The Hoabinhians and Their Island Contemporaries

I will now turn to the archaeologists’ record of the Late Pleistocene and early Holocene periods in the Indo-Malaysian Archipelago. In this time-span there are a number of long-occupied, radiocarbon-dated and stratified sites and all associated human remains are of modern physical type. Prior to about 5000 years ago all industries in the region comprised exclusively flaked and not ground stone tools (although Niah in Sarawak may be an exception), and the rather limited economic evidence at present available suggests a universal economy at hunting and gathering. During the Holocene the stone tool industries became more varied, and after 3000 BC new items, of which pottery and fully ground stones adzes are the most visible, spread through the region. As I shown in Chapter 4 the linguistic evidence clearly attests a slow expansion of Austronesian-speaking agricultural groups during the last five millennia, but as I have stressed before this expansion was not a geographically unified process of replacement. The hunting and gathering lifestyle has been progressively eroded but it has certainly never disappeared entirely, and flaked stone tools continued to be used by both hunting and gathering and agricultural groups until the recent past in some areas. Hence in recent millennia different technologies and economics could and did occur in neighbouring and contemporary sites in a mosaic-like fashion (Hutterer 1976). The archaeological record has to be considered partly in terms of synchronic regional variation, and not totally in terms of pan-archipelagic technological and economic phases.

Before going further, it is unnecessary to give a brief recapitulation of the palaeoenvironmental evidence from archipelago, particularly that which pertains to the last 20,000 years within which the great bulk of the dated sites lie. The most dramatic environmental changes would undoubtedly have been caused, particularly in the Sundaland region, by the rise in sea-level of perhaps 130m between approximately 15000 and 8000 years ago. Apart from drowning an unknown number of coastal archaeological sites, to the obvious detriment of modern prehistorc studies, this change carved the Sundaland continent into the islands which exist today, and in the presumed absence at this time of a developed voyaging technology it would have produced a number of separate pools of human population. Economically the change would have had certain benefits, since it certainly increased the length and environmental variety of coastline, but there might also have been less favourable changes, particularly through the expansion of rainforest with the prevailing warmer and wetter climate.

Although the post-glacial climatic amelioration perhaps had little effect in the core regions of the equatorial rainforest, it would presumably have had more impact on the fringing areas of seasonal climate, where monsoon forest or parkland vegetation may have been more extensive during the last glacial period. An increasing density of vegetation in these areas would have affected hunting populations through a diminution in mammal biomass, which decreases dramatically as one moves from optimal savanna conditions, through parkland, towards rainforest. For instance, modern densities of wild banteng cattle range from about 10-15 animals per ha in Javan savanna grasslands down to only 1-2 animals per ha in rainforests. Rainforest faunas present additional problems in that animals rarely herd together and many species are arboreal, and so are more difficult to hunt. This atomistic pattern also characterises rainforest vegetation, with many species mixed in a mosaic of small numbers of individuals, rather than in large stands. Such patterns tend to promote non-specialised economies and low population densities amongst hunting and gathering populations, and if the archaeological record is taken at face value it suggests strongly that rainforest occupation is very much a Holocene adaptation, perhaps as a result of increasing population pressure in the more favourable but shrinking monsoon forest and parkland regions, and perhaps also owing to the development of new methods of trapping and subsistence.

This may not mean that the equatorial rainforests were entirely unoccupied during the Pleistocene. However, there are some interesting lacunae in the archaeological evidence from these regions, and I suspect that modern Indo-Malaysian hunters and gatherers, who have survived in the more densely forested equatorial parts of the archipelago, are not entirely representative of the bulk of the hunting societies who would have been concentrated more in the regions of seasonal climate in the Late Pleistocene. Perhaps it is no coincidence that most of the flaked stone assemblages come from such seasonally dry regions as central Java, southern Sulawesi, the Lesser Sundas and part of the Philippines. In Borneo there clearly was last glacial occupation (perhaps under drier climatic conditions) in near-coastal caves and around at least one near-coastal lake, but I am inclined to doubt that the densely forested interior was inhabited very much, if at all, prior to Austronesian settlement – there are no physical or cultural traces of earlier peoples, and so far no Pleistocene archaeological assemblages in the deep interior regions of the island (e.g. Harrisson 1970a:21; Hanbury-Tenison 1980). A similar situation applies in Sumatra and Malaya, although the inland forests of Malaya were settled widely by Hoabinhians from the beginning of the Holocene.

I.             Malaya and Mainland South-East Asia: The Hoabinhian

Prior to Austronesian settlement the Malay Peninsula and the adjacent coasts of north-eastern Sumatra belonged culturally to the mainland of South-East Asia, rather than to the islands. From about 10 000 years ago Hoabinhian assemblages appeared throughout this region, and there seems little reason to doubt that in Malaya they were made by populations ancestral to the present Austro-Asiatic-speaking orang asli (Negritos and Senoi, Solheim 1980). These groups ceased to make flaked stone tools long before recorded history, but the Negritos have preserved a hunting and gathering way of life which may be regarded as a modified descendant of the inland Hoabinhian economy.

The term “Hoabinhian” has ben in use since the 1920s to refer to a stone tool industry characterized by distinctive pebble tools. Hoabinhian sites are found all over the mainland of South-East Asia, westwards to Burma and northwards to the southern provinces of China and perhaps Taiwan.so far, all radiocarbon-dated Hoabinhian assemblages fall between 14 000 and 3000 years ago, and it is possible that some Hoabinhian tool manufacture continued into even more recent times in some regions. Unfortunately, the Late Pleistocene ancestry of the Hoabinhian remains obscure, except in northern Vietnam where antecedent pebble tool industry termed the Sonviian has been dated back to about 23 000 bc (Ha Van Tan 1978, 1980, 1985a). another ancestral industry may occur in the undated lower levels of the cave of Sai Yok in western Thailand (Heekeren and Knuth 1967).

The Hoabinhian is therefore very much a terminal Pleistocene and Holocene phenomenon, and while its ultimate ancestry may be vague there seems little reason to doubt that it is indigenously South-East Asian. In Malaya and Sumatra the Hoabinhian does not appear to extend back in time for more than 10 000 years, and the prehistoric record prior to this date is virtually a complete blank. Again the question arises of whether the rainforests of this region were settled before this time, or were people living only along the now drowned coastlines?

Hoabinhian sites are found mostly in rock-shelters, but there are a few coastal shell middens in Sumatra, Malaya and northern Vietnam which seem to belong to the present period of sea-level (that is post 6000 bc). In addition, some inland non-midden open sites have been reported from Sumatra, eastern Malaya and northern Thailand. However, the excavation record is highly skewed towards inland shelters, and coastal middens in Malaya and Sumatra have never been satisfactorily investigated; most have now been destroyed for lime-kilns.

The characteristic tool types of the Hoabinhian are unifacially or bifacially flaked flat river pebbles of an approximate fist size, often with cutting edges all around their peripheries. They come in a variety of shapes from oval through rectangular to triangular, and some occasionally have waisted forms. Bifacially worked tools appear to predominate in some Malayan sites, but unifacial forms, according to the rather inadequate reports available, predominate elsewhere. The industry (or technocomplex, after Gorman 1971) has been excavated most prolifically in the limestone massifs of northern Vietnam, where it is associated with flake tools, bone points and spatula, stone mortars and pounders of various sizes, and flexed burials often dusted with hematite. In Vietnam there is considerable industrial variation within the Hoabinhian time-span, and many sites also have edge-ground tools and pottery in their upper layers. These present some problems in interpretation, and it is possible that many major cases of stratigraphic disturbance have gone unrecorded (Matthews 1965), particularly in the older excavations. However, taking the published record at face value, it must be accepted that egde-ground tools and pottery did appear in late Hoabinhian contexts prior to the appearance of the more formal polished adze and pottery associations which most South-East Asian archaeologists, including myself, term “Neolithic”.

