POST-COLONIZATION INTERACTION BETWEEN VANUATU AND FIJI RECONSIDERED: THE RE-ANALYSIS OF OBSIDIAN...

POST-COLONIZATION INTERACTION BETWEENVANUATU AND FIJI RECONSIDERED: THE

RE-ANALYSIS OF OBSIDIAN FROM LAKEBA ISLAND, FIJI*

C. REEPMEYER

School of Archaeology and Anthropology, The Australian National University, Canberra ACT 0200, Australia

and G. CLARK

Archaeology and Natural History, The Australian National University, Canberra ACT 0200, Australia

PIXE–PIGME analysis of 19 obsidian artefacts from Lakeba Island in east Fiji identifiedcontact with northern Vanuatu in the post-colonization period (c. 2500–1000 BP) of Fiji. TheLakeba obsidian is the only physical evidence for interaction across the 850 km water gapseparating the archipelagos of Vanuatu and Fiji in the first millennium AD. New research onthe Vanuatu obsidian sources with laser ablation – inductively coupled plasma – massspectrometry (LA–ICP–MS) casts serious doubt on the validity of a long-distanceinter-archipelago connection in the post-Lapita era. This paper presents the re-analysis of18 obsidian artefacts from Lakeba using LA–ICP–MS and radiogenic isotope results thatdemonstrate that the Lakeba obsidian is not from Vanuatu, and it most likely derives fromthe Fiji–Tonga region. Geochemical evidence for long-distance interaction and migrationbetween the West and Central Pacific in the post-Lapita era has yet to be identified.

KEYWORDS: POST-LAPITA, FIJI, OBSIDIAN, GEOCHEMISTRY, SOURCING

INTRODUCTION

Provenance studies of prehistoric artefacts is undertaken to identify the origin of non-localitems and to establish ancient migration and interaction patterns. This is particularly the casefor islands in the Pacific Ocean where studies of exotic non-perishable materials such asbasalt, chert, obsidian and ceramics offer concrete evidence for long-distance ocean voyaging,and continuing interaction within and between numerous islands in the past (Best 1987;Sheppard 1996; Weisler 1997; Dickinson 2006; Summerhayes in press). Understandingprehistoric interaction in Oceania is progressively refined by the use of new analytical techniquesfor determining artefact composition and improved knowledge of insular geology (Kennettet al. 2004; Collerson and Weisler 2007; Dickinson and Burley 2007). When taken together,these approaches can reduce the number of potential source areas, and tighten the intra-sourceorigin of a transported artefact. In addition, exotic artefacts with a provenance that iscontroversial or anomalous in relation to the archaeological record can be re-analysed, leading toa more accurate understanding of human mobility and inter-archipelago contacts in the past(Dickinson et al. 1999).

In the present case, the issue is the attribution of obsidian artefacts found on Lakeba Islandin east Fiji to a source in the Banks Islands of northern Vanuatu (Fig. 1). The obsidian flakes

*Received 29 October 2008; accepted 28 January 2009

Archaeometry 52, 1 (2010) 1–18 doi: 10.1111/j.1475-4754.2009.00465.x

© University of Oxford, 2009

were found in two rock shelters in levels dating to c. 660–1750 bp, and were chemicallycharacterized with proton-induced X-ray emission (PIXE) and proton-induced gamma–rayemission (PIGME) (Best 1984, 2002). Both Fiji and Vanuatu were first colonized around3100–3000 years ago by people of the Lapita culture, which has a distribution from the BismarckArchipelago to Samoa (Bedford et al. 2007). In contrast to the early transport of non-perishableitems during the colonization phase, most notably the 6500 km distribution of obsidian fromthe Kutau/Bao source on West New Britain (Bellwood and Koon 1989; Summerhayes in press),there is a marked absence of archaeological evidence for long-distance interaction in thepost-Lapita era (c. 2500–1000 bp). The Lakeba obsidian is the only physical evidence forinteraction across the 850 km water gap separating the archipelagos of Vanuatu and Fiji in thepost-colonization era. Thus, the transfer of obsidian from Vanuatu to Fiji has been pivotal toarguments about Fiji’s past, and to hypotheses about the scale and frequency of post-colonizationinteraction in the West Pacific (Wahome 1995; Bedford and Clark 2001; Spriggs 2003).

This is particularly evident in the ceramic sequence of Fiji where points of significant stylisticchange are viewed as the result of a migration from Vanuatu to Fiji, primarily becauseof the identification of Vanuatu obsidian on Lakeba Island, east Fiji (Best 2002, 31; Burley2005; Cochrane 2008). More importantly, the postulated Vanuatu migration to Fiji in thepost-Lapita period supports an intrusionist explanation for the greater amount of biological,cultural and linguistic variation seen in West Pacific populations, which contrasts with thecomparative homogeneity of Polynesian groups in the East Pacific (Clark 2003).

In this paper, we report new laser ablation – inductively coupled plasma – mass spectrometry(LA–ICP–MS) analyses and radiogenic isotope results for the Lakeba obsidian. These demonstratethat the artefacts previously sourced to northern Vanuatu do not derive from Vanuatu, nor dothey match any of the major obsidian sources in the West and Central Pacific. The newgeochemical data suggest that the Lakeba obsidian is most likely of local origin, and derivesfrom a small and currently unknown source in the Fiji–Tonga region.

Figure 1 The location of Lakeba Island in eastern Fiji (�) and major obsidian sources of the West and CentralPacific ( ).

2 C. Reepmeyer and G. Clark

© University of Oxford, 2009, Archaeometry 52, 1 (2010) 1–18

ARTEFACT CONTEXT AND PREVIOUS PIXE-PIGME RESULTS

The Lakeba obsidian collection consists of 20 flakes (Table 1), with 17 flakes derived fromtwo rock shelter sites (Laselase site 2(b) and Qaranipuqa site 197), and three flakes from twoopen sites nearby (Ulunikoro site 47 and Wakea site 196). Nineteen of the flakes were originallyanalysed with PIXE–PIGME for 18 elements at the ANSTO facility in Lucas Heights, Sydney.Sixteen artefact element/element ratios were then compared to element/element ratios from 66obsidian sources in Indonesia, Australia and the Pacific (Best 1984, 433–4) using a multivariate

Table 1 Lakeba obsidian artefact analysis codes used previously, and new codes used in the present study.‘Previous source’ refers to artefact source determined with PIXE–PIGME, and ‘New sourcing’ refers to an artefact

source identified in this paper

Site Context Table 6.7 andFigure 6.17 ID

(Best 1984, 434)

Appendix M(Best 1984, A79)

