Australia: The Land Where Time Began

A biography of the Australian continent 

The Archaeology of Sahul or Greater Australia - The lowlands of Melanesia

Near Fortification Point on the Huon Peninsula, on the northern coast of PNG, waisted axes were found in situ that were between layers of volcanic ash on the uplifted coral terraces, for which Groube obtained TL dates of about 40,000 BP (Groube et al., 1986). There are several buried axes and many stone axes on the surface, broken and complete. Groube mainly based his functional analysis on the 70 stone axes that were found in this site. At Bobongara Point, the Peninsula is comprised of 7 raised coral terraces, the upper terrace being about 400 m above sea level at the present. The coral terraces, some up to hundreds of metres wide, were formed when glacioeustatic sea level rises overtook land, that was also being uplifted by tectonic activity. The unnamed site that has been excavated, adjacent to Jo's Creek, an ephemeral stream. It is surrounded by surface finds of flaked stone artefacts, and more than 100 waisted and grooved flaked axes, often more than 20 cm long, and often more than 1 kg, over a more extensive area (Allen & O'Connell, Source 3).

An archaeological focus has been provided for the Pleistocene by these artefacts that are relatively specialised, though one that remains enigmatic. In the Highlands of Melanesia they have been found dating from 40,000-6,000 BP, while in the lowlands and the islands they are surface finds that are undated, as in the Solomon Islands (Groube, 1986: 172). The finding of such artefacts near Mackay, north Queensland (McCarthy, 1949; Lampert, 1983) in rainforest/open forest locations has been suggested to possibly be accommodated as part of a single geographical distribution that includes the Melanesian tools in the north. Their inclusion in the Kartan Culture on Kangaroo Island, at the opposite end of the Greater Australia (Sahul) continent poses some obvious questions.

It has been suggested that waisting was an aid in hafting that might be a significant technological aspect of the archaeological record that occurred on both sides of the Wallace Line (Golson, 1971b: 131-5). Lampert later addressed the question of any relationship between waisted axes in Australia and New Guinea, seeking to extend the comparison to Australia (Lampert, 1983, 145) The results of his multivariate analysis comparing the 2 Australian sets with that from Kosipe indicated that the 3 sites were unrelated to each other, the only common trait being waisting. He suggested that, at least in Australia, the waisted axes resulted from independent invention, though his argument that waisting was a universal method of hafting as support for his argument for independent invention, appears at odd with his argument that waisting was found only at 2 sites about 2000 km from each other in a continent such as Australia that is relatively well known (Lampert, 1983: 151).

Groube came to a diametrically opposite conclusion to that of Lampert as a result of a similar comparison he carried out using the Huon waisted axes and tools of similar shape from Botel Tobago and late Jomon Japan, that are believed to be hoes (Groube, 1986: 169), using a different statistical approach. He concluded that the waisted axes from Australia and New Guinea are distinct from the set from north Asia that were included in the comparison, being part of a single population, and on this basis, suggested that the waisted axes were invented in Greater Australia, no Asian influence being present (Grouble, 1986: 174).

Fragmentary evidence, that is often separated by large distances and often through time, and poor chronological resolution, are examples of the type of problems encountered that are highlighted by these 2 analyses. Allen suggests these analyses lack control over the variabilities within each data set, both of the analyses having been constructed to measure the similarities between the data sets, and not their variability. He asks a number of questions regarding the data sets used, such as the time frames of the collections, the different physical properties of the different raw materials used explain what differences between sets, the different collecting procedures have created what differences between the data sets (he says this point was raised by Groube, 1986: 170), and what variability of use of these implements may have occurred over space and time. Allen also suggests that though the functional explanation proposed by Groube for northern Greater Australia may be extended to cover the waisted axes from Mackay, it would appear to require a lateral shift, at least, of the use of these tools from opening patches in the canopy to allow enough sunlight to the surface to grow food plants, to a need for the clearing of forest for other purposes in the case of Kangaroo Island. The use of these tools for the clearance of forest could explain the distribution pattern that has been found, being supported by observations that waisting is widespread in New Guinea but not Australia (Lampert, 1983: 151). With regard to hafted tools in Australia see Stone Tools

The marine transgression that occurred following the Last Glacial Maximum about 18,000 years ago, submerged most Melanesian coastal sites from the Pleistocene, though some have been preserved as a result of what Allen describes as "idiosyncratic geological events". Some archaeological sites were preserved from inundation by the coastal uplift of about 3 m/1000 years that occurred before and during the human occupation of the Huon Peninsula (Groube, 1986: 171, 1988, 295). Several sites on New Ireland have been found in caves in the limestone terraces that were exposed along much of the east coast by a similar process. It has been suggested that it may have been steep underwater coastal contours that kept the sites close to the coast during the Last Glacial Maximum, at which time the sea level dropped to about 130 m below the present level (Chappell & Shackleton, 1986), and possibly not resulting so much from the uplift. The Kilu Rockshelter on Buka Island in the northern Solomon Islands, is in the same category (Wickler & Spriggs, 1988: 704).

