Themes in the prehistory of tropical Australia.

Morwood, M.J. ; Hobbs, D.R.

The wetter tropical zones of northern Australia are linked by their monsoonal climates. Their archaeology shows its own distinctive pattern as well, and rock-art is an important source of evidence and insight. This study focusses on a part of Queensland, setting this local sequence alongside Arnhem Land (reported by the paper of Tacon & Brockwell) and in the northern pattern as a whole.

No matter what route was taken by the first Australian immigrants, the earliest sites in Australia should be in the tropical north (Birdsell 1977). This same region is likely to document continued cultural and genetic input from adjacent island southeast Asia and New Guinea. Certainly in the historic period, the impact of Indonesians in the Kimberley and Arnhem Land (MacKnight 1976), and of Papuans across Torres Strait (McCarthy 1939) is well documented. Furthermore, the sudden appearance of the dingo in Australia c. 4000 b.p. is indisputable evidence for Asian contact in the mid Holocene (Gollan 1984). It also suggests that Asian contact has occurred at least sporadically throughout the entire Australian cultural sequence.

Parts of northern Australia also have longstanding technological, artistic and linguistic traits, which clearly distinguish them from more southern areas, and warrant more detailed archaeological investigation. Edgeground axes, found in Pleistocene contexts in the Kimberley, Arnhem Land and Cape York Peninsula (as well as in the New Guinea Highlands and parts of southeast Asia), only appear in the rest of Australia in the mid to late Holocene. What this early north-south difference in Australian stone artefact technology means is unclear, but Arnhem Land, the Pilbara and the Kimberley also have a degree of artistic and linguistic complexity not evident in the rest of the continent. These areas have complex figurative rock-art styles, which contrast markedly with the geometric and track emphasis of central Australian art and with the recent Simple Figurative rock-art styles found elsewhere around the margins of the continent (Maynard 1979). In addition, there seems to be a close association between the `Northwest Australian Rock-art Province' and linguistic complexity: of the (possibly) 29 Australian Aboriginal language phylla, 28 are found only in the Arnhem Land and Kimberley regions (McConvell 1990). Dated rock-art evidence indicates that the distinctiveness of symbolic systems in northwest Australia may have a Pleistocene antiquity.

This paper examines the evidence for generalized economic, material and demographic responses across northern Australia to the long-term climatic changes which occurred between 15,000 and 7000 b.p. However, interpretations of change in the archaeological record are seldom straightforward or unambiguous, even where there appears to be good correlation with climatic fluctuations. Inadequacies in the available data-base also preclude detailed reconstructions of developments in Aboriginal land-use in the tropics, a point emphasized by recent research in southeast Cape York Peninsula.

The Australian tropics

The Australian humid and semi-arid tropical zones are here defined as those areas north of the Tropic of Capricorn (23[degrees] 26' 5" S) with annual rainfall greater than 500 mm. Most of the region is characterized by a highly seasonal, summer (monsoonal) rainfall pattern. Differences in the predicability and magnitude of summer rains between say, the rainforest areas of northeast Queensland, the flood-plains of western Arnhem Land and the eucalypt savanna of the North Queensland highlands mean that Aboriginal adaptations to specific environments were similarly varied.

Recent Aboriginal land-use

In parts of the tropics, high and predictable rainfall has a major impact on the geographical and seasonal availability of faunal and floral resources, and Aboriginal communities responded accordingly; thus, coastal areas normally have higher population densities than inland regions (e.g. Tindale 1979). Aboriginal people in these regions perceive the year in terms of seasonal attributes, resource availability and economic activities, and their diet, social life, material culture and land-use patterns respond to these regular changes in resource distribution. The earliest study of this behaviour was by Thomson (1939) in western Cape York, where the Wik Monkan classify all parts of their country on the basis of botanical communities and resources, and divide the year into five seasons, each with characteristic conditions, foods and occupations (Thomson 1939: 211):

It cannot be emphasized too strongly that these movements, each circumscribed and conducted within well-defined limits and definitely related to a season and a food supply, take on a very different aspect; seen in true perspective, they form a regular and orderly annual cycle carried out systematically, and with a rhythm parallel to, and in step with, the seasonal changes themselves.

