Quick viewing(Text Mode)

The Torpedo Bay Excavations: Volume 1, the Pre-European Maori Site (HPA Authority 2009/275)

The Torpedo Bay Excavations: Volume 1, the Pre-European Maori Site (HPA Authority 2009/275)

The Torpedo Bay excavations: Volume 1, the pre-European Maori site (HPA authority 2009/275)

report to Heritage New Zealand Pouhere Taonga and The New Zealand Defence Force

Matthew Campbell, Mica Plowman, Emma Brooks, Arden Cruickshank, Louise Furey, Mark Horrocks, Marianne Turner, Rod Wallace and Richard Walter

The Torpedo Bay excavations: Volume 1, the pre-European Maori site (HPA authority 2009/275)

report to Heritage New Zealand Pouhere Taonga and The New Zealand Defence Force

Prepared by: Matthew Campbell

Reviewed by: Date: 10 December 2018 Louise Furey Reference: 17-0773

This report is made available by CFG Heritage Ltd under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-sa/4.0/.

CFG Heritage Ltd. 132 Symonds St Eden Terrace 1010 ph. (09) 309 2426 [email protected] Hard copy distribution

Heritage New Zealand Pouhere Taonga, Auckland New Zealand Archaeological Association (file copy) CFG Heritage Ltd (file copy) New Zealand Defense Force Marutuahu Collective Ngai Tai ki Tamaki Ngati Paoa Ngati Whatua o Orakei Te Kawerau a Maki Auckland War Memorial Museum Auckland Coucnil Cultural Heritage Inventory University of Auckland General Library University of Otago Anthropology Department This report is provided electronically Please consider the environment before printing Contents

Summary...... v 1 Introduction...... 1 2 Background...... 3 A Maori history of Tamaki...... 3 The archaeology of Aotearoa / New Zealand ...... 8 The archaeology of Tamaki...... 21 Acknowledgements...... 27 3 Archaeology...... 29 Methodology...... 29 Stratigraphy...... 31 Chronology...... 43 Summary: site formation processes ...... 47 4 Material Culture ...... 49 Formal tools...... 49 Flaked stone ...... 51 Discussion...... 56 5 Faunal Analysis...... 59 Methods ...... 59 Bird...... 60 Reptile...... 62 Mammal...... 62 Fish...... 63 Shellfish...... 66 Interpretation...... 68 Conclusion...... 71 6 Environment ...... 73 Charcoal ...... 73 Microfossils...... 75 Discussion...... 76 7 Discussion and Conclusion...... 81 The early archaeology of the upper North Island...... 81 Torpedo Bay ...... 101 References...... 103 iv Torpedo Bay Torpedo Bay v

Summary

• In 2009 site R11/1945, the Torpedo Bay Submarine Mining Base was excavated under archaeological authority 2009/275 . • An undisturbed pre-European Maori midden was discovered beneath the base office on an old beach terrace – an area of 5 5. x 4 7. m was excavated . • This midden had two main Phases of occupation: Layer 5 (Phase 1), containing extinct bird including moa, sea mammal and a diverse faunal assemblage; and Layers 3 and 2 (Phase 2), containing mostly shell with some fish and occasional dog . • These two layers were separated by Layers 4 (also Phase 1) and 3A, both of which appeared to be slopewash events and may be related to gardening on the slopes of Maungauika / North Head, just above the site . • Phase 1 dates to the 14th century AD; Phase 2 dates to the late 15th to mid-17th centuries AD . • At the time of excavation, this was the only 14th century site known from the Tamaki mainland and was considered highly significant; • Material culture was sparse but included: • three shell fishhook points from Phase 1; • a fragment of wooden tapa beater from Phase 1, preserved as charcoal; • an adze and a sinker from Phase 2; • and 73 flakes of obsidian, mostly form Tuhua / Mayor Island and mostly from Phase 1 . • Juvenile fur seal in Phase 1 indicated a summer occupation . • Spotted shag was the most common bird species, probably indicating a nearby rookery . • Snapper was the most common fish species, and the proportions of vertebrae to head bones indicated that snapper bodies had been preserved and transported off site in Phase 1 but in Phase 2 fish had been consumed on site . • Shellfish exploitation in Phase 1 centred on rocky shore species, while in Phase 2 soft shore species predominated . • The site is interpreted in relation to other 14th century upper North Island sites and with respect to recent archaeological thinking about the period AD 1300–1450 .

1 Introduction

In 2009 the New Zealand Defence Force (NZDF) began redeveloping their facilities at the Torpedo Bay Boat Yard in Devonport, Auckland, as the new site of the Royal New Zealand Navy Museum . The Torpedo Bay Boat Yard is recorded as archaeological site R11/1945 in the New Zealand Archaeological Association (NZAA) Site Recording Scheme (SRS) . This includes both mid-19th century ship building (the Beddoes and Holmes Shipyard) and the subsequent military activities of the Torpedo Bay Submarine Mining Base which began construction in 1886 in response to the perceived threat of a Russian invasion . The site was originally recorded in 1996 by Tony Packington-Hall and related only to the historic mining base . Pre-European Maori sites are also recorded in Torpedo Bay: R11/1819, a working floor / find spot at the west- ern end of the bay; and R11/2169, a find spot of flakes and an adze at the eastern end, close to the mining base . A historic kainga, Haukapua, is recorded as site R11/2401 half way along the bay . The Torpedo Bay Submarine Mining Base began operations in 1886 and, following two periods of reclamation and building, was completed in 1899 . The site represents the last intact 19th century submarine mining base in New Zealand and forms part of a much larger com- plex relating to the coastal defence of Auckland in the late 19th century . Extant archaeological remains relating to the formation and occupation of the mining base are regionally and nation- ally significant . An archaeological assessment of the effects of the museum project on the archaeology of the site was prepared in 2009 by Opus International Consultants Limited (Opus) (Plowman 2009) . The report primarily assessed the effects of the project in the 19th century mining base but it did acknowledge the remote possibility that pre-European or historic Maori mate- rial might be also be present . NZDF then applied to the New Zealand Historic Places Trust (NZHPT) for an archaeological authority to modify, damage or destroy site R11/1945 under section 14 of the Historic Places Act 1993 . Authority 2009/275 was granted on 11 August 2009 . Excavation was undertaken in November 2009 under the direction of Mica Plowman from Opus . Numerous structural features associated with the submarine mining base were exca- vated, recorded in detail and sampled as appropriate . Recording included photography, detailed sections and plans of all identified internal and external subsurface structural features tied into an overall theodolite site plan . Following excavation, archaeological monitoring of stormwater trenching in the north west corner of the site revealed a pre-European Maori site consisting of several layers of midden, located on what would have been the original beach terrace that formed the eastern extent of Torpedo Bay prior to the military reclamations that have substantially altered the original shore- line and cliff face . Further excavations of the pre-European Maori site were undertaken under the direction of Mica Plowman between 24 March and 5 April 2010 . Preliminary reports on the excavations were prepared by Mica Plowman as a require- ment of Condition 19 of authority 2009/275 but she subsequently left her employment at Opus and the final reports, as required under Condition 20 of the authority, were never completed . In April 2013 NZHPT granted an extension to the deadline for reporting and agreed that the pre-European Maori site and the historic mining base could be reported separately, with measured drawings of the buildings supplied as an appendix to the latter . A draft report for the mining base was completed by Ken Phillips but no final report was prepared . In 2017 NZDF approached Matthew Campbell of CFG Heritage Ltd and requested that he prepare the final reports . This volume presents the results of the pre-European Maori excavations; Volume 2 will present the results of the 19th century historic excavations . 2 Background

The historical and archaeological background to the Torpedo Bay excavation are provided here in two parts . Firstly, an account of Maori tradition is taken from public documents, includ- ing published works and unpublished reports, which between them present a complex and often contradictory and confusing story . This account has been condensed into a short, relatively com- prehensible account – mana whenua may have other traditions and other views that do not agree with parts of this account . A general Maori history of Tamaki is given before a more specific Maori history of Devonport . Secondly, the basis of the archaeology of Torpedo Bay is outlined from a number of perspectives: the nature of early archaeology in New Zealand (the lower layers of the excavated site date to this period); the general sequence of pre-European Maori archaeol- ogy; the archaeology of Tamaki; and the archaeology of Devonport . No concerted attempt is made to align the traditional and archaeological histories – while both deal with the same past and can inform each other, they are different kinds of knowledge about the past and any statements about archaeological processes and events made on the basis of traditional history must be made with great care .

A Maori history of Tamaki

Auckland was known to Maori as Tamaki Makau Rau, Tamaki of the hundred lovers 1. Fertile, well drained volcanic soils and a mild climate were perfectly suited to Maori horticul- ture; the volcanic cones were ideal places to establish pa as statements of political power and places of refuge in times of trouble; the Waitemata and Manukau harbours provided abun- dant fish and shellfish; and the narrow isthmus meant that portages between the two harbours allowed canoes to move between the east and west coasts without going the long way around Northland . Among the first occupants of Tamaki were a people known as Turehu, or ‘the people from the earth’ (Murdoch 1990: 13) . A manuscript of Te Ao o Te Rangi, a nineteenth century chief from Raglan (cited and translated in Sullivan n d. ). says that when Kupe discovered the west coast harbours south of Manukau “Mamoe (phantoms), Turehu (ghosts), Tahurangi (elves),” etc ,. known collectively as Ngati Matakore, inhabited the land . Kupe himself seems not to have entered the Manukau but returned to Hawaiki as a prelude to the Tainui voyage . Subsequently Tini o Maruiwi migrated north from Taranaki, conquering Turehu and settling in the Waitakere Ranges . Oho Mata Kamokamo conquered the Tamaki isthmus, settling at Rarotonga / Mt Smart (a volcanic cone and pa now entirely quarried away) and his descendants, known now as Nga Oho, lived as far north as the Kaipara and as far south as Waikato and Tauranga (Smith 1898a: 33; Murdoch 1990: 13; Daamen et al . 1996) . Nga Oho eventually developed sep- arate identities including Ngati Poutukeka who occupied Mangere/Puhinui and Nga Iwi in the area north of Papakura (Fenton 1879 [1994]: 58) . The early origin traditions are many, varied and contradictory and, as Sullivan (n d. ). points out, they are all ultimately sourced from outside Tamaki; their primary emphasis is on other places . None of the named groups can necessarily be considered the first inhabitants and the relationships between them are not always clear . This situation is further confounded by different

1 It is widely held that this was due to its being such a desirable situation and so frequently fought over (e g. ,. Graham 1922: 20), but there are many other explanations for the name, mostly involving being named after one historical figure or another (e g. ,. Daamen et al . 1996: 7), or it may be an old name that has been reinterpreted (Agnes Sullivan pers . comm . 2010) . 4 Torpedo Bay

peoples using the same name at different times and in different places . The following account is greatly simplified . In the mid-14th century a number Ngati Awa (Te Ati Awa) of Taranaki moved north and settled peacefully among Nga Oho (Murdoch 1990: 12) . Pa building on the isthmus may have commenced with Ngati Awa (Sullivan n d. ). . Following this Hua, a chief of Nga Iwi, gathered together people of different origins into Te Wai o Hua; on account of this he was known as Hua Kai Waka “Hua the eater of canoes ”. By other accounts Hua was a man of Ngati Poutukeka – either way, he seems to have united some of the descendants of Nga Oho with other groups . In some accounts Te Wai o Hua were already in occupation when Ngati Awa arrived . Te Wai o Hua occupied the isthmus at the same time that Kawerau are recorded as occupying from Manukau Heads to Karangahape (Cornwallis) (Waitangi Tribunal 1985: 10) . Kawerau are variously described as the original inhabitants of the Waitakeres or as descend- ants, along with Te Wai o Hua, of the Moekakara waka that landed between and Whangarei (Daamen et al . 1996) . Still, with the presence of these two groups we are on firmer ground (though the names ‘Te Wai o Hua’ and ‘Kawerau’ may be of later origin, Agnes Sullivan pers . comm ). . The canoes that are associated with Tamaki include Te Arawa, Aotea, Tainui and Mataatua (Simmons 1987) . Of these, Tainui has the strongest association . Tainui originally made land- fall near East Cape and journeyed up the east coast of the North Island, making landfall near Devonport . It then ventured up the and was dragged across the Tauoma portage at to the Manukau from where it travelled down to Kawhia . Some of the new arriv- als settled and intermarried with people already in residence so that from the earliest times the people of Tamaki and Manukau were closely related to Tainui (Waitangi Tribunal 1985: 10) . Eventually Tainui spread from their base in Kawhia to the Waikato, Hauraki and back to Tamaki . People continued to move around the North Island in relative peace, with groups from Waikato and the Bay of Plenty coming to live at Tamaki, and people from Tamaki moving to the Bay of Islands . The population of Tamaki in the 17th and mid-18th centuries is described as high compared to the 1820s and 30s when endemic warfare resulted in the virtual desertion of the region . At most stages, however, there was low-level conflict resulting from internal dissen- sion and external raids, occasionally escalating into more serious war (Sullivan n d. ). . Around this time the Kaipara chief Hauparoa urged Maki of Ngati Awa to conquer lands to both their profit . Maki subsequently received permission to settle in Tamaki where he became involved in local feuding, eventually siding with the people of Takapuna, attacking and defeat- ing his Tamaki hosts . This was only a short-lived conquest and his dominance seems to have been unsustainable . Remembering Hauparoa’s words he then moved on to conquer lands from the Manukau to the Kaipara . This was a less peaceful episode than the migrations of Ngati Awa some two centuries earlier and involved the conquest of most of Nga Oho and their incorpora- tion into Te Kawerau a Maki (Murdoch 1990: 12; Sullivan n d. ). . By the 18th century, then, the peoples of Tamaki were principally Te Wai o Hua and Te Kawerau a Maki . Ngai Tai, of Tainui origin, also lived at Tamaki and had Te Wai o Hua connections . In the late 18th century the scale of conflict and severity of its outcomes seems to have increased markedly in Tamaki . This increase in warfare, however, may be more apparent than real – there is more detailed information available which colours the history differently . The most important and reliable source is a manuscript of Tuhaere held in the Auckland Public Library, probably composed around 1868 as a memorandum for the Orakei Native Land Court case . Tuhaere was a tohunga whose manuscript survives in his own handwriting and in his own Background 5

language . Other surviving manuscripts give less complete accounts but do not substantially con- tradict Tuhaere’s history (Agnes Sullivan pers . comm . 2010) . The most accessible and complete modern accounts are Ballara (2003) and Stone (2001) . In the 18th century Ngati Whatua completed their migration from the far north to Kaitaia and Kaipara where they encountered Te Kawerau a Maki . In their battles against Te Kawerau they were led by Kawharu, who was of Ngati Whatua and Tainui descent and pre- viously lived at Aotea Harbour . Te Kawerau were badly beaten in the battles known as Te Raupatu Tihore, ‘the stripping conquest’, and all their pa in the Waitakeres are said to have been destroyed (Smith 1898a: 68) . Under Kawharu, Ngati Whatua attacked Te Wai o Hua in Tamaki and took various smaller pa but could not take Maungawhau / Mt Eden or Maungakiekie / One Tree Hill . Kawharu returned to Kaipara where he was later killed . Te Kawerau’s rohe shrank through the 18th century as various hapu merged with other groups through intermar- riage and conquest, so that only those at Waitakere and Mahurangi retained the name . When Ngati Whatua later conquered Tamaki they left Te Kawerau alone in Waitakere due to mar- riage relationships between the two iwi . Te Kawerau were by then a junior ally of Te Wai o Hua (Simmons 1987: 29; Murdoch 1990: 14; Stone 2001: 22) . In the late 18th century2 Te Wai o Hua, under their great ariki Kiwi Tamaki, occu- pied Tamaki / Manuka (Manukau) . At this time Te Taou were becoming established in the Kaipara, intermarrying with Nga Oho and warring with Ngati Whatua and their allies who were attempting to drive them out . Te Taou quarrelled with a chief to the south of Kaipara, Te Raraku, over a boundary . Te Raraku sent tokens of war to Kiwi who gathered a taua3 of more than 1000 men . After Kiwi’s surprise attack on Te Taou at Waituoro, Te Taou moved north to Okahukura in the north Kaipara for several years . Afterwards Te Taou obtained utu4 from Te Wai o Hua at the pa at Taurere / and sacked the whole of Tauoma (the wider Panmure area) . Kiwi then attacked the people of Mimihanui near Helensville, killing men of Ngati Whatua (Stone 2001: 40–45; Ballara 2003: 206–209) . Te Taou then raised a taua under Te Wahi Akiaki and sacked Tarataua, a pa of Te Wai o Hua on the Awhitu Peninsula and then attacked Te Wai o Hua at Titirangi . Kiwi raised a taua of 3000 or 4000 men and attacked Te Taou as they neared Te Whau . Te Wahi Akiaki led his taua (Tuhaere’s manuscript says there were only “120 fighters but they were all champions”) in a feigned retreat uphill toward Titirangi and, once they reached the ridge top, turned and broke the Wai o Hua advance . Te Wai o Hua fled down to the Manukau to Paruroa where Kiwi was killed and Te Wai o Hua were comprehensively routed . Kiwi’s remains were taken to Kaipara and many Te Wai o Hua fled to the Waikato . Te Taou then took Tamaki as their own, but even- tually some returned to the Kaipara and Te Wai o Hua began to return home, occupying Orakei and . Ngati Whatua then decided to seek utu for their kin formerly killed by Kiwi, sailing down the Waitemata to surprise and defeat Te Wai o Hua at Kohimarama . Te Wahi Akiaki also attacked Puketutu pa and besieged the Te Wai o Hua inhabitants who attempted to escape to Mangere pa under the cover of darkness . Te Taou attacked them

2 Fenton (1879 [1994]: 62) places the beginning of events described here in 1740, and most commentators since have accepted this, but Ballara (2003: 493) argues convincingly that the date for the conquest of Tamaki would have been in the 1780s or 90s . 3 Armed war party or expedition (Ballara 2003: 17) . 4 Utu is often glossed in English as ‘revenge’, but in fact it refers to reciprocity, like-for-like, repayment of debt, re-balancing of obligations . Utu can refer to reciprocal gifting as much as reciprocal warfare . Utu in the con- text of war meant balancing an offence against the mana of a person or group . In war this was not for vengeance, but in order to achieve a balance and resolve the take, or proximate cause of action . If the take was the taking of a life then the balance was often achieved through the reciprocal taking of life (Ballara 2003: Chapter 6) . 6 Torpedo Bay

on the way with great loss to both sides though Te Taou were driven off . This battle probably occurred after the sack of Tarataua and before Paruroa . Te Taou then returned to the Kaipara and further Te Wai o Hua were allowed to return to Tamaki where they intermarried with Te Taou . The next generation became known as Nga Oho (their relationship to the former hapu of the same name is not entirely clear) and Te Uringutu . Nga Oho lived at Maungakiekie, Mangere and Ihumatao, further marrying with people from the Waikato . Nga Oho / Te Wai o Hua remained in control of Mangere until it was conquered in the later 18th century by Ngati Whatua and Te Wai o Hua fled south to their Tainui relatives . (Fairfield 1938: 126; Tonson 1966; Ballara 2003: 211) . At the same time various Hauraki tribes were pressing claims in Tamaki, conquering and occupying the North Shore and offshore islands, and sacking and Maungawhau . Conflict soon arose between Ngati Whatua and Hauraki as the latter spread along the west- ern shore of the but in two important battles Hauraki were defeated and Ngati Whatua established their hegemony on the isthmus, with Hauraki maintaining a presence in Panmure . They remained there until driven out by musket armed Nga Puhi in the early 19th century (Simmons 1987: 30; Stone 2001: 49–52) . By this time the northern Kaipara was a battleground between the confederations of Nga Puhi and Ngati Whatua . In about 1807 a battle was fought at Moremonui between Nga Puhi and Te Roroa, supported by the Ngati Whatua confederation under Murupaenga, where Nga Puhi were defeated, losing several chiefs . This battle was known as Te Kai a te Karoro (the seagull’s feast) . From 1814 Nga Puhi under Hongi Hika began to acquire muskets and monop- olise contact with traders and missionaries in the Bay of Islands . In contrast, Kaipara Maori had little interaction with Pakeha until the 1830s (Waitangi Tribunal 2006: 15) . Among those killed at Te Kai a te Karoro were Hongi’s brother Houwaea and his sister Waitapu (Waitangi Tribunal 2006: 45) . In 1825 Hongi sought utu and defeated the Ngati Whatua confederation in a series of battles around Kaiwaka known as Te Ika a Ranganui . The survivors fled in several directions: some up the Kaihu valley to Te Roroa, some to Te Parawhau in the northern Wairoa (although Te Parawhau had fought with Nga Puhi) while others retreated south where they met a taua of their kin from Tamaki who had arrived too late for the battle . Together they fled to the Waikato though even here they were subject to Nga Puhi attacks . For around 10 years Tamaki and much of the southern Kaipara remained only very sparsely inhabited, with the few remaining behind living in fear of further Nga Puhi raids . Ngati Whatua only began returning in the mid-1830s . Nga Puhi did not occupy the Kaipara lands after Te Ika a Ranganui . Ulrich Cloher (2003: 33) points out that the desire, or necessity, for utu for his brother and sister had been the driving force in Hongi’s campaigns up to this time and Te Ika a Ranganui was the climactic event in this process . He had no need to establish a take raupatu (title by conquest) in the Kaipara and some Ngati Whatua were able to remain behind and maintain their ahi ka 5. Eventually they were able to return, so that when Lieutenant–Govenor Hobson arrived in Auckland in 1840 Ngati Whatua were back in occupation . This relationship between Tainui and the peoples of Tamaki and the Manukau was strengthened after Te Ika a Ranganui as the latter fled to the Waikato, becoming more closely tied to Tainui by marriage and more closely integrated into the wider Tainui–Waikato confeder- ation . Inland Tainui had customary rights on the Manukau, and presumably Manukau peoples had reciprocal rights elsewhere in the Tainui rohe (Waitangi Tribunal 1985: 11) . The European

5 Burning fires as a symbol of right of occupation . Background 7

settlement of Auckland in 1840 was essentially under the protection of Tainui rangatrira Te Wherowhero (Stone 2001: 136) . The traditional history of Tamaki and Manukau is confusing and contradictory and no attempt is made to make a formal, European historical sense of it . The conquests by Te Taou, Ngati Whatua and Nga Puhi had the effect of erasing much of the earlier history and what remained is only a broken remnant, lacking internal consistency and external corroboration (Stone 2001: 12) . Named groups (iwi and hapu) came and went through processes of budding off, assimilation through marriage and alliance, or conquest . The history of the region is one of movement and fluidity (this is true of the pre-European history of Aotearoa in general) . Connections between Tamaki and the north, particularly Kaipara, and the south, particularly Tainui and Waikato, were one constant among all this movement and it is often the case that the groups described as interacting in various ways were closely related .

A Maori history of Devonport

The following account of Maori traditional history is taken from research by Tony Packington-Hall, recorded in the site record form for site R11/2401, the Te Haukapua kainga, and Graham (1924) . Traditional accounts record Torpedo Bay as one of the landing places of the Tainui waka commanded by Hoturoa and as an area that was occupied from the earliest period of Maori settlement . Maki’s occupation of Takapuna is described in the previous section and the invasions of Ngati Whatua, Te Taou and Hauraki iwi affected the North Shore as well as the Tamaki isthmus . Following these events, or perhaps as part of them, Kapetaua, a chief of Ngati Paoa, assembled a taua from the Hauraki tribes and attacked various pa around the Waitemata Harbour to avenge an insult suffered in his youth . Graham (1924: 9) says this occured around AD 1650 but he has it following Te Raupatu Tihore in which case it must have been in the early 18th century; Graham’s timeline is rather confusing . During this campaign, the pa sites around Te Haukapua (Devonport), including Maungauika / North Head, were devastated and many people were killed . In the ensuing decades, the survivors intermarried extensively with Kapetaua’s people . Graham (1924: 9) suggests that the Takapuna–Devonport area was abandoned after this as there is no further reference to the pa at Te Haukapua . However, another tradition (Packington- Hall dates this to around AD 1690) mentions a chief called Taihua from Takapuna, suggesting that the area may have continued to be occupied . After this the next record is the occupation and fortification of Maungauika by the Ngati Paoa chief Te Rangikaketu around 1790 . Around this time a dispute between Ngati Paoa and Ngati Whatua over fishing rights at Mahurangi escalated into war . Ngati Whatua were defeated leaving Ngati Paoa in control of this part of the North Shore (Graham 1924: 10) . Further disputes over Mahurangi followed between Ngati Paoa and Nga Puhi, with Ngati Paoa eventually forced out of Maungauika which was briefly occupied by Nga Puhi . This campaign included a battle at Torpedo Bay . Ngati Paoa do not appear to have reoccupied the Devonport area . In 1827 Dumont D’Urville landed in Devonport Bay and climbed Takuranga / Mt Victoria . He observed no smoke from any cooking fires or other signs of habitation, indicating the area remained largely unpopulated into the early 19th century . Graham (1924: 10) suggests that the area was too vulnerable to attack and too hard to defend in a time of constant unrest, but much of Tamaki and Hauraki was abandoned at this time . 8 Torpedo Bay

The archaeology of Aotearoa / New Zealand

The Torpedo Bay excavation demonstrated two phases of pre-European Maori occupation: Phase 1 dating to the 14th century, and Phase 2 dating to the late 15th to mid-17th (Chapter 3) . While sites with dates similar to the Phase 2 are common in Tamaki, Phase 1 was, at the time of excavation, unique in the built up part of Auckland City 6. This is a highly significant find . Other early sites in Auckland are largely restricted to the Hauraki Gulf islands . This chapter provides the environmental and archaeological context for the Torpedo Bay excavations with particular emphasis on the time to which the lower phase is dated .

The idea of the ‘Archaic’

Two phases or periods of pre-European Maori archaeology have been recognised since the 19th century . Haast (1871) proposed that the earliest inhabitants of New Zealand were a race of “moa hunters”, recognisable as a Palaeolithic people who used flake stone tools and were respon- sible for the of the moa . This race was then followed by the Neolithic Maori whose migration to New Zealand bought kumara horticulture . This distinction between Moa-hunter and Maori mirrored the relatively recent recognition of the Palaeolithic in Europe, where peo- ples with flaked stone tools hunted fauna that had since become locally extinct (reindeer, rhinoc- eros, elephant, etc ). and were then supplanted by Neolithic agriculturalists . Subsequent recon- structions of Maori origins based on selective readings of oral tradition proposed that the first inhabitants, the Maruiwi or Moriori, were a peaceful Melanesian race who were subsequently displaced by the war-like Polynesian Maori of the Great Fleet, and who introduced kumara horticulture (e g. ,. Smith 1898b) . Archaeology and ethnography, as they were then understood, appeared to be in agreement . Roger Duff (1947) proposed that the differences between the Moa-hunters and the agri- cultural Maori were due to two waves of migration from Polynesia: the first from tropical East Polynesia by a people who had originated in Micronesia but lost their domesticated (with the exception of dogs, as the archaeological record demonstrated) and food plants on the way (citing Te Rangi Hiroa) and who became the Moa-hunters; and a second wave of people who followed once domesticates were reintroduced to East Polynesia from West Polynesia, and kumara from South America, and who became the Maori . Both migrations for Duff, then, were from East Polynesia and both used ground stone tools, a marker of the Neolithic, but the first wave of migrants had no agriculture and lived by hunting . They had no fortifications and so “one suspects that the Moa-hunters lived at peace” (Duff 1947: 283) . The second wave “doubtless from their mana as introducers of the taro and kumara … exercised an influence out of all proportion to their numbers and, like the Normans in England, rapidly founded a new hierarchy of tribes and tribal power” (Duff 1947: 282) . He subsequently modified this simple picture somewhat, allowing for multiple migrations to the North Island “right down to the great Fleet migration of 1350 A .D ”. (Duff 1977: 246), some of which introduced root crops and were a stimulus for the development of a North Island culture, while the South Island culture remained conservative . He concluded: “the Fleet arrival in 1350 did not so much introduce the new culture which we know as Maori culture as mark the final cessation of contact with Polynesia, and accelerate the already well marked trend of local change” (Duff 1977: 246) . However, he also hypothesised that

6 Since then two sites of similar age have been excavated in Auckland: The Masonic Tavern (R11/2517), 500 m west of Torpedo Bay in Devonport, but not yet fully reported (Russell Gibb pers . comm . 10 June 2017); and the Long Bay Restaurant (R10/1374) which, while not in a developed part of the city, is on the mainland and within the urban boundary, and is also not yet fully reported . Background 9

Maori culture was an indigenous development of the Moa-hunter culture (Duff 1977; see also Allen 1987) – his reliance on continued migration was clearly at odds with this view and he was unable to reconcile the two . Golson (1959b) recognised that Duff’s differentiation of Moa-hunter from Classic Maori implied a false dichotomy between early hunters and later horticulturalists, and that the conse- quent change in technology and lifestyles bought in by new immigrants was not supported by the archaeological evidence; particularly that horticulture had been present from the beginnings of human settlement . He noted that early archaeological sites often contain a very different suite of artefacts as well as evidence of moa butchery and consumption . Golson proposed that the early period, from the first settlement of New Zealand by people from tropical East Polynesia to around AD 1450, be referred to as the Archaic Phase . The Archaic was followed by the Classic Phase . The change between the two was technological, the two periods were defined by arte- fact typology rather than economy (although ultimately Duff’s two stages were also defined by changes in artefact types, Groube 1967: 8), and there was only a single migration with subse- quent cultural development occurring in isolation (Allen 1987: 17) . Green (1963a), more directly influenced by the North American tradition, proposed a more complex sequence of five stages for the upper North Island based on changes in settle- ment pattern: Settlement, Developmental, Experimental, Proto-Maori and Maori . Green’s five phase model was more explicitly evolutionary in nature (Groube 1967); the Settlement and in particular the Development Phases (equivalent to Golson’s Archaic) were where the first settlers from tropical East Polynesia went through the process of adapting to temperate New Zealand . Although the settlement types proposed by Green, evolving from highly mobile to nucleated settlement, have not been supported by subsequent archaeological evidence, his work introduced both the idea of cultural adaptation, and settlement pattern as a major research focus into New Zealand archaeology . While Haast (1871) proposed that the first inhabitants had a Palaeolithic level culture and Duff (1947, 1977) proposed that, while they had polished stone tools they lacked agriculture, in other words they had only some of the markers of the Neolithic, later commentators under- stood that both Archaic and Classic Maori had agriculture and were to be understood, insofar as cultural stages developed in Europe and the Near East can be universally applied to Polynesia, as Neolithic 7. The nature of the later period Classic Maori was, for both Duff and Golson, well under- stood and non-controversial – it was its antecedent and how it was derived from that antecedent that was at issue . Where 19th and early 20th century scholars had relied on oral tradition to outline Maori history, Duff and Golson relied on the evidence of economy and material cul- ture derived from archaeology, very much in line with then current international trends . Green (1975), acknowledging that horticulture was present in New Zealand from first settlement, re-emphasised economy as a major explanatory factor, alongside environment: New Zealand was heavily forested throughout when Polynesians first arrived and climatically unsuited to horticul- ture in much of the South Island . Changes in material culture were closely linked to economy and to Maori adaptation to the environment . Green, having by then abandoned his earlier five phase scheme, still accepted the division of pre-European archaeology into Archaic and Classic (he referred to the latter as Maori) but by now archaeologists were focussing on the idea of the cultural adaptation, in a quasi-evolutionary sense, of tropical Polynesians to the temperate envi-

7 They did lack the third classic marker of the Neolithic: pottery . Their Lapita antecedents in fact had pottery but it had been lost during the subsequent migration into Polynesia, where suitable clays were less common and food tended to be cooked in earth ovens (Irwin 1981) . 10 Torpedo Bay

ronment of New Zealand as an explanation for both the Archaic and the subsequent develop- ment of the Classic in isolation . The stage models of Duff, Golson and Green were essentially ‘culture-historical’ models, strongly influenced (though indirectly in Duff’s case) by similar models developed by V . Gordon Childe in Europe and Gordon Willey and Philip Phillips in North America (Groube 1967: 12; Allen 1987: 15) (the terms Archaic and Classic were derived from Mesoamerican Archaeology and ultimately from Classical Greek studies) . These models, insofar as they allowed for gradual cultural development rather than sudden replacement, were also evolutionary . An early critique from Groube (1967) pointed out that Duff’s model, where a Moa-hunter period, or stage, was followed by a Maori period (with some ill-defined transitional period between) was not an evolutionary model where a culture moved from simple to complex, like the familiar Stone Age ~ Bronze Age ~ Iron Age model for instance, though Duff’s (and Haast’s) thinking was based on these kinds of model . Golson’s subsequent Archaic ~ Classic model was, for Groube, an improvement on Duff’s model; in particular his use of the term ‘phase’ didn’t imply an evolutionary stage-based model . Instead, for Groube, the two phases of Duff and Golson could be incorporated into a single cultural-evolutionary stage, with limited evidence of, or need for, evolution . One problem with evolutionary models is that they rely on an analogy with biological evolution, where organisms evolve in response to environmental changes . Despite the devel- opment of more sophisticated approaches to evolution within cultural contexts (e g. ,. Shennan 2002) the evolutionary approach to New Zealand archaeology remains focussed on adaptation to the environment: the adaptation of tropical Polynesian culture to temperate New Zealand; and subsequent adaptation to environmental changes such as climate change (Anderson 2016), even if some environmental changes such as deforestation, siltation and extinction are human induced (Anderson et al . 2014: 30) .

