) An attempted post-Tertiary Variations of the northwest Australian sum- chronology for Australia. Proceedings of the Lin- mer monsoon over the last 300,000 years: nean Society of New South Wales 70: v–xxv. The paleohydrological record of the Gregory BrummA(2010) ‘The falling sky’: Symbolic and (Mulan) Lakes System. Quaternary International cosmological associations of the Mt William 83–85: 63–80. greenstone axe quarry, central Victoria, Aus- Bowler PJ (1992) From ‘savage’ to ‘primitive’: tralia. Cambridge Archaeological Journal 20: 179– Victorian evolutionism and the interpretation 96. of marginalized peoples. Antiquity 66: 721– Burbidge AA and McKenzie NL (1989)Pat- 29. terns in the modern decline of Western Aus- Bowman I (1924) Desert trails of Atacama.New tralia’s vertebrate fauna: Causes and conser- York: American Geographical Society. vation implications. Biological Conservation 50: Bowman DMJS (1998) Tansley Review No. 101. 143–98. The impact of Aboriginal landscape burning Burbidge N (1960) The phytogeography of Aus- on the Australian biota. New Phytologist 140: tralia. Australian Journal of Botany 8: 75–209. 385–410. Burrows ND, Burbidge AA, Fuller PJ and Behn Bradshaw E (1995) Dates from archaeological G(2006) Evidence of altered fire regimes in excavations on the Pilbara coastline and islands the Western Desert region of Australia. Con- of the Dampier Archipelago, Western Aus- servation Science Western Australia 5: 272–84. tralia. Australian Archaeology 41: 37–8. Byrne M (2008) Evidence for multiple refu- Brady LM and Carson A (2012) An archaic gia at different time scales during Pleis- face from the Woodstock Abydos Protected tocene climatic oscillations in southern 354 References

Australia inferred from phylogeography. Qua- of old ages: Essays in honour of Rhys Jones, ternary Science Reviews 27: 2576–85. pp. 141–65. Canberra: Pandanus Books, Aus- Byrne M, Yeates DK, Joseph L, Kearney M, tralian National University. Bowler J, Williams MAJ et al. (2008)Birth Capell A (1937) The structure of Australian lan- of a biome: Insights into the assembly and guages. Oceania 8: 27–61. maintenance of the Australian arid zone biota. Capell A (1940) The classiﬁcation of languages Molecular Ecology 17: 4398–417. in north and north-west Australia. Oceania 10: Callen RA (1984) Quaternary climatic cycles, 241–72, 404–33. Lake Millyera region, southern Strzelecki Capell A (1956) A new approach to Australian Desert. Transactions of the Royal Society of South linguistics. Oceania Linguistic Monographs 1. Australia 108: 163–73. Sydney: University of Sydney. Callen RA, Wasson RJ and Gillespie R (1983) Capell A (1962) Some linguistic types in Aus- Reliability of radiocarbon dating of pedogenic tralia. Handbook of Australian languages, part carbonate in the Australian arid zone. Sedimen- 2, Oceania Linguistic Monograph 7. Sydney: tary Geology 35: 1–14. University of Sydney. Campbell JB (1982) Automatic seafood retrieval Chen XY and Barton CE (1991) Onset of aridity systems: The evidence from Hinchinbrook and dune-building in central Australia: Sed- Island and its implications. In S Bowdler (ed) imentological and magnetostratigraphic evi- Coastal archaeology in eastern Australia,pp.96– dence from Lake Amadeus. Palaeogeography, 107. Canberra: Department of Prehistory, the Palaeoclimatology, Palaeoecology 84: 55–73. Australian National University. Chen XY, Bowler JM and Magee JW (1991) Campbell JB (1984)Extendingthearchaeologi- Aeolian landscapes in central Australia: Gyp- cal frontier: A review of work on the prehis- siferous and quartz dune environments from tory of north Queensland. Queensland Archae- Lake Amadeus. Sedimentology 38: 519–38. ological Research 1: 173–84. Chen XY, Bowler JM and Magee JW (1993) Campbell TD (1925) Detailed notes on the Abo- Late Cenozoic stratigraphy and hydrologic riginal intaglios near Burra. Transactions of the history of Lake Amadeus, a central Australian Royal Society of South Australia 49: 123–7. playa. Australian Journal of Earth Sciences 40: 1– Campbell TD (1960) The Pirri – An interesting 14. Australian Aboriginal implement. Records of Chen XY, Chappell J and Murray AS (1995) the South Australian Museum 13: 509–24. High (ground) water levels and dune devel- Cane S (1984) Desert Camps: A case study of stone opment in central Australia: TL dates from artefacts and Aboriginal behaviour in the West- gypsum and quartz dunes around Lake Lewis ern Desert. Unpublished PhD thesis, Australian (Napperby), Northern Territory. Geomorphol- National University. ogy 11: 311–22. Cane S (1987) Australian Aboriginal Subsistence Childe V Gordon (1928) The most ancient east: in the Western Desert. Human Ecology 15: The oriental prelude to European prehistory. 391–434. London: Kegan Paul. Cane S (1989) Australian Aboriginal seed grind- Chivas AR and Bowler JM. (1986)Introduc- ing and its archaeological record: A case study tion – The SLEADS Project. Palaeogeography, from the Western Desert. In DR Harris Palaeoclimatology, Palaeoecology 54: 3–6. and GC Hillman (eds), Foraging and farming: Clark PU, Dyke AS, Shakun JD, Carlson AE, The evolution of plant exploitation,pp.99–119. Clark J, Wohlfarth B et al. (2009) The last London: Unwin Hyman. glacial maximum. Science 325: 710–14. Cane S (1995) Nullarbor antiquity: Archaeolog- Clarke J (1976) Two Aboriginal rock art pig- ical, luminescent and seismic investigations on ments from Western Australia. Their proper- the Nullarbor Plain. Report to the National ties, use and durability. Studies in Conservation Estate Grants Program, Australian Heritage 21: 143–2. Commission, Canberra. Clarke J (1983) An Aboriginal engraving site in Cane S (2001) The Great Flood: Eustatic change the south-west of Western Australia. Records of and cultural change in Australia during the the Western Australian Museum 11: 63–7. late Pleistocene and Holocene. In A Ander- Clarke J (1989) Further comments (on Nobbs son, I Lilley and S O’Connor (eds) Histories and Dorn, ‘Age determinations for rock References 355

varnish formation within petroglyphs’). Rock in ancient Australians: Implications for mod- Art Research 6: 63–5. ern human origins’. Archaeology in Oceania 36: Clarkson C (2007) Lithics in the land of the 168–9. Lightning Brothers: The archaeology of Wardaman Comtesse S (2003) Mt Newman sites re-analysed: Country, Northern Territory. Terra Australis 25. Newman Ore body XXIX, Newman. Rockshel- Canberra: ANU E Press. ter and P0959. Unpublished BSc (Hons) thesis, Clegg J (1984) Comments on MF Nobbs, ‘Rock University of Western Australia. art in Olary province, South Australia’. Rock Conkey MW (1980) The identification of Art Research 1: 111–17. prehistoric hunter-gatherer aggregation sites: Clegg J (1987) Style and tradition at Sturt’s The case of Altamira. Current Anthropology 21: Meadows. World Archaeology 19: 236–55. 609–30. Clement E (1904) Ethnographical notes on the Conroy JL, Overpeck JT, Cole JE, Shanahan Western-Australian Aborigines. Archives Inter- TM and Steinitz-Kannan M (2008) Holocene nationals D’Ethnographie 16: 1–29. changes in eastern tropical Pacific climate Clendon M (2006) Reassessing Australia’s lin- inferred from a Galapagos lake sediment guistic prehistory. Current Anthropology 47: 39– record. Quaternary Science Reviews 27: 1166– 61. 80. Clobert J, Danchin E, Dhondt AA and Nichols Corbett L (1995) The dingo in Australia and JD (2001) Dispersal. Oxford: Oxford Univer- Asia. Australian Natural History series. Syd- sity Press. ney: UNSW Press. Clune G and Harrison R (2009) Coastal shell Cosford J, Qing H, Eglington B, Mattey D, Yuan middens of the Abydos coastal plain, Western D, Zhang M and Cheng H (2008) East Asian Australia. Archaeology in Oceania 44 (Supple- monsoon variability since the Mid-Holocene ment): 70–80. recorded in a high-resolution, absolute-dated Codding BF (2012) ‘Any kangaroo?’ On the ecol- aragonite speleothem from eastern China. ogy, ethnography and archaeology of foraging in Earth and Planetary Science Letters 275: 296– Australia’saridwest. Unpublished PhD thesis, 307. Stanford University. Cracraft J (1991) Patterns of diversification Codding BF, Bird DW and Bird R Bleige (2010) within continental biotas: Hierarchical con- Interpreting abundance indices: Some zoo- gruence among the areas of endemism of Aus- archaeological implications of Martu foraging. tralian vertebrates. Australian Systematic Botany Journal of Archaeological Science 37: 3200–10. 4: 211–27. CoggerHGandHeatwoleH(1981) The Aus- Crawford IM (1963) Report of excavation of a site tralian reptiles: Origins, biogeography, distri- at Wilgie Mia in February 1963. Unpublished bution patterns and island evolution. In A report. Australian Institute of Aboriginal and Keast (ed) Ecological biogeography of Australia, Torres Strait Islander Studies, Canberra. AIAS pp. 1331–73. The Hague: Junk. Doc. No. 63/7. Cohen TJ, Nanson GC, Jansen JD, Jones BG, Crisp MD, Isagi Y, Kato Y, Cook LG and Bow- Jacobs Z, Larsen JR et al. (2012)LateQua- man DMJS (2010) Livistona palms in Australia: ternary mega-lakes fed by the northern and Ancient relics or opportunistic immigrants? southern river systems of central Australia: Molecular Phylogenetics and Evolution 54: 512– Varying moisture sources and increased con- 23. tinental aridity. Palaeogeography, Palaeoclimatol- Crisp MD, Linder HP and Weston PH (1995) ogy, Palaeoecology 356–57: 89–109. Cladistic biogeography of plants in Australia Cohen TJ, Nanson GC, Larsen JR, Jones BG, and New Guinea: Congruent pattern reveals Price DM, Coleman M, and Pietsch TJ (2010) two endemic tropical tracks. Systematic Biology Late Quaternary aeolian and fluvial interac- 44: 457–73. tions on the Cooper Creek fan and the asso- Crocker RL (1946) Post-Miocene climatic and ciation between linear and source-bordering geologic history and its significance in the dunes, Strzelecki Desert, Australia. Quaternary genesis of the major soil types of South Aus- Science Reviews 29: 455–71. tralia. CSIR Bulletin 193. Melbourne: Coun- Colgan DJ (2001) Commentary on Adcock GJ cil for Scientific and Industrial Research et al. 2001 ‘Mitochondrial DNA sequences (Australia). 356 References

Crocker RL (1959) Past climatic fluctuations and prehistory. Memoirs of the Queensland Museum their influence upon Australian vegetation. In 37: 357–28. A Keast, RL Crocker and CS Christian (eds) David B, Clayton E and Watchman A (1993) Biogeography and ecology in Australia,pp.283– Initial results of PIXE analysis of Northern 90. Den Haag, Netherlands: W Junk. Territory ochres. Australian Archaeology 36: 50– Crocker RL and Cotton BC (1946) Some raised 7. beaches of the lower south-east of South Aus- David B, McNiven I, Attenbrow V, Flood J tralia and their significance. Transactions of the and Collins J (1994) Of Lightning Brothers Royal Society of South Australia 70: 64–82. and White Cockatoos: Dating the antiquity Crocker RL and Wood JG (1947) Some his- of signifying systems in the Northern Terri- torical influences on the development of the tory, Australia. Antiquity 68: 241–51. South Australian vegetation communities and David B, Watchman A, Goodall R and Clayton their bearing on concepts and classification in E(1995) The Maytown ochre source. Memoirs ecology. Transactions of the Royal Society of South of the Queensland Museum 38: 441–5. Australia 71(1): 91–136. David B, Geneste J-M, Whear RL, Delannoy Crowley GM (1996) Late Quaternary mangrove J-J, Katherine M, Gunn RG et al. (2011) distribution in northern Australia. Australian Nawarla Gabarnmang, a 45,180±910 cal BP Systematic Botany 9: 219–25. site in Jawoyn country, southwest Arnhem Cundy BJ (1989) Formal variation in Australian Land plateau. Australian Archaeology 73: 73–7. spear and spearthrower technology.Oxford:BAR David TW Edgeworth (1932) Explanatory notes International Series 546. to accompany a new geological map of the Com- Cupper ML (2005) Last glacial to Holocene evo- monwealth of Australia. Sydney: Common- lution of semi-arid rangelands in southeastern wealth Council for Scientific and Industrial Australia. The Holocene 15: 541–53. Research. Cupper ML and Duncan J (2006) Last glacial DavidTWEdgeworthandBrowneWR(1950) megafaunal death assemblage and early human The geology of the Commonwealth of Australia. occupation at Lake Menindee, southeastern London: E. Arnold. Australia. Quaternary Research 66: 332–41. Davidson DS (1928) The chronological aspects of cer- Curnoe D (2009) Possible causes and significance tain Australian social institutions as inferred from of cranial robusticity among Pleistocene–Early geographical distribution. Philadelphia: Univer- Holocene Australians. Journal of Archaeological sity of Philadelphia. Science 36: 980–90. Davidson DS (1937) A preliminary consideration Curnoe D and Thorne A (2006) Human ori- of Aboriginal Australian decorative art. Memoirs gins in Australia: The skeletal evidence. Before of the American Philosophical Society, vol. 9. Farming 2006/1 Article 5: 1–28. Philadelphia: American Philosophical Society. Darwin C (1859) The origin of species by means Davidson DS (1938) Stone axes of Western of natural selection. London: John Murray Australia. American Anthropology 40: 38–48. [Penguin Classics edition, 1985]. Davidson DS (1949a) The interlocking key Davenport S, Johnson P and Yuwali (2005) design in Aboriginal Australian decorative art. Cleared out: First contact in the Western Desert. Mankind 4: 85–98. Canberra: Aboriginal Studies Press. Davidson DS (1949b) Mourning-caps of the DavidB(1990) Echidna’s Rest, Chillagoe: A Australian Aborigines. Proceedings of the Amer- site report. Queensland Archaeological Research ican Philosophical Society 93: 57–70. 7: 73–94. Davidson DS (1952) Notes on the pictographs David B (2002) Landscapes, rock-art and the Dream- and petroglyphs of Western Australia and a ing: An archaeology of preunderstanding. London: discussion of their affinities with appearances Leicester University Press. elsewhere on the continent. Proceedings of the David B, Chant D and Flood J (1992). Jalijbang American Philosophical Society 96: 76–117. 2 and the distribution of pecked faces in Aus- Davidson DS and McCarthy FD (1957) The tralia. Memoirs of the Queensland Museum 32: distribution of stone implements in Western 61–77. Australia. Anthropos 52: 390–458. David B and Chant D (1995)Rockart Davidson I (1983) On the edge of the Simp- and regionalisation in north Queensland son: Recent additions to understanding of References 357

prehistoric artefact distributions in arid Aus- Hudson. Fondation Cartier pour l’art con- tralia. Australian Archaeology 17: 27–37. temporain. Davidson I (1990) Prehistoric Australian demog- DeVogel, SB, Magee, JW, Manley, WF and raphy. In B Meehan and N White (eds) Miller, GH (2004) A GIS-based reconstruc- Hunter-gatherer demography,pp.41–58. Sydney: tion of late Quaternary paleohydrology: Lake University of Sydney, Oceania monograph Eyre, arid central Australia. Palaeogeography, 39. Palaeoclimatology, Palaeoecology 204: 1–13. Davidson I, Cook N, Fischer M, Ridges M, Dias A, Veitch B and Hook F (2006) Report of Ross J and Sutton S (2005) Archaeology in the test excavation of four Aboriginal archaeologi- another country: Exchange and symbols in cal sites, on the West Angelas mining lease, Pil- north-west central Queensland. In I Macfar- bara, Western Australia. Unpublished report by lane, M-J Mountain and R Paton (eds) Many Archae-Aus Pty Ltd to Pilbara Iron. exchanges: Archaeology,history,community and the Dickins J (1996) Change and continuity in work of Isabel McBryde,pp.103–30. Canberra: Central Australian graphic systems. Journal of Aboriginal History Monograph 11. Anthropological Archaeology 15: 20–40. Davidson I, Sutton SA and Gale SJ (1993) The Dickman C and Woodford Ganf RW (2007) A human occupation of Cuckadoo 1 rockshel- fragile balance: The extraordinary story of Aus- ter, northwest central Queensland. In MA tralian marsupials. Fishermans Bend: Craftsman Smith, M Spriggs and B Fankhauser (eds) House. Sahul in review: Pleistocene archaeology in Aus- Dickson FP (1981) Australian stone hatchets: A tralia, New Guinea and Island Melanesia,pp. study in design and dynamics. Sydney: Academic 164–72. Canberra: Department of Prehistory, Press. Research School of Paciﬁc Studies, The Aus- Di Lernia S (1999) The Uan Afuda Cave: Hunter- tralian National University. Occasional Papers gatherer societies of central Sahara. Firenze: Arid in Prehistory 24. zone archaeology monographs 1. Davidson I, Tarrago´ I and Sullivan T (2004)Mar- Dix W (1977) Facial representations in Pilbara ket forces. In V Donovan and C Wall (eds) rock engravings. In PJ Ucko (ed) Form in Making connections: A journey along Central Aus- indigenous art: Schematisation in the art of Aborig- tralian Aboriginal trading routes,pp.12–23.Bris- inal Australia and prehistoric Europe,pp.277–85. bane: Queensland Government, Arts Queens- Canberra: Australian Institute of Aboriginal land. Studies. Davis MA (2009) Invasion biology.Oxford: Dix WC and Virili FL (1977) Prehistoric pet- Oxford University Press. roglyphs of the Dampier Archipelago, North DavisRC(ed)(2002) The Central Australian expe- Western Australia. Bollettino del Centro Camuno dition 1844–1846: The journals of Charles Sturt. di Studi Preistorici 16: 87–110. London: Hakluyt Society. Dixon RMW (2002) Australian languages: Their Dawson TJ and Dawson L (2006)Evolution nature and development. Cambridge: Cam- of arid Australia and consequences for verte- bridge University Press. brates. In JR Merrick, M Archer, GM Hickey Dodson JR and Wright RVS (1989) Humid and MSY Lee (eds) Evolution and biogeography to arid to subhumid vegetation shift on Pil- of Australasian vertebrates,pp.51–70. Oatlands: liga Sandstone, Ulungra Springs, New South Australian Scientiﬁc Publishing. Wales. Quaternary Research 32: 182–92. De Deckker P (2001) Late Quaternary cyclic Doelman T (2008) Time to quarry: The archaeology aridity in tropical Australia. Palaeogeography, of stone procurement in northwestern New South Palaeoclimatology, Palaeoecology 170: 1–9. Wales, Australia. BAR International Series De Deckker P, Magee JW and Shelley JMG 1810. Oxford: Archaeopress. (2011) Late Quaternary palaeohydrological Donaldson M (2009) Burrup Rock Art: Ancient changes in the large playa Lake Frome in cen- Aboriginal rock art of Burrup peninsula and tral Australia, recorded from the Mg/Ca and Dampier archipelago, Western Australia. Mount Sr/Ca in ostracod valves and biotic remains. Lawley WA: Wildrocks Publications. Journal of Arid Environments 75: 38–50. Dorn R (1997) Uncertainties in 14Cagesfor Depardon R, Khemir M, Thesiger W and Virilio petroglyphs from the Olary province, South P(2000) The desert. New York: Thames and Australia. Archaeology in Oceania 32: 214–15. 358 References

Dorn RI, Nobbs M and Cahill TA (1988). totemic system in Australia]. Paris: Librairie Cation-ratio dating of rock-engravings from Felix Alcan. the Olary Province of arid South Australia. Dury GH and Langford-Smith T (1970) A Pleis- Antiquity 62: 681–9. tocene Aboriginal camp fire from Lake Yan- Dortch CE (1977) Early and late stone industrial tara, northwestern New South Wales. Search phases in Western Australia. In RVS Wright 1: 73. (ed) Stone tools as cultural markers,pp.104–32. Dyason F (2007) Identification and dental cemen- Canberra: Australian Institute of Aboriginal tum analysis of Macropodoidea and Potoroidae teeth Studies. from the ‘Silver Dollar’ archaeological site, Shark Dortch CE (1984) Errata notice: Walga Rock Bay, Western Australia. Unpublished BSc thesis, and Billibilong Spring. Australian Archaeology University of Western Australia. 18: 91. Earl JW and McCleary BV (1994)Mysteryof Dortch CE (2002) Preliminary underwater sur- the poisoned expedition. Nature 368: 683–4. vey for rock engravings and other sea floor Eden W [Baron Auckland] (1787) The history of sites in the Dampier Archipelago, Pilbara New Holland from its first discovery in 1616 to the region, Western Australia. Australian Archae- present time. London: John Stockdale. Second ology 54: 37–42. edition. Douglas K (2006) ‘Forsaken spot’ to ‘classic Edwards DA and O’Connell JF (1995)Broad ground’: Geological heritage in Australia and spectrum diets in arid Australia. Antiquity 69: the recuperative power of the deep past. Envi- 769–83. ronment and History 12: 269–96. Edwards E and Murphy A (2001) A preliminary Dragovich D (1984) Minimum age for desert report on archaeological investigations at Malea varnish in the Broken Hill area, NSW: A pre- Rockshelter, near Newman, Western Australia. liminary estimate. Search 15: 113–15. Perth: McDonald, Hales and Associates. Dragovich D (1986) Minimum age of some Edwards E and Murphy A (2003) A prelimi- desert varnish near Broken Hill, New South nary report on archaeological investigations at Wales. Search 17: 149–51. Malea Rockshelter, Pilbara region, Western Dragovich D (1994) Fire, climate, and the per- Australia. Australian Archaeology 56: 44–6. sistence of desert varnish near Dampier, West- Edwards R (1963) Rock engravings of Panarami- ern Australia. Palaeogeography, Palaeoclimatology, tee station, north-eastern South Australia. Palaeoecology 111: 279–88. Transactions of the Royal Society of South Aus- Dragovich D (1998) Microchemistry of small tralia 86: 131–46. desert varnish samples, western New South Edwards R (1965) Rock engravings and incised Wales, Australia. Earth Surface Processes and stones: Tiverton station, North-east South Landforms 23: 445–53. Australia. Mankind 6: 223–31. Dragovich D (2000) Rock engraving chronolo- Edwards R (1966) Comparative study of rock gies and accelerator mass spectrometry radio- engravings in south and central Australia. carbon age of desert varnish. Journal of Archae- Transactions of the Royal Society of South Aus- ological Science 27: 871–6. tralia 90: 33–8. Draper N (1989) Arkaroo rock painting sites con- Edwards R (1968) Prehistoric rock engravings servation – Final report. Unpublished report, at Thomas Reservoir, Cleland Hills, western pMS4657, Australian Institute of Aboriginal Central Australia. Records of the South Australian and Torres Strait Islander Studies, Canberra. Museum 15: 647–70. Dubreuil L (2004) Long-term trends in Natu- Edwards R (1971) Art and Aboriginal prehistory. fian subsistence: A use-wear analysis of ground In DJ Mulvaney and J Golson (eds) Aborigi- stone tools. Journal of Archaeological Science 31: nal man and environment in Australia,pp.356– 1613–29. 67. Canberra: Australian National University Duncan-Kemp AM (1968) Where strange gods call. Press. Brisbane: WR Smith & Paterson. Elias M (1982) Belele, Western Australia, 1:250,000 Durkheim E (1912) Les formes el´ ementaires´ de la vie Geological series – Explanatory notes. Sheet religieuse: Le systeme` totemique´ en Australie. [The SG50–11. Perth: Geological Survey of West- elementary forms of the religious life: The ern Australia. References 359

Elkin AP (1931) The social organization of South Favenc E (1893) The influence of geographic Australian tribes. Oceania 2: 44–73. position on the development of the Australian English P, Spooner NA, Chappell J, Questiaux natives. The Geographical Journal 2: 316–20. DG and Hill NG (2001) Lake Lewis basin, Ferry J (1999) Colonial Armidale. St Lucia: central Australia: Environmental evolution Queensland of University Press. and OSL chronology. Quaternary International Field J and Dodson J (1999) Late Pleistocene 83–85: 81–101. megafauna and archaeology from Cuddie Evans N (2005) Review article. Australian lan- Springs, south-eastern Australia. Proceedings of guages reconsidered: A review of Dixon the Prehistoric Society 65: 275–301. (2002). Oceanic Linguistics 44: 242–86. Field JH, Dodson JR and Prosser IP (2002)A Evans N and Jones R (1997) The cradle of the late Pleistocene vegetation history from the Pama-Nyungans: Archaeological and linguis- Australian semi-arid zone. Quaternary Science tic speculations. In P McConvell and N Evans Reviews 21: 1023–37. (eds) Archaeology and linguistics: Aboriginal Aus- Field JH, Fullagar R, Dortch J, Dutkiewicz A tralia in global perspective,pp.385–417.Mel- and Gordon P (2003) Sandstone quarries and bourne: Oxford University Press. grinding stone manufacture: Survey and exca- Eyre EJ (1845) Journals of expeditions of discovery vation at Yambacoona Hill in south-eastern into Central Australia and overland from Adelaide Australia. Australian Archaeology 56: 46–7. to King George’s Sound in the years 1840–1. 2 Field J, Fullagar R and Lord G (2001) A large area volumes. London: T.and W.Boone [Adelaide: archaeological excavation at Cuddie Springs. Friends of the State Library of South Australia, Antiquity 75: 696–702. 1997]. Fillios M, Field J and Charles B (2010) Investigat- Fairbridge RW (1961) Eustatic changes in sea ing human and megafauna co-occurrence in level. Physics and Chemistry of the Earth 4: 99– Australian prehistory: Mode and causality in 185. fossil accumulations at Cuddie Springs. Qua- Falkenberg J (1962) Kin and totem: Group relations ternary International 211: 123–43. of Australian Aborigines in the Port Keats district. Finlayson HH (1936) The Red Centre: Man and Oslo: Oslo University Press. beast in the heart of Australia. Third edition. Fanning PC (1999) Recent landscape history in Sydney: Angus and Robertson. arid western New South Wales, Australia: A Fitzsimmons KE, Bowler JM, Rhodes EJ and model for regional change. Geomorphology 29: Magee JW (2007) Relationships between 191–209. desert dunes during the late Quaternary in the Fanning PC, Holdaway SJ, Rhodes EJ and Lake Frome region, Strzelecki Desert, Aus- Bryant TG (2009) The surface archaeologi- tralia. Journal of Quaternary Science 22: 549–58. cal record in arid Australia: Geomorphic con- Fitzsimmons KE, Rhodes EJ, Magee JW and trols on preservation, exposure and visibility. Barrows TT (2007) The timing of linear dune Geoarchaeology 24: 121–46. activity in the Strzelecki and Tirari Deserts, Farrand WR (2001) Sediments and stratigraphy Australia. Quaternary Science Reviews 26: 2598– in rockshelters and caves: A personal perspec- 616. tive on principles and pragmatics. Geoarchae- Flannery TF (1994) The future eaters: An ecological ology 16: 537–57. history of the Australasian lands and people.Port Faulkner P (2009) Focused, intense and long- Melbourne: Reed Books. term: Evidence for granular ark (Anadara gra- Flenniken JJ and White JP (1985) Australian nosa) exploitation from late Holocene shell flaked stone tools: A technological perspec- mounds of Blue Mud Bay, northern Australia. tive. Records of the Australian Museum 36: 131– Journal of Archaeological Science 36: 821–34. 51. Faulkner P and Thorley P (2009) Stone artefact Florek SM (1993) Archaeology of the Mound Spring assemblages from Kulpi Mara rockshelter, Central Campsites near Lake Eyre in South Australia. Un- Australia: A final report prepared for Australian published PhD thesis, University of Sydney. Institute of Aboriginal and Torres Strait Islander Flower WH and Lydekker R (1891) An introduc- Studies. Unpublished report, AIATSIS, Can- tion to the study of mammals living and extinct. berra. London: Adam and Charles Black. 360 References