Generally, ethnographic records from across the region indicate that there were common responses to short-term and/or seasonal fluctuations in resource availability. Maximum population dispersal and mobility occurred during the early dry season, when surface water was freely available, and a progressive `falling back' to permanent waters occurred as the country dried out. The wet made travel difficult, and often more-or-less permanent camps were established in that season. This was also the time when rock-shelters were used most intensively (e.g. Trezise 1971: 7). At the end of the wet, these camps were abandoned, and groups became progressively smaller and more mobile as the dry progressed (e.g. Anderson 1984; von Sturmer 1978; Tacon 1989).

In similar fashion, there appear to have been common responses across the region to long-term climatic fluctuations, although the archaeological manifestations vary according to local resource structures, particularly the availability of water.

Predicted human responses to palaeo-environmental

changes

Prior to 30,000 b.p., when the Australian tropics were first occupied, conditions were wetter and temperatures were slightly depressed (Bowler 1976). As a result there would have been few constraints imposed by availability of water sources. The earliest archaeological manifestations of human presence in the region are probably ephemeral sites documenting transitory visits by wide-ranging, exploratory groups.

During the Last Glacial Maximum, between 25,000 and 15,000 b.p., the sea was up to 140 m lower than today (Chappell & Grindrod 1983: 67). Most of the north coast of Australia was connected to New Guinea by a land-bridge, and elsewhere the coastline was up to 400 km further out than present shore-lines. The last glacial was a time of reduced rainfall, strong winds and lower mean annual temperatures, which would have limited water availability and biological productivity (Bowler 1976). Thermoluminescence dates for an extensive sandsheet in southeast Cape York Peninsula indicate that sands were most mobile during this time, with sand accumulation rates being highest c. 18,000 b.p. (Morwood et al. 1995a). Geologically permanent water sources would have been crucial to the pattern of human land-use at this time. Some regions/sites may have been abandoned. Others may show an increase in occupational intensity as people became more tethered to permanent waters.

From 15,000 b.p. the climate warmed and sea-levels rose. Greater rainfall also increased the number and distribution of reliable water sources and biological productivity, especially during the early Holocene, when conditions were more favourable than today. Some regional differences are apparent in timing. Between 13,000 and 14,000 b.p. there is general evidence for increased rainfall, possibly from the southward movement of the monsoons, but this increase is delayed in the Atherton Tableland sites of northeast Queensland until about 10,000 b.p. (Hiscock a Kershaw 1992: 54). Post-Pleistocene sea-levels reached a mid-Holocene high about 6000 b.p. Since this time, sea-level has fallen about 1 metre (Chappell 1983). Such relative stability has led to the progressive development of estuarine environments of high biological productivity in some coastal areas.

Considerable variation between regions is evident in the timing and nature of specific changes in resource levels and structures. In western Arnhem Land, the stabilization of sea-levels from 6000 p.b. was followed by a progradation of coastline, then the formation of freshwater swamps over the past 2000 years. These swamps, important resource areas for water and plant foods, significantly increased the potential human carrying capacity of this region (e.g. Hope et al. 1985; Tacon & Brockwell, this volume). Post-Pleistocene hydrological changes in some `inland' regions also seem to have increased water availability and allowed (more sustained) human occupation. The activation of Mickey Springs on the upper Flinders River in North Queensland was probably the factor responsible for first use of adjacent rock-shelters in the terminal Pleistocene (Morwood 1990: 20).

A common end-result in many regions was potential for human population expansion. In many cases, post-Pleistocene increases in carrying capacity are likely to have been progressive and cumulative (e.g. western Arnhem Land, southeast Cape York Peninsula). Population densities in some coastal regions also may have risen abruptly in response to the rises in sea-level and the consequent reduction in land area. Furthermore, the development of social mechanisms, including more complex alliance networks and systems of economic reciprocity, may have led to more efficient use of geographically and seasonally dispersed resource abundances, and so further increased effective environmental carrying capacity for local hunter-gatherers. These social developments are also likely to have been progressive and cumulative.

In summary, from about 15,000 b.p. to the stabilization of sea-level 6000 years ago, the archaeological record should show various indications of population increase. Although some of the associated changes may be threshold-linked and therefore abrupt, some should reflect processes rather than events and should therefore be gradual. Possible measures of population expansion might include increased site numbers, more intensive use of sites, first use of marginal habitats, more intensive types of resource-use, and implementation of new social mechanisms for territorial bounding (cf. Lourandos 1983).

