Volume 95 1986 > Volume 95, No. 4 > Unea: an obsidian non-source in Papua New Guinea, by W. R. Ambrose and R. W. Johnson, p 491-498
                                                                                             Previous | Next   

- 491

The Quaternary volcanic island of Unea in the Witu group (Fig. 1), 65km north-west of Willaumez Peninsula in New Britian, Papua New Guinea, has been identified by Blythe (1984) as a source of obsidian used for trade or exchange in prehistoric times. Papua New Guinea has three areas of major rhyolitic Quaternary volcanism where glassy obsidians can be found (Smith and Johnson 1981), including Talasea on Willaumez Peninsula, so recognition of a fourth obsidian source in the Witu Islands is of both archaeological and geological interest.

FIGURE 1. Part of northern Papua New Guinea, showing places referred to in the text.
- 492

Availability of obsidian to prehistoric users is a matter that requires detailed examination by archaeologists and geologists, particularly as some obsidian flakes in Melanesian archaeological collections have no currently known source. Rhyolitic volcanism, erosion, and sedimentation are continuing geological processes in Papua New Guinea, meaning that prehistoric people probably had impermanent access to some sources through time: new sources may have appeared, old ones may have become obscured, preventing prehistoric use, or may not be accessible today for comparison with obsidian artefacts. Take, for example, Admiralty Islands obsidian: collections from dated sites, and specimens of the known sources, were analysed chemically (Duerden et al. 1980) and the entire collection subgrouped by cluster analysis; known sources provided an internal crosscheck on the validity of the derived clusters. This procedure pointed to the presence of unlocated sources for the archaeological collections of obsidian artefacts, and these appear to have a chronological as well as a chemical basis (Ambrose and Duerden 1982). Leach and Manly (1982:80), however, took the view that because of the complexity of the multivariate statistical problems inherent in sourcing studies, the question of unlocated sources should be approached by working first from the known chemically definable sources and only secondly by addressing the problem of defining the characteristic chemistry of an unlocated source. There is also the nagging possibility that chemical analysis may not adequately discriminate between geographically separate sources.

Whatever chemical and statistical approach is adopted, there can be no substitute for locating and sampling sources in the field, and the search for additional sources therefore continues. Specht (1981) recently reported important occurrences of artefact-quality obsidian from West New Britain which need to be chemically characterised and included in the regional inventory (see also Specht and Hollis 1982). An additional source in the Manus province was located in 1981 by I. E. M. Smith, J. Kennedy, and Ambrose at the western end of Manus Island near Likum village (W. R. Ambrose, unpublished data). This source still has to be chemically analysed so that it too can be included in the suite of Melanesian sources. The report by Blythe (1984) of another source is of special interest in the contexts of improvement in the chemical analysis of the known obsidian sources, of a better knowledge of their geographic location, and of a refinement in the geological definition of their likely occurrence.

The reported obsidian on Unea also has implications for the known geology of the islands in the Witu group (Johnson and Blake 1972; Johnson and Arculus 1978). Unea is one of the two largest islands in the group. It consists of the remnants of a caldera about 6km in diameter, and has three principal post-caldera volcanoes — namely (in order of decreasing height), Kumbu (about 600m above sea level), Tamogone, and Kumburi (Fig. 2). No summit craters are present on any of these three youthful cones, and rock exposure is, in general, poor on the lower flanks. However, there are excellent cliff-forming rock outcrops tens of metres high on the higher parts of Kumbu and Tamogone volcanoes close to the summits. These outcrops are of massive, crudely columnar jointed, andesite lava which in places has broken away from the cliffs in slabs, columns and blocks.

- 493
FIGURE 2. Unea Island, showing the three main post-caldera volcanoes and other localities referred to. Stippling represents the older, pre-caldera parts of Unea. Adapted from the Royal Australian Survey Corps map (Bali, Sheet 8788, Papua New Guinea).

Dilated joints are quite common in the cliffs, and large detached rocks have tumbled down on to the flatter slopes of the volcanoes. The lava of the upper parts of both Kumbu and Tamogone was evidently extruded in a pasty or viscous form, forming bulbous coulées or cumolodomes rather than mobile flows that moved stream-like for long distances downslope. The sluggish nature of the extrusions evidently favoured the development of prominent joints in the thick, stagnant masses of lava.

Fifty-one rock samples from Unea Island were studied petrographically (Johnson and Blake 1972:74) and nine of these were chemically analysed for both major and trace elements. The analysed rocks are all basalt or andesite, ranging in SiO2 content from 50.4 (in a basalt from an offshore islet) to 60.1 weight percent in a sample from Mount Tamogone (Johnson and Arculus 1978). Kumbu andesite contains 58.0 percent SiO2. Dacite and rhyolite are apparently absent - 494 from Unea, and the SiO2 contents of the analysed samples are well below those of lavas capable of producing glassy obsidian. Glassy selvadges (tachylite) are known rarely on some basaltic lava flows, but none have been found on Unea.