First settlement: long vs short chronology

The date of the first settlement of New Zealand by canoe borne colonists from East Polynesia is not yet fully settled . While some proposed a pre-AD 500 date (Sutton 1987), based on no actual evidence or radiocarbon dates, this is no longer accepted by archaeologists . Anderson (1991) examined the available early radiocarbon dates for New Zealand and applied a process of “chronometric hygiene” to determine which were reliable on the basis of criteria such as the material dated, the context it came from, etc ,. and determined that there were no reliable dates prior to the 12th century AD . In fact, the date of settlement is probably later . Wilmshurst and Higham (2004) dated the seed cases of miro (Prumnopitys ferruginea) and other species that had been gnawed by kiore (Polynesian rat, Rattus exulans), using the arrival of kiore as a proxy for the arrival of humans . They concluded that the first evidence of kiore dated to the late 13th or early 14th century . Some of these early seeds were present in the Kaharoa tephra, a volcanic ash that has been tightly dated to AD 1314 ± 12 (Hogg et al . 2003) . This tephra originated from an eruption of Mt Tarawera and is dispersed in a band over the eastern northern North Island with the thickest deposits from the Bay of Plenty to Gisborne . There is some evidence of human activity immediately preceding the Kaharoa tephra from pollen cores in swamps, where a sudden rise in charcoal and bracken spores indicates burning of the vegetation and subsequent colonisa- tion by bracken (Newnham et al . 1998: 540; Lowe et al . 2000: 865), which would again support a late 13th century date for human arrival . While these environmental indicators of human presence (kiore-gnawed seeds and pollen cores) make a sound argument for human arrival just prior to the deposition of the Kaharoa tephra it is notable that no archaeological evidence, that is Background 11

direct evidence of human presence, has been found beneath the tephra . Furey et al . (2008) have suggested that the natural layer overlying Layer 9 at the Cross Creek site (T10/399) is Kaharoa tephra, although this has not been confirmed by direct observation (it was not recognised as such during the initial 1983 excavations and no sample was retained) . Layer 9 dates suggest a human presence in the mid-13th to late 14th century . On balance, a date of first human arrival between AD 1280 and 1320 seems most likely; Walter et al . (2017) propose a post-1300 date . When New Zealand archaeologists accepted a settlement date of ca AD 900–1000, this gave a period of 500–600 years for a small founding population of ‘Archaic East Polynesian’ settlers to grow, adapt to life on a temperate continental landmass and by ca . 1450, develop into the Classic Maori encountered by Cook in 1769 . Accepting a shorter chronology has several implications that archaeologists are only now acknowledging . The numerous 14th century sites in New Zealand must have all been occupied nearly contemporaneously – there isn’t time for population growth off a low base to account for the observed pattern of occupation . Exploration of New Zealand and exploitation of its stone resources must have been rapid and systematic since all high quality stone sources were transported widely from an early date (Walter et al . 2010) . Walter et al . (2017) propose that initial colonisation was a well planned, large scale event, which they refer to as a “strategic migration model ”. Genetic evidence indicates a minimum of 190 females in the founding population and it is probable that this population contained more males than females . A founding population of roughly 500, arriving in several canoes around AD 1300, would reasonably account for a population of around 100,000 by the time of the Cook voyages as well as allowing a colonising community large enough to be viable . While this accepts Cook’s estimate of 100,000 as accurate (which might be regarded as dubious, since Cook was a sea captain, not a demographer) it does seem likely that the founding population would have been higher than a single canoe or just a few canoes could account for . A planned or strategic migration implies an initial discovery of New Zealand and a return voyage in order to inform the homeland that New Zealand existed . Polynesians had a sophisti- cated oceanic voyaging capability . Part of a canoe with ocean-going capabilities, with a sea turtle carved on it, recently discovered at Anaweka in the north west South Island is made from New Zealand wood (matai, Prumnopytis taxifolia) and caulked with the bark of New Zealand totara (Podocarpus totara) . The caulking is dated to cal AD 1313–1415 (Johns et al . 2014; Irwin et al . 2017) . The authors point out that the caulking may be younger than the hull . If canoes were con- structed in New Zealand from local trees, which are much bigger than tropical Polynesian trees, with two-way voyaging capabilities, as seems to be the case for the Anaweka canoe, this implies that any strategic migration may have involved repeat return voyages . Despite this, it is notable that no artefacts have been found in tropical Polynesia that can be sourced to New Zealand, and repeat voyages did not establish either the pig or chicken in New Zealand . Another implication of the Anaweka canoe and planned migration is the presence of a full complement of specialist among the first settlers . Some would have been specialist tool makers; not only were all the high quality stone sources found early but there were people with the skills to assess the stone for suitability and make very specific tools with it . Among these tools were adzes for making canoes, and there were also people with the skills to make vessels like the Anaweka canoe out of trees much larger than they were used to, allowing hulls to be carved in one piece or fewer pieces rather than lashed together from multiple pieces of wood . These were master craftsmen, able to readily adapt their knowledge of Polynesian stone sources or trees to unfamiliar materials (it is quite possible that specialist canoe makers also made their own stone tools) . Not only were the voyages planned, but also the settlement and maintenance of new com- munities with the full complement of skills necessary to reproduce a new society in a new land isolated from the homeland . 12 Torpedo Bay

Another implication of the short chronology is that culture change from the Archaic to the Classic was achieved very rapidly, within 200 years . Getting rid of a cultural and population replacement model means is that there is no easy transition between the Archaic and the Classic, especially when these phases are conceived in terms of material culture, where one set of arte- fact forms appears to abruptly replace another . It “leads to the assumption that all the important cultural changes in New Zealand occurred in the brief transitional interval between the ear- lier and later stage and not in a series of steps over the entire sequence” (Green and Shawcross 1962: 212) . Davidson (1984: Chapter 10) acknowledged “that changes from East Polynesian to Maori traits took place gradually and at different times and places ”. The opposition of Archaic and Classic (or Moa-hunter and Maori, etc ). had the effect of polarising New Zealand prehis- tory between two extremes, useful in highlighting the differences between the two ends of the sequence but obscuring the continuities and changes from one to another . Davidson’s solution was a three-part sequence, based on settlement patterns, economy and technology, that might differ regionally: a Settlement Period, from first settlement to AD 1200; a period of Expansion and Rapid Change, AD 1200–1500; and a Traditional Period, AD 1500–1769 . She was reluc- tant to provide date of first settlement but her Settlement Period still implies a long chronology . Anderson (2016) proposes that the transitional period between Archaic and Classic8 was marked by an expansion of population, movement into the interior and a marked increase in the extent and intensity of gardening, accompanied by changes in group size and an increased organisation for defence . This occurred around AD 1500 when, for instance, large-scale garden- ing commenced in the Waikato, peaking after AD 1650 (Campbell and Hudson 2013; Warren Gumbley pers . comm . 31 August 2017) . In the Western Bay of Plenty a similar movement away from a primary settlement focus on the Tauranga Harbour to the Papamoa lowlands and the inland valleys occurred slightly earlier, around AD 1450 (Campbell et al . 2009) . Pa construction also commenced around this time (Schmidt 1996) . Anderson proposes that a major trigger for these changes was climate change as the lower North Island and upper South Island became too cold to support kumara horticulture, accompanied by population growth . Adding an additional transitional stage, as Davidson (1984) and Anderson (2016) have proposed, puts an even greater strain on the short chronology . Groube (1967) had earlier pro- posed that a model of New Zealand prehistory based on stages was unnecessary and implied a cultural evolution that had not taken place . virtually every archæologist who uses stages to organise his data thereby builds into them certain assumptions about cultural development without being aware that he is doing so . Later in making his cultural interpretations, he discovers the pattern of cultural development which was assumed in his system of organisation and thinks he is deriving it empirically from the data . The argument becomes perfectly circular (Rowe, cited in Groube 1967: 8) . Duff’s view of cultural change was, for Groube (1967: 9) non-adaptive change, “clearly analogous to genetic drift [that failed] to demonstrate ‘improvement’ or increase in efficiency ”. To this type of change Groube opposed: ‘adaptive change’, where culture adapts directionally to changing environmental conditions; and ‘evolutionary change’ which is “self-transcending” (citing Huxley) with a clear change from one kind of socio-political organisation to another . The latter justifies the use of a stage model (e g. ,. Palaeolithic ~ Neolithic ~ Bronze Age; or Archaic ~ Classic) but as he pointed out (1967: 11):

8 Anderson refers to a middle period between the early and late periods; archaeologists have tended recently to replace the Archaic ~ Classic nomenclature (with capital letters) with early and late (uncapitalized) . Background 13

the first people who came here (East Polynesian) were a neolithic, fishing, agricul- tural people… When Cook came to these shores the New Zealand Maori were still a neolithic, fishing, agricultural people … demonstrably there was no change in the economic status of the people, although practically all of their items of material culture, possibly their art styles, and probably their social organization were trans- formed from that of the first migrants . No evolutionary change, as he defined it, had taken place and a true stage model could not be applied . Groube proposed that the proper model for pre-European New Zealand archaeology was one that depended on adaptive change and incorporated the period from first settlement to the arrival of Europeans in a single stage . He proposed what he called the “strophic model”9 that “emphasizes the points of change rather than the platforms of conservatism” (1967: 22) . Strophes were peaks in the rate of cultural change: in an evolutionary strophic model the rate of change varied but on average was ever-increasing until a distinct new level of social, political, economic and technological organisation was reached; in a non-evolutionary strophic model the rate of change returns to a low level between specific peaks of change, in other words, adaptive change, directional but not ‘self-transcending ’. The usefulness of this model in New Zealand prehistory is that it does not require defined stages such as Archaic and Classic . Since Duff’s recognition that horticulture was not bought to New Zealand by a wave of later immigrants (Duff 1947) the two-stage model developed in the 19th century (Haast 1871) has been an awkward fit, but attempts to remedy it have consisted of either redefining the stages (Duff 1947; Golson 1959b) or adding a third, transitional stage (Davidson 1984; Anderson 2016) . Groube did away with separate stages altogether and fitted all change within a single stage . However, apart from his seeming invention of a terminology (strophic) that has not been subsequently picked up by other archaeologists, he neither specifies how the rate of cultural change could be measured, nor does he provide any examples of strophes from the pre-European New Zealand (or any other) archaeological record . It is productive to revisit Groube’s (1967) strophic model here, though rather than stro- phes, it is more useful to think of these archaeological visible things as events . The first of these events was the settlement of New Zealand by people originating in Central East Polynesia around AD 1280–1320 . This is an event of considerable significance, and where it is visible in the archaeological record, it is highly visible due to the presence of megafaunal remains and the wealth of distinctive material culture items . It is not, however, a signal of a particular ‘Archaic’ or ‘early’ culture; it was followed by a series of other events through which we can trace a con- tinued process of cultural change . One was the extinction of the moa, an event that occurred at different times in different places and, given the regional variation in the economic importance of megafauna, also had different impacts in different places – moa were never as economically significant in the warmer, horticultural north as they were in the south, where their population densities were significantly higher and horticulture was not possible . Another highly significant event was the beginning of pa construction . Schmidt (1996, 2000b) proposes, broadly, that moa exploitation ceased in the North Island by AD 1450 and pa construction began around AD 1500, while in the South Island both moa-hunting ceased and pa construction began around AD 1650 10. Other events were more local, such as the expansion of settlement from sheltered

9 From the Ancient Greek strophē, στροφή, a turn, bend or twist, which also relates to alternating syllabic rhythms in Greek poetry . 10 Holdaway and Jacomb (2000) have proposed that extinction occurred within a century, or even 50 years, of human settlement . In response, Anderson (2000) proposed that extinction occurred no earlier than the 15th 14 Torpedo Bay

coastal locations to open coast or inland localities . For instance, in the Western Bay of Plenty settlement extended east from an early focus on the Tauranga Harbour to the Papamoa dune plain, where it is marked primarily by shell middens (Felgate 2005; Campbell et al . 2009), and inland to the valleys that drain into the harbour, where it consists largely of pit storage sites, implying gardening though garden soils are rarely found (Kahotea 1983; Campbell and Harris 2012) . Occupation of the Oropi, Ohauiti and Kopurererua Valleys has been dated between AD 1450 and 1650, while the occupation of Papamoa began at the same time but continued for another hundred years, until around AD 1750 . Occupation of the valleys is assumed to have moved further inland as horticultural soils became depleted, though few sites have been excavated and dated further inland – Ruahihi Pa, located close to the bushline at the time of European contact, is dated to AD 1650–1800 (McFadgen and Sheppard 1984) . Occupation of the dune plain ceased with the historically well attested invasion of Tauranga by Ngaiterangi from the Eastern Bay of Plenty . These two contrasting events demonstrate that the process of ‘adaptation’ is a complex interplay of environmental and social factors, although the social fac- tors at play are more difficult to distinguish further back in time . Another significant event was the colonisation of the Waikato Basin, far from the coast and requiring specialised horticultural techniques (Gumbley et al . 2003; Campbell 2012) . These various events are often the result of events that are less visible archaeologically – the building of pa signals a response to a (hypothesised) phase of population growth which, coupled with environmental decline (for which there is good archaeological and palynologi- cal evidence) led to growing competition over resources and increased warfare . Archaeological cultures do not change at a constant rate (Groube 1967: 22) and it is the visibility of peaks of cultural activity that form the basis of archaeological models . This is Groube’s (1967: 22–25) ‘Strophic Model’ where the events in the archaeological record is equivalent to a strophe .

The archaeology of the first settlers

There is no doubt that the archaeology of the first East Polynesian settlers in New Zealand, from ca AD 1300 to perhaps 1450, is notably different from the archaeology of the Maori observed and documented by, for instance, Captain Cook in 1769, or by numerous mis- sionaries, traders and explorers from the early 19th century . Clearly there had been considerable change over the intervening 500 or so years but the Archaic ~ Classic model focusses attention on the two ends of the sequence and implies a single episode of dramatic change between the two . The archaeology of the 14th century11 and of the first settlers in New Zealand is undoubt- edly important . It tells the founding story of the nation but in many respects the first settlers are not markedly different to later Maori . Subsuming pre-European Maori archaeology into a single phase allows us to think in terms of more gradual cultural change, and allows us to look for the rhythms and patterns of this change . However, it is beyond the scope of this report to undertake such a project . Instead, this section provides a conventional thematic outline of pre-European Maori archaeology, beginning with the early 14th century sites that demonstrate first settlement, century, but he relied on a settlement date in the late 12th or early 13th century . Accepting a settlement date of around AD 1300, the early 15th century is within about a century of settlement . Moa may have been functionally extinct by ca AD 1400–1450, that is, if any survived their populations were no longer sustainable and they were no longer economically significant to Maori . 11 This is a ‘long’ 14th century running from, perhaps, AD 1280 to ca AD 1450 – some of the sites dis- cussed here are tightly dated to the 14th century but others have dates that span into the mid-15th century . The archaeology of this approximately 150 year period provides our insights into the initial settlement of New Zealand and is also the period of moa hunting and the ‘Archaic’ suite of artefacts . That there are changes around the end of this period is clear but, following Groube (1967), it is preferable to see these changes as occurring within a single period of pre-European Maori occupation rather than as a phase change between Archaic and Classic . Background 15

rapid acclimatisation of tropical Polynesian lifeways to temperate New Zealand conditions and the exploitation of abundant natural resources . This is the period to which the lower layers at Torpedo Bay belong . The timing of first settlement has already been discussed, and there is no site that demon- strates a ‘first footprint’, and very little material from East Polynesia has been identified: most famously, a tropical pearl shell (Pinctada margaritifera) lure shank was excavated at Tairua by Roger Green in 1964 (Smart and Green 1962; Green 1967a); while a shell tool from Wairau Bar, excavated by Jim Eyles in 1946, has been recently identified as tropical Acus crenulatus (Davidson et al . 2011) . Otherwise, any artefacts that have been identified to source are all from New Zealand sources, indicating that the first settlers familiarised themselves with the available stone, bone and shell resources very quickly . The first settlers would have been met with an abundance of terrestrial and marine resources in a pristine environment . As Haast (1871) noted in the 19th century, this included the moa as well as several other large, flightless bird species that have since become extinct, along with the species that Maori continued to exploit . Moa were more commonly exploited in the South Island, where the open country of Otago and Canterbury favoured larger populations and extensive sites developed at river mouths . In the North Island moa were exploited but not on the same scale – at least three species of moa are represented at Torpedo Bay, though in small num- bers (but see discussion of moa identificatons in Chapter 5) . Moa bone was an important indus- trial material, used to make tools and ornaments . Sea mammals, particularly various seal species, were also exploited – small quantities of sea lion (rāpoka, Phocarctos sp ). and fur seal (kekeno, Arctocephalus forsteri) were found at Torpedo Bay . Other mammals exploited included the com- mensal dog (kuri, Canis familiaris) and kiore, both of which are found at Torpedo Bay (Chapter 5) . In much of the North Island fishbone assemblages are dominated by snapper (tāmure, Chrysophrys auratus) while in relatively sheltered waters on the east coast of the South Island a specialised barracouta (makā, Thyrsites atun) fishery developed at an early date (Leach 2006; Anderson 1981), although other species were also commonly taken with baited hooks, lures, nets and spears . Shellfish were also taken, with a notable preference for large rocky shore species like oyster and mussel . While bone and shell preserve best in archaeological contexts, wild and cultivated plants would have formed an essential component of the diet . Horticulture has been present in New Zealand from the beginnings of human settlement when the East Polynesian colonists bought several tropical crops with them: taro (Colocasia esculenta), yam (uwhi, Dioscorea spp .), gourd (hue, Lagenaria siceraria), paper mulberry (aute, Broussonetia papyrifera) and most suc- cessfully kumara (Ipomoea batatas), which became a staple in warmer parts of the country . If the North Island lacked large moa populations, this was compensated for by favourable climates for horticulture . In the tropics kumara can be grown year round and even in frost-free subtropical climates it can overwinter in the ground, but in New Zealand it must be harvested before the first frosts . While kumara may have initially been grown in frost-free areas of the upper North Island (Yen 1961), such conditions cannot be guaranteed and the development of propagation and storage techniques must have been rapid . Leach (1984: 58) suggests that root propaga- tion and storage techniques of yams in the tropics could be readily adapted to kumara in New Zealand . Storage pits have been found in association with early sites, though few have been reliably or directly dated . Simple sub-rectangular or oval pits at Hahei had firescoops overlying them that dated to the 14th to mid-15th centuries (Harsant 1983) . Pits at Moikau Valley and Palliser Bay are simple semi-subterranean structures (Leach 1979) but those at Skippers Ridge on the from a proposed early date are quite complex (Davidson 1984: 16 Torpedo Bay

123, see discussion of these pits in Chapter 7) . Pits seem to be uncommon in early sites but this may in part be due to a self-fulfilling assumption (see quotation from Rowe, above) that ‘Archaic’ sites are generally coastal hunting and fishing sites rather than horticultural sites . Pits must have been present from an early date . Early sites often contain rich and distinctive artefact assemblages, in fact it was this that Golson (1959b) used to define the Archaic . Flakes are often abundant, and adzes, fishing gear and the by-products of their manufacture and repair are common . The one-piece ‘U’ shaped fish hook made in moa bone, sea mammal bone, ivory or occasionally shell (usually Cook’s turban, Cookia sulcata) was the main form, though composite two-piece trolling hooks, often with stone shanks, were not uncommon . These shanks are themselves distinctively early (Furey 2004: 39) . Wood-working tools (adzes) were highly specialized in the early period, and were designed predominantly to make the complex deep hulled sailing canoes of the kind that ena- bled East Polynesians to make successful landfall here . There was a variety of shapes and sizes and adzes often had tangs to facilitate attachment to wooden hafts . They were shaped using a flaking technology, and high quality, fine-grained stone was required to make them . Few rocks met this need and adzes often had to be imported over long distances from a small number of quarries (Best 1975; Turner 2000) . Personal ornaments were distinctive . Reel necklace units in ivory, shaped whale teeth or imitation whale teeth, and disc pendants with elaborate surface treatment of relatively soft Nelson serpentine are all assumed to be early (Golson 1959b; Prickett 1999, 2007), though many are of uncertain and undated provenance . These various early arte- fact forms derive from clear East Polynesian precedents (Furey 2004: 38) . Distinctive chevroned pendants in whale ivory or bone lack clear East Polynesian antecedents and may be a late 14th or 15th century innovation (Prickett 1999); the only example from an archaeological excavation is a fragment from Kahukura (G47/128), Foveaux Strait, which is dated to the mid-14th to mid-15th centuries (Brooks et al . 2010; Jacomb et al . 2010) . As Groube (1969: 1) points out, changes in artefact assemblages are often “quantitative rather than qualitative”; for instance, one-piece hooks are present in Classic assemblages and two-piece hooks in Archaic assemblages but the proportions change . This is probably due as much to the loss of moa bone as a raw material for one-piece hooks as anything else . Similarly, the loss of several distinctive Archaic adze forms that were made by flaking is a response to changing types of raw material – pounamu was suited to sawing and polishing while greywacke was suited to hammer dressing and polishing – and because there was no longer a need for specialised adze forms used to make ocean-going canoes . These apparent changes in the artefact assemblage are a response to local conditions and changes in social relations but are not evidence of the development of a new culture . Also typical of early sites is the range of sources of stone material found at them . Adzes and adze flakes from several different sources are usually present even when the site is located close to one particular source . from Tahanga on the Coromandel Peninsula and argil- lite from Nelson–Marlborough and Riverton, Southland are widely distributed at this time . The same is true of obsidian; while Tuhua / Mayor Island obsidian often dominates, obsidian from several other sources is usually present (Walter et al . 2010) . This suggests that most good quality stone sources were found very quickly; that either access to them was not tightly con- trolled or that there were extensive trade networks; and that people, perhaps including specialist adze manufacturers, continued to travel widely . The presence of side-hafted adzes designed for making ocean-going sailing canoes suggests that people continued to engage in long-distance travel . While Green (1963a) had proposed a change in settlement patterns from highly mobile to settled villages as part of his five phase model of pre-European Maori history in northern New Background 17

Zealand, kainga (villages) with elaborately carved wharenui and pataka storehouses appear to have developed as a response to the presence of Europeans in the 19th century (Groube 1967: 17; Sissons 2010; Holdaway and Wallace 2013) and pre-European settlement patterns can be characterised by seasonal mobility . The best developed regional synthesis of early settlement patterns remains Andersons’s (1982) examination of economy and settlement in southern New Zealand . Early multi-function sites (such as the large, well-known sites at Shag River or Waitaki River mouths) served as bases for travel to inland hunting and stone exploitation sites, as well as specialist coastal sites nearby . As big game – moa and seals – became less common, settle- ment switched to sites with more restricted function, centred on the exploitation of small game, fish and shell fish . It isn’t clear that this settlement pattern is replicated in the North Island as few large scale early sites have been well excavated and reported . Furey (2002) proposes that Houhora in the far north was a permanent village, occupied year round with an emphasis not on moa but on horticulture and exploitation of marine resources from which people would have moved out to temporary gardens, stone sources and fishing camps . However, there is no simi- lar base camp for the well studied, small scale early sites of the Coromandel Peninsula, which are generally associated with obsidian sources, or the Western Bay of Plenty . It may be that the settlement pattern was different here, or that any base camp has been destroyed or is yet to be found . The same can be said of Auckland and the Hauraki Gulf, though fewer sites have been studied in any detail . Overall, the impression of life at this time was one where there were plenty of resources to go around and little competition for them – in fact the first settlers lived in an age of free lunches . People occupied prime sites, often near river mouths, that provided ready access to resources and communication routes . Duff’s (1947) proposition that they lived at peace is prob- ably generally true, and people lived in undefended settlements on a permanent or semi-perma- nent basis, in what Anderson and Smith (1996) refer to as transient villages . Excavated early sites in the upper North Island, providing context for the early phase of occupation from the Torpedo Bay excavation, are discussed in greater detail in Chapter 7 .

The archaeology of later pre-European Maori

This situation, of free lunches, relative peace and semi-settled life, did not last . Major change is evident from around AD 1450–1500, though whether all the changes seen archaeo- logically in pre-European Maori culture occurred all at once, and had the same root causes, is unclear . It is very often assumed that this is the case, that the change from Archaic to Classic was an all-encompassing phase change . As we have seen, attempts to introduce a third temporal period acknowledge that this transition may have taken some time; Anderson (2016) suggests a transitional period from AD 1450–1650 . However, by accepting Groube’s assertion that there are no separate periods in pre-European Maori archaeology, it becomes possible to look for both longer term-patterns in the archaeological record without being constrained by the Archaic ~ Classic model, and to propose that these changes are, firstly, not necessarily directly con- nected; secondly, that they occurred at different rates in different places; and thirdly, that change occurred throughout the sequence and not just in a transitional period between two phases that by definition saw little or no change themselves . A fully developed model is beyond the scope of this report, but this section will briefly conclude by looking at a couple of projects that provide an outline of what is possible . One of the first and most obvious economic changes for early Maori, and one that was undoubtedly a cause of some of other changes visible in the archaeology, though not neces- sarily a primary driver, was the extinction of the moa and other large flightless birds, and the 18 Torpedo Bay

contraction in the range of seals from throughout New Zealand to the southern South Island . These were the free lunches, that were no longer available . Small birds continued to be exploited along with dog and kiore as well as fish and shellfish . Fish catches did not change markedly (but see discussion of Smith and James-Lee 2010, below), with snapper still dominant in many North Island sites and barracouta in the South Island . However, as settlement patterns changed and people began to rely more heavily on different coastal environments species like mackerel (Trachurus sp ). begin to dominate the assemblages from sites on open coasts in the North Island (Campbell et al . 2009) . It has often been suggested that there is a notable change in the shellfish exploited, with an early focus of rocky shore species (mussels and oysters) and a later focus on soft shore species, particularly tuangi (Austrovenus stutchburyi), pipi (Paphies australis) and tuatua (Paphies subtriangulata) . This change may be more apparent than real, however: Smart and Green (1962: 255) proposed that there was a clear shift from rocky shore to soft shore species at Tairua, but later analysis showed that shellfish were taken from both rocky and soft shore environments throughout the sequence (Davidson 1979: 194) and the work of Smith and James-Lee (2010) shows that this is generally the pattern . It is more probably the case that rocky shore species dominate early sites because mussels and oysters grow relatively large and are easily harvested but were quickly overexploited, while soft shore populations were more sustainable . Another factor that is often seen as a driver for cultural change was population growth which, accompanied by the loss of large prey species, put increasing pressure on resources and lead to several interlinked changes in economy and settlement pattern . The first of these was an increasing reliance on kumara horticulture and an accompanying population decrease in those areas of the South Island that were too cold to grow kumara . In the North Island populations moved away or expanded out from their early favoured coastal locations to good gardening soils in inland situations . From around AD 1450 settlement around Tauranga expanded east into the Papamoa dune plains and south to inland valleys . There was some gardening at Papamoa while the inland valleys were the focus of intensive gardening, with numerous pit storage sites recorded and excavated (Campbell et al . 2009; Campbell and Harris 2012) . At around AD 1500 there was a similar movement of horticulturalists from Kawhia to the inland Waikato Basin, with a peak of activity around 1650 (Campbell 2012; Anderson 2016; Warren Gumbley pers . comm . 31 August 2017) . The volcanic tephras being exploited in inland Tauranga and the Waikato were ideal for kumara horticulture . Another consequence of pressure on resources was a social change where warfare became more common . Settlement moved away from a costal focus to more easily defended elevated positions and from AD 1500 pa begin to appear in the landscape (Schmidt 1996) . Settlement centred around pa, which were political statements and monuments as well as defensive positions (Sutton et al . 2003) . People did not generally live on pa, though they stored kumara there and turned to them in times of trouble . Formal artefact forms, initially similar to their East Polynesian antecedents, began to change very quickly as people adapted to new materials (Furey 2004: 39) . For instance, the pro- portions of one and two-piece fishhooks changed as moa bone for large one-piece hooks became less available . The use of new materials led to changes in artefact design and artefact densities are much lower in later sites . Drill points, used to make one-piece fishhooks from bone, become much less common the record once moa bone was no longer available . The greatest change was in personal ornaments (Furey 2004: 41) . Bone necklace units and disc pendants were replaced with ear pendants and hei tiki, often utilising pounamu which was worked by sawing and grind- ing . Weapons in stone and bone begin to appear in archaeological sites from around AD 1500 at the same time that defended pa develop; they are unknown from early contexts but presum- Background 19

ably moa and seal were killed with heavy wooden clubs which could equally have been used on humans . Increasing warfare was probably associated with increasing control over resources and trade and exchange systems seem to have been disrupted . The distribution of stones resources such as Nelson–Marlborough argillite, Tahanga basalt and Tuhua obsidian becomes more restricted and a smaller range of local stone sources are more commonly used . Adze manu- facturing techniques also changed as easily flaked fine-grained and argillites were no longer widely available and adzes began to be made in simpler forms using local rocks, often using hammer dressing and grinding techniques . Artefact distributions will reflect the ways in which people moved around and as many formal tools, or more, may have been made but will be more thinly spread across the landscape . On the other hand, this apparent pattern may be an artefact of archaeological methodologies: early sites, viewed a priori as highly significant, tend to be more thoroughly investigated, with all deposits sieved, usually by academic archaeologists undertaking targeted research projects; later sites tend to be investigated using less thorough methodologies, often with minimal sampling, and often in the context of commercial archaeo- logical projects in response to development . At some later sites, such as Oruarangi (T12/192) on the Hauraki Plains (Furey 1996) or Panau (N36/72) on Banks Peninsula (Jacomb 2000), there are rich artefact assemblages, while swamp sites such as Kohika (V15/80, Irwin 2004) or Kauri Point Swamp (U13/41, Shawcross 1976) in the Bay of Plenty preserve remarkable assemblages of wooden artefacts, but there is a general impression that sites are less rich in artefacts . None of the four sites mentioned above are costal middens and all were intensively excavated (or, in the case of Oruarangi and Panau, fossicked) and it may be that archaeological research strate- gies and methodologies have biased these results, with large scale excavations on early coastal middens yielding large artefact assemblages while later pa and transient villages tend not to be excavated to the same degree . These two contrasting investigations methodologies will yield dif- ferent quantities of both finds and data . Excavations at the NRD site (R11/859) near Auckland Airport for example, where the two main middens dated to between the late 15th and late 18th centuries, sieved all midden deposits and yielded a rich artefact assemblage – although this was a burial site this assemblage was related to everyday activities such as fishing and woodworking (Campbell 2011) . Simple, expedient flake tools remain common throughout the sequence . While long range mobility appears to have become restricted, in general settlements became more ephemeral as people moved on a seasonal basis between gardens, fishing camps and other resources . Pa acted as socio-political foci for hapu, an anchor in the landscape to which they returned, as well as places to store and protect garden produce . Access to fisheries and resources was often dependant on pre-existing use rights and constantly renegotiated politi- cal alliances . Social and historical factors would always have been important influences on economy and settlement . The evolutionary paradigm underlying the conventional Archaic ~ Classic model has often led archaeologists to emphasise resource depletion and population growth as the drivers of change, in other words, external environmental factors, but internal factors would have been as, if not more, important . As mentioned above, subsuming pre-European Maori archaeology into a single phase allows us to escape this underlying evolutionary premise and to focus on social and historical factors . While, as also mentioned, it is beyond the scope of this report to fully formu- late an alternative model, some existing examples show what an outline of such a model might look like . Anderson and Smith (1996) outlined a settlement model for southern New Zealand where the multi-function bases that served as the centres of early settlement patterns (Anderson 1982) were redefined as transient villages that were no longer sustainable once moa were no longer 20 Torpedo Bay

available . Late in the prehistoric sequence transient villages were once again established (though the comparison between early and late villages is a little forced) . Anderson and Smith propose that this was a response to resource scarcity as people from these villages moved out to local resource extraction zones and then traded those resources to villages where they were scarce but had local resources of their own . Their model is located within the evolutionary environmental paradigm of the conventional Archaic ~ Classic model but “distribution of resources was facil- itated by a social system in which the entire region had come under the political control of a tribal leadership” (Anderson and Smith 1996: 368) from around AD 1750 . It would be better to view the late transient village as an economic and settlement response to a unified polity that was able to facilitate trade, the nature of which may have been constrained by the environment but was not determined by it . A second example is Smith and James-Lee’s (2010) study of marine resource exploitation in two areas of New Zealand: Greater Hauraki, which includes Torpedo Bay; and the Otago- Catlins area . In order to study temporal changes, they divided the prehistoric sequence up into three broad time periods based on a chronometric hygiene analysis (Smith 2010) of available radiocarbon dates: Early (ca AD 1250–1450), Middle (ca AD 1450–1650), and Late (ca AD 1650–1800) . In practice, many dates overlapped these ranges so that many assemblages were classified as Early/Middle or Middle/Late, giving five time slices . These were developed “with- out any prior expectation or prediction that they might also represent divisions in cultural, eco- nomic or environmental change” (Smith 2010: 185) allowing them to compare fish and shellfish catches through time, search for historical patterns in the data and construct empirical models based on the analysis independent of any predetermined model . This project provides a basis for the discussion of the faunal assemblage from Torpedo Bay (Chapter 5) but is not extended more widely in this report . Smith and James-Lee have not extended their analysis to artefacts or set- tlement patterns and chronometric hygiene excludes sites that are, on the basis of artefact forms or the presence of moa or seal bone, clearly early but lack robust radiocarbon dates (Chapter 7) .

The archaeology of historic period Maori

The archaeology of Maori in the historic period, after the arrival of the Cook voyages in 1769 and particularly following the increasing presence of missionaries, explorers and traders from the early decades of the 19th century, is one of the least studied aspects of New Zealand archaeology (Smith et al . 2014: 72; Campbell 2016b: 57) . Some important work has been done on the early missionaries and their interactions with Maori in the Bay of Islands from 1814 (Middleton 2008, 2013; Smith et al . 2012; Smith et al . 2014) and several excavations have been reported elsewhere such as Foster and Sewell’s (1995) excavations at Papahinu and Bedford’s (2013) excavations at Pohue Pa . There are several more excavations of sites post-dating the signing of the . On the Hauraki Plains, European material culture became progressively more common through time so that from a late 19th century house on the Puriri River no ‘tradi- tional’ material culture was found, but settlement and subsistence patterns remained ‘traditional’ (Bedford 2004: 149; Phillips and Allen 2013) . Maori were quick to adapt European material culture, technology and subsistence strategies as it suited them, though often at different tempos in different places (Anderson et al . 2014: 165) . European items that directly filled traditional roles were adopted first, while others were repurposed – ceramics were used as pendants and iron nails as fishhooks (Bedford 2004) but by the end of the 19th century there is often little to distinguish the archaeology of sites occupied by Maori or Pakeha (Campbell 2016b) . Background 21

Groube (1969: 5–10) points out that the settlement patterns and many of the structures and artefacts attributed to the Classic Maori were first recorded in the historic period following contact with Europeans and adoption of European technologies and economies . Even the early records of Maori living in pa, for instance from the Cook voyages, are a result of people flee- ing to their pa when they saw European ships – pa were not normally permanently inhabited . Settled villages (kainga) arose largely as a response to European trade, and with them carved wharenui and pataka also developed, the latter perhaps as a response to the presence of European rats that were more likely to burrow into underground storage pits than the kiore . Ornaments also changed, in particular the pounamu hei tiki became increasingly common as pounamu adzes, now superseded by iron, were converted to tiki, which were then traded to Europeans, often for muskets . This in turn led to the development of the ‘gunfighter pa’, a small ring ditch fortification more easily defended from musket attack than the large and elaborate pre-European pa . If many of the elements assigned by Duff and Golson to the Classic Maori are from a later date, this further reinforces the proposition that the culture of the first settlers ca AD 1300 was not markedly different to the culture of the Maori that Cook encountered in AD 1769 and the span of pre-European Maori history can be described without recourse to Stages, Periods or Phases . It was the arrival of Europeans, and particularly European crops and technology, that initiated a true phase change in Maori archaeology .

The archaeology of Tamaki

As noted at the start of this chapter, Tamaki Makau Rau, Tamaki of the hundred lovers, was an ideal settlement site for pre-European Maori horticulturalists, with fertile soils, produc- tive harbours, the volcanic cones on which impressive pa were built, and portages connecting the east and west coasts . A good quality adze rock, Motutapu greywacke, was available from several islands in the Hauraki Gulf; cherts, sandstones and basalts were also locally available; and most major obsidian sources were not far away . Several factors have impacted on the archaeological record of Tamaki, the most obvious one being the growth of Auckland City itself . In recent years this has resulted in numerous archaeological investigations in mitigation of development effects but for much of the city’s history there has been no archaeology carried out and major sites have vanished without record . The exception is the volcanic cone pa which remain fairly well preserved, but a single class of sites cannot be made to stand for an entire archaeological landscape . On the fringes of the city and in its hinterland of farm, forest, mountain and beach the situation is somewhat healthier . However, the sheer volume of archaeological excavation and reporting makes a full summary of the archaeology of Tamaki difficult . The archaeology of some smaller areas within the wider Tamaki region can be briefly sum- marised based on recent summaries provided for major excavation reports . These areas include South Auckland, which was a major focus of archaeological research, particulalry in the 1980s and is the best understood region, although excavation has tended to concentrate on garden sites (Campbell 2011); and the Tamaki River / Otahuhu area (Campbell and Ross-Sheppard 2013; Felgate and Opus 2014; Felgate 2017a), where excavation has focussed more on habitation and pa sites . To a lesser extent the archaeology of the East Coast Bays, and area which includes Torpedo Bay, can also be outlined . The archaeology of the wider Tamaki area will be broadly similar to these local summaries but the rich archaeological record of Auckland awaits a comprehensive synthesis, which is beyond the scope of this report . This summary begins with an overview of early sites in Tamaki . 22 Torpedo Bay

Early sites in Tamaki

Early sites in Tamaki are discussed in greater detail in Chapter 7 but are briefly summa- rised here . Few sites have been excavated that definitively demonstrate early occupation, but several unexcavated sites are regarded as probably early on the basis of artefacts forms or the presence of moa or seal bone . Often artefacts are found in the intertidal zone, implying a beach occupation now at least partly eroded or damaged by roading and housing . Like other early sites around the country, those in Tamaki contain a wide range of stone materials form both local and imported sources . Tuhua obsidian usually predominates but obsidian from the Coromandel Peninsula, and Northland is often present . Despite Tamaki having its own source of adze rock (Motutapu greywacke), adzes made of Tahanga basalt are common, while Nelson–Marlborough argillite is less common . The early sites of Tamaki would have had close links to sites of similar age in Northland, or on the Coromandel Peninsula and Western Bay of Plenty but these links remain unexplored (Chapter 7) . The best known early sites in Tamaki are the Pig Bay (R10/22) and Sunde (R11/25) sites on (Golson and Brothers 1959; Scott 1970; Davidson 1978; Smith 1985; Nichol 1981, 1988; Anderson 1991; Davidson and Leach 2017) . Artefacts are in early styles and rocky shore shellfish predominate . These sites were blanketed with volcanic tephras from nearby Rangitoto but the tephra provided fertility to the clay soils . Another early site at Motunau Bay (S11/20) on Ponui Island produced a similar array of cultural materials . The site dates to the 15th century (Nicholls 1964; Schmidt 2000a: 72; Sheppard et al . 2011: 52) . The only other excavated early site is Matatuahu (Q11/344) on the Manukau south head near Wattle Bay . A large and important collection of predominantly early artefacts had been made by the Brambley family prior to excavation, but this was small in scale and the site remains undated (Ambrose 1961; Prickett 1987) . The settlement of Tamaki was probably as early and extensive as anywhere else in New Zealand . Urban development since the European settlement of Auckland City has unfortunately removed many visible traces of these sites, which makes the finds at Torpedo Bay all the more important .

South Auckland

The stone fields sites of South Auckland are nationally significant archaeological sites and have figured heavily in discussions of pre-European Maori gardening . Otuataua is the best sur- viving example . Many of the tupuna maunga also had associated stone fields but these have been quarried away or built over . Some archaeological excavations have been carried out, particularly in South Auckland, but analysis and reporting have been of a decidedly mixed standard . The stone field at Ambury Farm Park, associated with Mangere Mountain, was mapped by archaeologists who interpreted surface stone alignments and mounds as evidence of housing and horticulture (Rickard et al . 1983) . Excavations at Ambury Farm Park, sites R11/1123 and R11/1129, in 1982 significantly modified this interpretation (Lilburn 1982) . Two areas previ- ously interpreted as stone structures were shown to be natural features; in another there was evidence of limited clearance of stone to create gardens . Brassey and Adds (1983) also excavated site R11/736 at Ambury Farm Park which, prior to excavation, had been interpreted as a small coastal settlement with shell midden, house sites, pits and a possible stone wall and crop marks . The midden was shown to be European in origin and many of the features previously recorded Background 23

were shown to be natural . Pre-European middens were found that were not previously visible on the surface . Occupation was temporary and not associated with gardening . Similar results have come from the Wiri oil terminal site (R11/1187) and the Wiri railway site (R11/1188); what appeared to be complex archaeology on the surface was very much less so when excavated (Cramond et al . 1982; Bulmer 1983; Veart et al . 1984; Coates 1992) . The best reported excavations of this type are at Puhinui (R11/25) on the / McLaughlins Mountain stone field . Excavated features included terraces with rock retain- ing walls; intensive cooking areas; postholes outlining small huts or shelters; and stone align- ments in the creek bed that were interpreted as a fish trap; but no evidence of any permanent or semi-permanent villages . Occupation at Puhinui dates to the 16th or 17th centuries (Lawlor 1981) . The only volcanic cone pa excavation in South Auckland was at / Elletts Mountain (R11/31) but only very brief reports have been prepared (McKinlay 1974, 1975) that indicate that stone faced terraces, pits and living platforms were excavated, with at least two phases of occupation . The NRD site (Northern Runway Development, R11/859) site was located on a sandy beach beside a freshwater stream, the only canoe landing for some distance . There were three main phases of occupation with numerous koiwi (burials) found in the final two phases, as well as an extensive midden containing a wide range of artefacts . The first phase, dating to the mid- 15th to 16th centuries, consisted primarily of kumara storage pits cut into the beach terrace . In the second phase, dating to the mid-17th century, several burials (Area A) were placed in rua kopiha, deep round pits that are known ethnographically but had not been described archaeo- logically before . Burials were also found in other features and several rua kopiha did not con- tain human bone, while two contained dog burials and others contained artefacts, large rocks and whale bone that had been deliberately buried . The same large rocks were often included with burials in rua kopiha and were not sourced locally . Burial practice was extremely varied, including multiple burials and secondary burials bought on site some time after death . The site was interpreted as reinforcing the group identity of the people buried there, some of whom may have lived and died some distance from the site while others were represented by other things buried in place of their physical bodies . In the final phase, dating to the late 17th to 18th cen- turies, burials were simpler (Area B) and located about 50 m south of Area A, indicating that the occupants were aware of the earlier burials and avoided them (Campbell 2011; Hudson and Campbell 2011) . Several sites were excavated at Timberly Road near Auckland Airport in 2015 . Most were simple midden deposits but R11/2379 was considerably more extensive . Three patches of shell midden, also containing fishbone, were situated around a cooking area, with 15 pits of various sizes located nearby . Three houses with associated rectangular hearths were found at the north end of the site along with numerous postholes, some of which may have been structures such as drying racks . Obsidian was found from Aotea / Great Barrier Island, Tuhua and Hahei . Analysis of charcoal samples suggested that intact bush was still located in the area at the time of occupa- tion . Seven radiocarbon dates were collected for the project, of which five were from R11/2379 . The site dated to the 16th–17th centuries, but one isolated hearth, separate from the main occu- pation, had a date from the late 14th century (Farley et al . 2015) . Other sites have been excavated near Auckland Airport for the Landing Development but are not yet fully reported . R11/2978 and R11/3056 were both single pits with associated midden; R11/2940 was an extensive palisaded or fenced site containing several probable whare, pits and 24 Torpedo Bay

midden; R11/3055 contained midden and posthole alignments indicating whare or shelters; R11/3112 contained several pits; while R11/311 was another midden (Farley et al . 2017) . Excavations at Papahinu (R11/ 229) on the Pukaki Creek revealed an early 19th century, historic period Maori occupation overlying an earlier midden layer, dated to to the 16th or 17th centuries . The site was occupied from at least the early 19th century up until 1823 and then reoccupied from 1835 until 1863 when Te Akitai refused the oath of allegiance to the Crown and departed to the Waikato . There are clear differences between pre-1823 and post-1835 house construction as well as changes in artefact types demonstrating Maori adoption of European technology in the 19th century (Foster and Sewell 1995) .