Forbes S (1982) Aboriginal rock engravings at Ma, dated by cosmogenic 21Ne-10Be. Geology N’Dahla Gorge: An analysis of a Central Aus- 33: 993–6. tralian ‘Panaramitee Style’ rock art site. Unpub- Fujioka T, Chappell J, Fifield K and Rhodes lished BLitt thesis, Australian National Uni- EJ (2009) Australian desert dune fields initi- versity. ated with Pliocene-Pleistocene global climatic Ford B, MacLeod I and Haydock P (1994)Rock shift. Geology 37: 51–4. art pigments from Kimberley region of West- Fullagar R and Field J (1997) Pleistocene seed- ern Australia: Identification of the minerals grinding implements from the Australian arid and conversion mechanisms. Studies in Con- zone. Antiquity 71: 300–7. servation 39: 57–69. Furby JH (1995) Megafauna under the microscope: Forman RTT (1995) Land mosaics: The ecology of Archaeology and palaeoenvironment at Cuddie landscapes and regions. Cambridge: Cambridge Springs. Unpublished PhD thesis, School of University Press. Geography, University of Sydney. Frank R (1971) Cave sediments as palaeoen- Gale F (1978) Woman’s role in Aboriginal society. vironmental indicators, and the sedimentary Third edition. Canberra: Australian Institute sequence in Koonalda Cave. In DJ Mulvaney of Aboriginal Studies. and J Golson (eds) Aboriginal man and environ- Galloway RW (1965) Late Quaternary climates ment in Australia,pp.94–104. Canberra: Aus- in Australia. Journal of Geology 73: 603–18. tralian National University Press. Gallus A (1968) Excavations at Koonalda Cave, Franklin NR (2004) Explorations of variability in South Australia. Current Anthropology 9: 324– Australian prehistoric rock engravings. BAR Inter- 5. national Series 1318. Oxford: British Archae- Gallus A. (1971) Results of the exploration of ological Reports. Koonalda Cave, 1956–68. In RVSWright (ed) Frawley S and O’Connor S (2010)A40,000 year Archaeology of the Gallus site, Koonalda Cave,pp. wood charcoal record from Carpenter’s Gap 87–133. Australian Aboriginal Studies No. 26. 1: New insights into palaeovegetation change Canberra: Australian Institute of Aboriginal and indigenous foraging strategies in the Kim- Studies. berley, Western Australia. In S Haberle, J Galt-Smith B (1997) Motives for motifs: Identify- Stevenson and M Prebble (eds) Altered ecolo- ing aggregation and dispersion settlement patterns gies: Fire, climate and human influence on terres- in the rock art assemblages of Central Australia. trial landscapes,pp.299–321. Terra Australis 32. Unpublished BA (Hons) thesis, University of Canberra: ANU. New England. Frazer JG (1899) The origin of totemism. Fort- GambleC(1982) Interaction and Alliance in nightly Review 71: 647–65. Palaeolithic Society. Man n.s. 17: 92–107. Frederick UK (1997) Drawing in differences: Gamble C (1993) Timewalkers: The Prehistory Changing social contexts of rock art production in of Global Colonisation. Phoenix Mill, United Watarrka (Kings Canyon) National Park, Central Kingdom: Alan Sutton Publishing. Australia. Unpublished MA thesis, Australian Gamble C (1999) The Palaeolithic societies of National University. Europe. Cambridge: Cambridge University French P, James E and Walsh N (2012) Analy- Press. sis of genetic variation in a disjunct, narcotic Gamble C, Davies W, Pettitt P, Hazelwood L producing, population of Dubosia hopwoodii (F. and Richards M (2005) The archaeological Muell.) F. Muell. Muellaria 20: 65–71. and genetic foundations of the European pop- Friedel MH (1991) Range condition assessment ulation during the late glacial: Implications for and the concept of thresholds: A viewpoint. ‘agricultural thinking’. Cambridge Archaeologi- Journal of Range Management 44: 422–6. cal Journal 15: 193–223. Friedlaender JS, Friedlaender FR, Hodgson JA, Garcea EAA (2004) Crossing deserts and avoid- Stoltz M, Koki G, Horvat G et al. (2007) ing seas: Aterian North African-European Melanesian mtDNA complexity. PLoS ONE relations. Journal of Anthropological Research 60: 2(2): e248. 27–53. Fujioka T, Chappell J, Honda M, Yatsevich I, Gardner GJ, Mortlock AJ, Price DM, Readhead Fifield K and Fabel D (2005) Global cooling ML and Wasson RJ (1987) Thermolumines- initiated stony deserts in central Australia 2–4 cence and radiocarbon dating of Australian References 361

desert dunes. Australian Journal of Earth Sciences AMS radiocarbon dating in the study of 34: 343–57. arid environments: Examples from Lake Eyre, Gason S (1874) TheDieyerietribeofAustralian South Australia. Palaeogeography, Palaeoclimatol- Aborigines. Adelaide: WC Cox, Government ogy, Palaeoecology 84: 333–8. Printer. Gilligan I and Bulbeck D (2007) Environment Gason S (1879) The manners and customs of and morphology in Australian Aborigines: A the Dieyerie tribe of Australian Aborigines. Re-analysis of the Birdsell Database. American In JD Woods (ed) The native tribes of South Journal of Physical Anthropology 134: 75–91. Australia,pp.253–307. Adelaide: ES Wigg and Glasby P, O’Flaherty A and Williams MAJ Son [Facsimile edition, Friends of the State (2010) A geospatial visualization and chrono- Library of South Australia, 2009]. logical study of a late Pleistocene fluvial wet- Gason S (1882) Letter to AW Howitt. Howitt land surface in the semi-arid Flinders Ranges, Papers, 1052/1(c) MS 9356, State Library of South Australia. Geomorphology 118: 130–51. Victoria, Melbourne. Glover IC and Lampert RJ (1969) Puntutjarpa Gason S (1895) Of the Tribes, Dieyerie, Rockshelter excavations by RA Gould: A Auminie, Yandrawontha, Yarawuarka, Pil- critical review. Archaeology and Physical Anthro- ladapa. Journal of the Anthropological Institute of pology in Oceania 4: 223–8. Great Britain and Ireland 24: 167–76. Goddard RH (1936) Kopi: Funerary skull caps. Gentilli J (1949) Foundations of Australian bird Mankind 2: 25–7. geography. Emu 49: 85–129. Gollan K (1984) The Australian dingo: In the Gentilli J (1961) Quaternary climates of the shadow of man. In M Archer and G Clay- Australian region. Annals of the New York ton (eds) Vertebrate zoogeography and evolution Academy of Science 95: 465–501. in Australasia,pp.923–7. Carlisle, WA: Hes- Gentilli J (1972) Australian climate patterns. perian Press. Melboure: Nelson. Golson J (1971) Australian Aboriginal food Gerritsen R (2008) Australia and the origins of agri- plants: Some ecological and culture-historical culture. Oxford: Archaeopress. BAR Interna- implications. In DJ Mulvaney and J Gol- tional Series 1874. son (eds) Aboriginal man and environment in Gibbs M. and Veth P. (2002) Ritual engines and Australia,pp.196–238. Canberra: Australian the archaeology of territorial ascendancy. Te m - National University Press. pus 7: 11–19. Goodall R, David B and Bartley J (1996) Non- Giles E (1889) Australia twice traversed: The romance destructive techniques for the analysis and of exploration, being a narrative compiled from characterisation of pigments from archaeolog- the journals of five exploring expeditions into and ical sites: The case of Fern Cave. In S Ulm, I through Central South Australia, and Western Lilley and A Ross (eds) Australian Archaeology Australia, from 1872 to 1876. Vols. I and II. ‘95: Proceedings of the 1995 Australian Archaeo- London: Sampson Low, Marston, Searle and logical Association Annual Conference,pp.183– Rivington. [Facsimile edition, 1964,Libraries 7. St Lucia, Brisbane: Anthropology Museum, Board of South Australia]. University of Queensland. Gill ED (1965) The paleogeography of Australia Gorbushina AA (2007) Life on the rocks. Envi- in relation to the migrations of marsupials and ronmental Microbiology 9: 1613–31. men. Annals of the New York Academy of Science Gordon RJ (1984) The !Kung in the Kalahari Series II, 28: 5–15. exchange: An ethnohistorical perspective. In Gillespie R and Brook BW (2006)Isthere C Schrire (ed) Past and present in hunter gatherer a Pleistocene archaeological site at Cuddie studies,pp.195–224. Orlando, Florida: Aca- Springs? Archaeology in Oceania 41: 1–11. demic Press. Gillespie R, Fifield LK, Levchenko V and Wells Gorecki P, Grant M, O’Connor S and Veth R(2008)New14C ages on cellulose from P(1997) The morphology, function and Diprotodon gut contents: Explorations in oxi- antiquity of Australian grinding implements. dation chemistry and combustion. Radiocar- Archaeology in Oceania 32:141–50. bon, 50: 75–81. Goudie AS, Stokes S, Livingstone I, Bailiff IK Gillespie R, Magee JW, Luly JG, Dlugo- and Alison RJ (1993) Post-depositional mod- kencky E, Sparks RJ and Wallace G (1991) ification of the linear sand ridges of the West 362 References

Kimberley area of North-West Australia. The Grayson DK (2001) The archaeological record of Geographical Journal 159: 306–17. human impacts on animal populations. Journal Gould RA (1968) Preliminary report on exca- of World Prehistory 15: 1–68. vations at Puntutjarpa Rockshelter near the Greathouse EA (1985) Archaeological investigations Warburton Ranges, Western Australia. Archae- at Kuyunba (Pine Gap), Northern Territory, Aus- ology and Physical Anthropology in Oceania 3: tralia. Unpublished MA thesis, California State 161–85. University, Stanislaus. Gould RA (1969a) Subsistence behaviour among Gregory AC (1887) Appendix II. Memoranda the Western Desert Aborigines of Australia. on the Aborigines of Australia. In H. Ling Oceania 39: 253–74. Roth ‘On the origins of agriculture’. Journal Gould, RA (1969b) Puntutjarpa Rockshelter: A of the Anthropological Institute of Great Britain reply to Messrs. Glover and Lampert. Archae- and Ireland 16: 102–36. ology and Physical Anthropology in Oceania 4: Gregory JW (1906) The Dead Heart of Australia: A 229–37. journey around Lake Eyre in the summer of 1901– Gould RA (1969c) Yiwarra: Foragers of the Aus- 1902. London: John Murray. [Facsimile edi- tralian Desert. Sydney: Collins. tion, North Adelaide: Corkwood Press, 1997]. Gould RA (1971a) The archaeologist as ethno- Griffiths ML, Drysdale RN, Gagan MK, Zhao grapher: A case from the Western Desert of J-X, Ayliffe LK, Hellstrom JC et al. (2009) Australia. World Archaeology 3: 143–77. Increasing Australian–Indonesian monsoon Gould RA (1971b) Uses and effects of fire among rainfall linked to early Holocene sea-level rise. the Western Desert Aborigines of Australia. Nature Geoscience 2: 636–9. Mankind 8: 14–24. Grocke¨ DR (1997) Distribution of C3 and C4 Gould RA (1977) Puntutjarpa Rockshelter and the plants in the late Pleistocene of South Australia Australian desert culture. Anthropological Papers recorded by isotope biogeochemistry of colla- of the American Museum of Natural History gen in megafauna. Australian Journal of Botany 54 (1): 1–187. New York: American Museum 45: 607–17. of Natural History. Groube L, Chappell J, Muke J and Price D. Gould RA (1978a) James Range East Rock Shel- (1986)A40,000-year-old human occupation ter, Northern Territory, Australia: A summary site at Huon Peninsula, Papua New Guinea. of the 1973 and 1974 investigations. Asian Per- Nature 324: 453–5. spectives 21: 85–126. Grove M (2010) Logistical mobility reduces sub- Gould RA (1978b) The anthropology of human sistence risk in hunting economies Journal of residues. American Anthropologist 80: 815–35. Archaeological Science 37: 1913–21. Gould RA (1980) Living archaeology. Cambridge: Groves, C. (2001) Lake Mungo 3 and his DNA. Cambridge University Press. Archaeology in Oceania 36: 166–7. Gould RA (1991) Arid-land foraging as seen Grun¨ R, Eggins S, Aubert M, Spooner N, Pike from Australia: Adaptive models and behav- AWG and Muller¨ W (2010)ESRandU- ioral realities. Oceania 62: 12–33. series analyses of faunal material from Cud- Gould RA (1996) Faunal reduction at Puntut- die Springs, NSW, Australia: Implications for jarpa Rockshelter, Warburton Ranges, West- the timing of the extinction of the Aus- ern Australia. Archaeology in Oceania 31: 72– tralian megafauna. Quaternary Science Reviews 86. 29: 596–610. Government of South Australia (1860)Explo- Gunn RG (1994) The rock art of Kwerlpe, James ration of the interior. Parliamentary Debates Range, Central Australia. Unpublished report. (Hansard), Legislative Council and House of Aboriginal Areas Protection Authority, Alice Assembly. 9 October, 1006–20. Springs. Graham R and Thorley P (1996) Central Aus- Gunn RG (1995) Regional patterning in the tralian Aboriginal stone knives: Their cul- Aboriginal rock-art of Central Australia: A tural significance, manufacture and trade. In preliminary report. Rock Art Research 12: 117– SR Morton and DJ Mulvaney (eds) Exploring 28. Central Australia: Society, the environment and Gunn RG (1997) Rock art, occupation and the 1894 Horn Expedition,pp.74–89. Chipping myth: The correspondence of symbolic and Norton (Sydney): Surrey Beatty & Sons. archaeological sites within Arrernte rock art References 363

complexes of central Australia. Rock Art Hayden B (1973) Analysis of a ‘Taap’ composite Research 14:124–36. knife. Archaeology and Physical Anthropology in Gunn RG (2000) Central Australian rock-art: A Oceania 8: 116–26. second report. Rock Art Research 17: 111–26. Hayden B (1975) Dingoes: Pets or producers? Gunn RG (2003) Arrernte rock-art of Central Mankind 10: 11–15. Australia: Interpreting physical permanence in Hayden B (1977a) Stone tool functions in the a changing social landscape. Australian Aborig- Western Desert. In RVS Wright (ed) Stone inal Studies 2003/1: 52–73. tools as cultural markers: Change, evolution and Gunn RG (2004) The rock-art of Central Aus- complexity,pp.178–88. Canberra: Australian tralia: An overview. Australian Aboriginal Stud- Institute of Aboriginal Studies. ies 2004/1: 54–68. Hayden B (1977b) Sticks and stones and ground Haberlah D, Glasby P, Williams MAJ, Hill SM, edge axes: The Upper Palaeolithic in South- Williams F, Rhodes EJ et al. (2010) ‘Of east Asia. In J Allen, J Golson and R Jones (eds) droughts and flooding rains’: An alluvial loess Sunda and Sahul: Prehistoric studies in Southeast record from central South Australia spanning Asia, Melanesia and Australia,pp.73–109. Lon- the last glacial cycle. In P Bishop and B Pillans don: Academic Press. (eds), Australian landscapes,pp.185–223.Spe- Hayden B (1979) Palaeolithic reflections: Lithic tech- cial Publication 346. London: Geological nology and ethnographic excavations among Aus- Society of London. tralian Aborigines. Canberra: Australian Insti- Habgood PJ and Franklin NR (2008) The rev- tute of Aboriginal Studies. olution that didn’t arrive: A review of Pleis- Haydon DT, Friar JK, and Pianka ER (2000) tocene Sahul. Journal of Human Evolution 55: Fire-driven dynamic mosaics in the Great Vic- 187–222. toria Desert I: Fire geometry. Landscape Ecology Haffer J (1969) Speciation in Amazonian forest 15: 373–81. birds. Science 165: 131–7. Heatwole H (1987) Major components and dis- Haffer J (1982) General aspects of the refuge the- tributions of the terrestrial fauna. In GR Dyne ory. In GT Prance (ed) Biological diversification and DW Walton (eds) Fauna of Australia,Vol. in the tropics,pp.6–24. New York: Columbia 1A, pp. 101–35. Canberra: Australian Govern- University Press. ment Publishing Service. Hale HM and Tindale NB (1925) Observations Hercus LA (1980) ‘How we danced the Mud- on Aborigines of the Flinders Ranges and lunga’: Memories of 1901 and 1902. Aboriginal records of rock carvings and paintings. Records History 4: 4–31 of the South Australian Museum 3: 45–60. Hercus L (2005) A hole in the ground and Hale K (1964) Classification of northern Paman a quarry. In I Macfarlane, M-J Mountain languages, Cape York peninsula, Australia: and R Paton (eds) Many Exchanges: Archae- A research report. Oceanic Linguistics 3: 248– ology, history, community and the work of Isabel 65. McBryde,pp.185–98. Canberra: Aboriginal Hall FJ, McGowan RG and Guleksen GF (1951) History Monograph 11. Aboriginal rock carvings: A locality near Hercus L and Koch G (1997) Old yet always new: Pimba, S.A. Records of the South Australian Song traditions of southern central Australia. Museum 9: 374–80. Strehlow Research Centre Occasional Paper 1: 83– Hamilton A (1972) Aboriginal man’s best friend? 106. Mankind 8: 287–95. Hesse PP (1994) The record of continental dust Hamilton A (1980) Dual social systems: Tech- from Australia in Tasman Sea sediments. Qua- nology, labour and women’s secret rites in the ternary Science Reviews 13: 257–72. eastern Western Desert of Australia. Oceania Hesse PP (2010) The Australian desert dune- 51: 4–19. fields: Formation and evolution in an old, flat, Harrison R (2009) The archaeology of the dry continent. In P Bishop and B Pillans (eds) Port Hedland coastal plain and implications Australian landscapes,pp.141–64. Special Pub- for understanding the prehistory of shell lications. London: The Geological Society of mounds and middens in northwestern Aus- London. tralia. Archaeology in Oceania 44 (Supplement): Hesse PP, Luly JG and Magee, JW (2005) 81–98. The beating heart: Environmental history of 364 References

Australia’s deserts. In MA Smith and P Hesse Australia: A reader,pp.257–65. St Leonards, (eds) 23oS: Archaeology and environmental history NSW: Allen and Unwin. of the Southern Deserts,pp.56–72. Canberra: Hiscock P (2002) Pattern and context in National Museum of Australia Press. the Holocene proliferation of backed arte- Hesse PP, Magee JW and van der Kaars S (2004) facts in Australia. In RG Elston and SL Late Quaternary climates of the Australian Kuhn (eds) Thinking small: Global perspectives arid zone: A review. Quaternary International on microlithization,pp.163–77.Archaeolog- 118–119: 87–102. ical Papers of the American Anthropologi- Hesse PP and McTainsh GH (1999) Last glacial cal Association, Number 12. Malden, MA: maximum to early Holocene wind strength American Anthropological Association. in the mid-latitudes of the Southern Hemi- Hiscock P (2005) Standardised axe manufacture sphere from aeolian dust in the Tasman Sea. at Mount Isa. In I Macfarlane, M-J Mountain Quaternary Research 52: 343–9. and R Paton (eds) Many exchanges: Archae- Hesse PP and McTainsh GH (2003) Australian ology, history, community and the work of Isabel dust deposits: Modern processes and the Qua- McBryde,pp.287–300. Canberra: Aboriginal ternary record. Quaternary Science Reviews 22: History Monograph 11. 2007–35. Hiscock P and Allen H (2000) Assemblage vari- Hiatt LR (1962) Local organisation among the ability in the Willandra Lakes. Archaeology in Australian Aborigines. Oceania 32: 267–86. Oceania 35: 97–103. Hiatt LR (1996) Arguments about Aborigines: Aus- Hiscock P and Attenbrow V (1998) Early tralia and the evolution of social anthropology. Holocene backed artefacts from Australia. Cambridge: Cambridge University Press. Archaeology in Oceania 33: 49–63. Hickman B (2000) Song and dance over desert Hiscock P and Attenbrow V (2004) A revised land. The Australian, Wednesday 29 Novem- sequence of backed artefact production at ber, p. 7. Capertee 3. Archaeology in Oceania 39: 94–9. Higham CFW, Kijngam A and Manly BFJ Hiscock P and Hall HJ (1988) Technologi- (1980) An analysis of prehistoric canid remains cal change at Bushranger’s Cave (LA:A11), from Thailand. Journal of Archaeological Science southeast Queensland. Queensland Archaeolog- 7: 149–65. ical Research 5: 90–112. Hill B (2002) Broken song: TGH Strehlow and Hiscock P and Hughes PJ (1980) Backed blades Aboriginal possession. Sydney: Random House in Northern Australia: Evidence from North- Australia. west Queensland. Australian Archaeology 10: Hill RS (ed) (1994) History of the Australian veg- 86–95. etation: Cretaceous to Recent.NewYork:Cam- Hiscock P and Mitchell S (1993) Stone artefact bridge University Press. quarries and reduction sites in Australia: Towards Hills ES (1939) The physiography of north- a type proﬁle. Canberra: Australian Heritage western Victoria. Proceedings of the Royal Society Commission. Technical Publications Series 4. of Victoria 51: 297–323. Hiscock P and Veth P (1991) Change in the Aus- Hills ES (1940) The lunette, a new land form tralian desert culture: A reanalysis of tulas from of aeolian origin. The Australian Geographer 3: Puntutjarpa rockshelter. World Archaeology 22: 15–21. 332–45. Hiscock P (1988a) Prehistoric settlement patterns Hiscock P and Wallis LA (2005) Pleistocene set- and artefact manufacture at Lawn Hill, Northwest tlement of deserts from an Australian perspec- Queensland. Unpublished PhD thesis, Univer- tive. In P Veth, M Smith and P Hiscock (eds) sity of Queensland. Desert peoples: Archaeological perspectives,pp.34– Hiscock P (1988b) A cache of tulas from the 57. Oxford: Blackwell Publishers. Boulia district, western Queensland. Archaeol- Hodgkinson WO (1877) North-west explo- ogy in Oceania 23: 60–70. rations by W.O. Hodgkinson, Esq. Queensland Hiscock P (1994) Technological responses to risk Votes and Proceedings 3: 203–27, 218. in Holocene Australia. Journal of World Prehis- Holdaway S (1995) Stone artefacts and the tran- tory 8: 267–92. sition. Antiquity 69: 784–97. Hiscock P (1998) Revitalising artefact analysis. Holdaway SJ, Fanning PC, Jones M, Shiner J, In T Murray (ed) Archaeology of Aboriginal Witter DC and Nicholls G (2002) Variability References 365

in the chronology of late Holocene Aboriginal 183–96. Occasional Papers in Prehistory 24. occupation on the arid margin of southeast- Canberra: The Australian National Univer- ern Australia. Journal of Archaeological Science sity. 29: 351–63. Hope JH, Dare-Edwards A and McIntyre ML Holdaway SJ, Fanning P and Rhodes E (2008) (1983) Middens and megafauna: Stratigraphy Challenging intensification: Human-envi- and dating of Lake Tandou lunette, western ronment interactions in the Holocene geo- New South Wales. Archaeology in Oceania 18: archaeological record from western New 38–45. South Wales, Australia. The Holocene 18: 403– Horne G and Aiston G (1924) Savage life in Cen- 12. tral Australia. Macmillan: London. Holdaway SJ, Fanning P and Shiner J (2005) Horsfall N (1982) Stone tools from the Keppel Absence of evidence or evidence of absence? Islands. Australian Archaeology 14: 72–8. Understanding the chronology of Indigenous Horton DR (1981) Water and woodland: The occupation of western New South Wales, peopling of Australia. Australian Institute of Australia. Archaeology in Oceania 40: 33–49. Aboriginal Studies Newsletter 16: 21–7. Holdaway SJ, Fanning PC, Rhodes E, Marx Horton DR (1984) Red kangaroos: Last of the S, Floyd B and Douglass M (2010) Human Australian megafauna. In PS Martin and RG response to palaeoenvironmental change and Klein (eds) Quaternary extinctions: A prehistoric the question of temporal scale. Palaeogeography, revolution,pp.639–80. Tucson: University of Palaeoclimatology, Palaeoecology 292: 190–200. Arizona Press. Holdaway S and Stern N (2004) Arecordin Howchin W (1909) The geography of South Aus- stone: The study of Australia’sflaked stone artefacts. tralia, including the Northern Territory: Historical, Melbourne: Museum Victoria/Canberra: physical, political and commercial. Christchurch: Aboriginal Studies Press. Whitcombe and Tombs. Hollands C, Nanson GC, Jones BG, Bristow Howchin W (1921) On the occurrence of Abo- CS, Price DM and Pietsch TJ (2006) Aeolian- riginal stone implements of unusual types fluvial interaction: Evidence for Late Qua- in the tableland regions of central Australia. ternary channel change and wind-rift linear Transactions of the Royal Society of South Aus- dune formation in the northwestern Simpson tralia 45: 206–30. Desert, Australia. Quaternary Science Reviews Howitt AW (1862) AW Howitt’s diary [Victorian 25: 142–62. Exploring Party], 3 July 1862 to 21 August Holmgren M, Stapp P, Dickman CR, Gracia C, 1862. State Library of Victoria, MS13071, Graham S, Gutierrez´ JR et al. (2006)Asyn- Box 2085/6a, Item 4. Online version: http:// thesis of ENSO effects on drylands in Aus- www.burkeandwills.net.au/Journals/Howitt/ tralia, North America and South America. Howitt July 1862.htm Advances in Geosciences 6: 69–72. Howitt AW (1878) Appendix D. Notes on the Hook F, Sinclair L and Wright N (2012) Adden- Aborigines of Cooper’s Creek. In R Brough dum to: Sinclair L and Wright N (eds) Report Smyth, The Aborigines of Victoria: With notes of an Indigenous Archaeological Assessment of 19 relating to the habits of the natives of other parts Sites within Fortescue Metals Group’s Christmas of Australia and Tasmania,Vol.2,pp.300–9. Creek Mining and Infrastructure Phase 22 Project Melbourne: Government Printer. Area. Unpublished report for Fortescue Metals Howitt AW (1891) The Dieri and other kin- Group by Archae-aus, Perth. dred tribes of Central Australia. Journal of the Hope G, Kershaw AP, van der Kaars S, Xiangjun Anthropological Institute of Great Britain and Ire- S, Liew P-M, Heusser LE et al. (2004)His- land 20: 30–104. tory of vegetation and habitat change in the Howitt AW (1904) The native tribes of south- Austral-Asian region. Quaternary International east Australia. London: Macmillan [Facsimile 118–119: 103–26. edition, Aboriginal Studies Press, Canberra, Hope J (1993) Pleistocene archaeological sites 1996]. in the central Murray-Darling basin. In MA Hudjashov G, Kivisild T, Underhill PA, Endicott Smith, M Spriggs and B Fankhauser (eds) P, Sanchez JJ, Lin AA et al. (2007) Reveal- Sahul in review: Pleistocene archaeology in Aus- ing the prehistoric settlement of Australia by tralia, New Guinea and Island Melanesia,pp. Y chromosome and mtDNA analysis. PNAS 366 References

