Printer Friendly

A reanalysis of the Tikopia obsidians.


In 1982 an initial sourcing of 13 obsidians and volcanic glasses from Tikopia in the Solomon Islands suggested that four specimens came from Bismarcks sources, with Talasea in West New Britain being the most likely, and the rest came from the Banks Islands. Reanalysis now attributes ten pieces to Banks Islands sources and three to sources in the Admiralty Islands.

Keywords: obsidian, Tikopia, Banks Islands, density, PIXE-PIGME


Kirch and Yen's (1982) now-classic monograph on the prehistory of the Polynesian outlier island of Tikopia in the south-east Solomons presented a three-phase prehistoric sequence: the Kiki Phase c.900-100 BC, the Sinapupu Phase 100 BC-1200 AD, and the Tuakamali Phase 1200-1800 AD. A few dentate-stamped sherds and other aspects of material culture linked the early part of the Kiki phase to the Lapita tradition, while the pottery of this phase was predominantly a plain ware. The Sinapupu phase was associated with pottery decorated with incised and applied relief designs, and shows some links with northern Vanuatu pottery of the same time period, while the Tuakamali phase was associated with the introduction of Polynesian language and culture.

During each phase Tikopia maintained exchange relations with areas either/or both to the north or south (see Kirch 1986 for details; corrected as concerns the pottery in Dickinson 2006:63, 66). During the Kiki phase obsidian came from the Bismarck Archipelago, metavolcanic adzes and chert from the main Solomons, and further obsidians from the Banks Islands to the south. During the Sinapupu phase chert continued to come from the Solomons, pottery was imported from either northern Vanuatu or from Vanikoro in the SE Solomons, with one sherd sourced to Fiji, and obsidian continued to come from the Banks. In the Tuakamali phase we have evidence of stone adzes coming from Samoa, with Banks Islands' obsidian imports continuing. Ethnohistoric records show exchange relations with the closest inhabited islands of Anuta and Vanikoro and with Vanua Lava in the Banks Islands to the south.

An analysis by Larry Olson of 13 specimens of obsidian and/or volcanic glass flakes from Tikopia was presented by Kirch and Yen (1982:256-60), using both petrography and density measurements (the latter measurement referred to there as 'specific gravity'). It was concluded that some samples could be sourced to the Bismarck Archipelago with Talasea in West New Britain identified as the likely source, while the rest of the samples were assigned to the Banks Islands sources. In the late 1980s we decided to re-analyse this material using PIXE-PIGME as part of the Lapita Homeland Project in order to see if more definite source attributions could be made, obsidian technology and exchange being one focus of the Project (Allen and Gosden 1991). Density was measured at the ANU, and this allowed a comparison with the results obtained by Olson as well as a further test of inter-investigator comparability.

Methods and sample selection

University of Hawaii study

An initial division of volcanic glass flakes was made by Kirch on hand specimen into 'obsidians' and 'basaltic glasses'. Fourteen 'obsidians' were identified, being 'relatively clear, gray glasses, highly isotropic, and quite transparent near the edges', along with 625 'basaltic glasses', defined as 'darker glasses, sometimes with phenocrysts, or banded, more opaque glasses of andesitic and/or basaltic grade' (Kirch and Yen 1982:257).

To investigate the source of the flakes density measurements and examination of petrographic thin sections were carried out on 13 of them by Larry Olson, at that time a doctoral candidate at the University of Hawaii. Although not explicitly stated in the Kirch and Yen monograph, the selection of these 13 flakes was clearly designed to include samples from all three phases of prehistoric occupation of the island (1). The 13 flakes were divided by Olson into seven groups on the basis of petrographic analysis of thin-sections, and density measurements were carried out using toluene (density 0.867g/ml) as the immersion liquid (UH 1980). The source data for the density measurements in Figure 104 (Kirch and Yen 1982:257), which are attributed to Roger Green in the text, come from an early version of Ambrose's work on separation of obsidian sources by density (Ambrose n.d.), as described below.

ANU Study

The ANU system for measuring density is based on the heavier liquid perfluoro-methyl-decalin (density 1.967 g/ml at 20[degrees]C). This gives greater separation between the air weight and immersed weight of an object, and therefore produces better density measurements (Ambrose n.d., detailed by Ambrose and Stevenson 2004). Density is defined in this work as the density in g/[cm.sup.3] of an obsidian flake relative to the maximum density of water taken as 1g/ml at 4[degrees]C).

An appreciation of the difficulty of relying on density for source attribution can be seen in the statistic of each known source. Table 1 displays the problem of separating the Talasea and Lou-Pam islands sources from the other Papua New Guinea obsidians deriving from the West Fergusson Island sources of Fagalulu and Iaupolo-Igwageta, and East Fergusson sources of Dobu, Sanaroa and Numanuma. On the other hand there is a usefully wide density separation between the Talasea (Groups 4,5) and Banks Island sources (Groups 10,11).