Tamaki River / Otahuhu

Maori occupation on the Otahuhu isthmus was centred on two maunga, Otahuhu / Mt Richmond and Te Apunga o Tainui / McLennan Hills . The latter was quarried away in the mid-20th century – a single date in the 16th or 17th century has been obtained from a remnant midden . Several excavations have been undertaken in the Otahuhu area . Mutukaroa / Hamlin’s Hill (R11/142) is a large non-volcanic hill that once held a large open undefended village on its southern end (Bulmer 1994; Foster 1984) . Hawkin’s Hill (R11/1394) has been interpreted as a small undefended hamlet . Although no physical evidence of gardening was found, the pres- ence of storage pits indicate that gardens would have been located nearby, possibly within the Te Apunga o Tainui lava field, which is now built on (Coates et al . 1996) . Near Fisher Road three archaeological sites, R11/887, R11/888 and R11/889, have been interpreted as the remains of an open undefended hamlet or hamlets (Foster and Sewell 1988), as was the Westfield Site (R11/898, Furey 1983, 1986; Sewell 1992) . R11/1201 and R11/1506 are located on the banks of the Tamaki River (Foster and Sewell 1993) . R11/1201 was essentially similar to other undefended sites but R11/1506 was a defended settlement or flatland pa with a clear palisade uncovered at one end that may have extended around the rest of the site . Inside the palisade were cooking and stone working areas as well as several pits in an alignment and several house structures . The dates from these sites indicate that occupation began around AD 1550, although this does not preclude earlier less intensive settlement in the region along the lines of small scale seasonal gardening (Foster and Sewell 1993) . These dates combined with data from the sites excavated in the area enables some hypotheses to be formed regarding the settlement pattern in the area . Other pa further north along the Tamaki River have also been investigated . Excavation at Taurere / Taylor’s Hill (R11/96) took place in the mid-1950s and were reported in 1991 (Leahy 1991) . Leahy notes a lack of evidence of defences so Taurere may not have been a pa . Excavation revealed several pits on terraces and artefacts included several adzes of basalt, greywacke and pounamu, hammerstones, grinders, and bone tattoo chisels, fishhooks and needles . Faunal remains included numerous dog bones, bird, a small fish assemblage dominated by snapper and a shell assemblage dominated by tuangi . The excavations were dated to the mid-15th to mid-17th centuries . Investigations were carried out on Maungarei / Mt Wellington (R11/12) in the 1960s and 70s and have recently been reported in full (Davidson 2011) . Various types of kumara stor- age pit were described, some stone lined to retain the loose scoria matrix, and ten radiocarbon dates indicated the main occupation was between the 15th and 17th centuries, with two main phases . Obsidian from afar afield as Taupo was found . One notable finding was that tuangi were Background 25

always small but got smaller over time, as a result of a combination of overharvesting and silta- tion in the Tamaki Estuary following Maori land clearance . Recent investigations on Maunagrei have included plant microfossil analysis, which found bracken fern spores dominant in all sam- ples, indicating extensive forest clearance and abandonment of gardens (Foster et al . 2012) . Recent extensive investigations have taken place at several sites as part of the AMETI roading project in Panmure / Mt Wellington . R11/2774 was a midden recorded in the wall of a trench . At R11/2880 two Tahanga basalt adze fragments, one with a ‘horned’ poll, an early form, were found as well as seven burials . At R11/2881 seven pre-European houses were found overlain with colonial era material including a Fencible road . Numerous obsidian flakes were found, mostly from Aotea and Tuhua, with some from Coromandel sources . A further four burials were found . R11/2882 consisted of a number of shell middens which included an adze of Nelson–Marlborough argillite . R11/2883, R11/2884 and R11/2950 were further shell middens . R11/2897 contained midden, firescoops, one within a probable rectangular shelter or windbreak, and four kumara storage pits . One pit had an internal drain lidded with stone and a rectangular sump . The dates from these excavations generally indicate forest clearance from AD 1430 with the main occupation between AD 1450 and 1625 (Felgate 2017a; Hudson 2016) . Mokoia Pa (R11/98) was occupied by Ngati Paoa in the early 19th century and was sacked by Nga Puhi in 1821 . Investigations between 1978 and 1980 undertaken by the New Zealand Historic Place Trust have never been reported, but a preliminary examination of the excavation field notes and plans indicates that the 19th century defences overlie earlier defences (Felgate 2017b) . Other undefended sites have also been excavated along the Tamaki . Excavations at the Waipuna site (R11/1436) revealed a series of intercutting pits demonstrating repeated use and reuse of the site in the 15th to 17th centuries . The pits had internal and external drains and the site was partly palisaded (Clough and Turner 1998) . At R11/1935 lines of postholes were inter- preted as windbreaks for a cooking area, and a possible house . Charcoal here and at R11/943 indicated that bracken and shrub species were dominant, with some remnant puriri, while microfossil analysis showed that kumara had been gardened (Bacquie et al . 2007) . The Cryers Road site (R11/1519) was located on the Matanginui / Green Mount stone field . Evidence of houses and gardening was found and radiocarbon dates indicated a long period of, probably intermittent, occupation from the mid-15th century into the early historic period (Fredericksen and Visser 1989) . The settlement pattern in the Tamaki River / Otahuhu area seems to have centred on undefended sites, often described as villages or hamlets, though the terms imply a permanence of settlement that is not supported by the archaeological evidence . These places were probably occupied seasonally, where gardening, stone working, cooking and other domestic and utilitar- ian actives were carried out, while also being part of the larger political units focused on the pa at Otahuhu, Te Apunga o Tainui and Maungarei . Gardens were located along river edges and throughout the lava field, which at this time was lightly forested . Defended flatland sites indicate a degree of control of the Tamaki River, though this control may have ebbed and flowed with the political fortunes of the inhabitants . Those wishing to use the portages would have to have sought permission from those living in the area, travelling up the defended Tamaki and across the defended Otahuhu isthmus . The flatland and volcanic cone pa were strategic as much as defensive pa, which would have provided a very visible focus for the groups living there . 26 Torpedo Bay

The North Shore

There are long stretches of the East Coast Bays coastline with little or no recorded archaeology, while sites are clustered in the inner Waitemata Harbour and Okura River estu- ary . This site distribution does not accurately reflect pre-European Maori occupation patterns but is largely a function of the history of site recording and urban expansion . As the East Coast Bays were first developed in the late 19th to mid-20th centuries archaeological sites would have been destroyed or built over without being recorded . There is an increase in site density around Takapuna / Devonport, in part due to recent redevelopment triggering the archaeolog- ical requirements of the Historic Places Act 1993 and its successor the Heritage New Zealand Pouhere Taonga Act 2014 . The greater density of recorded sites at Long Bay and the Okura River is in part due to recent greenfields developments proposals . It is certain that site densities along the East Coast Bays would have originally been higher than the current record suggests, although the clay soils here would not have been as attractive to Maori as the volcanic soils of Takapuna and Devonport . It is equally certain that numerous as yet unrecorded sites survive, albeit damaged by housing . Pa are recorded on the volcanic cones of Devonport (Maungauika R11/ 97; Takuranga R11/109; Takararo / Mt Cambria R11/110; and Takamaiwaho / Duder’s Hill R11/2402 though the status of the latter two as pa is uncertain, they are both now quarried away) and on coastal headlands, though again there are headlands where pa might be expected but none are recorded, perhaps due to former development . Several sites are recorded as containing koiwi . These sites are almost all coastal, with burials discovered close to the shore, often just behind the low foredune . They include the Long Bay Restaurant site (R10/1374) excavated by CFG Heritage in 2015–16 but not yet fully reported and the Masonic Tavern site (R11/2517) excavated by Geometria in 2010 and 2013 and also not yet fully reported (Russell Gibb pers . comm . 10 June 2017) . Both these sites contain an early component comparable in date to the lower layers at Torpedo Bay .

Summary

The three sub-regions of Tamaki summarised above would appear to have rather different archaeological signatures, but this is due more to the history of archaeological endeavour than to any real differences between them . South Auckland was an early focus of archaeology with an emphasis on evidence of gardening and occupation on the stonefields, but more recent excava- tions such as those at the NRD site and The Landing have demonstrated a wider range of occu- pations and activities . Excavations at Otahuhu and the Tamaki River have been less research focussed and more associated with large scale brownfields developments (even the Maungarei excavations were in response to water reservoir and road development) and have tended to con- centrate on pa and occupation sites; the soils would have been essentially the same as those at South Auckland and the occupations would have been associated with gardens, but highly visible stone fields such as Otuataua or Ambury Fram Park do not survive . Less substantial archaeology has been carried out on the North Shore and, away from the rich volcanic soils of Devonport gardening would have been less intense but still important . Recent small-scale redevelopment is revealing partially damaged sites with intact archaeology in places . An overview of the archaeology of Tamaki is long overdue . The most substantial and sig- nificant sites are the tupuna maunga but even these lack any recent overview (since perhaps Fox 1977) . Without such an overview it becomes difficult to provide a full context for excavations, to frame research objectives or to understand how one site can usefully be compared to another . Background 27

This chapter has provided a summary of the theoretical basis of our understanding of early (the long 14th century) settlement as well as an outline of the problems of what is now an outmoded approach; Chapter 7 of this report attempts an initial foray into providing some context for the early occupations at Torpedo Bay with a summary of excavated early sites in the upper North Island . Any more than this is beyond the scope of this report .

Acknowledgements

This chapter is based on an early draft prepared by Marianne Turner . It has benefitted from discussion with Richard Walter and Rod Clough .

3 Archaeology

The pre-European excavation, which was a small part of the wider excavation of the 19th century naval base at Torpedo Bay, covered an area of 5 5. x 4 7. m beneath the concrete floor of the 1886 base office in the northern corner of the project area, which had preserved the lower layers . The pre-European layers were located on what had been a narrow, low beach terrace at the eastern end of Torpedo Bay, visible in 19th century photos (Figures 3 1. and 3 .2) . To the north east of the beach terrace are the lower slopes of Maungauika / North Head, a pre-European pa (R11/97) located on a volcanic cone . The archaeology of Maungauika has been heavily impacted by military installations since the 19th century and the lower slopes have houses on them while the beach terrace is obscured by the naval base, now the Royal New Zealand Navy Museum, and King Edward Parade to the west .

Methodology

The entire base, except where there were standing buildings, was covered in a layer of asphalt and aggregate base course that was laid in the 1960s . This was removed by machine and was not further recorded . Removal of the base course exposed the floor of the base office, which was mapped, photographed and removed by machine . The archaeology of this layer is described in full in Volume 2 of this report and briefly in this chapter . The pre-European layers were discovered when the archaeologist monitored excavation of utilities trenches in order to trace the old shoreline and reclamation . Trench 1 ran east–west for 30 m and captured a profile through

Figure 3.1. 1870s photograph by James Richardson, looking east from the summit of Takuranga / Mt Victoria showing Torpedo Bay and the beach terrace prior to naval base construction. Beddoes Boat Shed is shown beneath the low cliff to the east of the terrace. The left middle ground shows typical ‘stonefields’ associated with the volcanic cone occupations, which would have been gardened (Sir George Grey Special Collections, Auckland Libraries, 4-2966). 30 Torpedo Bay

Figure 3.2. 1886 photograph by James Richardson looking east to Torpedo Bay and the beach terrace in the centre and left midground at the time the naval base was starting to be built. Beddoes Boat Shed is at the right of the image. Stone alignments running down the slope of Maungauika are probably pre-European Maori garden boundaries and the European stone wall running across the slope below them was probably built from stone scavenged from similar alignments (Sir George Grey Special Collections, Auckland Libraries, 4-2998).

Maungauika / North Head

Edward King Parade Pre-European Maori excavation

Torpedo Bay Royal New Zealand Navy Museum

N

0 50 metres

Figure 3.3. Aerial photo of Torpedo Bay and Maungauika showing the Royal New Zealand Navy Museum and the location of the pre-European excavation. The shipwright’s shop, now the museum office, is directly to the south east of the excavation. Archaeology 31

the centre of the original beach terrace and most of the base . Trench 2 ran north for 8 m from the intersection of Trench 1 to the west of the shipwright’s shop and provided a profile of the landward extent of the beach terrace . Trench 1 defined the southern edge of the pre-European excavation while Trench 2 and the shipwright’s shop defined the eastern edge . The northern edge was defined by the extent of the overlying concrete office floor and the western by the office chimney base that had disturbed the underlying layers . Profiles of Trenches 1 and 2 were drawn by hand . Several features in Layers 3 and 5 were only observed in the trench profiles; these were numbered and recorded but not otherwise excavated (Figures 3 .6 and 3 10). . Within this area the concrete floor of the base office was removed by machine and the underlying layers were exca- vated by hand . Features were numbered (numbering was restarted with each layer), described in the excavation notebook, mapped off the excavation baseline and photographed . North in this discussion is excavation north, 50° east of grid north .

Stratigraphy

The site stratigraphy was uniform across the excavation and the interface between the stratigraphic layers was for the most part clearly defined . This stratigraphy is shown in Figure 3 5. . Top, profile of Trench 1, north baulk (A–B); bottom, profile of Trench 2, west baulk (B–C) . The lowest layer was Layer 5, which was marked by numerous postholes and firescoops cut into the unmodified beach terrace, containing moa and sea mammal bone . Overlying this was Layer 4, a slopewash of cultural material from further up the slopes of Maungauika, that radi- ocarbon dates indicate is contemporaneous with Layer 5 and quite probably originated from the same occupation . Layer 3A was another slopewash event with features comprising Layer 3 cut into its surface, including firescoops, numerous postholes and stakeholes (i e. ,. where small stakes are driven in to the ground rather than dug), some of which formed clear alignments . Dates taken on the material from these firescoops may not be reliable as they probably incorporate older material from the Layer 3A matrix, which was not dated directly . A thin sand layer sepa- rated Layer 3 from Layer 2, a cultural layer that had been truncated and levelled in the historic

Figure 3.4. The concrete floor of the submarine mining base office after excavation and prior to the discovery of the pre-European layers. Photo scale = 1 m. 32 Torpedo Bay B C Trench 2 Trench C 45 45 B 13 13 44 44 1 e met r A 0 Asphalt Base course Concrete 1886 reclamation 2 Layer 3A Layer 4 Layer 5 Layer ll Feature Sand Modern disturbance 9 9 1 m 10 18 18 38 38 B A

Figure 3.5. Top, profile of Trench 1, north baulk (A–B); bottom, profile of Trench 2, west baulk (B–C). Archaeology 33

period to construct the concrete floor of the base office that was Layer 1 . Some shallow truncated firescoops remained on the surface of Layer 2 and historic period artefacts were also found here . Figure 3 5. . Top, profile of Trench 1, north baulk (A–B); bottom, profile of Trench 2, west baulk (B–C) . shows the profile of the excavated area in Trench 1 – Layers 3 and 4 could be traced intermittently for another 20 m to the east of the excavation and shovel test pits revealed that Layer 5 also continued to the east .

The beach terrace

The beach terraceconsisted of a well-sorted marine sand with grains of shell, probably deposited during the Holocene high-stand between 6500 and 2000 years ago when sea level was 1–2 m above the present level (Bruce Hayward, informal report to Mica Plowman 2011) . This sand was observed to be highly mobile on windy days during the excavation and was probably equally mobile when the terrace was first occupied . On the surface of the beach terrace and up to 150 mm below it were several basaltic scoria rocks typical of the cooking stones from the overly- ing Layer 5 . These are not native to the beach, though they could derive from very close by, and indicate some deflation or disturbance of the occupation deposit (or possibly the complete defla- tion of an earlier occupation) . Large, water-worn green-lipped mussel (Perna canaliculus), dog cockle (Tucetona laticostata) and occasional siphon whelk (Penion sulcatus) shells were also found in this layer, along with blue mussel (Mytilus galloprovincialis) shells in near pristine condition . This appears to be a natural shell assemblage; the blue mussels were sampled for radiocarbon dating .

Layer 5

Layer 5 was located on the natural beach terrace which also contains some evidence of deflation of either additional Layer 5 material or of an earlier occupation, this evidence is uncertain . Toward the north in Layer 5 was a cooking area consisting of five firescoops cut into the beach terrace, some intercutting, ranging in size from 600 mm diameter x 180 mm deep (Feature 2) to 1000 x 700 mm x 150 mm deep (Feature 9) . The fill of the firescoops was -gen erally a black, greasy, charcoal-rich sand with varying amounts of heat cracked rock and faunal material . Immediately south of these was Feature 4, a mounded deposit of midden, predominantly shell but including other faunal classes, and heat-cracked rock in a charcoal-stained sand matrix that was often burnt orange, located directly on the beach terrace and measuring about 2000 x 1200 mm and up to 200 mm deep, which probably originated as rakeout from the firescoops to the north and several shallow (less than 50 mm) firescoops cut into the beach terrace to the south (Features 12, 1, 5 and 29 and Features A–F) . These firescoops had a layer of volcanic rock in their bases . They generally showed some evidence of burning in the bases of the features but were filled with a clean, mottled beach sand, probably because they were raked out; they are not thought to be deflated features . They are interpreted as possibly having been heated, raked out but the heated rocks left in place, and then used to steam or dry fish for preservation . There is evidence that snapper at least were headed and preserved on site – this is discussed in greater detail in Chapter 5 . Feature 13 was the largest excavated feature from Layer 5 in the south baulk of Trench 2, an 1100 mm diameter x 400 mm deep firescoop with a layer of volcanic rocks lining the base 34 Torpedo Bay

N N Z TM excavation

0 1 metre

6

7 52

56 53 28 55 31 54

26 9 27 51 50 14 24 49 47 46 48 45 11 4 23 2 F 33 22 21 E B 58 15 20 32 8 57 43 19 16 A 17 12 5 44 29

1 D C 39 42 40 10 41 25

13

Figure 3.6. Layer 5 in plan. Postholes that were only located beneath overlying features are shown in grey. Features are numbered. Possible posthole alignments are highlighted. and a fill of black, greasy, charcoal-rich sand, particularly dense towards the base, quite different in form and content from the other shallow fire features in Layer 5 . Numerous postholes and a few stakeholes were evenly scattered across Layer 5 . Only one of these features, Feature 11, was observed to cut through a firescoop, while several of them were first observed beneath the firescoops, hearths or Feature 4 midden . Postholes are either infilled with residue from the subsequent cooking activities or a sterile grey sand . This possibly indicates Archaeology 35

Figure 3.7. Feature 4, Layer 5, excavated in half section with Feature F in the left foreground, looking east. Photo scale = 1 m.

Figure 3.8. Layer of stones at the base of Feature 4, Layer 5, typical of nearby shallow scoops. 36 Torpedo Bay

Figure 3.9. Feature 13, Layer 5, excavated in profile in the south wall of Trench 1. Photo scale = 1 m.

Figure 3.10. Alignment of stakeholes (Features 15–23) in the south west of Layer 5. Photo scale = 1 m. Archaeology 37

two phases of activity within the Layer 5 occupation: the postholes, particularly those filled with clean grey sand, which would be the stratigraphically earliest features on site; and subsequent cooking activities . However, it is very difficult to pick up postholes filled with the same matrix as the firescoops they cut through and some that appear to pre-date the firescoops may in fact post- date them . Many postholes appear to form alignments, although the density of features means some of these alignments may be spurious . In the west of the excavation in particular were sev- eral stake holes forming possible alignments that may indicate racks or small scale structures . Artefacts include three shell fish hook points or shanks along with lithic flaking debris such as chert and greywacke, Tahanga basalt and Tuhua / Mayor Island and Coromandel obsid- ian (Chapter 4) . The faunal material was the most numerous and varied from the site and includes: several species of land and sea birds including five extinct species; fur seal, sea lion and other unidenti- fied sea mammal; dog and kiore; tuatara; fish, predominantly snapper; and shell, primarily rocky shore species (Chapter 5) .

Layer 4

Layer 4 was a slopewash event that sealed the Layer 5 material beneath it and occurred after the Layer 5 occupation was abandoned . The matrix of Layer 4 was a compacted, soft, fine, red-brown soil deposit derived from weathered volcanic scoria, ranging from 100–200 mm deep and extending for the most part across the entire beach terrace, becoming quite thin in Trench 1 in the south of the main excavation area . Layer 4 contained a relatively sparse amount of evenly distributed charcoal, crushed shell and lithic and faunal material, including moa bone . It is a homogenous deposit with no evidence of internal stratification and no archaeological features, and the sharp interface between Layers 4 and 5 indicates that Layer 4 is a single rapid erosion event derived from a cultural layer on the adjacent slopes of Maungauika . The sparseness of cultural material indicated that either the matrix may have originated in a gardened soil that had cultural material cultivated into it, or it may have included archaeologically sterile soils that had become intermixed with the cultural material as it moved downslope . The dates for Layers 4 and 5 are identical (see below) so Layer 4 quite probably derived from an occupation on the slope contemporaneous with Layer 5 . Microfossil analysis (Chapter 6) indicates forest clearance prior to the Layer 5 occupation and gardening of the Layer 4 soil . It is probable that this clearance and gardening disturbed the soils on the slope and left them vulnerable to erosion . The articulated wing bones of a spotted shag (Phalacrocorax punctatus) and dog coprolites (turds) were found on the surface of Layer 4 . Artefacts included flakes of Motutapu greywacke, Tahanga basalt, Tuhua and Coromandel obsidian, and chert, as well as a carbonised chip of wood from a tapa beater (Chapter 4) . The faunal material consisted of: several species of bird, including two moa species; sea lion, dog and kiore; fish, predominantly snapper; but very little shell (Chapter 5) .

Layers 3A and 3

Layer 3A was another slopewash event that extended across the entire beach terrace and consisted of a 300–400 mm thick, red-brown soil, similar to Layer 4, somewhat lighter in colour and more compact in texture, and with more shell and charcoal fragments than Layer 4 mixed into it . This soil is derived from weathered volcanic scoria and, like Layer 4, probably originates on the adjacent slopes of Maungauika . The mixing of shell and charcoal is uneven across the site: 38 Torpedo Bay

N N Z TM excavation

0 1 metre 45

44 1 23 36

34 3 2 29

25 24 33 35 21 31 6

20 5 32 Trench 2 8 22 19

7 18 9 28 37 17 27 26 47 16 30 14 13 4 15 38 12 10 11

Trench 1

43

42 41 40

Figure 3.11. Layer 3 in plan. Features are numbered. Possible posthole alignments are highlighted. in the main excavation area there are substantially more inclusions than in the trench profiles to the east (Trench 1) and south (Trenches 3–4) . The lack of any internal stratification suggests that, like Layer 4, it represents a single rapid deposition event . It may have followed quite soon after Layer 4 was deposited as there is a sharp demarcation between Layer 3A and the underly- ing Layer 4 and no evidence of topsoil buildup on the surface of layer 4 . The deposition of Layer 3A preserved the articulated bird bone and the dog coprolites on the surface of Layer 4, indicat- ing possible use of the surface of Layer 4 after it was deposited and immediately prior to Layer Archaeology 39

3A being deposited . This activity may not have involved humans, however; it may just mark the presence of wandering dogs, perhaps scavenging the bird . Following the deposition of the Layer 3A slopewash matrix several large, often intercut- ting, firescoops and numerous stake and post holes recorded in the main excavation area were cut into the surface of the matrix (Figure 3 11. . Layer 3 in plan . Features are numbered . Possible posthole alignments are highlighted ). . Firescoops were generally circular and ranged in size from 500 mm diameter x 350 mm deep (Feature 27) to 1250 mm diameter x 250 mm deep (Feature 9) . The firescoop fill was generally a dark, charcoal stained, often quite greasy soil with some inclusions of shell and occasionally fish and other faunal classes, and moderate quantities of scoria oven stones, often denser towards the base (Figures 3 11. and 3 12). . Feature 5 contained a large imported river or beach cobble but all other stone was scoria . Some firescoops had a very clean fill, similar in colour and texture to the Layer 3A matrix, with or without moderately dense shell midden . None showed evidence of high heat such as reddening of soil at the base . Numerous large postholes around 200–300 mm in diameter and up to 300 mm deep, some containing post moulds, were found across the excavated area . Some of these (Features 15, 16, 18, 22, 20, 24 and 25) formed a north–south alignment but many others appeared to be isolated features . None were very deep and it is possible that many of them originated at a higher level and had been truncated by subsequent activity or by natural processes . It isn’t always clear what these postholes represent, but the alignment of postholes is substantial enough to have been a fence or a windbreak / cookhouse for the firescoops .

Figure 3.12. Feature 1, Layer 3, excavated in profile in the north baulk of the excavated area. Feature 2 is visible unexcavated in the right foreground, facing north. Photo scale = 1 m. 40 Torpedo Bay

Figure 3.13. Feature 2, Layer 3, excavated in half section, with scoria rocks typical of the Layer 3 firescoops excavated from the feature piled above it, facing west. Photo scale = 1 m.

25 24

20

19 22 17 18

7 47 13 16 14 12

15 11

Figure 3.14. Alignments of postholes and stake holes, Layer 3, facing north. Photo scale = 1 m. Archaeology 41

A line of stakeholes 40–50 mm in diameter and less than 50 mm deep (Features 1, 12, 13 and 17) ran north west–south east across the southern part of the excavation, with another stakehole (Feature 14) offset 500 mm to the south of this alignment . This collection of features plausibly represents a drying rack or similar structure . Artefacts included a stone net sinker and stone flakes of Motutapu greywacke, Tahanga basalt, and Tuhua and Kaeo obsidian (Chapter 4) . The faunal material consisted of small quantities of bird and mammal; five fish taxa, primarily snapper, which is the most for any layer; and the densest shell midden, primarily soft shore species (Chapter 5) . Two human bones were also recovered from Layer 3, both finger bones, one with some evidence of arthritis .

Layer 2

There is intermittent evidence across the site within the main excavation area and Trenches 1 and 2 of a 20–30 mm thick lens of crushed beach shell and sand overlying Layer 3A/3 and partially infilling some features . This material would have been wind deposited over the site following (or perhaps associated with) the abandonment of the Layer 3 occupation, sug- gesting that the Layer 2 occupation was a separate event . Layer 2 was a homogeneous, approximately 150–250 mm deep layer of compacted dark grey/black soil, evenly and densely mixed with charcoal and fragmented shell midden including minor amounts of other faunal classes, mostly fish . It directly underlies the 1886 office floor and the reclamation fill and extends over the entire northwest corner of the beach terrace from the original cliff line to the edge of the terrace . The profile of Trench 1 (Figure 3 5. . Top, profile of Trench 1, north baulk (A–B); bottom, profile of Trench 2, west baulk (B–C) ). showed that the upper portion of this layer has been truncated by levelling for the construction of the 1886 office and redeposited to the south where it overlay an undisturbed Layer 2 matrix . Some 19th century European artefacts were mixed into this redeposited material . The midden matrix appeared to be generally uniform across the site and the extent of this redisposition was not always clear . In both north and south profiles of Trench 1 were small but distinct sand lenses approx- imately 500–600 mm in extent at different levels within Layer 2 . These could have been intro- duced by people but could also be evidence of storm surge events, indicating that Layer 2 possi- bly did not accumulate as a single event . Outside of the excavation area and to the east in Trench 1, Layer 2 has been profoundly disturbed by 19th and 20th utility trenches – it does not extend beyond 18 m in the trench profile . Four firescoops were excavated in the upper surface of Layer 2 (with a fifth firescoop visible in profile in Trench 3, outside the main excavation) in the northern part of the excavated area, ranging in size from 500 mm diameter (Feature 4) to 1000 m in diameter (Feature 1) . These features were all shallow, no deeper than 70 mm, and had all been truncated by levelling for the office floor . The fill of each was generally similar: dark, charcoal stained soil, numerous heat cracked rocks and some midden shell and small quantities of fishbone . Artefacts included a single flake of Tuhua obsidian (Chapter 4) along with several intru- sive historic glass and ceramic fragments, a nail and a coin . During the initial excavation of Trench 1 an adze of Motutapu greywacke was recovered from the edge of a utility trench – it probably relates to either Layer 2 or Layer 3 . The faunal material contained no bird or mammal, apart from a pig tooth which is assumed to be a historic intrusion, some fish and mostly estuarine shell species (Chapter 5) . 42 Torpedo Bay

N 4 N Z TM excavation

0 1 metre 1

2 3

Trench 2

Trench 1

Figure 3.15. Layer 2 in plan. Features are numbered.

Layer 1

Layer 1 was the 19th century historic material associated with the construction of the 1886 submarine mining base office (see Volume 2 of this report) . This includes the office floor, foundations and chimney base and a reclamation fill to the south of the office . The lime cement floor, which was laid directly on the levelled surface without a base course, remained largely intact and acted to protect the lower pre-European Maori occupation Archaeology 43

Figure 3.16. Feature 2, Layer 2, excavated in half section, looking south. Photo scale = 1 m. layers from numerous 19th and 20th century utility trenches that skirted rather than penetrated it (some of these utilities are visible in Figures 3 9. and 3 10). . The building chimney base on the west side of the building however had significantly disturbed the underlying pre-European stra- tigraphy of the site and this has provided the logical western extent of the area excavated . To the south of the office floor the beach terrace had been levelled with a reclamation fill of red volcanic soil, probably imported either from elsewhere on Maungauika during other con- struction works, or from quarries on nearby Takararo / Mt Cambria or Takamaiwaho / Duder’s Hill . The fill ranged in depth from 400 mm at the base office to 1800 mm at the extent of the 1886 base, marked by the sea wall 15 m to the south of the excavated area .

Chronology

Fifteen radiocarbon dates were taken from the site, several from each layer, on a variety of materials where possible . These dates are listed in Table 3 1. and shown graphically in Figure 3 18. . One date (Wk 30313) was taken on the fresh-looking (as opposed to the other water-worn shell) mussel shell from the beach terrace below Layer 5, which provided a date of cal AD 1230– 1350 at a 68% confidence interval . A similar date (Wk 30626) was obtained from the Feature 4 midden in Layer 5, also on mussel shell, and it seems probable that this was intrusive beach shell and the date should be discounted . Four dates from Layer 5 (Wk 31981, Wk 31982, Wk 31983, Wk 31984) from moa bone and charcoal date to from around cal AD 1300–1400, placing the Layer 5 occupation firmly in the 14th century, while another date (Wk 30925) is about 100 years later and appears to be anomalous, dating some sort of intrusion, whether natural or human is not clear . 44 Torpedo Bay

N N Z TM excavation

0 1 metre

Figure 3.17. Layer 1 in plan.

The two moa bone dates from Layer 4 (Wk 31979, Wk 31980) are identical to the 14th century dates from layer 5, indicating that the Layer 4 material is quite probably derived from the same occupation as Layer 5, although when Layer 4 was redeposited is unknown . It is likely to have been contemporaneous with the Layer 5 occupation or soon after abandonment as it is reasonable to assume that if left undisturbed the slope above the beach terrace would have soon stabilised . Archaeology 45 1920– 1600–1608 1375–1420 1520–1820 1485–1693 1330–1545 1450–1650 1455–1630 1181–1402 1410–1499 1300–1405 1190–1420 cal AD 95.4% AD cal 1655–1905 1300–1410 1310–1360 1310–1405 1285–1395 1315–1360

1380–1430

1580–1620 1380–1400 1385–1405 1380–1400 1380–1400 1350–1390 1385–1410 1560–1710 1540–1655 1410–1495 1470–1585 1455–1510 1231–1331 1425–1463 1315–1355 1230–1350 cal AD 68.2% AD cal 1680–1830 1315–1355 1320–1350 1315–1355 1295–1320 1325–1345

± 35 636 ± 33 698 ± 35 875 ± 25 773 ± 29 398 ± 36 1098 ± 36 485 648 ± 29 ± 36 1083 CRA BP ± 35 550 ± 30 643 ± 28 624 ± 28 649 ± 34 689 ± 29 611

1 firescoop Feature 5 firescoop Feature 3 firescoop Feature firescoop 31 Feature 3 firescoop Feature 4 midden Feature 4 midden Feature 4 midden Feature shell Beach Feature sample bulk layer General layer General 3 firescoop Feature firescoop 13 Feature firescoop 13 Feature . Radiocarbon dates from Torpedo Bay. Torpedo from dates . Radiocarbon 1 . 3 2 3 3 3 3 5 5 5

Layer 2 4 4 5 5 5 Table Table (pipi) shell (tuangi) shell (pipi) shell (pipi) shell mahoe) hebe, (tutu, charcoal (mussel) shell (olearia) charcoal moa bone (mussel) shell Material Material (pipi) shell moa bone moa bone moa bone moa bone (mahoe) charcoal

Lab number Lab number Wk 31974 Wk 31975 Wk 31101 Wk 31976 Wk 31978 Wk 31977 Wk 31979 Wk 31980 Wk 30626 Wk 30925 Wk 31981 Wk 31982 Wk 31983 Wk 31984 Wk 30313 46 Torpedo Bay

Of the four dates from Layer 3, three (Wk 31101, Wk 31977 and Wk 31978) generally date to the late 15th to mid-17th centuries, while one (Wk 31976) is somewhat earlier . Although it overlaps with the other three at a 95% confidence interval, it is also possible that the dated material incorporates older shell from the Layer 3 matrix into which the firescoop was cut and the date should be discounted . The two dates from layer 2 (Wk 31974, Wk 31975) are quite different, although they overlap at a 68% confidence interval . The archaeology of the site indicated that the occupation of Layer 2 followed soon after that of Layer 3 and it seems that the later date, taken from a bulk midden sample, may have been contaminated with modern material during site levelling associ- ated with the construction of the base office .

Marine data from Hughen et al (2004);Delta_R -7±45; OxCal v3.10 Bronk Ramsey (2005); cub r:5 sd:12 prob usp[chron] Wk 31974 636 ± 35 BP Layer 2

Wk 31975 550 ± 35 BP

Wk 31101 698 ± 33 BP

Wk 31976 875 ± 35 BP Layer 3

Wk 31978 773 ± 25 BP

Wk 31977 398 ± 29 BP

Wk 31979 643 ± 30 BP Layer 4

Wk 31980 624 ± 28 BP

Wk 30626 1098 ± 36 BP

Wk 30925 485 ± 36 BP

Wk 31981 648 ± 29 BP Layer 5

Wk 31982 649 ± 28 BP

Wk 31983 689 ± 34 BP

Wk 31984 611 ± 29 BP

Wk 30313 1083 ± 36 BP

1000 1500 2000 moa bone Calibrated date AD shell

charcoal

Figure 3.18. Multiplot of the radiocarbon dates from Torpedo Bay. Archaeology 47

In summary, the 15 dates are not always consistent, but the likelihood of contamination rules out three of the inconsistent dates, while Wk 30925 is anomalous and could also be dis- counted as it cannot be explained from the archaeology of the site . Layer 5 dates to the 14th century . The dates, other than Wk 31983, are not as early as some other sites, and while they may relate to the initial settlement of East Polynesian voyagers they could equally plausibly relate to an occupation by first or second generation Maori . Layer 4 dates to the same time as Layer 5, and is derived from slopewash material from the slopes of Maungauika to the north . Layer 3 is less well dated: the Layer 3A matrix that the firescoops are cut into is not dated but almost certainly pre-dates the Layer 3 occupation, which dates to the late 15th to mid-17th centuries . The Layer 2 occupation, although disturbed by 19th century levelling, appears from the archaeology to be related to the Layer 3 occupation (an intermittent layer of storm or wind redeposited beach sand separates the two, indicating at least a brief hiatus in occupation) and so can be dated to the same time . Layers 5A, 5 and 4 can be considered to represent Phase 1 of occupation; Layers 3 and 2 represent Phase 2; but the slopewash Layer 3A matrix remains undated and unphased .