Let me illustrate this in more detail. In Vietnam, a large number of recent radiocarbon dates, mostly from freshwater shells, indicate that the Hoabinhian falls between about 10 000 and 5000 bc (Meacham 1976-8;Bayard 1980; Ha Van Tan 1980). Overlapping with this is a related industry known as the Bacsonian, which is in reality little more than an aspect of the Hoabinhian with an occurrence of edge-ground tools, and this dates from about 9000 bc onwards. In Vietnam edge-grinding is clearly an early Holocene innovation, and I will return to the question of the early distribution of this technique later. Pottery (mostly plain or vine/mat impressed rather than cord-marked) was also widespread in Vietnam by at least 4000 bc, but at present there is no good evidence to suggest how far back it goes into the Hoabinhian-Bacsoninan matrix.

Another major question concerns the role of the Hoabinhian in the development of agriculture in South-East Asia. It should be noted that the Hoabinhian technocomplex covered a vast area, extending virtually from the equator in Sumatra to beyond the Tropic of Cancer in southern China. In the far southern regions I remain fairly convinced that it had no agricultural status, but there can be less certainty for northern Vietnam and northern Thailand. On this question there are still only the results of Gorman’s excavations in Spirit Cave in north-western Thailand (Gorman 1970,1971;Glover 1977b:11-17), where remains of a number of edible but not necessarily cultivated fruits and legumes appeared in terminal Pleistocene Hoabinhian levels. Furthermore, an important Neolithic assemblage of cord-marked pottery, polished adzes, and slate knives was added to the continuing wongsa (in press) have tentatively suggested that cultivation phase of the northern Hoabinhian, and there may be a hint here of a very important stage in South-East Asian prehistory – a stage of initial cereal cultivation which attained its most coherent archaeological expression amongst the Neolithic cultures of southern China, south of the Yangtze River.

A.  Malaya

Having given this contextual introduction to the Hoabinhian, I will now turn to its most southerly expressions in Malaya and Sumatra. In Malaya a number of inland Hoabinhian caves and shelters have been excavated in the many limestone massifs scattered through the northern states of Perlis, Kedah, Perak, Pahang and Kelantan, and coastal shell middens once existed in the north-western states of Pinang and Perak. The majority of sites were excavated during the 1920s and 1930s, and the reports can only be described as brief. However, the very important site of Gua Cha in Kelantan, excavated in 1954 and more recently by a Malaysian National Museum team in 1979, has produced a firm record.

This massive limestone rock-shelter lies in a remote inland region of equatorial rainforest on the bank of the Nenggiri River, a tributary of the Kelantan River which flows into the sea at Kota Bharu. In 1954 three large trenches were excavated in the shelter by Sieveking, who published a detailed report on the contents of the Neolithic layers (Sieveking 1954), but gave only stratigraphical observations on the underlying Hoabinhian. In order to throw light on a number of questions concerning the Hoabinhian the shelter was excavated on a small scale again in 1979, by Adi Taha of the Malaysian National Museum, who at that time was doing postgraduate research under my supervision at the Australian National University (Adi 1981;Bellwood and Adi 1981). I will combine the results of both excavations here.

The Hoabinhian layers at Gua Cha are up to 170 cm thick, and rest on sterile alluvial deposits. According to sediment analyses by Philip Hughes the Hoabinhian deposit itself is also of alluvial origin, and was clearly formed by occasional flooding of the shelter by the neighbouring river. The industry is a surprisingly homogeneous collection of bifacially flaked flat river pebbles, and it has a minority component of cruder pebble tools together with a few utilized and waste flakes, and a number of river pebbles which may have served for crushing and pounding – some have hematite stains. Bone tool were absent, despite their occurrence at other Malayan sites such as Gua Bintong in Perlis. The homogeneity and emphasis on bifacial working of the Gua Cha industry are both quite striking, and radiocarbon dates indicate a commencement soon after 10 000 years ago, and a fairly decisive termination a little before 1000 bc. A number of primary flexed or secondary burials had been placed in the Hoabinhian deposits; none contained certain grave goods, but one flexed young male excavated in 1979 had a stone-slab pillow and a body cover of tufa chunks dusted with hematite, and another unexcavated burial lay beneath two limestone slabs.

The diet and economy of the Gua Cha Hoabinhians was investigated from three angles. Firstly, an examination by Bulbeck (1982) of the occurrence of caries in the teeth of the burials excavated in 1979 suggested considerable consumption of sweet foods such as fruits and honey. Secondly, flotation of the deposits produced a large quantity of charcoal but unfortunately no recognizable plant-food remains, and in this regard it is important to note that a large quantity of carbonised rice was found in an upper layer of the site dated to about 900 years ago, so this cereal would have been detected had it been present earlier. Therefore there is no evidence to suggest cereal cultivation at Gua Cha, either Hoabinhian or Neolithic. Finally, large numbers of animal bones (identified by Groves and Weitzel, see Adi 1981:Appendix) were found throughout the Hoabinhian layers; pigs (Sus scrofa and the bearded pig Sis barbatus) were the most commonly killed animals, and it is possible that large number of pigs were killed during mass river crossings, as described by Hislop (1954) for Malaya, and by St John (1974. I:138) for northern Borneo. Sieveking (1954) found about 25 small heaps of jaws and skull fragments of juvenile pigs in one sector of the Hoabinhian deposits, and young animals seem to have been favoured as prey in other species too. The latter included deer of several species, Malayan bear, monkeys and gibbons, rats, squirrels, flying foxes, and (more rarely) rhinoceros and cattles. This species list is very similar to that found in Hoabinhian sites as a whole (Gorman 1971: Table 2). In one of my earlier papers (Bellwood 1976a:162-3) I suggested very tentatively that the predominance of juvenile pig bones at Gua Cha could suggest some form of domestication, but this statement was made before I was able to work at the site and I now consider an explanation of simple hunting to be more likely. It is also worth noting that small quantities of fesh-water shellfish were found in the site, but marine shells were absent.

Within the South-East Asian Hoabinhian in general, the question of overlap with Neolithic assemblages characterised by pottery and fully-ground stones adzes has always been a particularly vexed one, partly because the necessary stratigraphic details were simply not recorded in the earlier excavations. At Spirit Cave, as noted above, Hoabinhian tools were stated by Gorman (1970) to overlap quite definitely with the Neolithic assemblage of potsherds and stone adzes, and in northern Vietnam there are indications that similar situations prevailed. So in these northern regions it is quite possible that the Hoabinhians did grade slowly into a fairly coherent and presumably agricultural array of Neolithic cultures. However, in Malaya the situation appears to be different. According to Sieveking (1954), the Hoabinhian at Gua Cha was separated by a gapfrom the overlying Neolithic sectioned adzes like those in the top layer of Spirit Cave, and later continues with a series of burials which I will describe in Chapter 8. The 1979 work at Gua Cha tended to support Sieveking, although there can be no doubt that a few Hoabinhian tools do occur in the pottery-bearing upper layer from which the Neolithic burial goods themselves, which do include several fully-ground stone adzes.

It seems, however, that Sieveking may have overemphasised his concept of a gap between Hoabinhian and the Neolithic; a more likely explanation is that rapid cultural change took place in the region of the site, and that this change, according to the skeletal evidence reviewed in Chapter 3, Section HIB, involved no replacement of population. I strongly suspect that the whole of the Gua Cha sequence belongs to ancestral orang asli populations who, as I will document later, were brought fairly rapidly into the Neolithic world of the Thai-Malayan peninsula at about 1000 bc.

Come of the complexity of the overall Malayan situation for the Hoabinhian can be estimated from a brief review of other sites. At the cave of Gua Kechil in Pahang, Dunn (1964, 1966) found Hoabinhian tools together with cord-marked and plain pottery in a lower occupation layer about 40 cm thick. This was overlain by a layer with polished adzes and pottery similar to that of the main Neolithic layer at Gua Cha, dated rather surpirisingly by radio-carbon to the earlier fourth millennium bc. Hoabinhian tools were absent in this upper layer, but the situation beneath suggests an overlap situation of Hoabinhian tools and pottery which can certainly not be recognised in the sequence from Gua Cha.