General archaeological evidence

Reconstruction of Aboriginal prehistory in the Australian tropics is based on a relatively small number of sites in a restricted number of areas. In addition, the nature of archaeological research in the Australian tropics has closely mirrored methodological and theoretical developments in Australian archaeology generally. Earlier emphasis on establishing antiquity and culture sequence has led to a fragmented body of data of limited value in writing regional prehistories and unsuitable for making general comparisons, except in very limited fashion. Even so, the oldest dates for human occupation of Australia now come from the tropics: radiocarbon determinations indicate initial use of Nurrabullgin Cave in northeast Queensland prior to 37,000 b.p. (David 1993; 1994), while thermoluminescence (TL) and optically stimulated luminescence (OSL) dates suggest parts of tropical Australia were first occupied between 53,000 and 61,000 b.p. (e.g. Roberts et al. 1990; 1994a; 1994b; cf. Allen 1994; Allen & Holdaway 1995). In addition, there is now sufficient evidence on archaeological sites and their environmental contexts in some regions to investigate human responses to climatic changes in the terminal Pleistocene/early Holocene. These areas include the west Kimberley, western Arnhem Land, the Queensland Gulf country and the North Queensland highlands.

Site numbers

Few parts of tropical Australia have had sufficient research to quantify changes in site numbers over time, although there is circumstantial evidence to suggest an increase during the terminal Pleistocene and Holocene. One notable exception is western Arnhem Land.

Western Arnhem Land shows an accelerating increase in site numbers from the terminal Pleistocene (Figure 3; Jones 1985; Schrire 1982; Allen & Barton 1989); this trend cannot be interpreted in detail because many of the basal dates are inferred from radiocarbon determinations higher in the sequence (e.g. Anbangbang, Jimeri II). In addition, late-Holocene changes in resource levels and structure appear to have resulted in major transformations in local Aboriginal land-use and population distribution. Over the past 2000 years one focus of occupation was on open sites adjacent to freshwater swamps, but most archaeological excavations have been in rock-shelters (Tacon & Brockwell, this volume; Hiscock 1992). The comparative increase in actual site numbers in western Arnhem Land during the late Holocene is likely to have been far greater than is evident in Figure 3.

Intensity of site-use

The intensity of site occupation can be monitored by examining changes in the artefact discard or sediment accumulation rates, and by looking at the degree of trampling or fire damage on stone artefacts (e.g. Hiscock 1985). O'Connor et al. (1993: 101) have argued that consistent changes in artefact discard rates in a number of sites provide reliable signals of regional population change. Their point is illustrated in these examples: artefact discard rates at Mandu Mandu on the Cape Range Peninsula, and at Widgingarri Shelter 2 and Koolan Shelter 2 in the west Kimberley rose dramatically between 25,000 and 18,000 b.p. as aridity increased and populations contracted. The sites then appear to have been abandoned at the height of the Last Glacial Maximum. When the sites were reoccupied in the early to mid Holocene, artefact discard rates were generally lower than those of late-Pleistocene levels, indicating a population decline during the Glacial Maximum. Population levels may have reached their pre-Glacial Maximum levels only in the mid to late Holocene.

The reverse trend is evident at Colless Creek in the Lawn Hills area of the Queensland Gulf country (Magee & Hughes 1982; Hiscock 1984; 1985). Hiscock in particular has focussed on the taphonomic and technological aspects of the distribution of stone sources in relation to palaeo-environmental change. He shows how raw material use and rationing behaviour reflects changes in resource accessibility, both chronologically and geographically.

The Colless Creek Cave sequence spans a minimum of 17,000 years. When the site was first occupied rainfall was higher than today (Macgee & Hughes 1982). However, rates of artefact discard, stone artefact breakage and fire damage, were highest between 16,000 and 13,500 b.p., suggesting that the site was used most intensively in a period of pronounced aridity (Hiscock 1984; 1985: 87-9). At the same time, occupants of the site mainly used stone obtainable within the gorge systems, while previously used stone sources on the surrounding plateau were abandoned. Both lines of evidence suggest that, at a time of pronounced aridity, people were more closely tethered to the permanent water sources in the gorges.