The unstable volcanic landscapes that produce the obsidian are a constant variable in the history of obsidian resource location, production, and availability. A corollary is that unless there is supporting independent chemical analysis a confident density separation of an archaeological collection is constrained within the measured ranges of the modern field samples. This is shown in a comparison between the limited known source collections and the more numerous archaeologically-found artefacts given the density ranges for measurements made up to the late 1980s (Table 1). The density values of the chemically differentiated Talasea and Lou sources can be compared with the chemically sorted artefacts (Figure 1). It can be seen that although the range within the artefacts overlaps (items 1,4), the modern reference source specimens are more convincingly separated by density (items 2,5). This arises from the simple observation that prehistoric long-term resource exploitation was unlike sampling conditions available to present day archaeology.


There is a potential for wider density variation accompanying any broader field sampling because of the volcanic origin of obsidian. All volcanic glasses begin as liquid magma that can carry microscopic crystal inclusions and gas vesicles. There will also be differences in water content and cooling rates resulting from local eruption conditions. All these factors will alter the final density range of an obsidian collection. The very limited number of Banks Islands specimens (items 10,11) overlap with the range of the Lou Island examples. The D'Entrecasteaux group encompasses the ranges of the Talasea-Lou-Pam-Banks examples. There may be some value in the density results for separating East and West Fergusson Islands sources when collections in the local region require preliminary sorting prior to more detailed chemical analysis. Thus overall it can be seen that density measurement has some value only when the possible sources of an archaeological collection are already known, or as a preliminary sorting procedure for large collections of artefacts to be followed by chemical characterization of selected specimens that cannot be clearly separated by density (for examples of the latter procedure see Green 1987; cf. Torrence and Victor 1995).

The Tikopian sample available to us consisted of fragments remaining after the petrographic thin sections had been taken from the 13 flakes by Olson. In most cases this results in two and in one case three pieces of the original flake. This 'expands' the database to 23 specimens. Density was re-measured on these at the ANU by two different operators prior to analysis at ANSTO using PIXE/PIGME.

PIXE-PIGME study at Lucas Heights

The techniques used in conducting Proton Induced X-Ray Emission (PIXE) and Proton Induced Gamma Ray Emission (PIGME) became standardized during the 1980s. They have been described many times before (Ambrose et al. (1981), Bird et al. (1981), Duerden et al. (1979) and Summerhayes et al. (1998). All analyses were carried out at the Australian Nuclear Science and Technology Organisation (ANSTO) facility at Lucas Heights, NSW, Australia. The results of the ANSTO PIXE-PIGME analyses are given in Tables 2 and 3.


University of Hawaii (UH) Study

As noted above, Kirch had first separated the glasses by visual examination of hand-specimens into 'obsidians' and 'basaltic glasses'. With one exception the 14 'obsidians' all came from the TK-4 site, eight from stratigraphically-secure layer II and five from Layer I, 'probably also from the Kiki Phase, having been displaced by gardening' (Kirch and Yen 1982:257). The one other 'obsidian', from site TK-36, was deemed to be intrusive into Tuakamali Phase deposits from the underlying late Kiki Phase levels of the site. This attribution as intrusive appears to have been made solely on the basis of the specimen's identification as being of obsidian grade (ibid:101). It receives no support from our reanalysis below.

Of the 'obsidians', six were submitted to the UH Archaeology Laboratory for examination: two from TK-4 Layer II, three from Layer I and the single specimen from TK-36. In addition, seven of the 625 'basaltic glass' flakes from Tikopian sites were also submitted.

The UH Laboratory analyses yielded a more complex picture than that gained from hand-specimen examination. Of Olson's seven petrographic groups, Groups III and VI, totaling four specimens, were grouped together after 'specific gravity' measurements and interpreted as being 'from the west with Talasea as a likely source' (Kirch and Yen 1982:260). The other groups (I, II, IV, V and VII) were reinterpreted as andesitic and/or basaltic grade glass coming from the Banks Islands or other as yet unrecorded sources in Vanuatu. Thus the hand-specimen identifications were not completely confirmed by petrography and density measurements. This appears to have been overlooked by Kirch in a later paper where in a discussion of the 14 'obsidians' (hand-specimen identification), it is stated that, 'standard petrographic sectioning and measurement of specific gravity suggest that the probable source of this material was Talasea' (1986:38).

In fact, of the six 'obsidians' tested by Olson, only three were defined as obsidian (Groups III and VI) and tentatively attributed to Talasea, and a further flake which on hand-specimen was assigned to the 'basaltic glass' category was re-assigned to an obsidian group (Group VI) (2). When attribution to source using density was made, Admiralty Islands sources do not appear to have been fully considered although on the source data presented in Figure 104 (Kirch and Yen 1982:257), eight of the specimens could have been assigned to those sources. As reported by Kirch and Yen (ibid:258,260), only tentative indications of sources for the obsidians were suggested on the basis of the density measurements, and more detailed analyses were recommended involving major and minor element composition.

ANU study

The results of PIXE-PIGME analysis (Tables 2 and 3) show that 10 out of 13 flakes could be sourced to the Banks Islands of Northern Vanuatu, while three were attributed to the Admiralty Islands (Marius Province) of Papua New Guinea. Of the three flakes sourced to the Admiralties, two were from secure Early Kiki Phase deposits (Layer II) from what was probably the initial occupation of the island at site TK-4 and the other was from the upper layer (Layer 1) of the same site, presumably representing disturbance of the early deposits.

