Аннотация:Abstract This paper argues that obsidian movement in the Bismarck Archipelago at the start of the Lapita period shows continuity with earlier patterns and that changes in sources used occur later than the introduction of ceramics. We base our argument on the analysis of obsidian from two ceramic sites in the Duke of York Islands. The sites contain very differently decorated pottery, but both radiocarbon and obsidian hydration dating show they are very close in time, dating to around 3000 years ago. Density and PIXE‐PIGME analysis show that nearly all the obsidian from the putatively older site (SEE), containing ‘classic’ Lapita, came from West New Britain, mostly the Kutau/Bao sources. The possibly slightly later site (SDP), with very thin walled pottery decorated only by rim notching, was initially supplied exclusively from the Umrei source in the Admiralties, with a subsequent reversion to West New Britain. The first two stages in this history appear to be repeated in the similarly dated sequence from EKQ on Mussau Is. Pre‐Lapita data from New Ireland, Nissan and the Papua New Guinea mainland show an extensive distribution of obsidian, exclusively using West New Britain sources, while in the same period Admiralty obsidian has not been found beyond the Admiralties. Thus the use of West New Britain sources in the probably oldest Lapita levels in the Duke of Yorks and Mussau suggests continuity in obsidian distribution with the preceding period. Some other evidence of continuity is noted.