Summary: site formation processes

The 5 5. x 4 7. m excavation on the beach terrace is small and almost certainly represents only a small part of the occupations on the beach terrace during Phases 1 and 2 . The terrace, as shown in the 1886 photo for instance (Figure 3 .2 . 1886 photograph by James Richardson look- ing east to Torpedo Bay and the beach terrace in the centre and left midground at the time the naval base was starting to be built . Beddoes Boat Shed is at the right of the image . Stone alignments running down the slope of Maungauika are proba- bly pre-European Maori garden boundaries and the European stone wall running across the slope below them was probably built from stone scavenged from similar alignments (Sir George Grey Special Collections, Auckland Libraries, 4-2998) ),. was considerably more extensive than this and was presumably also more extensive during the Phases 1 and 2 occu- pations . It is probable that the range of activities carried out at the site would also have been more extensive than can be observed from this small excavation, but these remain unknown . Some of this further extent of the site has been badly damaged and in places destroyed by activ- ities associated with the naval base, particularly building foundations and utilities trenches, but significant portions of it may remain intact beneath reclamation and the shipwright’s shop . The slopewash material that comprised Layer 4 has an identical date to Layer 5 and almost certainly derives from material on the adjacent slopes of Maungauika . This strongly implies that the Layer 5 occupation extended up the slope to the north and east and involved activities that left the ground open and vulnerable to erosion . Such activities could potentially relate to forest removal and burning (Chapter 6) although there is not sufficient charcoal to really support this . It is more likely to relate to gardening the fertile volcanic soils on the slopes of the maunga . Given that the sand on the beach terrace was highly mobile, the Layer 4 slopewash was probably deposited quite soon after the abandonment of Layer 5 and may in fact be all that has preserved the Layer 5 features, several of which may have been lost to wind deflation . The matrix of Layer 3A into which the archaeological features are cut is another slope- wash deposit . Some activity took place on the surface of Layer 4 just prior to the deposition of Layer 3A, as shown by the coprolites and articulated bird bone that were preserved by the Layer 3A deposit . Also, the Layer 3A slopewash matrix has not been dated and the interval between 48 Torpedo Bay

the two slopewash events isn’t known: Layer 3A may belong to Phase 1 or some intermediate Phase that derives from a gardening occupation on the slopes of the maunga without an associ- ated beach terrace occupation . It is unlikely to belong to Phase 2 as older material from Layer 3A appears to have contaminated the Layer 3 dating samples . The Layer 3 occupation is securely located in Phase 2 and is younger than the Layer 3A matrix into which it is cut . The depth of the Layers 3A and 4 slopewash deposits meant that the Layer 3 features did not extend into Layer 5 . Layer 3 contained the densest midden on the site although this was not very dense compared to many middens in New Zealand . This midden probably built up from rakeout of the numerous firescoops and general deposition of shell . A thin, intermittent lens of wind or wave deposited sand separated Layers 2 and 3, but Layer 2 was heavily truncated by the subsequent construction of the 1886 submarine mining base . This small excavation has revealed two phases of pre-European Maori occupation on a dynamic and mobile location and hinted at wider relationships to the surrounding environment, but its limited scope somewhat restricts interpretation . 4 Material Culture

tawheowheo (Quintinia serrata). The latterSix formal two species tools were are recoverednot closely from related the excavationbut are not – a carbonised chip of a wooden tapa distinguishable under the microscopbeater,e. an adze, a sinker and three shell fish hook pieces – along with 73 stone flakes . Louise Furey, Marianne Turner, Rod Wallace and Arden Cruickshank analysed the formal tools, and Discussion Marianne Turner and Arden Cruickshank analysed the flaked stone assemblage . Clearly at the time Moa were being hunted the local vegetation around North Head was closed canopy broadleaf podocarp forestFormal right down tools to the waterline. This forest type is characterised by very large conifers rising over a dense understory of somewhat smaller hardwood trees. Shrubs and smaller tree species are a completely different suite to those in the later assemblage and The formal artefacts have been registered as taonga tuturu with the Ministry of Culture were probably those limited to theand forestHeritage floor under or, more the Protectedlikely, on Objectsthe forest Act edge 1975 along . the shoreline itself.

Tapa beater The late charcoal assemblage appears to reflect a landscape largely cleared of forest with the few remaining tree species being restricted to coastal cliffs. Away from these cliffs vegetation seems to have consisted mainly of bracken fernDuring with charcoal the woody identification species being an mainlyunusual tutu,piece hebe, of charcoal was noted from Layer 4 coprosma, fivefinger and manuka.which, though damp and encrusted with a layer of sandy soil, appeared to have an abnormal sur- face texture . Close examination revealed that the surface had been incised with parallel grooves, Tapa Beater fragment and almost certainly was detached from the distal end of a tapa beater . The fragment was some During the identification process a10 piece x 8 ofmm charcoal x 2 mm (from thick a bagand labelled was made “Layer from 4 – a Feature dense, resinousSpit 1 kauri (Agathis australis) probably north of section – Catalogue # 67”)from which, branch though wood, damp which and encrusted was the preferred with a layer wood of sandy for all soil types, of beaters where it was available appeared to have an abnormal surface(Wallace texture 1989). This. The was fragment photographed had detached the download from the images beater, either during use or, possibly, when the whole artefact was burnt . revealed it was a chip from the surface of a wooden artefact, almost certainly from the working surface of a tapa beater. The fragment wasTapa some barkcloth 10mm long beaters by 8mm have wide a characteristic by 2mm thick working and was surface consisting of closely spaced made from a dense, resinous typeparallel of kauri grooves found on inone branches. or more Kaurisides whichbranch are wood designed was the to compact and felt the fibres of aute preferred wood for all types of beaters(Broussonetia where it papyrifera was available, paper (Wallace, mulberry) 1989). bark The sheets fragment into had a flexible cloth-like material . Most detached from the working surfaceexamples of the front from of acrossthe beater, the Pacific either during are very use similaror, possibly, to the when fifteen known New Zealand exam- the whole artefact was burnt. ples (Neich 1996, 2002) . With the exception of one from Taranaki, all are from the upper half of the North Island . One other beater has been recovered from an archaeological context at These beaters have a very characteristicMangakaware deeply (S15/18, grooved Bellwoodworking surface 1978) fromthat lateis designed in the Maori to sequence . compact and felt sheets of bark into a flexibleThe clothPolynesian-like mate settlersrial. Mostintroduced examples aute from to acrossNew Zealand the but it is a tropical plant that Pacific are very similar as are thedid fourteen not thrive originating in the intemperate New Zealand New thatZealand has been climate described . Although still surviving at the time of from museum collections (Neich, EuropeanPendergrast arrival, and Pfeiffer, the low 2004). number This of is observations, the first ever confinedrecorded to the upper North Island, suggest from and excavated archaeologicalthe site plants. were few in number and grew in a restricted geographical area .

Figure 4.1. Tapa beater fragment as it was when first found – its condition has since deteriorated (Z20932).

50 Torpedo Bay

Adze

A single Duff type 2C adze with rectangular cross section (Duff 1977: 168) was recovered from the edge of a utility trench – it probably relates to either Layer 2 or Layer 3 . It is small, at 98 mm and has a flared blade . Because it is still quite symmetrical, its length and blade width have probably been gradually reduced as a result of ongoing blade use and repair, suggesting it was once considerably larger and has had a long use life . The sides are steep and the corners well defined, with a low-angled bevel suited to timber trimming or ‘true adzing ’. This adze form is common in Motutapu greywacke, from which it is fashioned . The bevel and blade are well ground while the butt half has been hammer dressed exten- sively to prevent damaging the lashing when the adze was tied to the wooden haft . The poll of the adze has considerable haft polish suggesting that the length of the adze was reduced to the extent that it may have required being set it in a recessed or even a composite wooden haft . The blade shows evidence of use wear and rejuvenation . Use-striations are visible, par- ticularly on the front . The blade corners curve up somewhat asymmetrically and this, together with uneven grinding facets, suggest the repair of a number of corner snaps and chips . While the blade currently has a few minute chips, it would take little time to restore it to operational condition . The adze was still a valuable tool with a many more years of use in it at the time it

0 50 mm

Figure 4.2. Left, adze (Z20932); right, stone sinker (Z20929). Material Culture 51

entered the archaeological record . Because of this, it would be unlikely to have been discarded deliberately so must have been either lost accidently or cached away for use at a later date but never retrieved .

Sinker

A stone sinker, probably a net weight, was recovered from Feature 9 in Layer 3 . It is quite small at 56 mm length, and has a hammer dressed groove around the shorter circumference . The groove is approximately 18 mm wide, and is not very deep, but would have been adequate to secure a lashing . It is made from a sandstone cobble and has bruising all over on the upper surface . The base is naturally flattened . The style of sinker, with a single hammer dressed groove, is the most common sinker style in museum collections .

Shell fishhooks

Three shell parts from two-piece hooks were recovered from Layer 5, Feature 4 . One, Z20928, is an obvious point from the narrowing of the shank at one end, and the curvature, but the point limb is broken at both ends . The other two fragments, Z20926 and Z20927, may be points or shanks as the ends which distinguish whether point or lashing head and shank, is miss- ing . There is also a small shell tab from which a hook may have been fashioned . Cooks turban (Cookia sulcata) shell has been used with the cortex present on the reverse side . The shell tab has been ground and the rough cortex abraded away . The method of manufacture of shell fish hooks involving breaking the shell and chipping out a blank from a section of shell . The blanks are easily distinguished as is the pattern of break- age of the shell . However there was no evidence of manufacture in the deposit where they were found .

0 20 mm

Figure 4.3. Shell fishhooks (left to right: Z20926, Z20927, Z20928). 52 Torpedo Bay

Shell hooks are fragile and deteriorate quickly . They occur in low numbers in early sites . They are known from Pig Bay(Davidson and Leach 2017); Long Bay (Matthew Campbell pers . comm );. and the Masonic Hotel in Devonport (Russell Gibb pers . comm ). . Many early Coromandel sites also have small numbers of shell fish hook points or shanks and, rarely, one- piece hooks .

Flaked stone

Seventy-three flaked stone artefacts of obsidian, chert, basalt and greywacke were analysed following methodologies outlined in Beyin (2010), Holdaway and Stern (2004), Turner (2005) and Cruickshank (2011) .

Obsidian

Obsidian is a volcanic glass, usually of rhyolitic composition that owes its vitreous struc- ture to rapid cooling of a viscous high silica lava (Ward 1973) . It is widespread throughout many volcanic regions of the world including New Zealand (Green 1964) . Forty-six flakes of obsidian were recovered from the site . Obsidian played a significant role in Maori society prior to the arrival of metal implements . It is probable that a core of obsid- ian or other high quality local stone was frequently on hand to carry out a range of day to day tasks such as butchery, scraping and finishing wooden tools, carving and producing muka (flax fibre used for making cordage) (Turner 2005) . This was an expedient technology; a flake would be struck off a core, perhaps used until it was too blunt to keep or retouch and then discarded . Eventually the core would get too small to create useful flakes and would also be discarded . Because of the brittle nature of the material and the multitude of uses it had, it was discarded frequently and as a result it is present in nearly every type of pre-European site in the country (Green 1967) . Obsidian is found in four regions of New Zealand associated with tertiary and quater- nary rhyolitic volcanism: Northland, Mayor Island, the Taupo Volcanic Zone (TVZ) and the Coromandel Volcanic Zone (CVZ) (Sheppard et al . 2011) . Although these four geological areas are restricted to the top half of the North Island, complex exchange networks and access rights allowed its movement throughout the country, as far afield as the Chatham and Kermadec Islands (Leach et al . 1986; Walter et al . 2010) . The sources display distinct geochemistry which, when compared to pieces found in an archaeological context, can provide information about the source of the raw material, exchange networks and use of lithic resources . Geochemical analysis of the obsidian uncovered was undertaken in 2011 at the Department of Anthropology, University of Auckland using an InnovX Alpha series porta- ble Energy Dispersive X-Ray Fluorescence (EDXRF) machine . The flattest and most uniform surfaces of the flakes were placed on the detector window which is mounted on a test stand provided by the manufacturer . Any loose soil was removed where necessary, but no cleaning agents were used . Two of the flakes were too small to analyse . The results were then compared to a source database held at the Department of Anthropology using multivariate statistical analysis . The results are given in Table 4 1,. which includes an assemblage from a shovel test square con- taining 13 flakes that could not be assigned to layer . Obsidian from five sources was identified: Tuhua / Mayor Island off the coast of the Bay of Plenty; Hahei and Whangamata from the mainland CVZ; Te Ahumata from Aotea / Great CHANGE THROUGH TIME

occurring as detrital deposits in streams, in colluvium derived from in place rhyolitic formations upstream and inland, or as volcanic bomb deposits (Bell and Clark 1909:72, Ward 1972:123-27, 172-73, Appendix 4). The major exception is Mayor Island where high quality obsidian is readily available throughout most of this large island. It seems likely that the abundance, quality and accessibilityMaterial Cultureof Mayor Island obsidian accounts for its role as the premier source in the country 53 as was noted by the earliest geologists to survey the island (Sladden 1926, Thompson 1926).

N

1 NORTHLAND 2 Obsidian Source 3 Areas and Regions 4 1 - Weta 5 2 - Waiare/Pungaere 6 3 - Huruiki 7 4 - Burgess Island Coromandel 5 - Fanal Island 6 - Awana 9 8 10 Mayor Island 7 - Te Ahumata 11 8 - Cooks Beach 12 9 - Purangi 13 27 Taupo Volcanic Zone 10 - Hahei 14 11 - Tairua 12 - Whangamata 26 13 - Maratoto 22 24 14 - Waihi 18 15 - Ben Lomond 19 23 16 - Maraetai 15 17 21 16 20 17 - Whakamaru 14 18 - Ongaroto 19 - Ngongotaha 20 - Hemo Gorge 21 - Whakarewarewa 22 - Tarawera 23 - Lake Rotokawau 24 - Lake Okataina 25 - Lake Rotoiti 26 - Maketu 27 - Mayor Island

obsidian source Figure 4.4. The North Island showing obsidian sources and sites mentioned in the text (from 0 100 200 300 km Sheppard 2004: Figure 7.1).

Figure 7.1 Source regions of New Zealand Obsidian (compiled with permission from Jones (2002) Figures 3.1, 3.2, 3.3, 3.4). Moore (pers comm. 2004) no longer accepts Burgess or Purangi as Barrier meaningfulIsland, also sources. from They the are CVZ; retained and here forKaeo historical in Northland reference. . There were no flakes present from the Taupo Volcanic Zone . The flakes were measured categorically, based on maximum dimensions . All pieces were larger than 10 x 10 mm, with more than152 half being larger than 30 x 30 mm . The 12 angular fragments and eight core fragments identified indicate that flaking took place on site, while the lack of small waste flakes (smaller than 10 x 10 mm) is probably the result of the collection strat- egy utilised during the excavation rather than their absence from the site . There were three flakes which had cortical surfaces present . Two of these were from Whangamata, and the third from Te Ahumata . Use wear was identified macroscopically using a 10 x hand lens . Caution must be employed when making inferences about an assemblage based on macroscopic use wear analysis . Although macroscopic damage can and does form on artefacts, many activities where these artefacts are used in a New Zealand context would not show macroscopic damage . Often, especially with obsidian, a flake will be used for a single task (perhaps gutting a fish or bird) and then that flake 54 Torpedo Bay

Table 4.1. EDXRF results for obsidian by source and Layer. Layer 2 3 4 5 Test sq. Hahei 1 1 Kaeo 1 Te Ahumata 1 1 Tuhua 1 13 10 1 Whangamata 3 2 11 Total 1 2 17 13 13 will be discarded . It is not necessarily the case that flakes would always be kept and reused later, unless access to high quality material was limited . Generally, the situation in which macroscopic damage will form is when a thin-edged tool is used to work a hard material such as bone or wood, particularly when used for scraping . It rarely forms on thicker-edged tools that have pro- longed use times (Lambert-Law de Lauriston 2015) . Because of the inherent issues with identifying use wear, it was only recorded on those flakes with obvious signs of use wear . The most easily identifiable use wear is from scraping, when a flake is dragged at an obtuse angle across an item, causing damage to the trailing edge . This results in visible uni-facial micro-flaking of the surface of the flake . Use wear was positively identified on nine pieces of obsidian (Table 4 .2) . Seven of the pieces displayed evidence of having been used for scraping; one sourced to Whangamata from Layer 5 for cutting; and one from the test square for both cutting and scraping . The remaining flakes could have been used as tools, but the lack of identifiable use wear means it is not possible to infer what the flakes were used for .

Chert

There were 13 chert flakes . The term ‘chert’ is used here as described by Moore (1977) and Cruickshank (in Campbell 2011) as all material that appears to be a highly siliceous (such as flint, chalcedony and jasper) but cannot readily be classified into other well-known stone types in hand specimen .

Table 4.2. Obsidian flakes displaying use wear by source and Layer. Layer 2 3 4 5 Test sq. Te Ahumata 1 Tuhua 1 1 4 1 Whangamata 1 Total 1 1 1 5 1 Material Culture 55

Good, flake-quality chert occurs in three main areas in New Zealand: Marlborough, the East Coast of the North Island and the Coromandel and Auckland / Northland Peninsulas (Moore 1977: Figure 1) . These sources overlap with the other high quality lithic resources, par- ticularly Tahanga basalt, Motutapu greywacke and obsidian from Northland and the CVZ . In Tamaki, chert occurs on Motutapu, Waiheke and Ponui Islands and on the mainland in South Auckland and the Hunua and Waitakere Ranges . Thirteen flakes of chert were recovered from Layers 4 and 5 . Six chert types were identi- fied on the basis of colour and quality . • Type A was a medium to high quality stone that appeared to be black in reflected light, yet was slightly translucent and appeared brown in transmitted light . Black chert has been identified in the Tupou Complex of sandstones around Whangaroa Bay in Northland, but other sources cannot be ruled ou (Moore 1977; Edbrooke and Brook 2009) . It is also found on Ahuahu / Great Mercury Island • Type B was a medium to high quality stone, dark reddish-brown in colour . • Type C was a medium to high quality stone, dark grey in colour . • Type D was a poor to medium quality stone, white or yellow in colour . The single flake recovered was probably discarded due to its poor structural integrity . • Type E was a high quality stone, dark reddish-grey or greenish-grey in colour . The single flake recovered retained some rough colluvial cortex . • Type F was a medium to high quality stone, dark grey and dark yellowish-brown in colour . It is, however, difficult to identify source from colour . Cobbles from a single source area can grade through multiple colours, sometimes in a single cobble . The distribution of chert types in Layers 4 and 5 is shown in Table 4 .3 . The six chert types cannot be assigned to any particular source . No use wear was recorded on any of the flakes but it is a hard material unlikely to show evidence of light use .

Basalt

Four flakes of basalt were identified from Layers 3, 4 and 5 . All have been sourced to Tahanga on the basis of their high magnetism (Turner 2000: 38) . None showed any use wear, or any polish to indicate that they had been flaked off an adze .

Table 4.3. Chert types by layer. Layer 4 5 Type A 3 4 Type B 1 Type C 1 Type D 1 Type E 1 Type F 2 Total 6 7 56 Torpedo Bay

Greywacke

In addition to the adze described above, 10 flakes of fine-grained Motutapu greywacke were recovered from Layers 3, 4 and 5 . None showed any use wear, or any polish to indicate that they had been flaked off an adze . This greywacke occurs on several islands in the Hauraki Gulf including Motutapu, Motuihe and Rakino . Working floors are present in the inter-tidal areas on all islands .

Discussion

There is not much information that can be inferred from this lithic assemblage . Compared to other similar sites in Tamaki, the assemblage is small, coming from a limited excavation of a larger site . Although no specific tool manufacturing area was identified during the excavation, the presence of multiple sources of lithics, most notably the obsidian and basalt imported from outside the Hauraki gulf indicates the use of exchange networks for the procurement of tools . The Torpedo Bay obsidian assemblage, though small, can be compared to other similar sites in Tamaki . Obsidian from the 15th century Motanau Bay site on Ponui Island (S11/20) was analysed by Sheppard et al . (2011): 65% of the assemblage was sourced to Tuhua with the rest coming from the CVZ including Te Ahumata and Fanal Island, the TVZ including Maketu and a single flake from Northland . Although Te Ahumata and Fanal sources are the closest to the site, they make up less than 7% of the assemblage . Similarly, at the Sunde Site (R10/25) 45 5%. of the assemblage was from Tuhua while at nearby Pig Bay (R10/22) 83% was from Tuhua . In each case the balance was from various mainland CVZ sources (Cruickshank 2011) . At later sites in Tamaki post-dating AD 1500 the proportions of Tuhua and CVZ, par- ticularly Te Ahumata, obsidians change markedly . The NRD site (R11/859) 32% is from Tuhua while 65% is from Te Ahumata; and from Taputiketike Pa (R11/348) 20% is from Tuhua and 80% from Te Ahumata (Cruickshank 2011) . Early obsidian assemblages from a small sample of sites in Tamaki are dominated by the Tuhua source with a notable amount of mainland CVZ obsidian; after around AD 1500 the assemblages are dominated by Te Ahumata obsidian with lesser amounts from Tuhua and Northland obsidian appears in low quantities (Davidson 2011, 2013; Cruickshank 2011; Moore 2012) . A study by Moore (2012) concluded that after AD 1500 Te Ahumata obsidian probably had a primary distribution area of approximately 100 km in which it would have been directly procured, and beyond that it would have been exchanged . Prior to AD 1500 Tuhua obsidian was a valued material that was transported throughout the main islands of New Zealand and as far afield as Norfolk Island, the Kermadec Islands, Chatham Islands and Auckland Island . After AD 1500 it is rarely found more than 500 km from the source (Walter et al . 2010) . Moore (2012) agrees that there was a significant decline the distribution of Tuhua obsidian after AD 1500 even at sites within 150 km of the source . Moore concludes that this could be an indication of more complex distribution networks, as opposed to direct access to the source, being devel- oped after this time . At the same time, Te Ahumata became a major source in northern New Zealand and minor sources such as Kaeo, which is within the primary distribution area of Te Ahumata, are seen more often . Torpedo Bay fits this pattern: although the sample is small the Phase 1 assemblage is dominated by Tuhua obsidian (77%), with a minor component from the CVZ (Hahei and Whangamata); while Kaeo (Northland) and Te Ahumata obsidian only appear in Phase 2 . Material Culture 57

A similar pattern is generally evident in the sources of adze making stone: Tahanga basalt often dominates early assemblages in the upper North Island, including the Hauraki Gulf, with a minor percentage of Nelson–Marlborough argillite, and some local sources including Motutapu greywacke; in later assemblages argillite is generally absent (highly curated adzes, by now reduced to chisels may be found) and Tahanga basalt is much less widely spread while local stone, Motutapu greywacke, in the case of Tamaki, dominates . However, the assemblage from Torpedo Bay is too small to determine if this pattern is replicated at the site .

5 Faunal Analysis

This chapter is based on the faunal report prepared by Brooks et al . (2012) with some rea- nalysis of fishbone by Matthew Campbell .

Methods

Some faunal material was collected by hand during excavation but all deposits were sieved through a 3 5. mm screen and all screened bone and a subsample of shell was retained for anal- ysis . All retained material was returned to the lab where it was wet sieved through nested 6 4. and 3 .2 mm screens . All shell from the 6 4. mm fraction and all bone from both fractions was retained . All material was then sorted to faunal class, rebagged and relabelled, and sent to the relevant specialist for analysis: fishbone was analysed by Matthew Campbell of CFG Heritage using comparative collections housed at CFG Heritage; shell was analysed by Nick Cable of Opus International Consultants; and all other faunal classes – bird including moa, reptile and mammal including sea mammal – was analysed by Sheryl McPherson and Emma Brooks of South Pacific Archaeological Research using comparative collections housed at the Department of Anthropology and Archaeology, University of Otago . Moa identifications were carried out using the Natural Sciences Reference Collection in the Otago Museum . Additional references used for the shell identifications included Crowe (1999) and Raven and Bracegirdle (2011) . With the exception of the shell, the faunal material was then run through standard pro- tocols which involve a two stage process: firstly, the material was sorted into primary anatomical units which are defined as the sided element; secondly, these units were identified to the lowest possible taxonomic level (e g. ,. Leach 1986) . The fish assemblage was analysed following a revised methodology that builds on the conventional methodology developed by Leach (1986) and widely used in New Zealand and the Pacific . This method counts the five main paired mouth bones: the left and right dentary, articular, quadrate, maxilla and premaxilla – along with ‘special’ bones for some species . These bones were selected by Leach because they are generally distinctive to a low taxonomic level and survive well in archaeological contexts . Leach’s method is widely applied in New Zealand and can be used to make comparisons between different sites and assemblages . Recently several researchers working in the tropical Pacific have extended the number of bones that are identified (Vogel 2005; Walter 1998; Weisler et al . 2010) . They found that with an expanded range of elements, taxa not identified by the conventional method could now be identified; that some of these taxa were actually quite common in the assemblage; and that the relative abundances of taxa changed as more elements were identified . An extended set of fish bone was identified for the Torpedo Bay fishbone assemblage, including: the paired sub-cranial elements palatine, hyomandibular, ceratohyal, epihyal, opercular, preopercular, cleithrum, scap- ula, supracleithrum and postemporal; the unpaired cranial elements vomer and parasphenoid; and vertebrae, identified to atlas (first vertebra), thoracic vertebrae, caudal vertebrae and urostyle (last vertebra) . This extended methodology has since been more fully described and formalised (Campbell 2016) . A total of 1263 vertebrate and 1579 invertebrate elements were analysed and assigned to basic faunal categories (Table 1) . However, it must be noted that unidentifiable fish and shellfish fragments were not counted, whereas unidentifiable bird and mammal bones, where they could be assigned to class, were counted, and where they could not be assigned to class, were classed together as ‘unidentified ’. The majority of the identifiable vertebrate fauna was bird . Of the 1263 60 Torpedo Bay

vertebrate elements, many were unidentifiable beyond the basic faunal classes shown in Table 1, including 358 fragments of bird bone (some of which might have been identifiable given enough time) and one bird long bone that wasn’t in the reference collection . Fewer fish bones were iden- tified to a low taxonomic level but many other bones could be identified if an even wider set of elements was chosen for analysis . Two quantification methods are used, NISP and MNI . NISP (Number of Identified Specimens Present) is a simple count of identified elements for any taxonomic group . MNI is the Minimum Number of Individuals, and is calculated in the most common element by side for each taxon, e g. ,. right femur . There are problems associated with each count (Grayson 1978; Reitz and Wing 2008) but in general, NISP is preferred . MNI counts are given in order to facil- itate comparisons with other published assemblages but in general they are not discussed fur- ther; MNI for fish is based on the conventional method developed by Leach (1986); MNI only is given for shellfish, calculated as NISP for gastropods and NISP/2 for bivalves (left and right siding was not undertaken for bivalves) . The diagnostic element identified for shell was the hinge for bivalves and the apex or aperture for gastropods . Each layer is generally considered to be a discrete assemblage and no effort is made to subdivide these assemblages on the basis of excavated features or other zones within the layers . During discussion, Layers 4 and 5 are generally considered together as Phase 1 and Layers 2 and 3 as Phase 2 (Chapter 3) .

Table 5.1. NISP values for basic faunal classes. Faunal class NISP Bird 766 Fish 339* Mammal 53 Reptile 89 Shell 1579* Unidentified 18 Total 2842 * non-diagnostic elements and fragments not counted

Bird

A total of 766 bird bones were identified in the assemblage . Twelve bird taxa are present including five extinct species, three of which are moa, though we note that with the development of DNA analysis there has been some doubt on the accuracy of moa species identification to low taxonomic levels on morphology alone (Bunce et al . 2003) . The results are summarised by layer in Tables 5 .2 and 5 .3 . The range of species represented is not large and the majority of the bird bone (760 from a total NISP of 766) is from layers 4 and 5 . The most common bird in the assemblage is the spotted shag (Phalacrocorax punctatus) making up over 70% of the identified bird bone by NISP . One spotted shag bone showed evidence of cut marks and several have been gnawed by kiore (Polynesian rat, Rattus exulans) . Spotted shag wing bones were also found in articulation on the surface of Layer 4, indicating that they were deposited very soon before the Layer 3 slope wash event (in this context they may not necessarily be cultural in origin) . Faunal analysis 61

Table 5.2. NISP values for the bird bone assemblage by layer. Layer Common name Taxon 3 4 5 Albatross/mollymawk Diomedeidae 2 Coastal moa* Euryapteryx curtus 4 1 Fluttering shearwater Puffinus gavia 1 Kaka Nestor meridionalis 1 25 36 Little bush moa* Anomalopteryx didiformis 1 cf. Mappin’s moa* cf. Pachyornis geranoides 1 5 Moa sp* Dinornithiformes 11 20 New Zealand raven* Corvus antipodum 1 North Island snipe* Coenocorypha barrierensis 2 Salvin’s prion cf. Pachyptila salvini 1 Spotted Shag Phalacrocorax punctatus 59 235 Weka Gallirallus australis 1 1 Unidentified 3 50 305 Total 6 151 609 * extinct taxa

Table 5.3. MNI values for the bird bone assemblage by layer. Layer Common name Taxon 3 4 5 Albatross/mollymawk Diomedeidae 1 Coastal moa* Euryapteryx curtus 1 1 Fluttering shearwater Puffinus gavia 1 Kaka Nestor meridionalis 1 2 3 Little bush moa* Anomalopteryx didiformis 1 cf. Mappin’s moa* cf. Pachyornis geranoides 1 1 Moa sp* Dinornithiformes 1 1 New Zealand raven* Corvus antipodum 1 North Island snipe* Coenocorypha barrierensis 1 Salvin’s prion cf. Pachyptila salvini 1 Spotted Shag Phalacrocorax punctatus 3 12 Weka Gallirallus australis 1 1 Total 3 9 23 * extinct taxa

Although it is not a large assemblage and the range of bird species is not great, it is clear that the species come from both coastal and forest habitats which is consistent with other bird assemblages from well-described sites of a similar age (e g. ,. , Allen and Holdaway 2010; Cook’s Cove, Walter et al . 2011) indicating a broad range of small bird hunting . The higher number of spotted shags probably indicates that a nearby colony was being targeted and that the remaining bird species were being taken opportunistically . There are five extinct bird species present . Of these five, it should be noted that three are moa and as we have noted above the identification of moa species on morphology alone is prob- lematic . Nevertheless, the presence of even a single moa species indicates that the lower layers of this site are likely to date to the earliest period of human settlement . Moa species were never as abundant in the North Island as in the South Island (Worthy and Holdaway 2002) and it 62 Torpedo Bay

is unlikely that the populations survived for long following human arrival . North Island snipe (Coenocorypha barrierensis) has only been identified in one other North Island site but 13 New Zealand raven (Corvus antipodum) specimens have been identified (Worthy 1997) . North Island snipe did not become extinct until the late 19th century (Tennyson and Martinson 2006) . All of the identified moa bone is from the leg, implying that the birds were being killed and butchered elsewhere with the fleshy parts returned to the site, though numbers are small and this could be a sampling issue . This includes not only the femur which would have carried a large amount of meat but also phalanges which are unlikely to have had any meat value . This suggests that whole legs were being brought on the site as food rather than just long bones being acquired for industrial purposes . All of the long bones were broken and it may be that they were used for fish hook manufacture after they had been stripped of the meat . We note, however, that no bone fish hooks or other bone artefacts were recovered during the excavation, nor any drill points or sandstone files used to make bone hooks .

Reptile

Tuatara (Sphenodon punctatus) is the last surviving member of a group of reptiles that first appeared in the and which died out elsewhere 60 million years ago (Worthy and Holdaway 2002: 459) . They are the largest reptile in New Zealand and were once widespread throughout both the North and South Islands particularly on well-drained soils . Fossil remains are common in sand dunes (Worthy and Holdaway 2002: 461) . Several tuatara elements were identified from a single feature from Layer 5 with a total NISP of 89, but an MNI of only 1 . Many of the identified elements are paired suggesting that an almost complete skeleton was present in this feature . It may have been naturally deposited .

Mammal

There was only a small amount of mammal bone recovered during the excavation (NISP = 47) and it was not possible to identify all of it to species . The species identified are summarised by layer in Tables 5 4. and 5 5. . The majority of the mammal bone is from Layers 4 and 5 (NISP = 36) and the range of species present, i e. ,. dog (Canis familiaris), kiore, sea lion (Phocarctos sp .)1 and fur seal (Arctocephalus forsteri), is typical of sites of this period . The single fur seal element is an unfused vertebral epiphysis from a juvenile, suggesting a nearby breeding colony . From around AD 1500 fur seals no longer bred in the North Island (Smith 2002) . The most unusual element in Layer 5 is a single pig (Sus scrofa) phalange; the presence of this bone is discussed in more detail below . Three of the Layer 5 dog bone fragments were burnt and one displayed evi- dence of gnawing . Although dog and sea lion are both present in Layer 3 they are only represented by five fragments of rib (dog) and a fragment of cranium (sea lion) . A single pig tooth is also present but is interpreted as a historical intrusion . The pre-human breeding ranges of fur seals and sea

1 Recent DNA and osteological analysis has shown that the sea lion endemic to mainland New Zealand (Phocarctos sp ). became extinct soon after human arrival and was replaced by subantarctic populations of geneti- cally distinct P. hookeri (Collins et al . 2014) . Reference to sea lion in this chapter is to this extinct lineage (Collins et al . do not refer to ‘species’) . The same is true of the yellow-eyed penguin Megadyptes( antipodes), which replaced the extinct Waitaha penguin (M. waitaha) (Collins et al . 2014) and the black swan (Cygnus atratus), an Australian species that replaced the closely related but genetically and morphologically distinct Poūwa (C. sumnerensis) (Rawlence et al . 2017) . Faunal analysis 63

Table 5.4. NISP values for the mammal bone assemblage by layer. Layer Common name Taxon 3 4 5 Dog Canis familiaris 5 9 4 Fur seal Arctocephalus forsteri 1 Pig Sus scrofa 1 1 Kiore Rattus exulans 1 9 Sea lion Phocarctus sp. 1 2 1 Unidentified 2 7 Unidentified sea mammal 2 1 Total 11 12 24

Table 5.5. MNI values for the mammal bone assemblage by layer. Layer Common name Taxon 3 4 5 Dog Canis familiaris 1 1 1 Fur seal Arctocephalus forsteri 1 Pig Sus scrofa 1 1 Kiore Rattus exulans 1 1 Sea lion Phocarctus sp. 1 1 1 Total 3 3 5 lions included all of the North Island but by AD 1500 the evidence suggests fur seals at least no longer bred north of about Marlborough (Smith 2002; Collins et al . 2013) . The single sea lion bone from Layer 3 may be from a non-breeding straggler or it may be intrusive from Phase 1, though it was noted that no Layer 3 features cut as deep as Layer 5 . Two human finger bones were also identified from Layer 3 . One of these displayed well defined muscle features along the shaft and wear at the distal end which is probably from flexing . There was also some form of possibly arthritic bone change on the condyle at the distal end .

Fish

The fish bone from the assemblage has a total identified NISP of 339, although this does not include unidentified elements . A total of seven taxa were identified, although one taxa could not be identified beyond subclass (Chondrichthyes) and the ray in Layer 2 was identified from a tail spine . NISP, calculated on the extended set of elements are given in Table 5 .6, and MNI, calculated on the conventional set, are given in Table 5 7. . Snapper (Chrysophrys auratus) dominate all assemblages, with occasional kahawai (Arripis trutta) and gurnard (Chelidonichthys kumu) representing bony fish (gurnard in Layer 3 and kaha- wai in Layer 4 were only identified from vertebrae); along with eagle ray Myliobatis( tenuicau- datus) identified from toothplates in Layer 3, school shark Galeorhinus( galeus) in Layer 3, a ray identified from a tail spine in Layer 2 and unidentified shark in Layers 3 and 5 representing cartilaginous fish . The counts for cranial elements and vertebrae for snapper are given in Table 5 .8 . A com- parison of these two counts allows us to detect differential treatment of body parts, i e. ,. that heads were being treated differently to bodies on site . Both Shawcross (1972) at Houhora and Nichol (1988) at Houhora and the Sunde site compared cranial and vertebral counts and came to 64 Torpedo Bay

Table 5.6. Extended NISP values for the fish bone assemblage by layer. Layer Common name Taxon 2 3 4 5 Eagle ray Myliobatis tenuicaudatus 2 Gurnard Chelidonichthys kumu 4 Kahawai Arripis trutta 1 2 Ray sp. Myliobatoidei 1 School shark Galeorhinus galeus 1 Shark sp. Chondrichthyes 1 6 Snapper Chrysophrys auratus 36 42 48 193 Total 37 50 49 201

Table 5.7. Conventional MNI values for the fish bone assemblage by layer. Layer Common name Taxon 2 3 4 5 Kahawai Arripis trutta 1 Snapper Chrysophrys auratus 2 6 8 14 Total 2 6 8 15 similar conclusions but Torpedo Bay is the first site in New Zealand where this analysis has been formalised (Campbell 2016a) . The chi-squared test is used to compare the observed frequencies of a set of variables, in this case counts of cranial bones and vertebrae, with what the expected frequencies would be in an assemblage of that size, i e. ,. the proportions of cranial bones to verte- brae in a live fish . There are 32 cranial elements in the extended set2 while snapper have 24 vertebrae (Roberts et al . 2015: 1288) . As Table 5 shows, there are highly significant differences in the proportions of cranial bones and vertebrae in Layers 4 and 5 while the differences are less sig- nificant for Layers 2 and 3, though still positive . Some of these differences may be due to tapho- nomic factors, i e. ,. vertebrae and cranial elements may not survive equally well in archaeological contexts . However, the highly significant χ2 scores for Layers 4 and 5 (Phase 1) indicate that heads and bodies were being treated differently . While some snapper would have been consumed on site, others were probably being preserved for later, off site consumption . The preservation process involved removing the heads, which remained on site and the flesh on them probably eaten, while keeping the vertebrae with the preserved flesh . The shallow firescoops cut into the beach terrace in Layer 5 with a layer of volcanic rocks in their bases (Features 12, 1, 5 and 29

Table 5.8. Chi-squared scores comparing the counts of cranial elements to vertebrae for snapper from all layers. Layer Cranial Vertebrae χ2 p 2 15 20 2.917 0.087 3 32 10 6.222 0.012 4 38 10 9.507 0.002 5 141 38 34.19 >0.001

2 Including the vomer and cleithrum, which were not identified in the assemblage but were potentially part of the set . Faunal analysis 65

and Features A–F, Figures 3 .6 and 3 .8) may have been heated, raked out but the heated rocks left in place, and then used to steam or dry fish for preservation (Chapter 3) . Beattie (1994: 116) records: “at Otago Heads, said a Maori, the people used to put makaa (barracouta) on a stage, remove the bones, split the fish into layers two inches thick, plait these together, steam and hang up for the sun to dry . It was then packed away and would keep a long time ”. Anon . (1837: 90) at Hokianga describes the preservation of mackerel: “the natives prepare them on hot stones; they keep for months; they never attempt salting them ”. The firescoops in Layer 5 and the proportions of snapper cranial elements and vertebrae indicate that a similar process may have been used to preserve snapper at Torpedo Bay, either directly on the stones or on racks overhead . Apart from bony fish, some elements from cartilaginous fish (Chondrichthyes, sharks and rays) were also recovered . The vertebral centra of shark and ray are partly calcified but do not usually survive well in archaeological contexts, though shark and ray are known ethnographically to have been important food resources for Maori . The tooth plates of eagle ray Myliobatus( ten- uicaudatus), and the tail spines of stingrays (Dasyatis sp ). and eagle rays are much more calcified than cartilaginous vertebrae and do survive well . Of the vertebral centra that were recovered, one from Layer 3 was identified as school shark Galeorhinus( galeus) while the remaining seven verte- brae in Layers 3 and 5 were not identified to any lower taxonomic level, but they appeared to be all the same species, probably of medium size (roughly 1 5–2. m) . The standard paired mouth bones of snapper were measured in order to reconstruct the size of the live fish (fork length), following a method developed by Leach and Boocock (1995) . In order to get meaningfully large assemblages, Layers 4 and 5 were combined as Phase 1, and Layers 2 and 3 were combined as Phase 2 . From the 146 mouth bones in the assemblage, a total of 80 measurements were made on 48 bones from Phase 1, and 10 from Phase 2 . The recon- structed fork lengths of snapper in Phase 1 follow a roughly normal distribution, ranging from the 250 mm size class (250–300 mm) up to 550 mm with a median of 350 mm, in other words, neither particularly large nor particularly small . While 10 from Phase 2 is too low a number to say anything meaningful about, the pattern and size range is similar to that from Phase 1

6

Phase 1 3

15

12 Phase 2

9

6

3 Figure 5.1. Histograms of recon- 250 300 350 400 450 550 600 structed snapper mm fork lengths for Phase 1 and Phase 2. 66 Torpedo Bay

(Figure 5 1). . At the NRD site (R11/859, Campbell 2011) some very small snapper were found in the assemblage indicating that the snapper nursery in the Manukau Harbour was being exploited . The Waitemata was also likely to be a snapper nursery but there is no evidence of exploitation of very small snapper at Torpedo Bay . The snapper were all of a size to take a baited hook; three shell fishhook points were recovered from Layer 5 (Phase 1) and stone net sinker from Layer 3 (Phase 2) so both methods seem to have been used . The range of fish species in the assemblage is very restricted . Snapper were the dominant species and these can be caught either by baited hook or in drag or set nets . Kahawai were also probably taken with lures or in nets . Snapper spawn in bays and estuaries in spring and early summer (Paul 2000: 96) . The seasonal movements of kahawai are not well understood but they appear to spawn in late summer in inshore locations (Paul 2000: 93) . It would seem that line fishing from canoes for snapper and kahawai was undertaken possibly while they were shoaling through the harbour entrance for spawning . If this was the case, we can infer that the Torpedo Bay site was at least occupied over spring or summer .

