Our entire knowledge of the Early Period comes from Phases III and IV o f the Pamwak sequence (Chapter 4). Clearly this is a limiting factor in deriving any broad interpretation from the material. Whether or not the evidence revealed at Pamwak is indicative o f a regional pattern awaits the discovery o f further sites o f this age on Manus.
Evidence from Pamwak shows that obsidian began to be used and distributed in the period 13,786-12,627 BP. Prior to this cherts, presumably obtained from local stream deposits, were utilised for unmodified flake tools. The chronological depth o f this chert technology is unknown, but may extend back 10,000 years before the appearance o f obsidian, on the basis o f an age/depth extrapolation o f the Pamwak deposits. The sudden and virtually complete replacement of chert with obsidian in the terminal Pleistocene marks a major shift in lithic resource procurement. For the first time people were utilising stone available outside the local site catchment and moving this material around the landscape.
A strong candidate for the source of the first obsidian transported to Pamwak is the Pam Islands. On available bathymetric and geological evidence it seems unlikely that the larger Pleistocene Manus landmass was connected to land now present as the two small islands constituting the Pams. Watercraft must have been used to access obsidian. Today the Pam Islands are 35 km from Manus but at the time o f lower late Pleistocene sea levels a water crossing o f approximately 20 km would be needed, still a significant distance.
A number o f possibilities can be advanced for the terminal Pleistocene appearance o f obsidian in Pamwak; (1) obsidian carried inland by hunter-gatherer or hunter- horticultural groups (Spriggs 1993) was totally “consumed” before they arrived at inland localities such as Pamwak, (2) obsidian-bearing islands may have simply not existed before the terminal Pleistocene, and (3) the appearance o f obsidian reflects the first development or introduction o f watercraft sufficiently seaworthy to make regular water-crossings between the Pam Islands and Manus. On current knowledge o f rates o f sea level rise and an assumption o f minimal tectonic uplift, at various times during the terminal Pleistocene Pamwak would have been somewhere between 6 km and 10 km from the coast. This represents no more than half a day’s journey inland, which is an insufficient distance to account for the total consumption o f obsidian cores before arrival at the shelter. With regard to the second proposition, there is a possibility that obsidian deposits were formed by volcanic activity at the very end o f the Pleistocene. The geological composition o f Lou, Pam Lin and Pam Mandian indicates a late Pleistocene or early Holocene origin for these islands. However, at least two obsidian “sources” were utilised by the late Pleistocene occupants o f Pamwak; one (Source X) possibly deriving from one o f the Pam Islands and the other from Lou. It seems rather unlikely for compositionally discrete obsidians from different islands to be coincidently formed in the same geological instant, while the period before was marked by a complete absence o f obsidian.
This brings us to the possibility that the appearance o f obsidian is associated with the development or introduction o f watercraft capable o f making regular two-way crossings between Manus and the southern islands o f the archipelago. In connection with this, recognition must be made that at least two episodes o f voyaging to Manus had taken place by the end o f the Pleistocene, one o f which may correspond with the first appearance of obsidian in Pamwak (Fredericksen et al. 1993:151). Note must also be taken o f recent proof o f regular Pleistocene voyaging between New Britain and New Ireland (Summerhayes and Allen 1993). In connection with the issue of maritime capability, it must also be said that the Pleistocene settlement o f Manus did not follow the trend for the rest o f western Island Melanesia o f colonisation of intervisible islands (Irwin 1992:214). Clearly, Pleistocene peoples in this part o f the world were more adept mariners than previously supposed, an adaptation undoubtedly related to a need to navigate along coastlines and between neighbouring islands.
When obsidian appears in the Pamwak sequence it almost entirely replaces locally occurring cherts. At no stage was the flow o f this material sufficiently disrupted to necessitate a reversion to the sole use o f local stone. A significant feature o f obsidian use at Pamwak is that none was obtained from deposits in southwest Manus. There is no geological information on the Manus deposits and their discovery has come about only through archaeological research. We therefore have no direct information on the age o f the deposits. The absence o f southwest Manus obsidian in the Pamwak assemblage may mean this source was formed or became accessible late in the prehistoric sequence. Perhaps this occurred in the latter part o f the Holocene by which time access to obsidian from Lou and the Pam Islands had become formalised within a system of inter-community exchange.
manufactured o f pitchstone. At least some in Pamwak shelter may have been cached, presumably for intended future use. A number were reworked or resharpened. These tools must have formed a component o f the toolkit o f hunter-gatherer or hunter- horticultural groups which frequented the shelter at this time. The presence o f such implements demonstrates that the late Pleistocene occupants o f Manus possessed the capability to prepare cores, remove both equilateral and blade-like flakes, and undertake skilled percussion retouching to manufacture implements o f a fairly uniform shape. Such a technology is unknown from anywhere else in Island Melanesia at this time, although less refined retouched tools have been reported from a Pleistocene context on New Britain (Pavlides 1993). There are morphological parallels with contemporaneous tools from the Malaysian/Island Southeast Asian region (e g. Anderson 1987), but I see no obvious reason to ascribe the appearance o f retouched pitchstone implements to an external source. These probably represent an indigenous innovation. The appearance o f pitchstone tools in the Pamwak sequence may have been associated with a change in the way in which the shelter was used, perhaps involving a greater frequency o f visits.
A second technological change occurs with the appearance o f edge-ground stone tools and Tridacna shell adzes. The presence o f axes in terminal Pleistocene deposits demonstrates forest clearance or other heavy woodworking activity was carried out near the shelter. Spriggs (1993) has, partly on evidence from Pamwak, concluded that the possibility o f Pleistocene agriculture in Island Melanesia should not be dismissed. The presence of edge ground axes at Pamwak could very well be related to gardening activity. Shell adzes make an appearance later in the sequence, between the early and mid Holocene. These implements are more likely to have been used in woodworking than tree felling or other agricultural activity. Their appearance is associated with the deposition o f a mangrove swamp midden and, given this marine
focus at the shelter, the possibility is there that the adzes were used in canoe construction.