Blythe (1984) referred to a single obsidian source in the village of Tamogone on the western slope of Mount Tamogone. The mouth of the obsidian mine was said to be overgrown although still visible in 1976, but no specimens of obsidian were seen by her. This is a curious situation because under normal circumstances the procurement of obsidian by mining operations would be expected to produce a certain amount of waste material, either as discarded, low-grade obsidian, or debitage from trimming the raw material preliminary to artefact production, or rejected flakes and implements from the production of artefacts.

Blythe also reported traditional information on the ownership of the mine which, as well as producing obsidian, is said to have yielded three mortars and a slit gong. The use of the mine is said to have lasted for at least seven generations before the present grandparental generation, and obsidian from this long-term exploitation was circulated throughout Unea. Some obsidian was exported to other islands of the Witu group in exchange for pigs. Blythe summed up the situation by stating (page 202) that

the obsidian mine on Uneapa [Unea] was really of only local significance. The amounts that it produced were small, and little left the island. Nevertheless, its existence does illustrate that small local sources might exist and be utilised, even in areas where obsidian from Talasea is known to have circulated.

However, the existence of mining as noted above, seems to us to be dubious because of the absence of debris from this reputed long-term extraction of obsidian. Nevertheless, the claims made by Blythe for the distribution and history of exploitation of the obsidian were sufficiently detailed to prompt us to recover material for geological and chemical study, and to reconsider the geological prognostication that Unea is unlikely to be an obsidian source area.

We visited Unea in early September 1985 and contacted the traditional owner of the reported obsidian mine, Gabriel Gegilo. Gegilo was born around 1900 and was baptised in 1942, five years after the Sacred Heart mission was established on the island. Gegilo, despite his remarkable age and frailty, was adamant that the hole from which the obsidian came was near the summit of Mount Kumbu at Bibim, and not at Tamogone village as reported by Blythe. We were guided by a group of Unea men up a ridge on the north-eastern flank of Mount Kumbu along the route of a water pipeline that serves the mission and villages. Occasional flakes of obsidian and pottery sherds present on the lower slopes appeared to diminish as we climbed the track. We finally reached the spring supplying the water and the nearby mine that reportedly produced the obsidian.

The “mine” is one of the vertical clefts in the andesite cliffs that fracture into large columnar and tabular blocks. The andesite lava is more than 10m thick near - 495 the Mount Kumbu summit at Bibim, and is extensively jointed vertically and horizontally, evidently providing the aquifer conditions necessary for the spring. One of the larger vertical fissures near the spring is about 4m high narrowing inwards to a cleft about 30cm wide and diminishing in height to around 40cm. The floor of this narrow defile is flat and dry. No obsidian was observed. Our informants reported that the passage once extended into the mountain where it opened out to a cave in which the obsidian could be found. This was in the old times when correct ritual could cause the crevice to open, but unfortunately no one was able to produce the desired result nowadays. This site is of further importance because of the fresh water springs, and is marked by the faint weathered remains of rock carving on vertical rock faces around the spring.

The difference in opinion about the location of the alleged obsidian source — whether Mount Kumbu or Tamogone village — was complicated further when we were informed also of the presence of caves on Mount Tamogone. The reported owner of the Tamogone caves site was Bito, who died early in 1985. The geology of the area, called Nganda, is similar to that of Bibim or Kumbu: large tabular and columnar blocks have broken away from the cliffs forming clefts and dilated joints. No obsidian was observed anywhere in the vicinity of the Tamogone cliffs which, as on Kumbu, are made up of andesite — again precluding the likelihood of glassy obsidian being present.

The roads on the lower slopes of the volcanoes and to the coast at the mission are laid on weathered red ashes and lack any evidence in exposed section of pumiceous or other felsic tephras that might be expected to accompany the occurrence of rhyolite and obsidian. A few flakes of obsidian from garden areas and roads on the lower slopes of Unea were collected randomly and will be subjected to chemical analysis. Obsidian flakes in low concentration are ubiquitous in similar situations throughout the Bismarck Archipelago; their presence does not necessarily indicate that an obsidian source is nearby.

Nineteen flakes (2704–2722) were examined in handspecimen. Eighteen are glassy and translucent in thin edges and are probably from a rhyolitic source, and one (2719) is opaque and probably from a more mafic source. The 18 glassy flakes are all artefacts having typical percussion bulb flake scars on multiple faces. The rhyolitic glasses are roughly distinguishable from basaltic selvadge glasses on the basis of their density. The density of basaltic glasses from Melanesian sources range above 2.410 g/cc (Ambrose 1976), whereas the 18 Unea volcanic glasses have a mean density of 2.342 ± .014 which is within the density range of the Talasea obsidians (2.350 ± .008 g/cc).

The one opaque glass specimen (2719) has been rounded by weathering and may not be an artefact. The density of this opaque glass is 2.498 which is at the extreme edge of the range of any of the basaltic glasses that have been analysed from Melanesian sources. For instance, the Banks Islands basaltic selvadge glasses have a mean density of 2.445 ± .042 (Ambrose 1976:371). Specimen 2719, in common with the selvadge glasses, has a very poor conchoidal fracture, and is unlikely to have been used for a cutting implement when the more vitreous rhyolite sources were readily available from Talasea.