If we take the UH density figures and petrographic groups and the updated source data from Table 1 then, as shown in Table 4, two specimens would probably be from Talasea (both Group VI), five from the Admiralties (Groups I, IV, VII), and two from the Banks Islands (Groups II and V), with four samples giving anomalous values (Group III and one each of V, VI, VII) (3). These results, however, were not confirmed when the 'expanded' database of 23 specimens was re-analysed for relative density in the then-ANU Prehistory Department. As can be seen in Table 4 the UH and ANU density measurements on the same specimens were quite different. On the first set of ANU measurements carried out by technical officer Geoff Deeble, no specimens were assigned to Talasea, eight representing six original flakes were attributed to the Admiralties, ten representing six original flakes were assigned Banks Islands sources, four anomalous readings were recorded and one sample was too small to measure (ANU-D in Table 4). PhD student Michael Hanslip's independent measurements (Hanslip 2001:230) were closely comparable to Deeble's in all but one case: one of Deeble's attributions to Lou was changed to a Banks source (ANU-H in Table 4).

PIXE-PIGME results

The PIXE-PIGME analyses (Tables 2 and 3) showed that two Admiralty Islands sources were represented: Umrei on Lou Island and the source on Pam Lin Island. Similarly, two Banks Islands sources were possibly represented, Vanua Lava Island and Gaua Island (For source locations see Bird et al. 1981). Two samples were sourced to Gaua, but the other samples from the same individual flakes gave Vanua Lava readings. The eight other Banks Islands flakes came unequivocally from Vanua Lava sources.

These results allow re-assessment of the methods used in the original analyses undertaken by Olson at the UH Archaeology Laboratory. If the UH and ANU results are compared to the PIXE/PIGME analyses only one out of nine flakes attributable to source was correctly assigned by the UH Laboratory (11% success), while 13 out of 18 pieces were correctly identified at the ANU by Deeble (72% success) and 14 by Hanslip (78% success). This study thus confirms Torrence and Victor's (1995) remarks, both on the usefulness of density (when calculated correctly) and its limitations if not backed up by a systematic program of source attribution using more advanced techniques.


There is clearly a need for further analyses of the Tikopia material, both of the eight other 'obsidians' identified in hand-specimen and of a sample of the remaining 618 volcanic glass specimens. Of the 13 obsidian flakes analysed by PIXE/PIGME only three turned out to be from western sources, the rest coming from the Banks Islands to the south. Kirch has suggested (1986:39) that the western-source obsidians might represent items brought to Tikopia on an initial colonization voyage. The analyses reported here support this idea, as the Admiralties obsidians are indeed only found in the earliest Kiki Phase site on the island and occur there in very small quantities. Utilization of the relatively close Banks Islands sources appears to have started as soon as the island was settled and, in contrast, supplies from these sources were maintained throughout Tikopia's prehistory.

This pattern of distribution from Banks Islands and Admiralties obsidian sources is in such contrast with that pertaining at the nearest recorded Lapita sites in the Reefs-Santa Cruz Islands that it requires comment. Of the three sites so far published in any detail, Green (1991) originally contended that the earliest was SZ-8, occupied over some 300 years starting in the 12th to 11th centuries BC, followed by RF-2 dating to before the end of the 10th century BC and a short-term occupation of a generation or so, and the later site of RF-6 dating to the earlier 7th century BC. At this latter site it is suggested that the small quantity of obsidian results from scavenging of older deposits and so may not represent continuing contacts with Bismarcks obsidian sources at the time of occupation. From the 3 sites 97.5% (948 pieces) of obsidian are from West New Britain sources, with only 1.12% (11 pieces) coming from the Admiralties and 1.13% (12 pieces) from the Banks Islands, and a single piece (0.1%) from West Fergusson (Green 1987; the figures in Sheppard 1992 appear to be slightly different).

More recently further dates have been obtained from the three sites and Bayesian analysis has been used in an attempt to tie down the period of their occupation (Green et al. 2008; Jones et al. 2007). For SZ-8, based on a 'much sounder appraisal' (Green et al. 2008:50), it has now been decided that the span of occupation would have been only about 100 years, with about 50 years or less for RF-2. The results of the Bayesian analysis of SZ-8 are not particularly helpful with a start range of 3650-3000 BP and an end date of 3300-2600 BP at the 95% confidence interval, leading to a possible duration of occupation of 0-950 years! This is then narrowed down to a start date of 3300-3200 BP based on the fact that "few would advocate ages for Lapita sites in Remote Oceania as being greater than 3200-3300 BP" (ibid:59). But this is a somewhat circular point as the dates usually suggested are themselves largely based on previous assessments of the dating of SZ-8 and RF-2. For our purposes though the important date is the youngest Bayesian possible start date of 3000 BP. If one accepts the suggestion of only 100 years duration for SZ-8 then the very latest it could have been abandoned would therefore be 2900 BP or 850 BC.

The earliest current settlement date for Tikopia is from the base of TK-4 at 2680 [+ or -] 90 bp (UCR-964) which calibrates using CALIB 5.0.2 (Reimer et al. 2004) to 972-777BC at one standard deviation, with a probability of 0.96 that the date is between 936-777BC (cf. Kirch and Yen 1982:11 l- 125, 313). This suggests that the entire occupation of SZ-8 is probably older. A similar exercise for RF-2 would also suggest, given its likely overlap in age with SZ-8 based on pottery decoration, that it is again earlier than the initial settlement of Tikopia.