The number of sites known from the Pleistocene in Melanesia had increased by 7 between 1985 and 1998, when Allen's article in Source 2 was written. 2 m of cultural material continue down below a layer with a radiocarbon date of 12,000 BP in Pamwak, a limestone cave (Ambrose, pers.comm. to Allen). Among the animal remains reported from the site are an introduced bandicoot, 1 rat species, bats, reptiles and fish. Canarium nuts have been preserved as macroscopic charcoal in the Pamwak and Kilu sites, and at the Kilu site there were artefacts on which were residues suggesting they had been used in the processing of root vegetables (Wickler, 1990). Among the animal remains found at Kilu were lizards, fish and marine shellfish (Flannery & Wickler, 1990; Wickler,1990: 140-1), and bats, birds and 5 endemic species of rat (Spriggs, Source 2).

On the east coast side of New Ireland there are 5 limestone cave or rockshelter sites, covering about 200 km from Matenbek and Matenkupkum, the most southerly of the sites that are 70 m apart, to Panakiwuk, about 40 km from the northern end of the island, and between these sites at the opposite ends of New Ireland there are 2 other sites, Balof 2, about 50 km southeast of Panakiwuk, and Buang Marabak, 50 km further southeast of Balof 2. Of these sites, all but Buang Marabak have been reasonably reported, according to Allen (Allen et al., 1988; Allen et al., 1989; Marshall & Allen, 1991; Gosden & Robertson, 1991; White et al., 1991). A basal date of 31,990 830 BP (ANU-6614) has been obtained from Buang Marabak, and shell middens are present throughout the deposit (Balean, 1989: 7). Matenkupkum Cave also has shell middens throughout the deposit and basal dates cluster around 33,000-32,000 BP. These 2 sites were the oldest of the Melanesian island sites to be excavated, and Kilu being dated to 29,000 BP. On New Ireland, Matenbek has given 4 dates in the range of 20,000 - 18,000, and Panakiwuk and Balof 2, the 2 northern sites, appear to have been occupied for the first time 15,000-14,000 BP. Allen suggests 3 qualifications are necessary to understand the importance of these dates, accessibility, proximity and location.

Accessibility. There would have been no problem crossing the sea between New Guinea and New Ireland for people who had already crossed wider expanses of ocean to reach Greater Australia, as has been discussed (Irwin, 1991). The same would apply to the crossing from New Ireland to the Solomon Islands, New Ireland remaining in sight after Buka Island can be seen, with the result that the Solomon Islands were occupied soon after the arrival of people in Greater Australia. The crossing to Manus Island is different, in that crossing there is a gap of 60-90 km before Manus Island comes into view, Irwin suggesting it could have delayed the occupation of Manus. Radiocarbon dates taken from charcoal, shell and the seeds of Celtis are 12,000 BP (14,000 BP calibrated) See Link 2.

Proximity. Allen suggests that as Matenbek is only 70 km from Matenkupkum, in this respect they might be seen as 2 foci of 1 site. The dates from Matenbek come from the back of the cave, as a collapsed cave mouth has buried the front of the cave. Based on this, Allen suggests that Matenbek may have been used earlier than is implied by the available dates, as he believes that at Matenkupkum the materials are distributed more towards the front of the cave. He also suggests that Matenbek may have been a subsidiary site to Matenkupkum. A problem is posed by the 2 sites being taken in combination, as the occupation of Matenbek from the Pleistocene partially fills a gap that has been proposed in the sequence between 21,000 BP and 14,000 BP at Matenkupkum. Allen points out that this detracts from the suggested abandonment of the Matenkupkum Site at this time being a result of lowered sea levels. There is difficulty interpreting the relevant dated portion of the stratigraphy at this point as has been discussed (Gosden & Robertson, 1991).