Shellfish

Numbers of shellfish were low, with a total MNI of 1570 . This is in part due to only a sample of the shell assemblage being retained for analysis, but also reflects the sparse nature of the excavated middens . Thirty-two taxa were identified but several of these are small, non-eco- nomic species that would have been a bycatch (Table 5 9). . The majority of the shellfish was recovered from Layer 3 (63% of the total assemblage) and there is considerable variation in the composition of the shellfish assemblage between Phase 1 and Phase 2 (Figure 5 .2) . Rocky shore species green-lipped mussel (Perna canaliculata) and blue

40

35

30 Phase 1 Phase 2 25

20 percent 15

10

5

0 Pipi Oyster Tuangi Tuatua Scallop Cat's eye Cat's Mudsnail Horn shell Sea urchin Turret shell Turret Dog cockle Globisinum Blue mussel Fine dosinia Fine Lined whelk Morning star Arabic volute Arabic Purple cocklePurple Siphon whelk Cook's turban Cook's Dark rock shell Dark rock Speckled whelk Knobbed whelk Mud at top shell top Mud at Micromollusc sp? Angel wing borer Angel Oblong Venus Oblong Venus shell Ribbed slipper shell Circular slipper shell Circular Red mouthed whelk Green-lipped mussel Green-lipped Smooth white slipper Smooth white Purple mouthed whelk Purple

Figure 5.2. Percentage of MNI of shellfish by Phase Faunal analysis 67

Table 5.9. MNI values for the shellfish assemblage by layer. ‘Non-economic’ species are separated out by size. Layer Common name Taxon 2 3 4 5 Bivalves Angel wing borer* Barnea similis 1 Blue mussel Mytilus galloprovincialis 6 7 53 Dog cockle Tucetona laticostata 1 4 Fine dosinia Dosina subrosea 2 5 Green-lipped mussel Perna canaliculata 4 6 3 76 Morning star Tawera spissus 5 14 6 Oblong Venus shell Ruditapes largillerti 4 18 2 11 Oyster Saccostrea glomerata 62 38 38 Pipi Paphies australis 171 354 4 Purple cockle Venericardia purpurata 1 Scallop Pecten novaezealandiae 1 5 Tuangi Austrovenus stutchburyi 26 179 1 Tuatua Paphies subtriangulatum 1 1 2 Gastropods Arabic volute Alcithoe arabica 1 2 Cat’s eye Turbo smaragdus 23 268 7 Cook’s turban Cookia sulcata 1 4 Dark rock shell Haustrum haustorium 1 Globisinum Globisinum drewi 1 Lined whelk Buccinulum sp. 1 Mudflat top shell Diloma subrostrata 1 Mudsnail Amphibola crenata 44 19 Sea urchin Evechinus chloroticus 2 1 Siphon whelk Penion sulcatus 6 Speckled whelk Cominella adspersa 8 Non-economic gastropods Circular slipper shell Sigapatella novaezealandiae 2 Horn shell Zeacumuntus lutulentus 14 1 Knobbed whelk Austrofusus glans 13 Purple mouthed whelk Cominella glandiformis 2 8 1 Red-mouthed whelk Cominella virgata 4 Ribbed slipper shell Credipula costata 2 3 Smooth white slipper Credipula monoxyla 2 Turret shell Maoriculus roseus 14 Echinoderms† Sea urchin Evechinus chloroticus 2 1 Total 351 982 8 229 *Angel wing borer is large enough to be economic but burrow into soft rock and would be very difficult to harvest. †Echinoderms are not molluscs but are included here with shellfish. 68 Torpedo Bay

mussel (Mytilus galloprovincialis) and oyster (Crassostrea glomerata) are the most abundant species in Phase 1, while estuarine species tuangi (Austrovenus stutchburyi) and pipi (Paphies australis) are dominant in Phase 2 along with cat’s eye (Turbo smaragdus) .

Interpretation

The distinctive differences in the composition of the faunal assemblage between Layers 2 and 3 and Layers 4 and 5 supports the interpretation of the stratigraphy and chronology and justifies the division of the site into two phases . The lowest two layers (Phase 1) contain species that are typical of early period sites in New Zealand and a marked change is evident in the upper layers (Phase 2), with the loss of moa and a significant shift from rocky shore to estuarine shell- fish species . The excavated assemblage is small but nonetheless exhibits a wide diversity of taxa from a range of habitats . It is important to describe and understand this assemblage because the Torpedo Bay site is one of only a handful of sites representing initial settlement, or very soon after, in the upper North Island . The corpus of excavated sites from the 14th century is concen- trated on the east and south coasts of the South Island (this may be partly an artefact of aca- demic research at the University of Otago) and the faunal assemblages from a number of these sites have been well described . Indeed these have been used to describe what has been called a Moa-hunter phase which was focused on big-game hunting – moa, seals, etc . (Duff 1977) . Sites of this age from the more temperate north and out of the main habitat zones for moa are less well described . One of the difficulties in carrying out comparative studies with other sites from this time period is the fact that many of the early period assemblages were excavated at a time before the development of rigorous archaeozoological techniques . Consequently far less is known about faunal assemblages or the full range of diversity in fishing and hunting practices in the early settlement period . It is now well understood that archaeozoology international best practice requires a well-designed programme of methods and research from the field into the lab . Of par- ticular relevance is screen size which can have a significant impact on the range of fauna recov- ered from sites (Reitz and Wing 2008: 118; Allen 2014) . The use of a fine mesh screen during the Torpedo Bay excavation means that bones from fauna such as kiore, small fish and small bird were recovered, providing a good sample of the range of species present at the site . The following sections of this report consider the faunal assemblage at the Phase level and deal with issues of ecological adaptation, catch strategies and change through time, followed by a consideration of the Torpedo Bay faunal assemblage in its regional context .

Phase 1

The fish assemblage is dominated by snapper which could have been caught either inside or outside the harbour . The analysis of snapper vertebrae indicated that during this phase fish were probably being preserved for later consumption . This type of analysis has not been com- monly carried out in New Zealand (see, for instance, Nichol 1988 for the Sunde site), but similar conclusions have since been developed for fishbone assemblages at Parton Road, Papamoa and at Urquharts Bay, Whangarei Heads (Campbell 2016) . The most abundant shellfish taxa are mussel and oyster making up 71 per cent of the assemblage . These are rocky shore species and would have been available locally . This is a Faunal analysis 69

common pattern at many early sites where large rocky shore bivalves dominate early assemblages in suitable environments and some at least were used for fishhooks . The birds derive from a range of habitats . Forest species are represented by kaka (Nestor meridionalsi), weka (Gallirallus australis) and the extinct raven . Marine birds include shag and a prion and there is a single wetland species, North Island snipe . The habitat range suggests that the small birds were opportunistically hunted rather than specifically targeted . The one excep- tion is the spotted shag which is the most abundant species in the assemblage . Spotted shags are known to nest around the Waitemata Harbour and there may have been a nearby colony that was easily exploited . There are three possible species of moa present which is high diversity for such a small assemblage but as we discuss above moa is difficult to identify to species level based on bone morphology alone (Bunce et al . 2003) . Almost all of the moa bone is from the leg suggesting that the birds were being butchered elsewhere and only the meat bearing parts brought back to the site . This bone is also commonly used for the manufacture of items such as fish hooks although no bone hooks or other artefacts were found . It is interesting that processed (butchered) moa was being brought on-site for consumption while snapper were being processed and taken off-site . The range of mammals present is typical of other early period faunal assemblages com- prising dog, kiore and seal, albeit in low numbers . The fur seal is represented by a single unfused vertebral epiphysis from a juvenile . Since fur seals live in breeding colonies over the summer this may provide further support for summer occupation of the site . There was no evidence that any of the dog bone was from sub-adults . It is especially unusual to see pig in the early phase of this site . Pigs were part of the Austronesian domesticates that travelled throughout the Pacific but New Zealand is one of the rare places to not have pig and chicken . Pig does appear in layers from this period in some sites but in all cases this is considered to be intrusive . It is highly unlikely that the single pig bone relates to pre-European Maori and it is also considered to be intrusive .

Phase 2

The assemblage is still dominated by snapper in Phase 2, although there are less of them . The analysis of vertebrae does not indicate as clearly as for Phase 1 that snapper were being pre- served for off-site consumption . The data indicates that subsistence strategies regarding snapper changed between Phase 1 and Phase 2, from a strategy of preservation for later off-site con- sumption to one that may have combined both on-site and off-site consumption . There is a significant shift in focus from the rocky shore shellfish species of mussel and oyster to estuarine pipi and cockle supported by cat’s eye . This may be indicative of a decline in availability of rocky shore species as a result of over predation during Phase 1, a pattern found in many other multi-phase sites in the upper North Island . There are only four bird species present and these are represented by only a handful of bones . There is also very little mammal bone . This shift from a broad spectrum foraging strategy to a more restricted range of species has been well-documented in New Zealand particularly in the South Island where there is a marked shift from big-game hunting to an increasing dependence on fishing and shellfish collecting (Nagaoka 2002) . As Allen (2012) notes, however, in the more temperate north the situation may be more complex because of the availability of horticultural crops which would have tempered the demands on wild resources . 70 Torpedo Bay

In sum, we note four main changes from Phase 1 to Phase 2: first, there is a decline in the exploitation of birds – extinct species including moa are restricted to Phase 1; second, there is a decrease in mammal from Phase 1 to Phase 2; third, there is a change from preservation of snap- per in Phase 1 to on site consumption; lastly, there is a shift from mostly rocky shore shellfish species from Phase 1 to soft shore species in Phase 2, although cat’s eye becomes common .

Regional Context

The patterns described above are common to many of the Hauraki Coromandel sites . These patterns are discussed below in relation to the early period, restricting the comparative analysis to those sites which have been identified by Smith and James-Lee (2010) as having been investigated recently and for which there are reliable radiocarbon dates 3. For the purpose of this analysis the early period is considered to terminate at ca AD 1450 . For data tables see Smith and James-Lee (2010; counts are presented as MNI, following the original sources) and Brooks et al . (2012) . Snapper is the most abundant fish species among the comparative sites with the exception of Hahei where leatherjacket (Parika scaber) is the most common and Port Jackson where kaha- wai dominates (Brooks et al . 2012: Table 13) . A characteristic of the sites in the Hauraki/Coromandel region is the broad diversity of bird species present . In the majority of sites there is no one species that stands out as particu- larly abundant although little blue penguin (Eudyptula minor), kaka (Nestor meridionalis) and tui (Prosthemadera novaeseelandiae) are recurring species in several of them . This seems to sug- gest that during the early period small bird hunting was largely opportunistic although certain forest birds may have been more desirable for their feathers . As Davidson (1979: 188) observes, the pattern of avian exploitation on the Coromandel Peninsula is characterised by “generalised hunting without marked concentration on particular species ”. At sites where there is a dominant species, such as shag at Sunde (Brooks et al . 2012: Table 14), this is most likely a result of the presence of a nearby colony that could be easily exploited . In the case of Torpedo Bay Layer 5, the 235 shag bones (MNI = 12) could result from a single hunting trip . Moa is present in several sites but never in any quantity and in some cases it is only present as industrial bone with no evidence of butchery for food . As we discuss above, moa populations in northern New Zealand were never particularly abundant and moa hunting did not contribute significantly to subsistence practices in the Hauraki/Coromandel region . This in apparent con- trast to Houhora in Northland where there is a moa NISP of at least 1385 (Furey 2002: 115) . The Hauraki/Coromandel area would have provided few suitable habitats for supporting large populations of moa . Mammal bone is present in all of the comparative sites although the minimum numbers of the various species are low (Brooks et al . 2012: Table 16) . Osteological analysis of the mammal bone suggest that both adult and juvenile seals and dogs are present in these sites, although sub- adult dogs are more common than adults (Smith 1981; Furey 2002: 114) . Breeding populations of sea lions and fur seals occurred in the Hauraki region during the early period but seem to have disappeared by about AD 1500 (Smith 2002) .

3 The seven sites accepted by Smith and James Lee are: the Sunde site (R10/25), Port Jackson (S09/53), Cross Creek (T10/399, Layers 7 and 9), Hahei Beach (T11/242), Hot Water Beach (T11/115), Tairua (T11/62) and Whitipirorua (T12/16) . These sites and others are discussed in detail in Chapter 7 . Faunal analysis 71

Broadly speaking, the shellfish assemblages indicate a preference for rocky shore species among the comparative sites (Brooks et al . 2012: Table 15) . The most frequently occurring rocky shore species are green-lipped mussel, cat’s eye, white rock shell (Dicathais orbita), black nerita (Nerita atramentosa) and limpets (Cellana sp ). . Later sites are usually dominated by soft shore species . Torpedo Bay fits this pattern comfortably . It must be noted, however, that the environ- mental contexts of these sites will have some impact on the range of species present . The Port Jackson site, for instance, is unusual in the dominance of tuatua (Paphies subtriangulata) over rocky shore species . Tuatara is known from early period sites, e g. ,. at Houhora (Furey 2002), Cook’s Cove (Walter et al . 2011), Port Jackson, Hotwater Beach, Sarah’s Gully and Opito (Davidson 1979), but its role in subsistence is not well-described in the literature .

Conclusion

The Torpedo Bay faunal assemblage has provided an important contribution to the small suite of well described early period North Island sites . Although hunting strategies and subsist- ence practices are reasonably well understood for the South Island, information about what was happening at the same time in the North Island has been hampered by a lack of excavations from this period and by the quality of the analysed data from those sites that have been exca- vated . Recent reanalyses (Furey 2002; Allen and Holdaway 2010; Smith and James-Lee 2010) have begun to address this lack of comparative data . The patterns of northern New Zealand subsistence strategies described by Davidson (1979) still seem to generally hold true . Moa were hunted, perhaps opportunistically, but did not play a significant role in the diet . Instead, hunting of small birds, fishing and shellfish gathering along with hunting of sea mammal was carried out . Snapper were the preferred fish species and rocky shore shellfish such as mussels were preferentially selected . These wild-caught resources would have been supported by meat from domestic dogs . In these North Island sites horticultural crops also played a significant role but this is much more difficult to quantify . Although the quantifiable faunal assemblage from the later phase of the Torpedo Bay site is small, it too seems to reflect the broad patterns of change noted for northern New Zealand with a shift to soft-shore shellfish species and a decline in both mammal and small bird hunting .

6 Environment

Two analyses are used to contribute to our understanding of the local environment at the time of human occupation . Charcoal analysis provides evidence on what type of wood were burnt on the site; there is generally a reasonable assumption that firewood was gathered locally, while charcoals incorporated into soils will reflect vegetation prior to burning . Charcoal analysis was undertaken by Rod Wallace of Auckland University . Microfossil analysis of pollens, starch grains, phytoliths and other persistent small plant parts reflects the plants actually growing on or around the occupation surface, including evidence of garden crops . Wind-borne pollens may reflect vegetation over a wider area . Microfossil analysis was undertaken by Mark Horrocks of Microfossil Research Ltd .

Charcoal

Charcoal was identified by its distinctive cell patterns under high magnification . The split and snapped sections of charcoal were examined with an incident light compound microscope at 50–500 times magnification . Between 20 and 30 pieces of charcoal were identified from each bag until no ew species were found .

Results

All samples were reported to have been associated with cooking structures and are assumed to be the remains of firewood collected by the occupants from the local landscape . Most firewood would have been sourced locally, either from local trees, stumps remaining from dead trees or driftwood (Wallace and Holdaway 2017) . The site was on a beach immediately adjacent to a tidal channel that is quite capable of delivering driftwood originating from many kilometres away to the site, as mangrove (Avicennia marina) charcoal from one sample probably indicates . The results are summarised in Table 6 1. by Phase . The Phase 1 assemblage is dominated (71%) by large tree species with the remainder being a range of shrubs and small trees typical of coastal environments . Kauri (Agathis australis) was the dominant conifer but significant amounts of totara Podocarpus( totara), matai (Prumnopitys taxifolia) and kahikatea (Dacrycarpus dacrydioides) were also present . The hardwoods were domi- nated by pohutukawa (Metrosideros excelsa) which is to be expected of shoreline vegetation . Puriri (Vitex lucens) and hard beech (Nothofagus truncata) also make a significant contribution along with small amounts of hinau/pokaka (Elaeocarpus dentatus or E. hookerianus) and titoki (Alectryon excelsus) . Pohutukawa and puriri are the most common species in the whole assemblage but are equally abundant in the early and late period layers and are still the dominant trees on the coastal cliffs in the area today . They are persistent species tolerant of forest clearance, fire and other disturbance and by themselves provide little indication of the vegetation history of this site . Hard beech was abundant in Phase 1 but completely absent in Phase 2 . It is virtually extinct in the metropolitan area of Auckland today, occurring only in stands in the Hunua ranges in association with kauri (Collins and Burns 2001) . In the 19th century, however, it was abundant on down to sea level (Kirk 1878: 445) . On Motutapu it was the dominant species in a burn zone in a palaeosol sealed under the volcanic ash from the nearby Rangitoto volcano (Ladefoged and Wallace 2010) . This charcoal originated from vegetation burnt when the volcano erupted around 600 years ago . Beech was also the most common tree species in charcoal samples from the Pig Bay site on Motutapu but this is probably natural char- 74 Torpedo Bay

coal that has washed downstream to be incorporated into the lowest layers of the site (Davidson and Leach 2017: Appendix 1) . The Phase 2 assemblage is dominated (60%) by a suite of species characteristic of veg- etation that follows forest clearance . This includes bracken fern Pteridium( esculentum), tutu (Coriaria arborea), Hebe, Coprosma, manuka (Leptospermum scoparium) and fivefingerPseudopanax ( arboreus) . Bracken is not a woody plant and yields fragile charcoal that survives poorly in sites so the amount present in archaeological assemblages probably seriously underestimates its abun- dance on the prehistoric landscape . Its occurrence is strongly associated with charcoal fragments in pollen cores as it is a pioneering coloniser of land after fire (McGlone and Wilmshurst 1999) . Tutu is small straggling shrub with narrow stems rising from a multi-branched base . It is a nitro- gen-fixing species that colonises bare, disturbed ground and is typically seen today growing on the banks of road cuttings . It is also strongly associated with bracken and charcoal fragments in pollen cores . Hebe and Coprosma are small shrubs typically abundant in charcoal samples in association with bracken, tutu and a range of other small shrub species and signal an open, tree free landscape . Pohutukawa and puriri make up most of the rest of the late assemblage but, as discussed above, provide little indication of the vegetation in the wider area . The tidal mudflat mangrove is present in one of the Phase 2 samples . It doesn’t grow within 6 km of the site today but as all its branches are shed into the sea and become driftwood it was almost certainly collected for firewood from local beaches . It does, however, signal that wood of other non-local species might also have entered the assemblage in this way . This may apply to the only conifer in the late

Table 6.1. Summary of charcoal results by Phase. Plant type Species Phase 1 % Phase 2 % Bracken Bracken 6 2.5% Tutu 31 Shrubs typical of Hebe 2 37 open landscapes Coprosma 6 3% 42 57% Manuka 16 Pseudopanax 2 10 Ngaio 12 2 Olearia 35 4 Other shrubs and Wharangi 4 26% 1 4% small trees Mapou 11 Mahoe 35 2 Persistent Pohutukawa 109 43 39% 29% broadleaf trees Puriri 36 25 Titoki 3 Kowhai 2 Other broadleaf Kohekohe 1 8% 3% trees Hinau/Pokaka 1 3 Towai 1 Beech 25 Matai 14 Kahikatea 7 Conifers 24% 4% Totara 16 Kauri 49 10 Intertidal Mangrove 1 0.4% Totals 367 100% 237 100% Environment 75

assemblage, kauri . Alternatively, kauri branches and stumps are so resinous they can survive on land surfaces for centuries after the trees have died . Other broadleaf tree species were present but only in very small amounts . These include kowhai Sophora( microphylla), kohekohe (Dysoxylum spectabile), hinau/pokaka and towai (Weinmannia silvicola) or tawheowheo (Quintinia serrata) . The latter two species are not closely related but are not distinguishable under the microscope .

Summary

During the Phase 1 occupation the local vegetation around North Head was closed canopy broadleaf podocarp forest right down to the waterline . This forest type is characterised by very large conifers rising over a dense understory of somewhat smaller hardwood trees . Shrubs and smaller tree species are a completely different suite to those in the later assemblage and were probably those limited to the forest floor or, more likely, on the forest edge along the shoreline itself . The Phase 2 assemblage appears to reflect a landscape largely cleared of forest with the few remaining tree species being restricted to coastal cliffs . Away from these cliffs vegetation seems to have consisted mainly of bracken fern with the woody species being mainly tutu, Hebe, Coprosma, fivefinger and manuka .

Microfossils

Nine samples from archaeological deposits at Torpedo Bay (including soil layers, fires- coops, midden and dog coprolites) were analysed for pollen, phytolith and starch remains to provide a record of past vegetation, environments and human activity .

Results

All samples contained microscopic fragments of charcoal, reflecting human activity around the sites from burning of vegetation and cooking fires . All samples, except the one from Feature 13, Layer 5, had sufficient pollen for analysis . Fern spores, especially bracken, dominate the samples (Figure 6 1. . Pollen percentage diagram from Torpedo Bay (wb = west baulk, + = found after count) ). . Bracken spores in association with hornwort (Anthocerotales) spores and paucity of tall tree pollen indicate burning of vegetation . Bracken, an invasive ground fern with widely dispersed spores, is often abundant in New Zealand pollen spectra of the last millen- nium, almost always associated with large scale repeated burning of forest by early Maori . It may form dense stands, averaging 1–2 m tall, over extensive areas . Hornworts are small inconspicu- ous plants that colonise freshly exposed soils and are also associated with forest burning . Cyathea tree ferns, spores of which show a high value in the sample from Feature 4, Layer 5, typically colonise gullies in fernland . Ferns with monolete spores also feature . Many of New Zealand’s numerous fern species have this spore type, which is difficult to differentiate between genera . However, all are ground ferns with many species favouring disturbed forest environments . Pollen of manuka/kanuka (Leptospermum scoparium/Kunzea ericioides), tutu, grasses (Poaceae) and puha (Sonchus kirkii) also feature in some of the samples, also indicating forest dis- turbance . The native puha has edible shoots and leaves and featured in prehistoric (and modern) Maori diet . The pollen evidence also indicates that remnants of conifer-hardwood forest at the sites included kauri, matai, pohutukawa, rewarewa ( excelsa), rimu (Dacrydium cupressi- num), tanekaha/toatoa (Phyllocladus trichomanoides or P. toatoa), totara and Coprosma . 76 Torpedo Bay

Figure 6.1. Pollen percentage diagram from Torpedo Bay (wb = west baulk, + = found after count). Environment 77

The presence of pollen of European introduced plantain (a highly invasive rosette weed of disturbed environments) in the sample from Feature 5, Layer 3, suggests some disturbance to the deposit though this is not supported by the archaeology . It is more likely that pollen has perco- lated down to Layer 3 prior to excavation or has contaminated the soil or sample during excava- tion . This highlights a difficult problem with microfossil analyses where movement of pollen and starch grains through the deposit is difficult to control for . Phytoliths provide further insight into the local environment (Figure 6 .2 . Phytolith percentage diagram from Torpedo Bay (other biogenic silica shown in red, wb = west baulk, + = found after count) ). . The phytolith assemblages of most of the samples are dominated by trees and shrubs, including nikau (Rhopalostylus sapida) . This forest palm is one of the few New Zealand species that allows phytolith identification to species level (Kondo et al . 1994) . Nikau phytoliths are spherical spinulose and this type occurs only in palms (Arecaceae) and bromeliads (Bromeliaceae) (Piperno 2006) . New Zealand has no indigenous bromeliads and nikau is the only species of palm . The very large amounts of nikau phytoliths in most of the samples may in part reflect local activities such as weaving and thatching . Grass phytoliths also show some high values, increasing first in the coprolites from Layer 4 and increasing further in subsequent layers . This most likely reflects intensified forest clearance . Other types of biosilicate remains identified, namely (generally brackish water types), radiolarians and sponge spicules, reflect prox- imity to the sea and use of marine resources at the site . Starch consistent with the tuberous root of kumara (Ipomoea batatas) was identified in the sample from the soil layer, Layer 4 (Figure 6 .3 . Starch concentration (per c c. ). diagram from Torpedo Bay (• = found in pollen preparations) ). . This was present as individual starch grains, all highly degraded, in a sufficient amount to count . In addition, well preserved kumara starch grains were found in small quantities in the pollen preparation of the coprolite sample .

Summary

Changes between Phases 1 and 2 are not as marked for the microfossil assemblages as they are for the charcoal . Tree pollens are not common at any stage while fern spores are common throughout the sequence . Tree phytoliths are common throughout the sequence while grass phy- toliths increase . Puha pollen also increases . Generally, the microfossil evidence shows that the site and its general environs were cleared of forest prior to occupation .

Discussion

The charcoal and microfossil results appear to give somewhat different readings of the local environment at the time of the Phase 1 and Phase 2 occupations, however these are not incompatible with the archaeology of the site . It was proposed that Layer 4 was a slopewash of cleared and gardened soil from the adjacent slope of Maungauika / North Head . This is sup- ported by the high concentrations of kumara starch in the soil sample from this layer . Tree pollen was scarce throughout the sequence while brackens and hornworts were present through- out, indicating that the land was already cleared of forest vegetation at the time of the Layer 5 occupation . Conversely, charcoal from forest trees and understorey shrubs dominated the Phase 1 assemblages while charcoal from open landscape shrubs and bracken did not become common until Phase 2 . These results indicate that the beach terrace and particularly the slopes above it were cleared of forest vegetation before the terrace was first occupied (noting the highly mobile nature of the beach terrace sands we cannot rule out an earlier occupation evidence for which subsequently blew away), but that forest woods remained on the landscape, and were probably 78 Torpedo Bay

Figure 6.2. Phytolith percentage diagram from Torpedo Bay (other biogenic silica shown in red, wb = west baulk, + = found after count). Environment 79

Figure 6.3. Starch concentration (per c.c.) diagram from Torpedo Bay (• = found in pollen preparations). available as driftwood, and were exploited for firewood . The Layer 5 occupation, then, followed the clearance of the slopes and establishment of gardens . If the forest cover on Maungauika / North Head was cleared by burning we might expect much more charcoal in the assemblages, particularly of forest understorey, especially the Layer 4 soil samples which would be expected to incorporate charcoal lying on the surface from clearance fires on the slopes of the maunga, whereas the Layer 5 samples would be more likely to reflect wood bought to the beach terrace for cooking . However, this is not the case; there is little difference between the Layers 5 and 4 charcoal assemblages (Table 6 .2) . This doesn’t mean that the forest was necessarily cleared with tools; it rather indicates that the complexity of human–environment interactions cannot always be reduced to simple processes .

Table 6.2. Summary of charcoal results by Phase 1 Layers. Plant type Layer 5 % Layer 4 % Open landscape shrubs 3.1% Other shrubs and small trees 28.4% 12.5% Persistent broadleaf trees 39.4% 37.5% Other broadleaf trees 6.7% 17.5% Conifers 22.4% 32.5%

7 Discussion and Conclusion

The small excavation at Torpedo Bay has revealed significant information about the early settlement of Tamaki . It was the first excavation of an early site to be reported from the Tamaki mainland – others have been reported from Motutapu and Ponui Islands in the Hauraki Gulf, and subsequent excavations at the Masonic Tavern site and the Long Bay Restaurant site have also found early material, but these have not yet been reported . Other early sites are known or at least suspected in Tamaki, often on the basis of artefacts or moa bone found in association with them in the intertidal zone, but have not been excavated or reported . To provide an appropriate framework for the discussion of the Torpedo Bay site, particularly the significant early com- ponent, it is necessary to extend the context beyond Tamaki to other reported early sites in the upper North Island . Ideally, chronometric hygiene protocols would be applied to the available radiocarbon dates and only well dated sites used in ths exercise . This approach has been adopted by Smith and James-Lee (2010; Smith 2010, 2013), discussed in Chapter 2, where it was suggested that this is one way to approach the problem of the Archaic ~ Classic model, by looking for patterns in the data without the expectation that they must fit a pre-existing model . Thus sites that were securely dated to no later than AD 1450 (Smith and James-Lee’s Early period) would be eligible for inclusion in this discussion . This has also provided the basis for the discussion of the regional context of the Torpedo Bay faunal assemblages in Chapter 5 . There are, however, some problems with this approach . The reliance on strict chronometric hygiene protocols means that most reported sites in the upper North Island that are clearly early, based on artefact styles or the presence of moa or seal bone, are not captured . Radiocarbon dates are not the only criterion for assigning a site to Early, Middle of Late periods . Artefact styles in particular are highly diagnostic of early occupation . Unfortunately, artefact styles cannot be used to differentiate the Middle and Late periods recognised by Smith and James-Lee . At the same time, an uncritical reliance on artefact styles alone risks merely restating the Archaic ~ Classic model . It is, however, beyond the scope of this report to develop a fully realised alternative to the two stage model, no matter how outdated it is . The following discussion focusses on the archae- ology of the upper North Island up to around AD 1450, the period equivalent to the Archaic, while bearing in mind the critique of this model presented in Chapter 2 .

The early archaeology of the upper North Island

Early sites have been excavated in a number of places in the upper North Island, primar- ily on the east coast . Sites may not have survived on much of the west coast where high energy seas have potentially destroyed much early occupation evidence (as at Tauroa Point, below), but it seems probable that sites survive in the wide harbours between Kawhia in the south and Herekino in the north (Matatuahu and Kokohuia are examples of these, below) . This discussion is split into four geographic areas, partly because research into pre-1450 archaeology has largely concentrated on one of these areas (the Coromandel Peninsula) while other areas have been less studied, and in different ways; and partly because there appear to be real differences between these areas .