ANU no. Previous source New sourcing

197 AD5753-T-1(b)-E A

a FL357 ANU9149 ?Tonga Tonga

197 AD5833-O-R(inc) B.T.N.E. Ext C

c FL365 ANU9150 Tonga Tonga

196 B-2 (5) d FL516 ANU9151 Vanuatu/Tonga Tonga196 101/7 Sq 24, Lyr. B-1 e FL517 ANU9152 ? Tonga197 AD572

3-F-1 f171(a) Df FL354 ANU9153 Vanuatu Fiji/Tonga

197 AD5733-F-1 f171(a) E

g FL355 ANU9154 Vanuatu Fiji/Tonga

197 AD5773-F-1 f171(e)-a G

i FL359 ANU9155 Vanuatu Fiji/Tonga

2(b) AD5801-H3-f25 H

j FL362 ANU9156 Vanuatu Fiji/Tonga

2(b) AD5811-J3-4 J

k FL363 ANU9157 Vanuatu Fiji/Tonga

2(b) AD5761-J3-3 K

l FL358 ANU9158 Vanuatu Fiji/Tonga

2(b) AD5851-J3-4 L

m FL367 ANU9159 Vanuatu Fiji/Tonga

2(b) AD5861-J3-4 M

n FL368 ANU9160 Vanuatu Fiji/Tonga

2(b) AD5841-J2-3 O

o FL366 ANU9161 Vanuatu Fiji/Tonga

2(b) AD5781-J1d-2 P

p FL360 ANU9162 Vanuatu Fiji/Tonga

2(b) AD5791-J1d-2 Q

q FL361 ANU9163 Vanuatu Fiji/Tonga

2(b) AD5871-J3-4 N

r FL369 ANU9164 ? Tonga

2(b) 101/7 Sq 3, Lyr. J1, Spit 16 s — ANU9165 — Fiji/Tonga47 19-A-4 t FL518 ANU9166 ?Vanuatu Tonga197 3-F-1 F h FL356 — Vanuatu —197 3-T-1 B b FL364 — Tonga —

The re-analysis of obsidian from Lakeba Island, Fiji 3


discrimination method known as ‘Popper’s Razor’. The method takes account of unequalwithin-source variance and uses a modified Mahalanobis (D2) distance to calculate a sample’sdistance from the statistical centroid or group centre of an obsidian source (Leach andManly 1982).

Qaranipuqa site 197: Seven obsidian flakes were excavated from the Qaranipuiqa rockshelter, with three flakes from the deepest layers (Layer T, R-O) sourced to Tafahi in northernTonga (Best 1987). A radiocarbon date from Layer T has an age of 2870–2330 cal. bp(NZ4596, 2540 1 127 bp), indicating movement between east Fiji and Tonga during the earlysettlement phase. The remaining four obsidian flakes from Layer F1 were sourced to northernVanuatu. Layer F1 was not radiocarbon dated, but Layer F3 about 20 cm below F1 was 14Cdated to 1420–1920 cal. bp (NZ4592, 1770 1 90 bp), while Layer E2 above F1 was dated to700–920 cal. bp (NZ4905, 892 1 63 bp).

Laselase site 2(b): Ten obsidian flakes were excavated from the Laselase rock shelter, withmost in Layer H and Layer J1. All of the flakes were sourced to northern Vanuatu (Best 1984,434, table 6.7). Layer J1 was dated by a marine shell determination to 660–850 cal. bp (NZ5182,1200 1 40 cal. bp).

Ulunikoro site 47 and Wakea site 196: Three obsidian flakes from upper levels of theUlunikoro fortification and the Wakea coastal flat could not be unambiguously identified to asource (Best 1984, 434).

The three obsidian flakes from the lower layers of the Qaranipuqa site (c. 2500 cal. bp)sourced to Tonga were a ‘glossy black vesiculated glass, one with faint layering, and noneshowing cortex’ (Best 1984, 434). The obsidian from the middle and upper layers of Qaranipuqaand Laselase sourced to Vanuatu was different, with 13 of 14 flakes described as ‘a dense greyblack lustrous glass, and eight of these show a flat rough textured striated cortex’ (Best 1984,433). The remaining three flakes of analysed obsidian from Ulunikoro and Wakea that couldnot be attributed to a source with PIXE–PIGME were similar in hand specimen to the earlyobsidian from Qaranipuqa that was sourced to Tonga (Best 1984, 433).

To summarize, the elemental comparison of the Lakeba obsidian collection identified threegroups. The first group, dated to c. 2500 years ago, came from northern Tonga. The secondgroup of obsidian flakes, dating to c. 1700 years ago, was identified as coming from northernVanuatu. The third group of three flakes could not be identified to a source.

On the basis of the PIXE–PIGME obsidian results, Best (2002, 31) outlined a Vanuatu–Fijiconnection at around 1700 cal. bp. The archaeological distribution of Vanuatu obsidian wasconfined mainly to the north of the archipelago, suggesting contact between people fromnorthern Vanuatu and Fiji. Further, Best identified ceramic traits in Fiji that had parallels withceramic designs in Vanuatu, suggesting that Fijian pottery had been influenced by post-colonizationcontact with Vanuatu.

Serious doubt about a Vanuatu–Fiji connection arose after new research with LA–ICP–MSby the lead author on the two obsidian sources in the Banks Islands in northern Vanuatu(Reepmeyer 2008). The research involved field survey and analysis of multiple samplescollected from primary and secondary obsidian source localities. As a result, it is probable thatno new significant obsidian deposit exists in northern Vanuatu, and we now have a much clearerpicture of intra-source elemental variation within the Banks Islands obsidian sources (Ambrose1976; Smith et al. 1977; Duerden et al. 1987). The work also included the characterization ofobsidian from archaeological sites throughout the West and Central Pacific, including theLakeba assemblage, as it represented the eastern limit of Vanuatu obsidian in the Pacific(Reepmeyer in prep.).



GEOCHEMICAL ANALYSIS OF LAKEBA OBSIDIAN

A total of 18 obsidian flakes from Lakeba (Table 1) were analysed with LA–ICP–MS in theResearch School of Earth Science at the Australian National University (Falkner et al. 1995;Longerich et al. 1996; for detailed experimental set-up and methods used, see Reepmeyer2008). The obsidian artefacts were first examined with a JEOL JSM6400 Scanning ElectronMicroscope equipped with an Oxford ISIS energy-dispersive X-ray analyser (SEM–EDXA,Oxford instruments Link ISIS 3.3 software). Calibration was with the NIST612 and theANU2000 standard, which is a high-quality obsidian from Wekwok, Admiralty Islands(Summerhayes in press). Samples were then analysed in an AGILENT 7500S inductive coupledplasma – mass spectrometer combined with an EXIMER laser ablation system (LA–ICP–MS),with the laser diameter set at 86 mm. Counts for 39 isotopes were determined by calculatingthe mean counts for each isotope from three analysis runs per sample. For multivariatestatistical analysis, the C2 package was employed (Juggins 2005). Absolute ppm counts of32 isotopes (31P, 45Sc, 49Ti, 51V, 55Mn, 85Rb, 88Sr, 89Y, 90Zr, 93Nb, 95Mo, 118Sn, 133Cs, 138Ba, 139La,140Ce, 141Pr, 144Nd, 147Sm, 153Eu, 158Gd, 159Tb, 162Dy, 166Er, 169Tm, 174Yb, 175Lu, 181Ta, 186W, 208Pb, 232Thand 238U) were log (lg) transformed and examined with correspondence analysis (CA;Benzécri 1992).