David (1990: 47) notes a similar development at Fern Cave in the Chillagoe area of North Queensland. Fern Cave reflects marginal use before 29,000 b.p., but between 17,000 and 13,000 b.p. the site was used intensively. With the return to higher rainfall conditions at the end of the Pleistocene, the site was abandoned, probably because it became too damp to occupy. The site demonstrates a pattern of site-use and abandonment reflecting not only regional trends but also very site-specific circumstances.

Elsewhere in the tropics, changes in the intensity of site-use are more difficult to discern. At Miriwun in the East Kimberley, for instance, a near-basal radiocarbon determination of c. 18,000 b.p. was obtained just 20 cm below a determination of 3000 b.p. in deposits 110 cm deep (Dortch 1977). The nature of cultural deposition between the Pleistocene and Holocene deposits at the site is unclear. A similar situation occurred at Malangangerr and Nawamoyn in western Arnhem Land, where Schrire (1982) found occupation extending back for a minimum of 25,000 b.p. and of 22,000 b.p. respectively. At both sites, the Pleistocene levels comprised a sand matrix capped by a midden dating to c. 6000-7000 b.p. (Tacon & Brockwell, this volume). Schrire (1982: 85) was uncertain whether the significant time-gap between these Pleistocene and Holocene dated occupations reflected the limited number of radiocarbon determinations, deflation of the deposits or a lengthy period of abandonment. At other sites in the region, such as Nauwalabila I and Malakunanja II, however, the paucity of dated occupation for the Glacial Maximum is almost certainly due to choice of radiocarbon samples by the excavator. There is no stratigraphic or cultural evidence for a hiatus in occupation (Kamminga & Allen 1973; Jones & Johnson 1985).

Resource-use

Very few sites in tropical Australia have evidence for use of faunal and floral resources between 15,000 and 7000 b.p. One exception is the Koolan Shelter in the West Kimberley, where a developed maritime economy is evident from c. 10,400 b.p., when the sea reached the vicinity of the site. It suggests that before this time, people were living on the Pleistocene coast and fell back as the sea rose (O'Connor 1990: 170). At the base of the Koolan midden layers, terrestrial fauna, especially macropods, comprised the bulk of the diet, but with a further rise in sea-level and the isolation of Koolan Island, shellfish and fish became far more important (O'Connor 1990: 166-7). These changes in faunal exploitation reflect resource context rather than economic intensification in response to continued population growth.

Stone artefacts provide the most common evidence for changes in resource-use during the Pleistocene/Holocene transition. However, most Australian stone artefact analyses have identified only a general two-part sequence, with the major change occurring abruptly in the mid to late Holocene. Dortch (1977) noted that the cultural sequence at Miriwun in east Kimberley consisted of an early industry, which included amorphous unretouched and retouched flakes and edge-ground axes, and a late industry, distinguished by the addition of blade technology and bifacially and unifacially flaked points, which appeared about 3000 b.p. In the west Kimberley, O'Connor (1990) recognized the same two-part sequence, with evidence for edge-ground axes back to 28,000 b.p. at Widgingarri.

In western Arnhem Land, Schrire (1982) identified a similar stone artefact sequence, an early industry comprising scrapers, core scrapers, utilized flakes, grind-stones and edgeground axes, and a later industry characterized by unifacial and bifacial points, adzes, utilized flakes and edge-ground axes. Dates of c. 5000 b.p. (Schrire 1982: 239) and c. 5700 b.p. (Jones & Johnson 1985: 206) have been suggested for this change. The abrupt appearance of points into Arnhem Land and Kimberley sequences suggest that they were introduced rather than an in situ development (Allen & Barton 1989: 119).

Frameworks such as this basic two-part sequence are of limited utility in understanding developments in Aboriginal resource-use. The potential of a different approach is illustrated by Cundy (1990), who undertook technological analysis of a two-part industrial sequence from Ingaladdi in the Victoria River region. Cundy suggests that the lower material, dating in this site from c. 7000 to c. 3000 b.p., reflects local patterns of resource-use and distribution. The later industry is associated with a more formalized pattern of procurement involving more off-site reduction and production of standardized, lancet flakes (Cundy 1990: 353). Similar analyses are now required for earlier stone artefact assemblages spanning periods of major climatic change in the late Pleistocene and early Holocene.