Location. The 2 sites that are the most northerly are also further from the coast. At Balof 2 marine resources are present throughout the sequence. At Panakiwuk marine resources first appear in the sequence about 8,0000 BP when the sea approached its present position. Based on Allen's previous discussion of the sites in the Highlands, the 'inlandness' of Panakiwuk and Balof 2, 4 km and 2 km respectively from the coast, can be seen as only a short distance. The faunal lists for sites such as Yuku and Nombe, as opposed to those from Pamwak and Kilu, indicated, according to Allen, further adaptation as people moved to the Melanesian islands. Green has noted that, excluding extinct animals that were present 40,000 years ago, in Papua New Guinea there are presently 2 species of anteaters (echidnas), 5 wallaby species, as well as a range of bandicoots and phalangers. On the far side of the Vitiaz Strait there is 1 bandicoot, 1 wallaby and 2 phalanger species. Of the species of bird, there are 225 in eastern Papua New Guinea and on West New Britain there are 80. There is also a reduction of plant species across this divide (Spriggs, Source 1). Allen says it is not well known (in 1998 when Source 1 was written)  how the colonisation of the Bismarck Archipelago was affected by the the pauperisation, suggesting that they be seen as points of discussion based on the evidence from a small number of sites, insufficient evidence to construct a definitive prehistory.

A strong dependence on coastal resources in seen in the earliest levels of such sites as Matenkupkum, Matenbek, and Allen assumes, Buang Marabak at 32,000 BP. The marine fish bones that were present in the earliest levels ay Matenkupkum were claimed to be the oldest fishbones in an archaeological site in the world (in 1998), a few bones found not suggesting a specialised technology for catching fish, such as nets, lines, poisons or fish spears, not even deliberate pursuit. The few bones would be accounted for by the accidental or intentional trapping of fish on the outgoing tide on reefs, when the fish were opportunistically speared. The data available did not allow the dating of the beginnings of deliberate pursuit of fish. The site at Balof 2, a younger site from the Pleistocene, contained the best evidence for fishing, fish bones, that have been identified as 5 families of reef fish, being found throughout the deposit. The families found are Acanthuridae, Carangidae, Balistidae, Scaridae and Pomacanthidae. In Balof 2 there were 3 species of small shark, found only in the levels dating from the Holocene, strengthening the evidence for deliberate fishing as these sharks are most often found in the open sea, though they do enter lagoons (White et al., 1991).

According to Allen, echinoderms and shellfish were the most common remains of food found in the earliest levels, suggesting that the search for food was centred on slow moving organisms on the reef. Allen suggests that the strandlooper strategy apparent in New Ireland may be an indication of the food gathering strategy adopted by the earliest colonists in Greater Australia, a strategy that would require the least change in behaviour from their previous home in South East Asia, though if this is the case it raises the question of whether Matenkupkum and Buang Marabak, that have similar earliest dates, are actually sites of earliest colonisation of New Ireland. This bears directly on the question of minimalist explanations. Allen suggests that such a coast, that has familiar climate and resources, could be expected to be colonised early in the colonisation, probably before mid-montane forests, assuming other things were equal. Allen asks the question, was the occupation of this coast significantly delayed by the comparatively pauperate nature of its terrestrial biota in comparison to the northern coastlines of Greater Australia further west?

Allen argues that the initial colonisation of central eastern New Ireland is represented by Matenkupkum, Matenbek and Buang Marabak based on the shell evidence in the earliest levels of the 3 sites (Balean, 1989:33-4), Spriggs strictures in Source 1 not withstanding. Large individuals of a large species of Turbo were prominent for the 10,000 years of the record at Matenkupkum and Matenbek, predominating the earliest levels, indicating a long period of low-level human predation. At Matenkupkum there are clear changes in the subsequent exploitation of shell (Gosden & Robertson, 1991), especially in the period after the Last Glacial Maximum. Between 30,000 BP and 20,000 BP there is an apparent lack of change in the shellfish remains that allows for the possibility that the same low level of predation was occurring for 10,000 years prior to its commencement at Matenkupkum and Buang Marabak. Allen suggests, based on the coincidence in the dates of commencement at these sites strengthens the view that it did not occur earlier.

In the terminal Pleistocene these changes in shell use occur at the same time as other changes in the archaeological record, such as the presence of obsidian originating in the Talasea area of West New Britain, that is found in small amounts, though continuously in levels at Matenbek from 20,000-18,000 BP. The earliest published age of the appearance of obsidian in the cave at Matenkupkum was about 12,000 BP (Allen et al., 1989: 554), though in 1998 this was being reviewed following further excavations there in 1988. Allen suggests that in light of the apparent gap in the record at Matenkupkum between about 21,000 BP and about 14,000 BP, the discrepancy between Matenkupkum and Matenbek appears to be of a stratigraphical nature, so likely to be resolved. Allen makes 3 points about Talasea obsidian distribution in the sites on New Ireland.