As mentioned above, the original PIXE–PIGME analysis identified three differentprovenance groups in the Lakeba obsidian assemblage: one group of obsidian was sourced toTonga, one group to Vanuatu and the last group to one or several unknown sources. The LA–ICP–MS analysis divided the samples into only two provenance groups (Table 2). Group 1,including the obsidian flakes previously identified to Tonga (Table 1, ANU9149, ANU9150,ANU9151, ANU9152, ANU9166) and the single artefact ANU9164. These artefacts weresourced to two different obsidian outcrops in Tonga, along with the three flakes that could notbe identified to a source with PIXE–PIGME. These samples are not discussed further, as a geo-chemical analysis of the Tongan obsidian including artefacts from several sites in Tonga is inpreparation.

The second group contains 12 flakes, 11 of which were previously sourced with PIXE–PIGME to Vanuatu (ANU9165 was not previously analysed—see Table 1 and Best 1984,table 6.7). The elemental data for the Lakeba obsidians were compared to 117 obsidian samplesfrom the known West and Central Pacific obsidian sources that have been recently analysedwith LA–ICP–MS (Reepmeyer in prep.). A correspondence analysis (CA) was used to examinesimilarities between the Lakeba artefacts and potential sources, with Figure 2 displaying a plotof the first and second eigenvalues that represent a combined variance of 0.596.

The different obsidian sources from the West and Central Pacific plot away from one ano-ther in the CA plot (Fig. 2), with some overlap on the first eigenvalue for the West Fergussonsources and the Vanua Lava source (cf., Reepmeyer 2008). Isotopes that effect the location ofthe sources in the two-dimensional plot are 133Cs, 238U, 118Sn, 232Th, 93Nb, 175Lu, 153Eu, 51V, 181Taand 186W. The CA plot of the LA–ICP–MS results from the Lakeba Group 2 samples (includ-ing the single outlier ANU9158), which were previously sourced to Vanuatu, shows that theyhave a different chemical composition from all other West and Central Pacific obsidian sources,including those from northern Vanuatu (Gaua and Vanua Lava), ruling out a Vanuatu origin forthe obsidian found in east Fiji.

To further investigate differences in the chemical composition of the Lakeba artefacts, selectedisotopes and isotope ratios (K2O, Pb, Rb/Sr, Th/Ta, Nb/Th) were plotted (Figs 3 (a) – 3 (d)).Particularly distinctive is the high K2O content of the Banks Islands sources, which matches



Table 2 Summary statistics and counts of major elements (EDXA), trace elements (LA–ICP–MS) abundances(in mg g-1) and selected trace-element ratios for obsidian artefacts from Lakeba Island in east Fiji. Elemental

abundance and standard deviation (SD) for the ANU2000 obsidian standard is also reported. Group 1, sourced toTonga, does not include ANU9164, which has a composition indicating an origin from an unknown source in the

Tonga Islands. Additionally, counts of the outlier ANU9158 are given

Element Group 1 (n = 5) Group 2 (n = 11) ANU9158(ppm)

ANU2000 (n = 10)

Mean (ppm) SD Mean (ppm) SD Mean (ppm) SD

Na 20 809.09 350.91 32 833.92 791.88 35 015.65 35 824.15 387.03Mg 6 747.97 481.20 2 039.36 518.10 844.25 1 814.88 1 055.22Al 67 040.22 743.45 70 460.15 1 827.92 73 142.49 71 689.01 1 259.54Si 323 322.88 3 187.91 333 292.51 2 228.23 333 657.96 342 859.54 1 467.77Cl 1 390.00 155.72 2 500.00 206.16 1 700.00 730.91 1 098.35K 9 596.50 364.70 26 632.63 808.55 30 673.93 33 422.64 978.68Ca 35 928.02 1 219.08 15 658.45 1 174.14 13 150.50 7 930.52 333.73Fe 54 847.06 2 534.98 32 707.13 2 538.33 27 439.08 13 745.18 201.93P 1 004.94 23.32 574.53 18.15 394.22 189.09 3.84Sc 29.46 1.08 13.51 1.20 12.89 6.50 0.68Ti 2 816.28 60.35 3 204.74 247.21 2 318.83 1 990.84 22.90V 46.95 17.90 6.33 1.42 1.12 5.63 0.09Mn 1 410.73 18.57 843.34 95.36 998.98 450.63 12.59Co 9.94 1.30 4.17 0.64 1.82 1.37 0.04Ni 0.70 0.22 0.52 0.08 0.33 0.32 0.11Cu 41.11 25.69 45.82 1.48 21.88 3.37 0.26As 3.50 0.10 6.95 0.06 6.41 2.09 0.15Rb 19.56 0.44 55.28 0.94 65.11 144.70 4.88Sr 216.09 2.72 131.75 10.13 108.43 57.11 1.77Y 15.14 0.36 36.33 0.82 46.05 31.32 0.83Zr 31.23 0.76 186.78 2.77 212.34 277.17 6.76Nb 3.98 0.10 3.11 0.05 3.36 42.36 0.23Mo 1.91 0.04 3.10 0.08 2.97 3.52 0.07Sn 0.51 0.03 1.89 0.05 2.24 3.11 0.09Cs 0.53 0.01 1.77 0.04 2.04 1.94 0.07Ba 237.20 5.09 743.18 20.04 835.74 619.75 30.34La 6.42 0.17 11.00 0.15 12.77 34.83 1.23Ce 12.82 0.30 26.48 0.35 31.14 69.74 2.61Pr 1.61 0.04 3.62 0.06 4.27 7.17 0.32Nd 6.64 0.14 15.49 0.29 18.00 23.39 0.83Sm 1.97 0.06 4.63 0.11 5.48 4.97 0.15Eu 0.64 0.01 1.05 0.04 1.06 0.85 0.03Gd 2.37 0.07 5.38 0.18 6.40 5.03 0.16Tb 0.37 0.01 0.88 0.02 1.09 0.78 0.03Dy 2.60 0.04 6.16 0.14 7.69 5.25 0.14Er 1.75 0.03 4.14 0.12 5.30 3.42 0.07Tm 0.27 0.01 0.63 0.02 0.82 0.53 0.02Yb 1.92 0.06 4.45 0.16 5.81 3.71 0.10Lu 0.30 0.01 0.67 0.02 0.90 0.57 0.02Ta 0.18 0.01 0.20 0.00 0.22 3.01 0.05W 0.37 0.01 0.35 0.01 0.33 1.25 0.02208Pb 5.08 0.14 7.44 0.25 9.23 6.35 0.23Th 1.36 0.05 2.07 0.12 2.50 10.43 0.80U 0.46 0.02 1.09 0.05 1.22 2.99 0.23K/Nb 2 408.74 8 563.33 9 130.56K/Rb 490.54 481.79 471.12La/Yb 3.34 2.47 2.20Na/K 2.17 1.23 1.14Nb/Th 2.93 1.50 1.35Rb/Zr 0.63 0.30 0.31Rb/Sr 0.09 0.42 0.60Th/Ta 7.47 10.12 11.39Zr/Nb 7.84 60.06 63.21Zr/Ba 0.13 0.25 0.25



the overall high alkaline geochemical composition of the Northern New Hebrides Arc(Barsdell et al. 1982).