There are currently two plausible models that might explain why obsidian from New Britain dominates at some Lapita sites while at others the Admiralties sources are the major supplier. The first was articulated by Specht (2002:44) who saw early obsidian use at colonizing sites as part of a conscious pattern to replicate the ancestral societies. The pattern of source distribution therefore would reflect the regular sources of supply of the ancestral communities responsible for the colony:
   Thus, the prominence of Manus obsidian at the Buka
   sites might indicate that the ancestral communities had
   some kind of privileged (or restricted) access to Manus
   obsidian rather than to Willaumez sources ... In contrast,
   the communities ancestral to the Reef-Santa Cruz sites
   might have had links that derived obsidian primarily from
   Willlaumez Peninsula. If this were so, the archaeological
   distributions of obsidian from the various source regions
   could have significant historical meaning indicating at
   least two originating ancestral communities or areas for
   the colonizing process into Remote Oceania, each
   seeking to replicate itself (ibid.).

The alternative has been argued by Summerhayes (2003, 2004) who sees different patterns of source region distribution as representing temporal phases within a relatively homogeneous Lapita culture divided in the Bismarcks region at least into Early Lapita (1550-1050/950 BC), Middle Lapita (1050/950-750/650BC) and Late Lapita (750/650-c.250 BC). In the early and Late Lapita phases it is the Willaumez Peninsula sources in New Britain that dominate, while in the Middle Lapita phase "assemblages in the eastern Bismarck Archipelago region are dominated by Admiralties obsidian" (2004:154) with New Britain obsidian only remaining dominant in assemblages close to the sources. Summerhayes' Late Lapita phase is probably not pertinent outside of the Bismarck Archipelago and the adjacent Near Oceanic parts of the Solomon Islands as there is no evidence from Remote Oceania of any continuing transport of Bismarcks' obsidian during this time period. But the contrast between initial settlement of the Reefs-Santa Cruz taking place during the Early Lapita phase, and that of Tikopia taking place during the Middle Lapita phase does seem relevant. Wickler (2001:178) dates the earliest Buka Lapita site at the northern end of the Solomons to about 800 BC and the obsidians are overwhelmingly from Admiralties sources. Tikopia at the southern end was first settled at about the same (Middle Lapita) time, just before all direct contact with the Bismarcks sources ceased.

Summerhayes (2003) discusses why these shifts occurred, comparing the evidence for pottery with that of obsidian within the Bismarck Archipelago. He rejects a model that the development of separate obsidian networks in the Middle Lapita perod "equates with a regionalisation of the area or social break up of Early Lapita communities" (2003:143). He sees the separate networks as relating in some way to a higher level of mobility of people between settlements in the Early Lapita phase as compared to the Middle phase. The dominance of West New Britain obsidian in the Early phase is seen as "an expression of the direction of initial impetus for Austronesian expansion" (2003:142), which he notes comes from New Britain rather than Manus on linguistic grounds. Procurement of obsidian could have been direct or through a small number of hands. In contrast, in the Middle phase, Summerhayes sees down-the-line exchange as the dominant means of distribution operating among more stable communities, with closeness to source determining the amount of obsidian in a site.

Alternatively, the changing distribution of obsidian from New Britain and Manus sources may relate to changing prestige of the two supply areas over time. This would not necessarily have been in relation to obsidian as such, but to other products perhaps leaving no material traces today--including ceremonies or dances (cf. Terrell and Welsch 1997:568)--or to political realignment. The switch in the eastern Bismarcks could relate to disruption of the supply routes out of New Britain, perhaps caused by hostile intervening groups. The later resurgence of West New Britain as the dominant obsidian source could relate to the reestablishment of the status quo, or perhaps a further sociopolitical shift once regular connections with the more distant Lapita sites such as those in the SE Solomons were broken.

A third explanation might simply be chance. There was always a small percentage of Admiralties obsidian traveling through the exchange system, even when West New Britain sources were dominant in the early Lapita phase. It is therefore of course possible that only pieces of Admiralties obsidian reached Tikopia, rather than the predominant West New Britain material. We can in fact calculate the possibility of this being the case using the percentages from the Reefs-Santa Cruz sites as representing the norm of transported obsidian. On this basis the chances of 100% Admiralties obsidian reaching Tikopia at the time would be about 1 in 690,000 (0.000145%). If we include the up to three pieces of Banks obsidian that might also come from Early Kiki phase deposits--alternatively they could come from the much later Tuakamali phase--then the chance of 50% Admiralties and 50% Banks obsidian reaching Tikopia is approximately 1 in 18,300,000,000 (0.00000005454%). Chance does not therefore seem to be a likely explanation for the presence of Admiralties obsidian on Tikopia.