Closure of social networks and territorial

bounding

Symbolic evidence such as rock-art, and/or the movement of exotic items and materials, such as edge-ground axes and pigments, will provide crucial evidence for the nature of past social networks and exchange. Unfortunately, at this stage the paucity of well-dated rock-art chronologies and exotic items recovered from excavations means that this potential cannot yet be realized.

Art and decoration are evident in the earliest sites for Aboriginal occupation of tropical Australia. Quantities of high-quality pigment occur in the basal levels of Malakunanja II, dated to between 53,000 and 61,000 years b.p. (Roberts et al. 1994b), but the way these pigments were used is unknown. The earliest claimed date for rock-engravings in the tropics comes from Gum Tree Valley in the Dampier region of Western Australia (Lorblanchet 1992). At this site, marine shell in a fissure among deeply patinated engravings and dated to c. 18,500 b.p., suggests that adjacent rock-engravings may have been produced at the height of the Glacial Maximum. The engravings include ghost-like figures, kangaroos and geometric designs. Even if the contemporaneity between the shell and these engravings is disputed, their patination state strongly suggests that they are Pleistocene in age. Much later, a range of engravings depicting large humans, kangaroos, animal - human composites, birds, and boomerangs was produced by people responsible for deposition of shell middens. These latter engravings are either marginally patinated or fresh in appearance, and the oldest date to c. 7000 b.p.

In the Kimberley and western Arnhem Land there are also rock-paintings and engravings of assumed Pleistocene and early-Holocene age - but this still remains to be proven. The rock-art of western Arnhem Land is particularly well known (e.g. Brandl 1973; Chaloupka 1984; 1994; Lewis 1988; Tacon 1988). Chaloupka (1984) has established a relative sequence for this region based on changes in depicted fauna. Recent X-ray rock-paintings feature wetland and estuarine species (e.g. barramundi fish, estuarine crocodile), while earlier Dynamic Style paintings include depictions of terrestrial and freshwater fauna (e.g. emu, thylacine, possum, species of macropod).

Chaloupka argues that the Dynamic style substantially pre-dates present environmental conditions and the establishment of the modern sea-level some 6500 years ago. Lewis (1988: 84 agrees, noting similarities between the Dynamic figures of western Arnhem Land and the Bradshaw paintings of the Kimberley. He infers that these paintings date to a time of environmental stress, when regional integration would have had survival value. Lewis concludes that these early paintings represent a widespread Pleistocene rock-painting tradition which may date to the Last Glacial Maximum, with 9000 b.p. a minimum age. Tacon & Brockwell reiterate in detail the arguments for a Pleistocene age for the older western Arnhem Land art. Their assumption is that there are correlations between the subject matter of rock-paintings and their environmental contexts, which provide general ages for successive styles. Current research projects involving absolute dating of rock-art in the Kimberley and western Arnhem Land should resolve the chronological problem (e.g. Morwood et al. 1994; Nelson et al. 1995).

There is evidence for extensive exchange networks operating in parts of tropical Australia during the Pleistocene: at Widgingarri baler shell was found in deposits c. 26,000 years old, while pearlshell was dated to c. 18,000 b.p., when the site was some 200 km inland (O'Connor 1990). However, the evidence is too sparse to monitor the way in which these networks may have changed in response to later climatic fluctuations.

Southeast Cape York Peninsula. a case-study

Recent research in southeast Cape York Peninsula provides comparative data for the above general overview of Aboriginal archaeology in the tropics. The study involved excavations at nine sites, analysis and dating of rock-art, resource mapping and collection of a range of information for palaeo-environmental change (Morwood 1989; Morwood a Hobbs 1995.

The study area

Cape York Peninsula forms the northeast corner of the Australian mainland. The southern part of the Peninsula has a diverse range of environments reflecting differences in climate and geology (Morgan et al. 1995. It is dominated by unconsolidated Tertiary and Quaternary plains, but bedrock forms hills along a central spine and in higher areas to the east and southeast (Figure 5).

Vegetation in southeast Cape York Peninsula today is primarily eucalypt woodland. Study of pollen from swamp cores and archaeological sites indicates that similar woodland occurred over the past 32,000 years, but during the Pleistocene understoreys were more open than those found today (Stephens & Head 1995). However, Stephens & Head have evidence for an increase in the number of freshwater swamps in both the coastal and inland sections of the region throughout the Holocene. In historic times, such swamps were the focus for exploitation of a range of plant foods. This progressive increase in carrying capacity and distribution of freshwater sources would have permitted more permanent occupation of many areas.