Both of the Vanuatu obsidian sources have a significantly higher K2O and Pb content thanall Group 2 Lakeba samples, although the Vanua Lava source has similar Th/Ta and Nb/Thratios, but a higher Rb/Sr ratio. In contrast, the Gaua source has very different Th/Ta and Nb/Thratios, but a similar Rb/Sr ratio. Further support for a non-Vanuatu origin of the Group 2Lakeba obsidian is the absence of plagioclase phenocrysts, which are prominent in the VanuaLava source material (Ward 1979, 8–13; Reepmeyer 2008). The Group 2 Lakeba material(including the ANU9158 outlier, which has the same macroscopic appearance as the otherGroup 2 artefacts) resembles obsidian from Gaua, but the glass is less reflective and has a finertexture. Compared to obsidians from Tafahi in Tonga, the Lakeba Group 2 artefacts are signifi-cantly more alkaline, and they have a distinctively lower SiO2 and higher Fe content than theWest New Britain sources of Kutau/Bao, Gulu and Mopir.

ORIGIN OF THE LAKEBA GROUP 2 ARTEFACTS

The elemental composition of the Lakeba Group 2 artefacts demonstrates that they do notderive from Vanuatu, with the artefacts plotting away from all of the known obsidian sources

Figure 2 Correspondence analysis of 117 source samples from West and Central Pacific sources and artefacts fromLakeba, Fiji (�, West New Britain sources; , Lou sources; �, Manus source; �, West Fergusson sources; �,East Fergusson sources; +, Gaua source; ¥, Vanua Lava source; *, Tafahi sources; , Lakeba artefacts). Please notethat the ANU9158 outlier, which shows similarities with the WNB sources, has a significant different major elementdistribution than WNB obsidian, as shown in Table 2.



in the West and Central Pacific. The question of where these artefacts come from is thereforesignificant, as the obsidian may be from an unknown source within Fiji–West Polynesia, or asource which lies to the west or east of the Fiji–Tonga–Samoa region.

To investigate possible artefact sources, we first consider the trace-element data to deter-mine if their composition indicates an origin from Ocean Island Basalts (OIB) settings thatoccur east of Tonga, or if their composition is suggestive of derivation in an island arc setting.

Figure 3 A comparison of K2O abundances (wt%) and absolute counts of (a) Pb (ppm) and ratios of (b) Rb/Sr,(c) Th/Ta and (d) Nb/Th (�, West New Britain sources; , Admiralty Islands sources; �, Manus source; �, WestFergusson sources; �, East Fergusson sources; +, Gaua source; ¥, Vanua Lava source; *, Tafahi sources; ,Fiji artefacts). The location of the ANU9158 outlier is marked separately.



We then present heavy isotope results from two of the Lakeba Group 2 artefacts that indicatea regional source for the Lakeba artefacts in Fiji–northern Tonga.

Incompatible element abundance diagrams

The geochemistry of volcanic rocks from island arc environments in the Pacific has been thefocus of significant research directed at understanding the origin and processes involved in theproduction of volcanic rocks (cf., Cole et al. 1990; Peate et al. 1997; Ewart et al. 1998; Smith

Figure 3 Continued



et al. 2003; Raos and Crawford 2004). The considerable body of geochemical data from thesestudies can be used to examine potential provenance areas for the Lakeba Group 2 obsidian(cf., Collerson and Weisler 2007).

Mantle-normalized diagrams are commonly used to display relative elemental abundancesin specific rocks to identify fractionation processes in the melt (Sun and McDonough 1989). Suchdiagrams are arranged so that incompatible elements (elements that tend to not fractionateeasily in the melt) are plotted to the left of the diagram, with elements in decreasing order ofincompatibility to the right.

It is well known that island arc magmas are systematically depleted in Ta, Nb, Zr and Ti(high field strength elements, HFSE) (Albarède 2003), and as a result their magmas have atrace-element distribution that is readily distinguished from those of Ocean Island Basalts(OIBs). OIBs are formed from weak zones (hot spots) in oceanic plates and intra-plate volcan-ism, rather than subduction of one or more tectonic plates. As a result, volcanic rocks formedby intra-plate volcanism are generally less depleted in their trace elements (Albarède 2003). Agood example of obsidian sources with an OIB element pattern are those of Samoa (Sheppard1989). In contrast to OIB obsidian, the trace-element distribution of obsidian from the BanksIslands sources in Vanuatu clearly has an island arc trace-element pattern (Fig. 4).

The Group 2 artefacts from Lakeba also have an island arc trace-element distribution, andall of the intra-plate volcanic islands in the Pacific can therefore be ruled out as an origin(Smith et al. 1977; Collerson and Weisler 2007). It is worth noting that the Group 2 artefactshave an anomalous Sr depletion and Zr enrichment in contrast to the normal HFSE depletionin island arc magmas. In this, the artefacts are similar to most of the West Pacific obsidiansources, but differ from known obsidians in the Tongan Arc (Fig. 5).

The trace-element distribution of major obsidian sources in the West Pacific has an expectedisland arc signature, but no source is a close match to the Lakeba Group 2 artefacts (Figs 2 and5). Additional support for the view that the Group 2 obsidians cannot come from Vanuatu is

Figure 4 Incompatible element diagrams, primitive mantle-normalized, of selected obsidian sources and a calculatedOcean Island Basalts (OIB) source (Sheppard 1989; Sun and McDonough 1989) in comparison to Group 2 artefactcomposition.



that the volcanic rocks of the Central New Hebrides Arc (Peate et al. 1997; Raos andCrawford 2004) are more depleted in the majority of elements than the Group 2 artefacts,particularly the incompatible ones (Fig. 5).

In the Tongan–Kermadec Arc, volcanic rocks are divided into the Tongan volcanic rocks inthe northern part of the Arc and the Kermadec volcanic rocks in the southern part of the Arc(Ewart et al. 1977; Collerson and Weisler 2007). The Tongan rocks are, on average, moredepleted in almost all elements, but have a less prominent Ti depletion (Ewart and Hawkes-worth 1987). Most of the rocks from the southern Kermadec Arc are element depletedcompared to the Lakeba Group 2 artefacts. While some basalt samples from Macauley Islandin the Kermadec Arc have a similar trace-element pattern to the Lakeba Group 2 artefacts(especially sample Macauley 10384, Ewart et al. 1998, 367, table A1), the obsidian fromMacauley Island (including sample AH594, re-analysed with LA–ICP–MS in this study) isvery different (Smith et al. 2003) and rules out a Kermadec Arc source for the Group 2 artefacts.