  1. The transportation, at least 18,000 years ago, of a useful raw material across a straight line distance of about 350 km
  2. The transportation involved a water crossing from New Britain to New Ireland, the earliest known demonstration of canoe transport in the region that was not accidental-illustrating patterning in another aspect of human behaviour.
  3. Talasea obsidian has not been found in Balof 2 or Panakiwuk, the 2 northern sites, in the Pleistocene, though is present in levels from the Holocene, 8,000-7,000 BP in Balof 2, and probably at the same time in Panakiwuk, signalling a change of some sort.

Obsidian doesn't appear to have reached Manus Island or the Solomon Islands in the Pleistocene. Archaeologists found it easier to accept the transport of stone raw material at an early time than the transport of living animals across biogeographic bundaries. Rattus mordax, found in the earliest levels of all sites, that is believed to be locally extinct, possibly replacing Rattus praetor, a species found in levels from the Holocene in the sites at Panakiwuk and Balof 2, though it not found in Matenkupkum and Matenbek. Another species that appears to have been introduced is Phalanger orientalis, that is found in the earliest layers at Matenbek but not in the sites of Matenkupkum, Panakiwuk or Balof 2. Allen suggests P. orientalis may have appeared earlier in southern than northern New Ireland. In the Holocene layers in the northern sites the thylogale makes its first appearance, Thylogale brunii. Evidence from Balof 2 suggests this was separate from, and earlier than the first appearance of domestic animals such as the pig in New Ireland. In the Pamwak sequence on Manus Island the bandicoot is present.

Allen considers the evidence compelling for the implication that wild animals were transported by humans across water barriers in the terminal Pleistocene and Early Holocene, though conceding there are inconsistencies (Allen et al., 1989: 556). He suggests that the establishment of breeding populations may not necessarily have been the intention of the humans, suggesting it is more probable that the introduction of the animals was a by-product of colonisation. The transportation of useful products over relatively long distances across water by the terminal Pleistocene, based on evidence from obsidian and fauna, increases the probability that useful elements, and possibly whole horticultural systems for producing food were occurring in New Ireland as early as they did in highlands of New Guinea. It has been observed (Groube, 1988: 298) that the manipulation of swamps for food production was occurring at Kuk (Golson, 1998) about 9,000 years ago, as soon as it was permitted by the improvement of the climate at the end of the Pleistocene 'suggests it may have been practised at lower altitudes during the Pleistocene. The small amount of evidence that he considers might support this view (Allen, 1989: 558), though criticised by Spriggs (Source 1). Allen disputes Spriggs' treatment of the data, suggesting Spriggs' explanations are not in any way more parsimonious or compelling.

The lithic assemblages from these sites that have been described (see Freslov, 1989; Allen et al., 1989: 552-4; Marshall & Allen, 1991; White et al., 1991) appear to vary considerably between sites suggesting different raw material resources rather than cultural continuities in regard to their manufacture and use. The record clearly indicates that changes took place that were quite distinctive during the last 20,000 years of the Pleistocene, though data from the islands of Melanesia are fragmentary.

According to Allen, it is possible to see in the data a progression from initial occupation that was of low density that was coastally oriented, to a more intensive, extensive use of the region. Archaeologically, Matenbek, that is 18,000 years old appears different from Matenkupkum, that is 32,000 years old, and Balof 2 that is 8,000 years old, displays a different, more intensive usage than Panakiwuk that is 14,000 years old. There are also hints of even greater difference between the Papua New Guinea highlands and the islands of Melanesia that cannot easily be explained by differences in the environment. An increasing degree of divergence in human behaviours between the islands and the Highlands of Melanesia throughout the Pleistocene may have been dictated by human strategies predicated on sea travel rather than land travel, the result being a broad spectrum and extensive solutions to the problems of acquiring a subsistence living, as well as more specialised, intensive solutions.

See Stone Tools

See Source 1 for more information and illustrations.

Sources & Further reading

  1. Jim Allen, in Murray, Tim, 1998, Archaeology of Aboriginal Australia, Allen & Unwin.
  2. Allen. J & O'Connell, J.F., 2003, The long and the short of it: archaeological approaches to determining when humans first colonised Australia and New Guinea, Australian Archaeology, No. 57, 2003.


  1. Pamwak Rockshelter: A Pleistocene site on Manus Island, Papua New Guinea
  2. Faunal Composition of Pamwak Site, Manus Island, PNG 


Author: M. H. Monroe
Last updated: 30/09/2013



Journey Back Through Time
Experience Australia
Aboriginal Australia
National Parks
Photo Galleries
Site Map
                                                                                           Author: M.H.Monroe  Email:     Sources & Further reading