Abstract 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. 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. Archaeol. Oceania 45 (2010) 31–38 31 A reanalysis of the Tikopia obsidians MAtthEW SPRIGGS, ROGER BIRD and WAl AMBROSE Keywords: obsidian, tikopia, Banks Islands, density, PIXE-PIGME 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.