Volcanic rocks from the Lau Arc in east Fiji are more diverse than rocks from the adjacentTongan–Kermadec Arc (Cole et al. 1990). The Mago Volcanic Group has an element patternwith a broadly OIB distribution, while rocks from the Korobasaga and Lau Volcanic Groupshave trace elements that are more strongly island arc distributed, but with significantly differentNb/Ta ratios (Ewart et al. 1998; Peate et al. 2001). No Lau Arc rocks yet reported have a trace-element distribution similar to the Lakeba Group 2 artefacts.

Overall, rocks from the Tongan–Kermadec Arc and the Lau Arc are more depleted inpotassium than the Lakeba Group 2 samples. However, there are rocks from Ono-I-Lau (Coleet al. 1990) with unusually high K2O values compared to other parts of the Lau Arc, but withheavy isotope ratios different from the Lakeba artefacts (see below).

Figure 5 Incompatible element diagrams, primitive mantle-normalized (Sun and McDonough 1989) from majorobsidian sources of the Western and Central Pacific and Lakeba Island Group 2 artefacts. The diagram supports theresults of the correspondence analysis, as Group 2 artefacts show no resemblance with the other West Pacific obsidiansources; readily detectable is the mentioned Zr anomaly, which distinguishes Group 2 from the known Tonga Arcobsidian source on Tafahi.



Radiogenic isotopes

The geochemistry of the Lakeba Group 2 artefacts indicates an island arc origin, but none ofthe known West and Central Pacific obsidian sources appears to be the source. A potentialsource area could not be identified from a comparison of the LA–ICP–MS trace-element datafrom volcanic rocks in the Tongan–Kermadec Arc and the Lau Arc. As a result, two obsidiansamples from Lakeba Group 2 (ANU9154 and ANU9161) were submitted for radiogenicisotope analysis (87Sr/86Sr, 143Nd/144Nd, 206Pb/204Pb, 207Pb/204Pb and 208Pb/204Pb) with a Multi-Collector LA–ICP–MS at the research facility of the School of Earth Science, MelbourneUniversity. The results are presented in Table 3.

The Pb-isotope distribution of the two Lakeba Group 2 samples display unusually high206Pb/204Pb and 208Pb/204Pb ratios (Fig. 6 (b)), that plot between classic OIBs such as Samoanvolcanic rocks and Indian and Pacific MORB (Mid-Ocean Ridge Basalt) signatures (Wrightand White 1987; Zhu 2007), as well as having 206Pb/204Pb and 207Pb/204Pb ratios in the upperrange of the Pacific MORB array (Fig. 6 (a)). In this, they are unlike volcanic rocks from theNew Hebrides Arc, which have an Indian and Pacific Ocean MORB distribution (Peate et al.1997, 1343; Raos and Crawford 2004, 50, fig. 8). This further supports the view that theprovenance of Group 2 artefacts lies outside the New Hebrides Arc.

One explanation for the high Pb-ratio of the Group 2 samples is the impact of mantleplumes, or mixture of both MORB and OIB, in the genesis of the parent rock, which has beendescribed for the northern Tongan Arc islands of Tafahi and Niuatoputapu (Wendt et al. 1997;Falloon et al. 2007), and for some parts of the Fiji Islands such as the Mago Volcanic Group(Cole et al. 1990, 544, table 2). Several researchers have also mentioned an OIB influence inthe genesis of the younger Fiji volcanic rocks of Viti Levu and adjacent islands (Gill 1984;Gill and Whelan 1989b), and both processes result in unusual high 206Pb/204Pb, 207Pb/204Pb and208Pb/204Pb ratios like those seen in the Group 2 obsidian. The high Pb-isotope ratios are alsoassociated with more radiogenic Sr and low Zr/Nb ratios (16–21, compared to MORBs with40–120) in the northern Tongan Arc (Wendt et al. 1997, 613) and the Mago Volcanic Group(Cole et al. 1990).

We have established that the Lakeba Group 2 artefacts did not originate from the Tafahiobsidian source in northern Tonga (Fig. 2). The Pb-isotope ratios point toward northern Tonga,

Table 3 Isotopic composition (MC–LA–ICP–MS) for two Group 2 obsidian artefacts found on Lakeba: for artefactprovenance, see Table 1

ANU9154 ANU9161

207Pb/204Pb 18.916 18.918206Pb/204Pb 15.571 15.572208Pb/204Pb 38.637 38.640207Pb/206Pb 0.82282 0.82277208Pb/206Pb 2.04259 2.0425687Sr/86Sr 0.703669 1 18 0.703757 1 13143Nd/144Nd 0.513025 1 8 0.513026 1 8

Pb mass bias corrected using thallium doping; 87Sr/86Sr normalized to 86Sr/88Sr = 0.1194 and reported relative to SRM987 =0.710230; 143Nd/144Nd normalized to 146Nd/145Nd = 2.0719425 (equivalent to 146Nd/144Nd = 0.7219) and reported relative to LaJolla =0.511860.



but the Group 2 artefacts have a MORB or island arc Zr/Nb ratio (60.1) that, along with theirhigh potassium, suggests that northern Tonga is unlikely to be the source.

The possibility of a local origin for the Lakeba artefacts from the back-arc spreading areaof the Lau basin (Ono-I-Lau, Lau Volcanic Group and Korobasaga Volcanic Group) is also not

Figure 6 The radiogenic isotope composition of two samples from Lakeba Island (ANU9154 and ANU9161: ) andcompositional fields of several Polynesian OIBs and Indian and Pacific Ocean MORBs taken from Clift and Vroon(1996), Collerson and Weisler (2007) and Falloon et al. (2007) (changed); the dashed line represents the compositionalfield of New Hebrides Arc volcanic rocks taken from Raos and Crawford (2004). (a) This reflects variation in U/Pb ratios(206Pb/204Pb and 207Pb/204Pb). (b) This reflects additional variation in Th/Pb (208Pb/204Pb). Based on their isotopiccomposition, the two analysed artefacts show an OIB impact, but with an island arc trace-element distribution,similar to northern Tonga volcanic rocks and selected rocks from western Fiji.



supported, as these rocks have significantly lower Pb-isotope ratios, which plot in the dis-tribution of Indian Ocean and Pacific Ocean MORBs (Ewart et al. 1998, 344–6, fig. 6), and Srratios that are less radiogenic than the Group 2 obsidian (Cole et al. 1990, 547, table 5; Peateet al. 2001; Sun et al. 2003).

Intermediate to high Pb-isotope ratios similar to those of the Group 2 artefacts are found inseveral volcanic rock series in Fiji. Gill and Whelan (1989b) attribute the high Pb-isotope ratiosof these younger Fijian volcanic rocks to the impact of OIBs in the genesis of the islands. Thebest matches with the Lakeba Group 2 artefacts are the shoshonitic and transitional basalts,and calcalkaline andesites reported from Viti Levu, Ovalau, Beqa, Ngau, Moala, Totoya andMatuku (e.g., Sambeto or Mamosi Quarry—cf., Gill 1984, 448, table 2; Gill and Whelan1989a, 4563, fig. 1). These rocks developed after the North Fiji basin split, in the transitionto the younger Ocean Alkali Basalt volcanism around 5–6 million years ago (Gill and Whelan1989b, 4579; Squire and Crawford 2007, 311, fig. 11). Gill (1984, 445) describes these rocksas medium rich in K, but not Fe enriched, with La/Yb > 2 (2.47), K/Rb ~ 475 (481) and K/Nb >8000 (8563) (Group 2 values in brackets—cf., Gill 1984, 446, table 1; Gill and Whelan1989a, 4563, table 1; Gill and Whelan 1989b, 4582, table 2). Until now, no volcanic glasses havebeen recorded on these islands. However, the occurrence of rocks in Fiji that have similarisotope ratios and trace-element composition to the Group 2 artefacts strongly suggests anorigin in the Fiji Islands from a small and currently undiscovered obsidian source.

CONCLUSION

The re-analysis of 18 obsidian artefacts from archaeological sites on Lakeba Island has shownthat they were transported from two different provenances, one of them (Group 1) located inthe Tonga archipelago. The focus of this paper was on the source of the remaining group ofartefacts (Group 2), which had previously been sourced to Vanuatu.

The starting-point for examining the provenance of the Group 2 artefacts was a multivariatestatistical analysis of 39 isotopes determined with LA–ICP–MS to compare the geochemicalcomposition of these artefacts with the known obsidian sources in the West and CentralPacific. Results demonstrated that the chemical composition of the Group 2 artefacts did notmatch with the northern Vanuatu obsidian sources, nor with any of the other obsidian sources.

To examine the geological environment in which the Group 2 artefacts were formed, severalmethods from Earth Science research were employed. The mantle-normalized trace-elementdistribution showed that the parent rock of the Group 2 artefacts developed in an island arc. Allobsidian sources with an Ocean Island Basalt (OIB) pattern—that is, all obsidian sources east of the‘Andesite Line’(Collerson andWeisler 2007, 1907, fig. 1A)—could be excluded as potential sources.

The trace-element concentration and distribution of Group 2 artefacts has some similaritywith several samples from the Tongan–Kermadec Arc and the Lau Arc–Back-arc region.However, these areas can be excluded on the basis of their low potassium values, which aresignificantly different to the high potassium values of the Group 2 artefacts.

Radiogenic isotopes determined with MC–LA–ICP–MS showed an unusual pattern outsideof the typical Indian or Pacific MORB composition seen in island arcs of the West Pacific,supporting the interpretation that the Group 2 artefacts cannot originate from anywhere in theNew Hebrides Arc. The high Pb ratios of the Group 2 artefacts can best be explained by anOIB influence in the genesis of the parent rock. This unusual pattern has been detected inrocks from the northern Tongan Arc and Fiji Islands, making these areas the most likely sourceof the Group 2 artefacts.



Because the significant low potassium value and low Nb/Zr ratio ruled out the Tongan–Kermadec Arc as the Group 2 source, the composition of several younger volcanics from westFiji was examined. Several of these have similar trace-element, and radiogenic isotope, ratiosto the Group 2 artefacts, leading to the conclusion that the Group 2 obsidian probably derivesfrom a small and as yet unknown source in west Fiji. Regardless of the precise location of theGroup 2 artefacts in Fiji–Tonga, they do not originate from Vanuatu as previously thought.

A source of the Group 2 artefacts in Fiji–Tonga has several implications for our understanding ofthe post-Lapita period, which covers about half of the entire prehistoric sequence in Remote Oceania.The post-Lapita ceramics of Fiji display episodes of significant stylistic change that have beeninterpreted as the result of interaction/migration from Vanuatu (Best 2002; Burley 2005; Cochrane2008). From a material culture perspective, there are two essential tests of this proposition.First, geochemical evidence for prehistoric interaction should be present in the archaeologicalrecord, particularly the presence of imports such as basalt, obsidian, chert and ceramics, that areknown to have been transported in the Pacific in the past (Best 1987; Dickinson 2006; Collerson andWeisler 2007). The re-analysis of the Lakeba obsidian indicates that it derives from an unknownsource in Fiji–Tonga rather than from Vanuatu. This finding removes the only direct evidencefor contact between Vanuatu and Fiji in the post-colonization era (c. 2500–1000 cal. bp).

Second, the prehistoric ceramics of Vanuatu should be stylistically similar to those of Fiji,and be of comparable size and form, and employ similar decorative methods and designs.While a substantial migration could reasonably be expected to result in the transmission of alarge part of a ceramic complex—with high similarity between the ceramics found at migrantsource and destination (e.g., Ashikhmina 1997)—the effects of interaction, trade and exchangeon a pottery assemblage are more complicated, and might result in only the partial transmissionof the ceramic repertoire. Until recently, the prehistoric cultural sequence and pottery ofVanuatu was known from only a handful of excavations, and a comparison with Fijian cera-mics was based on a small number of vessel attributes found in highly fragmented and poorlydated collections (Bedford and Clark 2001; Bedford 2006). As knowledge of Vanuatu ceramicshas improved it has become apparent that the post-Lapita ceramics of Vanuatu differ substan-tially from those of Fiji, ruling out the possibility of a large-scale transfer of a Vanuatuceramic assemblage to Fiji (Bedford 2006; Bedford and Spriggs 2008).

The possibility that low-level interaction and contact between Vanuatu and Fiji influencedthe ceramic trajectory of Fiji should not be entirely ruled out, because archaeological under-standing of culture contact and its effect on ceramic transmission is still in its infancy in thePacific (Clark and Murray 2006; Cochrane 2008), and the geochemical database of archaeo-logical items from the West and Central Pacific is relatively small and focused on obsidians andceramics. Nonetheless, a probable origin for the Group 2 obsidian in Fiji–Tonga, combinedwith diminishing ceramic support for a connection between Fiji and Vanuatu, demonstratesthat no large-scale network spanned the West and Central Pacific in post-Lapita times (Spriggs1997, 2004; Wahome 1997). Thus, hypotheses that focus on internal archipelago-based culturalprocesses (e.g., Clark 2000; Field 2004, 94) are likely to have greater explanatory potential forunderstanding post-Lapita societies than do those that emphasize the significance of long-distance interaction and migration between the islands of the West and Central Pacific.

ACKNOWLEDGEMENTS

We would like to thank Simon Best for providing the obsidian artefacts from Lakeba forre-analysis. In addition, we would like to thank William Dickinson, Steve Eggins, James Gill



and Foss Leach for discussing the geochemical data from the Lakeba obsidians. Special thanksto Peter Sheppard, who provided an obsidian sample from Macauley Islands for analysis, andAtholl Anderson for discussing the distribution and antiquity of the Kermadec Islands obsidian.Lastly, our thanks go to Robin Torrence, Matthew Spriggs, Iona Flett and Stuart Bedford fortheir many useful comments on an earlier draft of this paper.