The temporal model is the simpler of the two reasonable possibilities. It requires only a single homogeneous ancestral community rather than two separate networks, for which there is no concrete evidence other than the obsidian itself. Differences in source region would thus reflect differences in the chronology of occupation. By the time Tikopia was settled, towards the end of the Middle Lapita phase, the convenience of obtaining obsidian from a much nearer source region in the Banks Islands outweighed the need to replicate ancestral Lapita society by maintaining homeland links. The few pieces of Admiralties obsidian on Tikopia would thus reflect the final, much-attenuated phase of long-distance links.


The reanalysis of the Tikopia obsidians has shown the source region identification to be more interesting than first appeared in 1982 when Talasea in New Britain was assumed to be the Bismarcks source represented there. The favoured conclusion suggests that differences between the pattern in the Reefs-Santa Cruz and Tikopia are related to temporal difference. This has wider implications for the interpretation of source region distributions in areas such as Buka and elsewhere (cf. Summerhayes 2004). The region in which significant quantities of Bismarcks obsidian occurred in Lapita times has recently been extended further south beyond the Reefs-Santa Cruz down to central Vanuatu (Bedford et al. 2006; Galipaud and Swete Kelly 2007). Further testing of obsidian source distribution patterns in Vanuatu should illuminate better the temporal implications suggested above and what they can tell us about communication networks of the time.


Roger Bird (1927-2001) passed away in November 2001. He is much missed.

The authors are grateful to Patrick Kirch for providing the original UH Archaeology Laboratory data. The samples for PIXE/PIGME analysis came from the UH Archaeology Laboratory, and the Department of Anthropology, UH Manoa is acknowledged for providing access to them. Michael Hanslip and Geoff Deeble are thanked for the density measurements carried out at ANU. Analysis of these obsidians was originally undertaken as part of the ANU-National Geographic Society Lapita Homeland Project. Una Vidarsdottir of Durham University kindly calculated probabilities for us. Glenn Summerhayes of Otago University provided helpful comments on an earlier draft.


Allen, J. and Gosden, C. (eds) 1991. Report of the Lapita Homeland Project. Occasional Papers in Prehistory 20. Canberra: Department of Prehistory, Research School of Pacific Studies.

Ambrose, W.R.n.d. A Note on Obsidian Density Measurement for Separating Melanesian Obsidian Sources. Manuscript, Department of Prehistory, RSPacS, Australian National University.

Ambrose, W., Duerden, P. and Bird, J.R. 1981. An archaeological application of PIXE-PIGME analysis to Admiralty Islands obsidians, Nuclear Instruments and Methods 191:397-402.

Ambrose, W.R. and Stevenson C.M. 2004. Obsidian density, connate water and hydration dating. Mediterranean Archaeology and Archaeometry, 4 (2):5-16.

Bedford, S., M. Spriggs and Regenvanu, R. 2006. The Teouma Lapita Site and the Early Human Settlement of the Pacific Islands. Antiquity 80:812-828.

Bird, J.R., W.R. Ambrose, L.H. Russell and Scott, M.D. 1981. The Characterisation of Melanesian Obsidian Sources and Artefacts Using the Proton Induced Gamma-Ray Emission (PIGME) Technique. AAEC/E510. Lucas Heights: Australian Atomic Energy Commission Research Establishment.

Dickinson, W.R. 2006. Temper Sands in Prehistoric Oceanian Pottery: Geotectonics, Sedimentology, Petrography, Provenance. Geological Society of America Special Paper 406. Boulder, Colorado.

Duerden, P., D.D. Cohen, E. Clayton, J.R. Bird, W.R. Ambrose, and Leach, B.F. 1979. Elemental analysis of thick obsidian samples by Proton Induced X-ray Emission Spectrometry, Analytical Chemistry 51:2350-4.

Galipaud, J.C. and Swete-Kelly, M.C. 2007. New Evidence relating to the Transport of Talasea Obsidian to Vanuatu. In S. Bedford, C. Sand and S. Connaughton (eds) Oceanic Explorations: Lapita and Western Pacific Settlement, pp. 151-162. Canberra: ANU E-Press.

Green, R.C. 1987. Obsidian Results from the Lapita sites of the Reef/Santa Cruz Islands. In W.R. Ambrose and J.M.J. Mummery (eds) Archaeometry: Further Australasian Studies, pp. 239-249. Canberra: Department of Prehistory, RSPacS, ANU and Australian National Gallery.

Green, R.C. 1991. A Reappraisal of the Dating for some Lapita Sites in the Reef/Santa Cruz Group of the Southeast Solomons. Journal of the Polynesian Society 100:197-207.

Green, R.C. and Bird, J.R. 1989. Fergusson Island Obsidian from the D'Entrecasteaux Group in a Lapita Site of the Reef Santa Cruz Group. New Zealand Journal of Archaeology 11:87-99.

Green, R.C., Jones, M. and Sheppard, P. 2008. The Reconstructed Environment and Absolute Dating of SE-SZ-8 Lapita site on Nendo, Santa Cruz, Solomon Islands. Archaeology in Oceania 43:49-61.

Hanslip, M. 2001. Expedient Technologies? Obsidian Artefacts in Island Melanesia. Unpublished PhD Thesis, The Australian National University, Canberra.

Jones, M.D., F.J. Petchey, R.C. Green, P.J. Sheppard and M. Phelan 2007. The Marine AR for Nenumbo (Solomon Islands): a case study in calculating reservoir offsets from paired sample data. Radiocarbon 49(1):95-102.