Northland

There have been several excavations of early sites in Northland though some have only been partly reported . The best known, and the largest, is Houhora (N03/59) on the mouth of 82 Torpedo Bay

Tom Bowling Bay

N03/59 Houhora M02/162 Twilight Beach P04/639 Tauranga Bay

Q05/682 Mangahawea Bay N05/302 Tauroa Point

Tawhiti Rahi O06/317 Kokohuia

T08/5 Harataonga

Figure 7.3

Figure 7.2

Q11/344 Matatuahu R11/2379 Timberly Road

U13/875 Bowentown U14/363 Pilot Bay V14/40 Te Tumu N V14/187 Maketu

0 50 km

Figure 7.1. Excavated early sites in the upper North Island. the Houhora Harbour at the base of Mt Camel on the Aopouri Peninsula . The site was exca- vated in 1965–66 and 1972 under the direction of Noel Roe and Wilfred Shawcross, but only fully reported more recently (Furey 2002) . Much of it has been destroyed by a shingle quarry but roughly 33 x 21 m was excavated in 2 5. x 2 5. m squares separated by 0 5. m baulks, for a total of 245 m2 . The layout and stratigraphy of the site have been reconstructed from the excavation notebooks and drawings but record keeping was poor, limiting interpretation . There are three cultural layers, Layers 2–3, with Layer 2 subdivided into Layers 2a–d . The cultural layers date Discussion and conclusion 83

to the 14th century, except Layer 2a which relates to a reoccupation a few hundred years later . Numerous artefacts were recovered during excavation and others were collected by the Wagener family during the shingle extraction operation . The collections include imitation whale tooth pendants, reel units, bone and ivory chisels and teka (dart heads), bone awls, tattoo chisels, numerous bone one-piece fishhooks, tabs and shanks, stone adzes, chisels, drill points, scrapers and abraders . Fauna included sea mammals, a large assemblage of 17 fish taxa, including some very large snapper (Chrysoprys auratus) (Leach 2006: 80), moa (Dinornithiformes), 26 bird taxa from forest, swamp and marine habitats, tuatara (Sphenodon punctatus), dog (Canis familiaris), rat (Rattus exulans), and shell, which has also not been well analysed but what has been analysed came from rocky, estuarine and open coast habitats . The moa bone assemblage was recorded as dominated by meat bearing and industrially useful femurs and lower leg bones and much of it had been broken for industrial use . While Euryapteryx curtus may have been common locally (on the Aupouri Peninsula), other species represented would more likely have been brought on site from kill / butchery sites further south . Furey (2002: 115) suggests that the amount of moa bone recovered does not account for the quantity of fishhooks alone, and that much of the bone may have been brought on site as industrial material rather than as food . Furey (2002) interprets Houhora as a village (“village” implies a level of permanence that is not necessarily supported by the archaeology, and a more functionally neutral term like base camp might be more applicable) . The range of artefacts, such as teka and tattoo chisels, suggests settled occupation with games, ritual and a balance of sex and ages present . In this it is similar to the multi-function base villages proposed by Anderson (1982) for southern New Zealand, though the economic basis at Houhora was horticulture and fishing rather than big game hunting . The Booth family have surface collected a number of artefacts of early form from Tom Bowling Bay in the Far North (Booth et al . 2017, no site number is given), including moa bone fishhooks and shanks, a stone shank and three shanks of what they identify as toheroa Paphies( ventricosa) shell, although it is notable that toheroa does not grow here and it may well be a different, thick-shelled species . They suggest that this shell was experimented with, and soon discounted as unsuitable, by early colonists in the absence of tropical pearl shell . It might be rea- sonably assumed that experimentation was carried out in the earliest period of colonisation and, although no excavations have been undertaken or dates obtained, the Tom Bowling Bay site may be as early as any in the country . The Kokohuia site (O06/317) on the Hokianga Harbour was excavated in 1987 by a Task Force Green team under the direction of Michael Taylor (Taylor n d. ). . Four midden layers were observed, with Layers IV and Va dating to the 15th century and Layers II and III to the 16th century (Leach et al . 1997; Schmidt 2000a: 68; Higham et al . 2004: 144) . This early phase is not as early as other Northland sites under consideration in this section, and perhaps should not be considered an early site at all . The site does, however, indicate continuity of fishing practice into the 1500s . Excavations revealed a house on the crest of a ridge and associated food storage, cooking, food preparation and midden dumping areas (Taylor n d. ). . On the slope below the house was a midden with evidence of cooking, containing estuarine, sandy shore and rocky shore shellfish species; 20 fish taxa, dominated by snapper and wrasse (Labridae) (Leach et al . 1997); carbon- ised seeds of hinau (Elaeocarpus dentatus), miro (Prumnopitys ferruginea) and tawa (Beilschmiedia tawa); forest bird, though the bird bone assemblage is not yet fully analysed (Michael Taylor pers . comm . 11 December 2017); dog and rat bone; and a small amount of whale bone, possibly scavenged from a stranding . Artefacts included four stone adzes and one chisel, 31 shell two- piece fishhook points, two fishing lures, miscellaneous stone artefacts and about 1500 flakes, 84 Torpedo Bay

many of which were obsidian, which is not available locally (Taylor n d. ). . Kumara storage pits were probably associated with this occupation, indicating gardening nearby (Taylor n d. ). . Taylor interprets the site as a kainga (or village) with a range of activities carried out and exploiting a variety of nearby habitats but it does not seem to have been as large or complex as Houhora and might be better interpreted as a seasonal base camp . Other Northland sites have less diverse assemblages and have been interpreted as hunt- ing or fishing camps, which implies that they were satellites of semi-sedentary base camps like Houhora – while Houhora is the only such site currently known it is entirely possible that there were others in Northland . One hunting site is Twilight Beach (M02/162), a deflated site rep- resented, when it was excavated in 1984 under the direction of Michael Taylor, by a remnant pedestal of midden measuring 6 x 1 5. m (Taylor 1984) . There was a single cultural layer of dense shell and bone midden with some internal stratigraphy . The entire deposit was excavated and returned to the lab for analysis . The midden was dominated by mussel Perna( canaliculus) and toheroa with snapper the dominant fish species (MNI = 592), while several bird taxa were iden- tified but not counted . A minimum number of 67 fur seals (Arctocephalus forsteri) were excavated, and this species would have contributed by far the most meat by weight, with five pilot whales also making a major contribution . The bones of these species showed clear marks of butchering and cooking, and Taylor (1984) suggests that the Twilight Beach was a specialist hunting site, occupied in summer as the presence of fur seal pups indicates, for single season . The whales were on the basal layer and are presumed to have stranded . Following the butchery and preservation of the whales the camp was established to exploit a nearby seal colony . The presence of breeding seals shows that the site is undoubtedly early while the single date places it in the late 13th or 14th centuries (Coster 1989: 56; McFadgen 2007: 265) . The Tauroa Point site N05/302 was badly affected by wind, wave and quad bike erosion and was discontinuously spread across the face and at the foot of the foredune (Allen 2005) . Allen suggested that the site was only occupied once in the early period, though the general vicinity was subsequently reoccupied on several occasions with other adjacent sites dating to later than the main N03/302 midden and indicating, as Taylor observed at Kokohuia, continuity in fishing practice into the 1500s . Few formal artefacts were found – a moa bone fishhook and two drill points – but numerous stone flakes were recovered, many of them from the deflated dune surface . There was a diversity of stone types from a number of sources, both regional (within 200 km) and from as far afield as Tahanga and Tuhua / Mayor Island (Phillipps et al . 2016) . In fact, Tahanga basalt and Tuhua obsidian appear to have been preferred over more local equivalents, probably because they were better quality materials although as Walter et al . (2010) suggest procurement from distant sources may have contributed to maintaining social networks . Fauna included seal, possibly small whale, two moa bones from the deflated surface (which may predate the occupation), along with dog, fish, dominated by snapper but with a significant component of reef fish like tarakihi Nemadactylus( macropterus) and wrasse (Labridae) (Wichman 2006), and shellfish . Phillipps et al . (2016: 119) suggest that “the occupants primarily focused on food gath- ering and processing, and the production and maintenance of tools needed for these activities ”. The same is evident at Tauranga Bay (P04/639) where 4 2m was excavated in the erod- ing dune face in 2004 . The main midden dated to the 14th century and contained fur seal and sea lion (Phocarctos sp ). bone as well as dog and rat; several bird taxa from a variety of habitats, primarily tui (Prosthemadera novaeseelandiae); numerous fish, primarily snapper and leatherjacket (Parika scaber); in a dense shell midden (unpublished data, Matthew Campbell) . The site repre- sents a generalised camp analogous to the Twilight Beach and Tauroa Point hunting and fishing camps . Tauranga Bay is close to the Kaeo obsidian source which may have been a major attrac- tion to occupation (and compares to the Coromandel sites, discussed below) . Discussion and conclusion 85

Mangahawea Bay (Q05/682) on Moturua Island, Bay of Islands, was excavated by James Robinson in 2017 revealing early artefact forms along with moa and seal bone and Cellana denticulata1 (James Robinson pers . comm . 5 March 2018) . The excavations have not yet been reported . Artefacts of early form, “including two very fine ‘side-hafted’ specimens as well as several ‘hogbacks’” (Coster and Johnston 1977: 263) were collected by Forest Service personnel “from the sand dunes of Ninety Mile Beach ”. Coster and Johnston describe the collection as “well provenanced” but no site numbers are given . Coster (1989: 70) proposes that there was early settlement on the open beaches of the Aupouri Peninsula but that this has left “little or no trace, possibly because of severe erosion of the coastal dunes as result of deforestation ”. It is hard to argue from negative data and, while the artefact collection demonstrates that Aupouri was occu- pied as early as anywhere in Northland, such open beach occupations are not known from else- where in the North Island . In the Western Bay of Plenty (discussed below), for instance, early occupation seems to have been restricted to the Tauranga Harbour and Maketu, including the Kaituna River mouth and Maketu Peninsula, and only expanded to the open coasts of Papamoa around AD 1450 . It seems probable that early occupation in Northland also emphasised estuar- ies and sheltered bays but, as an exception to what seems to be a general rule, also occupied the open beaches of the Aupuori Peninsula . On the island of Tawhiti Rahi in the Three Kings, pollen cores revealed a sudden increase in charcoal, decrease in forest cover and an approximately 500 year succession of bracken and fern . This burning, certainly human induced, began around the time of the deposition of the Kaharoa tephra (AD 1314 ± 12) or even just before it (see the discussion of Cross Creek in Chapter 2, and below) (Robinson 2016: 107) . While no 14th century sites were located on Tawhiti Rahi, these results show that evidence of early settlement is not dependant on artefacts or moa bone and reinforces the assumption that early settlers travelled and explored widely .

Tamaki

No early sites had been investigated within the built up area of Auckland City prior to the Torpedo Bay excavations . Since then the nearby Masonic Tavern site and the Long Bay Restaurant site to the north of the city have been excavated but not yet fully reported . The few reported sites are either at the Manukau Heads or on the Gulf Islands although there are several sites around the Waitemata Harbour where artefact styles indicate an early occupation . The Matatuahu, or Wattle Bay, site (Q11/344) on the Manukau Harbour South Head is the source of an important and extensive collection of artefacts (Prickett 1987), including a wide variety of adzes and chisels, a twin-lobed pendant, harpoon point, drill points and moa bone one and two-piece fishhooks and tabs . All moa bone on the site appears to be industrial, while limited sea mammal (probably fur seal), dog and fish bone (mostly snapper) have also been collected . A small excavation by the Auckland Archaeological Society in 1961 was briefly reported by Ambrose (1961) . Two occupation layers were described . The artefacts from the lower layer included drill points, a file, an adze and a fishhook blank, indicating an early occupa- tion, but Ambrose was unable to definitively associate this layer as the source of the Brambley Collection . Prickett (1987) cut back the dune face during a site visit in 1982 to expose a 100 mm thick midden of fragmentary mussel with lesser quantities tuangi (Austrovenus stutchburyi), pipi

1 C. denticulata requires cold conditions to breed and populations in the upper North Island are considered to be relicts or stragglers and were extirpated soon after people arrived . They are a marker of early sites in this region . 86 Torpedo Bay

N

R10/1374 Long Bay Restaurant

R10/22 Pig Bay 0 5 R10/25 Sunde site km

R11/1945 Torpedo Bay

R11/2517 Masonic Tavern

S11/20 Motanu Bay Figure 7.2. Excavated early sites in Tamaki.

(Paphies australis), paua (Haliotis iris) and cat’s eye (Turbo smaragda), with fishbone, charcoal and heat cracked rock also present . Where the shell ended a thin black layer continued the occupa- tion surface . It isn’t clear that this layer represents the early occupation at Matatuahu . The remaining reported early archaeological sites in Tamaki are all from islands in the Hauraki Gulf . The Sunde Site (R10/25) on Motutapu was excavated in 1981 and 1982 under the direction of Reg Nichol (1988) . The ‘oyster lens’ is a layer of midden sitting directly below the Rangitoto ash . The site is well known for the human and dog footprints on the surface of this ash indicating visitation to the location before the ash hardened (Nichol 1981) . The oyster lens was dominated by oyster (Saccostrea glomerata) and other rocky shore species, with numerous kina (Evenchinus cloroticus), and fish, bird and dog bone . The small amount of moa was almost certainly imported from the mainland as industrial material . Artefacts included unfinished adzes and chisels of Motutapu greywacke, drill points and files . The site is not well dated: the Rangitoto eruption lasted discontinuously for around 1000 years (Shane et al . 2013 cited in Davidson and Leach 2017) . It seems to have erupted twice in the early to mid-15th century, depositing a layer of volcanic ash over the northern part of Motutapu and parts of the North Shore (Needham et al . 2011) which is the eruption relating to the Sunde occupation . Other dates are based on obsidian hydration (Dodd 2007), which is not as useful as radiocarbon . However, the artefacts forms support a 14th or early 15th century date . The Pig Bay site (R10/22), also on Motutapu Island just north of the Sunde site, was excavated by Jack Golson in 1958 and 1959 but only recently reported (Davidson and Leach 2017) . The site is not well dated; dates from beneath the Rangitoto ash on both shell and char- coal give 12th–13th century dates, which is too early . The stratigraphy is complex, with several occupation layers cut into layers of naturally redeposited Rangitoto ash and windblown sand . Artefacts include a few finished adzes and numerous roughouts mostly of greywacke, probably local Motutapu greywacke though there appears to be more than one source in use (Davidson and Leach 2017: 17) along with one example of Nelson–Marlborough argillite . The latter, and three greywacke adzes, are of “Archaic” form . Most fish hooks and trolling lure shanks are Discussion and conclusion 87

sea mammal bone with one hook possibly made from moa bone, and others made from shell . Obsidian has not been sourced but more than 50% of the small assemblage shows signs of use wear . Most of the faunal material has been lost in the intervening 60 years . Smith (1981a: 103) noted that the range of mammalian fauna was limited but Davidson and Leach (2017: Footnote 4) report that over 500 catalogued ‘seal/dog’ entries are among those missing, so the interpre- tation of Pig Bay remains incomplete . The bird assemblage is small and dominated by seabirds, reflecting the probable loss of forest on Motutapu following the Rangitoto eruption . The surviv- ing fish assemblage is dominated by snapper . Davidson and Leach (2017: 34) conclude that the site was mostly occupied between the Rangitoto ashfall and the end of the 15th century, thus largely post-dating the oyster lens at the Sunde site though artefact forms and the presence of seal certainly indicate some occupation in the 14th to mid-15th centuries . The Motanau Bay site (S11/20) on the south coast of Ponui Island in the Hauraki Gulf was excavated by Vic Fisher in the late 1950s and subsequently reported by Molly Nicholls (1964) . Three occupation layers were described . Artefacts included moa bone fishhooks and basalt and chert drill points, indicating an early occupation . Some seal bone was identified, but no moa bone other than the bone artefacts . Dog, bird and fish were recorded, but have not been analysed further . These excavations were not dated . The site was subsequently re-excavated by the University of Auckland field school in 1994 under the direction of Geoff Irwin . These excavations have not been reported but are summarised briefly by Sheppard et al ,. who describe “several shell midden layers of Archaic age” (2011: 52) containing a small amount of moa bone as well as bone of other extinct bird species . X-ray fluorescence analysis of obsidian indicated that about two thirds came from Tuhua, with most of the rest from Aotea / Great Barrier Island (the closest source) or the Coromandel Peninsula . They interpreted the site as representing “a substantial hunting and fishing camp of a group of mobile and maritime people” (2011: 52) . The site dates to the 15th century (Schmidt 2000a: 72; Sheppard et al . 2011: 52), based on the 1994 excavation material . A firescoop at Timberly Road (R11/2379) near Auckland Airport yielded a late 14th century date although the rest of the excavated site dated to the 16th–17th centuries (Farley et al . 2015; Farley and Bickler 2017) . This was an isolated feature roughly 20 m south of the main excavation; such finds, along with pollen evidence such as that from Tawhiti Rahi, serve to indi- cate that people spread out over the landscape from an early date even if their use of it remained at a low level for many years . Several other sites around Tamaki have been described as early, including others from the Hauraki Gulf Islands, on the basis of artefacts forms or the presence of moa or seal bone, though none have been systematically excavated or dated . Often artefacts and bone are found in the intertidal zone, implying a beach occupation now at least partly eroded by the sea and often damaged by roading and housing . The artefacts are therefore not in their initial context . Most of the early sites are characterised as middens, since they are marked by visible shell, but they are often more than just simple dumps of food waste . Site activities may include gardening, cooking, storage, manufacture of artefacts in bone, stone and perishable materials, and hunting, fishing and shellfish gathering . The condition of these sites is generally unknown and is probably quite variable, but it seems probable that some good evidence of the early settlement of Tamaki sur- vives, even if some of it may not be readily accessible . On the North Shore, early adze forms or moa bone have been found at Takapuna, Northcote, Bayswater and Devonport . Prior to the 2010 Torpedo Bay excavations stone arte- facts had been found in the intertidal zone between the present Devonport ferry building in the south to the Naval Base in the north, indicating that coastal erosion has been affecting sites in Torpedo Bay for some time . Motutapu greywacke flakes and preforms representing early adze 88 Torpedo Bay

manufacture dominate in these surface collections, but obsidian and chert artefacts are also present (Turner 2000) . Some of these finds may relate to the Torpedo Bay site, while others may relate to the Masonic Tavern site . On the south side of the harbour similar surface assem- blages are known at Kohimarama, Ladies Bay and St Heliers, and at the mouth of the Tamaki River at and . These collections are again dominated by Motutapu greywacke early adze preforms and flakes . An early ornament form, a pendant with notches on the edges and made from whale tooth is also recorded from Howick (Duff 1977: 135) . Evidence of early settlement is also found on several of the Gulf islands, including Waiheke, Rakino, Motuihe, Tiritiri Matangi, Otata in the Noises and (Turner 2000; Dodd and Turner 2008) . Like other early sites around the country, those in Tamaki contain a wide range of stone source materials both local and imported . Tuhua obsidian usually predominates but obsidian from the Coromandel Peninsula, Aotea / Great Barrier Island and Northland is often pres- ent . Despite Tamaki having its own source of adze rock (Motutapu greywacke), adzes made of Tahanga basalt are common and Nelson–Marlborough argillite less so . The early sites of Tamaki would have had close links to sites of similar age throughout the upper North Island but these links remain unexplored .

The Coromandel Peninsula

The Coromandel has long been a focus of research into the early archaeology of the North Island . As early as 1979 Davidson listed 14 sites that had been investigated (several of these have no real evidence of being early and are not discussed here), all located on the ocean-facing east coast of the peninsula, and several have been excavated since . These are discussed here from north to south . Only six of the seventeen sites discussed here meet the chronometric hygiene tests of Smith and James-Lee (2010; Smith 2010) . Several are not well described but they con- tain moa bone and / or artefact assemblages that are clearly early . They are included in this discussion despite the lack of firm dating . Harataonga (T08/5) is located on a dune adjacent to a small estuary on the east coast of Aotea / Great Barrier Island, the main island in the outer Hauraki Gulf, but is discussed with the Coromandel sites rather than Tamaki as the island is geographically part of the Coromandel Peninsula, it has its own obsidian source and the environmental setting is also similar to the Coromandel sites . The site was first excavated by the University of Auckland Archaeological Society in 1962 under the direction of Roger Green and Wynne Spring-Rice (Law 1972) . They excavated three 2 5. m squares, and recognised two occupation layers, each a diffuse midden, separated by 200 mm of clean sand . In one square a firescoop was also excavated . They recov- ered one-piece moa bone fishhooks, a two-piece point and shank, and drill points, but the site was not radiocarbon dated at this time (some obsidian hydration rim dates are reported by Law but are problematic) . It was subsequently excavated by the University of Auckland field school in 1999–2000 under the direction of Doug Sutton . These later excavations have not been reported but the site has been radiocarbon dated to the 14th century . Law (1972) reports the faunal assemblage from the 1962 excavation: the shell is dominated by rocky shore species, largely cat’s eye and Cook’s turban; there is some dog and seal bone; one piece of moa bone; several bird species from a variety of habitats, mostly represented by only one or a few individuals; and a fish bone assemblage dominated by snapper . Allen and Holdaway (2010) and Allen (2014) report the bird and fish from the 1999 excavations . The bird assemblage is generally similar to the 1972 assemblage: there are a few less seabird species reported, which Allen and Holdaway (2010: 18) suggest may be due to shortcomings in the analytical methods used in the earlier study; Discussion and conclusion 89

S09/53 Port Jackson N

0 10 km T10/171 Sarah's midden T10/167 Sarah's Gully T10/164 Arthur Black midden T10/399 T10/161 Opito Beach midden Cross Creek T10/165 Skipper's Ridge

T10/166 Tahanga T11/914 Taputaupatea Stream T11/997 Mill Road T11/242 Hahei Beach T11/326 Hahei

T11/115 Hot Water Beach

T11/62 Tairua

T12/20 Opoutere

T12/16 Whitipirorua

T12/2 Whangamata Wharf T12/3 Cabana lodge

T12/500 Wheritoa Figure 7.3. Excavated early sites on the Coromandel Peninsula. and more forest species which they suggest may be due to the use of smaller screens . Similarly, smaller screens allowed more fish taxa to be identified for the upper layer in 2000, but curiously there was little difference in the lower 14th century layer though proportions changed with snap- per less dominant (Allen 2014) . Numerous artefacts were collected by local residents from the deflated Port Jackson site (S09/53) and Davidson (1979) briefly reported a collection of faunal material . The site was exca- vated by Susan Bulmer in 1981 and reported in his MA thesis by Russell Foster (1983) . Much of 90 Torpedo Bay

the site, located on a dune, was deflated by farming activities and wind and in fact Foster (1983: 10) reports that within a year of the excavation very little of the site remained in situ . At the time of excavation several exposed mounds of sand with caps of midden remained and were excavated, along with deflated areas with high concentrations of bone and artefacts . Area A revealed three layers of midden separated by clean sand . Each layer was quite thin: top to bottom, 60 mm, 30–40 mm and ‘scattered’ respectively . In Area B an pebble surface was recorded on the deflated dune surface, with some moa and small bird bone overlying it – although reported as artificially laid it may be a lag deposit . Several firescoops and postholes were found beneath the pebble layer although it seems probable that they were contemporaneous, with wind erosion pushing the pebbles around to cover the features . Artefacts, including those collected by a local resident, included shell, ivory and bone lure shanks, two-piece points and one-piece hooks of bone, ivory reel, adzes and tattoo chisels . Most shell samples were generally dominated by tuatua . Bird, both surface collected and excavated, was dominated by kaka (Nestor meridionalis), a forest parrot, with several other forest, swamp and coastal species also present with clear evidence of on-site butchery . Several species of moa were also represented, dominated by leg bones with clear signs of butchery suggesting limb bones were brought to the site as food, while there is little evidence of industrial use of moa bone . Bone from sea lion and fur seal, dog, rat and tuatara was alos found while a small assemblage of fishbone was dominated by snapper . The site may have been a specialist forest bird hunting site . It is not well dated; a single 14th century radiocarbon date was obtained by Phil Millener in 1978 on surface collected moa bone (cited by Foster 1983: 50, recalibrated with OxCal 4 .3) . This bone shows signs of butchery identical to those from within the archaeological deposit and so can probably be associated with that occupation (Russell Foster pers . comm . 21 December 2017) . Additionally, the cultural layer overlies an unweathered pri- mary deposit of Kaharoa tephra, which has been dated to AD 1314 ± 12 (Hogg et al . 2003) and so cannot date much later than this . A 14th or early 15th century occupation is also indicated by the artefact forms . The greatest concentration of excavated early sites is located at Otama and Opito on the Kuoatunu Peninsula, initially excavated by the Auckland Archaeological Society under the direction of Jack Golson over three summer seasons between 1956 and 1959 (Golson 1959a) . Some sites were excavated more carefully than others, with some excavations poorly recorded and better described as fossicks by local amateurs . Sarah’s Gully (T10/167) was excavated between 1956 and 1959 (Golson 1959a) and briefly reported (Birks 1960; Birks and Birks 1970; Birks and Birks 1973; Green 1963, 2004), but only recently reported in full (Davidson 2018) . This was one of Golson’s type sites for the North Island Archaic . Several areas totally around 1550 m2 were excavated in the dune, where multiple layers of shell midden were found, and on a terrace immediately inland, where buri- als, kumara storage pits and other features were found, and on a pa higher up the slope . Golson traced one of the dune midden layers up the terrace where he considered it to overlie the pits, and thus assigned the pits to the ‘moa-hunter’ occupation (Davidson 2018: 101), but this is uncer- tain . It has been suggested that these, along with those of nearby Skipper’s Ridge (T10/165), demonstrate the early development of kumara storage technologies (Golson 1959a; Davidson 1984: 125) but Davidson (2018: 149) now considers this unlikely . The lower dune layers dated to the 14th century . Finds included personal ornaments, one-piece moa bone fishhooks and tabs, lure shanks, adzes and preforms, mostly of Tahanga basalt although some were of Nelson- Marlborough argillite, drill-points, abraders and numerous obsidian flakes, some carefully flaked into scrapers and awls . Moa bone was found in low numbers, probably industrial material for fishhook manufacture, while small birds included coastal and forest species . The assemblage was small, with shearwaters and storm petrels indicating a summer occupation . Few mammals were found, mostly dog with occasional sea mammal (Smith 1981b) . Fish was dominated by snapper Discussion and conclusion 91

with some leatherjacket and wrasse, but the assemblages have not survived and so cannot be rea- nalysed using more recently developed methodologies . The middens were not particularly dense and were dominated by pipi, with a range of soft and rocky shore species represented . One aspect of the excavations is the temporal span from the 14th century dune middens to the 18th century pa, and Davidson (2018) concludes that there was little change at Sarah’s Gully and the Opito sites generally throughout the sequence, most marked by a slight change in emphasis from rocky shore to soft shore shellfish and the later appearance of fortifications . Green (1963: 67) describes “worked and unworked moa bone” in the lower layer at Sarah’s midden (T10/171), excavated by Lawrie and Helen Birks in 1960 on the beach at the base of the Sarah’s Gully pa . Moa bone was found in association with cooking features (Davidson 2018: 105), small bird from a variety of habitats, fish (recently reanalysed) dominated by snapper and leatherjacket, small numbers of rocky and soft shore shellfish (though collection and retention strategies mean the sample is small) . The site was described as in very poor condition in 1974 (NZAA site record T10/171) and could not be relocated in 1990 (Sewell 1990: 200) . The Opito Beach or Skipper’s midden (T10/161) was excavated while Jack Golson was excavating at Sarahs Gully in 1957–58 . It is an eroding midden at the mouth of a small creek towards the northern end of Opito Bay . The excavations have not been fully reported but Golson (1959a: 17) describes “an 18-inch series of layers quite remarkably rich in faunal and artefactual specimens ”. Non-industrial moa bone and fragments of moa egg shell suggest that moa lived and were hunted locally (Golson 1959b: 44; Scarlett 1974) . Golson (1959b) briefly discusses the artefacts and they are described more fully by Boileau (1980) . She lists 134 one-piece fishhooks, four two-piece points, five lure shanks, 115 adzes and roughouts, mostly of Tahanga basalt (the Tahanga quarry, T10/166, is only 2 5. km distant) and many in distinctively early forms, files, drill points and 78 bone beads . There is no reliable radiocarbon date but the “material culture … is unequivocally Archaic” (Boileau 1980: 92) . The site was re-excavated by Trower (1962) who describes artefacts and fishbone scattered in pockets of darker sand and it is probable that the site was effectively destroyed by this time . While the archaeology is not well described, and the faunal remains noted by Golson, other than moa, have not been analysed, the Opito Beach midden is an important site for its contribution to the development of the idea of the Archaic . The Arthur Black Midden (T10/164) was excavated in 1957–58 . Golson (1959a: 18) states that “moa bone was again found in considerable quantity” but no further description is availa- ble . Smith (1981b: 109) analysed the mammal bone and noted that it was bagged in two layers: sea mammal were found in both layers, indicating a probable early occupation, along with dog . Sewell (1990: 198) describes the “manufacture of moa bone fishhooks”, and notes that the site was severely damaged by cattle . In response to continued erosion of the dune the site was exca- vated by Louise Furey in 2001 . Layer 4 was dated to the mid-14th to mid-15th centuries while Layer 2 was dated to the mid-15th to mid-16th centuries, but no moa bone was found (Mann 2009: 52), suggesting that the layers excavated by Golson relate to an earlier 14th century occu- pation . The lower layer contained a large basalt flaking area . The 2001 fishbone assemblage was analysed by Sarah Mann (2009): the lower layers were dominated by snapper while the upper layers were dominated by leatherjacket . The best described of the Opito sites from the Auckland Archaeological Society pro- gramme is Skipper’s Ridge (T10/165), excavated by Ham Parker as part of the same project in 1959–60 . While briefly mentioned in publication by Parker and others it was subsequently fully described by Davidson (1975, 1976; Davidson and Green 1975) . The site is located on a low ridge above the Opito Beach midden and is notable for its complex storage pits . Four phases of occu- pation were recognised on the basis of intercutting features, and the most complex pits, including two with fully underground pits entered from adjacent rectangular pits, were assigned to Phase 92 Torpedo Bay

1 . Later pits were generally simpler rectangular features . Several adzes, roughouts and flakes were recovered along with files, hammerstones, a small assemblage of obsidian and two pieces of possibly worked shell . Davidson (1975: 36) gives a single date from a Phase 1 pit of AD 1170 ± 60 but Anderson (1991: 782) dismisses this date as unreliable because of the potential inbuilt age and uncertain identification of the charcoal used . These excavations on the Kuaotunu Peninsula were all carried out at an early date – in the late 1950s and early 1960s modern excavation techniques were still being introduced to New Zealand and adapted to local conditions . The excavations are not always well described in the literature, analysis was often haphazard and rigorous radiocarbon dating protocols had not been developed at the time . Even so, we can accept that the sites are early on the basis of moa and seal bone and artefact forms . In fact, these excavations were critical in the development of the idea of the Archaic but it is fair to say that their importance in the historical development if New Zealand archaeology outweighs the contribution they can make to our current understanding of pre-European Maori history , though it is likely in many cases that reanalysis of the finds and interrogation of the excavation notes and plans would allow more complete descriptions of the sites as has recently been the case for the Sarah’s Gully sites (Davidson 2018) . One other early site has since been excavated on the Kuaotunu Peninsula and here descrip- tion, analysis and dating are more reliable . The Cross Creek site (T10/399) at Sarah’s Gully was excavated by Brenda Sewell (1984) in 1983 along with extensive collecting of material on the deflated dune surface . Five occupation layers, including the deflated midden of Layer 1, were found separated by sterile sand although, as discussed in Chapter 2, Furey et al . (2008) have suggested that the sterile layer overlying Layer 9 is the Kaharoa tephra . Compared to the other layers, Layer 9, the earliest layer, is rather sparse . Rocky shore species dominated the midden in every layer . One notable trend is the decrease in exploitation of the Cellana denticulata over time . They are a marker of early sites in the upper North island (though none were found at Torpedo Bay, Chapter 5) and Cross Creek exemplifies this pattern . Snapper were the predominant fish species in most layers, with leatherjacket becoming more important over time by number, if not by weight . The variety of fish indicated baited hooks and netting . Dog and rat were found in all layers and, although present in layer 1, seals were found in small numbers mostly in the lower layers . Small numbers of coastal birds were found in all layers, while forest birds were largely restricted to the lower layers . Moa bone in Layer 9 was not butchered and may have been present naturally . Fishing gear of moa bone is more common in the lower layers while the upper layers contain more shell fishhooks . Layers 7 and 9 date to the late 13th or early 14th centuries while Layers 3 and 5 date to the mid-15th to early 16th centuries (Furey et al . 2008) . The Tahanga basalt quarry has been recorded as 13 separate sites, mostly as ‘work- ing floors’ – T10/166 was the earliest record and in this discussion stands for the entire set of recorded working floors . Tahanga was the source of a high quality, fine-grained basalt, well suited to adze manufacture . This stone was distributed throughout the North Island and into the upper South Island (Turner 2000; Walter et al . 2010) . Basalt was quarried at Tahanga from boulders and flaked to roughout stage and finished off site (Turner and Bonica 1994: 27) . The quarry was used from an early date but production declined after about 1500, although it may have continued in use into the 18th century (http://www heritage. org. nz/the-list/details/9419). . Tahanga and its basalt would have been a major attraction to early settlement, which largely accounts for the numerous 14th century sites found on the Kuaotunu Peninsula, although the Auckland Archaeological Society research programme in the late 1950s recorded and investi- gated sites prior to the types of development that may have destroyed them elsewhere . South of the Kuoatunu Peninsula, the Taputapuatea Stream site (T11/914) in Whitianga was excavated by Andrew Hoffmann in 2014 (Hoffmann 2016) . This was an extensive excavation Discussion and conclusion 93

for a greenfields subdivision . Several contexts in three phases were identified: Phase 1 provided 14th century dates while others date up to the mid-17th century . Shellfish follow the familiar pattern, with early contexts dominated by rocky shore species while later contexts are dominated by soft shore species . Snapper dominate the early fish assemblage, while kahawai and mackerel dominate the later assemblages . Few bird or mammal bones were found . Artefacts included 3318 flakes of Tahanga basalt, 89 flakes of Nelson–Marlborough argillite, numerous flakes of chert and obsidian, 20 Tahanga basalt adzes and roughouts, 21 hammerstones, ten abraders, three poorly preserved broken moa bone fishhooks and four tabs of moa, or possibly sea mammal, bone . The Mill Road site (T11/997) in Whitianga was excavated by Jaden Harris in 2011 (Harris and Campbell 2012) . The site had been badly affected by 19th and 20th century devel- opment . It contained a shell midden up to 300 mm deep, with firescoops cut into it and pits above and below it . In places this midden overlay clean beach shell which in turn overlay a dark grey ‘occupation surface’ about 50 mm thick containing shell, stone and fishbone, with firescoops cut into it . The fishbone was almost all snapper, and only a few vertebrae were found; Harris and Campbell (2012: 17) concluded that it was likely that the fish were being preserved for off-site consumption . Some dog bone and unidentified bird bone was also found . Artefacts included a broken moa bone fishhook, the bevel of an adze of Tahanga basalt that probably came for an early form ‘hog-back’ adze, and a chert drill point . Fourteen flakes of obsidian from both local Coromandel sources and from Tuhua were found . The site dated to the 14th century . Site T11/326 at Hahei was surface-collected and excavated by Steve Edson and Dorothy Brown in 1976 following subdivision earthworks (Edson and Brown 1977) . Scattered human remains were found, and 21 ivory reel units and two elephant seal teeth believed to be the mate- rial from which the reels were made, were recovered nearby . The excavators interpreted these as grave goods but they were no longer in situ . The site dates to the 14th century (Edson 1980: 41, recalibrated with OxCal 4 .3) . Artefacts included Tahanga basalt adze roughouts, cores and drill points of ‘hard siliceous material’, fishhook tabs and flakes of argillite, basalt, greywacke and obsidian . Edson and Brown suggested that drill points were being manufactured on site . Midden and faunal material was sparse and was not fully analysed, but included moa bone, dog, seal, fish and soft shore shell species . Another part of the same site was excavated by Wendy Harsant in 1981 (Harsant 1984, 1985) . Six storage pits were recorded, sub-rectangular or oval in plan . Firescoops overlying the pits dated to the 14th to mid-15th centuries (Harsant 1983: 60, recalibrated with OxCal 4 .3), although one firescoop also gave a mid-15th to mid-17th century date . The artefact assemblage, though larger, was essentially similar to that found by Edson and Brown, dominated by Tahanga basalt roughouts and adze production waste with flakes of Tahanga basalt, Nelson–Marlborough argillite and a chip of pounamu indicating adze maintenance . Drill points were common, along with 114 fishhook tabs and 19 broken hooks; two stone reels and a shark tooth . Again, very little midden and faunal material were recovered, with a small shellfish assemblage dominated by rocky shore species and a small fish assemblage dominated by snapper . The site appears to have been a large site with specialist activity areas such as adze and fishhook manufactories, and containing at least one burial . The Hahei Beach site (T11/242), 800 m north along the beach, also dates to the 14th to mid-15th centuries (Nichol 1986: 197, recalibrated with OxCal 4 .3) . The shellfish assemblage is dominated by rocky shore species including low numbers of Cellana denticulata and greater num- bers of Cook’s turban (Cookia sulcata) which tended to be large, while the fishbone assemblage is dominated by leatherjacket followed by wrasse, although snapper dominated Nichol’s meat weight calculations . 94 Torpedo Bay

Hot Water Beach (T11/115) was excavated by Anne Leahy in 1969 (Leahy 1974) . There were four cultural layers (Layers 3b–6) with no sterile layers separating them . Moa bone fish- ing gear was mainly found in the lower two layers . A stone reel unit was recovered from Layer 5 . Seven adzes and roughouts were found along with abraders, worked pumice and drill points . Other artefacts, including ivory reel units and an Antalis bead2, were collected by locals . There was a considerable assemblage of stone flakes: 276 obsidian flakes are described as from Tuhua and ‘other sources’; while 867 basalt flakes are ‘mainly’ Tahanga basalt . The fish assemblage is dominated by snapper but leatherjacket is an important component of Layer 5, which also has the highest numbers and highest diversity of fish remains . Numerous bird species were found in small numbers throughout the excavation from coastal, forest, swamp and open country habitats . Moa was only present in small quantities and identifications are uncertain . Rocky shore shellfish species dominate all layers except Layer 4 and C. denticulata is present throughout the sequence, though its numbers decline . Dog and seal are also present in all layers . The site is not dated but the lower three layers at least would seem to belong to the 14th century while Layer 3b, lacking industrial moa bone, is probably a little later . In 2017 the public toilets at Hot Water Beach were replaced and the site was excavated by Warren Gumbley (Gumbley et al . 2018) . Here Gumbley found only a single cultural layer, which he suggests is probably the same as Leahy’s Layer 4 . Rocky shore species, including C. dentic- ulata, dominated the shell midden . Artefacts included a broken adze and 15 flakes of Tahanga basalt, 2 flakes each of obsidian and chert and 2 moa bone fishhook cores . The site was dated to the 14th–mid-15th centuries; Gumbley suggests dating material excavated by Leahy would allow for a finer grained dating of the site . The Tairua site (T11/62), located on the tombolo connecting Paku to the mainland, was excavated over several seasons in the late 1950s and early 1960s by Roger Green, Colin Smart and Bob Jolly . Layers 2, 4, 6 and 8 were the cultural layers, each separated from the other by clean dune sand . The early period layer is Layer 2, which was where a pearl shell lure was found during the 1964 excavations (Smart and Green 1962; Green 1967a), at that time the only known artefact from tropical Polynesia known for New Zealand excavations (a tropical shell chisel has since been identified from Wairau Bar museum collections, Davidson et al . 2011) . As one of the first North Island sites for which a faunal analysis was carried out (Rowland 1977: 135) the site has an important place in the history of New Zealand archaeology . For these reasons the site has been well studied although it was not a particularly large or intensive occupation . Layer 2 dates to the mid-13th to 14th centuries, which is earlier than most accepted dates from the upper North Island (Schmidt and Higham 1998), though the date is statistically indistinguishable from dates more firmly in the late 13th to 14th centuries and, other than the find of a single highly significant item, Tairua does not stand out from other similar sites . Apart from the pearl shell lure, one adze and four roughouts of Tahanga basalt, abraders, moa bone fishhooks and tabs as well as needles and an awl were recovered . Bone of four species of moa was recovered . Four sea mammal species, mostly fur seal, were identified (Smith 1978) . Age/sex categories indicate at least nine fur seal individuals were present, while pups indicate a nearby breeding population . Twenty-four small bird taxa from coastal, forest and wetland environments were found, includ- ing pelican, though in generally small numbers (data in author’s possession) . The small fish assemblage was dominated by snapper with a significant proportion of wrasse and tarakihi . Shell was dominated by rocky shore species with a significant component ofCellana denticulata, in contrast to the dense Layer 6 pipi-cockle midden (Davidson 1964) .