(Proceedings of the National Academy of Sciences) Jennings JN (1968) A revised map of the desert 104: 8726–30. dunes of Australia. Australian Geographer 10: Hughes PJ (1977) The geomorphology of archaeolog- 408–9. ical sites on the south coast of New South Wales. Jennings JN (1975) Desert dunes and estuarine Unpublished PhD thesis, University of New fill in the Fitzroy estuary (north-western Aus- South Wales. tralia). Catena 2: 215–62. Hughes P, Hiscock P, Sullivan M and Mar- Jercher M, Pring A, Jones PG and Raven MD wick B (2011) Archaeological investigations (1998) Rietveld X-Ray diffraction and X-Ray at Olympic Dam in arid northeast South Aus- fluorescence analysis of Australian Aboriginal tralia. Journal of the Anthropological Society of ochres. Archaeometry 40: 383–401. South Australia. 34: 21–37. Johnson BJ, Miller GH, Fogel ML, Magee JW, Hughes PJ and Lampert, RJ (1980) Pleistocene Gagan MK and Chivas AR (1999) 65,000 years occupation of the arid zone in south-east of vegetation change in Central Australia and Australia: Research prospects for the Cooper the Australian summer monsoon. Science 284: Creek–Strzelecki Desert region. Australian 1150–2. Archaeology 10: 53–67. Johnson C (2006) Australia’s mammal extinctions: Hughes PJ and Quartermaine G (1992) Investiga- A 50,000 year history. Port Melbourne: Cam- tions of Aboriginal archaeological sites in the Mesa bridge University Press. J development area, Pannawonica. Unpublished Johnson CN and Brook BW (2011) Recon- report to Robe River Iron Associates, Perth, structing the dynamics of ancient human pop- Western Australia. ulations from radiocarbon dates: 10000 years Hughes PJ, Quartermaine G and Harris J (2011) of population growth in Australia. Proceedings Pleistocene rockshelters J23 and J24, Mesa J, of the Royal Society B 278: 3748–54. Pilbara, Western Australia. Australian Archaeol- Johnson CN and Wroe S (2003) Causes of ogy 73: 58–61. extinction of vertebrates during the Holocene Hutchins AR (1998) Retreat to Tinga Tingana. of mainland Australia: Arrival of the dingo, or Unpublished BA (Hons) thesis, University of human impact? The Holocene 13: 941–8. New England. Johnson I (1979) The getting of data: A case study Jacobs SWL (1982) Relationships, distribu- from the recent industries of Australia. Unpub- tion and evolution of Triodia and Plechtra- lished PhD, Australian National University. chne (Gramineae). In WR Barker and PJM Johnson JE (1958) Manuscript of archaeology of Greenslade (eds) Evolution of the flora and S-W Centralian Desert. Unpublished ms, 2 fauna of arid Australia,pp.278–90. Australian March 1958.SeriesAA159/4 (Manuscripts Systematic Botany Society. Frewville, SA: and papers by JE Johnson) – South Australian Peacock Publications. Museum Archives. Jacobson G (1996) The interrelationship of Johnson JE (1964) Observations on some Abo- hydrogeology and landform in central Aus- riginal campsites in South Australia and tralia. In SR Morton and DJ Mulvaney (eds) adjoining States (Part 2). Mankind 6: 154– Exploring Central Australia: Society, the environ- 81. ment and the 1894 Horn Expedition,pp.249–66. Johnston H (1993) Pleistocene shell middens Chipping Norton (Sydney): Surrey Beatty. of the Willandra Lakes. In MA Smith, M Jacobson G, Calf GE, Jankowski J and McDonald Spriggs and B Fankhauser (eds) Sahul in review: PS (1989) Groundwater chemistry and palae- Pleistocene archaeology in Australia, New Guinea orecharge in the Amadeus basin, Central Aus- and Island Melanesia,pp.197–203. Canberra: tralia. Journal of Hydrology 109: 237–66. Department of Prehistory, Research School of Jennings JD (1957) Danger Cave. University of Pacific Studies, The Australian National Uni- Utah Anthropological Papers 27. Salt Lake versity. Occasional Papers in Prehistory 24. City: University of Utah Press. Johnston H, Clark P and White JP (eds) (1998) Jennings JD (1973) The short useful life of a Willandra Lakes: People and palaeoenviron- simple hypothesis. Te b iwa 13: 1–9. ments. Archaeology in Oceania 33 No. 3. Jennings JD and Norbeck E (1955)GreatBasin Johnston TH and Cleland JB (1933–4) The his- prehistory: A review. American Antiquity 21:1– tory of the Aboriginal narcotic, Pituri. Oceania 11. 4: 201–23, 268–89. References 367

Jones F Wood and Campbell TD (1925)A Kefous KC (1977)‘We have a fish with ears, and contribution to the study of eoliths: Some wonder if it’s valuable.’ Unpublished BA (Hons) observations on the natural forces at work in thesis, Australian National University. the production of flaked stones on the Cen- Kelly JD (1968) Hut sites, rock engravings, stone tral Australian Tablelands. Journal of the Royal arrangements, and tjurunga. Mulligan River, Anthropological Institute of Great Britain and Ire- Queensland. Mankind 6: 563–6. land 55: 115–22. Kelly RL (1995) The foraging spectrum: Diversity Jones P (2007) Ochre and rust: Artefacts and encoun- in hunter-gatherer lifeways. Washington: Smith- ters on Australian frontiers. Adelaide: Wakefield sonian Institution Press. Press. Kendrick GW and Morse K (1982)AnAbo- Jones R (1969) Fire-stick farming. Australian riginal shell midden deposit from the War- Natural History 16: 224–8. roora coast, northwestern Australia. Australian Jones R (1979) The fifth continent: Problems Archaeology 14: 6–12. concerning the human colonisation of Aus- Kendrick GW and Porter JK (1973) Remains tralia. Annual Review of Anthropology 8: 445– of a Thylacine (Marsupialia: Dasyuroidae) and 66. other fauna from caves in the Cape Range, Jones R (1984) Hunters and history: A case study Western Australia. Journal of the Royal Society from western Tasmania. In C Schrire (ed) Past of Western Australia 56: 116–22. and present in hunter gatherer studies,pp.27–65. Kershaw AP and van der Kaars S (2007)Qua- Orlando, Florida: Academic Press. ternary pollen records: Australia and New Jones R (1987) Pleistocene life in the dead heart Zealand. Encyclopedia of Quaternary Science, of Australia. Nature 328: 666. 2613–23. Jones W (1979) Up the creek: Hunter-gatherers in the Kershaw AP, van der Kaars S and Moss P (2003) Cooper basin. Unpublished BA (Hons) thesis, Late Quaternary Milankovitch-scale climatic University of New England. change and variability and its impact on mon- Jouzel J and Masson-Delmotte V (2007)Ice soonal Australia. Marine Geology 201: 81–95. core records – Antarctic stable isotopes. In KershawP,vanderKaarsS,MossP,Opdyke SA Elias (ed) Encyclopedia of Quaternary Science, B, Guichard F, Rule S and Turney C (2006) pp. 1242–50. Online: Elsevier. Environmental change and the arrival of peo- Jutson JT (1934) The physiography (geomorphology) ple in the Australian region. Before Farming of Western Australia. Western Australia Geolog- 2006/1(2): 1–24. ical Survey Bulletin 95.Perth:FWSimpson, Kimber R (1983) Black lightning: Aborigines Government Printer. and fire in Central Australia and the Western Kamminga J (1980) A functional investigation Desert. Archaeology in Oceania 18: 38–45. of Australian microliths. The Artefact 5:1– Kimber RG (1984) Resource use and manage- 18. ment in central Australia. Australian Aboriginal Keast A (1959) The Australian environment. In Studies 2: 12–23. A Keast, RL Crocker and CS Christian (eds) Kimber RG (1985) Comments on B Wright, Biogeography and ecology in Australia,pp.15–35. ‘The significance of hand motif variations in Den Haag, Netherlands: W Junk. the stencilled art of the Australian Aborigines’. Keble RA (1947) Notes on Australian Quater- Rock Art Research 2: 14–15. nary climates and migration. Memoirs of the Kimber RG (1988) Pre-European history. In N National Museum of Victoria 15: 28–81. Amadio and R Kimber (eds) Wildbird dream- Keeling JL (1984) Report Book No. 831. Bookartoo ing,pp.47–53. Melbourne: Greenhouse Pub- ochre deposit – Section 85, Hundred Parachilna. lications. DME 36/61. Adelaide: Department of Mines Kimber RG (1989) Journey to Alalya. Unpub- and Energy (South Australia). lished ms. Keeley LH (1982) Hafting and retooling: Effects Kimber RG (1990) Hunter-gatherer demogra- on the archaeological record. American Antiq- phy: The recent past in Central Australia. In uity 47: 798–809. B Meehan and N White (eds) Hunter-gatherer Keen I (2004) Aboriginal economy and society: Aus- demography: Past and present,pp.160–70.Syd- tralia at the threshold of colonisation. Melbourne: ney: University of Sydney. Oceania Mono- Oxford University Press. graph 39. 368 References

Kimber RG (2006) Kapi! – Contact experiences Kukla GJ, Bender ML, de Beaulieu J-L, Bond in the Western Desert 1873–1984. In P Batty G, Broecker WS, Cleveringa P et al. (2002) (ed) Colliding worlds: First contact in the West- Last interglacial climates. Quaternary Research ern Desert 1932–1984,pp.4–28 Melbourne: 58: 2–13. Museum Victoria. Kuklick H (2006) ‘Humanity in the chrysalis Kimber RG and Smith MA (1987)AnAranda stage’: Indigenous Australians in the anthropo- ceremony. In DJ Mulvaney and JP White (eds) logical imagination, 1899–1926. British Journal Australians to 1788,pp.220–37. Sydney: Fair- for the History of Science 39(4): 535–68. fax, Syme and Weldon. Lambeck K, Yokoyama Y and Purcell T (2002) King D (1956) The Quaternary stratigraphic Into and out of the Last Glacial Maxi- record at Lake Eyre North and the evolu- mum: Sea-level change during Oxygen Iso- tion of existing topographic forms. Transactions tope Stages 3 and 2. Quaternary Science Reviews of the Royal Society of South Australia 79: 93– 21: 343–60. 103. Lampert RJ and Hughes PJ (1987) The Flinders King PP (1827) Narrative of a survey of the intertrop- Ranges: A Pleistocene outpost in the arid ical and western coasts of Australia, performed zone? Records of the South Australian Museum between the years 1818 and 1822. London: John 20: 29–34. Murray. [Australiana facsimile editions, no. 30, Lampert RJ and Hughes PJ (1988) Early human Adelaide, Libraries Board of South Australia, occupation of the Flinders Ranges. Records of 1969] the South Australian Museum 22: 139–68. Kingsford RT, Curtin AL and Porter J (1999) Lancaster N (2008) Desert dune dynamics and Water flows on Cooper Creek in arid Aus- development: Insights from luminescence dat- tralia determine ‘boom’ and ‘bust’ periods ing. Boreas 37: 559–73. for waterbirds. Biological Conservation 88: 231– Lang A (1903) Social origins. London: Longmans 48. Green and Co. Kirk RL (1971) Genetic evidence and its impli- Lang A (1904) A theory of Arunta totemism. cations for Aboriginal prehistory. In DJ Mul- Man 4: 67–9. vaney and J Golson (eds) Aboriginal man and Lang A (1910) The ‘historicity’ of Arunta tradi- environment in Australia,pp.326–43. Canberra: tions. Man 10: 118–21. Australian National University Press. Latz P (1975) Notes on the relict palm Livistona Kitching RL (1981) The geography of the Aus- mariae F. Muell. in Central Australia. Transac- tralian Papilionoidea. In A Keast (ed) Ecologi- tions of the Royal Society of South Australia 99: cal biogeography of Australia,pp.977–1005. The 189–96. Hague: W Junk. Latz P (1995) Bushfires and bushtucker: Aborigi- Kocurek G (1998) Aeolian system response to nal plant use in Central Australia. Alice Springs: external forcing factors – a sequence strati- IAD Press. graphic view of the Saharan region. In AS Latz P (2007) The flaming desert: Arid Australia – Alsharhan, KW Glennie, GL Whittle and CG A fire shaped landscape. Alice Springs: Self- St C Kendall (eds) Quaternary deserts and climate published. change,pp.327–37. Rotterdam: AA Balkema. Latz PK and Griffin GF (1978) Changes in Abo- Kolig E (1978) Aboriginal dogmatics: Canines in riginal land management in relation to fire theory, myth and dogma. Bijdragen tot de Taal-, and to food plants in Central Australia. In BS Land- en Volkenkunde 134: 84–115. Hetzel and HJ Frith (eds) The nutrition of Abo- Kondo T, Crisp MD, Linde C, Bowman DMJS, rigines in relation to the ecosystem of Central Aus- Kawamura K, Kaneko S and Isagi Y (2012) tralia,pp.77–85. Melbourne: CSIRO. Not an ancient relic: The endemic Livistona Laughren M (1994) Warlpiri dictionary. Alice palms of arid central Australia could have been Springs: Institute for Aboriginal Develop- introduced by humans. Proceedings of the Royal ment. Society B, 279: 2652–61. Law WB, Cropper DN and Petchey F (2010) KotwickiV(1986) FloodsofLakeEyre. Adelaide: Djadjiling Rockshelter: 35,000 14Cyearsof Engineering and Water Supply Department. Aboriginal occupation in the Pilbara, West- Kretchmar WH (1936) ‘Wilgie Mia’: An Abo- ern Australia. Australian Archaeology 70: 68– riginal red ochre mine. Walkabout 2: 40–1. 71. References 369

Lawrence D (1994) Customary exchange across pp. 262–70. Canberra: Australian National Torres Strait. Memoirs of the Queensland University Press. Museum 34: 241–6. Long J, Archer M, Flannery T and Hand S Layton R (1986) Uluru: An Aboriginal history of (2002) Prehistoric mammals of Australia and New Ayers Rock. Canberra: Australian Institute of Guinea: One hundred million years of evolution. Aboriginal Studies. Baltimore: Johns Hopkins University Press. Layton R (1992) Australian rock art: A new synthe- Lorblanchet M (1977) Summary report of field- sis. Cambridge: Cambridge University Press. work,Dampier,WAAustralian Institute of Abo- Lee RB and DeVore I (eds) (1968) Man the hunter. riginal Studies Newsletter n.s. 7: 36–40. Chicago: Aldine. Lorblanchet M (1992) The rock engravings of Letnic M and Dickman CR (2006) Boom Gum Tree Valley and Skew Valley, Dampier, means bust: Interactions between the El Western Australia: Chronology and function Nino/Southern Oscillation (ENSO), rainfall of the sites. In J McDonald and IP Haskovic and the processes threatening mammal species (eds) State of the art: Regional rock art studies in in arid Australia. Biodiversity and Conservation Australia and Melanesia,pp.39–59. Occasional 15: 3847–80. AURA Publication No. 6. Melbourne: Aus- Letnic M, Dickman CR, Tischler MK, Tamayo tralian Rock Art Research Association. BandBehC-L(2004) The responses of small Lorblanchet M and Jones R (1979) Les premieres` mammals and lizards to post-fire succession fouilles a` Dampier (Australie occidentale), et and rainfall in arid Australia. Journal of Arid leur place dans l’ensemble Australien. Bulletin Environments 59: 85–114. de la Societe´ Prehistorique Francaise 76: 463–87. Lewis D (1988) The rock paintings of Arnhem Lourandos H (1983) Intensification: A late Land, Australia: Social, ecological and material Pleistocene-Holocene archaeological se- culture change in the post-glacial period.Oxford: quence from southwestern Victoria. Archaeol- Archaeopress. BAR International Series 415. ogy in Oceania 18: 81–94. LewisDandRoseD(1995) The shape of the Lourandos H (1985) Intensification and Aus- Dreaming: The cultural significance of Victoria tralian prehistory. In TD Price and JA Brown River rock art. Canberra: Aboriginal Studies (eds) Prehistoric hunter-gatherers: The emergence Press. of cultural complexity,pp.385–423. Orlando, Lewis JW (1875) Report on the Lake Eyre expe- Florida: Academic Press. dition. South Australian Parliamentary Papers Lourandos H (1997) Continent of hunter-gatherers: 1875/114. Adelaide: Government Printer. New perspectives in Australian prehistory.Cam- Lindsell P (2001) Bergmann, Allen and Birdsell: bridge: Cambridge University Press. Patterns of ecogeographic adaptation in Aboriginal Love JRB (1936) Stone-age bushmen of today: Life Australians. Unpublished PhD thesis, Univer- and adventures among a tribe of savages in North- sity of New England. Western Australia. London: Blackie and Son. Lioubimtseva E and Henebry GM (2009) Lowry JWJ and Merrilees D (1969)Ageofthe Review: Climate and environmental change desiccated carcase of a Thylacine (Marsupialia, in arid Central Asia: Impacts, vulnerability, Dasyuriodea) from Thylacine Hole, Nullarbor and adaptations. Journal of Arid Environments Region, Western Australia. Helictite 7: 15–16. 73: 963–77. Lubbock J (1875) Pre-historic times, as illustrated by Loffler¨ E and Sullivan ME (1979) Lake Dieri ancient remains, and the manners and customs of resurrected: An interpretation using satellite modern savages. New York: D Appleton & Co. imagery. Zeitschrift fur¨ Geomorphologie 23: 233– [Lubbock J] Lord Avebury (1912) Pre-historic 42. times, as illustrated by ancient remains, and the Lomax J, Hilgers A, Wopfner H, Grun¨ R, manners and customs of modern savages. Sixth edi- Twidale CR and Radtke U (2003) The onset tion. London: D. Williams & Norgate. of dune formation in the Strzelecki Desert, Luly JG (2001) On the equivocal fate of Late South Australia. Quaternary Science Reviews 22: Pleistocene Callitris Vent.(Cupressaceae) wood- 1067–76. lands in arid South Australia. Quaternary Inter- Long JPM (1971) Arid region Aborigines: The national 83–85: 155–68. Pintupi. In DJ Mulvaney and J Golson (eds) Mabbutt JA (1971a) The Australian arid zone as a Aboriginal man and environment in Australia, prehistoric environment. In DJ Mulvaney and 370 References

J Golson (eds) Aboriginal man and environment Colless Creek archaeological site in north- in Australia,pp.66–79. Canberra: Australian western Queensland. In W Ambrose and P National University Press. Duerden (eds) Archaeometry: An Australasian Mabbutt JA (1971b) Denudation chronology in perspective,pp.120–8. Canberra: Australian Central Australia: Structure, climate and land- National University. form inheritance in the Alice Springs area. In Magee JW and Miller GH (1998)LakeEyre JN Jennings and JA Mabbutt (eds) Landform palaeohydrology from 60 ka to the present: studies from Australia and New Guinea,pp.144– Beach ridges and glacial maximum aridity. 81. Canberra: Australian National University Palaeogeography, Palaeoclimatology, Palaeoecology Press. 144: 307–29. Mabbutt JA (1977) Desert landforms: An introduc- Magee JW, Miller GH, Spooner NA and tion to systematic geomorphology,Vol.2.Can- Questiaux D (2004) A continuous 150 ky berra: Australian National University Press. monsoon record from Lake Eyre, Australia: Macaulay V, Hill C, Achilli A, Rengo C, Clarke Insolation-forcing implications and unex- D, Meehan W et al. (2005) Single, rapid pected Holocene failure. Geology 32: 885–8. coastal settlement of Asia revealed by analysis Main AR and Bakker HR (1981) Adaptation of of complete mitochondrial genomes. Science macropod marsupials to aridity.In A Keast (ed) 308: 1034–6. Ecological biogeography of Australia,pp.1490– Macintosh NWG (1971) Analysis of an Abo- 519. The Hague: W Junk. riginal skeleton and a pierced tooth necklace Maroulis JC, Nanson GC, Price DM and Pietsch from Lake Nitchie, Australia. Anthropologie 9: T(2007) Aeolian–ﬂuvial interaction and cli- 49–62. mate change: Source-bordering dune devel- Macintosh NWG, Smith KN and Bailey AB opment over the past 100 ka on Cooper (1970) Lake Nitchie skeleton – unique Abo- Creek, central Australia. Quaternary Science riginal burial. Archaeology and Physical Anthro- Reviews 26: 386–404. pology in Oceania 5: 85–101. Martin HA (1973) Palynology and historical Macphail MK, Alley NF, Truswell EM and ecology of some cave excavations in the Aus- Sluiter IRK (1994) Early Tertiary vegetation: tralian Nullarbor. Australian Journal of Botany Evidence from spores and pollen. In RS Hill 21: 283–316. (ed) History of the Australian vegetation: Creta- Martin HA (1990) The palynology of the ceous to recent,pp.189–261.NewYork:Cam- Namba Formation in the Wooltana-1 bore, bridge University Press. Callabonna Basin (Lake Frome), South Aus- Madigan CT (1936) The Australian sand-ridge tralia, and its bearing on grasslands in central deserts. Geographical Review 26: 205–27. Australia. Alcheringa 14: 247–55. Magee JW (1997) Late Quaternary environments Martin HA (2006) Cenozoic climatic change and and palaeohydrology of Lake Eyre, arid central the development of the arid vegetation in Aus- Australia. Unpublished PhD thesis, Australian tralia. Journal of Arid Environments 66: 533–63. National University. Martinson DG, Pisias NG, Hays JD, Imbrie J, MageeJ(2007) Lake level studies – Australia. In Moore TC and Shackleton NJ (1987)Age SA Elias (ed) Encyclopedia of Quaternary Science, dating and the orbital theory of the ice pp. 1359–66. Online: Elsevier. ages: Development of a high-resolution 0 to MageeJW(2009) Palaeovalley groundwater resources 300,000-year chronostratigraphy. Quaternary in arid and semi-arid Australia: A literature Research 27: 1–29. review. Geoscience Australia Record 2009/03. Marun LH (1972) The Mirning and their predeces- Canberra. sors on the coastal Nullarbor plain. Unpublished Magee JW, Bowler JM, Miller GH and PhD thesis, University of Sydney. Williams DLG (1995) Stratigraphy, sedimen- Marwick B (2002a) Inland Pilbara archaeology: A tology, chronology and palaeohydrology of study of variation in Aboriginal occupation over time Quaternary lacustrine deposits at Madigan and space on the Hamersley plateau. Unpublished Gulf, Lake Eyre, South Australia. Palaeogeogra- MA thesis, University of Western Australia. phy, Palaeoclimatology, Palaeoecology, 113: 3–42. Marwick B (2002b) Milly’s Cave: Evidence for Magee JW and Hughes PJ (1982) Thin-section human occupation of the inland Pilbara during analysis and the geomorphic history of the the Last Glacial Maximum. Te m p us 7: 21–33. References 371

Marwick B (2005) Element concentrations and McBrearty S and Brooks AS (2000) The revolu- magnetic susceptibility of anthrosols: Indica- tion that wasn’t: A new interpretation of the tors of prehistoric human occupation in the origin of modern human behavior. Journal of inland Pilbara, Western Australia. Journal of Human Evolution 39: 453–63. Archaeological Science 32: 1357–68. McBryde I (1977) Determinants of assemblage Marwick B (2009) Change or decay? An inter- variation in New England prehistory: Envi- pretation of late Holocene archaeological evi- ronment, subsistence economies, site activ- dence from the Hamersley plateau, Western ities, or cultural tradition? In RVS Wright Australia. Archaeology in Oceania 44 (Supple- (ed) Stone tools as cultural markers,pp.225–50. ment): 16–22. Canberra: Australian Institute of Aboriginal Marx SK, McGowan HA and Kamber BS (2009) Studies. Long-range dust transport from eastern Aus- McBryde I (1978) Wil-im-ee Moor-ring: Or, tralia: A proxy for Holocene aridity and where do axes come from? Mankind 11: 354– ENSO-type climate variability. Earth and Plan- 82. etary Science Letters 282: 167–77. McBryde I (1982) Coast and estuary: Archaeo- Masey TA (1882) The red ochre caves of the logical investigations on the north coast of New blacks. The Port Augusta Despatch and Flinders South Wales at Wombah and Schnapper Point. Advertiser, 9 June 1882, 3. Canberra: Australian Institute of Aboriginal Masson V, Vimeux F, Jouzel J, Morgan V, Del- Studies. motte M, Ciais P et al. (2000). Holocene cli- McBryde I (1987) Goods from another coun- mate variability in Antarctica based on 11 ice- try: Exchange networks and the people of core isotopic records. Quaternary Research 54: the Lake Eyre Basin. In DJ Mulvaney and JP 348–58. White (eds), Australians to 1788,pp.252–73. Masson-Delmotte V, Stenni B, Pol K, Bracon- Broadway, NSW: Fairfax, Syme and Weldon not P, Cattani O, Falourd S et al. (2010) Associates. EPICA Dome C record of glacial and inter- McBryde I (1997) ‘The landscape is a series of glacial intensities. Quaternary Science Reviews stories’: Grindstones, quarries and exchange 29: 113–28. in Aboriginal Australia: A Lake Eyre case Mathew J (1889) The Australian Aborigines. study. In A Ramos-Millan and A. Bustillo Royal Society of New South Wales – Journal and (eds) Siliceous rocks and culture,pp.587–607. Proceedings 23: 335–449. Granada: University of Granada. Mayewski PA, Rohling EE, Stager JC, Karlen´ McBryde I (2000) Travellers in storied land- W, Maascha KA, Meekere LD et al. scapes: A case study in exchanges and heritage. (2004) Holocene climate variability. Quater- Aboriginal History 24: 152–74. nary Research 62: 243–55. McBryde I and Watchman A (1976) The dis- Maynard L (1976) An archaeological approach to the tribution of greenstone axes in southeastern study of Australian rock art. Unpublished MA Australia: A preliminary report. Mankind 10: thesis, University of Sydney. 163–74. Maynard L (1979) The archaeology of Australian McCarthy FD (1938–40) ‘Trade’ in Aboriginal Aboriginal art. In SM Mead (ed) Exploring the Australia, and ‘trade’ relationships with Torres visual art of Oceania,pp.83–109. Honolulu: Strait, New Guinea and Malaya. Oceania 9: University Press of Hawaii. 405–38, 10: 80–104, 171–95. Maynard L (1980) A Pleistocene date from McCarthy FD (1958) Australian Aboriginal rock an occupation deposit in the Pilbara region, art. Sydney: Australian Museum. Western Australia. Australian Archaeology 10: McCarthy FD (1961) The rock engravings of 3–8. Depuch Island, North-west Australia. Records Maynard L and Edwards R (1971) Wall mark- of the Australian Museum 25: 121–48. ings. In RVS Wright (ed) Archaeology of the McCarthy FD (1962) The rock engravings at Gallus site, Koonalda Cave,pp.61–80. Aus- Port Hedland, Northwestern Australia. Kroe- tralian Aboriginal Studies No. 26. Canberra: ber Anthropological Society Papers 26: 1–70. Australian Institute of Aboriginal Studies. McCarthy FD (1976a) Australian Aboriginal stone Mawson D (1950) Occurrence of water in Lake implements. Second edition. Sydney: The Aus- Eyre, South Australia. Nature 166: 667–8. tralian Museum. 372 References