REFERENCES

Albarède, F., 2003, Geochemistry: an introduction, Cambridge University Press, Cambridge.Ambrose, W. R., 1976, Obsidian and its prehistoric distribution in Melanesia, in Ancient Chinese bronzes and South

East Asian metal and other archaeological artifacts (ed. N. Barnard), 351–78, National Gallery of Victoria,Melbourne.

Ashikhmina, L., 1997, Ancient migrations in the northern sub-Urals: archaeology, linguistics and folklore, inArchaeology and language I. Theoretical and methodological orientations (eds. R. Blench and M. Spriggs), 282–307, Routledge, London.

Barsdell, M., Smith, I. E. M., and Spoerli, K. B., 1982, The origin of reversed geochemical zoning in the NorthernNew Hebrides volcanic arc, Contributions to Mineralogy and Petrology, 81, 148–55.

Bedford, S., 2006, Pieces of the Vanuatu puzzle: archaeology of the north, south and centre, Terra Australis, vol. 23,Pandanus Books, Canberra.

Bedford, S., and Clark, G. R., 2001, The rise and rise of the incised and applied relief tradition: a review and re-assessment, in The archaeology of Lapita dispersal in Oceania: papers from the fourth Lapita Conference, June2000 (eds. A. Anderson, G. R. Clark and T. Vunidilo), 61–74, Terra Australis, Pandanus Books, Canberra.

Bedford, S., and Spriggs, M., 2008, Northern Vanuatu as a Pacific crossroads. The archaeology of discovery, interaction,and the emergence of the ethnographic present, Asian Perspectives, 47(1), 95–120.

Bedford, S., Sand, C., and Connaughton, S. P. (eds.), 2007, Oceanic explorations: Lapita and Western Pacific settlement,Terra Australis, vol. 26, ANU E Press, Canberra.

Bellwood, P., and Koon, P., 1989, Lapita colonists leave boats unburned! The questions of Lapita links with IslandSouth East Asia, Antiquity, 63, 613–22.

Benzécri, J.-P., 1992, Correspondence analysis handbook, Marcel Dekker, New York.Best, S., 1984, Lakeba: the prehistory of a Fijian island, Ph.D. thesis, University of Auckland, Auckland.Best, S., 1987, Long-distance obsidian travel and possible implications for the settlement of Fiji, Archaeology in

Oceania, 22, 31–2.Best, S., 2002, Lapita: a view from the east, New Zealand Archaeological Association Monographs, vol. 24, New

Zealand Archaeological Association, Auckland.Burley, D., 2005, Mid-sequence archaeology at the Sigatoka sand dunes with interpretive implications for Fijian and

oceanic culture history, Asian Perspectives, 44(2), 320–48.Clark, G. R., 2000, Post-Lapita Fiji: cultural transformation in the mid-sequence, Ph.D. thesis, Australian National

University, Canberra.Clark, G. R., 2003, Shards of meaning: archaeology and the Melanesia–Polynesia divide, The Journal of Pacific History,

38(2), 197–215.Clark, G. R., and Murray, T., 2006, Decay characteristics of the eastern Lapita design system, Archaeology in

Oceania, 41, 107–17.Clift, P. D., and Vroon, P. Z., 1996, Isotopic evolution of the Tonga Arc during Lau Basin rifting; evidence from the

volcaniclastic record, Journal of Petrology, 37(5), 1153–73.Cochrane, E. E., 2008, Migration and cultural transmission: investigating human movement as an explanation for

Fijian ceramic change, in Cultural transmission and archaeology: issues and case studies, (ed. M. J. O’Brian),132–45, Society for American Archaeology, The SAA Press, Washington, DC.

Cole, J. W., Graham, I. J., and Gibson, I. L., 1990, Magmatic evolution of Late Cenozoic volcanic rocks of the LauRidge, Fiji, Contributions to Mineralogy and Petrology, 104, 540–54.

Collerson, K. D., and Weisler, M. I., 2007, Stone adze compositions and the extent of ancient Polynesian voyagingand trade, Science, 317(5846), 1907.

Dickinson, W. R., 2006, Temper sands in prehistoric Oceanian pottery: geotectonics, sedimentology, petrography,provenance, Geological Society of America Special papers vol. 406, The Geological Society of America, Boulder, CO.

Dickinson, W. R., and Burley, D. V., 2007, Geoarchaeology of Tonga: geotechtonic and geomorphic controls, Geo-archaeology: an International Journal, 22(2), 229–59.



Dickinson, W. R., Sinoto, Y. H., Shutler, R. Jr., Shutler, M. E., Garanger, J., and Teska, T. M., 1999, Japanese Jomonsherds in artifact collections from Mele Plain on Efate in Vanuatu, Archaeology in Oceania, 34(1), 15–24.

Duerden, P., Clayton, E., Bird, J. R., Ambrose, W. R., and Leach, B. F., 1987, Obsidian composition catalogue, inArchaeometry: further Australasian studies (eds. W. R. Ambrose and J. M. J. Mummery), 232–8, OccasionalPapers in Prehistory, RSPAS and ANU, Canberra.

Ewart, A., and Hawkesworth, C. J., 1987, The Pleistocene–Recent Tonga–Kermadec Arc lavas: interpretation of newisotopic and rare earth data in terms of a depleted mantle source model, Journal of Petrology, 28(3), 495–530.

Ewart, A., Brothers, R. N., and Mateen, A., 1977, An outline of the geology and geochemistry, and the possible petro-genetic evolution of the volcanic rocks of the Tonga–Kermadec–New Zealand Island Arc, Journal of Volcanologyand Geothermal Research, 2, 205–50.

Ewart, A., Collerson, K. D., Regelous, M., Wendt, J. I., and Niu, Y., 1998, Geochemical evolution within the Tonga–Kermadec—Lau Arc—Back-arc systems: the role of varying mantle wedge composition in space and time, Jour-nal of Petrology, 39(3), 331–68.

Falkner, K. K., Klinkhammer, G. P., Ungerer, C. A., and Christie, D. M., 1995, Inductively coupled plasma massspectrometry in geochemistry, Annual Review of Earth and Planetary Sciences, 23, 409–49.

Falloon, T. J., Danyushevsky, L. V., Crawford, A. J., Maas, R., Woodhead, J. D., Eggins, S. M., Bloomer, S. H.,Wright, D. J., Zlobin, S. K., and Stacey, A. R., 2007, Multiple mantle plume components involved in thepetrogenesis of subduction-related lavas from the northern termination of the Tonga Arc and northern LauBasin: evidence from the geochemistry of arc and backarc submarine volcanics, Geochemistry, Geophysics,Geosystems, 8(9), 1–45.

Field, J. S., 2004, Environmental and climatic considerations: a hypothesis for conflict and the emergence of socialcomplexity in Fijian prehistory, Journal of Anthropological Archaeology, 23, 79–99.

Gill, J. B., 1984, Sr–Pb–Nd isotopic evidence that both MORB and OIB sources contribute to oceanic island arcmagmas in Fiji, Earth and Planetary Science Letters, 68, 443–58.

Gill, J. B., and Whelan, P., 1989a, Early rifting of an oceanic island arc (Fiji) produced shoshonitic to tholeiiticbasalts, Journal of Geophysical Research, 94(B4), 4561–78.

Gill, J. B., and Whelan, P., 1989b, Postsubduction ocean island alkali basalts in Fiji, Journal of GeophysicalResearch, 94(B4), 4579–88.

Juggins, S., 2005, C2 data analysis, University of Newcastle, Newcastle.Kennett, D. J., Anderson, A. J., Cruz, M. J., Clark, G. R., and Summerhayes, G. R., 2004, Geochemical characteriza-

tion of Lapita pottery via inductively coupled plasma-mass spectrometry (ICP–MS), Archaeometry, 46(1), 35–46.Leach, B. F., and Manly, B., 1982, Minimum Mahalanobis distance functions and lithic source characterisation by

multi-element analysis, New Zealand Journal of Archaeology, 4, 77–109.Longerich, H. P., Jackson, S. E., and Günther, D., 1996, Laser ablation inductively coupled plasma mass spectrometric

transient signal data acquisition and analyte concentration calculation, Journal of Analytical Atomic Spectrometry,11, 899–904.

Peate, D. W., Kokfelt, T. F., Hawkesworth, C. J., van Calsteren, P. W., Hergt, J. M., and Pearce, J. A., 2001, U-seriesisotope data on Lau Basin glasses: the role of subduction-related fluids during melt generation in back-arc basins,Journal of Petrology, 42(8), 1449–70.

Peate, D. W., Pearce, J. A., Hawkesworth, C. J., Colley, H., Edwards, C. M. H., and Hirose, A. K., 1997, Geochemicalvariations in Vanuatu Arc lavas: the role of subducted material and a variable mantle wedge composition, Journalof Petrology, 38(10), 1331–58.

Raos, A. M., and Crawford, A. J., 2004, Basalts from the Efate Island Group, central section of the Vanuatu arc, SWPacific: geochemistry and petrogenesis, Journal of Volcanology and Geothermal Research, 134, 35–56.

Reepmeyer, C., 2008, Characterising volcanic glass sources in the Banks Islands, Vanuatu, Archaeology in Oceania,43(2), 120–7.

Reepmeyer, C., in prep., Obsidian sources and distribution systems emanating from Gaua and Vanua Lava on theBanks Islands of Vanuatu, Ph.D. thesis, Australian National University, Canberra.

Sheppard, P. J., 1989, Samoan volcanic glass, Archaeology in Oceania, 24, 70–4.Sheppard, P. J., 1996, Hard rock: archaeological implications of chert sourcing in Near and Remote Oceania, in

Oceanic culture history: essays in honour of Roger Green (eds. J. Davidson, G. Irwin, B. F. Leach, A. Pawley andD. Brown), 99–115, New Zealand Journal of Archaeology Special Publications, Auckland.

Smith, I. E. M., Steward, R. B., and Price, R. C., 2003, The petrology of a large intra-oceanic silicic eruption: the Sandy BayTephra, Kermadec Arc, Southwest Pacific, Journal of Volcanology and Geothermal Research, 124, 173–94.

Smith, I. E. M., Ward, G. K., and Ambrose, W. R., 1977, Geographic distribution and the characterization of volcanicglasses in Oceania, Archaeology and Physical Anthropology in Oceania, 12, 173–201.



Spriggs, M., 1997, The island Melanesians, Oxford, Blackwell.Spriggs, M., 2003, Post-Lapita evolutions in Island Melanesia, in Pacific archaeology: assessments and prospects.

Proceedings of the International Conference for the 50th anniversary of the first Lapita Excavation, Koné-Nouméa2002 (ed. C. Sand), 213–20, Les Cahiers de L’Archéologie en Nouvelle-Calédonie, Museum of New Caledonia,Noumea.

Spriggs, M., 2004, Is there life after Lapita, and do you remember the 60s? The post-Lapita sequences of the WesternPacific, in A Pacific odyssey: archaeology and anthropology in the Western Pacific: papers in honour of Jim Specht(eds. V. Attenbrow and R. Fullagar), 139–44, Records of the Australian Museum, Supplement, Sydney.

Squire, R. J., and Crawford, A. J., 2007, Magmatic characteristics and geochronology of Ordovician igneous rocksfrom the Cadia–Neville region, New South Wales: implications for tectonic evolution, Australian Journal of EarthSciences, 54, 293–314.

Summerhayes, G. R., in press, Melanesian obsidian—its sources, characterisation and distribution, in Proceedings ofthe International Obsidian Summit (eds. W. R. Ambrose and G. R. Summerhayes), Rikkyo University, Tokyo.

Sun, S.-S., and McDonough, W. F., 1989, Chemical and isotopic systematics of oceanic basalts: implications for mantlecomposition and processes, in Magmatism in the ocean basins (eds. A. D. Saunders and M. J. Norry), 313–45,Geological Society Special Publications, Blackwell Scientific, Oxford.

Sun, W., Bennett, V. C., Eggins, S. M., Arculus, R. J., and Perfit, M. R., 2003, Rhenium systematics in submarineMORB and back-arc basin glasses: laser ablation ICP–MS results, Chemical Geology, 196, 259–81.

Wahome, E. W., 1995, Ceramics and prehistoric exchange in the Admiralty Islands, Papua New Guinea, Ph.D. thesis,Australian National University, Canberra.

Wahome, E. W., 1997, Continuity and change in Lapita and post-Lapita ceramics: a review of evidence from theAdmiralty Islands and New Ireland, Papua New Guinea, Archaeology in Oceania, 21(1), 118–23.

Ward, G., 1979, Prehistoric settlement and economy in a tropical small island environment: the Banks Islands,Insular Melanesia, Ph.D. thesis, Australian National University, Canberra.

Weisler, M. I. (ed.), 1997, Prehistoric long-distance interaction in Oceania: an interdisciplinary approach, NewZealand Archaeological Association monograph, vol. 21, New Zealand Archaeological Association, Auckland.

Wendt, J. I., Regelous, M., Collerson, K. D., and Ewart, A., 1997, Evidence for a contribution from two mantleplumes to island-arc lavas from northern Tonga, Geology, 25(7), 611–14.

Wright, E., and White, W. M., 1987, The origin of Samoa: new evidence from Sr, Nd, and Pb isotopes, Earth andPlanetary Science Letters, 81, 151–62.

Zhu, B., 2007, Pb–Sr–Nd isotopic systematics of mantle-derived rocks in the world, Earth Science Frontiers, 14(2),24–36.



POST-COLONIZATION INTERACTION BETWEEN VANUATU AND FIJI RECONSIDERED: THE RE-ANALYSIS OF OBSIDIAN...

Documents

Transcript of POST-COLONIZATION INTERACTION BETWEEN VANUATU AND FIJI RECONSIDERED: THE RE-ANALYSIS OF OBSIDIAN...