Resource mapping showed that the upper sections of plateau scarps and sandy outwash plains were particularly rich in plant foods, which formed the staple of the local Aboriginal diet (Morgan et al. 1995). Springs and perched swamps also occur around the plateau scarps, and many are geologically permanent. These would have enabled Aboriginal occupation of some sections of the Laura Basin sandstone country to continue throughout the most arid period of the Last Glacial Maximum. In contrast, water sources in the extensive alluvial and outwash plains of southern Cape York Peninsula are not geologically permanent; Aboriginal occupation of these zones is likely to have fluctuated in response to long and shortterm water availability.

The study concentrated on rock-shelters in the dissected sandstone plateaux, which form the uplands on the southern rim of the Laura River Basin. These plateaux run south from Princess Charlotte Bay then west below the township of Laura. Adjacent resource provinces include outwash Tertiary and marine plains to the north, alluvial plains associated with major rivers, folded sediments to the south, and wet tropics to the southeast.

Archaeological evidence in southeast Cape

York Peninsula

Site numbers

A histogram of dated sites for southeast Cape York Peninsula shows initial occupation of the region occurred prior to c. 32,000 b.p. at Sandy Creek 1 (FIGURE 6). In addition, extrapolation of radiocarbon dates at Sandy Creek 1 and TL dates at Mushroom Rock West suggest human presence by 50,000 b.p. (Morwood et al. 1995a; b). Evidence for local occupation prior to 25,000 b.p. is sparse. The evidence from Sandy Creek 1 and Yam Camp indicates fleeting visits to sites. Mushroom Rock West offers a more detailed picture of changes in the intensity of occupation during this early period, but this must await further dating of the basal deposits.

A date of c. 16,900 b.p. associated with nonbasal occupation at Yam Camp shows that human use of some sections of the plateau scarps occurred during the height of the Last Glacial Maximum (Morwood & Dagg 1995: 112). An increase in site numbers then occurred from 15,000 b.p., the approximate time when Early Man and Magnificent Gallery were first used. Both these sites have underlying sterile deposits, meaning that initial human use is well-documented. The increase in dated sites accelerated throughout the Holocene. In some cases, the first accumulation of cultural deposits can be explained in terms of a geomorphic event, such as the formation of a rockfall dam to retain deposits (e.g. Giant Horse), but in most cases the beginnings of cultural deposition appear to reflect the beginnings of site-use.

In general terms, the pattern of dated sites seems to indicate an increase in local population beginning around 15,000 b.p., and continuing right up to the European contact period. Beaton (1985) has also argued that the mid- to late-Holocene expansion of settlement indicated by first use of the coast around Princess Charlotte Bay, the development of specialized maritime economies and the first use of adjacent islands, reflect population growth @sites such as Waleamini, Endaen, Alkaline Hill, South Mound: see Figure 6).

Intensity of site-use

We argue that the pattern of response at individual sites to changes in climate in southeast Cape York Peninsula should vary depending upon their specific, as well as general, resource contexts (cf. O'Connor et al. 1993). * Sites close to geologically permanent water sources are likely to show increased artefact discard rates as people became more tied to such areas at the height of the Last Glacial Maximum. From 15,000 b. p. such sites may actually show a reduction in intensity of occupation, as the mobility options of local populations expanded.

* Sites distant from permanent water sources away from the Laura Plateau scarps may show abandonment between 25,000 and 15,000 b.p.

* After 15,000 b.p. the general pattern at sites in the region should be one of increased occupational intensity; since subsequent population increase is likely to have occurred progressively, there may be major differences between sites in the timing of this increase. Graphs of occupation intensity at Sandy Creek i and Magnificent Gallery are informative (Figures 7a-b). The former was used sporadically from well before 32,000 b.p., but consistent use of the site did not commence until about 25,000 years ago: Sandy Creek is near a geologically permanent spring, which must have provided a focus for occupation of the area. From the terminal Pleistocene, the pattern of occupation at the site is one of low-amplitude fluctuations, with alternate periods of light and heavy use. Over the past 4000 years, closely spaced, high-amplitude fluctuations in site-use occurred.