McCarthy FD (1976b) Rock art of the Cobar McDonald J and Veth P (2006)Rockartand pediplain in central western New South Wales. social identity: A comparison of graphic sys- Canberra: Australian Institute of Aboriginal tems operating in arid and fertile environ- Studies. ments in the Holocene. In I Lilley (ed) Archae- McCarthy FD and Macintosh NWG (1962) The ology of Oceania: Australia and the Paciﬁc Islands, archaeology of Mootwingee, western New pp. 96–115. Oxford: Blackwell. South Wales. Records of the Australian Museum McDonald J and Veth P (2008) Rock-art of the 25: 249–98. Western Desert and Pilbara: Pigment dates McCarthy L and Head L (2001) Holocene vari- provide new perspectives on the role of art in ability in semi-arid vegetation: New evidence the Australian arid zone. Australian Aboriginal from Leporillus middens from the Flinders Studies 2008/1: 4–21. Ranges, South Australia. The Holocene 11: McDonald J and Veth P (2011) Information 681–9. exchange among hunter-gatherers of the McCarthy L, Head L and Quade J (1996) Western Desert of Australia. In R Whallon, Holocene palaeoecology of the northern WA Lovis and RK Hitchcock (eds) Information Flinders Ranges, South Australia, based on and its role in hunter-gatherer bands,pp.221–93. stick-nest rat (Leporillus spp.) middens: A pre- UCLA Berkeley CA: The Cotsen Institute of liminary overview. Palaeogeography, Palaeocli- Archaeology. matology, Palaeoecology 123: 205–18. McIntyre ML and Hope JH (1978) Procoptodon McCaskill DL (1968) Aboriginal rock engravings fossils from the Willandra Lakes, western at Boom Boom spring, Waldburg Range. The New South Wales. The Artefact 3: 117– Western Australian Naturalist 11: 25–39. 32. McConnell K (1998) The prehistoric use of McKinlay J (1862) McKinlay’s journal of exploration Chenopodiaceae in Australia: Evidence from in the interior of Australia: Burke Relief Expedi- Carpenter’s Gap shelter 1 in the Kimberley, tion. Melbourne: FF Bailliere. Australia. Vegetation History and Archaeobotany McLaren S and Wallace MW (2010) Plio- 7: 179–88. Pleistocene climate change and the onset of McConnell K and O’Connor S (1997) 40,000- aridity in southeastern Australia. Global and year record of food plants in the southern Planetary Change 71: 55–72. Kimberley Ranges, Western Australia. Aus- McNickle HP (1984) Variation in style and dis- tralian Archaeology 45: 20–31. tribution of rock engravings in the Pilbara McConnochie KR (1996) The archaeology of the region (Western Australia). Rock Art Research Coongie Lakes area. Unpublished MLitt thesis, 1: 5–24. University of New England. McNiven IJ (1988) Brooyar Rockshelter: A McConvell P (1985) The origin of subsections late Holocene seasonal hunting camp from in northern Australia. Oceania 56: 1–33. southeast Queensland. Queensland Archaeolog- McConvell P (1996) Backtracking to Babel: The ical Research 5: 133–60. chronology of Pama-Nyungan expansion in McNiven IJ (1993) Tula adzes and bifacial points Australia. Archaeology in Oceania 31: 125–44. on the east coast of Australia. Australian Archae- McDonald J (1993) The depiction of species in ology 36: 22–33. macropod track engravings at an Aboriginal McNiven IJ (2000) Backed to the Pleistocene. art site in western New South Wales. Records of Archaeology in Oceania 35: 48–52. the Australian Museum (Supplement) 17: 105– Meehan B (1982) Shell bed to shell midden. 11. Canberra: Australian Institute of Aboriginal McDonald J (2005) Archaic faces to headdresses: Studies. The changing role of the rock art across the Meggitt MJ (1962) Desert people: A study of the arid zone. In P Veth, M Smith and P Hiscock Walbiri Aborigines of Central Australia. Sydney: (eds) Desert peoples: Archaeological perspectives, Angus and Robertson. pp. 116–41. Oxford: Blackwell. Megirian D, Murray PF, Latz PK and Johnson McDonald JM, Donlon D, Field JH, Fullagar KA (2002) The Mygoora local fauna: A late RLK, Coltrain JB, Mitchell P and Rawson Quaternary vertebrate assemblage from Cen- M(2007) The ﬁrst archaeological evidence tral Australia. The Beagle: Records of the Muse- for death by spearing in Australia. Antiquity 8: ums and Art Galleries of the Northern Territory 877–85. 18: 77–93. References 373

Megirian D, Murray P and Wells R (1996) The Moorehead A (1985) Cooper’s Creek: The story of late Miocene Ongeva Local Fauna of Central Burke and Wills. Melbourne: Nelson (paper- Australia. The Beagle: Records of the Museums and back edition). Art Galleries of the Northern Territory 13: 9–38. Moros M, De Deckker P, Eystein J, Perner Merrilees D (1970) A check on the radiocar- K and Telford RJ (2009) Holocene climate bon dating of desiccated Thylacine (Marsupial variability in the Southern Ocean recorded ‘wolf ’) and dog tissue from Thylacine Hole, in a deep-sea sediment core off South Aus- Nullarbor Region Western Australia. Helictite. tralia. Quaternary Science Reviews 28: 1932– 8: 39–42. 40. Middleton NJ and Thomas DSG (1997) World Morphy H (1989) From dull to brilliant: atlas of desertiﬁcation. Second edition. London: The aesthetics of spiritual power among the Arnold. Yo l n g u . Man n.s. 24: 21–40. Mildrexler DJ, Zhang M and Running SW Morphy H (1997) Gillen – Man of science. In DJ (2006) Where are the hottest spots on Earth? Mulvaney, H Morphy and A Petch (eds) ‘My Eos – Transactions, American Geophysical Union Dear Spencer’: The letters of FJ Gillen to Bald- 87: 461. win Spencer,pp.23–50. Melbourne: Hyland Milham P and Thompson P (1976) Relative House. antiquity of human occupation and extinct Morphy H (1998). Aboriginal art. Art and Ideas fauna at Madura cave, southeastern Western series. London: Phaidon. Australia. Mankind 10: 175–80. Morse K (1988) Mandu Mandu Creek rockshel- Miller AH (1963) The fossil ﬂamingos of Aus- ter: Pleistocene human coastal occupation of tralia. The Condor 65: 289–99. North West Cape, Western Australia. Archae- Miller GH, Fogel ML, Magee JW, Gagan MK, ology in Oceania 23: 81–8. Clarke SJ and Johnson BJ (2005) Ecosystem Morse K (1993a) West Side story: Towards a prehis- collapse in Pleistocene Australia and a human tory of the Cape Range Peninsula, Western Aus- role in megafauna extinction. Science 309: 287– tralia. Unpublished PhD thesis, University of 90. Western Australia. Miller GH, Magee JW, Johnson BJ, Fogel ML, Morse K (1993b) Shell beads from Mandu Spooner NA, McCulloch MT and Ayliffe Mandu Creek rock-shelter, Cape Range LK (1999) Pleistocene extinction of Geny- peninsula, Western Australia, dated before ornis newtoni: Human impact on Australian 30,000 b.p. Antiquity 67: 877–83. megafauna. Science 283: 205–8. Morse K (1993c) New radiocarbon dates from Miller GH, Magee JW and Jull AJT (1997)Low- North West Cape, Western Australia: A pre- latitude glacial cooling in the Southern Hemi- liminary report. In MA Smith, M Spriggs sphere from amino-acid racemization in emu and B Fankhauser (eds) Sahul in review: Pleis- eggshells. Nature 385: 241–4. tocene archaeology in Australia, New Guinea Miller GH, Mangan J, Pollard D, Thompson and Island Melanesia,pp.155–63. Occasional S, Felzer B and Magee J (2005) Sensitivity of Papers in Prehistory 24. Canberra: The Aus- the Australian monsoon to insolation and veg- tralian National University. etation: Implications for human impact on con- Morse K (2009) Introduction: Emerging from tinental moisture balance. Geology 33: 65–8. the abyss – archaeology in the Pilbara region Mitchell PJ (1996) Prehistoric exchange and of Western Australia. Archaeology in Oceania interaction in southeastern Southern Africa: 44: 1–5. Marine shells and ostrich eggshell. African Morton SR, Short J and Barker RD (1995) Refu- Archaeological Review 13: 35–76. gia for biological diversity in arid and semi-arid Mitchell SR (1949) Stone-Age Craftsmen: Stone Australia. Biodiversity Series Paper 4. Canberra: tools and camping places of the Australian Aborig- Biodiversity Unit, Department of Environ- ines. Melbourne: Tait. ment, Sport and Territories, Commonwealth Mix AC, Bard E and Schneider R (2001)Envi- of Australia. ronmental processes of the ice age: Land, Morton SR, Stafford Smith DM, Dickman CR, ocean, glaciers (EPILOG). Quaternary Science Dunkerley DL, Friedel MH, McAllister RRJ Reviews 20: 627–57. et al. (2011) A fresh framework for the ecology MooreMW(2004) The Tula adze: Manufacture of arid Australia. Journal of Arid Environments and purpose. Antiquity 78: 61–73. 75: 313–29. 374 References

Morwood MJ (1979) Art and stone: Towards a timescales during the Holocene epoch. Nature prehistory of central western Queensland. Unpub- 420: 162–5. lished PhD thesis, Australian National Uni- Mulvaney DJ (1960) Archaeological excavations versity, Canberra. at Fromm’s Landing on the Lower Murray Morwood MJ (1981) Archaeology of the central River, SA Proceedings of the Royal Society of Vic- Queensland highlands: The stone component. toria 72: 53–85. Archaeology in Oceania 16: 1–52. Mulvaney DJ (1961) The stone-age of Aus- Morwood MJ (1984). The prehistory of the cen- tralia. Proceedings of the Prehistoric Society 27: 56– tral Queensland highlands. Advances in World 107. Archaeology 3: 325–80. Mulvaney DJ (1964) The Australian Aborigines Morwood M (1988) Comment on FD 1606–1929: Opinion and fieldwork. Parts I McCarthy ‘Rock art sequences: A matter of and II. In JJ Eastwood and FB Smith (compl.) clarification’. Rock Art Research 5: 32–3. Historical studies, Australia and New Zealand: Morwood MJ (2002) Visions from the past: The Selected articles,pp.1–56. First series. Mel- archaeology of Australian Aboriginal Art.Crows bourne: Melbourne University Press. Nest (Sydney): Allen and Unwin. Mulvaney DJ (1969) The prehistory of Australia. Morwood MJ and Godwin L (1982)Aboriginal First edition. London: Thames and Hudson. sites in the Hughenden region, north Queens- Mulvaney DJ (1975) The prehistory of Australia. land highlands: Research prospects. Australian Revised edition. Ringwood, Vic.: Penguin Archaeology 15: 49–53. Books. Morwood MJ and Hobbs DR (1995) Themes in Mulvaney DJ (1976) ‘The chain of connec- the prehistory of tropical Australia. Antiquity tion’: The material evidence. In N Peterson 69: 747–68. (ed) Tribes and boundaries in Australia,pp.72– Mountford CP (1929a) Aboriginal rock carvings 94. Social Anthropology Series 10. Canberra: in South Australia. Report – Australian and New Australian Institute of Aboriginal Studies. Zealand Association for the Advancement of Science Mulvaney DJ (1985) Australian backed blade 1929: 337–66. industries in perspective. In VN Misra and Mountford CP (1929b) A unique example P Bellwood (eds) Recent advances in Indo-Pacific of Aboriginal rock carving at Panaramittee prehistory: Proceedings of the international sym- north. Transactions of the Royal Society of South posium held at Poona, December 19–21, 1978, Australia 53: 245–8. pp. 211–17. New Delhi: Oxford and IBH. Mountford CP (1955) An unrecorded method of Mulvaney DJ and Calaby JH (1985) ‘So much that Aboriginal rock marking. Records of the South is new’, Baldwin Spencer 1860–1929: A biography. Australian Museum 11: 345–52. Melbourne: Melbourne University Press. Mountford CP (1960) Simple rock engravings in Mulvaney DJ and Joyce EB (1965)Archaeolog- Central Australia. Man 60: 145–7. ical and geomorphological investigations on Mountford CP (1965) Ayers Rock: Its people, their Mt. Moffat Station, Queensland, Australia. beliefs and their art. Honolulu: East-West Cen- Proceedings of the Prehistoric Society 31: 147–212. tre Press. Mulvaney DJ and Kamminga J (1999) Prehis- Mountford CP (1968) Winbaraku and the myth of tory of Australia. St Leonards NSW: Allen and Jarapiri. Adelaide: Rigby. Unwin. Mountford CP (1976) Nomads of the Australian Mulvaney DJ, Lawton GH and Twidale CR Desert. Adelaide: Rigby. (1964) Archaeological excavation of rockshel- Mountford CP and Edwards R (1962)Aborig- ter No. 6, Fromm’s Landing, South Australia. inal rock engravings of extinct creatures in Proceedings of the Royal Society of Victoria 77: South Australia. Man 62: 97–9. 479–516. Mountford, CP and Edwards R (1964)Rock Mulvaney DJ, Morphy H and Petch A (eds) engravings in the Red Gorge, Deception (1997)‘My Dear Spencer’: The letters of FJ Creek, northern South Australia. Anthropos Gillen to Baldwin Spencer. Melbourne: Hyland 59: 849–59. House. Moy CM, Seltzer GO, Rodbell DT and Ander- Mulvaney KJ and Gunn RG (1995)‘Likeanews- son DM (2002) Variability of El Nino/˜ paper’. A report on the sacred and archaeologi- Southern Oscillation activity at millennial cal site Kurutiti for the Aboriginal custodians and References 375

Australian Heritage Commission. Unpublished Palaeogeography, Palaeoclimatology, Palaeoecology report, Aboriginal Areas Protection Author- 113: 87–102. ity, Darwin. Nanson GC, Price DM, Jones BG, Maroulis JC, Mulvaney KJ (1997) More than a chip off the old Coleman M, Bowman H et al. (2008) Alluvial block. A prehistoric sandstone quarry: The notion evidence for major climate and flow regime of production and exchange. Unpublished MA changes during the middle and late Quater- thesis, Northern Territory University. nary in eastern central Australia. Geomorphol- Mulvaney KJ (2010) Murujuga Marni-Dampier ogy 101: 109–29. petroglyphs: Shadows in the landscape, echoes across Nanson GC, Price DM and Short SA (1992) time. Unpublished PhD thesis, University of Wetting and drying of Australia over the past New England. 300 ka. Geology 20: 791–4. Munn ND (1986) Walbiri iconography: Graphic Nanson GC, Young RW, Price DM and Rust representation and cultural symbolism in a Cen- BR (1988) Stratigraphy, sedimentology and tral Australian society. Second edition. Chicago: late Quaternary chronology of the Chan- University of Chicago Press. nel Country of western Queensland. In RF Murgatroyd W (1991) Djaperi: A prestige item in Warner (ed) Fluvial geomorphology of Australia, Aboriginal exchange. Unpublished BA (Hons) pp. 151–75. Sydney: Academic Press. thesis, Northern Territory University. Napton LK and Greathouse EA (1985)Archaeo- Murphy BP, Williamson GJ and Bowman DMJS logical investigations at Pine Gap (Kuyunba), (2012) Did central Australian megafaunal Northern Territory. Australian Archaeology 20: extinction coincide with abrupt ecosystem 90–108. collapse or gradual climate change? Global Napton K and Greathouse E (1996)Archaeo- Ecology and Biogeography 21: 142–51. logical investigations in the Simpson Desert, Murray P (1984) Extinctions Downunder: A Northern Territory. In P Veth and P His- bestiary of extinct Australian late Pleistocene cock (eds) Archaeology of northern Australia,pp. monotremes and marsupials. In PS Martin and 106–36,Tempus4, University of Queensland: RG Klein (eds) Quaternary extinctions: A prehis- St Lucia (Brisbane). toric revolution,pp.600–28. Tucson: The Uni- Nelson DJ, Webb RH and Long A (1990) Anal- versity of Arizona Press. ysis of stick-nest rat (Leporillus: Muridae) mid- Murray PF and Vickers-Rich P (2004) Magnifi- dens from central Australia. In JL Betancourt, cent Mihirungs: The colossal flightless birds of the TRV Devenden, and PS Martin (eds) Packrat Australian Dreamtime. Bloomington: Indiana middens: The last 40,000 years of biotic change, University Press. pp. 428–34. Tucson: University of Arizona Murray-Wallace CV (2007) Sea level studies – Press. Eustatic sea-level changes since the last glacia- Newnham RM, Lowe DJ, Giles T and Alloway tion. In SA Elias (ed) Encyclopedia of Quaternary BV (2007) Vegetation and climate of Auck- Science,pp.3034–43. Online: Elsevier. land, New Zealand, since ca. 32 000 cal. yr Myers FR (1986) Pintupi country, Pintupi self. ago: Support for an extended LGM. Journal of Canberra: Australian Institute of Aboriginal Quaternary Science 22: 517–34. Studies. Nicholson AF (1994) Archaeology on an arid coast: Myers FR (1988) Locating ethnographic prac- Environmental and cultural influences on subsis- tice: Romance, reality and politics in the Out- tence economies on the west coast of South Australia. back. American Ethnologist 15: 609–24. Unpublished MA thesis, Australian National Myers FR (1993) Place, identity, and exchange University. in a totemic system: Nurturance and the pro- Nobbs MF (1983) A preliminary report on cess of social reproduction in Pintupi society. 7 radiocarbon age determinations from In J Fajans (ed) Exchanging products: Producing Plumbago Historic Reserve. Journal of the exchange,pp.33–57. Oceania monographs 33. Anthropological Society of South Australia 21: Sydney: University of Sydney. 3–7. Nanson GC, Chen XY and Price DM (1995) Nobbs M (1984) Rock art in Olary province, Aeolian and fluvial evidence of changing cli- South Australia. Rock Art Research 1: 91–118. mate and wind patterns during the past 100 Nobbs MF and Dorn RI (1988) Age deter- ka in the western Simpson Desert, Australia. minations for rock varnish formation within 376 References

petroglyphs: Cation-ratio dating of 24 motifs O’Connor S and Fankhauser B (2001)Artat from the Olary region, South Australia. Rock 40,000 BP? One step closer: An ochre cov- Art Research 5: 108–46. ered rock from Carpenter’s Gap Shelter 1, Nobbs M and Dorn RI (1993)Newsurface Kimberley Region, Western Australia. In A exposure ages for petrogylphs from the Olary Anderson, I Lilley and S O’Connor (eds) His- province, South Australia. Archaeology in Ocea- tories of old ages: Essays in honour of Rhys Jones, nia 28: 18–39. pp. 287–300. Canberra: Pandanus Books. O’Connell JF (1974) Spoons, knives and scrap- O’Connor S, Ono R and Clarkson C (2011) ers: The function of Yilugwa in central Aus- Pelagic fishing at 42,000 years before the tralia. Mankind 9: 189–94. present and the maritime skills of modern O’Connell JF (1977) Aspects of variation in humans. Science 334: 1117–21. central Australian lithic assemblages. In RVS O’Connor S and Veth P (1996) A Preliminary Wright (ed) Stone tools as cultural markers: report on recent archaeological research in the Change, evolution and complexity,pp.269–81. semi-arid/arid belt of Western Australia. Aus- Canberra: Australian Institute of Aboriginal tralian Aboriginal Studies 1996/2: 42–50. Studies. O’Connor S and Veth P (2000) The world’s first O’Connell JF and Allen FJ (2004) Dating the mariners: Savannah dwellers in an island con- colonization of Sahul (Pleistocene Australia- tinent. In S O’Connor and P Veth (eds) East New Guinea): A review of recent research. of Wallace’s Line: Studies of past and present mar- Journal of Archaeological Science 31: 835–53. itime cultures of the Indo-Pacific region,pp.99– O’Connell J and Allen J (2012) The restaurant 137. Rotterdam: Balkema. at the end of the universe: Modelling the col- O’Connor S and Veth P (2006) Revisiting the onization of Sahul. Australian Archaeology 74: past: Changing interpretations of Pleistocene 5–16. settlement, subsistence and demography in O’Connell JF and Hawkes K (1981) Alyawara northern Australia. In I Lilley (ed) Archaeol- plant use and optimal foraging theory. In ogy of Oceania, Australia and the Pacific Islands, B Winterhalder and E Smith (eds) Hunter- pp. 31–47. Oxford: Blackwell. gatherer foraging strategies: Ethnographic and O’Connor S, Veth P and Barham AJ (1999)Cul- archaeological analyses,pp.99–125. Chicago: tural versus natural explanations for lacunae University of Chicago Press. in Aboriginal occupation deposits in North- O’Connell JF, Latz P and Barnett P (1983)Tra- ern Australia. Quaternary International 59: 61– ditional and modern plant use among the 70. Alyawara of Central Australia. Economic Botany O’Connor S, Veth P and Campbell C (1998)Ser- 37: 80–109. pent’s Glen Rockshelter: Report of the first O’Connell JF and Madsen DB (1982) Man and Pleistocene-aged occupation sequence from environment in the Great Basin: An introduc- the Western Desert. Australian Archaeology 46: tion. In DB Madsen and JF O’Connell (eds) 12–22. Man and environment in the Great Basin,pp.1– O’Connor S, Veth P and Hubbard N (1993) 7.SAAPapers2. Washington DC: Society for Changing interpretations of postglacial human American Archaeology. subsistence and demography in Sahul. In O’Connor S (1995) Carpenter’s Gap Rockshel- MA Smith, M Spriggs and B Fankhauser ter 1: 40,000 years of Aboriginal occupation (eds) Sahul in review: Pleistocene archaeology in the Napier Ranges, Kimberley, WA. Aus- in Australia, New Guinea and Island Melane- tralian Archaeology 40: 58–9. sia,pp.95–105. Occasional Papers in Prehis- O’Connor S (1999a) 30,000 years of Aborigi- tory 24. Canberra: The Australian National nal occupation: Kimberley, North West Australia. University. Terra Australis 14. Canberra: ANH Publica- Odum EP (1971) Fundamentals of ecology. tions, The Australian National University. Third edition. Saunders College Publishing, O’Connor S (1999b) A diversity of coastal Philadelphia. economies: Shell mounds in the Kimberley O’Grady GN (1979) Preliminaries to a proto- region in the Holocene. In J Hall and IJ nuclear-Pama Nyungan stem list. In SA Wurm McNiven (eds) Australian coastal archaeology, (ed) Australian linguistic studies,pp.107–39. pp. 37–50. Canberra: ANH Publications. Canberra: Pacific Linguistics. References 377

O’Grady GN (1996) Lexicostatistic and other Pearce RH (1974) Spatial and temporal distribu- evidence for the homeland of the West- tion of Australian backed blades. Mankind 9: ern Desert language. Unpublished paper, The 300–9. Third Australian Linguistic Institute Work- Pennington R (2001) Hunter-gatherer demog- shop, ‘Where did the Western Desert language raphy. In C Panter-Brick, RH Layton come from?’, Canberra, July 1996. and P Rowley-Conwy (eds) Hunter-gatherers: Olley JM, Roberts RG, Yoshida H and Bowler An interdisciplinary perspective,pp.170– JM (2006) Single-grain optical dating of grave- 204. Cambridge: Cambridge University infill associated with human burials at Lake Press. Mungo, Australia. Quaternary Science Reviews Peron´ F (1824) Voyage of discovery to the South- 25: 2469–74. ern Lands. Book IV.Second edition [Adelaide: Oppenheimer S (2012) A single southern exit of Friends of the State Library of South Aus- modern humans from Africa: Before or after tralia, 2003, Vol. II, English translation by C To b a ? Quaternary International 258: 88–99. Cornell]. Pack SM, Miller GH, Fogel ML and Spooner Peterson N (1972) Totemism yesterday: Senti- NA (2003) Carbon isotopic evidence for ment and local organisation among the Aus- increased aridity in northwestern Australia tralian Aborigines. Man n.s. 7: 12–32. through the Quaternary. Quaternary Science Peterson N (1976) The natural and cultural areas Reviews 22: 629–43. of Aboriginal Australia: A preliminary analysis Palacio-Perez E (2010) Salomon Reinach and of population groupings with adaptive signif- the religious interpretation of Palaeolithic art. icance. In N Peterson (ed) Tribes and bound- Antiquity 84: 843–63. aries in Australia,pp.50–71. Social Anthropol- PardoeC(1993) The Pleistocene is still with us: ogy Series 10. Canberra: Australian Institute Analytical constraints and possibilities for the of Aboriginal Studies. study of ancient human remains in archae- Peterson N (1986) Australian Territorial Organisa- ology. In MA Smith, M Spriggs and B tion. Oceania Monograph 30. Sydney: Uni- Fankhauser (eds) Sahul in review: Pleistocene versity of Sydney. archaeology in Australia, New Guinea and Island Peterson N and Lampert R (1985) A Central Melanesia,pp.81–94.OccasionalPapersin Australian ochre mine. Records of the Australian Prehistory 24. Canberra: Australian National Museum 37: 1–9. University. Petherick GL, McGowan H and Moss P (2008) Pardoe C (1994) Bioscapes: The evolutionary Climate variability during the last glacial max- landscape of Australia. Archaeology in Oceania imum in eastern Australia: Evidence of two 29: 182–90. stadials? Journal of Quaternary Science 23: 787– Pardoe C (2006) Becoming Australian: Evo- 802. lutionary processes and biological variation Petit JR, Jouzel J, Raynaud D, Barkov NI, from ancient to modern times. Before Farm- Barnola J-M, Basile I et al. (1999) Climate and ing 2006/1: article 4. atmospheric history of the past 420,000 years Partridge J (1967)A3,300-year-old Thylacine from the Vostok ice core, Antarctica. Nature (Marsupialia: Thylacinidae) from the Nullarbor 399: 429–36. Plain, Western Australia. Journal of the Royal Petri HE and Schulz AS (1951) Felsgravierungen Society of Western Australia 50: 57–9. aus Nordwest-Australien. Zeitschrift fur¨ Eth- Paton R (1994) Speaking through stones: A study nologie 76: 70–93. from northern Australia. World Archaeology 26: Pianka ER (1981) Diversity and adaptive radi- 172–84. ations of Australian lizards. In A Keast (ed) Patton PC, Pickup G and Price DM (1993) Ecological biogeography of Australia,pp.1375–92. Holocene palaeofloods of the Ross River, The Hague: W Junk. Central Australia. Quaternary Research 40: 201– Pickering M (1990) Backed blades from 12. the McArthur River, Borroloola, Northern Peake-Jones K (ed) (1955) Lake Eyre, South Aus- Te r r i t o r y. Australian Archaeology 31: 83–5. tralia: The great flooding of 1949–50. Adelaide: Pickup G, Allan G and Baker VR (1988)History, The Royal Geographical Society of Australa- palaeochannels and palaeofloods of the Finke sia, (South Australian Branch). River, Central Australia. In RF Warner (ed) 378 References