A less certain case of overlap is illustrated by Peacock (1971) for the unpublished excavations at Kota Tongkat, another shelter in Pahang. The Hobinhian levels here have high quantities of occupation materials such as animal bones and shells, and there does appear to be some slight overlap between the Hoabinhian tools and upper level characterised mainly by pottery and apparently little else. Howeve, I must confess from Peacock’s illustrations that I can see litte real differences in the overall sequences of Gua Cha and Kota Tongkat, and it may in fact be the overlap situation in Gua Kechil which is unusual in the Malayan context – unfortunately the necessary data for a firm assessment are not available.

It may also be recalled that edge-ground tools are very characteristic of later stages of the Hoabinhian (that is the Bacsoninan) in Vietnam, yet such tools are to my knowledge quite absent from the three Malayan sites which I have just described. However, they have been reported from old excavations at Gua Madu in Kelantan (Tweedie 1940), and from Gua Baik (Gol Ba’it, Steun Callenfels and Noone 1940) and Gua Kerbau (stein Sallenfels and Evans 1928) in Perak. At the last two sites they were reported as occurring down to the bases of the deposits, but I am inclined to accept this with considerable reservations, especially since the Gua Kerbau deposit was excavated in 75 cm thick spits and was apparently riddled with animal holes, and Gua Baik produced only one specimen.

The Malayan shell middens now tell rather a sorry tale of destruction, although Adi (1983:53) has recently reported a new discovery at Seberang Perak near Telok Anson in Perak. It has also been known since 1869 that large middens of marine bivalves (formerly published as Meretrix and Arca, but I believe more probably Batissa and Anadara in modern classification) once occurred on old beach ridges in the mainland portion of Pinang state (formerly Province Wellesley). These have now been destroyed apart from small remnants, but the remains of three were excavated long ago by Stein Callenfels at a location about three miles inland called Guar Kepah (Stein Callenfels 1936a, the sites were then called Guak Kepah).

According to Stein Callenfels these middens, whish were originally up to 5 m thick, contained hearths, secondary burials dusted with hematite (one jaw was classified as “Palae-Melanesian” by Mijsberg 1940, pig and estuarine fish bones, Hoabinhian tools, necked and apparently hammer-dressed axes, and small quantities of cord marked and incised pottery. No stratigraphics order for these items was clearly established, but Tweedie (1853:69) thought that the pottery may have post-dated the Hoabinhian tools. It appears that the Hoabinhian and the necked axes (often described as “ground”, but I suspect from illustrations that they were simply hammer-dressed) did belong together with the midden deposits, which presumably date from the present phase of sea-level, and thus somewhere within the Holocene. These sites clearly pose a number of unresolved problems; they appear to represent a coastal Hoabinhian adaptation and thus an aspect of Hoabinhian life which cannot be found in the inland shelters, but which may now be virtually lost as a result of the terminal Pleistocene rise in sea-level and the activities of lime burners.

B.   Sumatra

The only Hoabinhian sites found within modern political boundaries of Indonesia lie inland from the north-eastern coast of Sumatra between Lhokseumawe and Medan (Witkamp 1920; Kupper 1930; Heekeren 1972:85-92; Brandt 1976; Glover 1978b). Many of the sites are large shell middens up to 30 m in diameter and 5 m deep, with interstratified lenses of shells, soil and ash. They appear to be located at approximately present sea-level on an early Holocene strandline which now lies between 10 and 15 km inland, and most have been buried under sediments deposited along this rapidly aggrading coast during the past few millennia. None have been systematically excavated or dated, although a radiocarbon date within the sixth millennium bc has recently been quarried away for lime burning (Bronson, in press). Like the Pinang middens of Malaya, the Sumatran middens must be of Holocene date and may have been occupied at any time between 10 000 and perhaps as recently as 3000 years ago. The region has no caves or shelters, but other non-midden Hoabinhian sites have been reported from inland terraces and flat limestone rises to about 150 m above sea-level.

It appears that virtually all of the shell middens have now been destroyed for lime manufacture, but many archaeological collections have been made from them and these are described by Heekeren (1972). The majority of the tools appear to be unifacially flaked oval or elongated pebbles, often flaked all over one surface. Bifacial tools and edge-ground tools appear to be rather rare, as are retouched flakes, and this industry gives the impression of being technologically simpler than that of the Malayan sites as a whole. Grind-stones, mortars, hematite and human burials also occur in the middens, and faunal remains include elephant, rhinoceros, bear, deer, and presumably many smaller species. The shellfish illustrated by Heekeren (1972: Plate 36) appear to belong to the genera Batissa, Anadara and Telescopium; the same estuarine species which once formed the Pinang middens. Pottery appears to be universally absent, at least in confirmed association with the Hoabinhian deposits.

C.  Further Comments on the Hoabinhian

As reported from Malaya and Sumatra the Hoabinhian seems to have had a coastal and inland hunting and gathering mode of economic orientation, and I feel it is pushing the evidence too far to suggest a local development of agriculture in these regions. As I have stated, the Hoabinhian of the intermediate tropical zone in northern Thailand, northern Vietnam and southern China may hold more significant evidence in this regard, although a lot more material needs to be excavated if this possibility is to be substantiated. The edge-grinding of stone tools and the making of pottery may be innovations from within a Hoabinhianculture matrix, but again I think the evidence is at present inconclusive, partly because shelter deposits are so prone to those types of hidden stratigraphic disturbance which will always perhaps cloud the issue.

Hoabinhian sites do not occur in the Indo-Malaysian islands outside north-eastern Sumatra. I suspect they are present in Taiwan, perhaps in the so-called “Changpinian” of the eastern coast (Sung 1979), and also perhaps in some aceramic assemblages of “chipped hoes” reported by Koyama (1977) from the western coast. In the Philippines, assemblages termed “Hoabinhian” have been reported by Kress (1977a, b) for Palawan and by Peterson (1974) for the Pintu shelter in northern Luzon, but the illustrations provided by Peterson (1974: Plate 1) do not convince me that the tools from this site are really any different from the contemporary pebble and flake industries characteristics of many of other Indo-Malaysian islands. Naturally, signs of a gradation from the classic mainland Hoabinhian into the very different stone tool expressions found in the islands might be expected, and it is possible that the Philippines and Sumatra are in such gradation areas. But my own experience of handling Malayan and Sumatran Hoabinhian stone tools is that they represent a dominance of pebble tools as opposed to flake tools which sets them well apart from contemporary industries in all the island regions, including the Philippines.

II.          Island South-East Asia: The Later Pebble and Flake Industries, With Variations

In the Philippines, East Malaysia and Indonesia the record of flaked stone tool production goes back about 40 000 radiocarbon years – considerably longer than in the Hoabinhian region (this is, of course, ignoring the undated but possibly much older Tampanian industry). Basically it comprises a widespread pebble and flake technocomplex which was also, on present evidence, carried by at least one of the first populations to settle in Australia and New Guinea. The technocomplex includes the majority of Indo-Malaysian flaked stone assemblages, and in its most basic form is characterised by varying proportions of simple pebble toolsm cores and flakes with non-standardised shapes. As I have noted, the Hoabinhian tendency towards a dominance of pebble tools with regular bifacial or all-over unifacial working is not present, and while this circumstance may reflect important cultural differences I suspect there may also be some geological reasons behind it. For instance, the blocky cherts and obsidians used more commonly in the island regions may have been more amenable to a flake technology than the flat coarse-grained pebbles which occur in Malayan rivers, although this is only a subjective opinion drawn from my own rather limited geological observations.