In contrast, occupation began at Magnificent Gallery c. 15,000 b.p., but the most intensive use of the site occurred only in the past millennium (Morwood & Jung 1995). Such obvious differences between the long-term patterns of site-use at Magnificent Gallery, Sandy Creek 1 and other sites in the region reinforce the interpretation offered for the cumulative increase in site numbers and more intensive occupation of sites. From the terminal Pleistocene there is more intensive use of sites, as part of a developing pattern of land-use. The archaeological evidence reflects a process not an event.

Resource-use: changes in diet

Archaeological evidence for diet is uncommon in southeast Cape York Peninsula, but at Red Horse, near Cooktown, pandanus drupes in the deposits date from the early Holocene to the European-contact period (Morwood & L'Oste-Brown 1995a). Of particular significance at this site, however, is the evidence for consumption of Cycas nuts, as well as for use of grindstones to prepare plant materials, that is restricted to the uppermost deposits spanning the last 1100 years. Cycas nuts, which contain toxins and require elaborate processing before consumption, by historic times were a late-dry-season staple for local people (Anderson 1984: 102-7; Beaton 1982). The Red Horse evidence indicates a major increase in the intensity of plant exploitation between the early and late Holocene periods. In turn this suggests a corresponding growth in social and domestic demands upon Aboriginal production systems.

Resource-use: changes in stone artefacts

In all excavated sites in southeast Cape York Peninsula, flaked stone artefacts are the dominant evidence for past human occupation and activities. The manner in which the procurement and use of stone for knapping has changed is crucial for inferring changes in more general land and resource-use.

Stone artefact assemblages are complex databases which allow a number of possible analytical approaches. Models are necessary to explain variability in the procurement, use and discard of stone artefacts. Although based on possibly simplistic and incorrect assumptions, such models can be used to make predictions that can be tested against the archaeological record.

In general, low-intensity, episodic use of `early' key sites should be reflected in stone artefact assemblages which indicate opportunistic use of local raw materials, with some refurbishing of curated items, possibly made on exotic materials. From the terminal Pleistocene, however, climatic amelioration was conducive to population growth and more intensive use of specific tracts with associated basecamps, as documented in historic times. Raw materials consistently used for stone artefact manufacture in southeast Cape York Peninsula are abundant and fine-grained in distribution. Suitable water-rolled pebbles are available at most water sources. Even so, increased demand for knapping stone from the terminal Pleistocene may have brought pressure to cut associated procurement costs and for more efficient use. Pressure would probably have been greatest at sites such as base-camps, which were occupied intensively over extended periods. Indices of efficiency could include: * Greater selectivity in procuring suitable stone, with emphasis on better-quality materials with more predictable flaking properties. This would involve more 'testing, and reduction of materials at the source. * More successful knapping strategies involving greater care in platform preparation, overhang removal, establishing core morphology, and predictable flake detachment (e.g. focalized platforms). * Less waste by increasing reduction, rotation and bipolar working of cores; reducing the size of tools; and getting more cutting edge per given weight of raw material by producing blades. * Extending the use-life of artefacts with more retouch. * The development of hafted stone tools. Odell (1994) has argued that when mobility options become more constrained, the hafting of tools to increase their reliability becomes a way of reducing risk. Many of these indices of knapping efficiency are interdependent (Hayden 1989: 10-11) and may correlate with general evidence for environmental carrying capacity and human population levels. Individually, or in combination, they are likely to have characterized the stone artefact sequence in southeast Cape York Peninsula during the Holocene.

Analyses of stone artefact assemblages from southeast Cape York Peninsula have shown a general three-part sequence, comprising the Early, Middle and Recent industries (Morwood & L'Oste-Brown 1995b). The analyses of cores, tools and flakes/pieces, showed very similar trends (Figure 8).

The Early Stone Artefact Industry

This industry is characterized by large flake and core implements made on quartz, silcrete, quartzite and chert (Figure 9). Many pieces have heavy retouch and/or use-wear indicating use in wood chopping and scraping. Some large tools made on local material appear to have been manufactured off-site. Small tools were a major concern of knappers (on the basis of the core evidence), and the small size of some examples indicates probable hafting of quartz flakes. Edge-ground axes were used by the earliest occupants of the region but were relatively uncommon (Morwood &, Trezise 1990; Morwood et al. 1995b).