2 Antalis nana, a tusk shell (Scaphopoda), was previously referred to as Dentalium nanum . Archaeologists are more familiar with the term ‘Dentalium bead ’. Discussion and conclusion 95

The Opoutere midden (T12/20) extends approximately 100 m along the Wharekawa Harbour foreshore and is up to 1 m deep, with several visible layers, all superficially similar . Several column samples were excavated under the direction of Louise Furey in 2008, with eight layers identified, giving a 14th century date for the lower layer and late 18th century date for the upper layer . A large quantity of basalt flakes in the intertidal zone shows that much of the site has eroded (Louise Furey pers . comm . 4 December 2017) . Two column samples have been analysed: there was little change in shellfish species abundance between the upper and lower layers, with soft shore species predominating throughout, reflecting a local environment lacking rocky outcrops, in contrast to the usual pattern of rocky shore species dominating early middens (Mason 2009; Louise Furey pers . comm . 30 January 2018) . The Whitipirorua site (T12/16) was subject to surface collecting by Bob Jolly between 1962 and 1973 (Jolly 1978) and others have also surface collected artefacts . Furey undertook two small excavations in 1986 and 1988 (Furey 1991) . Jolly’s description indicates an upper layer and a lower layer dominated by mussel shell while Furey found three layers separated by sterile sand . The lower layer of the site dates to the 14th century while the upper layer dates to the mid-14th to early 16th centuries (Furey 1990: 56) . The collected artefacts include ivory and bone reels, fossil Antalis beads, shark teeth and other ornaments; 51 one-piece fishhooks, several two-piece fishhook shanks and points, 8 lure shanks and several fishhook tabs; several hundred drill points, mostly chert; adzes and roughouts of Tahanga basalt, a Nelson–Marlborough argillite adze as well as adzes made on flakes of basalt and chert . From the 1986–88 excavations, rocky shore shellfish dominate the midden with a significant fraction of soft shore species . Fish was not fully analysed but Furey (1991: 23) noted that snapper was dominant throughout . Little bird or mammal was found . Furey (1990) suggests that the amount and distribution of moa bone and eggshell indicates that moa were being hunted for food . The Whangamata Wharf site (T12/2) was excavated by Jan Allo in 1969 (Allo 1972) when she opened up seven 2 x 2 m squares . There were two midden layers: the upper dominated by shell and containing European and Maori artefacts indicating a date of roughly AD 1870; the lower dominated by bone, including of an immature moa and a moa bone one-piece fishhook, along with a triangular section adze roughout, all indicating an early date . The lower midden contained very little shell, a few fish bones, a few bird bones from coastal and forest habitats, a minimum number of 2 seals, 172 dog bones representing at least 17 individuals, and some burnt and fragmented human bone . Allo did not date the site . She concluded that the occupants of the lower midden relied on seal, moa and dog for their protein rather than shellfish or fish, and that the human bone also represents food . The Cabana Lodge site (T12/3) at Whangamata was excavated by Wilfred Shawcross in 1963 but not reported at the time . It was subsequently excavated by Warren Gumbley in 2007, who included Shawcross’s excavations in his report (Gumbley 2014) . It should be noted that the Whangamata Wharf site excavated by Allo and the Cabana Lodge site are only about 100 m apart and are probably the same site . Gumbley opened up an area of 270 m2 . He describes two occupation layers across much of the site, Layers 2 and 3, both occupied towards the middle of the 14th century . In places this was overlain, and partly disturbed by, later pre-European Maori horticulture . There were several large, deep firescoops, particularly in Layer 3, some containing seal bone – Gumbley interprets these features as specifically for cooking seal . The midden was dominated by pipi and tuangi, while the fishbone assemblage was dominated by snapper with a significant proportion of blue moki Latridopsis( ciliaris) and tarakihi (Nemadactylus macropterus) . There were few birds, mostly little blue penguin Eudyptula( minor), along with small quantities of fur seal and dog . Fragments of moa bone were identified as tool manufacturing debris (James- Lee and Gumbley 2012) . Artefacts, mostly from Layer 2, included 41 adzes and preforms of 96 Torpedo Bay

Tahanga basalt and 2 of Nelson–Marlborough argillite, drill points, hammerstones, abraders, 2 net weights and 3 possible pumice net floats, as well as numerous flakes primarily of obsidian and Tahanga basalt . Eighty-two bone one-piece fishhooks, some complete, along with tabs and the drill points, indicate fishhook manufacture on site . Three bone trolling lure points and 2 trolling lure shanks, 1 bone and 1 stone, were also recovered . Recent excavations at the site have revealed a mid-14th century occupation with gardened soils overlying a small rectangular storage pit cut into the dune sand (Laumea and Gumbley 2018; Warren Gumbley pers . comm ). . Wheritoa (T12/500) was excavated in the late 1950s and early 1960s by Bob Jolly . Crosby (Crosby 1963, 1977) reconstructed the stratigraphy from Jolly’s incomplete notes; there are four cultural layers without any intervening sterile deposits of which the upper layer, Layer 3, contains most of the cultural material . Moa bone is present but Crosby suggests that this was industrial material and she records 109 moa bone fishhooks along with numerous tabs and 21 drillpoints . Wheritoa remains undated, though Crosby suggests that it is not a very early site but offers no real evidence for this . Kaka was the most numerous bird along with some seabird, seal and dog present . Fish bone was numerous but was not identified, although Crosby noted that snapper appeared to predominate; while mudsnail (Amphibola crenata) and pipi dominate what was probably a relatively sparse shell midden .

The Western Bay of Plenty

South of the Coromandel Peninsula the coast opens out to be essentially one long sandy beach running from Waihi to Whakatane with brief interruptions at the Tauranga Harbour heads and Maketu Peninsula . This is an attractive environment, as the name Bay of Plenty suggests . Few excavations of early sites have been undertaken along this coast but it would seem to have, at least potentially, a strong association with the Coromandel, and to offer resources not readily available there, particularly extensive fertile soils . Site U13/875 at Bowentown on the Tauranga Harbour north head was recorded in profile in a pipeline trench (Moore 2004) . Three cultural layers overlay each other, the lowest of which directly overlay the Kaharoa tephra with several features excavated through the tephra . The basal layer dated to the mid-14th to 15th centuries . Rocky shore shells species were present but not dominant, though they were virtually absent from nearby 16th century sites . Small quantities of snapper bone were identified . Charcoal represented a broadleaf podocarp forest with under- growth species in the lower layers, probably quickly cleared away so that the upper layer occu- pants relied on mature trees or driftwood for firewood . Tuhua obsidian and an adze of Nelson– Marlborough argillite were found, though most artefacts were recovered from the trench spoil . At the other end of Tauranga harbour is Waikoriri / Pilot Bay (U14/363) on the harbour side of the tombolo connecting Mauao to the mainland . The site is badly affected by 20th cen- tury housing development but recent redevelopment has exposed several small windows into the site (Mallows and Cable 2006; Mallows 2007; Hooker 2010; Phillips and McCaffrey n d. ;. Holmes et al . 2014) . These excavations have revealed two layers of midden: an upper shell- rich midden dating to the 16th to 17th centuries, and an artefact-rich, charcoal-stained sand layer dating to the 14th to mid-15th centuries, in places directly overlying the Kaharoa tephra (Hooker 2010) . The upper layer of midden is almost entirely soft shore species but, despite the proximity of reefs and rocks, the lower layers are 25% rocky shore species at most . Little fish, bird or mammal has been reported . Obsidian is almost all from Tuhua with one flake from Hahei . Artefacts include moa bone fishhooks, drill points, tattoo chisels and adzes . The site appears to indicate initial occupation of the area . Discussion and conclusion 97

Test excavations at V14/40 on the west bank of the Kaituna River mouth revealed a 17th–18th century occupation above a truncated pit fill containing a midden deposit with a 14th century date (Campbell 2013) . This midden was exclusively soft shore species . Charcoal from the feature was exclusively matai (Prumnopitys taxifolia) indicating occupation in an undisturbed environment but little can be said about this feature other than that the Kaituna mouth was an ideal site for early occupation . At Maketu site V14/187 was a midden dating to the 14th century, recorded in profile in a utilities trench, containing several moa bones representing at least three individuals, as well as seal, dog and bird bone, but “no shell and remarkably little fish bone” (Moore et al . 2009: 94) .

Discussion

The preceding sections have summarised the archaeology of 38 excavations in the upper North Island . Several of these excavations were undertaken in the 1950s and 60s and the stand- ard of excavation, sampling, recording, analysis and reporting would not meet modern standards . Some, such as Houhora (Furey 2002), have been reported long after excavation by researchers other than the original excavators . Furey (2002: 13) acknowledges these limitations and yet Houhora is one of the better reported, and most extensive, of these excavations . The delay in reporting has its advantages – Davidson’s and Leach’s recent reporting of Pig Bay (2017) bene- fits from a better understanding of the context of the site than was available to Golson in 1959 . Often more recent excavations have been small in scope, keyholes into much wider occupation sites – if 14th century occupation at Torpedo Bay extended across the entire beach terrace and up the gardened slopes of Maungauika, as the evidence of slopewash layers suggests, then the 5 5. x 4 7. m excavated here is very much a keyhole and activities carried out elsewhere on site such as gardening can only be inferred, while others remain unknown . It would be fair to say that our knowledge of early settlement in the upper North Island is biased and incomplete . Even so, we can make some preliminary observations . The excavation and reporting of at least seventeen 14th century sites on the Coromandel, nearly half the sites discussed here, has led several commentators to suggest that the Coromandel was a major focus of early settlement (e g. ,. Smith 2010: 186) . The Coromandel has been a major focus of archaeological excavation beginning with the Auckland University excavations at Opito and Sarah’s Gully in the 1950s and 60s through to recent commercial mitigation excavations for greenfields developments . In Tamaki recent redevelopments such as Torpedo Bay, and also the Masonic Tavern and the Long Bay Restaurant, have revealed unexpectedly well preserved 14th century sites and it might be expected that more will be found in future . Additionally, there are numerous recorded sites around the Waitemata Harbour and Hauraki Gulf where artefact types, or moa or seal bone, indicate an early date of occupation . Northland, on the other hand, has been less of an area of focus for archaeologists, both research and commercial . There are, how- ever, also numerous recorded sites there which have similar indications of an early date . All these areas seem to have been settled early and probably quite extensively, and neither the Coromandel nor anywhere else were exclusive foci of settlement . The exception here might be the Western Bay of Plenty, which has been included in this review precisely to provide this sort of contrast . The long sandy beaches and dune plains were not the type of environment that Maori first occupied, which were more typically sheltered estuaries and bays . Settlement expansion out to the dune plains dates from AD 1450, at the end of the long 14th century of early settlement . While the extent of V14/187 at Maketu is not known, and there may be unrecorded early sites around the Tauranga Harbour or on Matakana Island, the Western Bay would appear not to have been a major focus of early settlement . 98 Torpedo Bay

Among the sites described in this chapter, Houhora stands out as particularly large, a central place that Furey (2002) refers to as a village . Houhoura was a place occupied for longer duration than a seasonal hunting camp and where a wider range of activities than just food or resource procurement clearly took place, but it was not occupied on a permanent or semi-perma- nent basis as the term village suggests; base camp might be a more appropriate term . This implies that the settlement pattern described for southern New Zealand by Anderson (1982), where coastal villages served as bases for the exploitation of animal and stone resources in the interior, can potentially be mapped on to Northland, with sites like Houhora serving as bases for the exploitation of other specialist sites such as fishing camps . Such a settlement pattern has been proposed for Marlborough by Edwards (2016) where Wairau Bar served as the base from which coastal sites south of the Wairau River were occupied . Walter et al . (2017) explicitly interpret Wairau Bar as a central place that played a key social, economic and ritual role in the emerging colonial network of Polynesian settlement . No such central places or base sites are known for either Tamaki or the Coromandel . It may be that this model does not apply universally to all of New Zealand; it may be that Houhora served as a central place for the upper North Island, though a catchment this large seems unlikely; or it may be that the relevant central place archaeological sites have not been located, if indeed they survive . It is probable that all the sites investigated on the Coromandel Peninsula would not equate in extent to Houhora (not accounting for past erosion of coastal sites prior to recording) so there is a significant difference in magnitude between Houhora and all the other sites discussed here . This is certainly a question that deserves further investigation . The recent full description of the Sarah’s Gully excavations, however, indicates that the occupations away from the dune midden, the settlement and pa sites, are later, that is the pits of the settlement site do not relate to the 14th century occupation as Golson originally thought . An analogue of Houhora has yet to be found on the Coromandel and, given the restricted geographical situation of the Coromandel sites, such an analogue probably does not exist here . Areas like the Tauranga Harbour remain a possibility . Defining pre-European Maori history in terms of Archaic and Classic (or early, middle/ transition and late) obscures important continuities . There are certainly changes in artefact form, economy and settlement pattern, but these changes did not all occur at once and certainly do not mark a change between one culture and another – Maori were always a hunting, fishing, hor- ticultural East Polynesian people . As Shawcross (Green and Shawcross 1962: 213) noted “The nature of the artefacts selected for study and the conditions under which all but a few samples have been gathered, has caused research to be dissipated upon broad generalisations, both in terms of culture and of time ”. Davidson has recently summed up the Opito excavations which “appear to reflect a regional sequence that is hardly a sequence at all” and “the people in the Coromandel had no need to change . They were in occupation of a region that was ideally suited to their way of life” (Davidson 2018: 149–150) . Such an analysis reflects Groube’s (1967) view that there were no stages or phases in New Zealand prehistory but rather a series of events (stro- phes) within Maori culture such as a change from rocky to soft shore species, or the development of defended pa . “Changes in tempo in late … prehistory have been camouflaged by continuity in artefact style” (Irwin 2013: 325) There. were developments in economy, technology and settle- ment patterns and changes in artefact form are a marker of some of these, but a collective fail- ure of New Zealand archaeologists to move beyond reliance on artefacts and stagal models has obscured much of the detail that would provide an outline of pre-European Maori history . Dating is crucial to any arguments about the origins of the first New Zealanders and the subsequent technological, social and cultural changes that can be detected through archaeol- ogy . Unfortunately, the calibration curve for 14th century, and for much of the period since, is Discussion and conclusion 99

particularly wiggly (Hogg et al . 2013; Walter et al . 2017) and tight dating resolution is difficult to achieve . Consequently, all early dates tend to get lumped together into a broad pre-AD 1450 time span . Of necessity this is the approach this report has taken but this default position risks restating the Archaic ~ Classic model . Where the New Zealand sequence is conceived of in terms of stages, it is often a simple matter to assign a site or assemblage to a stage on the basis of its contents . Artefacts of ‘Archaic’ form, moa bone and, in the upper North Island, seal bone and C. denticulata, all indicate an ‘Archaic’ site, with a probable pre-AD 1450 date . Consequently, change is conceived as occurring all at once at the end of this period, if not a fully cultural change from Moa-hunter to Maori, at least a phase change from Archaic to Classic . That change probably occurred at different rates in different places has long been recognised and has led Davidson (1984) or Anderson (2016) to propose intermediate stages during which change took place . Early on both Shawcross (Green and Shawcross 1962) and Groube (1969) recognised that artefact types changed at different times – not everything changed all at once at some putative Archaic ~ Classic boundary . It follows then that not all changes in technology, economy and society are necessarily directly related . Some changes may occur in parallel, having different or at least not closely related causes and the predominant narrative of population growth and environmental damage driving change is too simplistic . But disentangling ultimate cause from proximate cause, linked from parallel change, depends on tight dating which is currently not available . Dating is also crucial for theories about technological change . Booth et al . (2017), for instance, propose that toheroa (but see discussion of Tom Bowling Bay, above) was initially considered as a material for fishing lure shanks and quickly rejected in favour of more suita- ble materials such as paua shell . Such experimentation, necessitated by the absence of familiar tropical Polynesian materials, is likely to have occurred early (although the Tom Bowling Bay site is undated, other surface collected artefacts are of early form) . Experimentation and adaptation would have been common, even crucial, but evidence is currently sparse . On the other hand, kumara storage pits would have been present from an early date and disentangling their technological and stylistic development in the absence of tight dates has proven problematic . Davidson (1984: 125) has proposed that the Occupation 1 pits at Skipper’s Ridge are early experiments in storage technology . The Skipper’s Ridge pits are unusual (Davidson 1975: Figures 2 and 3), but pits are a very variable feature type and assigning them to a time frame on the basis of morphology alone is not feasible . Also, there is no good evidence that the site is early . Anderson (1991) rejects the single available date as being from charcoal that was not securely identified and could have a significant inbuilt age . For nearby Sarah’s Gully, Anderson (1991) accepts only two of the seven radiocarbon dates: cal AD 1283–1398 and cal AD 1266–1393 . He rejects the date on the Phase 1 pit and so it isn’t certain that these pits are also early (Davidson 1984: 125) . Accepting Sarah’s Gully as an early settlement site, it seems reasonable that the pits may date to the 14th century, but if so they are in no way remarkable (Davidson 1984: Figure 83) . It might be better to turn Davidson’s argument on its head: early kumara storage pits would have been careful, conservative structures adapted from tropical yam storage technologies (Leach 1984: 58), simple rectangular features, as most remained throughout the pre-European period . In fact the pits described by Harsant (1983, 1984), which are overlain with 14th century firescoops, are rather crude, subrectangular or oval features . The single mid- 14th century pit found by Laumea and Gumbley at the Cabana Lodge site (2018) is a simple rectangular structure . It was perhaps only later that Maori were confident to experiment with pit development and baroque examples such as those at Skipper’s Ridge were built . Elaborate pits would potentially have had social or political functions rather than merely economic, as Law proposes for ‘pits long, large and prestigious’ (Law 1999) . But, lacking tight dates, this model 100 Torpedo Bay

cannot be substantiated any more than can Davidson’s . Similar problems beset virtually all the other changes in Maori culture that are proposed to have taken place around AD 1400–1500 . There is a suggestion in much of the literature that sites with moa bone as food remains, indicated when whole legs including phalanges are bought on site, are earlier, while sites where moa bone appears to only be brought on site as industrial material are later . Whether this can be substantiated or not, it begs the question of the source of the bone . Primary butchery sites, common on the South Island east coast, are not recorded in the study area though small popu- lations moa would have been present . If moa were extirpated early in the upper North Island in the long 14th century then industrial material would either have been subfossil bone, which may have survived in the landscape at butchery and consumption sites, natural deposits and perhaps even cached, or it would have been imported from somewhere further south . Unfortunately, the lower North Island is the least investigated area when it come to the archaeology of early Maori, so evidence of long distance communications remains elusive . The movement of moa bone may have paralleled the movement of Tuhua obsidian and other stone resources that Walter et al . (2010) have interpreted as maintaining social networks . While doubtless the materials themselves – obsidian, basalt or moa bone – were valuable, they also represent connections with distant kin and allies, and these connections were equally valuable . Other problems arise when changes in excavation and analytical techniques result in finer-grained data . As just a single example, Wichman (2006) identified at least 17 fish taxa at Tauroa Point and 17 were found at Tauranga Bay (author’s unpublished data), while 17 were also identified from the much larger Houhora assemblage (Furey 2002: 113) . That the same number were found at these three excavations is not due to the fact that only 17 taxa were targeted, but to improved excavation, collection and analytical techniques – finer mesh sizes when sieving (Allen 2014) and analysis of a greater range of skeletal elements (Vogel 2005; Campbell 2016a) . In general, the larger the assemblage the greater the number of taxa we would expect to see in it so we would expect Houhora to have a greater range of taxa than the very small Tauranga Bay excavation . It is, therefore, difficult to make direct comparisons between sites excavated in the mid-20th century and the early 21st century . Some patterns, recognised in the 1950s and 60s, seem to hold up . For example, most early shellfish assemblages follow the general pattern of selecting rocky shore species over sandy shore and estuarine species, with soft shore species becoming dominant over time . There are exceptions such as at Opoutere and Port Jackson, where rocky habitats are not available, or Pilot Bay where soft shore species predominate despite the presence of rocky reefs around the foot of Mauou . These exceptions aside, the general pattern is that large, easily harvested species, whether they are moa and seal, or rocky shore shellfish, are preferentially targeted by the first occupants in any environment . Some species become at least locally extirpated if not, as in the case of moa, extinct . The loss of large prey would have been mitigated in the upper North Island by horticulture – in the South Island it was followed by massive depopulation . Almost all the sites discussed here are on the east coast in sheltered bays and harbours, rather than on the open, high energy west coast . Kokohuia and Matatuahu are both west coast sites just inside harbour mouths and so are relatively sheltered . The exceptions are Twilight Beach and Tauroa Point, both of which were actively eroding when excavated and Twilight Beach at least is now completely destroyed (by the excavators rather than by erosion, though its imminent destruction was the reason that they fully cleared the site) . This does suggest that recorded site distributions may be biased towards the east coast though, as noted for the west- ern Bay of Plenty, early settlers probably did not occupy open sandy beaches . The occupation at Twilight Beach, for instance, seems to have centred around an expedient beaching of pilot whales; it was not therefore a typical settlement (though what ‘typical’ might be is not really Discussion and conclusion 101

clear) . Despite this, there are likely to be unrecorded early sites in the large sheltered harbours of the west coast . It is interesting to note that no early sites are known from the Coromandel west coast or the Firth of Thames although there would seem to be suitable sheltered bays north of Coromandel township at least . Artefacts in the Auckland Museum collections certainly indicate early occupation here (Louise Furey pers . comm . 4 May 2018) . Another component of the settlement pattern that is missing is gardening . Horticultural sites are often ephemeral and where they survive are difficult to interpret and date . We are therefore often reliant on indirect evidence in the form of kumara storage pits but, as outlined above, assigning these to the 14th century has often proven difficult . There has been a tendency to conceive of the ‘Archaic’ in terms of technology such as artefact forms and materials, but only in a very limited way in terms of economy, specifically moa-hunting . It is acknowledged that gardening was an important part of the 14th century economy of the upper North Island and kumara storage pits have received some discussion (Davidson 1975, 1984), along with theoreti- cal considerations of the initial establishment of tropical kumara horticulture in temperate New Zealand (Yen 1961; Leach 1984), but there has been little research on gardens themselves . Sites are very often located on islands or at the tips of peninsulas . Houhora, for instance, is located in a sheltered bay at the end of the peninsula that forms the eastern, outer edge of the Houhora Harbour . It is notably not located to facilitate overland travel but is well positioned for canoe travel indicating that the 14th century settlers continued to be a maritime people . Matatuahu, Kokohuia, Port Jackson and indeed Torpedo Bay are all in similar locations, while Harataonga, Pig Bay, the Sunde site and Motanu Bay are located on islands . Walter et al . (2010) propose that the maintenance of long distance contacts between small, spatially separated settler communities was vital for establishing their social, economic and reproductive viability . In fact, it makes sense that communication would have been by sea rather than by what was still heavily forested land, and, as the Anaweka canoe demonstrates (Johns et al . 2014; Irwin et al . 2017), the first settlers retained the capability to construct ocean going canoes . Links between the four regions described above are not fully explored . These regions are partly based on modern conceptions of New Zealand administrative boundaries and may have had little reality to pre-European Maori, particularly a highly mobile maritime population . There are environmental differences between the regions, particularly the opposition of sheltered bays and estuaries in Northland, Tamaki and the Coromandel and open beaches in the western Bay of Plenty . Resources are also not evenly distributed . The Tahanga basalt quarry is located on the Coromandel Peninsula and it is close to the Tuhua obsidian source as well as having several high and medium quality obsidian sources of its own . These resources are often considered to be a major attraction to settlement on the Coromandel . Tamaki, on the other hand, may have lacked the highest quality stone sources but had rich volcanic soils well suited to kumara hor- ticulture . Torpedo Bay shows that these may have been gardened from an early date and these soils would have been an attraction equal to the Coromandel stone sources . There are also dif- ferences at a smaller scale that can affect the type of occupation in any specific sheltered bay or estuary . A fine grained comparative study of the local topography and soils adjacent to each site would potentially be instructive .

Torpedo Bay

This chapter has focused on the early pre-European Maori occupation of the upper North Island to provide a wider context for the Phase 1 occupations at Torpedo Bay . This is very much the most significant component of the site, though the Phase 2 occupations are also important . 102 Torpedo Bay

The excavations only covered a small area and it is probable that occupation extended across much of the beach terrace . It is likley that different occupations and activities took place elsewhere on the terrace but that not all were represented in this small area, or that archaeolog- ical evidence of them did not survive . The various occupations probably extended beyond the beach terrace to the slopes of Maungauika and the fertile soils and stone fields on the adjacent level ground (Figure 3 1). . Figure 3 .2 shows stone alignments on the slopes of Maungauika above the excavated site – although these do not relate directly to the site they are indicative of the type of pre-European Maori gardening that is probably responsible for the Layers 4 and 3a slope- washes . The excavation is only a keyhole into the occupation of Torpedo Bay and Maungauika and offers only a very limited glimpse . As noted, the site is located at the end of a peninsula where overland communications are difficult but is well positioned for canoe travel (this is still the case, as anyone who has compared a rush hour trip between Devonport and the central city by car and one by ferry can attest) . The nature of the relationship between this site and the nearby Masonic Tavern site, or to the Sunde site on Motutapu or Motanu Bay on Ponui, is not immediately clear . Dating is not sufficiently fine grained for us to be able to say f the sites were, or were not, contemporaneous . They may have been occupied successively by the same group of people, or some at least may have been contemporaneous and occupied by closely related or allied groups . It seems less likely that the inhabitants of Torpedo Bay would have been allowed to establish gardens on Maungauika if they were not in some way related to their neighbours, whether through bonds of kinship or though some other pre-existing social obligation . Graham (1924: 10) suggests that the area was too vulnerable to attack and too hard to defend in a time of unrest and so was abandoned in the late 18th and early 19th centuries but in the absence of conflict the fertile volcanic soils of the Devonport Peninsula and ready access to the resources of the Hauraki Gulf and Waitemata Harbour would have made the area an ideal site for occupation . It was generalised conflict throughout the northern North Island that led to the abandonment of not just Devonport but most of Tamaki . A summer occupation for Phase 1 at Torpedo Bay is indicated by the presence of snap- per and kahawai as well as the juvenile fur seal, which congregate in breeding colonies over the summer . Summer is also the season for kumara cultivation . This leaves unanswered the question of where people were living over winter, but evidence of summer occupation does not rule out a winter occupation at Torpedo Bay . An important contribution that the Torpedo Bay excavations have made is the evidence of gardening on the adjacent slopes of Maunauika . The slopewash origin of Layer 4 implies clearance and resulting instability of the lower slopes above the beach terrace and the presence of kumara starch shows that this clearance would have been associated with gardening . In situ garden soils have not been located and their extent can only be guessed at . It remains uncertain, though very probable, that any gardening on Maunaguika was contemporaneous with the Layer 5 beach terrace occupation . The other main contribution of the excavations is that Torpedo Bay is the first 14th cen- tury site excavated on the Tamaki mainland . The subsequent excavations of the Masonic Tavern and Long Bay Restaurant sites means we are now in a position to re-examine the 14th century history of Tamaki . References

Allen, H . 1987 . Moa-hunters and Maoris: a critical and Native Inhabitants of New Zealand . John discussion of the work of Roger Duff and W . Parker, London . later commentators . New Zealand Journal of Bacquie, B ,. M . Horrocks and R . Clough 2007 . Archaeology, 9: 5–23 . Archaeological excavation of R11/1935 & Allen, M .S . 2005 . Periodicity, duration, and function R11/943, Highbrook Business Park, East of occupation at Tauroa Point, Northland, Tamaki, Manukau City: investigation report New Zealand . New Zealand Journal of (Authority No . 2005/243) . Unpublished Archaeology, 27: 19–62 . Clough and Associates report to Highbrook Allen, M .S . 2012 . Molluscan foraging efficiency Developments Ltd . and patterns of mobility amongst foraging Ballara, A . 2003 . Taua: ‘Musket Wars’, ‘Land Wars’ or agriculturalists: a case study from northern Tikanga? Warfare in Māori Society in the Early New Zealand . Journal of Archaeological Science, Nineteenth Century . Penguin, Auckland . 39: 295–307 . Beattie, J .H . 1994 . Traditional Lifeways of the Southern Allen, M .S . 2014 . Variability is in the mesh-size of Maori . Otago University Press, Dunedin . the sorter: Harataonga Beach and spatio-tem- Bedford, S . 2004 . Tenacity of the traditional: the poral patterning in northern Maori Fisheries . first hundred years of Maori–European settler Journal of Pacific Archaeology, 5(1): 21–38 . contact on the Hauraki Plains, Aotearoa/New Allen, M .S . and R . Holdaway 2010 . Archaeological Zealand . In T . Murray (ed ). The Archaeology avifauna of Harataonga, Great Barrier Island, of Contact in Settler Societies, 144–154 . New Zealand: implications for avian palae- Cambridge University Press, Cambridge . ontology, Maori prehistory, and archaeofau- Bedford, S . 2013 . From Paeroa to Pohue Pa: rem- nal recovery techniques . Journal of the Royal nant landscapes of events that once shook Society of New Zealand, 40(1): 11–25 . the world . In M . Campbell, S . Holdaway Allo, J . 1972 . The Whangamata Wharf site (N49/2): and S . Macready (eds) Finding Our Recent excavations on a Coromandel coastal midden . Past: Historical Archaeology in New Zealand . Records of the Auckland Institute and Museum, New Zealand Archaeological Association, 6: 61–79 . 29: 59–76 . New Zealand Archaeological Ambrose, W . 1961 . Excavations at Wattle Bay, Association, Auckland . Manukau South Head . New Zealand Bellwood, P . 1978 . Archaeological research at Lake Archaeological Association Newsletter, 4(2): Mangakaware, Waikato, 1969–1970 . New 85–86 . Zealand Archaeological Association Anderson, A J. . 1981 . Barracouta fishing in prehistoric Monograph, 9 . New Zealand Archaeological and early historic New Zealand . Journal de la Association, Auckland . Société des Océanistes, 34(72–73): 145–158 . Best, S . 1975 . Adzes, rocks, and men . Research essay, Anderson, A J. . 1982 . A review of economic patterns University of Auckland . during the Archaic Phase in southern New Beyin, A . 2010 . Use-wear analysis of obsidian Zealand . New Zealand Journal of Archaeology, artifacts from Later Stone Age shell midden 4: 45–75 . sites on the Red Sea Coast of Eritrea, with Anderson, A J. . 1991 . The chronology of colonization experimental results . Journal of Archaeological in New Zealand . Antiquity, 65: 767–795 . Science, 37(7): 1543–1556 . Anderson, A J. . 2000 . Defining the period of moa Birks, L . 1960 . Pa at Sarah’s Gully . New Zealand extinction . Archaeology in New Zealand, 43(3): Archaeological Association Newsletter, 3(2): 195–200 . 16–20 . Anderson, A J. . 2016 . The Making of the Māori Birks, L . and H . Birks 1970 . Radiocarbon dates Middle Ages . Journal of New Zealand Studies, for a pa site at Sarah’s Gully . New Zealand NS23: 2–18 . Archaeological Association Newsletter, 13(2): 63 . Anderson, A J. ,. J . Binney and A . Harris 2014 . Birks, L . and H . Birks 1973 . Additional dates Tangata Whenua: An Illustrated History . for Sarah’s Gully pa site . New Zealand Bridget Williams Books, Wellington . Archaeological Association Newsletter, 16(2): 73 . Anderson, A J. . and I W. G. . Smith 1996 . The tran- Boileau, J . 1980 . The artefact assemblage from the sient village in southern New Zealand . World Opito Beach midden, N40/3, Coromandel Archaeology, 27(3): 359–371 . Peninsula . Records of the Auckland Institute and Anon . 1837 . The British Colonization of New Zealand; Museum, 17: 65–95 . Being an Account of the Principles, Objects, and Booth, J ,. W . Booth, C . Booth, R . Booth and B . Plans of the New Zealand Association; Together Marshall 2017 . Evidence for early – yet short- with Particulars Concerning the Position, lived – use of toheroa (Paphies ventricosa) shell Extent, Soil and Climate, Natural Productions, in the manufacture of trolling-lure shanks? Archaeology in New Zealand, 60(1): 31–44 . 104 Torpedo Bay

Brassey, R . and P . Adds 1983 . Archaeological investi- Campbell, M . and B . Hudson 2013 . The Thornton gations at Mangere, site N42/779 . New Zealand Road Pa (S15/66) and the Swayne Road site Historic Places Trust, Auckland, 1983/5 . (S15/324), Cambridge Section of the Waikato Unpublished report . Expressway: final report (HPA authority Brooks, E ,. M . Campbell and R . Walter 2012 . Torpedo 2013/55) . Unpublished CFG Heritage Ltd Bay faunal analysis . Unpublished South Pacific and Opus International Consultants report to Archaeological Research and CFG Heritage The New Zealand Historic Places Trust, Opus Ltd report to OPUS International . International and The New Zealand Transport Brooks, E ,. C . Jacomb and R . Walter 2010 . Interim Agency . report on archaeological investigations at Campbell, M . and C . Ross-Sheppard 2013 . Kahukura (G47/128), Southland . Unpublished Springpark, Panama Road, Otahuhu – Te Southern Pacific Archaeological Research Apunga o Tainui McLennan Hills: archaeo- report to the New Zealand Historic Places logical assessment . Unpublished CFG Heritage Trust and the SCHIP Partners . Ltd report to Panama Road Development Ltd Bulmer, S . 1983 . Preliminary report of archaeolog- and Barker and Associates . ical investigation at Wiri Oil Terminal site Cloher, D U. . 2003 . Hongi Hika Warrior Chief . Viking, (N42/1224) . New Zealand Historic Places Auckland . Trust, Auckland, 1983/7 . Unpublished report . Clough, R . and M . Turner 1998 . The archaeology of Bulmer, S . 1994 . Sources for the archaeology of the the South Eastern Arterial: the Waipuna site Maaori settlement of the Taamaki volcanic district . R11/1436 . Unpublished Clough and Associates Science and Research Series, 63 . Department of report to the New Zealand Historic Places Conservation, Wellington . Trust . Bunce, M ,. T .H . Worthy, T . Ford, W . Hoppitt, E . Coates, J . 1992 . An Experimental Approach to the Willerslev, A . Drummond and A . Cooper Archaeology of Earth and Rock Mounds in 2003 . Extreme reversed sexual size dimorphism New Zealand . Auckland Conservancy in the extinct New Zealand moa Dinornis . Historic Resource Series, 4 . Department of Nature, 425: 172–175 . Conservation, Auckland . Campbell, M . 2011 . The NRD site: I the archaeol- Coates, J .M ,. R .S . Foster and B .M . Sewell 1996 . ogy . Unpublished CFG Heritage Ltd report Excavations at R11/1394 (Hawkins Hill), to the New Zealand Historic Places Trust and Tamaki, South Auckland . Auckland Conservancy Auckland International Airport Ltd . Historic Resource Series, 13 . Department of Campbell, M . 2012 . Cambridge Section of the Conservation, Auckland . Waikato Expressway: archaeological desktop Collins, C J. ,. N J. . Rawlence, S . Prost, C .N .K . study . Unpublished CFG Heritage Ltd report Anderson, M . Knapp, R .P . Scofield, B C. . to The New Zealand Transport Agency and The Robertson, I W. G. . Smith, E .A . Matisoo- New Zealand Historic Places Trust . Smith, B .L . Chilvers and J .M . Waters 2014 . Campbell, M . 2013 . Section 18 investigation of site Extinction and recolonization of coastal mega- V14/40, Te Tumu, Bay of Plenty . Unpublished fauna following human arrival in New Zealand . CFG Heritage Ltd report to the New Zealand Proceedings of the Royal Society B, 281: 20140097 . Historic Paces Trust . Collins, C J. ,. N J. . Rawlence, T .H . Worthy, R .P . Campbell, M . 2016a . Body part representation and the Scofield, A J. .D . Tennyson, I W. G. . Smith, M . extended analysis of New Zealand fishbone . Knapp and J .M . Waters 2013 . Pre-human New Archaeology in Oceania, 51: 18–30 . Zealand sea lion (Phocarctos hookeri) rookeries Campbell, M . 2016b . The Tawhiao Cottage and the on mainland New Zealand . Journal of the Royal archaeology of race and ethnicity . Journal of Society of New Zealand,, 43(1): 1–37 . Pacific Archaeology, 7(2): 43–58 . Collins, L . and B . Burns 2001 . The dynamics of Campbell, M ,. W . Gumbley and B . Hudson 2009 . Agathis australis–Nothofagus truncata forest in The Tara Road sites, Papamoa . Unpublished the Hapuakohe Ecological District, Waikato CFG Heritage Ltd report to The New Zealand Region . New Zealand Journal of Botany, 39: Historic Places Trust, The LS Johnson 423–433 . Trust, Tauranga City Council, The Ministry Coster, J . 1989 . Dates from the dunes: a sequence of Education, MTEC Ltd, Tauranga and for the Aupouri Peninsula, Northland, New Harrison Grierson Ltd, Tauranga . Zealand . New Zealand Journal of Archaeology, 11: Campbell, M . and J . Harris 2012 . Mataraua U14/2351, 51–71 . Tauriko, Western Bay of Plenty: Stage II Coster, J . and G . Johnston 1977 . The Aupouri archaeological investigations . Unpublished Forest collection . New Zealand Archaeological CFG Heritage Ltd report to The New Zealand Association Newsletter, 20(4): 263 . Historic Places Trust and Comanche Holdings Cramond, B ,. S . Bulmer and K . Lilburn 1982 . Limited . Archaeological survey of the Wiri Oil Terminal References 105

site (N42/1224) . Unpublished report to The Davidson, J .M . and R C. . Green 1975 . A locality map New Zealand Historic Places Trust . for Skipper’s Ridge (N40/7), Opito . Records of Crosby, E . 1963 . Preliminary report on Whiritoa . New the Auckland Institute and Museum, 12: 43–46 . Zealand Archaeological Association Newsletter, Davidson, J .M . and B .F . Leach 2017 . Archaeological 6(1): 46–49 . excavations at Pig Bay (N38/21, R10/22), Crosby, E . 1977 . Wheritoa: a post-settlement dune Motutapu Island, Auckland, New Zealand, in midden on the Coromandel Peninsula . 1958 and 1959 . Records of the Auckland Institute Microfiche, Oceanic Prehistory Records, 2 . and Museum, 52: 9–38 . Crowe, A . 1999 . Which Seashell?: A Simple Guide to the Dodd, A . 2007 . Management of the Motutapu archae- Identification of New Zealand Seashells . Penguin, ological landscape . Archaeology in New Zealand, Auckland . 50(4): 253–274 . Cruickshank, A . 2011 . A qualitative and quantitative Dodd, A . and M T. . Turner 2008 . Motuihe archae- analysis of the obsidian sources on Aotea (Great ological landscape and recent investigations . Barrier Island), and their archaeological signifi- Archaeology in New Zealand, 51(3): 188–205 . cance . MA Thesis, University of Auckland . Duff, R .S . 1947 . The evolution of native culture in Daamen, R ,. P . Hamer and B . Rigby 1996 . Rangahaua New Zealand: Moa Hunters, Morioris, Maoris . Whanui District 1: Auckland . Waitangi Mankind, 3(10): 281–291, 313–322 . Tribunal report . Duff, R .S . 1977 . The Moa-Hunter Period of Maori Davidson, J .M . 1964 . The physical analysis of refuse Culture . 3rd edition . E C. . Keating, Government and New Zealand archaeological sites . MA Printer, Wellington . thesis, University of Auckland . Edbrooke, S W. . and F J. . Brook 2009 . Geology of the Davidson, J .M . 1975 . The excavation of Skipper’s Ridge Whangarei Area . GNS Science, Lower Hutt . (N40/7), Opito, Coromandel Peninsula, in 1959 Edson, S . 1980 . A radiocarbon date for the Archaic and 1960 . Records of the Auckland Institute and burial context (N44/97) at Hahei . Records of the Museum, 12: 1–42 . Auckland Institute and Museum, 16: 41–43 . Davidson, J .M . 1976 . Additional evidence from the Edson, S . and D . Brown 1977 . Salvage excavation of an excavations at Skipper’s Ridge, (N40/7), Opito, Archaic burial context, N44/97, Hahei . Records Coromandel Peninsula . Records of the Auckland of the Auckland Institute and Museum, 14: 25–36 . Institute and Museum, 13: 1–7 . Edwards, B . 2016 . Early settlement sites south of Davidson, J .M . 1978 . The prehistory of Motutapu the Wairau Bar – an interconnected land- Island, New Zealand: five centuries of scape? Paper presented at the New Zealand Polynesian occupation in a changing landscape . Archaeological Association Conference, Journal of the Polynesian Society, 87(4): 327–338 . Blenheim . Davidson, J .M . 1979 . Archaic middens of the Fairfield, F G. . 1938 . Puketutu pa on Weekes’ Island, Coromandel region: a review . In A J. . Manukau Harbour . Journal of the Polynesian Anderson (ed ). Birds of a Feather: Osteological Society, 47(2): 119–128 . and Archaeological Papers from the South Pacific Farley, G . and S . Bickler 2017 . The Timberly Road in Honour of R.J. Scarlett . New Zealand excavation . Archaeology in New Zealand, 60(4): Archaeological Association Monograph, 31–42 . 11: 183–202 . New Zealand Archaeological Farley, G ,. Z . Burnett and J . Low 2015 . Archaeological Association, Oxford . investigations at Timberly Road, Mangere: final Davidson, J .M . 1984 . The Prehistory of New Zealand . report in fulfilment of NZHPT authority no . Longman Paul, Auckland . 2014/573 . Unpublished Clough & Associates Davidson, J .M . 2011 . Archaeological investigations at report to Auckland International Airport Ltd . Maungarei: a large Māori settlement on a volca- Farley, G ,. J . Low, S . Bickler and R . Clough 2017 . nic cone in Auckland, New Zealand . Tuhinga, The Landing Development (Precinct C, Stage 22: 19–100 . 3), Auckland Airport, Mangere: interim Davidson, J .M . 2013 . Archaeological excavations at the archaeological report in fulfilment of NZHPT Station Bay pā, Motutapu Island, inner Hauraki Authority No . 2015/322 . Unpublished Clough Gulf, New Zealand . Tuhinga, 24: 5–47 . and Associates report to RCP Limited and Davidson, J .M . 2018 . In search of the North Island Auckland International Airport Limited . Archaic: archaeological excavations at Sarah’s Felgate, M . 2005 . Just another Papamoa pipi Gully, Coromandel Peninsula, New Zealand, midden? Archaeological investigation at the from 1956 to 1960 . Tuhinga, 29: 90–164 . BOP Supacenta development site (including Davidson, J .M ,. A . Findlater, R . Fyfe, J . MacDonald NZAA Sites U14/2888, U14/2889, U14/2890, and B . Marshall 2011 . Connections with U14/2891 & U14/2892): Papamoa Dune Plain, Hawaiki: the evidence of a shell tool from Tauranga District . Unpublished Felgate & Wairau Bar, Marlborough, New Zealand . Associates report to Jonmer Developments . Journal of Pacific Archaeology, 2(2): 93–101 . Felgate, M . 2017a . Archaeological investigations during construction of AMETI Stage 1, 106 Torpedo Bay