Fluvial geomorphology of Australia,pp.177–200. cock (eds) Desert peoples: Archaeological per- Sydney: Academic Press. spectives,pp.177–205. Oxford: Blackwell Pillans B (2007) Quaternary stratigraphy – Publishers. Chrono-stratigraphy. In SA Elias (ed) Ency- Pulleine R (1925) Rock carvings (petroglyphs) clopedia of Quaternary science,pp.2811–19. and cave paintings at Mootwingee, NSW Online: Elsevier. Transactions of the Royal Society of South Aus- Pillans B and Fiﬁeld K (2010) Weathering his- tralia 50: 180–2. tory of rock surfaces associated with Abo- Pumpelly R (ed) (1908) Explorations in Turkestan. riginal rock art, Burrup Peninsula, Western Washington: Carnegie Institution. Australia. Abstract. Australian Earth Sciences Quartermaine G (1994) Report on an archaeologi- Convention 2010 – Earth systems: Change, cal excavation programme at Aboriginal site MG2, Sustainability, Vulnerability – 4–8 July 2010, Marandoo. Unpublished report for Hamersley National Convention Centre, Canberra. Can- Iron Pty Ltd. berra: Geological Society of Australia. Quigley MC, Horton T, Hellstrom JC, Cupper Pledge NS (1994) Fossils of the lake: A history ML and Sandiford M (2010) Holocene cli- of the Lake Callabonna excavations. Records of mate change in arid Australia from speleothem the South Australian Museum 27: 65–77. and alluvial records. The Holocene 20: 1093– Pratt ML (1992) Imperial eyes: Studies in travel writ- 104. ing and transculturation. Abingdon: Routledge. Ratcliffe FN (1936) Soil drift in the arid pastoral Prescott JR, Williams FM and Hunt CD (2007) areas of South Australia. CSIR Pamphlet 64. Comparison of TL multiple aliquot, single Melbourne: Council for Scientiﬁc and Indus- grain GLSL SAR and C-14 ages for the trial Research (Australia). Puritjarra, Australia, rock shelter. Quaternary Ratcliffe F (1938) Flying fox and drifting sand: The Geochronology 2: 344–9. adventures of a biologist in Australia. London: Pretty GL (1968) Excavation of Aboriginal graves Chatto and Windus. at Gidgealpa, South Australia. Records of the Reeves JM, Chivas AR, Garcia A, Holt S, South Australian Museum 15: 671–7. Couapel MJJ, Jones BG et al. (2008) The sed- Price GJ (2008) Taxonomy and palaeobiology of imentary record of palaeoenvironmental and the largest-ever marsupial, Diprotodon Owen sea-level change in the Gulf of Carpentaria, 1838 (Diprotodontidae, Marsupialia). Zoolog- Australia, though the last glacial cycle. Quater- ical Journal of the Linnean Society 153: 389– nary International 183: 3–22. 417. Reinach S (1903) L’art et la magie: aproposdes´ Price TD and Brown JA (eds) (1985) Prehistoric peintures et des gravures de l’ageˆ du renne. hunter-gatherers: The emergence of cultural com- L’Anthropologie 14: 257–66. plexity. Orlando, Florida: Academic Press. Reneau SL, Oberlander TM and Harrington Prideaux GJ, Ayliffe LK, DeSantis LRG, CD (1991). Accelerator mass spectrometry Schubert BW, Murray PF, Gagan MK and radiocarbon dating of rock varnish: Discussion Cerling TE. (2009) Extinction implications and reply. Geological Society of America Bulletin of a chenopod browse diet for a giant Pleis- 103: 310–14. tocene kangaroo. PNAS 106 (28): 11646– Reuther JG (1981)[1914] The Diari. [AIAS 50. Microform 2, trans. PA Scherer, Australian Prideaux GJ, Long JA, Ayliffe LK, Hellstrom Institute of Aboriginal Studies, Canberra, JC, Pillans B, Boles WE et al. (2007)Anarid- 1981]. adapted middle Pleistocene vertebrate fauna Reuther JG and Strehlow C (trans.) (1897) from south-central Australia. Nature 445: Testamenta marra: Jesuni Christuni ngantjani 422–5. jaura ninaia karitjimalkana wonti Dieri jaurani. Przywolnik K (2002) Patterns of occupation in Cape Tanunda, SA: G. Auricht. Range peninsula (WA) over the last 36,000 years. Rich PV (1979) The Dromornithidae. Bulletin Unpublished PhD thesis, University of West- of the Bureau of Mineral Resources, Geology and ern Australia. Geophysics 184: 1–196. Przywolnik K (2005) Long-term transitions in Rich PV (1985) Genyornis newtoni Stirling and hunter-gatherers of coastal northwestern Aus- Zietz, 1896. A Mihirung. In PV Rich and GF tralia. In P Veth, M Smith and P His- van Tets (eds), Kadimakara: Extinct vertebrates of References 379

Australia,pp.188–94. New Jersey: Princeton Robertson GV, Attenbrow V and Hiscock P University Press. (2009) Multiple uses for Australian backed Ride WDL, Crawford IM and Storr GM (1964) artefacts. Antiquity 83: 296–308. Report on the Aboriginal engravings and flora and Robin L, Dickman C and Martin M (2010) fauna of Depuch Island, Western Australia. West- Desert channels: The impulse to conserve.Mel- ern Australian Museum, Special Publication bourne: CSIRO Publishing. No. 2.Perth. Robins RP (1993) Archaeology and the Currawinya Ridges M (2003) Numerous indications: The archae- Lakes: Towards a prehistory of arid lands of ology of regional hunter-gatherer behaviour in north- southwest Queensland. Unpublished PhD the- west central Queensland, Australia. Unpublished sis, Griffith University PhD thesis. Armidale NSW: University of Robins RP (1998) Archaeological investigations New England. at Youlain Springs, southwest Queensland. Riedinger MA, Steinitz-Kannan M, Last WM Memoirs of the Queensland Museum, Cultural and Brenner M (2002)A∼6100 14Cyrrecord Heritage Series 1(1): 57–74. of El Nino˜ activity from the Galapagos´ Islands. Rockman M and Steele J (eds) (2003) Colonisa- Journal of Paleolimnology 27: 1–7. tion of unfamiliar landscapes: The archaeology of Rindos D and Webb E (1992) Modelling the adaptation. London: Routledge. initial human colonization of Australia: Per- Rosenfeld A (1990) Rock art in Watarrka National fect adaptation, cultural variability, and cul- Park: Report on fieldwork in the Watarrka tural change. Proceedings of the Australasian Soci- National Park 1990. Unpublished report to the ety for Human Biology 5: 441–54. Conservation Commission of the Northern Robbins LH, Murphy ML, Brook GA, Ivester Territory, Alice Springs. AH, Campbell AC, Klein RG et al. Rosenfeld A (1991) Panaramittee: Dead or alive? (2000) Archaeology, palaeoenvironment and In P Bahn and A Rosenfeld (eds) Rock art and chronology of the Tsodilo Hills White prehistory,pp.136–44. Oxbow Monograph 10. Paintings Rock Shelter, Northwest Kalahari Oxford: Oxbow Books. Desert, Botswana. Journal of Archaeological Sci- Rosenfeld A (1993) A review of the evidence ence 27: 1085–113. for the emergence of rock art in Australia. Robbins LH, Murphy ML, Campbell AC and In MA Smith, M Spriggs and B Fankhauser Brook GA (1998) Intensive mining of specular (eds) Sahul in review: Pleistocene Archaeology hematite in the Kalahari ca. AD 800–1000. in Australia, New Guinea and Island Melane- Current Anthropology 39: 144–50. sia,pp.71–80. Occasional Papers in Prehis- Roberts RG, Flannery TF, Ayliffe LK, Yoshida tory 24. Canberra: Department of Prehistory, H, Olley JM, Prideaux GJ et al. (2001) Research School of Pacific Studies, Australian New ages for the last Australian megafauna: National University. Continent-wide extinction about 46,000 years Rosenfeld A (1997) Archaeological signatures of ago. Science 292: 1888–92. the social context of rock art production. In Roberts RG, Jones R and Smith MA (1990) M Conkey, O Soffer, D Stratman and NG Thermoluminescence dating of a 50,000- Jablonski (eds) Beyond art: Pleistocene image and year-old human occupation site in northern symbol. Memoirs of the California Academy Australia. Nature 345: 153–6. of Sciences No. 23. San Francisco: California Roberts RG, Jones R, Spooner NA, Head Academy of Sciences. MJ, Murray AS and Smith MA (1994) The Rosenfeld A (2002) Rock-art as an indicator of human colonisation of Australia: Optical dates changing social geographies in Central Aus- of 53,000 and 60,000 years bracket human tralia. In B David and M Wilson (eds), Inscribed arrival at Deaf Adder Gorge, Northern Terri- landscapes: Marking and making place,pp.61–78. tory. Quaternary Science Reviews 13: 575–83. Honolulu: University of Hawaii Press. Roberts RG, Spooner NA, Jones R, Cane S, Rosenfeld A, Horton D and Winter J (1981). Olley JM, Murray AS and Head MJ (1996) Early man in North Queensland: Art and archaeol- Preliminary luminescence ages for archae- ogy in the Laura area. Canberra: Terra Australis ological sediments on the Nullarbor Plain, 6, Australian National University. South Australia. Australian Archaeology 42: 7– Rosenfeld A and Mumford W (1996) The 16. Thuiparta rock engravings at Erowalle, 380 References

Wallace Rock Hole, James Range, Northern Roth WE (1897) Ethnological studies among the Territory. In S Ulm, I Lilley and A Ross (eds) north-west-central Queensland Aborigines.Bris- Australian Archaeology ’95: Proceedings of the bane: E Gregory, Government Printer. [Fac- 1995 Australian Archaeological Association Annual simile, Hesperian Press, Carlisle, Western Conference,pp.247–55. Brisbane: University of Australia, 1984] Queensland. Te m p us 6. Roth WE (1904) Domestic implements, arts, and Rosenfeld A and Smith MA (2002) Rock-art manufactures. North Queensland Ethnography and the history of Puritjarra Rock Shelter, Bulletin No. 7. Brisbane: Department of Pub- Cleland Hills, Central Australia. Proceedings of lic Lands. [Facsimile, Hesperian Press, Carlisle, the Prehistoric Society 68: 103–24. Western Australia, 1984] Ross A (1981) Holocene environments and Rowley-Conwy P (2001) Time, change and the prehistoric site patterning in the Victorian archaeology of hunter-gatherers: How orig- Mallee. Archaeology in Oceania 16: 145–54. inal is the ‘Original Afﬂuent Society’? In C Ross A, Donnelly T and Wasson R (1992) Panter-Brick, RH Layton and P Rowley- The peopling of the arid zone: Human- Conwy (eds) Hunter-gatherers: An interdis- environment interactions. In J Dodson (ed) ciplinary perspective,pp.39–72. Cambridge: The Naive Lands: Prehistory and environmen- Cambridge University Press. tal change in Australia and the Southwest Paciﬁc, Royal Society of South Australia (1901–02). pp. 76–114. Melbourne: Longman Cheshire. Abstract of Proceedings of the Royal Society Ross J (1997) Painted relationships: An archaeologi- of South Australia for 1901–02. Transactions of cal analysis of a distinctive anthropomorphic rock art the Royal Society of South Australia 26: 322–30. motif in northwest central Queensland. Unpub- Rudwick MJS (2008) Worlds before Adam: The lished BA (Hons) thesis, University of New reconstruction of geohistory in the age of reform. England. Chicago: University of Chicago Press. Ross J (2002) Rocking the boundaries, scratch- Ryan I and Morse K (2009) Towards a late ing the surface: An analysis of the relationship Holocene archaeology of the inland Pilbara. between paintings and engravings in the Cen- Archaeology in Oceania 44 (Supplement): 6–15. tral Australian arid zone. In S Ulm, C West- Sadr K (2005) Hunter-gatherers and herders of cott, J Reid, A Ross, I Lilley, J Prangnell and the Kalahari during the late Holocene. In P L Kirkwood (eds) Barriers, borders, boundaries. Veth, M Smith and P Hiscock (eds) Desert Proceedings of the 2001 Australian Archaeological peoples: Archaeological perspectives,pp.206–21. Association Annual Conference,pp.83–9.Bris- Maldon: Blackwell. bane: University of Queensland. Te m p us 7. Sagona AG (ed) (1994) Bruising the red earth: RossJ(2003) Rock art, ritual and relationships: An Ochre mining and ritual in Aboriginal Tasmania. archaeological analysis of rock art from the central Melbourne: Melbourne University Press. Australian arid zone. Unpublished PhD thesis, Salmon D (1998) Hamersley Plateau: Land University of New England. between two rivers. Technological analysis of Cleft Ross J (2005) Rock art of the Red Centre. In Rockshelter, North-west Australia. Unpublished M Smith and P Hesse (eds) 23oS: Archaeology BSc (Hons) thesis, University of Western and environmental history of the Southern Deserts, Australia. pp. 217–30. Canberra: National Museum of Sandall R (1967) Walbiri ritual at Gunadjari. Film, Australia Press. AIAS-016, Australian Institute of Aboriginal Ross J and Davidson I (2006) Rock art and rit- and Torres Strait Islander Studies, Canberra. ual: An archaeological analysis of rock art in Sandall R (1972) Pintupi revisit Yaru Yaru. Film, arid Central Australia. Journal of Archaeological AIAS-Film/Video-42, Australian Institute of Method and Theory 13: 305–41. Aboriginal and Torres Strait Islander Studies, Ross J and Smith MA (2007) Research report: Canberra. A late Pleistocene site on Watarrka Plateau, Savolainen P, Leitner T, Wilton AN, Matisoo- Central Australia. Australian Aboriginal Studies Smith E and Lundeberg J (2004) A detailed 2007/1: 138–43. picture of the origin of the Australian dingo, Ross J and Smith MA (2009)Anengraved obtained from the study of mitochondrial ‘archaic face’ in the northeastern Simpson DNA. Proceedings of the National Academy of Desert. Australian Archaeology 69: 68–70. Science 101: 12387–90. References 381

Schall A (1985) Aboriginal use of shell on Cape Shulmeister J and Lees B (1995) Pollen evidence York Peninsula. Cultural Resource Manage- from tropical Australia for the onset of an ment Monograph 6. Brisbane: Archaeol- ENSO-dominated climate at c. 4000 BP. The ogy Branch, Department of Community Holocene 5: 10–18. Services. Silcock JL, Tischler M and Smith MA (2012) Schama S (1996) Landscape and memory. London: Quantifying the Mulligan River Pituri, Fontana. Duboisia hopwoodii ((F. Muell.) F. Muell.) Schneider H and Pryer KM (2002) Structure and (Solanaceae), trade of Central Australia. Eth- function of spores in the aquatic heterosporous nobotany Research and Applications 10: 37–44. fern family Marsileaceae. International Journal of Singh G (1981) Late Quaternary pollen records Plant Sciences 163: 485–505. and seasonal palaeoclimates of Lake Frome, Schmidt PW (1919) Die Gliederung des australis- South Australia. Hydrobiologia 82: 419–30. chen Sprachen. Vienna: Mechitharisten-Buch Singh G and Luly J (1991) Changes in vegetation [English translation, D Clark, typescript, Aus- and seasonal climate since the last full glacial tralian Institute of Aboriginal and Torres Strait at Lake Frome, South Australia. Palaeogeog- Islander Studies, Canberra, 1972]. raphy, Palaeoclimatology, Palaeoecology 84: 75– Shackleton NJ, Chapman M, Sanchez-Goni˜ 86. MF, Pailler D and Lancelot Y (2002) The Slack MJ (2007) Between the Desert and the Gulf: Classic Marine Isotope Substage 5e. Quater- Evolutionary anthropology and Aboriginal prehis- nary Research 58: 14–16. tory in the Riversleigh/Lawn Hill Region, North- Sharp L (1952) Steel axes for stone-age Aus- ern Australia. Unpublished PhD thesis, Uni- tralians. Human Organisation 11: 17–22. versity of Sydney. Sharpe CE and Sharpe KJ (1976) A prelimi- Slack M, Fillios M and Fullagar R (2009) nary survey of engraved boulders in the art Aboriginal settlement during the LGM at sanctuary of Koonalda Cave, South Australia. Brockman, Pilbara Region, Western Aus- Mankind 10: 125–30. tralia. Archaeology in Oceania 44 (Supplement): Shawcross FW and Kaye M (1980) Australian 32–9. archaeology: Implications of current inter- Slack MJ, Fullagar RLK, Field JH and Border disciplinary research. Interdisciplinary Science A(2004) New Pleistocene dates for backed Reviews 5: 112–28. artefact technology in Australia. Archaeology in Sheridan G (1979) Tulas and Triodia: A multi- Oceania 39: 131–7. disciplinary investigation of the mechanics and Slaytor WJH (ed) (2005) Archaeology of the Lake antecedents of the Koondi Tuhla and their impli- Eyre South region. Adelaide: Royal Geograph- cations for prehistory. Unpublished MA Qual. ical Society of South Australia. Lake Eyre thesis, Australian National University. Monograph Series 6. Shigesada N and Kawasaki K (1997) Biological Sloss CR, Murray-Wallace CV and Jones BG invasions: Theory and practice.Oxford:Oxford (2007) Holocene sea-level change on the University Press. southeast coast of Australia: A review. The Shiner J (2008) Place as occupational histories: An Holocene 17: 999–1014. investigation of the deflated surface archaeological Smith C (1992) Colonising with style: Review- record of Pine Point and Langwell Stations, western ing the nexus between rock art, territoriality New South Wales, Australia. Oxford: Archaeo- and the colonisation and occupation of Sahul. press. BAR International Series 1763. Australian Archaeology 34: 34–42. Shiner J, Holdaway S, Allen H and Fan- Smith MA (1977) Devon Downs reconsidered: ning P (2007) Burkes Cave and flaked stone An exercise in bioarchaeology. Unpublished BA assemblage variability in western New South (Hons) thesis, Australian National University. Wales, Australia. Australian Archaeology 64: 35– Smith MA (1983) Central Australia: Prelimi- 45. nary archaeological investigations. Australian Shulmeister J (1999) Australasian evidence for Archaeology 16: 27–38. mid-holocene climate change implies preces- Smith MA (1985) A morphological comparison sional control of Walker Circulation in the of Central Australian seed-grinding imple- Pacific. Quaternary International 57/58: 81– ments and Australian Pleistocene-age grind- 91. stones. The Beagle, Occasional Papers of the 382 References

Northern Territory Museum of Arts and Sciences country,pp.147–83. Collingwood, Victoria: 2: 23–38. CSIRO Publishing, Smith MA (1986a) The antiquity of seed- Smith MA (2009b) Late Quaternary landscapes grinding in arid Australia. Archaeology in Ocea- in Central Australia: Sedimentary history and nia 21: 29–39. palaeoecology of Puritjarra rock shelter. Jour- Smith MA (1986b) A revised chronology for nal of Quaternary Science 24: 747–60. Intirtekwerle (James Range East) rockshelter, SmithMA(2010) Puritjarra Rock Shelter: Final Central Australia. The Beagle: Occasional Papers report on excavations 1986–1990. Canberra: of the Northern Territory Museum of Arts and Sci- National Museum of Australia Press, Tech- ences 3: 123–30. nical Report 1 (on CD). Smith MA (1987) Pleistocene occupation in arid Smith MA, Bird MI, Turney CSM, Fifield LK, Central Australia. Nature 328: 710–11. Santos GM, Hausladen PA and di Tada ML Smith MA (1988) The pattern and timing of prehis- (2001) New ABOX AMS-14Cagesremove toric settlement in Central Australia. Unpublished dating anomalies at Puritjarra rock shelter. PhD thesis. University of New England. Australian Archaeology 53: 45–7. SmithMA(1989a) Seed gathering in inland Aus- Smith MA and Clark PM (1993) Radiocar- tralia: Current evidence from seed-grinders bon dates for prehistoric occupation of the on the antiquity of the ethnohistorical pattern Simpson Desert. Records of the South Australian of exploitation. In DR Harris and GC Hill- Museum 26: 121–7. man (eds) Foraging and farming: The evolution of Smith MA and Cundy BJ (1985) Distribution plant exploitation,pp.305–17. London: Unwin maps for flaked stone points and backed blades Hyman. in the Northern Territory. Australian Aborigi- Smith MA (1989b) The case for a resident human nal Studies 1985/2: 32–7. population in the Central Australian Ranges Smith MA and Fankhauser BL (2009) Geochem- during full glacial aridity. Archaeology in Ocea- istry and identification of Australian red ochre nia 24: 93–105. deposits. Palaeoworks Technical Papers 9. Smith MA (1993) Biogeography, human ecol- http://palaeoworks.anu.edu.au/publications ogy and prehistory in the sandridge deserts. .html Australian Archaeology 37: 35–50. Smith MA, Fankhauser B and Jercher M (1998) SmithMA(2000) ‘The opening chapter of the The changing provenance of red ochre at romance of excavation in Australia’: Reflec- Puritjarra rock shelter, Central Australia: Late tions on Norman Tindale’s archaeology. His- Pleistocene to present. Proceedings of the Prehis- torical Records of Australian Science 13: 151– toric Society 64: 275–92. 60. Smith MA and Hesse P (eds) (2005) 23oS: Archae- Smith MA (2004) The grindstone assemblage ology and environmental history of the Southern from Puritjarra rock shelter: Investigating the Deserts. Canberra: National Museum of Aus- history of seed-based economies in arid Aus- tralia Press. tralia. In T Murray (ed) Archaeology from Smith MA, McBryde I and Ross J (2010) The Australia,pp.168–86. Melbourne: Australian economics of grindstone production at Nar- Scholarly Publishing. coonowie quarry, Strzelecki Desert. Australian Smith MA (2005) Desert archaeology, linguis- Aboriginal Studies 2010/1: 92–9. tic stratigraphy and the spread of the Western Smith MA, Prescott JR and Head MJ Desert language. In P Veth, MA Smith and (1997) Comparison of 14C and lumines- P Hiscock (eds) Desert peoples: Archaeological cence chronologies at Puritjarra rock shelter, perspectives pp. 222–42. Oxford: Blackwell. Central Australia. Quaternary Science Reviews Smith MA (2006) Characterising late Pleistocene (Quaternary Geochronology) 16: 299–320. and Holocene stone artefact assemblages from SmithMAandRossJ(2008a) What happened Puritjarra rock shelter: A long sequence from at 1500–1000 BP in Central Australia? – Tim- the Australian desert. Records of the Australian ing, impact and archaeological signatures. The Museum 58: 371–410. Holocene 18: 379–88. Smith MA (2009a) Genyornis: Last of the Dro- Smith MA and Ross J (2008b) Glen Thirsty: mornithids. In L Robin, R Heinsohn and L The history and archaeology of a desert well. Joseph (eds) Boom and bust: Bird stories for a dry Australian Archaeology 66: 45–59. References 383

Smith MA, Vellen L and Pask J (1995) Vegeta- Spencer B and Gillen FJ (1897) The Engwura, tion history from archaeological charcoals in or fire ceremony of certain Central Australian central Australia: The late Quaternary record tribes. Nature 56: 136–9. from Puritjarra rock shelter. Vegetation History Spencer B and Gillen FJ (1899) The native tribes of & Archaeobotany 4: 171–7. Central Australia. London: Macmillan [Ooster- Smith MA and Veth PM (2004) Radiocarbon hout NB, The Netherlands: Anthropological dates for baler shell in the Great Sandy Desert. Publications, 1969]. Australian Archaeology 58: 37–8. Spencer B and Gillen FJ (1904) The northern tribes Smith MA, Watchman A and Ross J (2009) of Central Australia. London: Macmillan [Oost- Direct dating indicates a mid-Holocene age erhout NB, The Netherlands: Anthropologi- for archaic desert rock engravings in arid cal Publications, 1969]. Central Australia. Geoarchaeology 24: 191– Spencer B and Gillen FJ (1927) The Arunta: A 203. study of a stone age people. London: Macmillan. Smith MA, Williams A, Turney C and Cup- Spooner MI, Barrows TT, De Deckker P and per ML (2008) Human-environment inter- Paterne M (2005) Palaeoceanography of the actions in Australian drylands: Exploratory Banda Sea, and Late Pleistocene initiation of time-series analysis of archaeological records. the northwest monsoon. Global and Planetary The Holocene 18: 389–401. Change 49: 28–46. Smith MA, Williams E and Wasson RJ (1991) Sprigg RC (1964) Consequences of Quaternary The archaeology of the JSN site: Some impli- climatic fluctuations in Australia. Geological cations for the dynamics of human occupation Society of America Special Paper 82: 193–4. in the Strzelecki Desert during the late Pleis- Sprigg RC (1979) Stranded and submerged sea- tocene. Records of the South Australian Museum beach systems of southeast South Australia and 25: 175–92. the aeolian desert cycle. Sedimentary Geology Smith-White S (1982) Summary and redinte- 22: 53–96. gration. In WR Barker and PJM Greenslade Spriggs M (1989) The dating of the island south- (eds) Evolution of the flora and fauna of east Asian neolithic: An attempt at chrono- arid Australia,pp.371–9. Australian System- metric hygiene and linguistic correlation. atic Botany Society. Frewville, SA: Peacock Antiquity 63: 587–613. Publications. Stafford Smith M and Cribb J (2009) Dry times: SmythRBrough(1878) The Aborigines of Vic- Blueprint for a red land. Melbourne: CSIRO toria: With notes relating to the habits of the Publishing. natives of other parts of Australia and Tasmania. Stafford Smith M, West J and Thiele K (1996) Melbourne: Government Printer. Biogeographic patterns in inland Australia: Sollas WJ (1911) Ancient hunters and their modern The role of isolation and uncertainty. In SR representatives. London: Macmillan. Morton and DJ Mulvaney (eds) Exploring Cen- Sowell J (2001) Desert ecology: An introduction to life tral Australia: Society, the environment and the in the arid Southwest. Salt Lake City: University 1894 Horn Expedition,pp.368–78. Chipping of Utah Press. Norton (Sydney): Surrey Beatty & Sons. Spencer B (ed) (1896) Report on the work of Stanner WEH (1933–4) Ceremonial economics the Horn scientific expedition to Central Aus- of the Mulluk Mulluk and Madngella tribes of tralia,Parts1–4. Melbourne: Melville, Mullen the Daly River, north Australia. A preliminary and Slade. [Facsimile, Corkwood Press, Bund- paper. Oceania 4: 156–75, 458–71. aberg, 1994]. Stanner WEH (1965) Aboriginal territorial orga- Spencer WB (1901) Guide to the Australian ethno- nization: Estate, range, domain and regime. graphical collection in the National Museum of Vic- Oceania 36: 1–26. toria. Melbourne: Government Printer. Stanner WEH (2009) The dreaming (1953). In Spencer B (1921) Presidential address. Report of R Manne (comp.) WE Stanner: The Dreaming the Fifteenth Meeting of the Australasian Associ- and other essays,pp.57–72. Melbourne: Black ation for the Advancement of Science,pp.1–39. Inc. Agenda. Melbourne: Albert J Mullett. Stephens CG and Crocker RL (1946) Composi- Spencer B (1928) Wanderings in Wild Australia. tion and genesis of lunettes. Transactions of the London: Macmillan. Royal Society of South Australia 70: 302–12. 384 References

Stephenson B (2011) In the groove: An integrated Australian Aboriginal anthropology,pp.92–140. functional analysis of arid zone millstones from Nedlands: University of Western Australia Queensland. Unpublished BA (Hons) thesis, Press. University of Queensland. Strehlow TGH (1971) Songs of Central Australia. Stevens BPJ (1991) Some aligned claypans in the Sydney: Angus and Robertson. Strzelecki dunefield, central Australia. Aus- Strehlow TGH (1978) Central Australian religion: tralian Journal of Earth Sciences 38(4): 485–95. Personal monototemism in a polytotemic commu- Stirling EC (1893) Report to Board of Governors, nity. Adelaide: Australian Association for the 28 August 1893, South Australian Museum – Study of Religions. Archives, Adelaide. Sturman A and Tapper N (2006) The weather and Stirling EC (1894) The recent discovery of fossil climate of Australia and New Zealand. Second remains at Lake Callabonna, South Australia. edition. Melbourne: Oxford University Press. Nature 50: 184–8, 206–11. Sturt C (1849) Narrative of an expedition into Stirling EC (1896) Part IV – Anthropology. In B Central Australia. London: T and W Boone. Spencer (ed) Report on the work of the Horn scien- [Reprint, Greenwood Press, New York, tific expedition to Central Australia. Melbourne: 1969]. Melville, Mullen and Slade. [Facsimile, Cork- Sturt C (1984) Journal of the Central Australian wood Press, Bundaberg, 1994]. expedition 1844–45. ( J Waterhouse [ed]). Lon- Stirling EC (1900) The physical features of Lake don: Caliban. Callabonna. Memoirs of the Royal Society of Summerhayes GR, Leavesley M, Fairbairn A, South Australia 1:i–xv. Mandui H, Field J, Ford A and Fullagar R Stirling EC and Zietz AHC (1899–1913) Fossil (2010) Human adaptation and plant use in remains of Lake Callabonna. Memoirs of the Highland New Guinea 49,000 to 44,000 years Royal Society of South Australia 1(1–4): 1–178. ago. Science 330: 78–81. Stockton ED (1971) Investigations at Santa Sutton P (ed) (1988) Dreamings: The art of Abo- Teresa, Central Australia. Archaeology and Phys- riginal Australia. Melbourne: Viking. ical Anthropology in Oceania 6: 44–61. Sutton P (2003) Native title in Australia: An ethno- Stodart E and Parer I (1988) Colonisation of Aus- graphic perspective. Cambridge: Cambridge tralia by the rabbit, Oryctolagus cuniculus (L). University Press. Project Report No. 6. Canberra: Division of SuttonPandKochH(2008) Review article. Wildlife and Ecology, CSIRO. Australian languages: A singular vision? Journal Stott L, Cannariato K, Thunell R, Haug GH, of Linguistics 44: 471–504. Koutavas A and Lund S (2004) Decline of sur- Surovell TA, Finley JB, Smith GM, Brantingham face temperature and salinity in the western PJ and Kelly R (2009) Correcting temporal tropical Pacific Ocean in the Holocene epoch. frequency distributions for taphonomic bias. Nature 431: 56–9. Journal of Archaeological Science 36: 1715–24. Strehlow C (1907–20) Die Aranda- und Loritja- Tacon PSC (1994) Socialising landscapes: The Stamme¨ in Zentral-Australien. Frankfurt: Joseph long-term implications of signs, symbols and Baer. marks on the land. Archaeology in Oceania 29: Strehlow TGH (1947) Aranda traditions.Mel- 117–29. bourne: Melbourne University Press. Tate R (1879) The anniversary address of Strehlow TGH (1964) The art of circle, line and the president. Transactions and Proceedings and square. In RM Berndt (ed) Australian Aborigi- Report of the Philosophical Society of Adelaide, nal art,pp.44–59. Sydney: Ure Smith. South Australia 1878–79: xxxix–lxxv. Strehlow TGH (1965) Culture, social structure Tate R (1886) Post-Miocene climate in South and environment in Aboriginal Central Aus- Australia. Transactions and Proceedings and Report tralia. In RM Berndt and CH Berndt (eds) of the Royal Society of South Australia 8: 49–59. Aboriginal man in Australia: Essays in honour Tate R (1896) Botany. In B Spencer (ed) Report of Emeritus Professor AP Elkin,pp.121–45. on the work of the Horn scientific expedition to Sydney: Angus and Robertson. Central Australia, Part III: Geology and Botany, Strehlow TGH (1970) Geography and the pp. 117–94. Melbourne: Melville, Mullen and totemic landscape in Central Australia: A Slade. [Facsimile, Corkwood Press, Bund- functional study. In RM Berndt (ed) aberg, 1994]. References 385

Taylor L (1979) Ancestors into art: An analysis of Thorley PB (1998b) Shifting location, shifting scale: Pitjantjatjara kulpidji designs and crayon drawings. Towards an archaeology of place in the Palmer River Unpublished BA (Hons) Thesis, Australian catchment, Central Australia. Unpublished PhD National University. Thesis, Northern Territory University. Taylor TG (1919) Climatic Cycles and Evolu- Thorley PB (1999) A resin-hafted stone imple- tion. Geographical Review 8: 289–328. ment from the middle Palmer River catch- Te d f o r d R ( 1984) The Diprotodons of ment, Central Australia. Australian Archaeology Callabonna. In M Archer and G Clayton (eds) 48: 45. Vertebrate zoogeography and evolution in Australa- Thorley P, Faulkner P and Smith M (2011)New sia (Animals in space and time),pp.999–1002. radiocarbon dates for Kulpi Mara rock shel- Carlisle: Hesperian Press. ter, Central Australia. Australian Archaeology TedfordRHandWellsRT(1990) Pleistocene 72: 47–9. deposits and fossil vertebrates from the ‘Dead Thorne AG (1977) Separation or reconciliation? Heart of Australia’. Memoirs of the Queensland Biological clues to the development of Aus- Museum 28: 263–84. tralian society. In J Allen, J Golson and R Telford RJ, Heegaard E and Birks HJB (2004) Jones (eds) Sunda and Sahul: Prehistoric stud- The intercept is a poor estimate of a cali- ies in Southeast Asia, Melanesia and Australia, brated radiocarbon age. The Holocene 14: 296– pp. 187–204. London: Academic Press. 8. Thorne A and Curnoe D (2000) Sex and sig- Testart A (1988) Some major problems in the niﬁcance of Lake Mungo 3:ReplytoBrown social anthropology of hunter-gatherers. Cur- ‘Australian Pleistocene variation and the sex rent Anthropology 29: 1–31. of Lake Mungo 3’. Journal of Human Evolution Thackeray AI (1992) The Middle Stone Age 39: 587–600. south of the Limpopo River. Journal of World Thorne A, Grun¨ R, Mortimer G, Spooner NA, Prehistory 6: 385–440. Simpson JJ, McCulloch M et al. (1999) Aus- Thackway R and Cresswell ID (1995) An interim tralia’s oldest human remains: Age of the Lake biogeographic regionalisation for Australia: A frame- Mungo 3 skeleton. Journal of Human Evolution work for establishing the national system of reserves. 36: 591–612. Canberra: Australian Nature Conservation Tibbett K (2002) Archaeological analysis of stone Agency. axe exchange networks in the Lake Eyre Basin The South Australian Register (1864) The northern during the mid- to late Holocene. Australian blacks. 26 July, 7. Archaeology 55: 22–9. The South Australian Register (1893) Editorial, 24 Tindale NB (1932) Notes on the supposed prim- May, 4. itive stone implements from the Tableland Thomas I, Enright NJ and Kenyon CE (2001) regions of Central Australia. Records of the The Holocene history of Mediterranean-type South Australian Museum 4: 483–8. plant communities, Little Desert National Tindale NB (1940)ResultsoftheHarvard- Park, Victoria, Australia. The Holocene 11: Adelaide Universities Anthropological Expe- 691–7. dition, 1938–1939. Distribution of Australian Thomson DF (1949) Economic structure and the Aboriginal tribes: A ﬁeld survey. Transactions ceremonial exchange cycle in Arnhem Land.Mel- of the Royal Society of South Australia 64: 140– bourne: Macmillan. 231. Thomson DF (1964) Some wood and stone Tindale NB (1941) The hand axe used in the implements of the Bindibu tribe of central Western Desert of Australia. Mankind 3: 37– western Australia. Proceedings of the Prehistoric 41. Society 17: 400–22. Tindale NB (1951) Comments on supposed rep- Thomson DF (1977) Bindibu country.Mel- resentations of giant bird tracks at Pimba. bourne: Thomas Nelson. Paperback edition Records of the South Australian Museum 9: 381– of 1975 publication. 2. Thorley PB (1998a) Pleistocene settlement in the Tindale NB (1957) Culture succession in south Australian arid zone: Occupation of an inland eastern Australia from late Pleistocene to the riverine landscape in the Central Australian present. Records of the South Australian Museum ranges. Antiquity 72: 34–45. 13: 1–49. 386 References

Tindale NB (1959) Ecology of primitive Aborig- Turney CSM, Bird MI and Roberts RG (2001a) inal man in Australia. In A Keast, RL Crocker Elemental δ13C at Allen’s Cave, Nullarbor and CS Christian (eds) Biogeography and ecology Plain, Australia: Assessing post-depositional in Australia,pp.36–51.DenHaag,Nether- disturbance and reconstructing past environ- lands: W Junk. ments. Journal of Quaternary Science 16: 779– Tindale NB (1972) The Pitjandjara. In MG Bic- 84. chieri (ed) Hunters and gatherers today,pp.217– Turney CSM, Bird MI, Fifield LK, Roberts 68. New York: Holt, Rinehart and Winston. RG, Smith MA, Dortch CE et al. (2001b) Tindale NB (1974) Aboriginal tribes of Aus- Early human occupation at Devil’s Lair, south- tralia. Canberra: Australian National Univer- western Australia 50,000 years ago. Quaternary sity Press. Research 55: 3–13. Tindale NB (1977) Adaptive significance of the Twidale CR, Prescott JR, Bourne JA and Panara or grass seed culture of Australia. In Williams FM (2001) Age of desert dunes near RVS Wright (ed) Stone tools as cultural mark- Birdsville, southwest Queensland. Quaternary ers: Change, evolution and complexity,pp.340– Science Reviews 20: 1355–64. 9. Canberra: Australian Institute of Aboriginal Tzedakis PC, Lawson IT, Frogley MR, Hewitt Studies. GM and Preece RC (2002) Buffered tree pop- Tonkinson R (1978) The Mardudjara Aborigines: ulation changes in a Quaternary refugium: Living the dream in Australia’sDesert.NewYork: Evolutionary implications. Science 297: 2044– Holt, Rinehart and Winston. 7. Tooth S (1997) The morphology, dynamics and United Nations Environment Programme Late Quaternary sedimentary history of ephemeral (UNEP) (2006) Global deserts outlook. Nairobi: drainage systems on the northern plains of Central United Nations Environment Programme. Australia. Unpublished PhD thesis, University Vallee P (2004–6) God, guns and government on the of Wollongong. Central Australian frontier. Canberra: Restora- To r r e n c e R ( 2001) Hunter-gatherer technol- tion Books. ogy: Macro- and microscale approaches. In van de Graaff WJE, Crowe RWA, Bunting JA C Panter-Brick, RH Layton and P Rowley- and Jackson MJ (1977) Relict early Cainozoic Conwy (eds) Hunter-gatherers: An interdis- drainages in arid Western Australia. Zeitschrift ciplinary perspective,pp.73–98. Cambridge: fur¨ Geomorphologie N.F. 21: 379–400. Cambridge University Press. van der Kaars S and De Deckker P (2002) A late Trigger BG (1989) A history of archaeological Quaternary pollen record from deep-sea core thought. Cambridge: Cambridge University Fr10/95 GC17 offshore Cape Range penin- Press. sula, northwestern Western Australia. Review Troilett G (1982) Report on the Ethel Gorge sal- of Palaeobotany and Palynology 120: 17–39. vage project. Department of Aboriginal Sites, van der Kaars S, De Deckker P and Gingele FX Western Australian Museum, Perth. (2006)A100,000-year record of annual and Trueman CNG, Field JH, Dortch J, Charles B seasonal rainfall and temperature for north- and Wroe S (2005) Prolonged coexistence of western Australia based on a pollen record humans and megafauna in Pleistocene Aus- obtained offshore. Journal of Quaternary Science tralia. Proceedings of the National Academy of Sci- 21: 879–89. ence 102: 8381–5. van Etten EJB (2009) Inter-annual rainfall vari- Trueman JWH (2001) Does the Lake Mungo 3 ability in arid Australia: Greater than else- mtDNA evidence stand up to analysis? Archae- where? Australian Geographer 40: 107–18. ology in Oceania 36: 163–5. van Holst Pellekaan S (2008) Origins of the Aus- Truswell EM and Harris WK (1982) The Cain- tralian and New Guinean Aborigines. In Ency- ozoic palaeobotanical record in arid Aus- clopedia of Life Sciences. Chichester: John Wiley tralia: Fossil evidence for the origin of an and Sons. arid-adapted flora. In WR Barker and PJM van Holst Pellekaan S and Harding R (2006) Greenslade (eds) Evolution of the flora and fauna Excavating the mitochondrial genome iden- of arid Australia,pp.67–76. Australian System- tifies major haplogroups in Aboriginal Aus- atic Botany Society. Frewville, SA: Peacock tralians. Before Farming 2006/1(article 3): 1– Publications. 26. References 387

Veitch B (1996) Evidence for mid-Holocene (eds) Desert peoples: Archaeological perspectives, change in the Mitchell Plateau, northwest pp. 100–15. Oxford: Blackwell Publishers. Kimberley, Western Australia. In P Veth and Veth P (2005b) Conclusion: Major themes and P Hiscock (eds) Archaeology of Northern Aus- future research directions. In P Veth, M Smith tralia,pp.66–89. St Lucia: Anthropology and P Hiscock (eds) Desert peoples: Archaeologi- Museum, University of Queensland. Te m p us cal perspectives,pp.293–300. Oxford: Blackwell 4. Publishers. Veitch B (1999) Shell middens on the Mitchell Veth P (2006) Social dynamism in the archae- Plateau: A reflection of a wider phenomenon? ology of the Western Desert. In B David, In J Hall and IJ McNiven (eds) Australian B Barker and IJ McNiven (eds) The social coastal archaeology,pp.51–64. Research Papers archaeology of Australian indigenous societies, in Archaeology and Natural History 31.Can- pp. 242–53. Canberra: Aboriginal Studies berra: ANH Publications, Australian National Press. University. Veth P, Aplin K, Wallis L, Manne T, Pulsford T, Veitch B, Hook F and Bradshaw E (2005) A note White E and Chappell A (2007) The archaeology on radiocarbon dates from the Paraburdoo, of Montebello Islands, north-west Australia: Late Mount Brockman and Yandicoogina areas of Quaternary foragers on an arid coastline.British the Hamersley Plateau, Pilbara, Western Aus- Archaeological Reports International Series, tralia. Australian Archaeology 60: 58–61. 1668. Oxford: Archaeopress. Veth P (1987) Martujarra prehistory: Variation Veth P, Hamm G and Lampert RJ (1990) in arid zone adaptations. Australian Archaeology The archaeological significance of the lower 25: 102–11. Cooper Creek. Records of the South Australian Veth PM (1989a) The prehistory of the sandy deserts: Museum 24: 43–66. Spatial and temporal variation in settlement and Veth P, Fullagar R and Gould R (1997) Residue subsistence behaviour within the arid zone of Aus- and use-wear analysis of grinding implements tralia. Unpublished PhD thesis. University of from Puntutjarpa Rockshelter in the Western Western Australia. Desert: Current and proposed research. Aus- Veth PM (1989b) Islands in the Interior: A model tralian Archaeology 44: 23–5. for the colonisation of Australia’s arid zone. Veth P, Hiscock P and Williams A (2011)Are Archaeology in Oceania 24: 81–92. Tulas and ENSO linked in Australia? Australian Veth PM (1993) Islands in the Interior: The dynam- Archaeology 72: 7–14. ics of prehistoric adaptations within the arid zone of Veth P, McDonald J and White B (2008) Dating Australia. Ann Arbor, Michigan: International of Bush Turkey Rockshelter 3 in the Calvert Monographs in Prehistory. Ranges establishes early Holocene occupation Veth P (1995a) Aridity and settlement in north- of the Little Sandy Desert, Western Australia. west Australia. Antiquity 69: 733–46. Australian Archaeology 66: 33–44. Veth P (1995b) Marginal returns and fringe ben- Veth P and O’Connor S (1996) A preliminary efits: Characterising the prehistory of the low- analysis of basal grindstones from the Carnar- land desert of Australia (a reply to Smith). von Range, Little Sandy Desert. Australian Australian Archaeology 40: 32–8. Archaeology 43: 20–2. Veth P (1999) The occupation of arid coastlines Veth P, Smith MA, Bowler J, Fitzsimmons K, during the terminal Pleistocene of Australia. Williams A and Hiscock P (2009)Excava- In J Hall and IJ McNiven (eds) Australian coastal tions at Parnkupirti, Lake Gregory, Great Sandy archaeology,pp.65–72. Canberra: ANH Pub- Desert: OSL dates for occupation before the lications. Last Glacial Maximum. Australian Archaeology Veth P (2000) Origins of the Western Desert lan- 69: 1–10. guage: Convergence in linguistic and archae- Veth P, Smith MA and Haley M (2001) Kaalpi: ological space and time models. Archaeology in The archaeology of an outlying range in the Oceania 35: 11–19. dunefields of the Western Desert. Australian Veth P (2005a) Cycles of aridity and human Archaeology 56: 9–17. mobility: Risk minimisation among late Pleis- Veth P, Smith M and Hiscock P (eds) tocene foragers of the Western Desert, Aus- (2005) Desert peoples: Archaeological perspectives. tralia. In P Veth, M Smith and P Hiscock Oxford: Blackwell Publishers. 388 References von Brandenstein CG and Thomas AP (1974) dunefields. Geographical Review of Japan B, 59: Taruru: Aboriginal song poetry from the Pilbara. 55–67. Adelaide: Rigby. Watchman AL (1991) Age and composition of Wallace P and Wallace N (1977) Killing me oxalate-rich crusts in the Northern Territory, softly: The destruction of a heritage. Melbourne: Australia. Studies in Conservation 36: 24–32. Nelson. Watchman A (1993) Perspectives and potentials Wallis LA (2001) Environmental history of for absolute dating prehistoric rock paintings. northwest Australia based on phytolith analysis Antiquity 67: 58–65. at Carpenter’s Gap 1. Quaternary International Watchman A (2000). A review of the history of 83–85: 103–17. dating rock varnishes. Earth Science Reviews 49: Walshe KA (1994) A taphonomic analysis of the 261–77. vertebrate material from Allen’s Cave: Implications Watchman A (2001) Dating oxalate minerals for Australian arid zone archaeology. Unpublished in rock surface deposits. In M Jones and PhD thesis, Australian National University. P Sheppard (eds) Australasian connections and Walshe KA (1998) Taphonomy of Mungo B new directions: Proceedings of the 7th Australasian assemblage: Indicators for subsistence and Archaeometry Conference,pp.401–11. Research occupation of Lake Mungo. Archaeology in Papers in Anthropology and Linguistics 5. Oceania 33: 200–6. Auckland: University of Auckland. Walshe K (2000) Carnivores, taphonomy and Watson P (1983) This precious foliage: A study of dietary stress at Puntutjarpa, Serpent’s Glen the Aboriginal psychoactive drug Pituri.Ocea- and Intitjikula. Archaeology in Oceania 35: 74– nia Monographs 26. Sydney: University of 81. Sydney. Walshe K (2005) Aboriginal occupation at Watson PL, Luanratana O and Griffin WJ (1983) Hawker Lagoon, southern Flinders Ranges, The ethnopharmacology of Pituri. Journal of South Australia. Australian Archaeology 60: 24– Ethnopharmacology 8: 303–11. 33. Watanabe O, Jouzel J, Johnsen S, Parrenin F, Walshe K (2012) Port Augusta hearth site dated Shoji H and Yoshida N (2003) Homogeneous to 40,000 years. Australian Archaeology 74: 106– climate variability across East Antarctica over 110. the past three glacial cycles. Nature 422: 509– Wanner H, Beer J, Butikofer¨ J, Crowley TJ, 12. Cubasch U, Fluckiger¨ J et al. (2008)Mid-to Webb SG (1989) The Willandra Lakes Hominids. Late Holocene climate change: An overview. Canberra: Department of Prehistory, Aus- Quaternary Science Reviews 27: 1791–1828. tralian National University. Ward I and Larcombe P (2003) A process- Webb S (2006) The first boat people. Cambridge: orientated approach to archaeological site for- Cambridge University Press. mation: Application to semi-arid Northern Webb S (2007) Further research of the Willandra Australia. Journal of Archaeological Science 30: Lakes fossil footprint site, southeastern Aus- 1223–36. tralia. Journal of Human Evolution 52: 711–15. Warner WL (1958) A Black civilization: A social Webb S (2008) Megafauna demography and late study of an Australian tribe.NewYork:Harper Quaternary climatic change in Australia: A and Row. (1969 reprint of revised edition). predisposition to extinction. Boreas 37: 329– Wasson RJ (1982) Landform development in 45. Australia. In WR Barker and PJM Greenslade Webb S (2009) Late Quaternary distribution and (eds) Evolution of the flora and fauna of arid Aus- biogeography of the southern Lake Eyre basin tralia,pp.23–33. Australian Systematic Botany (SLEB) megafauna, South Australia. Boreas 38: Society. Frewville, SA: Peacock Publications. 25–38. Wasson RJ (1983) The Cainozoic history of the Webb S, Cupper ML and Robins R (2006) Pleis- Strzelecki and Simpson dunefields (Australia), tocene human footprints from the Willandra and the origin of the desert dunes. Zeitschrift Lakes, southeastern Australia. Journal of Human fur¨ Geomorphologie Supplementband 45: 85– Evolution 50: 405–13. 115. Webeck K and Pear son SG (2005) Stick-nest rat Wasson RJ (1986) Geomorphology and Qua- middens and a late-Holocene record of White ternary history of the Australian continental Range, central Australia. Holocene 15: 466–71. References 389

Webster J (1982) Faunal analysis and the James among the Kalahari San. Journal of Anthropo- Range East site. Unpublished BA (Hons) thesis, logical Archaeology 3: 190–234. University of Sydney. Wiessner P (2002) Hunting, healing, and hxaro Wells FE and Lalli CM (2003) Aspects of the exchange – A long-term perspective on !Kung ecology of the mudwhelks Terebralia palustris ( Ju/‘hoansi) large-game hunting. Evolution and T. semistriata in northwestern Australia. In and Human Behavior 23: 407–36. FE Wells, DI Walker and DS Jones (eds) The Wills WJ (1863) A successful exploration through marine flora and fauna of Dampier, Western Aus- the interior of Australia, from Melbourne to the tralia,pp.193–208. Perth: Western Australian Gulf of Carpentaria. London: Richard Bentley Museum. [Facsimile, 1996, The Friends of the State Wells RT and Callen RA (eds) (1986) The Lake Library of South Australia]. Eyre Basin – Cainozoic sediments, fossil vertebrates Williams AN (2012) The use of summed radio- and plants, landforms, silcretes and climatic impli- carbon probability distributions in archaeol- cations. Australasian Sedimentologists Group Field ogy:Areviewofmethods.Journal of Archaeo- Guide Series No. 4. Sydney: Geological Society logical Science 39: 578–89. of Australia. Williams E (1988) The archaeology of the Wells RT and Tedford RH (1995) Sthenurus Cooper Basin: Report on fieldwork. Records (Macropodidae: Marsupialia) from the Pleis- of the South Australian Museum 22: 53–62. tocene of Lake Callabonna, South Australia. Williams E (1998) The archaeology of lake sys- Bulletin of the American Museum of Natural His- tems in the middle Cooper Basin, north- tory 225: 1–111. eastern South Australia. Records of the South Wendorf F, Schild R and Close E (and associates) Australian Museum 30: 69–91. (1993) Egypt during the last interglacial: The Mid- Williams GE and Polach HA (1969) The evalu- dle Paleolithic of Bir Tarfawi and Bir Sahara East. ation of 14C ages for soil carbonate from the New York: Plenum Press. arid zone. Earth and Planetary Science Letters 7: Wendt WE (1972) Preliminary report on an 240–2. archaeological research programme in South Wilmsen EN (1989) Land filled with flies: A politi- West Africa. Cimbebasia (B) 2: 1–61. cal economy of the Kalahari. Chicago: University Westoby M, Walker B and Noy-Meir I (1989) of Chicago Press. Opportunistic management for rangelands not Wilson S (1982) Appendix A. In S Forbes (ed) at equilibrium. Journal of Range Management Aboriginal rock engravings at N’Dahla Gorge: 42: 266–74. An analysis of a Central Australian ‘Panarami- White E (2007) Lithics from Noala Cave 1aand tee Style’ rock art site. Unpublished BLitt thesis, Haynes Cave 4, Montebello Islands. In P Veth, Australian National University. K Aplin, L Wallis, T Manne, T Pulsford, E Wingate MTD, Pirajno F and Morris PA (2004) White and A Chappell (eds) The archaeology Warakurna large igneous province: A new of Montebello Islands, north-west Australia: Late Mesoproterozoic large igneous province in Quaternary foragers on an arid coastline,pp.63– west-central Australia. Geology 32: 105–8. 84. British Archaeological Reports Interna- Winnecke C (1884) Mr Winnecke’s explorations tional Series, 1668. Oxford: Archaeopress. during 1883. South Australian Parliamentary White JP and O’Connell JF (1982) A prehistory Papers 39: 1–16. of Australia, New Guinea and Sahul. Sydney: Winterhalder B (2001) The behavioural ecol- Academic Press. ogy of hunter-gatherers. In C Panter-Brick, White N (1997) Genes, languages, and land- RH Layton and P Rowley-Conwy (eds) scapes in Australia. In P McConvell and N Hunter-gatherers: An interdisciplinary perspective, Evans (eds) Archaeology and linguistics: Aborig- pp. 12–38. Cambridge: Cambridge University inal Australia in global perspective,pp.45–81. Press. Oxford: Oxford University Press. Winterhalder B and Kennett DJ (2006)Behav- Whitehouse FW (1940) Studies in the late geologi- ioral ecology and the transition from hunting cal history of Queensland. University of Queens- and gathering to agriculture. In DJ Kennett land Papers – Geology 2(1), Brisbane. and B Winterhalder (eds) Behavioral ecology and Wiessner P (1984) Reconsidering the the transition to agriculture,pp.1–21.Berkeley: behavioural basis for style: A case study University of California Press. 390 References

Wischusen JDH, Fifield LK and Cresswell RG Wright BJ (1972) Rock engravings of striped (2004) Hydrogeology of Palm Valley, central mammals: The Pilbara region, Western Aus- Australia: A Pleistocene flora refuge? Journal of tralia. Archaeology and Physical Anthropology in Hydrology 293: 20–46. Oceania 7: 15–23. Wobst HM (1978) The archaeo-ethnology of Wright RVS (ed) (1971) Archaeology of the Gallus hunter-gatherers or the tyranny of the ethno- site, Koonalda Cave. Australian Aboriginal Stud- graphic record in archaeology. American Antiq- ies No. 26. Canberra: Australian Institute of uity 43: 303–9. Aboriginal Studies. Wobst M (1977) Stylistic behaviour and informa- Wroe S, Crowther M, Dortch J and Chong J tion exchange. In CE Cleland (ed) Papers for (2004) The size of the largest marsupial and the Director: Research essays in honour of James B why it matters. Proceedings of the Royal Society Griffin,pp.317–42. Ann Arbor MI: Museum London B 271: 534–6. of Anthropology, University of Michigan. Wur m SA (1972) Languages of Australia and Woodburne MO, Macfadden BJ, Case JA, Tasmania. The Hague: Mouton. Springer MS, Pledge NS, Power JD et al. Wyrwoll K-H, Hopwood J and McKenzie NL (1993) Land mammal biostratigraphy and (1992) The Holocene paleohydrology and cli- magnetostratigraphy of the Etadunna forma- matic history of the northern Great Sandy tion (late Oligocene) of South Australia. Jour- Desert – Fitzroy Trough: With special ref- nal of Vertebrate Paleontology 13: 483–515. erence to the history of the northwest Aus- Woodward HP (1914) A geological reconnais- tralian monsoon. Climatic Change 22: 47– sance of a portion of the Murchison goldfields. 65. Bulletin of the Geological Survey of Western Aus- Wyrwoll K-H and Miller GH (2001) Initiation tralia 57: 71–89. of the Australian summer monsoon 14,000 Worrall RH (1994) Rock art survey, Roma Gorge years ago. Quaternary International 83–85: 119– vicinity, West MacDonnell National Park, Central 28. Australia. Unpublished report. Conservation Yengoyan AA (1968) Demographic and ecolog- Commission of the Northern Territory, Alice ical influences on Aboriginal Australian mar- Springs. riage sections. In RB Lee and I DeVore (eds) Worsnop T (1897) The prehistoric arts, manu- Man the hunter,pp.185–99. Chicago: Aldine factures, works, weapons, etc., of the Aborigines Publishing Company. of Australia. Adelaide: Government Printer, Yuan D, Cheng H, Edwards RL, Dykoski South Australia. CA, Kelly MJ, Zhang M et al. (2004) Wright BJ (1968) Rock art of the Pilbara region, Timing, duration, and transitions of the North-West Australia. Canberra: Australian Last Interglacial Asian monsoon. Science 304: Institute of Aboriginal Studies. 575–8. INDEX

Pagenumberswithan‘f’indicateafigure abandonment of sites, 110–114, 131f, Aboriginal societies (cont.) 149–150 desert cultural systems, 208–211 Aboriginal art.Seerock art development of, 22 Aboriginal ceremonies.Seeceremonies dynamics of hunter-gatherer societies, Aboriginal languages. 29–30, 202–205, 305–308 209–210, 333–335 economy and population changes, 35, 42 Arandic, 30 foraging risks, 196–197 dialects, 340 formation of desert culture, 208–211 groups, 270, 333–335, 335f fragmentation during LGM, 114–115 Karrwan, 203f, 204 gender roles in, 210 Pama-Nyungan, 202–205, 263 gerontocratic privilege, 300 Ta n g k i c , 203f, 204 gift exchange, 270 Aboriginal ontology, 220–221 Holocene, 163–164, 174–176 Aboriginal people, 9, 12, 34–36, 92–94 impact of LGM on, 110, 156 adapting to food supply, 12 influence of increased social activity on ceremonies, large, 258–259 production systems, 42 colonisation of deserts by, 76–78 internal social dynamics, 305–308 as ecological agents, 40–41 intragroup exchanges of goods, 270–271 genetic data, 92, 334 population growth of, 208–209 impact of LGM on, 154–156 role of dogs in, 208 large gatherings of, 101 role of mythology in, 216–217 length of desert occupation, 39–40 shared graphic systems of, 214 physical variation among, 28 stylistic behaviour in, 224 late Pleistocene range of, 98–99 subsections (‘skins’), 336 perspective on fossil bones, 69 trade and exchange within, 266–270 responses to environmental stresses, 149 uniformity of, 21 skeletal remains of, 92–94, 153, 183–184 use of fire by, 324 useofochresby,182, 276 Western Desert, 333–335 use of stone tools by, 136–138 Aboriginal tribal or ethnographic groups as vector for spread of seeds, 105 Arrernte, 10, 17–18, 19f, 20, 22, 23, 42 Aboriginal societies, 18, 44, 70, 106, 107–108, Diyari, 10, 69, 266, 277, 293 207–208, 211.Seealsohunter-gatherers Jandrowonta, 284 conservatism of, 31 Jauruworka, 284 demographic trends, 338–339 Jingili, 270

391 392 Index

Aboriginal tribal or ethnographic groups (cont.) archaeological research (cont.) Kalkadoon, 259 misunderstanding of rock art, 221 Mudburra, 270 political aspects of research, 14–15 Pintupi, 10, 109, 136, 154, 270, 324, 334, 335 problems with identifying seed-grinding Pitjantjatjara, 10, 15, 334 tools, 200–201 Thirrari, 69 recurring themes in, 43–44 Wangkangurru, 10, 69, 182, 277, 293 with bush communities, 15 Warlpiri, 217, 220, 270, 280 archaeological sites, 13, 15, 32, 35, 46f, Warrumungu, 20 78–92.Seealsoindividual site names, Western Desert, 333–335 e.g., Puritjarra site Yandruwantha, 10, 277, 286, 293 difficulties in dating, 225–227 Yankunytjatjara, 10, 15, 334 late Holocene, 310–312 access to country, 123 late Pleistocene, 124–133 adhesives, 195, 287 management of, 15 adzes, 29, 31, 164, 185, 186f, 186, 189, 190f, significance of rock art at, 224–225 194–196.Seealsotula adzes ‘archaic faces’ rock art, 229f, 242, 245f, 252–253 aeolian landforms, 6, 25, 36 arid rivers region, 5f, 7, 8, 57–58 aeolian responses to climate change, 59f arid zone, 5f, 5, 14, 60, 73f, 75, 124f, 144–146, aeolian sediments, 122 190f.Seealsodeserts aerial mapping, 27 biota, 38–40 Africa, 1–3, 11, 105, 265, 300 boundaries, 76 Akerman, K, 22, 274, 275 changes to, 26, 76 Allen, H, 34, 71, 72, 74 defined, 6 Allens Cave site, 73f, 88–89, 111f, 151, 152, ecology of, 43 175–176, 177f lakes and saltlakes, 52–53 Amazon basin, 116 last glacial maximum, 110, 120 American deserts, 31, 105, 163, 265 megafauna extinction in, 103–104 Anadara middens, 172–175, 246, 250, 310 southeastern sector, sites, 322–323 ancestors, 22, 23, 214, 216–217, 220–221, 340. woodlands, 8 See also dreaming aridity Ancient Hunters and Their Modern Representatives, adaptations to by people, 34, 35 22 as barrier to settlement, 34 animal translocation, dogs, 206–207 aridity index, 5 Anna Creek grindstone quarry, 283–284 Arkaba site, 143 Antarctic Cold Reversal (ACR), 119 Arnhem Land, 71, 216, 254, 270–271, 295, 300 anthropogenic fires, 103–104, 303, 324–326, Arrernte people, 10, 17, 18, 19f, 20, 22, 23, 42 328f, 333, 336.Seealsofire artefacts.Seestone tools, wooden artefacts anthropomorphs, 232t–233t, 241–242, 247, Arunta: A Study of a Stone Age People, The, 23, 249, 250t, 251, 255, 259–261 304 Anvil Creek ochre quarry, 281 Ash, JE, 59 Arandic groups, 217 Ashwin, AC, 331 Arandic languages, 10, 30, 334 Atnetye site, 321 archaeological research, 1–4, 12–15, 46, 157. atyelpe ceremonies, 23, 258 See also research questions Australia Twice Traversed, 302 analysis of faunal bone, 327t Australian desert culture.SeeAboriginal analysis of rock art, 212–216, 221–225, 227, societies; desert cultures 252 Australian Institute of Aboriginal Studies, 157 approaches to trade and exchange, 271–274 axes.Seestone axes arid zones and stone artefacts, 322–323 Central Australian sites, 312–321 backed artefacts, 185–186, 187f classical ethnographies, 304–305 Balcoracana Creek site, 144, 183 fire and patch burning, 323–326 baler shell, 98, 274–275.Seealsoshell artefacts hunter-gatherer social dynamics, 305–308 Balme, J, 201 late Holocene sites, 310–312 Barton, H, 294 Index 393

Basedow, H, 23, 24, 212, 213 Central Australia (cont.) Beaton, JM, 74 graphic systems of, 258–259 behavioural ecology, human, 328–329 Holocene changes to groundwater, 121, 122, Bender, B, 306 160 Berndt, RM, 215f, 10, 24, 216, 220, 259, 270, lakes of, 56–57 271 rock art from, 227–243, 254–258, 255f biogeography, 89, 113, 115–117 sites in 89–91, 133–139, 134f, 181–182, models, 40 312–318 refugia, 7 Central Australian Expedition (C Sturt), 20, 21f zones, 38 ceremonies, 17, 210, 218, 258–259, 305–307, biota, 3, 8, 40 321.Seealsoingkwere biotic activity in deserts, 9 atyelpe, 23, 258 Bird, DW, 41, 325, 326, 337 ‘increase’, 216 Birdsell, JB, 28, 72, 75, 297f, 334 initiation, 216, 217 boat-building, 71 kunapipi, 217 body painting, 218, 337f rain-making, 274 Bookartoo.SeePukardu for regulating knowledge 216, 217 boomerangs, 186, 194, 298 for social integration, 224 Bowdler, S, 34, 48 chenopods, 7, 65, 76, 104, 143, 150, 200 Bowler, JM, 32, 36, 52, 53f, 94, 110 chronologies, 14, 22, 26, 42, 72, 78–92, 312, 334 Brachina Gorge site, 121 of axes, 289f, 291–292 Brook, BW, 312 of deserts, 43, 48 Brooks, AS, 1 of dunefields, 37, 43, 58–59 Brown, HYL, 212 of rock art, 234–238, 243–245, 261–262 burials, 14, 32, 81, 92–93, 93f, 153, 183, 299f clan estates, 12, 280 Burke and Wills (explorers), 20, 288, 292, 330 ‘classical refugia’, 116 Burkes Cave site, 323 Clendon, M, 204 burning of desert vegetation.Seeanthropogenic climate change, 13, 36, 53f, 59f, 110, 158–159, fires; fire 306, 308–309, 324 Burrup Peninsula sites, 172, 241, 244, 245f, and anthropogenic fires, 103 246–247, 251, 253 Pleistocene, 51 bush foods, 12, 164, 196, 202, 209, 308, 330 and the LGM, 119–120 Bush Turkey 3 site, 115, 145–146, 179, 246, 321 climate indices, 5 Byrne, M, 117 climatic belts, 27f, 26–28 climatic shifts, 4, 52–53, 113 Calvert Range sites, 240–242, 246, 251 climatic variability, 51f, 74–75 Cane, S, 156, 176 ‘climbing men’ motif, 242, 243f, 251 canoes, 288 ‘clinal model’ of re-occupation, 115 Cape Range sites, 86–87 ‘coastal colonisation hypothesis’, 34 Capell, A, 30 coastal regions and resources, 6–7, 35, 102, 174, carbon-14 (14C) dating.Seeradiocarbon 262 dating Codding, BF, 326–329 carbon isotope analysis of teeth, 65 Colless Creek Cave site, 147–148 Carnarvon Range, 91, 146 colonial research, 18, 42–43 Carpentaria region, 116, 147, 149 colonisation, 34–36, 70–74, 76–78, 106–107, Carpenters Gap sites, 98, 102, 160, 201 177–178, 335 Carpenters Gap 1, 78, 107, 143, 200 ecological impacts of, 102–105 Carpenters Gap 3, 143 limiting factors for, 72–74 cation-ratio dating, 235–236 routes, 77–78 Central Australia, 5f, 6–7, 14, 121, 133–139, patterns of settlement, 117, 320 314f, 334 translocation of plant and animal species, environmental changes in, 50, 308–310 104–105 fluvial history of, 57 composite tools, 185, 194 glacial refugia in, 116 Conkey, MW, 224 394 Index

‘contact zones’, 110, 153 desert people.SeeAboriginal people continental desiccation, 19, 22, 31, 34, 49–50 desert refugia, 38, 117 contraction desert societies.SeeAboriginal societies of deserts, 160–162 desert songs, 216, 266 of settlement, 110–114 ‘desert transformation’ model, 47–48 convex scrapers, 97–98, 135 desert varnish.Seemineral skins Coongie Lakes sites, 225, 322, 323 desertification of Australia, 48–51 Cooper Creek, 20, 57–58, 145, 266, 284, 322, deserts, 5f, 4–8, 23, 43, 48–51, 177–184.Seealso 329–331 arid zones corporate descent groups, 10, 75, 156, 217 as barriers to settlement, 39–40 corridors of dispersal, 66, 77, 113, 117 boundaries, 2f, 4 Crawford, IM, 279f, 279–280, 282 contraction of, 160–162 cremation, 81, 92–93 defined, 4–8 Cribb, J, 5, 8 dunes, 24–26, 58–60 Crocker, RL, 26, 34, 38, 116, 119 ecosystems, 122–123, 158–163 cryptic refugia, 112f, 116, 117, 119, 156 fauna, 50, 205–208, 210 Cuckadoo site, 143, 182, 292 as human artefacts, 40–41 Cuddie Springs site, 85f, 83–86, 104, 107, 111f, as limiting factor on settlement, 31 200 lowlands, 44, 112–116 cultural development, autonomous, 22 margins, 6, 35, 43–44, 146–154 cultural ecology, 11 modern distribution of, 2f cultural succession models, 29 plant distribution in, 38 Panara, 198 rock-art culture of, 213–214 cultural variability, 21 as ‘spread zone’, 209–210 uplands, 44, 117–119, 143–144 Dampier, W (explorer), 20 Devils Lair site, 71, 77, 78 Dampier region, 6, 171–173, 239, 240, dialects.SeeAboriginal languages 243–244, 246–247, 260, 262 Diamantina River, 7, 36, 57, 77 Darling River, 5f, 22, 34, 80–83, 152–153, 183, diaspora following European contact, 1, 321–323 106 Darwin, C, 37 Dickins, J, 218 Davidson, DS, 23, 213, 271, 278, 280 differentiation of tribes, 21 Davidson, I, 272, 294 digging sticks, 194, 210, 279 dead heart of Australia, 18–20 diluvialist theories, 25 Dead Heart of Australia, The, 18–19, 46 dingoes, 205–208, 210, 339 De Deckker, P, 103 Diprotodon optatum, 25f, 45, 46, 46f, 50, 61, deflation of land surfaces, 13, 32, 56, 61, 84–85, 65 120 dispersal, 1, 74–76, 105 deglacial period, 158–159 exotic animal models of, 75 demographic growth, 208–209, 273, 306–307, of plant species, 104–105 335 distribution of natural resources, 11 Dempseys Lake site, 82–83 diversity of deserts, 6 ‘desert culture’ model, 157–158, 163 division of labour, by gender, 210, 329, desert cultures, 31, 208–211.SeealsoAboriginal 340 societies; hunter-gatherers Diyari people, 266, 277, 293 affected by environmental change, 308 Djadjiling site, 87, 140f, 141 and fire, 323–326 dogs in Aboriginal societies, 208.Seealso foraging, 326–329, 336–338 dingoes seed-based economies, 329–333 ‘down-the-line’ exchange systems, 271, 274, social history of, 338–341 296–297 subsection systems, 336 dreaming, 15, 22, 44, 157, 219, 220, 264 desert ethnographies, 43, 304–305 altyerrenge, 22–23 desert exchange systems, 296–301.Seealsotrade defined, 216 and exchange honey ant, 218f Index 395

Kardimarkara (the Rainbow Serpent), 69 endemic plant species, 7, 37 and proprietary relationships, 340 endscrapers, 186, 192, 193 and seed-foods, 210–211 engraved stone plaques, 217 significance of red ochre in, 276 environmental change, 308–310.Seealso totemic geography, 216 climate change tjukurpa, 220 environmental stochasticity, 115 Dromaius novaehollandiae (emu) eggshell. 61, 99, ephemeral waters, 11, 56, 122 326 ‘Eremian’ zones, 38 at Silver Dollar site, 87 Etadunna formation, 49, 50 and carbon isotopes, 61, 103 ethnographic distribution patterns, goods, 267f, from Lake Menindee site, 82 272 revealing vegetation change, 103 ethnographies, 304–305, 329 dromornithids, 50, 66 Etten, EJB, van, 8 drylands, 4, 120–122.Seealsoarid zones; eucalypt woodlands, 103, 175 deserts Europe, glacial-age, 116 fauna of, 61–67 European contact and exploration (in interior), Duboisia hopwoodii (source of pituri), 292–294 42–43, 109, 292, 294, 302–303 dunefields, 6, 24–26, 37, 59f, 58–60, 68, 73f European settlement (of interior), aerial mapping of, 27 277 as biogeographic barriers, 115 eustatic changes, impact of, 151 problems of archaeological investigation in, Evans, N, 204 145–146 evaporation rates, 122–123 dust evolutionary biodiversity, 18, 50, 117 emissions, 309 exchange.Seetrade and exchange fluxes, 121 expedient grindstones.Seegrindstones mantles, 121 explorers, European, 18, 20–22, 197, 302–303, transport patterns, 122 322, 329 dynamics of hunter-gatherer societies, 305–308 extinct fauna, 22, 24, 41 extinction of arid-zone megafauna, 65–66, early Holocene epoch, 38 103–105, 324 defined, 158 eyes, depicted in rock art, 252–253 language change during, 204–205 Eyre, Lake.SeeLake Eyre rock-art traditions, 228–253 stone assemblages, 184–185 Fanning, PC, 323 early Pleistocene epoch, 23, 48, 51 faunal assemblages, 326–329 ecological adaptations, 8 ‘fern’ headdresses in rock art, 259 ecological impacts, 102–105 fertility cults, 216–217 economic intensification, 332 fire, 40–42, 102–103, 302, 303f, 323–326 economies, 99–102, 149, 222, 332–333, ‘firestick farming’, 40 336–338 first contact, Pintupi Nine, 109–110 foraging, 326–328, 332–333, 336–338, fish otoliths, 101 339 fishing, 77, 102, 171 Holocene changes in, 209 flaked-stone assemblages.Seestone artefact Holocene coastal, 174 (lithic) assemblages political, 300 Flinders Ranges, 43, 112, 116, 121, seed-based, 202, 329–333 143 Edgeworth David, TW, 19 Holocene sites, 182–183 Eight Mile Creek site, 234 Pukardu mine, 276–278 El Nino–Southern˜ Oscillation (ENSO), 158, flint quarrying, 150, 152.Seealsoquarries; 162–163, 308–310, 335–336 Koonalda Cave site embedded procurement, 296 flooding, 8, 57, 77, 121, 309 Emily Gap site, 220, 221f, 256 floodplains, 7 emu eggshell.SeeDromaius novaehollandiae fluvial sediments, 25 (emu) eggshell food exchange, 271 396 Index food preparation, 197–198, 320, 329–331 global diaspora of anatomically modern food sources, 9, 11–12, 34, 99, 320.Seealso humans, 106 bush foods; hunting global warming, 158–159.Seealsoclimate affected by fire, 324–325 change macropods, 327–329 gorge systems, 78–80, 99, 143, 147–149, 178, mangrove, 165–170 232t seeds, 197–198, 330–331 Gould, RA, 12, 30, 31, 157–158, 163, 179 small game, 326–327 granaries, 332 food storage, 196, 331–332 granivory, 197 footprints, fossilised, 151f, 153–154, 176 graphic traditions, 213f, 214, 218–220, 262, foraging, 4, 12, 99–102, 321f, 323, 324–329. 305, 340.Seealsorock art See also hunter-gatherers; seed grass seeds as food source, 40, 197–198, 282. harvesting and preparation See also seed harvesting and long-term changes in, 339 preparation risks, 196–197 grasslands, 50, 60, 68 territories, 11, 123 GRE8 site, 80, 99, 147, 148–149 Fortescue River, 88, 140 ‘Great Australian Arid Period’, 24–28, 38 fossil beds, 22 Great Sandy Desert, 6, 89 fossil landforms, 26 Great Victoria Desert, 6 fossils.SeealsoDromaius novaehollandiae (emu) Gregory, AC, 330, 331 eggshell; Genyornis newtoni; skeletal Gregory River, 80 remains Griffith Taylor, T, 26 bones, 55, 69, 84, 153, 326–327 grindstone quarries, 282–287 fauna, 18–19 chronology of, 279f, 287 human footprints, 153–154 grindstones, 32, 97–98, 146, 173, 182, 328f. trackways, 65, 154 See also seed harvesting and freshwater environments, 32, 49, 67, 68–69, preparation 152–154 from Cuddie Springs, 83, 98 distribution of, 272 Gamble, C, 74, 107, 116, 222 expedient, 199–200 Gason, S (Mounted Constable), 266, 276, 277, matrilineal ownership of millstones, 210 293 millstones, 190f, 198, 201, 282–287 gender division of labour, 210, 329 mortars, 199, 332 genetic data, 92, 205, 334 Puritjarra grindstone assemblage, 201–202 genetic drift, 94 replacing grinding tools, 282 Gentilli, J, 27 seed-grinders, 332–333 Genyornis newtoni, 66–67, 104 tjungguri topstones (mullers), 282 fossilised eggshell, 41 ground cover, reduction of, 121 and human populations, 68 groundwater, 121, 122, 160 geometric microliths, 185–186 group compositions, 12 gerontocratic privilege, 300 gibber plains, 6 Haffer, J, 116 Gibson Desert, 6 hafted tools, 185, 186–191, 193–195, 295f, 329. gift economies, 197, 267 See also stone axes Australia vs Kalahari, 300 hafting cement, 188–191, 193, 195 Giles, E (explorer), 302–303 Hamersley plateau, 140–141 gill nets, 101 Hamilton, A, 210 Gillen, FJ, 4, 22, 43 hand stencils, 220, 256f, 256 glacial aridity, 32 Harrison, R, 172 glacial maximum harvesting seeds, 197–198 glacial refugia, 40, 112f, 114–117 hatchets.Seestone axes glaciers, 121 Hawker Lagoon site, 143 Glen Thirsty site, 317f, 320, 337f Hayden, B, 136, 288 Index 397 hayricks, 331 hunter-gatherers (cont.) hearths, 13, 32, 82, 132, 134f, 322, 323 and ephemeral lakes, 56 Allens Cave, 175 equilibrium models for, 31 JSN site, 145 foraging patterns of, 326–329 Lake Eyre basin, 144–145, 182 granivory, 197 Pilbara, 142 Holocene, 163–164, 208 Riwi, 78 mobility of, 11–12, 306 Heatwole, H, 117 number of contacts, for exchange systems Helen Springs quarry.SeeKurutiti 297f Hiscock, P, 48, 113, 186, 196, 272, 289–290, population density and colonisation, 75, 294 105–106, 307–308 historical linguistics, 29–30 production of axes by, 290 history of deserts, written, 42–44 responses to foraging risks, 196–197 Hodgkinson, WO, 330 seed-grinding economies, 202 Holdaway, SJ, 184, 322–323 and shellfish, 55 Holmgren, M, 309 social history of, 338–341 Holocene epoch, 26, 31, 158.Seealsoearly, subsection systems, 336 mid-, and late Holocene epochs use of Allens Cave, 88–89 adaptations, 115 use of sandy deserts, 146 Anadara middens, 172–175 hunting, 196, 303f.Seealsomacropods; arid west coast, 165–172 marsupials aridity, 31 dogs, 209, 339 artefacts, 82–84, 184–197 spears, 194, 195 contraction of deserts, 160–162 hydrological systems, 51f–53f, 121 desert interior, 177–184 development of desert culture, 166, 208 ice-core records, 52, 158 formation of mineral skins during, 226–227 iconic motifs in rock art, 251, 262 hunter-gatherer societies, 163–164 ‘increase’ ceremonies, 216 Nullarbor coast and hinterland, 175–177 India, 106 rock-art traditions, 227–254 Indo-Malaysian archipelago, 70–71 sites, 14, 23 infertile soils, 9 vegetation responses in deserts, 162 ingkwere, 17, 19f, 20, 42, 44, 341.Seealso homelands movement, 109 ceremonies honey ant ‘dreaming’, 218f initiation ceremonies, 216, 217 Hookina Creek site, 143 inland environments, 39, 56–58, 119–120 Horn Scientific Expedition, 20, 37, 294 inland Pilbara rock-art sites, 246 ‘horsehoof’ cores, 96 Innamincka grindstone quarries, 284–285 Horton, DR, 35 Inscribed landscape, questions about, 263 Howitt, AW, 28, 331 interglacial, 48, 51–52, 60–61, 67–68.Seealso Hughes, PJ, 112 marine isotope stages human behavioural ecology, 328–329 interpretation of sites and artefacts, 130–133, human colonisation of deserts, 56, 68 200–201 human ecology, 10–12, 102–105, 112f, 122–124 Intirtekwerle site, 295, 314, 317f, 318, 327, 328 human fossils.Seeskeletal remains invasion biology, 74–76 human settlement, 34–36 invasion pressure, 74 hummock grasslands, 8, 38, 120, 121, 162, 320, invertebrates, desert, 9 331 Irtikiri site, 317 hunter-gatherers, 10, 31, 69, 74, 98, 295f, Islands in the Interior, 39 305–308 adaptations by, 114, 308–310 James Range East site.SeeIntirtekwerle site Aterian, 3 Jandrowonta people.SeeYandruwantha people distinguishing features of Australian, 305 Jansz rockshelter, 86, 149 during LGM, 123, 124 Jauruworka people, 284 398 Index

Javanese vs Australian human fossils, 153 Lake Eyre basin, 144–145, 163 Jingili people, 270 grindstone quarries, 282–287 Johnson, CN, 311 Holocene sites in, 181–182 Johnston, TH, 99 Panaramitee-style rock art, 228–241 Jones, R, 40, 70, 204 quarries in, 267f JSN site, 145 Lake Frome, 37, 52, 57, 112, 121, Juukan sites, 87, 141 160 Lake Garnpung, footprints, 154 Kaalpi site, 246, 321 Lake Gregory, 60, 106 Kaars, S, van der, 103 Lake Lewis, 52, 53, 121 Kalkadoon people, 259 Lake Mackay, 109 Kardimarkara (the Rainbow Serpent), Lake Menindee, 66, 82 69 Lake Moondarra 1 (LM1) axe quarries, Karolta (Mannahill) site, 212, 235 289–290 Karrku ochre mine, 280–281 Lake Mulligan.SeeLake Callabonna karst plains, 7, 88, 150 Lake Mungo, 31–34, 68, 80–82, 93, 101 Katapiri fauna, 62–64, 65–66, 103 Lake Nitchie, 183–184 Katapiri formation, 58 Lake Tandou, 101, 107, 153 Keast, A, 27 lakes.Seealsoindividual lake names, e.g., Keble, RA, 27 Willandra Lakes Kimber, RG, 114 Bowler’s ‘hydrological threshold’ for Kimberley, 78, 160 Australian, 51f King Sound, 121 prehistoric history of, 52–58 Kings Canyon, 162 Lambeck, K, 119 kinship systems, 305, 336 Lampert, RJ, 112 Kirk, RL, 205 land ownership and tenure, 12, 305, 340 Kocurek’s model of aeolian responses to climate landesque capital, 41 change, 55f, 59 Landscapes, Rock-art and the Dreaming, 265 Kondo, T, 104–105 land use, 90, 112f, 113f, 124–130, 142–143, 154, Koonalda Cave site, 88, 150–151, 312 152 Lang, A, 23 kopi mourning caps, 299f language groups, 270, 319f, 332–336 Kulpi Mara site, 90–91, 134f, 133–134, 135f, languages, 199f, 210 317 dialects, 333 kunapipi ceremonies, 217, 253 Karrwan, 204 Kurutiti grindstone quarry, 239, 286–287 Pama-Nyungan, 202–205 Ta n g k i c , 204 lacustrine resources, 35, 99–101, 114 Wati language group, 30, 319f, 332–336 Lake Amadeus, 52 last glacial maximum (LGM), 111f, 110–112, Lake Callabonna, 46f, 45–46, 64–65, 69 116, 117–119, 339 Lake Carpentaria, 159 archaeological sites from, 124–130 Lake Eyre, 27, 53f, 54f, 53–56, 67, 120, 160 defined, 119 drainage system, 7 impact of, 118, 120–123, 154 dune-building in, 120 impact of, in Australian desert vs Europe, during MIS3, 68 40 Etadunna formation, 49 and the Pilbara, 141 exploration of, 18 as unstable phenomenon, 119–120 extinction of megafauna, 103–104 last millennium as fossil ‘dead’ landscape, 43 foraging economy during, 336–339 Holocene occupation of, 182–184 rock art of, 254–259 influence on early research, 43 late Holocene epoch, 42, 164, 311f, 310–312, palaeovegatation record of, 61 314f rainfall at, 6 defined, 158 and SLEADS program, 36–37 foraging risk in, 196–197 Index 399

rock art, 222, 260–261 mangrove forests, 160, 165–172 seed harvesting, 197–198 Anadara middens, 172–175 small-tool phase, 196 economies based on, 250, 251–252 stone assemblages, 184–193 Mannahill (Karolta) site, 212, 235 late Pleistocene epoch, 4, 6, 14, 24, 32, 34, 37, Marillana A site, 178 39.SeealsoCuddie Springs site, marine isotope stages, 43, 51f, 51 Puritjarra site MIS2, 119 arid-zone sites, 73f MIS3, 68–69, 71, 76, 244 colonisation during, 72, 105–106 MIS5, 52, 56, 57, 61, 105 firing of vegetation, 102 marine resources, 101–102, 174, 176–177 hunter-gatherer groups, 39, 98 maritime air masses, 120 Panaramitee-style rock engravings, 222 marsupials, 9, 21f, 45, 165, 324, 327–329 residential mobility of groups, 98 ‘radiation’ of, 50 late Quaternary megafauna, 61, 62f.Seealso Marun, LH, 175 megafauna Marwick, B, 141 lateritic gravel plains, 6 material culture, 340.Seealsostone artefact Lawrence, D, 272 (lithic) assemblages, stone tools layer interfaces, 131–132 Maynard, L, 228, 231 leilira blades, trade in, 294–296 megafauna, 41, 50, 83 LGM.Seelast glacial maximum Diprotodon optatum, 25f, 45, 46, 46f, 50, 61, limestone gorge systems, 78, 80 65 linguistic changes, 203, 205 distribution of, 46f–47f linguistics, historical 29–30 of drylands, 61–67 lithic assemblages.Seestone artefact (lithic) extinct fauna, 22, 24, 41 assemblages extinction of, 41, 65–66, 103–105, 324 lithic economy, 195 fossils, 34, 84 Little Sandy Desert, 6, 259 Genyornis newtoni, 41, 66–67, 68, 104 Livistona palm trees, 49, 104–105, 140 Katapiri fauna, 61–64, 65–66, 103 local social evolution, 23 Lake Callabonna fossils, 18 ‘locales’ of activity vs ‘places’, 107 population ecology of megafauna, 65 long-distance exchange systems, 272, 296–301 mega-lakes, 34, 46 Lorblanchet, M, 172 metamorphic rocks, 7 Louie Creek Cave site, 147, 148 microhabitats, 116 Lourandos, H, 42, 273 microliths, 29, 31, 157 luminescence dating methods, 14, 48 backed artefacts, 186f, 185–186, 187f lunettes, 33f, 32–34, 143, 153 geometric microliths, 185–186 middens, 32, 173f, 172–175 Macaulay, V, 106 mid-Holocene epoch, 14, 29 McBrearty, S, 1 defined, 158 McBryde, I, 271–272 fauna, 175–176, 205–208 McCarthy, FD, 271 mangrove forests, 165 McConvell, P, 204 Pama-Nyungan language, 202–205 MacDonnell Ranges, 6–7, 39, 89 rock-art traditions, 221f, 227–254 Macintosh, NWG, 183 stone assemblages, 184–185 McKinlay, J (explorer), 322 migration, 21–22, 28, 34, 334 macropods, 62, 66, 99 earliest settlement of Australia, 70–71 hunting of, 34, 176, 298, 327, 329, from Africa to Australasia 339 linguistic theories of, 29–30 in rock art, 241, 249, 251 mikiri (Simpson Desert) wells, 182 Madigan Gulf (MG4)site,145 Miller, G, 41, 103, 104 Magee, J, 54–55 millstones, 190f, 198, 201, 282–287 Malakunanja II site, 71 Millys Cave site, 141–143 Malea site, 140 mineral skins, 225–227, 236, 238 Mandu Mandu site, 86, 98, 102, 149, 150 oxalate crusts, 226–227, 238 400 Index mines, red-ochre, 276–282.Seealsoquarries Namib Desert, 105 mining, commercial, 2f, 15, 140, 279 Narcoonowie grindstone quarry, 272, 285–286, Miocene epoch, 49 287 missions, 43, 109 narcotics.Seepituri mitochondrial DNA lineages, 92, 106, 153 nardoo (ngardu) grain, 330–331 of dingoes, 206 native title, 14–15 mobility of populations, 11–12, 79, 107, 196 Native Tribes of Central Australia, The, 4, 17, 19f models of desert prehistory, 72–74 Nauwalabila 1 site, 71 ‘clinal model’ of re-occupation, 115–116 Nawarla Gabarnmang site, 71 ‘coastal colonisation hypothesis’, 34 Neanderthals, 23 cultural succession model, 29 New Guinea, 3, 70, 71 ‘desert culture’ model, 157–158, 163 Newman site, 87, 88 ‘desert transformation’ model, 47–48 ngardu (nardoo) grain, 330–331 Islands in the Interior, 39 Ngillipidji quarry, 295 ‘oasis theory’ of colonisation, 34 ngirntaka (ngintaka) mythology, 157, 210 molecular taxonomy, 116, 117 Nicholson, AF, 177 monsoons, 8, 52, 61, 120, 122, 150, 308 Noala Cave site, 99 changes influencing aridity, 37, 41 North African colonisation, 104 historical strength of, 69 North American desert culture, 163 Holocene, 160 North American refugia, 117 influence of, 5f–9f northern desert fringe, 78, 80 and onset of ENSO, 162–163 Northern Tribes of Central Australia, The, 4 montane refugia, 37, 117 Northwest Cape, 86, 101, 149–150 Montebello Islands, 87, 99, 159 northwestern Australia, 6, 15, 21, 60, 121 mortars, 199, 332 Nullarbor Plain, 5f, 7, 88–89, 150–152, 161, Morton, SR, 117 328 mortuary rites, 106, 107 caves, 63 Morwood, MJ, 259 coast and hinterland, 175–177 mosaics, successional vegetation, 41 dingoes, 205 mourning rituals, 297f–299f Mousterians, 4, 22 ‘oasis theory’ of colonisation, 34 Mt Isa sites oasis-type environments, 34, 112f, 148 axe quarries, 272, 289f, 289–292 occupation of sites, patterns of, 14, 312–318, graphic traditions, 258–259 320 Mt William axe quarry, 288 ocean crossings, 3 Mudburra people, 270 ochre mines, 276–282.Seealsored ochre Mudukian phase, 29 oil and gas exploration, 2f, 15 mullers, 198, 201 ontology, Aboriginal, 220–221 Mulligan River, 266, 267, 292, 293, 294 open-site excavations, 13 multilingualism, 205 organic artefacts, lack of, 13 Mulvaney, DJ, 34, 47, 193, 243, 246, 266, 271 Overland Telegraph Line, 17, 20, 302, 331 Mungo 1 (WLH1) cremation-burial, 92–93 oxalate crusts, 226–227, 238.Seealsomineral Mungo 3 (WLH3) burial, 83f, 93f, 93–94, 101, skins 107–108 O’Connell, JF, 74 Mungo B site, 99 Mungo, Lake.SeeLake Mungo painted art, 255 Munn, ND, 218 palaeohydrology, 54, 58 Murray Valley, 153, 184 palaeolakes, 19, 32, 36, 52, 68, 76.Seealso Murray-Darling region, 104, 183, 200 mega-lakes glacial refugia, 116 palaeomagnetic dating, 48, 49 Murundian phase, 29 palaeovegetation, 41, 50 Myers, FR, 109, 304, 340 Palmer River, 91 mythologies, Aboriginal, 216–218, 260, 264 palynology, 103 Index 401

Pama-Nyungan languages, 30, 202–205, 263, plant fossils, 49, 50 334 plant processing, 143 origins of, 199f, 203–204 playas, 52 Panara culture, 198 Pleistocene coastline, 73f, 86, 88, 116, 149 Panaramitee-style rock art, 202, 213, 222, 229f, Pleistocene economy, 34 228–241, 254–255 Pleistocene fossil footprints, 153–154 chronology of, 234–238 Pleistocene as ice age, ideas about, 25, 38 defined, 228–234 Pleistocene landmass, 76, 120 engravings, 213f, 252, 264, 340 Pleistocene pluvial period, 19, 22, 27, 31 form and composition, 233–234 Pleistocene sites, 123–133 history of the tradition, 238–239 baler shell from, 274–275 social context of, 239–241 Poaceae, 162, 163 Papunya Tula art movement, 255 point-to-point subsistence patterns, 112f, 123, patch burning, 324–326.Seealsoanthropogenic 136, 143, 156 fires; fire point-to-point trade routes, 272 patination, 227 political aspects of research, 14–15 patri-clans, 217 political economy, 214–218, 219, 264–307 paua kudu (seed-storage pits), 332 pollen, 37, 60, 61, 103, 116, 120, 151 pearl shells, 275f, 274–276 population densities, 10, 42, 114, 296–297, 304, pelagic fishing, 77 307–308 Peterson, N, 10 population growth, 208–209, 310–312, 336, 339 petroglyphs, 23–24, 232t, 235, 240, 241–244, post-glacial environmental changes, 24 260, 286 post-Pleistocene period, 31 petrological examinations of stone axes, potable water, 69, 76, 177.Seealsowater 271 prehistory of desert rock art, 219, 261 phylogeographic studies, 117 projectile points on tools, 185 physical anthropology, 28 proliferation events, 186 phytoliths, 117, 136, 162 Pukardu ochre quarry, 266–267, 276–278, 284 Pilbara, 5f, 15–16, 87–88, 137f, 140–143, Puli Tjulkura quarry, 135 321–322 Puntutjarpa site, 30, 159f, 179–181, 201, 234, chronology of rock art, 243–245 321, 327 defined, 6 Gould’s excavation of, 157–158 glacial refugia in, 116 Puritjarra site, 90f, 91f, 89–91, 177f, 201, 230f, Holocene occupation of, 178–179 317 inland rock-art sites, 246 artefacts, 97f, 135f, 201–202 late Holocene rock art, 260–261 excavation history of, 70 natural environment of, 49 and Karrku ochres, 280–281 rock art, 241–254 natural environment of, 60, 68, 117, 162 shell middens, 172–174 occupation phases at, 135–136 Skew Valley midden, rock art, 245–246 rock art, 234, 237f, 238, 255f, 256 social context of art, 251–254 site use at, 107, 134–140 Pintupi dialect, 333–334 stone assemblages from, 184–185, 191–194 ‘Pintupi Nine’, 109–110 Pintupi people, 136, 154, 334 quarries, 267f, 268f, 271–273, 286–287.Seealso pirri points, 190f, 191, 195 millstones; red ochre pirrian phase, 29 regional vs ‘long transfer’, 282–283 pitfall faunal assemblages, 63 Lake Eyre basin, 283–286 Pitjantjatjara dialect, 333–334 grindstone quarries, 279f, 282–287 Pitjantjatjara people, 334 Anna Creek, 284 pituri (narcotic), 266, 293–294 Innamincka, 284–285 active ingredient in, 293 Kurutiti, 286–287 preparation of, 293 Narcoonowie, 279f, 285–286, 287 source of, 292–293 Tooth’s Nob, 283–284 402 Index quarries (cont.) regional surveys, 15 stone-axe quarries Reinach, S, 214 Lake Moondarra 1, 289–290 Rainbow Serpent (Kardimarkara), 69 Mt Isa, 289f, 289–292 relict flora, 38 Mt William, 288 religion and rituals, 42, 213f, 214–218, ochre quarries, 276–282 220–221, 265, 340–341 Anvil Creek, 281 reptiles, desert, 9 Karrku, 280–281 research questions Pukardu, 276–278 chronology of rock art, 261–262 Ulpunyali, 281 graphic systems, 261–263 Wilgie Mia, 275f, 278f, 278–280 inscribed landscapes, 261–263 stone-blade quarries pituri trade, 294 Ngillipidji, 295 political economy, 219, 264 Quaternary dating methods, 36, 42 population growth, 310 Quaternary sub-era, 52 residence patterns, 12 resins, as adhesives, 188–191, 287, 294 Radcliffe-Brown, AR, 304 Reuther, JG, 43 radiocarbon chronologies, 14 Rindos, D, 74 radiocarbon dating, 29, 32, 132, 159f, 161f, 243, ritual property, 305 280, 303f–311f, 322–323 ritual-political system, 217 of continent dunefield chronologies, 37 rituals.Seereligion and rituals first use of in arid Australia, 19 river catchments, 2f, 7, 57–58, 77 of quarries, 287 riverine resources, 69, 76, 114 of rock art, 43, 225, 226, 227, 259 Riwi site, 78, 102 rainfall patterns, 2f, 6, 8, 9f, 121, 155, 309 shell beads from, 98 affecting colonisation, 77 rock art, 25f, 29, 98, 181, 182, 212–213, and glacial aridity, 31–32, 60, 122 318–320 Holocene monsoons, 160 analytical approaches to, 212–213, 221–225, impact of uncertain, 12 253 rain-making ceremonies, 274 anthropomorphs, 233 reaping knives, 331 archaeological approaches to, 221–225 reburial of artefacts, 13 ‘archaic faces’, 252–254 Red Centre.SeeCentral Australia from Burrup-Pilbara region, 214 red ochre, 93f, 98, 107, 135, 140, 228, 271 from Calvert Range, 246 distribution of, 272 from Central Australian margins, 258–259 from Pukardu, 277–278 dating rock art, 225–228 geochemical sourcing of, 272 early to mid-Holocene traditions, 227–254 mines, 278f, 276–282 as evidence of social networks, 115 preparation of, 276–277 figurative art, 214 used in burials, 93 graphic system of, 218–220 refuge areas, 11 graphic traditions, 213f, 214, 219–220, ‘refuge theory’, 116 261–263, 305, 340 refugia, 7, 37–39, 112f, 110–119 hand stencils, 220, 256f, 256 ‘classical’, 116 iconic motifs, 251, 255 cryptic, 116, 117, 156 at Koonalda Cave site, 150 desert, 37–39, 117 at Kulpi Mara site, 133–134 European, 116 marine subjects, 251–252 glacial, 40, 112f, 114–119 mid-Holocene, 171 montane, 37, 117 paintings, 255–256 North American, 117 from Pilbara, 241–254 relict flora, 38 prehistory of, 219, 261 and species diversity, 116–117 at Puritjarra site, 136 uplands as, 117–119 rates of production of, 222 Index 403

red ochre, 98 seed-grinding implements, 31, 35, 164, 190f, regions, 213–214 198–200, 201–202, 332.Seealso research questions, 219, 261 grindstones; millstones secular vs sacred, 219–220 difficulties with identifying, 200–201 significance of, 224–225 mullers, 198, 201 and stylistic behaviour, 224 quarries for, 282–287 of the last millennium, 254–259 seed-storage pits (paua kudu), 332 totemic geography and political economy, serological data, 334 214–221 Serpents Glen site, 91–92, 141f, 145–146, 321, ‘track and circle’, 229–231 328 from Western Desert, 259–261 settlement.Seecolonisation rock varnishes.Seemineral skins sexual dimorphism in Pleistocene humans, 153 rockshelters, 13, 14, 86, 170 shell artefacts, 149 Allens Cave site, 88 beads, 98 Cape Range sites, 86 pendants, 275f, 274–275 evidence of occupation at, 130–131, shell middens, 32, 82, 150, 153, 164, 172–175 312 mangrove forest, 165 Kulpi Mara site, 133–134 Silver Dollar site, 87, 165–170 Pilbara, 142, 178–179 Willandra, 99 Puntutjarpa site, 157, 179–181 Shigesada, N, 74 Puritjarra site, 89–91, 134–140 silcrete, 50, 182, 272 Serpents Glen site, 146 Silver Dollar site, 87, 99, 150, 165–170 rocky-shore ecosystems, 149 Simpson Desert, 37, 60, 139, 182 Rosenfeld, A, 229 Singh, G, 37 Ross, A, 184 sinkholes, 88, 150, 151f Ross, J, 231, 318 site abandonment vs regional abandonment, Roth, WR, 43 130 Runton Range, 259 site assemblages, 224 site data, interpretation of, 14 sacred boards, 217 site-formation processes, 46–47 Sahara Desert, 105 site histories, 130–133, 314f, 312–315, 318 Sahul, 71, 94, 99 site inventories, 318 saline environments, 32 site locations and settlement patterns, 320–321 saltlakes, 52–58, 122 skeletal remains, 92–94, 153, 183–184 salvage and survey contracts, 15 showing osteoarthritis, 94 sand drawings, as storytelling, 218 showing tooth avulsion and wear, 94 sandhill country sites, 124 Skew Valley site, 171f, 173f, 184–185, 245–246 Sandover River, 121 ‘skins’.Seesubsection systems sandplains, 7 sky gods, 216, 217 sandy deserts, 144, 145–146.Seealsodeserts Slack, MJ, 142 sea cliffs, as barrier to marine resources, SLEADS program (ANU), 36–37 176 small-tool phase, 177f, 185, 191–194, 196 sea levels, 119, 120, 159–160, 244–245 Smith, MA, 112, 164, 286, 318 sea surface temperatures, 121 Snowy Mountains, 32 sea voyages, 105 social anthropology, 222, 304–305 seasonal drought, 32 social context of Panaramitee rock art, 239–243 secular vs sacred rock art, 219–220, 257 social development and history of Aboriginal sedimentary histories, 130–132 societies, 44, 338–341 seed harvesting and preparation, 197–201, 209, social exchange networks, 273 329–330, 331–332 social interactions, 336 seed-food species, 198, 325–326 societies, hunter-gatherer, 305–308.Seealso seed-grinding economies, 202, 210–211, Aboriginal societies, hunter-gatherers 287 Sollas, WJ, 4 404 Index songlines, 216 knives and blades, 289f, 294–296, 331 southeastern Australia, 13–14, 43, 61, 68, 78, pirri points, 190f, 191, 195 121 pirrian implements, 29, 191 spear armatures, 185 problems with applying typology to artefacts, spear-throwers, 186, 194, 195 201 spearblades, 195 scrapers, 135 species distribution and ‘refuge theory’, 116 stone tool systematics, 13, 29 species diversity, refugia, 116–117 tula adzes, 187f, 186–190, 192, 193 Spencer, B, 4, 17, 23, 44, 303 tula adze slugs, 186–191 spinifex, 188–191, 195, 330–331.Seealso type artefacts, 29 hummock grasslands ‘type fossils’, 29 Sprachbund, 30 white chert, 135 ‘spread zones’, 209–210 stony deserts, 6, 50 Stafford Smith, M, 8 storytelling, sand drawings as, 218 Stanton, J, 275 stratigraphy, 22, 124f–131f, 130–133, 148, 317f starch, 201 of deposits, 13 state-and-transition models, 257f–263f, 263, 264 of desert sites, 225 Stephenson, B, 201 of palaeo-shorelines, 19 step-wise settlement, 72 of Wilgie Mia mine, 278f stone artefact (lithic) assemblages, 93f, 94–99, of Kulpi Mara site, 134f–135f 140f, 141f, 184–197, 289f, 328 Strehlow, TGH, 216 arid zone, studies of, 322–323 Strzelecki Desert, 21f, 37, 43, 286, 322 classification of, 29 Sturt, C (explorer), 20, 24–25, 329 Cuddie Springs, 83 Sturts Meadows site, 234–235 discard rates of, 185 stylistic behaviours and rock art, 224 discard zones, 14 sub-aerial weathering, 227 flaked stone, 94–97, 135f, 184–197 subincision, 23 Puritjarra, 90, 182 subsection systems, 336 Holocene, 176, 209 subsistence patterns, 99–102, 336 Lake Eyre basin, 144 summer monsoons.Seemonsoons Millys Cave, 141–142 ‘super-totemic’ ancestral beings, 216 paucity of early, 107 superimposition in rock art, 227 Pilbara, 321–322 supernatural beings, 217 rates of use, 136–140 site inventories, 318 taphonomy, 13, 255 trade in, 287–292 Tar tangan phase, 29 stone axes, 195, 258, 285f–288f, 292 Tasman Sea, 60, 121 chronology of, 288f–289f Tasmania, 3, 71, 77, 121, 229 distribution of, 272, 290–291 Tate, R, 18, 37 hafted, 195, 285f–288f, 292 Taylor, L, 218 petrological examinations of, 271 Taylor, TG, 26 stone tools, 13, 14, 28–29, 132, 135f, 136–148, technology, Aboriginal, 11, 196, 209, 339 294–296 tektites, 176 backed artefacts, 186f, 185–186, 187f temperature chalcedony tools, 188, 193 changing in LGM, 122 chert tools, 142, 188, 193 extremes of, 8 circumcision knives, 23 temporary water supplies, 12 composite tools, 185, 194–195 Terebralia middens, 172 convex scrapers, 97, 135 territoriality and land use, 10, 12, 136 endscrapers, 193 Tertiary sub-era, 48 ‘horsehoof’ cores, 96 The Dreaming.Seedreaming ironstone flakes, 141 The Pilbara.SeePilbara geometric microliths, 29, 31, 157, 185–186 Therreyererte site, 258–259, 321 Index 405

Thirrari people, 69 Veth, P, 39–40, 106, 115, 116, 222, 305 Thomas Reservoir site, 252 Walga Rock site, 178 Thomson, DF, 154, 270, 300 Walinya site, 260 Thorley, PB, 48, 114 Wallis, LA, 48 Timor Sea, 3 Wanga East (W04)site,181–182, 230f, Tindale, NB, 28–29, 40, 41 236–238, 317 Tingarri traditions, 253–254, 260 Wangamirri, JR, 266 Tjapaltjarri, Piyiti, 109 Wangkangurru people, 182, 277, 293 Tjapananga, Pinta Pinta, 109 Warbur ton River, 37, 61 Tjungkupu site, 302, 314, 315f Warburton sites, 157 TNL sites, 101 Warlpiri people, 217, 280 Tonkinson, R, 12, 44 Warrumungu people, 20 tool kits, 196–197, 209, 210.Seealsostone Wasson, RJ, 37, 59 artefact (lithic) assemblages water, 10, 12, 69, 112–113, 115, 160 wooden, 209, 288 water tables, 122 tooth avulsion, 94, 183 watercraft, 77 tooth-enamel analysis, 65 waterhole death assemblages, 83 Tooth’s Nob grindstone quarry, 284 Wati Kutjarra tradition, 260 topographical barriers, 77 Wati language group, 30, 319f, 332–336 totemism, Australian, 23, 216, 241, 258 Webb, S, 62, 74, 154 ancestors, 214–216, 220 wells, 84, 113, 182 designs, 218f, 219–220, 335f west coast, 86–87, 165–172 geography, 214–221, 222, 241, 254, 265–341 Western Desert, 6, 30, 144, 259–261, 321–322, sites, 320–321 324 ‘track-and-circle’ engravings, 214, 229–231, Aboriginal societies from, 10, 332–336 261.Seealsorock art cultural stability in, 163–164 trade and exchange, 11, 99, 266–270, 274, rock art of the margins of, 259–261 296–301 scarcity of stone for tools, 282 archaeological approaches to, 271–274 Serpents Glen site, 91–92 grindstone quarries and millstones, Wati language group, 332–336 282–287 wetlands, 121, 143, 178, 322 leilira blades, 294–296 Whitehouse, FW, 26 pituri (narcotic), 292–294 Wiessner, P, 224 red-ochre mines, 275–282 wildﬁres, role in desert ecosystems, 8 shells, 274–276 Wilgie Mia ochre mine, 276, 278f, 278–282 stone axes, 287–292 Willandra Lakes, 31–34, 48, 52, 55, 56, 81f, stone knives, 294–296 80–82, 152–153 systematics, 270–271 geomorphology, 33f transhumance, 11 Williams, AN, 311 translocation of species, 104 Williams, E, 322 tropical cyclones, 8 Winbarku camp, 109 tula adzes, 187f, 186–190, 192, 193 wind erosion, 26 tula adze slugs, 186–191 wind systems, 27 type artefacts, 29 winnowing trays, 194, 195, 197 ‘type fossils’, 29 WLH1 (Mungo 1), 92–93 WLH135, 92 Ulpunyali ochre quarry, 136 WLH3 (Mungo 3)site,93f, 93–94, 101, Uluru site, 260 107–108 unifacial pirri points, 29, 190, 191 Wood, JG, 26, 34, 38, 116 uplands as refugia, 117–119.Seealsorefugia wooden artefacts, 186–188 dishes, 194, 195 vegetation, 37, 59, 60–61, 162 implements, 136, 194–195, 196–197, 279 changes, 103, 120, 121, 162–163 tool kits, 209, 288 406 Index woodlands, 50, 60, 120, 162 Yengoyan, AA, 336 Yirra site, 143 Yandruwantha people, 277, 286, Younger Dryas stadial, 158 293 Yankunytjatjara people, 334 Zietz, ACH, 45