Kirch, P.V. 1986. Exchange Systems and Inter-island contact in the Transformation of an Island Society: the Tikopia Case. In P.V.

Kirch (ed.) Island Societies: Archaeological Approaches to Evolution and Transformation, pp. 33-41. Cambridge University Press, Cambridge.

Kirch, P.V. and D.E. Yen 1982. Tikopia: The Prehistory and Ecology of a Polynesian Outlier. Bernice P. Bishop Museum Bulletin 238. Bishop Museum Press, Honolulu.

Reimer, P.J., M.G.L. Baillie, E. Bard, A. Bayliss, J.W. Beck, C.J.H. Bertrand, P.G. Blackwell, C.E. Buck, G.S. Burr, K.B. Cutler, P.E. Damon, R.L. Edwards, R.G. Fairbanks, M. Friedrich, T.E Guilderson, A.G. Hogg, K.A. Hughen, B. Kromer, F.G. McCormac, S.W. Manning, C.B. Ramsey, R.W. Reimer, S. Remmele, J.R. Southon, M. Stuiver, S. Talamo, F.W. Taylor, J. van der Plicht, and C.E. Weyhenmeyer, 2004. IntCal04 Terrestrial radiocarbon age calibration, 26-0 ka BP. Radiocarbon 46, 1029-1058.

Sheppard, P.J. 1992. A Report on the Flaked Lithic Assemblages from Three Southeast Solomons Lapita Sites (SE-SZ-8, SE-RF-2 and SE-RF-6). In J.C. Galipaud (ed.) Poterie Lapita et Peuplement: Actes du Colloque Lapita, Noumea, NouvelleCaledonie, Janvier 1992, pp. 145-153. Noumea: ORSTOM.

Specht, J. 2002. Obsidian, Colonising and Exchange. In S. Bedford, C. Sand and D. Burley (eds) Fifty Years in the Field. Essays in Honour and Celebration of Richard Shutler Jr's Archaeological Career, pp. 37-49. New Zealand Archaeological Association Monograph 25. Auckland: NZAA.

Summerhayes, G.R. 2003. Modelling Differences between Lapita Obsidian and Pottery Distribution Patterns in the Bismarck Archipelago, Papua New Guinea. In C. Sand (ed.) Pacific Archaeology: Assessments and Prospects, pp. 135-145. Les Cahiers de l'Archeologie en Nouvelle-Caledonie 15. Noumea: Departement Archdologie, Service des Musees et du Patrimoine de Nouvelle-Caledonie.

Summerhayes, G.R. 2004. The Nature of Prehistoric Obsidian Importation to Anir and the Development of a 3,000 year old Regional Picture of Obsidian Exchange within the Bismarck Archipelago, Papua New Guinea. In V. Attenbrow and R. Fullagar (eds) A Pacific Odyssey: Archaeology and Anthropology in the Western Pacific. Papers in Honour of Jim Specht, pp. 145-156. Records of the Australian Museum, Supplement 29. Sydney: Australian Museum.

Summerhayes, G.R., J.R. Bird, R. Fullagar, C. Gosden, J. Specht and R. Torrence 1998. Application of PIXE-PIGME to Archaeological Analysis of Changing Patterns of Obsidian Use in West New Britain, Papua New Guinea. In M.S. Shackley (ed.) Archaeological Obsidian Studies: Method and Theory, pp. 129-158. New York: Plenum Press.

Terrell, J. and R.L. Welsch 1997. Lapita and the Temporal Geography of Prehistory. Antiquity 71:548-572.

Torrence, R. and K.L. Victor 1995. The Relativity of Density. Archaeology in Oceania 30:121-131.

UH 1980. Archaeology Laboratory Standard Procedures. Typescript, 24 January 1980.2 pp.

Wickler, S. 2001. The Prehistory of Buka: A Stepping Stone Island in the Northern Solomons. Terra Australis 16. Canberra: ANH and CAR, Australian National University.


(1.) Data on the contexts for the sample and the full specific gravity figures were not presented in the publication and were kindly provided to us by Patrick Kirch.

(2.) In the description of the Groups (Kirch and Yen 1982:258). the Group III sample is wrongly labelled TK-1 whereas it is actually from TK-4, and the Group IV specimen is wrongly attributed to TK-4 rather than TK1. Correct attributions are made elsewhere in the text and in Figure 104.

(3.) In attributing source here by density no account is taken of the various Fergusson Islands (PNG) sources whose measurements overlap all of the other regional sources. This is because after several hundred PIXE/PIGME results from areas south and east of New Britain only one sample has been identified from a Fergusson source (see Green and Bird 1989), and so any occurrence of material from this source on Tikopia is highly unlikely, as was confirmed by the PIXE/PIGME results from the Island.

MS: School of Archaeology and Anthropology, College of Arts and Social Sciences, Australian National University, Canberra, ACT 0200; The late RB: formerly of the Australian Nuclear Science and Technology Organisation (ANSTO), Lucas Heights, NSW 2234; WA: Department of Archaeology and Natural History, College of Asia-Pacific, Australian National University, Canberra, ACT 0200.
Table 1. Density measurements on obsidians used as
reference values for the Tikopia artefacts.