- 496

The conclusion we draw is that Unea Island provides no evidence for obsidian quarrying. Neither is it a likely source for obsidian given the range of known rock compositions.

Unea is nevertheless an archaeologically fascinating island in that it appears to have had a distinctive prehistory. The artefacts of an early occupation are to be seen in the finely carved stone sculptures on the lower slopes of Mount Kumbu (Riebe 1967), in the widespread use of heavy rock slabs mounted on other standing stones as seats or flat tables, and in the presence of larger standing structures and arrangements formed from tabular and columnar rocks. Three lateral ditches on the ridge rising to Kumbu from Tamogone village are said to have been boundaries between hostile neighbours. The lowermost ditch is about 5m between upcast crests. It is now reduced to a shallow ditch by siltation but is said to have once been deeper than a man's height, and this is confirmed by present surface features. This ditch was evidently a substantial feature, extending to stream gullies on either side of the ridge. Further up the ridge are two tabular blocks about 1.5 X 0.7m resting on smaller stones and forming tables said to have been used for butchering people for ritual purposes. There are other erected blocks of andesite on the ridge. The track down the eastern slope of Mount Kumbu to Mount Tamogone follows the saddle between the two mountains, and at about the centre of the saddle are groups of large columns and slabs of andesite arranged in a roughly circular fashion about 12m in diameter. One rock at the centre, about 2m high, has been raised upright, as is another on the perimeter to the east. The prolific prehistoric surface remains on Unea warrant further archaeological investigation.


We are indebted to Gabriel Gegilo of Tamogone for assistance in locating the alleged “mine” which was reported as an obsidian source on Unea. We are also grateful for guidance on Unea from Anton Vabura, Pengeti Kamir, George Makua, Vincent Ruku, Leo Koroi, Michael Uva, Bruno Kataka, Vincent Buo, John Lomboi, and Bernard Darau. In particular, we should like to thank Johnny Banzak from the Government District Office, Nigilani, for introducing us at short notice to Gegilo and others, and Father Wilhelm Woestheinrich MSC for his hospitality at the mission. R.W.J. publishes with permission of the Director of the Bureau of Mineral Resources.

  • Ambrose, W. R., 1976. Obsidian and its Prehistoric Distribution in Melanesia, in N. Barnard (ed.), Ancient Chinese bronzes and Southeast Asian metal and other archaeological artefacts. Melbourne, National Gallery of Victoria. pp.351–78.
  • ——, J. R. Bird, and P. Duerden, 1981. The Impermanence of Obsidian Sources in Melanesia, in F. Leach and J. Davidson (eds), Archaeological studies of Pacific stone resources. British Archaeological Reports International Series 104. pp.1–19.
  • —— and P. Duerden, 1982. PIXE Analysis in the Distribution and Chronology of Obsidian Use in the Admiralty Islands, in W. R. Ambrose and
- 497
  • P. Duerden (eds), Archaeometry: an Australasian perspective. Canberra, Australian National University.
  • Bird, J. R., W. R. Ambrose, L. H. Russell, and M. D. Scott, 1981. The Characterisation of Melanesian Sources and Artefacts using the Proton Induced Gamma-ray Emission (PIGME) Technique. Australian Atomic Energy Commission (AAEC/E510).
  • Blythe, J., 1984. An Obsidian Source in the Witu Islands, West New Britain. Journal of the Polynesian Society 93:199–203.
  • Duerden, P., J. R. Bird, M. D. Scott, E. Clayton, L. H. Russell, and D. Cohen, 1980. PIXE-PIGME Studies of Artefacts. Nuclear Instruments and Methods 168:447–52.
  • Johnson, R. W., and D. H. Blake, 1972. The Cape Hoskins Areas, Southern Willaumez Peninsula, the Witu Islands, and Associated Volcanic Centres, New Britain: Volcanic Geology and Petrology. Bureau of Mineral Resources, Australia — Record 1982/133 (unpublished).
  • —— and R. J. Arculus, 1978. Volcanic Rocks of the the Witu Islands, Papua New Guinea: The Origin of Magmas above the Deepest Part of the New Britain Benioff Zone. Bulletin Volcanologique 41:609–55.
  • Leach, F., and B. Manly, 1982. Minimum Mahalanobis Distance Functions and Lithic Source Characterisation by Multi-element Analysis. New Zealand Journal of Archaeology 4:77–109.
  • Riebe, I., 1967. Anthropomorphic Stone Carvings on Unea Island. Journal of the Polynesian Society 76:374–8.
  • Smith, I. E. M., and R. W. Johnson, 1981. Contrasting Rhyolite Suites in the Late Cenozoic of Papua New Guinea. Journal of Geophysical Research 86:10257–72.
  • Specht, J., 1981. Obsidian Sources at Talasea, West New Britain, Papua New Guinea. Journal of the Polynesian Society 90:337–56.
  • —— and J. D. Hollis, 1982. A New Obsidian Source in West New Britain. Mankind 16:4124–7.

- 498 Page is blank