Volume 1: final report on archaeological inves- Furey, L . 1990 . The artefact collection from tigations undertaken for authorities 2013/424, Whitipirorua (T12/16), Coromandel Peninsula . 2013/550, 2013/573, 2014/550 . Unpublished Records of the Auckland Institute and Museum, 27: Maatai Taonga report to Auckland Transport . 19–60 . Felgate, M . 2017b . Archaeological assessment: Eastern Furey, L . 1991 . Excavations at Whitipirorua, T12/16, Busway 1 . Unpublished Maatai Taonga report Coromandel Peninsula . Records of the Auckland to Auckland Transport . Institute and Museum, 28: 1–32 . Felgate, M W. . and Opus 2014 . Archaeology in the Furey, L . 1996 . Oruarangi: The Archaeology and shadow of Maungarei: AMETI Phase 1: Material Culture of a Hauraki Pa . Bulletin of the annual report on archaeological investiga- Auckland Institute and Museum, 17 . Auckland tions undertaken for authorities 2013/424, Museum, Auckland . 2013/550, 2013/573, 2014/550 . Unpublished Furey, L . 2002 . Houhora: A Fourteenth Century Maori Opus International Consultants report to the Village in Northland . Bulletin of the Auckland Heritage New Zealand Pouhere Taonga . Museum, 19 . Auckland Museum, Auckland . Fenton, F .D . 1879 [1994] . Important Judgments Furey, L . 2004 . Material culture . In L . Furey and S . Delivered in the Compensation Court and Native Holdaway (eds) Change Through Time: 50 Years Land Court, 1866–1879 . Southern Reprints, of New Zealand Archaeology . New Zealand Auckland . Archaeological Association Monograph, Foster, R . 1983 . Archaeological investigations at 26: 29–54 . New Zealand Archaeological site N35/88, Port Jackson, Coromandel . MA Association, Auckland . Thesis, University of Auckland . Furey, L ,. F . Petchey, B . Sewell and R C. . Green Foster, R . 1984 . Test excavations at Hamlins Hill . 2008 . New observations on the stratigraphy Unpublished report to The New Zealand and radiocarbon dates at the Cross Creek site, Historic Places Trust, Auckland . Opito, Coromandel Peninsula . Archaeology in Foster, R ,. M . Felgate and M . Horrocks 2012 . Mt New Zealand, 51(1): 46–64 . Wellington Water Supply; Section 3c, Mt Golson, J . 1959a . Excavations on the Coromandel Wellington Domain, archaeological report: Peninsula . New Zealand Archaeological NZ Historic Places Trust Authority 2010/137 . Association Newsletter, 2(2): 13–18 . Unpublished report to Watercare Services Ltd Golson, J . 1959b . Culture change in prehistoric New and the New Zealand Historic Places Trust . Zealand . In J .D . Freeman and W .R . Geddes Foster, R . and B . Sewell 1988 . An open settlement in (eds) Anthropology in the South Seas: Essays Tamaki, Auckland, New Zealand: excavation of presented to H.D. Skinner, 29–74 . Avery, New sites R11/887, R11/888 and R11/899 . Science Plymouth . and Research, 4 . Department of Conservation, Golson, J . and R .N . Brothers 1959 . Excavations Wellington . at Motutapu . New Zealand Archaeological Foster, R . and B . Sewell 1993 . The Tamaki River Sites: Association Newsletter, 2(2): 5–8 . excavations at sites R11/1201 and R11/1506, Graham, G . 1922 . A Maori history of the Auckland Tamaki, Auckland, New Zealand . Auckland isthmus . In J . Barr (ed ). The City of Auckland, Conservancy Historic Research Series, 6 . New Zealand, 1840–1920, 1–32 . Whitcombe & Department of Conservation, Wellington . Tombs Ltd, Auckland . Foster, R . and B . Sewell 1995 . Papāhīnu: the archaeol- Graham, G . 1924 . Before the Pakeha: days of Maori ogy of an early 19th century Maori settlement might: the invasions of Takapuna . In T . Walsh on the bank of the Pukaki Creek, Manukau (ed ). An Illustrated Story of Devonport and the Old City . Unpublished report to The Department of North Shore from 1841 to 1924, 9–10 . Devonport Conservation . Library Associates, Devonport . Fox, A . 1977 . Pa of the Auckland Isthmus: an archaeo- Grayson, D .K . 1978 . Minimum numbers and sample logical analysis . Records of the Auckland Institute size in vertebrate faunal analysis . American and Museum, 14: 1–24 . Antiquity, 43(1): 53–65 . Fredericksen, C .F .K . and E .P . Visser 1989 . Green, R C. . 1963a . A Review of the Prehistoric Sequence Archaeological Investigations at Site R11/1519, in the Auckland Province . New Zealand Cryers Road, East Tamaki, Auckland, New Archaeological Association Monograph, 2 . Zealand. Science and Research Series, 21 . New Zealand Archaeological Association, Department of Conservation, Wellington, Auckland . Wellington . Green, R C. . 1963b . Summaries of sites at Opito, Furey, L . 1983 . Excavation of N42/941 Westfield Sarah’s Gully and Great Mercury Island . New South Auckland . Unpublished report to the Zealand Archaeological Association Newsletter, New Zealand Historic Places Trust, Auckland . 6(1): 57–69 . Furey, L . 1986 . The excavation of Westfield (R11/898), Green, R C. . 1964 . Sources, ages and exploitation of South Auckland . Records of the Auckland New Zealand obsidian: an interim report . New Institute and Museum, 23: 1–24 . References 107

Zealand Archaeological Association Newsletter, Higham, C . Bronk Ramsey and C . Owen 7(3): 134–143 . (eds) Radiocarbon and Archaeology, Proceedings Green, R C. . 1967a . Sources of New Zealand’s East of the Fourth Symposium, Oxford 2002 . Oxford Polynesian culture: the evidence of a pearl shell University School of Archaeology Monograph, lure shank . Archaeology and Physical Anthropology 62: 135–151 . Oxbow Books, Oxford . in Oceania, 2: 81–90 . Hoffmann, A . 2016 . Final report: investigation of Green, R C. . 1967b . Characterisation of New Zealand archaeological site T11/914, Taputapuatea obsidians by emission spectroscopy . New Stream, Whitianga, Mercury Bay – Stage Zealand Journal of Science, 10: 675–682 . 4E, 2014 . HNZPT authority 2014/973 . Green, R C. . 1975 . Adaptation and change in Maori Unpublished report to Pacific Estates Limited culture . In G . Kuschel (ed ). Biogeography and & Heritage New Zealand . Ecology in New Zealand, 591–641 . Dr . W . Junk Hogg, A ,. Q . Hua, P . Blackwell, M . Niu, C . Buck, T . b v. ,. The Hague . Guilderson, T . Heaton, J . Palmer, P . Reimer, R . Green, R C. . 2004 . Where is the map, Roger? Reimer, C . Turney and S . Zimmerman 2013 . Archaeology in New Zealand, 47(4): 97–107 . SHCal13 Southern Hemisphere calibration, Green, R C. . and W . Shawcross 1962 . The cultural 0–50,000 years cal BP . Radiocarbon, 55(4): sequence of the Auckland Province . New 1889–1903 . Zealand Archaeological Association Newsletter, Hogg, A G. ,. T .F G. . Higham, D J. . Lowe, J G. . Palmer, 5(4): 210–220 . P J. . Reimer and R .M . Newnham 2003 . A wig- Groube, L .M . 1967 . Models in prehistory: a consider- gle-match date for the Polynesian settlement of ation of the New Zealand evidence . Archaeology New Zealand . Antiquity, 77: 116–125 . and Physical Anthropology in Oceania, 2(1): 1–27 . Holdaway, R . and C . Jacomb 2000 . Rapid extinction Groube, L .M . 1969 . From Archaic to Classic Maori . of the moas (Aves: Dinornithiformes): model, Auckland Student Geographer, 6: 1–11 . test and implications . Science, 287(5461): Gumbley, W . 2014 . The Cabana site (T12/3), 2250–2254 . Whangamata: results of the 2007 investigation . Holdaway, S . and R . Wallace 2013 . A materialisation Unpublished report to Heritage New Zealand . of social organisation: the 19th century occupa- Gumbley, W ,. T .F G. . Higham and D J. . Lowe 2003 . tion of Te Oropuriri, Taranaki, New Zealand . Prehistoric horticultural adaptation of soils in In M . Campbell, S . Holdaway and S . Macready the Middle Waikato Basin: review and evidence (eds) Finding Our Recent Past: Historical from S14/201 and S14/185, Hamilton . New Archaeology in New Zealand . New Zealand Zeeland Journal of Archaeology, 25: 5–30 . Archaeological Association Monograph, Gumbley, W ,. M . Laumea and M . Sutton 2018 . 29: 77–96 . New Zealand Archaeological Archaeological report for site T11/115 . Association, Auckland . Unpublished report to Thames-Coromandel Holdaway, S J. . and N . Stern 2004 . A Record in Stone . District Council and Heritage New Zealand Aboriginal Studies Press, Canberra . Pouhere Taonga . Holmes, P ,. A . Cruickshank and M . Campbell 2014 . Haast, J . 1871 . Moas and moa hunters: address to Mt Maunganui North Ultrafast Fibre instal- the Philosophical Institute of Canterbury . lation (HNZPTA authority 2014/1028) . Transactions and Proceedings of the Royal Society of Unpublished CFG Heritage Ltd report to New Zealand, 4: 66–107 . Heritage New Zealand Pouhere Taonga and Harris, J . and M . Campbell 2012 . The Mill Road site, Transfield Services Ltd . Whitianga . Unpublished CFG Heritage Ltd Hooker, R . 2010 . Report on archaeological inves- report to The New Zealand Historic Places tigation, 37 The Mall, Mount Maunganui, Trust, John and Cheryl Howse and Paul and Tauranga . Unpublished ArchSearch report to Dianne White . Glover Family Trust . Harsant, W . 1983 . Radiocarbon dates from N44/97, Hudson, B . 2016 . Archaeology in the shadow of Hahei, Coromandel Peninsula . New Zealand Maungarei: AMETI Phase 1 . report on kōiwi Archaeological Association Newsletter, 26(1): tangata uncovered by archaeological excava- 59–61 . tions . R11/2880 and R11/2881 . Unpublished Harsant, W . 1984 . Archaic storage pits at N44/97, ArchOs Archaeology report to Mātai Taonga Hahei, Coromandel Peninsula, New Zealand . Ltd ,. Auckland Transport and Heritage New New Zealand Journal of Archaeology, 6: 23–35 . Zealand Pouhere Taonga . Harsant, W . 1985 . The Hahei (N44/97) assemblage Hudson, B . and M . Campbell 2011 . The NRD site: of Archaic artefacts . New Zealand Journal of II the koiwi . Unpublished CFG Heritage Ltd Archaeology, 7: 5–37 . report to the New Zealand Historic Places Higham, T .F G. ,. C . Bronk Ramsey, F . Petchey, C . Trust and Auckland International Airport Ltd . Tompkins and M . Taylor 2004 . AMS radio- Irwin, G . 1981 . How Lapita lost its pots: the question carbon dating of Rattus exulans bone from of continuity in the colonisation of Polynesia . the Kokohuia site (New Zealand) . In T . Journal of the Polynesian Society, 90(4): 481–494 . 108 Torpedo Bay

Irwin, G . 2013 . Wetland archaeology and the study of Law, R G. . 1999 . Pits long, large and prestigious: recog- late Māori settlement patterns and social organ- nition of varieties of Māori kūmara storage pits isation in northern New Zealand . Journal of the in northern New Zealand . New Zealand Journal Polynesian Society, 122(4): 311–332 . of Archaeology, 21: 29–45 . Irwin, G . (ed ). 2004 . Kohika: The Archaeology of a Late Lawlor, I . 1981 . Puhinui excavation report N42/17 . Maori Village in the Ngati Awa Rohe, Bay of Department of Anthropology, University of Plenty, New Zealand . Auckland University Auckland . Unpublished report . Press, Auckland . Leach, B .F . 1986 . A method for the analysis of Pacific Irwin, G ,. D . Johns, R G. J. . Flay, F . Munaro, Y . Sung Island fishbone assemblages and an associ- and T . Mackrell 2017 . A review of archaeologi- ated database management system . Journal of cal Māori canoes (waka) reveals changes in sail- Archaeological Science, 13: 147–159 . ing technology and maritime communications Leach, B .F . 2006 . Fishing in pre-European New in Aotearoa/New Zealand, AD 1300–1800 . Zealand . New Zealand Journal of Archaeology Journal of Pacific Archaeology, 8(2): 31–43 . Special Publication . New Zealand Jacomb, C . 2000 . Panau: The Archaeology of a Banks Archaeological Association, Wellington . Peninsula Māori Pa . Canterbury Museum Leach, B .F ,. A J. . Anderson, D G. . Sutton, R . Bird, P . Bulletin, 9 . Canterbury Museum, Christchurch . Duerden and E . Clayton 1986 . The origin of Jacomb, C ,. R . Wallace and C . Jennings 2010 . Review prehistoric obsidian artefacts from the Chatham of the archaeology of Foveaux Strait . Journal of Islands and Kermadec Islands . New Zealand the Polynesian Society, 119(1): 25–59 . Journal of Archaeology, 8: 143–170 . James-Lee, T . and W . Gumbley 2012 . Patterns of Leach, B .F . and A .S . Boocock 1995 . Estimating live faunal resource use at an early prehistoric fish catches from archaeological bone fragments settlement at Whangamata on the Coromandel of snapper, Pagrus auratus . Tuhinga, 3: 1–28 . Peninsula, North Island, New Zealand . Journal Leach, B .F ,. J .M . Davidson and L .M . Horwood 1997 . of Pacific Archaeology, 3(2): 33–51 . Prehistoric Māori fishermen at Kokohuia, Johns, D ,. G . Irwin and K . Sung 2014 . An early Hokianga Harbour, Northland, New Zealand . sophisticated East Polynesian voyaging canoe Man and Culture in Oceania, 13: 99–116 . discovered on New Zealand’s coast . Proceedings Leach, H .M . 1979 . The significance of early horticul- of the National Academy of Sciences, USA, 111(41): ture in Palliser Bay for New Zealand prehistory . 14728–14733 . In B .F . Leach and H .M . Leach (eds) Prehistoric Jolly, R G. W. . 1978 . Brief record of work at Man in Palliser Bay . National Museum of Whitipirorua Beach (N49/16) and nearby New Zealand Bulletin, 21: 241–249 . National area . New Zealand Archaeological Association Museum of New Zealand, Wellington . Newsletter, 21(4): 129–134 . Leach, H .M . 1984 . 1,000 Years of Gardening in New Kahotea, D . 1983 . The interaction of Tauranga hapu Zealand . A .H . & A W. . Reed, Wellington . with the landscape . MA thesis, University of Leahy, A . 1974 . Excavations at Hot Water Beach Auckland . (N44/69), Coromandel Peninsula . Records of the Kirk, T . 1878 . Notes on the botany of Waiheke, Auckland Institute and Museum, 11: 23–76 . Rangitoto, and other islands in the Hauraki Leahy, A . 1991 . Excavations at Taylor’s Hill, R11/96, Gulf . Transactions and Proceedings of the Royal Auckland . Records of the Auckland Institute and Society of New Zealand, 11: 444–454 . Museum, 28: 33–68 . Kondo, R ,. C . Childs and I . Atkinson 1994 . Opal Lilburn, K . 1982 . Ambury Farm Park archaeological Phytoliths of New Zealand . Manaaki Whenua investigation, Stage 1: N42/1143, 1137 & 1251 . Press, Lincoln . Unpublished report to New Zealand Historic Ladefoged, T .N . and R T. . Wallace 2010 . Excavation of Places Trust, Auckland . undefended site R10/494 on Motutapu Island, Lowe, D J. ,. R .M . Newnham, B G. . McFadgen and New Zealand . Archaeology in New Zealand, T .F G. . Higham 2000 . Tephras and New 53(3): 170–184 . Zealand archaeology . Journal of Archaeological Lambert-Law de Lauriston, T . 2015 . An exploration Science, 27(10): 859–870 . of use-wear analysis on Acheulean large cutting Mallows, C . 2007 . Archaeological monitoring at The tools: the Cave of Hearths’ Bed 3 assemblage . Mall, Mount Maunganui of U14/363: The MSc Thesis, University of the Witwatersrand . Mall Infrastructure Improvements TC44/05 . Laumea, M . and W . Gumbley 2018 . The Cabana Unpublished Opus International Consultants Site: review of evidence for horticulture in the report to Tauranga City Council . early Settlement Phase of New Zealand . Paper Mallows, C . and N . Cable 2006 . Archaeological presented at the NZAA – AAA conference, assessment of midden U14/363 at The Mall, Auckland, 2018 . Mt Maunganui, Tauranga . Unpublished Opus Law, R G. . 1972 . Archaeology at Harataonga Bay, International Consultants report to Tauranga Great Barrier Island . Records of the Auckland City Council . Institute and Museum, 9: 81–123 . References 109

Mann, S .R . 2009 . How to catch a leatherjacket: Needham, A J. ,. J .M . Lindsay, I .E .M . Smith, P . prehistoric fishing strategies at Arthur Black’s Augustinus and P .A . Shane 2011 . Sequential Midden, Opito Bay, New Zealand . MA Thesis, eruption of alkaline and sub- University of Otago . alkaline from a small monogenetic Mason, T . 2009 . Shellfish use and subsistence patterns volcano in the , New at Opoutere, Coromandel Peninsula . Research Zealand . Journal of Volcanology and Geothermal Portfolio, University of Auckland . Research, 201: 126–142 . McFadgen, B G. . 2007 . Hostile Shores: Catastrophic Neich, R . 1996 . New Zealand Maori barkcloth and Events in Prehistoric New Zealand and their barkcloth beaters . Records of the Auckland Impact on Maori Coastal Communities . Auckland Institute and Museum, 33: 111–158 . University Press, Auckland . Neich, R . 2002 . Another Maori barkcloth beater from McFadgen, B G. . and R .A . Sheppard 1984 . Ruahihi the Kaipara Harbour . Records of the Auckland Pa - A Prehistoric Defended Settlement in the Institute and Museum, 39: 13–15 . South-Western Bay of Plenty . National Museum Newnham, R .M ,. D J. . Lowe, M .S . McGlone, J .M . of New Zealand Bulletin, 22, and New Wilmshurst and H . T .F G. 1998 . The Kaharoa Zealand Historic Places Trust Publication, 19, Tephra as a critical datum for earliest human Wellington . impact in northern New Zealand . Journal of McGlone, M .S . and J .M . Wilmshurst 1999 . Dating Archaeological Science, 25(6): 533–544 . initial Māori environmental impact in New Nichol, R .A . 1981 . Preliminary report on excava- Zealand . Quaternary International, 59: 5–16 . tions at the Sunde Site, N38/24, Motutapu McKinlay, J . 1974 . Elletts Mountain excavation, Island . New Zealand Archaeological Association 1973–74 . New Zealand Historic Places Trust Newsletter, 24(4): 237–256 . Newsletter, 4: 4–6 . Nichol, R .A . 1986 . Analysis of midden from N44/215: McKinlay, J . 1975 . Elletts Mountain 1974–75 . New hard times t Hahei? In A J. . Anderson (ed ). Zealand Historic Places Trust Newsletter, 5: 6 . Traditional Fishing in the Pacific: Ethnographical Middleton, A . 2008 . Te Puna – A New Zealand Mission and Archaeological Papers from the 15th Pacific Station . Springer, New York . Science Congress . Pacific Anthropological Middleton, A . 2013 . Mission archaeology in New Records, 37: 179–198 . Zealand . In M . Campbell, S . Holdaway and Nichol, R .A . 1988 . Tipping the feather against a scale: S . Macready (eds) Finding Our Recent Past: archaeozoology from the tail of the fish . PhD Historical Archaeology in New Zealand . New Thesis, University of Auckland . Zealand Archaeological Association, 29: 33–28 . Nicholls, M .P . 1964 . Excavations on Ponui Island . New Zealand Archaeological Association, Records of the Auckland Institute and Museum, Auckland . 6(1): 23–38 . Moore, P .R . 1977 . The definition, distribution and Paul, L . 2000 . New Zealand Fishes: Identification, sourcing of chert in New Zealand . New Zealand Natural History and Fisheries. Reed, Auckland . Archaeological Association Newsletter, 20(2): Phillipps, R .S ,. A J. . McAlister and M .S . Allen 2016 . 51–85 . Occupation duration and mobility in New Moore, P .R . 2004 . Archaeological investigation of Zealand prehistory: insights from geochem- site U13/874, 875 & 876, Bowentown, Waihi ical and technological analyses of an early Beach . Unpublished Peninsula Research report . Maori stone artefact assemblage . Journal of Moore, P .R . 2012 . Procurement and cultural distri- Anthropological Archaeology, 42: 105–121 . bution of obsidian in Northern New Zealand . Phillips, C . and H . Allen 2013 . Archaeology at Opita: Journal of Pacific Archaeology, 3(2): 17-32 . Three Hundred Years of Continuity and Change . Moore, P .R ,. G . Taylor, B . Gill and T . James-Lee Research in Anthropology & Linguistics, elec- 2009 . Faunal remains from two sites at Maketu, tronic series, 5 . Department of Anthropology, Bay of Plenty . Archaeology in New Zealand, University of Auckland, Auckland . 52(2): 90–100 . Phillips, K . and C . McCaffrey n d. . Preliminary report: Murdoch, G . 1990 . Ngo Tohu o Waitakere: the Maori monitoring and excavations for Pilot Bay place names of the Waitakere River Valley and boardwalk and storm water upgrade, NZHPT its environs; their background history and an Authority 2013/881 . Unpublished Archaeology explanation of their meaning . In J . Northcode- B O. .P . report to the New Zealand Historic Bade (ed ). West Auckland Remembers: A Collection Places Trust . of Historical Essays for the 1990 Commemoration, Piperno, D .R . 2006 . Phytoliths: A Comprehensive Guide 9–32 . West Auckland Historical Society, Glen for Archaeologists and Paleoecologists . Altamira Eden . Press . Lanham . Nagaoka, L . 2002 . The effects of resource depres- Plowman, M . 2009 . New Zealand Defence Force sion on foraging efficiency, diet breadth, and Torpedo Bay Boat Yard, Devonport, Auckland patch use in southern New Zealand . Journal of (R11/1945): archaeological assessment of Anthropological Archaeology, 21(4): 419–442 . the proposed Royal New Zealand Navy 110 Torpedo Bay

Museum, Torpedo Bay, Devonport, Auckland . Scott, S .D . 1970 . Excavations at the “Sunde Site”, Unpublished Opus International Consultants N38/24, Motutapu Island, New Zealand . Limited report to the New Zealand Defence Records of the Auckland Institute and Museum, 7: Force . 13–30 . Prickett, N . 1987 . The Brambley collection of Maori Sewell, B . 1984 . The Cross Creek site (N40/260) artefacts, Auckland Museum . Records of the Coromandel Peninsula: a study of an archae- Auckland Institute and Museum, 24: 1–66 . ological investigation in spatial analysis and Prickett, N . 1999 . Nga Tohu Tawhito: Early Maori continuity in the New Zealand Archaic . MA Ornaments . David Bateman, Auckland . thesis, University of Auckland . Prickett, N . 2007 . Early Maori disc pendants . In A . Sewell, B . 1990 . Opito and Otama sites revisited . Anderson, K . Green and F . Leach (eds) Vastly Archaeology in New Zealand, 33(4): 189–202 . Ingenious: The Archaeology of Pacific Material Sewell, B . 1992 . Further Excavations at the Westfield Culture, in Honour of Janet M. Davidson, 29–42 . site (R11/898), Tamaki, Auckland . Auckland Otago University Press, Dunedin . Conservancy Historic Research, Series 1 . Raven, J . and S . Bracegirdle 2011 . New Zealand Department of Conservation, Auckland . Seashells Visual Guide . Creatus Design, Shane, P ,. M . Gehrels, A . Zawalna-Geer, P . Wellington . Augustinus, J . Lindsay and I . Chaillou 2013 . Rawlence, N J. ,. A . Kardamaki, L J. . Easton, A J. .D . Longevity of a small shield volcano revealed Tennyson, R .P . Scofield and J .M . Waters 2017 . by crypto-tephra studies (Rangitoto volcano, Ancient DNA and morphometric analysis New Zealand): Change in eruptive behavior reveal extinction and replacement of New of a basaltic field . Journal of Volcanology and Zealand’s unique black swans . Proceedings of the Geothermal Research, 257: 174–183 . Royal Society B, 284: 20170876 . Shawcross, W . 1972 . Energy and ecology: thermody- Reitz, E J. . and E .S . Wing 2008 . Zooarchaeology . namic models in archaeology . In D .L . Clarke 2nd edition . Cambridge University Press, (ed ). Models in Archaeology, 577–622 . Methuen Cambridge . & Co Ltd, London . Rickard, V .A ,. D G. . Veart and S . Bulmer 1983 . A Shawcross, W . 1976 . Kauri Point Swamp: the ethno- review of archaeological stone structures of graphic interpretation of a prehistoric site . In G . South Auckland . New Zealand Historic Places Sieveking, I .H . Longworth and K .E . Wilson Trust, Auckland, 1983/4 . Unpublished report . (eds) Problems in Economic and Social Archaeology, Roberts, C .D ,. A .L . Stewart and C .D . Struthers (eds) 277–395 . Duckworth, London . 2015 . The Fishes of New Zealand . Te Papa Press, Shennan, S . 2002 . Genes, Memes and Human History: Wellington . Darwinian Archaeology and Cultural Evolution . Robinson, J J. . 2016 . Tawhiti Rahi: “Nga Poito o te Thames & Hudson, London . Kupenga o Toi te Huatahi” [A float of the fish- Sheppard, P J. . 2004 . Moving stones: comments on ing net of Toi te Huatahi]: A multi-disciplinary the archaeology of spatial interaction in New study of Māori settlement of Tawhiti Rahi, an Zealand . In L . Furey and S J. . Holdaway (eds) offshore island in northern New Zealand . PhD Change Through Time: 50 Years of New Zealand thesis, University of Otago . Archaeology . New Zealand Archaeological Rowland, M J. . 1977 . Seasonality and the interpretation Association Monograph, 26: 147–168 . New of the Tairua site, Coromandel Peninsula, N .Z . Zealand Archaeological Association, Auckland . Archaeology and Physical Anthropology in Oceania, Sheppard, P J. ,. G J. . Irwin, S C. . Lin and C .P . 12(2): 135–150 . McCaffrey 2011 . Characterization of New Scarlett, R J. . 1974 . Moa and man in New Zealand . Zealand obsidian using PXRF . Journal of Notornis, 21(1): 1–12 . Archaeological Science, 38(1): 45–56 . Schmidt, M . 1996 . The commencement of pa construc- Sheppard, P J. ,. G J. . Irwin, S C. . Lin and C .P . tion in New Zealand prehistory . Journal of the McCaffrey 2011 . Characterization of New Polynesian Society, 105(4): 441–460 . Zealand obsidian using PXRF . Journal of Schmidt, M . 2000a . Radiocarbon dating New Zealand Archaeological Science, 38: 45–56 . prehistory using marine shell . BAR International Simmons, D .R . 1987 . Maori Auckland: Including the Series, 842 . John and Erica Hedges and Maori Place Names of Auckland Collected by Archaeopress, Oxford . George Graham . Gordon Ell The Bush Press, Schmidt, M . 2000b . Radiocarbon dating the end Auckland . of moa-hunting in New Zealand prehistory . Sissons, J . 2010 . Building a house society: the reorgan- Archaeology in New Zealand, 43(4): 314–329 . isation of Maori communities around meeting Schmidt, M . and T . Higham 1998 . Sources of New houses . Journal of the Royal Anthropological Zealand’s East Polynesian culture revisited: Institute, 16(2): 372–386 . the radiocarbon chronology of the Tairua Smart, C .D . and R C. . Green 1962 . A stratified dune archaeological site, New Zealand . Journal of the site at Tairua, Coromandel . Dominion Museum Polynesian Society, 107: 395–403 . Records in Ethnology, 1(7): 243–266 . References 111

Smith, I W. G. . 1978 . Seasonal sea mammal exploita- Sullivan, A . n d. . Maori gardening in Tamakai before tion and butchering patterns in an Archaic site 1840, Volume 1: traditional, ethnographic (Tairua, N44/2) on the Coromandel Peninsula . and other historic documentary sources . Records of the Auckland Institute and Museum, 15: Unpublished typescript . Unpublished report . 17–26 . Sutton, D G. . 1987 . A paradigmatic shift in Polynesian Smith, I W. G. . 1981a . Mammalian fauna from an prehistory: implications for New Zealand . New Archaic site on Motutapu Island, New Zealand . Zeeland Journal of Archaeology, 9: 135–155 . Records of the Auckland Institute and Museum, 18: Sutton, D G. ,. L . Furey and Y . Marshall 2003 . The 95–105 . Archaeology of Pouerua . Auckland University Smith, I W. G. . 1981b . Prehistoric mammalian fauna Press, Auckland . from the Coromandel Peninsula . Records of the Taylor, M . 1984 . Bone refuse from Twilight Beach . Auckland Institute and Museum, 8: 107–1125 . MA Thesis, University of Auckland . Smith, I W. G. . 1985 . Sea mammal hunting and prehis- Taylor, M . n d. . A preliminary report on the excavation toric subsistence in New Zealand . PhD thesis, at Kokohuia, South Hokianga . Unpublished University of Auckland . draft report . Smith, I W. G. . 2002 . Retreat and resilience: fur seals Tennyson, A . and P . Martinson 2006 . Extinct Birds of and human settlement in New Zealand . In New Zealand . Te Papa Press, Wellington . G G. . Monks (ed ). The Exploitation and Cultural Tonson, A .E . 1966 . Old Manukau . Tonson Publishing Importance of Sea Mammals. Proceedings of the House, Onehunga . 9th Conference of the International Council of Trower, D . 1962 . Opito Beach: two sites . New Zealand Archaeozoology, Durham, August 2002, 6–18 . Archaeological Association Newsletter, 5(1): 43–46 . Oxbow Books, Oxford . Turner, M T. . 2000 . The function, design and distri- Smith, I W. G. . 2010 . Protocols for organising radio- bution of New Zealand adzes . PhD thesis, carbon dated assemblages from New Zealand University of Auckland . archaeological sites for comparative analysis . Turner, M T. . 2005 . Notes on the analysis of usewear in Journal of Pacific Archaeology, 1(2): 184–187 . flake assemblages . New Zealand Archaeological Smith, I W. G. . 2013 . Pre-European Maori exploitation Association Newsletter, 48(4): 314–325 . of marine resources in two New Zealand case Turner, M T. . and D . Bonica 1994 . Following the flake study areas: species range and temporal change . trail: adze production on the Coromandel east Journal of the Royal Society of New Zealand, coast, New Zealand . New Zealand Journal of 43(1): 1–37 . Archaeology, 16: 5–32 . Smith, I W. G. . and T . James-Lee 2010 . Data for an Veart, D G. ,. R . Foster and S . Bulmer 1984 . archaeozoological analysis of marine resource Archaeological mapping of the Wiri Railway use in two New Zealand study areas (revised Site N42/1225 . New Zealand Historic Places edition) . Otago Archaeological Laboratory Trust, Auckland, 1984/7 . Unpublished report . Report, 7 . Vogel, Y . 2005 . Ika . MA thesis, University of Otago . Smith, I W. G. ,. A . Middleton, J . Garland and N . Waitangi Tribunal 1985 . Report of the Waitangi Woods 2012 . Excavations at the Hohi Mission Tribunal on the Manukau claim . Wai-8 . Station, Volume I: The 2012 Excavations . Government Printer, Wellington . Unpublished University of Otago Studies in Archaeology, 24 . report . Department of Anthropology and Archaeology, Waitangi Tribunal 2006 . The Kaipara Report . University of Otago, Dunedin . Wai-674 . Legislation Direct, Wellington . Smith, I W. G. ,. A . Middleton, J . Garland and T . Unpublished report . Russell 2014 . Excavations at the Hohi Mission Wallace, R . 1989 . A preliminary study of wood types Station, Volume II: The 2013 Excavations . used in pre-European Maori wooden arte- University of Otago Studies in Archaeology, 26 . facts . In D G. . Sutton (ed ). Saying So Doesn’t Department of Anthropology and Archaeology, Make it So: Papers in Honour of B. Foss Leach . University of Otago, Dunedin . New Zealand Archaeological Association Smith, S .P . 1898a . The Peopling of the North: Notes on the Monograph, 17: 222–232 . The New Zealand Ancient Maori History of the Northern Peninsula, Archaeological Association, Dunedin . and Sketches of the History of the Ngati-Whatua Wallace, R . and S J. . Holdaway 2017 . Archaeological tribe of Kaipara, New Zealand: “Heru-Hapainga” . charcoal analysis in New Zealand . Journal of The Polynesian Society, Wellington . Pacific Archaeology, 8(2): 17–30 . Smith, S .P . 1898b . Hawaiki: the whence of the Maori: Walter, R . 1998 . Anai’o: The Archaeology of a being an introduction to Rarotonga history . Fourteenth Century Polynesian Community in Journal of the Polynesian Society, 7(3): 137–177 the Cook Islands . New Zealand Archaeological Stone, R C. J. . 2001 . From Tamaki-Makau-Rau Association Monograph, 22 . New Zealand to Auckland . Auckland University Press, Archaeological Association, Auckland . Auckland . Walter, R ,. H . Buckley, C . Jacomb and L . Matisoo- Smith 2017 . Mass migration and the Polynesian 112 Torpedo Bay

settlement of Aotearoa/New Zealand . Journal of World Prehistory Walter, R ,. C . Jacomb and S . Bowron-Muth 2010 . Colonisation, mobility and exchange in New Zealand prehistory . Antiquity, 84: 497–513 . Walter, R ,. C . Jacomb and E . Brooks 2011 . Excavations at Cook’s Cove, Tolaga Bay, New Zealand . Journal of Pacific Archaeology, 2(1): 1–27 . Ward, G . 1973 . Obsidian source localities in the North Island of New Zealand . New Zealand Archaeological Association Newsletter, 16(3): 85–103 . Weisler, M .I ,. R . Bollt and A . Findlater 2010 . Prehistoric fishing strategies on the makatea island of Rurutu . Archaeology in Oceania, 45(3): 130–143 . Wichman, V .L . 2006 . Prehistoric Māori fishing at Tauroa Point, Northland, New Zealand . MA thesis, University of Auckland . Wilmshurst, J .M . and T .F G. . Higham 2004 . Using rat- gnawed seeds to independently date the arrival of Pacific rats and humans in New Zealand . The Holocene, 14(6): 801–806 . Worthy, T . 1997 . What was on the menu? Avian in New Zealand . New Zealand Journal of Archaeology 19: 125–160 . Worthy, T .H . and R . Holdaway 2002 . The Lost World of the Moa . Indiana University Press, Bloomington . Yen, D .E . 1961 . The adaptation of kumara by the New Zealand Maori . Journal of the Polynesian Society, 70(3): 338–348 .