Source Count         Mean    Std D    Min     Max    Range   Group
Artefacts      157   2.394   0.011   2.361   2.425   0.064       1
Source          50   2.393   0.005   2.378   2.401   0.022       2
Pam Is.         11   2.379   0.005   2.373   2.387   0.014       3

Artefacts       89   2.352   0.007   2.326   2.370   0.045       4
Source          21   2.351   0.004   2.342   2.355   0.013       5

D'Entrecasteaux Is
East Ferg.      15   2.402   0.017   2.380   2.438   0.058       6
Sanaroa         11   2.415   0.009   2.408   2.434   0.026       7
West Ferg.      20   2.372   0.010   2.348   2.381   0.033       8
Fagalulu        10   2.359   0.004   2.348   2.364   0.016       9

Banks Is
Vanua Lava      21   2.414   0.010   2.396   2.444   0.047      10
Gana            13   2.441   0.012   2.419   2.455   0.036      11
Quartz Std      39   2.650   0.001   2.649   2.653   0.004

The single crystal quartz standard is used as a calibration monitor for
comparison with the large variation expected in the volcanic glass
source samples collected by scientists and the even wider density
range of the artefacts collected by prehistoric people from unknown
field locations. Figures for the obsidian density values are from
geologically collected Group samples except for artefact Groups 1 and
4. All the items listed have been chemically analyzed by either
PIGME or PIXE at the Lucas Heights facility.

Table 2. Artefact location and phase compared with
PIXE-PIGME results. A space below a site and artefact
number denotes the other part of the same cut flake,
given a separate ANU number.

ANU    Site and         Depth (cm)   Cultural         PIXE/
no.    artefact no.     below        phase            PIGME
                        surface                       source

3418   TK-4-No. 1-5     15           Tuakamali or     Banks (Gaua)
                                     Early Kiki

3419                                                  Banks (VL)

3420   TK-4 S1611-35    0-30         Tuakamali or     Banks (VL)
                                     Early Kiki

3421                                                  Banks (VL)

3422   TK-4 R16111-39   0-30         Tuakamali        Lou (Umrei)
                                     or Early Kiki

3423   TK-4 N17-6       30           Early Kiki       Pam Lin

3424   TK-4 S 1611-42   65           Early Kiki       Lou (Umrei)

3425                                                  ?Lou (Umrei)
                                                      or ?Pam Lin

3426   TK-1 J5-64       0-20         Tuakamali        Banks (VL)

3427                                                  Banks (VL)

3428                                                  Banks (VL)

3429   TK-1 J5-72       0-20         Tuakamali        Banks (VL)

3430                                                  Banks (Gaua)

3431   TK-1 J5-94       20-40        Tuakamali        Banks (VL)

3432   TK-36 A3-4       0-25         Tuakamali or     Banks (VL)
                                     Late Kiki

3433                                                  Banks (VL)

3434   TK-TP-49-25      65           Late Kiki        Banks (VL)

3435                                                  Banks (VL)

3436   TK-1 J5-101      75-95        Late Sinapupu    Banks (VL)

3437                                                  Banks (VL)

3438   TK-36 CI-108     100          Late Kiki        Banks (VL)

3439                                                  Banks (VL)

3440   TK-TP-46-22      195          Early Sinapupu   Banks (VL)

Table 3. PIXE-PIGME elemental data for Tikopia specimens.

ANU    ANSTO      Na%    Al%    Si%     K%     Ca%    Fe%    F
no.    file no.

3418   11347      2.82   7.11   15.40   2.35   0.86   1.60   789
3419   11348      2.86   7.26   18.80   2.75   0.47   1.63   732
3420   11349      2.96   7.33   17.90   2.57   0.60   1.66   660
3421   11350      2.66   6.99   18.50   2.69   0.50   1.66   651
3422   11351      3.57   6.88   29.70   3.27   1.00   2.02   1320
3423   11353      3.01   6.41   37.50   4.33   0.62   1.58   1410
3424   11354      3.24   6.93   29.70   3.45   0.93   1.93   1360
3425   11352      2.60   6.49   27.20   3.11   0.44   1.34   1420
3426   11356      2.69   7.18   19.60   2.79   0.52   1.57   727
3427   11357      3.05   7.52   18.50   2.63   0.62   1.60   697
3428   11358      2.72   7.28   19.50   2.82   0.50   1.55   746
3429   11359      2.70   7.10   17.70   2.59   0.63   1.62   747
3430   11360      2.78   7.18   17.90   2.58   0.68   1.59   737
3431   11361      2.94   7.37   19.40   2.74   0.51   1.62   708
3432   11673      3.14   7.12   29.20   4.71   0.96   2.62   742
3433   11429      2.85   7.07   18.10   2.64   0.57   1.64   713
3434   11442      2.87   7.18   19.10   2.72   0.51   1.63   739
3435   11433      2.74   6.86   18.10   2.62   0.63   1.60   688
3436   11443      2.96   7.00   20.00   2.74   0.51   1.61   684
3437   11362      2.98   7.13   18.90   2.73   0.53   1.61   713
3438   11444      2.81   7.10   18.30   2.73   0.58   1.55   735
3439   11445      2.92   7.17   16.80   2.43   0.75   1.63   737
3440   11432      2.78   7.01   19.10   2.76   0.55   1.57   739