The best way to visualize the prehistoric record of flaked stone within the last 40 000 years in the Indo-Malaysian islands is in terms of periodic and normally highly localised accretions on to the basic pebble and flake technocomplex, which in its basic form underwent very little change over this period. Thus, sporadic and short-lived occurrences of prepared core, bifacial lanceolate, edge-grinding and blade technologies occur, each in a restricted region and over a different period of time. The final stage is of course the widespread appearance of fully-ground adzes and axes together with pottery after 3000 bc, and I will return to this in the following chapter. However, it should be noted that the older flaked stone technologies continued with no obvious changes until they finally faded in the face of metal tools from the late first millennium bc onwards. The flaked stone traditions do not in themselves record the spread of an agricultural lifestyle in the region.

In organising this chapter I have decided to proceed along geographical lines from Borneo through eastern Indonesia. However, the industries with a blade or “microlithic” component, all dating within the last 7000 years, are described separately in the final section.

A.  The West Mouth, Niah, Sarawak

The huge West Mouth of the Niah Caves in Sarawak contains the longest stratified record of human occupation in Island South-East Asia. The caves themselves form a network of hugh and awe-inspiring passages, with an area of about 10.5 ha, inside the Gunung Subis limestone massif near Niah in northern Sarawak. The system has many outlets, of which the West Mouth is the largest, being about 250 m across and 60 m high. Most of the system is floored with continuously-deposited wet guano, but an area high and dry at the northern end of the West Mouth was used for habitation and burial from about 40 000 until perhaps 2000 years ago or later. This area was excavated on a fairly massive scale by the late Tom Harrisson between 1954 and 1967.

Harrisson produced many impressionistic articles and a few detailed typological studies based on his research at Niah, but no proper plans or stratigraphic drawings were ever made. Since most of the site was destroyed by the excavations it is now too late to make full amends, although Majid (1982) has recently attempted to piece together some of the information which can be gleaned from the earlier records. In addition, other publications since 1967 by Barbara Harrisson, Lord Medway (the Earl of Cranbrook) and by biological anthropologists have filled in many lacunae in the records of animal faunas and human burials.

Harrisson’s reconstructions of the cultural sequence at Niah were based partly on the idea that depths and ages could be correlated regularly across the site. However, the site has an uneven surface, and arbitrary levels of excavation up to 24 inches thick, plus a set of partially contradictory radiocarbon dates recorded only by depths below surface, clearly do now encourage much confidence in finer details of the “Niah area phaseology” which he published and revised from time to time. His last version appeared in his 1970a paper, in which he favoured a basal flake industry with pebble tools appearing intermittently, and becoming edge-ground after about 10 000 bc. I will consider the Neolithic tools and the pottery (the latter comes in at 2500 bc) in a later section, but wish now to examine the preceramic West Mouth sequence in terms of artefacts, faunas and human burials.

I.             Artefacts at Niah

Since Niah has no sources of fine-grained stone such as chert or obsidian in its vicinity, the tools were made mainly on various types of sandstone which had to be smashed rather than flaked to give sharp-edged pieces. Hence the industry comprises mainly an unretouched array of chunks and chips, without coherent core forms and with few conchoidal flakes. There is little systematic retouch. Pebble tools also occur, but apparently not in the oldest levels. Bone spatula and points do apparently occur to the base of the site, some made on pig tusks or mammal longbones (Harrisson and Medway 1962; Majid 1982: Appendix 3). Stone mortars and edge-ground pebble axes appear later in the sequence; dates unfortunately are not clear, but Majid suggests that both could have appeared somewhere between 20 000 and 10 000 years ago, although Harrisson preferred the latter date for the edge-ground axes.

The edge-grinding of pebble axes is of course a significant technological development (Hayden 1977b). it occurs from the end of the Pleistocene in Vietnam, from dates in excess of 20 000 years ago in northern Australia (Schrire 1982), from about 30 000 years ago in Japan (Ikawa-Smith 1979;Serizawa 1982), and possibly before 14 5000 years ago in the New Guinea Highlands (Mountain 1983:94-5). Hence, if Niah is included, there are four separate Late Pleistocene occurrences, to which may be added a possibly early Holocene appearance on Palawan in the Philippines (kress 1977a; see also Peterson et al. 1979 for Luzon). The technique clearly recedes the development of blade technologies, even in Japan where blades appear by perhaps 26 000 years ago, and it occurs in similar pebble- and flake-based industrial backgrounds in each region. At Niah it was probably adopted because of the difficulty of finding good stone for flaking, and it is totally absent in many contemporary or later Indo-Malaysian industries where good cherts were available (for instance, in neighbouring Sabah). Hence there are two rather contradictory aspects of distribution – on the one hand a wide-sread distribution of the edge-grinding technique in Late Pleistocene times around the eastern fringes of the Old World, but on the other hand a very spotty occurrences within this territory (within Australia, for instance, it remained strangely restricted to the region of Arnhem Lang until about 5000 years ago). It thus seems hat equal cases can be made for multiple independent development of the technique or for its diffusion from one source, and the real answer may combine both processes.

2. The Niah Economy

The animal bones from the West Mouth indicate fairly eclectiv hunting patterns. Medway (1977a) lists 58 species of mammals found in the cave, and apart from bats, which may have fallen naturally into the deposits, there are numerous species of rodents and insectivores seven species of primates, eleven carnivores (excluding the Neolithic dog and a tooth of the non-native tiger from the top of the site), and ten other large native mammals. Wild pigs (Sus barbatus; Medway 1978) were the most popular prey throughout, together with porcupines at depth but with an increasing emphasis on monkeys in higher levels. Other large mammals from the oldest levels include giant pangolin (now extinct), Malay pangolin, Malayan tapir, orang utan, deer, and perhaps bovids (Hooijer 1963; Medway 1977a). the Sumatran rhinoceros (Medway 1965) and the Malayan bear also make rare appearances at higher levels. Apart from the mammals, fish, birds, monitor lizards, snakes and crocodiles were also brought into the cave.

I discussed some of the faunal evidence which suggests that the Niah region may have had a drier and more seasonal climate during the period of the last glacial maximum. Some of the larger mammals such as rhinoceros and bear seem to have been more common at this time, and during the early Holocene a number of species (such as orang utan, rhinoceros, monkeys, and rats; Medway 1978) commenced a slight decline in size. The Malayan tapir also declined into local extinction, and Medway (1977a) attributes these Holocene changes to the spread of dense forest and its non-clearance by human agency. Estuarine shellfish also increased in numbers in both the West Mouth and neighbouring Lobang Angus as a result of the rise in sea-level. Majid (1982) has reported a few plant remains from early levels, particularly of the large edible nuts (toxic until soaked in water) of the forest tree Pangium edule.

3. Human Burials

The possible racial affinities of the Niah human remains have been discussed and I will only add cultural details here. The single “deep skull” associated by Tom Harrisson (1975b:161) with a radiocarbon age of about 40 000 years, was found together with some longbones under a large stone. No stratigraphic details have ever been published, but Barbara Harrisson (1967:143) has stated that the dated charcoal was taken from directly above the skull. Nevertheless, the fact that it lies come 125 cm below all other human remains in the site must at least suggest burial from a higher level and consequent mixing of deposits of different ages. The problem cannot really been solved since all deposits surrounding the skull have now been excavated away.

The other burials from the preceramic levels comprised flexed, sitting, and disturbed fragmentary remains. A number of radiocarbon dates on bone collagen have now been published (Harrisson 1975b; Brooks et al. 1977) and between 12 000 bc and 6000 bc, while the flexed ones run from perhaps 9000 onwards. Several of the flexed burisals occurred in later levels with pottery, so this mode is not of course a guarantee of an early date. Hematite powder and traces of burning occurred on several burials (Harrisson 1967), and goods included an edge-ground pebble (unfortunately not with a dated burial), a rhinoceros femur pillow, a bone point and an estuarine shell with hematite. It should be noted that the only buried dated twice, number 147, has one published collagen determination of 7020 +_ 135 bp (Brooks et al. 1977), and another very contradictory determination of 13 6000 +_130 bp (Harrisson 1975b).

The task of summarizing 35 000 years of pre-agricultural life around the Niah Cave is a difficult one – the flaked stone tool tradition has a stunted appearance due to the poverty of raw material, and there are insufficient data to study trends in economy within these 35 long millennia in any very useful way. The sequence clearly lacks the specialised flaking technologies which appeared in other nearby regions, yet the early adoption of edge-grinding suggests for Niah a fairly innovative prehistory somewhat unique to itself.

B.   Sites in Eastern Sabah (Northern Borneo)

Since 1980, an excavation project carried out in eastern Sabah under the aegis of the Sabah Museum in Kota Kinabalu has documented a number of cave and open site with deposits extending back for perhaps 30 000 years (Bellwood 1084). The situation of these sites; the caves and shelters are found in the Madai and Baturong limesone massifs, both of which contain networks of solution tunnels, some of which emerge into the open air as dry habitable locations (as in the Niah Caves). Baturong is in turn surrounded by a large are of water-laid deposits which are presumed to have been laid down in the bed of an extinct lake, formed by the damming of an old course of the Tingkayu River by a lava flow extruded from the flanks of nearby Mount Mostyn. Although these sites are near the coast now, the low sea-level conditions of the Late Pleistocene may have placed them up to 150 km inland.

Flexed burial (undated) from a preceramic level at the West Mouth, Niah, Courtesy Sarawak Museum have shown the approximate boundaries of the old Tingkayu lake as identified by previous soil and geological surveys, and by fieldwork undertaken in 1981 with Philip Hughes. The lake covered perhaps 100 km before it drained away as the outlet for the present Tingkayu River was downcut into a gorge just north of Tingkayu village. The date of formation of the lake is probably indicated by a radiocarbon determination of 28 300+750 bp from charcoal sealed beneath the end of the lave flow, which outcrops into the side of the exit gorge near Tingkayu village, and active research to verify this surprising antiquity is currently underway. In addition, very weathered water-laid sediment at the base of the Hagop Bilo shelter in the Baturong massif were exposed to subaerial weathering as a result of draining of the lake some considerable time before 17 000 years ago (John Magee, personal communication.) These dates are highly significant because a number of open sites lie directly on the shoreline of the old lake, and on locational grounds they may be considered as contemporary with lake-full stage, and thus between 28 000 and 17 000 years old.

     The major lake-edge sites, which is Tingkayu, Baturong and Madai in eastern Sabah lie close together on a small promontory which juts into the old bed close to the lake outlet. Another site, at Tingkayu lies on top of what must have been a small island – today it is a hill amongst the oil palm plantations which occupy the western half of the lake bed.

     The Tingkayu stone industry shows a unique level of skill for its time period in South-East Asia, and the tools are mostly made on a locally quarried tabular chert (the precise source is not known, and may no longer exist, or it may be buried somewhere in the vicinity of the site), together with a few riverine pebbles of chert. The basic aspects of the industry are not exceptional, and comprise a range of large pebble tools, multi- and single-platform (horsehoof) cores, and utilised flakes. However, many of the tabular blanks were worked into large bifaces, and into smaller and quite remarkable lanceolate knives. Only one of the latter has been found complete and it has very fine surface flaking, but broken segments and points with varying degrees of finish are common.

     At the present time this bifacial industry is quite unique in the whole of South-East Asia, except for one apparent lanceolate found years ago in a tin mine in Kedah in Malaya (Stein Callenfels 1936b), and although similar forms do occur in northern Australia they all seem to postdate 4000 BC. It seems likely that this tradition was develop locally, perhaps to meet a specific need in this rather unusual lacustrine environment. In north-eastern Asia there are remote parallels for the lanceolates in the Late Pleistocene Diuktai tradition of north-eastern Siberia (Chard 1974; Ikawa-Smith 1982), and in several regions of Japan after about 18 000 years ago (Ikawa-Smith 197; Aikens and Higuchi 1981), but these occurrences are so distant that they can be no more than noted at the present time.

     During the lake period the Baturong massif formed a towering limestone island and the rock-shelters along the base of its southern cliff were all drowned. After the lake drained away the open sites were abandoned, and an occupation hiatus, which may have lasted for abut 10 000 years, now intervenes in the sequence. In the shelter of Hagop Bilo, the basal and culturally-sterile lake sediment were overlain, after a long break characterised by considerable weathering, by midden deposits in riverine alluvium dating between about 15 000 and 10 000 bc. These midden layers contain mainly three species of lacustrine gastropods in the genera Balanocochlis, Sulcospira and Brotia, and marine shells are absent. The animal faunas, provisionally identified by the Earl of Cranbrook, include pig, sambhur deer, mouse deer, porcupine, monkey, rat, snake, tortoise, monitor lizard, birds, and probably other small unidentified species. The stone tools of this period lack of bifaces, and comprise a fairly typical Indo-Malaysian pebble and flake industry with single- and multi-platform cores, utilised flakes, and flat-based and steep-edged domed scraper-like tools, all made on chert. Characteristics of some interest include a class of long blade like-knives, perhaps functional descendants of the Tingkayu lanceolates, and also the presence of an opal phytolith gloss on some tool working-edges. This gloss is widely reported from other sites of this period and later in South-East Asia, but was absent in the Tingkayu industry where open situations may not have been conducive to its survival, had it once existed. Another tool of interest from Hagop Bilo is a large bone spatula, similar to those from Niah, and pieces of scratched hematite were also recovered.

     Soon after 10 000 bc the Hagop Bilo shelter was in turn abandoned. The absence of marine shells in the deposits suggests that its inhabitants were mainly inland dwellers, and with the shoreline originally perhaps 100-150 km away this interpretation is clearly sensible. However, by 9000 bc the Madai Caves may have been coming within easy reach of the approaching coastal resources, and the cave users apparently moved seawards from Baturong to Madai at this time. The largest of the Madai Caves, Agop Atas, today contains a substantial Idahan village occupied seasonally for bird’s-nest collection. Above Agop Atas lies the smaller Agop Sarapad; both these caves were intensively inhabited by hunters during the early Holocene, between about 9000 and 5000 bc.

     The early Holocene human deposits in Agop Atas lie in a very acidic guano deposit, and, as at Tingkayu, only stone tools survive with some charcoal, all animal bones having totally dissolved. But the Agop Sarapad cave has much better conditions, for here the people deposited  a large shell midden which has created and maintained its own alkaline environment; bone survives in quite good condition, although both caves are too damp for any plant organic matter to have survived. The Agop Sarapad midden thus tells the best story, and has yielded thousands of stone tools of local river pebble chert, of an industry similar to that Hagop Bilo but lacking the blade-like knives. There is a heavy emphasis on pebble tools, large steep-edged tools, multi-platform and horsehoof (single-platform) cores, and utilised flakes, many of which have glossed edges. A number of large pitted anvils or grindstones occur, some coated with red ochre, and hammerstones are also common, either for stone tool making or for food or ochre preparations of the anvils.

     The food remains in the midden include many shells of the estuarine mangrove shellfish Batissa and Anadara, and clearly the inhabitants were visiting the encroaching coast frequently.  Most shells, however, are of the three same riverine shellfish species which were eaten earlier at Hagop Bilo. The animals hunted were also similar, with the addition of larger creatures such as the orang utan, cattle and rhinoceros; these appear to have been absent at Hagap Bilo, but the small sample size and the provisional nature of this identifications makes this uncertain. Remains of Javan rhinoceros have also been identified from Agop Sarapad by Cranbrook; this is the only report of this animal from Borneo, and it evidently survived in parts of Sabah until the early Holocene.

     After 7000 years ago the two Madai caves were abandoned, and I am unable to see any clear explanation for this, except to suggest that the inhabitants may have moved to a coastal location, or perhaps dwelt elsewhere in an unexcavated part of the cave system. However, the sweet waters of the Madai stream flow directly through the lower cave at Madai (Agop Alag), and I have seen no other caves suitable for long-term habitation in the massif; this impilies strongly that the population did move away. For about 4000 years the caves remained unoccupied, and then a new and totally different cultural assemblage makes its appearance; I will describe this in the next chapter.

C. The South-western Arm of Sulawesi

The south-western arm of spider-like Sulawesi has produced one of the best preceramic sequences of Late Pleistocene and Holocene stone tool working in the whole Indo-Malaysian Archipelago. This region is the homeland of the Makassarese and Buginese peoples who have played such major roles in the recent history of Indonesia, and archaeologically I hav already looked at the rather enigmatic open-site industry of the region around Cabenge (Chapter 2, Section IIID5). Archaeological excavations have also been carried out since early this century in caves and shelters in the tower-like karst topography which is particularly well-developed in the Maros region inland from Ujungpandang (Makassar), and many sites have now produced assemblages belonging to the industry of backed flakes and microliths known as the Toalian, which I will describe in the next section. The Toalian postdates 7000 years ago and overlaps with the appearance of pottery in the region. However, earlier assemblages are now known as a result of several seasons of fieldwork by Glover, and these come from the shelters of Leang Burung 2 and Ulu Leang in the Maros limestone.

     The shelter of Leang Burung 2 (Presland 1980; Glover 1981) is the most important site, and has produced an industry characterised by unretouched flakes and small multi-platform cores of chert from levels dated between 29 000 and 17 000 bc from radiocarbon determinations on freshwater shell. Some flakes have the typical opal phytolith edge-gloss which is found so widely in this region, and Glover suggests that it may result here from the working of matting or basketry, possibly of palm leaves (see also Kamminga 1979). In addition, there are at least four elongated blade-like flakes with facetted striking platforms which Glover regards as similar to those of the Levalloisian prepared-core technology associated with the late Acheulian and Mousterian technocomplexes of the northern and western territories of Eurasia (generally prior to about 30 000 years ago). These tools are significant because they do indicate a degree of conscious core preparation prior to flake removal which is not otherwise found in any other South-East Asian industry, although rather strangely it does appear again, perhaps quite independently, as recently as 4000 years ago in the north-western parts of Australia (Dortch 1977; Dortch and Bordes 1977). However, there are no recognisable Levalloisian cored in Leang Burung 2 and the technique seems here to have played quite a minor role. Its invention may have been independent of occurrences elsewhere, and it does not appear to continue on into the Holocene period in the Sulawesi sites.

     Also found in Leang Burung 2 are pieces of the ubiquitous hematite, but bone points are absent, as are fishbones and marine shells since the sea was presumably very far from the site at this time. The industry seems to continue (after a possible gap) into the lower levels of the shelter of the Ulu Leang (Glover 1976) which date from the early Holocene, and here there is a distinctive range of steep-edged domed tools and horsehoof-shaped cores of white chert, very similar to the Agop Sarapad industry of the same date from Sabah. Bone spatulae also appear in basal Ulu Leang, and this bone tool tradition is elaborated in the succeeding Toalian industry.

D. The Northern Arm of Sulawesi: The Paso Midden

The Paso shell midden (Bellwood 1976b) lies close to the shoreline of Lake Tondano, in the inland volcanic terrain of the Minahasa Peninsula. The midden is about 30 m in diameter and averages 1 m in depth, and consists of lenses of loose lacustrine shell interstratified with occupation layers. The latter contain an obsidian flake industry, bone points, hematite, and prolific faunal remains. The site is radiocarbon dated to about 6500 BC.

The obsidian, collected locally, is vesicular and coarse, and lumps were mostly smashed to obtain sharp chips and chunks, although flakes were struck individually from multi-platform cores as well. There are no pebble tools (one would perhaps not expect them in a raw material of this type), and no edge-gloss have been observed, perhaps due to low visibility rather than total absence. A few chunks and flakes were retouched, often into steep-edged and high backed forms like those of Agop Sarapad and basal Ulu Leang.

     The faunal remains from Paso and from the contemporary (Pre-Toalian) layers at Ulu Leang have been identified by Clason (1980; in press). Pigs (Sus celebensis, not babirusa) were most popular in both sites, and occurred with anoa, monkeys, rodents and the two Sulawesi species of marsupial cuscus (Phalanger celebensis and P. ursinus). The lake-edge situation of Paso allowed for considerable hunting of birds, while the karst riverine situation of Ulu Leang led to more frequent catches of tortoises, snakes and occasional fish. In neither assemblage are there indications of animal domestication.

E. Eastern Timor and Flores

From four caves in eastern (formerly Portuguese) Timor, Glover (1971, 1972a, 1977a) has excavated an industry with basal dates of about 13 000 years ago, which continued with little change into the ceramic period which began here during the third millennium BC. The tools are primarily chert flakes (there is also a little obsidian) and the retouched forms are mainly steep-edged scrapers. A number of the unretouched flakes have an edge-gloss and there are also a few long thick blades.

     The fauna of the period prior to 3000 BC is dominated by several species of extinct giant rats which survived until about 2000 years ago, together with fruit bats, snakes, reptiles, fish and shellfish (other placental mammal species such as pigs and deer were all introduced into Timor after 3000 BC). Remains of gathered plants in the early levels included seeds or fragments of the perennial cereal Job’s tears, betel vine and Areca nut (the ingredients of betel chewing), and candlenut (Aleurites) (Glover 1977b:18). Basically, this Timorese industry has much in common with those just described from Sabah and Sulawesi, but it does seem to be a little distinctive in its minor predilection for long blade-like artefacts.

     An undated industry from several open sites on Flores may be related to the Timor material, and I have already referred to these industries in Chapter 2, section IIID6, since some components may be of considerable antiquity, possibly contemporary with extinct species of Stegodon. However, none of this material is dated, and with scattered surface finds of this type it is quite possible that the collections cover an enormous timespan.

F. Comments

I should first add to the above list the only dated industry from Sumatra, which comes from the cave of Tianko Panjang in the Sumatran highlands near Lake Kerinci (Bronson and Asmar 1975), and which comprises unretouched obsidian flakes and chips which date from about 9000 bc onwards. Many other scattered industries of this kind have been found across Indonesia, particularly in Sumatra and Java (e.g. Franssen 1941; Heekeren 1972:137-9) and eastern Indonesia, and they have continued in production almost into the ethnographic present amongst the Tasaday (here with some edge-grinding; Fox 1976), and in the New Guinea Highlands and central Australia. The long industrial sequence from the Tabon Caves on Palawan (Fox 1970, 1978) also fits well within the Indo-Malaysian pebble and flake repertoire.

     All the chert and obsidian-based industries I have looked at (that is excluding the more intractable raw materials used at Niah) are characterised by flake production with stone hammers from multi- or single-platform cores. Retouch is often steep, and high-domed “scraper” forms are quite common, as is the edge-gloss on retouched flakes. In some sites core smashing appears to have been as important as systematic flaking (especially with obsidian), and this has been emphasised by Coutts and Wesson (1980) for the Philippines, where they refer rather colourfully to a category of “smash and grab” industries. Although flakes often have blade-like proportions, there is no sign of any systematic attempt at blade production until after 6000 BC.

     The basic core, flake and pebble characteristics of these Late Pleistocene and Holocene island industries are on a similar technological level to some of the Mousterian or Middle Palaeolithic industries of northern Asia (including Japan) and India, and they find fairly close parallels in the “core tool and scraper tradition” (Bowler et al. 1970) of Australia and New Guinea, both of which were first settled from Indonesia before or around 40 000 years ago. These similarities are not surprising, and the localised variations which relate to raw materials and chance are often overstressed by prehistorians with purely local concerns. It is true that unusual variants do occur, and apart from those at Leang Burung 2 and lake Tingkayu there is also the unusual significance of large waisted axe-like tools (“waisted blades”) in the New Guinea Highlands and parts of Australia. These latter items do not occur in the Indo-Malaysian islands, but they appear occasionally and perhaps independently in some Hoabinhian sites.

III. The Flake and Blade Technocomplex of the Mid-Holocene

In parts of the Philippines, Sulawesi and Java there are a number of assemblages dated to after 6000 BC which demonstrate regionally-varied emphases on the production of small blades, and sometimes on other specialised tools such as “microliths” with blunted backs and small projectile points. In all cases these new technologies appear as accretions on to the old and continuing tradition of unprepared flake production.

     In a previous book (Bellwood 1978a:71) I quoted a definition by Morlan (1971:143) of blades as “elongate parallel-sided flakes with parallel arrises or parallel-sided facets on their dorsal faces. Blades are struck (by indirect percussion) from prepared, polyhedral cores…”. The Upper Palaeolithic industriesof much of the Old World were focussed on the production of blades of this type, and in Japan, northern-China and north-western India it is now clear that they were widespread by at least 20 000 years ago. However, in Island South-East Asia and Australia blades form only  a small minority component of most assemblages in which they occur, and true cylindrical or conical blade cores are generally very rare. Many of the “blades” found in this region fall into Morlan’s category of blade-like flakes, which are less symmetrical than true blades and which lack the parallel ridges. Nevertheless, since cores of a subprismatic shape do occur in small numbers, and since I believe that both blades and blade-like flakes were produced intentionally in some sites, I will refer to the industries described in this section as the “flake and blade technocomplex”.

     At present there appear to be two kinds of industry within the flake and blade technocomplex; the unretouched blade industries of the Philippines and the Talaud Islands of north-eastern Indonesi, and the backed flake-blade and microlithic industry, termes the Toalian, in south-western Sulawesi. At the end of this section, I will consider some undated industries, perhaps related to the Toalian, which have been found on Java. The Philippines, Sulawesi, Java, Australia, and possibly southern Sumatra encompass the distribution of industries in this technocomplex, and they seem to be completely absent from Borneo, eastern Indonesia and New Guinea (although Timor does have a later Neolithic industry of tanged blade-like points).

A.  The Philippines and Talaud

 Industries in which a component of small retouched blades is added to a continuing flake tradition occur in the Philippines, and in the Talaud Islands south of Mindanao. In Duyong Cave, near Tabon Cave on Palawan, Fox (1970) has excavated an industry with small blade-like flakes struck from what he originally termed “prepared cores”, although he has since stated that such cores were not present (Fox 1979:236). The tools occurred in a midden of marine and estuarine shell dated to the sixth millennium BC. Possibly similar industries have been described from Cebu by Tenazas (1985), and from a small island called Buad off the western coast of Samar by Cherry (1978). No clear chronologies are available for these sites, and both late preceramic and Neolithic associations seem likely. Blades account for up to 50% of the Buad collections and some pieces have glossed edges or traces of a mastic used in hafting.

     The most detailed information on an industry of this type comes from the rock-shelter of Leang Tuwo Mane’s on Karakellang Island in the Talaud group (Bellwood 1976b). this shelter was originally cut by the sea into a cliff of coral limestone, and was then uplifted to a habitable level by about 4000 BC. The basal deposits produced an industry comprising blades and blade-like flakes (about 50% of all flaked stone), together with some rather rudimentary prismatic cores, made on a grey chert. Retouch is virtually absent, but some edge-gloss occurs. Around 2500 BC pottery appeared in the site, and at this time there was a surprising and unexplained change away from the grey chert towards a nodular brown chert which was used for the production of a much less refined flake industry, which continued on into the upper layers of the site. The blade industry is thus restricted mainly to the preceramic levels, and the date of its actual appearance in Talaud cannot be determined from this site. Both preceramic and ceramic layers contained large numbers of shellfish of no less than 94 species from a wide range of reef and mudflat habitats (Heffernan 1980), but apart from pig in the ceramic layers no mammal bones were present; the Talaud Islands had only a limited native fauna of rats, bats and a species of Phalanger.

B.   The Toalian

The most important industry of the flake and blade technocomplex is undoubtedly the Toalian of south-western Sulawesi, which commenced about 6000 BC with an array of microliths (small backed flakes and geometrics) of types seemingly unique in the Indo-Malaysian Archipelago. Toalian assemblages have been excavated since 1902 from caves and shelters scattered across the southern half of the south-western peninsula of Sulawesi, and during the 1930s and 1940s some rudimentary typological successions were established by Stein Callenfels (1938) and Heekeren (1949). In Heekeren’s last major summary (1972), by which time about 20 sites had been excavated, he felt justified in supporting a three-phase sequence commencing with plain blades, followed by a second phase with backed flakes and blades and geometric (crescentic and trapezoidal) microliths, and culminating finally in a third phase with bone points, serrated and hollow-based stone points, and pottery.

     As a result of two recent excavations there is now much more information on the Toalian. Both sites are located in the Maros limestone region north of Ujungpandang; one is called Leang Burung 1 (Mulvaney and Soejono 1970, 1971; Chapman 1981) and the other, Ulu Leang, has received detailed treatment over several seasons by a team led by Glover (1976, 1977a, 1978a; Glover and Presland 1985). The latter site has the most complete sequence.

     I have already considered the basal industry of flakes and steep-edged tools at Ulu Leang, dated to between 8000 and 6000 bc. The Toalian tool types appear in higher levels dated from about 6000 BC within a continuing basic industry of flake tools and bone points, although the steep-edged tools fade in importance. The most important new tool type is a small flake or blade with straight or oblique blunting down one side and often around the butt (that is similar to a “backed blade” in Australian terminology), and some of these backed forms have distinctly crescentic or trapezoidal shapes (geometric microliths). The present trend (Chapman 1981; Glover and Presland 1985) is to refer to all these new Toalian forms as “microliths”. Sometime after 4200 BC another type of microlith, the “Maros” serrated and hollow-based point, appeared in the sequence, perhaps as a result of local innovation. Other artefacts which occur throughout the Ulu Leang sequence include glossed flakes, small bipolar cores, bone points, and bivalve shell scrapers (William 1939).

     The site of Leang Burung 1 is later than Ulu Leang, and the deposits appear to date within the last two millennia BC and have pottery throughout. Hence there is no long sequence here, but the microliths and Maros points are still present and Chapman (1981) stresses the importance of the edge-gloss (on 24% of used tools), and also notes the absence of steep-edged tools, pebble tools and edge-ground tools (the latter, to my knowledge, appear to be absent in all Toalian sites). At Leang Burung 1 the microliths (including Maros points) comprise about 35% of all retouched tools, which in turn comprise only 6% of the total of stone artefacts, and at Ulu Leang the corresponding percentages ar 20% and 6%.

     The Ulu Leang sequence generally supports the essentials of Heekeren’s second and third phases (that is backed microliths succeeded by Maros points), but the site has no earlier phase with plain blades. This need not mean that this earlier phase does not exist in all other Toalian sequences, and as I have shown it is present without later typological elaboration in the Talaud Islands. The question of intentional blade production in the Toalian is in fact a difficult one. Chapman feels that it was practised at Leang Burung 1, and about 10% of flakes here had blade proportions, although no prismatic core were found. Heekeren (1972: Plate 63b) also illustrates a series of blades from the site of Panganreang Tudea (near the southern tip of the peninsula) which remind me closely of the blades from Talaud. However, Glover and Presland (1985) deny the existence of a blade technology at Ulu Leang, although Glover has often used the term “blade” in his published discussions of the site. After analysing the stone tools from the much older site of Leang Burung 2, together with those from Ulu Leang and Presland (1980) concluded that a blade technology was not evident in these particular Maros sites, although flakes did become smaller by about 20% in average size over the whole period (over 20 000 years). However, no one has challenged the significance of the sudden appearance of the new microlithic forms after 6000 BC. These raise the question of outside parallels, which I will consider later.

     The Toalian industry continues on well into the pottery phase at most excavated sites, and in itself it reveals no clear evidence for the agricultural economy which must surely have been developing in some parts of Sulawesi after 2500 BC. Even as late as AD 1000 the retouched and glossed flakes still occur at Batu Ejaya near the southern end of the peninsula (Chapman 1981), although most of the microliths have by now disappeared and the serrated Maros points appear to have been replaced by plainer round-based forms. The late survival of the Toalian here may document a continuing tradition of hunting and gathering, perhaps amongst indigenous non-Austronesian speakers contemporary with cultivators elsewhere, or perhaps as a sporadic activity by village-based agriculturalists.

     The economic evidence from the Toalian sites includes a range of hunted and gathered resources. Riverine shellfish are very common, and Glover (1977b:52) found remains of wild seeds and nuts at Ulu Leang, although carbonised rice grains only appeared in the site after AD 500 (Glover 1985). In Section IIE I described the animal remains from the lower levels of Ulu Leang, and the faunas from other Toalian sites (Hooijer 1950a) were drawn from a similar range of Sulawesi mammals; the two species of Phalanger, macaque monkey, civet cat, anoa, Sus celebensis and babirusa. A small quantity of art found on the limestone walls of the Toalian sites has been described by Keekeren (1950b, 1972:119-20), although none can be dated. Hand stencils and wild pigs were depicted in red hematite, and the former are of great interest because of the world-wide occurrence and great antiquity of this motif. Some of the hand stencils, interestingly, lack one or more digits.

C.  The Sampung Industry of Eastern Java

This industry, called the “Sampung bone industry” by Heekeren (1979:92), is best known from a cave called Gua Lawa near the village of Sampung, between the Lawu and Liman volcanoes in central Java. The site was excavated by Stein Callenfels in 1931, and although the methods were rather crude and the site obviously disturbed, he did provide a section drawing in his report (1932) which shows the vertical distribution of all the artefacts found within a 2 m thick occupation deposit. The lowest of three apparent levels of occupation produced a number of stone projectile points with hollow and round bases, but without the serrated edges characteristic of Maros points. Hollowed stone grindstones and spherical rubbing stones with traces of hematite also appeared in this layer, but no records were made of the basic flaked stone industry which was also present.

     Above the lowest level, and apparently extending over about half of the excavated area within the cave, Stein Callenfels found a lens of bone and antler points and spatulae with more stone mortars, and a few possibly downwards-disturbed postherds and a polished adze. Pottery, metal, and a general mixture of other material then occurred in the top layer, with four bone fishhooks. Flexed burials, at least one under a stoneslab and including one child with  shell necklace, were indicated as stratified within the middle layer with the bone tools; all have been classified as Australoid or Melanesian (see Chapter 3, Section IIIB).

     At face value (and there is no other way to interpret this site now) the Gua Lawa sequence indicates a definitely preceramic industry of bone tools and stone arrowheads very similar to the non-serrated Toalian types, together with “flakes and blades without secondary working and many retouched shell scrapers” (Heekeren 1972:94). In a previous account of this site (Bellwood 1978a:76) I included the bone tools with the top pottery-bearing layer, but I am now inclined to regard the latter as little more than a surface skin and to regard the bone tools as basically preceramic. The fauna, apparently mainly from the middle layer with the bone tools (Dammerman, Annexe I in Callenfels 1932), comprised a broad range of big mammals such as banteng cattle, deer, pig (Sus vittatus), monkeys, Indian elephant, buffalo, and several large felines. Monitor lizards were especially frequent.

     This type of industry is known from sites scattered all over the eastern end of Java, but all from very old excavations. South of Gua Lawa a rock-shelter in Gunung Cantalan has produced round-based stone points and many facial, masklike parts of macaque monkeys (Heekeren 1972:99). Other sites occur to the north around Bojonegoro and inland from Tuban. Heekeren (1935-6;1937) also excavated three caves, called Petpuruh, Sodong and Marjan near Puger in south-eastern Java, about 200 km east of Gua Lawa. These caves produced the same preceramic assemblage of round-based stone points, bone tools, shell scrapers and rings, and flexed burials. Pebble tools and flakes of quartz, chert and obsidian also occurred, together with hammerstones and grindstones. Heekeren seemed to favour Hoabinhian affinities for these industries, but I find this most unlikely and would rather stress their close similarities with Toalian, except for the obvious absences in eastern Java of the backed flakes and geometrics. A similar mid-Holocene date also seems very likely.

D.  Other Flake and Blade Industries in Java

A large number of sites in western Java, especially on the Bandung Plateau, have produced an undated but presumably preceramic industry of obsidian (Bandi 1951; Heekeren 1972:133-7; Anggraeni 1976). The sites still await detailed archaeological investigation, and the existence of a definite blade element seems to be rather uncertain. However, the available illustrations leave no doubt that backed flakes, round-based and unifacially retouched projectile points, and a few crescentic and trapezoidal geometrics are present. The sites may in fact be mixtures of several time periods, for a number of more archaic domed and steep-edged tools also occur, but the overall impression is of an industry which may be fairly similar to the Toalian. Some surface collected absidians from sites around Lake Kerinci and Jambi in south-central Sumatra (Hoop 1940) may also contain point and microliths, and Glover and Presland (1985) report backed crescents from some of these sites, although such forms were absent in the excavated Tianko Panjang cave.

E.   Comments

Reasons for the appearances of these new flake and blade technologies, both in the Indo-Malaysian region and in Australia, are topics which cause considerable dispute amongst prehistorians. One important fact which should be noted is that in Island South-East Asia and Australia these new technologies are always grafted on to old ones – there are no wholesale replacements of flake industries by blade industries, and no good reasons from the stone tool evidence to invoke large-scale human expansions into the region prior to that of the agricultural Austronesians. Nevertheless, it is difficult to decide for individual regions whether the new technologies developed independently or whether they were adopted from an outside source, an in some cases the answer may be a combination of both processes. In company with Glover and Presland (1985) and Dortch (1977) I feel that some significance mut be allocated to diffusion, although it should not be forgotten that the ability to produce elongated blade-like flakes was present at the sites of Leang Burung 1 and Hagop Bilo well before the end of the Pleistocene. Indeed, many Australian prehistorians today regard the “small tool industries” of Holocene Australia as local developments quite unrelated to anything in Indonesia (e.g. Hallam 1977; White and O’Connell 1982).

     I rather doubt that these questions of independence versus diffusion will ever be resolved, since stone tools of this kind, with the possible exception of the Maros points, do not carry a large component of cultural as opposed to purely functional information. However, the simple fact of coincidence in dates can hardly be overlooked, and the points, backed blades and geometrics so characteristic of later Australian prehistory do appear there from about 4000 BC onwards (Schrire 1982; Mulvaney 1985). Possibly Heekeren (1972:125) was correct when he suggested that these tool types actually diffused to Sulawesi from Australia. Radiocarbon dates as known at present do not in my opinion provide much support for this view, but it should not be dismissed entirely.

     Perhaps the most likely region of outside influence on the island of South-East Asia in terms of these flake and blade industries is Japan. India is only a remote possibility, since the intervening Hoabinhian industries show no signs of any of the developments under consideration. However, during the Initial, Early and Middle Jomon periods of Holocene Japan (c. 7500 bc onwards) there is a range of hollow-based projectile points and blade tools which look a little like some of the South-East Asian examples, although geometric microliths of the Toalian types are to my knowledge not found there. An undated but presumably Holocene industry with some blades has recently been reported from a site near Guangzhou (Canton) in Guangdong province in southern China (Huang et al. 1982), but there seems to be no material of this type from Taiwan or the Ryukyu Islands. Nevertheless, Japan does provide a number of interesting but vague possibilities, and no prehistorian with competence in both South-East Asia and Japan has ever really looked into this matter in any detail. As with the appearance of edge-grinding, a development obviously quite unrelated to the appearance of the flake and blade industries, there are a lot of question marks when looking at questions of origin.