Dating the Appearance of Lapita Pottery in the Bismarck Archipelago and Its Dispersal to Remote Oceania

Article excerpt

Abstract

The Bayesian calibration program OxCal v.4.1.5 is applied to two chronological datasets for early Lapita derived from two pcomprehensive reviews. The two datasets are supplemented by published ages for early Lapita sites in two key island groups within Remote Oceania: Vanuatu and Fiji. The analyses provide statistically robust chronologies for the emergence of Lapita on Mussau at 3470-3250 cal BP and in the rest of the Bismarck Archipelago at 3360-3240 cal BP. After a period of 130-290 years, Lapita dispersed to Vanuatu by 3250-3100 cal BP and to Fiji by 3130-3010 cal BP.

Keywords: Lapita pottery, Bayesian calibration, Bismarck Archipelago, Remote Oceania, dispersal

The appearance of Lapita pottery, primarily dentatestamped, in the Bismarck Archipelago is a major event in Pacific history (e.g. Kirch 1997), yet it is relatively poorly dated (Specht 2007). Lapita pottery is generally considered to be derived from red-slipped pottery in Island Southeast Asia (Bellwood 1997; Kirch 1997), with dentate-stamped decorative innovations emerging in the Bismarck Archipelago. In this paper, the focus is upon the chronology of early or formative Lapita pottery in the Bismarck Archipelago and the timing of its dispersal to the islands of Remote Oceania, where this dispersal represented the first human colonization. Several key debates associated with Lapita pottery, principally those focused on cultural associations, geographical and chronological variations, social practices and ultimate demise, are not considered here (Green 1979, 1991a; Anson 1986; Kirch 1997; Spriggs 1997; Summerhayes 2000).

Surprisingly, given the importance of Lapita pottery to Pacific archaeology and the ways in which archaeologists have been 'ensnared' by radiocarbon dating in the region (Bedford and Sand 2007), there has been no systematic attempt to derive an explicitly chronological model for its appearance in the New Guinea region and subsequent dispersal into Remote Oceania. Two exceptions are the application of Bayesian approaches 1) at the Nanggu site in the southeast Solomons (Green et al. 2008), and 2) to dates on human bone from a range of Lapita pottery contexts (Petchey et al. 2011). Despite the lack of precision in regional syntheses of radiocarbon dates, given that they have been derived from ad hoc interpretative approaches rather than Bayesian modelling, various conclusions regarding the nature of the dispersal of Lapita pottery have been drawn, especially a fast dispersal rate that implies a structured process (Kirch and Hunt 1988; Anderson 2001).

There have been various reviews of the radiocarbon dates for early Lapita sites, especially within the Bismarck Archipelago (Kirch and Hunt 1988; Specht and Gosden 1997; Kirch 2001; Summerhayes 2001,2010; Spriggs 2003; Specht 2007). The resultant dates proposed for the appearance of Lapita pottery, however, are impressionistic assessments based largely on the range of individual age determinations through an effective 'eye-bailing' of calibrated age ranges, e.g. 3300-3200 cal BP (Specht and Gosden 1997: 189; Spriggs 2001: 240; Summerhayes 2010), 3350-3250 cal BP or ca. 3550 cal BP 'at the earliest' (Kirch 2001: 219), and 3450-3350 cal BP (Specht 2007: 54). Overlaps between date ranges have enabled only tendencies and approximate chronologies to be ascertained. Other proposed dates for the emergence of Lapita have relied heavily upon the results from individual sites or island groups (Kirch 2001; Summerhayes 2001). Although Kirch's (2001) application of summed probability distributions is an attempt to interpret data in summary form, it is not a model-based approach that includes statistical interrogation of sets of geographically and historically restricted radiocarbon age determinations.

Bayesian statistical programs are model-based applications for the analysis of radiocarbon age determinations that are being increasingly applied to archaeological problems in order to generate higher resolution and statistically valid calibrated date ranges. …