ANU    Ti     Mn       Zn   Rb    Sr   Y    Zr    Nb
no.           x1.123

3418   1030   628      79   49    83   29   158   0
3419   1100   661      63   68    66   28   170   5
3420   1150   650      55   56    75   23   159   7
3421   1060   676      69   60    55   24   152   0
3422   2030   593      55   153   80   24   418   46
3423   1330   559      62   189   31   34   264   32
3424   2010   578      65   146   73   27   409   51
3425   1280   418      67   115   29   35   183   35
3426   1020   655      53   57    57   23   171   0
3427   1010   647      65   60    83   20   174   0
3428   1030   649      54   54    54   21   166   0
3429   1060   664      66   64    75   19   192   0
3430   1060   645      71   54    80   25   164   0
3431   1040   677      58   65    63   35   170   0
3432   1720   n.d.     88   108   95   41   327   6
3433   991    685      65   76    77   24   181   0
3434   1030   691      54   62    58   23   170   0
3435   1020   621      52   56    71   19   170   8
3436   1050   659      62   61    58   19   168   0
3437   1060   655      63   58    56   32   170   4
3438   1070   626      56   57    59   19   154   7
3439   1090   641      69   52    65   22   146   4
3440   1040   663      58   55    53   23   141   0

Table 4. University of Hawaii and ANU density measurements
(g/[cm.sup.3]) compared with PIXE-PIGME results.

UH no.   UH density   UH revised              ANU no.   (ANU-D)
                      source (Group)                    density

35-1     2.386        Lou (VII)               3418      2.419
                                              3419      2.422

35-2     2.393        Lou (1)                 3420      2.408
                                              3421      2.414

35-3     2.430        Banks: Gaua (II)        3422      2.395

35-4     2.268        Anomalous (III)         3423      2.371

35-5     2.364        ?Talasea (VI)           3424      2.397
                                              3425      2.397

35-6     2.405        Banks: VL (IV)          3426      2.417
                                              3427      2.419
                                              3428      2.415

35-7     2.386        Lou (VII)               3429      2.408
                                              3430      2.400

35-8     2.483        Anomalous (VII)         3431      2.421

35-9     2.337        Anomalous (VI)          3432      Not run
                                              3433      2.414

35-10    2.477        Anomalous (V)           3434      2.418
                                              3435      2.323

35-11    2.403        ?Lou ?Banks: VL (VII)   3436      2.426
                                              3437      2.386

35-12    2.469        Anomalous (V)           3438      2.440
                                              3439      2.437

35-13    2.364        ?Talasea (VI)           3440      2.403

UH no.   (ANU-H)   ANU source         PIXE/PIGME
         density                      source

35-1     2.420     Banks (VL)         Banks (Gaua)
         2.419     Banks (VL)         Banks (VL)

35-2     2.405     Banks (VL)         Banks (VL)
         2.411     Banks (VL)         Banks (VL)

35-3     2.392     Admiralties        Lou (Umrei)

35-4     Not run   Anomalous          Pam Lin

35-5     2.392     Admiralties        Lou (Umrei)
         2.392     Admiralties        ?Lou (Umrei) or ?Pam Lin

35-6     2.415     Banks (VL)         Banks (VL)
         2.415     Banks (VL)         Banks (VL)
         2.412     Banks (VL)         Banks (VL)

35-7     2.404     Banks (VL)         Banks (VL)
         2.397     Admiralties        Banks (Gaua)

35-8     2.413     Banks              Banks (VL)

35-9     *         --                 Banks (VL)
         Not run   Banks (VL)         Banks (VL)

35-10    2.417     Banks (VL)         Banks (VL)
         Not run   Anomalous          Banks (VL)

35-11    2.423     ?Banks (VL)        Banks (VL)
         2.417     Admiralties [D],   Banks (VL)
                   Banks (VL)[H]

35-12    2.437     Banks (Gaua)       Banks (VL)
         2.434     Banks (Gaua)       Banks (VL)

35-13    2.400     ?Admiralties       Banks (VL)

? before UH and ANU sources denotes density readings lying close to
but just outside the ranges reported for the particular sources. Lou
Island is the most commonly used Admiralties source. * Although
considered by Deeble [D] to be too small for density measurement,
Hanslip [H] attempted a measurement which came out as 2.391 (within
the Admiralties range), but this was considered unreliable given
specimen size. A space below a UH number denotes the other part of
the same cut flake, given a separate ANU number. VL refers to the
Vanua Lava source within the Banks Islands.
COPYRIGHT 2010 Blackwell Publishing Limited, a company of John Wiley & Sons, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2010 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Spriggs, Matthew; Bird, Roger; Ambrose, Wal
Publication:Archaeology in Oceania
Article Type:Report
Geographic Code:80OCE
Date:Apr 1, 2010
Previous Article:Pacific obsidian sourcing by portable XRF.
Next Article:Radiocarbon and linguistic dates for occupation of the South Wellesley Islands, Northern Australia.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters