The Ancient Mediterranean Environment Between Science and History Columbia Studies in the Classical Tradition

The Ancient Mediterranean Environment between Science and History Columbia Studies in the Classical Tradition

Editorial Board William V. Harris (editor) Alan Cameron, Suzanne Said, Kathy H. Eden, Gareth D. Williams, Holger A. Klein

VOLUME 39

The titles published in this series are listed at brill.com/csct The Ancient Mediterranean Environment between Science and History

Edited by W.V. Harris

LEIDEN • BOSTON 2013 Cover illustration: Fresco from the Casa del Bracciale d’Oro, Insula Occidentalis 42, Pompeii. Photograph © Stefano Bolognini. Courtesy of the Soprintendenza Archeologica di Pompei.

Library of Congress Cataloging-in-Publication Data

The ancient Mediterranean environment between science and history / edited by W.V. Harris. pages cm. – (Columbia studies in the classical tradition, ISSN 0166-1302 ; volume 39) Includes bibliographical references and index. ISBN 978-90-04-25343-8 (hardback : alk. paper) – ISBN 978-90-04-25405-3 (e-book) 1. Human ecology–Mediterranean Region–History. 2. Mediterranean Region–Environmental conditions–History. 3. Nature–Effect of human beings on–Mediterranean Region–History. I. Harris, William V. (William Vernon) author, editor of compilation.

GF541.A64 2013 550.937–dc23 2013021551

This publication has been typeset in the multilingual “Brill” typeface. With over 5,100 characters covering Latin, IPA, Greek, and Cyrillic, this typeface is especially suitable for use in the humanities. For more information, please see www.brill.com/brill-typeface.

ISSN 0166-1302 ISBN 978-90-04-25343-8 (hardback) ISBN 978-90-04-25405-3 (e-book)

Copyright 2013 by The Trustees of Columbia University in the City of New York. Koninklijke Brill NV incorporates the imprints Brill, Global Oriental, Hotei Publishing, IDC Publishers and Martinus Nijhoff Publishers.

All rights reserved. No part of this publication may be reproduced, translated, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission from the publisher.

Authorization to photocopy items for internal or personal use is granted by Koninklijke Brill NV provided that the appropriate fees are paid directly to The Copyright Clearance Center, 222 Rosewood Drive, Suite 910, Danvers, MA 01923, USA. Fees are subject to change.

This book is printed on acid-free paper. CONTENTS

List of Tables and Figures ...... vii Notes on Contributors ...... xiii Abbreviations...... xvii Preface ...... xix

What Kind of Environmental History for Antiquity? ...... 1 W.V. Harris

PART ONE FRAMEWORKS

Energy Consumption in the Roman World ...... 13 Paolo Malanima Fuelling Ancient Mediterranean Cities: A Framework for Charcoal Research ...... 37 Robyn Veal

PART TWO CLIMATE

What Climate Science, Ausonius, Nile Floods, Rye, and Thatch Tell Us about the Environmental History of the Roman Empire ...... 61 Michael McCormick Megadroughts, ENSO, and the Invasion of Late-Roman Europe by the Huns and Avars ...... 89 Edward R. Cook The Roman World and Climate: Context, Relevance of Climate Change, and Some Issues ...... 103 Sturt Manning vi contents

PART THREE WOODLANDS

Defining and Detecting Mediterranean Deforestation, 800bce to 700ce...... 173 W.V. Harris

PART FOUR AREA REPORTS

Problems of Relating Environmental History and Human Settlement in the Classical and Late Classical Periods: The Example of Southern Jordan ...... 197 Paula Kouki Human-Environment Interactions in the Southern Tyrrhenian Coastal Area: Hypotheses from Neapolis and Elea-Velia ...... 213 Elda Russo Ermolli, Paola Romano, and Maria Rosaria Ruello Large-Scale Water Management Projects in Roman Central-Southern Italy ...... 233 Duncan Keenan-Jones

PART FIVE FINALE

The Mediterranean Environment in Ancient History: Perspectives and Prospects ...... 259 Andrew Wilson

Bibliography ...... 277 Index ...... 327 LIST OF TABLES AND FIGURES

Malanima

Tables 1 Energy consumption in the early Roman Empire ...... 17 2 Energy consumption in advanced regions of the West and East according to I. Morris. 8000bc–2000ad ...... 35

Figures 1 Dated remains of coal in England 1–500ad ...... 23 2 Oxygen isotopes in the ice carrot GISP2 (Greenland glacier ice core) 60bc–350ad...... 25 3 Intensity of precipitation between 400bc and ad400 (and range of error) ...... 26 4 Estimates of forest clearance in Central Europe (Germany, North-Eastern France) from archaeological wood remains 200bc–400ad ...... 28 5 Food consumption by modern populations according to age ...... 31

Veal

Figures 1 Factors affecting the wood supply, which underpin the types of archaeological charcoals found...... 41 2 Examples of charcoals from excavation ...... 46 3 Modern charcoal stack ready for covering with mud, leaves and charring residues, which was then set to char by insertion of a burning log ...... 48 4 Summary results of diachronic study of wood fuel of Pompeii c. third c. bc to ad79 ...... 56 viii list of tables and figures

McCormick

Tables 1 Nile Floods: overview of broader qualities as classified by Bonneau 1971 ...... 77 2 Detailed categories of flood qualities of the Nile according to Bonneau 1971 ...... 78 3 Recording quality as assessed by Bonneau ...... 80

Figures 1 Reconstructed precipitation anomalies (mm/day), April, May, June, 367–378ad, northeast France...... 69 2 Percentages of Nile flood qualities, early vs. later Roman Empire . . 78

Cook

Figures 1 The Dulan-Wulan annual tree-ring chronology from north-central China and the occurrence of severe droughts during the times of the Hun-Avar migrations into late-Roman Europe...... 90 2 Correlations between December–February Niño 3.4 sea surface temperatures (a measure of ENSO variability) and March–June total precipitation from 1951 to 2003 ...... 94 3 Two 2,000 year long annual tree-ring chronologies from ENSO sensitive regions in the Northern and Southern Hemispheres: Douglas fir from northwest New Mexico and Kauri from the North Island of New Zealand ...... 95 4 Correlations between the Douglas fir and Kauri tree-ring chronologies and Hadley Centre global sea surface temperatures (HadISST1) for the winter season: 1871–2003...... 96 5 The average (A) and difference (B) of the Douglas fir and Kauri tree-ring chronologies ...... 97 6 Correlations between December–February average SSTs and the average and difference of the Douglass fir and Kauri annual tree-ring chronologies, each with an identified ENSO signal ...... 98 7 Comparisons of correlation patterns of March–June precipitation with actual and tree-ring ENSO indices ...... 99 list of tables and figures ix

8 The Douglas fir—Kauri average interhemispheric ENSO index with drought inducing La Niña periods indicated around the times of the Hun-Avar migrations into late-Roman Europe ...... 100

Manning

Figures 1a Comparisons of general northern hemisphere temperature covering the past millennium and the often differing (almost opposite) precipitation records from the west and east Mediterranean ...... 110 1b A comparison of periods noted in the analyses of Nicault et al. (2008) of decadal or longer intervals of wetter (more negative PDSI) and drier (more positive PDSI) for Italy, Greece and the Levant versus reconstructed winter NAO indices...... 111 2 A. The standard radiocarbon calibration curve for the period 3000bc to ad1950 from known-age trees. B. The Δ14C record per mille (‰) from A—this is the relative 14C content decay corrected and normalized. C. The residual Δ14C record per mille (‰) after a 1000-year moving average is removed ...... 123 3 Bottom: observed sun-spot numbers (SSN) per year. Top: The annual Δ14C record per mille (‰) ad1600–1900 and an 11-year moving average of this record. Middle: The Δ14C record per mille (‰) from IntCal09 and IntCal04 and two models of 14C production ...... 124 4 Top: A. The residual annual Δ14C with 2 point (pt) FFT smoothing from the data shown in Figure 3 from Stuiver et al. (1998) calculated minus a 22pt FFT smoothing to emphasise the change around the longer-term trend. B. The residual annual production of 14C (iterative method—see Figure 3) with 2pt smoothing ...... 125 5 High resolution 10Be data from Greenland for the most recent six centuries ...... 127 6a Comparison of the Total Solar Irradiance (TSI) reconstructions of Vieira et al. (2011) from the 14C record versus Steinhilber et al. (2009) from the 10Be record ...... 128 6b Top: Total Solar Irradiance (TSI) reconstruction from ice-core 10Be records from Steinhilber et al. (2009)—see Figure 6a. Bottom: 14C production from Marmod09 (Reimer et al. 2009) ...... 129 x list of tables and figures

7 Top: detail of the Total Solar Irradiance (dTSI) reconstruction from ice-core 10Be records from Steinhilber et al. (2009) for the period 300bc to ad800. Bottom: two 14C production models: (a) from Marmod09 (Reimer et al. 2009) and (b) the iterative model from Usoskin and Kromer (2005) ...... 130 8 Top: the two 14C production models in Figure 7—now not inverted—for the period 300bc to ad800. Bottom: the SSN reconstruction from Solanki et al. (2004) for 300bc to ad800 ...... 131 9 The main trends of the solar proxy records in Figures 7 and 8 for the period 300bc to ad700 ...... 134 10 Top: Extra-tropical Northern Hemisphere temperature record. A. 50-year smoothed curve; B. A 10pt FFT smoothed curve. Middle: IntCal09 radiocarbon calibration curve. Bottom: Ring widths of ANG-7B ...... 138 11 Top: the mean tree-ring widths record for the tree-ring series from Istanbul. Bottom: the reconstructed temperature and precipitation records from central European oak time-series shown in Büntgen et al. (2011, Figure 4) are shown for the time interval covered by the Istanbul tree-ring data at the top of the figure...... 140 12 Reconstructed precipitation (April–May–June = AMJ) in mm with respect to the instrumental ad1901–2000 period record from the Büntgen et al. (2011) study for central Europe and for Germany ...... 141 13 Reconstructed summer (June–July–August = JJA) temperature anomalies with respect to the instrumental ad1901–2000 period record from the Büntgen et al. (2011) study for central Europe and for Switzerland ...... 142 14 Comparison of the temperature reconstruction for the extra-tropical Northern Hemisphere for the last 1000 years by Christiansen and Lungqvist (2011) with the temperature anomalies reconstruction over the same period by Büntgen et al. (2011; 2006) ...... 144 15 A comparison of the Büntgen et al. (2011) precipitation record from Central and Northern Europe versus two east Mediterranean precipitation reconstructions and the PDSI reconstruction by Esper et al. (2007) from Morocco ...... 145 list of tables and figures xi

16 Three δ18O plots from Mediterranean locations. A: Soreq Cave from Bar-Mathews et al. (2003); B: Bucca della Renella from Drysdale et al. (2006); and C. the record from planktonic foraminifera (Globigerinoides ruber) from the southeast Mediterranean off Israel ...... 148 17 Four δ13C records from Mediterranean region speleothems ...... 150 18aComparison of two temperature proxy records derived from speleothems (from northern Spain and Austria), with the proxy temperature records for central Europe and for Switzerland from tree rings produced by Büntgen et al. Top: the Marmod09 14C production model ...... 152 18bDetail from Figure 18a of the summer temperature reconstructions from (top) the central Alps speleothem from Spannagel Cave (Mangini et al. 2005) with (bottom) the central European temperature reconstruction based on tree-rings in Büntgen et al. (2011) ...... 153 19aTwo records of past major volcanism relevant to the Northern Hemisphere from two Greenland ice-cores ...... 155 19bBottom: number of ring-width growth minima in the Bristlecone pine (BCP) record of Salzer and Hughes (2007) associated within 5 years of a volcanic signal in an ice-core record. Top: Packages of years where there are notable decreases in Bristlecone pine growth and ice-core volcanic eruption signals ...... 156 20 Two temperature reconstructions for high northern latitudes of the northern hemisphere ...... 159 21 A comparison of a selection of records or events discussed in the text for the period 300bc to ad800 ...... 167

Kouki

Tables 1 The relevant archaeological periods in Jordan ...... 200 2 A reconstruction of the climate history of southern Levant and the rural settlement in the Petra region ...... 206

Figures 1 The Petra region, southern Jordan ...... 199 2 A modern barrage in the Jabal Harun area in September 2011 ...... 201 xii list of tables and figures

3 Environmental zones in the Petra region ...... 205 4 The distribution of rural settlement in different environmental zones in the 1st century ad ...... 207 5 The distribution of rural settlement in different environmental zones in the 3rd century ad ...... 208 6 The distribution of rural settlement in different environmental zones in the 6th century ad ...... 209 7 The distribution of rural settlement in different environmental zones in the 7th century ad ...... 210

Ermolli et al.

Figures 1 Location map of the study area...... 214 2 Pollen diagram from the C106 core in the Salerno Gulf ...... 217 3 Three-step scheme of the evolution of the Neapolis area...... 221 4 Pollen diagram from the Neapolis port sediments...... 223 5 Three-step scheme of the evolution of the Velia area ...... 227 6 The road to Porta V at Velia ...... 228 7 Section of the alluvial deposits at Velia ...... 230

Keenan-Jones

Figures 1 The Italian Peninsula ...... 236 2 The Bay of Naples in the Roman Period showing the Aqua Augusta...... 241 3 The area of the proposed Tiber flood prevention works of 15ad. . . . 249 NOTES ON CONTRIBUTORS

Edward R. Cook is Ewing Research Professor at Lamont-Doherty Earth Observatory of Columbia University and is the Director of the Tree-Ring Laboratory there, which he co-founded in 1975. His primary research activities have been oriented around the development of statistical methods of tree-ring analysis, with an emphasis on time series modeling and robust estimation of tree-ring chronologies, all with the goal of producing the best reconstructions of past climate possible. He co-edited and contributed to a book entitled Methods of Dendrochronology: Applications in the Environ- mental Sciences (1990). Cook has over 150 peer review publications and was elected Fellow of the American Geophysical Union in 2011. His most recent publication in 2012 is a reconstruction of past summer temperatures from tree rings in East Asia for the past 1,200 years.

Elda Russo Ermolli is a Researcher in Physical Geography and Geomor- phology and Professor of Quaternary Geology and Geoarchaeology at the University of Naples Federico II, and also an Associate Researcher in the Pre- history Deptartment of the Natural History Museum in Paris, specializing in Quaternary palynology. Her main research interests concern the climatic and environmental history studied through pollen analysis of marine, tran- sitional and continental records. The aims of her work are the reconstruction of the vegetation cover variations, and their connection to natural and /or anthropogenic causes. She is the author of more than fifty journal articles as well as of book chapters and many congress reports.

W.V. Harris is Shepherd Professor of History and Director of the Center for the Ancient Mediterranean at Columbia University. His most recent book is Rome’s Imperial Economy (2011), and he was the editor of and a contributor to Mental Disorders in Classical Antiquity (2013). For his other publications see http://www.columbia.edu/history/faculty/Harris.html.

Duncan Keenan-Jones is currently a Mellon Postdoctoral Research Asso- ciate at the Illinois Program for Research in the Humanities and Visiting Assistant Professor in the Department of the Classics, University of Illinois at Urbana-Champaign. He is interested in the use of historical, archaeological, quantitative and scientific approaches in the study of ancient society, xiv notes on contributors technology and environment. He has recently published on the lead contamination in the drinking water supply of Pompeii, and he has in the press a quantitative study of lead-pipe inscriptions in Campania and a geoarchaeological study of mineral deposits in the aqueducts of Rome.

Paula Kouki has recently finished her PhD at the University of Helsinki. She is currently working as a researcher for the Finnish Jabal Harun Project at the University of Helsinki, and is the editor of the forthcoming publication of the FJHP survey results, Petra—The Mountain of Aaron: The Archaeologi- cal Survey. She has fieldwork experience in Jordan, Greece and Finland. Her research interests include Near Eastern Classical and post-Classical archaeology, landscape archaeology, human-environment relationships, the perception of time in archaeology and archaeological field survey.

Paolo Malanima is the Director of the Institute of Studies on Mediter- ranean Societies (ISSM) of the Italian National Council of Research (CNR), based in Naples. He has been Professor of Economic History and Economics at the University of Pisa (1977–1994) and University “Magna Graecia” in Catanzaro (1994–2002). He is a member of the editorial board of the jour- nals Società e Storia, Rivista di Storia Economica, Economic History Review and Scandinavian Economic History Review. He is author of The Pre-Modern European Economy (Brill, 2009), and (with A. Kander and P. Warde) Power to the People. Energy in Europe over the last Five Centuries (Princeton, 2013).

Sturt Manning is the Goldwin Smith Professor of Classical Archaeology and Director of the Malcolm and Carolyn Wiener Laboratory for Aegean and Near Eastern Dendrochronology at Cornell University. His research interests cover Mediterranean archaeology (especially the Aegean, Cyprus and east Mediterranean), and archaeological science (especially dendrochronology, radiocarbon, and climate). For more information on his publications, see http://cornell.academia.edu/SturtWManning

Michael McCormick is Francis Goelet Professor of Medieval History at Harvard University. His books include Origins of the European Economy (Cambridge University Press, 2002) and Charlemagne’s Survey of the Holy Land (Dumbarton Oaks-Harvard University Press, 2011); he is senior editor of the Digital Atlas of Roman and Medieval Civilizations (http://darmc.harvard .edu/). He currently co-directs the excavation of a late Roman settlement in eastern France, and chairs the Science of the Human Past, a new Harvard research and teaching network of natural scientific approaches to the past notes on contributors xv including human health and the environment, and applying computer science to the study of ancient texts.

Paola Romano is Associate Professor at the Earth Science Department of the “Federico II” University, Naples. She teaches geomorphology and geoarchaeology to both undergraduate and graduate geology students. Her main research interests fall in the field of morpho-dynamic processes in continental and coastal environments. In the last decade she has been applying her expertise to the environmental and morphological reconstruction of ancient and historical human settlements in southern Italy.

Maria Rosaria Ruello is a post-doctoral researcher in geomorphology and geoarchaeology at the University of Naples “Federico II”. Her main research interests are the paleoenvironmental and geomorphological evolution of natural contexts and their interaction with archaeological sites.

Robyn Veal earned her PhD at the University of Sydney with a dissertation on the fuel economy of Pompeii (2009), which will appear as Fuelling Pom- peii (Accordia, London). Her interests include natural resource economics in the ancient world; archaeological theory and whole assemblage analysis; and GIS and information management systems for archaeology. She has been the Ralegh Radford Rome Fellow at the British School at Rome, and is currently the Anniversary Research Fellow at the McDonald Institute for Archaeological Research at Cambridge.

Andrew Wilson, Professor of the Archaeology of the Roman Empire, Uni- versity of Oxford. Recent and forthcoming publications include: Quantifying the Roman Economy: Methods and Problems (ed. with Alan Bowman, Oxford, 2009), Settlement, Urbanization and Population (ed. with Alan Bowman, Oxford, 2011); The Roman Agricultural Economy: Organization, Investment, and Production (ed. with Alan Bowman, Oxford, 2013); Maritime Archaeol- ogy and Ancient Trade in the Mediterranean (ed. with Damian Robinson, Oxford 2011); and ‘Saharan trade in the Roman period: short-, medium- and long-distance trade networks’, Azania: Archaeological Research in Africa 47, 4 (2012), 409–449.

ABBREVIATIONS

AÉ L’Année Épigraphique AJA American Journal of Archaeology AJPh American Journal of Philology Annales HSS Annales: histoire, sciences sociales AO Arctic Oscillation AP arboreal pollen AS Anatolian Studies BGU Aegyptische Urkunden aus den Königlichen/Staatlichen Museen zu Berlin CIL Corpus Inscriptionum Latinarum C.Ord.Ptol. Corpus des Ordonnances des Ptolémées ed. M.T. Lenger C.Th. Codex Theodosianus DJF December January February ENSO El Niño-Southern Oscillation FJHP Finnish Jabal Harun Project GPCC Global Precipitation Climatology Centre IAWA International Association of Wood Anatomists IGLS Inscriptions Grecques et Latines de la Syrie ILS Inscriptiones Latinae Selectae ed. H. Dessau ITRDB International Tree-Ring Data Bank JFA Journal of Field Archaeology JRA Journal of Roman Archaeology JRS Journal of Roman Studies KNMI Koninklijk Nederlands Meteorologisch Instituut LIA Little Ice Age LibStud Libyan Studies MCA Medieval Climate Anomaly NAO North Atlantic Oscillation NSA Notizie degli scavi di antichità ORom Opuscula Romana OSL Optically Stimulated Luminiscence P. Berl. Leihg. Berliner Leihgabe griechischer Papyri ed. T. Kalén et al. PBSR Papers of the British School at Rome PDSI Palmer Drought Series Index P. Tebt. The Tebtunis Papyri ed. B.P. Grenfell et al. Sammelbuch Sammelbuch griechischer Urkunden aus Aegypten ed. F. Preisigke et al. SEG Supplementum Epigraphicum Graecum SHA Scriptores Historiae Augustae SSN Sun Spot Number TSI total solar irradiance ZPE Zeitschrift für Papyrologie und Epigraphik

PREFACE

Historians and natural scientists are two separate tribes, both of them to a great extent endogamous, intellectually speaking. It is true that ever since Auguste Comte (1798–1857) there have been historians who have wanted history to be more scientific in the sense that it would conform to a supposedly higher epistemology, true also that some of the earliest Annales historians were interested in, for instance, psychology, as well as more obviously in physical geography. And there have always been historians of medicine, and, in a sense, of the environment. It also remains true, however, that few members of one tribe have been much attracted by the other. But now a certain change is taking place. It features some extremists, who can with some inaccuracy be labeled ‘technological determinists’, and no doubt it will leave a great number of historians untouched even in English- speaking and French-speaking lands (not to mention Germany or Italy, where genuine environmental history has made virtually no headway to date among ancient historians). And it is obviously only in certain scientific fields that anyone is interested in the historical past. The change in the attitude of historians that I am referring to has nothing to do, be it noted, with the notion that history should become in some extended sense an experimental science. In so far as the language of experimentation merely serves in such books as Natural Experiments of History (2010), edited by Jared Diamond and James A. Robinson, as window-dressing for comparative history it will do no harm;1 not so, however, if it gives anyone the idea that a new and superior way of doing history is involved. No, what is changing for the better is that the subject-matter of history has widened still further in recent decades to include problems that have also been, and continue to be, the objects of widespread scientific attention. Actual collaboration between scientists and historians is still rather rare—or at least inconspicuous—, not least because scientists with historical interests commonly suppose, with varying degrees of justifica- tion, that they can do without real historians. But whatever the future of historical-scientific collaboration may be, we historians now have on our

1 For a critical review of this book and its significance see Roth 2012/3. xx preface agendas all sorts of problems that require us to leap over the traditional disciplinary boundaries. The reluctance of historians to engage with the natural sciences has roots so much out in the open that there is no need to describe them. But some growing fields of historical interest, the most conspicuous of which is probably environmental history, are making this attitude obsolete. Scholars still write books about the ancient environment that are essentially digests of what Greek and Roman writers said about the environment or about natural features (rivers or mountains), but if we want to know what the environment in antiquity was actually like, and why it developed as it did, we turn to scientific archaeology, to geology, to climate studies, to palynology, to botany, and so on. This book is largely the work of others, but it also springs from my con- viction that ancient historians, and indeed historians more generally, can benefit greatly at the present juncture from making more use of some of the natural sciences and of technological expertise. There are many productive alliances to be formed. The present collection, as I think all contributors to it would agree, is full of historical and scientific questions that are far from being resolved. We shall only get closer to resolving them if we make use of all relevant methodologies and bodies of knowledge. For ancient history, in particular, there is a serious challenge here. The field can look inwards, re-hash well-known textual sources and concoct ‘new’ biographies of famous Romans. Or it can put its energy into exploring fresher subjects, with fresher methods. Of course the choice is not a stark as that, and no one deplores more than I do the fact—and it is a fact—that there are professional Graeco-Roman historians at large who have a poor knowledge of Greek and Latin (the great anxiety of the tradi- tionalists) and are prepared to publish books and articles without know- ing the bibliographies of the subjects that they write about. No one said that writing good scholarly history was easy. But if we want to deserve intellectual respect we must confront modern times as well as ancient ones.

For the most part this book consists of papers that were given in earlier forms at a conference entitled “History and Environment in the Ancient Mediter- ranean”, held in Rome on June 15th and 16th, 2011. I should like to thank first of all the speakers who are represented here for their commitment and their willingness to engage in a somewhat unconventional project. Special thanks go to Duncan Keenan-Jones, a young scholar whose work was not known to me when I organized the Rome conference but who afterwards kindly preface xxi agreed to contribute a paper about water management, a subject that the conference neglected. Some of the conference took place at the American Academy in Rome, and I should like to thank its officials Carmela Franklin, Christopher Celenza, Corey Brennan and Anne Coulson for making it possible to meet there in extraordinarily pleasant surroundings. I owe an equal debt to the director of the Institutum Romanum Finlandiae, Katariina Mustakallio, for the opportunity to hold one of the conference sessions at that remarkable institution. Several other people gave useful advice or provided practical help with this project, and I most of all thank Roger Bagnall (NYU), Saskia Hin (Rostock), Joe Manning (Yale), and Nate Pilkington (a former Columbia student who is about to take up a post-doctoral fellowship at Cornell) for their assistance. But the person who more than anyone has helped to beat this volume into shape has been my long-suffering collaborator Emily Cook, ABD in the Art History and Archaeology Department at this university, to whom I offer my deep appreciation. Finally, none of this would have been possible without generous funding provided by the Andrew W. Mellon Foundation, to whose officers I once again offer my sincere thanks.

William Harris Columbia University, February 2013

WHAT KIND OF ENVIRONMENTAL HISTORY FOR ANTIQUITY?

W.V. Harris

What sort of fresh potential does environmental history have for our understanding of the Greek and Roman world? (Others must answer the sym- metrical question: what sort of fresh potential do Greek and Roman history have for our understanding of changes in our planet?). Environmen- tal scientists, climatologists, and archaeologists, together with historians economic and otherwise, have recently been converging on a number of major issues in the environmental history of the late Holocene period, that is to say, roughly speaking, historical times. The Mediterranean area has rightly attracted especially intense attention, partly at least because, as far as ancient times are concerned, it uniquely combines extensive written records with an ocean of archaeological information, while also offering the results of numerous palynological and geological studies. This convergence seems to promise that before long we shall be able to construct an environmental history of the region that will be more satisfying to both scientists and historians. One obstacle of course is, to put it bluntly,that no historian knows enough science and no scientist knows enough history. In both camps, however, there are outstanding individuals who have made vigorous efforts to cross the great divide; and there exist ambidextrous archaeologists who keep up with both fields as far as that is humanly possible. Not that one should give way to optimism. The technical complexities of every field grow and grow, and collaboration between humanists and scientists, while it may be on the increase, faces all sorts of barriers. And it is by no means clear that when we are considering the late-Holocene environment we are usually pursuing the same questions. This book is somewhat experimental in combining historical and scientific approaches, and the experiment reveals disparities of aims and methods that will continue to require discussion. We need to clarify our agendas. It is worth recalling here the distinction usefully stressed by Horden and Purcell between history in the Mediterranean and the history of the Mediterranean.1 By the former Horden and Purcell meant the history of

1 Clarified in Horden and Purcell 2005, 357. 2 w.v. harris politics, society, the economy, religion and warfare that takes up most of the space in Mediterranean histories. But the Mediterranean world has had its own physical history, it is not just a ‘setting’.2 By the history of the Mediter- ranean Horden and Purcell meant the history of interactions between man and the environment, hence a history of human use and misuse of natural resources, a history of cultivars and rivers, a history of landscapes, settlement patterns, demography and ‘connectivity’. The Corrupting Sea, which was the first book to carry out a part of this agenda (but demography and climate were reserved for Part II), did not attempt to explain contingent events, indeed it largely ignored them. Some of the contributors to this volume, by contrast—notably Malanima, Cook and McCormick—aim to fit together climatic change and historical events of various kinds.3 Malanima takes a Roman ‘economic decline’ to have resulted from population growth (which meant that inferior land had to be used) and the end of the so-called Roman Warm Period (a decline in temperatures shortened growing seasons).4 This is a coherent story, but the facts on which it is based are fragile, and there were other factors at work. If there was an empire-wide economic decline, it probably set in after the beginning of the Great Pestilence (the Antonine Plague), in other words in a period of (possibly sharp) population shrinkage, not growth. And the pestilence itself was probably a major factor in any economic decline that did occur.5 As for climate change, ‘the scientific study of the ancient climate has barely begun’ (McCormick, below, p. 63),6 which means that this is an inop- portune time for anything like environmental determinism.7 There are now multiple sources of information about the ancient climate, and the strong

2 Not that one should complain when historians use this term. What is a ground for com- plaint is the publication, even within the last decade, of general histories of classical antiquity that pay no attention to the natural world: see Cambridge Ancient History volume XII, for example (2005). When, on the other hand Chakrabarty 2009, 201, announces that ‘anthropogenic explanations of climate change’ are spelling ‘the collapse of the distinction between natural history and human history’, he seems to be catching up with Braudel. 3 Here I discuss approaches and methods and do not attempt to summarize any of the papers included here. 4 The warmth of the Roman Warm Period must have been quite relative, at least in some places: Pliny’s description of his Tifernum villa argues strongly for a climate there colder than the present one (see Letters 5.6.4, with Grove and Rackham 2001, 142). 5 Harris 2012. 6 See also Ermolli et al., this volume p. 213. 7 As is unwittingly demonstrated by the self-described neo-determinists Issar and Zohar 2007. what kind of environmental history for antiquity? 3 impression I receive from this literature (see especially Finné et al. 2011, Manning, this volume) is that the Roman Empire was full of micro-climates, and that no narrow date can be assigned to the end of the Roman Warm Period. If, as some suppose, that occurred much later than 150ad,8 it has to compete with other possible causes of economic deterioration such as repeated civil war and a severe currency crisis in the 270s. Nonetheless Malanima’s paper implicitly calls into question the viability of the ‘history in’/‘history of ’ distinction, or rather it firmly places the economic history of an agrarian society in the ‘history of ’ category. Bresson, similarly, found it impossible to separate off the exchange of commodities in antiquity from the history of.9 Malanima’s paper sets out without unprecedented clarity the likely energy needs of the Roman world, and might lead in any of several interesting directions, such as an attempt to calculate the carrying capacities of a variety of regions. The fuel history of the Roman world certainly counts as ‘history of ’ too. One of the material bases of the whole culture was charcoal, which was the essential means for producing the high temperatures needed for iron- and copper-working. This topic is not of course new,10 but it is now receiving more attention from archaeologists, thanks in good part to the enthusiasm of Robyn Veal. Her paper in this volume raises the question how much charcoal was employed in other processes and practices. If there is to be a clear answer, it will presumably come from further excavation reports and the new methods of analysing charcoal that she outlines. The most intriguing result she reports, from Pompeii, is the predominance of beech-wood (fagus sylvatica) charcoal from the third century bc to the date of the eruption, with other taxa, including fruit and nut woods ‘in small, but then increasing amounts’, becoming commoner in the first centuries bc and ad. Was this change simply a matter of preference, or were the beech woods thinning out?11

8 As late as 400 perhaps: see the bibliography in Manning, this volume, p. 135. Manning’s conclusion leaves a supposed end of the Roman Warm Period c. 150 in limbo, dating the change to some time in the third century. It is not clear whether the third century was drier or wetter in central-Mediterranean lands: see on the one hand McCormick (p. 70) and Manning (p. 163), and on the other Keenan-Jones (p. 234). 9 Bresson 2005, 99–100. See further Horden and Purcell 2005, 357. 10 The book edited by Fiorentino and Magri 2008 gives an idea of the state of this field. 11 Incidentally I note that beeches, which do not now grow below 800 metres in Italy, sometimes grew at notably lower altitudes in the past (Theophrastus, Hist.Plant. 5.8.3, with Grove and Rackham 2001, 142; Ferrari Fontana et al. 2008), which makes it less likely that Pompeii was running out of them. Beeches can be coppiced. 4 w.v. harris

McCormick’s paper complements Malanima’s. Having amusingly settled the question of the exact year in the 370s when Ausonius wrote his poem Mosella,12 he generously presents the results of his Dumbarton Oaks research group’s work on the Mediterranean climate, 100bc to 600ad. This leads primarily to some specific questions about agricultural production, though McCormick’s scholarly caution and rigor sometimes give way to the more general hope that climate change can be correlated with ‘imperial fortunes’. The results proposed so far are regional but intriguing: one is a refinement of data collected forty years ago by Danielle Bonneau about the annual flooding of the Nile, which appear to show that (evidently because of lower precipitation in the river’s headwater region) the Nile flooded less well in the period 155 to 299ad than it had in previous centuries, with the overall effect of lowering agricultural production. Whether this had any impact on the grain supply of the city of Rome is unknown, but it must certainly have made life more difficult for the Egyptian poor. Another possibility canvassed by McCormick is that the cultivation of rye (secalecereale) in central Europe, inside and outside the Roman Empire, was a response to climatic change, either to lower temperatures, as previously proposed, or to lessened precipitation, as proposed by McCormick himself. Like the spread of other cultivars, the spread of rye, a grain known to but looked down on by classical Mediterranean writers, is a matter of controversy; the very concept of domestication is a complex one.13 We seem to lack an up-to-date account of the diffusion of rye in Europe, but K.-E. Behre’s impressively detailed study of 1992 appears to show that it was cultivated in free Germany as early as the second century ad. ‘The decisive reason for rye becoming a crop has to be seen in the change of the harvesting method and not in climatic deterioration’, he claimed.14 But it would certainly be tempting to associate the spread of rye cultivation in central and northern Europe with a colder climate if a close chronological fit could be established. McCormick briefly floats the idea, and Edward Cook develops it in full, that the Hunnic and Avar invasions of Europe in late antiquity were ‘probably encouraged’ (McCormick) by aridification in central Asia. Or that aridification ‘may have contributed to’ the impulse to move westwards (Cook). What Cook contributes to strengthening this old theory is recent dendrochronological information that indicates a ‘megadrought’ in central China

12 Unless, that is to say, Ausonius’ description of the physical conditions of the Moselle valley was—as seems to me quite likely—largely unlocated in time. 13 Cf. Weiss et al. 2006. 14 Behre 1992, 142, q.v. what kind of environmental history for antiquity? 5

(present Qinghai province) about 360ad. With his wide knowledge of climate systems, he is able to delineate the climate change in question with a new level of detail. One obvious difficulty is that the Qinghai sites are a good 4,000 kilometres from the earliest known home of the Huns, to the east of the lower reaches of the Volga. Furthermore, Huns are reasonably well attested further west, to the north and west of the Black Sea, as early as the 370s.15 These two papers raise the wider question whether climatic variation has ever been a detectable major cause of social or political events. It is probably fair to say that most historians are congenitally disinclined to accept such ideas, fond as they are of human agency and specific cases. Some recent work may make this scepticism harder to defend. In particular, it has been argued that ENSO (El Niño/Southern Oscillation) (see Cook, p. 92, for an explanation of this concept) has been one of the causes of civil conflicts in tropical countries since 1950.16 The very modesty of the claim in question (‘ENSO may have had a role in 21% of all civil conflicts since 1950’) is appealing. The paper contributed by Sturt Manning, which amounts to a comprehensive review of the current state of knowledge concerning the ancient Mediterranean climate, points out some of the reasons why such are hypotheses may be problematic:17 on the one hand, obviously, lie the complexities of historical causation,18 on the other the present uncertainties of climate history. When we talk about the historical effects of climate change, it is essential that we consider both alternative explanations and the likely effects on human beings—and these particular human beings—of climate instability, aridification, and changes such as short and long periods of decreased or increased temperatures, and of decreased or increased precipitation. Should we think of ‘the Huns’ leaving central Asia in desperate economic straits and arriving in Europe as, fairly briefly, a highly formidable force? Think of other large-scale migrations, such as the Arab expansion of the seventh century

15 Orosius 7.33.10. But where exactly Huns were to be found in the 370s and 380s continues to be debated. 16 Hsiang et al. 2011. 17 Manning, this volume pp. 114–115. 18 Manning 2010 associates the large 14C production peak centered on 765bc with population growth in Greece and all the complex cultural changes that began in that period. Such a climate would have resulted in longer and more reliable growing seasons. While well aware of the difficulties of establishing any causal relationship (42), he concludes that climate change ‘may have created conditions that actively promoted development and change in human societies’ (44). See Danti in the same volume, esp. 140, for a properly sceptical approach. 6 w.v. harris or the European occupation of the Americas: when we know about their underlying motives in detail, they concern the obvious material advantages of conquering your weaker neighbours, together with religious considerations of various kinds. Which is not to exclude the possibility that traditional historical methods have under-estimated the influence of climatic change. It can be hypothesized that climatic events have much more potential for explaining economic and other consequent difficulties in past societies than for explaining economic, cultural or any other kind of success—or migrations. Thus it has been plausibly argued—though the argument is far from closed—that a basic cause of the collapse of Classic Mayan civilization was a period of recurrent major droughts.19 And it may not, on the other hand, be in any way an anomaly when Paula Kouki discovers that in southern Jor- dan, in the region of Petra, ‘the intensification of settlement and agriculture lagged behind … improved climatic conditions by at least two centuries’.20 Manning mentions several examples of proposed correlations between climate change and large-scale successes or advances, none of which has been adequately demonstrated.21 He seems willing to attribute the so-called Greek Renaissance that began in the eighth century bc to a climatic change favourable to agriculture; the point, he says, is ‘not who is right or wrong— rather how the same case can be and is differently interpreted’. I entirely disagree with this last claim, because I want to save climate scientists from chasing historical mirages: we could possibly say that a climate reasonably favourable to agriculture was a necessary condition for the ‘Greek Renaissance’, but it has zero explanatory power.22 ‘Axial age’ or not, this was a unique succession of events, which can to a considerable extent be explained—although some would say that the great fascination of classical Greek history is that its successes remain to a considerable extent unex- plained. Manning is, however, generally cautious about climatic causes that have been attributed to historical events and processes. As he notes, several scholars have argued that climate change was ‘a factor, even a significant

19 Hammond 2010. 20 Kouki, this volume pp. 205–206. 21 Compare the argument of Sallares 2007, 19, based on Speranza et al. 2002, that ‘a decline in solar activity leading to colder and more humid conditions c. 850bc may well have been the critical factor underlying the simultaneous development of Iron Age cultures around the Mediterranean, which is otherwise difficult to explain’, an assertion, this last, which seems crucially to under-estimate Mediterranean connectivity. 22 A historian is not allowed to say that a climatic event ‘marks the beginning of’ the Greek Renaissance (Manning, this volume, p. 132). what kind of environmental history for antiquity? 7 factor, in the decline and re-organization of the later Roman world’,23 but the nature, scale and chronology of the climate changes themselves are still most unclear. What difference does it make if average temperatures change by one degree or by several? How much less precipitation means difficulties—or disaster? We need more data, obviously, but, more importantly still, we need better thinking about the possible range of differences between local climates within the same region and within the Roman Empire (Mediterranean and otherwise) as a whole; and we also need better thinking about the likely effects of climatic change on human, or rather Roman, behaviour. Many questions about the environmental history of antiquity seem ripe for scientist-humanist collaboration.24 Scientists sometimes frame the subject simply as the effect of mankind and climate on landscapes,25 but it is the vital interactions that demand attention. What would grow where? Grains (not only rye) and olive trees may raise the largest problems here, but there is also much to say about other fruits and vegetables,26 and about the timber- productivity of various areas from Spain to Egypt. What levels of population, urban and rural, could the resources of various areas support? The Mediter- ranean has many desert areas and marshes and mountains—how many of them were productive? And the sea itself has a multifaceted history. One scholar has it the Mediterranean ‘is relatively poor in fish’27—should we think that this was so in antiquity, or should we allow that many coastal populations depended heavily on harvesting the sea?28 Then there is the vast array of the mineral resources, widely studied in the last half-century in particular, but in the case of metals at least still needing Mediterranean-wide studies. What effects, direct and indirect, did humans in turn have on natural resources, in particular on woodlands, soil, fresh water and wild life? Sub-topics here include wood fuel, deforestation and erosion, fallow fields, drainage and irrigation projects, hunting, and the consequences of tax

23 Below p. 158. 24 See also the list of questions in Harris 2005b, 12–20, and the bibliography mentioned there, which I do not attempt to bring up to date in this introduction. 25 E.g. Sadori and Giardini 2008, 229. An obvious difference is that scientists generally want to deal with longer periods, in this case since the mid-Holocene, i.e. the last 4500 years, whereas few Mediterranean (as distinct from Ancient Near Eastern) historians want to look back much beyond 1000bc. 26 See Sallares 2007, 29, for some brief remarks about the diffusion of fruit trees. 27 Sallares 15. 28 Cf. Horden and Purcell 2000, 576–577. See now Boardman 2011. 8 w.v. harris regimes.29 How large a share of ancient economies was pastoral, and how did pastoralism affect landscapes? The list can go on: one of the most contested topics in ancient history at present is osteological: the relationship between surviving human bones and stature, and more generally well- being.30 This issue leads to the wider question of disease patterns. Hence we need, among others, agronomists, food scientists, geologists, botanists, met- allurgists, hydraulic engineers, and physiologists. Arching over all these questions is the perennial problem of general- ization, already mentioned in the context of climate. Horden and Purcell attempted to circumvent this problem by concentrating on four micro- regions. In retrospect, I believe that this was in essence a good idea but that we need two or three times as many micro-regions, at least. Many scholars have devoted themselves to studying Greek regions and Roman provinces over the last forty years, but unfortunately few of them have pursued environmental questions in detail if at all (there have been some splendid exceptions). My contribution to the body of this book is an account of deforestation that attempts to combine historical and scientific information and comes to a mixed conclusion—quite a bit of deforestation, but not a Mediterranean- wide crisis. We then proceed to local studies, that is in a sense to micro-regions. Paula Kouki’s study takes us to Petra. Her highly suggestive conclusion is, in brief, that the patterns of settlement there do not conform to what the climate data might lead one to expect. When precipitation somewhat increased in the second and third centuries ad, rural settlement thinned out; a drying climate in the sixth century was accompanied by the enlargement of settlements. She considers various possible explanations, concluding that the notion of a ‘favourable climate’ is simplistic even in a region that was so heavily influenced by a single climatic variable, rainfall. In particular, much will always depend on the plants and technologies available, and indeed on local traditions of land use. Our last two studies take us to Italy. The approach of Elda Russo Ermolli and her collaborators is geoarchaeological. They found that both at Neapo- lis and Velia it was probably a combination of climatic events and human neglect that led to sedimentation and urban degradation in and after the third century ad. In both places the coastline moved forward more quickly

29 On the latter see especially Purcell 2005. 30 See among others Steckel 2009, Wheeler 2012, Pitts and Griffin 2012. what kind of environmental history for antiquity? 9 in the third century. The old port of Neapolis ceased to function, and flooding rendered part of the site of Velia uninhabitable. Our authors conclude that ‘the effects of particular land use conditions were enhanced by extreme climatic events’, creating ‘episodes of severe erosion’.31 Roman officials intervened from early times to influence land-use (by taking control of all coastal woodlands: Cicero, De Republica 2.58), and perhaps even earlier to exercise some control over water resources. Keenan-Jones’ paper considers their ambitions and achievements in the latter department, concentrating on central and south-central Italy in early imperial times. His first case is the Aqua Augusta in Campania, a longer construction, taken as a whole, than any of the aqueducts of the capital itself. Whoever diverted the water from the springs at the head of the River Sabato (see the map on p. 241) to the towns to the north and west of Naples as far as Misenum—it was probably Augustus who was responsible for most of it—did a great favour for their permanent and temporary residents, including of course members of the imperial household and of the senatorial-equestrian elite. In so doing, the person or persons responsible also did a disfavour to the inhabitants of the basin of the Sabato itself. It would be worth speculating further how grave this damage is likely to have been, and whether the decision is likely to have been an entirely political and social one. For Frederiksen, ‘the feeling behind [Augustus’] massive corrections of nature in Campania is not in the least utilitarian’; rather it was a matter of ‘spectacular ostentation’.32 Since the aqueduct served the interests of productive towns such as Puteoli, however, its net economic effect may well have been positive. And we have been reminded by the Hadrianic ‘Bronze of Agón’ that Roman authorities often took thought for utilitarian irrigation schemes as well as spectacular aqueducts.33 Keenan-Jones’s second case concerns the aborted project of 15ad to make the waters of the River Clanis flow northwards into the Arno instead of southwards into the Tiber, to prevent the Tiber from flooding again. It could probably have been done (since it was in fact done by 1700), but as why it was not, we are at the mercy of Tacitus. The most interesting aspect of the matter, it seems to me, is the inability of Rome’s famed hydraulic engineers to find a solution that would have prevented serious floods in the capital.

31 It might be tempting to associate this erosion with local deforestation, but the pollen data from the port sediments at Naples, which show an increase in cabbage cultivation during the third century (Ermolli et al., this volume, fig. 4) and the spread of wild vegetation, seem to exclude this. 32 Frederiksen 1984, 334. 33 Beltrán Lloris 2006, with bibliography on other cases (166–167, 192–193). 10 w.v. harris

These case studies, together with those incorporated in the papers by McCormick and Veal, have shown how much hard work is necessary to produce results in this field. All the more reason for scholars in diverse fields to attempt to collaborate. But an even more pressing conclusion is that all of us who are concerned with such questions about the past as these need to debate our objectives and consider carefully which questions most need answering. PART ONE

FRAMEWORKS

ENERGY CONSUMPTION IN THE ROMAN WORLD*

Paolo Malanima

Economic development has been supported, over the last two centuries, by a technical revolution in the use of power and energy. The introduction of modern machines, able to deliver huge quantities of work per unit of time on the one hand, and the availability of cheap fossil energy sources on the other, have enormously increased productive capacity. Both changes were the necessary although not sufficient conditions for the notable discontinuity in the economic history of the human populations and were the main determinants of a huge increase of output. The scarce availability both of mechanical power and energy set a limit to the growth potential of previous agricultural economies from the 5th millennium bc until the start of modern growth two centuries ago, and was the direct determinant of phases of decline or collapse. We cannot but agree with the view presented by E.A. Wrigley on pre-modern agricultural or ‘organic’ societies. His opinion is that ‘societies before the Industrial Revolution were dependent on the annual cycle of plant photosynthesis for both heat and mechanical energy. The quantity of energy available each year was therefore limited, and economic growth was necessarily constrained’.1 This was the main reason why decreasing returns to labour prevailed in past agricultural civilisations, as the English classical economists maintained. The topic of energy consumption as a whole has been only marginally investigated in the case of the Roman world (though there has been some attention to particular energy sources such as wood). Previous attempts to quantify energy consumption do not allow one to understand the procedures followed.2 It is obviously impossible to present definite figures of energy consumption, since local conditions and the relations between human beings and the environment differed so much within the Roman

* I thank Elio Lo Cascio for his comments on a previous draft of this paper. I also thank the participants in the conference ‘Growth and Factors of Growth in the Ancient Economy’,January 28–29, 2011, held in Chicago (with the support of the Federal Reserve Bank of Chicago), and particularly Alain Bresson and Joel Mokyr, for their comments. 1 Wrigley 2013, 1. See also Wrigley 2010 on the same topic. 2 See the Appendix. 14 paolo malanima

Empire. It is possible, however, to present plausible data and plausible confidence intervals around the figures. This is a first step towards a comparison of energy consumption within past societies and between past societies and the present world. The purpose of the present work is to focus on energy consumption in the early Roman Empire; and, in particular, to identify the energy sources (§1), to quantify their exploitation (§2–3), and their constraints to the growth potential (§4–5). The last section (§6) will be devoted to the dynamics of the ancient energy systems, that is the innovations in the technical exploitation of energy and its availability. The Appendix will present the procedure followed in the quantification of energy consumption in the Roman Empire and discuss alternative estimates.

1. The Input of Energy

Often it is not completely clear what actually were the sources of energy in past agrarian civilizations.3 The consequence is that any quantification becomes imprecise or, indeed, quite impossible. Although certainty is unat- tainable on the subject, a plausible order of magnitude is not out of reach.4 There were three main inputs of energy in pre-modern agrarian civilizations from about 5000bc until 1800ad: food, firewood and fodder for working animals.5 Food has been the primary source of energy since the beginning of the human species. A second source, firewood, began to be exploited as fuel between 1,000,000 and 500,000 years ago. From then until the Industrial Rev- olution it was the main provider of heat.6 The third source, fodder for draft animals, began to supply mechanical work in the agricultural civilizations between 5000 and 4000bc, that is since the exploitation of animal power on a wide scale in agriculture and transportation. These were still the main

3 Here I refer to the energy sources with a cost (often an opportunity cost). Solar light is important for our survival, but is free and then excluded from our calculations. The same holds true for the vegetation of a forest, when not exploited by the humans. Water and wind power, when exploited through mills and sails (expensive to build), is included, while it is excluded when not exploited for some productive activity. See, however, the Appendix for more information on the subject. 4 I have discussed this topic in greater depth in Malanima (forthcoming). See the following Appendix on the quantification of energy consumption in the early Roman Empire. 5 I have examined the transitions among energy systems in greater depth in Malanima 2010. 6 Perlès 1977; Goudsblom 1992. energy consumption in the roman world 15 energy carriers of ancient Mediterranean civilization. The discovery of fire on the one hand and the exploitation of draft animals on the other, marked two main changes in the history of technology. The most recent change has been the spread of thermal machines over the last two centuries. In the long period between the first exploitation of animal power in agriculture and the steam engine, so for almost seven millennia, no radical change, or macroinvention,7 occurred in the exploitation of energy, although several minor changes took place. Food consumption has not changed so very much during the long history of mankind, at least in term of calories. Even in the case of ancient Greek and Roman civilizations, we can assume a daily average consumption of 2–3,000 calories;8 as recent estimates indicate. In particular, ‘the diet of the Mediterranean region with its high population density was probably marked by much lower overall meat consumption’.9 Pork meat ‘was a promi- nent food of the urban high-income strata of society, whereas the poorer ancient Roman population consumed primarily vegetarian food’.10 Although within a wide geographic area such as the Roman Empire differences in diet were remarkable, the intake of calories was necessarily similar.11 Regional variations in firewood consumption were much wider and depended on two main variables: temperature and industrial demand. In Mediterranean civilizations the amount of 1kg. of wood (that is about 3,000 kcal.) per head per day can be assumed as the lower margin of a likely range, given the relatively high temperature in this area of the world. Calculations of industrial consumption by metallurgy and other industries (such as pottery, glass and tile production) and services (such as baths) suggest that another half kg. could be added to this daily amount, at least in regions with widespread industrial activity. This half kg. more is, however, a relatively high estimate, based on what we know on early Modern Europe.12 For the early Roman Empire only rough estimates on wood consumption by metallurgy are possible.13 Differences in firewood consumption certainly existed within the Roman world and derived from the regional differences

7 I use here the word ‘macroinvention’ following Mokyr 1990. 8 Here I use the terms of kilocalorie (kcal.) or calorie as synonyms, although they are not. Actually, a kilocalorie (the correct unit of measure when we speak of food or heat) is 1,000 calories. 9 Koepke and Baten 2008, 132. 10 Koepke and Baten 2008, 142. 11 See Jongman 2007b. 12 Kander et al. 2013. 13 See the Appendix. 16 paolo malanima in temperature and industrial development. A range between 1 and 2kg., that is between 3,000 and 6,000 calories per head per day, seems plausible.14 According to a calculation of biofuels consumption on a world scale about 1850, that is when wood was still the main fuel, the per capita average was 2.3kg. and this average was far lower in the South.15 When taking into account the high temperatures in the Southern Mediterranean and the existence of regions with poor industrial activity, a lower estimate of firewood consumption of about 3,000 kcal. per head per day, that is 1kg., seems plausible for the Roman Empire. A consumption of 6,000 kcal, equivalent to 2kg. of wood, could however have been reached in cold regions, in the mountains, or in areas with relatively high industrial activity.16 As to the contribution by draft animals to the energy balance, an estimate can be based on the ratio between their consumption of fodder (expressed in some energy measure) and population. We follow, in this case, the same procedure we use today to establish the average consumption of oil in a country: that is, dividing the oil consumed among the population. The only difference being that in pre-modern agrarian civilizations, we are mainly dealing with biological converters and that their fuel is food intake. From the available information on the size of ancient working animals17 and the draft animals-population ratio,18 we then estimate how much energy was consumed per head dividing the calories of fodder intake by the population. The range of a plausible consumption is 1,000–2,000 kcal. per head per day. The only energy carriers not provided by the land through photosynthesis in ancient agricultural civilizations were wind, used to drive sailing ships, and water, exploited for mills as from the 3rd century bc.19 An estimate of the consumption of the energy of wind and water is difficult.20 We know, however, for the early Modern Age, that their contribution to the energy balance hardly represented more than 1 percent of the total energy consumed. It seems plausible to assume that watermills and sailing ships were not more numerous in the Roman Empire than in medieval and early modern Europe.

14 The article by Harris 2011a is important for the quantification of firewood consumption. 15 Fernandes et al. 2007 (see the auxiliary material for the article in http://onlinelibrary .wiley.com/doi/10.1029/2006GB002836/suppinfo). The consumption of biofuels in the Medi- terranean regions was lower than the average. 16 See the lower energy consumption proposed by Smil 2010, reported in the Appendix to this paper. 17 On the topic see in particular Kron 2000, 2002, and 2004. See also Ward-Perkins 2005, Ch. VII and Fig. 7.3. 18 This ratio is hard to establish for ancient economies. See, however, the Appendix. 19 Wilson 2002a and 2008b and Lo Cascio and Malanima 2008. 20 But see the Appendix. energy consumption in the roman world 17

In mere quantitative terms, the role of wind and water in pre-modern agrarian societies was negligible, although they were very important from the technological viewpoint. Actually, sailing ships and watermills were the only engines whose mechanical work did not derive from the metabolism of food.21 Together these engines provided 100 percent of the mechanical energy by non-biological converters.

2. A Quantification

Table 1 presents a likely consumption range for the ancient Mediterranean in the age of the early Roman Empire, that is the 1st century and the first half of the 2nd, up until the Antonine Plague. As we see, energy consumption is comprised between 6,000 and 11,000 kcal. per capita per day (or 9.2–18.4 Gigajoules per year). We see also that half of consumption consisted of food for humans and draft animals, the other half of firewood.

Table 1. Energy consumption in the early Roman Empire (in Gj. per capita per year and kcal. per capita per day).

Gj/year Kcal/day % Sources of energy Min. Max. Min. Max. Min. Max. Food for humans 3.1 4.6 2,000 3,000 33 27 Fuel 4.6 9.2 3,000 6,000 50 55 Fodder for animals 1.5 3.0 1,000 2,000 17 18 Total 9.2 16.8 6,000 11,000 100 100 Sources: see text and Appendix.

Today World energy consumption is 50,000 kcal. per capita per day or 76.5 gigajoules (Gj.) per year. In Europe it is notably higher: 100,000 kcal. per day (153 Gj. per year). At the beginning of modern growth, in the early decades of the 19th century, World average consumption per capita was 7–10,000 kcal. per day (10–15 Gj. per year) and the European 15,000 kcal. per day (23 Gj. per year).22 Around 1850, consumption per head of the three main sources of energy (food, firewood and fodder) in Northern Mediterranean

21 Technical change in maritime technology was continuous and certainly contributed to enhance the exploitation of wind power, although, in mere quantitative terms, the energy consumed by sailing ships remained modest. See now, on changes in maritime technology, Harris and Iara 2011. 22 Malanima 1996 and 2010. 18 paolo malanima countries (Portugal, Spain, France, Italy) ranged between 11,500 and 13,500 kcal. per day.23 A Mediterranean average including Northern Africa and the Near East (for which we have no data until 1970) would certainly be lower.24 Thus, a plausible result is that per capita energy consumption in the ancient Roman world was 5–6 times less than the World average in 2000 and 10 times less than the European average at the same date. It was also a little lower than that of the Northern Mediterranean countries at the beginning of industrialisation. The Roman Empire included many Southern regions, where the consumption of firewood was certainly lower than in the Mediterranean countries of Europe at the start of industrialisation. It is hard to specify the impact of the production of energy on the environment in the early Roman Empire. If we assume that food production required half a hectare per capita,25 firewood half a hectare of forest and fodder for draft animals another half hectare, then per capita requirement was 1.5 hectares. This estimate is nothing but a plausible average (based mainly on late medieval-early modern European examples, where the productivity of fields, meadows and forests was quite similar to that in Roman antiquity). In around 165ad, the Roman Empire measured 3,800,000km2.26 Accept- ing the previous calculations regarding consumption and soil per head, to provide energy for the 70 million inhabitants living in the Empire 1,050,000 km2 were necessary, which is 25–30 percent of the total. If we assume a population of 100 million, plausible as well for the middle of the 2nd century ad, the need of soil to support energy production becomes 1,500,000km2, which is 40 percent of the Empire. If we exclude the mountains (lands more than 600 metres high), which in the Mediterranean regions cover 20–25 percent of the total area and were hard to exploit, the extent of the agrarian soil in the Roman Empire becomes about 3,000,000km2. In this case, according to the two previous population estimates, the share covered by fields, exploitable woods and meadows becomes respectively 33 and 50 percent of the total area. These shares naturally rise if we subtract from the total extent not only the mountains, but also hilly lands hard to cultivate, marshes, lakes and urban areas.

23 Kander et al. 2013. 24 For these countries the series elaborated by IEA (International Energy Agency) start only from the 1970s. 25 Fallow land is not included. 26 I take both the extent of the Empire and the inhabitants from Scheidel 2007, 48. energy consumption in the roman world 19

3. Efficiency and Energy Intensity

Only a part of energy input is actually transformed into useful energy (or energy services, that is mechanical work, light and useful heat). How great this share is depends on the efficiency of the converters of energy, that is labour (L) and capital goods (K). The thermodynamic efficiency (η) of the system of energy can be represented through the following ratio between the energy services (Eu) and the total input of energy (Ei):

Eu η Ei = Today, in our developed economies, this ratio is about 0.35; that is 35 percent of the input of energy becomes actual mechanical work, light or useful heat. In past agricultural civilizations, the efficiency was much lower. A plausible calculation is easier for the past, when biological converters prevailed, than for the present. Today, in fact, the variety of machines, with diverse yields, make any estimate hard. The ratio between useful mechanical work and input of energy into biological converters, such as humans and working animals, is around 15–20 percent.27 Part of the intake of energy in the form of food is not digested and is expelled as waste, whilst the main part is utilized as metabolic energy in order to repair the cells, digest and preserve body heat. A human being or animal consumes even when inactive. The use of firewood is even less efficient. The greater part of the heat is dispersed without any benefit for those who burn the wood. Its yield is about 5–10 percent. Overall, the efficiency of a vegetable energy system based on biological converters, such as that of ancient civilizations, was around 15 per cent at the most: that is 1,000–1,500 kcal. were transformed into useful mechanical work or heat; the rest was lost. Thermal machines are much more efficient than biological converters such as animals and humans. Another measure of efficiency in the use of energy is the ratio between the energy input and output, that is GDP. It represents the energy intensity, or the quantity of energy we need to produce a unit of output (Y ):

Ei i Y =

27 See the useful Herman 2007. 20 paolo malanima

This ratio depends on the efficiency of the converters, but, contrary to the previous ratio, it also depends on the structure of production, that is the relative importance of the different sectors and subsectors within the economy. Some sectors (e.g. industry and especially heavy industry) consume much more energy per unit of output than others (e.g. some services). If there is a change in the relative importance of any specific sector, energy intensity changes as well, even without any change in the thermodynamic efficiency of the converters. It is apparent that the impact of energy use on the environment depends both on the amount of energy exploitation and on energy intensity; higher intensity implying a higher impact on the environment. In past agricultural civilizations, for any unit of GDP (e.g. 1 dollar), the expense of energy was higher than today. Around 2000, in Western Europe, energy intensity was 7–8 Megajoules per dollar.28 In past agrarian economies it was at least twice as much, since mechanical converters of energy are more efficient than biological converters. In 1800 Western Europe, that is before the start of industrialization, it was 12–14 Megajoules per dollar. Assuming that in the early Roman Empire energy intensity was the same as in pre-modern European societies, the level of per capita GDP would be about 1,000 dollars (1990 intern. $ Purchasing Parity Power).29

4. The Energy Constraints

Vegetable energy carriers, such as those exploited in past pre-modern civilizations, are reproducible. The sun’s energy enables a continuous flow of exploitable phytomass and the circulation of water and wind. Although the availability of these carriers was and is endless,30 and the energy system based on them was and is sustainable, their increase was hard and time- consuming. A large part of working time in pre-modern economies was aimed at providing energy. All in all, the expense31 for energy (food, firewood and fodder) could represent 60–70 percent of the average income. In pre- modern economies consumption represented, at least, 80 percent of GDP.32

28 International 1990 Geary-Khamis dollars Purchasing Parity Power. 29 See on the topic Lo Cascio and Malanima 2009; forthcoming. 30 Actually, it is not endless, but the Sun’s light will still reach the Earth for 5 billion years. 31 Including the opportunity cost when a source of energy is provided directly by the consumer himself. 32 Malanima 2009, chap. VII. energy consumption in the roman world 21

Although this 80 percent was not devoted completely to providing energy, the expense for food and firewood was remarkable. Since all sources of energy came from the soil and soil is not endless, the consequence during epochs of demographic rise was a fall in soil per worker and then decreasing returns to labour. The main change taking place from the start of modern growth has been the elimination of the dependence of the energy system on the soil’s constraint. When demand increases, it is much easier to provide coal, oil or natural gas, than the vegetable carriers utilized in past agrarian economies. Since in pre-modern organic vegetable energy systems, the transformation of the Sun’s radiation by plants into phytomass, thanks to photosynthesis, was central and climatic conditions can heavily influence the output of energy, climatic phases marked the past history of mankind. Short-term deviations from the average temperature or precipitations resulted in dramatic increases or falls in energy availability: the well-known years of plenty and the frequent famines of the agricultural economies. Long-run changes were much less felt or were even unnoticed, although they influenced agricultural production, thus the overall availability of energy, and, consequently, total output and population trends. The second important constraint of all pre-modern energy systems was the low power of the converters, which resulted in a low working capacity per unit of time. The high standard of living of modern societies is the result of the higher output per unit of time or higher labour productivity. The power of a man in everyday work is the same as a 40-watt lamp, or 0.05–0.07 Horse Power (HP). The power of a horse is 15–20 times higher. In pre-modern civilizations, the most powerful engines were watermills, whose power was about 3 HP, and sailing ships, which could even reach 50 HP.33 To clarify this central point about the differences between past and modern energy systems, we must remember that the power of an average car (80 kilowatts) is equal to the power of 2,000 people and that the power of a big generating electric station (800 megawatts) is the same as that of 20 million people. The electric power of a medium sized nation such as Italy in 2000 equals 80,000 megawatts, which is the same power as that of 2 billion people. Today, a nuclear plant or a nuclear bomb can concentrate millions of HP, or the work of many generations of humans and draft animals, into a small space and a fraction of time.

33 I neglect here the employment of power for military purposes. A catapult was an ingenious concentration of power. 22 paolo malanima

While the adoption of new energy carriers in the past two centuries has greatly expanded the quantity of energy at our disposal, an equally key development has been new technology (machinery) able to concentrate large amounts of work in particular locations in order to carry out specific tasks. This concentration of work allows humans to accomplish tasks that were barely imaginable just a few lifetimes ago. It was the first step toward a new control of the natural forces at a level inconceivable in past agrarian civilizations.

5. Innovations

The progress of technology in the ancient Mediterranean world did not reveal interruptions or declines:34 ‘the use of machines was more widespread in ancient Greece and Rome, together with ancient China, than in any other civilization until certainly the 12th or perhaps the 14th century A.D. in West- ern Europe’.35 On the other hand, looking at the problem of technical innovation from the viewpoint of energy, Roman technology consisted primarily, as J-P. Vernant wrote, ‘in the application of the human and animal force through a variety of tools, and not in the utilisation of the forces of nature through the use of machines’.36 The introduction of new tools, that is micro- inventions, was continuous. In a sense this flow of innovations made human work more efficient, although this increase in efficiency, from the specific viewpoint of energy and power, was modest indeed. As suggested by A. Bresson, in the 1st century ad,37 Hero’s work demonstrates the knowledge of all the main elements for constructing a steam engine, such as the conversion from rotatory to alternating movement, the cylinder and piston, non-return valves and gearings: ‘the main technical elements embodied in the Newcomen engine were, if not in function at least well known in the Hellenistic age’.38 We can wonder, however, how widespread this knowledge actually was. With the exception of Hero’s work, no other mention of the use of steam is available in ancient literary texts or archaeological remains. We know that in England coal began to be used on a wide scale from the 1st century ad both for domestic usage and for the melting of metals.

34 Greene 2000; Schneider 2007. 35 Wilson 2008b, 362. 36 Vernant 1957, 207. 37 Described in Pneumatica 2.11. 38 Bresson 2006, 72. energy consumption in the roman world 23

Fig. 1. Dated remains of coal in England 1–500ad (% of the total dated remains every 50 years). Source: based on data in Smith 1997.

Coal has been recovered from 70 archaeological sites in England and Wales. Its chemical analysis has allowed these remains to be dated (Figure 1).39 Although we cannot quantify the level of consumption, we can specify the chronology of its exploitation. When the Romans conquered England, coal was already exploited. Its utilization spread and attained a maximum level from the 2nd until the 4th century. At least until the 5th century ad, coal continued to be used on a wide scale. Later it almost disappeared. Coal, however, is very unevenly distributed across the globe, and, apart from Australia, is almost entirely found in a few parts of the Northern hemisphere, that is, North America, North-Western and Eastern Europe, Russia and China. The centres of ancient civilizations and especially the Mediterranean regions are not comprised in the geography of coal. The high price of firewood on the one hand and the lack of coal on the other did not allow the transition towards a new energy system in a Mediterranean civilization.40

39 The decline of the curve in Figure 2 coincides with economic decline in Britain. See the trend of the British economy described by Ward-Perkins 2005, Ch.V. 40 Bresson 2006, 77. 24 paolo malanima

6. An Energy Crisis?

It is still hard to quantify the rise in population during the millennium spanned by ancient Mediterranean civilizations. While historians do not agree on the figures, they do agree, on the trend of population. In 800bc, some 20 million people lived around the Mediterranean Sea, whereas in 150ad the population of the Roman Empire numbered 70 million,41 although the estimate of 100 million could be equally plausible, given the uncertainty of any estimate for that period. Such a level of population was again attained by the European continent (without Russia), only in the early modern centuries. Although a calculation of the carrying capacity of the Mediterranean world is risky, the estimates proposed above regarding the extent of land necessary to support the population in energy sources do suggest that the rising population put pressure on resources. Data on decreasing returns to labour are, however, scanty and uncertain. It has been suggested that body size diminished in Western Europe from 150ad, after a period of rise.42 On the topic, however, there is no certainty at all. Koepke and Baten write that ‘during Roman times we have more or less stagnating heights’.43 If stature actually diminished, probably it diminished later in Central and Northern Europe (e.g. Germany) than in the Mediter- ranean regions.44 A wider knowledge begins to be available on climate and we can start to speculate on the possible influence of climatic changes on the availability of energy sources. On this topic as well, the evidence is still contradictory, however. For a long time the rising pressure of population was supported by rising temperatures in the Mediterranean and the whole of the Northern hemisphere, during the Ancient Climatic Optimum.45 Historians agree on the existence of a Roman Warm Period.46 Research on ice carrots from Greenland ice core and the ratio of two oxygen isotopes (18O/16O) provides a record of ancient water temperature and then climatic oscillations. On this basis changes in temperature have been reconstructed over several million years. Annual changes from the 1st century bc are represented in Figure 2.

41 Scheidel 2007, 47. 42 This is the opinion expressed by Jongman 2007a, based on data collection by Geertje M. Klein Goldewijk. See also Kron 2005 and 2008. 43 Koepke and Baten 2008, 150. 44 Koepke n.d. 45 Haas 2006, 147–150. 46 Sallares 2007, 19. See also the long-term view in Blender et al. 2006. energy consumption in the roman world 25

Fig. 2. Oxygen isotopes in the ice carrot GISP2 (Greenland glacier ice core) 60bc– 350ad. Source: Rossignol 2012, 97.

We can see that the two centuries bc were favourable from a climatic viewpoint. Temperatures were high during that period and remained so until the middle of the 2nd century ad. Some historians suggest that, after 150ad temperatures diminished remarkably, as the curve in Figure 2 shows. Very little, however, is known about the evolution of climate in the Mediter- ranean. Rossignol has claimed that ‘a remarkable worsening of the climatic conditions’ occurred from about 150ad. The middle of the 2nd century ‘witnesses the end of a warm period during which the ratio of the oxygen isotopes had attained levels which would only be reached again in the 20th century’.47 The presence in the ice carrots of sulphuric acid, dated between 153 and 162, reveals the influence of volcanic eruptions on the fall in temperatures.48 Higher temperatures mean that the season for harvesting vegetables is longer; that land can be cultivated at higher altitudes and further North.49 Soil per worker rises when temperatures are milder. The opinion expressed by S.W. Manning is more cautious: ‘A range of records indicate that a stable and reasonably positive (warm, and in a number of areas or cases also mainly moist) climate regime was in place for the period from about the 2nd century bc through the 2nd century ad. This

47 Rossignol 2012, 96. See also Manning this volume, Fig. 8. 48 Rossignol and Durost 2007. 49 See also Weinstein 2009. 26 paolo malanima

Fig. 3. Intensity of precipitations between 400bc and ad400 (and range of error) (mm. per year). Source: Büntgen et al. 2011, 581. unusual status, reducing some of the typical variability, uncertainty and risks of the Mediterranean climate regime for farming, would have been conducive to the growth of the Roman world. It was also an especially favourable time (warm, moist) for both agricultural and demographic expansion in central and northern Europe’.50 According to Manning, ‘the stability of the previous several centuries ended; agricultural uncertainty and bad years would have increased’. It is hard, however, to specify the turning point towards decreasing temperatures. The 2nd century does not reveal, in his opinion, a clear declining trend. Precipitation has been reconstructed for the region of Israel51 and for Germany and Switzerland.52 We know that it diminished and the climate became drier when the temperature was falling (Figure 3). In Central Eu- rope, precipitation peaked in 100bc, but from then on diminished, reaching a minimum in ad300 (100 millimetres less than in the second century bc). The climate became ‘increasingly dry’.53 According to Manning, ‘the 2nd to

50 Manning this volume. 51 Orland et al. 2009. 52 Büntgen et al. 2011, 581. 53 Schmidt and Gruhle 2003a and 2003b. energy consumption in the roman world 27

5th or 6th centuries ad seem to be relatively arid in several areas of the eastern Roman empire, and the indications of less favourable climate conditions further East into central Asia may have been one of the forcings behind the movements of populations that led to invasions/migrations into the late Roman world’.54 The pressure of population on the energy resources both to provide food (and then widen the arables) and firewood resulted in a decline of the forested areas.55 ‘By the end of the Republic, most of the areas of Italy that were accessible to Rome had lost most of their stands of tall trees, but except for some metal-working centres, most places had stabilized their fuel supplies. Patches of eroded land continued to multiply, however, all the way through the high-imperial period of prosperity’.56 In Spain, ‘climate deterioration’ would have hampered ‘vegetation recovery after fire and exacer- bate[d] human impact (deforestation) in general’.57 In such cases, because of the need to meet the inelastic demand for food, the livestock and meadows diminish (although for the ancient world nothing certain can be said on the matter). Intensification occurred in agriculture and convertible husbandry spread to support the demographic rise at least in Italy.58 For a comparison, in Europe, between 1500 and 1700, the 40 percent rise in population, from 80 to 120 million,59 resulted in a 20 percent decrease in agricultural product per capita (that is energy, since the greater part of energy came from the fields).60 Population pressure on the energy sources diminished certainly after the Antonine Plague, that spread between 160 and 170ad,61 as archaeological wood remains from Central Europe seem to suggest (Figure 4). By themselves, neither population rise nor climatic changes are necessarily connected to phases of economic decline. Their coincidence can, however, deeply influence the economy and provoke destructuration and finally collapse.

54 Manning this volume. 55 See, however, the reconstruction by Kaplan et al. 2009. See also Ruddiman and Ellis 2009. 56 Harris 2011a, 139. 57 Kaal et al. 2011, 172. 58 Forni and Marcone 2002. 59 Russia is included in these estimates of population. 60 Kander et al. 2013. 61 See especially Lo Cascio 2012. 28 paolo malanima

Fig. 4. Estimates of forest clearance in Central Europe (Germany, North-Eastern France) from archaeological wood remains 200bc–400ad (decadal data; any point of the diagram represents the intensity of the felling). Source: Büntgen et al. 2011, 580.

Conclusion

The energy system of ancient Mediterranean civilizations was the same as that of all agrarian societies. Despite the increase in useful knowledge and the extensive development of the agrarian energy basis, supported by a favourable climatic phase, this system was finally unable to support the increasing needs of the rising population (as always in agrarian civilisations). If we follow the economic approach by the classical economists, rephrased by E.A. Wrigley with particular reference to energy, an increasing pressure on the resources by the rising population would have been followed by decreasing returns and then diminishing energy availability, after some centuries of rising population. Data showing a clear economic trend for the first centuries of the Empire are almost entirely lacking, but an unfavourable climatic phase, beginning probably, but not certainly, in the second half of the 2nd century ad, contributed to a decline. Much later, during the Little Ice Age, in the early modern centuries, the reaction to a similar crisis was a much wider use of coal.62 This main change developed in England since the 16th century. Then, in the 18th century, the steam engine began to interact with the new, rising input of energy. This

62 The topic is discussed in Malanima 2010 and 2011. energy consumption in the roman world 29 interaction initially began to involve the Central and Northern European regions and subsequently also the regions far from the centre of the great change then in progress. The combination of changes in power and energy was the basis of modern growth. Just as in many other pre-modern societies, the structure of the energy system prevented ancient Mediterranean civilizations from following a similar path. Ancient growth found in its energy basis a main constraint to its further economic progress. 30 paolo malanima

Appendix

Estimates of Energy Consumption in the Early Roman Empire A wider analysis of per capita energy consumption in the early Roman Empire is presented in Malanima (forthcoming). The topic of energy consumption in pre-modern economies is also discussed in Malanima 1996, 2006, 2009, 2010, 2011 (www.paolomalanima.it). As seen above (§1–2), sources of energy of pre-modern, agricultural economies are the following: 1. food; 2. fuel (almost always firewood); 3. fodder for working animals; 4. water and wind power.

1. Food Food consumption has always been the most stable energy carrier ever exploited since the beginning of the human species. In the following diagram (Figure 5), I report the series presented by Jongman (2007b, 599), on calorie consumption in present day populations. Taking into account the age structure in Roman antiquity, with more young people than today, the range of 2–3,000 calories seems plausible. Considering yields per hectare, to cover the needs of a family of 5 people, about 5 hectares were necessary, including fallow lands. Thus a family needed between 2.5–3.3 hectares of cultivated land (excluding fallows): i.e. from half to two-thirds of a hectare per person (for data on yields, and soil per capita necessary to satisfy food demand, see Forni and Marcone 2002, on agriculture in Roman Italy).

2. Firewood As said in §1, firewood consumption depends on temperature and industrial use. One kg. of wood can be seen as the lowest possible level of consumption (as also stated by Harris 2011a; see also data in Pireddu 1990, 27). Although hard to quantify, firewood consumption was low where temperatures were high and high where temperatures were low (see, for instance, data in Warde 2006, referring to early modern Europe). If, to simplify, we assume that in a Mediterranean climate, each individual consumed 1 cubic metre of wood per year, that is 625kg., including industrial uses as fuel (1.7kg. per day), this amount of wood could be provided by the yearly growth of half a hectare of forest (Chierici 1911, 232–233). Assuming that the population of the Roman Empire in around 165ad was 70 million inhabitants for an energy consumption in the roman world 31

Fig. 5. Food consumption by modern populations according to age (kcal.) Source: Jongman 2007b, 599. area of 3,800,000km2, and that every inhabitant consumed 1 cubic metre of firewood (including wood from prunings), then the total requirement was 70 million cubic metres. It could be provided by a wooded area of 350,000km2, or 9–10 percent of the total inhabited surface of the Empire. With a population of 100 million inhabitants, the wooded area rises to 500,000km2, or about 13 percent of the total. A city such as Rome, with 1 million inhabitants in the age of Augustus, needed 50km2 of forest to cover its needs. As to industrial consumption, we can only provide some calculations from what we know about the output of metallurgy. Let us assume that iron production was between 80,000 and 160,000 tons per year (cf. Harris 2011a) and, at the lowest, a consumption level of 30kg of firewood (transformed into charcoal) per kg. of iron (Smil 1994, 144–156). Charcoal, known in Egypt as early as the 3rd millennium bc, was widely used in Greek-Roman antiquity (Wikander 2008, 138). For the production of 80,000 tons of iron, the quantity of firewood would thus be 2,400,000 tons (converted into charcoal). In cubic metres, the requirement was 3,840,000 (assuming 625kg. per cubic metre, and then dividing 2,400 million kg. by 625). With a yearly productivity of half a cubic metre per hectare of forest, in order to produce 3,840,000 cubic metres, 1,920,000 hectares or 19,200km2 were necessary.Assuming iron output being twice as high, the need amounts to 38,400km2 of forest. This 32 paolo malanima area is only 5–10 percent of the total forest required by the population for heating and cooking. As said before (§1), other industries (such as pottery, glass and tile production) and services (such as baths) exploited wood. An estimate is, in this case, impossible. Fuels different from firewood represented a negligible share of the total. Thus, our estimate for a Southern, Mediterranean civilization such as the Roman Empire is between 1 and 2kg. of wood, that is 3,000–6,000 calories per capita per day.

3. Fodder The estimate of fodder consumed by draft animals is more complex. From the viewpoint of energy, an ox or some other working animal is like a machine. It metabolizes vegetables to accomplish a task. In order to establish the average consumption in energy sources per head, the input of energy by a draft animal must be divided by the family members that exploit it. We know that improved fodder management and nutrition determined a remarkable increase in the size of animals during Graeco-Roman antiquity. Ley farming and meadows supplied animals with better fodder than in the late Middle Ages and early modern times (Kron 2000). Oxen were taller and heavier than in Medieval and early modern Europe: about 400kg instead of 2–300 (Kron 2002 and 2004). We can establish a ratio between working animals and population in ancient Mediterranean civilizations from the technical relationship suggested by ancient agronomists between land and working animals. In the 1st century bc, Varro recalls the opinions of Cato and Saserna about the need of a yoke for every 80–100 iugera (20–25 hectares) (On agriculture 1.21–22). Since a yoke is composed of two oxen, the relationship is therefore a working animal per 10–12.5 hectares. A century later, Columella tells of two yokes of oxen for a farm of 200 iugera (or 50 hectares) (On agriculture 2.12.1–7). Again we find a ratio similar to that suggested by Varro and relatively close to the animal-land ratio found in early Modern Europe. Since a peasant family required a farm of about 3–5 hectares to support its living (as shown in §1 of this App.), we could divide among the 10–15 members of two average families endowed with a farm of 3–5 hectares each, the calories from fodder consumed by oxen (25–30,000 kcal per animal per day) and we would obtain the result of 1,700–3,000 kcal. per head. We would have to add to this estimate horses (on which see Vigneron 1968), mules, donkeys and camels, and we would also have to include urban inhabitants (excluded from the previous draft animals-peasant families ratio) in the denominator of our ratio. All things considered, a range of 1,000–2,000 calories per day per capita seems plausible. energy consumption in the roman world 33

4. Wind and Water The only possibility of estimating the consumption of water and wind power is to start from power (work done per unit of time -1 second-). In the case of a large sailing ship, with a carrying capacity of 400 tons, a rare example in the ancient world, where the majority of sailing ships were below 100 tons (Greene 1986, 26), a relationship existed between tonnage and power. The power of such a ship (400 tons) was about 50 HP (Malanima 2006). Assuming (absurdly!) that this power was exploited fully for 24 hours and 365 days per year, energy per year would be 438,000 HPh (Horse Power hour is a measure of energy), that is, 770,000 kcal. per day. We would now need a plausible ratio between ships and boats on one hand and population on the other. Even assuming the ratio existing in early modern Europe to be correct, the result would be less than 1 percent of the entire energy consumption per capita. The watermill was the most powerful engine existing on land. Generally its power did not exceed 2–3 HP, although examples of big mills (Munro 2003) or the combination of several mills in powerful sets of engines are not lacking (Brun 2006; Wikander 1979, 2000 and 2008). The mechanical work produced by a watermill endowed with the power of 2 HP is about 64,749 kj. (15,000 kcal.) per day, and since a man consumes 2,550–3,000 kcal. per day as food, consumption of gravitational energy by a watermill is 6 times the energy consumption of food per capita. In late medieval and early modern Europe, a ratio existed between watermills and population: 1 watermill every 250 people. Otherwise stated, any small village of 50 families had its own mill (on the topic Makkai 1981 is important). If we divide a mill’s energy consumption by 250, the result is 60 kcal. Certainly, the use of mechanical energy to grind cereals was a remarkable achievement of ancient civilization. Its contribution to the energy balance was, however, modest in mere quantitative terms. Although we do not know the inhabitant-watermill ratio in the ancient world, and even allowing for the existence of the same late medieval ratio, which seems too high for antiquity, as early as the first centuries of the Roman Empire, the result is that the contribution to the energy balance was indeed modest (Reynolds 1983; Lo Cascio and Malan- ima 2008). Let us consider that previous calculations on mills and ships assume full-time work (24 hours per day), which is implausible. Contributions to the energy balance assuming more realistic working time imply a reduction of the available energy per head. 34 paolo malanima

The Estimates by Ian Morris Different estimates of energy consumption have been provided by Ian Mor- ris (2010a and 2010b). According to Morris (2010a, 28), the sources to be taken into account for a calculation of energy consumption (including the ones used in modern economies) are the following: Food (whether consumed directly, given to animals that provide labour, or given to animals that are subsequently eaten); Fuel (whether for cooking, heating, cooling, firing kilns and furnaces, or powering machines, and including wind and waterpower as well as wood, coal, oil, gas, and nuclear power); Raw materials (whether for construction, metalwork, pot making, clothing or any other purpose). We can see that there is a similarity between this list and the sources taken into account in this paper. However: 1. I do not include feed ‘given to animals that are subsequently eaten’, since it is already included in the 2–3,000 kcal. of food for humans (and it would be a duplication of the same source in our calculations). These animals certainly put a high pressure on carrying capacity. If agricultural produce is not consumed directly by the population, but consumed by animals which are then eaten by the humans, the pressure on land is higher. In any case those animals are only used as food and are not exploited in agriculture or transport. They are part of human food; which enters the energy balance. As a consequence, I include only feed for working animals; 2. it is not clear how Morris computes the contribution by wind and water power; 3. raw materials cannot be considered as energy carriers and are not included in my estimates (or in those of the International Energy Agency or the US Energy Information Administration). Morris follows, however, Cook 1971, who includes ‘vegetable fiber’, which brings ‘solar energy into the economy through photosynthesis’ (134). See also Cook 1976, 51 and 135. Raw materials, however, are not used as providers of energy. Fire- wood, is also generated by photosynthesis, hence when used as an energy carrier I include it in my calculations. When timber is used as raw material for construction, it is not included, despite being produced by photosynthesis. It is not an energy carrier in this case. The results by Morris are quite different from those presented in the previous pages. In the following Table 2 some data are reported from two series presented by Morris (2010b, 628). energy consumption in the roman world 35

Table 2. Energy consumption in advanced regions of the West and East according to I. Morris. 8000bc–2000ad (thousands of kcal. per capita per day). West East (000)(000) 2000 230 104 1900 92 49 1800 38 36 1700 32 33 1500 27 30 1000 26 29.5 200ad 30 26 1ad 31 27 200bc 27 24 8000 6 5 Source: Morris 2010b, 628.

Around 2000, the average world energy consumption was 50,000 calories per day. According to Morris’ estimate, in 200bc some parts of the World already exceeded this level even without fossil fuels. In both works by Morris (2010a and 2010b), previous data (reported in Table 2) for the year 2000 actually refer to the most advanced countries in the West (USA) and in the East (Japan). In addition, data for previous years refer to ‘the most developed core within the West’ (Morris 2010b, 42), whose borders, however, are not clearly defined. In any case, Morris’ results are too high. In 1800, according to recent research, energy consumption in Western Europe (a highly developed part of the globe) was not 38,000 kcal. (as maintained by Morris), but about 15,000 (average for Sweden, Norway, The Netherlands, Germany, France, Spain, Portugal and Italy) (Kander et al. 2013 and data published in Gales et al. 2007). In 1900, for the same countries of Western Europe, the average was 41,500 kcal. per day per capita, and not 92,000 (as in the previous Table 2). In England it was 95,000. Morris’ estimate for 1900 is only plausible if by ‘West’ we refer only to England. As we see, data for the Roman Empire are also quite different from ours. Even if we take the most advanced part of the Roman Empire, Italy, in 1861, that is, the year of the Unification of the country, energy consumption per capita was 11–12,000 calories (Malanima 2006), less than half the estimate proposed by Morris for the West (31,000) in 1ad. Energy intensity represents the ratio between energy consumption and GDP. In Western Europe from 1800–1820 it was 12–15 Megajoules per 1 dollar (1990 international Gery-Khamis dollars), when per capita GDP was 1,200 36 paolo malanima dollars (according to the series by Maddison 2007, in 1990 international Geary-Khamis dollars PPP). If we assume the very high estimate of 1,500 dollars for Roman Italy (taking into account that recent estimates hardly exceed 1,000 dollars, as shown in Lo Cascio and Malanima 2009 and forthcoming), the resulting estimate of energy intensity, taking Morris’ estimate of 31,000 kcal. per head per day (and then 11,315,000 kcal. per year, or 47,342 Gigajoules), is 32 Mj. per dollar, and thus more than twice that ascertained in 1800 for Western Europe. With a GDP per capita of 1,000 dollars in the early Roman Empire, the implied energy intensity becomes 47 Mj. per dollar. For a comparison, in 2000, World energy intensity was 11.5 Mj. per dollar (1990 Geary-Khamis int. dollars) and in Europe it was 5.5 Mj. per dollar. Vaclav Smil (2010, 107–113) proposed estimates of energy consumption in ancient Rome that are far lower than those by Morris. Here is the comment by Morris on Smil’s views: ‘Roman total energy capture would be somewhere between 4,600 and 7,700 kcal/cap/day [according, that is, to Smil’s calculations]; if we assume that roughly 2,000 kcal/cap/day of this was food (which means ignoring the archaeological evidence for relatively high levels of expensive calories from meat, oil, and wine), that leaves just 2,600–5,700 kcal/cap/day to cover all other energy consumption’. To justify this estimate, Smil suggests that Roman fuel use was just 180–200kg. of wood equivalent per capita per year, or ‘roughly 1,750–2,000 kcal/cap/day’. Smil’s estimate of firewood consumption certainly seems too low. On the whole, however, Smil’s estimates are closer to mine than are those by Morris. FUELLING ANCIENT MEDITERRANEAN CITIES: A FRAMEWORK FOR CHARCOAL RESEARCH*

Robyn Veal

Introduction

Fuel in the ancient Mediterranean has to date received little detailed analysis. Humans in the Mediterranean consumed fuel in socio-culturally con- ditioned ways (i.e. history ‘in’ the Mediterranean); but that they could consume fuel at all, and which fuels were available in which areas, is very much a ‘history of ’ topic.1 Quantitative and qualitative studies of the economy have focused on production and trade of goods and slave labor, but the fuel economy has been difficult to trace in the historical sources, mentions being more incidental than material. The most important archaeological evidence, i.e. that of the archaeological charcoal, is not yet routinely collected by all excavators. This is an omission that begs attention, as ancient settlements could not function without fuel. The gathering of wood for fuel occasionally resulted in dramatic changes in the environment when over- exploitation occurred (for example, on islands), while in other places, more sustainable practices appear to have occurred. (Wood was not the only fuel in many parts of the Mediterranean: animal dung and agricultural waste such as chaff and olive lees were also consumed.) Geology, topography, and climate determine which trees may grow in a particular location; but

* The author thanks the conference organizer, W.V. Harris for the invitation to speak and to contribute this paper; and also for his helpful comments upon the draft. All conference participants are thanked for their enormous warmth and intellectual generosity. Access to at the time unpublished manuscripts by both W.V. Harris and P. Malanima is also gratefully acknowledged. I also thank A. Wilson for helpful discussion. A fellowship at the British School at Rome provided the optimal environment for writing this offering, which is also based on work completed at the Australian Centre for Microscopy and Microanalysis, and the Department of Archaeology (University of Sydney). 1 Cf. Horden and Purcell 2000. Fuel as a natural resource neatly straddles the nuanced distinction Horden and Purcell make between examining the (natural) history of the Medi- terranean, as opposed to human history conducted in it. We may examine these from prehistory to the modern period using all data at our disposal for one any period: scientific, archaeological and historical. 38 robyn veal politics, land ownership, cultural mores and agricultural practice moderated the physical factors. This contribution provides a framework for examining ancient Mediter- ranean cities’ fuel supplies. Archaeological charcoal is at the heart of this approach but aspects of the historic sources are also considered and a case study of Pompeii’s fuel economy c. third c. bc to ad79 is briefly overviewed in line with the methodology suggested. New scientific techniques beyond simple charcoal identification as to wood type have started to appear and are discussed here in terms of their usefulness for examining forest management and consumption. Further aims of this contribution are to encourage researchers to collect charcoal, and to show the detailed ways in which it can now be used to examine a city’s fuel supply. In time, with sufficient further research, it may be possible to synthesize regional patterns of supply and consumption for the Mediterranean (and the ancient world as a whole). Indeed the relevance of studying ancient wood fuel remains appears to have become greater today as we consider modern problems of climate change, and the potential of pelletized wood (at perhaps 70% of the calorific value of coal)2 as a part of our fuel future.

Studying the Fuel Economy: Modern and Ancient Difficulties

Fuel is a central part of most production processes, and as such forms part of a city or state’s, economic consumption. An economy’s size may be estimated by its Gross Domestic Product, or GDP. Two methods are routinely used to calculate GDP: i) the income method, or ii) the expenditure/consumption method. For the income (of households) method, GDP is defined as labor income + capital interest + rent. Under the expenditure/consumption model, GDP is equivalent to consumption + investment + government spending + net exports. In the ancient world, the latter is often expressed in terms of grain equivalents.3 These two methods implicitly include some sort of value for fuel. They are meant to arrive at the same value, and they are

2 Wynn 2011. See also Collaborative Partnership on Forests 2011. 3 The range of modern scholars’ calculations of the size of the ancient Roman GDP has been summarized recently by Lo Cascio and Malanima forthcoming. See also Scheidel et al. 2007. fuelling ancient mediterranean cities 39 usually expressed on a per capita basis.4 Besides estimating population, other difficulties in measuring both modern and ancient GDP include the fact that both methods omit ‘unpriced goods’, typically natural resources such as wood and water (or they may be underpriced). Both methods also omit ‘non-market’ activities, for example: barter and villa type non- monetized activities (operating outside regular markets), patronage, and euergetism. In ancient studies, a third method for calculating GDP has been more recently described, the so-called ‘new institutional economics.’5 This method more explicitly addresses the problems of calculating economic values in a system that is so moderated by social, religious and political mores. It is not the intention to review ancient economic methodology in detail here, but merely to work towards understanding the part in the ancient GDP fuel might have constituted, since, as already intimated, it appears to have been overlooked. In the twentieth century, data for the United States of America provide some food for thought. Energy as a proportion of GDP varied between 8% and 14% between 1970 and 1998, dropping back to roughly 9% in 2006.6 So, in recent history, there has been a more efficient use of energy per GDP dollar over time. Our modern energy sources are mostly nuclear and fossil fuel (petroleum) based (i.e. high calorific fuels by weight). In the ancient world wood and wood converted into charcoal predomi- nated, although there were many alternatives, such as peat, coal (in limited amounts), and the agricultural wastes mentioned in the introduction. Of these, olive lees and pits provide a very high calorific potential fuel, while peats and chaff provide much lower rates of return by weight.7 Many of these alternatives were often dried, pressed and even pre-charred prior to use, as they are even today in some locations. Charring increases calorific, i.e. heat, potential, which is an indicator of the quality of a fuel. How can we estimate fuel energy as a percentage of GDP in the ancient world? In the ancient economy, agricultural activities dominated, but are rarely thought to have been greatly efficient.8 Manufacturing also operated

4 Here lies one of the first problems of such estimates, which population figures should be used in ancient calculations? Also see Scheidel and Friesen 2009. 5 Bang 2009. 6 Institute for Energy Research 2010. 7 Modern work on biochar (any kind of charred organic material) is providing data for comparison with ancient equivalents (Sohi et al. 2009). At Leptiminus (Mediterranean Africa), amphorae of pre-charred olive waste were found outside a kiln and olive pits present elsewhere in some areas of the site were interpreted as fuel waste (Smith 2001, 434–435). Smith quotes several other examples of this phenomenon in the region. 8 Cf. Spurr 1986 and White 1970, 47–52, for Roman agricultural efficiency. 40 robyn veal at a basic level compared with the twentieth century, so if fuel is currently nearly 10% of the equation today, and relative efficiency in the ancient world is anything to go by, we might conclude that the value of fuel inputs might be higher than 10%. Perhaps as high as 20%, or more. In assessing the ‘value’ fuel has in the modern world we may refer to extensive modern pricing data that explicitly allow a comparison between the cost of fuel and the cost of other inputs to the economy. In the ancient world, we have little data on the actual volumes of fuel consumed, and even less on the prices (and in any event these will likely have been underestimated, as previously noted).9 We still lack the data to make more than an educated guess, but nothing will be lost if we attempt to qualify and quantify the fuel supply.

Sources of Information about Fuel

Most historical information about wood (written, epigraphic, sculptural) relates to timber for construction: of buildings, ships, and war machines, to name a few of the largest uses. Meiggs’ overview, Trees and Timber in the Ancient Mediterranean World, synthesizes the ancient historic sources, and the archaeological data available at its publication date, and focuses substantially on wood use for timber. At that time, charcoal analysis had been established as a discipline, but was as yet a fairly obscure specialization. Meiggs discusses fuel consumption mostly in relation to exceptional cases, for example, Delos, where all wood fuel for temple use was imported from about the 3rd c. bc.10 The amount of forest consumed proportionately for fuel (as opposed to timber) is, and was, relatively high. In modern day developing countries that are still substantially wood dependent (much of Africa), over 90% of harvested wood is used for energy, while about 10% is consumed for timber and other purposes.11 The focus on large timbers in the ancient world is unsurprising since the evidence for long distance trade

9 Diocletian’s Edict (with all of its incumbent difficulties, not least of which is its fairly late date) provides some useful data that tells us of the relative values of timber for building, as opposed to raw fuel wood, or ‘cuttings’ (kindling) (Graser 1959). Timber has the highest value, while kindling is rather highly valued. Relative transport costs are also provided. Of course it is not valid to assume the relative values of these commodities remained constant through time. Other small inferences about pricing can be gleaned from the historic sources, but these do not provide anything more than a broad indication. 10 Meiggs 1982, 441–457; see also Reger 1994. 11 Food and Agricultural Organisation 2011. We are not able to calculate this proportion for the ancient world at this point but the volume of wood used for construction in any current developing society has rarely been much greater than 20%. fuelling ancient mediterranean cities 41

Fig. 1. Factors affecting the wood supply, which underpin the types of archaeological charcoals found. The factors are named in general terms in the top left-hand part of the diagram, while at the bottom right, described are the ways in which these factors may be exemplified. in this commodity is well documented. The large cedars of Lebanon which were exported all over the ancient world to build large palaces and temples, stir the imagination, while the remains of fuel in the archaeological record, i.e. those small black fragments that often actually fall through an archaeologist’s sieve, seem plain and insignificant. However, charcoal remains, and their analysis, are the keys to unlocking the wood fuel economy.

Factors Affecting the Wood Supply

Physical Factors Figure 1 depicts a diagrammatic representation of the factors that affect the wood supply (and consequently the charcoals we might find in an archaeological setting). At the broadest level, landscape characteristics (geology/ soils, topography), and climate are the first determinants of which trees may grow in a location. These are moderated in the longue durée by human agency, which can improve or damage the broader environment, making it more, or less, hospitable for trees. Land clearance for agriculture in particular can lead to soil erosion. Over use of water for irrigation leads to aridification of an environment ultimately changing the types of trees that can survive. The interaction between man and the environment, especially in relation to the wood fuel supply, is both iterative and complex. 42 robyn veal

Woodland Use and Management Woodland use and management also of course altered the landscape in various ways. In the historical sources, dramatic examples of state or imperial command for a forest to be cut for one purpose or another12 suggest whole-scale deforestation might have been occurring in parts of the Roman period, but except for specific localized instances, the pollen record shows no large scale deforestation until the Mediaeval period.13 Pollen studies are our main source of scientific information about forest cover. They contribute to knowledge of the palaeo-environment through evaluation of archaeological soil samples from excavations (i.e. quite local data), and/or longer records taken from bogs, ponds, lakes or other damp environments which provide a more regional signature.14 A renewed interest in pollen theory and method has meant that the reliability of pollen studies for landscape reconstruction is now being re-evaluated.15 Scale is an issue in examining pollen evidence. Views of short term historical periods (e.g. the Roman Imperial period) are difficult to find in long scale pollen studies (e.g. the Holocene—from 12,000 years ago until today). Also, we know broadly that continual reduction of forest cover occurred from the beginning of agriculture, and accelerated with metal smelting and smithing, especially of iron, but this is not always obvious in a pollen diagram. The more commonplace management of woodlands belonging to vici being carefully husbanded for generations rarely attracts comment in the

12 For example the building of the Roman navy to confront the Carthaginians in the First Punic War would have taken considerable resources (Polybius, Hist. I.20). 13 ‘Deforestation’ requires definition, see W.V. Harris, this volume, and also Harris 2011a, 108–109. Harris moves to a more nuanced view on a regional basis. Grove and Rackham 2001, Chs 10–11 provide an overview of the history of forests in the Mediterranean and the factors which led to changed forest cover over time. Meiggs 1982, 377, suggests real deforestation could not have occurred until railways were built in the 19th c. in Italy. This is rather later than the pollen evidence now suggests. For pollen, for example in south-east Italy, see Russo Ermolli and di Pasquale 2002. Modern Europe is now afforesting at a slow rate (while Asia and South America are losing forest cover). 14 Dincauze 2000, 377–380. 15 Giesecke et al. 2010 overview research being done to evaluate inter-annual variability of pollen fall, the effect of weather, and adequacy of sampling methods, among other variables. The results suggest that absence of a pollen type from a diagram does not necessarily mean absence in the environment, and concentrations of just one pollen type can vary wildly, even when multiple samples are carried out in the same area. For an accurate regional (i.e. Mediterranean-wide) view then, we must carry out many pollen studies over time and space and compare these. The European Pollen Database (available at www .europeanpollendatabase.net), was established in 2007 just for this purpose but not all studies completed are yet recorded here. fuelling ancient mediterranean cities 43 historical sources, as this type of sustainable management was the norm. Local communities had to preserve their woodlands for fuel or potentially perish. Imperial consumption might have been, from time to time, on a large scale, but as Grove and Rackham point out, forests can grow back, and will do so providing soils remain, however species patterns and diversity may change as colonisers (i.e. those tree types which are stronger competitors) move in. Over-use of forests depends not just on the presence of the normally suspect heavy fuel use industries such as smelting, but more particularly on a range of factors including the carrying capacity of an area, and the woodland management practices.16 It should be noted however, that deciduous forests usually recover relatively spontaneously, while conifer forests may not recover without planting seed. In areas of marginal environment, overexploitation has always been a problem for the survival of broadleaf deciduous trees, which were the most desirable and among the most commonly used types used as fuel in the Mediterranean.17 Such marginal areas include those with poor or thin soils, those with low rainfall, or both conditions. Areas, especially islands, which were in early history verdant with forest, became marginal or deforested more quickly if they were commandeered for special purposes (such as smelting). Some areas were never heavily wooded (much of Greece, for example), but occasionally scholars assume the lack of woodland means the environment has been denuded.18 Taken together, the ecological and socio-political factors, which we may call ‘macro’ factors, likely influenced the wood fuel supply (and the resultant archaeological charcoal remains) in some places more than others. But in considering a strategy for examining the fuel supply in any one part of the Mediterranean, consideration of the macro factors helps to frame the way in which the work may begin: the type(s) of fuel that might be expected, as indicated by the physical environment (and thus what reference materials are needed); the historical evidence that exists for woodland use and management (and political control), and the types of archaeological approach required to find the fuel remains. Demographics and fuel consumption levels are key to subsequent quantitative modeling.19

16 Grove and Rackham 2001, Ch. 11. 17 The low growing heat tolerant shrubs of the macchia being the other type. We don’t know in what portions (yet) these two basic woodland types provided fuel for the Mediter- ranean; nor indeed, the proportions used for alternative fuels. 18 Greece’s woodland cover, both past and present is discussed by Rackham 1982. 19 In some areas, wood fuel remains are not the only ones we should be seeking, in particular in Africa. 44 robyn veal

Local Silvicultural Practices On a more local scale, silvicultural practices further modified woodlands. The particulars of Greek and Roman forest management strategies in Greece and Italy can only be broadly inferred from the ancient sources. What happened in the more distant parts of the Graeco-Roman world is even more obscure.20 Building and military requirements were the highest priorities, and for these purposes the state owned or commandeered resources during the Roman period. The indigenous forerunners to the Romans, and the Romans themselves, had a sophisticated knowledge of cropping strategies for the various fuel products required, and of timber performance for various uses. Coppicing (cutting of standard sized round-wood at regular time intervals at ground level), and pollarding (cutting above animal grazing height, i.e. about 2–3m), were well-known strategies. Cutting cycles are attested in the Roman period as ranging from 5 to 7 years, for chestnut, and from 7 to 10 years for oak,21 though the charcoal evidence varies greatly on this point. Both methods of harvesting wood increased a tree’s life. Other methods of harvesting were known.22 Because the physical requirements for different tree types have not changed in time,23 we can use modern scientific studies on forest management to assist our interpretation of the ancient data. We now know, for example, at what altitude in a given climate a particular tree’s photosynthesis is maximized (so we can scientifically infer a tree’s favored growing conditions).24

Cultural Uses of Wood and Wood Charcoal Fuel

In considering how much and what type of fuel might be employed in an ancient city or town, the range of cultural activities employing fuel need

20 A detailed account of the ancient writers’ advice on raising various tree types for fuel is provided in Veal 2009, I, 17–24, to be published in Veal (forthcoming). 21 Columella, De Re Rustica 4.33.1; Pliny, NH 17.147. 22 Grove and Rackham 2001, 48 discuss different methods of producing small, so-called ‘round-wood’ (i.e. of c. 10–30mm diameter). In the archaeological record, small branches produced by different cropping methods are indistinguishable from each other, and in fact, from cropping of whole trees, where the larger sized branches and trunk have been sorted and removed for other purposes (i.e. building). See also Visser 2010. 23 See, for example, Scagel et al. 1969, 2–6. 24 In modern Italy, beech (Fagus sylvatica) photosynthesizes optimally between 1200– 1500m when light and water are sufficient (Pignatti 1997, 487–490), although the tree’s range is 0–2,000m in all of Italy. Lack of water is the factor that severely hampers photosynthesis. fuelling ancient mediterranean cities 45 to be reviewed, together with other aspects of cultural influence, and the practicalities of making and sustaining a fire. Historical evidence, when available, is a starting point, but the analysis of archaeologically collected fuel-remains from firmly identified cultural contexts provides proof (if not always conclusive proof). Processes that in general need only a regular fire temperature in (more or less) an open fire, with the possibility of varying the temperature, e.g. general cooking, fulling, and tanning, more often used wood. Processes which required higher temperatures, and which obtained these by means of containing the heat in a kiln or oven type structure usually also used wood: ceramics manufacture (pottery, brick and tile), lime-slaking, glass-making. Roman bathing falls somewhere between the two, in that hypocaust systems were designed to trap and conduct heat efficiently, even though the heat required was only for heating water, as well as heating the ambient air in the rooms. On the other side of the scale, processes requiring high sustained temperatures of c. 1100°C or more, required the use of charcoal. The industry consuming the most charcoal fuel was iron-smelting (and smithing, although temperatures required are lower for smithing).25 Many of the cultural activities named above as using wood, could con- ceivably have used charcoal as well as wood (and also fire accelerants such as olive oil, and/or the alternative fuels already mentioned). Most fires would have been started with kindling (if embers were not retained). Addition- ally to be considered are public or private rituals, especially cremation, and the heating of rooms in closed braziers. Lower temperature metal-smelting and smithing activities (for copper, bronze, silver and gold), could conceiv- ably have used wood, but charcoal’s cleaner burning properties (and more constant temperature) suggest its likely dominance in any metal processing, since obtaining as pure a product as possible would have been easier with charcoal fuel. Other considerations might include the possible avoidance or inclusion of some wood types, e.g. for religious reasons, or because of the odour or toxicity of smoke.26

25 Sim 1998, 7. 26 Many of the woods of the Rosaceae family have a pleasant perfume/smoke when burnt. Other uses of some woods would have limited the amount available for burning: e.g. laurel (Laurus nobilis) was used extensively in wreaths, for celebrations, medicinally and in cooking. Cuttings for these purposes would keep any laurel tree naturally trimmed, with little waste available for burning. Some woods have a noxious smoke: e.g. oleander (Ner- ium oleander). Other examples of fragrant woods identified by the author in association with burials or ritual include Cistus sp., juniper (Juniperis spp.), and myrtle (Myrtus commu- nis). 46 robyn veal

Fig. 2. Examples of charcoals from excavation—i) a typical assemblage from one context containing dust and charcoals of mixed sizes; ii) a partial small branch: heart and bark are visible, and even ancient cropping marks. Photos: the author; and Jennifer Stephens, respectively. fuelling ancient mediterranean cities 47

Which Fuel: Raw Wood or Manufactured Charcoal?

Throughout this discussion, reference has been made to both (raw) wood fuel, and wood charcoal fuel. In the archaeological record, these appear to be the same thing, but discrimination between the two will ultimately be essential. Figure 2 shows some examples of archaeological charcoal, including a typical assemblage from one context of study in Pompeii, and an example of a partial small branch. There is no accepted test currently available to discriminate between archaeological charcoal that is the result of burning raw wood, and archaeological charcoal that arises from wood converted into charcoal that has been ‘re-burnt,’ although a new method called ‘reflectance’ is under study (discussed in further detail later).27 Peoples of the ancient Mediterranean used both types of fuel. Charcoal was made in charcoal heaps or pits of suit- ably cut or collected logs and branches, which were then covered in soil and leaves or cereal residue and sealed with a mixture of ash and soil (to exclude oxygen). The wood heap was then slowly ‘charcoalified’ rather than burnt. Modern ethnographic studies28 help to illuminate this process, which is also documented in the ancient sources.29 Charcoal making typically occurred in the forest where woods were sourced, and charcoals were then bagged up and transported to private homes or markets nearby. Figure 3 shows a modern charcoal heap that has been prepared (by careful stacking of branches of equal length on a flattened mound of charred residue). Charcoal making ‘piazze’ had to be located near water so the charcoal maker could dampen down the mound if necessary to prevent its full combustion (or indeed acci- dental burning of the surrounding forest). Naturally, the extensive use of charcoal (as opposed to raw wood) in a society has a much greater impact on that society’s woodland resources. To gain a full understanding of forest consumption for fuel, we need therefore to understand how much charcoal was being consumed, and also, how efficient a conversion process was in place for making charcoal from wood. This is one aspect of the fuel economy for which archaeological science and ecological modeling together may provide some answers. Ethnographic data suggest that conversion from raw wood to charcoal can be as efficient as 3 or 4 portions of wood to make one portion of charcoal, to a ratio of

27 Work has been undertaken by a number of authors, only a few of which are listed here: McParland et al. 2009; Scott 2005; McParland et al. 2010; Scott and Veal 2010. 28 Veal 2009, I, 142. 29 See for example Theophrastus, Hist. Pl. 5.9.2–4. 48 robyn veal

Fig. 3. Modern charcoal stack ready for covering with mud, leaves and charring residues, which was then set to char by insertion of a burning log. The charcoal maker intermittently checks progress by inserting a thin stick progressively through the mound: white gases indicate water vapor escaping; blue indicate emission of organic volatile gases. After some time no gas or vapor is emitted and the charcoal maker knows the process is finished. Charring of a stack this size takes about one week. (Photo: the author, Borgo Pace, Le Marche). as much as 10 or even 15:1.30 Some of these data arise from tropical and/or very dry climates so their relevance to the Mediterranean needs careful interpolation.

Charcoal Analysis

Charcoal is analyzed by identifying charred wood cellular structures under reflective light microscopy.31 The structures closely resemble those of wood in its natural state and so identification is sometimes possible to species

30 For a full review see Veal 2009, I, 142–146. 31 Leney and Casteel 1975. fuelling ancient mediterranean cities 49 level. Where this is not possible, identification is made to the most refined level possible (genus, sub-family, or family level). Knowledge of regional vegetation from pollen and phytolith studies and from flora guides can also contribute. Identification is made by comparison with wood atlases,32 and modern reference charcoal.33 Charcoals results are computed either by fragment count (per wood type), or by weight. Both methods have drawbacks: charcoal keeps on frag- menting post-excavation with any type of handling, while weights can be biased substantially by the amount and variety of inter-cellular mineral inclusions, and any attached soils (which often far outweigh the charcoal itself). The process of recovery by flotation34 helps avoid the problem of weight bias, but it also tends to fragment the charcoal further, increasing laboratory analysis time for little return, and occasionally resulting in the loss of smaller wood types, although this rarely affects major trends. Frag- ment count is now the mode most often used, as it is the most convenient and efficient. Methodology in charcoal analysis has been fairly strongly influenced by pre-historic archaeological method, where flotation has led to collection and identification of charcoal fragments over 2mm.35 Hand collected charcoals and those collected over dry sieves of 4–5mm are also examined. In urban environments, city-wide syntheses, other than that for Pompeii (discussed below), are rare.

32 The IAWA handbooks for hardwood and softwood identification (Wheeler et al. 1989; Richter et al. 2004), are used as the basis of nomenclature for recognizing the possible macro and microscopic wood structures. Schweingrüber’s 1990 atlas is the main standard for European charcoal identification. Many other publications relating to localized flora have been used. An online database that is being constantly augmented is likely to become the flagship point of first reference: Schoch et al. 2004. 33 Difficulties can arise if the modern environment is substantially different (i.e. lacking the ancient wood types). The use of reference collections maintained in the northern part of Europe to identify Mediterranean assemblages is also an issue. Climate affects growth and cellular structural habit. Collection of locally curated specimens (certified by a herbarium) is highly desirable, and from differing parts of a tree, as well as (ideally) from different altitudes. 34 This is a method of archaeological recovery where a fixed amount of soil is washed over a fine mesh and the floating material (in the ‘light’ fraction) is recovered separately from the sinking material (in the ‘heavy’ fraction). This method complements regular dry sieving of soils and/or hand collection during excavation, and allows collection of small and fragile materials such as seeds, charcoal and small bones. 35 Asouti 2004a provides a good overview of the history of the discipline. Methodology continues to be refined, see for example, Asouti 2004b and more recently, Théry-Parisot et al. 2010. 50 robyn veal

Emerging Techniques in Charcoal Analysis

Charcoal analysis is still a young discipline. Charcoal remains are not always collected by excavators, sometimes for reasons of cost, and sometimes because its usefulness is not fully appreciated. Even when collected it may reside in storage and be the last artifact type to be analysed, with seeds and other macro-remains being privileged over charcoal. However, in the last twenty years or so, significant theoretical and experimental work has been carried out. Much of this work has related to fragmentation rates and representativeness of archaeological charcoals, and has provided a stronger base from which we may now be more confident about the reliability of charcoal reports. The discipline now needs to move forward with the use of newer scientific techniques to characterize the quality of fuels employed, and the relative ratios of wood to charcoal fuel. An overview of progress and emerging tools follows.

General Studies on Fragmentation Rates and Representativeness of Charcoal Assemblages Aspects of fragmentation patterns, and the representativeness of collected archaeological assemblages have been under discussion for some time, and are now being specifically tested. Whereas in the past focus has been on collection of charcoals sized in classes from 1–2mm, 2–4mm and above 4mm, recent work has begun to demonstrate that in the very small classes (i.e. 1–2mm) oak can be especially over-represented (in the European flora).36 This work is paving way for greater confidence in examining the >4mm sized charcoals, i.e. those which may be collected by dry sieving (and not only flotation). Larger charcoal fragments are faster and easier to fracture for identification and also generally provide a greater level of security of identification, although much is possible even with 2mm fragments. Other work suggests the required number of fragments per archaeological level is far fewer than had previously been the accepted practice, although the partic- ularities of any one site have always to be considered.37

36 Chrzavzez et al. 2011. 37 Typically a quantity of 350–450 fragments ‘per level’ has been recommended (Chabal et al. 1999, 66), based on extensive work in pre-historic excavations. A ‘level’ is a notion that requires definition according to the type of archaeological contextual environment. Charcoal work in the UK and Ireland is simplified by lower floristic diversity by comparison with the Mediterranean. Even so, recent work demonstrates that fewer fragments (60–80) may often be sufficient to ascertain major trends (Veal 2009, I, 87; O’Carroll and Mitchell 2011; Py 2006, 41). These developments mean charcoal analysis is becoming a more cost-effective operation. fuelling ancient mediterranean cities 51

Sampling of sites for charcoal is regularly limited to hearths or kilns and other areas of concentrated charcoal deposits. However this strategy merely provides a view of the last (or last few) burn episodes(s), and often means a restricted number of wood types are identified. Random sampling and collection over all context types, will, in the long run, provide a greater view of wood diversity (and thus the breadth of the fuel economy).38 Collection of soils and analysis from ‘off-site’ areas may complement the archaeological site charcoals, providing data to assist vegetation reconstruction (since they will more likely reflect ambient woodland). ‘On-site’ urban charcoals will reflect those woods from the environment specifically selected for fuel.39 Sometimes the two goals of environmental reconstruction and analysis of the fuel economy are improperly conflated. Rarely can both goals be met by the same dataset because: (1) environmental reconstruction from wood fragment identification (only) is incomplete (non woody plant forms are rarely preserved); and (2) selection of wood for different economic uses cannot be ascertained if the archaeological traces of different wood consuming activities are not observable in the archaeological record. In pre-historic studies, the process of human selection of woods for burning cannot necessarily be considered to have ranged across all woods available in the environment (as has often been traditionally accepted).40 In urban environments, on the other hand, with specialized industries and various options for transport, more selective use of woods for different purposes appears probable. In some cases we have sufficient data to confirm such trends but few studies are yet able to attempt the level of integration required of the archaeological record with the charcoal results to provide regular proof.

Tree Ring Counting and Curvature ‘Dendroanthracology’ or the science of tree ring analysis specifically in the service of charcoal research has arisen as the new standard for estimation of tree cropping indicators. Methods for the measurement of fragment diameters (which permit us to infer the raw wood’s metrics) include (1) placement of a charcoal fragment on the familiar ceramics rim and base measurement template, and (2) estimation of fragment size and diameter using

38 Py 2006, 41. 39 They may also reflect timbers chosen for construction (or furniture), if a holistic burn event is identified, which of course, muddies interpretation as far as fuel analysis goes. 40 Gelabert et al. 2011 provides a detailed review of the principle of least effort (PLE) for firewood collection by pre-historic peoples and offers evidence for the need for a much more nuanced approach. 52 robyn veal trigonometry, both manually and digitally.41 The former method is fast and economical, but potentially more error-prone, the latter is more expensive and time-consuming but less error-prone. Having estimated diameters of charcoals (and having allowed for their c. 15% shrinkage from raw wood), one can construct a picture of the sizes of the branches used and of cropping strategies. We can complement this method by estimating ring curvature, which falls broadly into three groups: flat (indicating use of mature large branches or trunk wood); moderately curved (indicating use of medium- sized branches); and highly curved (indicating use of small branches or twigs).42 Not all fragments in an archaeological assemblage may lend themselves to this type of analysis (they may be too small), but major trends are usually discernible. Observation of repeated use of small woods (in conjunction with a robust presence of the same type of wood) is usually interpreted as being indicative of sustainable forest management. Use of larger scale woods, especially trunk wood, indicates exploitation at a higher opportunity cost (since whole trees, or even very large branches, take longer to regrow than small wood). This may, or may not be a less sustainable practice, depending on the cropping cycle and time allowed for regrowth (which cannot always be inferred from the charcoal). Reference to nearby pollen studies (which show large scale changes in woodland makeup), and/or large- scale changes in wood types in the charcoal over time (e.g. from large deciduous trees to smaller colonizing or macchia types), together may be the markers of over-exploitation. There are many other types of observation that allow inferences to be made about whether or not freshly cut wood, or older, aged wood has been used, about stresses that the growing tree may have been under, about fungal or insect infestations, and about aspects of the combustion process.43 However, these usually relate to occasional observations about individual fragments, and can rarely provide statistically valid information.

Heat Values of Different Woods The calorific potential (or heat value) of different woods is a useful tool for examining wood utility, and also for comparing the heat potentials of alternative fuel types. Heat value is not a fixed value for any one wood type since it will vary with moisture conditions, size and shape of faggots burnt, and

41 Théry-Parisot et al. 2010, 145. Marguerie 2011 has more recently tried to qualify and quantify limits to dendro-anthracological methods. 42 Marguerie and Hunot 2007. 43 Carriòn 2006, 86. fuelling ancient mediterranean cities 53 other ambient factors in combustion.44 However, specific gravity (at a specific moisture content) is a rough proxy for expected heat values.45 Average specific gravities for dry wood, for most of the common broadleaf trees of the European flora, range from around 0.5 (for riparian wood types, i.e. water- loving species like alder (Alnus spp.)), to about 0.75 for beech (Fagus sylvatica) and oak (Quercus spp.).46 The implication is that if inferior heat-value woods are consumed, then a greater volume will be required to achieve the same temperature in a fire or kiln. The naturally higher volume of water in a riparian wood must first be driven off (and this consumes calories). Consumed as charcoal, fuel heat outputs of different woods become more similar, and charcoal in general provides about double the calorific potential (by weight) of wood. However, in the process of charcoalification of wood, the water content is removed (which constitutes part of the weight of the wood), and so for the riparian types, again, a greater weight or volume of wood is required to make an equivalent weight of charcoal than for a denser wood. Only potential calorific value can be considered, as actual heat value is dependent on a range of different factors, and the calorific values reached in a fire never achieve their theoretical maxima.47 The matter is further complicated by the presence of resins in some woods. These tend to be highly combustible, and tend to increase heat potential. Also, so-called lower heat potential woods (often called ‘soft woods’) may be advantageous if they ignite and burn quickly (for example, in the process of initiating a fire).

Measuring the Absolute Burn Temperature of Charcoals: the Reflectance Technique The reflectance technique is a method borrowed from coal studies that allows the estimation of the absolute temperature to which a charcoal fragment has been subjected. Experimentation by L. McParland on various common (modern) wood types charred at various temperatures and for various times has proved the linear relationship between reflectance and burn temperature, although the effects of taphonomy48 and the usefulness

44 Lyons et al. 1985. 45 Bootle 2004, 202. 46 Macchia vegetation types, which by form are scrubbier and denser, usually have quite high heat values. However, they can be more difficult to crop and manage (being hard to cut and sometimes thorny). 47 Lyons et al. 1985. 48 Taphonomy is the archaeologists’ term to describe every type of process that happens to an artifact after deposition, whether environmental or anthropogenic. See a recent study 54 robyn veal of this method for archaeological charcoal are just now being tested.49 It may provide a complement to other temperature estimation methods, and it is already revealing more about the taphonomy of archaeological charcoal in various contexts.50 Reflectance can also provide an indication as to whether raw wood, or charcoal fuel, has been employed. Measurements of wood fires will tend to exhibit low reflectance of a range of temperatures (often on a Bell curve) of c. 100°C to c. 450–500°C, while charcoal-only fires will show temperatures starting from c. 350°C upwards (the approximate temperature at which charcoal starts to form) and will lack low temperature readings. Of course, the matter is complicated by the fact that both fuel types may have been used in a fire. Studies are continuing on a range of archaeological charcoals associated with particular cultural uses.

Charcoal Quality Analysis: Measuring ‘Purity’ of Charcoal Modern fuel economists spend much time on estimating wood charcoal consumption in developing countries where wood is still a major fuel. In these economies, different types of charcoal are produced for different purposes. In particular charcoal destined for domestic use (manufactured at a seemingly ‘good’ conversion ratio) is only lightly charred in order to leave in some of the organic volatiles (so combustion later on is not too difficult); while charcoal destined for industrial use (especially iron smelting) is heavily charred in production in order to produce as high a quality product as possible. The difference between the two types may be measured by their absolute carbon content.51 For domestic use, charcoal of about 65% carbon is desirable, while for industrial use, charcoal of about 80–85% carbon will provide a hotter fire. The charring process in the latter instance is longer, and

(Théry-Parisot et al. 2010) which attempts, confusingly, to redefine taphonomy as all processes associated with the chain of events that leads to an archaeological assemblage, including human selection of the wood. 49 McParland et al. 2009. More recently the method has been used to test cremated bones with less success (Veal et al. 2011) and iron smelting charcoal, with good success (these results will be reported shortly). 50 Taphonomic effects in charcoal typically increase fragmentation, or even reduce it to dust, thus rendering it difficult to separate from the soil matrix, let alone identify it as to wood type. Early trends are suggesting that, especially for charcoal remains associated with industrial processes, the charcoal collected may well sometimes only be unconsumed charcoal fuel, and will reflect the temperature of formation of the charcoal (in charcoal production), rather than that of the industrial process under study. 51 A further difference is that woods charred for a lower period will produce some smoke (when the remaining organic materials are burnt), while highly charred woods will produce very little. Both produce far less smoke than raw wood. fuelling ancient mediterranean cities 55 the conversion rate seemingly ‘poorer’,i.e. a lower volume of charcoal is produced using the same volume of wood, but the product has a higher calorific potential.52 It is likely the same differentiation occurred in the ancient world, and work has begun on charcoals identified from industrial and domestic environments in Pompeii to test this hypothesis.

Pompeii As a Case Study

The author has completed a city-wide synthesis of Pompeii’s fuel supply from the third century bc to ad79, which is published elsewhere in summary form and will be published in detail shortly.53 It is useful to review here how this study fits the model described above. Pompeii was a coastal site on fertile soils, with a climate conducive to the growth of a large range of broad-leaf deciduous and evergreen trees in the city, in the plain and in the nearby hinterland. Baths, bread baking, and iron-smithing, to name only a few activities, consumed fuel on a large scale. Domestic consumption at the elite level would likely have included both wood and especially charcoal, for cooking and for heating braziers, among other purposes. Charcoal fuel’s capacity to burn at a higher and more consistent temperature, with less or no smoke, was likely useful both in the kitchen and the triclinium. Recipes from Apicius suggest both fuels were used in the kitchen,54 as do the ranges of ceramics and stove and oven types found in the city. In the study about 4,000 charcoal fragments were examined. These were dated by known relative typologies for ceramics and coins from the third century bc to ad79, and

52 Schenkel et al. 1998. 53 A preliminary investigation of the House of the Vestals (Veal and Thompson 2008), provides an overview of the approach; my doctoral thesis (Veal 2009) will appear shortly in revised form (Veal forthcoming). A summary of some of the major trends appears in Veal 2012. 54 Apicius (cf. the Grocock and Grainger 2006 edition) provides a large number of verbs to describe cooking methods including baking (in the oven, or possibly in special closed baking dishes on the stovetop called clibani); boiling, frying, steaming, and smoking. Most recipes describe cooking in a pot over a fire, and in these instances we can’t tell precisely if a wood or a charcoal fire is intended. Some give us more information, for example, his recipe for Lucanian sausages (II.iv) requires them to be hung up ad fumum. Smoking requires fresh (usually fruit) wood, for appropriate flavours to be imparted. But his recipe for pisa farsilis (V.iii.2), a complicated recipe for peas layered with cooked meats, is finally finished in a baking dish over a ‘slow fire’ (lento igni imponis). We infer this to be charcoal, as raw wood smoke would be very undesirable in an oven, or even over a closed baking dish on the open stove. This is discussed further by Renfrew 2004, 23. 56 robyn veal

Fig. 4. Summary results of diachronic study of wood fuel of Pompeii c. third c. bc to ad79.55 they were excavated from four separate locations by three different excavation teams. The excavations were carefully documented, and recorded by these teams using substantially similar methods of recovery. Detailed examination of the associated archaeological records allowed close integration of charcoal analysis with the archaeology permitting a very nuanced view of the results on a century-by-century basis. The architectural history of these sites suggests that as many as eleven different property owners, from a range of socio-economic classes, may have been responsible for this material. The results showed a clear use of one dominant fuel, beech (Fagus sylvatica), through the whole period, with secondary use of oaks (Quercus spp.), maples (Acer spp.) and hornbeams (Carpinus spp.), and a smattering of conifer, riparian and other types of woods. The beech constituted 50–75% of the assemblage depending on time period and location, and it diminished over time to be replaced in part by the secondary woods, but also by some fruit and nut woods (e.g. Prunus spp., Castanea sativa, and Vitis vinifera). Fruit and nut woods were observed in small, but then increasing amounts in the 1st century bc and 1st century ad, i.e. from about the time of Roman colonization. A summary graph is provided in figure 4. Work is ongoing in two

55 From Veal 2012. fuelling ancient mediterranean cities 57 further locations in the city, and the fuel economy trends appear to be the same, and so, now, with confidence, we may speak of the city’s fuel economy on a holistic basis. The primary analysis of the charcoal data led to economic modeling of the quantities of wood and charcoal fuel consumed in the city, using a limited range of population figures, over various relative proportions of fuel types, and ranges of charcoal making efficiency ratings.56 Factoring in the preferred growth niches for the wood types has allowed for basic assessment of forest growth areas on mountain and plain required to support the fuel quantities inferred. Subsequent analysis also explored cost/benefit possibilities for different types of transport (road, river, sea).57 A picture of a large and complex fuel supply system emerged where the city was greatly dependent on its somewhat distant, (i.e. 15–25km) hinterland as the major provider of its fuel. The social and economic structures that supported and sustained the wood fuel economy can in part be inferred. Work is ongoing to refine this model through (especially) estimating the ratio of wood to charcoal use (through reflectance). Tests to compare absolute carbon content are also in train, both on the archaeological charcoals, and some obtained from a con- trolled modern charring experiment.

Conclusions

With respect to Mediterranean studies, W.V. Harris has called for a ‘natural history, articulated through periods’,58 and he names fuel as one of the three physical necessities that specifically require study. It is to be hoped that this discussion points the way forward. Recent scholarship on energy consumption in the ancient Mediterranean has examined data holistically, using a top down approach (cf. Malanima, this volume). Wood fuel, con- stituting perhaps 50% or more of the total energy that fuelled the ancient economy, has been broadly examined using assumptions about ‘average wood consumption’ and ‘average calorific return’ from wood. This top down approach has, at least, shed light on the importance of fuel. These averages were naturally moderated by considerations of temperatures required, extent of industrial activity, and other cultural concerns, but at present we have not yet qualified these factors in detail.

56 The base economic model was presented at the conference from which this paper arises and may be examined in Veal 2012. This is a base model, and is currently being refined. 57 The economics of the transport options will be discussed in Veal forthcoming. 58 Harris 2005b, 12. 58 robyn veal

Consumption of wood and wood charcoal is often discussed in terms of metal smelting and smithing, and the fuelling of baths, but many other industrial activities used wood and charcoal. Some were great consumers, such as lime-slaking, glass-making, and tile- and brick-making. A closer examination of the different fuel consuming processes is needed, together with a much closer understanding of the ratio of raw wood to charcoal consumed, and the productivity achieved in charcoal making, for different cultural purposes. Charcoal identification, together with the developing associated scientific studies, are the key to revealing the finer details about raw wood and wood charcoal fuel consumption. The variability in the supply system, in terms of quality of fuel and charcoal, was potentially much greater than the averages we have been able to consider to date. Pompeii offers the unique example of a well-documented city that has provided a base for moving forward to analyze other city’s fuel supplies, but the differing geological, climatic, political and cultural frameworks of settlements around the Mediterranean need to be analyzed in order to accurately examine their fuel supplies. Charcoal offers the potential to view ancient forest husbandry. It provides a ‘bottom up’ view of the actual fuel consumption patterns of a city or town. With time, and future studies, the potential to provide a much more nuanced view of Mediterranean-wide practice may be realized. PART TWO

CLIMATE

WHAT CLIMATE SCIENCE, AUSONIUS, NILE FLOODS, RYE, AND THATCH TELL US ABOUT THE ENVIRONMENTAL HISTORY OF THE ROMAN EMPIRE

Michael McCormick

Recent scientific advances are transforming our understanding of how, why, and how fast climate systems change. That understanding is still imper- fect. Yet what has become alarmingly visible today invites us to explore how societies and environments interacted in the past. Modern climate scientists seek signals in natural scientific ‘proxy data’ such as tree rings and ice cores that testify indirectly to past climate conditions. A rich if incomplete memory of climate change also lies buried in the written records and archaeological remains of earlier civilizations. In some real sense, a human being who attests a climate event in a particular place and time constitutes the ultimate proxy testimony to past climates. But because this testimony is subject to all the complexities of human recording, historians and archaeologists must take the lead in the urgent study of this ‘human proxy data’. The human record of western Eurasia is deep, incomparably rich and well-studied; it has deservedly attracted the attention of climate scientists working on the early modern period.1 The medieval European records also offer much to the climate scientist, and a project to assemble and analyze the abundant Arabic evidence for medieval Middle Eastern climate is underway.2 While valuable work has been accomplished in assembling the

1 See in particular ‘Euro-Climhist. A Data-Base on past Weather and Climate in Europe and its Human Dimension’ elaborated at the University of Bern, Abteilung für Wirtschafts-, Sozial- und Umweltgeschichte (WSU), a new version of which is supposed to be available online in Spring 2012: http://www.hist.unibe.ch/content/institut/abteilungen/wsu/index _ger.html. A good overview of the evidence and research to that date for Europe from the Middle Ages forward is Brázdil et al. 2005. See also the next notes. 2 Pioneering studies of the European Middle Ages came from H.H. Lamb and Emmanuel Le Roy Ladurie; see the important summations of their work: Lamb 1995 and Le Roy Ladurie 2004. The most valuable printed collections of climate events for western Europe are Alexan- dre 1987 and Buisman and van Engelen 1995–. Vogt et al. 2011 describe a study from medieval Arabic sources of some 5,000 reports of weather phenomena between 800 and 1500ad, but make no reference to plans to make the data set more broadly available. For two important collections of climate evidence from the late antique and medieval Byzantine records, see the next note. 62 michael mccormick data from late antiquity and the Byzantine Empire, there is so far no extensive and critical corpus of ancient climate evidence.3 To fix that, a small group of scholars at Harvard has begun creating a first geodatabase of written evidence on the Roman and post-Roman climate.4 Yet, by themselves, the written records rarely suffice. As the real possibility of accurate reconstructions of past climates takes shape, we must pick up the challenge of leveraging the independent testimony of the natural ‘archive’ of scientific proxy data by comparing it to the existing historical and archaeological evidence. Of course the different sorts of evidence are immensely complicated. How many climate scientists can scan the meter of Ausonius’ allusive poetry to determine the ambivalent grammatical case of a particular Latin word, and so understand his phrase correctly? How many philologists can assess tree rings’ testimony on precipitation patterns or the

δ18O records from multiple ice cores that testify to temperatures in central Greenland? The solution must be for scientists and scholars to collaborate and to prepare, explain, and share our evidence in ways that are comprehensible and useful to specialists working in very different fields. Only then can the archaeologist and historian consult, for instance, the results of dendroecologists’ analyses, and the dendroecologists use those of the scholars. That is a huge challenge, but it is only the beginning. Once we have identified the historical and natural scientific evidence, it is essential to compare and connect them, an undertaking that presents its own problems. Then comes the most delicate yet crucial task: to identify possible cases of climate change and human response from the combined evidence of the natural scientific, historical, and archaeological records. This paper explores these issues through a few case studies. The first considers the long-standing and intricate debate on the date of a famous Latin poem in the light of new palaeoclimatic data. The implications of new scientific data for a traditional philological problem illustrate the complexities and potential of coordinat- ing the different kinds of evidence. Next, a summary of the present, preliminary state of knowledge about climate conditions across the Roman Empire will introduce some reflections on pressing areas for further investigation as well as pitfalls to avoid when interpreting the data that is already available. Finally, we will explore some potential cases of human response to climate change in the late Roman and early medieval archaeological record. It is almost needless to add that this essay’s case studies are offered in a spirit

3 Climate reports from written sources in the Byzantine Empire, beginning in 300ad, are comprehensively catalogued by Teleles 2004 and Stathakopoulos 2004. 4 McCormick, Harper, et al. 2012. See further below. the environmental history of the roman empire 63 of experiment and exploration. Scientific study of the ancient climate has barely begun; new data arrive every day with the potential to revise and deepen our knowledge on every aspect of the phenomena discussed here.

1. The Poet and the Tree Ring: Dendroecological Dating of Ausonius’ Mosella

In the later fourth century, the Gallic teacher and statesman Ausonius (ca. 310–ca. 395) composed one of the most exquisite poems ever penned about the environment and Roman civilization. Mosella describes the way- farer’s wonder as he rides from the mighty new fortifications along the Rhine across the Hunsrück, the hilly plateau that separates the Rhine from the Moselle River, and reaches the Moselle’s terraced vineyards.5 Water courses throughout the poem. The traveler hears the water-powered saw- mills, whose screaming teeth slice the stone (vv. 361–364) likely intended for splendid buildings rising around the imperial capital at Trier; he muses on the river fishermen and, famously, on their succulent fish (vv. 75–149). The proud Roman breathes finally free in the bright sun that shines across the Roman space, civilized now that he has moved away from the trees that darken the light, not to mention that air fouled by forests (vv. 10–22). So finely detailed is the poem that, imaginary or not, it has proven possible to track Ausonius’ trip along a well-known Roman road. Station by station we can follow him through what the courtier depicts as Roman Gaul’s renewed prosperity.6 Ausonius talks about more than just rivers. He notes water conditions along the way. As he crosses the small settlement known today as Kirch- berg, Ausonius remarks ‘I pass Dumnissus, drought-stricken (arentem) with, all around, its fields parched’: ‘praetereo arentem sitientibus undique ter- ris / Dumnissum.’ Striking juxtapositions pervade Ausonius’ artistry, and he continues here by contrasting dry Dumnissus with the next place he tra- verses: ‘Tabernae, watered by an ever-flowing spring,’ ‘riguasque perenni

5 Ausonius, Mosella, ed. Green 1991, 115–130. 6 Opinions have varied on whether in lines 1–22 Ausonius is describing an actual or an imaginary trip. For instance, Green 1991, 463 (but cf. 451) left open the possibility of an imaginary trip or an actual one retouched, but later Green 1997, 214, reckoned it wiser to assume it ‘bears no relationship to any historical event’; Shanzer 1998b, 228–230, seems inclined to think it a real trip. The new climate evidence suggests that whether reimagined or more or less recorded, Ausonius’ description reflects a specific experience in a specific year. 64 michael mccormick fonte Tabernas.’ In fact, Kirchberg sits on a waterless ridge, and so relied on rainfall for irrigating its fields. Its parched or, literally, ‘thirsty’ fields seem to have been suffering from a drought.7 The modern identity of Tabernae is less certain, but the proposed location at the place known as Heidenpütz, 25km west of Kirchberg, fits the ever-flowing groundwater of the poem.8 The very name ‘Heidenpütz’ seems to me to confirm the identification, for the local dialect word Pütz derives from Latin puteus and designates a deep well from which water could be drawn.9 And ‘Heiden,’ in the sense of ‘pagan’ is a common element of German place names. When the word occurs in toponyms in areas such as the Hunsrück where the Germanic newcomers’ culture took root inside the Roman Empire, it typically refers to structures or ruins inherited from the Romans. So the locals identified this site as ‘the Roman well.’10 For centuries, classicists have sifted through Ausonius’ masterpiece and weighed the potential chronological implications of every allusion. It refers to, and so surely was composed after a victory over the barbarians celebrated by Valentinian I and his son Gratian. This very likely was the campaign of 368 or 369 in which both emperors participated according to Ammianus Mar- cellinus.11 The poem was clearly written before Valentinian I died on Novem- ber 17, 375 since, even if he is not named, this emperor and his doings are

7 Ausonius, Mosella, 5–11, Green 115–116. See Wightman 1970, 130, who describes the geography in the light of the Peutinger Table and reports the waterless character of the site. Kirchberg does indeed sit on a ridge which the Geological Service of the German federal state of the Rhineland-Palatinate maps as the boundary between two groundwater zones or aquifers: see the website Rheinland-Pfalz, Landesamt für Geologie und Bergbau, LGB Hydro- logische Karten http://mapserver.lgb-rlp.de/php_hydro/index.phtml selecting the layer for Grundwasserkörpergrenzen, and setting the search for the Gemeinde of Kirchberg. 8 Wightman 1970, 131, notes that it was still ‘rather marshy’ in 1970. The Rhineland- Palatinate Geological Service mapping site cited in the previous note is less revealing for the two leading possible sites of Tabernae, identified as either Belginum near Wederath or the site about two km southwest of Wederath known as Heidenpütz; cf. Green 1991, 465. 9 Rheinisches Wörterbuch ed. Müller et al. 1928–1971, vol. 6, 1248–1251, s.v., consulted online, November 30, 2011, Universität Trier, http://woerterbuchnetz.de/RhWB/; cf. Bach 1952–1956, vol. 2, 1.287. 10 Bach 1952–1956, vol. 2, 1.356–357. 11 See in general on Ausonius’ career, Jones et al. 1971–1992, vol. 1, 140–141, ‘Ausonius 7,’ and the detailed study of Coşkun 2002a. Ausonius’ summons from Bordeaux to the imperial court and the Moselle region to serve as tutor to the young Augustus Gratian has been placed somewhere in the mid-360s: e.g., Green 1991, xxviii. Coşkun 2002a, 37–40, concludes it could have occurred as early as 364, but happened more likely after August 367 and the summer of 369. Ausonius manifestly wrote Mosella after he had come to the imperial court as tutor: see, e.g., on his expression ‘mea maxima cura’ (v. 450, Green 129), below, note 15. The campaign: Ammianus Marcellinus, Res gestae, 27, 10, 6–16, ed. Seyfarth, 2, 1978, 51–54; cf., e.g., Seeck 1920, 5, 37–38. the environmental history of the roman empire 65 mentioned repeatedly and unambiguously.12 Within this seven-year span, over the last two decades alone, four different dates have been proposed for the poem’s composition. The later fourth century’s rich documentation allows good arguments for all of them. The enduring debate also reflects the finely wrought wording and allusive quality of Ausonius’ poetry. Thus he clearly alludes to the high honors recently awarded to specific individuals, but Ausonius does so in a way that leaves exquisitely ambiguous the exact person or persons to whom he is referring. As his most comprehensive editor has observed, this could well be part of the sophisticated author’s calculation: Ausonius aimed to flatter a particular highly placed minister, but chose to do so in such a way that more than one such official might delight in recognizing himself in the pattern of praise.13 The two earliest datings (368 or a little earlier; late 369 through mid-370), have won little support.14 One or more allusions that might seem inconsistent with an origin in 370–371 have been adduced in favor of the late date, in 375, or possibly, in favor of a slight retouching in 378 or 379 of a poem origi- nally composed at the earlier date.15 The most substantial of these supposed

12 Mosella alludes (vv. 420–426, Green 128) to the triumphs the imperial father and son celebrated together at Trier for the victories achieved beyond the Rhine and the Neckar, witnessed by the personified Moselle; the laurel draped letter of victory had arrived recently, and more are predicted to come soon: ‘… sed Augustae veniens quod moenibus Urbis / spec- tavit iunctos natique patrisque triumphos / hostibus exactis Nicrum super et Lupodunum / et fontem Latiis ignotum annalibus Histri / haec profligati venit modo laurea belli.’ On these victories, see, e.g., Green 1991, 506–508, and below, n. 15. On the litterae laureatae announcing imperial victories to the population in this period, see McCormick 1990, 39–43, and, in general, 190–196. On Valentinian I’s death as an unavoidable terminus ante quem, see, e.g., Mondin 2003, 189–190. 13 Green 1978. 14 Sivan 1990’s early date of 368 has been carefully and comprehensively refuted by Green 1997, Shanzer 1998a, and Shanzer 1998b. Even if one rejects with him the common identification of the mysterious figure alluded to in vv. 409–411 as Petronius Probus, consul in 371, Coşkun 2002b’s learned arguments that Ausonius could have finished the poem by the middle of 370 are suggestive rather than compelling. In favor of the identification with Probus see, e.g., the subtle and persuasive observations of Shanzer 1998b, 216–228. See further for a date in 370–371, Mondin 2003. 15 Drinkwater 1999 argues with some reason for hidden complexities in Ammianus’ account of the military campaigns with which Mosella is usually connected, and therefore concludes for a terminus post quem of 370. He adduces in favor of his late date of 375: Auso- nius’ expectation of a consulate, on which see further below; the success of Valentinian I’s border campaigns implied by the statement that castles are becoming granaries (vv. 456–457, Green 129: ‘addam praesidiis dubiarum condita rerum / sed modo secures non castra sed horrea Belgis’); and Ausonius’ laying claim to tutoring Valentinian II. No specific evidence is adduced for the granaries assertion; lacking that, I see no way of dating such a propagandistic claim early or late within this period. Drinkwater’s assessment is not universally shared that it 66 michael mccormick inconsistencies comes when Ausonius refers to the insignia of the consulate in connection with the imagined future time of his retirement (vv. 448–453, Green 129). These words have been interpreted to mean that he announces here his expectation of receiving the ordinary consulate, the supreme honor of the senatorial class. Not injudiciously, such an announcement has been considered implausibly tactless for an accomplished courtier if the emperor had not already committed to designating Ausonius as ordinary consul. That designation in turn has been viewed as unlikely, given that Ausonius actually held his ordinary consulate only in 379, i.e., some eight or nine years after the possible original composition of the poem ca. 370 or 371. Hence, it is thought, the allusion must have been added to the poem only at the time when Ausonius was sure he was going to be consul, presumably in 378.16 On the other hand, since at least the nineteenth century, some have observed that Ausonius’ words do not explicitly specify the ordinary consulate. The poem mentions only the insignia of consular rank, the fasces and the chair of state, the sella curulis shared by various high-ranking officials, including consuls. Thus Ausonius might be referring only to the consular honors that were routinely conferred on higher ranking officials and which entailed the use of these insignia.17 Such consular honors were prestigious but fell well short of the ordinary consulate, whose holders paid for and led the New Year’s celebrations in the capitals, gave their name to identify that year ever after, and rose to the highest levels of precedence available to a non-emperor.18 In fact, a law of 381 cites precisely these same insignia as sym- bolic of the proconsular rank of the head imperial notary.19

would be farfetched for Ausonius to project himself in 371 as tutor to a yet to be born or new- born imperial child, as would be implied by the plural nati (vv. 448–453, Green 129: ‘ast ego, quanta mihi dederit se vena liquoris, / Burdigalam cum me in patriam nidumque senectae / Augustus pater et nati, mea maxima cura, / … / mittent emeritae post tempora disciplinae, / latius Arctoi praeconia persequar amnis.’) When that distant day comes when Ausonius will be dismissed by the emperor to his ‘retirement nest’ in Bordeaux, he foresees that he will have accomplished his maxima cura of educating the imperial children (cf. Venus in Aen. 1.678 and Ausonius’ own use of the expression in reference to his imperial student Gratian in Cento nuptialis 8, Green 134). Green 1991, 511, finds it natural that Ausonius should expect to teach a new child as well as the older sibling. One could go further and see in this statement a quiet assertion by Ausonius of his ambition to do so. 16 See, e.g., Sivan 1990, Shanzer 1998a and Shanzer 1998b. 17 de la Ville de Mirmont 1889, 130–132; Green 1991, 511; further Mondin 2003, 191–192; Cavarzere 2003, 178, ad locum, who notes nevertheless that Ausonius uses almost exactly the same language about his ordinary consulate: Praefationes, 1, 37–38 (Green, 4): ‘et, prior indep- tus fasces Latiamque curulem, / consul, collega posterior fui.’ See also however below, n. 19. 18 Bagnall et al. 1987. 19 Codex Theodosianus, 6, 10, 3: ‘Notariorum primicerium in numero proconsulum habe- the environmental history of the roman empire 67

But was it in fact so unlikely that Ausonius could have been promised an ordinary consulate seven years ahead of taking the office? When celebrating his own consulate in 379, he seems to refer to Valentinian I’s promise of the appointment; that necessarily occurred before the latter’s death in 375.20 We know relatively little about how far in advance consulates might be decided, but several contemporary instances identify an individual as consul designatus a year in advance. In one extreme case, a future consul, Lollianus, is called ordinarius consul designatus before May 337, when actually he only became consul eighteen years later.21 In fact, the games sponsored by the consuls cost a fortune; holders of the lesser office of praetor were designated ten years in advance to allow them to assemble the wealth necessary to pay for their praetorian games.22 Ausonius might well have appreciated a substantial lead time to prepare to finance his own ordinary consulate, a celebration which, in the sixth century, cost 2000 Roman pounds of gold.23 He was by this time well-to-do, but he had not been born to immense wealth and surely was not close to the front ranks of the wealthiest Roman senators.24 So, whether we take the consular insignia Ausonius imagined in his future as referring to the ordinary consulate or merely to consular honors, those lines need not have been written near 379. Ausonius’ reference to Kirchberg’s parched fields offers an opportunity to consider a moment in the environmental history of the Roman Empire in the light of climate science, and to add a new consideration to the debate about dating the poem. In a small but tangible way, it exemplifies the new evidence from the natural sciences that illuminates the ecology and history of the ancient world. European dendroecologists have recently produced a remarkable research resource, a record of spring and early summer precipitation in western central Europe. Its power derives from the fact that different—potentially competing—laboratories came together and pooled their data of 7284 oak tree ring records collected in northeastern France and northeastern and southeastern Germany.25 With a database of this size and quality, the scientists could subject the growth records to statistical

mus, tamquam comitis ei semper fasces cum curulibus dederimus.’ Constantinople, Decem- ber 13, 381. 20 Gratiarum actio, 22, addressing Gratian about why he might have deserved the consulate, includes ‘seu fideicommissum patris exsolvis;’ Green 149.31. 21 Bagnall et al. 1987, 19–20; Jones et al. 1971–1992, 1, 513, ‘Lollianus 5.’ 22 Bagnall et al. 1987, 18. 23 Procopius, Historia arcana, 16, 13, ed. Haury and Wirth 1963, 159.24–27. 24 Green 1991, xxvi. 25 Büntgen et al. 2011. 68 michael mccormick procedures in order to reduce the ‘noise’ arising from phenomena such as the differing growth patterns of young and old trees, varying sample sizes in different periods, etc.26 Because the annual growth patterns of oak trees in the forests they studied reflect first and foremost how much rain the trees got in April, May and June, the new data base allowed the group to reconstruct the early summer precipitation in these areas for each year from 398bc to 2000ad. The link between spring precipitation and tree ring growth was deduced in the usual way, by comparing with actual ring widths the record of precipitation established since the advent of reliable instrumental recording in these regions, in this case over the period from 1901 to 1980. Some have challenged this kind of extrapolation from modern data, given the dramatic changes the Industrial Revolution has introduced into the modern ecosystem. Such concern proves unfounded here. A study of extreme excess or shortfall in precipitation deduced from the tree ring records between 1000 and 1504ad confirmed the precipitation events from contemporary historical records in 32 out of 34 extreme years.27 The tree rings analyzed in northeastern France come from the region immediately adjacent to the one that Ausonius describes.28 Figure 1 displays the daily early summer precipitation as reconstructed from tree ring growth for the years during which the poem must have been written. The new dendroecological database shows that, in the region immediately adjacent to the Hunsrück, two of the years in play, 371 and 375, were marked by sharp shortfalls of spring and early summer rain, indeed by drought conditions. If the thick web of contemporary allusions that Ausonius has woven into his poem extended to the weather conditions of the journey he depicts, then these two years should lead the candidates for the date of composition, regardless of the other deductions drawn from the poem’s content. Since the strongest arguments have been adduced for composition in the period 370–371, in my eyes the new dendroecological data decisively tip the balance toward 371. Quite unexpectedly, completely independent new data from climate science reinforce arguments derived from the poem’s internal characteristics and Roman political history that the work refers to and likely stems from 371. This convergence of ancient Roman poet and tree rings analyzed by sophisticated scientific study represents a fine example of what E.O. Wilson

26 See Büntgen et al., ‘Supplementary Online Materials.’ 27 Büntgen et al. 2011. 28 See the map of sample locations in Büntgen et al. 2011, Figure 1. the environmental history of the roman empire 69

Fig. 1. Reconstructed precipitation anomalies (mm/day), April, May, June, 367– 378ad, northeast France. Source: Büntgen et al. 2011 with Supplementary Online Materials, 9, and Fig. S4. Data on deposit at NOAA Paleoclimatology: http://www .ncdc.noaa.gov/paleo/pubs/buentgen2011/buentgen2011.html. calls ‘consilience,’ a term coined in the 19th-century. Consilience occurs when evidence of two completely different origins ‘jumps together.’ Epis- temologically distinct, the two different pieces of evidence come together because the reality from which they stem is one. The independent findings of dendroclimatologists and of classicists converge and reinforce one another. Remarkable though this one case is, it is likely only the first of its kind. The promise for future investigations on time scales useful to historians and archaeologists is even more exciting. But beyond that early summer of 371, what light does today’s climate science shed on more general environmental conditions under the Roman Empire?

2. What We Know Today about Climate Conditions under the Roman Empire

A group of climate scientists, archaeologists and historians met at Har- vard University’s Dumbarton Oaks in Washington D.C., to discuss what we can know about climate conditions in the Roman and post-Roman world. The result was a kind of white paper, a first synthesis of conditions across the Roman space as they could be established from the latest available 70 michael mccormick scientific research.29 It used eleven high-resolution (that is, relatively precise in chronological terms) multi-proxy indicators—independent natural ‘archives’ such as tree rings, ice cores, pollen deposits in lake valves, and isotopic signals in speleothems exploited by modern climate science—and our new geodatabase of ancient written evidence about climate events to delineate broad patterns of change and continuity. Here I will only summarize a few salient features that our group detected in three broad phases between ca. 100bc and 600ad. Although some have been claimed before, none have ever been demonstrated with this rigor and relative comprehensiveness. Even so, this is just a provisional balance sheet from an area of scientific inquiry that is now undergoing exponential development. First: an age of stability characterized Rome’s maximum expansion. Ex- ceptionally steady climate conditions prevailed over the territory of the Roman Empire from ca. 100bc to 150ad. Precipitation was fairly even and, in the Levant, wetter than usual. The northwest provinces were exceptionally warm, enjoying conditions perhaps comparable even to those of the 1990s. Nile floods were unusually favorable, as we will see in more detail below. Second: instability followed by partial recovery characterized 200 to 400ad. Stability began to dissipate starting about 150ad. Broader cooling occurred in the western provinces, as glaciers in the Alps ended a couple of centuries of retreat and began growing again. Extreme volcanic eruptions with their risk of rapid climate forcing and volcanic winters and summers became more frequent, peaking ca. 250–290.30 Spring and early summer precipitation declined sharply in northeastern Gaul from about 235 to 310ad. In the Empire’s eastern provinces, precipitation also declined significantly in the third century, to judge from some written evidence and Dead Sea levels. Then, as imperial fortunes recovered in the fourth century, Gaul became wetter and, later in the century, its climate warmed; in the east, wetter conditions returned sometime between 300 and 400ad. Two major climate developments that originated outside the Roman Empire likely worked important and negative consequences inside it. Ex- ceptionally dry climate conditions developed in Central Asia, possibly under the influence of the ENSO (El Niño-Southern Oscillation) phenomenon. These conditions probably encouraged in the later fourth century the migration into Europe of the nomadic federation known as the Huns. The move-

29 McCormick, Büntgen et al. 2012, where the details and substantiating materials will be found. 30 Using the same GRIP-2 data, Rossignol and Durost 2007 have come to similar conclusions on volcanic forcing. the environmental history of the roman empire 71 ment into Europe of course triggered the Gothic crisis on the Danube. This ultimately led to the destruction of the Roman army and emperor at Adrianople that permanently established inside the Empire large groups of Goths who escaped from imperial control. The Nile’s annual floods reflect precipitation outside the empire in eastern Africa. We will return to the details in the next section, but a tabulation based on Danielle Bonneau’s study of Nile flood levels between 261bc and 299ad argues that the early and later Roman Empire experienced different qualities of Nile floods.31 Written records indicate that unusually favorable climate conditions for Egyptian food production prevailed over the first two centuries of the Roman Empire, while the conditions underpinning food production appear to have been consistently less good from 156 to 299ad.32 In a third phase, instability returned between 400 and 600 in ways that, on today’s data, appear to differ between the western and eastern parts of the empire and its margins. In the west, signals are mixed but seem to indicate fluctuating temperatures interspersed with periods of warming and cooling. Two developments may have had particularly negative consequences for agrarian production and the economy. The fifth century may have been nearly as volcanically active as the third, implying considerable potential for disruptive volcanic winters and summers regardless of the overall temperature trend. Secondly, in northeast France and in northeastern and southeastern Germany, early summer precipitation was very wet until about 450. Then it shifted to very dry and continued in this mode for the next two centuries.33 In the eastern Roman Empire, on the other hand, generally humid conditions resumed no later than about 400, as the eastern provinces approached their political and cultural apogee.34 Dead Sea levels testify to a steep increase in Levantine precipitation. That changes dramatically in the sixth

31 Bonneau 1971. 32 So far, Nile flood records are easily available or can be deduced for only eight years out of the next three centuries that the Romans retained Egypt (300–618ad): see McCormick, Harper, et al. 2012 (on which see further, below, n. 44). Reconstructing for these centuries the kind of Nile flood record that has been developed for the earlier period should be an urgent priority. 33 For the precipitation record, see Büntgen et al. 2011. 34 Bookman (Ken-Tor) et al. 2004 differ in their datings from Migowski et al. 2006. The discrepancy in the proposed starting date of the wetter conditions (and therefore of the onset of the subsequent dry conditions) at ca. 300 or ca. 400ad lies well within the range of the two-hundred-year accuracy of the radiocarbon dating of the samples at two standard deviations (see especially Migowski et al. 2006, Appendix A), but both studies concur that drying set in sometime between ca. 500 and 600ad. 72 michael mccormick century when mentions of eastern droughts and heat events equal or exceed precipitation reports in the written sources; Roman water works in Palestine appear inordinately concentrated in the first half of the sixth century. The summer water shortages that affected Constantinople in the 520s could also reflect a decline in precipitation. More decisively, the two centuries of favor- ably wet Levantine conditions documented by Dead Sea levels came to an end at some point in the sixth century and, according to both key studies, dry conditions persisted through the eighth century.35 Finally, and most spectacularly, written sources across the empire document the seriousness of a veiling of solar radiation in 536 and 537 that caused crop failures in different areas.36 The scientific proxy records appear to converge with the historical evidence of the 536 event. Northern tree rings, Greenland sea ice, oxygen isotopes all signal cooling that peaked around 540; summer temperatures dropped in the Alps, and a significant glacier advance is proven in the Swiss Alps for the sixth century.37 Overall, the sixth century also looks cooler in the post-Roman west, and difficult in the late Roman east. In sum, the initial indications suggest that the early centuries of the Roman Empire’s existence occurred against the backdrop of remarkably stable and favorable climate conditions. Around 150ad the stability ebbed from a broad spectrum of climate indicators. Agricultural difficulties could well have attended more fluctuating conditions, even as less favorable temperature and precipitation conditions became more common in some regions of the empire. In the west, the fifth century may have been particularly unset- tled, while the sixth century looks less positive in the east. In a general way, this provisional sketch appears to fit the overall trend of imperial fortunes. But it is only a first delineation at a time when new data and advances in understanding the climate evidence and the mechanisms they imply appear nearly weekly. Even though this description relies on the most comprehensive survey of the available evidence, those data are sparsely scattered across the territory of an empire that sprawled across three continents and climate zones. Further work and more data are urgently needed to correct and improve this initial sketch by adding regional and chronological detail. The appearance of very general correlations between imperial success or difficulties and general climate trends does not yet decide the essential question

35 For more details on these developments, see McCormick, Büntgen, et al. 2012. 36 Gunn 2000; Arjava 2005. 37 For the tree ring evidence, see Larsen et al. 2008; for the rest, McCormick, Büntgen, et al. 2012. the environmental history of the roman empire 73 of cause and effect, of specific types of climate change in specific times and places, and specific human responses. Let us turn next to what work and data we need most and follow with two case studies of what we already have: Nile flood qualities; and archaeological signs of human response to climate change in the empire’s northwestern provinces.

3. Opportunities and Challenges for the Study of the Ancient Climate

What’s next? First we need more, and more refined, data, preferably high resolution in chronological terms, and preferably from the underrepresented imperial heartland: Italy, Spain, North Africa, the southern Balkans, Greece, the Danubian provinces, and Asia Minor. In Italy, a wonderful first step would be for the dendrochronological labs operating there to bring their precious evidence to the quantitative and qualitative levels demanded by international science by working together and pooling their data. That alone will create the preconditions in Italy for the kind of environmental and archaeological breakthroughs that are now occurring in northern European dendroecology. Valuable work has been done on lake deposits, but much more is needed, and that work needs to supply for these centuries very high resolution limnological data in order to be helpful to the environmental history of the Roman Empire.38 Hopefully ongoing high-resolution work from Lake Van and the Dead Sea will yield insight into the detailed climate situation in the Levant.39 If the chronological resolution issues can be addressed, speleothems hold considerable promise. They have already yielded some insights into temperature and precipitation conditions e.g., in Spain, the Alps, Asia Minor, and the Levant, albeit with very uneven coverage for our period; potentially very important work for the Roman Empire is ongoing in Africa’s Atlas Mountains.40 Ice cores are another potential source. The Greenland ice cores GISP2 and GRIP have given precious high-resolution

38 See, e.g., Allen et al. 2002. 39 For an introduction to the various kinds of palaeoclimatic proxy evidence mentioned here, see McCormick, Büntgen, et al. 2012, Appendix. 40 For an overview of the Alpine, Turkish, and Israeli data and further references, see McCormick, Büntgen, et al. 2012; although the published chronological resolution is at present unsuitable for integration into Roman climate history, see for first indications from Spain, Railsback et al. 2011, and for a summary initial report on ongoing work in the Atlas Mountains, Wassenburg et al. 2010. 74 michael mccormick climate proxy data at the local and hemispheric scales. It would be very desirable to obtain similar data from glaciers closer to the Roman Empire, although there are of course precious few peaks over the ca. 4,000m required for permanent ice preservation in mid-latitude glaciers. Nor should we neglect the climate data that still lies concealed in the written sources. The ‘Geodatabase of Historical Evidence on Roman and Post-Roman Cli- mate’ now contains over 700 environmental events documented in written sources from 100bc to 800ad. As our publications have appeared, we have put the ‘Geodatabase’ up on our free, internet-based Digital Atlas of Roman and Medieval Civilizations (http://darmc.harvard.edu), so others can use, add to, and improve it. We anticipate that a considerable number of potential new pieces of evidence will emerge from climate-minded scrutiny of ancient letters, scientific treatises, inscriptions (recording e.g., repairs of flood damage), and the papyri. But data alone do not take us very far. We need understanding too. Cli- mate scientists and specialists of ancient environments face two challenges. The first concerns chronology and dating; the second, understanding. Schol- ars must pay careful attention to the dating methods and their precision in various climate science investigations. For the tasks of historical analysis and explanation, high-resolution data are indispensable. If we really do not know whether a particular climate event or shift began before or after some human phenomenon that may be connected to climate developments, we cannot begin to claim causality. Historians work in years, months and days. Climate scientists, however, often do important work on geological time scales for which even millennia might be a flash in the pan. What is for the historian low resolution data is prevalent in climate science studies, and dangerous for historical reconstructions of ancient climates. On the spectrum of precision, the kinds of proxy evidence climate scientists use range from precise to very imprecise in historical and archaeological terms. Examples of the former are yearly tree rings, absolutely annually dated varve-counted—counting back from the year when the sample was taken, as is possible in rare cases, or from some indubitably dated deposit event—lake deposits, extremely well-preserved ice cores, or unambiguously dendrodated—not radiocarbon dated—glacier movements. Examples of imprecise data are any phenomena which rely on radiocarbon dating, including geomorphological data such as lake levels, soil transport and so on, notwithstanding their potential contribution to multiple debates, such as the controversy over Roman-era soil erosion. These must be considered with caution. the environmental history of the roman empire 75

Scientific methods such as calibrated A.M.S. (accelerated mass spectrometry) radiocarbon dating offer a high level of reliability within the parameters of their accuracy. This entails what is often, for historians, an inherent level of imprecision that is very high. In most cases presently available, the highest confidence level attaches to radiocarbon datings expressed within two standard deviations (the measure of dispersion of individual results from the average result), that is to say, within an error margin that, for our period, runs typically to over a hundred years. A confidence level of 95% for a two standard deviation calibrated AMS dating of 419–591ad, for instance, means that we can be assured that there is a 95% chance that the object sampled dates to any single year within the 172-year period in question. 172 years is not usually a date range that historians will find helpful. A newer method, uranium-thorium dating, is generally credited with greater precision for the calcium carbonate materials that are central to speleothems, but it does not appear to be widely available.41 It is true that under certain circumstances, the radiocarbon date range can sometimes be narrowed considerably, but here we must be wary of the effects of what we might call the economic imprecision that occurs in some studies using scientific dating. Radiocarbon dating is relatively expensive, about $400 or more for a typical sample. I cannot even approximate the cost of uranium-thorium dating. This leads scientists to limit the numbers of samples that are actually dated scientifically, and to arrive at more precise dates by extrapolating precise dates of undated samples taken between dated samples by calculating the physical distance between the dated samples. This works, but yields dates which can be very approximate, for it assumes a regular, uniformly linear growth rate of the deposits that con- stitute the speleothems, bog layer, or lake bottom (in the absence of datable annual or seasonal varves). When one observes a climate phenomenon dated only approximately in these ways, it is sometimes tempting to try to integrate it into the much more precisely dated historical record by assuming a kind of coherence which quickly can lead to circular reasoning and to correlations or even assertions of cause and effect that amount to a house of cards. These problems can be overcome with money, good samples, and good methods, but it is well to be aware of them. We should receive with circumspection the sometimes extravagant claims based on important and interesting data whose chronological resolution will not sustain the claims.

41 This radiometric dating method measures age on the basis of the relationship of the radioactive isotope thorium-230 and its parent isotope uranium-234. As of this writing, I am unaware of any commercial service that will perform this dating. 76 michael mccormick

We need not only to multiply the dating precautions, we need to multiply the types of independent proxy evidence on which we base our climate reconstructions. Multiple independent data sets of proxy evidence—multi- proxy in the jargon—are the best guarantee in these early days of fast developing science that the phenomena we detect are real climate developments rather than artifacts of proxy evidence formation. In climate science as in history, conclusions based on consilience, on multiple, independent lines of evidence, are more robust. As high resolution climate data from multiple sources and places relevant to the Roman Empire’s environmental conditions become more abundant, we shall have to join with our colleagues in climate science to understand the new data qua climate data: how does each emerging piece of the puzzle fit into bigger schemes of global and hemispheric climate systems, change, stability and instability? How does it fit into the regional picture? How does it fit into the micro-regional picture? In coming years we may even hope to define polygons across the Roman Empire whose essential climate features and much of whose weather should share similar features at similar dates. Yet the greatest challenge lies beyond establishing and verifying the accuracy of our understanding of climate mechanisms and proxy data. It comes with the daunting question of climate conditions and human response. Let us turn to two such cases, one based on the historical record of Nile flooding, and one which compares to archaeological materials the new dendroecological record of northwestern European precipitation.

4. Nile Flood Qualities under the Roman Empire: A Closer Look

Environmental conditions that were capable of affecting the Mediterra- nean’s breadbasket must have weighed heavily on the imperial enterprise and the Roman economy. More abundant floods—more land inundated with the Nile’s fertilizing waters—generally mean more abundant harvests. As we noted above, the Nile’s productivity may have undergone a subtle but hitherto unnoticed change in the second century ad. At least that is what emerges from a tabulation based on Danielle Bonneau’s study of Nile flood levels between 261bc and 299ad as evidenced by the papyri, coin issues, and other records.42 Bonneau was able to identify records which allowed her to

42 Bonneau 1971. On Bonneau’s oeuvre, see Bernand 1993. the environmental history of the roman empire 77

Table 1. Nile Floods: overview of broader qualities as classified by Bonneau 1971.

Total docu- Good to Poor to Period mented floods average floods % bad floods % 30bc–155ad 112 72 64.3 40 35.7 156–299ad 87 56 64.4 31 35.6 Source: McCormick, Harper, et al. 2012. classify the quality of floods for 199 of the years from the Roman annexation of Egypt in 30bc down to 299ad. Her treasure trove of data has never been used for environmental history.43 Naturally imperfections are inevitable in any such effort. Nevertheless a simple quantification of her data produces a highly interesting series of observations that invite more intensive investigation and testing by specialists of the intricacies of the Egyptian documentary record and of global climate systems. The data also suggest the extraordinary untapped riches the papyri hold for the environmental history of the ancient world. At first blush, the proportion of positive floods compared to negative ones as assessed by Bonneau looks virtually identical on either side of 155ad (Table 1).44 However, closer scrutiny of specific flood levels suggests a more complex picture. Viewed in this way (Figure 2), the data (Table 2) indicate that under the later Roman Empire a larger proportion (34.5%) of annual floods reached normally productive levels than had been the case (25%) before 156ad. Across the entire period under review, somewhat better than normal floods occurred about once every five floods, that is, with the same frequency before and after 155. The most significant differences concern the best and worst floods, and therefore the best and worst years of agrarian productivity. Between the annexation in 30bc and 155ad, the very best level of floods also occurred about 20% of the time: conditions allowing a superbly

43 McCormick, Büntgen, et al. 2012. 44 For the original data for the Roman period, see: Bonneau 1971, 231–258. She starts her record in 261bc (221). Prof. Kyle Harper (University of Oklahoma), Dr. Alex M. More (Harvard University), and I created a geodatabase of Nile floods from her flood data and, for the period after 299ad, from other materials. Since the studies founded on this material have been published, we have made the Nile geodatabase available for others to use and improve as part of McCormick, Harper, et al. 2012, at McCormick et al. 2010, at http://darmc.harvard.edu. The descriptive values Bonneau assigns to the floods in these years have been numerically coded as follows: ‘M/mauvaise’ (‘bad’): -3; ‘F/faible’ (‘low’): -2; ‘Méd/médiocre’ (‘mediocre’): -1; ‘N/normale’ (‘normal’): 0.06 [this arbitrary value was assigned to make normal floods visible in the graphing functions of Excel 2010]; ‘B/bonne’ (‘good’): 1; ‘TB/très bonne’ (‘very good’): 2; ‘Ab/abondante’ (‘very high’): 3. For her group of ‘forte’ (‘strong,’ ‘sudden’), see next note. 78 michael mccormick

Fig. 2. Percentages of Nile flood qualities, early vs. later Roman Empire.

See Table 2 for details.

Table 2. Detailed categories of flood qualities of the Nile according to Bonneau 1971. Total Best floods Better than normal floods (values 2–3) % floods (value 1) % 30bc–155ad 112 22 19.6 22 19.6 156–299ad 87 7 8.0 19 21.8

Normal Worst floods Sudden rise at floods % (values -1 to -3) % some point (-0.5) % 30bc–155ad 28 25 24 21.4 16 14 156–299ad 30 34.5 27 31.0 4 4.6 Source: McCormick, Harper et al. 2012. abundant harvest were recorded on average every five years. However, under the later Roman Empire, similar conditions occurred only in eight percent of recorded years, that is, on average, extremely favorable conditions for cereal production occurred more than twice as rarely, only once every twelve and a half years. Thus the early Roman emperors unwittingly enjoyed a positive advantage in the conditions of Nile food production, quite independently of the overall more favorable climate conditions that we have detected. The early imperial advantage was probably only enhanced by the different the environmental history of the roman empire 79 patterns of poor or failed floods that prevailed on either side of 155ad.45 The least favorable floods—and therefore poorest harvests—occurred about as often as extremely favorable floods down to 155ad, that is, about once every five years. But after 155, poor or failed floods recurred almost every third year. The new observation is highly interesting. It appears to accord with the other, entirely independent evidence both for greater climate stability in a basically favorable regime under the earlier Roman Empire, and for increased climate instability in the first centuries of the later Roman Empire. This raises questions of potential pattern shifting in global climate regimes that should be of interest to climate scientists. The Nile data’s apparent evidence for a changing food supply for the Roman Empire’s cities and armies holds considerable explanatory power. Precisely for this reason, it will require detailed scrutiny from the vantage point both of the written records and whatever proxy data and potential climate mechanisms can be developed. In particular, I would like to spell out a little more explicitly than perhaps is customary among historians the caveats that should attend this proposition. Two in particular will need more consideration before so consequential an environmental change can be accepted as established, and integrated into analyses of the Roman economy. The first caveat arises from the uneven quality of the written record over this period. Bonneau classified the quality of the records and the deductions she drew from them at three levels: those in which she considered the evidence so explicit that the flood quality can be considered as completely certain; the floods that she has deduced with a level of confidence acceptable enough to leave her classification as unqualified; and those for which the documentation left enough uncertain

45 One final group of entries has been left out of this discussion. That is Bonneau’s category ‘Fo/forte (à quelque moment)’ (‘Strong [or sudden] (at some point)’), the values for which are supplied in Graph 2 and Table 2. These sudden floods could have been damaging: see Bonneau, 1971, 66–76. But this is not necessarily the implication, particularly if they did not last long or dissipated quickly: Bonneau, 76. Thus the very sudden and strong flood of 90ad seems to have been reckoned a very good one: Bonneau, 238 with 156, n. 759 and on 263, Graphique IV; the sudden flood of 125ad was commemorated by a Nile coin celebrating an abundant grain supply: cf. Bonneau 1971, 242 and Bonneau 1964, 330–331; in 131ad, although the flood may have been satisfactory, the strength of the flood caused destruction at Oxyrhyncus: Bonneau 1971, 243; in 141ad, despite some signs of difficulties, Elephantine recorded an excellent level, as did Alexandria: Bonneau, 245; cf. 183–184. Out of an abundance of caution, I classified such floods as negative in Table 2, at the value of ‘-0.5.’ They were more abundant in the earlier period (16 occurrences; 14%) against 4 (4.6%) in the later period. Could this decline in sudden floods be another indicator of the phenomenon of diminished floods that seems to prevail in the later period? The turning point around 155ad seems clear from a graph of the floods: see McCormick, Büntgen, et al. 2012, Figure 10. 80 michael mccormick

Table 3. Recording quality as assessed by Bonneau. Years Some Period documented Certain % Unqualified % doubt % 30bc–155ad 112 28 25.0 33 29.5 51 45.5 156–299ad 87 2 2.3 29 33.3 56 64.4 that she recorded doubt.46 As is not uncommon in ancient history, the top level of certainty is always the exception (Table 3); as is also not uncommon, that first level is more substantial in the earlier imperial period. Nearly a quarter of all the earlier floods (28; 25%) are attested explicitly and therefore with complete certainty, whereas that is the case only for a small minority (2; 2.3%) of the later Roman floods. For about a third of both data sets, Bonneau was satisfied enough with the evidence to leave her deduction without further qualification, even if it fell short of the exceptional certainty of her first group, as is, again, not unusual in ancient and medieval history. Finally, as in any rigorous examination of ancient evidence, a substantial portion of the deductions leaves room for some doubt: nearly a half of the early imperial floods, and nearly two thirds of the later ones fall into this category. The superior floods of the period down to 155, in other words, are evidentially more secure than the poor ones after that date. But there is real evidence for both phenomena. The second caveat also pertains to the nature of the written record. Over part of the later period in which the best quality floods appear rarer, and poor floods more common, a new kind of document emerges, the ‘declara- tion of unflooded land’ (apographē abrochias). This new type of document appears around 150ad and continues to crop up until 245. Why it appears, and why only in this period, is not known. Papyrologists’ speculation on its origins has mostly focused on a change in the mentality of documentary practice or the possibility that such a document could procure tax relief.47 Could the appearance of the new type of document explain the quantitative difference between early and late periods with respect to poor floods?

46 See Bonneau 1971, 217–218; cf. ibid. 14–16. Nevertheless the detailed presentation of her cautious classification method inspires confidence in the overall reliability of her results: Bonneau 1966. 47 Former explanation: Bonneau 1971, 183–187; she also considers that it may have been an effort by the emperor to make the farmer feel that he was participating in Egypt’s economic life; tax relief: Préaux 1963; good list but no explanation: Parassoglou 1987. I am grateful to Roger S. Bagnall for his thoughtful comments and guidance on these records. Naturally he bears no blame for what I make of them. the environmental history of the roman empire 81

Probably not. Bonneau adduced this type of document for only eighteen years between 157 or 158 and 239; six of those years had floods that Bonneau rated as normal.48 Another flood she rated as strong or sudden, and certainly so.49 Of the eleven years for which Bonneau uses the declarations to characterize poor floods, all except two are documented by other types of records as well, and one of the two also preserves a coin issued in connection with the flood, usually a sign of an at least partially successful flood.50 At most then, two additional poor floods of the total of 27 from this period could be due exclusively to the appearance of a new type of record. This would not change the overall picture. In fact, I cannot help but wonder if it were not, on the contrary, a change in the environment—the increased frequency of poor floods—that itself elicited the new type of document, as the tax authorities of Egypt were confronted with increasing appeals from farmers whose fields had not received the flood necessary to produce their usual taxable crop. In sum, even with careful caveats, it appears that considered as an aggre- gate, the data on Nile floods carefully assembled by Bonneau signal an important shift in the reliability and overall productivity of the Roman Empire’s Nile granary after 155ad. This certainly merits closer scrutiny by papyrologists and by climate scientists considering the precipitation patterns over eastern Africa and their possible global teleconnections; this new insight may also shed light on the increasing role of Proconsular Africa in supplying the city of Rome in the later empire. But these are questions for another day. Let us turn, rather, to what may be Roman responses to climate change documented in the archaeological record.

5. Is Tile More Comfortable Than Thatch? Climate Change and Human Response in the Archaeological Record

Within the tangled, incomplete web of what we know about ancient and medieval civilization, it can be terribly hard to observe and distinguish cause and effect for events big and little. This is true whether we approach the problem with the tools of the historian or of the archaeologist, or both. It is true in spades when we turn to the complexities of climate change.

48 See Bonneau 1971, 247–248, 249, 252–253, 254, 255, respectively for the floods of ad157, 163, 189, 201, 203, 225, 239. 49 Bonneau, 252, ad194. 50 Bonneau 1966, 384–385. 82 michael mccormick

The words ‘climate change’ actually designate different types of change that it is important to distinguish. Compared to short-term rapid change, gradual long-term environmental shifts seem a priori more likely to generate graduated and successful human responses. Nevertheless, we must recognize that context is everything: gradual change toward cooler temperatures may affect agrarian practices very differently in temperate areas of the Roman Empire compared to marginal areas. For instance, at higher altitudes even a slight shift in temperature might change definitively the growing season or the hardiness range. One easily imagines that rapid change such as volcanic cooling could have had more immediately disruptive effects. Never- theless, when those effects arose from short-term developments, they could well have been cushioned by a society that had the capacity to seek alternate food supplies until conditions returned to normal. We can expect that the well-organized Roman transport sector will have been capable of addressing and overcoming some climate-induced famines, especially in years in which the Nile produced abundant harvests in response to climate systems outside the Roman Empire. Distinguishing between trend and fluctuation captures another set of differences. Trends may be faster or slower and we may surmise differing consequences for slow versus swift shifts to new patterns, especially if the changes were lasting. Fluctuation is different: it is repeated change back and forth and it can be quick. One could imagine that for many kinds of farms, short-term fluctuation would have affected planting, growing and harvesting. This could have represented the most difficult type of climate change for a society. To explore the historical problem of societal response to climate change, we need to proceed on at least two levels. We need to select examples of specific, securely demonstrated and understood climate change with high chronological resolution in specific geographic regions, and analyze in their light relevant historical and archaeological data, for instance, about food production or water supply.51 Let me suggest some questions for future research. Regarding exceptionally good or bad Nile flood levels, can we detect responses within Egypt in the surviving, well dated papyri: explicit responses such as price changes, labor shortages, surges or declines in land transactions, and the like? Beyond Egypt, can we observe direct or indirect signs of supply difficulties, for instance in the capitals?52

51 See for example Schmidt et al. 2005; for a possible exploitation of climatic variation by Roman engineers, see Schmidt 2010. 52 See the brief but suggestive comments of Bonneau 1966, 394–395. the environmental history of the roman empire 83

Another approach would be to investigate the impact of different types of climate change on farming practices as documented by written records, the archaeology of ancient farming technology and food preservation and storage, and archaeobotany. Do the surviving agrarian calendars give any hints of such adaptation? It might be worthwhile to examine in this light the fifth- century Palladius in the western Empire, or the late antique agrarian writers preserved in the Byzantine Geoponica.53 In any case, actual agrarian production systems as they developed in antiquity and the Middle Ages must have reflected the changing environment just as they are acknowledged to have reflected—and inflected—changing social, cultural and economic structures.54 Because it is destined to keep growing, the archaeobotanical evidence is very promising. For example, that declining early summer precipitation in northeastern France and Germany from about 230 to at least 315, and again from around 425 to 650, affected the most sensitive season for many food plants.55 Now palaeobotanical research has made clear that cultivation of rye, normally sown as a winter crop, expanded in the late Roman period. Scholars have suspected a link between rye’s cold hardiness and possible climate cooling; rye in fact needs a hard frost to be able to sprout, lessening its value in warmer climates.56 However the new dendroecological evidence shows two long periods of reduced spring precipitation. Could it have been this particular abrupt and lasting climate change that elicited this human agrarian response? In fact it has not been sufficiently appreciated that rye

53 Palladius Rutilius Taurus Aemilianus, Opus agriculturae, ed. Rodgers 1975; Geoponica, ed. Beckh 1895. 54 Henning 2009. 55 Under modern climate conditions, the temperature and precipitation sensitivity of winter grains peaks in western and central Europe during the heading phase that occurs in May and June: United States Department of Agriculture, Foreign Agricultural Service, ‘Monthly Crop Growth Stage and Harvest Calendars,’ http://www.fas.usda.gov/pecad/ weather/Crop_calendar/crop_cal.pdf, consulted January 4, 2012. Cf. Food and Agricultural Organization of the United Nations, Land and Water development division, Crop Water Man- agement, ‘Wheat, Water Supply and Crop Yield’: http://www.fao.org/landandwater/aglw/ cropwater/wheat.stm#supply, consulted January 4, 2012. 56 For Roman rye and cold hardiness, see Henning 1987, 100. See in general the comprehensive and critical review in Behre 1992 and, on hard freezing and sprouting, ibid. 145. More recently, for instance, a survey of carbonized fruits and seeds identified from rescue excavations in France’s Aisne valley from the Neolithic to the early Middle Ages yielded rye only in deposits dated to the third-fourth and fifth-sixth centuries ad: Bakels 1999, at 74, Table 1. Rye was found stored in the Roman fort at Dichin, Bulgaria from the destruction which occurred ca. 480. I am most grateful to Andrew G. Poulter for his kind shar- ing of details of his excavation in advance of publication. In the meantime, see Grinter 2007. 84 michael mccormick enjoys a special ability to withstand spring drought. The new precipitation record suggests to me that this drought hardiness may have driven rye’s emergence as a food crop at least as much as cooling (whose exact parameters remain unclear) in late Roman times.57 I can easily imagine how such a shift might have occurred. Rye grains first show up on archaeological sites in very small quantities and associated with other cereals.58 Palaeobotanists interpret this as signifying that rye was present as a weed plant. Some farmers would surely have noticed if a spring drought killed off all the wheat they had sown, but another adven- titious cereal weed—rye—continued to thrive in an otherwise devastated field. Dire necessity could well have encouraged them to experiment with the food preparation and cultivation of a new crop if they survived the first bad harvest.59 Continuing archaeological investigation should deepen and reinforce the link between changing precipitation patterns and a shift to cul- tivating rye over more delicate forms of wheat. If so we have identified a specific and broad-ranging change in diet and the agrarian economy as human response to climate change. A shift originating in the changing environmental conditions of the late Roman Empire will have shaped the spectrum of cereals sown in medieval Europe, including the much-noted consequences of rye’s propensity to ergotism.60 And that late Roman climate change will have continued to shape the German diet down to the twentieth century.61 Other case studies have already been adumbrated. They now need ex- panding in the light of the new climate data. In the British Isles and on the Continent, opinions continue to develop about the culinary or climate implication of the late Roman proliferation of corn (grain) drying ovens.62 But what about the impact of cooling on clothing, domestic architecture and living arrangements?63 Rooms may have gotten smaller to facilitate heating,

57 Rye’s deep root system mostly explains this resistance; see Peltonen-Sainio et al. 2009, at 77–78; for details, [University of Florida, the Food and Agricultural Organization of the United Nations, and the National Museum of Natural History of the Smithsonian Institution], ‘Eco- port,’ Secale cereale, ‘Description: Physiology,’ http://ecoport.org/ep?Plant=1929&entityType =PL&entityDisplayCategory=full. 58 Behre 1992, 141, and 143 for Roman examples. 59 Although ancient farmers’ practice is of course only very imperfectly recorded, exper- imenting with different types of crops is in fact explicitly documented in the Geoponica in the light of astrological influences on crops and soil: Geoponica 2, 15, ed. Beckh 55.1–13; trans. Dalby 2011, 82. 60 See Haller 1993 and Carmichael 1993. 61 Wheat supplanted rye as the dominant bread cereal in the Federal Republic of Germany only in the 1960s: Körber-Grohne 1994, 43. 62 Compare Dark 2000, 83–84 and Haas 2006, 256–258, with further references. 63 For clothing, see Zabehlicky 1994, 465–466 and below, n. 67. the environmental history of the roman empire 85 whether by braziers which, because they were portable, are nearly invisi- ble archaeologically, or by more expensive built in heating systems.64 The famous Roman hypocausts—floors and sometimes walls heated by hot air from nearby furnaces—offer the advantage of being fairly well known and archaeologically hard to miss. Conventional wisdom is that central heating spread in the later Empire.65 It is certainly possible that centrally-heated hypocaust dwelling structures became more common in the Roman Empire’s northwestern provinces in the third century. But that cannot be considered proven. One difficulty lies in the often imprecise dating of construction; another in judging whether the proportion of villas with central heating increases in a given region; and a third in discerning whether wealthy people began to build villas with centrally heated rooms in response to newly colder temperatures, to something more cultural such as the spread of a metropolitan fashion, to increased disposable wealth, or to some combination of all these.66 It has been argued that the strongest case for a connection between climate cooling and the building of central heating systems may come from pre-existing dwellings to which heating systems were added, and there are hints that more resources were devoted to building higher performance heating systems in the northwestern provinces in the third and fourth centuries.67 If true, this would accord with palaeoclimatic signals of cooling in central Europe around 150 and particularly from ca. 200, albeit with some periods of warming, especially ca. 365.68 For this argument to go beyond an impression would require a systematic quantitative survey of all villas with or without heating systems in a given region, with careful attention to dating and also to possible changes in the mode of dwelling room sizes. Such a study

64 See the discussion in Haas 2006, 252–256, with further references. 65 E.g., DeLaine 1996, 737. 66 For numerous late Roman heating installations in northern Gaul as well as problems of dating, see, e.g., Van Ossel 1992, 128–130. 67 Remodeling with heating systems: Haas 2006, 255; Zabehlicky 1994 argued that a volcanic eruption of 186ad caused a cooling of the Roman Empire that lasted into the third century, and detects a cultural response in the form of warmer uniforms for Roman soldiers (that, as he observes, at least in part antedate the eruption), and in an increase of the proportion of villas with central heating. The latter argument is made provisionally and based on a selection of 18 British and central European villas, chosen because he reckoned them securely dated between the early second and late third centuries (466); naturally this can be no more than suggestive. A few more possible indications from Ephesus are added in Vet- ters and Zabehlicky 2001. For a suggestive overview of Roman built-in heating technology, including experimental archaeology, see Schiebold 2010. 68 McCormick, Büntgen, et al. 2012. 86 michael mccormick could illuminate the capacity of a particular socioeconomic group to insu- late themselves from the effects of cooling climate, and perhaps shed light on differing patterns of climate change and human response in different regions of the vast empire. A medieval analogue may turn out to be the spread of fireplaces or stoves with chimneys in castles and lordly residences as Europe’s medieval warming came to an end.69 The built sources of heat are but one aspect of the archaeological detec- tion of human responses to climate change. How could cold peasants shiver less in their hearth-heated houses in an age after the Roman braziers but before the later medieval chimneys, fireplaces and stoves? Insulation is key. Archaeologists and historians have lately made much historical hay, as it were, from the early medieval replacement of Roman fired tiles by the organic roofs of thatched straw. This has been depicted as prima facie evidence of decline in physical comfort, and plausibly interpreted as signaling an economy so deeply in reverse that it could no longer afford or organize to bake clay roof tiles for its homes.70 But could changing climate have helped drive this change also? Anyone who has lived, for instance, in a nineteenth-century Belgian worker’s cottage with a kerosene space heater in the age before retrofitting with insulation knows that its tile roofs and brick walls did not allow the heat from a liter of kerosene to last long beyond its burning. Modern thatchers claim for their roofs extreme thermal efficiency comparable to several inches of modern fiberglass insulation. English Heritage confirms: ‘Thatch has a much greater insulating value than any other traditional roof covering.’71 With respect to durability, the same organization notes that some of the 50,000 thatched buildings in modern Britain ‘retain thatch over 600 years old.’ But that is clearly very exceptional, when modern estimates for a thatch roof’s lifespan run between 10 and 50 years, depending on the type of straw and many other considerations.72

69 Although the beginning of their diffusion, ca. 1000, may seem rather early in view of the conventional wisdom on the medieval warm phase and the development of later medieval cooling: Meyer 1995. For more details about the development of medieval fireplaces, chimneys and stoves, see Sirot 2011. Again, rigorous quantification is needed to clarify the evidentiary potential of medieval chimneys, open fireplaces and stoves. 70 Ward-Perkins 2005, 94–96; cf. 110, where he recognizes the superior thermal quality of thatch. 71 Ogley et al. 2010, 3; and 7–8. The report is available at http://www.english-heritage.org .uk/content/publications/docs/eehb-insulating-thatched-roofs.pdf. 72 Quote: Ogley et al., 3. English Heritage 2000, 9, on the longevity spans of English thatched roofs. This report is available at http://www.english-heritage.org.uk/content/ publications/docs/thatchandthatching.pdf. the environmental history of the roman empire 87

So did the spread of early medieval thatch—detectible archaeologically by the absence of roof tiles, the lighter roof structures it implies and, occasionally, from the thatch itself—result from economic collapse, or did it represent something more complex? In the latter case, it may have reflected changing economic structures and new climate conditions, that is, the likely broader cooling phases presently detected from ca. 200 to 365, in the early fifth century, and throughout the sixth century down to about 650; the old- fashioned tiles would simply not have been worth the presumed extra cost when they did such a poor job of countering the new climate conditions. It may only be a happy coincidence in light of the spread of rye, but that cereal’s stalks are supposed to make good thatch.73 To test this hypothesis will require examining roofing patterns in similar dwellings within different regions of northwest Europe as climate science clarifies the timing, duration, and seasonal and spatial reach of cooling. One could continue, but the basic point should be clear. Ongoing and future research in archaeology and climate science will furnish more and more data that will allow us to investigate in detail whether and how change in the natural environment forced changes on the built and farming environment, and that will be the first step in a much longer and exciting archaeological investigation into the interaction of climate, human health, economies and culture.

This first reconnoitering of a large and complex theme provides a few provisional conclusions. The era of consilient research is at hand. It draws on the advances of the natural sciences alongside the humanities and social sciences to address crucial questions of historical change. The Roman and early medieval world offers archaeological and historical knowledge that illuminates most remarkably those human societies and their changes. Now the rapidly accumulating data of climate science is adding a new dimension to what is knowable about the environmental conditions of antiquity and the Middle Ages. This new data can be brought to bear on traditional questions, as we saw in the case of dating Ausonius’ Mosella, even as they expand the range and breadth of investigations we can devise to deepen our understanding of how Roman and post-Roman civilization interacted with its natural environment. Secondly, there are signals of rapid shifts toward favorable or unfavorable climate conditions in different areas of the Roman Empire. These changes

73 Paillet 2005, 53. 88 michael mccormick may be correlated and even, to some degree, causative of archaeologically documentable phenomena such as the spread of rye cultivation, built heating systems in dwellings, and thatched roofs. Some changes toward less favorable climate conditions also coincided with periods of crisis in the Roman Empire. Of course co-occurrence is not cause, even if it is a necessary precondition for establishing causality. Today the evidence is insufficient to claim that climate anomalies determined the course of empires and civilization. Nevertheless climate change certainly was capable of playing a role in the complex unfolding of human history by changing the background environmental conditions of food production, animal and human health, and in the seasonal patterns of daily life. Establishing the details will prove delicate. But even short-term sudden climate anomalies such as the tremendous volcanic winter of 536 must have had serious economic and human consequences, including mental ones, although that event only weakened and did not destroy the Roman Empire of Justinian. The task before us is to identify and date rigorously such moments, factors and actors in the growing record of ancient and medieval environmental history. Only then can we begin to understand their interaction and implications for our broader task of explaining the lives of the women and men who have preceded us. In broader terms, we will need to devise innovative ways of measuring climate change to test and analyze its possible correlation with bigger developments of political, social, and economic change. But for that measure to be truly useful, we will also have to find novel ways of measuring political and economic change in ancient and medieval civilization, and of correlating those two measures.74 There is much to be done.

74 For recent efforts to create ‘instability indexes’ for ancient Rome and the Byzantine Empire, see Turchin and Scheidel 2009, with the (online) ‘Supporting Information’ at http:// www.pnas.org/content/106/41/17276/suppl/DCSupplemental; and Preiser-Kapeller 2010, esp. 29–34, on his instability index and related approaches to Byzantium, with further references. The working paper is available at http://oeaw.academia.edu/JohannesPreiserKapeller/ Papers/506625/Complex_historical_dynamics_of_crisis_the_case_of_Byzantium. Cf. e.g., the various indexes devised and deployed by Morris 2010b, 150–169, etc. MEGADROUGHTS, ENSO, AND THE INVASION OF LATE-ROMAN EUROPE BY THE HUNS AND AVARS

Edward R. Cook

Introduction

The degree to which climate played a role in the invasion of late-Roman Europe by the Huns in the 4th–5th centuries ad and the Avars in the 6th century ad has long been debated, e.g. from Huntington (1907) to McCormick, Büntgen, et al. (2012). The importance of climate in shaping this critical period of European history has been investigated in some detail using millennia-long, precisely dated, annual tree-ring chronologies from central Europe (e.g. Büntgen et al. 2011), but these superb records of past environmental change provide no direct insights into what climate was like around the same time in the home territories of the Huns and Avars in central Asia. This limitation is difficult to overcome because of the paucity of millennia- long tree-ring records directly from the central Asian steppes and western China where these nomadic “barbarian” tribes are thought to have migrated from. However, it is still possible to make some potentially useful inferences and speculations about what climate may have been like based on the limited tree-ring data available. This will be done through the presentation of a long tree-ring record from north-central China (Qinghai-Tibetan Plateau) that spans this interesting cultural period. In so doing, we will argue that prolonged periods of drought may have contributed to the migration of the Huns and Avars westward into Europe during late-Roman times. In addition, we will investigate the long-range teleconnection between the tropical Pacific Ocean El Niño-Southern Oscillation (ENSO) system and central Asian climate as a possible contributor to the development of droughts there.

Megadroughts in North-Central China

Presently, only two annual tree-ring chronologies from north-central China are sufficiently long to tell us what climate may have been like in the homelands of the Huns and Avars at around the times of their westward 90 edward r. cook

Fig. 1. The Dulan-Wulan annual tree-ring chronology from north-central China and and the occurrence of severe droughts during the times of the Hun-Avar migrations into late-Roman Europe. Three multi-decadal droughts, among the worst of the past 2,000 years, are indicated in the 4th, 5th, and 6th centuries ad at around known times of invasion by these nomadic peoples from central Asia. The lower plot illustrates this more clearly. migrations. The first record is the millennia-long juniper (Juniperus przewalskii) tree-ring record from the Dulan-Wulan region in northeastern Qinhai Province, which extends back over 2,000 years (Zhang et al. 2003; Sheppard et al. 2004). It has been used to reconstruct annual precipitation (Sheppard et al. 2004) and thus provides a useful reflection of local hydroclimatic variability there. This record, expressed as standard normal deviates, is shown in Fig. 1 over its entire length of record and only for the ad200–700 time period containing the Hun and Avar migrations for easier assessment. A second multi-millennial tree-ring chronology produced by Shao et al. (2010) is located about 100km west of Dulan-Wulan and is likewise an expression of moisture variability and change there. It basically supports the story shown in Fig. 1 during the Hun-Avar period and, therefore, will not be described further here. megadroughts, enso, and the invasion by the huns and avars 91

The Dulan-Wulan record (Fig. 1) reveals three periods of intense multi- decadal drought during the Hun-Avar migration periods in the mid-4th, 5th, and 6th centuries, with the first ‘megadrought’ being the worst such event of the past 2,000 years. The 4th century megadrought centered around ad360 occurred at about the same time as the first migration of the Huns westward into Roman Europe (McCormick, Büntgen, et al. 2012; Heather 1998). It is conceivable that this period of intense aridity spurred the nomadic Huns to seek better living conditions westward of their home territory to as far as the eastern Roman Empire, with invasion and conquest a natural part of this migratory process. The second major drought in the mid-5th century occurred at around the time of the second Hunnic invasion of Roman Europe in ad447–451 (McCormick, Büntgen, et al. 2012; Heather 2000), once again suggesting that drought may have played a role in inciting that invasion too. However, this period of conquest happened mainly in the early stages of this megadrought, which suggests that it may not have been the primary inciting mechanism. Interestingly, the 4th and 5th century megadroughts are also separated by about 50 years of mostly above average wetness. This ‘pluvial’ period is likely to have produced better living conditions for the Huns in their central Asian homelands, thus allowing them to build up their capacity for the invasion and conquest in Roman Europe. The final drought period indicated in Fig. 1 is centered on ad550 at about the time of the Avars’ invasion of late-Roman eastern Europe (McCormick, Büntgen, et al. 2012; Whitby 2000). While not as extreme as the previous two megadroughts, this period of dryness may again have incited the nomadic Avars to migrate westward in search of better conditions and plundered wealth. And like the wet period preceding the invasion of Roman Europe by the Huns, there is a similar period of above average wetness that preceded the invasion of eastern Europe by the Avars. This again may have allowed the Avars to build up their capacity for invasion. In summary, the evidence presented here suggests that climate may have played a role in the invasions of late-Roman Europe by the Huns and Avars. This evidence is in the form of three multi-decadal droughts indicated in the Dulan-Wulan tree-ring chronology at around the times of the Hun-Avar migrations. Above average moisture supply may have also contributed to the latter two invasions by improving the capacity for migration and invasion by these nomadic pastoralists. It must be emphasized, however, that the Dulan-Wulan record is from north-central China and is therefore unlikely to fully represent what climate was like in the homelands of the Huns and Avars on the central Asia steppes. However, severe droughts can cover 92 edward r. cook very large areas (Andreadis et al. 2005). Thus, we suggest that these three multi-decadal droughts could easily have extended into the steppes of central Asia as well. Assuming this to be true, how might these persistent depar- tures in moisture availability have happened? In the next section we will present evidence for the long-range teleconnection between precipitation over central Asia and China and the El Niño-Southern Oscillation (ENSO) in the equatorial Pacific as a plausible mechanism for how the Hun-Avar megadroughts occurred. This will be done through the use of two millennia- long ENSO sensitive tree-ring chronologies from New Mexico in the USA and the North Island of New Zealand.

ENSO as a Hypothesized Cause of the Hun-Avar Megadroughts

The modern link between ENSO and drought in central Asia. The El Niño- Southern Oscillation (ENSO) is the most important mode of internal climate variability on earth (Rasmusson and Wallace 1983; Cane 2005). Its global influence on climate was first described by Walker and Bliss (1932), and more recently for precipitation and drought by Ropelewski and Halpert (1987), Dai and Wigley (2000), Diaz et al. (2001), and Vicente-Serrano et al. (2011). Physical mechanisms why precipitation patterns occur as they do under the influence of ENSO can be found in Seager et al. (2005), and the robustness of ENSO as an important tool for forecasting climate up to two years ahead can be found in Chen et al. (2004). So what is the modern day influence of ENSO on precipitation and drought over central Asia? Walker and Bliss (1932) were unable to report on anything there because of insufficient preciptation data over large areas of central Asia at the time. Indeed, it is only since 1951 that one can readily claim sufficient spatial coverage of precipitation stations over central Asia for rigorously testing for the influence of ENSO on precipitation and drought there (cf. Cook et al. 2010). Therefore, we will only report on correlations between ENSO and preciptation over Europe and Asia since 1951. To this end, the measure of ENSO variability we will use is the winter (DJF) season Niño 3.4 sea surface temperature (SST) index from the eastern equatorial Pacific, a region and season that are highly representative of ENSO (Tren- berth 1997). The precipitation data set used is the 0.5° gridded GPCC global land-surface monthly precipitation data set (Rudolf and Schneider 2005). All correlations and the resulting maps shown were produced using KNMI Climate Explorer (http://climexp.knmi.nl/). megadroughts, enso, and the invasion by the huns and avars 93

Figure 2 shows the corrrelations between DJF Niño 3.4 SSTs and March– June GPCC precipitation data over Europe and Asia. They were calculated using the original data online and after the data were first-differenced to eliminate the influence of trend on the results. Only those correlations significant at the 90% level of significance (p<0.10, 2-tailed) are shown. Positive correlations indicate that warm SSTs (El Niños) produce wetter conditions and cold SSTs (La Niñas) produce drier conditions over central Asia and China under modern day conditions. The March–June precipitation season produced the strongest spatial pattern of significant correlations with DJF Niño 3.4 SSTs overall. This is highly significant because moisture supply during the spring months could be critical to the quality of livestock grazing and the overall pastoralist lifestyle of the Huns and Avars. Correlations based on both the original and first-differenced data both show a strong patterns between ENSO and precipitation over north-central China and central Asia. These results indicate that ENSO could have played a role in producing the Hun-Avar megadroughts in the 4th–6th centuries ad. The question now is: what evidence do we have to more directly implicate ENSO that far back in the past? Are there any proxy records of ENSO that extend back over the past 2,000 years? Next we will show that such records do exist and the appear to lend further support for the ENSO/Hun-Avar megadrought hypothesis.

Paleo-estimates of ENSO variability during the Hun-Avar period. Two millennia-long tree-ring chronologies from ENSO-sensitive regions in the North- ern and Southern Hemispheres will be used here. The first is a Douglas fir (Pseudotsuga menziesii) chronology from the El Malpais lava field in northwestern New Mexico USA. This Northern Hemisphere record was produced by Henri Grissino-Mayer and the data are lodged in the International Tree-Ring Data Bank (ITRDB; http://www.ncdc.noaa.gov/paleo/metadata/ noaa-tree-8517.html). Grissino-Mayer (1996) produced an annual (water- year) precipitation reconstruction from this record in an area of the Amer- ican Southwest known to be strongly influenced by ENSO (Ropelewski and Halpert 1987). The Southern Hemisphere tree-ring chronology used here is a Kauri (Agathis australis) record from the North Island of New Zealand. These data were kindly provided for this study by Gretel Boswijk, Anthony Fowler, and Jonathan Palmer, the primary developers of this chronology. Subsets of the data used here can be found in the ITRDB (http://www.ncdc .noaa.gov/paleo/treering.html), but the long multi-millennial record used here is not yet publically available. Boswijk et al. (2006) describe the development of this long record based on a combination of cross-dated tree rings from living trees, archeological timbers, and sub-fossil wood. Fowler et al. 94 edward r. cook

Fig 2. Correlations between December–February Niño 3.4 sea surface temperatures (a measure of ENSO variability) and March–June total precipitation from 1951 to 2003. Significant (p<0.10, 2-tailed) correlations are indicated over central Asia and China in approximately the regions where the Huns and Avars would have come from. This illustrates the modern sensitivity of precipitation there to ENSO variability, with warm eastern equatorial Pacific sea surface temperatures (El Niños) producing wetter conditions and cold sea surface temperatures (La Niñas) in the same region producing drier conditions. megadroughts, enso, and the invasion by the huns and avars 95

Fig. 3. Two 2,000 year long annual tree-ring chronologies from ENSO sensitive regions in the Northern and Southern Hemispheres: Douglas fir from northwest New Mexico and Kauri from the North Island of New Zealand.

(2000 as well as Fowler 2008 and Fowler et al. 2008) provide the case for the sensitivity of Kauri to ENSO. The exact cause for this sensitivity is not well understood, but as we will show here it appears to be highly robust. Figure 3 shows these chronologies in standard normal deviate form. Each of these records is 2,000 years long and is thought to express the effects of ENSO variability and forcing on climate over the regions where the trees are growing. In order to test this hypothesis, we used KNMI Climate Explorer to produce maps of correlations between these tree-ring records and Hadley Centre HadISST1 gridded global SSTs (Rayner et al. 2003) for the winter (DJF) season from 1871 to 2003. This includes the exact region in the eastern equatorial Pacific where the Niño 3.4 ENSO index is estimated from the SSTs. Figure 4 shows the results of these correlation analyses, with the locations of the chronologies indicated by red stars. In each case, the correlations are shown (p<0.10, 2-tailed) based on the original data, after the data were detrended, and after the data were first-differenced. This was done to test the statistical robustness of the putative ENSO signal in each chronology. A meaningful ENSO signal should result in a robust pattern of signficant correlations between tree rings and SSTs in the eastern equatorial Pacific and this pattern should be relatively insensitive to the way the data were 96 edward r. cook

Fig. 4. Correlations between the Douglas fir and Kauri tree-ring chronologies and Hadley Centre global sea surface temperatures (HadISST1) for the winter season: 1871–2003. The locations of the tree-ring chronologies are indicated by red stars. Only regions with correlations significant at the 90% level (p<0.10, 2-tailed) are shown. A meaningful ENSO signal should result in a robust pattern of signficant correlations between tree rings and SSTs in the eastern equatorial Pacific, with additional correlations found in the Indian Ocean sector, and this pattern should be relatively insensitive to the way the data were treated. In each case, a robust ENSO signal is indicated regardless of how the data were treated for presence of trend. megadroughts, enso, and the invasion by the huns and avars 97

Fig. 5. The average (A) and difference (B) of the Douglas fir and Kauri tree-ring chronologies. If a true ENSO signal of the same kind exists in both, as suggested by Fig. 4, then the average should preserve that ENSO signal as well and perhaps even strengthen it. The difference between the chronologies is a test for an interhemispheric gradient in ENSO variability.

treated. This is exactly what was found. In every case, a robust ENSO signal over the eastern equatorial Pacific is indicated in both the Douglas fir and Kauri chronologies located on opposite sides of the equator. While somewhat expected for Douglas fir in New Mexico based on known ENSO/climate teleconnections (Ropelewski and Halpert 1987), the Kauri result is equiva- lently strong, which validates the claims of Fowler et al. (2000; Fowler 2008 and Fowler et al. 2008). The robust ENSO signal in each tree-ring chronology suggests that an even better result might be found in the average of these two records. Testing the average is also a stringent additional examination of ENSO signal robustness because if a true ENSO signal is present in each, it should be preserved or even strengthened in the average. Figure 5a shows the average of the two tree-ring chronologies and Fig. 5b shows the difference. The latter is included here to test for a possible interhemispheric gradient in ENSO variability as well. These two series were used in another correlation analysis with SSTs of the kind done previously for the individual chronologies, and their results are shown in Fig. 6. The average series has an ENSO signal in it that is at least as strong as any of the individual chronologies. This 98 edward r. cook

Fig. 6. Correlations between December–February average SSTs and the average and difference of the Douglass fir and Kauri annual tree-ring chronologies, each with an identified ENSO signal. The locations of the chronologies used in the average and difference are indicated by red stars. The average has an ENSO signal that is at least as strong as any of the individual chronologies. This suggests that this interhemispheric signal is real and not a statistical artefact. The lack of any meaningful correlations in the chronology differences indicates no gradient in the interhemispheric ENSO signal. megadroughts, enso, and the invasion by the huns and avars 99

Fig. 7.Comparisons of correlation patterns of March–June precipitation with actual and tree-ring ENSO indices. The top figure shows the correlations with the actual DJF Niño 3.4 ENSO index. The bottom figure shows the correlations with the tree- ring ENSO index based on the average of the Douglas fir and Kauri chronologies. While somewhat weaker, the pattern of significant correlations (p<0.10) in central Asia based on the tree-ring index is essentially the same as that based on the actual Niño 3.4 index. 100 edward r. cook

Fig. 8. The Douglas fir—Kauri average interhemispheric ENSO index with drought inducing La Niña periods indicated around the times of the Hun-Avar migrations into late-Roman Europe. The lower plot illustrates this more clearly. The timing of these drought inducing periods is not one-to-one with the indicated megadroughts in the Dulan-Wulan record from north-central China (Fig. 1). However, the general timing of these drought inducing La Niñas overlaps sufficiently well to suggest that ENSO did in fact contribute to the megadroughts that may have incited the Huns and Avars to migrate west and invade late-Roman Europe. suggests that there is a true interhemispheric ENSO signal in these series that might tell us what ENSO variability was like during the Hun-Avar megadroughts. In contrast, the lack of any meaningful correlations in the chronology differences indicates that there is no interhemispheric gradient in the strength of the ENSO signal. Given the strong ENSO signal in the average tree-ring series, the logical next question to answer is how well this tree-ring ENSO index produces the pattern of corrrelations found in Fig. 2 between the actual DJF Niño 3.4 ENSO index and March–June precipitation over central Asia and parts of northern China. Figure 7 answers this question in a generally affirma- tive way. The pattern of positive correlations in central Asia based on the megadroughts, enso, and the invasion by the huns and avars 101 tree-ring ENSO index (Fig. 7b) matches those based on the Niño 3.4 index (Fig. 7a) reasonably well. The strength of the tree-ring ENSO index correlation pattern is weaker overall, but the geographic location of those correlations matches the Niño 3.4 pattern in central Asia quite well. This results means that we can use the multi-millennial tree-ring ENSO index as a reasonably accurate surrogate for ENSO variability during the Hun-Avar migration periods and use it to infer what the March–June moisture conditions in central Asia were like during those times. So what does this 2,000 year long proxy of ENSO tell us about the causes of the 4th, 5th, and 6th century megadroughts in central Asia and China? Figure 8 indicates that unusual periods of drought inducing La Niña-like conditions occurred around the times of the Hun-Avar migrations. The timing of these drought inducing periods is not one-to-one with the indicated megadroughts in the Dulan-Wulan record from north-central China (Fig. 1). This is not unexpected given that the Dulan-Wulan record itself is probably a locally biased expression of drought variability over central Asia and the tree-ring ENSO index is not perfect either. However, the general timing of these drought inducing La Niña-like conditions overlaps sufficiently well with the megadroughts at Dulan-Wulan to suggest that ENSO did in fact contribute to their development. Consequently, megadroughts caused in part by ENSO are viable hypotheses for explaining what might have incited the Huns and Avars to migrate west and invade late-Roman Europe.

Concluding Remarks

A hypothesis has been presented here for a climate mechanism that could have incited the Huns and Avars to migrate west and invade late-Roman Europe in the 4th, 5th, and 6th centuries ad. This hypothesis uses a millennia-long tree-ring chronology from Dulan-Wulan in north-central China to show that a series of megadroughts probably occurred around those times in the homelands of the Huns and Avars. The cause of these megadroughts is then shown to be plausibly related to long-range ENSO forcing of climate, which influences March–June precipitation amounts in central Asia and northern China in the modern era. Two multi-millenial ENSO-sensitive tree- ring chronologies from New Mexico and New Zealand support this hypothesis by indicating that persistent drought-inducing La Niña-like conditions occurred around the time of the megadroughts and Hun-Avar migrations. To verify these hypotheses, more multi-millennial paleo records of hydroclimatic variability in central Asia are needed. 102 edward r. cook

Acknowledgements

This research presented here was stimulated by an invitation to give a talk at the Dumbarton Oaks Workshop of Climate Change under the late Roman Empire, April 24–25, 2009, organized by Zachary Smith (aka Mike McCormick). I thank Henri Grissino-Mayer for his remarkable El Malpais Douglas fir tree-ring data and Jonathan Palmer, Gretel Boswijk, and Anthony Fowler for generously allowing me to use their amazing Kauri data, much of it unpublished. Lamont-Doherty Earth Observatory Contribution Num- ber 7700. THE ROMAN WORLD AND CLIMATE: CONTEXT, RELEVANCE OF CLIMATE CHANGE, AND SOME ISSUES*

Sturt W. Manning

Introduction

This essay seeks to develop what is now a burgeoning discussion on the role and relevance of climate for the history of the Roman world (especially McCormick, Büntgen et al. 2012 and literature cited; Büntgen et al. 2011).1 The essay begins by reviewing some of the background issues, questions of focus (timescales and resolution), contradictions, and problems which are all inherent when investigating climate context in the Mediter- ranean region in the later Holocene (last few thousand years). It then considers the main forcings on climate over this recent and relatively short timescale—namely, the sun and volcanic eruptions (Shindell et al. 2003; Wanner et al. 2008, 1802–1804; Wanner et al. 2011, Figure 5)—and reviews what records of changing solar activity tell us for the Roman period, and the parameters these records suggest for climate and climate change in the Roman period. To address some specifics, I then review some of the high- resolution climate archives available and relevant to the Roman world (from tree-rings and from speleothems), before integrating some of these observations with some of the other published data for climate and environmental

* I thank William Harris for inviting me to the American Academy in Rome, and for orga- nizing a very enjoyable event including even a lunar eclipse, and for his editing skills. I also thank Michael McCormick, both for an invitation to an earlier workshop at the Dumbarton Oaks which led to focus on climate and the Roman world, and for his comments on a draft of this text. I thank Jürg Luterbacher, Eleni Xoplaki, Charlotte Pearson and Catherine Kearns for comments and reviews of drafts at various stages, and thank them very much for making many good suggestions and improvements. The remaining errors and mistakes are by the author. 1 In this essay the term ‘Roman world’ refers approximately to the time period from the 3rd century bc to the end of the 7th century ad—so around about a millennium. In terms of space, I mainly refer primarily to the Mediterranean area including the Levant. At various times central and northern Europe—incorporated into the High Roman Empire—are also mentioned, but usually as contrasts with the general Mediterranean zone. 104 sturt w. manning change in the Roman Mediterranean. Some patterns and implications are noted; some contradictions and problems are highlighted. At present we may usefully, but only generally, begin to describe the climate context of the Roman world—more data from a wider spatial network, and especially good chronological control on these data, are essential to overcome current contradictions and uncertainties.

Some Issues

Climate is complex, and depending on the scale of resolution—whether in terms of geographical area or temporal precision (and longer-term trends v. short-term data)—and place of observation, different and even apparently contradictory observations may be made. Climate forcings may have varying effects on human history, depending on a society’s environmental setting, social, economic and political structures, and those of its neighbours. As Trigger (2003, 279) notes, ‘there is no basis for theories that attribute the rise of civilization to the influence of a single type of environment or climatic event’. The correlation or path from climate to human history, if present, is instead a complex, multifaceted, multiscalar interaction contingent on social context and human agency (e.g. Rosen and Rosen 2001; Fisher et al. 2009; Wossink 2009). Rare instances of major stress and challenge—trigger points—are sometimes easier to diagnose than stability or opportunity. Within the generally rather mild and stable Holocene era (an interglacial period) of the last ca. 12,000 years, some key periods of significant or rapid climate change (linked primarily to aridity—water being the key to life in temperate to hot loci) are widely noted as relevant (at least) to SW Asia and the Mediterranean region in prehistory (e.g. episodes ca. 6200bc/8200bp,2 3200bc/5200bp, 2200bc/4200bp, and 1200bc/3200bp), and these seem to translate via various processes into historical change-point periods of relevance to prehistory (e.g. Weiss 2000; Weiss and Bradley 2001; deMenocal 2001; Chew 2002; Staubwasser and Weiss 2006; Weninger et al. 2009). But even these few now well-known episodes, which have become almost standard issues to mention in the archaeological literature, are often not entirely without problems if examined critically.Looking to an earlier period

2 bc = Before Christ or Before Common Era (bce); bp = Before Present (which in radiocarbon contexts refers to dates before ad1950; but in some other palaeoclimate contexts it refers to dates before ad2000, or even other dates stated by authors); ad = Anno Domini or Common Era (ce). the roman world and climate 105 before, around, and following the development of agriculture (23,000 to 8,000 years Cal bp), where climate has long been held to be some sort of prime-mover, the study of Maher et al. (2011) offers an excellent example of this: they find that there is in fact a lack of close correlation between rapid climate change episodes and changes in the archaeological record. Turning to the later Holocene, the review and discussion of Finné et al. (2011, 3163–3164, 3166) with regard to the 2200bc event is another good example: they highlight that this event is only clear in a few records, and not really present in others, and thus the event’s critical status may be over-played (see also Wossink 2009). Yet the same study points to another fundamental issue with respect to trying to investigate past climate regimes. The data they review are of very different types, qualities, and dating resolutions. Some of the differing if not contradictory findings which they note may indicate regional variations, but many reflect data derived from different types of archives formed by various (often unique) processes, with a range of complications (some natural, some likely anthropogenic, some a mixture of both in recent millennia). The end result is considerable noise, and either almost nothing historically useful is revealed or one is left with very general conclusions which all but overlook any shorter, even century-scale events, such as: 4600–1400 yrs bp: Drier conditions mainly dominate the climate picture of the eastern Mediterranean, but there are periods of increased moisture creating more benign climate situations at times, e.g. 1600–1200 yrs bp. (Finné et al. 2011, 3169) And one can note studies cited in the Finné et al. (2011, 3164) review which suggest contrary findings for some of this period. Further, in terms of the topic of this essay, concerning Rome and Climate, the 3200-year interval of ‘4600–1400 yrs bp’, considered as a vast whole, entirely fails to consider the shorter-term but potentially key changes, such as those in the 3rd–7th centuries ad argued to be of considerable relevance by the recent study of Büntgen et al. (2011). If there can be problems agreeing on instances of regionally-relevant major negative or stress episodes, then it is almost inevitable that it will be even harder to identify instances of potential climate opportunity for particular regions within the overall fairly benign Holocene era (where, with the exception of the 8200bp event—Rohling and Pälike 2005, there are no really dramatic rapid climate change episodes in contrast to the dramatic interglacial/glacial changes to be observed over the previous 100,000+ years). It is more likely we shall be able to identify episodes of climate 106 sturt w. manning stress (e.g. drought or substantial reduction/increase in temperature)—for example when a tree all but dies because of water deficit or cold—whereas the response to improved conditions is never quite so clearly represented (Hughes 2002, 103). Instances of positive conditions may well have only regional, versus hemispheric/global, impacts, and they may be expressed as general changes, or even a stable period, and not an easily spotted short- term spike (Caseldine and Turney 2010, 90). It is also important to remember that climate impacts are as relevant to given regions, and the same climate forcing may be both negative and also positive in different areas and under different circumstances. Despite the additional element of massive anthropogenic inputs in the last couple of centuries (making this now the Anthropocene era on Earth after Crutzen and Stoermer 20003), the differential impacts, perceptions and challenges of modern global warming around the Earth—and the ensuing debates and uncertainties—offer a good example of the multiple contemporary scenarios which can apply at different scales and in different regions, and the complications involved in any attempt to write a single large regional, hemispheric, let alone global, narrative (Watson et al. 1998; McCarthy et al. 2001; Parmesan and Yohe 2003; IPCC 2007; Cline 2007). In pre-modern times significant global warming and cooling events are another clear example of differential impacts according to scale and area, with the Medieval Climate Anomaly (MCA—see Diaz et al. 2011) and the Little Ice Age (LIA—generally, see Grove 1988; Fagan 2000) offering our best attested evidence. For example, at a large area and multi-century scale, the warming MCA sees a sustained period of a positive North Atlantic Oscil- lation (NAO, or Arctic Oscillation = AO4), and drier conditions in the west Mediterranean but wetter conditions in the eastern Mediterranean, and the cooling LIA sees the opposite with a sustained negative NAO phase with

3 Ruddiman (2003), however, argued that anthropogenic climate forcing started with the origins of agriculture. This hypothesis and the ensuing debate are discussed in a set of papers in a special issue of The Holocene (Ruddiman et al. 2011). Certini and Scalenghe (2011) make an attractive case for a date for the start of the Anthropocene about 2000 years ago (mid Roman period) when anthropogenic soils can first be recognized widely as a significant feature. 4 The AO (Arctic Oscillation) encompasses the NAO (North Atlantic Oscillation), and they are closely related. The AO is regarded as the more fundamental and wide-ranging climate structure by those who proposed the term (Thompson and Wallace 1998), whereas they regard the NAO as regional, and an historical accident deriving from the greater information available regarding the north Atlantic. I employ the term NAO primarily as it has greater currency in much of the literature cited, Mediterranean climate relates to the Atlantic, and it is almost indistinguishable from the AO and vice versa (Deser 2000). the roman world and climate 107 wetter conditions in the west Mediterranean and drier conditions in the east Mediterranean (Roberts et al. 2012; Kaniewski et al. 2011). But, within this macro pattern, there are then important regional and shorter time period subtleties (see also further discussion below). For example, the study of Shindell et al. (2001) nicely shows how the effect of a major cooling episode (solar driven) like the Maunder Minimum (one of the two grand solar minima of the LIA) plays out in quite different ways. Northern Europe cooled quite significantly, but the central and east Mediterranean and SW Asia were either stable or even slightly warmer, and especially over the winter period which is in turn key to crop growth in this region (see Shindell et al. 2001, Figures 1 right and 3). And despite the generally drier LIA multi-century period in the east Mediterranean, the specific 50-year period ad1700–1750 after the Maunder Minimum episode sees sustained increased precipitation across much of the Mediterranean including many areas of the east Mediter- ranean except part of central Anatolia (Luterbacher et al. 2006; Nicault et al. 2008). Such changes or non-changes are only evident according to the region and scale of observation and analysis. As the Shindell et al. (2001) paper highlights, despite the temperature drops in Europe, the average global temperature change through the Maunder Minimum period was in fact very small—thus the global or hemispheric average may have little bearing on observations in a specific region, and vice versa. Altogether, as the critique of Caseldine and Turney (2010) notes, we need attention to chronology and resolution, appropriate and often locally-based data, and new ways of integrating palaeoclimate data with historical and archaeological sequences and stories. In much of the Mediterranean, where lack of water is the key risk for agriculture and so for human food supply (see e.g. Garnsey 1988; Issar and Brown 1998; Mithen and Black 2011), precipitation response is the more vital statistic, and I concentrate mostly on water availability in this essay. For the recent period Hurrell (1995; 1996; Hurrell and van Loon 1997) demonstrated how the state of the North Atlantic Oscillation (NAO) relates to European- Mediterranean climate. During the period of instrumental records, the NAO has strongly influenced inter-annual precipitation variations in the western Mediterranean, while some eastern parts of the basin have shown an anti- phase relationship in precipitation and atmospheric pressure (see also Trigo et al. 2000; Xoplaki 2002, Figure 6.1; Hughes et al. 2001, esp. p. 71; Roberts et al. 2012). In a recent study, Roberts et al. (2012) explored how the NAO and other atmospheric circulation modes operated over the longer timescales of the MCA and LIA using high-resolution palaeolimnological evidence from opposite ends of the Mediterranean basin, supplemented by other 108 sturt w. manning palaeoclimate data. Iberian lakes show lower water levels and higher salin- ities during the 11th to 13th centuries ad synchronous with the MCA, and generally more humid conditions during the LIA (15th–19th centuries ad) (Roberts et al. 2012). This pattern is clearly evident in tree-ring records from Morocco (Esper et al. 2007) and from marine cores in the western Mediter- ranean Sea. In the eastern Mediterranean, palaeoclimatic records (Turkey, Greece and the Levant: see Roberts et al. 2012 and references therein) indicate generally drier hydro-climatic conditions during the LIA and a wetter phase during the MCA (Roberts et al. 2012). This implies on longer timescales that a bipolar climate see-saw has operated in the Mediter- ranean for the last 1100 years. However, while western Mediterranean aridity appears consistent with a persistent positive NAO state during the MCA, the pattern is less clear in the eastern Mediterranean (Roberts et al. 2012). Results indicate that the long timescale LIA/MCA hydroclimatic pattern in the Mediterranean was determined by a combination of different climate modes along with major physical geographical controls, and not by NAO forcing alone, or, alternatively, that the character of the NAO and its teleconnections have been non-stationary (Roberts et al. 2012, 30–31). But perhaps the most striking observation to come from the Roberts et al. (2012) paper is that there appears to be a key difference according to the timeframe of study. They observe (2012, 30–31) that on shorter timescales (annual to decadal) there is in fact no opposition between the west Mediter- ranean (Iberia) and the east Mediterranean (Anatolia), and that studies have usually found a general correlation of a negative winter NAO (and especially a negative AO—see Unal et al. 2012, 402–403) with increased precipitation in the Aegean region and western and some other areas of Anatolia (but less so eastern and southeastern Anatolia and on the Black Sea coast) and other parts of the east Mediterranean, and the reverse (e.g. Unal et al. 2012 and references cited therein; Türkeş and Erlat 2003; Xoplaki 2002, Figure 6.1; Hughes et al. 2001, esp. p. 71). It is only when the timeframe of observation and analysis moves to century and greater scale that the opposite pattern of general aridity then links with the sustained overall negative NAO period of the LIA in the east Mediterranean (contrary short-term associations of opposite character), and the generally positive NAO period of the MCA links with overall increased precipitation in the east Mediterranean. Hence there are two different patterns depending on the timeframe of study. We may plausibly assume a fairly similar regime for the later Holocene, roughly the last 6000 years, and in contrast to the earlier Holocene when evidence indicates a wetter (and so different) climate regime for the Mediter- ranean (Robinson et al. 2006; Brooks 2006; Wanner et al. 2008; Black et al. the roman world and climate 109

2011). However, the appreciation of a differential relationship between the NAO index and precipitation in the Aegean-Anatolia-east Mediterranean region according to the timescale of observation/analysis, creates two possible scenarios and so may affect some previous studies which have simply assumed a (single) stable linkage between negative NAO and precipitation for the east Mediterranean, such as that by Lamy et al. (2006), who found (or assumed) a correlation between drier conditions in the east Mediterranean and a positive NAO, and between wetter conditions in the east Mediter- ranean and a negative NAO (or AO). Over the decadal scale, yes, but at the longer century scale then perhaps no following the findings of Roberts et al. (2012), or Kaniewski et al. (2011). And between the opposite shorter-scale and longer-term NAO associations, a variety of other forcings may also be relevant (Roberts et al. 2012, 30–31). In particular, we may note again that, despite a generally cooler, drier LIA and overall more negative NAO across this period (Trouet et al. 2009), during the half century (ad1700–1750) which followed the Maunder grand solar minimum (ca. 1645–1715)—the major cooling episode of the past five centuries—Luterbacher et al. (2006) and Nicault et al. (2008) observe a significant switch to wetter conditions in the Mediterranean generally (except parts of central-east Anatolia), and Nicault et al. (2008, 240) find this half-century to form the longest stable wet period observed in their 500-year record. And, generalizing from the Lamy et al. study (2006, esp. Figure 8 and pp. 8–10), there is a plausible association of wetter conditions in the periods around major solar minima. Figures 1a and 1b illustrate some of the patterns, complications, and contradictions from a hemispheric scale to a regional scale (west versus east Mediterranean). The warmer v. cooler MCA v. LIA sees a general shift in the Palmer Drought Series Index values from drier to wetter in Morocco (southwest Mediterranean), but the precipitation record from the east Mediter- ranean (Touchan et al. 2005) offers often an anti-correlation versus the Moroccan record (see also Figure 15 below). Comparison of the reconstructed winter NAO by Trouet et al. (2009) and Cook et al. (2002) versus some decade or longer periods of more negative (drier) PDSI or more positive (wetter) PDSI observed in the analysis of Mediterranean tree-ring series ad1500–2000 by Nicault et al. (2008) for Italy, Greece and the Levant in Figure 1b further highlights differences and patterns. The longest period when all these regions (and much of the Mediterranean) enjoyed positive PDSI values and increased water availability, ad1700–1750, correlates neither with a noticeable positive or negative NAO; indeed, it is if anything a relatively stable period (in both the Trouet et al. 2009 and Cook et al. 2002 reconstructions), and, as noted above, it follows 110 sturt w. manning

Fig. 1a. Comparisons of general northern hemisphere temperature covering the past millennium (MCA and the LIA and recent warming period) and the often differing (almost opposite) precipitation records from the west and east Mediterranean. Top: Temperature reconstructions for the extra-tropical Northern Hemisphere for the last 1000 years from a multi-proxy synthesis by Christiansen and Ljungqvist (2011) and for central Greenland from the GISP2 ice-core based on stable isotope analysis and ice accumulation data (Alley 2000; Cuffey and Clow 1997). Bottom: Palmer Drought Series Index (PDSI) reconstruction from Morocco from tree-ring data by Esper et al. (2007)—more negative values = drier, positive = wetter; and reconstructed precipitation in the eastern Mediterranean from tree-ring data by Touchan et al. (2005). Data are shown with 10 and 40 point (pt) (= years in this case) FFT (Fast Fourier Transform) smoothing (which removes the high frequency noise).5

5 Data for top from: ftp://ftp.ncdc.noaa.gov/pub/data/paleo/contributions_by_author/ christiansen2011/christiansen2011.txt and ftp://ftp.ncdc.noaa.gov/pub/data/paleo/icecore/ greenland/summit/gisp2/isotopes/gisp2_temp_accum_alley2000.txt. Data for bottom from: the roman world and climate 111

Fig. 1b. A comparison of periods noted in the analyses of Nicault et al. (2008) of decadal or longer intervals of wetter (more negative PDSI) and drier (more positive PDSI) for Italy, Greece and the Levant versus reconstructed winter NAO indices from (at bottom) Cook et al. (2002) and (middle) Trouet et al. (2009). The annual Cook et al. (2002) reconstructed data (light grey) are shown with a 5 year running average (bold black line). The Trouet et al. (2009) data as published have had a 30-year smoothing (spline) applied to them.6 the Maunder Minimum solar activity episode. The other common wetter period, ad1870–1900, is associated with a more positive NAO in the Trouet al. (2009) record (and the Cook et al. 2002 record includes a major positive NAO episode in the early ad1880s also). At other times there are marked regional differences. Wetter periods in Greece or Italy can be the opposite

http://www.ncdc.noaa.gov/paleo/metadata/noaa-recon-8712.html and http://www.ncdc .noaa.gov/paleo/metadata/noaa-recon-6381.html. Note: all webpages cited in this paper were accessed in January–February 2012. Sources of data are only cited once even if employed several times. 6 Cook et al. (2002) data from: ftp://ftp.ncdc.noaa.gov/pub/data/paleo/treering/ reconstructions/nao_cook2002.txt and the Trouet et al. (2009) data from: ftp://ftp.ncdc.noaa .gov/pub/data/paleo/treering/reconstructions/nao-trouet2009.txt. 112 sturt w. manning in the Levant, and the reverse. Hence some observed apparent contrasts in precipitation histories for the Roman period between the western and central Mediterranean and for the northern coasts of the Mediterranean, versus the Levant, which we will encounter in later discussions, become plausible regional phenomena.

8th Century bc Greek Renaissance: Sun, Climate Change and Patterns

Such complexities and contradictions make the attempt to link climate change closely with the ancient historical and prehistoric records even more challenging. There will always be exceptions, and different interpretations focusing on different elements will very easily reach diverse conclusions. Eighth century bc Greece offers an interesting but salutary case. In big picture terms, there is a major solar minimum dated ca. 765bc (Usoskin et al. 2007)7—arguably the largest of the past 3000 years—and this is plausibly linked to climate change observed widely around the Earth in the 9th to 8th centuries bc (e.g. van Geel et al. 1996; 1998; 1999; 2000; 2004; van Geel and Renssen 1998; Speranza et al. 2002; Blaauw et al. 2004; Hall et al. 2004; Chambers et al. 2007; Swindles et al. 2007; Plunkett and Swindles 2008; Arnaud-Fassetta et al. 2010, 108; Wang et al. 2012). Often cited as the 2800 years bp event (or, in Europe, as e.g. the Iron Age Cold Period, or the Ini- tial/First Phase of the Subatlantic—a cooler time marked, for example, by glacier advances in the Alps: Wanner et al. 2008; Holzhauser et al. 2005; Haas et al. 1998; Denton and Karlén 1973), this forcing can in fact be observed over a period of time. This is because in some locations there will be what Plun- kett and Swindles (2008) describe as ‘time-transgressive climate responses’ over the following several decades to about a century. With regard to the Mediterranean, our discussions above indicate in general terms that such major solar minima and cooling episodes (e.g. Bond et al. 2001) seem to link with high storm activity (negative NAO) in the north Atlantic and high storminess and precipitation in the northwest Mediter- ranean (Sabatier et al. 2012) and increased precipitation and wetter condi-

7 This is the date for the peak of the episode; Usoskin et al. (2007) estimate the duration at 90 years. The Usoskin et al. (2007) dates are derived from the radiocarbon record. The recent Beryllium-10 analysis of Steinhilber et al. (2009) places the peak minimum around 817bc and a sustained period of very low solar activity around 830–740bc (see further discussion on radiocarbon and Beryllium-10 and solar proxy records in the text below). the roman world and climate 113 tions in the Black Sea and northern Red Sea and so wider east Mediterranean region (Lamy et al. 2006).8 Although, looking at the 2nd millennium ad, prolonged periods of such conditions (i.e. the LIA) in general lead to drier conditions in the eastern Mediterranean (Roberts et al. 2012), and, especially we might predict, in the cooling period leading to a solar minimum (and, for example, Crete appears notably cool and dry in the mid-16th to mid-17th centuries ad heading towards the Maunder Minimum: Grove 2001), we can also observe that on a shorter-term basis the opposite may be the case, and especially in the period following the solar minimum (as warming occurs). The ca. 50-year period (ad1700–1750) following such a major solar minimum (the Maunder Minimum case) saw increased precipitation and thus better growing conditions in the Mediterranean (Luterbacher et al. 2006; Nicault et al. 2008). And, on the short-term in the instrumental period, there is some correlation of a negative NAO with increased precipitation in the Aegean region (Xoplaki 2002, Figure 6.1; Hughes et al. 2001; Unal et al. 2012).9 Thus, in macro-terms, such a major solar minimum should/could see increased aridity in the period leading into the minimum (and, for example, Kaniewski et al. 2011 find an increased aridity trend about 800–750bc for coastal Syria), but should/could bring in the following period generally moister conditions in much of the Mediterranean (so longer and more reliable growing seasons), as observed for the period following the Maun- der Minimum (Luterbacher et al. 2006; Nicault et al. 2008 and references therein), and as a variety of palaeoenvironmental records from the west (especially Esper et al. 2007; Roberts et al. 2012 and references therein), central, and east Mediterranean suggest for the period after about 750bc (e.g.

8 Magny et al. (2003) suggest an approximate correlation of the Bond et al. (2001) cooler episodes (and reduced solar activity) with drier phases in Lake Siles in southern Spain. But the chronology seems both critical to a correct interpretation, and less than precise. For the 2800bp event, in particular, the dry phase at Lake Siles is in fact shown as ending just before 2800bp (see Magny et al. 2003, Figure 3). Thus we might assume that the period from after 2800bp was not especially dry. Magny et al. (2003) indicate that a multi-century wet phase in the central European lakes begins following 2800bp. 9 The exceptions to this generalisation, of course, could be plant species and contexts in generally fairly arid areas which are susceptible to cold. For example, Ferrio et al. (2006, 1264) suggest that cooler temperatures may lead Pinus halepensis trees in Iberia to increase their reliance on summer precipitation, and, if this was poor, they would seem to record water stress, and so arid conditions, even if there was in fact generous winter precipitation and ‘good’ conditions in agricultural areas and for crops which were not especially intolerant of cold. Ferrio et al. (2006) suggest this may explain why their study of δ13C from archaeological charcoals indicates an arid period ca. 900–300bc whereas other evidence including for Spain (as reviewed in this paragraph) generally indicates otherwise: a cooler and wetter period. 114 sturt w. manning

Neuman 1985; Heim et al. 1997; Lemcke and Sturm 1997; Issar 2003, 24; Drysdale et al. 2006; Lamy et al. 2006; Martín-Puertas et al. 2009; 2008; Martín-Chivelet et al. 2011; Kaniewski et al. 2011; Sabatier et al. 2012; Roberts et al. 2012).10 Vita-Finzi (2008) has further argued that the Mediterranean should receive an increase in small, non-erosive, rainfall (during winter and spring) associated with such a period of major solar cooling, and this too should be beneficial for agriculture.11 In sum, all these conditions should provide a more positive regional context for agriculture (and especially contrasted to the indications of aridity in the preceding period from the 12th to the 8th centuries bc in the eastern Mediterranean: e.g. Kaniewski et al. 2010; 2011; Drake 2012), and so be good for human populations in this region (as for example argued by van Geel et al. 2004 with regard to the Scythian culture), and especially for those in the lower elevation and more arid-risk regions, such as Athens and central Greece (Manning 2010, 41–44). And so, there seems a potential climate context of relevance for the indications of dramatic population growth and concomitant demographic, social, economic, political and material culture change (labelled the ‘structural revolution’ by Snodgrass 1980, 15–84, or the ‘Greek Renaissance’ by Coldstream 1977)—including the phenomenon of Greek colonization—in the Greek world. This starts in the period ca. 750–700bc (e.g. Snodgrass 1977; 1980; Hägg 1983; Tandy 1997, 20–43; Morris 1998). Morris (1998, 75 and n. 118) further notes that ‘[i]n the eighth and seventh century, the whole Mediter- ranean basis experienced substantial population growth’, and observes that ‘[e]xplaining what happened in eighth-century central Greece will require both a general account of the sudden increase in the pace of life across the whole Mediterranean and a series of particularized accounts of what made each region unique’. From the previous paragraph, we might suggest that climate change forced by a grand solar minimum offers part of the ‘general account … across the whole Mediterranean’. However, other scholars postulate severe drought at the very same time (Camp 1979), and/or instead analyse the ancient sources to suggest the evidence is more concordant with the view that colonization, and the appar-

10 Some (but not all) records from the southern Levant indicate a relatively arid climate at this time (Rambeau and Black 2011, 99), and this may reflect the differences between the conditions for increased precipitation in the western Mediterranean and across the northern Mediterranean to the Aegean, versus in the southern Levant (see Black 2011). 11 See also Devillers and Lecuyer (2008), who in their Cypriot Little Ice Age case study find evidence of overall, or net, agriculturally positive change with increased water availability and moderate-scale floods, and a decrease in highly erosive flash floods. the roman world and climate 115 ent evidence of population growth, is a response to climatic disaster (e.g. drought) and other historical factors (Cawkwell 1992), rather than to marked population growth or favourable conditions. Finné et al. (2011, 3168) conclude that the palaeoclimate data they surveyed do not in fact show this event in the east Mediterranean, and, entirely contrary to the hypothesis in the previous paragraphs, a recent paper by Drake (2012) claims that the Iron Age recovery in Greece and the rise of Athens do not appear to be climate related. This is on the basis of Drake’s assessment of the limited and rather coarse information he employed, to the effect that both precipitation and sea surface temperatures (SST) continue to be low.12 The point, however, for this discussion is not who is right or wrong—rather how the same case can be and is differently interpreted. There are no easy generalizations in this field; we also observe the importance of chronology if specific correlations are to be made (also Maher et al. 2011; Rohling et al. 2002, 590–592).

Aims and Limitations

There are a number of works which address aspects of Holocene climate in considerable detail: e.g. Roberts (1998); Issar and Brown (1998); Issar (2003); Roberts et al. (2004; 2008; 2012; Roberts, Brayshaw, et al. 2011); Thompson

12 Some of Drake’s data plots, such as for paleo-rainfall in his Figure 2, are coarse resolution to the point of not being useful, whereas several much higher resolution records indicate the opposite of his assessment: see Figure 17 below; and, in fact, 3 of his 4 SST plots in his own Figure 3 indicate a rise in SST—contrary his assessment—by or from the earlier 1st millennium bc—this includes the data from Emeis et al. 2000, Figure 4.c from both the Levantine Sea and the Ionian Sea. The study of Rohling et al. (2002, Figure 1d), in particular, clearly shows the temperature trend in the Aegean Sea (from % warm species of foraminifera: Rohling et al. 2002, 589) as in fact rising through the early 1st millennium bc (from a cool period late in the 2nd millennium bc)—and, given that the dates for this record appear to be around 300 years too old by the mid second millennium bc (see Rohling et al. 2002, 590–591), the period of increasing temperature shown as ca. 3300–2800 Cal bp should in fact likely be dated as somewhat more recent, and so the rising temperature period will continue at least through the 8th century bc and perhaps even a little further. As noted in the text above, the potential for some warming in the specific east Mediterranean region during a wider general global cooling episode is entirely plausible based on the findings of the Shindell et al. (2001) study. Elsewhere, a fall in SST around the time of the grand solar minimum in the 8th century bc makes general sense (and, for example, the Sargasso Sea record indicates cooling in the early 1st millennium bc from a warm peak in the 10th century bc, until a sharp warming trend in 7th century bc, and then generally stable fairly warm conditions through to the 2nd century ad: Keigwin 1996). The available evidence is consistent with moister and more agriculturally favourable conditions in the Aegean region from the mid 8th century bc onwards following the 9th–8th century bc global cooling episode. 116 sturt w. manning et al. (2002; 2006); Mayewski et al. (2004); Staubwasser and Weiss (2006); Robinson et al. (2006); Rosen (2007); Wanner et al. (2008); Luterbacher et al. (2006; 2012); Finné et al. (2011); Rambeau and Black (2011). The overall trends in climate-environment in the Mediterranean through the Holocene are approximately clear (Roberts et al. 2012). There is something of an east- west division and see-saw relationship. The eastern Mediterranean saw a change to less precipitation around 6000 years ago (mid-Holocene—in line with a widely represented trend: Brooks 2006; Wanner et al. 2008, 1795; Ram- beau and Black 2011), whereas the west saw less dramatic changes generally, but maximum increased precipitation occurred in the mid-Holocene 6000–3000 years ago, before a shift to increased aridity leading to present day levels. For the Roman period, in particular, a recent paper by McCormick, Bünt- gen et al. (2012) reviews a number of the sources (scientific and historical) of data for climate, and climate change, relevant to the Roman world 100bc to ad800. See that paper for a first effort at a respectably comprehensive review. I have been much more selective here. My aim is to consider some of the general climate parameters for the wider historical trajectory of the Roman world. I focus on those sources of information which are indisputably relevant to climate, and not those likely to be affected by anthropogenic activities (so natural archives which can be linked to climate forcing, versus secondary records such as pollen sequences, etc., which can result from both or are a mixture of natural or human activities), since Roberts, Brayshaw, et al. (2011) conclude that by the mid-1st millennium bc human-induced landscape changes and activities were a significant element of the Mediterranean environment (and Certini and Scalenghe 2011 argue this case from about 1ad for several areas of the Earth). These impacts range from agricultural activities, to deforestation, to industrial activities.13

13 The issue of human impacts on the natural landscape (e.g. Goudie 2006; Redman 1999) is fraught and complex in the Mediterranean. Whereas some claims of dramatic impacts can be critiqued (e.g. Horden and Purcell 2000, 318, 324–326), and Grove and Rackham 2001 argue a general thesis that human activity has at most merely abetted climate change (but cf. Butzer 2003; 2005), there is also undeniably a major human element involved (we can side-step the issue of prime causality) in the environmental record of the Mediterranean region from several thousand years ago (e.g. Bottema et al. 1990; Butzer 2005; Eastwood et al. 1998; Leng et al. 2010; Luterbacher et al. 2012). On trees and timber in the ancient world generally, see Meiggs (1982), who did not view the ancient world as responsible for massive deforestation (in contrast to the 19th and 20th centuries ad); the recent studies of Harris (2011) and Harris (this volume) argue for a mixed responsibility. Even the general assumption of modern deforestation needs care, since in some areas recent rural depopulation and the roman world and climate 117

I have also not considered historical sources in this essay; appropriate use of such data requires its own study and methodological approaches.14 How the Roman world interacted with, and was impacted by, climate and possible climate change, is an obvious subject for interest and speculation. Climate does not cause or create history in some reductionist paradigm; but it does provide a context for human society, lives, interactions and decisions, and potentially provides opportunities or challenges which may influence both small scale and macro scale human history. Indeed, recent attention and reconsideration of the dynamic and recursive role landscape and its perceptions and projections play in producing and maintaining human social, political and economic organization (e.g. Smith 2003; Falconer and Redman 2009), only further foreground the relevance of climate and its part in the formation of the overall social landscape and environment context for any archaeological or historical study. Hence climate is of considerable relevance—and yet it has if anything been rather overlooked in scholarship on the Mediterranean until quite recently (as noted by e.g. Broodbank 2011, 31). More particularly, in light of fairly sweeping and general claims made in recent papers such as Büntgen et al. (2011)—where climate change is, more or less, suggested to be a primary driver in several key historical changes such as the decline of the Roman world (and previously see papers such as Perry and Hsu 2000 for even grander scale linkages of climate and human development)–I wish to assess some aspects of what we know and what we do not know.

changes in agricultural strategies have led to rapid re-forestation (Debussche et al. 1999). Nonetheless, there clearly were impacts on the natural landscape—whether through human silviculture (e.g. Conedera et al. 2004) or the harvesting of natural forests, as evident from the requirements of the pyrotechnical industries of the Bronze Age through Roman periods such as metallurgy, and, in the Classical era, the production of bricks and lime-cement, apart from needs for building and so on (Wertime 1983; Horden and Purcell 2000, 334–336; Meiggs 1982). The Roman period sees the first evidence of large-scale pollution of the environment and atmosphere from industrial-type anthropogenic activities (versus instances of burning of landscapes), in particular, there is evidence of widespread lead and other metals (mercury, copper) pollution from metallurgical activities (e.g. Mighall et al. 2009; De Vleeschouwer et al. 2007; Renberg et al. 2000; Shotyk et al. 1998; Martinez-Cortizas et al. 1999; 1997; Hong, Candelone, Patterson and Boutron 1996; 1994). 14 There is enormous potential in the extant textual record. For some discussions of such material, see e.g. Bonneau 1971; Hassan 1981, 2007; Neumann 1985; Garnsey 1988; Sallares 1991; Stathakopoulos 2000; 2004; Teleles 2004; Kondrashov et al. 2005; Little 2006; Lehoux 2007; McCormick et al. 2007; McCormick, Büntgen et al. et al. (2012). In particular, see the Digital Atlas of Roman and Medieval Civilization (McCormick et al. 2010 at http://darmc.harvard .edu/). 118 sturt w. manning

The key issues are the types of data, the scale, accuracy and precision of resolution, and the approaches taken. At present we can still begin by say- ing that we need (many) more high-resolution datasets and more regional studies to be able to advance understanding of human-climate relationships in specific terms (versus very general associations) for the Mediterranean region.15 The recent review of a range of climate data available from geological sources for Italy by Giraudi et al. (2011) is a good example both of how far we have come—namely there are quite a lot of data—but also of these problems, since the data reviewed offer some general long-term trends, and some specific insights (like picking up major multi-regional events such as the event at 4200 years bp), but no easy synthesis into a coherent high-resolution story for Holocene, let alone Roman, Italy. Chronological resolution and/or control is a, or the, key missing element in many cases. The excellent, thorough, but, ultimately, rather uninformative synthesis of Finné et al. (2011) on the last 6000 years of climate in the east Mediterranean is a similar case. Of over 80 data sources reviewed in that paper, only 3 approach high-resolution (that is around less than 10 (or even less than 20) calendar years resolution), and of these the Oman speleothem data do not relate to Mediterranean climate,16 leaving just two directly relevant high (1–5 years: Jones et al. 2006) or near-high (2–40 years: Kaniewski et al. 2007) resolution records from Turkey. In many other cases, the underlying chronologies are much less precise (with precisions of decades to several centuries), such that records are coarse and lacking the necessary detail to allow correlations with history and archaeology,or it is not even really certain exactly where key features are placed, with possible movement of decades to centuries potentially allowing entirely different syntheses and scenarios. Much better chronological resolution and control are necessary. However, this situation raises another fundamental topic: what type of resolution and what type of climate data are ancient historians and archaeologists seeking to uncover and address? Their goals are not necessarily the same as those of the palaeoclimate community. While sometimes as historians we may seek the specific year—a climate record linked to a key famine, for example—more often we actually want the trend on a more

15 The statement of Rosen (2007, 153) remains largely true: ‘… we still do not have fine enough resolution in the dates of climatic data to be able to clearly mesh climatic fluctuations with precise historical events in the Roman-Byzantine period’. 16 Instead they primarily reflect the movements of the Intertropical Convergence Zone, ITCZ, and the Indian Summer Monsoon (Fleitmann et al. 2003; 2007). the roman world and climate 119 decadal to century-level basis (but with precision on the time period this trend applies to). We accept and assume year to year variability whatever the climate regime (as observed in the recent instrumental era17 and our own lives). What we want to know is whether the trend over a decade to several decades to several generations (so to 100 years or so) is generally positive in terms of key resources (e.g. agriculture), or the reverse, or stable; this sort of trend can then act as an underlying forcing parameter which could shape sets of human decisions and actions. This does not have to be as sensational as a ‘mega-drought’ (e.g. Cook et al. 2004; 2010)—though this is just the sort of palaeoclimatic forcing data historians and archaeologists keenly want to know about for their regions of interest (e.g. Buckley et al. 2010). In palaeoclimate terms historians are more focused on low-frequency variations (the trends over multi-year and longer periods) and not the high-frequency variations year-by-year. Yet this has not always been the aim of palaeoclimate research. Tree-ring studies, despite offering high-resolution data, are an especial case in point. As Finné et al. (2011, 3164, 3168) note, most tree-ring studies so far in the Mediterranean region have employed statistical methods which in the effort to remove noise also remove trends in favour of year to year variation (e.g. Hughes 2002, 102–104). Thus good tree-ring studies can identify specific positive or negative years in terms of tree-ring growth, and, where the sources are available, these can sometimes be linked with historical records of, e.g., drought or high precipitation (e.g. Köse et al. 2011), but, as Finné et al. (2011, 3164, 3168) lament, many of the available tree-ring studies from the Mediter- ranean do not in fact pick up such obvious trend events as the MCA or the LIA … Clearly this is a serious short-coming. Of course scientists in the field are well aware of this issue, and considerable effort has gone into developing methods (such as Regional Curve Standardization, or RCS) to try to extract the low frequency signals from tree-ring series, and to overcome such limitations of these data (e.g. Esper et al. 2002; 2003; 2007; Briffa 2000; Briffa et al. 2001; Frank et al. 2007; Hughes et al. 2011). In the western Mediterranean

17 The period for which we have instrumental data on climate is short. I refer loosely to the last couple of centuries. The longest continuous temperature record, from central England, dates from ad1659 (monthly data) and ad1772 (daily data) (Manley 1974; Parker et al. 1992): data and further bibliography available from http://www.metoffice.gov.uk/hadobs/hadcet/. Despite some sporadic instrumental records from the 17th century onwards, national meteorological services were only established widely in Europe and the Mediterranean in the mid-19th century (Luterbacher et al. 2006; 2012), and there are many areas where meteorological records only begin in the second half of the 20th century. 120 sturt w. manning

Esper et al. (2007) employed both a large data set focused on long-lived trees, and a range of approaches, and were able to recognize trends indicative of the MCA and the LIA. Similar work is now needed elsewhere in the Mediterranean. The tree-ring study of Büntgen et al. (2011), though outside the Mediterranean, is also relevant to the Roman world. We can also hope for more high-resolution data sets from speleothems in coming years; in the right circumstances these can offer good correlation with instrumental data comparable to tree-rings (see Jex et al. 2010 and Göktürk et al. 2011 for such very high resolution studies in the recent period from Turkey).18 As one focuses down in scale to the region or site, there is the important added complication of the human subjects, and their role in the environment under study, especially for the later Holocene. Rosen (2007, 150–171) nicely highlights some of the issues and problems (and the potential for circularity in argument) in her review of the interpretations of evidence of increased and widespread human activity in the Roman-Byzantine period in arid areas like the Negev. She points out that climate is only one factor in historical explanation (see also Horden and Purcell 2000, 298–341).19 We cannot write climate history on the basis of the record of human activities alone, or vice versa.

What the Sun Was Doing in the Roman Period

For the Roman period, ca. 300bc to ad800, the main general (background) forcing factors on climate which operate at this scale are the Sun and volcanic eruptions. Volcanic activity relevant to the northern hemisphere seems to have been notably low for much of the Roman period, and certainly for the ca. 570 years from ca. 35bc to ad535: see Figures 19a and 19b below. This is an important absence of evidence and is noted further below. Here I therefore discuss the Sun.

18 There is however an important methodological issue of relevance here. There is a major difference between a record which is demonstrably annual and absolute (for example a robust securely crossdated tree-ring chronology) and a record which is only approximate (however high resolution). As Betancourt et al. (2002) argued, high-resolution speleothem records are not absolute or annual resolution unless they can achieve the robustness levels of tree-ring chronologies, and should be described instead as ‘near annual’ or ‘subdecadal’. 19 Barker (2002) offers another relevant critique, reviewing two cases (Tripolitanian pre- desert in Libya and the Wadi Faynan in Jordan) where marginal arid environments (then as now) were transformed by systems of floodwater farming through the centuries of Roman imperialism. The archaeological evidence thus reflects primarily economic and political trajectories (each very different), rather than climate. the roman world and climate 121

A linkage between variations in solar activity and the Earth’s climate became apparent with the examination of records of sunspot activity. Schol- ars such as Eddy (1976; 1977) observed an association between times of few or no sunspots (and so low solar activity) and cooler temperatures and climate anomalies on Earth, and the reverse, and Magny (1993)20 argued that solar variations were associated with (and were a main cause of) variations in European precipitation. The key example where there are historical data is the Maunder Minimum (about ad1645–1715). There was little or no observed sunspot activity—the open solar flux is calculated to have been a mere 11% of that observed during the very recent grand solar maximum (Lockwood and Owens 2011)—and this period is associated with some of the colder temperatures of the last four centuries on Earth (Bradley and Jones 1993). Hence for some time it has been agreed that solar forcing was the predominant factor in the observed cooler 17th and 18th century ad temperatures (e.g. Lean and Rind 1998)—although this may be especially, or particularly, in the case of Europe, where most of the historic Maunder Minimum-related data come from, based on analysis of both historic and recent data (Lockwood, Harrison et al. 2010; Luterbacher et al. 2001). In the pre-instrumental era there is more debate over the exact mechanisms, and relations are complicated, but again there is now generally agreed to be a clear correlation of the observed palaeoclimate record with changing solar activity (e.g. Lockwood 2006; Bond et al. 2001; Neff et al. 2001). Recent work has also started to offer plausible models to explain how changes in solar activity may force climate on Earth—a previously debated point (e.g. Shin- dell et al. 2001; Gray et al. 2010; Ineson et al. 2011; Lockwood, Bell, et al. 2010; Hegerl et al. 2011). Without considering the many complexities of this topic (see e.g. Gray et al. 2010 for a review), we can regard the changing activity of the Sun as a key climate parameter—with low (high) solar activity associated over the Holocene with low (high) temperatures—and general associations between solar activity, climate, and human historical trajectories have been noted by many scholars (e.g. Perry and Hsu 2000). Some cosmogenic isotopes—14C (radiocarbon) and 10Be (Beryllium-10)—provide approximate proxy records of changing solar activity for the period before historical or instrumental data (i.e. for before ad1600) (Vonmoos et al. 2006).21 The 14C record from

20 Recently, with further literature review, see Magny et al. (2010). 21 For discussion of the production of these cosmogenic isotopes, see Masarik and Beer (1999). 122 sturt w. manning known-age tree-rings offers an absolutely dated and highly resolved record of changing atmospheric 14C levels through the Holocene for the Northern Hemisphere: Figure 2. Although affected by other factors (e.g. changes in the oceans, or changes in the geomagnetic field), this record of changing 14C in recent millennia largely reflects changing solar activity (e.g. Stuiver and Quay 1980; Stuiver and Braziunas 1988; 1993; 1998; Stuiver et al. 1991). Given various assumptions about the carbon cycle, it is in turn possible to model 14C production as (largely) a contemporary proxy of changing solar activity (e.g. Usoskin and Kromer 2005; Siegenthaler et al. 1980). Figures 3–4 show how the changes in the 14C record, and the modeled production record of 14C, broadly reflect the activity of the Sun as reflected in observed sun-spot numbers (SSN) over the last 350 years. A model transforming such 14C production estimates into a total solar irradiance (TSI) reconstruction has been published recently by Vieira et al. (2011): see Figure 6a. 10Be records from (principally) polar ice-cores offer the other main long- term solar proxy record. Over recent centuries these have annual resolution, and there are only very small calendar errors to beyond the Roman period. As an example, Figure 5 shows two 10Be records for the recent centuries versus both observed sun spot numbers and the 14C record. Beer et al. (1990) demonstrated that the annual resolution Dye-3 record closely recorded solar activity and especially the 11-year solar cycle. A number of studies have demonstrated the good comparison of 10Be and 14C records (e.g. Beer et al. 1988; Bard et al. 1997; Beer 2000; Beer et al. 2006; Muscheler et al. 2007). The advantage of 10Be is that it is deposited at the Earth’s surface more quickly and directly after production in the atmosphere (typically 1–2 years) compared to 14C (typically a 15–20 year lag is expected comparing 10Be to 14C: Stuiver and Braziunas 1993; Bard et al. 1997), and it is not damped by the carbon cycle or complicated by changes in it (in contrast with 14C). However, the problems with 10Be are twofold. First, the 10Be records contain considerable noise reflecting transport (and associated mixing) and then deposition processes (local weather conditions: precipitation, wind speed), and so it can be difficult to disentangle climate-driven versus solar-driven changes (e.g. Field et al. 2006). Such issues have been reviewed by Heikkilä et al. (2011), who nonetheless find that 10Be offers a useful record when treated appropri- ately. Second, the current precision and accuracy of ice-core dating becomes less certain backwards over time (Baillie 2010), and various anomalies are clearly evident at decadal and greater levels over some periods, such as in the second millennium bc, indicating dating or other issues in various key records (e.g. Southon 2002; 2004; Muscheler 2009—see further in footnote 24). A recent record of modelled total solar irradiance (TSI) from 10Be for the roman world and climate 123

Fig. 2. A. The standard radiocarbon (14C) calibration curve for the period 3000bc to ad1950 from known age trees (IntCal04: Reimer et al. 2004—the tree-ring based 14C data are the same as in the more recent IntCal09: Reimer et al. 2009). INSET: The inset shows a section (1000bc to 500bc) of the standard 14C calibration curve (IntCal04) in more detail (1σ error band)—note the characteristic ‘wiggly’ shape. B. The Δ14C record per mille (‰) from A—this is the relative 14C content decay corrected and normalized (for the definition of this, see Mook and van der Plicht 1999): data from Reimer et al. (2004). C. The residual Δ14C record per mille (‰) after a 1000-year moving average is removed—this record thus highlights the relative changes in the Δ14C record around the underlying longer-term trend: data from Reimer et al. (2004). Note: the stated errors on lines A, B and C are not shown for reasons of clarity and scale.22

22 Radiocarbon data files are available from: http://www.radiocarbon.org/IntCal09.htm and http://www.radiocarbon.org/IntCal04.htm. 124 sturt w. manning

Fig. 3. Bottom: observed sun-spot numbers (SSN) per year. The marked declines reflect two known cooler climate episodes: the Maunder Minimum and the (much smaller) Dalton Minimum. The recurrent spikes in SSN reflect the well-known approximately 11-year cycles in solar activity (and a longer 22-year cycle). The SSN data from historical records of sun spot numbers come from Hoyt and Schatten (1998a; 1998b). Top: The annual Δ14C record per mille (‰) ad1600–1900 (Stuiver et al. 1998) (grey) and an 11-year moving average of this record (black). Middle: The Δ14C record per mille (‰) from IntCal09 and IntCal04 (Reimer et al. 2009; 2004) and two models of 14C production: (a) from the MarMod09 model in Reimer et al. (2009); and (b) from the iterative model in Usoskin and Kromer (2005). No errors shown for lines in Top or Middle plots.23

23 The SSN data from historical records of sun spot numbers come from Hoyt and Schatten (1998a; 1998b) and are available from http://www.ngdc.noaa.gov/stp/SOLAR/ ftpsunspotnumber.html. The annual Δ14C record per mille (‰) ad1600–1900 (Stuiver et al. 1998) is available from http://depts.washington.edu/qil/datasets/. The iterative 14C production model data (from Usoskin and Kromer 2005) were kindly provided by Bernd Kromer. The Suess Effect—first noted by Suess (1955; Revelle and Suess 1957)—refers to the post- Industrial Revolution anthropogenic situation where large amounts of fossil-fuel derived CO2 14 have been emitted into the atmosphere diluting the natural atmospheric CO2 concentration (and hence have noticeably modified what would have been the natural atmospheric radiocarbon record from about ad1900 onwards). the roman world and climate 125

Fig. 4. Top: A. The residual annual Δ14C with 2 point (pt) FFT smoothing from the data shown in Figure 3 (top in grey) from Stuiver et al. (1998) calculated minus a 22pt FFT smoothing to emphasise the change around the longer-term trend. B. The residual annual production of 14C (iterative method—see Figure 3) with 2pt smoothing. No errors shown. Bottom: Observed sun spot numbers (SSN) from Figure 3. Observe that the 14C derived residual records A and B closely reflect the peaks and troughs of the SSN record—indicating that they are largely a solar activity proxy. 126 sturt w. manning the past 9312 years from Steinhilber et al. (2009) is shown in Figures 6a and 6b, and compared against (i) the Vieira et al. (2011) TSI reconstruction from the 14C record in Figure 6a, and (ii) modeled 14C production in Figure 6b. A good general comparison is evident in each case: the Steinhilber et al. (2009) and Vieira et al. (2011) TSI records are generally similar in trend and timing (‘same variability range’ to quote Vieira et al. 2011). The absolute levels of reconstructed TSI are not relevant to our discussions in this essay—rather the trends and the solar activity extrema, and these are very similar for the last 3000 years (and so the Roman period): see Figures 6a and 6b.24 To consider what the Sun was doing during the Roman period, we can therefore look more carefully at both the 14C record, a derived reconstruction of Sun Spot Numbers (Solanki et al. 2004), and the 10Be record, in the period 300bc to ad800: see Figures 7 and 8.

24 Looking at Figure 6b in detail, the 10Be series (from ice cores) and the 14C production models (from known age trees and the carbon cycle model) correlate in terms of calendar placements relatively closely over the ad period and to the ca. 360bc minimum, and again in the 4th millennium bc, but there is something of a mismatch in the intervening period. There are apparently errors in the ice-core timescale especially in the period of the mid 1st millennium bc through 2nd millennium bc, and the ice-core chronology has been incorrectly placed as too old by around a couple of decades or a little more. The 14C record on known age tree-rings is regarded as likely the more solid record for the period of time shown in Figure 6b because the relevant dendrochronology and 14C have been replicated more than once in the course of different groups building the radiocarbon calibration sequences (e.g. Reimer et al. 2009; 2004; Muscheler 2009, 282–284), and also in various larger-scale tree-ring wiggle-match exercises (e.g. Kromer et al. 2001; Manning et al. 2010). In contrast, the ice-core records are less replicated and review of the literature indicates clear grounds for possible errors in the period indicated above, and indeed small possible errors even by the Roman period (e.g. Southon 2002; 2004; Baillie 2008; 2010). The issue of an offset of the 10Be ice-core record from the GRIP core versus the 14C timescale for the 2nd millennium bc has already been noted by Muscheler (2009). While minor in terms of assessing the general history of solar activity, this issue potentially has substantive implications for the exact dates for major volcanic eruptions in the ice-core records in the period from the mid 1st millennium ad through the 2nd millennia bc where a dating problem is identified: in particular, it suggests that it is likely that the ice-core dates are too old by a couple to a few decades. To take the best known case: the ice-core signal argued to perhaps represent the Santorini (Thera) eruption—following the arguments of Vinther et al. (2008); Muscheler (2009, 276–277)—which is dated from the ice-core records ca. 1642bc, might instead be considered to perhaps date ca. 20 years later (Muscheler 2009). This revision could lead to a consistent scenario where the 14C evidence for the dating of the Santorini eruption (from both short-lived sample material from the volcanic destruction level on the island and from the last extant growth increment of an olive branch buried by the eruption on the island: Manning et al. 2006; Manning and Kromer 2011; Friedrich et al. 2006), as well as the ice-core record of a major eruption which (following Vinther et al. 2008; Muscheler 2009, 276–277) could be Santorini, all point to a date in the late 17th century bc for this event (Manning and Kromer 2012). the roman world and climate 127

Fig. 5. High resolution 10Be data from Greenland for the most recent four centuries. The NGRIP record (Berggren et al. 2009) is annual and it is shown as a 3-year average.25 The Dye-3 data (Beer et al. 1990; 1998) are also annual and are shown smoothed on a 3pt basis (compare approximately Beer et al. 1990, Figure 1).

25 The Berggren et al. (2009) data are available from ftp://ftp.ncdc.noaa.gov/pub/data/ paleo/icecore/greenland/summit/ngrip/ngrip-10be.txt. 128 sturt w. manning

Fig. 6a. Comparison of the Total Solar Irradiance (TSI) reconstructions of Vieira et al. (2011) from the 14C record versus Steinhilber et al. (2009) from the 10Be record. For the definition of dTSI see Steinhilber et al. (2009).26

26 Data for the Vieira et al. (2011) TSI reconstruction available from https://www.mps .mpg.de/projects/sun-climate/data/tsi_hol.txt; data for the Steinhilber et al. (2009) dTSI reconstruction available from ftp://ftp.ncdc.noaa.gov/pub/data/paleo/climate_forcing/solar_ variability/steinhilber2009tsi.txt. The offset of 1365.6 watts/m2 to plot and compare the Vieira et al. (2011) TSI follows Vieira et al. (2011, Figures 7 and 8). the roman world and climate 129

Fig. 6b. Top: Total Solar Irradiance (TSI) reconstruction from ice-core 10Be records from Steinhilber et al. (2009)—see Figure 6a. Bottom: 14C production from Mar- mod09 (Reimer et al. 2009)—note the 14C production record has been inverted (contrast with Figures 3 and 5). The two curves thus appear very similar, but increased 14C production in fact correlates with decreased TSI. The grand solar minima and maxima are taken from Usoskin et al. (2007—note dates for major minima or maxima episodes from the 10Be record of Steinhilber et al. 2009 are similar and some comparisons are noted in my text and notes below). 130 sturt w. manning

Fig. 7. Top: detail of the Total Solar Irradiance (dTSI) reconstruction from ice-core 10Be records from Steinhilber et al. (2009)—see Figures 6a and 6b—for the period 300bc to ad800. Bottom: two 14C production models: (a) from Marmod09 (Reimer et al. 2009) and (b) the iterative model from Usoskin and Kromer (2005). Note that the 14C production record has been inverted (contrast with Figures 3 and 5). The curves at top and bottom thus appear similar in shape, but increased 14C production correlates with decreased dTSI. The grey panel indicates where the variation of the dTSI values are within one standard deviation of the average value between 300bc to ad800. Note the relative lack of signal, and stability, in the 2nd century bc to the 2nd century ad. the roman world and climate 131

Fig. 8. Top: the two 14C production models in Figure 7—now not inverted—for the period 300bc to ad800. Bottom: the Sun Spot Number (SSN) reconstruction from Solanki et al. (2004) for 300bc to ad800. Larger SSN are correlated with a more active sun. Observe that decreased 14C production corresponds to increased Sun Spot Numbers, and the reverse. The two grey panels indicate where variation of the Usoskin and Kromer (2005) iterative method 14C production values and the reconstructed SSN values of Solanki et al. (2004) are within one standard deviation of the respective average values between 300bc to ad800. Note the relative lack of signal, and stability, in the 2nd century bc to the 2nd century ad. 132 sturt w. manning

If we consider Figures 2 and 6a and 6b for the general context (looking at the last 6000 years), we see that in the Mediterranean climate context the grand solar minima ca. 2860bc and 765bc each equal a clear 14C production peak or TSI/dTSI minimum (see Usoskin et al. 2007 for dates of grand solar minima and maxima: the respective dates for the peak minima from the analysis of the 10Be record of Steinhilber et al. 2009 are slightly earlier at ca. 2920bc and 817bc—but, as evident in Figures 6a, 6b and 7 above, the major minima and maxima episodes are closely aligned and approximately of the same dates in both the 14C and 10Be derived records). Each also roughly demarcates the beginning of widely represented cultural floruits, and each in addition follows what appear to have been negative or challenging climate change episodes: (i) the sharp global downturn ca. 5200 years ago observed in several records (e.g. Thompson et al. 2006, 10542; Staubwasser and Weiss 2006, 379–380); and (ii) the changes at the end of the 2nd millennium bc to an episode of aridity (e.g. Kaniewski et al. 2010; Roberts, East- wood, et al. 2011, 153). In the case of the 3rd millennium bc, the grand solar minimum episode in the 29th century bc marks the beginning of the main high-point Early Bronze cultural periods in the east Mediterranean, such as the Early Bronze Age II ‘international spirit’ of the Aegean world (Renfrew 1972; Broodbank 2000) and the time of the formation of early states and the contemporary major early ‘world system’ of the wider Levant and Near East (van de Mieroop 2004; Sherratt and Sherratt 1991). In the case of the earlier 1st millennium bc, as noted above, the grand solar minimum in the 8th century bc marks the beginning of the Greek ‘Renaissance’ and related cultural developments, or the Italian Advanced Iron Age Orientalising period = (Transalpine early Iron Age) = Hallstatt C horizon (especially adopting the new higher radiocarbon-based chronology: e.g. Nijboer et al. 1999/2000; van der Plicht et al. 2009), and a general take-off of human activity across the Mediterranean region. The total ca. 1470 year period between (and without) grand solar minima (2860bc to 1390bc), notwithstanding four grand solar maxima, is notable as a largely extended interval of relatively stable solar conditions compared to more marked variations in the 4th millennium bc or the early 1st millennium bc or from the mid 3rd century ad to the present. The overall Roman period (or Greco-Roman period) occupies the one other long (1045 years) relatively stable period lying between two well-spaced grand solar minima (between ca. 360bc and ca. ad685).27

27 The dates for the major (extreme) solar activity minima from Steinhilber et al. (2009) in these cases are about 372–368bc and about ad657–672, with the two episodes thus closely and similarly placed from both the 14C and 10Be derived records. the roman world and climate 133

It is thus tempting, looking at the overall solar proxy record summarized in Figures 6a and 6b, to see these two long relatively stable intervals each between two well-spaced apart grand solar minima as providing especially favourable, or at least benign, conditions, and context, for the development of complex civilizations in the Mediterranean region. The first covers the Early Bronze Age takeoff through the main palatial periods of the civilizations of the eastern Mediterranean in the Late Bronze Age (there was a negative ‘blip’ around, and for a period, after ca. 2200bc—Weiss et al. 1993; Dalfes et al. 1997; Staubwasser and Weiss 2006, 380–383; cf. Finné et al. 2011)—but the large-scale picture is of the main Bronze Age civilization phase of the Mediterranean region covering some 1500 years in all (e.g. Mathers and Stoddard 1994; Van de Mieroop 2004). The second and arguably even more generally stable millennial-scale period, from the 4th century bc to the 7th century ad, covers the whole of the Roman period. Figure 9 summarises the approximate record of solar activity within this ‘Roman’ millennium. In detail—looking at Figures 6–9 and with the dates and approximate phase numbers after 300bc and before ad700 below from Figure 9—we see a grand solar minimum ca. 360bc (a time of change), then a major active solar peak around or just before 270bc followed by (1) a less active sun (from this peak) to around 200bc, but then, from about 200bc to around ad135 (2), there was an unusually stable period of over 300 years (so likely favourable). Variability then increases (with a sharp downturn ca. ad200–260, and then a reversal) from about ad135–305 (3–5). Solar activity around ad300 is as high as it has been for some centuries: a warm 4th century ad follows from this peak with fairly stable conditions ca. 305–370ad (6). From around ad370 to ca. ad690 there is then a long period with an overall trend down to a very inactive sun (a grand solar minimum about ad685—or ca. ad657–672 for the extreme minima in the 10Be based record of Steinhilber et al. 2009), and we might expect this to be a period of change. This process occurs in stages: decline in solar activity ca. ad370 to ca. ad435 (7), period of reversal and likely amelioration ca. ad435–515 (8), and then a period of marked decline in solar activity ca. ad515 to ad690 (9), with a reversal or plateau about ad540–630. The impact of these changes, especially the main change ca. ad515–690, will have varied according to context, but the previous era of relative stability was gone. West-East and North-South differences could be anticipated in a period of greater climate variability. While clearly there is some risk of gross simplification and over-generali- zation, the solar activity record seems to reflect a suitable overall context for the Greco-Roman period. Such a general Roman Climate Optimum (wet, 134 sturt w. manning

Fig. 9. The main trends of the solar proxy records in Figures 7 and 8 for the period 300bc to ad700 are summarized and compared from both the various 14C based data (from the two 14C production models and the reconstructed SSN data—rounded to nearest ca. 5 years and averaged) and the 10Be data in terms of (i) less active sun (grey), (ii) more active sun (white) and a long relatively stable period ca. 200bc to ad135 and a shorter relatively stable period ca. ad305–370 (both cross-hatched). The intervals are only approximate, and there is a ±5–10 year error, given the variations in the records, for the 14C summary. This interpretation is the author’s subjective assessment. The 9 phases identified on the right are taken from the 14C-based model data (see Figure 8); generally the 14C and 10Be records are quite similar as can be observed in Figure 7, but I have favoured the 14C record as the record to cite because it is a more robust general signal (from multiple measurements from different loci in the northern hemisphere all of absolute date), and it avoids some short-term noise. the roman world and climate 135 humid), or Roman Warm Period (RWP), has been recognized in various palaeoclimatic records, and is usually placed from the mid-later 1st millennium bc through to the earlier 1st millennium ad to at most about ad400 (e.g. Lamb 1995, 156–159; Hass 1996; Martinez-Cortizas et al. 1999; Desprat et al. 2003; McDermott 2004; Ji et al. 2005; Jiang et al. 2005; Seidenkrantz et al. 2007; García et al. 2007; Martín-Puertas et al. 2009; Patterson et al. 2010). Among other conspicuous observations which indicate a warm interval: the period ca. 400bc to ad400 is a notable long period of major glacier recession in the Alps (Holzhauser et al. 2005). In finer detail, the likely periods of benign or positive solar activity correlate approximately with, first, the development and floruit of the Roman world and empire in the 2nd century bc to the 2nd century ad, then, second, the recovery of the 4th century ad, and, third, the re-emergence of the Late Roman-Byzantine world in the mid-5th to early 6th centuries ad (P. Brown 1971; Cameron 1993). The likely less stable, and less positive to negative periods, like the 3rd century ad (and the so-called crisis of the 3rd century: Watson 1999, 1–20), and the mid-6th to 7th centuries ad, correlate broadly with periods of decline or change and re-orientation.

Some Other Records of Climate and Their Relevance to the Roman World

There are myriad possible resources one might cite by way of records or proxies indicative of global, hemispheric, regional or local climate and environment for all, or parts of, the past few millennia. A number of these other potential sources of information, of greater and lesser resolution, are surveyed by McCormick, Büntgen et al. (2012) focused on the Roman world and region. Papers by Luterbacher et al. (2004; 2006; 2012; Xoplaki et al. 2005; Guiot et al. 2010) provide comprehensive reviews of the current sources of data from both climate science, and from the mining of historical records, for the study of Mediterranean climate in the past 500 to 2000 years. To date, a number of researchers have produced—often differing—multi- proxy reconstructions especially of global or hemispheric temperature covering the last 1000 to 2000 years, and thus including the later two-thirds of the Roman period (see Jones et al. 2009). Future work will see further and more sophisticated statistical approaches applied to palaeoclimate data beyond the models employed so far aimed at richer and more robust outcomes (Tingley et al. 2012; Werner et al. 2013). The integration of advanced climate science is therefore close to becoming unavoidably relevant to the 136 sturt w. manning study of the Roman world; indeed, a data-rich synthesis of an impressive network of tree-ring data from central Europe has recently claimed to demonstrate this relevance in dramatic fashion (Büntgen et al. 2011).

Tree-Rings Tree-rings are the best available high-resolution (annual to subannual) climate archive for most parts of the world, especially for studying a specific region (e.g. Hughes et al. 2011; Hughes 2002). The problem for the Mediterranean region (south of the Alps)—contrast for example central Europe (Büntgen et al. 2011; Nicolussi et al. 2009)—is that we presently lack long absolute tree-ring chronologies which cover the time period of the Roman world.28 No current tree-ring chronology runs back continuously from the present beyond the mid to late 1st millennium ad, and no significant dendroclimate series extends beyond the late 1st millennium ad (where available, most Mediterranean dendroclimate analyses reveal hydro- climate information and not temperature). For the Mediterranean in the Roman period, we thus rely on extrapolation from areas elsewhere which have longer sequences and dendroclimate analyses—such as the large central Europe dataset analysed in Büntgen et al. (2011).29 It is worth noting first, however, that some potential long-lived trees and sources of wood may exist in the Mediterranean region and may offer future promise. For example, Cremaschi et al. (2006) report some long ring count, but not cross-matched (i.e. not dendro-dated), tree-ring samples of Cupres- sus dupreziana from the central Sahara region, which seem on the basis of a few radiocarbon dates to cover parts of the Roman period. These authors interpret briefly alternating periods of wide and narrow rings to perhaps indicate recurrent rainy periods in the 16th century bc to 5th century ad,

28 A number of areas of the circum-Mediterranean world have produced tree-ring records covering variously multi-century periods to the whole of the last millennium, but so far they do not reach back to the period of the Roman world (e.g. Touchan et al. 1999; 2003; 2005; 2007; 2010; Touchan, Anchukaitis, et al. 2008; Touchan, Meko, et al. 2008; Akkemik and Aras 2005; Akkemik et al. 2005; Griggs et al. 2007; Esper et al. 2007; Nicault et al. 2008; Köse et al. 2011; Seim et al. 2010; Serre 1978). A recent exception is a tentative oak dendrochronology from the Aegean region approximately placed as reaching from the present to ca. ad398 (Pearson et al. 2012)—see text below. 29 Further away from the Mediterranean world there are other long tree-ring series (to give just two examples from Eurasia, see e.g. Grudd et al. 2002; Sheppard et al. 2004), but only the central European datasets (i.e. those collected and analysed in Büntgen et al. 2011) offer a relatively proximate source for the Mediterranean region and for the European part of the Roman world during the Roman period. the roman world and climate 137 and then a shift to prevailing narrower rings to indicate drier conditions from about ad450 onwards. Another Cypress species, Cupressus semper- virens, from the upper timberline region of west Crete, may be another very long-lived but problematic resource. Rackham and Moody (1996, 189–190) note some examples with over 1000 rings (years) age and speculate that they might ‘provide a record of good and bad years back to Roman times’. To gain a hint of possible promise, the ring count on one living stem cored in the White Mountains (Lévka Ori) in 2007 by a team from the Malcolm and Carolyn Wiener Laboratory for Aegean and Near Eastern Dendrochronol- ogy was 926 (ANG-13), and the approximate test 14C wiggle-match dating of a sample from a previously cut branch of another tree (ANG-7) demonstrates the likely millennial scale age of the parent tree: Figure 10.30 But the cross-dating of these trees is a considerable challenge due to irregular and eccentric growth, missing rings, and, often, scarring or other injury-induced issues (human and animal). In terms of climate, the approximately-dated record shown in Figure 10 would indicate a general and rough correlation of increased growth during warmer periods in the northern hemisphere (and the reverse), when compared to the composite extra-tropical northern hemisphere temperature record of Christiansen and Ljungqvist (2011). This pattern, which mimics that of northern and central Europe, and not the arid Mediterranean (so positive growth conditions ca. ad1500–1650 when in general this is a dry and unfavourable period in much of the Mediterranean: Nicault et al. 2008), highlights the potential differences between controls on growth at the upper timber line in some areas of the Mediterranean, versus those operating at lower elevations where humans mainly live and farm. Another exciting recent development is the oak dendrochronology (Quercus sp.) for the 1st millennium ad that is starting to take shape from timbers from recent archaeological work in Istanbul (Pearson et al. 2012; Manning et al. 2012). Linking this material to the existing oak chronologies for north Turkey and north Greece (Griggs et al. 2007; 2009) and southeast Europe and further north offers the future prospect of a 2+ millennia oak dendrochronology from this region. The exact sources of these oaks are not yet known, and they may have been imported from local forests, from S.E. Europe, or from the Black Sea region. For example, at present, there is a good 213-year oak series running from ca. ad398–610, comprising 85 series

30 See also http://dendro.cornell.edu/reports/report2007.pdf pp. 3–4 and http://dendro .cornell.edu/reports/report2008.pdf pp. 1–2. 138 sturt w. manning

Fig. 10. Top: Extra-tropical Northern Hemisphere temperature record from Chris- tiansen and Ljungqvist 2011. A. 50-year smoothed curve; B. A 10pt FFT smoothed curve. Middle: IntCal09 radiocarbon calibration curve (Reimer et al. 2009) with the approximate test wiggle-match best placement of a time series of seven 14C dates on wood section ANG-7B according the ring counts between the sampled rings for this wood section (not crossdated) using the D_Sequence function of OxCal 4.17 (Bronk Ramsey et al. 2001). Vertical error bars show the 1σ 14C date errors; horizontal error bars indicate the 95.4% probability range on the wiggle-match found. The wiggle-match indicates a relatively good overall fit to the specific shape of the calibration curve, indicating that, despite likely issues of some missing or indistinct or eccentric tree-rings, the ring count recorded cannot be more than a few decades incorrect for the period of time represented by the branch section. The most recent date lies off (above) the calibration curve (and so appears as an outlier), but it in fact appears to be a ‘good miss’ for the upwards wiggle in the calibration curve peaking at ad1605 and especially if one allows for a few potential missing rings. With this date included the series narrowly fails a Chi-square test at the 5% level (although the OxCal Acomb value is 33.2 > An 26.7); with it excluded the series passes and has a good OxCal Acomb value of 80.5 > An 28.9 (quoted values with curve resolution set at 5).31 Bottom: Ring widths of ANG-7B in 100ths of a mm.

31 The 14C data, SDs, and the ring count gaps (= years) between the mid-points in order (older to more recent, according to position in the dendro sample) are: ETH-35577, 716±14bp; the roman world and climate 139

(samples) with a mean sample series length of 93.33 years, but only a much shorter interval offers a robust dendrochronology for potential climate analysis (see Figure 11). The controlling constraints on growth are not known for these trees; for example, a series of better growth years in the ca. ad480s to early 490s correlate both with a period of increased solar activity (Fig- ures 7–8) and decreased precipitation in central Europe as reconstructed by Büntgen et al. (2011: Figure 4)—see Figure 12 below—and generally there is some correlation between increased growth in the Istanbul record and decreased precipitation in central Europe—but nothing strong. One feature that does not appear represented is the ad536–541 event(s); although widely represented in cold-sensitive trees in northern latitudes (Larsen et al. 2008; Baillie 2008), and in historical records from Constantinople and elsewhere (Gunn 2000), this event or events does not appear to have noticeably affected the growing conditions (negatively or positively) of the oaks shown in Figure 11 (see also Pearson et al. 2012). The Istanbul record is of course very short and needs to be secure and much longer to be of potential climate use. Thus, for the present, we continue to lack a direct Mediterranean (to southeast Europe and Black Sea) Roman-era tree-ring record, but one may exist relatively soon. But let us return to the most important recent tree-ring based record and analysis for the Roman world which has been provided by Büntgen et al. (2011), based on the integration of a very large number of tree-ring series from central and northern Europe to provide both precipitation and temperature reconstructions for the last 2500 years. Figure 12 shows the April– May–June (AMJ) precipitation reconstruction from that paper; it was based on 7284 oak (Quercus spp.) tree-ring series and the correlation of the recent part of these records against instrumental data (e.g. ad1901–1980 in Büntgen et al. 2011, Figure 3), and then an extrapolation backwards from this. Fig- ure 13 shows the June–July–August temperature reconstruction from Bünt- gen et al. (2011) based on 1089 Stone Pine (Pinus cembra) and 457 Larch (Larix decidua) series and their correlation against instrumental data for ad1864–2003, and then again an extrapolation backwards from this.

Gap 60.5 years; ETH-35578, 616±13bp; Gap 41 years; ETH-35579, 618±14bp; Gap 57 years; ETH-35580, 466±13bp; Gap 43 years; ETH-35581, 373±13bp; Gap 56 years; ETH-35582, 307±14 bp; Gap 55 years; ETH-35583, 393±14bp. I thank Lukas Wacker, ETH, for the 14C data. 140 sturt w. manning

Fig. 11. Top: the mean tree-ring widths (mm) record for the tree-ring series from Istanbul (Pearson et al. 2012; Manning et al. 2012) for the interval where the Ex- pressed Population Signal (EPS) is above 0.85 (typically deemed a satisfactory level: Cook and Kairiukstis 1990) (data from ARSTAN: Cook and Holmes 1999); a 20pt (year) FFT smoothed curve is also shown. Calendar placement after Pearson et al. (2012) and as found to be compatible with 14C data and historical information (Man- ning et al. 2012). The residual curve showing the differences of the mean raw series versus the 20pt smoothed curve is then shown underneath. Bottom: the reconstructed temperature and precipitation records from central European oak time- series shown in Büntgen et al. (2011, Figure 4) are shown for the time interval covered by the Istanbul tree-ring data at the top of the figure. The curved lines through these data are 60pt FFT smoothed curves. The number of samples in the Istanbul chronology per year is shown at the very bottom of the figure.32

32 Büntgen et al. (2011) data from ftp://ftp.ncdc.noaa.gov/pub/data/paleo/treering/ reconstructions/europe/buentgen2011europe.txt. the roman world and climate 141

Fig. 12. Reconstructed precipitation (April–May–June = AMJ) in mm with respect to the instrumental ad1901–2000 period record from the Büntgen et al. (2011) study for central Europe and for Germany (dotted line—from the previous study of Büntgen et al. 2010). Two lines show 10pt and 60pt FFT smoothed curves through the Büntgen et al. (2011) data. 142 sturt w. manning

Fig. 13. Reconstructed summer (June–July–August = JJA) temperature anomalies with respect to the instrumental ad1901–2000 period record from the Büntgen et al. (2011) study for central Europe and for Switzerland (dotted line—from the previous study of Büntgen et al. 2006). Two lines show 10pt and 60pt FFT smoothed curves through the Büntgen et al. (2011) data. the roman world and climate 143

In each case we see that relatively stable and favourable climate conditions changed around the mid 3rd century ad to a period of 350 years or so marked by much greater variability and much less favourable conditions.33 There was a change towards cooler temperatures, with a notably cooler interval in the 6th century ad, and two periods of major drops in precipitation (3rd century and 6th–7th centuries ad). As Büntgen et al. (2011) argue, relatively wet and warm summers occurred during the periods of Roman and medieval prosperity. The period of increased climate variability and the change to drier and cooler conditions ca. ad250–600 broadly coincides with the demise of the western Roman Empire and the turmoil and changes of what is called the Migration Period (Halsall 2007).34 We can further check on the reliability of, and especially the relevance for the wider mid-latitudes of the Northern Hemisphere of the Büntgen et al. (2011) reconstructions, by comparing their last 1000 years of temperature record with the temperature reconstructions recently compiled from 40 different proxies from a range of source data (from ice-cores, tree-rings, sea sediments, documentary sources, varved sediments, lake sediments, pollen and a speleothem) by Christiansen and Lungqvist (2011) for the extra-tropical Northern Hemisphere: see Figure 14. The two reconstructions show considerable similarity. At the annual scale the correlation is 0.39, but when considered as, e.g., 10pt smoothed curves as in Figure 14, the correlation rises to a strong 0.61. This indicates that the general annual to decadal scale of temperature variation indicated in the Büntgen et al. (2011) curve is fairly robust for the earlier 1500 years. If we compare the precipitation record from central and northern Europe of Büntgen et al. (2011) with some precipitation records from the eastern Mediterranean (Griggs et al. 2007; Touchan et al. 2005), we see some potentially interesting patterns: see Figure 15. There are more or less opposite trends in precipitation in the period before ca. ad1500 (so before the LIA), but then there is a noticeable switch to similar trends in direction in

33 This finding is consonant with previous tree-ring work by Briffa (2000), which indicates a favourable climate for northern Sweden through to ca. ad200. 34 The precipitation reconstruction speaks of a deterioration (drier conditions) ca. ad220–340, then a reversion ca. 340–460ad, before an especially drier period ca. ad460–680. The temperature record indicates cooler weather ca. ad325–430, a partial reversion ca. ad430–530, and then an unusually cool period (similar to the worst of the LIA) ca. ad530– 650. These patterns correlate reasonably (given some delays in climate processes) with the solar activity record for the period (compare with Figures 6–9 above). 144 sturt w. manning

Fig. 14. Comparison of the temperature reconstruction for the extra-tropical North- ern Hemisphere for the last 1000 years by Christiansen and Lungqvist (2011) with the temperature anomalies reconstruction over the same period by Büntgen et al. (2011; 2006). precipitation from ca. ad1500 to ca. ad1840—i.e. for the main LIA period. This change broadly correlates with the change to less negative Palmer Drought Series Index values in the 15th century ad and through the LIA period in the Esper et al. (2007) record from Morocco. After the LIA, so from ca. ad1840 and since, there are again opposite trends in direction between the Büntgen et al. (2011) data and the Griggs et al. (2007) data (only partly so with the Touchan et al. 2005 data). All three precipitation records attest generally increasing precipitation in the post-Maunder Minimum 18th century ad. This might suggest that during Little Ice Age-like times and/or periods of major solar minima the Büntgen et al. (2011) precipitation record can indicate increasing or decreasing trends in precipitation in at least some parts of the Mediterranean, but not necessarily at other times, and, especially perhaps, it does not always give guidance for parts of the eastern Mediterranean. This complicates simple transfer of the Büntgen et the roman world and climate 145

Fig. 15. A comparison of the Büntgen et al. (2011) precipitation record from Central and Northern Europe versus two east Mediterranean precipitation reconstructions (Griggs et al. 2007 and Touchan et al. 2005) and the Palmer Drought Series Index reconstruction by Esper et al. (2007) from Morocco. All data shown after 30pt FFT smoothing applied. The three precipitation records (lower three lines) are separated and scaled on a relative scale on the Y axis to make them easier to see and compare.35 al. (2011) record to the Roman Mediterranean (whereas it is clearly directly relevant to Roman Europe in and north of the Alps).36

35 Griggs et al. (2007) data available from http://www.ncdc.noaa.gov/paleo/metadata/ noaa-recon-6379.html. 36 This issue is very noticeable for the period around ad400–700. Büntgen et al. (2011) reconstruct a period of reducing precipitation with a major low in the 6th century ad—yet Finné et al. (2011, 3168) conclude that a variety of east Mediterranean evidence appears to indicate rather wetter conditions at this time (see also p. 3164, Fig. 6A)—although one of the high resolution records they discuss (Bereket) in fact indicates the reverse (p. 3165)—high- lighting (again) the challenges of comparing different data types and high-resolution versus non-high-resolution data. 146 sturt w. manning

Speleothems The other important source of high-resolution terrestrial climate data available from the Roman region, or from close to it, comes from the analysis of speleothems and similar sources, primarily via the creation of δ18O and δ13C records. By high-resolution I mean annual to at worst decadal resolution data; I contrast non-high-resolution sources such as century-scale information from pollen records and related sources (such as the pollen-based data available from Lake Van in Wick et al. 2003). For a review and discussion of how climate proxy data are extracted from speleothems, see McDermott (2004).37 In brief, changes in the δ13C of speleothems can often be related to variations in vegetation and soil microbial activity above the cave, and these in turn relate to the temperature and precipitation in the region of the cave. Typically in temperate regions, higher temperatures and increased precipitation are expected to lead to lower δ13C values—although there are numerous complications (for a discussion of such complications and the Sofular Cave case—from coastal northern Turkey—see Göktürk et al. 2011, 2436–2437). Speleothem δ18O records can often be even more challenging (Lachniet 2009). In the right conditions, the recovered δ18O reflects the δ18O of precipitation in the cave catchment, and can also indicate temperature, but there are numerous complications in many cases. To date, only a relatively few high-resolution speleothem records have been published which are directly relevant to the Roman world and period.38 Göktürk et al. (2011, Figure 9) provide an overview of several of the main current records from the Mediterranean region, with only the Sofular Cave record offering a long high-resolution Holocene dataset.39 A new record from the Kocain Cave in southern Turkey covering the ad period is mentioned in Luterbacher et al. (2011; 2012), and this may be a useful additional resource when published.

37 A useful brief on-line review can also be found at www.ncdc.noaa.gov/paleo/reports/ trieste2008/speleothems.pdf. 38 Finné et al. (2011) include the impressive high resolution records from Oman (Fleit- mann et al. 2003; 2007). However, as Fleitmann et al. argue, the precipitation indicated by this δ18O record reflects processes primarily relating to the mean latitudinal position of the ITCZ and the dynamics of the Indian summer monsoon. The record is thus not relevant to the Mediterranean in any specific way. With reference to the Roman period, this speleothem record is also rather crucially missing the temporal period between 2700 and 1312 years ago— i.e. the entire Roman world time range. 39 I do not consider the recent Soreq record ca. 2.2ka to 0.9ka published in Orland et al. (2009) as suitable in this category, as the underlying dating precision does not in fact appear to be entirely high resolution and the temporal coverage is not regular: see esp. Orland et al. (2009, Table 1 and Figure 6 and comments p. 33). the roman world and climate 147

However, as noted, the interpretation of speleothem records is complicated and may reflect the processes of record formation, and the local environment, more strongly than regional or wider climate. In the case of the Sofular Cave, Fleitmann et al. (2009) demonstrate that the main source of moisture relevant to this record is the Black Sea. Thus, as Göktürk et al. (2011, 2442) conclude: ‘Consequently, the Sofular δ18O profile … show[s] no signs of an isotopic pattern that could be ascribed to Mediterranean-derived moisture.’ Another interesting record which seems of limited or no value for the Roman period is the Jeita Cave speleothem from Lebanon (Verheyden et al. 2008).40 Figure 16 shows some of the other better δ18O data available relevant to the Mediterranean region for the Roman period, from two speleothem sources (Soreq Cave, Israel = A and Bucca della Renella, in the Apuan Alps, Italy = B), along with another δ18O record (C) from analysis of planktonic foraminifera from a core from the southeast Mediterranean just off the coast of Israel. The Soreq Cave data (A) in Figure 16 (from Bar-Matthews et al. 2003) are uninformative for the Roman period beyond being noticeably stable (see Discussion below). The Bucca della Renella data are argued by Drysdale et al. (2006) to pick up the 4200bp/2200bc drier event, and also seem to be showing drier conditions for the 1st through 5th centuries ad with a peak in the earlier 5th century ad. In between, it seems moister in the early 1st millennium bc and relatively stable through to the earlier Roman period. The Schilman et al. (2001) record is compared with the Soreq Cave data and argued by Schilman et al. (2002; also Rosen 2007, 90) to show humid (wetter) periods peaking around 1200bc, 700ad and 1300ad and arid peaks at around 100bc, ad1100 and ad1700 (Schilman et al 2002; Rosen 2007, 90; see also Schilman et al. 2001, 165, 168–169). The age-depth model of the marine record of Schilman et al. (2001) is not, however, entirely solid.41

40 For the period ca. 1000bc to ad900 there is a general slightly drier or fairly stable trend, but no real signal. According to Verheyden et al. (2008, 380): ‘Between 3.0 and 1.1 ka, soil activity progressively decreased, as indicated by increasing δ13C values. δ18O values present more variability and it is therefore less clear if the δ13C increase is due to a progressive dryer climate (in which a δ18O increase would be expected) or if it is to be ascribed to a decrease in soil activity linked with increasing agriculture and/or grazing.’ [my italics]. 41 Its matching versus Soreq could be adjusted in at least one place; for example, there is something of an ‘excursion’ of the 14C data v. model in the early 1st millennium bc (depth 300–350cm) if one studies Schilman et al. (2001, Figure 3). 148 sturt w. manning

Fig. 16. Three δ18O plots from Mediterranean locations. They are plotted so ‘wetter’ is to the top and ‘drier’ is to the bottom. A: Soreq Cave from Bar-Matthews et al. (2003); B: Bucca della Renella from Drysdale et al. (2006); and C. the record from planktonic foraminifera (Globigerinoides ruber) from the southeast Mediterranean off Israel (Schilman et al. 2001).42

Figure 17 shows some of the available δ13C records from speleothems from the Mediterranean relevant to the Roman period. Martín-Chivelet et al. (2011) model their impressive high resolution record from North Spain as a temperature proxy over the last four millennia. They find the Iron Age Cold Period cooling event around and following ca. 800bc, then warmer conditions ca. 500bc to ad300 with a peak ca. 150bc to ad250 (the Roman Warm Period). There is then a trend to cooler conditions ca. ad350–600 with a cool minimum around ad500, before, clearly indicated, the sub-

42 Data available from: ftp://ftp.ncdc.noaa.gov/pub/data/paleo/speleothem/europe/ italy/renella2006.txt; ftp://ftp.ncdc.noaa.gov/pub/data/paleo/speleothem/israel/soreq_ peqiin_2003.txt; and from Schilman et al. (2001, Table 1). the roman world and climate 149 sequent warm conditions of the MCA, and, after this, the cooler period of the LIA (see Martín-Chivelet et al. 2011, 9–10 for summary and references to other literature). The Bucca della Renella record (Drysdale et al. 2006) indicates moister conditions in the first half of the 1st millennium bc and then drier conditions from ca. 500bc to ad300, with, just as the record ends, a suggestion of wetter conditions from the 4th century ad. The high-resolution Sofular Cave record (Fleitmann et al. 2009) shows a dip to wetter conditions in the 9th century bc, and then a fairly stable record through to the mid 1st millennium bc. There is a gradual trend to drier conditions overall, but largely a very stable record through to just after ad100, when a slight change to a wetter trend occurs through to around ad310. The stability of the main Greco-Roman period is a feature (as again during the period of the MCA, especially 10th–13th centuries ad). From ad310 there is a clear shift to drier conditions to about ad430, and then a clear shift from about ad430 through about ad640 to wetter conditions (peaking around ad600–640), before then shifting back to a drier trend from the mid 7th century ad.43 The Late Roman shifts are noticeable as major events in the relatively placid Sofular Cave Late Holocene record (see the Inset to Figure 17 for a detailed view). The Soreq Cave record in Bar- Matthews et al. (2003) is not very high resolution, but indicates a fairly stable Greco-Roman period, and a slight up/down change in the 5–7th centuries ad matches the Sofular Cave record. Whereas the Soreq Cave record in Schilman et al. (2002) also shows a humid/moister trend in the 1st–7th centuries ad, and Bar-Matthews and Ayalon (2004; Rosen 2007, Fig- ure 5.3) derive an increasing rainfall record for the 2nd–7th centuries ad from the Soreq Cave records, we may note that the analysis in Orland et al. (2009), employing a different methodology, argues for the exact opposite, with decreasing rainfall ad100–700 (with steep drops around ad100 and ad400).

43 There is an additional stalagmite δ13C effective moisture record from the Kocain Cave in southern Turkey shown by Luterbacher et al. (2011, Figure 1B), from work by Göktürk et al. However, it would appear that the graph is shown reversed (or the arrow for precipitation should point down) in Luterbacher et al. (2011, Figure 1B). On this assumption, this record seems to show stable conditions to after ad100, then a shift to a drier trend to ad300, then wetter to about ad400, a short drier shift in the early 5th century ad, and then a generally wetter period from the mid-5th century to late 6th century ad, before a major drier interval ca. ad600–800. (Note: the corrected version of the Kocain record showing the above may be found in Luterbacher et al. 2012: Figure 2.7.) 150 sturt w. manning

Fig. 17. Four δ13C records from Mediterranean region speleothems from Martín- Chivelet et al. (2011); Drysdale et al. (2006); Fleitmann et al. (2009) and Bar- Matthews et al. (2003). The Inset shows a detail ad250–750 of the Sofular Cave δ13C record. Wetter (or cooler for North Spain dataset) is downwards; drier (or hotter for North Spain dataset) is upwards.44

If we compare some proxy temperature records from two speleothems with available tree-ring records from central Europe (see Figure 18a), we see some general patterns, but also shorter-term variations between locations and data sources. All the records indicate a generally stable main Roman period through to around the earlier 3rd century ad, and then changes to cooler temperatures in the 3–5th centuries ad but with some reversal in either the 4th or 5th centuries ad (varying by record). The two

44 For the North Spain dataset of Martín-Chivelet et al. (2011), see ftp://ftp.ncdc.noaa.gov/ pub/data/paleo/speleothem/europe/spain/nspain2011d13ct.txt; and for the Sofular Cave dataset of Fleitmann et al. (2009), see ftp://ftp.ncdc.noaa.gov/pub/data/paleo/speleothem/ asia/turkey/sofular2009.txt. the roman world and climate 151 speleothem-based records and the two tree-ring-based records usually correlate fairly closely—allowing for some minor lags—with the solar activity proxy offered by the 14C production model (Marmod09 from Reimer et al. 2009).45 The ca. 360bc major solar minimum seems to be picked up in the available north Spain and Büntgen et al. (2011) records; the major solar minimum ca. ad685 is in the Spannagel and north Spain records but seems only a small deviation in the Büntgen et al. (2011) record; the ca. ad1040 solar minimum seems to be represented in all records as is the ad1305 solar minimum; the ad1470 solar minimum is in all except the north Spain record; and the ad1680 solar minimum is in all records. The solar maximum 445bc is in the available north Spain and Büntgen et al. (2011) records, and the sustained low 14C production (= active sun) period from the mid 8th to mid 13th centuries ad (the MCA) is clearly visible in all records, as is the subsequent LIA (dates for major solar minima/maxima from Usoskin et al. 2007—see also Figure 6b above). There is thus some encouraging consistency in general findings. Perhaps the key feature we can note across several records is the relative stability of these favourably warm conditions in the main Roman period compared with later times. The least coherent narrative comes from the two alpine temperature reconstructions in Figure 18a: comparing the Büntgen et al. (2011) temperature record from central (alpine) Europe with the Spannagel speleothem also from the central Alps (for the specific comparison of just these two datasets, see Figure 18b). Since the two reconstructions are both for temperature, and come from the same area, one might expect them to be relatively similar. But, although these two records do offer some general similarities in trends (so general direction and shape of the curves)—give or take some flexibility on about a decadal scale—for some periods like the MCA (e.g. ad800–1300), or again around ad1500–1700, at several other times there is no correlation, or even an opposite one (e.g. ad200–400, around ad1400, early 19th century ad). In the Roman period the two records indicate similar trends to warming conditions in the 5th century ad, before significant cooling in much of the 6th century ad, and then warming conditions from the late 6th or start of the 7th century ad—at other times there is less agree- ment in trends.

45 Note that the Marmod09 curve is inverted in Figure 18a so that production peaks (= less active sun and so likely cooler conditions) point downwards. 152 sturt w. manning

Fig. 18a. Comparison of two temperature proxy records derived from speleothems (a record from northern Spain from Martín-Chivelet et al. (2011) and a record from Spannagel Cave in Austria from Mangini et al. 2005) with the proxy temperature records for central Europe and for Switzerland from tree rings produced by Büntgen et al. (2011; 2006). Top: the Marmod09 14C production model (Reimer et al. 2009).46

46 For the Spannagel Cave data of Mangini et al. (2005), see ftp://ftp.ncdc.noaa.gov/pub/ data/paleo/speleothem/europe/austria/spannagel2005.txt. the roman world and climate 153

Fig. 18b. Detail from Figure 18a of the summer temperature reconstructions from (top) the central Alps speleothem from Spannagel Cave (Mangini et al. 2005) with (bottom) the central European temperature reconstruction based on tree-rings in Büntgen et al. (2011) (see also Figure 13).

Discussion

No dramatic or major global climate events are currently posited for the main Roman period—say from the 3rd century bc to the 5th century ad. Contrast for example the well-known and much discussed climate event 8200–8000 years ago (e.g. Cheng et al. 2009; Rohling and Pälike 2005; Alley and Ágústsdóttir 2005), or any other of the major or rapid Holocene climate change events noted in the literature (see e.g. Mayewski et al. 2004; Bond et al. 2001; Weiss 2000; Staubwasser and Weiss 2006; Thompson et al. 2002; 2006; Bar-Matthews and Ayalon 2011; Chew 2002; Weninger et al. 2009; etc.). The main Roman period lies after the pair of grand solar minima of ca. 765bc and 360bc and the Ice Rafted Debris (IRD) event 2 of ca. 700bc (Bond et al. 1997; 2001),47 and before the (next) grand solar minimum ca.

47 Ice Rafted Debris (IRD) events refer to periods when there seems to be an increase in 154 sturt w. manning ad685 and the IRD 1 event around ad500 (Bond et al. 1997; 2001) and the less favourable later 1st millennium ad time period (McCormick et al. 2007). It appears as a relative sustained era of stability, and can be contrasted with the scale of changes evident in the 2nd millennium ad during the MCA or the subsequent LIA (Bradley et al. 1993; Le Roy Ladurie 1971). The main Roman period is also conspicuous for a relative dearth of large climatically effective volcanic eruptions; indeed, it was an extended period of relative volcanic calm. There was no period of enhanced major volcanic activity or sets of eruptions of the sort which can be argued to change climate—contrast the arguments for just this type of scenario for the start of the LIA (Miller et al. 2012); nor were there several volcanic eruptions across a time period which might be argued to cause short-term climate forcings of the sort suggested for the 8th–10th century ad period by McCormick et al. (2007). To illustrate the situation, Figure 19a shows the volcanic eruption trace records from two Greenland ice-cores.48 There was a large eruption around 50bc (likely not of high northern latitude49), and then, for the next several centuries through to ad500, there was a low level of major volcanic activity compared to periods either before or afterwards. The Bristle- cone Pine (BCP) tree-ring record offers another reasonable volcanic activity proxy record for the northern hemisphere (Salzer and Hughes 2007). Fig- ure 19b shows the number of events per century between the 10th century bc and the 10th century ad where there was a ring-width minima year that corresponds with a volcanic signal in an ice-core (allowing the ice-core signals to be dated as ±5 years of ring-width minima) from Salzer and Hughes (2007, Table 2). Certain centuries stand out as volcanically active to a significant the discharges of icebergs which travel some distance into the sub-polar north Atlantic—the assumption is that this increased iceberg survivability occurs because of relatively cold water temperatures. This phenomenon was studied in particular by Gerard Bond, who further found that the source of the icebergs could be determined from petrologic tracers, and that the pattern of such tracers being found at some distance from source (indicating a notable ice-rafting event) corresponded with cooler climate episodes every few thousand years (Bond and Lotti 1995; Bond et al. 1997). Bond et al. (2001) further showed that the pattern of increased discharge events (cooler episodes) corresponded with a long-term pattern (roughly a 1500-year cycle) of solar activity through the Holocene. 48 One or two major volcanic events in the 530s ad to ca. ad541 should be added to the records shown in Figure 19a (see Larsen et al. 2008; Baillie 2008). It must also be remembered that the GISP2 ice-core was unluckily, but crucially, missing several meters of core around this time, which is probably why there is no record in this core of a major 6th century ad volcanic eruption (as first highlighted by Baillie 1994, 216). 49 Clausen et al. (1997, 26721). Some potential candidate volcanic eruptions are listed by the Global Volcanism Program on their website at http://www.volcano.si.edu/world/ largeeruptions.cfm. the roman world and climate 155

Fig. 19a. Two records of past major volcanism relevant to the Northern Hemisphere from two Greenland ice-cores (GISP2—see e.g. Zielinksi et al. 1996; and GRIP—see e.g. Clausen et al. 1997). The timescales of the two ice-core records agree fairly well (and with independent markers) from the present back through the 1st millennium bc (so to before the Roman period)—whereas there are issues before this point (see review and discussion of Southon 2004).50 degree, for example the 5th century bc and the 6th, and especially the 7th, and 10th centuries ad. Salzer and Hughes (2007, Table 4) note some packages of years with notable growth decreases and volcanic signals in the ice-core records which are indicated by the grey diamonds at the top of Figure 19b: 425–419bc, 282–280bc, 42–36bc, ad536–547, ad687–698 and ad899–903. The period from ca. 35bc to ad535 (= 570 consecutive years; and, excluding the mid-1st century bc event, the period 279bc to ad535 = 814 consecutive years) is conspicuous for the lack of evidence for major volcanic activity

50 Data with further references from: ftp://ftp.ncdc.noaa.gov/pub/data/paleo/icecore/ greenland/summit/gisp2/chem/volcano.txt and ftp://ftp.ncdc.noaa.gov/pub/data/paleo/ icecore/greenland/summit/grip/chem/gripacid.txt. 156 sturt w. manning

Fig. 19b. Bottom: number of ring-width growth minima in the Bristlecone pine (BCP) record of Salzer and Hughes (2007) associated within 5 years of a volcanic signal in an ice-core record (from Salzer and Hughes 2007, Table 2). Top: Packages of years where there are notable decreases in Bristlecone pine growth and ice-core volcanic eruption signals (from Salzer and Hughes 2007, Table 4). relevant to the mid latitudes of the northern hemisphere. There were of course volcanic eruptions—these happen regularly—and some during the Roman period had major impacts on human societies in different areas of the world (such as the eruption of Popocatépetl in central Mexico about 2000 years bp: Plunket and Uruñuela 2006; or, of course, Vesuvius in Italy in ad79: Sigurdsson and Carey 2002). But there is a notable lack of evidence for major climatically effective (thus climate forcing) volcanic eruptions relevant to the area of the Roman world over most of the Roman period until the 6th century ad. the roman world and climate 157

In sum, the Roman period seems to be a relative era of stability compared to the subsequent noisy 1500 years, and many of the periods that went before. It is also positive—relatively warm—as well as stable, and might be considered something of an optimum for much of the Europe- Mediterranean region. Not surprisingly, the Greco-Roman period as a whole has indeed been regarded for some time as something of a climate optimum (e.g. Perry and Hsu 2000; Denton and Karlén 1973). This era of stability was noted by Bar-Matthews et al. (1998), who observed from the Israeli speleothem record that the period from the end of the 2nd millennium bc to the late 1st millennium ad had the most stable δ18O and δ13C values in the past 6500 years—and hence stable rainfall (noted also by Rosen 2007, 91, 168 and her Fig. 5.3 on p. 82). Other indications are generally consistent with favourable conditions. For example, Ferrio et al. (2006) report a humid phase ca. 300bc to 300ad in northern Iberia from archaeological charcoal analysis, and, based on a the lake varve study, Martín-Puertas et al. (2009; 2008) report similar findings of humid/moist conditions broadly 600bc to ad400 (the Iberian-Roman Humid Period). There is something of a drier (more arid) dip in the middle ca. 190bc to ad150 which might correspond with the reconstructed temperature dips around this time in the central European data of Büntgen et al. (2011), centered ca. 40bc, and in the north Spain data of Martín-Chivelet et al. (2011)—see Figure 18a. Oth- erwise, Büntgen et al. (2011) find generally positive precipitation and temperature conditions for central and northern Europe ca. 300bc to ad200. Higher levels of fluvial activity (increased flood intensities) are also noted for the Roman ‘High Empire’ period (roughly 1st century bc through 2nd century ad) in studies of French rivers (Arnaud-Fassetta et al. 2010, 101, Fig- ures 9, 12), and similar findings are reported from several western to central Mediterranean areas (Luterbacher et al. 2012). McCormick, Büntgen et al. (2012) highlight that Nile flood data indicate good conditions in general for Egypt especially from 30bc to ad164—the height of the Roman world—and then a deterioration; and Dead Sea levels suggest a wet era in the century or two either side of the bc/ad transition (Bookman et al. 2004; Migowski et al. 2006), as does isotope analysis of wood from 1st century ad Masada by Yakir et al. (1994; Issar and Yakir 1997—but cf. Lev-Yadun et al. 2010). If we look more widely at the northern hemisphere, the central Green- land temperature proxy available from the GISP2 ice-core (Alley 2000), and, for the last 2000 years, the multi-proxy based Arctic summer temperature reconstruction of Kaufman et al. (2009), reveal that the Greco-Roman world enjoyed relatively warm temperatures not very dissimilar to the very late 158 sturt w. manning

2nd millennium ad, and potentially as warm as, or warmer, than the MCA— at least for high northern latitudes: see Figure 20.51 The Sea Surface Tem- perature proxy of Keigwin (1996, Figure 4) from the northern Sargasso Sea provides a comparison from the lower mid latitudes of the Atlantic region, and indicates broadly a similar pattern.52 Unfortunately, there is never complete consistency in our data, or at least there is something of difference between the west and east Mediterranean. Despite a fairly general picture of a warm and/or moist regime in the earlier Roman period in many areas, some of the available eastern Mediterranean records, especially, indicate gradually increasing dryness into and through the earlier Roman period. For example, the Lake Van oxygen isotope record (Wick et al. 2003, Figure 4) indicates increasing aridity from the end of the 3rd millennium bc through to a peak ca. 110bc (2100bp from ad1990 in the Wick et al. 2003 plot),53 as does the Sofular Cave δ13C record less dramatically (see Figure 17 above). Schilman et al. (2002), using Soreq Cave data and analysis of foraminifera from an east Mediterranean core, also identify an arid peak ca. 100bc (and then a change to more humid conditions). These records are consistent with a warm period, but with regionally more negative outcomes. It has been suggested by many scholars that climate change was a factor, even a significant factor, in the decline and re-organisation of the later Roman world (e.g. Issar and Zohar 2007; Büntgen et al. 2011; McCormick, Büntgen et al. 2012). The review of the Figures and other literature cited to this point has indeed found evidence of some substantive changes, but they are not all consistent. The available temperature proxies indicate a general decline in temperatures from the 3rd to the 7th centuries ad, but with some showing a short reversal variously in parts of the 4th to early 6th centuries

51 Humlum et al. (2011, Figure 8) achieve generally similar model results from hindcasting. The period from the 8th century bc through to the 1st century bc saw a rising temperature trend with a peak period (the Roman Warm Period) in the 4th–1st centuries bc. The Green- land temperature record (Figure 20) then shows the change to the general major cooling trend in the 5th–8th centuries ad (and especially the 7th–8th centuries ad), the subsequent warmer period of the MCA at the end of the 1st millennium ad and the start of the 2nd millennium ad, before the LIA era which only ended with the warming of the past 150 years. 52 The period ca. 300bc to ad200 sees fairly stable and relatively warm conditions, before cooling ca. ad200–500 (and then the warming trend of the MCA, and subsequent cooling of the LIA). 53 The earlier analysis of Lake Van data by Lemcke and Sturm (1997, Figure 5) also indicates relatively dry conditions in the last couple of centuries of the 1st millennium bc, before a sharp change to more humid conditions in the 1st century ad. the roman world and climate 159

Fig. 20. Two temperature reconstructions for high northern latitudes of the northern hemisphere. The plot in grey is the central Greenland temperature proxy record derived from the GISP2 ice-core based on stable isotope analysis, and ice accumulation data (Alley 2000; Cuffey and Clow 1997). The plot in black is the reconstructed Arctic summer temperature proxy derived from a multi-proxy suite of datasets in Kaufman et al. (2009).54 ad.55 The general solar forcing record offers some correlation. It indicates a quite active sun around ad300 and thus likely a fairly warm 4th century ad as this activity declined, then a cooler period ca. ad370–435 as solar activity reduces significantly, before a recovery ca. ad435–515, and then a return to significant cooling for most of the 6th and 7th centuries ad: see Figures 6–9. However, our information about precipitation in the Roman world from the first century ad onwards is more difficult to synthesize, since some of

54 GISP2 data (Alley 2000) from ftp://ftp.ncdc.noaa.gov/pub/data/paleo/icecore/ greenland/summit/gisp2/isotopes/gisp2_temp_accum_alley2000.txt. Kaufman et al. (2009) data from http://www.ncdc.noaa.gov/paleo/pubs/kaufman2009/kaufman2009.html. 55 A range of other evidence reviewed by McCormick, Büntgen et al. (2012), ranging from glacier advance histories to insect remains, supports some 4th to 5th century ad warming. 160 sturt w. manning it seems contradictory (and especially in the southern Levant—see for one recent summary Rambeau and Black 2011, 99–100—which may well reflect climate reality as different processes lead to precipitation here versus in the northern Mediterranean and further west: Black 2011). Bar-Matthews and Ayalon (2004) model increasing or positive precipitation ca. 600–200bc on the basis of the Soreq Cave data, and then more arid conditions ca. 200bc to ad200. As noted above, Schilman et al. (2002) see an arid peak ca. 100bc and then increasingly moister (humid) conditions to a peak about ad700. This is in line with a humid/moist phase in the first half of the 1st millennium ad in the Black Sea record of Lamy et al. (2006), and the evidence of the Lake Van δ13C studies (Lemcke and Sturm 1997, Figure 5; Wick et al. 2003, Figure 4) indicating an arid peak at the end of the 2nd century bc followed by more humid conditions for the next century or so. But the more recent study of Orland et al. (2009) on Soreq Cave material literally finds the opposite to the Schilman et al. (2002) paper, with a major arid trend ad100–700 (whereas Bar-Matthews and Ayalon 2004 had precipitation start to increase, slowly, from around ad200 and especially in the 4th to 7th centuries ad). Ehrmann et al. (2007) working on Aegean marine cores also find indications of likely drier conditions from the mid-2nd century ad peaking around ad650. The Dead Sea records are halfway between the above two opposites (following Bookman et al. 2004; Migowski et al. 2006; Rosen 2007, 93–94 and Figure 5.6). There is a wet phase around the end of the bc period and start of the ad period (so contra the arid peak ca. 100bc in Schilman et al. 2002), and then an arid trend (like Orland et al. 2009 and contra Schilman et al. 2002) before a short sharp reversal either in the 4th century ad (Bookman et al. 2004) or more with the 5th century ad (Migowski et al. 2006)—the opposite of the arid trend in Orland et al. (2009) and especially their sharp arid trend ca. ad400—but consonant with indications of a change to wetter conditions ca. 450–650ad in southwest Turkey reported by Kaniewski et al. (2007, 2212), or, more coarsely, around the 4th to 8th centuries ad in the record of Lemcke and Sturm (1997, Figure 5). Migowski et al. (2006, 426) report indications of arid conditions from about ad500, and the Dead Sea water-level reconstructions place a peak level either ca. ad500 or ca. ad600 (Migowski et al. 2006, Figure 3 comparing their data with Bookman et al. 2004), before a long-term arid trend (Bar-Matthews and Ayalon 2004 model precipitation as falling ca. ad600 to 1000). A coherent synthesis is clearly impossible, as some data are simply contradictory—whether in terms of dating, or the trends indicated. Whereas the synthesis of Wanner et al. (2011, Figure 4e) regards the period ad300–500ad as drier in the Levant, the more widely based synthesis assessment of the southern Levant by the roman world and climate 161

Rambeau and Black (2011, Figure 7.2) favours relatively moist conditions ca. 200bc to ad150, and then wetter conditions again ca. ad350–550 (thus contra the Wanner et al. 2011, Figure 4e synthesis), and then drier conditions ca. ad550–750. I use the Rambeau and Black (2011) assessment below in Figure 21. It is fair to say that the overall picture from the published literature is less than clear! Finné et al. (2011, 3168) come down in favour of a generally moister period ca. ad1–600 in the eastern Mediterranean, but their literature review also notes contrary evidence. It is of course not obvious how many of these variations and contradictions are real, and how much they are merely a product of the relatively loose, approximate, or potentially flexible chronologies which underlie several of the non-high-resolution to fairly low-resolution records. Some differences between regions are also likely, e.g. Aegean versus southern Levant, based on modern weather systems (Black 2011). Perhaps the main conclusion (adding the records noted in the next paragraph) is that the relative stability and coherence and positive conditions of the earlier Roman period seems to have ended as one moves beyond the 1st to 2nd centuries ad; instead, there seems both more variation in records, as well as some indications of aridity especially in the east. There are some indications of changing climate regimes in the later Roman empire in the east. There is evidence for a regional long-lasting drought in central Turkey ca. 400–540ad from the study of a diatom record from Nar Gölü (Nar Crater Lake) by Woodbridge and Roberts (2011), and generally for arid conditions from the mid 4th to early 6th centuries ad (Jones et al. 2006). The environmental study in the Bereket Basin, southwest Turkey, by Kaniewski et al. (2007) found indications of more arid conditions ca. 40bc to about ad450, but then found moister conditions ca. ad450–650 (as noted above). Sofular Cave shows drier conditions in the mid-4th to mid-5th centuries ad, and then a shift to wetter conditions in the later 5th to mid 7th centuries ad (Figure 17). And, looking more widely, likely more arid/drought conditions peaking in the 5th century ad (1500 cal bp from ad1950) are reported from Iran (Stevens et al. 2006), and there are indications of increased aridity in the mid 4th and 5th–7th centuries ad across central Asia based on the dendroclimatic study of Sheppard et al. (2004), which found long dry/drought periods ad426–500, 526–575, 626–700, and cite generally consonant indications from other sources (pp. 875, 877), and from other recent studies in western China (e.g. Qin et al. 2012).56 There

56 Cook (this volume) provides additional analysis and evidence of arid episodes in the 162 sturt w. manning appears something of a contrast (opposite regimes) between Anatolia and the southern Levant, therefore, in some of the period around the mid 4th to mid-5th centuries ad, with Anatolia dry and the southern Levant becoming wetter. Both areas seem to enjoy more favourable conditions (from several records, but not all) in the mid-5th to mid-6th centuries ad. The southern Levant then sees a sharp change to more arid conditions, somewhere by during the 6th century ad and certainly by about ad600 (see discussion above, and see Rosen 2007, 150–171; Rambeau and Black 2011, 100). Anatolia seems to enjoy more favourable wetter conditions for longer, through the mid 7th century ad, before a shift to more arid conditions. The period around ad400–600 represents a time of wider climate reorga- nization. This is the period of cooling in the north Atlantic (and IRD event 1) (Bond et al. 2001; 1997), cold temperatures around Iceland (Patterson et al. 2010), glacier advance in the Alps (Holzhauser et al. 2005), and lower sea surface salinity in the northeast Caribbean (Nyberg et al. 2010), as occurs again around the start of the LIA. In the northwest Mediterranean this is a stormier period (Sabatier et al. 2012, Figure 9c), and such stormier periods broadly correspond to the north Atlantic IRD and associated episodes over the Holocene. Increased fluvial (flood) activity occurred in several French rivers in the late antique period, ca. ad450–700 (Arnaud-Fassetta et al. 2010, Figures 9, 12), and ca. ad500–700 more widely across the Mediterranean (Luterbacher et al. 2012). The differential climate histories of Anatolia and the southern Levant in the later Roman period may thus represent a version of the Mediterranean see-saw at work. In the MCA Spanish lakes generally indicate aridity but then wetter conditions during the LIA (Corella et al. 2011, 364–365), whereas the opposite is reported from central Anatolia from the Nar Crater Lake (Roberts et al. 2012). Such patterns may also occur in the first millennium ad. The Zoñar Lake record indicates drier conditions ca. 190bc to ad150, matching the later Roman Warm Period (Martín-Puertas et al. 2009),57 before a shift to a more humid period ca. ad150–350 (end of record reported) which perhaps matches some of the stormier period in the northwest Mediterranean seabed record of Sabatier et al. (2012). But this same period (and especially the 4th to mid-5th centuries ad) sees evidence of aridity in central Anatolia (and in contrast moister conditions in the southern

4th, 5th and 6th centuries ad in central Asia, and suggests a possible long-range association with the El Nino-Southern Oscillation (ENSO) in the equatorial Pacific. See also discussion in McCormick, Büntgen et al. (2012). 57 In view of parallel changes in other records, Martín-Puertas et al. (2009, 117–118) see their lowered lake levels at this time as mainly climate driven, but accept some possible anthropogenic element from human pressures during the Roman period. the roman world and climate 163

Levant).58 As precipitation drops in the later 5th–6th centuries ad in central Europe (Figure 12) it appears to get wetter in parts of Anatolia (above). I conclude with an attempt to synthesize the data reviewed into a summary by Roman time periods.59 Refer to Figure 21, especially, unless otherwise noted.

200bc to ad200: Broadly stable solar activity, see Figures 6–8 (drop and then recovery in 1st century bc—especially in 10Be record—and an up/down in 2nd century ad); generally relatively warm but with a dip in the 1st century bc (also in central Greenland ice-core record: see Figure 20 above) and glaciers in recession; relatively moist in the western empire and the southern Levant (Central Europe trees, French rivers, Southern Levant, general Iberian charcoal δ13C and in southern Spain the Zoñar Lake overview— with, Zoñar Lake, slightly less moist/drier period ca. 190bc to ad150), but drier in Anatolia except in 1st century bc (when there seems a linkage with the marked temperature drop). Thus most areas positive including Anato- lia around the 1st century bc—which then returns to drying trend in the ad period—, while the western empire regions and the southern Levant are largely positive through the 2nd century ad. ad200 to 300: Warm generally (solar activity starts relatively high and then declines: Figures 6–8), and change (unfavourable) to drier conditions in central, and some of southern, Europe, and in the eastern empire in general; Spain in general still humid/moist (see also Figure 18a). ad300 to 430: Temperature trends seem to vary by region or be inconsistent. The central European tree data indicates a clear cooler shift. However, the Spannagel Cave in Austria indicates the opposite with warming through the 4th century ad to a peak about ad400 (see Figure 18a above), and there is also a rising temperature spike in the last couple of decades of the 4th

58 Such differences can be observed or reconstructed at periods in the more recent past. For example in the post-Maunder Minimum period ad1700–1750, Nicault et al. (2008, Figure 9) reconstruct wet conditions for most of the Mediterranean including the southern Levant except for central Anatolia. In support, Xoplaki (2002) finds that whereas some regions of Anatolia show a weak correlation with a positive NAO, the southern Levant (and much of the Mediterranean) exhibits a negative correlation. 59 The following summary by time periods—although varying in some of the details and data employed and emphasized—benefits from, and follows, the time horizon approach employed in McCormick, Büntgen et al. (2012), and is indebted to this larger and more comprehensive study. 164 sturt w. manning century ad to a peak about ad400 in the north Spain speleothem data (Fig- ure 18a). Solar activity is relatively high until late in the 4th century ad (Figures 6–8). We might wonder if these two rises in reconstructed temperatures perhaps also relate to increased precipitation, since this time interval is within the last part of the general humid/moist phase recognized in southern Spain by Martín-Puertas et al. (2009; 2008), and as there is increased precipitation evident at this time in central Europe (cf. Mangini et al. 2005, 746–747). Although there is not a clear correlation in the north Spain record between δ13C and precipitation (Martín-Chivelet et al. 2011, 6), there is sometimes something of a general correlation in trend based on visual observation of Figure 5 in Martín-Chivelet et al. (2011). It is wetter in central Europe and the southern Levant, but dry in Anatolia—note especially the opposite trends in Figure 21 between central Europe precipitation (D) versus Sofular Cave δ13C (E, where effective moisture = precipitation is upwards for drier and downwards for wetter—thus opposite in direction to D). The overall scenario seems similar to the general LIA pattern where a change to cooler conditions (and more negative NAO) means wetter in Europe and western/northern Mediterranean, but drier in Anatolia. Thus we might surmise somewhat more positive conditions in the western empire, and, by later 4th century ad, in the southern Levant, but less positive conditions in Anato- lia. ad430 to 500: A little warmer in central Europe but still cool versus typical range 200bc to ad300, and there is a more active sun through this interval (Figures 6–8, 13, 18a). It is notably drier in central Europe, but wetter at Sofular Cave in northwest Anatolia and from ca. ad450 in the Bereket Basin in southern Anatolia—these are the biggest and opposite shifts (from wet to dry and dry to wet respectively) in both central Europe and Anatolia evident in the timeframe under review—but it remains dry at Nar Crater Lake in central Anatolia. The north Spain speleothem indicates warmer temperatures (Figure 18a), and this might also indicate somewhat more humid/moist conditions (but no other clear evidence), which would be consistent with indications from southern France of increased fluvial activity. There are (still) wetter conditions in the southern Levant. ad500 to 550: Markedly colder in central Europe (see also Figures 13, 18a) (also Spannagel Cave record: Figure 18a60), and solar activity declines sharply

60 The Spannagel, Austria, speleothem record indicates falling temperatures for the 6th the roman world and climate 165 from a peak. This is a cold period in the north Atlantic (Figure 20 and text above), and a time of widespread glacier advance in the Alps. This was also a very dry period (driest 50-year period in timeframe under review) in central Europe (see also Figures 12, 18a). An opposite pattern continues with wetter conditions at Solfular Cave and in some of the rest of Anatolia (Bereket Basin)—but not everywhere, as this is the last few decades (to ca. ad540) of dry conditions at Nar Crater Lake. The southern Levant remains wetter to about this interval of time—to when exactly is not clear from the less than precisely dated palaeoclimate data. Historical and archaeological data indicate drier times in Palestine and parts of the eastern Roman world (McCormick, Büntgen et al. 2012), so it may be that drier conditions start even by the earlier 6th century ad in some of this region. There is fluvial (flood) activity in southern France and the final period of the earlier 1st millennium stormier phase in the northwest Mediterranean. There is a gradual apparent temperature increase in the north Spain speleothem record (Fig- ure 18a)—it is not known if this may also correspond to somewhat moister conditions (suggested by records in previous sentence). The last part of this interval correlates with the widespread effects of what seems to be one or two large volcanic eruptions, perhaps ca. ad536 and 541 (Baillie 2008 with Larson et al. 2008), and a package of poor conditions which seem plausibly associated across the period ca. ad536–550 (Gunn 2000). The Justinianic Plague breaks out ad541 (Little 2006; Stathakopoulos 2000), and spreads, and it is as yet unclear whether this links to generally stressed times, or to a specifically changing climate context. ad550 to 600: Still cold and fairly dry in central Europe (see also Figures 13, 18a), and the 7th century ad is a time of glacier advance in the Alps and colder northern hemisphere temperatures (e.g. Figure 20 above; Spannagel Cave in Figure 18a above). Arid conditions have definitely started in the southern Levant ending the more positive late Roman period. But Anatolia (in general) now seems to be enjoying wetter conditions (Sofular Cave, Nar Crater Lake, Bereket Basin). It is probably wetter in southern France. Rising temperatures are reconstructed from the north Spain speleothem (also Figure 18a).

century ad (and again in the earlier 7th through mid-8th centuries ad), but the North Spain speleothem indicates the opposite, only showing a fall in the 7th century ad: see Figure 18 above. There are thus some differences or contradictions comparing the tree-ring versus available speleothem derived data. 166 sturt w. manning ad600 to 700: As Figure 21 summarises, temperatures start cold but substantially improve through the century in central Europe (see also Figures 13, 18a—although Spannagel Cave indicates warming in the earlier part of interval, it then changes to cooling from mid 7th century ad in contrast to the central European tree data reconstruction: Figure 18a above). Precipi- tation remains fairly poor (central Europe) (also Figures 12, 18a). The north Spain speleothem indicates falling temperatures there—but this may also link to decreased precipitation—and this line might offer something of an explanatory route for the contrasting records from central European trees versus the two speleothem records just noted. Wet peak (in the timeframe under review) in the Sofular Cave record through to around ad670, and elsewhere in Anatolia conditions seem wetter also. Fluvial activity attested in southern French and other Mediterranean rivers. But dry in the southern Levant. ad700 to 800: Temperatures return to average and above average in central Europe (also Figures 13, 18a); precipitation improves towards average in central Europe (also Figures 12, 18a); and temperatures rise and perhaps precipitation improves in the north Spain speleothem (Spannagel Cave is the exception with its temperature record falling until after the mid 8th century ad—it only starts its long MCA climb about ad760: Figure 18a above). But the favorable wetter regime seems to end in Anatolia. Sofular Cave sees a drier trend from about ad670 and a dry phase right through the 8th century ad, and the wet phases end by ca. ad700 in the other available records from Anatolia. Dry in the southern Levant. Thus the 8th century ad sees a return to more positive conditions in much of the west of the Roman world, but less favourable conditions in most of the eastern Roman world.

Conclusions

We are a long way from a detailed climate map through time for the Roman world, but more and more, higher-quality and diverse data are being produced at a rapid rate. We may expect major progress within the foresee- able future. At present, the majority of proxy datasets from the Mediter- ranean region reflect changes in the hydrological cycle (critically relevant for relatively arid regions), whereas, with two high-resolution exceptions (from north Spain and the Austrian Alps: see Figure 18a—which may not really reflect much of the Mediterranean region), and some fairly coarse- the roman world and climate 167

Fig. 21. A comparison of a selection of records or events discussed in the text for the period 300bc to ad800. A. Total Solar Irradiance (dTSI) reconstruction from the GRIP 10Be record from Steinhilber et al. (2009). B. 14C production from Marmod09 (Reimer et al. 2009) inverted. C. Central Europe Temperature Anomalies from Bünt- gen et al. (2011) 10pt FFT smooth—dotted horizontal grey line is average 499bc to ad2003. D. Central Europe Precipitation Reconstruction from Büntgen et al. (2011) 10pt FFT smooth—dashed grey line is average 398bc to ad2008. E. δ13C record from Sofular Cave, NW Turkey from Fleitmann et al. (2009)—wetter downwards, drier upwards. F. Temperature Deviations from North Spain Speleothem from Martín- Chivelet et al. (2011) 10 year average. IRD Event 1 from Bond et al. (1997; 2001). Alpine Glacier Recession, Glacier Advance (GA) from Holzhauser et al. (2005). Northwest Mediterranean storms from Sabatier et al. (2012). Zoñar Lake from Martín-Puertas et al. (2009)—note record ends ca. ad350 and that H = Humid; Iberia charcoal δ13C from Ferrio et al. (2006). Rivers and fluvial activity for southern France from Arnaud- Fassetta et al. (2010). Southern Levant synthesis summary from Rambeau and Black (2011). Van from Wick et al. (2003); Bereket Basin from Kaniewski et al. (2007) and Nar Crater Lake from Jones et al. (2006) and Woodbridge and Roberts (2011). Change to wetter phase in 1st century bc: Wick et al. (2003). 168 sturt w. manning resolution Sea Surface Temperature proxies (e.g. Sangiorni et al. 2003; Rohling et al. 2002; Cacho et al. 2001; Emeis et al. 2000), we have much less information with regard to temperatures across most of the greater Mediterranean region. The scale of local to regional variability is clear from current often contrasting records. Where possible, creating data networks (versus isolated individual records) is key to spatial control. Tighter chronological control is also necessary in a number of cases. Many data are presented as rather convincing graphs and curves, but in a number of instances the underlying time placements—usually from radiocarbon data—remain quite flexible. This situation makes it difficult to compare such lower-resolution records with each other, or to align them with such high-resolution series which are available. It encourages a potentially false ‘tuning’ or align- ing of apparent features, and the general suck-in and smear problem (Casel- dine and Turney 2010, 90). This need not be the case going forward. With substantial series of quality accelerator mass spectrometry (AMS) radiocarbon dates allied with Bayesian analytical modelling or other approaches, it is possible to achieve much greater chronological control, and to better accommodate patterns and variations observed in age-depth time-series (e.g. Bronk Ramsey 2008; Blaauw 2010), with Blockley et al. (2008) and Blaauw et al. (2011) offering two examples. Use of such techniques can improve and allow interrogation even of existing time-series (e.g. Manning 2010, 27–32 and Figures 8–12). We should also focus on records where there is the potential to make rigorous linkages with human responses and history (Diaz and Stahle 2007). But, while a rigorous timescale is key, it is also important not to let precision become overly dominant (Caseldine and Turney 2010, 90). Some substantial climate forcings work themselves out over larger regions, like Europe and the Mediterranean, during periods of years to decades to even a century— as Plunkett and Swindles (2008) argue with respect to the 2800–2700bp event—whereas other forcings are much more immediate. We need robust knowledge of what happens when to enable us to see packages of events and processes independent of a subjective suck-in and smear tendency. While Diaz et al. (2011, 1495) encourage a move away from descriptive climatology towards physical climatology, we remain at present a little short of the necessary range of adequate and robust data for the Roman world to get beyond description. What may we conclude at present? The solar proxy records give us a broad outline for climate, within which other forcings (e.g. internal variation including ocean systems and atmospheric processes, volcanism, land use changes, etc.) then operate, on various scales. Several records exhibit rea- the roman world and climate 169 sonable linkages to the solar activity record (e.g. Spannagel Cave: Mangini et al. 2005, 745–746, Figure 7, 749). A range of records indicate that a stable and reasonably positive (warm, and in a number of areas or cases also mainly moist) climate regime was in place for the period from about the 2nd century bc through the 2nd century ad. This unusual situation, reducing some of the typical variability, uncertainty and risks of the Mediterranean climate regime for farming (Garnsey 1988; Gallant 1991), would have been conducive to the growth of the Roman world. It was also an especially favourable time (warm, moist) for both agricultural and demographic expansion in central and northern Europe. The overall climate context generally deteriorates in the centuries that follow—notwithstanding reverses/recoveries in the mid-4th and 5th centuries—eventually to a low (cool and arid) point in the later 7th century ad. In particular, the stability of the previous several centuries ended; agricultural uncertainty and bad years would have increased. The 2nd to 5th or 6th centuries ad seem to be relatively arid in several areas of the eastern Roman empire, and the indications of less favourable climate conditions further east into central Asia may have been one of the forcings behind the movements of populations that led to invasions/migrations into the late Roman world. Several records from the east indicate a change to a moister regime for a period in the 4th to 6th or even 7th centuries ad61—which may have

61 The switch to wetter conditions in the mid 5th or mid 6th centuries ad in the Sofu- lar Cave (see Figures 17, 21), or ca. ad450–650 in the Bereket Basin study (Kaniewski et al. 2007), is also evident in the Kocain Cave speleothem record (see Luterbacher et al. 2011, Figure 1B—see with footnote 43 above). However, the relationships with and between the speleothem records from the Sofular (Figure 17 above) and Kocain Caves in NW and S Turkey, respectively, vary (noting footnote 43 above). After ca. 1–150ad, when both records are relatively stable, Sofular trends wetter to ca. ad300 whereas Kocain trends drier. Sofular then shows a drier period in the 4th to mid 5th century ad showing no relationship with the precipitation jump in the later 4th through mid-5th centuries ad in central Europe, nor the matching the first half of a wetter period in the southern Levant. Kocain does however show a precipitation spike in the later 4th century ad, but then has a short drier phase around the start of the 5th century ad. Both the Sofular and Kocain records show trends to wetter conditions from the mid-5th through 6th centuries ad as noted at the start of this footnote (the opposite of the central Europe record). Both then have changes to much drier conditions: Sofular especially from the mid-7th to mid-8th centuries ad (and generally to the mid-9th century ad), and Kocain from the late 6th through 7th centuries ad (so a little earlier). Kocain thus seems to pick up the change to arid conditions in the southern Levant around ca. ad550–750, and Sofular only rather later. Further investigation and data will be required to judge how much these two speleothems reflect local conditions or their wider regions. The Black Sea (versus Mediterranean) driven climate regime primarily recorded in the Sofular Cave probably accounts for some divergence at times. The 170 sturt w. manning helped the revival and consolidation of the Eastern Roman Empire—before then a long arid and thus generally negative period. In contrast, central and northern Europe saw less favourable conditions arrive earlier than the east. Precipitation and temperatures both trend downwards from the mid-5th through the 6th centuries ad, which would have been less than optimal. This dichotomy of ‘less favourable’ west in the mid-5th through 6th centuries, and ‘more favourable’ east at around the same time, offers some context for the divergent trajectories between the eastern and western empires and their territories in the 5th through 7th centuries ad. The reverse then follows in the later 7th through 8th centuries ad, as the west recovers and the east declines. With increasing data and an interdisciplinary perspective it can be argued that climate is becoming an important element in the analysis of the history of the Roman world (McCormick, Büntgen et al. 2012). But we must nonetheless end by noting the weakness of the present story. We lack good data from many areas of the Roman world, and we lack good chronological control in many cases. A wider network of data is needed, and precise and accurate chronology. We have only an inkling so far of the regional mosaic underneath macro-level observations and generalizations. What role climate change played in the history of the Roman Empire will continue to be debated, but hopefully with more and better data.

most obvious divergence between the two records relevant to the Roman period is in the period before the common shift to wetter conditions noted at the start of this footnote: thus the Sofular record (for the 4th to mid-5th centuries ad) partly corresponds to other studies indicating arid conditions in areas of Turkey in the 4th/5th to 6th centuries ad (Jones et al. 2006; Woodbridge and Roberts 2011; Kaniewski et al. 2007), whereas the Kocain record does not (except perhaps briefly in the early 5th century). PART THREE

WOODLANDS

DEFINING AND DETECTING MEDITERRANEAN DEFORESTATION, 800BCE TO 700CE

W.V. Harris

The Problem

Did the ancient Mediterranean world experience real deforestation?* Opin- ion is divided.1 In this paper I shall first attempt to clarify the problem (definition is essential), describing the destructive forces that were at work in the period from the rise of the Greek polis to the first phase of the Mus- lim conquests. I shall then consider the evidence for possible shortages of fuel-wood and construction wood, and make some comments on the markets in each of these classes of commodity. Next I shall briefly review the most pertinent evidence provided by pollen analyses and by alluvial sedimentation. I shall evaluate a recent discussion based on demography. I shall attempt to demonstrate that the reason why the effects of heavy demand for wood were not more severe was probably active woodland management combined with an effective system of distribution. Finally I shall consider the likely effects of known climate changes. The substantive problem of deforestation arises here from the fact that the Greeks, Romans and other peoples of the Mediterranean certainly used and destroyed great quantities of trees. In the period 800bce to perhaps 165ce, they cleared huge amounts of arable land. They were always very heavy consumers of timber: under the Roman Empire especially, the demand was heavy both for fuel wood (including fuel for the production of metals and glass), and for timber to be used for innumerable kinds of building and manufacturing. From ships to writing tablets, from spears to ploughs, almost everything was made of wood. The vast majority of fuel was wood,2 often in the form of charcoal, and cremation was widespread until at least the second century ce. Those who do not believe that all this

* I thank Karl Butzer, Paolo Malanima and Robyn Veal for comments on an earlier draft of this article, and Milena Vasiljevic for useful information. 1 See Harris 2011a, 108. 2 But see Veal in this volume, p. 37. 174 w.v. harris wood use had severe effects on Mediterranean woodlands have to empha- size that woodlands, in temperate climates at least, can regenerate themselves.3 Both periodization and definition are crucial to what follows. Periodiza- tion first of all: people began clearing tree-covered land during the Neolithic, and the larger kind of settlement could have serious consequences. Investi- gators of a key site in Jordan, for instance, have written of ‘dramatic local deforestation’ before 4500bce.4 By the third millennium bce pharaohs were importing timber to Egypt by sea.5 Bronze Age changes in the environment have been widely recognized across the whole region. By 800bce, at all events, Mediterranean lands, especially in the east, are likely to have been much less tree-covered than they were at the beginning of the Holocene, though there is no way of quantifying this change. Population and economic production then intensified from the eighth century bce till the second or third century ce—and what that did to Mediterranean woodlands is the primary subject of this paper. What happened demographically and to production later on, say from 200 to 700ce, is hotly disputed; I shall hypothesize here that by 450 both population and production had greatly decreased in Mediterranean Europe but less so in the eastern and North-African parts of the Mediterranean zone. And there is no natural cut-off point in the seventh century either. Thus the period discussed in this paper, while it consists of a time of intensified timber use, was neither preceded nor succeeded by periods of zero demand; far from it. We must also remember that demand for wood could rise and fall dramatically in particular regions independently of the trends I have been describing: Attica, for example, and some other parts of Greece too, were substantially denuded of tall trees in the era of Athenian naval power in the fifth and fourth centuries bce, but Attica seems to have recovered to a notable extent when Athens lost its independence and its fleet.6

3 See Rackham 1996, etc. Much depends, naturally, on the local climate and geology, on how the land is treated (whether for example it is used for pasturage), and on the species in question (cedar forest, for instance, takes centuries to regenerate: cf. Rauh et al. 2009, 267). Conifers in general require reseeding/replanting from scratch. 4 ʾAin Ghazal, just north of Amman: Fall et al. 2004, 143; the climate was too arid to allow woodlands to renew themselves effectively. 5 Gardiner 1961, 42, Polzer 2011, 351. 6 Harris 2011a, esp. 123; Harris 2011b. defining and detecting mediterranean deforestation 175

Definition

But what is deforestation?7 Not simply, in any case, ‘anthropogenic distur- bance of the local forest ecosystem’,8 as when, because of human activity, one species supplants another.9 Part of the problem of definition is an excessively simple dichotomy between undisturbed woodland and cleared land. Everyone, or almost everyone, recognizes that landscape types are numerous in every climate zone, with differing quantities of open areas. We need at the very least a tri- partite categorization of ancient Mediterranean landscapes, which should include a lightly wooded category, such as would sustain the everyday needs of an agrarian population. But whenever woodland was cleared away and stayed cleared away, that must count as deforestation, whether the result was farmland, scrub, or eroded hills, or any combination of these conditions. It will only fail to count as deforestation if, because of natural causes or managed re-growth, the effect was short-lived. Normal usage, however, seems to reserve the term deforestation for cases that produce easily recognizable ecological crisis. And much of the argument about the Mediterranean environment is about who or what caused the ruined landscapes that characteristically occur when karstic limestone hills, stripped of their major vegetation, become seriously eroded. (Not that these are the only ‘badlands’). But some regrettable vagueness remains, and all that seems possible at the moment, in the context of ancient history, is to distinguish between deforestation that is more radical and that which is less so. It may be fair enough to describe the clearing of a certain area for farmland as an ‘episode of deforestation’,10 but episodes of this kind will only take us so far.

7 I have to leave aside here the differences that may exist between this term and similar concepts in other languages such as déboisement, Entwaldung, Abholzung, and disbosca- mento. 8 Hughes 2011, 49, appears to use this expression as an equivalent of deforestation. He claims (55) to have evidence of ‘major deforestation in many parts of the Mediterranean Basin in classical Greek and Roman times’,and that this deforestation ‘caused environmental damage contributing to the disruption of ancient economies’. For the latter kind of claim see above, pp. 2–3. There is indeed evidence of some deforestation in the Graeco-Roman Mediterranean, though as it happens the two studies from the 1980s that Hughes principally relies on do not provide evidence of this even for the localities in question (Planchais 1982; Lamb et al. 1989; see below, p. 184), but the reality is more complicated and less disastrous than he supposes. There have now been scores of palynological studies within the Mediterranean region. 9 With the exception, I suppose, that land planted with olive trees or fruit trees can be considered deforested. 10 Hughes 2011, 50. 176 w.v. harris

Destructive Forces

What special factors tended to destroy woodland in the ancient world? The most obvious factor is population growth. Let us suppose that the population of the Mediterranean world rose from 20 million to 50 million by the second century ad,11 and that each person required on average the product of from 1 to 2 hectares of arable land;12 that meant at a minimum an additional 300,000km2 under cultivation. We might think that the amount actually cleared was larger by a factor of at least two and probably more. The land area of the Mediterranean Roman Empire, mountains, marshes, deserts and other uncultivable land included, amounted to approximately 2,700,000km2. So it begins to be quite evident that land-clearance made a major difference. But it is a mistake to suppose that all pre-modern cultures have more or less the same per capita demand for wood.13 No one, admittedly, has so far succeeded in quantifying the Graeco-Roman demand for wood, or even for wood-fuel, in any impressively precise way. Consider some of the difficulties. It was metallurgy that required the truly lavish use of wood fuel in the Roman world, not the manufacturing of bricks or glass or the heating of baths, though all these processes required very large amounts. But we have only a quite vague idea of the overall scale of metal production. It is often said, for example, that the high Roman Empire is likely to have produced 80- to 85,000 tons of iron a year.14 80,000 tons would have required the fuel produced by approximately 26,000km2 of coppiced land.15 80–85,000 tons of production is unlikely to be a greatly exaggerated figure (the evidence of the Greenland ice-cores suggests that the production of metals in Europe did not reach Roman levels again until the Industrial Revolution),16 and it may well be too low, but in any case it is quite speculative. I see no way of estimating in any useful way the amount of wood fuel that was needed across the whole ancient Mediterranean for metallurgy, brick- and glass-production and the heating of baths, but it was obviously enormous and continuous. It could of course be argued that the ancients would

11 For the first figure see Morris et al. 2007, 9; I have not forgotten the sceptical comments of Cawkwell 1992 about population growth in archaic Greece. 12 For 1 hectare per head as the bare minimum needed for food and fodder, see Malanima this volume, p. 18. 13 Thus I disagree with Kaplan et al. 2009; see further below, p. 187. 14 The figure derives from a guess by Healy 1978, 196. 15 Harris 2011a, 119. 16 See the bibliography in Sallares 2007, 26. defining and detecting mediterranean deforestation 177 not have been able to continue certain cultural practices, such as heating water for baths and cremating the dead, if prices had risen precipitously (and eventually cremation became less popular—see below). Timber for construction sometimes seems to have severe effects, as we shall see below, especially during building-booms and when large navies had to be built. All the more so, because particular species were always preferred for shipbuilding17 and for any but the simplest on-land building. On the other hand, the volume of timber needed for fuel must vastly have exceeded the volume needed for construction.18 Landowners and farmers must often have faced the dilemma of choos- ing between trees and animals (a slow return versus a faster return). All ancient Mediterranean people who knew anything about the countryside— the great majority in other words—were doubtless aware of the fact that trees and goats do not get along with each other (the classic text is a vivid fragment of Eupolis’ comedy The Goats).19 There are many other traces of this conflict in Greek and Latin literature.20 The trees must often have lost.

Wood Shortages

As far as fuel wood is concerned, most Mediterranean micro-regions probably remained self-sufficient at most periods. There were three exceptions to this pattern, possibly reduceable to two: (1) some areas of intense and prolonged metal-working, (2) the great metropolitan centres, and (3) Egypt, where however the root of the problem may have been Alexandria. Athens, which had long before run short of ship-building timber, seems still to have produced much of not all of its fuel wood and charcoal in the late-fifth century and in the fourth,21 but it is known by a chance find to have imported fuel wood from Torone (on the north coast of the Aegean) in the third quarter of the fourth century bc,22 just in the period in which its Laurion silver mines became more active again. Together with other evidence, all this shows that Attica was probably quite severely deforested in the mid-fourth century,but several centuries later it seems to have recovered

17 Meiggs 1982, 118, Giachi et al. 2003, Guibal and Pomey 2009. 18 See Veal, this volume, p. 40. 19 Fragment 13 Kassel-Austin (Poetae Comici Graeci V, pp. 308–309). 20 See for instance Varro, RR 1.2.17–18, with J. Heurgon’s commentary (1978). 21 See Olson 1991, 414–419. 22 See the letter reported in SEG 43 (1993), no. 488, with Harris 2011a, 123 n. 89. 178 w.v. harris to a considerable extent.23 Simply to cite one other case, it is reasonably clear that Elba eventually ran short of fuel for processing its iron ore, so that the processing had to be carried out at least in part on the mainland.24 But the normal pattern of Greek and Roman production of metal goods was that ingots of the raw material were transported to centres of demand.25 That meant that every prosperous community, in every part of the region, required fuel for the actual manufacturing processes, Tebtunis with its much-differentiated metal-workers as well as the towns of high-imperial Italy. Rome and Alexandria had to draw on areas beyond their immediate hinterlands. The construction of the porticus inter lignarios outside the Porta Trigemina at Rome in 192bce26 occurred at a time when the city of Rome was growing very rapidly and its location suggests that the lignum in question was coming up the Tiber, not down it as we might have expected;27 it would be reasonable to conclude that the lower Tiber valley was no longer rich in timber (the era of carefully-managed Italian woodlands had perhaps not yet begun). We can only speculate about where the famous navicularii lignarii of Ostia found wood fuel to import near the end of the second century ce (Sardinia perhaps).28 In the fourth century, under Valens and Valentinian in the 360s, Rome was importing fuel from North Africa; it may have been the huge baths of Diocletian and Constantine that necessitated this.29 We shall see later that evidence from Antioch suggests that by the second century ce it was running out of local fuel, and the same may well have applied to other ‘second-tier’ cities such as Perga- mum. As for Egypt, there seems by Severan times to have been a chronic shortage,30 to such a degree that fuel wood was sometimes even imported from

23 See Pausanias 1.32.1. 24 Contrast Diodorus Siculus 5.13 with Varro (ap. Servius, Verg.Aen. 10.174) and Strabo 5.223 (cf. Ps.-Aristotle, On Marvellous Things Heard 93). Corretti and Firmati 2011, esp. 229, show that iron production on the island itself had largely ceased by about 50bce. 25 Harris 2000, 723. 26 Livy 35.41.10. 27 This view is fortified by Tucci’s demonstration (2004, 199) that the Pons Sublicius (and hence in all likelihood the Porta Trigemina) was lower down the river than recent doctrine has maintained. 28 The inscription: CIL 14.4549. The date: Meiggs 1982, 339. 29 The main texts are C.Th. 13.5.10, 14.5.1. See further Harris forthcoming. 30 See Ulpian in Digest 32.55.5 ‘in Aegypto, ubi harundine pro ligno utuntur …’,‘where they use reeds in place of wood’. defining and detecting mediterranean deforestation 179

Italy;31 the actual deforestation, however, may pre-date the period under investigation here. With regard to construction timber (Lat. materia), the story is quite different. The earliest signs of local shortage concern naval power. By the late fifth century, and probably earlier, Athenian naval hegemony depended on importing ship-building timber from Macedon and Thrace; it was out of the question to rely on Attica itself,32 all the more so because particular species were required. Here is the well-informed Theophrastus:33 fir, mountain pine and cedar are the standard ship-timbers. Triremes and long ships [i.e. warships] are made of fir because it is light, while merchant ships are made of pine because it does not rot. Some people, however, make their triremes of pine also, because they are short of fir …. These woods are used for the main timbers, but for the trireme’s keel oak is used …. They make the cutwater … and cat-heads, which require special strength, of ash, mulberry or elm. Theophrastus shows in fact that there was a serious lack of ship-building timber in Greece by the late fourth century bce. ‘There is only a small area [he means in the eastern and central Mediterranean] which produces wood suitable for ship-building’: in Europe, Macedon and certain parts of Thrace and ‘Italy’, in Asia, Cilicia, Sinope, Amisus, and Mounts Olympus and Ida (but they do not have much); Syria has cedars, which they use for warships’.34 The other large naval powers of the high-classical Greek world—Syracuse, Corcyra, and above all Corinth—are likely to have encountered problems similar to those of Athens.

31 P. Tebt. II.686 (second or third century). It may have been transported by grain ships returning from Ostia to Alexandria (via the Calabrian coast). Whether we can associate fuel shortage with a decline in brick production in Italy after Severus Alexander (there was a recovery under the tetrarchy) or with the large-scale change from cremation to inhumation is regrettably obscure (Meiggs asserts the one possibility while denying the other [1982, 504 n. 119 and ib. 257]; I incline to the opposite view). However the whole notion that cremation ceased, or largely ceased, in the Roman Empire as a whole is in need of revision: in some provinces it certainly did not – see for example Pop-Lazic 2002. 32 For the evidence for Attic deforestation to be found in Plato, Critias 111c, see Harris 2011b. 33 Hist. Plant. 5.7.1–3. 34 Hist. Plant. 4.5.5. This might be taken to imply that Xenophon’s report (Anabasis 6.4.4) of ample ship-building timber at Calpe, on the Bithynian coast far to the west of Sinope, was now out-of-date. Italy for Theophrastus still meant Calabria; he underestimated the resources of the rest of the peninsula, which he had not apparently visited. He wrote this very early in a period of about a century during which, because of the number and increasing size of warships, the long-timber supplies of the eastern Mediterranean were put under greater strain than ever before or later; cf. Meiggs 1982, 139. 180 w.v. harris

Theophrastus tells a somewhat similar story about long timber ‘for the builder’s needs’, but there is a significant difference, because here he is concerned with quality not availability: the main cities of European Greece had to import the best building timber from Macedon, the Black Sea and other places, but rustic Euboea produced some building timber even though it was of poor quality.35 Such evidence as we have for the sources of timber listed in various fourth-century temple-accounts shows that Arcadia and even such a centrally located place as Sicyon could still provide some long timber.36 We might conclude that most but not all of mainland Greece, Sicily, the Aegean islands and the west coast of Asia Minor had lost most of their tall trees by about 310bc, leaving them with simply enough coppiced trees and shrubs to provide fuel. Outside Attica, however, we have no strong reason to suspect ‘ruined landscapes’. Mount Lebanon meanwhile was rich in timber, cedar and cypress especially, when Antigonus set out to build a fleet on that coast in 315.37 Ptolemaic Egypt was far from self-sufficient in this respect. Since this had probably been true of Egypt since the second millennium bc if not earlier, there is no need to set out all the evidence. Naval power, not to mention the building of Alexandria itself, depended on timber from overseas pos- sessions, in particular Cyprus.38 In Egypt’s Greek period, the government’s concern is well attested from the middle of the third century, and one has the impression that in Greek and Roman times every tree in Egypt was under surveillance.39 Regulations promulgated by the king in 118bce show not only that it was illegal to fell a tree on private land without permission but that the occupiers of some legal categories of land had some obligation (not explained in detail) to plant trees.40 The most interesting fact, however, may be that a region possessing only minimal quantities of harvestable long timber could in fact manage reasonably well within the broader Hellenistic

35 Hist. Plant. 5.2.1. 36 Meiggs 1982, 423–457, reviewed this evidence in detail, but I do not share his confidence that a timber-seller’s geographical origin reveals where his timber came from (there are too many Corinthians). The most interesting evidence from our point of view comes from Delphi, showing among other things that the rebuilders of the temple of Apollo could obtain seventeen cypresses from Sicyon in 335, but at ‘exceptionally high’ prices (Meiggs 431). 37 Diodorus Siculus 19.58. 38 Meiggs 1982, 133–135. 39 On all this see Cadell 1976, 346, and Kramer 1995, 218–222. But note that the government was interested in trees partly because of their importance for the security of dykes: Drew-Bear 1995, 3–4, and P. Tebt. III.703 (second half of the third century bce). 40 Select Papyri II.210 = C.Ord.Ptol. 53, lines 200–206. (This depends on whether the ‘plant- ings’ referred to have to be trees). defining and detecting mediterranean deforestation 181 and Roman economies, helped by a wheat surplus, the de facto papyrus monopoly and other advantages. But ambitious ship-building required the control of territory elsewhere, and that was why M. Antonius gave Cleopatra the territory around Hamaxia in western Cilicia.41 Long before this time the Romans had come to dominate the Mediter- ranean Sea, having paid careful attention to the supplies that they needed for naval warfare.42 Their naval history started from small beginnings in the late fourth century and reaching a crescendo at the Battle of Myonnesos, off the Ionian coast, in 190. Fortunately for Rome and for the Mediterranean woodlands, the interests of imperial power seldom thereafter required the deployment of a large fighting navy. The evidence about Italy is a little mixed, but as it stands now it suggests an intensifying shortage of tall trees in early imperial times (further research on Campanian and Ostian architectural timber might possibly validate or invalidate this conclusion). When Dionysius of Halicarnassus, a worshipper of Rome, lauds the woodlands of Italy, claiming in particular that they provide ample timber for ship-building, he is demonstrably exaggerating, and committing something of an anachronism.43 Other evidence suggests that some areas of Italy became more and more dependant on long-distance imports.44 A text of Strabo about Pisa points in this direction: most of the timber in the Monti Pisani ‘is now [in Augustan times] being used up’ for construction at Rome and elsewhere.45 This is what was likely to happen when there was long timber fairly close to a good port. But Strabo also says that most of Rome’s building timber comes down the Tiber from ‘Tyrrhe- nia’.46 Somewhat later the situation seems to have changed: some of the fir and spruce (picea sp.) from the Campanian cities came from far away, according to one theory from the Austrian Alps.47

41 ‘Since it was suited to the building of fleets’, Strabo 14.669. He is somewhat vague about the extent of the region in question. This was also how she obtained Rough Cilicia (ibid. 670), presumably. 42 It is symptomatic of this concern that at some early date the state took title to all coastal woodlands (Cicero, De rep. 2.58—he attributes the measure to King Ancus Marcius). 43 Dionysius of Halicarnassus, Roman Antiquities 1.37.4 (taken literally by Nenninger 2001, 200). It was quite untrue, for example, that Italy had ‘mines of all kinds’, 1.37.5; cf. Brunt 1971, 128–129. And he is plainly inaccurate when he says (20.15) that the Sila suffices for the needs of Italy. 44 I leave aside the matter of super-luxurious tables made of citrus-wood (Callitris quadri- valvis) imported from Mauretania, a fashion that led to the deforestation of the ‘Mons Anco- rarius’ (Pliny, NH 13.95). 45 5.223, mistranslated by among others Grove and Rackham 2001, 173 n. 46 5.222. 47 Kuniholm 2002, 236–237 (dendrochronological evidence). It is a pity that the researchers in question did not check samples from Liguria, since that would be a more 182 w.v. harris

The elder Pliny thinks of hills freshly stripped of trees as a common phenomenon: ‘springs often arise when woods have been cut down … [he refers to an incident from Greek history] … harmful torrents often run together when the woods which used to hold and absorb the rains have been stripped from the hills’.48 It is quite wrong to cite him as a witness to the supposed fact that there had been widespread deforestation,49 but the frequent allusions in Natural History Book 16 to sources of timber such as Raetia, Histria and Corsica that are just outside the Italian peninsula,50 suggest that certain species at least were not readily to be had nearer at hand, even though he also mentions the Appennines. Elsewhere he alludes to logging in the upper reaches of the Tiber, and he mentions the silva Sila.51 The source areas mentioned by Pliny are thus, in the main, well removed from the densest concentrations of population.52 There may be other signs too of an eventual shortage in the area of the metropolis: Ulrich has pointed out a difference in construction practices between the Vesuvian cities in the first century and Ostia in the second—the latter used less timber, perhaps, as he suggests, because the supply of long timber was now under stress.53 Emperors always owned a good deal of woodland,54 but the only ambitious attempt by an emperor to look after the supply of long timber seems to have been an elaborate initiative undertaken by Hadrian in the province of Syria. In the northern half of Mount Lebanon, he had markers put up over vast areas of forest. There were at least 800 such markers (they were plausible source. Robyn Veal (pers.comm.) shares my doubts about the Alpine source, noting that Picea grows in Liguria and Tuscany (Pignatti 1982, I, 74–75). 48 NH 31.53. 49 Hughes 1983, 437, on the basis of NH 13.65 (together with Livy 9.36), completely misrep- resented (the Pliny passages refers only to a supposed silvestris regio near Memphis; actually the whole passage about trees in Egypt, sects. 60–65, is an intriguing one). 50 Raetia (sects. 66, 190), Histria (66), Corsica (71, 197). The other places he mentions are Macedonia, the Pyrenees, some specific zones of Asia Minor and Gaul, the Tyrrhenian coast of Italy (meaning Liguria?), the Alps and Appennines, Crete, Africa, Syria and the land of the Vaccaei in Spain. 51 NH 3.53, 74. 52 It is intriguing that a resident of Italy such as Hermas (first- or second-century) seems to have been familiar with tree-less mountains (see at length Parable 9.1 and 9.19–29 in the Shepherd); he sets the scene in Arcadia simply because it was famously mountainous. 53 Ulrich 2007, 121. An interesting topic which we are not (or not yet) in a position to clarify, is why certain places, Gades for example, became ship-building centres at certain periods. 54 But I would not go as far as Thonemann 2011, 280, according to whom ‘throughout antiquity … large and potentially profitable stretches of woodland were normally regarded as state property’. defining and detecting mediterranean deforestation 183 often numbered),55 reserving four types of trees as imperial property.56 This definitio silvarum obviously shows that the emperor or some powerful sub- ordinate regarded these forests as a valuable resource (and one recalls that Hadrian travelled in this region in 129–130). It seems unlikely that he would have intervened if he had not thought that the Mount Lebanon timber was at risk; the exact nature of his concern is unknown, but military prepared- ness, Mediterranean shipping and imperial building plans may all have come into it.57 Under the high Roman Empire, there is good deal of evidence about shortages of long timber in certain areas, but nothing on the other hand that would justify diagnosing widespread deforestation throughout the Mediter- ranean.

The Palynological Evidence

Pollen deposits may one day settle the issue at hand in a definitive fashion, and over some thirty plus years studies of this material have proliferated in many Mediterranean regions58—though not in some of the areas where they would be most interesting, such as Tunisia, Appennine Italy, Sicily, and Dalmatia. This evidence comes with various caveats, however. When woodland begins to be used for grazing rather than being displaced by cultivation, it may be hard to detect in the pollen record.59 Furthermore, many pollen studies concern more or less remote areas that are of secondary importance for the overall problem of deforestation. And it is only recently that it has become feasible to take proper account of three important variables, the pollen productivity of different species, the ‘fall speed’ of different kinds of pollen, and the prevailing winds at the various sites in question.60 Research that takes inadequate account of these variables may be next to useless. Then there is the problem of low-resolution chronology. However the following results, set out here very briefly, have seemed to emerge.61

55 See IGLS VIII, 3, edited by J.-F. Breton; at least 187 of these inscriptions have been recorded. There are now dozens more: AÉ 2006, 1572f. 56 Probably cedar, fir, and pine, and either cypress or juniper: Mikesell 1969, 20. 57 For other comments on this evidence see Harris 2011a, 130–131. 58 It is regrettable that there is no thorough up-to-date bibliography. Di Rita and Magri 2009 is useful from this point of view. 59 Atherden 2000, 64–66. 60 See Poska and Pidek 2010. 61 For a more detailed but already somewhat outdated survey see Harris 2011a, 133–136. 184 w.v. harris

(1) In a number of places, both the Bronze Age and the later spread of Greek agriculture sharply reduced arboreal pollen; there is good evidence to this effect from, for example, bronze-age Miletus and the early Hellenistic Golan Heights.62 In Acarnania about 800–600bce there was a marked change in vegetation, obviously to be connected with increased human presence.63 Whether any of this amounted to deforestation seems to me to be to some extent a matter of definition. Unfortunately the areas most likely to have suffered deforestation in classical Greek and Hellenistic times, such as Attica and the hinterland of Alexandria, have no palynological evidence to offer, as far as is currently known.

(2) The Roman Empire witnessed the serious depletion of some woodland resources, and the effective management of others. Here in fact is our central problem—was the former pattern widespread? We have already seen some textual evidence that may make us suspect that it was so. Recent pollen research has detected this effect at Miletus,64 in coastal Puglia,65 and in north-western Iberia.66 Older reports make similar claims about, for instance, sites on the coasts of Catalonia67 and Provence.68 Pollen research

62 See Knipping et al. 2008 (Bafa Gölü, Miletus) and Neumann et al. 2007 (Birkat Ram, northern Golan Heights). This is how I would interpret the ‘marked retreat of forest vegetation’ (Gerasimidis 2000, 35) at Lailias in the mountains of Macedonia (1420m. a.s.l.) in the second century bce; if the Romans had anything to do with it, it may have been because they brought this district into the timber market (the Strymon valley was not far away). 63 Jahns 2005 (Lake Voulkaria; the main feature was a big increase in Phillyrea, a genus of small flowering evergreen tree). 64 Knipping et al. 2008. 65 Di Rita and Magri 2009. They say that an accelerated decline in arboreal pollen took place at 2100bp, but their Fig. 3 seems to date the event about 2300–2200bp. They do not, however, apply the term deforestation to anything that occurred in Roman or late-Roman times. 66 Mighall et al. 2006. 67 Riera-Mora and Esteban-Amat 1994. 68 Planchais 1982. Hughes 2011 relies heavily, though indirectly, on this study as evidence for Graeco-Roman deforestation. Planchais showed that pollen recovered from a site near the Étang de Mauguio (43° 350 N) in Provence demonstrated a sharp contrast between a millennium-long period (roughly 2270–1300bp) and what came before: the former showed lots of beech and oak, whereas the only trees well represented in the latter period were walnut and olive, and non-arboreal pollen (Typhaceae and Cyperaceae as well as Gramineae) was abundant. In other words, the local population and/or the Romans massively changed the pattern of local vegetation (be it noted that there was a huge amount of Roman centuriation nearby, though not in the immediate vicinity). Another study cited by Hughes, Lamb et al. 1989, discussed a site in the Middle Atlas in northern Morocco; these authors unfortunately wrote a misleading abstract in which they asserted that ‘anthropogenic forest degradation dates from about 2250bp’,a date, be it noted, when the local economy had nothing to do with defining and detecting mediterranean deforestation 185 also seems to show, however, that at Sagalassos in Anatolia, at a site deep in the Libyan desert, and at another high up in the Alps, to name three relatively clear examples, Roman practices did not have a strongly negative effect, if any at all, on the local tree population;69 but the last two of these sites can easily be regarded as peripheral.

(3) Late antiquity presents a similarly complex array of evidence. In some places woodlands grew back: at the site in the Golan Heights already referred to, olive cultivation collapsed in the first half of the seventh century and Quercus calliprinos expanded.70 But the same period seems to have been quite harmful to trees in some places, for example in Spain, in spite of an economic shrinkage that might have permitted woodland regeneration: at a site in the north-west of the peninsula, both total arboreal pollen and the percentage of pollen attributed to Quercus seem to be declining in the approximate period 550–750ce;71 at sites on the Catalan coast, there is considerable evidence of a decrease in arboreal pollen in the period 500 to 700ce, and the reduction in woodland led to erosion (the investigators proposed to associate these changes with an increase in grazing).72 At the site in Puglia mentioned above, olive trees come to dominate the pollen evidence more and more in late antiquity, especially after about 500ce, while ‘most trees keep their declining trend’;73 was this trend owed to Byzantine demand for the region’s olive-oil? It seems obvious, in short, that as far as the pollen evidence is concerned we cannot yet generalize about the whole Mediter- ranean world in late antiquity. If Mediterranean woodlands suffered in the period 500–700ce, climate change may have been partly or wholly responsible (a more arid climate may have prevailed—see below), but we do not yet know enough to assert this with any confidence. Another possible explanation might be a decline in the quality of land-management. The violent disruption of the barbarian either the Greeks or the Romans; what actually happened at about that date was the decline of Fraxinus sp. and Quercus canariensis (72); ‘a more severe phase of forest degradation’,these authors says, ‘begins at about 1600 years ago’ (ibid.), that is to say with Roman power and influence in this area now on the decline. 69 See respectively Vermoere et al. 2003; Hunt et al. 2001; Moe et al. 2007 (the area in question is between 1830 and 2304m. a.s.l.). The same seems to be true at Lake Voulkaria in Acarnania (above, n. 63). 70 Neumann et al. 2007, 340. 71 Mighall et al. 2006, esp. 209 fig. 3. 72 Riera-Mora and Esteban-Amat 1994, esp. 20–21. The key sites were at Ullastret (decrease in AP values at 1500 +/- 80bp) and at Besós (‘not much before 1300 +/- 40bp’). 73 Di Rita and Magri 2009, 300. 186 w.v. harris invasion, especially in the early fifth century, may have made careful, long- term estate management a rarer phenomenon. (There is admittedly some dispute about the whole nature of these invasions).

(4) Being close to water-borne transport under the Roman Empire meant economic opportunity and also ecological trouble (contrast for example two sites in the Alps, the Lac de Praver near Grenoble,74 with excellent access to river transport, and Val Febbraro in the province of Sondrio,75 with very poor access). But this statement has an important corollary: much of the woodland of the Roman Empire, even in Italy, not to mention the Danube and other provinces, was beyond the range of affordable water-transport, and was therefore relatively safe from deforestation. A wooded area well away from water-borne transport, such as the Troodos Mountains in Cyprus, was to some extent protected from exploitation.76

Sedimentation and Erosion

Erosion is a very large topic in the geological literature, but here I simply want to ask whether any of the resulting sedimentation in the Graeco- Roman Mediterranean can be traced to deforestation within the period we are examining. This is widely assumed to be the case, but precise chronology has generally been lacking. Recent literature has provided some relatively precise sites in Turkey. Results from the hinterland of Antioch on the Orontes are especially interesting. A recent investigator has shown that in the Jebel al-Aqra region immediately to the east of Antakya, after millennia of human habitation, severe erosion set in from about 150ce onwards. In keeping with current trends, he attributes this partly to climate and partly to land-use practices,77 but in a moment of candour he admits that there is no correlation with anything we know about climate change in the Levant.78 It is reasonably obvious that for centuries after its foundation in 300bce Antioch managed its need

74 Nakagawa et al. 2000. 75 Moe et al. 2007. 76 Cf. Butzer and Harris 2007, 1951, with Theophrastus, Hist.Plant. 5.8.1. 77 Casana 2008. The ‘400-year lag between the initial settlement of upland areas and the first evidence of soil erosion suggest[s] that it may have been the intersection of extreme precipitation events with particular land use conditions of the Roman and late Roman periods which worked together to drive soil erosion’ (429). 78 Casana 437. defining and detecting mediterranean deforestation 187 for fuel-wood without severely deforesting the nearby hills (the area in question is between 7 and 14 kms. from the centre of the city), but eventually the situation got out of hand; given a population of perhaps 250,000 in the second century, that is hardly surprising—and the greatest density of settlement in this particular areas, together with the worst erosion, was still to come (from about 300 to 600ce).79 An earlier study, concentrating on south-western Turkey, detected similar effects in a more vaguely defined ‘classical period’, offering as an example the area between Burdur and Elmalı, that is in ancient terms between Lycia and Pisidia.80 Sedimentation at Lake Bafa, near to ancient Miletus in western Turkey and to the mouth of the River Maeander, reached a peak in the period from the first century bce to the fourth century ce.81

A Demographic Approach

In a previous article I implied that known population growth must have led to massive woodland clearance in the period from 800bce onwards, but also that this approach cannot by itself settle the major question about ancient Mediterranean deforestation.82 Meanwhile, in a paper entitled ‘The Prehistoric and Preindustrial Deforestation of Europe’, published after mine went to press, Kaplan, Krumhardt and Zimmermann made a resolute and more optimistic attempt to come at the deforestation problem from the angle of demography.83 They estimate the population of ancient Europe and most of the Mediter- ranean, posit the amount of arable land that would have been necessary to support these populations (taking account of climate and of soil properties); that, according to them, was the amount of land cleared and deforested. Their population estimates for ancient times are mostly reasonable ones (and we can overlook the incongruity of their claim to know what the population of, say, Belgium was in 1000bce; they insist on using modern territorial units).84 And it is a positive aspect of this study that its authors take account of soil properties and climate change.

79 Casana 433. 80 Roberts 1990. 81 Knipping et al. 2008. On possibly related changes in the rural economy of Miletus in the first century bce see Thonemann 2011, 293, and on Miletus’s prosperity for much of this period ib. 334–338. 82 Harris 2011a, 116–117. 83 Kaplan et al. 2009. 84 And some historical oddities are perhaps inevitable: ‘Maghreb regions of North Africa 188 w.v. harris

The most striking results of this approach, as far as the period 800bce to 700ce is concerned, are a major reduction in ‘forest coverage on usable land’ between 500bce and 1ce both in North Africa and in a zone designated as central and western Europe,85 with some recovery by 500ce, especially in North Africa and the zone designated as ‘eastern regions with consistently low forest cover’.86 Thus

1000bce 500bce 1ce 500ce 1000ce Central and western Europe 77.2 70.6 50.7 52.4 40.2 North Africa 93.9 75.7 33.3 43.6 27.8 ‘Eastern regions with consistently 48.7 40.4 32.5 41.0 36.5 low forest cover’

But there are several serious weaknesses in this approach.87 One is a vastly oversimplified idea of consumption patterns. Kaplan et al. shunt this problem aside, announcing that they ‘assume that population change is the primary driver of change in forest area for the years 1000bc to 1850’.88 Taken literally, this may be true, but their assumption ignores the enormously variable pressures on both timber for fuel and construction-timber within the period in question—just consider the single item of fuel for metallurgy, which I discussed earlier. The second serious weakness in this paper is that its authors hugely overestimate the amount of land necessary for subsistence in a pre-modern economy, by a factor of perhaps five with regard to Mediterranean populations. Their Figure 9,89 though it offers no speculation about the population density of the Mediterranean world in the period I am discussing here, implies, in conjunction with the rest of the article, that the cultivable land in that region could in antiquity have supported between 10 and 30 people per km2. The origin of this estimate is unclear, but it is generally held, on the basis of extensive research, that one to two hectares per display a distinct pattern of low clearance during early times with a relatively rapid increase at around 200B.C.’, Kaplan et al. 2009, 3025, referring to their Fig. 8, group 4. 85 North Africa means for them Morocco, Algeria, Tunisia and Libya; central and western Europe means Czechoslovakia, France, Germany, England-Wales, Ireland, Italy, Poland, Portugal and Spain. 86 This means Cyprus, Greece, Iraq, Palestine-Jordan, Syria-Lebanon and Turkey-in-Asia. 87 One might quarrel about many details here. For instance, it is not plausible to suppose that the central and western European zone had notably less forest coverage in 1000ce than in 1ce (the culprit here is clearly an indefensible ‘progressive’ view of history). Ancient Cyprus was certainly not a region of ‘consistently low forest cover’. And so on. 88 Kaplan et al. 2009, 3019. 89 Kaplan et al. 3028. defining and detecting mediterranean deforestation 189 person were sufficient for subsistence,90 i.e. population density on cultivable land lay in the range 50 to 100 persons per km2. Be it noted that at a density of 20 persons per km2, the population growth conservatively hypothesized for the Graeco-Roman Mediterranean (see above) would have required the clearance not of 350,000km2, but of perhaps 1,750,000km2, which is entirely impossible. These two criticisms of Kaplan et al. might very roughly cancel each other out: they underestimate some of the destructive forces tending to produce deforestation, as far as the Graeco-Roman period is concerned, but they overestimate others. What gets left out, however, is what is culturally specific. This matters in two ways (and raises again the question of definition). No ancient or mediaeval town or city could afford to be reckless about its fuel needs, and as noted above we have every reason to believe that with a few readily comprehensible exceptions ancient communities satisfied those needs within their own territories, which implies that they at worst maintained enough scrub, boscaglia, phrygana, and coppiced or pollarded trees to supply themselves with fuel. The need for long timber was not so manageable, or rather what made it manageable was a combination of trading networks and imperial power that left large areas denuded of tall trees over, at least in some cases, long periods. The relative importance of ship-building and on-land building varied greatly from period to period, and the effects of these needs on the landscape must have varied greatly as well. In some cases (fifth- and fourth-century Attica, for instance, and the lower Tiber valley in the high Roman Empire), the effects probably amounted to what by any standards should count as deforestation; in other cases, the effects must have been less drastic, with some tall trees surviving but little in the way of continuous woodland. I consider therefore that Kaplan et al. probably overestimate to a significant degree the relative decline of forest coverage in both central and western Europe and the ‘eastern regions with consistently low forest cover’ in the five centuries from 500bce onwards. Their article is undeniably useful, but it merits further discussion not unqualified citation.

Woodland Management

The totality of the evidence thus seems to suggest in short that there were probably some deforested areas in classical and Hellenistic Greece and

90 Malanima 2009, 106, with references. 190 w.v. harris the high Roman Empire. At the height of the Roman Empire, many areas that had been more or less heavily wooded in 800bce were farmland with scattered trees, or scrub-land, or in some cases heavily eroded land. Some areas that lost their trees were woodland once again by 700ce. There was, however, no general crisis of timber-supply in antiquity, not only because woodlands grow again in temperate climates and because demand created trading mechanisms: landowners also put a great deal of effort into looking after what they of course knew to be a valuable resource. Such an assertion does not imply any idealization of the classical world, nor should it obscure the law of unintended ecological consequences (we know, for example, about the ecological harm done by the Roman system of drainage in one section of the valley of the Rhône91). Our direct evidence about tree- and woodland-care in classical antiquity consists, admittedly, of advice from highly literate arboricultural experts such as Columella whose works cannot have been widely circulated: yet their advice arose ultimately from real experience and can therefore give some hints about actual practice. In real life, no doubt, practical woodsmen mostly passed on information and advice by word of mouth. You can in fact see knowledge and ambition increasing in the surviving texts. Theophrastus, writing in the late fourth century bce, is already impressive: consider for example his discussion of the merits and demer- its of various kinds of wood for charcoal-making,92 which also seems to reveal the existence of a sophisticated charcoal market. He apparently knew something about coppicing.93 He gives advice about grafting.94 But while he envisages the planting of fruit trees, he clearly sees almost all other trees as simply natural growths; he seems to make an exception for cypresses,95 presumably cultivated for the long timbers they provided for construction and shipbuilding. 150 years later, Cato is several degrees more interested in the cultivation of non-fruit trees; he explains how to plant willows, elms and pine-trees as well as cypresses (to which he too pays special attention).96 There is already a seminarium, a nursery, for young olive-trees on his ideal estate;97

91 Such at any rate is the argument of Van der Leeuw et al. 2005, 25. 92 Hist. Plant. 5.9. 93 I take this to be the reference of the word kolobon in Hist. Plant. 5.9.2. 94 Caus. Plant. 1.6, etc. Aristotle knew of this technique. 95 Hist. Plant. 2.7.1. 96 De agr. 9, 28, 151. See further Meiggs 1982, 262–263. Other references to woodland management include chs. 6, 7, 17, 37 end, 38 end, 55. 97 De agr. 45–46. defining and detecting mediterranean deforestation 191 this was a regular Roman institution. He gives detailed instructions about grafting olives, fig-, pear-, and apple-trees and vines.98 He recommends the owner of estates that are suburbani to produce firewood for the market (whereas the aim of planting tall trees is not the market but simply self- sufficiency).99 By the time we get to Varro, in the 30s bce, there has been another change: he refers to the Roman habit of planting big trees, specifically pines, cypresses and elms; he particularly recommends the latter—‘there is no better tree for planting; it is extremely profitable’.100 But why? Not, as we might have expected, because these are three of the four woods most used in ship-building,101 but for more complex reasons: an elm ‘often supports and gathers a certain number of baskets of grapes, yields most agreeable foliage for sheep and cattle, and provides stakes for fencing, and also for hearth and furnace’. At all events this passage demonstrates that Varro and his readers, who will have included substantial Roman landowners with estates in diverse parts of the Mediterranean, are completely familiar with the techniques necessary for planting such trees. Columella, like Cato, was a great enthusiast for vineyards, but he admits that there are ‘many’ landowners who preferred pasture land or silva cae- dua—timber-producing land for coppicing;102 he shows once again how familiar Roman landowners were with techniques for planting large trees.103 He also shows that by modern standards Roman landowners coppiced their trees on a very short cycle, a reflection of the powerful demand for wood fuel.104 In the same period the elder Pliny maintains that cypress branches, which go for a denarius apiece after twelve years of growth, are the most profitable kind of planting: this is because they can be used as vine-supports.105 He

98 De agr. 40–42. 99 De agr. 6 and 7. 100 RR 1.15 (‘maxime fructuosa’, correctly translated by Heurgon 1978 as ‘d’excellent rap- port’). Williams 2003, 97, was wholly mistaken to claim that there were no ‘examples of efforts to plant trees other than olive trees’. On Roman perceptions of the relative desirability of woodland and other kinds of land see Giardina 1981, 102–103. 101 Silver-fir would not come into his calculations in any case, since it is assumed that the ideal estate is not at a very great altitude. 102 De Re Rustica 3.3.1. 103 De Re Rustica 5.6 and De arboribus (though the latter concerns fruit-trees, and the author says at the beginning [1.1] that trees that provide materia, construction timber, grow without human aid). 104 See Meiggs 1982, 268. 105 NH 16.141. They are known as ‘a daughter’s dowry’, he says, twelve being his idea of the age at which a girl would marry. 192 w.v. harris takes it for granted that many different varieties of trees other than fruit trees are regularly propagated by landowners:106 Trees for coppicing, in addition to those which we have mentioned [willow, chestnut, aesculus oak and cypress] are the ash, laurel, peach, hazel, and apple, but these shoot more slowly and when fixed in the ground tolerate the soil, not to mention the damp, with difficulty. The elder, on the contrary, which is very strong timber for a vine-stake, is grown from cuttings like the poplar. Other species of non-fruit trees deemed worthy of cultivation include plane, elm, ash and alder.107 In short there is every reason to think that by the second century bce and even more in later times Roman estate-owners regarded many species of tree as useful assets, to be exploited and therefore to be propagated.

The Impact of Climate Change

In the temperate zone that makes up the majority of the area under discussion, the most important climatic factor that may have affected the lives of trees is the pattern of precipitation, and not only its volume but its seasonal distribution.108 Can we identify any Mediterranean climate changes in the period 800bce to 700ce that are likely to have had large effects on woodland, and in particular any changes in patterns of rain- and snow-fall? The sheer diversity of Mediterranean climates109 is bound to make this difficult. The papers by Michael McCormick and Sturt Manning in this volume both review the existing evidence. My reading of their work and of the work they cite is that as far as archaic Greece is concerned (the eighth and seventh centuries bce in particular), while it is very possible that precipitation increased and aridity declined across wide areas of the eastern Mediterranean, it is too early to assert this as a fact.110 Even if it were true, it is quite uncertain what the net effects on woodlands would have been,

106 NH 17.151. 107 NH 17.65–78. 108 Cf. Perry et al. 2008, 41–59, Manning this volume p. 107. 109 Cf. Grove and Rackham 2001, 25, who do not even consider the eastern or southern shores. 110 A recent review of the literature on Holocene climate in the eastern Mediterranean, which considered some eighty studies, concluded that ‘during the late Mid-Holocene (2800– 1800 yrs bp) diverging records prevent the emergence of a coherent picture’ (Finné et al. 2011, 3167). (bp for these authors means before 1950). There is still a chronic shortage of data with high-enough chronological resolution to be useful to a historian. defining and detecting mediterranean deforestation 193 since the hypothetical change in climate may have been a prime cause of the demographic expansion that undoubtedly took place in this period, at least among the Greeks (shades of Malthus). Greeks of this period greatly increased their production of both metal artefacts and ships. A modest increase in precipitation may perhaps have caused more loss of woodland than gain. It is a matter of scale, and we need more numbers. We seem to be on more solid ground in the third century ce and later. McCormick’s group considers that precipitation declined in Rome’s eastern provinces and in north-eastern Gaul in the third century and recovered in the fourth.111 Another dry period sets in in north-eastern France about 450 and continues for about two centuries.112 In the eastern provinces, according to this group, a humid climate continues until ‘it changes dramatically in the sixth century’.113 These changes coincide to some extent with the pollen evidence reviewed earlier (but they appear to be partly based on the pollen evidence). The most serious difficulty is that north-eastern France is merely a fragment of the area we are trying to generalize about, while the eastern provinces of the Roman Empire cover a huge and ecologically diverse area where we would need data from many different places. And a recent literature review seems to show that the eastern-Mediterranean climate grew wetter not drier from 550 to 750ce.114

Conclusions

Much woodland was degraded or disappeared in the Graeco-Roman Medi- terranean. But no extreme hypothesis about deforestation seems well- founded, and there is no reason to believe in a generalized crisis (though Italian wood being used for fuel in Egypt comes close to suggesting that). In truth, the uncertainties and unknowns are very extensive. Both scientists and historians should work harder to achieve a clearer definition of deforestation, or a typology of deforestations.115 And, the problem of gen- eralization is fundamental: we need more discussion of, on the one hand,

111 Büntgen et al. 2011; McCormick, this volume, p. 70. 112 Ibid. 113 McCormick, this volume, p. 71. 114 Finné et al. 2011 (see above, n. 110), 3168 (‘a widespread period of wetter conditions at 1400–1200 yrs bp’). See their Fig. 1. Their ‘eastern Mediterranean’ includes Italy, and they have no data from Syria or Egypt. More precipitation from 550ce onwards at a site in Switzerland: Roos-Barraclough et al. 2004. 115 It is depressing to read an account by a classicist who, while citing some scientific studies in his bibliography, makes no systematic use of them (Thommen 2012, 37–41, 85–89). 194 w.v. harris the representativeness of the scientific evidence and on the other the real value of the textual sources. Thus the following claims are tentative: (1) Classical Greek deforestation, in a weak sense of the term (land clearance, with some trees left behind) was widespread, but in a strong sense of the term it was probably restricted to Attica and the immediate supply area of a few other cities. Local fuel crises probably occurred wherever metal-smelting was intense (so for this reason too Attica was under pressure in the fifth and fourth centuries). (2) Hellenistic and Roman demand intensified in the third century bce. This demand was met to some extent by a stronger trading network and more careful woodland management, but ship-building and fuel- needs are likely to have deforested some areas quite seriously—most of all perhaps in Italy—, and left others with much less woodland coverage. (3) Stress on woodland resources continued to mount all the way through the high-imperial period of prosperity, perhaps down to Severan times. (4) Alleviation that came later was highly selective. Some areas regained woodland, but whatever the reasons were (unfavourable climate change, poorer management) others did not. PART FOUR

AREA REPORTS

PROBLEMS OF RELATING ENVIRONMENTAL HISTORY TO HUMAN SETTLEMENT IN THE CLASSICAL AND LATE CLASSICAL PERIODS— THE EXAMPLE OF SOUTHERN JORDAN*

Paula Kouki

Introduction

One of the central questions in Mediterranean archaeology is the relationship between humans and their environment. The issue of environmental change and the factors behind it, as well as its influence on human settlement and land use, is, therefore, a valid topic of research for historians and archaeologists working in the region. The dating of archaeological sites is usually based on pottery. For the Classical and Late Classical periods pottery is closely datable, often within a century, sometimes even within a few decades, which makes it possible to reconstruct relatively detailed settlement histories. From ancient written sources we may even obtain a picture of the political and economic conditions of a given region. However, the dating resolution of the results of palaeoenvironmental and climatological studies, such as sedimentary or pollen analyses, is generally within a couple of hundred years at best, which makes it complicated to relate environmental changes to those in human settlement and land use. Furthermore, different records of environmental data may produce widely differing pictures, for example, of climatic trends. Some of these problems may be overcome by means of high-resolution environmental data,1 but such data are not yet widely available. There are also regions where the availability of palaeoenvironmental data is limited due to local conditions, such as the lack of waterlogged deposits and sedimentary basins. This necessitates the use of proxies, which creates its own problems.

* Acknowledgements. This research has been carried out under the Finnish Academy’s Ancient Greek Written Sources Centre of Excellence in Research. I would like to thank the following foundations for funding my research: the Jenny and Antti Wihuri Foundation, the Finnish Cultural Foundation, the Emil Aaltonen Foundation and the Research Foundation of the University of Helsinki. 1 Manning, this volume. 198 paula kouki

In this article, I address these problems and discuss them in the light of research material from southern Jordan. After providing a background description of the environmental conditions and archaeological research there, I address the problems created by the lack of local palaeoenvironmental data as well as the limitations of the use of proxies and their reliability. I then briefly present the results of my study comparing the archaeological settlement patterns and evidence of climatic change for the Late Hellenistic, Roman and Byzantine periods in the Petra region, southern Jordan. Finally, I would like to call into question the notion of ‘favourable/unfavourable climate’, which has been rather indiscriminately used in archaeological and historical explanation in the past.

Background

The study discussed in this article is part of my recently finished PhD research, in which I studied the settlement and land-use change in the hinterland of the ancient city of Petra, southern Jordan (Fig. 1), from c. 300bc until c. ad700. The research was carried out under the Finnish Jabal Harun Project (henceforth the FJHP), an archaeological project under the research umbrella of the Finnish Academy’s Ancient Greek Written Sources Centre of Excellence in Research.2 The FJHP consists of two interrelated parts: the excavation of a Byzantine monastery/pilgrimage centre on the top plateau of Jabal Harun, c. 4km SW of Petra, and the intensive archaeological survey of the immediate surroundings of the mountain. The present-day climate of the Petra region is semi-arid, with dry, hot summers and winter rains. There is considerable change in the relief within a distance of only 15km from the floor of Wadi ʿAraba at c. 100m asl to the elevation of 1700m at the highest parts of Jabal ash-Shara on the western edge of the Jordanian Highland. Rainfall is mostly generated when moist air masses are lifted over the Jabal ash-Shara range, resulting in the highest mean annual rainfall, c. 200mm, in the area. Towards the west and east the rainfall decreases, being only c. 50mm on the floor of Wadi ʿAraba and in Maʾan in the eastern pre-desert zone.3 Due to the low rainfall, agriculture is not possible in most parts of the region without some means of irrigation. Extensive ancient systems of rainwater collection and runoff cultivation are known in the region, dating at least from the Nabataean period onwards.

2 2000–2005, 2006–2011. 3 Lane and Bousquet 1994, 22; Meteorological Department of the Hashemite Kingdom of Jordan 2004. relating environmental history to human settlement 199

Fig. 1. The Petra region, southern Jordan.

At the turn of the Common Era, southern Jordan belonged to the Naba- taean kingdom. Within the last couple of centuries bc, the formerly nomadic Nabataeans had established their capital city in Petra and had begun to settle the surrounding areas. In the 1st to early 2nd century ad an unprecedented expansion of sedentary settlement and agriculture took place in the Petra region, attested in the archaeological record by the appearance of numerous farmsteads, hamlets and villages. The Nabataean kingdom was annexed by Rome in ad106, and Petra became the capital of the province of Arabia. The annexation apparently did not cause any large-scale disruption in the land-holding system, and the region continued to prosper under the Roman rule. However, a considerable reduction in the number of rural sites is visible from the 3rd century ad onwards.4

4 Kouki 2009, 35–39. 200 paula kouki

In earlier scholarship, the initial expansion of settlement and agriculture by the Nabataeans has been related to a favourable climatic phase.5 The opposite phenomenon observed in the archaeological record for the Late Roman and particularly the Byzantine period (Table 1) has been explained by environmental degradation as the result of over-irrigation and erosion that resulted in the reversal of the population into nomadism.6 The assumption is reasonable, since the entire region can be considered environmentally marginal for agriculture and settlement, and a considerable part of its population has until recently consisted of mobile or semi-sedentary Bedouin. However, there has not been much actual research into the question. Therefore I made an attempt to find out whether climatic change could indeed explain the archaeologically observable changes in the settlement pattern.

Table 1. The relevant archaeological periods in Jordan. period dating Hellenistic 332–63bc Nabataean–Early Roman 63bc–ad106 Late Roman ad106–324 Early Byzantine ad324–491 Late Byzantine ad491–634 Islamic ad634 onwards

Environmental Data

The immediate problem when dealing with environmental change in a semi-arid to arid area such as southern Jordan is the lack of local sources for environmental data. No wide-ranging environmental studies have been carried out in the Petra region, and most of those dealing with certain localities or sites are concerned with earlier periods. The available data is mainly based on sedimentary records, and its temporal resolution is low.The same goes for southern Jordan in general: the data are sporadic, and their use as a climatic indicator is further complicated by the contexts where they are

5 MacDonald 2001, 375–376. 6 Hart 1986, 58; Fiema 2006, 82. relating environmental history to human settlement 201

Fig. 2. A modern barrage in the Jabal Harun area in September 2011. obtained, since these are often affected by human activity. High-resolution data from a tree-ring series are available from the Dana area, but they cover only the recent historical period.7 For example, there are pollen series from sediments deposited behind a large barrage or reservoir in a tributary of Wadi Faynan. The presence of algae spores in the sediments dated to the Byzantine period has led to the suggestion that the climate of southern Jordan must have been more humid at the time to enable the presence of standing water throughout the year.8 The hydrological role and purpose of this structure is uncertain,9 but there are other such walls in the same area. In the Jabal Harun area we have been able to observe that one heavy winter storm has been enough to fill a modern barrage in a wadi at the foot of the mountain to the extent that there is still shallow water behind the dam in late September (Fig. 2). Furthermore, the results from agricultural experiments in the Negev indicate that the

7 Touchan and Hughes 1999. 8 Hunt et al. 2007, 1329–1330. 9 Crook 2009, 2433. The suggestion is that these constructions are related to ore processing rather than agricultural pursuits, see Newson et al. 2007, 163–164. 202 paula kouki mean annual rainfall of c. 100mm can be transformed to c. 300–500mm of effective precipitation by intensive collection of runoff.10 Supposing that the Khirbat Faynan barrage system was maintained, it can be questioned whether the rainfall really had to be significantly more abundant than at present to provide water behind the barrage year round. Small-scale environmental studies were also carried out by the FJHP survey. In the Jabal Harun area the most significant group of structures in the area consists of those related to runoff farming and terrace cultivation: prac- tically every possible wadi and slope has been farmed at some point of time. Although the runoff-cultivation systems are notoriously difficult to date, the dating of the off-site scatter of pottery and the functional interdependence of the structures strongly suggest that at least the large barrage system to the west of the mountain was built as a planned enterprise in the Nabataean period, in the 1st century ad.11 By what is known about the economics of contemporary monastic communities in Palestine and Egypt,12 it could reasonably be expected that the use of the field system continued while the monastery was occupied, that is, from the 5th to the 9th century at least.13 However, very little off-site pottery datable to these later periods was found in the area of the field system. Therefore we attempted to obtain evidence of the erosion and sedimentation history of the area in order to recognize the potential construction phases of the runoff-cultivation structures. Sed- imentary structures were documented in three wadi terraces and sampled for OSL dating.14 The results of the OSL dating were disappointing: for the most part, the sediments were not completely bleached before deposition, and therefore the dates obtained were not indicative of the timing of the deposition processes. Furthermore, even in the case where we had apparently reliable dates from well-bleached sediments, the uncertainties of the dates—within a couple of hundred years at best—were too large to confidently associate the deposition of the sediment with any one period of human activity. The sedimentary structures were mostly indicative of very rapid depositional events, high sediment loads and short transport distances. The situation is typical of a semi-arid headwater area, which largely explains the poor results

10 Evenari et al. 1971, 109; Bruins 1986, 38–44. 11 Lavento et al. 2004, 166–167. 12 Hirschfeld 1992, 103–104. 13 Fiema 2008, 434–436. 14 Grün 2001. relating environmental history to human settlement 203 of the OSL dating15 and reduces the possibility of correlating the results with a wider-scale climate or environmental change. The unavailability of local palaeoenvironmental data necessitated the use of proxy data, which are mainly available from the Dead Sea and northern Israel.16 The use of such data is, of course, not straightforward because of the changes in climatic controls when the distance from the Mediterranean coast increases—the climate of Israel is much more affected by the circulation over the Mediterranean than that of southern Jordan, which besides the Mediterranean circulation is affected by the Red Sea Trough, resulting in greater climatic instability.17 Nevertheless, since at least the general trends in the climate over the Levant are considered to have been broadly similar in the past as in the present, with the coastal and northern areas receiving more rainfall than the southern parts, I considered it possible to infer regional-scale climatic changes from the proxies also for southern Jordan. However the proxy data are not without their problems, especially in the case of relatively recent periods like the Roman and Byzantine periods. The emphasis of palaeoenvironmental research in the Near East has largely been on the Late Pleistocene and earlier Holocene, which means that less information is available for the more recent times. The climatic changes of the historical period have also been of lesser magnitude than those of the Pleis- tocene and Early Holocene, and thus they are more difficult to discern in the palaeoclimatic records. A further problem is the resolution of the available data. While the resolution of a couple of hundred centuries may be an excellent achievement for palaeoclimatologists, it makes all the difference when working within the closely dated Classical and Late Classical periods. Finally, the results of different palaeoclimatic studies are ambiguous, and the scientists themselves disagree on the timing and scale of the aridification in southern Levant. While some propose that it began as early as in the fourth century ad,18 others push its start as late as the seventh century.19 On the other hand, the Talmudic sources point towards crop failures, hunger and pestilence in the third century,20 which is not reflected in the available environmental data.

15 For more detailed results of the OSL analyses, see Kouki 2006, 147–153. 16 E.g., Bar-Matthews et al. 1997, 1999; Bookman et al. 2004; Bruins 1994; Enzel et al. 2003; Frumkin 1997; Heim et al. 1997; Hirschfeld 2004; Orland et al. 2009. 17 Freiwan and Kadioglu 2008, 525–529, 533–534. 18 Bruins 1994, 308. 19 Hirschfeld 2004, 133. 20 Sperber 1978, 70–99; cf. Bruins 1986, 307–308. 204 paula kouki

Furthermore, the reliability of even these data is questionable. It is generally agreed that pollen is not a very reliable indicator of the climate for the historical periods in the Near East due to the widely spread and long-lasting influence of human activity on the landscape. If we consider the Dead Sea, the reliability of the sedimentological record and lake levels can also be questioned. It is well known that currently the level of the Dead Sea is lower than ever before due to the increased consumption of water by the neigh- bouring countries, which results in very little of the flow actually reaching the Dead Sea basin. If we consider the scale of agriculture and land use in the Levant in the Roman and Byzantine periods, could it not be possible that it also had some, albeit smaller, impact on the Dead Sea levels and the sedimentary record?

Settlement and Climatic Change in the Petra Region

The pottery sequences from the Petra region are relatively well known for the Nabataean-Roman and Byzantine periods, although there is some uncertainty concerning the pottery of the 2nd–3rd centuries and the Late Byzantine—Early Islamic transition (late 6th–early 7th century). Neverthe- less, the pottery chronology enables the dating of the archaeological sites within the accuracy of a century or two, which makes it possible to compare the trends in settlement distribution and climate on a centennial scale. To establish the potential influence of climate change on settlement patterns, I divided the Petra region into four rough environmental zones (Fig. 3). The Jabal ash-Shara zone is environmentally the most favourable, with the highest mean annual rainfall and most springs. It is the only part of the region where dry farming of cereals is possible under the current climatic regime. Until recently, the sedentary villages in the region have been mainly limited to this zone. The narrow Escarpment zone receives less rainfall, but it is fed water by the large wadis running east-west, and its topography and soils are particularly well-suited for runoff cultivation,21 for which there is also abundant archaeological evidence from at least the Nabataean period onwards. The environment of the Wadi ʿAraba and Eastern Highlands zones is the most unfavourable, with the lowest mean annual rainfall and the highest spatial and temporal diversity in the distribution of rainfall.22 As a result,

21 Bruins 1986, 39–43; Yair and Kossovsky 2002, 55. 22 Freiwan and Kadioglu 2008, 529. relating environmental history to human settlement 205

Fig. 3. Environmental zones in the Petra region. these zones are likely to be severely affected by a decrease in rainfall caused by the aridification of the climate. I then compared the centennial distribution of settlement through these zones, combined from three different surveys recently carried out in the Petra region,23 to the climate history reconstructed from the proxies. My hypothesis was that in the case of adverse climatic conditions (i.e., aridification), settlement and agriculture would withdraw particularly from the environmentally marginal zones of Wadi ʿAraba and Eastern Highlands. Accord- ing to the same principle, the Jabal ash-Shara zone should continuously be the most densely settled. The palaeoclimatic evidence for the relevant period is presented in Table 2 together with the settlement trends. The results were somewhat surprising. The expansion of rural settlement took place mainly in the Jabal ash- Shara and Escarpment zones as predicted (Fig. 4), but the intensification of settlement and agriculture lagged behind the improved climatic conditions

23 ʿAmr et al. 1998; ʿAmr and al-Momani 2001; Abudanh 2006; and the unpublished data from the FJHP survey. 206 paula kouki by at least two centuries. The increased rainfall24 set good conditions for the intensification of agricultural production in the Petra region. However, the intensification of rural settlement lagged behind the beginning of this humid period by a couple of centuries. The main expansion of rural settlement occurred only in the 1st century ad. If there was a drop in rainfall already at the turn of the 1st–2nd centuries ad, as has been suggested,25 it apparently did not have an immediate impact on the rural settlement.

Table 2. A reconstruction of the climate history of southern Levant and the rural settlement in the Petra region. time climate settlement in Petra region ad800 climate very arid further concentration of rural ad700 settlement into a few large sites

end of Petra as a city?

abandonment of many rural sites in ad600 gradually increasing aridity the late 6th–7th centuries through the 6th and 7th centuries increase of rural settlement in the eastern Jabal ash-Shara and Eastern Highlands during the 6th century ad500 aridification of climate starts in the 5th century agricultural sites established in the Eastern Highlands in the 5th century ad400 possible drop in rainfall around ad400 discontinuity of rural settlement drop in the number of sites ad300 a return to more humid conditions in the 3rd century widespread abandonment of rural ad200 settlements by the early 3rd century ad100 a possible drop in rainfall around ad100

24 Frumkin 1997, 244; Bookman et al. 2004, 566. 25 Frumkin 1997, 244; Orland et al. 2009, 34. relating environmental history to human settlement 207

Fig. 4. The distribution of rural settlement in different environmental zones in the 1st century ad.

time climate settlement in Petra region proliferation of rural settlement and intensification of land-use in the humid climatic phase culmi- the 1st–2nd centuries nates around the turn of the era first rural settlements elsewhere

100bc climate more humid than at present first rural settlements in the Jabal ash-Sharah area 200bc climate becomes more humid, first permanent settlement in Petra probable cooling 300bc before an arid climatic phase no permanent settlement 208 paula kouki

Fig. 5. The distribution of rural settlement in different environmental zones in the 3rd century ad.

However, although relative humidity apparently increased again in the late 2nd–3rd centuries, a widespread reduction of rural settlement appears to have taken place in all environmental zones over the same period (Fig. 5). By the 4th century, most agricultural settlements had withdrawn from both the western and the eastern peripheries of the region, and rural settlement had largely become concentrated in the Jabal ash-Shara zone. The palaeoclimatic data concerning the 4th century are somewhat ambiguous, but most studies support the interpretation that the climate was still relatively humid.26 Most palaeoclimatic data point towards increasing aridity from the 5th century onwards.27 The 5th and particularly 6th centuries witnessed a reduction of the number of rural settlements in the western part of the Jabal ash- Shara zone. However, this trajectory was accompanied with an expansion of

26 Frumkin 1997, 244; Orland et al. 2009, 33–34; Enzel et al. 2003, 268; Bookman et al. 2004, 566. 27 Frumkin 1997, 244; Orland et al. 2009, 34; Enzel et al. 2003, 268, cf. Bookman et al. 2004, 566. relating environmental history to human settlement 209

Fig. 6. The distribution of rural settlement in different environmental zones in the 6th century ad. settlement and agriculture into the Eastern Highlands zone, with the estab- lishment of large agricultural settlements in the desert fringe, particularly in the 6th century (Fig. 6). Such development seems to match poorly with the evidence for a drying climate, when an opposite phenomenon could be expected. Finally, a considerable reduction of settlement again took place in the 7th century (Fig. 7), with the concentration of settlement into a few large and medium sites. In the eastern part of the Jabal ash-Shara zone, most of the small settlements established in the previous century were abandoned. In the western part of Jabl ash-Shara, the village settlements seem to become localized in the areas of modern Wadi Musa and at-Tayyiba. However, the large agricultural settlements in the Eastern Highlands zone show continuity through the 7th century and even beyond, which was not to be expected in the light of the evidence of further aridification.28

28 Frumkin 1997, 244; Enzel et al. 2003, 268; Bookman et al. 2004, 566–567. 210 paula kouki

Fig. 7. The distribution of rural settlement in different environmental zones in the 7th century ad.

Conclusion

As can be seen from the above, I was not able to establish a connection between the archaeological settlement patterns and climatic change in the Petra region for the Nabataean-Roman through the Byzantine periods. Although the initial expansion of rural settlement and agriculture takes place during a period of ‘favourable’ climate, it only begins a couple of centuries later than this wetter climatic phase. On the other hand, the reduction of rural settlement begins during what can be considered a still relatively favourable climatic phase. Furthermore, although most scholars agree that aridification had begun by the later 5th century at the latest, there is a phenomenon of new agricultural settlements being established in the eastern part of the Petra region, in the area that lies in the desert margins, in the 5th and 6th centuries. Actually, the focus of the permanent settlement apparently shifted towards the east in the 5th century, and this eastern focus became even more pronounced in the 6th and 7th centuries. Two obvious alternative conclusions can be drawn from these results: relating environmental history to human settlement 211

1) The reconstruction of the past climate is correct. The settlement change observed is not related to climatic change, but to other factors such as political, economic and social changes. 2) The reconstruction of the past climate is faulty. The resolution of the available climatic data is too low or the proxies used are unsuitable for the reconstruction of the climate of southern Jordan, and thus no conclusions can be made of the relationship of climate and settlement based on the available data. A third possible explanation is that the notion of ‘favourable climate’ is in itself at fault. To conclude, I would like to consider this possibility a little more closely. The environmental data available for the southern Levant is mainly indicative of changes in rainfall and/or runoff. The periods for which increased rainfall is implied are usually considered ‘favourable’, while those with less rainfall are considered unfavourable. I would like to argue, however, that the definition of what is ‘favourable climate’ is not so straightforward, and that it is not possible to reduce it into a simple function of the mean annual rainfall. The timing, spatial distribution and mode of rainfall are equally important. The temperature both during and after the rainy season affects evaporation and thus influences the actual available moisture. Furthermore, local topography, geology and soils influence the outcome through the differential generation of runoff, absorption of rainwater into soils and erosion potential. Even short-lived storms can generate a lot of runoff and potentially catastrophic erosion and hazardous floods in a sparsely vegetated region, but they also enable the effective collection of runoff for later use. On the other hand, long-lasting but relatively low intensity rainfall is more useful in areas with thicker soils—provided that the temperatures are low enough that the moisture will be retained in the soil and does not evaporate. The Petra region exemplifies that these conditions can vary considerably even within a relatively small geographical area. Finally, and perhaps most importantly, we should not forget that the response of human communities to changing environmental conditions is not mechanistic and unchangeable. From the viewpoint of people, the optimal characteristics of rainfall depend on the cultivated plants and available technologies and techniques used for irrigation and water storage, as well as on traditions of land use. Thus we need not only more detailed environmental and palaeoclimatic data but also a more thorough picture of ancient societies to be able to evaluate the impact of climatic change on a given society with any confidence.

HUMAN-ENVIRONMENT INTERACTIONS IN THE SOUTHERN TYRRHENIAN COASTAL AREA: HYPOTHESES FROM NEAPOLIS AND ELEA-VELIA

Elda Russo Ermolli, Paola Romano and Maria Rosaria Ruello

1. Introduction

Reconstructing past landscapes is a challenging task, especially when one is dealing with those regions that have witnessed the presence of man since prehistoric times. In these areas the natural evolution of the environments, driven by climatic and/or endogenous factors, is strictly and deeply related to the continuous land use and management by man. On this assumption, it has become clear that any attempt of landscape reconstruction in archaeological contexts has to rely on the systematic integration of archaeology, geomorphology and other scientific disciplines. Many recent studies1 have shown, for instance, that humans have influenced large-scale sediment production and movement, and can be considered, like climate, as one of the main agents of sediment movement since the first millennium bc. Detailed climatic reconstructions of the last 2500 years were recently provided for central Europe, thanks to a tree ring-based methodology.2 Wet and warm summers occurred during periods of Roman and medieval prosperity while increased climate variability from ~250 to 600ad coincided with the demise of the western Roman Empire and the turmoil of the Migra- tion Period. Such detailed climatic reconstructions are not yet available for the Mediterranean area, where the climatic events could have had a different character and/or intensity from those recorded in central Europe. So, in this region, linking the response of geomorphic systems to the interaction between land use and extreme climatic events is not straightforward. Increasing the acquisition of knowledge about this matter is the only prag- matic way to come near to a possible solution to the debate, and the present contribution represents a further step towards the clarification of the long- lasting question: man, climate or both?

1 e.g. Casana 2008, 431–439. 2 Büntgen et al. 2011, 579–582. 214 elda russo ermolli, paola romano, maria rosaria ruello

Fig. 1. Location map of the study area.

2. The Study Areas

This study focuses on two archaeological sites of the Greco-Roman period located along the western coasts of southern Italy. In this part of the peninsula, facing the Tyrrhenian Sea, several archaeological sites of human settlements dating back to the period of Greek colonization are present.

2.1. Parthenope-Neapolis The first site is located in the modern town of Naples where the Greco- Roman town of Parthenope-Neapolis was situated. The landscape of the area is characterized by favorable conditions that have facilitated human settlements since Neolithic times: cliffed promontories alternating with narrow coastal plains offered timber sources and, at the same time, protected land- ing places. This landscape was inherited by the third epoch (4800–3800 yr bp) of activity of the Campi Flegrei volcanic field3 which led to the formation

3 Di Vito et al. 1999, 244. human-environment interactions: neapolis and elea-velia 215 of tuff ring and cones, as well as of straight fault scarps (Fig. 1). The volcanic activity of this area started about 50 thousand years ago and is still active today;4 the last eruption occurred as recently as 1538 and caused in the space of a few days the emergence of Monte Nuovo. To the east, the landscape is almost flat and gradually passes to the Vesuvius foot-slope. This volcano, more than 1,000 meters high, is also of course still active.5 Parthenope was settled by the Greeks during the eighth century bc on the hilltop of Monte Echia promontory6 as a small outpost belonging to the northern town of Cuma. Subsequently, at the end of the sixth–beginning of the fifth century bc,7 the major Greek town of Neapolis was established at the foot-slope of Monte Echia, leading to the progressive abandonment of the military outpost which was renamed Palaeopolis. Neapolis developed as a municipium and then as a colonia under the Roman Empire. Dur- ing Greco-Roman times port activities developed along the coastal strip of Neapolis, favored by the presence of a protected bay. The ancient port was discovered because of construction work for Naples’ new underground rail- way.8

2.2. Elea-Velia The second site is the Greco-Roman town of Elea-Velia on the coast of the Cilento, about 200km south of Neapolis. This sector of southern Italy is characterized by a gently-sloping hilly landscape, reaching hundreds of meters, made up by soft silica-clastic sediments of Mesozoic and Cenozoic age.9 Pliocene cemented conglomerates unconformably cover the soft bedrock, leading to the emergence of some rocky ridges. The town of Elea-Velia was founded around one of these rocky hills in the sixth century bc. The “acropolis” rose on the hilltop of a protruding cliffed promontory at 200m asl. Subsequently, the town developed in the lowland, at the mouth of the Frittolo river, where a densely urbanized area, namely the Southern Quarter, was built.10 The entire history of the Greco-Roman town was influenced by the Frittolo stream dynamics which caused several

4 De Vivo et al. 2010, 10–11. 5 Sigurdsson et al. 1985, 332–384. 6 Giampaola 2011. 7 D’Agostino and Giampaola 2005, 63–72. 8 Giampaola et al. 2006, 51; Amato et al. 2009, 26–30; Carsana et al. 2009, 17–21. 9 Bonardi et al. 1988. 10 Greco 2003. 216 elda russo ermolli, paola romano, maria rosaria ruello alluvial floodings in the Southern Quarter. Episodes of flood were recorded until late ancient times,11 but the absence of subsequent historical reports about Velia’s history points to the town’s disappearance beneath sediments and shrubs during the Middle Ages. Archaeological excavations carried out during the twentieth century removed some meters of the thick alluvial strata, unearthing the ruins of the Southern Quarter.

2.3. The Landscape In both areas the landscape was characterized by a rich vegetation cover. This conclusion relies on pollen data obtained from the analysis of a sea core (C106) in the Salerno Gulf.12 The position of this core is equidistant from the two studied sites (Fig. 1), and moreover this position allows the pollen rain from a very wide area to be represented in the analyzed sediments. Indeed, the source area for pollen can be considered the whole catchment basin of the river Sele and other minor basins. The pollen diagram of the Salerno core covers the last 30,000 years (Fig. 2). testifying to the transition from the Last Glacial period to the Holocene. Before Greek colonization, pollen data clearly show that a deciduous forest dominated by oaks probably occupied all the slopes surrounding the study sites. Mediterranean maquis was present as well on the steepest and sun- niest slopes and close to the coast on thin soils. On mountain tops above 1000m a fir forest was present which has now almost completely disappeared from the southern Apennines, to be replaced by beech woods.

3. Methodological Approaches

This study concerned different aspects of both the natural and human landscape and involved researchers in geological, environmental and archaeological sciences. The main goal was to reconstruct the ways and times of environmental changes and to understand their causes. For this purpose three steps in the evolution of these two coastal sectors were chosen, representing the main detectable changes in the environmental features.

11 Amato et al. 2010, 14–15. 12 Russo Ermolli and Di Pasquale 2002, 211–219; Di Donato et al. 2008, 153–156. human-environment interactions: neapolis and elea-velia 217 Fig . 2. Pollen diagram from the C106pollen; core LGM= in last the Salerno glacial Gulf. maximum; Taxa B/A= percentages Bolling/Allerod are plotted period; against YD= age Younger (ka Dryas = cold thousand spell years). (modified AP= after arboreal Di Donato et al., 2008) 218 elda russo ermolli, paola romano, maria rosaria ruello

The first step concerns the landscape setting prior to Greek colonization in the first millennium bc; the second step concerns the Greco-Roman period; while the third step focuses on the Late Ancient period, from the third century ad onwards. The methods used to achieve these goals can be summarized in three points: 1. the first approach is normally the collection of direct data from the excavation areas and from boreholes; 2. secondly, these data are integrated to reconstruct the 3-dimensional arrangement of the strata and their bounding surfaces and 3. finally, all the local reconstructions are integrated to obtain a wider paleogeographical reconstruction.

4. Evolution of the Neapolis Area

The environmental evolution of the Neapolis region is shown in Fig. 3 through three-dimensional schemes obtained from a digital elevation model (DEM) and from cross sections.

4.1. Step 1—First Millennium bc In the first step of environmental evolution, the coastal profile of the Neapo- lis region was located further inland with respect to the modern one (Fig. 3A). Promontories and inlets characterized the coast which was formed by tufaceous cliffs that in some places were tens of meters high. Slopes were cut by a dense fluvial network. The section of Fig. 3A’ was built along the line indicated in Fig. 3A thanks to the data coming from boreholes and trenches dug during the construction of the new underground line of Naples. In this section, cutting the piazza facing the modern port, it is possible to visualize the ancient coastline which was about 500 meters inland with respect to the present one. The tufaceous cliffs directly faced the sea without any beach strips. The bedrock and the cliffs were made up by the Neapolitan Yellow Tuff which was emplaced 15,000 years ago from one of the main eruptions of the Campi Flegrei volcanic fields.13

13 Deino et al. 2004, 168. human-environment interactions: neapolis and elea-velia 219

Data from boreholes and archeological sections14 allowed the morphol- ogy of the bedrock to be reconstructed as well as the thickness of marine deposits mainly concentrated in the depressions.

4.2. Step 2—Greco-Roman Period The second evolutionary step of the coastal environment of Neapolis (Fig. 3B) shows the landscape towards the end of the second century ad, after the town’s foundation. In the sixth century bc Neapolis was founded on the terraced surface gently dipping towards the sea. Its shape was driven by natural features: it was bounded by streams at the sides and by a cliff toward the sea. The city walls more or less followed these natural boundaries. In Fig. 3B it is also possible to see the main streams and the main Greco-Roman roads—at least those that have been discovered. They led towards the western region where other towns were established, such as Puteoli and Cuma (Fig. 1). The first nucleus of Neapolis now represents the historical center of Naples where the main streets still follow the traces of the Greco-Roman streets and where much archaeological material has been found. Concerning the coastal landscape, a sheltered bay was located at the cliff base in the western side of the gulf. A small tufaceous promontory protected the bay from the main sea storms coming from southwest. In this bay, port activities probably started from the fourth century bc and went on until the fourth century ad when this part of the bay dried up. The main environmental difference with respect to the previous stage is the formation of a sandy beach in the eastern coastal sector, which was occupied by human activities that developed in this period also outside the city walls. In fact, remains of a temple and of the gymnasium were found on the ancient beach.15 The section of Fig. 3B’ refers to this second stage and shows the beach that formed at the cliff toe and the construction of the port. The latter, as previously noted, was located in a sheltered inlet, and the promontory which protected the bay from the southwestern storms is also visible in the section. The beginning of port activities started between the fourth and the second century bc; traces of dredging carried out at this period were found on the sea floor (Fig. 3B’). Dredging was carried out to lower the sea bottom and to make the inner part of the basin suitable for shipping. A wooden dock is dated to the first century ad while two shipwrecks were found in the sediments dated to the turn of the first and second centuries ad (Fig. 3B’).

14 Amato et al. 2009, 25–30. 15 Giampaola 2004, 44. 220 elda russo ermolli, paola romano, maria rosaria ruello human-environment interactions: neapolis and elea-velia 221

Fig. 3. Three-step scheme of the evolution of the Neapolis area. A= Step 1—I millennium bc; A’= cross section related to Step 1; B= Step 2—Greco-Roman period; B’= cross section related to Step 2; C and D= Step 3—since the Late Ancient; C’ and D’= cross sections related to Step 3 (after Russo Ermolli et al., in prep.). 222 elda russo ermolli, paola romano, maria rosaria ruello

The many repairs to the timber of these ships, together with the absence of any cargo, suggests that these ships were probably abandoned on purpose.16 Wood analysis of the shipwrecks revealed the systematic use of walnut and cypress timber. This datum, that is the use of local plants, together with the modest dimension of the ships, suggests that the location of shipyard has to be restricted to the central-southern Tyrrhenian coasts.17

4.2.1. The Landscape of Neapolis through Pollen Analysis The sediments of the port were excavated for about 7 meters. The rich archeological content allowed the sediments to be dated from the second century bc to the fifth century ad.18 These deposits were studied in detail through different methods with the aim of reconstructing the underwater environment as well as the terrestrial landscape surrounding the site. For this purpose sediment samples were collected for pollen analysis. In Fig. 4 the results of pollen analysis are shown in a detailed diagram where all the recognized taxa are grouped in vegetation associations marked by different grey shades. It is necessary to underline that pollen spectra from very urbanized areas are the sum of natural and anthropogenic plants and thus their interpretation has to be very cautious, especially when one is dealing with plants that could be either wild or cultivated. This applies, for instance, to hazelnut, olive and chestnut. Another problem concerns the herbs which are only determined at the family rank and families such as Poaceae, Brassicaceae and Fabaceae could include both wild and cultivated varieties. On the whole, it is possible to say that from the first century bc to the second century ad a deciduous oak forest was present on the slopes surrounding the Neapolis harbor and that the Mediterranean maquis probably occupied the most sunny and rocky sectors, especially close to the sea coast. These data confirm the landscape reconstruction based on the pollen diagram from the Salerno Gulf (cfr. 2.3). The tree crops mainly consisted of walnut and secondly of chestnut and grapevine. Concerning the horticultural practices, the presumed cultivated varieties are dominated by the Bras- sicaceae family, most likely representing cabbage cultivation, which was rather common in Roman times and represented one of the main plant food sources for the Romans.

16 Giampaola et al. 2006, 63–76. 17 Allevato et al. 2010, 2373–2374. 18 Giampaola et al. 2006, 53–57. human-environment interactions: neapolis and elea-velia 223 port sediments. Taxa percentages are plotted against depth. DF= deciduous forest; TC= tree crops; Neapolis Fig . 4. Pollen diagram from the MM= Mediterranean maquis; MF= montaneErmolli et forest; al., HS= in herbs; prep.). HT= horticultural; WT= water plants; AP= arboreal pollen (after Russo 224 elda russo ermolli, paola romano, maria rosaria ruello

Very few pollen are representative of the mountain forest which was located rather far from Neapolis, on mountain tops over 1000 meters and probably also on Vesuvius which was higher than today.

4.3. Step 3—Since the Late Ancient In the third step of environmental evolution at Neapolis (Fig. 3C), the main episode is the closing of the port area and the formation of a lagoon, first connected to the sea and then isolated. This closure took place in the fifth century ad thanks to the growth of a beach ridge. Port activities in this part of the bay ended even before, during the fourth century, when the town walls expanded westward. Towards the end of the fifth century the bay was completely filled up, and the site was used as a farmland from the beginning of the sixth century. The section of Fig. 3C’ shows the lagoon and the beach ridge which formed above a peak in the tufaceous bedrock. The reduced water depth and the closing connection to the sea is evidenced by the microfaunal association which changed from marine to brackish.19 A third shipwreck was found in the sediments dated between the end of the second and the beginning of the third century. Since the definitive emergence of the area at the end of the fifth century, the lagoon deposits have been covered by continental sediments as shown in Fig. 3D. In the section of Fig. 3D’ the subsequent anthropogenic filling and the modern ground level are also shown.

4.3.1. Pollen Results from the Third to the Fifth Century ad Pollen data from the port sediments show a drastic decrease in cabbage cultivation during the third century ad (Fig. 4). This decline was accompanied by an increase in Mediterranean maquis and in some elements of deciduous forest. Such change can be interpreted as a phase of abandonment of vegetable gardens in this area of the town and the contemporaneous spread of wild vegetation. This could mean a decreased human presence and consequent decrease in horticultural activities which favored the increase in tree crops, requiring less care than vegetables. Indeed, the main peak of chestnut is recorded in this period. After this moment the situation appears to have been restored and in the fourth and fifth centuries the vegetation is very similar to that preceding the third century, apart from a generally higher

19 Bourillon 2005, 38. human-environment interactions: neapolis and elea-velia 225 presence of oaks and a continuous presence of walnut and grapevine. The olive peak, restricted to the third century, suggests that it was probably the wild variety. It is interesting to observe that at the same time, namely during the third century, port activity also had a period of stagnation.20

5. Evolution of the Velia Area

The environmental evolution of the Velia region is shown in Fig. 5 through three-dimensional schemes obtained from a digital elevation model (DEM) and from cross sections.

5.1. Step 1—First Millennium bc As in the Neapolis region, the coastline of the Elea- Velia region was located about 800 meters further inland with respect to the modern one (Fig. 5A). This paleo-coastline position, prior to the development of the Greco-Roman town, was reconstructed thanks to borehole data which highlighted the occurrence of marine sediments at ca. 0m asl21 at the cliff base, covering the fluvial and slope deposits forming the bedrock and the cliffs. The coastal profile of the area consisted of promontories and inlets with rocky cliffs made up of cemented fluvial conglomerates. The section of Fig. 5A’,crossing the slope and the lower quarters of the future Greco-Roman town, is based on data coming from some boreholes stratigraphies and archaeological excavations.

5.2. Step 2—Greco-Roman Period In the second step of environmental evolution in the Elea Velia area (Fig. 5B) the main difference with respect to the previous step is the progradation of the coastline and the formation of sandy beaches at the cliff toes. Elea-Velia rose in the sixth century bc on a hill surrounded to the westward by the sea. In the fifth century bc it developed downhill, as far as the narrow emerged littoral plain, with the building of the Southern Quarter. Some remains of the main structures are still visible, such as the Acropolis on the top hill, the Temple of Asklepius and the Roman baths, both located in the thalweg of the Frittolo stream (cfr. section 3.2), and the Southern

20 Giampaola et al. 2006, 60–62. 21 Ortolani et al. 1991, 167. 226 elda russo ermolli, paola romano, maria rosaria ruello human-environment interactions: neapolis and elea-velia 227

Fig. 5. Three-step scheme of the evolution of the Velia area. A= Step 1—I millennium bc; A’= cross section related to Step 1; B= Step 2—Greco-Roman period; B’= cross section related to Step 2; C= Step 3—since the Late Ancient; C’= cross section related to Step 3 (after Romano et al., in prep). 228 elda russo ermolli, paola romano, maria rosaria ruello

Fig. 6. The road to Porta V at Velia. The wall shows several restoration phases which were realized in order to compensate for the alluvial deposition at its back.

Quarter with the Necropolis in the plain. As in the Neapolis case, at Velia too the town walls followed the natural features and in particular the main divide of the stream basins. The life in the Southern Quarter of the town was strongly influence by the torrential regime of the Frittolo stream. In fact, phases of peak in discharge were followed by alluvial floodings of the town causing damaging to the urban structures. This influence is clearly shown in the section of Fig. 5B’ and in Fig. 6 where it is possible to see the buildings and walls subjected to several phases of restoration in order to compensate for the ground level aggradation. This aggradation is represented in section by the layers of alluvial sediments coming from the stream which alternate with the wall’s ruins. The coastline was probably located very close to the town as testified by the storm sediments covering some Roman tombs in the Necropolis,22 outside the town’s walls.

5.3. Step 3—Since the Late Ancient Starting from the third century ad several new phases of intensive over- flooding, mainly supplied by the Frittolo stream, completely covered the Southern Quarter (Fig. 5C and 5C’). The cross section of Fig. 5D’ shows the

22 Ruello 2008, 262. human-environment interactions: neapolis and elea-velia 229 present situation where thick alluvial levels were removed by the archaeological excavation carried out in the twentieth century in order to unearth the town ruins. The alternation of alluvial phases and restoration phases is clearly shown in several sectors of the town where the walls of buildings were rebuilt several times above the strata of alluvial deposits fed by the stream. In particular, the main rebuilding phases can be dated to the end of the first century bc, the second century ad, the beginning of the third century ad and the end of the third century ad. The sedimentary structures characterizing the alluvial deposits can be alternately referred to debris flow and flood processes. The thickness of each alluvial episode (i.e. the thickness of the alluvial strata) clearly rises upward, testifying to the increase in both sedimentary supply and stream power, particularly during the third century ad (Fig. 7). These traits are typical of flash flooding, that is rapid response events to intense precipitation episodes. The fact that the flooding from the beginning of the third century ad onwards was more severe is also clearly visible close to the apical zone of the alluvial deposition, where remnants of the road leading to Porta V (Fig. 6) are found. The wall beside the road displays levels of elevations dated to the second and third centuries ad, buried by alluvial strata increasing in thickness during the third century ad (1.7 meter/100 years). This exceptional increase in sedimentary inputs starting from the third century is most likely linked to declining land use management on the slopes behind the town, especially if we consider the size of the Frittolo basin, the major stream flowing toward the town, which is very small if compared to the huge amount of sediments produced.

6. Conclusions

The evidence collected up to now in the two study sites leads to certain conclusions. Throughout the time period analyzed, namely from the first millennium bc, both sites recorded a significant coastline progradation, which can be quantified in more than 500 meters at Neapolis and more than 800 meters at Velia. In the third century ad, evidence of acceleration in both progradation and aggradation processes is recorded in the two study sites. At Neapolis, this acceleration caused the formation of a beach ridge which isolated the port transforming it into a lagoon. At the same time, pollen data show the abandonment of vegetable gardens, at least in the port area, and the spread of wild vegetation. 230 elda russo ermolli, paola romano, maria rosaria ruello

Fig. 7. Section of the alluvial deposits at Velia. The thickness of the alluvial strata clearly rises in the III century ad testifying to the increase in both sedimentary supply and stream power. human-environment interactions: neapolis and elea-velia 231

At Velia, the sediments of the third century ad are the evidence of flash floods, that is: rapid response events to intense precipitation episodes. The alluvial strata show an increased thickness with respect to those of the preceding centuries, testifying to the occurrence of more severe flooding. It is known from proxy-based reconstruction of the north-central Euro- pean climate that a higher variability characterized the period from the third to the seventh century ad, but in the study sites it seems clear that the presence/absence of man played a crucial role in influencing the slope dynamics. It is thus rather probable that the effects of particular land use conditions were enhanced by extreme climatic events creating a “window of opportunity”23 in which episodes of severe erosion occurred.

23 Bintliff 2002, 417–435.

LARGE-SCALE WATER MANAGEMENT PROJECTS IN ROMAN CENTRAL-SOUTHERN ITALY*

Duncan Keenan-Jones

Tot aquarum tam multis necessariis molibus pyramidas videlicet otiosas compares aut cetera inertia sed fama celebrata opera Graecorum.1 One may plainly compare such indispensible and enormous structures, carrying so much water, with the indolent pyramids and the other useless yet renowned and celebrated works of the Greeks.

quod si quis diligentius aestumaverit abundantiam aquarum in publico, bali- neis, piscinis, euripis, domibus, hortis, suburbanis villis, spatia aquae venien- tis, exstructos arcus, montes perfossos, convalles aequatas, fatebitur nil magis mirandum fuisse in toto orbe terrarum.2 If anyone were to consider more carefully the vast amount of water for the use of the public, in baths, in pools, in channels, in homes, in gardens and in sur- burban villas, the distances traversed by the water, the arcades constructed, the mountains pierced and the valleys levelled, he would say that there was nothing more amazing in the whole world. Extensive alteration of hydrologic systems was a conspicuous feature of ancient Roman civilization. Water was channelled and directed for a myriad of different purposes: drinking, aesthetic display, bathing, irrigation, industrial uses, land reclamation and waste disposal. Rows of arcades supporting water channels were dominant features on the Italian plains. Such arcades maintained the elevation of water brought from the Apennines so that it could reach the very highest points of urban centres. Building on Etruscan practices, marshes and lakes were drained partially or entirely in order to reclaim land for farming.3 This paper considers two case studies, both involving inter-basin transfers of water, in order to investigate elite Roman attitudes towards the use

* The research for this paper was generously supported by a Macquarie Gale British School at Rome Fellowship, a Macquarie Gale Travelling Fellowship, a Friends of Hercula- neum Society Studentship and two Borse di Studio from the Italian Government. 1 Frontinus, De Aquis 1.16. 2 Pliny the Elder, Naturalis Historia 36.24.123. 3 Purcell 1996, 190–199. 234 duncan keenan-jones of water resources, particularly their attitudes towards deleterious consequences of such use on other communities and the wider environment. Before turning to the case studies, we must first consider how the hydrological cycle has changed in Central-Southern Italy since antiquity so that we can make use of modern data, and also consider the usage and legal control of water in rural areas.

Hydrological and Climate Change in Central-Southern Italy

Variables of particular interest to the hydrologic cycle are temperature, which affects evaporation and transpiration (evaporative loss of water from plants), and rainfall. These two variables exert a powerful influence on river flow (including flooding) and groundwater recharge, which, in turn, controls spring flow rates.4 The basic picture of the Mediterranean climate, that of a marked seasonality with dry summers and abundant rainfall only in winter, seems to have been begun between 550 and 50bc and has remained until the present day.5 This seasonal variation, and the resulting reliance on spring water during summer, is confirmed for the Apennine springs of Campania in the early fifth century ad by Paulinus of Nola.6 The numerical values of rainfall and temperature in antiquity, however, would not necessarily have been comparable to those of today. Even over the short period of time since the beginning of record keeping, annual rainfall in Central-Southern Italy has decreased by 18%.7 The rainfall pattern has also changed, resulting in fewer, more intense rain events, which lead to more run-off, decreased infiltration and hence decreased groundwater recharge. In general, in Central-Southern Italy over the last 3000 years, the climate has alternated between warm, dry periods and cool, wet periods.8 The warm, dry periods, through increased evapo-transpiration and decreased rainfall, would have seen reduced river and spring flow rates. The opposite would have been true for the cool, wet periods, which, through increased rainfall, saw increased flooding and sediment transport in rivers. Giraudi and co-workers have recently provided the first comprehensive reconstruc-

4 Chen, Grasby, et al. 2004; Fiorillo et al. 2007; Dragoni and Sukhija 2008, 4. 5 Peyron et al. 2011. 6 Carmen 21.752–753. 7 Fiorillo and Esposito 2006, 1. 8 Dragoni 1998, 261. large-scale water management projects 235 tion of climate in Central-Southern Italy over the past 10,000 years.9 They used Apennine glacial advances and retreats (which basically correlate during our period with those of other European mountain chains and with evidence for large amounts of ice in the North Atlantic) and related processes as proxies for temperature. Lake levels and δ18O (stable oxygen isotope composition) of carbonate deposits from caves (including stalagmites) and lakes (shells) were used to estimate the water balance (precipitation versus evapo-transpiration). These data are believed to represent climatic, rather than human-induced, changes, although this is less true with regard to lake levels (see below). The period encompassed by our case studies, that is to say between the late first century bc and the fifth century ad, lay between cool and wet periods dated to 850–650bc and ad550–750. Apart from some evidence for a minor increase in rainfall in the late 2nd/early 3rd century ad discussed below, Central-Southern Italy seems to have been undergoing a warm, dry climate at this time. The most recent cool, wet period was between ad1450–1650, during the period known as the Little Ice Age. Hence, the late 19th and 20th century were also a comparatively warm and dry period, happily rendering the period of instrumental records of climate a reasonable analogue for our period of interest.10 Temperatures and probably also rainfall, and hence spring discharge, were seemingly comparable to the only data we have available (but see the discussion of flood history below).

Flood History The situation regarding the flood history of Central-Southern Italy is more complex. The flood history of the Tiber River (and of the Po) and the levels of Lakes Fucino (east of Rome) and Accesa (in Tuscany) (Fig. 1) are well-correlated with Alpine glacial retreat and advance, a proxy for regional temperature change.11 The exception is the warm period between the 3rd century bc and the 3rd century ad, a time of increased flooding in the Tiber despite warmer temperatures. Giraudi claims that this period was one of increased rainfall, citing increases in the levels of Lake Fucino and Lake Ledro in northern Italy, and coeval flooding in the Po Basin, while also allowing that human action in the Tiber Basin may have played a major role.12

9 Giraudi et al. 2011. 10 Lamb 1995, 157–159. 11 Giraudi 2011, 8–9. 12 Giraudi 2005; Giraudi 2011, citing Giraudi 1998 for the Lake Fucino data. 236 duncan keenan-jones

Fig. 1. The Italian Peninsula. Based on data provided by ESRI Inc.

The major relevant human activity was probably land clearance and consequent deforestation. The Greco-Roman world witnessed massive land- clearance and engaged in heavy and prolonged consumption of wood, especially perhaps in Italy.13 Pollen evidence attests to deforestation in most parts of Central or Southern Italy starting from around 1700bc or even earlier, with the process seemingly reaching remote, elevated sited sites such as the Lago Grande di Monticchio (Fig. 1) around the turn of the era.14 River basins, especially the Tiber, were particular foci for the wood industry, since the rivers provided the means to transport the wood to market. Reduction of forest and other vegetation cover increases the frequency and height of flooding by reducing the amount of water retained in plants and the soil and increasing

13 See most recently Kaplan et al. 2009, Harris 2011a, and Harris this volume. 14 Ramrath et al. 2000; Allen et al. 2002. large-scale water management projects 237 erosion.15 Instead of being retained and slowed in the vegetation, more water runs off into streams and rivers more quickly. Increased erosion arising from the lack of vegetation also fills buffers, such as lakes, with silt, reducing their capacity to slow and minimize flooding. Silt also fills the channels on the flood-plains, reducing their capacity and preventing the quick movement of water out of the basin, and raises the height of the flood-plains, again increasing flood heights. Of course, similar historical processes to those affecting the Tiber Basin were also affecting that of the Po at this time. Significant siltation and progradation at the mouth of the Po testifies to the large sediment loads carried.16 During the same period, multiple rivers in Basilicata also show evidence of similar siltation processes that have been attributed to human activity.17 Slightly later, in the 3rd century ad, we see the beginning of an acceleration of siltation in rivers at Naples and Velia that has been attributed to a combination of human activity and climate events.18 Hence, there is clear evidence for anthropogenic factors that could explain the increased flooding of the Tiber and the Po, and those of other Mediterranean rivers such as the Arno (Fig. 1) and the Rhone,19 and we need not posit increased rainfall on the basis of this evidence. It is interesting that sediment accumulation in the Ombrone river delta (not far from Rome, Fig. 1) is well-correlated with those of the Tiber and Po (and with known climatic cool and wet periods) for the Late Holocene, except for the Roman period, where there is no large sediment accumulation.20 If this is not due to marine erosion, could it be that there was little anthropogenic influence on this particular catchment? The other evidence for increased rainfall during this period is also inconclusive. Lake levels are affected by changes in precipitation and evapotranspiration, but also by local non-climatic factors, such as changes in land use.21 As a result, only multiple synchronous movements in lake levels within a region should be taken as evidence of climatic change. In addition, use of lake levels as a proxy in Italy in the Holocene is hampered by a lack of comprehensive and continuous data at a sufficient number of different sites.

15 Hughes 1994, 82–84; Yin and Li 2001; Beven 2003, 529, 538; Walling 2006; Roberts, Brayshaw, et al. 2011. 16 Hughes 1994, 84. 17 Piccarreta et al. 2011. 18 See the contribution by Russo Ermolli and co-workers to this volume. 19 Magny et al. 2009. 20 Magny et al. 2007, citing Bellotti et al. 2004. 21 Magny 2004; Giraudi et al. 2011. 238 duncan keenan-jones

There are question marks over both lakes cited by Giraudi as evidence. Regarding Lake Fucino, it is strange that the contemporary publication by Giraudi and co-workers shows no high stands of Lake Fucino within approximately 500 years of the turn of the era.22 A possible explanation is that the increased level shown by Giraudi in his 1998 publication was too small and/or short-lived to be noted in the longer-term and, hence, lower- resolution 2011 study by Giraudi and co-workers. This could be because the increase in level at Fucino cited in Giraudi’s 1998 article was abruptly cut off by draining, that is to say by anthropogenic activity. The level of Lake Fucino could also reflect tectonic activity, adding another potentially confounding factor.23 Regarding Lake Ledro, there is clear evidence for high lake levels around the turn of the era.24 How relevant is this site on the southern slopes of the Alps (Fig. 1) to the climates of Lazio and Campania? If in fact we bring in a wider survey of lake levels around West-Central Europe, we see that our period of interest is bracketed by major lake high stands at 800–400bc and ad750 with a minor high stand at ad150–250.25 The third relevant lake is Lake Accesa. Its level steadily declined from c. 750bc, reaching a low around ad200 before a short-lived increase.26 The next major high stand was c. ad750. Unfortunately, the data available for Lake Mezzano are discontinuous at this period, partly due to the human action of draining the lake.27 There is also some discontinuous archaeological evidence that the level of Lake Martignano (near Rome, Fig. 1) was higher than present in 2bc and fell during the first two centuries ad to a much lower level.28 With one lake (Fucino) showing level increases, one showing low levels (Accesa), and some evidence that another (Martignano) dropped in level, it seems difficult at this stage to maintain that there are multiple synchronous movements of lake levels in Central-Southern Italy. The fact that larger closed lakes, such as Lake Fucino, are more likely to reflect climate change than smaller ones, such as Lake Accesa,29 is irrelevant, since multiple synchronous datasets are required. Given that

22 Giraudi et al. 2011, 109–110, Figure 3. The results for Accesa are given in Magny et al. 2007. 23 Giraudi et al. 2011. 24 Magny et al. 2009. 25 Magny 2004. 26 Magny et al. 2007. 27 Giraudi 2004. 28 Dragoni 1998, 245; Giraudi 1998, 11. 29 Giraudi et al. 2011. large-scale water management projects 239

1. the late Republican/Imperial Roman period seems to be the only exception in the late Holocene to an otherwise good correlation between low temperatures and higher rainfall, and that 2. the only evidence for this exception is ambiguous and derived from lake levels (the proxy least reliable, and most susceptible to local factors such as anthropogenic influence, among those considered by Giraudi and co-workers), the explanation of predominantly anthropogenic activity (which is well- attested for this period), rather than increased rainfall, for the increased flooding in the Tiber and Po valleys seems to resolve the problem nicely. This also puts the Central-Southern Italian climate in sync with that of West- Central Europe, which, with the exception of Lake Ledro, shows low lake levels at this time. It therefore seems prudent, until further data from more lakes results in a more univocal testimony of climatic effect on lake levels in Central-Southern Italy, to continue to regard the Roman period as another warm and somewhat dry period within the late Holocene history of Central- Southern Italy. I will conclude this section by noting that there are numerous problems in using written sources to reconstruct comparative flood histories. One is the uneven survival of written sources from different time periods. It has been suggested the 10th and 11th century ad Italian records are defective.30 Much depends on the flood’s impact on the literate elite, which may not correlate with actual flood magnitudes.31 Large floods whose impacts are felt largely in rural areas with low population density are less likely to be recorded. Nevertheless, for Rome from the late Republican to the middle of the imperial period we have some of the best written records of flood history from pre-modern Europe.

Rural Water Use and Access in Central-Southern Italy

A major use of water here would have been for irrigation.32 Frontinus describes the careful, time-based allotment of the water of the Aqua Crabra for irrigation of surrounding villae, including that of Cicero in earlier times, that was disrupted by the addition of this stream to the Aqua Iulia.33

30 Camuffo and Enzi 1996, 440. 31 Dragoni 1998, 243. 32 Thomas and Wilson 1994, 157–158. 33 Frontinus. Aq. 1.9, Cic. Leg. agr. 3.9. 240 duncan keenan-jones

Similar irrigation systems, also from central Italy, were depicted in CIL 6.1261 and CIL 14.3676.34 It is not clear whether they were supplied by rivers (as is more likely) or by aqueducts. The regulation of the use of water in this way shows that there was a high demand for water for irrigation and that there were concerns about scarcity.35 An example of proposed aqueduct irrigation is preserved in Cicero’s Ad Quintum Fratrem 3.1.3. Cicero urges his brother to expend 45,000 sestertii on the construction of an aqueduct in order to irrigate 50 iugera of his land, probably near Arpinum (Fig. 1). Another use of running water in rural Central-Southern Italy was for fishing. Fresh water fishing seems to have been an important and widespread practice in Roman Italy, although there is little trace of its presence, except in Roman art.36 Roman law supported public access to most large lakes and perennial rivers (flumina), which seem to have been treated as a commons.37 This included access for fishing and hauling or drawing off water.

Case Study 1: The Aqua Augusta The largest and most complex of the aqueducts carrying water from the Apennines to urban centres was the Aqua Augusta.38 It was Roman water management on a larger scale than ever before: 1. Rather than being a focused supply for a single urban centre, the Augusta supplied 8 or 9 different towns plus numerous villas through a network of branches (Fig. 2). 2. This network had a total length of between 139 and 145km, making the Augusta the longest single Roman aqueduct at the time of its construction in the late Republican or Augustan period, not to be significantly surpassed until the fifth-century aqueduct of Constantinople. Simultaneous supply (as it seems) of so many different locations suggests that the Augusta carried a large amount of water. Some of the towns supplied were significant in size: Puteoli may have been the third largest town in Italy in the early Imperial period and the third largest amphitheatre in Italy was constructed there under the Flavians.39 In addition to the resident

34 Cf. CIL 8.4440 and Pomponius in Dig. 43.20.2. 35 Bannon 2009, 48, citing Bruun 2000, 580–581. 36 Toynbee 1973, 209–215; Squatriti 2002, 98–101. One reference is Varro, Rerum Rusticarum 3.3.5. 37 Taylor 2000, 55–65; Bannon 2009, 221. 38 For a more detailed introduction to this aqueduct see Keenan-Jones 2010a, 2010b. 39 D’Arms 1974, 119 n. 122; Ostrow 1985, 73. large-scale water management projects 241

Fig. 2 The Bay of Naples in the Roman Period showing the Aqua Augusta. Based on data provided by ESRI Inc. population, as a major port it had many visitors as well as many vessels requiring water for long voyages. Several sets of public baths, at least 8 public fountains and numerous large cisterns have so far been uncovered in the city.40 Puteoli’s demand for water would have been large, and it is not clear whether Puteoli’s other aqueduct, the Campanian aqueduct, was functioning during the life of the Augusta.41 The other towns supplied were also of considerable size. Estimates of Naples’ population during the Roman period range up to 35,000.42 On the poorly-watered western end of the bay were Baiae and Misenum. Baiae, according to Strabo (5.4.7), was as large as Puteoli. Even allowing for exag- geration, the closely-spaced yet luxurious villas of the Roman elite for which Baiae was famous would have had a large (and conspicuous) consumption

40 Known public baths include Via Ragnisco (Republican, Gialanella 1993a, 88) and the ‘Tempio di Nettuno’ (early 2nd century ad, Sommella 1978, 32). Fountains: Sommella 1978, 26 (#12) & #39; Gialanella 1993b, 89; De Caro 1999, 651; De Caro and Gialanella 2002, 19; De Caro 2003, 593–594. Cisterns of note range from those on the scale of 2 or 3 interconnected vaulted rooms (Sommella 1978, #1, 11, 14 and 36) to the cavernous Piscina di Lusciano (Sommella 1978, 48), Cento Camerelle (De Caro 2003, 593–594) and Piscina Cardito (Sommella 1978, 61). 41 Maiuri (1958, 58) and Döring (2007, 110) suggested it pre-dated the Aqua Augusta. Opus reticulatum found in an access stair (Dubois 1907, 271) suggests the Campanian aqueduct was operational at some point during the first centuries bc and ad. 42 Arthur 2002, 22. 242 duncan keenan-jones of water. Misenum, the home of the western Imperial fleet, had an estimated population of 10,000 sailors and up to 4000 civilians of citizen rank during the early Empire, who were catered to by the enormous Piscina Mirabilis cistern.43 The demand for the water of the Augusta would have been great and it seems very unlikely that all these towns and villas could have been supplied to everyone’s satisfaction.44 This would have been particularly the case in summer, as we shall see below. One surviving piece of written evidence suggests greater demand for the Augusta’s water than could be supplied. A constitution promulgated in ad399 following a repair of the Augusta stipulated heavy penalties for use of its water by any private individual.45 The Aqua Augusta represents a conscious decision, almost certainly on the part of Augustus himself, to transfer a significant quantity of water out of an Apennine river basin to supply urban centres on the plains around the Bay of Naples. How large was this transfer and what were the effects of the removal of this water at the source? Quantifying the flow rate (or flux) of the Augusta is difficult. Like all ancient aqueducts, modelling or even estimation of ancient flows is frustrated by the partial survival and limited accessibility of the channel. In the case of the Augusta, it is complicated even further by bradyseism, the slow fluctuations in ground level due to the movement of water and magma beneath volcanoes, such as Vesuvius and the Campi Flegrei (Fig. 3). This renders the all-important gradient of the aqueduct very difficult to reconstruct, at least where the aqueduct crosses the coastal plains. One starting point in attempting to estimate the flow-rate is the flow available to the Augusta at its source, i.e. the maximum amount of water it could have carried. The Augusta was supplied by two main, perennial springs, Acquaro and Pelosi, and a minor spring, Acquarolo, all located near Santa Lucia di Serino. These springs were, and still are, fed by the Terminio aquifer, which is composed of calcareous (calcium carbonate-rich) and calcareous-dolomite (calcium magnesium carbonate-rich) strata.46 While travelling from the aquifer to the surface, the water passes through overlying Quaternary (i.e. more recent) deposits of alluvium and volcanic ash.

43 Starr 1941, 16; Keppie 1984, 103; D’Arms 2000, 134. 44 Keenan-Jones 2010a, 4, and 6 Fig. 2. This is implicit in the inscription commemorating the repairs carried out under Constantine, AÉ 1939 no. 151. 45 Codex Theodosianus 15.2.8. 46 Fiorillo and Esposito 2006, 2. large-scale water management projects 243

In order to reconstruct the operation of the Acquaro and Pelosi springs during the time the Augusta functioned, we need to consider the recent history of the springs. This has been laid out in detail by a team from the Università degli Studi del Sannio in two recent publications,47 upon which this section is largely based. The discharge data for Acquaro and Pelosi, and for another nearby spring supplied by the Terminio aquifer, Urciuoli, are available from 1887 to the present day. Rainfall data from the area are available from 1918 to the present day. Analysis of these data has revealed that over that period the mean flow of Acquaro was 300–800 litres per second (L/s) and that of Pelosi was 100–600 L/s, giving a combined mean flow rate of 400–1400 L/s (34,500m3/day—121000m3/day). The range is seasonal, as the spring discharge is related to the amount of water recharging the aquifer. The recharge is a function of the effective rainfall, or the amount of rainfall over the aquifer’s catchment area less any evapotranspiration. As discussed above, this part of Italy has had a standard Mediterranean climate over this period, where the seasonal cycles of rainfall and evapotranspiration (related to temperature and humidity) are almost exactly opposite. Rainfall is lowest and evapotranspiration highest in July and there is zero effective rainfall from June to August. After a time delay related to water per- colation from the surface into and through the aquifer and back to the surface, this leads to minimum spring discharge in November. From September, rainfall increases, peaking in November, and evapotranspiration decreases, reaching a minimum in December/January. First, the Quaternary deposits covering the aquifer are refilled, and once they have reached capacity, the aquifer is recharged. This and the effects of snowmelt lead to maximum spring discharge in April and May. The delay in aquifer recharge means that individual rainstorms do not affect the flow: discharge varies as a smooth curve throughout the year. The delay is a very useful feature for water supply, as minimum spring discharge does not coincide with minimum rainfall, when the spring would be most needed. It requires an effective rainfall of c. 250mm to fill the overlying Qua- ternary deposits and begin to recharge the aquifer. If recharge is less than spring discharge during the wet season, then the spring discharge continues to decrease where it would normally increase. This has only happened on two occasions since records began, 1948–1949 and 2001–2002. The discharge profiles for the Serino springs for these two years differ markedly from

47 Fiorillo and Esposito 2006; Fiorillo et al. 2007. 244 duncan keenan-jones the norm. In 2001–2002, the drought was exacerbated by large-scale groundwater extraction and dramatic temperature increases. Thus 1948–1949 is a better indicator of a typical Serino spring drought. The combined monthly discharge for this year decreased from around 1500 L/s before flatlining at a minimum figure of around 1000 L/s. This was the minimum even in 2001–2002 and thus probably represents the maximum effect of a short-term fluctuation such as a one- or two-year drought. Unfortunately, this figure does not show the relative contributions of Acquaro/Pelosi and Urciuoli for these two years. Given that Acquaro and Pelosi are far more sensitive to fluctuations in rainfall than Urciuoli, they were probably more severely affected during times of drought. The flow rate of Urciuoli at the beginning of the 1948–1949 drought would have been around 1200 L/s. The Acquaro and Pelosi springs (and the artesian wells near the springs) dried up completely during the 2001–2002 drought48 and hence it can be assumed that the minimum flow rate of Urciuoli is 1000 L/s even in the worst droughts. It seems that Acquaro and Pelosi were completely dry for 3–5 months during the 1948–1949 drought.49 The return time of these Serino Spring droughts has been calculated to be 70 years based on the data for 1887–2007. This return time seems to have begun to decrease (i.e. droughts have started to become more frequent) during the late 20th and early 21st centuries due to the large-scale groundwater extraction and dramatic temperature increases mentioned above. Now that some quantification of the water transferred by the Augusta has been carried out, the effects of this transfer will be considered. None of the towns supplied by the Augusta were in the Sabato (ancient Saba- tus) basin, which contained the Acquaro and Pelosi springs, and there is no archaeological evidence of supply to locations within that basin. The Augusta functioned as a continuous transfer of water from the Sabato basin to the Clanius basin (containing Nola, Acerrae and Atella) and the basins draining into (or, in the case of Cumae, very near to) the Bay of Naples (containing the other towns supplied) via the 6km long Forino tunnel. The long distance transport of water from a rural source to an urban centre obviously resulted in less water available at the source, with concomitant social and environmental impacts. The Acquaro, Pelosi and Urciuoli springs at Serino are far and away the largest springs along the course of the Sabato. The other springs, those

48 Personal communication from Azienda Risorse Idriche Napoli staff. 49 Fiorillo et al. 2007, Fig. 5. large-scale water management projects 245 at Sorbo Serpico and the Tornola (200–400 L/s and 3–5 L/s respectively) have only a fraction of the flow of those at Serino.50 The flow of the Serino springs is currently diverted to Naples for drinking water supply. If the springs were allowed to flow into the Sabato River, however, the Acquaro and Pelosi would account for 19% and Urciuoli 30% of the flow of the Sabato at Ceppaloni (Fig. 3), more than 20km downstream.51 Thus, they would together account for almost half the median flow of the river at this point and a greater share further upstream. The water from the springs would be especially important to river health due to the time lag between the springs’ flow cycle and that of rainfall, described above. Thus, the springs continue to provide water even during the dry summer period when rainfall is almost non-existent. The present diversion of the Acquaro and Pelosi often causes the river to be dry downstream of these springs during the summer. As discussed above, the modern data are a reasonable approximation for the Serino springs during the period of the Augusta’s operation. This requires the assumption that the vegetation cover above the aquifer was not drastically different in these two periods, which seems reasonable. Precipi- tation and hence river flow when the Augusta was constructed seem to have been similar to that of the 20th Century. Both the sets of springs at Serino were tapped, and possibly dammed, in the Roman period to supply urban centres. It is not clear how much of the available water from these springs was diverted to the aqueducts. Certainly, the diversion was not as comprehensive as the modern exploitation of the springs, which also draws water directly from the aquifer using artesian wells near the Acquaro and Pelosi springs. Remains of a sluice gate visible at the entrance to the aqueduct in the 16th century suggest that the proportion of water diverted to the Augusta in Roman times was variable.52 The summer would have been the period of highest demand for the water of the Augusta, however, so we should not expect that extra water was released to the river at this time. Evidence survives for the use of the water of the Sabato for irrigation. Upstream of Abellinum, signs of centuriation have been identified in the fertile valley and adjacent hills.53 The effects of the removal of a significant proportion of the flow of the river on irrigated agriculture in this area had the potential to be dramatic, particularly during summer.

50 Regione Campania A.G.C. Sviluppo 2011, 59. 51 Calculated from the figures provided in Provincia di Avellino 2009, 39–40, 48. 52 Lettieri 1560, 399. 53 Spadea 1998, 35. 246 duncan keenan-jones

The environmental effects on the river could also have been pronounced. Rapid reductions in the flow rate of a river can severely disrupt ecosystems that have developed on a longer timescale. This is particularly the case where certain reaches of the river become disconnected, even if only tem- porarily. Research on modern rivers around the world has shown that the fish numbers often drop and some species become extinct in rivers under these conditions.54 Since the Sabato River was probably a commons in the Roman period, as discussed above, damage to fish stocks in the Sabato River would have removed a source of income and/or dietary protein from those who could ill afford to lose it. In this way, the use of the water of the Acquaro and Pelosi and Urciuoli springs for urban populations was comparable to an enclosure of an ager publicus. Both these impacts, on water available for irrigation and on the river itself, would have been most severe closest to the springs at Serino. Further downstream, as the catchment area of the river increased, the impact of the removal of Serino spring water would have been lessened.

Case Study 2: Tiber Flood Control In the previous case study, the voices of upstream users of water from Central-Southern Italian rivers have remained largely mute. One case where we should be able to recover the voice of these upstream users is in the proposal to prevent the flooding of the Tiber narrated by Tacitus in his Annales 1.76 and 79. There is good reason to believe that Tacitus was drawing his account directly or at least indirectly from the acta senatus, the written records of senate business55 and thus that he was using records of the actual speeches of the delegations from these communities. Also that year (ad15), the rising of the Tiber, swollen by constant rain, inundated the level parts of the city. Large-scale loss of life and building damage followed the receding of its waters. Asinius Gallus therefore suggested that the Sibylline books be consulted. Tiberius indicated his denial with a nod of his head, drawing a veil over both human and divine affairs at the same time. However, the task of keeping the river within its banks was committed to Ateius Capito and L. Arruntius. (Annals 1.76) This flood was towards the end of one of the worst, sustained flooding episodes in the known history of the Tiber. The 1st centuries bc and ad were

54 Poff et al. 1997; Bunn and Arthington 2002; Richter et al. 2003. 55 Syme 1981, 366; Swan 1987. large-scale water management projects 247 second only to the Little Ice Age in terms of flood events,56 with the possible exception of a spike at the beginning of the 2nd century bc.57 The turn of the era even exceeded the Little Ice Age in the frequency of exceptionally large flood events.58 The ad15 flood was the 8th recorded in 45 years.59 There were major incentives for Tiberius and the Senate to seek to remedy the situation. Tacitus highlights the loss of life and damage to buildings. Important areas such as the Forum Romanum and the Forum Boarium would always have been prone to flooding. Moreover, the pressure for monumental space and housing in the growing city had led to an increasing number of permanent public structures in the Campus Martius, a marshy area that was one of the first to be flooded when the Tiber rose. Public building had begun in the southern part of the Campus Martius in the 3rd century bc with a series of temples associated in some way with, tellingly, water.60 More temples, a circus and porticoes followed in the late 3rd and 2nd centuries bc. The late Republic saw the grandiose building program of Pompey,including the Theatre of Pompey, and some activity by Caesar,61 but it was in the Augustan period that building in the Campus really accelerated. The list of Augustan buildings shows the importance of the area in the Augustan monumental program: the Baths of Agrippa, Ustrinum Agrippae, Saepta, Pan- theon, Mausoleum of Augustus, Porticus Argonautorum, temple of Bonus Eventus, Laconicum Sudatorium, Porticus Octavia and Octaviae, theatres of Marcellus and Balbus, Stagnum Agrippae, Ara Pacis and the horologium.62 In addition, the flooding of the Tiber also caused Rome’s sewers to overflow, inundating low-lying areas such as the Forum and the Campus Martius with sewage.63 Tacitus may have been alluding to the widespread health problems this would have caused when he stated that the loss of life followed the receding of the waters. Early attempts to mitigate the flooding of the Tiber included raising the height of low areas, such as the Forum, by the addition of large quantities of earth.64 As the monumental character of the Campus Martius grew (and

56 Bersani and Bencivenga 2001, as cited in Giraudi 2011, Fig. 7. 57 Camuffo and Enzi 1995, 1996. 58 Calenda et al. 2009, Fig. 2. 59 Aldrete 2007, 242–243. 60 Torelli 2006, 92, 96–99. 61 Coarelli 1997, 532–590. 62 Shipley 1931; Coarelli 1997, Fig. 140. 63 Hughes 1994, 163. 64 Aldrete 2007, 177–178. 248 duncan keenan-jones also the population of the city), so did the scale of the solutions to remedy the problem. Julius Caesar had apparently canvassed a plan to divert the Tiber alongside the Vatican in order to enable further construction on the Campus Martius.65 Probably in an attempt to do something about the flooding problem that was adversely affecting so many of his new constructions, Augustus further raised the level of the Forum, widened and deepened the Tiber and, according to Suetonius, seems to have appointed curatores alvei et riparum Tiberis (commissioners of the bed and banks of the Tiber).66 Dio, in discussing Tiberius’ response to this same flood, mentions his institution of something very similar: a permanent five-man committee chosen by lot and tasked with ensuring the even, constant flow of the Tiber.67 Of the two men entrusted with giving advice after the flood of 15ad, one at least is known to have possessed some expertise. C. Ateius Capito (a former consul and a legal expert) had been curator aquarum (water supply commissioner) for two years by ad15,68 so he was probably the senator with the most water-related legal and technical expertise. L. Arruntius was also a former consul but is not known to have possessed any relevant expertise. We next hear of them from Tacitus a few chapters later: Next an item was raised in the Senate by Arruntius and Ateius regarding whether, in order to moderate the floods of the Tiber, the rivers and lakes through which it was swelled should be diverted. Embassies from towns and colonies were heard, with the Florentines pleading that the Clanis not be removed from its normal bed and transferred into the Arno river, so that this would not bring disaster upon themselves. There was similarity in the matters that the Interamnates discussed, stating that the most fertile fields of Italy would be ruined if the river Nar, having been led away through channels (for this was being prepared), spread over them and formed a lake. Nor were the Reatines silent, rejecting the damming of Lake Velinus (which there pours itself out into the Nar) after which it would necessarily overflow into the surrounding areas. Nature, they argued, which gave to rivers their sources and their courses, had considered human affairs most carefully: as it gave a beginning, so it gave limits. They further argued that the cults of the allies, which have consecrated sanctuaries, sacred groves and altars to the streams of their native lands, should be respected, and that the Tiber itself did not want to flow onwards with reduced glory, deprived of its tributaries. Either the

65 Cicero, Ad Atticum 13.33.4. 66 Divine Augustus 37. Cf. Aldrete 2007, 177–178, 198–199. 67 Dio 57.14.8. 68 Frontinus, De Aquis 102.2. One wonders whether he was related to the Ateius Capito who told Cicero about Caesar’s plan to divert the Tiber (above, n. 66). large-scale water management projects 249

Fig. 3. The area of the proposed Tiber flood prevention works. Based on data provided by ESRI Inc.

pleas of the colonies, or the difficulty of the works, or superstition, prevailed such that at the motion of Piso, who had expressed the opinion that nothing should be changed, the plan was abandoned. (Annals 1.79) We are unfortunately given few details about the planned scheme. It involved the diversion of at least two flumina (also described as amnes) and one lacus (but possibly more) away from the Tiber to lessen its maximal flow. Writing closer to the event than Tacitus, Pliny regarded the Clanis as one of the two major tributaries in the upper reaches of the river.69 Figure 3 shows the Clanis after its comparatively recent diversion into the Arno. Pliny suggests that in his day the Clanis drained land as far north as Arretium into the Tiber. The Nar was reckoned by Pliny as the second largest of the Tiber tributaries below the Clanis. Hence, these were two of the major tributaries to

69 Naturalis Historia 3.5.53–54. 250 duncan keenan-jones the river, although neither was navigable (of all the Tiber tributaries, only the Anio claimed this distinction). The damming of the Velinus had ironically been advocated by the Interamnates in 54bc (presumably to lessen flooding around Interamna), at which time it had also been strenuously opposed by the Reatines.70 ad15 saw the two communities making common cause. The effectiveness of these measures, had they been carried out, is hard to judge. The diversion of the Clanis, eventually completed c. 1700, was followed by a drop in the frequency and severity of Tiber flood events.71 The major cause of this drop, however, was the end of the Little Ice Age, and numerous damaging floods still occurred after that time. As we saw above, flumina (larger, perennial rivers) were regarded legally as public, with public rights of access. Unlike in the first case study, access to public water does not form the basis of the complaints brought by communities upstream, however. The communities here faced the opposite problem: a surfeit rather than a shortage of water. Given the seriousness of the underlying problem, the communities needed to present the Senate with strong arguments in favour of the status quo. Central to the Roman elite’s perception of their relationship to the natural environment was an idea of continual improvement to make land more productive and profitable.72 Stoic philosophy, an important influence on Roman aristocratic thinkers, placed an emphasis on such agricultural improvement, as did Cicero and Strabo, among others.73 The flooding of prime agricultural land was the antithesis of this process. In particular, the outflow of Lake Velinus to be dammed had actually been created by Manius Curius Dentatus centuries before in order to reclaim agricultural land.74 The Reatines appealed to more abstract concepts that were helpful to their cause: natura, religio and the gloria of the Tiber itself. Their argument was that natura knows best and should not be tampered with. This line of reasoning, clearly in tension with the belief in improvement described above, also stretches back a long way in complex and varying world of Greco- Roman thought.75 It can also be seen in the writings of the Roman elite, with Cicero stating a few generations earlier that ‘nature’s … administration has no component that could be criticised. The best possible outcome has been

70 Cicero, Ad Atticum 4.15.5, Pro Scauro 27. 71 Alexander 1984; Camuffo and Enzi 1996; Giraudi 2011. 72 Hughes 1994, 196; Purcell 1996. 73 Hughes 1994, 68; Pothecary 2002. 74 Cicero, Ad Atticum 4.15.5. 75 Hughes 1994, 69. large-scale water management projects 251 produced from the existing elements. If someone were to think it could be better (although no one ever will), and desired to correct something, they would either make it worse or seek to do the impossible’.76 Scale and the relevance to agriculture seem to have been important to ancient thinkers in resolving this conflicting tradition. Large-scale projects criticised by ancient Greek and Roman writers (including Pliny the Elder) as offending nature and related religious entities, such as bridges and canals that created islands out of isthmuses,77 had benefits for transportation but not agriculture. Nevertheless, because of such transportation benefits such projects were still championed by Roman rulers such as Julius Caesar and Nero.78 The project put forward by Capito and Arruntius was certainly on a large scale and would have negatively impacted both agriculture and transportation. The Tiber was vital in the trade of goods to Rome.79 With the flow in the Tiber reduced, the navigable distance and navigable season of the river would also have been reduced. There is evidence of concerns at this time that occasionally the Tiber might not be navigable in summer. As we saw above, according to Dio, Tiberius’ five-man committee was to ensure also that the Tiber might not ‘fail in summer’.80 No doubt the Reatines knew what they were doing when they appealed to religio. Sacred groves, in theory at least, were strongly protected.81 While Livy cites a widespread abandonment of the ominous conception of floods,82 it remained a persistent theme in history and other writing. Livy himself ascribes divine causation to 5 of the 8 floods he describes.83 Julius Obse- quens (unsurprisingly), Aurelius Victor and Claudian (once each) as well as the Historia Augusta (twice) also see divine forces at work in floods. Dio Cassius, in his brief account of the ad15 flood, notes that many at the time

76 De Natura Deorum 2.34.86–87: ‘cuius quidem administratio nihil habet in se quod reprehendi possit; ex his enim naturis quae erant quod effici optimum potuit effectum est …’. 77 Pliny, Naturalis Historia 4.4.10, Hughes 1994, 50–51, 69. 78 Julius Caesar, Caligula and Nero favoured a canal though the Isthmus of Corinth (Sue- tonius, Divine Julius 44, Gaius 21, Nero 19; Pliny, Naturalis Historia 4.4.10), while canals linking Rome to better harbours at Terracina (Suetonius, Divine Julius 58.4) and Puteoli (Tacitus, Annals 15.42, Suetonius, Nero 31, Pliny, Naturalis Historia 14.8.61) were put forward by Cae- sar and Nero respectively. 79 Cicero, De Republica 2.10; Pliny, Naturalis Historia, 3.54, Brunt 1980, 92, esp. n. 60. 80 Dio 57.14.8. Compare Cicero’s description of the Tiber as a perennis amnis (De Republica 2.10). 81 Lucan, Pharsalia, 3.433–437; Hughes 1994, 169–177. 82 43.13. 83 Aldrete 2007, 221–223, 296 n. 18, which is also the source for the following authors. 252 duncan keenan-jones

(like Asinius Gallus) considered it a portent,84 as Dio himself does for 10 of the 12 floods he records. According to Pliny, Tiber floods made people more observant of religio.85 One first century ad Roman was sufficiently concerned about divine intervention while constructing an aqueduct tunnel through a mountain important to the Bona Dea that he repaired her temple after its completion, probably making good on an earlier vow.86 The political impacts of Capito’s and Arruntius’ project were also probably slightly against it. Clearly, there was political mileage to be made as the saviour of Rome from floods. Since the houses of the wealthy and powerful were concentrated on the hills out of reach of the floodwaters,87 the constituency was largely the masses. Augustus and Tiberius (and before them, Caesar), as well as the Senate, had a vested interest in keeping the numerous poor occupants of Rome satisfied as a way of maintaining public order. In addition to the corn dole, one method was by maintaining un- or semi-skilled employment in public building projects.88 These projects, of course, also distributed wealth to individuals across the socio-economic spectrum in Rome,89 developing patronage networks with political benefits. Augustus was clearly seeking to underscore this patronage in his Res Ges- tae, and it is telling that all the construction projects mentioned in the work were located at Rome.90 While presumably the considerable cost of the ad15 project would technically have been paid from funds under the control of the Senate (the aerarium),91 it is clear even from the passage discussed above that Tiberius could have decided the project’s future, if he had wished. This reality was surely known to those receiving the funds. Julius Caesar’s proposal for flood control would have secured a grateful Rome (if it had actually alleviated the flooding problem) but had the added advantage of conspicuously spending the funds in the city itself. The creation of the Aqua Augusta, considered above, while not benefiting the city of Rome, did spend its exorbitant funds at a critical place at a critical time. The proposal of Capito and Arruntius may have alleviated flooding, but the

84 Dio 57.14.7–8. 85 Pliny, Naturalis Historia 3.55. 86 CIL 14.3560. 87 Aldrete 2007, 211–217. 88 Brunt 1980; Wilson 2002a, 4; Wilson 2006, 230–231, citing Suetonius, Vespasian 18; Aldrete 2007, 297 n. 23. 89 Geiger 2009. 90 With the exception of the Via Flaminia (20.5), which at least started at Rome. 91 Jones 1950. large-scale water management projects 253 funds would not have been spent in Rome or even in some other critical area. It should be added that this form of patronage does not seem to have been particularly important to Tiberius. He spent little on public building projects, especially when compared to his predecessor Augustus, preferring to accumulate funds.92 The project would also have antagonized communities upstream. Floren- tia, Interamna and Reate probably had patrons in the Senate to act on their behalf. How much weight they carried we cannot know. If the reasons behind the decision of the Senate not to proceed could not be determined a century later by Tacitus, himself a senator, we can hardly hope to do so now, given the chronological and cultural gulf separating us. Of course, the decision-making may not have been rational. There is, however, at least a plausible rationale to the decision, whether or not it was the proximate cause. A number of different modes of analysis (philosophical/religious and political) seem to have contributed to this line of reasoning: impacts upstream outweighed the benefits to the urban centre.

Conclusion: The Return of Modernity

Judging from these two case studies, it seems that the Roman elite around the turn of the era was prepared to consider, and to execute, plans involving large-scale alteration to the hydrologic landscape in order to benefit coastal urban centres and to serve imperial aims. At the same time, they show some mindfulness of the impacts of such projects on communities upstream, particularly municipia and coloniae, and of religious and cultural impacts. There is also evidence of a tension between a desire for productive alterations to the landscape and an unwillingness to upset the existing natural balance. These considerations could sometimes be exploited by upstream communities seeking to avoid adverse effects on their way of life from proposed changes. The confident attitude to remaking the hydrologic landscape shown by the Roman elite prefigured, and, at least in Central-Southern Italy, served as a model for, the modernity of the 18th and following centuries.93 The modern ideals of progress seemingly behind such Roman projects contrast with

92 Thornton 1986; Thornton and Thornton 1990. 93 On modernity and water management, see Allan 2006; Balali et al. 2009; Molle et al. 2009. 254 duncan keenan-jones other, often earlier, conceptions of regress in history, particularly human history.94 Many of the later engineers who were involved in this work made explicit reference to Roman civilization. In the history of water management in Central-Southern Italy, the early Roman Empire is more similar to the 18th and 19th centuries than to any period between the two. There is, however, some continuity in the intervening period, with much effort being expended in maintaining more modest water supply systems. The last recorded repair of the Aqua Augusta was shortly before ad399 and it seems to have finally broken down even before the major eruption of Somma-Vesuvius in the late 5th/early 6th century buried much of its channel. A Roman aqueduct at Abella (Fig. 3), long out of commission, was repaired around ad406 at the instigation of Paulinus of Nola.95 The upheavals in the area during the 5th and 6th centuries probably prevented any further repairs. The construction of a new aqueduct at Salerno during the 9th century was unusual. Almost all aqueduct construction at this time was repair of Roman systems and the Salerno aqueduct was probably the last wholly new aqueduct built in Italy for several hundred years.96 At Rome, the last ancient aqueducts had been maintained, intermittently, until the 11th century due to Papal repairs.97 For a long time after the final breakdown of the Roman aqueducts, generally only very low-level aqueducts seem to have been maintained or constructed. The Acquedotto della Bolla probably supplied Naples during this time,98 but the Neapolitans accessed its water through wells, rendering it vulnerable to contamination. At Romavecchia, where the Aqua Claudia and Anio Novus had soared up to 32m above ground in superimposed, covered stone channels, the Acqua Marrana Mariana, constructed to supply Rome in 1122,99 was an uncovered canal running at ground level. The remains of the Roman aqueducts were sometimes reused in later ones. An early initiative was the reuse of a 1km section of the Aqua Clau- dia and of parts of the Aqua Alexandrina at Rome in the construction of the Acqua Marrana Mariana.100 A further step was taken with the repair of

94 Hughes 1994, 69. These include the widespread ‘metallic ages’ mythological motif found in Hesiod, the Book of Daniel and Ovid. 95 Paulinus of Nola, Carmen 21.706–750. 96 Squatriti 2002, 17. 97 Coates-Stephens 2003, 169; Hostetter et al. forthcoming. 98 Rasulo 2002; Riccio 2002. 99 Motta 1986b, 203–204. 100 Ashby 1935, 222–223. large-scale water management projects 255 the Aqua Virgo at Rome by a succession of Popes between 1453 and 1570.101 Subsequent Papal projects at Rome were the reuse of several ancient aqueducts in order to build the Acqua Felice (completed in 1589) and of the Aqua Traiana to build the Acqua Paola (completed in 1613).102 Sufficient cen- tralized authority to undertake such large-scale repairs and constructions seems to have reappeared at Naples rather later. The Spanish viceroy of Naples commissioned a four-year study (completed in 1560)103 of the feasibility of repairing the Aqua Augusta, but the repair was never undertaken. The same viceroy had, however, repaired the ancient aqueduct supplying Puteoli known as the Acquedotto Campano in 1540.104 Further repairs to this aqueduct are attested in 1640, 1695 and 1863..105 Outside Central-Southern Italy, the reunification of Italy seems to have given impetus to such projects, with the reactivation, or investigations to do so, of Roman aqueducts at Bologna and Forlì (1880) and Corfinium (1888–1900).106 Pietrantonio Abate was an engineer in the Kingdom of Naples in the mid-19th century and a leading proponent of the repair of the Aqua Augusta. After more than two decades of research into the feasibility of the project, he introduced his final report in this way: The art under consideration (water supply), the origin of which goes back to furthest antiquity, has two hey-days: classical Rome and the present. A protracted period of decline and misery, which extends into the early years of the current century, separates these two outstanding epochs.107 Abate’s proposal of the repair of the Aqua Augusta was also not taken up. Instead an entirely new aqueduct, the Acquedotto di Serino (completed in 1885)108 was built, which was symptomatic of the increasing technical confidence and capability in Central-Southern Italy (with British assistance in this case). The springs used for the Aqua Augusta were again transferred out of their drainage basin to supply Naples in 1938 when they were added to Acquedotto di Serino.109 The ancient Romans had been surpassed but

101 Ashby 1935, 170; Evans 2002, 72. 102 Cancellieri 1986; Motta 1986a. 103 Lettieri 1560. 104 Scherillo 1844, 101–103. 105 De Criscio 1881, 58–69. 106 Bologna and Forlì (Santarelli 1882), Corfinium (De Nino 1900, 642). Evidence outside Italy also comes from Nimes in the 19th Century (Espérandieu 1926; Fabre et al. 1991). 107 Abate 1864, 2. 108 Naples Water Works Company Ltd 1885. 109 Lettieri 1560; Sgobbo 1938. 256 duncan keenan-jones not forgotten: a new water system inaugurated in 1923 was commemorated by the addition of another Latin inscription in the already crowded Porta Maggiore.110 The diversion of the Clanis (Chiana) from the Tiber into the Arno, planned in ad15, was actually carried out piecemeal between about 1500 and the eighteenth century. By that time, the purpose was not flood-control, but land reclamation.111 Land reclamation on the Roman scale had to wait until the mid-19th century and it continued under Mussolini. The partial draining of Lake Fucino in the 1st century ad, mentioned above, was carried out again (and completed) between 1861 and 1875.112 Lake Mezzano was similarly drained during the Roman period and then partially drained again at some point after the 17th century.113 Thus the history of water management in Central-Southern Italy shows the persistent influence of Roman technical achievements and thought right through to the early 20th century.

110 Aicher 1995, 168. 111 Alexander 1984, 547. 112 Giraudi 1998, 2. 113 Giraudi 2004. PART FIVE

FINALE

THE MEDITERRANEAN ENVIRONMENT IN ANCIENT HISTORY: PERSPECTIVES AND PROSPECTS*

Andrew Wilson

The papers at the conference from which this volume is derived concentrated chiefly on energy, climate and climate change, and environmental questions of land use, deforestation, mountains and rivers. This reflects perhaps the current focus of historical research on the ancient environment— but we should recall that several important facets of the environment, including the marine environment, pollution, and natural disasters such as earthquakes and volcanic eruptions (on which more below) are absent from such treatments. Overall, they reflect not only a current interest in mankind’s impact on the natural environment, but also renewed interest, in the face of advances in climate science, in how the environment influences human action, and history.1

Energy

Paolo Malanima’s chapter stresses the importance of understanding the ‘energy budget’ when considering ancient economies. His questions are important and go to the heart of debates over the environment and ancient economic performance. Although he presents no real data for the ancient world, his model suggests some of the constraints on pre-industrial societies and explains why climate change would have had such a big impact on the Roman world and other pre-industrial societies. But the approach remains very general, with sweeping assumptions and questionable proxies. The graph of femur length in fig. 3 shows a rise initially because most British samples were excluded from the period ad 1–49 as Britain lay outside the Roman Empire for most of this period, but were included in the ad 50–99

* I am very grateful to William Harris for inviting me to be a respondent at the conference in Rome, and to all the participants for stimulating debate and ideas. I thank especially Sturt Manning, Mike McCormick and Kyle Harper for helpful discussion and references. 1 This is not to be confused with environmental determinism, which is the notion that the physical environment determines human character and intelligence. 260 andrew wilson and later periods; the rise is partly thus due to the inclusion of more northern European samples from populations with a high degree of pastoralism and thus milk and meat consumption, which typically correlate with greater stature. Archaeological wood remains cannot straightforwardly be used to estimate the extent of forest clearance, contra his fig. 6. Moreover, the discussion ignores cultural specifics, which for the Roman world in particular, would include the need for fuel generated by the widespread practice of public bathing in large, heated bathhouses. Nor can we safely assume that ‘fuels different from firewood represented a negligible share of the total’: archaeological evidence from North Africa, Syria and Italy shows that olive pressing waste was used to fire kilns and even bakers’ ovens at various sites, some but not all of which lay in wood-scarce regions.2 There is a methodological danger in borrowing figures (for fuel consumption per capita) from other periods or places and then using these to establish Roman performance or consumption or whatever—it denies us the possibility of finding out what was specific to or characteristic of the Roman world, which is precisely what we want to know. For example, the assumption that sailing ships were not more numerous in the Roman Empire than in medieval and early modern Europe is not self-evidently secure, nor were ships of 400 tons ‘rare’; there were clearly fewer of them than 100-ton ships, but that is not the same thing. Nor is it quite the case that water-mills and sailing ships between them provided 100% of the mechanical energy supplied by non-biological converters. Besides grain mills, water-power was also used to drive saws and ore-crushing devices, and also water-lifting wheels for irrigation. A consideration of the energy budget ought also to include the use of flowing water for downstream river transport, which led to significant differences between the costs of upstream and downstream transport.3 The potential for downstream river transport was of course regionally limited, but had a substantial effect on shaping regional economies of major river valleys, especially the Nile (100-ton barges were not uncommon), the Rhône and the Rhine.4 As to the role of animals in the ancient economy, we should remember that they were not used only for ploughing and transport, but for driving irrigation machines, at least from the third century bc onwards.5 There is clearly

2 Wilson 2012b, 149–150. Kilns at Leptiminus (Lamta, Tunisia): Smith 2001, 434–435; kiln at Androna (al-Andarin, Syria): Mango 2011, 108; bakeries at Pompeii: Monteix 2009, 7–8. Cf. McCormick 2012, 73. 3 Russell 2009, 113–114. 4 Cf. Franconi 2013. 5 Oleson 1984; 2000; Wilson 2012b; Malouta and Wilson 2013. the mediterranean environment in ancient history 261 much more to be done in teasing out the similarities and differences in the availability and use of the energy budget in different pre-industrial societies, Rome included.6

Climate Reconstructions

An important study in Science published in February 2011 presented a climate reconstruction for central Europe over the last 2,400 years, based on northern European tree-rings, including April–May–June precipitation levels and June–July–August averaged temperatures normalised to the twentieth-century averages.7 The study noted some suggestive parallels with major historical events: the period of Roman conquest coincided with a marked dip in summer temperatures in the late first century bc; the Roman Empire saw, by and large, favourable rainfall and temperatures, followed by a period of much greater climatic instability, including lower temperatures, in the late third to late sixth centuries. Given Malanima’s emphasis on the significant economic effect that a rise or fall in average temperatures of even 1°C could have, it therefore becomes important to ask: to what extent, and how, do these northern European data relate to the Mediterranean? Are they part of a more global trend in climate, that affected the Mediterranean world, and, in its time, the Roman Empire, as a whole? Interestingly, increasing precipitation in the fourth to sixth centuries shown in the northern European data seems to parallel a phase of increased runoff and erosion which has been identified (but not closely dated) in sites along the Tunisian coast by the Franco-Tunisian coastline survey which identified a late Roman erosive peak which the researchers attributed to climate rather than anthropic causes.8 But the continental European and Mediterranean climates are different and should not be expected to track each other precisely; and McCormick’s paper in this volume reminds us that there is not a close correlation between the northern European data and Mediterranean written records. We need much more work on Mediter- ranean climate data—large-scale analyses of tree-rings, to compare with the northern European study, and a coherent programme of pollen analyses, especially from regions around the southern shores of the Mediterranean. Nonetheless, a new synthetic review of multiple series of proxy climate

6 Cf. Wikander 2008; Wilson 2012b. 7 Büntgen et al. 2011. 8 Slim et al. 2004, 251–253. 262 andrew wilson data, while identifying regional variation, underscores some broad correlations between climate and general economic performance, and some specific events.9 In particular, it notes that the period from the first century bc to early third century ad, coinciding with the peak of the Roman imperial expansion and apparent prosperity (from the archaeological record), was unusually favourable from a climatic point of view. Sturt Manning’s reworking of some of the northern European dendrochronological data (his figs. 12–13), shows rises in both precipitation and temperature which appear to coincide with a peak in the expansion (numbers and sizes) of villas in Britain in the fourth century; this relationship is worth exploring further. When considering the possible inter-relationships between climate and history, attention has until recently largely been focused largely on catastro- phes and problems, especially the fall of the Roman Empire, and the large- scale population migrations of late antiquity, for example the migrations of the Huns and Avars discussed in Ed Cook’s chapter in this volume.10 But the tree-ring-based climatic reconstruction by Büntgen et al., and the wider multi-proxy syntheses in this book by Sturt Manning, and elsewhere by McCormick et al., remind us of the significance of the Roman warm period in Europe, which coincided with a period of general economic prosperity in the Roman world from the first to the mid-third century ad. Clearly climate is not a monocausal explanation for the success of the Roman Empire at its peak, but it does rather look like one of many contributing factors. The fall of the Roman Empire was not of course not a single event, but more of a process. What emerges as remarkable from the new climate evidence, if it is applicable to the Mediterranean too, is how resilient the Roman Empire was through the third century ad, because it now seems as though, in addition to everything else thrown at it in the course of that century, the Roman Empire also had to contend with a marked deterioration in climate which will have put agricultural production, and therefore the economy at large, under additional stress.

Climate, Settlement and the Economy

The relationship of the expansion and contraction of settlement, as detectable by archaeological survey, to climatic change is still very unclear. Do we see favourable climatic conditions leading to expansion, or was settle-

9 McCormick, Büntgen et al. 2012. 10 E.g., in a popular vein, Keys 1999. the mediterranean environment in ancient history 263 ment expansion underpinned by better irrigation technology and greater capital investment? Were past societies at the mercy of natural events, men hopping ineffectually on the earth’s crust powerless in the grip of unshake- able natural forces, or were people in some control of their environment, and able to control their destiny by shaping it, at least to a degree? Paula Kouki’s paper on the Jabal Harun is ambivalent on this point. So too the long- running debate over Roman North Africa is still not fully settled; its origins date back to the nineteenth century, when some argued that Roman Africa’s prosperity must have been facilitated by a more favourable climate,11 while others argued that it was achieved at the price of extensive investment in artificial irrigation and water management systems.12 What relatively little work has been done on the subject in recent decades has not resolved the question: we have a few, inconclusive pollen studies,13 and even the UNESCO Libyan Valleys Study concluded that it was unable to answer the question definitively; the climate in the Roman period was thought to be broadly similar to that today, although any detectable variation in ancient vegetation cover may have been due either to cultivation or to climate change, but the environment was so marginal that potentially even a small variation could have made a big difference.14 Whether there was such a variation, and whether it did make a difference, the project could not say. As Malanima points out, a change of one degree centigrade in the average temperature affects cultivation altitudes by 100m. Warmer climates thus mean that more upland areas can be cultivated, and cooler climates restrict the cultivated surface. Of course, things are rather more complicated than that because mountainsides are not totally abandoned when the climate becomes cooler; rather their landuse would change from vines or olives to forest, which might still be used for fuel, or timber. What we do see, in parts of North Africa at least, is a considerable expansion of the cultivated surface as marginal lands were increasingly brought into cultivation. This is particularly apparent in the Kasserine Survey, where the Roman period saw intensive exploitation of nearly all hillsides and mountainsides, which were terraced to enable their systematic cultivation with vines and olive trees.15 Pottery from the terraced slopes dated from the first to seventh centuries ad; there was no sign of the slopes being used

11 E.g. du Coudray la Blanchère 1895, 4. 12 Gsell 1921, 56–99; cf. Shaw 1981, 380, 383, 388–390; Wilson 1997, 33–35. 13 Rouvillois-Brigol 1985; 1986; cf. discussion in Wilson 1997, 34. 14 Barker et al. 1983, 82–84; Hunt et al. 1985, 13; Hunt et al. 1986, 14–16; Hunt et al. 1987, 4–6. 15 Hitchner 1988, 28; 1989; 1990; 1992; 1995b. 264 andrew wilson before the Roman period, and little afterwards. Bruce Hitchner has argued that these developments reflect the kind of incentivisation to expand the cultivated area that we can see on imperial estates in the first and early second centuries ad, in the lex Manciana and the lex Hadriana de rudibus agris, which provided for rent-free periods on uncultivated land that was brought under exploitation, while newly planted trees and vines came to maturity.16 The significance to the present debate is that this expansion was happen- ing in a climatically favourable period. Climatic warming was not necessary to enable cultivation of these hillsides, although it remains an open question whether this exploitation of marginal lands was at least facilitated by (slightly?) increased rainfall. It is possible, therefore, that the cultivation of these more marginal lands is to be explained as a result of population pressure, and perhaps economic growth, taking advantage of expanded or better connected markets, rather than as a result of climate change. Debates over Roman economic growth are tending to countenance scenarios where there was some per capita growth from the reign of Augus- tus to the mid second century ad, at which point it was stopped by an exogenous shock. There are dissenting voices—Walter Scheidel, for example, argues that such growth as there was from the late Republic to the second century ad was neither sustained or sustainable but rather a by- product of the ‘peace dividend’ following the end of the Civil Wars and the unification of the Mediterranean under direct or indirect Roman control, and was terminated by Malthusian constraints.17 Yet the growth/shock scenario has found considerable favour,18 and even Scheidel has accepted that the economy received a major shock in the later second century. So far the Antonine Plague has figured in debate as the best candidate for such a shock, and this is accepted by a number of historians, though by no means all; there remains in fact considerable debate over the impact of the Anto- nine Plague.19 But Mike McCormick’s contribution presents data for Nile flood events extracted from papyrological records20 which suggest a marked

16 Hitchner 1995a. 17 Scheidel 2002; 2009, 67–70. 18 E.g. Temin 2006; Jongman 2007a. 19 Severe: Duncan-Jones 1996; Scheidel 2002. Less severe: Bagnall 2000; 2002; Greenberg 2003; Bruun 2003; 2007. See now the various papers in Lo Cascio 2012, where the majority is in favour of the ‘severe’ opinion. 20 McCormick (this volume); see also McCormick, Büntgen et al. 2012; full dataset (on which the following relies) available online as McCormick, Harper et al. 2012: ‘Geodatabase of Historical Evidence on Roman and Post-Roman Climate’ at http://darmc.harvard.edu/icb/ icb.do?keyword=k40248&pageid=icb.page496495 (accessed 16 December 2012). the mediterranean environment in ancient history 265 change in the climate-related agricultural potential for Egypt in the mid- second century. Although the pattern of average and somewhat better than normal floods was much the same across the period, before ad155 exceptionally good floods occurred on average one year in five, but afterwards only once every twelve and a half years; while poor or failed floods occurred every five years or so before the mid second century, but more frequently—every three years—afterwards. This deterioration in the pattern of Nile floods would have depressed agricultural production, and exposed the Egyptian populace to greater risk of famine and malnutrition, especially as the state insisted on extracting annona grain even in poor years. McCormick chooses ad155 as the year dividing the two different sets of flood regimes, but ad162 or 166 would do equally well, i.e. at around the same time as the great epidemic. Is the deterioration of the Nile flood regime an additional, or alternative, perhaps even better, explanation, for the economic troubles observed in Egypt than depopulation caused by the plague? Did the epidemic ravage a population already to some extent weakened by poorer harvests? Or are the observed discontinuities in the proxy data from Egypt which have been used to argue for the effects of the plague in fact due not to the plague but to the effects of poor harvests consequent on bad or failed Nile floods? In general, and not just in the period of the epidemic, there is in fact a striking apparent correlation with recorded incidents of flight from tax-collection— to take only the examples mentioned by Gilliam in his discussion of whether or not the Antonine Plague had a serious effect,21 the anachoresis mentioned in a document of ad55/5922 would follow a poor decade: three consecutive years of poor or very poor flooding ad45–47, four years of average floods (ad49–52), three years of very poor floods (ad53–55), a normal flood in ad57 and a very poor flood in ad59, with no good recorded floods in the years 45–59. Further evidence for flight from villages in the Fayyum in ad162 correlates with a year in which the Nile flood was low;23 the depopulation of villages in the Mendesian nome in ad168/169 and the revolt of the Boukoloi in ad172 or 173 can be seen to have followed a series of bad years:24 a very poor flood in 166, an average flood in 167, three successive poor years in 168–170, and an average flood again in 171 (data for 172 and 173 themselves are not available).25

21 Gilliam 1961, 240–242. 22 P. Graux 2 = Sammelbuch IV, 7462. 23 P. Berl. Leihg. 7. 24 Mendesian nome: BGU 903. 25 Data from McCormick, Harper et al. 2012. 266 andrew wilson

The Nile flood records, which ultimately reflect rainfall in the Nile’s headwater catchment area in Ethiopia and Central and East Africa, remind us that events affecting the environment outside the Mediterranean world may have repercussions for the Mediterranean too—either directly through transport of precipitation, or less directly through their effects on neigh- bouring populations. Büntgen et al. have drawn renewed attention to the links between climate and the migrations of late antiquity in northern Europe.26 Recent research on the Garamantes of the Libyan Sahara has emphasised how the incipient urbanism, oasis agriculture and long- distance trading contacts with the Roman world were underpinned by irrigation from subterranean foggara systems tapping groundwater based on foggara irrigation; and when these irrigation systems failed in late antiquity the nexus of urban settlement and long-distance trade collapsed.27 The precise chronology remains uncertain, as does the question of to what extent this failure was due to over-exploitation of a fossil, non-renewable, aquifer, or to decreased local rainfall as a result of climate change;28 but the environmental changes that led to the break-up of Garamantian power over the Libyan Sahara seem to have had broader repercussions, and may be linked to migrations out of oases in the northern Sahara, and the incursions of Libyan tribes into late Roman, Vandal and Byzantine North Africa in the fourth to sixth centuries ad.29

Ancient Understanding of the Hydrological Cycle

To explore the extent to which past societies might have been able to control and manage their environment, it it perhaps useful to review Roman understanding of the hydrological cycle.30 Vitruvius (fl. c. 50–26 bc) gives a description of the hydrological cycle based on Greek sources (he cites Theophras- tus, Timaeus, Posidonius, Hegesias, Herodotus, Aristides and Metrodorus).

26 Büntgen et al. 2011. 27 Mattingly 2003; Wilson and Mattingly 2003; Mattingly and Wilson 2003; Mattingly and Wilson 2010; Wilson 2012a. 28 Cremaschi et al. 2006 argue for a dramatic drop in Saharan rainfall in the Ghat region around ad450; but their dendro-record is based on a single tree for large stretches of the reconstruction, and cannot be regarded as reliable without cross-matching (Sturt Manning, pers. comm.). Moreover, rainfall in the Ghat region may have had little connection with that in the Wadi al-Ajal, the Garamantian heartlands, 400 kms. to the north-east. 29 Fentress and Wilson forthcoming. 30 The following section is based on Wilson 1997, 35–39, which treats the sources at greater length. the mediterranean environment in ancient history 267

Rainfall is caused by the air acquiring moisture from the ground and from springs, rivers, marshes and the sea, which is evaporated by the heat of the sun. It is carried upwards in the form of clouds, and is precipitated when the clouds strike mountains (de Architectura 8.2.1–4). Long-distance transfer of water resources is described in the case of south winds, which desiccate hot areas and redeposit the moisture elsewhere (8.2.5). Water (including melt- water) infiltrates underground and emerges as springs (8.1.7). Seneca (c. 5/4 bc–ad 65) devotes Book 3 of his Naturales Quaestiones to a discussion of water, rivers and springs, and two partially preserved books (4A and 4B) to the Nile and to hail and snow. A lost work of Theophras- tus seems to have been one of his main sources. He discusses a number of observed phenomena, but his explanations for them are influenced by Stoic notions of cosmic flux and are often less accurate than those of some of his predecessors. He plays down the role of groundwater, believing that rain never penetrates the earth to a depth of more than ten feet, and most of it is carried off to the sea by rivers (Nat. 3.7.1). He notes that in arid regions wells may be driven 200 or 300 feet before encountering water, and calls this aqua viva, ‘living water’, as it cannot have come from rainwater infiltration (3.7.3–4). Rivers are formed not by groundwater but by condensa- tion of air in hollows within the earth (3.9.2–3); rain does not cause rivers, but only makes them flow faster (3.11.6). He does not give an account of the hydrologic cycle, and states that the sea has its own springs (3.14.3). However, some observations are more accurate; deforestation causes the emergence of springs because water is no longer consumed by vegetation (3.11.3–5). Pliny (ad 23/24–79) indicates an awareness of seasonal variations in groundwater levels; he says that wells generally run dry about the rising of Arcturus (17th September), and that springs vary their discharge according to season (Nat.Hist. 31.xxviii.50–51). Earthquakes may cause springs to dry up or appear (31.xxx.54). Springs can also arise as a result of deforestation, and tree cover inhibits erosive streams (31.xxx.53). Pausanias (fl. c. ad 160) displays awareness of the effects of agriculture on increasing erosion and the concomitant downstream deposition of alluvial fans (8.24.10–11). Certain ancient writers therefore display a basic understanding of the hydrological cycle, the formation and behaviour of groundwater, and of the effects of agriculture and deforestation on erosion. Knowledge was available to tap aquifers, locate springs, use dams to recharge groundwater reserves, and exploit mineral or thermal properties of water. The principle that subterranean reservoirs could store winter rains for discharge in the summer months was at least theoretically understood, even if Aristotle had refuted it 268 andrew wilson as the explanation for the origin of rivers (Mete. 349b3–15). There was clearly much interest in different types of water, their properties and their effects on health. We are familiar with the fact that Near Eastern, Greek and Roman hydraulic engineering was capable of remedying local deficiencies in water availability by the transfer of water resources from elsewhere via aqueducts, which could in the Roman and Byzantine periods achieve extraordinary lengths—98km for the second-century aqueduct feeding Carthage from the springs at Zaghouan, later increased to 132km by the addition of tributaries in the Severan period; and the aqueduct that supplied Constantinople, completed in ad373, had a main channel of 227km with a tributary of 41km.31 Duncan Keenan-Jones’ discussion of the Serino aqueduct in Campania in this volume emphasises the capability of Roman engineers not only to effect inter-basin transfers of water (something common in fact to many Roman aqueducts), but also to think on a regional scale in terms of supplying several cities. The highly urbanised region of the Bay of Naples demanded extraordinary solutions to the problem of supplying such a large and nucleated urban population with water.

Case Study: Controlling Tiber Floods A further illustration of the developing ancient understanding of total river basin management, which enables us to go further than the information given by the ancient authors mentioned above, is provided by the history of Roman attempts at controlling the flooding of the Tiber. Duncan Keenan- Jones’ chapter in this volume discusses two of the big proposed flood relief schemes, under Caesar and Tiberius; here I want to consider them in the light of the overall chronological development of attempts at Tiber flood control. In the last thousand years the Tiber has flooded to levels of 10–13 masl on average every year, and above 13 masl on average every two years.32 As Keenan-Jones notes, the wetter conditions suggested by climate reconstructions for between c. 100bc and ad200 would have exacerbated this pattern in the Roman period. The lowest-lying areas in antiquity, most exposed to flooding, were the Campus Martius, the Velabrum, forum, the Circus

31 Generally: Wilson 2008a. Constantinople: Crow et al. 2008; Crow 2012, 40. 32 Le Gall 1953, 62–65; Lugli 1953, 61–69; Ammerman 1990, 637; Belati 1999; Aldrete 2007. the mediterranean environment in ancient history 269

Maximus and Emporium areas below the Aventine, and Trastevere. The Pons Sulpicius had to be rebuilt on numerous occasions following damage by flooding.33 From the period of the Etruscan kings through to the High Empire, progressively more ambitious and effective means of flood control were considered, some of which were not carried out for political reasons. Coring work in the forum and the Velabrum by Albert Ammerman has shown that in the sixth century bc the ground level in these areas was artificially raised by between 1 and 4m by dumps containing gravel, tuff fragments and anthropic material, as a landfill scheme intended to bring the ground surface above the level of many floods.34 The ground surface would have needed to be raised some 2m in the centre of the valley, to bring it to around the 9m contour level that would put it above all but the worst floods. Ammerman estimates that at least c. 10,000m3 of fill would have been needed; possibly twice that amount. Such an exercise implies considerable manpower resources, and may have occurred over a period of several years; Ammerman notes parallels with other major landscaping works under the Etruscan kings. If the construction of the Cloaca Maxima to drain the forum area went hand in hand with raising of its surface above the level of the Tiber floods, the emphasis in the ancient writers on the arduous nature of the corvée work becomes easier to understand.35 These early solutions, involving considerable landscape transformations of the forum valley, made possible the creation of a public space in what was to become the heart of Rome, and laid the foundations for the later urban development of the city. Impressive and labour-intensive as they were, their effect was to raise certain areas only above the levels of many, but not all floods; the problem of Tiber flooding had been alleviated, but not solved. As Rome expanded in the mid and late Republic, Tiber floods began to affect a larger inhabited area. The development of the Campus Mar- tius, started by Pompey and Caesar and further continued under Augustus, compounded the problem.36 There were six severe floods in the reign of Augustus, and a further, disastrous, flood in ad 15, the year after his death. Tiberius, instead of consulting the Sibylline books as advised, appointed a

33 Tac. Ann. I.86 (23bc); SHA Antoninus Pius 8; Dio Cassius 1.3.20; 27.58 and 50.8; Desnier 1998, 515, 520. 34 Ammerman 1990; 1998. 35 Ammerman 1990, esp. 636–645. 36 Belati 1999, 13. 270 andrew wilson permanent commission of five senators, the curatores alvei Tiberis et riparum, who had responsibility for managing the banks of the Tiber—their duties included ensuring the dredging of the river channel and canalising its banks (marked by cippi). These measures contained some of the lesser floods, but still did not remove the problem; floods continued almost annually. Part of a bridge pier belong to the ancient Pons Aurelius, found below the Ponte Sisto in the nineteenth century, seems to have been marked with a Tiber flood gauge; it was graduated in Roman feet in reverse order, apparently therefore showing the height remaining until the river reached a danger or warning level.37 Up to this point, then, we have seen the development of attempts to contain flooding within the urban zone. Fundamentally different were efforts to tackle the cause of the problem, which we see for the first time in a scheme considered by Caesar, but never put into effect. This was the construction of a relief channel leaving the Tiber near the Mulvian bridge and running around the back of the Janiculum hill, to feed back into the Tiber below the city.38 Caesar’s assassination prevented the scheme being carried out. Serious flooding recurred almost annually, and throughout the course of the first century ad increasingly ambitious attempts started to deal with the root of the problem, rather than simply to contain the symptoms. After a major flood under Tiberius, a scheme was proposed to divert the Chiana, one of the Tiber’s tributaries on the right bank upstream from Rome, into the Arno to reduce the amount of water flowing through Rome, and that the waters of another tributary, the Nar, should be dispersed throughout its floodplain, and the artificial channel feeding the Nar from the Veline lakes be dammed up. This provoked a senatorial debate and the scheme was successfully opposed by the local communities (Florence, Interamna, and Reate) who would have been affected by this move, using a variety of arguments both practical and religious.39 At the downstream end of the Tiber Valley, Claudius’ works on the new harbour at Portus included the digging of relief channels to link the Tiber to the sea, above the sharp bend by Ostia which acted as a bottleneck and caused the waters to back up, affecting the river as far upstream as Rome. Claudius’ inscription makes explicit the link between the works at Portus and the risk of flooding at Rome (ILS 207 = CIL XIV.85):

37 Marchetti 1892. 38 Cicero, Att. 13.33.4; Aldrete 2007, 182–184. 39 Tacitus, Annals I.79. Aldrete 2007, 185–188; Campbell 2012, 118–119; Keenan-Jones (this volume). the mediterranean environment in ancient history 271

Ti. Claudius Drusi f. Caesar … fossis ductis a Tiberi operis portu caussa emis- sisque in mare urbem inundationis periculo liberavit Tiberius Claudius Caesar, son of Drusus, … freed the city from the danger of flooding by leading canals from the Tiber into the sea in connection with the building of the harbour. This scheme shows that Claudius’ engineers evidently understood the systemic nature of flood control, and that it might be necessary to effect works many miles distant from the point at which one desired to control the floods. An additional effect of Claudius’ works must also have been to reduce the extent to which Ostia was exposed to regular and severe flooding. Claudius’ boast that he had liberated the city from the danger of flooding was cruelly shown to be premature by the severe flood of ad. 69, recorded by Tacitus (Hist. I, 86.2), and also by the inscription commemorating Trajanic works (below). There was further raising of the ground level in the Campus Martius in the later first century ad: under Domitian the pavement of Augustus’ sundial was repaved at a level 1.5m higher than the original. Indeed, it may be that the flooding of the Tiber in this area caused the obelisk to settle out of true, explaining why Pliny says the sundial was inaccurate only 50 years after its construction. The Trajanic or Hadrianic pavement of the Pantheon lies 1.85m above that of its Agrippan predecessor.40 Trajan constructed further exit channels from the Tiber to the sea near Ostia and Portus; if the reconstruction of some of the missing text is correct, Claudius’ measures had not been entirely effective (ILS 5797): [Imp. Caesar divi] | Ne[rvae fil. Nerva] | Tra[ianus Aug. Germ.] | Dac[icus trib. pot.] … | im[p. …, cos. … p. p.] | fossam [fecit | q[ua inun[dationes Tiberis | a]dsidue u[rbem vexantes | rivo] peren[ni instituto arcerentur] Imperator Caesar Nerva Traianus Augustus Germanicus, son of the deified Nerva, … made a canal by which the floods of the Tiber which continually troubled the city were kept away with the construction of a permanent channel. Post-Trajanic measures seem to have been limited largely to repairs to the Tiber banks and dredging of its channel, for example under Aurelian (SHA Aur. 47, 1–3). The larger flood relief schemes of the first and early second centuries ad imply a recognition that effective solutions to the problem of the Tiber

40 Belati 1999, 16. 272 andrew wilson floods required a landscape-wide solution which addressed the whole catchment area of the river valley, and its estuary. However, political considerations prevented the most ambitious schemes (Caesar’s relief channel and the Chiana diversion), which were also the most likely to have provided an effective longer-term solution, from being put into action.

The Environment of Production

The differential costs of land, riverine and maritime transport influenced the geography of production in a manner clearly seen along coastlines and major river valleys. In the ancient Mediterranean the preferred container for long-distance transport was for centuries the amphora, although it gradually gave ground to the barrel. Ceramic amphorae were heavy, however, and their production was preferentially located along navigable rivers and near the coast. In Baetica, a major olive-oil producing region, the kilns which made the Dressel 20 amphorae in which the oil was shipped were located along the Guadalquivir and its main tributary, the Genil, and the oil was transported in skins from the farms to the kiln sites for bottling and riverine transport in amphorae down to Hispalis (Seville) and then loaded onto maritime ships for export to Rome and other distant markets. In the other major olive oil exporting regions, modern Tunisia and north-west Libya (Tripolitania), amphora production was chiefly concentrated at coastal sites which acted as the bottling centres for the produce of the hinterland. Amphora production, and ceramic production in general, was also strongly influenced by the availability of fuel, especially important where production had been scaled up for large-scale long-distance export. The otherwise puzzling location of the large-scale southern Gaulish samian production centre at La Graufesenque may be explicable to a large degree by the ready available of firewood (the clay deposits were some kilometres distant). In Tunisia, where firewood was scarce, and the waste from olive pressings was used as fuel, some co-location of olive oil production and amphora and cookware manufacture, often at coastal sites, is observable.41 Coastal sites also had the advantage that firewood could be imported by boat, as depicted in a mosaic from Sousse.42 Confirmation of the importance of a

41 Olive pressings as fuel at Leptiminus, Tunisia: Smith 2001, 434–435. Co-location of amphora and cookware production: Leitch 2011. 42 du Coudray la Blanchère and Gauckler 1897; Meiggs 1982, 10, Pl. I.6; Meiggs 1982, 529–530; Wilson 2012b, 149. the mediterranean environment in ancient history 273 long-distance maritime trade in wood fuel comes from a second- or third- century papyrus from Tebtunis, referring to the import of Italian firewood.43 William Harris’ contribution to this volume discusses the extent of possible deforestation in the Greek and Roman periods, suggesting hesitantly that such deforestation as there was in classical times was probably more in the nature of land clearance, and most visible in Attica; deforestation probably increased in the Hellenistic and Roman periods but the effects of demand were to some extent offset by improved woodland management and increased long-distance maritime trade that mitigated the impact on their immediate hinterlands of the fuel demand created by large cities. His discussion emphasises the scale of ancient fuel consumption and the trade both in firewood and in construction timbers.44 For firewood, it is not fully clear whether fuel wood was imported to certain regions because it had to be (i.e. was there serious deforestation in North Africa, Egypt, and the vicinity of Rome?), or because it could be, cheaply, for example as return cargoes. Robyn Veal’s chapter on fuels stresses the use of charcoal as well as firewood, and shows the potential for ancient fuel studies, but we should also consider the possible uses of non-wood fuels—as mentioned above, Pompeii, North Africa and Syria have all yielded evidence suggesting that the use of olive pressing waste as a fuel in commercial ovens and kilns was common; and the use of animal dung as a fuel for purposes not requiring very high temperatures, such as domestic hearths and bread ovens, has a long history in Asia Minor and the Near East, and has been identified in Numidian levels at Althiburos (Tunisia).45 The scale of the trade in construction timber will evidently need to be taken into account when we get to the stage of doing the tree-ring analysis for the Roman Mediterranean. What mixing of the dendro-record might it create? Can we use the tree-ring research to identify the sources of timber?

Future Directions

Our understanding of the scale and effects of ancient pollution still has a long way to go. Exemplary regional studies such as that in the Wadi Fay- nan, Jordan, have shown the local effects of Roman and Byzantine copper

43 P. Tebt. II.686 (cited by Harris, this volume). 44 Harris (this volume); cf. Wilson 2012b: 139–140. 45 W. Brown 1820, 297–300. Althiburos (dung in an oven of the 4th–2nd century bc): Portillo and Albert 2011, 3232. 274 andrew wilson smelting and mining activities there, and how the soil still remains massively polluted, with long-term effects on animals and the human population of the region even today.46 The identification of anthropogenic pollution in Greenland ice cores in the 1990s, principally from copper and lead/silver working, has attracted much attention, but the original studies used relatively few samples and produce graphs with very spiky peaks and troughs.47 Much work on ice-cores in both the Arctic and Antarctica has been done since the original studies, and we can soon expect to have a high-resolution record of atmospheric pollution over the last 3,000 years, with annual or subannual resolution and an absolute dating precision of ±2 years, that could enable further investigation of levels of ancient mining and metallurgical activity, and their environmental impact. Most of the papers in this volume concentrate on processes and long- term trends, rather than catastrophic events. Volcanic eruptions or their effects are mentioned in passing in several papers (McCormick, Keenan- Jones), but earthquakes are not mentioned at all. These perhaps deserve greater consideration in environmental history—not to overemphasise catastrophe in itself, but rather to assess the relative effects of sudden and long-term change. There is much work to be done here, ranging from assembling more comprehensive lists of ancient earthquakes (including not just those mentioned in ancient sources but also those inferred from archaeological excavations) and volcanic eruptions (again, including those detected principally through tephra or sulphate deposits in ice cores), to the consideration of large socio-economic effects.48 To what extent were ancient states able to respond to natural disasters such as earthquakes or volcanic eruptions, to afford disaster relief, or rebuilding, and how far did they try to do so? The clustering of earthquakes in the fourth–sixth centuries ad known to geologists as the ‘Early Byzantine Tectonic Paroxysm’,deserves greater attention in considerations of Byzantine history.49 Does a particular clustering of earthquakes within this period and around the ‘big one’ of 21 July ad365 with its epicentre off western Crete, help to explain the notable peak in rebuild-

46 Barker et al. 2007. 47 Original studies: Hong, Candelone, Patterson and Boutron 1994; Hong, Candelone, and Boutron 1994; Hong, Candelone, Patterson and Boutron 1996; Hong, Candelone, Soutif, and Boutron 1996; Rosman et al. 1997. Relation to ancient economic history: McCormick 2001, 53; Wilson 2002a, 25–29; 2007; 2009; de Callataÿ 2005; Jongman 2007a, 188–189; Scheidel 2009. 48 Lists of earthquakes: Guidoboni 1989; Guidoboni et al. 1994; Ambraseys et al. 2005; cf. Ambraseys 1994; 2006; Nur 2010. Much archaeological evidence remains to be assessed and added (e.g. a probable earthquake at Euesperides, Benghazi, 262–250bc: Wilson 2003). 49 Stiros 2001. the mediterranean environment in ancient history 275 ing inscriptions between ad360 and 375, especially in North Africa?50 The ice core record, which has trapped sulphates emitted by volcanic eruptions, can also help with reconstructing links between ancient eruptions and climate. A recent study of Scandinavian settlement and tree rings argued that the Norse myth of the Fimbulwinter, the harsh winter that pre- cedes the end of the world, with three years without summer, and the breakdown of social order, with successive wars and brothers killing brothers, reflects the dust veil event of ad536, observable in the dendrochronological record of Scandinavia and in the abandonment of numerous settlements in the mid sixth century.51 The dust veil event had a global reach, affecting the Mediterranean (Procopius speaks of the sun shining only weakly for 18 months) and China.52 The Greenland ice core record shows extremely high sulphate levels for ad533–534±2, which suggests that the dust veil event of 536 was caused by a volcano,53 possibly the Tierra Blanca Joven (TBJ) event of the Ilopango volcano in El Salvador.54 Through a combination of ice core studies, volcanology, archaeology, and the analysis of Norse poetry, we are now able to draw a link between the mythological motif of the Fimbulwinter and, possibly, a volcano in Latin America.

And lastly, what of the marine environment? The effects of eustatic and relative sea-level change have been identified along the coasts of Italy and Tunisia,55 and at other localised points—chiefly harbours and major ports—,56 but systematic study of the Mediterranean coastline, and especially Algeria and Libya where relatively little work has been done, is needed to assess the regional impact of sea-level change on coastal settlement and economies, including the effects on the viability of ports. Conversely, although a certain amount of attention has been paid by historians and archaeologists to human impact on the terrestrial environment, what of the impact on the marine environment and fish stocks. Did the large-scale fishing practised in the Roman period have any measurable impact on the size of particular fish species, or on fish stocks or populations?57

50 Cf. Wilson 2011, 163–165. 51 Gräslund and Price 2012. 52 Gunn 2000. 53 Larsen et al. 2008. 54 Dull et al. 2001; Dull et al. 2010. 55 Schmiedt 1972; Slim et al. 2004. 56 Blackman 1973. 57 Cf. Marzano 2013 on the scale of Roman fishing. 276 andrew wilson

The highest resolution reconstructions of ancient climate currently available pertain to northern Europe where the dendrochronological record is best, but the situation for the Mediterranean is improving, with multi- proxy reconstructions now being produced, albeit often reliant on lower- resolution data. There is a need for more data and higher-resolution studies for the Mediterranean, North Africa and the Near East if we are properly to address the question of how far climate affected history and settlement, and explore to what extent were ancient populations able to combat adverse climate change with adaptive agricultural technologies and water management systems, such as the runoff farming in the pre-desert of Tripolitania or in the Negev. Nevertheless, exciting advances have been made in recent years at an accelerating rate in the fields of ancient climate reconstruction, dendrochronology, and the studies of environmental pollution trapped in ice cores, and the increasing interplay between these kinds of science and historical research is opening up new possibilities for profoundly deepen- ing, and perhaps even transforming, our understanding of the relationship between history and the environment. BIBLIOGRAPHY

Abate, F. 1864. Studi sull’acquedotto Claudio e progetto per fornire di acqua potabile la citta di Napoli. Naples. Abudanh, F. 2006. Settlement Patterns and Military Organisation in the Region of Udhruh (southern Jordan) in the Roman and Byzantine Periods I–II. Unpublished PhD Dissertation, University of Newcastle upon Tyne. Aicher, P.J. 1995. Guide to the Aqueducts of Ancient Rome. Wauconda, IL. Akkemik, Ü., and A. Aras. 2005. “Reconstruction (1689–1994ad) of April–August precipitation in the southern part of central Turkey.” International Journal of Climatology 25: 537–548. Akkemik, Ü., N. Dagdeviren, and A. Aras. 2005. “A preliminary reconstruction (A.D. 1635–2000) of spring precipitation using oak tree rings in the western Black Sea region of Turkey.” International Journal of Biometeorology 49: 297–302. Aldrete, G.S. 2007. Floods of the Tiber in ancient Rome. Baltimore. Alexander, D. 1984. “The Reclamation of Val-di-Chiana (Tuscany).” Annals of the Association of American Geographers 74(4): 527–550. Alexandre, P. 1987. Le climat en Europe au Moyen Âge: contribution à l’histoire des variations climatiques de 1000 à 1425, d’après les sources narratives de l’Europe occidentale. Paris. Allan, J.A. 2006. “Millennial water management paradigms: making IWRM work.” http://www.mafhoum.com/press/53aE1.htm. Allen, J.R.M., W A. Watts, E. McGee, and B. Huntley. 2002. “Holocene environmental variability—the record from Lago Grande di Monticchio, Italy.” Quaternary International 88: 69–80. Allevato, E., E. Russo Ermolli, G. Boetto, and G. Di Pasquale. 2010. “Pollen-wood analysis at the Neapolis harbour site (1st–3rd century ad, southern Italy) and its archaeobotanical implications.” Journal of Archaeological Science 37: 2365–2375. Alley,R.B. 2000. “The Younger Dryas cold interval as viewed from central Greenland.” Quaternary Science Reviews 19: 213–226. Alley, R.B., and A.M. Ágústsdóttir. 2005. “The 8k event: Cause and consequences of a major Holocene abrupt climate change.” Quaternary Science Reviews 24: 1123–1149. Amato, L., V. Carsana, A. Cinque, V. Di Donato, D. Giampaola, C. Guastaferro, G. Irollo, C. Morhange, P. Romano, M.R. Ruello, E. Russo Ermolli. 2009. “Ricostru- zioni geoarcheologiche e morfoevolutive nel territorio di Napoli (Italia): l’evoluzione tardo pleistocenica-olocenica e le linee di riva di epoca storica.” Méditer- ranée 112:23–31. Amato, L., G. Bisogno, L. Cicala, A. Cinque, P. Romano, M.R. Ruello, E. Russo Ermolli. 2010. “Palaeo-environmental changes in the archaeological settlement of Elea- Velia: climatic and/or human impact signatures?” In Scienze naturali e archeologia. Il paesaggio antico: interazione uomo/ambiente ed eventi catastrofici, edited by A. Ciarallo and M.R. Senatore, 13–16. Rome. 278 bibliography

Ambraseys, N.N. 1994. “Material for the Investigation of the Seismicity of Libya.” LibStud 25: 7–22. ———2006. “Earthquakes and archaeology.” Journal of Archaeological Science 33.7: 1008–1016. Ambraseys, N.N., C.P. Melville, and R.D. Adams. 2005. The Seismicity of Egypt, Arabia and the Red Sea: A Historical Review. Cambridge. Ammerman, A.J. 1990. “On the origins of the Forum Romanum.” AJA 94.4: 627– 645. ———1998. “Environmental archaeology in the Velabrum, Rome: interim report.” JRA 11: 213–223. ʾAmr, K., A. al-Momani, S. Farajat, and H. Falahat. 1998. “Archaeological survey of the Wadi Musa Water Supply and Wastewater Project Area.” Annual of the Department of Antiquities of Jordan 42: 503–548. ʿAmr, K., and A. al-Momani. 2001. “Preliminary report on the archaeological component of the Wadi Musa water supply and wastewater project (1998–2000).” Annual of the Department of Antiquities of Jordan 45: 253–285. Andreadis, K.M., E.A. Clark, A.W. Wood, A.F. Hamlet, and D.P. Lettenmaier. 2005. “Twentieth-century drought in the conterminous United States.” Journal of Hydrometeorology 6: 985–1001. Arjava, A. 2005. “The mystery cloud of 536CE in the Mediterranean sources.” Dum- barton Oaks Papers 59: 73–93. Arnaud-Fassetta, G., N. Carcaud, C. Castanet, and P.-G. Salvador. 2010. “Fluviatile palaeoenvironments in archaeological context: Geographical position, methodological approach and global change—Hydrological risk issues.” Quaternary International 216: 93–117. Arthur, P. 2002. Naples, from Roman town to city-state: an archaeological perspective. Rome/Lecce. Ashby, T. 1935. The aqueducts of ancient Rome. Oxford. Asouti, E. 2004a. “Charcoal Analysis—a Short History of the Discipline.” London. Available from http://www.ucl.ac.uk/~tcrneas/History of charcoal analysis.htm. Asouti, E. 2004b. “Factors Affecting the Formation of an Archaeological Wood Char- coal Assemblage.” London. http://www.ucl.ac.uk/archaeology/staff/profiles/ asouti.htm. Atherden, M. 2000. “Human Impact on the Vegetation of Southern Greece and Problems of Palynological Interpretation: a Case Study from Crete.” In Landscape and Land Use in Postglacial Greece, edited by P. Halstead and C. Frederick, 62–78. Sheffield. Bach, A. 1952–1956. Deutsche Namenkunde. Heidelberg. Bagnall, R.S. 2000. “P.Oxy. 4527 and the Antonine plague in Egypt: death or flight?” JRA 13: 288–292. ———2002. “The effects of plague: model and evidence.” JRA 15: 114–120. Bagnall, R.S., A. Cameron, S.R. Schwartz, and K.A. Worp. 1987. Consuls of the later Roman Empire. Atlanta. Baillie, M.G.L. 1994. “Dendrochronology raises questions about the nature of the AD536 dust-veil event.” The Holocene 4: 212–217. ———2008. “Proposed re-dating of the European ice core chronology by seven years prior to the 7th century AD.” Geophysical Research Letters 35: L15813. bibliography 279

———2010. “Volcanoes, ice-cores and tree-rings: one story or two?” Antiquity 84: 202–215. Bakels, C. 1999. “Archaeobotanical investigations in the Aisne valley, northern France, from the Neolithic up to the early Middle Ages.” Vegetation History and Archaeobotany 8: 71–77. Balali, M.R., J. Keulartz, and M. Korthals. 2009. “Reflexive water management in arid regions: the case of Iran.” Environmental Values 18(1): 91–112. Bang, P.F. 2009. “The Ancient Economy and New Institutional Economics.” JRS 99: 194–206. Bannon, C.J. 2009. Gardens and Neighbors: Private Water Rights in Roman Italy. Ann Arbor, MI. Bar-Matthews, M., and A. Ayalon. 2004. “Speleothems as paleoclimate indicators, a case study from Soreq Cave located in the Eastern Mediterranean region, Israel.” In Past Climate Variability through Europe and Africa, Vol. 6, edited by R.W. Battarbee, F. Gasse, and C.E. Stickley, 363–391. Dordrecht. Bar-Matthews, M., and A. Ayalon. 2011. “Mid-Holocene climate variations revealed by high-resolution speleothem records from Soreq Cave, Israel and their correlation with cultural changes.” The Holocene 21: 163–171. Bar-Matthews, M., A. Ayalon, and A. Kaufman. 1997. “Late Quaternary paleoclimate in the eastern Mediterranean region from stable isotope analysis of speleothems at Soreq Cave, Israel.” Quaternary Research 47: 155–168. Bar-Matthews, M., A. Ayalon, and A. Kaufman. 1998. “Middle to Late Holocene (6500 years period) paleoclimate in the Eastern Mediterranean region from stable isotopic composition of speleothems from Soreq Cave, Israel.” In Water, Environment and Society in times of Climate Change, edited by A.S. Issar and N. Brown, 203–214. Dordrecht. Bar-Matthews, M., A. Ayalon, M. Gilmour, A. Matthews, and C.J. Hawkesworth. 2003. “Sea-land oxygen isotopic relationships from planktonic foraminifera and speleothems in the Eastern Mediterranean region and their implication for paleorainfall during interglacial intervals.” Geochimica et Cosmochimica Acta 67: 3181–3199. Bar-Matthews, M., A. Ayalon, A. Kaufman, and G.J. Wasserburg. 1999. “The eastern Mediterranean paleoclimate as a reflection of regional events: Soreq cave, Israel.” Earth and Planetary Science Letters 166: 85–95. Bard, E., G.M. Raisbeck, F. Yiou, and J. Jouzel. 1997. “Solar modulation of cosmogenic nuclide production over the last millennium: comparison between 14C and 10Be records.” Earth and Planetary Science Letters 150: 453–462. Barker, G. 2002. “A Tale of Two Deserts: Contrasting Desertification Histories on Rome’s Desert Frontiers.” World Archaeology 33: 488–507. Barker, G.W.W., D.D. Gilbertson, C.M. Griffin, P.P. Hayes, and D.A. Jones. 1983. “The UNESCO Libyan Valleys Survey V: sedimentological properties of Holocene wadi floor and plateau deposits in Tripolitania, north-west Libya.” LibStud 14: 69– 85. Barker, G.W.W., D.D. Gilbertson, and D.J. Mattingly. 2007. Archaeology and desertification. The Wadi Faynan Landscape Survey, Southern Jordan. Oxford. Beckh, H. 1895. Geoponica sive Cassiani Bassi scholastici De re rustica eclogae. Leip- zig. 280 bibliography

Beer, J. 2000. “Long-term indirect indices of solar variability.” Space Science Reviews 94: 53–66. Beer, J., A. Blinov, G. Bonani, R.C. Finkel, H.J. Hofmann, B. Lehmann, H. Oeschger, A. Sigg, J. Schwander, T. Staffelbach, B. Stauffer, M. Suter, and W. Wölfli. 1990. “Use of 10Be in polar ice to trace the 11-year cycle of solar activity.” Nature 347: 164– 166. Beer, J., U. Siegenthaler, G. Bonani, R.C. Finkel, H. Oescheger, M. Suter, and W. Wölfli. 1988. “Information on past solar activity and geomagnetism from 10Be in the Camp Century ice core.” Nature 331: 675–679. Beer, J., S. Tobias, and N. Weiss. 1998. “An Active Sun Throughout the Maunder Minimum.” Solar Physics 181: 237–249. Beer, J., M. Vonmoos, and R. Muscheler. 2006. “Solar variability over the past several millennia.” Space Science Reviews 125: 67–79. Behre, K.-E. 1992. “The History of Rye Cultivation in Europe.” Vegetation History and Archaeobotany 1: 141–156. Belati, M. 1999. “Le inondazioni del Tevere nell’antichità.” Forma Urbis 4.7/8: 12–17. Bellotti, P., C. Caputo, L. Davoli, S. Evangelista, E. Garzanti, F. Pugliese, and P. Valeri. 2004. “Morpho-sedimentary characteristics and Holocene evolution of the emer- gent part of the Ombrone River delta (southern Tuscany).” Geomorphology 61(1– 2): 71–90. Beltrán Lloris, F. 2006. “An Irrigation Decree from Roman Spain: the Lex Rivi Hibe- riensis.” JRS 96: 147–197. Berggren, A.-M., J. Beer, G. Possnert, A. Aldahan, P. Kubik, M. Christl, S.J. Johnsen, J. Abreu, and B.M. Vinther. 2009. “A 600-year annual 10Be record from the NGRIP ice core, Greenland.” Geophysical Research Letters 36: L11801. Bernand, E. 1993. “Une figure de la papyrologie: Danielle Bonneau.” ZPE 98: 97–102. Bersani, P., and M. Bencivenga. 2001. Le piene del Tevere a Roma dal V secolo a.C. all’anno 2000. Presidenza del Consiglio dei Ministri, Dipartimento per i Servizi Tecnici Nazionali, Servizio Idrografico, Mareografico Nazionale. Betancourt, J.L., H.D. Grissino-Mayer, M.W. Salzer, and T.W. Swetnam. 2002. “A Test of “Annual Resolution” in Stalagmites Using Tree Rings.” Quaternary Research 58: 197–199. Beven, K. 2003. “Surface runoff generation. Handbook of weather, climate, and water: atmospheric chemistry, hydrology, and societal impacts.” In Handbook of weather, climate and water: atmospheric chemistry, hydrology, and societal impacts, edited by T.D. Potter and B.R. Colman, 527–542. Hoboken, N.J. Bintliff, J. 2002. “Time, process and catastrophism in the study of Mediterranean alluvial history: a review.” World Archaeology 33 (3): 417–435. Blaauw, M. 2010. “Methods and code for ‘classical’ age-modelling of radiocarbon sequences.” Quaternary Geochronology 5: 512–518. Blaauw, M., B. van Geel, I. Kristen, B. Plessen, A. Lyaruu, D.R. Engstrom, J. van der Plicht, and D. Verschuren. 2011. “High-resolution 14C dating of a 25,000-year lake-sediment record from equatorial East Africa.” Quaternary Science Reviews 30: 3043–3059. Blaauw, M., B. van Geel, and J. van der Plicht. 2004. “Solar forcing of climatic change during the mid-Holocene: indications from raised bogs in The Netherlands.” The Holocene 14: 35–44. bibliography 281

Black, E. 2011. “The influence of the North Atlantic Oscillation and European circulation regimes on the daily to interannual variability of winter precipitation in Israel.” International Journal of Climatology 32(11): 1654–1664. Black, E., D. Brayshaw, S. Black, and C. Rambeau. 2011. “Using proxy data, historical climate data and climate models to investigate aridification during the Holocene.” In Water, Life and Civilisation: Climate, Environment and Society in the Jordan Valley, edited by S. Mithen and E. Black, 105–112. Cambridge. Blackman, D.J. 1973. “Evidence of Sea Level Change in Ancient Harbours and Coastal Installations.” In Marine archaeology (Colston Symposium Papers 23), edited by D.J. Blackman, 115–139. London. Blender, R., K. Fraedrich, and B. Hunt. 2006. “Millennial Climate Variability: GCM- Simulation and Greenland Ice Cores.” Geophysical Research Letters 33: 2–4. Blockley, S.P.E., C.B. Ramsey, C.S. Lane, and A.F. Lotter. 2008. “Improved age modelling approaches as exemplified by the Revised Chronology for the Central Euro- pean Varved Lake, Soppensee.” Quaternary Science Reviews 27: 61–67. Boardman, J. 2011. “Fish and the Mediterranean: The Nourishing Sea.” Ancient West and East 10: 1–9. Bonardi, G., B. D’Argenio, V. Perrone. 1988. Carta Geologica dell’Appennino meridionale in scala 1: 250.000. 74° Congresso della Siocietà Geologica Italiana, Sorrento, 13–17 settembre 1988. Memorie della Società Geologica, 41. Rome. Bond, G., B. Kromer, J. Beer, R. Muscheler, M. Evans, W. Showers, S. Hoffman, R. Lotti-Bond, I. Hajdas, and G. Bonani. 2001. “Persistent Solar Influence on North Atlantic Surface Circulation During the Holocene.” Science 294: 2130–2136. Bond, G.C., and R. Lotti. 1995. “Iceberg Discharges into the North Atlantic on Mil- lennial Time Scales During the Last Glaciation.” Science 267: 1005–1010. Bond, G., W. Showers, M. Cheseby, R. Lotti, P. Almasi, P. deMenocal, P. Priore, H. Cullen, I. Hajdas, and G. Bonani. 1997. “A pervasive millennial-scale cycle in North Atlantic Holocene and glacial climates.” Science 278: 1257–1266. Bonneau, D. 1964. La crue du Nil, divinité égyptienne, à travers mille ans d’histoire (332 av.-641 ap. J.-C.) d’après les auteurs grecs et latins, et les documents des époques ptolémaïque, romaine et byzantine. Études et commentaires, 52. Paris. ———1966. “Utilisation des documents papyrologiques, numismatiques et épi- graphiques pour la détermination de la qualité de la crue du Nil, chaque année de l’époque gréco-romaine.” Atti dell’XI Congresso internazionale di papirologia, Milano 2–8 settembre 1965, 1: 379–395. Milan. ———1971. Le fisc et le Nil. Incidences des irrégularités de la crue du Nil sur la fiscalité foncière dans l’Égypte grecque et romaine. Paris. Bookman (Ken-Tor) R., Y. Enzel, A. Agnon, and M. Stein. 2004. “Late Holocene lake levels of the Dead Sea.” Geological Society of America Bulletin 116: 555–571. Bootle, K.R. 2004. Wood in Australia. 2nd ed. Sydney. Boswijk, G., A. Fowler, A. Lorrey, J. Palmer, and J. Ogden. 2006. “Extension of the New Zealand kauri (Agathis australis) chronology to 1724BC.” The Holocene 16: 1–12. Bottema, S., G. Entjes-Nieborg, and W. van Ziest, eds. 1990. Man’s role in the shaping of the Eastern Mediterranean landscape. Rotterdam. Bourillon, J. 2005. Etude paléoenvironnementale du port antique de Naples: le site de Piazza Municipio. Mémoire de Master 2 Géographie Physique, Institut de Géographie, Université de Provence, France. 282 bibliography

Bradley, R.S., K.R. Briffa, J. Cole, M.K. Hughes, and T.J. Osborn. 2003. “The climate of the last millennium.” In Paleoclimate, Global Change and the Future, edited by K.D. Alverson, R.S. Bradley and T.F. Pedersen, 105–149. Berlin. Bradley, R.S., and P.D. Jones. 1993. “‘Little Ice Age’ summer temperature variations: Their nature and relevance to recent global warming trends.” The Holocene 3: 367–376. Brázdil, R., C. Pfister, H. Wanner, H. von Storch, and J. Luterbacher. 2005. “Historical climatology in Europe—The state of the art.” Climatic Change 70: 363–430. Bresson, A. “Ecology and Beyond: the Mediterranean Paradigm.” In Harris 2005a, 94–114. ———2006. “La machine d’Héron et le coût de l’énergie dans le monde antique.” In Innovazione tecnica e progresso economico nel mondo romano, edited by E. Lo Cascio, 55–80. Bari. Briffa, K.R. 2000. “Annual climate variability in the Holocene: interpreting the mes- sage of ancient trees.” Quaternary Science Reviews 19: 87–105. Briffa, K.R., T.J. Osborn, F.H. Schweingruber, I.C. Harris, P.D. Jones, S.G. Shiyatov, and E.A. Vaganov. 2001. “Low-frequency temperature variations from a northern tree ring density network.” Journal of Geophysical Research 106: 2929–2941. Bronk Ramsey, C. 2008. “Deposition models for chronological records.” Quaternary Science Reviews 27: 42–60. Bronk Ramsey, C., J. van der Plicht, and B. Weninger. 2001. “‘Wiggle matching’ radiocarbon dates.” Radiocarbon 43: 381–389. Broodbank, C. 2000. An Island Archaeology of the Early Cyclades. Cambridge. ———2011. “The Mediterranean and the Mediterranean World in the Age of Andrew Sherratt.” In Interweaving Worlds: Systemic Interactions in Eurasia, 7th to 1st Mil- lennia bc, edited by T.C. Wilkinson, S. Sherratt and J. Bennet, 27–36. Oxford. Brooks, N. 2006. “Cultural responses to aridity in the Middle Holocene and increased social complexity.” Quaternary International 151: 29–49. Brown, P. 1971. The world of late antiquity: from Marcus Aurelius to Muhammad. London. Brown, W. 1820. Antiquities of the Jews, Carefully Compiled from Authentic Sources and Their Customs Illustrated from Modern Travels, Vol. 2. London. Bruins, H.J. 1986. Desert Environment and Agriculture in the Central Negeb and Kadesh-Barnea during Historical Times. Ph.D. dissertation, The Agricultural Uni- versity of Wageningen. Nijkerk. ———1994. “Comparative chronology of climate and human history in the southern Levant from the Late Chalcolithic to Early Arab Period.” In Late Quaternary Chronology and Paleoclimates of the Eastern Mediterranean, edited by O. Bar- Yosef and R.S. Kra, 301–314. Tucson. Brun, J-P. 2006. “L’énergie hydraulique durant l’empire Romain: quel impact sur l’économie agricole?” In Innovazione tecnica e progresso economico nel mondo romano, edited by E. Lo Cascio, 101–131. Bari. Brunt, P.A. 1971. Italian Manpower, 225BC–14AD. Oxford. ———1980. “Free Labour and Public Works at Rome.” JRS 70: 81–100. Bruun, C.F.M. 2000. “The Roman World.” In Handbook of Ancient Water Technology, edited by O. Wikander, 675–604. Leiden. ———2003. “The Antonine plague in Rome and Ostia.” JRA 16: 426–434. bibliography 283

———2007. “The Antonine Plague and the ‘Third-Century Crisis.’” In Crises and the Roman Empire, (Impact of Empire 7), edited by O. Hekster, G. de Kleijn, and D. Slootjes, 201–217. Leiden. Buckley, B.M., K.J. Anchukaitis, D. Penny, R. Fletcher, E.R. Cook, M. Sano, L.C. Nam, A. Wichienkeeo, T.T. Minh, T.M. Hong. 2010. “Climate as a contributing factor in the demise of Angkor, Cambodia.” Proceedings of the National Academy of Sciences 107: 6748–6752. Buisman, J., and A.F.V. van Engelen, ed. 1995–. Duizend jaar weer, wind en water in de Lage Landen. Franeker. Bunn, S.E., and A.H. Arthington. 2002. “Basic principles and ecological consequences of altered flow regimes for aquatic biodiversity.” Environmental Man- agement 30: 492–507. Büntgen, U., D.C. Frank, D. Nievergelt, and J. Esper. 2006. “Summer temperature variations in the European Alps, AD 755–2004.” Journal of Climate 19: 5606–5623. Büntgen, U., W. Tegel, K. Nicolussi, M. McCormick, D. Frank, V. Trouet, J.O. Kaplan, F. Herzig, K.-U. Heussner, H. Wanner, J. Luterbacher, and J. Esper. 2011. “2500 Years of European Climate Variability and Human Susceptibility.” Science 331: 578–582. Büntgen, U., V. Trouet, D. Frank, H.H. Leuschner, D. Friedrichs, J. Luterbacher, and J. Esper. 2010. “Tree-ring indicators of German summer drought over the last millennium.” Quaternary Science Reviews 29: 1005–1016. Burton, R.F. 1875. “Notes on Rome.” Macmillan’s Magazine 31: 126–134. Butzer, K.W.2003. Review of A. Grove and O. Rackham, The Nature of Mediterranean Europe. Annals of the Association of American Geographers 93: 494–498. ———2005. “Environmental history of the Mediterranean world: cross-disciplinary investigation of cause-and-effect for degradation and soil erosion.” Journal of Archaeological Science 32: 1773–1800. ———2011. “Geoarchaeology, climate and sustainability: a Mediterranean Perspec- tive.” In Geoarchaeology. Climate and Sustainability, Special Paper 476, edited by A.G. Brown, L.S. Basell, and K.W. Butzer, 1–12. Denver. ———2012. “Collapse, Environment, and Society.” Proceedings, National Academy of Sciences USA 109: 3632–3639. Butzer, K.W., and S.E. Harris. 2007. “Geoarchaeological Approaches to the Environ- mental History of Cyprus: Explication and Critical Evaluation.” Journal of Archae- ological Science 34: 1932–1952. Butzer, K.W. and D.M. Helgren. 2005. “Livestock, land cover and environmental history: the Tablelands of New South Wales, Australia, 1820–1920.” Annals, Associa- tion of American Geographers 95: 80–111. Cacho, I., J.O. Grimalt, M. Canals, L. Sbaffi, N.J. Shackleton, J. Schönfeld, and R. Zahn. 2001. “Variability of the western Mediterranean Sea surface temperature during the last 25,000 years and its connection with the Northern Hemisphere climatic changes.” Paleoceanography 16: 40–52. Cadell, H. 1976. “P. Genève 60, B.G.U. II 456 et le problème du bois en Egypte.” Chronique d’Égypte 51: 331–348. Calenda, G., C.P. Mancini, and E. Volpi. 2009. “Selection of the probabilistic model of extreme floods: The case of the River Tiber in Rome.” Journal of Hydrology 371(1–4): 1–11. Cameron, A. 1993. The Mediterranean world in late antiquity, AD 395–600. London. 284 bibliography

Camp, J. McK. 1979. “A Drought in the Late Eighth Century B.C.” Hesperia 48: 397–411. Campbell, J.B. 2012. Rivers and the power of ancient Rome. Chapel Hill, NC. Camuffo, D., and S. Enzi. 1995. “Climatic features during the Sporer and Maunder Minima.” Paläoklimaforschung—Palaeoclimate Research 16: 105–124. ———1996. “The analysis of two bi-millenial series: Tiber and Po river floods.” In Climatic variations and forcing mechanisms of the last 2000 years, edited by P.D. Jones, R.S. Bradley and J. Jouzel, 432–450. Berlin; Heidelberg. Cane, M.A. 2005. “The evolution of El Niño, past and future.” Earth and Planetary Sciences Letters 230: 227–240. Cancellieri, C. 1986. “L’acquedotto Paolo (sec. XVII–XX).” In Il trionfo dell’acqua: Acque e acquedotti a Roma IV Sec. a.C.–XX Sec., edited by G. Pisani Sartorio and A. Liberati Silverio, 225–231. Rome. Carmichael, G.A. 1993. “St. Anthony’s Fire.” The Cambridge World History of Human Disease. Cambridge. Carriòn, Y. 2006. “Tres Montes (Navarra, Spain): Dendrology and wood uses in an arid environment.” In Charcoal Analysis: New Analytical Tools and Methods for Archaeology. Papers from the Table-Ronde held in Basel 2004, edited by A. Dufraisse, 83–94. Oxford. Carsana, V., S. Febbraro, D. Giampaola, C. Guastaferro, G. Irollo, and M.R. Ruello. 2009. “Evoluzione del paesaggio costiero tra Parthenope e Neapolis.” Méditerranée 112: 15–22. Casana, J. 2008. “Mediterranean valleys revisited: Linking soil erosion, land use and climate variability in the Northern Levant.” Geomorphology 101: 429–442. Caseldine, C.J., and C. Turney. 2010. “The bigger picture: towards integrating palaeoclimate and environmental data with a history of societal change.” Journal of Quaternary Science 25: 88–93. Cavarzere, A., ed. 2003. Decimus Magnus Ausonius: Mosella. Introduzione, testo, traduzione e commento. Amsterdam. Cawkwell, G.L. 1992. “Early Colonisation.” Classical Quarterly 42: 289–303. Certini, G., and R. Scalenghe. 2011. “Anthropogenic soils are the golden spikes for the Anthropocene.” The Holocene 21: 1269–1274. Chabal, L., L. Fabre, J.-F. Terral, and I. Théry-Parisot. 1999. “L’Anthracologie.” In La Botanique, edited by C. Bourquin-Mignot, J-E. Brochier, L. Chabal, S. Crozat, L. Fabre, F. Guibal, P.Marinval, H. Richard, J-F. Terral, and I. Théry-Parisot, 43–104. Paris. Chakrabarty, D. 2009. “The Climate of History: Four Theses.” Critical Inquiry 35: 197–222. Chambers, F.M., D. Mauquoy, S.A. Brain, M. Blaauw, and J.R.G. Daniell. 2007. “Glob- ally synchronous climate change 2800 years ago: proxy data from peat in South America.” Earth and Planetary Science Letters 253: 439–444. Chen, D., M. Cane, A. Kaplan, S. Zebiak, and D. Huang. 2004. “Predictability of El Niño over the past 148 years.” Nature 428: 733–736. Chen, Z., S.E. Grasby, and K.G. Osadetz. 2004. “Relation between climate variability and groundwater levels in the upper carbonate aquifer, southern Manitoba, Canada.” Journal of Hydrology 290(1–2): 43–62. Cheng, H., D. Fleitmann, R.L. Edwards, X.F. Wang, F.W. Cruz, A.S. Auler, A. Mangini, Y.J. Wang, S.J. Burns, and A. Matter. 2009. “Timing and structure of the 8.2 ky event bibliography 285

inferred from 18O records of stalagmites from China, Oman and Brazil.” Geology 37: 1007–1010. Chew, S.C. 2002. “Globalisation, ecological crisis, and Dark Ages.” Global Society 16: 333–356. Chierici, R. 1911. I boschi nell’economia generale d’Italia. Loro stima. Caserta. Christiansen, B., and F.C. Ljungqvist. 2011. “Reconstruction of the Extratropical NH Mean Temperature over the Last Millennium with a Method that Preserves Low-Frequency Variability.” Journal of Climate 24: 6013–6034. Chrzavzez, J., I. Théry-Parisot, J.-F. Terral, A. Ducom, and G. Fiorucci. 2011. “Differen- tial preservation of anthracological material and mechanical properties of wood charcoal, an experimental approach of fragmentation.” In 5th International Meet- ing of Charcoal Analysis. The charcoal as cultural and biological heritage, edited by E. Badal, Y. Carriòn, E. Grau, M. Macías, and M. Ntinou, 29–30. València. Clausen, H.B., C.U. Hammer, C.S. Hvidberg, D. Dahl-Jensen, J.P. Steffensen, J. Kipfs- tuhl, and M. Legrand. 1997. “A comparison of the volcanic records over the past 4000 years from the Greenland Ice Core Project and Dye 3 Greenland ice cores.” Journal of Geophysical Research 102 C12: 26,707–726,723. Cline, W.R. 2007. Global Warming and Agriculture: Impact Estimates by Country. Washington, D.C. Coarelli, F. 1997. Il Campo Marzio: dalle origini alla fine della repubblica. Rome. Coates-Stephens, R. 2003. “The Water Supply of Rome from Late Antiquity to the Early Middle Ages.” Acta ad Archaeologiam et Artium Historiam Pertinentia 17: 165–186. Coldstream, J.N. 1977. Geometric Greece. London. Collaborative Partnership on Forests. 2011. “Energy grows on trees: Global forest institutions highlight vital role of forests to cope with future demands for bioenergy, but call for balanced approaches.” International Union of Forest Research Organisations (IUFRO), United Nations Forum on Forests Secretariat, Interna- tional Year of Forests New Releases, 8 November, 2011. Available at: http://www .cpfweb.org/75548/en/. Conedera, M., P.Krebs, W.Tinner, M. Pradella, and D. Torriani. 2004. “The cultivation of Castanea sativa (Mill.) in Europe, from its origin to its diffusion on a continental scale.” Vegetation History and Archaeobotany 13: 161–179. Cook, E. 1971. “The Flow of Energy in an Industrial Society.” Scientific American 225, 3: 135–142. ———1976. Man, Energy, Society. San Francisco. Cook, E.R., K.J. Anchukaitis, B.M. Buckley, R.D. D’Arrigo, G.C. Jacoby, and W.E. Wright. 2010. “Asian monsoon failure and megadrought during the last millennium.” Science 328(5977): 486–489. Cook, E.R., R. D’Arrigo, and M.E. Mann. 2002. “A Well-Verified, Multiproxy Recon- struction of the Winter North Atlantic Oscillation Index since A.D. 1400.” Journal of Climate 15: 1754–1764. Cook, E.R., and R.L. Holmes. 1999. USERS MANUAL for Program ARSTAN. February 1999. Tucson, AZ. Cook, E.R., and L.A. Kairiukstis, eds. 1990. Methods of Dendrochronology: Applica- tions in the Environmental Sciences. Dordrecht. 286 bibliography

Cook, E.R., C.A. Woodhouse, C.M. Eakin, D.M. Meko, and D.W. Stahle. 2004. “Long- Term Aridity Changes in the Western United States.” Science 306: 1015–1018. Corella, J.P., A. Moreno, M. Morellón, V. Rull, S. Giralt, M.T. Rico, A. Pérez-Sanz, and L. Valero-Garcés. 2011. “Climate and human impact on a meromictic lake during the last 6,000 years (Montcortès Lake, Central Pyrenees, Spain).” Journal of Paleolimnology 46: 351–367. Corretti, A., and M. Firmati. “Metallurgia antica e medievale all’isola d’Elba: vecchi dati e nuove acquisizioni.” In Archeometallurgia: dalla conoscenza alla fruizione, edited by C. Giardino, 229–241. Bari. Coşkun, A. 2002a. Die gens Ausoniana an der Macht: Untersuchungen zu Decimius Magnus Ausonius und seiner Familie. Oxford. ———2002b. “Ein geheimnisvoller gallischer Beamter in Rom, ein Sommerfeldzug Valentinians und weitere Probleme in Ausonius’ Mosella.” Revue des études anci- ennes 104: 401–431. Cremaschi, M., M. Pelfini, and M. Santilli. 2006. “Cupressus dupreziana: a dendroclimatic record for the middle-late Holocene in the central Sahara.” The Holocene 16.2: 293–303. Crook, D. 2009. “Hydrology of the combination irrigation system in the Wadi Faynan, Jordan.” Journal of Archaeological Science 36: 2427–2436. Crow, J. 2012. “Ruling the waters: managing the water supply of Constantinople, AD 330–1204.” Water History 4.1: 35–55. Crow, J., J. Bardill, and R. Bayliss. 2008. The water supply of Byzantine Constantinople (Journal of Roman Studies Monograph). London. Crutzen, P.J., and E.F. Stoermer. 2000. “The ‘Anthropocene’.” Global Change Newslet- ter 41: 17–18. Cuffey, K.M., and G.D. Clow. 1997. “Temperature, accumulation, and ice sheet elevation in central Greenland through the last deglacial transition.” Journal of Geo- physical Research 102: 26383–26396. D’Agostino, B., and D. Giampaola. 2005. “Osservazioni storiche e archeologiche sulla fondazione di Neapolis.” In Noctes Campanae, studi di storia antica e archeologia dell’Italia pre-romana e romana in memoria di Martin W. Frederiksen, edited by W.V. Harris and E. Lo Cascio, 63–72. Naples. D’Arms, J.H. 1974. “Puteoli in the Second Century of the Roman Empire: A Social and Economic Study.” JRS 64: 104–124. ———2000. “Memory, Money and Status at Misenum: Three New Inscriptions from the Collegium of Augustales.” JRS 90: 126–144. Dai, A.G., and T.M.L. Wigley. 2000. “Global patterns of ENSO-induced precipitation.” Geophysical Research Letters 27: 1283–1286. Dalby, A., trans. 2011. Geoponika. Farm Work: a modern translation of the Roman and Byzantine farming handbook. Totnes, Devon. Dalfes, H.N., G. Kukla, and H. Weiss, eds. 1997. Third Millennium BC Climate Change and Old World Collapse. NATO ASI Series I, vol. 49. Berlin. Danti, M.D. “Late Middle Holocene Climate amd Northern Mesopotamia: Varying Cultural Responses to the 5.2 and 4.2 ka Aridification Events.” In Mainwaring, Giegengack, and Vita-Finzi 2010, 139–172. Dark, P. 2000. The environment of Britain in the first millennium AD. London. bibliography 287 de Callataÿ, F. 2005. “The Graeco-Roman economy in the super long-run: lead, copper and shipwrecks.” JRA 18: 361–372. De Caro, S. 1999. “L’attività della Soprintendenza archeologica di Napoli e Caserta nel 1998.” In L’Italiameridionaleinetàtardoantica.Attideltrentottesimoconvegno di studi sulla Magna Grecia. Taranto 2–6 ottobre 1998, edited by A. Stazio and S. Ceccoli, 635–661. Taranto. ———2003. “L’attività archeologica a Napoli e Caserta nel 2002.” In Ambiente e paesaggio nella Magna Grecia. Atti del quarantaduesimo convegno di studi sulla Magna Grecia. Taranto 5–8 ottobre 2002, 2: 569–621. Taranto. De Caro, S., and C. Gialanella, eds. 2002. Il Rione Terra di Pozzuoli. Naples. de Criscio, G. 1881. Notizie istoriche archeologiche topografiche dell’antica citta di Pozzuoli e dei suoi due acquidotti Serino e Campano. Naples. de la Ville de Mirmont, H. 1889. La Mosella d’Ausone. Édition critique et traduction française précédées d’une introduction, suivies d’un commentaire explicatif. Bor- deaux. De Nino, A. 1900. “Pentima—Indagini circa il percorso dell’antico acquedotto corfi- niese.” NSA: 642–643. De Vivo, B., P. Petrosino, A. Lima, G. Rolandi, and H.E. Belkin. 2010. “Research progress in volcanology in the Neapolitan area, southern Italy: a review and some alternative views.” Mineralogy and Petrology 99: 1–28. De Vleeschouwer, F., L. Gérard, C. Goormaghtigh, N. Mattielli, G. Le Roux, and N. Fagel. 2007. “Atmospheric lead and heavy metal pollution records from a Belgian peat bog spanning the last two millennia: Human impact on a regional to global scale.” Science of the Total Environment 377: 282–295. Debussche, M., J. Lepart, and A. Dervieux. 1999. “Mediterranean landscape changes: evidence from old postcards.” Global Ecology and Biogeography 8: 3–15. Deino, A.L., G. Orsi, S. de Vita, and M. Piochi. 2004. “The age of the Neapolitan Yellow Tuff caldera—forming eruption (Campi Flegrei Caldera—Italy) assessed by 40Ar/39Ar dating method.” Journal of Volcanology and Geothermal Research 133: 157–170. DeLaine, J. 1996. “Hypocaust.” In The Oxford Classical Dictionary, 3rd edn, edited by S. Hornblower and A. Spawforth. Oxford. deMenocal, P.B. 2001. “Cultural Responses to Climate Change During the Late Holo- cene.” Science 292: 667–673. Denton, G., and W. Karlén. 1973. “Holocene climatic variations—their pattern and possible cause.” Quaternary Research 3(2): 155–205. Deser, C. 2000. “On the Teleconnectivity of the “Arctic Oscillation”.” Geophysical Research Letters 27: 779–782. Desnier, J.L. 1998. “Les débordements du Fleuve.” Latomus 57: 513–522. Desprat, S., M.F. Sánchez-Goñi, and M.F. Loutre. 2003. “Revealing climatic variability of the last three millennia in northwestern Iberia using pollen influx data.” Earth and Planetary Science Letters 213: 63–78. Devillers, B., and N. Lecuyer. 2008. “Le Petit Âge Glaciaire en milieu semi-aride: le bassin versant du Gialias (Chypre) et ses relations avec l’occupation des sols.” Zeitschrift für Geomorphologie 52(2): 207–224. Di Donato, V., P. Esposito, E. Russo Ermolli, A. Scarano, and R. Cheddadi. 2008. “Cou- pled atmospheric and marine palaeoclimatic reconstruction for the last 35 kyr 288 bibliography

in the Sele Plain-Gulf of Salerno area (southern Italy).” Quaternary International 190: 146–157. Di Vito, M.A., R. Isaia, G. Orsi, J. Southon, S. de Vita, M. D’Antonio, L. Pappalardo, and M. Piochi. 1999. “Volcanism and deformation since 12,000 years at the Campi Flegrei caldera (Italy).” Journal of Volcanology and Geothermal Research 91: 221– 246. Di Rita, F., and D. Magri. 2009. “Holocene Drought, Deforestation, and Evergreen Vegetation Development in the Central Mediterranean: a 5500 Year Record from Lago Alimini Piccolo, Apulia, Southeast Italy.” The Holocene 19: 295–306. Diaz, H.F., M.P. Hoerling, and J.K. Eischeid. 2001. “ENSO variability, teleconnections, and climate change.” International Journal of Climatology 21: 1845–1862. Diaz, H.F., and D.W. Stahle. 2007. “Climate and cultural history in the Americas: an overview.” Climatic Change 83: 1–8. Diaz, H.F., R. Trigo, M.K. Hughes, M.E. Mann, E. Xoplaki, and D. Barriopedro. 2011. “Spatial and Temporal Characteristics of Climate in Medieval Times Revisited.” Bulletin of the American Meteorological Society 92: 1487–1500. Dincauze, D.F. 2000. Environmental archaeology. Principles and practice. Cambridge. Döring, M. 2007. Römische Häfen, Aquädukte und Zisternen in Campanien. Bestands- aufnahme der antiken Wasserbauten. Darmstadt. Dragoni, W. 1998. “Some considerations on climatic changes, water resources and water need in the Italian region south of the 43°N.” In Water, Environment and Society in Times of Climatic Change, edited by S. Issar and N. Brown, 241–271. Dordrecht; Boston. Dragoni, W., and B.S. Sukhija. 2008. “Climate change and groundwater: a short review.” In Climate change and groundwater, edited by W. Dragoni and B.S. Sukhija, 1–12. London. Drake, B.L. 2012. “The influence of climatic change on the late Bronze Age Collapse and the Greek Dark Ages.” Journal of Archaeological Science 39: 1862–1870. Drew-Bear, M. 1995. “Le bois en Égypte d’après les papyrus d’époque romaine.” In L’arbre et le forêt: le bois dans l’antiquité, edited by J.-C. Béal, 3–9. Paris. Drinkwater, J.F. 1999. “Re-Dating Ausonius’ War Poetry.” AJPh 120: 443–452. Drysdale, R.N., G. Zanchetta, J.C. Hellstrom, R. Maas, A.E. Fallick, M. Pickett, I. Cartwright, and L. Piccini. 2006. “Late Holocene drought responsible for the collapse of Old World civilizations is recorded in an Italian cave flowstone.” Geology 34: 101–104. du Coudray la Blanchère, R.M. 1895. L’aménagement de l’eau et l’installation rurale dans l’Afrique ancienne. Paris. du Coudray la Blanchère, R.M., and P. Gauckler. 1897. Catalogue du Musée Alaoui (Musées et collections archéologiques de l’Algérie et de la Tunisie). Paris. Dubois, C. 1907. Pouzzoles antique. Paris. Dull, R., J.R. Southon, S. Kutterolf, A. Freundt, D. Wahl, and P.D. Sheets. 2010. “Did the TBJ Ilopango eruption cause the AD 536 event?” In American Geophysical Union, Fall Meeting 2010, abstract #V13C-2370. Washington. Dull, R.A., J.R. Southon, and S. Payson. 2001. “Volcanism, Ecology and Culture: A Reassessment of the Volcán Ilopango Tbj eruption in the Southern Maya Realm.” Latin American Antiquity 12.1: 25–44. Duncan-Jones, R.P. 1996. “The impact of the Antonine plague.” JRA 9: 108–136. bibliography 289

Eastwood, W.J., N. Roberts, and H.F. Lamb. 1998. “Palaeoecological and archaeological evidence for human occupance in southwest Turkey: the Beyşhir Occupation Phase.” AS 48: 69–86. Eddy, J.A. 1976. “The Maunder Minimum.” Science 192: 1189–202. ———1977. “Climate and the changing sun.” Climate Change 1: 173–190. Ehrmann, W., G. Schmiedl, Y. Hamann, T. Kuhnt, C. Hemleben, and W. Siebel. 2007. “Clay minerals in late glacial and Holocene sediments of the northern and southern Aegean Sea.” Palaeogeography, Palaeoclimatology, Palaeoecology 249: 36–57. Emeis, K.C., U. Struck, H.M. Schulz, R. Rosenberg, S. Bernasconi, H. Erlenkeuser, T. Sakamoto, and F. Martinez-Ruiz. 2000. “Temperature and salinity variations of Mediterranean Sea surface waters over the last 16,000 years from records of planktonic stable oxygen isotopes and alkenone unsaturation ratios.” Palaeo- geography, Palaeoclimatology, Palaeoecology 158: 259–280. English Heritage. 2000. Thatch and thatching: a guidance note. http://www.english -heritage.org.uk/publications/thatch-and-thatching/. Enzel, Y., R. Bookman, D. Sharon, H. Gvirtzman, U. Dayan, B. Ziv, and M. Stein. 2003. “Late Holocene climates of the Near East deduced from Dead Sea level variations and modern regional winter rainfall.” Quaternary Research 60: 263–273. Ermolli, E.R., and G. Di Pasquale. 2002. “Vegetation Dynamics of South-western Italy in the Last 28kyr Inferred from Pollen Analysis of a Tyrrhenian Sea Core.” Vegetation History and Archaeobotany 11: 211–219. Esper, J., E.R. Cook, F.H. Schweingruber. 2002. “Low-Frequency Signals in Long Tree-Ring Chronologies for Reconstructing Past Temperature Variability.” Science 295: 2250–2253. Esper, J., E.R. Cook, P.J. Krusic, K. Peters, and F.H. Schweingruber. 2003. “Tests of the RCS method for preserving low-frequency variability in long tree-ring chronologies.” Tree Ring Research 59: 81–98. Esper, J., D.C. Frank, U. Büntgen, A. Verstege, J. Luterbacher, and E. Xoplaki. 2007. “Long-term drought severity variations in Morocco.” Geophysical Research Letters 34. Espérandieu, E. 1926. Le Pont du Gard et l’Aqueduc de Nîmes. Paris. Evans, H.B. 2002. Aqueduct Hunting in the Seventeenth Century: Rafaello Fabretti’s De aquis et aquaeductibus veteris Romae. Ann Arbor, MI. Evenari, M., L. Shanan, and T. Naphtali. 1971. The Negev: The Challenge of a Desert. Cambridge, MA. Fabre, G., J.-L. Fiches, and J.-L. Paillet. 1991. “Interdisciplinary research on the aqueduct of Nimes and the Pont du Gard.” JRA 4: 63–88. Fagan, B.M. 2000. The little ice age: how climate made history 1300–1850. New York. Falconer, S., and C.L. Redman, eds. 2009. Polities and Power: Archaeological Perspec- tives on the Landscapes of Early States. Tucson, AZ. Fall, P.L., S.E. Falconer, L. Lines, and M.C. Metzger. 2004. “Environmental Impacts of the Rise of Civilization in the Southern Levant.” In The Archaeology of Global Change: the Impact of Humans on their Environment, edited by C.L. Redman, S.R. James, P.R. Fish, and J.D. Rogers, 141–157. Washington. Fentress, E.W.B., and A.I. Wilson. Forthcoming. “The Saharan Berber Diaspora and the Southern Frontiers of Vandal and Byzantine North Africa.” In Rome Re- 290 bibliography

imagined: Byzantine and Early Islamic Africa, AD 500–800, (Dumbarton Oaks Byzantine Symposia and Colloquia), edited by J. Conant and S. Stevens. Wash- ington. Fernandes S.D., N.M. Trautmann, D.G. Streets, C.A. Roden, T.C. Bond. 2007. “Global Biofuel Use, 1850–2000.” Global Biogeochemical Cycles 21: 1–15. Ferrari Fontana, I., B.I. Menozzi, and C. Montanari. 2008. “Charcoal as Environmen- tal and Ethnological Evidence from Medieval Archaeological Sites in NW Italy.” In Fiorentino and Magri 2008, 105–109. Ferrio, J.P., N. Alonso, J.B. López, J.L. Araus, and J. Voltas. 2006. “Carbon isotope composition of fossil charcoal reveals aridity changes in the NW Mediterranean Basin.” Global Change Biology 12: 1253–1266. Field, C.V., G.A. Schmidt, D. Koch, and C. Salyk. 2006. “Modeling production and climate-related impacts on 10Be concentration in ice-cores.” Journal of Geophysi- cal Research 111: D15107. Fiema, Z.T. 2006. “City and countryside in Byzantine Palestine. Prosperity in question.” In Settlements and Demography in the Near East in Late Antiquity. Proceed- ings of the colloquium, Matera 27–29 October 2005. Biblioteca di ‘Mediterraneo Antico’ 2, edited by A.S. Lewin and P. Pellegrini, 67–88. Pisa; Rome. ———2008. “The Concluding Remarks.” In Petra—The Mountain of Aaron I. The Church and the Chapel, edited by Z.T. Fiema and J. Frösén, 425–441. Helsinki. Finné, M., K. Holmgren, H.S. Sundqvist, E. Weiberg, and M. Lindblom. 2011. “Cli- mate in the Eastern Mediterranean and Adjacent Regions, during the Past 6000 Years—A Review.” Journal of Archaeological Science 38: 3153–3173. Fiorentino, G., and D. Magri, eds. 2008. Charcoals from the Past: Cultural and Palaeo- environmental Implications (BARIS 1807). Oxford. Fiorillo, F., and L. Esposito. 2006. “The Serino Spring Discharge: Analyses of an Ultra- Centenarian Hydrological Series, Southern Italy.” BALWOIS Conference on Water Observation and Information Systems For Decision Support, http://balwois.mpl .ird.fr/balwois/administration/full_paper/ffp-632.pdf. [But apparently no longer accessible as of June 9, 2013]. Fiorillo, F., L. Esposito, and F.M. Guadagno. 2007. “Analyses and forecast of water resources in an ultra-centenarian spring discharge series from Serino (Southern Italy).” Journal of Hydrology 336: 125–138. Fisher, C.T., J.B. Hill, and G.M. Feinman, eds. 2009. The Archaeology of Environmental Change: Socionatural Legacies of Degradation and Resilience. Tucson, AZ. Fleitmann, D., S.J. Burns, A. Mangini, M. Mudelsee, J. Kramers, I. Villa, U. Neff, A.A. Al-Subbary, A. Buettner, D. Hippler, and A. Matter. 2007. “Holocene ITCZ and Indian monsoon dynamics recorded in stalagmites from Oman and Yemen (Socotra).” Quaternary Science Reviews 26: 170–188. Fleitmann, D., S.J. Burns, M. Mudelsee, U. Neff, J. Kramers, A. Mangini, and A. Matter. 2003. “Holocene Forcing of the Indian Monsoon Recorded in a Stalagmite from Southern Oman.” Science 300: 1737–1739. Fleitmann, D., H. Cheng, S. Badertscher, R.L. Edwards, M. Mudelsee, O.M. Göktürk, A. Fankhauser, R. Pickering, C.C. Raible, A. Matter, J. Kramers, and O. Tüysüz. 2009. “Timing and climatic impact of Greenland interstadials recorded in stalagmites from northern Turkey.” Geophysical Research Letters 36: L19707. Forni, G., and A. Marcone, eds. 2002. Storia dell’agricoltura italiana, I. Firenze. bibliography 291

Food and Agricultural Organization of the United Nations, Land and Water development division, Crop Water Management, “Wheat, Water Supply and Crop Yield”: http://www.fao.org/landandwater/aglw/cropwater/wheat.stm#supply, consulted January 4, 2012. ———2011. Forestry. Key Facts. Global Forestry Resources Assessment. Food and Agriculture Organisation of the United Nations. Available from: http://www.fao .org/docrep/014/am859e/am859e08.pdf. Fowler, A.M. 2008. “ENSO history recorded in Agathis australis (kauri) tree rings. Part B: 423 years of ENSO robustness.” International Journal of Climatology 28: 21–35. Fowler, A.M., J. Palmer, J. Salinger, and J. Ogden. 2000. “Dendroclimatic interpretation of tree-rings in Agathis australis (kauri). 2. Evidence of a significant relationship with ENSO.” Journal of the Royal Society of New Zealand 30: 277–292. Fowler, A.M., G. Boswijk, J. Gergis, and A. Lorrey. 2008. “ENSO history recorded in Agathis australis (kauri) tree rings. Part A: kauri’s potential as an ENSO proxy.” International Journal of Climatology 28: 1–20. Franconi, T. forthcoming 2013. “The power of rivers in the world of ancient Rome.” JRA 23.2. Frank, D., J. Esper, and E.R. Cook. 2007. “Adjustment for proxy number and coherence in a large-scale temperature reconstruction.” Geophysical Research Letters 34: L16709. Frederiksen, M.W. 1984. Campania. London. Freiwan, M., and M. Kadioglu. 2008. “Spatial and temporal analysis of climatological data in Jordan.” International Journal of Climatology 28: 521–535. Friedrich, W.L., B. Kromer, M. Friedrich, J. Heinemeier, T. Pfeiffer, and S. Talamo. 2006. “Santorini eruption radiocarbon dated to 1627–1600B.C.” Science 312: 548. Frumkin, A. 1997. “The Holocene history of Dead Sea levels.” In The Dead Sea. The Lake and Its Setting. Oxford Monographs on Geology and Geophysics 36, edited by T. Niemi, Z. Ben-Avraham, and J.R. Gat, 237–248. New York; Oxford. Gales, B., A. Kander, P. Malanima, and M. Rubio. 2007. “North versus South: Energy Transition and Energy Intensity in Europe over 200 Years.” European Review of Economic History 11: 219–253. Gallant, T.W. 1991. Risk and Survival in Ancient Greece. Cambridge. García, M.J.G., M.B.R. Zapata, J.I. Santisteban, R. Mediavilla, E. López-Pamo, and C.J. Dabrio. 2007. “Late holocene environments in Las Tablas de Daimiel (south central Iberian peninsula, Spain).” Vegetation History and Archaeobotany 16: 241– 250. Gardiner, A. 1961. Egypt of the Pharaohs: an Introduction. Oxford. Garnsey, P. 1988. Famine and food supply in the Graeco-Roman world: responses to risk and crisis. Cambridge. Geiger, J. 2009. “Rome and Jerusalem: Public Building and the Economy. Herod and Augustus.” In Papers presented at the IJS conference, 21st–23rd June 2005, edited by D.M. Jacobson and N. Kokkinos, 157–169. Leiden. Gelabert, L.P., E. Asouti, and E.A. Martí. 2011. “The ethnoarchaeology of firewood management in the Fang Village of Equatorial Guinea, central Africa: Impli- cations for the interpretation of wood fuel remains from archaeological sites.” Journal of Anthropological Archaeology 30: 375–384. 292 bibliography

Gerasimidis, A. 2000. “Palynological Evidence for Human Influence on the Vege- tation of Mountain Regions in Northern Greece: the Case of Lailias, Serres.” In Halstead and Frederick 2000, 28–37. Giachi, G., S. Lazzeri, M. Mariotti Lippi, N. Macchioni, and S. Paci. 2003. “The Wood of “C” and “F” Roman Ships Found in the Ancient Harbour of Pisa (Tuscany, Italy): the Utilisation of Different Timbers and the Probable Geographical Area which Supplied them.” Journal of Cultural Heritage 4: 269–283. Gialanella, C. 1993a. “Il Rione Terra alla luce dei nuovi scavi archeologici. Rione Terra, Pozzuoli.” Bollettino di Archeologia 22: 84–91. ———1993b. “La Topografia di Puteoli.” In Puteoli, edited by F. Zevi, 73–98. Napoli. Giampaola, D. 2004. “Dagli studi di Bartolomeo Capasso agli scavi della Metropoli- tana: ricerche sulle mura di Napoli e sull’evoluzione del paesaggio costiero.” Napoli Nobilissima 5(I–II): 35–56. ———2011. Archeologia e opere pubbliche: lo scavo della Stazione Chiaia della Linea 6 della Metropolitana. Conferenza Museo Archeologico Nazionale di Napoli, 20 ottobre 2011. Giampaola, D., V. Carsana, G. Boetto, M. Bartolini, C. Capretti, G. Galotta, G. Giachi, N. Macchioni, M.P. Nugari, B. Pizzo. 2006. “La scoperta del porto di Neapolis: dalla ricostruzione topografica allo scavo e al recupero dei relitti.” Archeologia Marit- tima Mediterranea, International Journal on Underwater Archaeology 2, 47–91, Istituti Editoriali e Poligrafici Internazionali MMVI. Pisa; Rome. Giardina, A. 1981. “Allevamento ed economia della selva in Italia meridionale: tras- formazioni e continuità.” In Società romana e produzione schiavistica, edited by A. Giardina and A. Schiavone, I, 87–113. Rome; Bari. Giesecke T., W.O. van der Knaap, and F. Bittmann. 2010. “Towards quantitative palynology: using pollen accumulation rates and models of pollen dispersal.” Vegetation History and Archaeobotany 19: 243–245. Gilliam, J.F. 1961. “The plague under Marcus Aurelius.” AJPh 82.3: 225–251. Giraudi, C. 1998. “Late Pleistocene and Holocene lake-level variations in Fucino Lake (Abruzzo, central Italy) inferred from geological, archaeological and historical data.” In Palaeohydrology as Reflected in Lake-Level Changes as Climatic Evidence for Holocene Times, edited by S.P. Harrison, B. Frenzel, U. Huckried, and M. Weiss, 1–17. Stuttgart. ———2004. “Le oscillazioni di livello del Lago di Mezzano (Valentino-VT): vari- azioni climatiche e interventi antropici.” Il Quaternario 17: 221–230. ———2005. “Late-Holocene alluvial events in the Central Apennines, Italy.” The Holocene 15(5): 768–773. ———2011. “The sediments of the ‘Stagno di Maccarese’ marsh (Tiber river delta, central Italy): A late-Holocene record of natural and human-induced environmental changes.” The Holocene 22: 1461–1471. Giraudi, C., M. Magny, G. Zanchetta, and R.N. Drysdale. 2011. “The Holocene climatic evolution of Mediterranean Italy: A review of the continental geological data.” The Holocene 21(1): 105–115. Göktürk, O.M., D. Fleitmann, S. Badertscher, H. Cheng, R.L. Edwards, M. Leuen- berger, A. Fankhauser, O. Tüysüz, and J. Kramers. 2011. “Climate on the southern Black Sea coast during the Holocene: implications from the Sofular Cave record.” Quaternary Science Reviews 30: 2433–2445. bibliography 293

Goudie, A.S. 2006. The Human Impact on the Natural Environment. 6th edition. Oxford. Goudsblom, J. 1992. Fire and Civilization. London. Graser, E.R. 1959. “The Edict of Diocletian on Maximum Prices (Appendix).” In An Economic Survey of Ancient Rome. Rome and Italy of the Empire, Vol. 5, edited by T. Frank, 305–422. Paterson, NJ. Gräslund, B., and N. Price. 2012. “Twilight of the gods? The ‘dust veil event’ of AD 536 in critical perspective.” Antiquity 86.322: 428–443. Gray, L.J., J. Beer, M. Geller, J.D. Haigh, M. Lockwood, K. Matthes, U. Cubasch, D. Fleitmann, G. Harrison, L. Hood, J. Luterbacher, G.A. Meehl, D. Shindell, B. van Geel, and W. White. 2010. “Solar Influences on Climate.” Reviews of Geophysics 48: RG4001. Greco, G. 2003. “Le nuove ricerche nel quartiere meridionale.” In Velia, le nuove ricerche, Atti del convegno di studi, Napoli, 14 dicembre 2001. Quaderni del Centro Studi Magna Grecia 1, Pozzuoli. Green, R.P.H. 1978. “The Éminence grise of Ausonius’ Moselle.” Res publica litterarum 1: 89–94. ———1991. Decimus Magnus Ausonius: The works of Ausonius. Oxford. ———1997. “On a recent redating of Ausonius’ Moselle.” Historia 46: 214–226. Greenberg, J. 2003. “Plagued by doubt: reconsidering the impact of a mortality crisis in the 2nd c. A.D.” JRA 16: 413–425. Greene, K. 2000. “Technological Innovation and Economic Progress in the Ancient World: M.I. Finley reconsidered.” Economic History Review 53: 29–59. —1986. The Achaeology of the Roman Economy. Berkeley-Los Angeles. Griggs, C.G., A. DeGaetano, P.I. Kuniholm, and M.W. Newton. 2007. “A regional high-frequency reconstruction of May–June precipitation in the north Aegean from oak tree rings, A.D.1089–1989.” InternationalJournalofClimatology 27: 1075– 1089. Griggs, C.B., P.I. Kuniholm, M.W. Newton, J.D. Watkins, and S.W. Manning. 2009. “A 924-year Regional Oak Tree-ring Chronology for North Central Turkey.” In Tree-Rings, Kings and Old World Archaeology and Environment: Papers Presented in Honor of Peter Ian Kuniholm, edited by S.W. Manning and M.J. Bruce, 71–79. Oxford. Grinter, P. 2007. “Seeds of destruction: conflagration in the grain stores of Dichin.” In The transition to late antiquity: on the Danube and beyond, edited by A.G. Poulter, 281–285. Oxford. Grissino-Mayer, H. 1996. “A 2129-year reconstruction of precipitation for northwestern New Mexico, U.S.A.” In Tree Rings, Environment and Humanity, edited by J.S. Dean, D.M. Meko, and T.W. Swetnam, 191–204. Tucson, AZ. Data are archived at the World Data Center for Paleoclimatology, Boulder, CO. Grocock, C., and S. Grainger, eds and trans. 2006. “The Texts.” In Apicius. A Critical Edition with an Introduction and an English Translation of the Latin Recipe Text Apicius, 125–326. Blackawton, Devon. Grove, A.T. 2001. “The Little Ice Age and its geomorphological consequences in Mediterranean Europe.” Climatic Change 48: 121–136. Grove, A.T., and O. Rackham. 2001. The Nature of Mediterranean Europe: an Ecologi- cal History. New Haven; London. 294 bibliography

Grove, J.M. 1988. The Little Ice Age. London. Grudd, H., K.R. Briffa, W. Karlén, T.S. Bartholin, P.D. Jones, and B. Kromer. 2002. “A 7400-year tree-ring chronology in northern Swedish Lapland: natural climatic variability expressed on annual to millennial timescales.” The Holocene 12: 657– 665. Grün, R. 2001. “Trapped charge dating (ESR, TL, OSL).” In Handbook of Archaeologi- cal Sciences, edited by D.R. Brothwell and A.M. Pollard, 47–62. Chichester. Gsell, S. 1921. Histoire ancienne de l’Afrique du Nord, vol. 1. Paris. Guibal, F., and P. Pomey. 2009. “Ancient Shipwrecks, Naval Architecture and Den- drochronology in the Western Mediterranean.” In Between the Seas: Transfer and Exchange in Nautical Technology. Proceedings of the 11th International Sympo- sium on Boat and Ship Archaeology, edited by R. Bockius, 219–226. Mainz am Rhein. Guidoboni, E. 1989. I Terremoti prima del Mille in Italia e nell’area mediterranea (Storia, Archeologia, Sismologia). Bologna. Guidoboni, E., A. Comastri, and G. Traina. 1994. Catalogue of ancient earthquakes in the Mediterranean area up to the 10th century, vol. 1. Rome. Guiot, J., C. Corona, and ESCARSEL members. 2010. “Growing Season Temperatures in Europe and Climate Forcings Over the Past 1400 Years.” PLoS ONE 5(4): e9972. Gunn, J.D., ed. 2000. The years without summer. Tracing A.D. 536 and its aftermath. (BAR International Series 872). Oxford. Haas, J. 2006. Die Umweltkrise des 3. Jahrhundert n. Chr. im Nordwesten des Imperium Romanum: interdisziplinäre Studien zu einem Aspekt der allgemeinen Reichskrise im Bereich der beiden Germaniae sowie der Belgica und der Raetia. Geographica historica 22. Stuttgart. Haas, J.N., I. Richoz, W. Tinner, and L. Wick 1998. “Synchronous Holocene climatic oscillations recorded on the Swiss Plateau and at timberline in the Alps.” The Holocene 8: 301–309. Hägg, R., ed. 1983. The Greek Renaissance of the Eighth Century BC: tradition and innovation. Stockholm. Hall, I.R., G.G. Bianchi, and J.R. Evans. 2004. “Centennial to millennial scale Holo- cene climate-deep water linkage in the North Atlantic.” Quaternary Science Re- views 23: 1529–1536. Haller, J.S., Jr. 1993. “Ergotism.” The Cambridge World History of Human Disease. Cambridge. Halsall, G. 2007. Barbarian Migrations and the Roman West, 376–568. Cambridge. Halstead, P., and C. Frederick, eds. 2000. Landscape and Land Use in Postglacial Greece. Sheffield. Hammond, N. 2010. “Climate, Crisis, Collapse, and the Ancient Maya Civilization: an Enduring Debate.” In Mainwaring et al. 2010, 189–196. Harris, W.V. 2000. “Trade.” In Cambridge Ancient History XI (second ed.), 710–739. Cambridge. ———ed. 2005a. Rethinking the Mediterranean. Oxford. ———2005b. “The Mediterranean and Ancient History.” In Harris 2005a, 1–42. ———2011a. “Bois et déboisement dans la Méditerranée antique.” Annales HSS 66, 1: 105–140. ———2011b. “Plato and the Deforestation of Attica.” Athenaeum 99: 479–482. bibliography 295

———2012. “The Great Pestilence and the Complexities of the Antonine-Severan Economy.” In L’impatto della peste antonina, edited by E. Lo Cascio, 331–338. Bari. ———forthcoming. “The Indispensable Commodity: Notes on the Economy of Wood in the Roman Mediterranean.” Harris, W.V.,and K. Iara, eds. 2011. Maritime Technology in the Ancient Economy: Ship- Design and Navigation. Journal of Roman Archaeology, Supplementary Series Number 84. Portsmouth, R.I. Hart, S. 1986. “Some preliminary thoughts on settlement in southern Edom.” Levant 18: 51–58. Hass, H.C. 1996. “Northern Europe climate variations during late Holocene: evidence from marine Skagerrak.” Palaeogeography Palaeoclimatology Palaeoecology 123: 121–145. Hassan, F.A. 1981. “Historical Nile floods and their implications for climatic change.” Science 212: 1142–1144. ———2007. “Extreme Nile floods and famines in Medieval Egypt (AD 930–1500) and their climatic implications.” Quaternary International 173–174: 101–112. Haury,J., and G. Wirth, ed. 1963 Procopius,OperaOmnia (revised edn), 3 vols. Leipzig. Healy, J.F. 1978. Mining and Metallurgy in the Greek and Roman World. London. Heather, P. 1998. “Goths and Huns, c. 320–425.” In Cambridge Ancient History XIII, 487–515. Cambridge. ———2000. “The Western Empire, 425–476.” In Cambridge Ancient History XIV,1–32. Cambridge. Hegerl, G., J. Luterbacher, F.J. González-Rouco, S. Tett, T. Crowley, and E. Xoplaki. 2011. “Influence of human and natural forcing on European seasonal temperatures.” Nature Geoscience 4: 99–103. Heikkilä, U., J. Beer, J.A. Abreu, and F. Steinhilber. 2011. “On the Atmospheric Trans- port and Deposition of the Cosmogenic Radionuclides (10Be): A Review.” Space Science Reviews. Heim, C., N.R. Nowaczyk, J.F.W. Negendak, S.A.G. Leroy, and Z. Ben-Avraham. 1997. “Near East desertification: evidence from the Dead Sea.” Naturwissenschaften 84: 398–401. Henning, J. 1987. Südosteuropa zwischen Antike und Mittelalter: archäologische Bei- träge zur Landwirtschaft des 1. Jahrtausends u. Z. Schriften zur Ur- und Früh- geschichte 42. Berlin. ———2009. “Revolution or relapse? Technology, agriculture and early medieval archaeology in Germanic Central Europe.” In The Langobards before the Frankish conquest: an ethnographic perspective, edited by G. Ausenda, P. Delogu, and C. Wickham, 149–173. Woodbridge, UK. Herman, I.P. 2007. Physics of the Human Body. Berlin. Heurgon, J., ed. 1978. Varron, Économie rurale I. Paris. Hirschfeld, Y. 1992. The Judean Desert Monasteries in the Byzantine Period. New Haven. ———2004. “A climatic change in the Early Byzantine period? Some archaeological evidence.” Palestine Exploration Quarterly 136: 133–149. Hitchner, R.B. 1988. “The Kasserine archaeological survey, 1982–1986.” Antiquités africaines 24: 7–41. 296 bibliography

———1989. “The organisation of rural settlement in the Cillium-Thelepte region.” In L’Africa Romana 6. Atti del VI convegno di studio, Sassari, 16–18 dicembre 1988, vol. 1, edited by A. Mastino, 387–402. Sassari. ———1990. “The Kasserine archaeological survey—1987.” Antiquités africaines 26: 231–259. ———1992. “The Kasserine Archaeological Survey 1982–1985.” Africa. Fouilles, mon- uments et collections archéologiques en Tunisie 11–12: 158–198. ———1995a. “Historical text and archaeological context in Roman North Africa: the Albertini Tablets and the Kasserine Survey.” In Methods in the Mediterranean. Historical and archaeological views on texts and archaeology, edited by D.B. Small, 124–142. Leiden. ———1995b. “Irrigation, terraces, dams and aqueducts in the region of Cillium (mod. Kasserine). The role of water works in the development of a Roman-African town and its countryside.” In Productions et exportations africaines: actualités archéologiques en Afrique du Nord antique et mediévale. VI e colloque international sur l’histoire et l’archéologie de l’Afrique du Nord (PAU, octobre 1993–118e congrès), edited by P. Trousset, 143–157. Paris. Holzhauser, H., M. Magny, and H.J. Zumbühl. 2005. “Glacier and lake-level variations in west-central Europe over the last 3500 years.” The Holocene 15: 789– 801. Hong, S., J.P. Candelone, and C.F. Boutron. 1994a. “Greenland ice history of the pollution of the atmosphere of the northern hemisphere for lead during the last three millenia.” Analusis 22: M38–M40. Hong, S., J.P. Candelone, C.C. Patterson, and C.F. Boutron. 1994b. “Greenland ice evidence of hemispheric lead pollution two millennia ago by Greek and Roman civilizations.” Science 265: 1841–1843. Hong, S., J.P. Candelone, C.C. Patterson, and C.F. Boutron. 1996a. “History of ancient copper smelting pollution during Roman and medieval times recorded in Green- land ice.” Science 272: 246–249. Hong, S., J.P. Candelone, M. Soutif, and C.F. Boutron. 1996b. “A reconstruction of changes in copper production and copper emissions to the atmosphere during the past 7000 years.” The Science of Total Environment 188: 183–193. Horden, P., and N. Purcell. 2000. The Corrupting Sea: a Study of Mediterranean History. Oxford. ———“Four Years of Corruption: a Response to Critics.” In Harris 2005a, 348–375. Hostetter, E., B.W. Fouke, and C.C. Lundstrom. Forthcoming. “The Last Flow of Water to the Baths of Caracalla: Age, Temperature and Chemistry.” Acts of the Terme di Caracalla conference, 1–2 March 2007, Soprintendenza Archeologica di Roma/Istituto Storico Austriaco. Hoyt, D.V., and K.H. Schatten. 1998a. “Group sunspot numbers: a new solar activity indicator. Part 1.” Solar Physics 181: 189–219. ———1998b. “Group sunspot numbers: a new solar activity indicator. Part 2.” Solar Physics 181: 491–512. Hsiang, S.M., K.C. Meng, and M.A. Cane. 2011. “Civil Conflicts Are Associated with the Global Climate.” Nature 476: 438–441. Hughes, J.D. 1983. “How the Ancients Viewed Deforestation.” JFA 10: 437–445. bibliography 297

———1994. Pan’s travail: environmental problems of the ancient Greeks and Romans. Baltimore. ———2011. “Ancient Deforestation Revisited.” Journal of the History of Biology 44: 43–57. Hughes, M.K. 2002. “Dendrochronology in climatology—the state of the art.” Den- drochronologia 20: 95–116. Hughes, M.K., P.I. Kuniholm, J. Eischeid, G. Garfin, C.B. Griggs, and C. Latini. 2001. “Aegean Tree-Ring Signatures Explained.” Tree-Ring Research 57: 67–73. Hughes, M.K., T.W. Swetnam, and H.F. Diaz. 2011. Dendroclimatology: Progress and Prospects. Developments in Paleoenvironmental Research, Vol. 11. Heidelberg. Humlum, O., J.-E. Solheim, and K. Stordahl. 2011. “Identifying natural contributions to late Holocene climate change.” Global and Planetary Change 79: 145–156. Hunt, C.O., S.J. Gale, and D.D. Gilbertson. 1985. “The UNESCO Libyan Valleys Sur- vey IX: Anhydrite and limestone karst in the Tripolitanian pre-desert.” LibStud 16: 1–13. Hunt, C.O., D.D. Gilbertson, and H.A. El-Rishi. 2007. “An 8000-year history of lan- scape, climate, and copper exploitation in the Middle East: the Wadi Faynan and the Wadi Dana National Reserve in southern Jordan.” Journal of Archaeological Science 34: 1306–1338. Hunt, C.O., D.D. Gilbertson, R.D.S. Jenkinson, M. van der Ween, G. Yates, and P.C. Buckland. 1987. “ULVS XVII: Paleoecology and agriculture of an abandonment phase at Gasr Mm 10, Wadi Mimoun, Tripolitania.” LibStud 18: 1–13. Hunt, C.O., D.J. Mattingly, D.D. Gilbertson, J.N. Dore, G.W.W. Barker, J.R. Burns, A.M. Fleming, and M. van der Ween. 1986. “ULVS XIII: Interdisciplinary approaches to ancient farming in the Wadi Mansur, Tripolitania.” LibStud 17: 7– 47. Hunt, C.O., G. Rushworth, D.D. Gilbertson, and D.J. Mattingly. 2001. “Romano-Libyan Dryland Animal Husbandry and Landscape: Pollen and Palynofacies Analyses of Coprolites from a Farm in the Wadi el-Amud, Tripolitania.” Journal of Archaeo- logical Science 28: 351–363. Huntington, E. 1907. The Pulse of Asia: A Journey in Central Asia Illustrating the Geographic Basis of History. Boston. Hurrell, J.W. 1995. “Decadal trends in the North Atlantic Oscillation: regional temperatures and precipitation.” Science 269: 676–679. ———1996. “Influence of variations in extratropical wintertime teleconnections on Northern Hemisphere temperature.” Geophysical Research Letters 23: 665–668. Hurrell, J.W., and H. van Loon. 1997. “Decadal Variations in Climate associated with the North Atlantic Oscillation.” Climatic Change 36: 301–326. Ineson, S., A.A. Scaife, J.R. Knight, J.C. Manners, N.J. Dunstone, L.J. Gray, and J.D. Haigh. 2011. “Solar forcing of winter climate variability in the Northern Hemi- sphere.” Nature Geoscience 4: 753–757. Institute for Energy Research. 2010. “A Primer on Energy and the Economy.” Available from http://www.instituteforenergyresearch.org/2010/02/16/a-primer -on-energy-and-the-economy-energys-large-share-of-the-economy-requires -caution-in-determining-policies-that-affect-it/. IPCC. 2007. Climate Change 2007: Synthesis Report. Contribution of Working Groups I, II and III to the Fourth Assessment Report of the Intergovernmental 298 bibliography

Panel on Climate Change. Core Writing Team, edited by R.K. Pachauri and A. Rei- singer. Geneva: IPCC. Available from http://www.ipcc.ch/publications_and_ data/publications_ipcc_fourth_assessment_report_synthesis_report.htm. Issar, A.S. 2003. Climate Changes during the Holocene and their Impact on Hydrolog- ical Systems. Cambridge. Issar, A.S., and N. Brown eds. 1998. Water, environment and society in times of climatic change. Dordrecht. Issar, A.S., and D. Yakir. 1997. “Isotopes from Wood Buried in the Roman Siege Ramp of Masada: The Roman Period’s Colder Climate.” Biblical Archaeologist 60 (2): 101–106. Issar, A.S., and M. Zohar. 2007. Climate Change: Environment and History of the Near East. 2nd edition. Berlin. Jahns, S. 2005. “The Holocene History of Vegetation and Settlement at the Coastal Site of Lake Voulkaria in Acarnania, Western Greece.” Vegetation History and Archaeobotany 14: 55–66. Jex, C.N., A. Baker, I.J. Fairchild, W.J. Eastwood, M.J. Leng, H.J. Sloane, L. Thomas, and E. Bekaroğlu. 2010. “Calibration of speleothem δ18O with instrumental climate records from Turkey.” Global and Planetary Change 71: 207–217. Ji, J., J. Shen, W. Balsam, J. Chen, L. Liu, and X. Liu. 2005. “Asian monsoon oscillations in the northeastern Qinghai-Tibet Plateau since the late glacial as interpreted from visible reflectance of Qinghai Lake sediments.” Earth and Planetary Science Letters 233: 61–70. Jiang, H., J. Eiríksson, M. Schulz, K.-L. Knudsen, and M.-S. Seidenkrantz. 2005. “Evidence for solar forcing of sea-surface temperature on the North Icelandic Shelf during the late Holocene.” Geology 33: 73–76. Jones, A.H.M. 1950. “The Aerarium and the Fiscus.” JRS 40: 22–29. Jones, A.H.M., J.R. Martindale, and J. Morris. 1971–1992. The prosopography of the later Roman Empire. Cambridge. Jones, M.D., C.N. Roberts, M.J. Leng, and M. Türkes. 2006. “A high-resolution late Holocene lake isotope record from Turkey and links to North Atlantic and monsoon climate.” Geology 34: 361–364. Jones, P.D., K.R. Briffa, T.J. Osborn, J.M. Lough, T.D. van Ommen, B.M. Vinther, J. Luterbacher, E.R. Wahl, F.W. Zwiers, M.E. Mann, G.A. Schmidt, C.M. Ammann, B.M. Buckley, K.M. Cobb, J. Esper, H. Goosse, N. Graham, E. Jansen, T. Kiefer, C. Kull, M. Küttel, E. Mosley-Thompson, J.T. Overpeck, N. Riedwyl, M. Schulz, A.W. Tudhope, R. Villalba, H. Wanner, E. Wolff, and E. Xoplaki. 2009. “High- resolution palaeoclimatology of the last millennium: a review of current status and future prospects.” The Holocene 19: 3–49. Jongman, W.M. 2007a. “Gibbon was Right: the Decline and Fall of the Roman Econ- omy.” In Crises and the Roman Empire. Proceedings of the Seventh Workshop of the International Network Impact of Empire, edited by O. Hekster, G. de Kleijn, and D. Slooties, 183–199. Leiden. ———2007b. “The Early Roman Empire: Consumption.” In Cambridge Economic History of the Greco-Roman World, edited by W. Scheidel, I. Morris, and R. Saller, 592–618. Cambridge. Kaal, J., Y. Carrion Marco, E. Asouti, M. Martin Seijo, A. Martinez Cortizas, M. Costa Casais, and F. Criado Boado. 2011. “Long-term Deforestation in NW Spain: linking bibliography 299

the Holocene Fire History to Vegetation Change and Human Activities.” Quater- nary Science Reviews 30: 161–175. Kander, A., P. Malanima, and P. Warde. 2013. Power to the People. Energy in Europe over the Last Five Centuries. Princeton. Kaniewski, D., E. Paulissen, V. De Laet, K. Dossche, and M. Waelkens. 2007. “A high resolution Late Holocene landscape ecological history inferred from an intra- montane basin in the Western Taurus Mountains, Turkey.” Quaternary Science Reviews 26: 2201–2218. Kaniewski, D., E. Paulissen, E. Van Campo, H. Weiss, T. Otto, J. Bretschneider, and K. Van Lerberghe. 2010. “Late second–early first millennium BC abrupt climate changes in coastal Syria and their possible significance for the history of the Eastern Mediterranean.” Quaternary Research 74: 207–215. Kaniewski, D., E. Van Campo, E. Paulissen, H. Weiss, J. Bakker, I. Rossignol, and K. Van Lerberghe. 2011. “The medieval climate anomaly and the little Ice Age in coastal Syria inferred from pollen-derived palaeoclimatic patterns.” Global and Planetary Change 78: 178–187. Kaplan, J.O., K.M. Krumhardt, and N. Zimmermann. 2009. “The Prehistoric and Preindustrial Deforestation of Europe.” Quaternary Science Reviews 28: 3016– 3034. Kaufman, D.S., D.P. Scheider, N.P. McKay, C.M. Ammann, R.S. Bradley, K.R. Briffa, G.H. Miller, B.L. Otto-Bliesner, J.T. Overpeck, and B.M. Vinther. “Artic Lakes 2k Project Members. 2009. Recent Warming Reverses Long-Term Arctic Cooling.” Science 325: 1236–1239. Keenan-Jones, D.C. 2010a. “The Aqua Augusta and control of water resources in the Bay of Naples.” ASCS 31 [2010] Proceedings, www.classics.uwa.edu.au/ascs31. ———2010b. The Aqua Augusta. Regional water supply in Roman and late antique Campania. Department of Ancient History. Macquarie University. PhD The- sis. Keigwin, L.D. 1996. “The Little Ice Age and Medieval Warm Period in the Sargasso Sea.” Science 274: 1504–1508. Keppie, L.J.F. 1983. Colonisation and Veteran Settlement in Italy, 47–14BC. London. ———1984. “Colonisation and Veteran Settlement in Italy in the first century AD.” PBSR 52: 77–114. Keys, D. 1999. Catastrophe: An Investigation into the Origins of the Modern World. London. Knipping, M., M. Müllenhoff, and H. Brückner. 2008. “Human Induced Landscape Changes around Bafa Gölü (Western Turkey).” Vegetation History and Archaeo- botany 17: 365–380. Koepke, N. (without date). “Anthropometric Decline of the Roman Empire? Re- gional Differences and Temporal Development of the Quality of Nutrition in the Roman Provinces of Germania and Raetia from the First Century to the Fourth Century AD.” http://eh.net/XIIICongress/Papers/Koepke.pdf. Koepke, N., and J. Baten. 2005. “Climate and its Impact on the Biological Standard of Living in North-East, Centre-West and South Europe during the Last 2000 Years.” History of Meteorology 2: 147–159. Koepke, N., and J. Baten. 2008. “Agricultural Specialization and Height in Ancient and Medieval Europe.” Explorations in Economic History 45, 2: 127–146. 300 bibliography

Kondrashov D., Y. Feliks, and M. Ghil. 2005. “Oscillatory modes of extended Nile River records (A.D. 622–1922).” Geophysical Research Letters 32: L10702. Körber-Grohne, U. 1994. NutzpflanzeninDeutschland:Kulturgeschichteund Biologie. Stuttgart. Köse, N., Ü. Akkemik, H.N. Dalfes, and M.S. Özeren. 2011. “Tree-ring reconstructions of May–June precipitation for western Anatolia.” Quaternary Research 75: 438–450. Kouki, P. 2006. Environmental change and human history in the Jabal Harun area, Jordan. Unpublished licentiate’s dissertation. Institute for Cultural Research, Department of Archaeology, University of Helsinki. Available in electronic for- mat: http://urn.fi/URN:NBN:fi-fe20071209. ———2009. “Archaeological evidence of land tenure in the Petra region, Jordan: Nabataean-Early Roman to Late Byzantine.” Journal of Mediterranean Archaeo- logy 22.1: 29–56. Kramer, B. 1995. “Arborikultur und Holzwirtschaft im griechischen, römischen und byzantinischen Ägypten.” Archiv für Papyrusforschung 41: 217–231. Kromer, B., S.W. Manning, P.I. Kuniholm, M.W. Newton, M. Spurk, and I. Levin. 14 2001. “Regional CO2 offsets in the troposphere: magnitude, mechanisms, and consequences.” Science 294: 2529–2532. Kron, G. 2000. “Roman ley-farming.” JRA 13: 277–287. ———2002. “Archeozoological Evidence for the Productivity of Roman Livestock Farming.” Münstersche Beiträge zur Antiken Handelsgeschichte 21, 2: 53–73. ———2004. “Roman Livestock Farming in Southern Italy: The Case against Environ- mental Determinism.” In Espacesintégrésetgestiondesressourcesnaturellesdans l’EmpireRomain, edited by M. Clavel-Léveque and E. Hermon, 119–134. Besançon. ———2005. “Anthropometry, Physical Anthropology, and the Reconstruction of Ancient Health, Nutrition, and Living Standards.” Historia 54: 68–83. ———2008. “The Much Maligned Peasant. Comparative Perspectives on the Pro- ductivity of the Small Farmer in Classical Antiquity.” In People, Land, and Politics: Demographic Developments and the Transformation of Roman Italy 300B.C.–A.D. 14, edited by S. Northwood and L. De Ligt, 71–119. Leiden. Kuniholm, P.I. 2002. “Dendrochronological Investigations at Herculaneum and Pompeii.” In The Natural History of Pompeii, edited by W.F. Jashemski and F.G. Meyer, 235–239. Cambridge. Lachniet, M.S. 2009. “Climatic and environmental controls on speleothem oxygen- isotope values.” Quaternary Science Reviews 28: 412–432. Lamb, H.H. 1995. Climate, history and the modern world. London. Lamb, H.F., U. Eicher, and V.R. Switsur. 1989. “An 18,000 Year Record of Vegetation, Lake-level, and Climatic Change from Tigalmamine, Middle Atlas, Morocco.” Journal of Biogeography 16: 65–74. Lamy, F., H.W. Arz, G.C. Bond, A. Bahr, and J. Pätzold. 2006. “Multicentennial-scale hydrological changes in the Black Sea and northern Red Sea during the Holocene and the Arctic/North Atlantic Oscillation.” Paleoceanography 21: PA1008. Lane, B., and B. Bousquet. 1994. “Jordan Petra National Park Management Plan.” Main Report. UNESCO. Larsen, L.B., B.M. Vinther, K.R. Briffa, T.M. Melvin, H.B. Clausen, P.D. Jones, M.-L. Siggaard-Andersen, C.U. Hammer, M. Eronen, H. Grudd, B.E. Gunnarson, R.M. bibliography 301

Hantemirov, M.M. Naurzbaev and K. Nicolussi. 2008. “New ice core evidence for a volcanic cause of the A.D.536 Dust Veil.” Geophysical Research Letters 35: L04708. Lavento, M., M. Huotari, H. Jansson, S. Silvonen, and Z.T. Fiema. 2004. “Ancientwater management system in the area of Jabal Haroun, Petra.” In Men of Dikes and Canals, Orient archäologie Band 13, edited by H.-D. Bienert and J. Häser, 163–172. Berlin. Lean, J., and D. Rind. 1998. “Climate Forcing by Changing Solar Radiation.” Journal of Climate 11: 3069–3094. Le Gall, J. 1953. Recherches sur le culte du Tibre. Paris. Lehoux, D. 2007. Astronomy, weather, and calendars in the ancient world: parapeg- mata and related texts in classical and Near Eastern societies. Cambridge. Leitch, V. 2011. “Location, location, location: characterizing coastal and inland production and distribution of Roman African cooking wares.” In Maritime Archae- ology and Ancient Trade in the Mediterranean, edited by D. Robinson and A. Wil- son, 169–195. Oxford. Lemcke, G., and M. Sturm. 1997. “δ18O and Trace Element Measurements as Proxy for the Reconstruction of Climate Changes at Lake Van (Turkey): Preliminary Results.” In Third Millennium BC Climate Change and Old World Collapse: 653– 678, NATO ASI Series I, vol. 49, edited by N. Dalfes, G. Kukla, and H. Weiss. Berlin. Leney, L., and R.W. Casteel. 1975. “Simplified Procedure for Examining Charcoal Specimens for Identification.” Journal of Archaeological Science 2: 153–159. Leng, M.J., M.D. Jones, M. Frogley, W.J. Eastwood, and C.N. Roberts. 2010. “Detri- tal carbonate influences on bulk oxygen and carbon isotope composition of lacustrine sediments from the Mediterranean.” Global and Planetary Change 71: 175–182. Le Roy Ladurie, E. 1971. Times of Feast, Times of Famine, a History of Climate Since the Year 1000. Trans. B. Bray. Garden City, NY. Le Roy Ladurie, E. 2004. Histoire humaine et comparée du climat. Paris. Lettieri, P.1560 [1803]. Discorso dottissimo del Magnifico Ms. Pierro Antonio de’ Lec- thiero cittadino, et Tabulario Napolitano circa l’anticha pianta, et ampliatione della Città di Nap. et del’itinerario del acqua che anticamente flueva, et dentro, et fora la pred. Città per aquedocti mjrabili quale secondo per più raggioni ne dimostra, era il Sebbetho celebrato dagli antichi auttori. Dizionario geografico- ragionato del Regno di Napoli. L. Giustiniani, 6: 382–411. Naples. Lev-Yadun, S., D.S. Lucas, and M. Weinstein-Evron. 2010. “Modeling the demands for wood by the inhabitants of Masada and for the Roman siege.” Journal of Arid Environments 74: 777–785. Little, L.K., ed. 2006. Plague and the End of Antiquity: the pandemic of 541–750. Cambridge. Lo Cascio, E., ed. 2012. L’impatto della peste antonina (Pragmateiai 22). Bari. Lo Cascio, E., and P. Malanima. 2009. “GDP in Pre-Modern Agrarian Economies (1–1820ad). A Revision of the Estimates.” Rivista di Storia Economica, n.s., 25: 391–419. Lo Cascio, E., and P. Malanima. 2008. “Mechanical Energy and Water Power in Europe. A Long Stability?” In Vers une gestion intégrée de l’eau dans l’Empire Romain, edited by E. Hermon, 201–208. Rome. 302 bibliography

Lo Cascio, E., and P. Malanima. Forthcoming. “Ancient and Pre-Modern Economies. GDP in the Roman Empire and Early Modern Europe.” Lockwood, M. 2006. “What do cosmogenic isotopes tell us about past solar forcing of climate?” Space Science Reviews 125: 95–109. Lockwood, M., C. Bell, T. Woollings, R.G. Harrison, L.J. Gray, and J.D. Haigh. 2010a. “Top-down solar modulation of climate: evidence for centennial-scale change.” Environmental Research Letters 5 (3): 034008. Lockwood, M., R.G. Harrison, T. Woollings, and S.K. Solanki. 2010b. “Are cold winters in Europe associated with low solar activity?” Environmental Research Letters 5: 024001. Lockwood, M., and M.J. Owens. 2011. “Centennial changes in the heliospheric mag- netic field and open solar flux: the consensus view from geomagnetic data and cosmogenic isotopes and its implications.” Journal of Geophysical Research 116 (A4): A04109. Lugli, G. 1953. Fontes ad topographiam veteris urbis Romae pertinentes, vol. 2: Libri V– VII. Rome. Luterbacher, J., and 48 coauthors. 2006. “Mediterranean Climate Variability over the Last Centuries: A Review.” In Mediterranean Climate Variability, edited by P. Lionello, P. Malanotte-Rizzoli, and R. Boscolo, 27–148. Amsterdam. Luterbacher, J., R. García-Herrera, S. Akcer-On, R. Allan, M.C. Alvarez-Castro, G. Ben- ito, J. Booth, U. Büntgen, N. Cagatay, D. Colombaroli, B. Davis, J. Esper, T. Felis, D. Fleitmann, D. Frank, D. Gallego, E. Garcia-Bustamante, R. Glaser, J.F. González- Rouco, H. Goosse, T. Kiefer, M.G. Macklin, S.W. Manning, P. Montagna, L. New- man, M.J. Power, V. Rath, P. Ribera, D. Riemann, N. Roberts, S. Silenzi, W. Tinner, B. Valero-Garces, G. van der Schrier, C. Tzedakis, B. Vannière, S. Vogt, H. Wan- ner, J.P. Werner, G. Willett, M.H. Williams, E. Xoplaki, C.S. Zerefos, and E. Zorita. 2012. “A Review of 2000 Years of Paleoclimatic Evidence in the Mediterranean.” In The Climate of the Mediterranean Region: From the Past to the Future, edited by P. Lionello. 87–185. Amsterdam. Luterbacher, J., D. Dietrich, E. Xoplaki, M. Grosjean, and H. Wanner. 2004. “European Seasonal and Annual Temperature Variability, Trends, and Extremes Since 1500.” Science 303: 1499–1503. Luterbacher, J., D. McCarroll, D. Fleitmann, F.J. Gonzalez-Rouco, E. Zorita, S. Sal- cedo, and B. Vinther. 2011. “The first Euro-Med2k regional workshop: Review of current knowledge, available data and plans for multproxy integration.” PAGES news 19: 75–76. Luterbacher, J., R. Rickli, E. Xoplaki, C. Tinguely, C. Beck, C. Pfister, and H. Wanner. 2001. “The Late Maunder Minimum (1675–1715)—A Key Period for Studying Decadal Scale Climatic Change in Europe.” Climatic Change 49: 441– 462. Lyons, G., F. Lunny, and H.P. Pollock. 1985. “A Procedure for Estimating the Value of Forest Fuels.” Biomass 8:283–300. MacDonald, B. 2001. “Relation between paleoclimate and the settlement of southern Jordan during Nabataean, Roman and Byzantine periods.” Amman: Department of Antiquities, Studies in the History and Archaeology of Jordan 7: 373–378. Maddison, A. 2007. Contours of the World Economy, 1–2030A.D. Essays in Macro- Economic History. Oxford. bibliography 303

Magny, M. 1993. “Solar influences on Holocene climatic changes illustrated by correlations between past lake-level fluctuations and the atmospheric 14C record.” Quaternary Research 40: 1–9. ———2004. “Holocene climate variability as reflected by mid-European lake-level fluctuations and its probable impact on prehistoric human settlements.” Quater- nary International 113(1): 65–79. Magny, M., C. Bégeot, J. Guiot, and O. Peyron, O. 2003. “Contrasting patterns of hydrological changes in Europe in response to Holocene climate cooling phases.” Quaternary Science Reviews 22: 1589–1596. Magny, M., J.-L. de Beaulieu, R. Drescher-Schneider, B. Vannière, A.-V. Walter- Simonnet, Y. Miras, L. Millet, G. Bossuet, O. Peyron, E. Brugiapaglia and A. Ler- oux. 2007. “Holocene climate changes in the central Mediterranean as recorded by lake-level fluctuations at Lake Accesa (Tuscany, Italy).” Quaternary Science Reviews 26(13–14): 1736–1758. Magny, M., D. Galop, P. Bellintani, M. Desmet, J. Didier, J.N. Haas, N. Martinelli, A. Pedrotti, R. Scandolari, A. Stock and B. Vannière. 2009. “Late-Holocene climatic variability south of the Alps as recorded by lake-level fluctuations at Lake Ledro, Trentino, Italy.” The Holocene 19(4): 575–589. Magny, M., F. Arnaud, H. Holzhauser, E. Chapron, M. Debret, M. Desmet, A. Leroux, L. Millet, M. Revel, and B. Vannière. 2010. “Solar and proxy-sensitivity imprints on paleohydrological records for the last millennium in west-central Europe.” Quaternary Research 73: 173–179. Maher, L., E.B. Banning, and M. Chazan. 2011. “Oasis or Mirage? Assessing the Role of Abrupt Climate Change in the Prehistory of the Southern Levant.” Cambridge Archaeological Journal 21: 1–29. Mainwaring, A.B., R. Giegengack, and C. Vita-Finzi, eds. 2010. Climate Crises in Human History. Philadelphia. Μaischberger, M. 1999. “Tiberis.” in E.M. Steinby (ed.), Lexicon Topographicum Urbis Romae 5, 69–73. Rome. Maiuri, A. 1958. The Phlegraean Fields. Rome. Makkai L. 1981. “Productivité et exploitation des sources d’énergie (XIIe–XVIIe siè- cle).” In Produttività e tecnologie nei secoli XII–XVII, edited by S. Mariotti, 165–181. Florence. Malanima, P. 1996. Energia e crescita nell’Europa preindustriale. Roma. ———2006. Energy Consumption in Italy in the 19th and 20th Centuries. Naples. ———2009. Pre-Modern European Economy. Leiden. ———2010. “Energy in History.” Encyclopaedia of Life Support Systems (Unesco). ———2011. “The Path Towards the Modern Economy. The Role of Energy.” Rivista di Politica Economica 100: 77–106. ———forthcoming. “Energy.” in Oxford Handbook of Economies in the Classical World, edited by A. Bresson, E. Lo Cascio, and F. Velde. Oxford. Malouta, M. and A.I. Wilson. 2013. “Mechanical irrigation: water-lifting devices in the archaeological evidence and in the Egyptian papyri.” In The Roman agricultural economy: organization, investment, and production (Oxford Studies on the Roman Economy), edited by A. Bowman and A. Wilson, 273–305. Oxford. Mangini, A., C. Spötl, and P. Verdes. 2005. “Reconstruction of temperature in the 304 bibliography

Central Alps during the past 2000 yr from a δ18O stalagmite record.” Earth and Planetary Science Letters 235: 741–751. Mango, M. 2011. “Byzantine settlement expansion in North Central Syria: the case of Androna/Andarin.” In Le Proche-Orient de Justinien aux Abbassides: Peuple- ment et dynamiques spatiales (Bibliothèque de l’Antiquité Tardive 19), edited by A. Borrut, M. Debié, A. Papaconstantinou, D. Pieri, and J.-P. Sodini, 93–122. Turn- hout. Manley, G. 1974. “Central England Temperatures: monthly means 1659 to 1973.” Quarterly Journal of the Royal Meteorological Society 100: 389–405. Mann, M.E., Z. Zhang, S. Rutherford, R.S. Bradley, M.K. Hughes, D. Shindell, C. Ammann, G. Faluvegi, and F. Ni. 2009. “Global signatures and dynamical origins of the Little Ice Age and Medieval Climate Anomaly.” Science 326: 1256–1260. Manning, S.W. 2010. “Radiocarbon dating and climate change.” In Climate Crises in Human History, edited by A.B. Mainwaring, R. Giegengack and C. Vita-Finzi, 25–59. Philadelphia. Manning, S.W., C. Bronk Ramsey, W. Kutschera, T. Higham, B. Kromer, P. Steier, and E.M. Wild. 2006. “Chronology for the Aegean Late Bronze Age 1700–1400B.C.” Science 312: 565–569. Manning, S.W., and B. Kromer. 2011. “Radiocarbon dating archaeological samples in the Eastern Mediterranean, 1730–1480BC: further exploring the atmospheric radiocarbon calibration record and the archaeological implications.” Archaeom- etry 53: 413–439. Manning, S.W., and B. Kromer. 2012. “Considerations of the Scale of Radiocarbon Offsets in the East Mediterranean, and Considering a Case for the Latest (Most Recent) Likely Date for the Santorini Eruption.” Radiocarbon 54 (3–4): 449– 474. Manning, S.W., B. Kromer, C. Bronk Ramsey, C.L. Pearson, S. Talamo, N. Trano, and J.D. Watkins. 2010. “14C Record and Wiggle-Match Placement for the Anatolian (Gordion Area) Juniper Tree-Ring Chronology ~1729 to 751 Cal bc, and typical Aegean/Anatolian (growing season related) regional 14C offset assessment.” Radiocarbon 52: 1571–1597. Manning, S.W., C.L. Pearson, C.B. Griggs, and B. Kromer. 2012. “Dendro-wiggle- match placement of an oak tree-ring chronology from mid-first millennium AD Constantinople.” Antiquity 86, Issue 331: http://antiquity.ac.uk/projgall/ manning331/. Marchetti, D. 1892. “Frammento di un antico pilastro per misurare le acque del Tevere, ed altre notizie topografiche.” Bullettino della Commissione archeologica comunale di Roma 20: 139–149. Marguerie, D. 2011. “Short tree ring series: the study materials of the dendro- anthracologist. In 5th International Meeting of Charcoal Analysis.” In The 5th International Meeting of Charcoal Analysis: The Charcoal as Cultural and Biolog- ical Heritage, edited by E. Badal, Y. Carriòn, E. Grau, M. Macías and M. Ntinou, 15–16. Valencia. Marguerie, D., and J.-Y. Hunot. 2007. “Charcoal Analysis and Dendrology: Data from Archaeological Sites in North-Western France.” Journal of Archaeological Science 34: 1417–1433. Martín-Chivelet, J., M.B. Muñoz-García, R.L. Edwards, M.J. Turrero, and A.I. Ortega. bibliography 305

2011. “Land surface temperature changes in Northern Iberia since 4000 yr bp, based on δ13C of speleothems.” Global and Planetary Change 77: 1–12. Martinez-Cortizas, A., X. Pontevedra-Pombal, J.C. Nóvoa-Muñoz, and E. García- Rodeja. 1997. “Four thousand years of atmospheric Pb, Cd and Zn deposition recorded by the ombrotrophic peat bog of Penido Vello (northwest Spain).” Water, Air, & Soil Pollution 100: 387–403. Martínez-Cortizas, A., X. Pontevedra-Pombal, E. García-Rodeja, J.C. Nóvoa-Muñoz, and W. Shotyk. 1999. “Mercury in a Spanish Peat Bog: Archive of Climate Change and Atmospheric Metal Deposition.” Science 284: 939–942. Martín-Puertas, C., B.L. Valero-Garcés, A. Brauer, M. Pilar Mata, A. Delgado-Huertas, and P. Dulski. 2009. “The Iberian-Roman Humid Period (2600–1600 cal yr bp) in the Zoñar Lake varve record (Andalucía, southern Spain).” Quaternary Research 71: 108–120. Martín-Puertas, C., B.L. Valero-Garcés, M.P. Mata, P. González-Sampériz, R. Bao, A. Moreno, and V. Stefanova. 2008. “Arid and humid phases in the southern Spain during the last 4000 years: the Zoñar Lake record, Córdoba.” The Holocene 18: 907–921. Marzano, A. 2013. Harvesting the sea: the exploitation of marine resources in the Roman Mediterranean (Oxford Studies on the Roman Economy). Oxford. Masarik, J., and J. Beer. 1999. “Simulation of particle fluxes and cosmogenic nuclide production in the Earth’s atmosphere.” Journal of Geophysical Research 104: 12099–12111. Masse, G., S.J. Rowland, M.A. Sicre, J. Jacob, E. Jansen, and S.T. Belt. 2008. “Abrupt climate changes for Iceland during the last millennium: Evidence from high- resolution sea ice reconstructions.” Earth and Planetary Science Letters 269: 565– 569. Mathers, C., and S. Stoddart, eds. 1994. Development and Decline in the Mediter- ranean Bronze Age, Sheffield Archaeological Monographs 8. Sheffield. Mattingly, D.J., ed. 2003. The Archaeology of Fazzan, vol. 1: Synthesis. London. Mattingly, D.J., and A.I. Wilson. 2003. “Farming the Sahara: the Garamantian contribution in Southern Libya.” In Arid lands in Roman times (Arid Zone Archaeology, Monographs), edited by M. Liverani, 37–50. Firenze. Mattingly, D.J., and A.I. Wilson. 2010. “Concluding thoughts: Made in Fazzan?” In The Archaeology of Fazzan, vol. 3, edited by D.J. Mattingly, 523–530. London. Mayewski, P.A., E.E. Rohling, J.C. Stager, W. Karlen, K.A. Maascha, L.D. Meekler, E.A. Meyerson, F. Gasse, S. van Kreveld, K. Holmgren, J. Lee-Thorp, G. Rosqvist, F. Rack, M. Staubwasser, R.R. Schneider, and E.J. Steig. 2004. “Holocene climate variability.” Quaternary Research 62: 243–255. McCarthy, J.J., O.F. Canziani, N.A. Leary, D.J. Dokken, and K.S. White, eds. 2001. Climate Change 2001: Impacts, Adaptation and Vulnerability. Cambridge. McCormick, M. 1990. Eternal Victory. Triumphal rulership in late antiquity, Byzan- tium, and the early medieval West. Cambridge. ———2001. Origins of the European economy. Communications and commerce AD 300–900. Cambridge. ———2012. “Movements and markets in the first millennium: information, contain- ers, and shipwrecks.” In Trade and Markets in Byzantium, edited by C. Morrisson, 51–98. Washington DC. 306 bibliography

McCormick, M., P.E. Dutton, and P.A. Mayewski. 2007. “Volcanoes and the Climate Forcing of Carolingian Europe, A.D. 750–950.” Speculum 82: 865–895. McCormick, M., G. Huang, K.L. Gibson, and M. Polk, eds. 2010. Digital Atlas of Roman and Medieval Civilizations: http://darmc.harvard.edu/. McCormick, M., U. Büntgen, M. Cane, E. Cook, K. Harper, P. Huybers, T. Litt, S.W. Manning, P.A. Mayewski, A.M. More, K. Nicolussi, and W. Tegel. 2012. “Climate Change under the Roman Empire and its Successors, 100bc–800ad. A First Syn- thesis Based on Multi-Proxy Natural Scientific and Historical Evidence.” Journal of Interdisciplinary History 43.2: 169–220. McCormick, M., K. Harper, A.M. More and K. Gibson. 2012. “Geodatabase of historical evidence on Roman and post-Roman climate.” DARMC Scholarly Data Series, Data Contribution Series # 2012–1, DARMC, Center for Geographic Analy- sis, Harvard University, Cambridge MA 01238: http://darmc.harvard.edu/icb/icb .do?keyword=k40248&pageid=icb.page496495. McDermott, F. 2004. “Palaeo-climate reconstruction from stable isotope variations in speleothems: a review.” Quaternary Science Reviews 23: 901–918. McParland, L.C., M.E. Collinson, A.C. Scott, and G. Campbell. 2009. “The use of reflectance for the interpretation of natural and anthropogenic charcoal assemblages.” Archaeological and Anthropological Sciences 1: 249–261. McParland, L.C., M.E. Collinson, A.C. Scott, G. Campbell, and R. Veal. 2010. “Is Vitrification in Charcoal a Result of High Temperature Burning of Wood?” Journal of Archaeological Science 37: 2679–2687. McParland, L.C., Z. Hazell, G. Campbell, M.E. Collinson, and A.C. Scott. 2009. “How the Romans got Themselves into Hot Water, Temperatures and Fuel Types used in Firing a Hypocaust.” Environmental Archaeology 14: 172–179. Meiggs, R. 1982. Trees and Timber in the Ancient Mediterranean World. Oxford. Meteorological Department of the Hashemite Kingdom of Jordan. 2004. “Tables: Total rainfall, Mean monthly and yearly temperature, Absolute maximum air temperature, Absolute minimum air temperature.” http://www.jmd.gov.jo/ .pdf documents accessed 24.10.2004; no longer available in the Internet. Documents in possession of P. Kouki. Meyer, W. 1995. “Burgen, Pfalzen, Herrensitze.” In Mittelalterarchäologie in Zen- traleuropa: Zum Wandel der Aufgaben und Zielsetzungen, edited by G.P. Fehring and W. Sage, 27–36. Cologne. Mighall, T.M., A. Martinez Cortizas, H. Biester, and S. Turner. 2006. “Proxy Climate and Vegetation Changes during the Last Five Millennia in NW Iberia: Pollen and Non-Pollen Palynomorph Data from Two Ombrotrophic Peat Bogs in the North Western Iberian Peninsula.” Review of Palaeobotany and Palynology 141: 203–223. Mighall, T.M., S. Timberlake, I.D.L. Foster, E. Krupp, and S. Singh. 2009. “Ancient copper and lead pollution records from a raised bog complex in Central Wales, UK.” Journal of Archaeological Science 36: 1504–1515. Migowski, C., M. Stein, S. Prasad, J.F.W. Negendank, and A. Agnon. 2006. “Holocene climate variability and cultural evolution in the Near East from the Dead Sea sedimentary record.” Quaternary Research 66: 421–431. Mikesell, M.W. 1969. “The Deforestation of Mount Lebanon.” Geographical Review 59: 1–28. Miller, G.H., A. Geirsdottir, Y. Zhong, D.J. Larsen, B.L. Otto-Bliesner, M.M. Holland, bibliography 307

D.A. Bailey, K.A. Refsnider, S.J. Lehman, J.R. Southon, C. Anderson, H. Björnsson, and T. Thordarson. 2012. “Abrupt onset of the Little Ice Age triggered by volcanism and sustained by sea-ice/ocean feedbacks.” Geophysical Research Letters 39: L02708. Mithen, S., and E. Black, eds. 2011. Water, Life and Civilisation: Climate, Environment and Society in the Jordan Valley. Cambridge. Moe, D., F.G. Fedele, A.E. Maude, and M. Kvamme. 2007. “Vegetational Changes and Human Presence in the Low-alpine and Subalpine Zone in Val Febbraro, Upper Valle di Spluga (Italian Central Alps), from the Neolithic to the Roman Period.” Vegetation History and Archaeobotany 16: 431–451. Mokyr, J. 1990. The Lever of Riches. Technological Creativity and Economic Progress. New York; Oxford. Molle, F., P.P. Mollinga, and P. Wester. 2009. “Hydraulic bureaucracies and the hydraulic mission: Flows of water, flows of power.” Water Alternatives 2(3): 328– 349. Mondin, L. 2003. “La data di pubblicazione della Mosella.” In Cavarzere 2003, 189– 218. Monteix, N. 2009. “Pompéi, recherches sur les boulangeries de l’Italie romaine.” Fasti Online Documents & Research: The Journal of Fasti Online. Available from: http:// eprints.bice.rm.cnr.it/1129/1/FOLDER-it-2009--168.pdf. Mook, W.G., and J. van der Plicht. 1999. “Reporting 14C Activities and Concentra- tions.” Radiocarbon 41(3): 227–239. Morris, I. 1998. “Archaeology and archaic Greek history.” In Archaic Greece: New Approaches and New Evidence, edited by N. Fisher and H. van Wees, 1–91. London. ———2010a. Social Development. Available in www.ianmorris.org. ———2010b. Why the West Rules—for now. The Patterns of History, and what they reveal about the Future. New York. Morris, I., R. Saller, and W. Scheidel. 2007. “Introduction.” In The Cambridge Eco- nomic History of the Greco-Roman Worlds, edited by W. Scheidel, I. Morris, and R. Saller, 1–12. Cambridge. Motta, R. 1986a. “L’acquedotto Felice.” In Il trionfo dell’acqua: Acque e acquedotti a Roma IV Sec. a.C.–XX Sec., edited by G. Pisani Sartorio and A. Liberati Silverio, 220–225. Roma. ———1986b. “La decadenza degli acquedotti antichi e la conduzione dell’acqua Mariana.” In Il trionfo dell’acqua: Acque e acquedotti a Roma IV Sec. a.C.–XX Sec., edited by G. Pisani Sartorio and A. Liberati Silverio, 203–205. Roma. Müller, J., K. Meisen, H. Dittmeier, and M. Zender, eds. 1928–1971. RheinischesWörter- buch, 9 vols. Bonn. Munro, J.H. 2003. “Industrial Energy from Water-Mills in the European Economy, 5th to 18th Centuries: the Limitations of Power.” In Economia e energia Secc. XIII–XVIII, Istituto Internazionale di Storia economica “F. Datini”, edited by S. Cavaciocchi, 223–269. Firenze. Muscheler, R. 2009. “14C and 10Be around 1650 cal BC.” In Time’s Up! Dating the Minoan eruption of Santorini. Acts of the Minoan Eruption Chronology Workshop, Sandbjerg November 2007 initiated by Jan Heinemeier & Walter L. Friedrich. Mono- graphs of the Danish Institute at Athens Volume 10, edited by D.A. Warburton, 275–284. Athens. 308 bibliography

Muscheler, R., F. Joos, J. Beer, S. Müller, M. Vonmoos, and I. Snowball. 2007. “Solar activity during the last 1000 yr inferred from radionuclide records.” Quaternary Science Reviews 26: 82–97. Nakagawa, T., J.-L. de Beaulieu, and H. Kitagawa. 2000. “Pollen-Derived History of Timber Exploitation from the Roman Period Onwards in the Romanche Valley, Central French Alps.” Vegetation History and Archaeobotany 9: 85–89. Naples Water Works Company Ltd. 1885. Acquedotto di Napoli. Padova. Neff, U., S.J. Burns, A. Mangini, M. Mudelsee, D. Fleitmann, and A. Mater. 2001. “Strong coherence between solar variability and the monsoon in Oman between 9 and 6 kyr ago.” Nature 411: 290–293. Nenninger, M. 2001. Die Römer und der Wald: Untersuchungen zum Umgang mit einem Naturraum am Beispiel der römischen Nordwestprovinzen. Stuttgart. Neumann, J., 1985. “Climate change as a topic in the classical Greek and Roman literature.” Climate Change 7: 441–454. Neumann, F., C. Schölzel, T. Litt, A. Hense, and M. Stein. 2007. “Holocene Vegetation and Climate History of the Northern Golan Heights (Near East).” Vegetation History and Archaeobotany 16: 329–346. Newson, P., G. Barker, P. Daly, D. Mattingly, and D. Gilbertson. 2007. “The Wadi Faynan field systems.” In Archaeology and Desertification: The Wadi Faynan Land- scapeSurvey,SouthernJordan, edited by G. Barker, D. Gilbertson and D. Mattingly, 141–174. Oxford. Nicault, A., Alleaume, S., Brewer, S., Carrer, M., Nola, P. and Guiot, J. 2008. “Mediter- ranean drought fluctuation during the last 500 years based on tree-ring data.” Climate Dynamics 31: 227–245. Nicolussi, K., M. Kaufmann, T.M. Melvin, J. van der Plicht, P. Schießling, and A. Thurner. 2009. “A 9111 year long conifer tree-ring chronology for the European Alps: a base for environmental and climatic investigations.” The Holocene 19: 909–920. Nijboer, A.J., J. van der Plicht, A.M. Bietti Sestieri, and A. De Santis. “1999/2000. A high chronology for the early Iron Age in central Italy.” Palaeohistoria 41/42: 163–176. Nur, A. 2010. “Destructive Earthquakes in Alexandria and Aboukir Bay?” In Alexan- dria and the North-Western Delta. Joint Conference Proceedings of Alexandria: City and Harbour (Oxford 2004) and The Trade and Topography of Egypt’s North- Western Delta, 8th Century bc to 8th Century ad (Berlin 2006) (Oxford Centre for Maritime Archaeology Monograph 5), edited by D. Robinson and A. Wilson. Oxford. Nyberg, J., Malmgren, B.A., Kuijpers, A. and Winter, A. 2002. “A centennial-scale variability of tropical North Atlantic surface hydrography during the late Holocene.” Palaeogeography, Palaeoclimatology, Palaeoecology 183: 25–41. O’Carroll, E., and F. Mitchell. 2011. “Methodological approaches towards quantification and identification of charcoal samples retrieved from archaeological sites.” In 5th International Meeting of Charcoal Analysis. The charcoal as cultural and biological heritage, edited by E. Badal, Y. Carriòn, E. Grau, M. Macías and M. Nti- nou, 25–26 Valencia. Ogley, P., D. Pickles, I. Brocklebank and C. Wood. 2010. Energy efficiency in historic buildings. Insulating thatched roofs. n. pl. bibliography 309

Oleson, J.P. 1984. Greek and Roman mechanical water-lifting devices: the history of a technology, (Phoenix Supplementary Volume). Toronto. ———2000. “Water-lifting.” In Handbook of ancient water technology (Technology and change in history), edited by Ö. Wikander, 207–302. Leiden. Olson, S.D. 1991. “Firewood and Charcoal in Classical Athens.” Hesperia 60: 411–420. Orland, I.J., M. Bar-Matthews, N.T. Kita, A. Ayalon, A. Matthews, and J.W. Valley, 2009. “Climate deterioration in the eastern Mediterranean as revealed by ion microprobe analysis of a speleothem that grew from 2.2 to 0.9 ka in Soreq Cave, Israel.” Quaternary Research 71: 27–35. Ortolani, F., S. Pagliuca, and R.M. Toccaceli. 1991. “Osservazioni sull’evoluzione geo- morfologica olocenica nella piana costiera di Velia (Cilento, Campania) sulla base di nuovi rinvenimenti archeologici.” Geografia fisica e dinamica quaternaria 14: 163–169. Ostrow, S.E. 1985. ““Augustales” along the Bay of Naples: A Case for Their Early Growth.” Historia 34: 64–101. Paillet, A. 2005. Archéologie de l’agriculture moderne. Paris. Parassoglou, G.M. 1987. “Return of uninundated land.” Chronique d’Égypte 62: 205– 218. Pareto, R. 1877. “On the Works Proper to Prevent the Inundations of the Tiber in the City of Rome.” Minutes of Proceedings of the Institution of Civil Engineers 49(3): 334–338. Parker, D.E., T.P. Legg, and C.K. Folland. 1992. “A new daily Central England temperature series, 1772–1991.” International Journal of Climatology 12: 317–342. Parmesan, C., and G. Yohe. 2003. “A globally coherent fingerprint of climate change impacts across natural systems.” Nature 421: 37–42. Patterson, W.P., K.A. Dietrich, C. Holmden, and J.T. Andrews. 2010. “Two millennia of North Atlantic seasonality and implications for Norse colonies.” Proceedings of the National Academy of Sciences 107: 5306–5310. Pearson, C.L., C.B. Griggs, P.I. Kuniholm, P.W. Brewer, T. Wazny, and L. Canady. 2012. “Dendroarchaeology of the mid-first millennium AD in Constantinople.” Journal of Archaeological Science 39(11): 3402–3414. Peltonen-Sainio, P., A. Rajala, H. Känkänen and K. Hakala. 2009. “Improving farming systems in Northern European conditions.” In Crop Physiology: Applications for Genetic Improvement and Agronomy, edited by V.O. Sadras and D. Calderini, 71–97. Amsterdam. Perlès, C. 1977. Préhistoire du feu. Paris. Perry, C.A., and K.J. Hsu. 2000. “Geophysical, archaeological, and historical evidence support a solar-output model for climate change.” Proceedings of the National Academy of Sciences 97: 12433–2438. Perry, D.A., R. Oren, and S.C. Hart. 2008. Forest Ecosystems, 2nd ed. Baltimore. Peyron, O., S. Goring, I. Dormoy, U. Kotthoff, J.R. Pross, J.-L. de Beaulieu, R. Drescher- Schneider, B. Vannière, and M. Magny. 2011. “Holocene seasonality changes in the central Mediterranean region reconstructed from the pollen sequences of Lake Accesa (Italy) and Tenaghi Philippon (Greece).” The Holocene 21(1): 131–146. Piccarreta, M., M. Caldara, D. Capolongo, and F. Boenzi. 2011. “Holocene geomorphic activity related to climatic change and human impact in Basilicata, Southern Italy.” Geomorphology 128(3–4): 137–147. 310 bibliography

Pignatti, S. 1982. Flora d’Italia, 3 v. Bologna. ———1997. I Boschi d’Italia. Sinecologia e biodiversità. Torino. Pireddu, G. 1990. L’energia nell’analisi economica. Milan. Pitts, M., and R. Griffin. 2012. “Exploring Health and Social Well-being in Late Roman Britain.” AJA 116: 253–276. Planchais, N. 1982. “Palynologie lagunaire de l’étang de Mauguio: Paléoenviron- nement végétal et évolution anthropique.” Pollen et Spores 24: 93–118. Plunket, P., and G. Uruñuela. 2006. “Social and cultural consequences of a late Holocene eruption of Popocatépetl in central Mexico.” Quaternary International 151: 19–28. Plunkett, G., and G.T. Swindles. 2008. “Determining the Sun’s influence on Late- glacial and Holocene climates: a focus on climate response to centennial-scale solar forcing at 2800 cal. BP.” Quaternary Science Reviews 27: 175–184. Poff, N.L., J.D. Allan, M.B. Bain, J.R. Karr, K.L. Prestegaard, B.D. Richter, R.E. Sparks and J.C. Stromberg. 1997. “The Natural Flow Regime.” BioScience 47(11): 769–784. Polzer, M.E. 2011. “Early Shipbuilding in the Eastern Mediterranean.” in The Oxford Handbook of Maritime Archaeology, edited by A. Catsambis, B. Ford and D.L. Hamilton, 349–378. Oxford. Pop-Lazic, S.P. 2002. “Nekropole rimskog Singidunuma.” Singidunum 3: 7–100. Pope, K.O., and T.H. Van Andel. 1984. “Late Quaternary alluviation and soil formation in the southern Argolid: its history, causes and archaeological implications.” Journal of Archaeological Science 11: 281–306. Portillo, M., and R.M. Albert. 2011. “Husbandry practices and livestock dung at the Numidian site of Althiburos (el Médéina, Kef Governorate, northern Tunisia): the phytolith and spherulite evidence.” Journal of Archaeological Science 38.12: 3224–3233. Poska, A., and I.A. Pidek. 2010. “Pollen Dispersal and Deposition Characteristics of Abies alba, Fagus sylvatica and Pinus sylvestris, Roztocze Region (SE Poland).” Vegetation History and Archaeobotany 19: 91–101. Pothecary, S. 2002. “Strabo, the Tiberian Author: Past, Present and Silence in Strabo’s Geography.” Mnemosyne 55: 387–438. Préaux, C. 1963. “Déclaration d’inondation déficitaire du Brooklyn Museum (P. Brooklyn gr. 5).” Chronique d’Égypte 38: 117–133. Preiser-Kapeller, J. 2010. Complex historical dynamics of crisis: the case of Byzantium: Working Paper. “Historical Dynamics of Byzantium” 4 (Version November 2010): http://oeaw.academia.edu/JohannesPreiserKapeller/Papers/506625/Complex_ historical_dynamics_of_crisis_the_case_of_Byzantium. Provincia di Avellino. 2009. “5. Inquadramento fisico-geografico.” Piano di Emergen- ze Provinciale. Piano stralcio rischio meteorologico: crisi idriche, 32–62 http:// www.provincia.avellino.it:8180/documents/11528/14908/Relazione_Crisi+ Idriche.pdf. Purcell, N. 1996. “Rome and the management of water: environment, culture and power.” In Human landscapes in classical antiquity: environment and culture, edited by G. Shipley and J.B. Salmon, 180–212. London; New York. ———2005. “The Ancient Mediterranean: the View from the Customs House.” In Harris 2005a, 200–232. bibliography 311

Py, V.2006. “Mine charcoal deposits: methods and strategies. The Mediaeval Fournel silver mines in the Hautes-Alpes (France).” In Charcoal Analysis: New Analytical Tools and Methods for Archaeology. Papers from the Table-Ronde held in Basel 2004, BAR International Series 1493, edited by A. Dufraisse, 35–46. Oxford. Qin, X., J. Liu, H. Jia, H. Lu, X. Xia, L. Zhou, G. Mu, Q. Xu, and Y. Jiao. 2012. “New evidence of agricultural activity and environmental change associated with the ancient Loulan kingdom, China, around 1500 years ago.” The Holocene 22: 53–61. Rackham, O. 1982. “Land-use and the Native Vegetation of Greece.” In Symposia of the Association for Environmental Archaeology No. 2, 4–6 September, 1981 (BAR International Series No. 146), edited by M. Bell and S. Limbrey, 177–198. Bristol. ———1996. “Ecology and Pseudo-ecology: the Example of Ancient Greece.” In Human Landscapes in Classical Antiquity: Environment and Culture, edited by G. Shipley and J. Salmon, 16–43. London; New York. Rackham, O., and J. Moody. 1996. The making of the Cretan landscape. Manchester. Railsback, L.B., F. Liang, J.R. Vidal Romaní, A. Grandal-d’Anglade,M. Vaqueiro Rodrí- guez, L. Santos Fidalgo, D. Fernández Mosquera, H. Cheng, and R.L. Edwards. 2011. “Petrographic and isotopic evidence for Holocene long-term climate change and shorter-term environmental shifts from a stalagmite from the Serra do Cou- rel of northwestern Spain, and implications for climatic history across Europe and the Mediterranean.” Palaeogeography, Palaeoclimatology, Palaeoecology 305: 172–184. Rambeau, C., and S. Black. 2011. “Palaeoenvironments of the southern Levant 5,000 bp tp present: linking the geological and archaeological records.” In Water, Life and Civilisation: Climate, Environment and Society in the Jordan Valley, edited by S. Mithen and E. Black, 94–104. Cambridge. Ramrath, A., L. Sadori and J.R.F.W. Negendank. 2000. “Sediments from Lago di Mezzano, central Italy: a record of Lateglacial/Holocene climatic variations and anthropogenic impact.” The Holocene 10(1): 87–95. Rasmusson, E.M., and J.M. Wallace. 1983. “Meteorological aspects of the El Niño/ Southern Oscillation.” Science 222: 1195–1202. Rasulo, M. 2002. “Bolla aqueduct: a two-thousand-year lasting service.” In Hydraulic Information Management (Ecology and the Environment 52), edited by C.A. Breb- bia and W.R. Blain, 369–378. Southampton. Rauh, N.K., R.F. Townsend, M.C. Hoff, M. Dillon, M.W. Doyle, C.A. Ward, R.M. Rothaus, H. Caner, Ü. Akkemik, L. Wandsnider, F.S. Ozener, and C.D. Dorner. 2009. “Life in the Truck Lane: Urban Development in Western Rough Cilicia.” Jahreshefte des Österreichischen Archäologischen Institutes in Wien 78: 253–312. Rayner, N.A., D.E. Parker, E.B. Horton, C.K. Folland, L.V. Alexander, D.P. Rowell, E.C. Kent, and A. Kaplan. 2003. “Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century.” Journal of Geophysical Research 108(D14), 4407. Redman, C.L. 1999. Human impact on ancient environments. Tucson. Reger, G. 1994. Regionalism and Change in the Economy of Independent Delos, 314–167 BC. Berkeley. Regione Campania A.G.C. Sviluppo Attività Settore Primario Settore Foreste Caccia e Pesca, G. de Filippo, M. Fraissinet and M.F. Caliendo. 2011. “Rapporto Ambien- tale sui possibili impatti ambientali significativi derivanti dall’attuazione 312 bibliography

del Piano Faunistico Venatorio, Regione Campania.” http://agricoltura.regione .campania.it/VAS/rapporto_ambientale_2011.pdf. Reimer P.J., M.G.L. Baillie, E. Bard, A. Bayliss, J.W. Beck, C.J.H. Bertrand, P.G. Black- well, C.E. Buck, G.S. Burr, K.B. Cutler, P.E. Damon, R.L. Edwards, R.G. Fairbanks, M. Friedrich, T.P. Guilderson, A.G. Hogg, K.A. Hughen, B. Kromer, G. McCormac, S. Manning, C. Bronk Ramsey, R.W. Reimer, S. Remmele, J.R. Southon, M. Stu- iver, S. Talamo, F.W. Taylor, J. van der Plicht, and C.E. Weyhenmeyer. 2004. “Int- Cal04 terrestrial radiocarbon age calibration, 0–26 cal kyr BP.” Radiocarbon 46: 1029–1058. Reimer, P.J., M.G.L. Baillie, E. Bard, A. Bayliss, J.W. Beck, P.G. Blackwell, C. Bronk Ramsey, C.E. Buck, G.S. Burr, R.L. Edwards, M. Friedrich, P.M. Grootes, T.P. Guilderson, I. Hajdas, T.J. Heaton, A.G. Hogg, K.A. Hughen, K.F. Kaiser, B. Kromer, G. McCormac, S.W.Manning, R.W.Reimer, D.A. Richards, J.A. Southon, S. Talamo, C.S.M. Turney, J. van der Plicht and C.E. Weyhenmeyer. 2009. “IntCal09 and Marine09 radiocarbon age calibration curves, 0–50,000 years cal BP.” Radiocar- bon 59(4): 1111–1150. Renberg, I., M.-L. Brännvall, R. Bindler, and O. Emteryd. 2000. “Atmospheric Lead Pollution History during Four Millennia (2000BC to 2000AD) in Sweden.” Ambio 29 (3): 150–156. Renfrew, C. 1972. The Emergence of Civilization. The Cyclades and the Aegean in the third millennium B.C. London. Renfrew, J. 2004 [1985]. Roman Cookery. Recipes and History. London. Revelle, R., and H.E. Suess. 1957. “Carbon dioxide exchange between atmosphere and ocean and the question of an increase of atmospheric CO2 during the past decades.” Tellus 9: 18–27. Reynolds, T.S. 1983. Stronger than a Hundred Men. A History of the Vertical Water Wheel. Baltimore; London. Riccio, A. 2002. “L’antico acquedotto della Bolla. L’acqua e l’architettura.” In Acque- dotti e fontane del regno di Napoli, edited by F. Starace, 115–180. Lecce. Richter, H.G., D. Grosser, I. Heinz, and P.E. Gasson. 2004. IAWA list of microscopic features for softwood identification. Leiden. Richter, B.D., R. Mathews, D.L. Harrison, and R. Wigington. 2003. “Ecologically Sus- tainable Water Management: Managing River Flows for Ecological Integrity.” Eco- logical Applications 13 (1): 206–224. Riera-Mora, S., and A. Esteban-Amat, A. 1994. “Vegetation History and Human Activity during the Last 6000 Years on the Central Catalan Coast (Northeastern Iberian Peninsula).” Vegetation History and Archaeobotany 3: 7–23. Roberts, N. 1990. “Human-induced Landscape Change in South and Southwest Turkey during the Later Holocene.” In Man’s Role in the Shaping of the Eastern Mediterranean Landscape, edited by S. Bottema, G. Entjes-Nieborg and W. van Zeist, 53–67. Rotterdam and Brookfield, VT. ———1998. The Holocene: An Environmental History. Oxford. Roberts, N., D. Brayshaw, C. Kuzucuoğlu, R. Perez, and L. Sadori. 2011. “The mid- Holocene climatic transition in the Mediterranean: causes and consequences.” The Holocene 21: 3–13. Roberts, N., W.J. Eastwood, C. Kuzucuoğlu, G. Fiorentino, and V. Caracuta. 2011. bibliography 313

“Climatic, vegetation and cultural change in the eastern Mediterranean during the mid-Holocene environmental transition.” The Holocene 21: 147–162. Roberts, N., M.D. Jones, A. Benkaddour, W.J. Eastwood, M.L. Filippi, M.R. Frog- ley, H.F. Lamb, M.J. Leng, J.M. Reed, M. Stein, L. Stevens, B. Valero-Garcés, and G. Zanchetta. 2008. “Stable isotope records of Late Quaternary climate and hydrology from Mediterranean lakes: The ISOMED synthesis.” Quaternary Sci- ence Reviews 27: 2426–2441. Roberts, N., A. Moreno, B.L. Valero-Garcés, J.P. Corella, M. Jones, S. Allcock, J. Wood- bridge, M. Morellón, J. Luterbacher, E. Xoplaki, and M. Türkes. 2012. “Palaeolim- nological evidence for an east-west climate see-saw in the Mediterranean since AD900.” Global and Planetary Change 84: 23–34. Roberts, N., T. Stevenson, B. Davis, R. Cheddadi, S. Brewster, and A. Rosen. 2004. “Holocene Climate, Environment and Cultural Change in the Circum-Medi- terranean Region.” In Past Climate Variability through Europe and Africa, volume 6, edited by R.W. Battarbee, F. Gasse, and C.E. Stickley, 343–362. Dor- drecht. Robinson, S.A., S. Black, B.W. Sellwood, and P.J. Valdes. 2006. “A review of palaeocli- mates and palaeoenvironments in the Levant and Eastern Mediterranean from 25,000 to 5000 years bp: setting the environmental background for the evolution of human civilization.” Quaternary Science Reviews 25: 1517–1541. Rodgers, R.H., ed. 1975. Palladii Rutilii Tauri Aemiliani viri inlustris Opus agriculturae de veterinaria medicina de insitione. Leipzig. Rogers, R.S. 1931. “Lucius Arruntius.” Classical Philology 26: 31–45. Rohling, E.J., P.A. Mayewski, R.H. Abu Zied, J.S.L. Casford, and A. Hayes. 2002. “Holocene atmosphere-ocean interactions: Records from Greenland and the Aegean Sea.” Climate Dynamics 18: 587–594. Rohling, E.J., and H. Pälike. 2005. “Centennial-scale climate cooling with a sudden cold event around 8,200 years ago.” Nature 434: 975–979. Ropelewski, C.F., and M.S. Halpert. 1987. “Global and regional scale precipitation patterns associated with the El Niño/Southern Oscillation.” Monthly Weather Review 115: 1606–1626. Rosen, A.M. 2007. Civilizing Climate: Social Responses to Climate Change in the Ancient Near East. Lanham, MD. Rosen, A.M., and S.A. Rosen. 2001. “Determinist or Not Determinist? Climate, Envi- ronment, and Archaeological Explanation in the Levant.” In Studies in the Archaeology of Israel and Neighboring Lands in Memory of Douglas L. Esse. Stud- ies in Ancient Oriental Civilization 59, edited by S.R. Wolff, 535–549. Chicago. Roos-Barraclough, F., W.O. van der Knapp, J.F.N. van Leeuwen, and W. Shotyk. 2004. “A Late-glacial and Holocene record of climatic change from a Swiss peat humification profile.” The Holocene 13: 7–19. Rosman, K.J.R., W. Chisholm, S. Hong, J.P. Candelone, and C.F. Boutron. 1997. “Lead from Carthaginian and Roman Spanish mines isotopically identified in Green- land ice dated from 600B.C. to 300A.D.” Environment, Science and Technology 31: 3413–3416. Rossignol, B. 2012. “Le climat, les famines et la guerre: éléments du contexte de la Peste Antonine.” In L’impatto della peste antonina, edited by E. Lo Cascio, 87–122. Bari. 314 bibliography

Rossignol, B., and S. Durost. 2007. “Volcanisme global et variations climatiques de courte durée dans l’histoire romaine (Ier s. av. J.-C.–IVème s. ap. J.-C.): leçons d’une archive glaciaire (GISP2).” Jahrbuch des römisch-germanischen Zentralmuseums Mainz 54: 395–438. Roth, R. 2012/3. “Scientific History and Experimental History.” Journal of Interdisci- plinary History 43: 443–458. Rouvillois-Brigol, M. 1985. “La steppisation en Tunisie depuis l’époque punique: déterminisme humain ou climatique?” Bulletin archéologique du Comité des tra- vaux historiques et scientifiques 19.B: 215–224. ———1986. “Quelques remarques sur les variations de l’occupation du sol dans le Sud-Est algérien.” Histoire et archéologie de l’Afrique du Nord. Actes du 3e colloque international, Montpellier, 1985, 35–53. Paris. Ruddiman, W.F. 2003. “The Anthropogenic Greenhouse Era Began Thousands of Years Ago.” Climatic Change 61: 261–293. Ruddiman, W.F., M.C. Crucifix, and F.A. Oldfield, eds. 2011. “Holocene Special Issue: The early-Anthropocene hypothesis.” The Holocene 21(5): 713–879. Ruddiman, W.F., and E.C. Ellis. 2009. “Effects of per-capita Land Use Changes on Holocene Forest Clearance.” Quaternary Science Reviews 28: 3011–3015. Rudolf, B., and U. Schneider. 2005. “Calculation of gridded precipitation data for the global land-surface using in-situ gauge observations.” Proceedings of the 2nd Workshopof the InternationalPrecipitationWorkingGroupIPWG, Monterey Octo- ber 2004, EUMETSAT, 231–247. Monterey, CA. Ruello, M.R. 2008. Geoarcheologia in aree costiere della Campania: i siti di Neapolis ed Elea-Velia. Tesi di dottorato, XX ciclo, Università di Napoli Federico II. www .fedoa.it. Russell, B. 2009. “The dynamics of stone transport between the Roman Mediter- ranean and its hinterland.” Facta 2: 107–126. Russo Ermolli, E., and G. di Pasquale. 2002. “Vegetation dynamics of South western Italy in the last 28 kyr inferred from pollen analysis of a Tyrrhenian Sea core.” Vegetation History and Archaeobotany 11: 211–219. Sabatier, P., L. Dezileau, C. Colin, L. Briqueu, F. Bouchette, P. Martinez, G. Siani, O. Raynal, and U. Von Grafenstein. 2012. “7000 years of paleostorm activity in the NW Mediterranean Sea in response to Holocene climate events.” Quaternary Research 77: 1–11. Sadori, L., and M. Giardini. 2008. “Environmental History in the Mediterranean Basin: Microcharcoal as a Tool to Disentangle Human Impact and Climate Change.” In Charcoals from the Past: Cultural and Palaeoenvironmental Implica- tions (BARIS 1807), edited by G. Fiorentino and D. Magri, 229–236. Oxford. Sallares, R. 1991. The Ecology of the Ancient Greek World. London. ———2007. “Ecology.” In The Cambridge Economic History of the Greco-Roman World, edited by W. Scheidel, I. Morris and R. Saller, 15–37. Cambridge. Salzer, M.W., and M.K. Hughes. 2007. “Bristlecone pine tree rings and volcanic eruptions over the last 5000 yr.” Quaternary Research 67: 57–68. Sangiorni, F., L. Capotondi, N. Combourieu Nebout, L. Vigliotti, H. Brinkhuis, S. Giunta, A.F. Lotter, C. Morigi, A. Negri, and G.J. Reichart. 2003. “Holocene seasur- face temperature variations in the southern Adriatic Sea inferred from a multi- proxy approach.” Journal of Quaternary Science 18: 723–732. bibliography 315

Santarelli, A. 1882. “Forlì.” NSA 1882: 41. Scagel, R.F., R.J. Bandoni, G.E. Rouse, W.B. Schofield, J. Stein, and T.M.C. Taylor. 1969. PlantDiversity:An EvolutionaryApproach (The Wadsworth Botany Series), edited by W.A. Jensen and L.G. Kavaljian. Belmont, CA. Scheidel, W. 2002. “A model of demographic and economic change in Roman Egypt after the Antonine plague.” JRA 15: 97–114. ———2007. “Demography.” In The Cambridge Economic History of the Greco-Roman World, edited by W. Scheidel, I. Morris, R. Saller, 38–86. Cambridge. ———2009. “In search of Roman economic growth.” JRA 22: 46–70. Scheidel, W., and S.J. Friesen. 2009. “The Size of the Economy and the Distribution of Income in the Roman Empire.” JRS 99: 61–91. Scheidel, W., I. Morris, and R.P. Saller. 2007. The Cambridge Economic History of the Greco-Roman World. Cambridge. Schenkel, Y., P. Bertaux, S. Vanwijnbserghe, and J. Carre. 1998. “An Evaluation of the Mound Kiln Carbonization Technique.” Biomass and Bioenergy 14: 505–516. Scherillo, G. 1844. Dell’aria di Baja a tempo dei Romani e di una meravigliosa spelonca scoperta nelle vicinanze di Cuma. Naples. Schiebold, H. 2010. Heizung und Wassererwärmung in römischen Thermen: historische Entwicklung—Nachfolgesysteme—neuzeitliche Betrachtungen und Untersu- chungen. Siegburg. Schilman, B., M. Bar-Mathews, A. Almogi-Labin, and B. Luz. 2001. “Global climate instability reflected by eastern Mediterranean marine records during the late Holocene.” Palaeogeography, Palaeoclimatology, Palaeoecology 176: 157– 176. Schilman, B., A. Ayalon, M. Bar-Matthews, E.J. Kagan, and A. Almogi-Labin., 2002. “Sealand paleoclimate correlation in the Eastern Mediterranean region during the late Holocene.” Israel Journal of Earth Sciences 51: 181–190. Schmidt, B. 2010. “Der Jahrhundert Sommer im Jahr 49 n. Chr. im Rheinland.” Kölner Jahrbuch 43: 695–699. Schmidt, B., and W. Gruhle, W. 2003a. “Klimaextreme in Römischer Zeit. Eine Struk- turanalyse dendrochronologischer Daten.” Archäologisches Korrespondenzblatt 33: 421–426. ———2003b. “Niederschlagsschwankungen in Westeuropa während der letzten 8000 Jahre. Versuch einer Rekonstruktion mit Hilfe eines neuen dendrochro- nologischen Verfahrens (Grad der Wuchshomogenität).” Archäologisches Korre- spondenzblatt 33: 281–300. Schmidt, B., W. Gruhle, A. Zimmermann, and T. Fischer. 2005. “Mögliche Schwan- kungen von Getreideerträgen—Befunde zur rheinischen Linienbandkeramik und römischen Kaiserzeit.” Archäologisches Korrespondenzblatt 35: 301–316. Schmiedt, G., ed. 1972. Il livello antico del Mar Tirreno. Testimonianze dei resti archeologici. Firenze. Schneider, H. 2007. “Technology.” In Cambridge Economic History of the Greco- Roman World, edited by W. Scheidel, I. Morris, R. Saller, 144–171. Cambridge. Schoch, W., I. Heller, F.H. Schweingrüber, and F. Kienast. 2004. “Wood Anatomy of Central European Species.” In Swiss Federal Research Institute WSL. http://www .woodanatomy.ch/ (accessed 2004 onwards). 316 bibliography

Schweingrüber, F.H. 1990. Anatomie europäischer Hölzer: ein Atlas zur Bestimmung europäischer Baum-, Strauch-, und Zwergstrauchhölzer. Bern. Scott, A.C. 2005. “Charcoal Reflectance as a Proxy for the Emplacement Tempera- ture of Pyroclastic Flow Deposits.” Geology 33(July): 589–592. Scott, A.C., and R. Veal. 2010. “New Methods and Applications in Roman Charcoal Studies.” In Scienze Naturali e Archeologia, Il paesaggio antico: uomo/ambiente ed eventi catastrofoci. Proceedings of the Conference, Naples 14–16 October, 2010, edited by M.R. Senatore and A.M. Ciarallo, 203–206. Rome. Seager, R., N. Harnik, W.A. Robinson, Y. Kushnir, M. Ting, H.P. Huang, and J. Vélez. 2005. “Mechanisms of ENSO-forcing of hemispherically symmetric precipitation variability.” Quarterly Journal of the Royal Meteorological Society 131: 1501– 1527. Seeck, O. 1920. Geschichte des Untergangs der antiken Welt. Stuttgart. Seidenkrantz, M.-S., S. Aagaard-Sørensen, H. Sulsbrück, A. Kuijpers, K.G. Jensen, and H. Kunzendorf. 2007. “Hydrography and climate of the last 4400 years in a SW Greenland fjord: implication for Labrador Sea palaeoceanography.” The Holocene 17: 387–401. Seim, A., K. Treydte, U. Büntgen, J. Esper, P. Fonti, H. Haska, F. Herzig, W. Tegel, and D. Faust. 2010. “Exploring the potential of Pinus heldreichii Christ for long-term climate reconstruction in Albania.” Tree Rings in Archaeology, Climatology and Ecology 8: 75–82. Seyfarth, W., ed. 1978. Ammiani Marcellini Rerum gestarum libri qui supersunt. Leip- zig. Serre, F. 1978. “The dendroclimatological value of the European larch (Larix decidua Mill.) in the French Maritime Alps.” Tree-Ring Bulletin 38: 25–34. Sgobbo, I. 1938. “L’acquedotto romano della Campania: “Fontis Augustei Aquaeduc- tus”.” NSA 14: 75–97. Shanzer, D. 1998a. “The date and literary context of Ausonius’ Mosella: Ausonius, Symmachus, and the Mosella.” In Style and Tradition: studies in honor of Wendell Clausen, edited by P. Knox and C. Foss, 284–305. Stuttgart. ———1998b. “The date and literary context of Ausonius’s Mosella.” Historia 47: 204–233. Shao, X., Y. Xu, Z.-Y. Yin, E. Liang, H. Zhu, and S. Wang. 2010. “Climatic implications of a 3585-year tree-ring width chronology from the northeastern Qinghai-Tibetan Plateau.” Quaternary Science Reviews 29: 2111–2122. Shaw, B.D. 1981. “Climate, environment and history: the case of Roman North Africa.” In Climate and history: studies in past climates and their impact on man, edited by T.M.L. Wigley, M.J. Ingram, and G. Farmer, 379–403. Cambridge. Sheppard, P., P. Tarasov, L. Graumlich, K. Heussner, M. Wagner, H. Osterle, and L. Thompson. 2004. “Annual precipitation since 515bc reconstructed from living and fossil juniper growth of Northeast Qinghai Province, China.” Climate Dynam- ics 23: 869–881. Sherratt, A., and S. Sherratt. 1991. “From Luxuries to Commodities: The Nature of Mediterranean Bronze Age Trading Systems.” In Bronze Age Trade in the Mediter- ranean (Studies in Mediterranean Archaeology 90), edited by N.H. Gale, 351–386. Göteborg. bibliography 317

Shindell, D.T., G.A. Schmidt, M.E. Mann, D. Rind, and A. Waple. 2001. “Solar Forcing of Regional Climate During the Maunder Minimum.” Science 294: 2149–2152. Shindell, D.T., G.A. Schmidt, R.L. Miller, and M.E. Mann. 2003. “Volcanic and solar forcing of climate change during the preindustrial era.” Journal of Climate 16: 4094–107. Shipley, F.W. 1931. “Chronology of the Building Operations in Rome from the Death of Caesar to the Death of Augustus.” Memoirs of the American Academy in Rome 9: 7–60. Shotyk, W., D. Weiss, P.G. Appleby, A.K. Cheburkin, R. Frei, M. Gloor, J.D. Kramers, S. Reese, and W.O. Van Der Knapp. 1998. “History of Atmospheric Lead Deposi- tion Since 12,370 14C yr bp from a Peat Bog, Jura Mountains, Switzerland.” Science 281: 1635–1640. Siegenthaler, U., M. Heimann, and H. Oeschger. 1980. “14C variations caused by changes in the global carbon cycle.” Radiocarbon 22: 177–191. Sigurdsson, H., and S. Carey. 2002. “The Eruption of Vesuvius in A.D. 79.” In The Natural History of Pompeii, edited by W.F. Jashemski and F.G. Meyer, 37–64. Cambridge. Sigurdsson, H., S. Carey, W.Cornell, and T. Pescatore. 1985. “The eruption of Vesuvius in A.D.79.” National Geographic Research 1: 332–387. Sim, D. 1998. Beyond the Bloom. Bloom Refining and Iron Artifact Production in the Roman World, BAR International Series 725. Edited by Isabel Ridge. Oxford. Sirot, E. 2011. Allumer le feu: Cheminée et poêle dans la maison noble et le château du XII e au XVI e siècle. Paris. Sivan, H.S. 1990. “Redating Ausonius’ Moselle.” AJPh 111: 383–394. Slim, H., P. Trousset, R. Paskoff, and A. Oueslati. 2004. Le littoral de la Tunisie. Étude géoarchéologique et historique (Études d’Antiquités africaines). Paris. Smil, V. 1994. Energy in World History. Boulder, CO; San Francisco; Oxford. ———2010. Why America is not a New Rome. Cambridge, MA. Smith, A.H.V. 1997. “Provenance of Coals from Roman Sites in England and Wales.” Britannia 28: 297–324. Smith, A.T. 2003. The Political Landscape: Constellations of Authority in Early Com- plex Polities. Berkeley. Smith, W. 2001. “Environmental Sampling.” In Leptiminus (Lamta) Report No. 2. The East Baths, Cemeteries, Kilns, Venus Mosaic, Site Museum, and other Studies, edited by L.M. Stirling, D.J. Mattingly, and N.B. Lazreg. Portsmouth, RI. Snodgrass, A.M. 1977. Archaeology and the rise of the Greek state. Cambridge. ———1980. Archaic Greece; the age of experiment. London. Sohi, S., E. Lopez-Capel, E. Krull, and R. Bol. 2009. “Biochar, climate change and soil: A review to guide future research.” In CSIRO Land and Water Science Report 05/09. Sydney. Solanki, S.K., I.G. Usoskin, B. Kromer, M. Schüssler, and J. Beer. 2004. “Unusual activity of the Sun during recent decades compared to the previous 11,000 years.” Nature 431: 1084–1087. Sommella, P. 1978. Forma e urbanistica di Pozzuoli romana. Naples. Southon, J.R. 2002. “A first step to resolving the GISP2 and GRIP ice-core chronologies, 0–14,500 yr BP.” Quaternary Research 57: 32–37. 318 bibliography

———2004. “A Radiocarbon Perspective on Greenland Ice-Core Chronologies: can we use ice cores for 14C calibration?” Radiocarbon 46: 1239–1259. Spadea, R., ed. 1998. La Campania antica dal pleistocene all’eta romana ritrovamenti archeologici lungo il gasdotto transmediterraneo. Naples. Speranza, A., B. van Geel, and J. van der Plicht. 2002. “Evidence for Solar Forcing of Climate Change at ca. 850 cal. BC from a Czech Peat Sequence.” Global and Planetary Change 35: 51–65. Sperber, D. 1978. Roman Palestine, 200–400—the Land: Crisis and Change in Agrarian Society as Reflected in Rabbinic Sources. Ramat-Gan. Spurr, M.S. 1986. Arable Cultivation in Roman Italy c. 200B.C.–c. A.D.100. London. Squatriti, P. 2002. Water and Society in Early Medieval Italy, 400–1000. Cambridge. Starr, C.G. 1941. The Roman imperial navy: 31B.C.—A.D.324. Ithaca, NY. Stathakopoulos, D.C. 2000. “The Justinianic Plague Revisited.” Byzantine and Mod- ern Greek Studies 24: 256–276. ———2004. Famine and pestilence in the late Roman and early Byzantine empire: a systematic survey of subsistence crises and epidemics. Aldershot; Burlington, VT. Staubwasser, M., and H. Weiss. 2006. “Holocene climate and cultural evolution in late prehistoric–early historic West Asia.” Quaternary Research 66: 372–387. Steckel, R.H. 2009. “Height and Human Welfare: Recent Developments and New Directions.” Explorations in Economic History 46: 1–23. Steinhilber, F., J. Beer, and C. Fröhlich. 2009. “Total solar irradiance during the Holocene.” Geophysical Research Letters 36: L19704. doi: 10.1029/2009GL040142. Stevens, L.R., E. Ito, A. Schwalb, and H.E. Wright. 2006. “Timing of atmospheric precipitation in the Zagros Mountains inferred from a multi-proxy record from Lake Mirabad, Iran.” Quaternary Research 66: 494–500. Stiros, S.C. 2001. “The AD 365 Crete earthquake and possible seismic clustering during the fourth to sixth centuries ad in the Eastern Mediterranean: a review of historical and archaeological data.” Journal of Structural Geology 23.2–3: 545–562. Stuiver, M., and T.F. Braziunas. 1988. “The Solar Component of the Atmospheric 14C Record.” In Secular Solar and Geomagnetic Variations in the Last 10,000 Years, edited by F.R. Stephenson and A.W. Wolfendale, 245–266. Dordrecht. 14 ———1993. “Sun, ocean, climate and atmospheric CO2: an evaluation of causal and spectral relationships.” The Holocene 3: 289–305. ———1998. “Anthropogenic and solar components of hemispheric 14C.” Geophysical Research Letters 25: 329–332. Stuiver, M., T.F. Braziunas, B. Becker, and B. Kromer. 1991. “Climatic, solar, oceanic, and geomagnetic influences on late-glacial and Holocene atmospheric 14C/12C change.” Quaternary Research 35: 1–24. Stuiver, M., and P.D. Quay. 1980. “Changes in atmospheric carbon-14 attributed to a variable sun.” Science 207: 11–19. Stuiver, M., P.J. Reimer, and T.F. Braziunas. 1998. “High-precision radiocarbon age calibration for terrestrial and marine samples.” Radiocarbon 40: 1127–1151. Suess, H.E. 1955. “Radiocarbon Concentration in Modern Wood.” Science 122: 415– 417. Swan, P.M.1987. “Cassius Dio on Augustus: A Poverty of Annalistic Sources?” Phoenix 41: 272–291. bibliography 319

Swindles, G., G. Plunkett, and H.M. Roe. 2007. “A delayed climatic response to solar forcing at 2800 cal. bp: multiproxy evidence from three Irish peatlands.” The Holocene 17: 177–182. Syme, R. 1981. “Vibius Rufus and Vibius Rufinus.” ZPE 43: 365–376. Tandy, D.W. 1997. Warriors into Traders: The Power of the Market in Early Greece. Berkeley; Los Angeles. Taylor, R.M. 2000. Public needs and private pleasures: water distribution, the Tiber river and the urban development of ancient Rome. Rome. Teleles, I.G. 2004. Meteōrologika phainomena kai klima sto Vyzantio. Athens. Temin, P. 2006. “The economy of the early Roman empire.” Journal of Economic Perspectives 20.1: 133–151. Théry-Parisot, I., L. Chabal, and J. Chrzavzez. 2010. “Anthracology and taphonomy, from wood gathering to charcoal analysis. A review of the taphonomic processes modifying charcoal assemblages, in archaeological contexts.” Palaeogeography, Palaeoclimatology, Palaeoecology 291: 142–153. Thomas, R.G., and A.I. Wilson. 1994. “Water supply for Roman farms in Latium and South Etruria.” PBSR 62: 139–196. Thommen, L. 2012. AnEnvironmentalHistoryofAncientGreeceandRome. Translated by P. Hill. Cambridge. Thompson, D.J.W., and J.M. Wallace. 1998. “The Arctic Oscillation signature in wintertime geopotential height and temperature fields.” GeophysicalResearchLetters 25: 1297–1300. Thompson, L.G., E. Mosley-Thompson, M.E. Davis, A. Henderson, H.H. Brecher, V.S. Zagorodnov, T.A. Mashiotta, P.-N. Lin, V.N. Mikhalenko, D.R. Hardy, and J. Beer. 2002. “Kilimanjaro Ice Core Records: Evidence of Holocene Climate Change in Tropical Africa.” Science 298: 589–593. Thompson, L.G., E. Mosley-Thompson, H. Brecher, M. Davis, B. León, D. Les, P.-N. Lin, T. Mashiotta, and K. Mountain. 2006. “Abrupt tropical climate change: Past and present.” Proceedings of the National Academy of Sciences 103: 10536–543. Thonemann, P. 2011. The Maeander Valley: a Historical Geography from Antiquity to Byzantium. Cambridge. Thornton, M.K. 1986. “Julio-Claudian Building Programs: Eat, Drink, and Be Merry.” Historia 35: 28–44. Thornton, M.K., and R.L. Thornton. 1990. “The Financial Crisis of A.D. 33: A Keyne- sian Depression?” Journal of Economic History 50(3): 655–662. Tingley, M.P., P.F. Craigmile, M. Haran, B. Li, E. Mannshardt, and B. Rajaratnam. 2012. “Piecing together the past: statistical insights into paleoclimatic reconstructions.” Quaternary Science Reviews 35: 1–22. Torelli, M. 2006. “The Topography and Archaeology of Republican Rome.” In A Com- panion to the Roman Republic, edited by N. Rosenstein and R. Morstein-Marx, 81–101. Malden, MA; Oxford. Touchan, R., and M.K. Hughes. 1999. “Dendrochronology in Jordan.” Journal of Arid Environments 42: 291–303. Touchan, R., D.M. Meko, and M.K. Hughes. 1999. “A 396-year reconstruction of precipitation in Southern Jordan.” Journal of the American Water Resources Associa- tion 35: 45–55. Touchan, R., G.M. Garfin, D.M. Meko, G. Funkhouser, N. Erkan, M.K. Hughes, and 320 bibliography

B.S. Wallin. 2003. “Preliminary reconstructions of spring precipitation in southwestern Turkey from tree-ring width.” International Journal of Climatology 23: 157–171. Touchan, R., E. Xoplaki, G. Funkhouser, J. Luterbacher, M.K. Hughes, N. Erkan, Ü. Akkemik, and J. Stephan. 2005. “Reconstructions of spring/summer precipitation for the Eastern Mediterranean from tree-ring widths and its connection to large-scale atmospheric circulation.” Climate Dynamics 25: 75–98. Touchan, R., Ü. Akkemik, M.K. Hughes, and N. Erkan. 2007. “May–June precipitation reconstruction of southwestern Anatolia, Turkey during the last 900 years from tree rings.” Quaternary Research 68: 196–202. Touchan, R., K.J. Anchukaitis, D.M. Meko, S. Attalah, C. Baisan, A. Aloui. 2008. “Long term context for recent drought in northwestern Africa.” Geophysical Research Letters 35: L13705. Touchan, R., D.M. Meko, and A. Aloui. 2008b. “Precipitation reconstruction for Northwestern Tunisia from tree rings.” Journal of Arid Environments 72: 1887– 1896. Touchan, R., K.J. Anchukaitis, D.M. Meko, M. Sabir, S. Attalah, and A. Aloui. 2010. “Spatiotemporal drought variability in northwestern Africa over the last nine centuries.” Climate Dynamics 37: 237–252. Toynbee, J.M.C. 1973. Animals in Roman life and art. London. Trenberth, K.E. 1997. “The definition of El Niño.” Bulletin of the American Meteoro- logical Society 78: 2771–2777. Trigger, B.G. 2003. Understanding early civilizations: a comparative study. Cam- bridge. Trigo, I.F., T.D. Davies, and G.R. Bigg. 2000. “Decline in Mediterranean rainfall caused by weakening of Mediterranean cyclones.” Geophysical Research Letters 27: 2913– 2916. Trouet, V., J. Esper, N.E. Graham, A. Baker, J.D. Scourse, and D.C. Frank. 2009. “Persis- tent Positive North Atlantic Oscillation Mode Dominated the Medieval Climate Anomaly.” Science 324: 78–80. Tucci, P.L. 2004. “Eight Fragments of the Marble Plan of Rome Shedding New Light on the Transtiberim.” PBSR 72: 185–202. Turchin, P., and W. Scheidel. 2009. “Coin hoards speak of population declines in ancient Rome.” Proceedings of the National Academy of Sciences 106: 17276–279. Türkeş, M., and E. Erlat. 2003. “Precipitation Changes and Variability in Turkey Linked to the North Atlantic Oscillation During the Period 1930–2000.” Interna- tional Journal of Climatology 23: 1771–1796. Ulrich, R.B. 2007. Roman Woodworking. New Haven, CT. Unal, Y.S., A. Deniz, H. Toros, and S. Incecik. 2012. “Temporal and spatial patterns of precipitation variability for annual, wet, and dry seasons in Turkey.” International Journal of Climatology 32: 392–405. United States Department of Agriculture, Foreign Agricultural Service. “Monthly Crop Growth Stage and Harvest Calendars.” http://www.fas.usda.gov/pecad/ weather/Crop_calendar/crop_cal.pdf, consulted January 4, 2012. University of Florida, the Food and Agricultural Organization of the United Nations, and the National Museum of Natural History of the Smithsonian Institution. bibliography 321

“Ecoport,” Secale cereale, “Description: Physiology,” http://ecoport.org/ep?Plant =1929&entityType=PL&entityDisplayCategory=full. Usoskin, I.G., and B. Kromer. 2005. “Reconstruction of the 14C Production Rate from Measured Relative Abundance.” Radiocarbon 47: 31–37. Usoskin, I.G., S.K. Solanki, and G.A. Kovaltsov. 2007. “Grand minima and maxima of solar activity: new observational constraints.” Astronomy & Astrophysics 471: 301–309. van Andel, T.H., E. Zangger, and A. Demitrack. 1990. “Land use and soil erosion in prehistoric and historical Greece.” JFA 17: 379–396. van de Mieroop, M. 2004. A History of the Ancient Near East, ca. 3000–323bc. Oxford. van der Leeuw, S.E., and the Archaeomedes Research Team. 2005. “Climate, Hydrol- ogy,Land Use, and Environmental Degradation in the Lower Rhone Valley during the Roman Period.” Comptes Rendus. Geoscience 337 (1–2): 9–27. van der Plicht, J., H.J. Bruins, and A.J. Nijboer. 2009. “The Iron Age around the Mediterranean: a High Chronology perspective from the Groningen radiocarbon database.” Radiocarbon 51: 213–242. van Geel, B., N.A. Bokovenko, N.D. Burova, K.V. Chugunov, V.A. Dergachev, V.G. Dirk- sen, M. Kulkova, A. Nagler, H. Parzinger, J. van der Plicht, S.S. Vasiliev, and G.I. Zaitseva. 2004. “Climate change and the expansion of the Scythian culture after 850BC: a hypothesis.” Journal of Archaeological Science 31: 1735–1742. van Geel, B., J. Buurman, and H.T. Waterbolk. 1996. “Archaeological and palaeoecological indications for an abrupt climate change in The Netherlands and evidence for climatological teleconnections around 2650bp.” Journal of Quaternary Science 11: 451–460. van Geel, B., C.J. Heusser, H. Renssen, and C.J.E. Schuurmans. 2000. “Climatic change in Chile at around 2700bp and global evidence for solar forcing: a hypothesis.” The Holocene 10: 659–664. van Geel, B., O.M. Raspopov, H. Renssen, J. van der Plicht, V.A. Dergachev, and H.A. Meijer. 1999. “The role of solar forcing upon climate change.” Quateranry Science Reviews 18: 331–338. van Geel, B., and H. Renssen. 1998. “Abrupt climate change around 2,650bp in North- West Europe: evidence for climatic teleconnections and a tentative explanation.” In Water,EnvironmentandSocietyinTimesofClimaticChange, edited by A.S. Issar and N. Brown, 21–41. Amsterdam. van Geel, B., J. van der Plicht, M.R. Kilian, E.R. Klaver, J.H.M. Kouwenberg, H. Renssen, I. Reynaud-Farrera, and H.T. Waterbolk. 1998. “The sharp rise of Δ14C ca. 800 cal bc: possible causes, related climatic teleconnections and the impact on human environments.” Radiocarbon 40: 535–550. Van Ossel, P. 1992. Établissements ruraux de l’Antiquité tardive dans le nord de la Gaule. Paris. Veal, R. 2009. The Wood Fuel Supply to Pompeii, 3rd c. BC to AD79: an environmental, historical and economic study based on charcoal analysis. Doctoral thesis, Depart- ment of Archaeology, University of Sydney, Sydney. ———2012. “From Context to Economy: Charcoal as an Archaeological Interpreta- tive Tool. A Case Study from Pompeii (3rd c. B.C. to A.D. 79).” In More than just numbers? The role of science in Roman archaeology, Journal of Roman Archaeol- ogy Supplement 91, edited by I. Schrufer-Kolb, 19–52. Portsmouth, RI. 322 bibliography

———forthcoming. Fuelling Pompeii. London. Veal, R., L. O’Donnell, and L. McParland. 2011. “Measuring burn temperatures from charcoal using the reflectance method, first results from an Irish Bronze Age cremation site.” In 5th International Meeting of Charcoal Analysis. The charcoal as cultural and biological heritage, edited by E. Badal, Y. Carriòn, E. Grau, M. Macías and M. Ntinou. 21–22. Valencia. Veal, R., and G. Thompson. 2008. “Fuel Supplies for Pompeii. Pre-Roman and Roman charcoals of the Casa delle Vestali.” In Charcoals From the Past: Cultural and Palaeoenvironmental Implications. Proceedings of the Third International Meet- ing of Anthracology, Cavallino—Lecce (Italy) June 28th–July 1st 2004, BAR Intern- taional Series 1807, edited by Girolamo Fiorentino and Donatella Magri, 287–298. Oxford. Verheyden, S., F.H. Nader, H.J. Cheng, L.R. Edwards, R. Swennen. 2008. “Paleoclimate reconstruction in the Levant region from the geochemistry of a Holocene stalagmite from the Jeita Cave, Lebanon.” Quaternary Research 70: 368–381. Vermoere, M., S. Six, J. Poblome, P. Degryse, E. Paulissen, M. Waelkens, and E. Smets. 2003. “Pollen Sequences from the City of Sagalassos (Pisidia, Southwest Turkey).” AS 53: 161–173. Vernant, J-P. 1957. “Remarques sur les formes et les limites de la pensée technique chez les Grecs.” Revue d’histoire des sciences et de leurs application 10, 3: 205–225. Vetters, W., and H. Zabehlicky. 2001. “Eine Klimakatastrophe um 200 n. Chr. und ihre archäologisch-historische Nachweisbarkeit.” In Archäologie, Naturwissenschaf- ten, Umwelt: Beiträge der Arbeitsgemeinschaft “Römische Archäologie” auf dem 3. Deutschen Archäologenkongress in Heidelberg, 25.5–30.5.1999, edited by M. Frey and N. Hanel, 9–12. Oxford. Vicente-Serrano, S.M., J.I. López-Moreno, L. Gimeno, R. Nieto, E. Morán-Tejeda, J. Lorenzo-Lacruz, S. Beguería, and C. Azorin-Molina. 2011. “A multiscalar global evaluation of the impact of ENSO on droughts.” Journal of Geophysical Research 116, D20109. Vieira, L.E.A., S.K. Solanki, N.A. Krovova, and I. Usoskin. 2011. “Evolution of the solar irradiance during the Holocene.” Astronomy & Astrophysics 531 (A6). Vigneron, P. 1968. Le cheval dans l’Antiquité greco-romaine. Nancy. Vinther, B.M., H.B. Clausen, S.J. Johnsen, S.O. Rasmussen, J.P. Steffensen, K.K. Ander- sen, S.L. Buchardt, D. Dahl-Jensen, I.K. Seierstad, A.M. Svensson, M.-L. Siggaard- Andersen, J. Olsen, and J. Heinemeier. 2008. Reply to comment by J.S. Denton and N.J.G. Pearce on “A synchronized dating of three Greenland ice cores throughout the Holocene”. Journal of Geophysical Research 113: D12306. Visser, R.M. 2010. “Growing and Felling? Theory and Evidence Related to the Appli- cation of Silvicultural Systems in the Roman Period.” In TRAC 2009: Proceedings of the Nineteenth Annual Theoretical Roman Archaeology Conference, edited by A. Moore, G. Taylor, P. Girdwood, E. Harris, and L. Shipley, 11–22. Ann Arbor, MI; Southampton. Vita-Finzi, C. 2008. “Fluvial solar signals.” In Landscape Evolution: Denudation, Cli- mate and Tectonics over Different Time and Space Scales, Geological Society, Lon- don, Special Publications 296, edited by K. Gallagher, S.J. Jones, and J. Wain- wright, 105–115. London. Vogt, S., R. Glaser, J. Luterbacher, D. Riemann, G.H. Al Dyab, J. Schoenbein, and bibliography 323

E. Garcia-Bustamante. 2011. “Assessing the Medieval Climate Anomaly in the Middle East: the potential of Arabic documentary sources.” PAGES News 19: 28– 29. Vonmoos, M., J. Beer, and R. Muscheler. 2006. “Large variations in Holocene solar activity: Constraints from 10Be in the Greenland Ice Core Project ice core.” Journal of Geophysical Research 111: A10105. doi: 10.1029/2005JA011500. Walker, G.T., and E.W. Bliss. 1932. “World Weather V.” Memoirs of the Royal Meteoro- logical Society 4(36): 53–84. Walling, D.E. 2006. “Human impact on land-ocean sediment transfer by the world’s rivers.” Geomorphology 79(3–4): 192–216. Wang, T., D. Surge, and S. Mithen. 2012. “Seasonal temperature variability of the Neoglacial (3300–2500bp) and Roman Warm Period (2500–1600bp) reconstructed from oxygen isotope ratios of limpet shells (Patella vulgata), Northwest Scotland.” Palaeogeography, Palaeoclimatology, Palaeoecology 317–318: 104–113. Wanner, H., J. Beer, J. Bütikofer, T.J. Crowley, U. Cubasch, J. Flückiger, H. Goosse, M. Grosjean, F. Joos, J.O. Kaplan, M. Küttel, S.A. Müller, I.C. Prentice, O. Solom- ina, T.F. Stocker, P. Tarasov, M. Wagner, and M. Widmann. 2008. “Mid- to Late Holocene climate change: an overview.” Quaternary Science Reviews 27: 1791– 1828. Wanner, H., O. Solomina, M. Grosjean, S.P. Ritz, and M. Jetel. 2011. “Structure and origin of Holocene cold events.” Quaternary Science Reviews 30: 3109–3123. Ward-Perkins, B. 2005. The Fall of Rome and the End of Civilization. Oxford. Warde, P. 2006. “Fear of Wood Shortage and the Reality of the Woodland in Europe, c. 1450–1850.” History Workshop Journal 62: 28–57. Wassenburg, J., J. Fietzke, D. Richter, and A. Immenhauser. 2010. “A Holocene speleothem record from Morocco, NW Africa.” EGU General Assembly 2010, held 2–7 May, 2010 in Vienna, Austria. [Katlenburg-Lindau]. Watson, A. 1999. Aurelian and the Third Century. London. Watson, R.T., M.C. Zinyowera, and R.H. Moss, eds. 1998. The Regional Impacts of Climate Change: An Assessment of Vulnerability. A special report of IPCC Working Group II. Project administrator: David J. Dokken. Cambridge. Weinstein, L. 2009. “Limitations on Anthropogenic Global Warming.” In noconsensus.wordpress.com/2009/05/27/the-truth-about-arctic-and-greenland -ice/. Weiss, H. 2000. “Beyond the Younger Dryas: Collapse as Adaptation to Abrupt Cli- mate Change in Ancient West Asia and the Eastern Mediterranean.” In Environ- mental Disaster and the Archaeology of Human Response, edited by G. Bawden and R.M. Reycraft, 75–98. Albuquerque. Weiss, H., and R. Bradley. 2001. “What Drives Societal Collapse.” Science 291: 609–610. Weiss, H., M.-A. Courty, W. Wetterstrom, F. Guichard, L. Senior, R. Meadow, and A. Curnow. 1993. “The Genesis and Collapse of Third Millennium North Mesopo- tamian Civilization.” Science 261: 995–1004. Weiss, E., M.E. Kislev, and A. Hartmann. 2006. “Autonomous Cultivation before Domestication.” Science 312 (16 June): 1608–1610. Weninger, B., L. Clare, E.L. Rohling, O. Bar-Yosef,U. Böhner, M. Budja, M. Bundschuh, A. Feurdean, H.G. Gebel, O. Jöris, J. Linstadter, P. Mayewski, T. Mühlenbruch, A. Reingruber, G. Rollefson, D. Schyle, L. Thissen, H. Todorova, and C. Zielhofer. 324 bibliography

2009. “The impact of rapid climate change on prehistoric societies during the Holocene in the eastern Mediterranean.” Documenta Praehistorica 36: 7–59. Werner, J.P., J. Luterbacher, and J.E. Smerdon. 2013. “A Pseudoproxy Evaluation of Bayesian Hierarchical Modelling and Canonical Correlation Analysis for Climate Field Reconstructions over Europe.” Journal of Climate 26(3): 851–867. Wertime, T.A. 1983. “The Furnace versus the Goat: The Pyrotechnologic Industries and Mediterranean Deforestation in Antiquity.” JFA 10: 445–452. Wheeler, E.A., P. Baas, and P.E. Gasson. 1989. IAWA list of microscopic features for hardwood identification. Leiden. Wheeler, S.M. 2012. “Nutritional and Disease Stress of Juveniles from the Dakhleh Oasis, Egypt.” International Journal of Osteoarchaeology 22: 219–234. Whitby, M. 2000. “The Balkans and Greece, 420–602.” Cambridge Ancient History XIV: 701–730. White, K.D. 1970. Roman Farming. Ithaca, NY. Wick, L., G. Lemcke, and M. Sturm. 2003. “Evidence of Lateglacial and Holocene climatic change and human impact in eastern Anatolia: high-resolution pollen, charcoal, isotopic and geochemical records from the laminated sediments of Lake Van, Turkey.” The Holocene 13: 665–675. Wightman, E.M. 1970. Roman Trier and the Treveri. London. Wikander, Ö. 1979. “Water-mills in Ancient Rome.” ORom: 13–36. ———2000. Handbook of Ancient Water Technology. Leiden; Boston. ———2008. “Sources of Energy and Exploitation of Power.” In The Oxford Hand- book of Engineering and Technology in the Classical World, edited by J.P. Oleson, 136–157. Oxford. Williams, M. 2003. Deforesting the Earth. Chicago. Wilson, A. 1997. Water management and usage in Roman North Africa: a social and technological study. D. Phil. Thesis, University of Oxford. ———2002a. “Machines, Power and the Ancient Economy.” JRS 92: 1–32. ———2002b. “Marine resource exploitation in the cities of coastal Tripolitania.” In L’Africa Romana XIV, vol. 1, edited by M. Khanoussi, P. Ruggeri, and C. Vismara, 429–436. Roma. ——— 2003. “Une cité grecque de Libye: fouilles d’Euhespéridès (Benghazi).” Comptes rendus des séances de l’Académie des inscriptions et belles-lettres,147e année, N. 4, 2003: 1648–1675. ———2006. “The economic impact of technological advances in the Roman construction industry.” In Innovazione tecnica e progresso economico nel mondo romano, edited by E. Lo Cascio, 225–236. Bari. ———2007. “The metal supply of the Roman Empire.” In Supplying Rome and the empire: the proceedings of an international seminar held at Siena-Certosa di Pontignano on May 2–4, 2004, on Rome, the provinces, production and distribution, (JRA Supplement), edited by E. Papi and B. Scardigli, 109–125. Portsmouth, RI. ———2008a. “Hydraulic Engineering.” In Handbook of Engineering and Technology in the Classical World, edited by J.P. Oleson, 285–318. Oxford. ———2008b. “Machines in Greek and Roman Technology.” In The Oxford Hand- book of Engineering and Technology in the Classical World, edited by J.P. Oleson, 337–366. Oxford. bibliography 325

———2009. “Indicators for Roman economic growth: a response to Walter Scheidel.” JRA 22.1: 71–82. ———2011. “City sizes and urbanization in the Roman Empire.” In Settlement,Urban- ization, and Population, (Oxford Studies on the Roman Economy), edited by A. Bowman and A. Wilson, 160–195. Oxford. ———2012a. “Saharan trade in the Roman period: short-, medium- and long- distance trade networks.” Azania: Archaeological Research in Africa 47.4: 409– 449. ———2012b. “Raw materials and energy generation.” In The Cambridge Companion to the Roman Economy, (Cambridge Companions to the Ancient World), edited by W. Scheidel, 133–155. Cambridge. Wilson, A.I., and D.J. Mattingly. 2003. “Irrigation technologies: foggaras, wells and field systems.” In The Archaeology of Fazzan, vol. 1., edited by D.J. Mattingly, 235–278. London. Woodbridge, J., and N. Roberts. 2011. “Late Holocene climate of the Eastern Mediter- ranean inferred from diatom analysis of annually-laminated lake sediments.” Quaternary Science Reviews 30: 3381–3392. Wossink, A. 2009. Challenging Climate Change: Competition and cooperation among pastoralists and agriculturalists in northern Mesopotamia (c. 3000–1600BC). Lei- den. Wrigley, E.A. 2010. Energy and the English Industrial Revolution. Cambridge. ———2013. “Energy and the English Industrial Revolution.” Philosophical Transac- tions of the Royal Society 371: 1–10. Wynn, G. 2011. “Analysis: wood fuel poised to be next global commodity.” http://uk .reuters.com/article/2011/05/19/us-energy-biomass-commodity-idUSTRE 74I3NK20110519 Xoplaki, E. 2002. Climate variability over the Mediterranean. Ph.D. thesis, Univer- sity of Bern. Available from: http://sinus.unibe.ch/klimet/docs/phd_xoplaki.pdf [acccessed January 2012]. Xoplaki, E., J. Luterbacher, H. Paeth, D. Dietrich, N. Steiner, M. Grosjean, and H. Wan- ner. 2005. “European spring and autumn temperature variability and change of extremes over the last half millennium.” Geophysical Research Letters 32: L15713. Yair, A., and A. Kossovsky. 2002. “Climate and surface properties: hydrological response of small arid and semiarid watersheds.” Geomorphology 42: 43–57. Yakir, D., A. Issar, J. Gat, E. Adar, P. Trimborn, and J. Lipp. 1994. “13C and 18O of wood from the Roman siege rampart in Masada, Israel (AD70–73): Evidence for a less arid climate for the region.” Geochimica et Cosmochimica Acta 58: 3535–3539. Yin, H., and C. Li. 2001. “Human impact on floods and flood disasters on the Yangtze River.” Geomorphology 41(2–3): 105–109. Zabehlicky, H. 1994. “Kriegs- oder Klimafolgen in archäologischen Befunden?” Markomannenkriege: Ursachen und Wirkungen, edited by H. Friesinger, J. Tejral, and A. Stuppner, 463–469. Vienna. Zhang, Q.B., G.D. Cheng, T.D. Yao, X.C. Kang, and J.G. Huang. 2003. “A 2326-year tree- ring record of climate variability on the northeastern Qinghai-Tibetan Plateau.” Geophysical Research Letters 30(14), 1739. Zielinski, G.A., P.A. Mayewski, L.D. Meeker, S. Whitlow, and M.S. Twickler. 1996. “A 110,000-yr record of explosive volcanism from the GISP2 (Greenland) ice core.” Quaternary Research 45:109–118.

INDEX

Ancient and modern place names are listed indiscriminately, according to the usage of the various contributors. All lake and river names are listed under ‘Lake’ and ‘River’.

Abate, Pietrantonio, 255 Iulia, 239 Abella, 254 Traiana, 255 Abellinum, 245 Virgo, 255 Acarnania, 184 and see Acquedotto Acer spp.: see maples Arabia, 199 Acquaro spring, 242–246 arboreal pollen, 183–185 and see Serino and see palynological evidence Acquedotto: Arcadia, 180 della Bolla, 254 aridification, 4–5, 41, 203, 208–210 Campano, 254 Aristotle, 267–268 di Serino, 255 Arpinum, 240 Aegean, 160 Arruntius, L. (curator aquarum), 248, Islands, 180 252 Africa, north, 178, 188, 263, 273 Asia, central, 4–5, 89–102 passim Proconsularis, 81 Asia (Minor), 179 agriculture, beginnings of, 105 Ateius Capito, C. (curator aquarum), 248, alder, 53 252 Alexandria, 178, 180, 184 Athens, 115, 177, 179 Alps, 151, 185 Attica, 177, 184, 194 Althiburos (Numidia), 273 Augustus, 242, 248, 252–253, 269 Amisus, 179 Aurelian, 271 Ammerman, A., 260 Ausonius, 4, 63–68, 87 amphorae, 272–273 Avars, 5, 89–93, 101 Anatolia, 161–163 Ancient Climatic Optimum, 24 Baiae, 241 animals, working, 32 Basilicata, 237 Anio Novus, 254 beech (Fagus sylvatica), 3, 53, 56, 216 anthropogenic factors: see human impact Behre, K.-E., 4 on landscape and rivers, pollution Bereket Basin (Turkey), 161 Antigonus the One-Eyed, 180 Beryllium-10 records, 121–123, 126–127, 130, Antioch, 178, 186 132, 134 Antonine Plague, 27, 264–265 Bologna, 255 Antonius, M., 181 Bonneau, D., 4, 71, 76–82 Apicius, 55 Boswijk, G., 93, 102 Appennines, 182, 216, 233, 235 bradyseism, 242 aqueducts, 9–10, 233, 254–255, 268 Bresson, A., 3, 22 Acqua Felice, 255 Britain, Roman, 23, 259–260 Marrana Mariana, 254 Bronze Age, Early, 132 Paola, 255 Bucca della Renella (cave in the Apuan Aqua Alexandrina, 254 Alps), 147–148, 149, 150 Augusta, 9, 240–246, 252, 254–255 Büntgen, U., 89, 103, 105, 116, 117, 139–145, 157, Claudia, 254 262, 266 Crabra, 239 328 index

Caesar, C. Iulius, 248, 251–252, 269–271 coastline changes, 218–221, 225–228, Calpe (Bithynia), 179 n. 229–231, 275 Campania, 9, 233–246, 253–256, 268 colonization, Greek, 114–115, 219 Campi Flegrei, 214–215, 218 Columella, 32, 190–191 Carbon-14: see Radiocarbon Comte, Auguste xiii Carpinus spp.: see hornbeams Constantinople, 72, 240, 268 Carthage, 268 construction, timber for, 177, 179–183 Catalonia, 184 consular honors, consulate, 65–67 Cato, 32, 190–191 Cook, E., 2, 4, 262 cedars, 40, 180 coppicing, 44, 190, 191–192 cereals, 83–84 Corcyra, 179 chaff, 39 Corfinium, 255 Chakrabarty, D., 2 Corinth, 179 charcoal, 3, 37–58, 190 Corsica, 182 varieties of, 54–55 cremation, 179 n. China, climate in, 4–5, 89–101, 161, 275 Crete, 137 use of machines in ancient, 22 Cumae, 215 Cicero, 239–240, 250 curatores alvei Tiberis et riparum, 270 Cilicia, 179 Curius Dentatus, M’., 250 Claudian, 251 cypresses, 136–137, 180, 190–191 Claudius, 270–271 Cyprus, 180 Cleopatra, 181 climate, climate change, 1–10, 21, 24–29, Dalton Minimum, 124 61–170 passim, 185, 192–193, 200–211, Danube provinces of Roman Empire, 186 261–266, 276 Dead Sea, 70–72, 157, 160, 203–204 central-European, 4, 24, 26–29, 67, 85, 89, ‘declarations of unflooded land’ (apographai 136–143, 145, 150, 152–153, 157, 163–165, abrochias), 81–82 167, 169–170, 188–189, 213 deforestation, 8, 27–28, 42, 116, 173–194, 236, defectiveness of concepts of ‘favourable’, 273 ‘unfavourable’, 211 Delos, 40 indifference of rural settlement to, δ13C, 146, 148, 150, 167 197–211 δ18O, 146, 147–148, 235 methods for studying, 61–63, 74–76, 82, demography: see population 168–170 dendochronology, 73 regional differences within the Mediter- and see tree rings ranean world, 109–112, 113–170 passim Dio Cassius, 251–252 ‘see-saw’ relationship between eastern Diocletian’s Price Edict, 40 n. and western Mediterranean, 116, 162 Dionysius of Halicarnassus, 181 short- and long-term, 82 Douglas fir (Pseudotsuga menziesii), 93, stability during Roman period, 26, 70, 96–100 104, 133, 149, 151, 154, 157, 161, 169 Drake, B.L., 115 timescales in, 108–109, 197; droughts, 89–101 passim, 114, 243–244 and see Ancient Climatic Optimum, Dulan-Wulan (north-central China), 90–91, Dalton Minimum, Greenland temper- 100–101 ature proxy record, instability, Iron Dumbarton Oaks, 4, 69, 102, 103 n. Age Cold Period, Little Ice Age (LIA), Dumnissus (Kirchberg), 63–64, 67 Maunder Minimum, Medieval Cli- mate Anomaly, microclimates, North earthquakes, 274–275 Atlantic Oscillation, precipitation, Egypt, 4, 31, 71, 82, 174, 178–180, 265 Roman Climate Optimum, Roman and see River Nile Warm Period Elba, 178 coal, 22–23, 28 Elea: see Velia index 329 elm, 191 Greek ‘Renaissance’, 6, 112–115, 132 El Niño/Southern Oscillation (ENSO), 5, Greenland ice cores, 24–25, 110, 127, 154, 155, 70–71, 89, 92–101 159, 274, 275 energy, 12–36, 259–261 Grissino-Mayer, H., 93, 102 energy-converters, efficiency of, 19, 21 Grove, A.T., 43 Ermolli, E. Russo, 8 growth, economic, 13, 28–29, 264–265 erosion, 185–187, 237 Euboea, 180 Hadrian, 182–183 Eupolis, 177 Hamaxia (in Cilicia), 181 Euro-Climhist, 61 n. Harris, W.V., 57, 273 harvest yields in Egypt, 78 Finnish Jabal Harun Project (FJHP), 198 heating systems, domestic, 85–86 firewood, 14, 15–16, 19, 30–32, 34, 36, 272–273 Hero of Alexandria, 22 and see fuel, wood history in vs. history of, 1–2, 3, 37 fish, fishing, 7, 64, 240, 246, 275 Histria, 182 flood gauge in Rome, 270 Hitchner, B., 264 flooding, 9, 74, 157, 162, 165, 216, 228–229, Horden, P., 8, 120 235–239, 246–253, 268–272 hornbeams (Carpinus spp.), 56 and see River Nile Hughes, J.D., 175 Florentia, 253 human impact on landscape and rivers, fodder, 14, 16, 32 173–256 passim food consumption in antiquity, 15, 30, 34 Huns, 4–5, 70–71, 89–93, 101 foraminifera, use in climate history, 147–148 Huntington, E., 89 forest, 64 hydrological cycle, ancient understanding husbandry: see management of wood- of, 266–272 land hypocausts, 85 Forino aqueduct tunnel (Campania), 244 Fowler, A., 93, 102 Iberia, 108, 157, 184 Frontinus, 239 and see Spain fruit trees, wood of, 56 Ida, Mount, 179 fuel, 12–36, 37–40 Ilopango volcano (El Salvador), 275 and see charcoal, coal, firewood, olive instability, climatic, 5, 70–71, 203, 261 lees and pits, wood indexes, 88 n. and see climate Garamantes, 266 Interamna, 250, 253 Gaul, 63, 193 International Tree-Ring Data Bank (ITRDB), GDP, 19–20, 35–36, 38–40 93 Geoponica, 83 Iran, 161 Germany, ancient, 4, 26 Iron Age Cold Period, 148 Gilliam, J.F., 265 iron production, Roman, 31–32, 176, 178 Giraudi, C., 234–235, 238–239 irrigation, 260 glaciers, 70, 112, 135, 235 in Italy, 239–240, 245 global warming, 106 in Nabataean territory, 201–202, 211 goats, 177 in North Africa, 266 Golan Heights, 184–185 Israel, 26, 203 government regulation of trees, forest, 180, and see Soreq Cave 182–183 Istanbul, timber excavated at, 137, 139–140 grafting, 190 Italy, climate history of, 118 Gratian, 64 ecological science in, 73 La Graufesenque pottery, 272 environmental case-studies, 213–256 Greece, environmental conditions in woodlands of, 181 ancient, 180, 194 Iulius: see Caesar 330 index

Jabal Harun (Jordan), 201–202, 263 maples (Acer spp.), 56 Jabal ash-Shara (Jordan), 198, 204–209 Masada, 157 Jebel al-Aqra (Turkey), 186–187 materia, 179 Jongman, W., 30 Mauguio, Étang de (Provence), 184 n. Jordan, 174, 198–211, 273 Maunder Minimum, 107, 111, 113, 121, 144 juniper (Juniperus przewalskii), 90 McCormick, M., 2, 4, 10, 89, 192–193, 261–262, 264–265, 274 Kaplan, J.O., 187–189 McParland, L., 53 Kasserine Survey, 263 medieval climate, 165–166 kauri (Agathis australis), 93–100 Medieval Climate Anomaly (MCA), Keenan-Jones, D., 9, 268, 274 106–110, 119–120, 149, 151, 154, 158, Kocain Cave (southern Turkey), 146, 149 n. 162 Kouki, P., 6, 8, 263 and see late antiquity Krumhardt, K.M., 187–189 Meiggs, R., 40 metal-smelting, 194 Lake Accesa, 235, 238 and see iron production Bafa, 187 microclimates, 3 Fucino, 235, 238, 256 micro-regions, 8 Ledro, 238–239 migration, 5–6, 27, 70–71 Martignano, 238 Migration Period, 143, 213 Mezzano, 238, 256 Miletus, 184 Monticchio (Lago Grande di Montic- Misenum, 241 chio), 236 Morocco, 108, 109, 144 Nar (Nar Gölü), 161, 162 Morris, I., 34–36 Praver, 186 Van, 158, 160 Nabataean kingdom, 198–201 Velinus, 248, 250, 270 Naples, 8–9, 214–225, 229, 237, 240, 244–245, Zoñar (southern Spain), 162 268 lake levels, 235, 237–238 natura, 250–251 land clearance, 173–176, 236 navicularii lignarii, 178 land reclamation, 256 navies, 177, 179–181 land requirements per capita, 18, 30 Neapolis: see Naples landscapes, ‘ruined’, 175, 180 Negev, 201–202 typology of, 175 Nero, 251 larch (Larix decidua), 139 New Mexico, tree rings from, 92, 93, 95–98 late antiquity, 89–101, 158–162, 163–166, New Zealand, tree rings from, 92, 93–98 169–170, 185–186, 194, 203, 209, 242, North Atlantic Oscillation, 106–111 266 nut trees, wood of, 56 Laurion silver mines, 177 Lebanon, Mount, 180, 182–183 oaks (Quercus spp.), 50, 53, 56, 68, 137–138, Levant, southern, 161–163 139, 185 lignum, 177–179 olive lees and pits, 39, 260 Little Ice Age (LIA), 106–110, 119–120, 144, trees, 185 149, 151, 154, 162, 235 Olympus, Mount, 179 OSL (optically stimulated luminescence) Macedon, 179–180 dating, 202–203 marginal land, 43, 200, 205, 263–264 Ostia, 178, 182, 270–271 Malanima, P., 2–4, 30, 57, 259–261, 263 Malthusian constraints, 264 palaeoclimate research, 120–121 management of woodland, 42–44, 52, 58, Palladius, 83 178, 185–186, 189–192 Palmer, J., 93, 102 Manning, S., 3, 5–6, 25–27, 192, 262 Palmer Drought Series Index, 109–111, 145 index 331 palynological evidence, 42, 52, 183–185, 193, Ombrone, 237 203, 216–218, 222–225, 236 Po, 235, 237, 239 Parthenope, 214–215 Rhone, 237 Paulinus of Nola, 234, 254 Sabato, 9, 244–246 Pausanias, 267 Tiber, 9, 235–237, 239, 246–253, 256, Pelosi spring, 242–246 268–272 and see Serino rivers, 233–256, 260 Petra, 6, 8, 198–211 Roman Climate Optimum, 133–135, 157 pines: stone pine (Pinus cembra), 139 Roman Empire, total cultivated area of, 18 Pisa, 181 Roman Warm Period, 2–3, 24, 70, 134–135, place-names, 64 148, 158 n., 162, 235, 262 Pliny the Elder, 182, 191–192, 249, 251–252, Romans and hydrology, 233–256 267 Rome, 178, 268–272 Pliny the Younger, 2 n. Baths of Constantine and Diocletian, 178 pollarding, 44 Campus Martius, 247–248, 268–271 pollen: see palynological evidence Cloaca Maxima, 269 pollution, ancient, 273–274 Forum Romanum, 248, 269–270 Pompeii, 3, 55–57, 273 Forum Boarium, 247 and see Vesuvian cities fuel needs of, 31 Pompey (Cn. Pompeius Magnus), 269 Horologium, 271 population growth, 24, 27, 114, 176, 187–189, Pantheon, 271 192 Porticus inter lignarios, 178 pottery manufacture at La Graufesenque, Pons Aurelius, 270 272 Pons Sulpicius, 269 in the Petra region, 204 Porta Trigemina, 178 precipitation in Roman times, 26–27, 68, 70, Velabrum, 269–270 71–72, 159–161, 167, 234, 235 Rossignol, B., 25 progress, 253–254 runoff cultivation, 202, 204, 276 Provence, 184 rye, 4, 83–84 public-works projects, 252–253 Puglia, 184–185 Sagalassos, 185 Purcell, N., 1–2, 8, 120 Sahara, 136, 266 Puteoli, 240–241, 255 Salerno, 254 Bay of, 216 Quercus spp.: see oaks Sannio, Università degli Studi del, 243 S. Lucia di Serino: see Serino Rackham, O., 43 Santorini (Thera) eruption, 126 n. radiocarbon dating, 75, 121–122, 123–125, Saserna (Roman agronomist), 32 126–127, 129–132, 134, 138, 151, 167, Scheidel, W., 264 168 sedimentation, 186–187, 213 Raetia, 182 Seneca, 267 Reate, 250–253 Serino springs, 243–246, 268 reflectance, 47, 53–54 ships, sailing-, 21, 33, 260 religio, 251–252 ship-building, timber for, 42, 177, 179, 180, 181, River Arno, 9, 237, 248–249 194 Clanis (Chiana), 9, 248–250, 256, 270, Sicily, 180 272 Sicyon, 180 Frittolo, 215–216, 225–229 silt, 237 Guadalquivir, 272 Sinope, 179 Moselle, 63–64 Smil, V., 36 Nar, 248–249, 270 Sofular Cave (Turkey), 146–147, 149, 150,158, Nile, 70–71, 76–81, 82, 157, 264–266 161 332 index solar activity, 103 Trajan, 271 influence of on climate, 120–135, 139, 144, transport, cost of, 57 151, 159, 168–169 water-borne, 186 solar minimum about 2860bc, 132 tree rings, 51–52, 67–68, 89, 90–100, 119–120, solar minimum about 765bc, 112, 132 136–140, 152–153, 154, 156 solar radiation, veiling of about 536/7ad, Trier, 64 72, 139, 165, 275 Trigger, B.G., 104 and see Maunder Minimum, Sun Spot Troodos Mountains (Cyprus), 186 Number, Total Solar Irradiance Soreq Cave (Israel), 147–149, 150, 158, 160 Ulrich, R.B., 182 Spain, 148, 150, 151, 152, 167, 185 UNESCO Libyan Valleys Study, 263 and see Iberia uranium-thorium dating, 75 Spannagel Cave (Austria), 151–153, 163 Urciuoli spring, 243–246 speleothems, use of for climate history, 73, 120, 146–153 Val Febbraro (Sondrio), 186 state, Roman, and the environment, 9, 250, Valentinian I, 64–67 253, 269 Varro, 32, 191 states, formation of in third millen- Veal, R., 3, 273 niumbc, 132 Velia (Elea), 8–10, 215–217, 225–231, 237 and see government regulation Vernant, J.-P., 22 stature, variations in, 24, 259–260 Vesuvian cities, 182 steam power, 22, 28 and see Pompeii Stoicism, 250, 267 vines, 191 Strabo, 181, 250 Vitruvius, 266–267 sun: see solar volcanic eruptions, 70, 71, 88, 103, 120, 154, Sun Spot Number, 131 155–156, 214–215, 274–275 Switzerland, 26 Syracuse, 179 Wadi ʿAraba (Jordan), 198, 204–205 Syria, 179, 182–183 Wadi Faynan (Jordan), 201, 273 water mills, 16–17, 21, 33, 64 ‘Tabernae’ (Roman Germany), 63–64 Wilson, E.O., 68–69 Tacitus, 246, 250, 253 wind power, 16–17 Talmud, 203 wood: atlases, 49 tax flight, 265 calorific value of various types of, 52–53 Tebtunis, 178 demand for, 174, 176–177 Terminio aquifer, 242–243 as fuel, 38, 45, 47, 176–179 thatched roofs, 86–87 supply, 41–44 Theophrastus, 179–180, 190, 266–267 and see charcoal, firewood, fuel Thrace, 179 woodland, over-exploitation of, 52 Tiberius, 246–248, 268–270 and see deforestation, management Tierra Blanca Joven event, 275 Wrigley, E.A., 13, 28 Torone, 177 Total Solar Irradiance, 122, 126, 128–129, 130, Zimmerman, N., 187–189 167