The Middle Stone Artefact Industry

The first evidence for substantial changes in stone artefact technology occurred at some sites c. 15,000 b.p. The Middle Industry is marked by far greater emphasis on use of fine-grained, good-quality materials, particularly chert. This coincides with the appearance of a new type of core, the burin core (see Cundy 1990: 118; Flenniken and White 1985: 136). Burin cores were used for production of small, thin blades of triangular cross-section, probably for hafting as spear barbs. All other core types also show more emphasis on fine-grained, good-quality materials. Cores and flakes indicate far more care in establishing and maintaining core shapes for efficient and predictable flake production.

Small and medium-sized retouched flake scrapers were the dominant tool type of the Middle Industry (Figure 10). On average these were smaller than those of the Early Industry, and with much more emphasis on the use of chert. Their size and edge shape indicates use as spokeshaves. The first evidence for the hafting and retouching of small chert flakes for wood-working occurs in the upper part of this industry, dating from c. 6000 b.p. (Flood g, Horsfall 1986: 54; Rosenfeld et al. 1981:15). Evidence for use and maintenance of edge-ground axes is relatively common.

Of particular significance is that dates for the appearance of the Middle Industry vary between sites in the region. It appears c. 15,000 b.p. at Mushroom Rock West and Sandy Creek 2, about 8300 b.p. at the near-by Sandy Creek i shelter, and much later at Yam Camp shelter. In fact, at Yam Camp, sporadic low-intensity occupation and expedient use of stone artefacts, characteristic of the Early Industry, continue up to the most recent millennium, when occupation suddenly becomes intensive, chert use increases markedly, and adzes appear (see Pearson 1989).

The Recent Stone Artefact Industry

This industry is characterized by increasing emphasis on use of small, chert flakes for heavy wood-working. These must have been hafted, and some still retain traces of resin (Figure 11). Kamminga (1978: 341-3) examined 25 examples from Mushroom Rock, and concluded that these were for scraping wood. These adzes, became more common and more standardized in form over time@ in particular burren adzes, made on blades, progressively became the pre-dominant wood-working implement.

As with the Middle Industry, dates for the first appearance of the Recent Industry at specific sites appear to have been determined by the context of site-use. Where occupation intensity is high, this industry dates from c. 3800 b.p. (e.g. Mushroom Rock West, Sandy Creek 2). and seems to continue trends already evident in the Middle Industry (i.e. more intensive chert use for wood-working). Where the preceding mid-Holocene occupation is sparse, the Recent Industry does not appear until a late Holocene increase in site-use - and the change is abrupt.

Assessment of the stone artefact sequence

The predictions made about changes in the stone artefact sequence anticipate that from c. 15,000 b.p. stone artefact assemblages in southeast Cape York Peninsula would show greater selectivity in the procurement of stone for knapping, the adoption of more successful knapping strategies, more economical use of cores, greater use of retouch, and more emphasis on hafted curated tools. It was argued also that quantitative measures of efficiency in raw material use would correlate with greater environmental carrying capacity, local population and intensity of individual site-use. Given the evidence for continuing increases in biological productivity and number of reliable water sources from the terminal Pleistocene, this is the expected time-frame for major changes in use of stone for artefact manufacture.

Quantitatively, the Middle Industry, appearing at some sites c. 15,000 b.p., shows far greater selectivity in choice of stone for tool manufacture than the Early Industry - in particular, greater use was made of fine-grained materials with more predictable flaking qualities (see Pearson 1990: 164-5). The reduction in the number of decortication flakes associated with the Middle Industry also suggests that more primary core reduction/testing was done at the source, while the proportion of small flakes, flakes with dorsal median ridges and focalized platforms, show that this was the time when knappers exercised most care in setting up core morphology and detaching flakes. In addition, average core size decreased markedly.

These changes in knapping coincided with greater emphasis upon use of chert for manufacture of hafted spear barbs and wood-working tools. The latter also were retouched more intensively. Later in the upper part of this industry, a few small chert flakes with particularly heavy retouch must have been hafted as adzes. They are functionally equivalent to the large chopper/scraping tools of the Early Industry.

In the Recent Industry, from 3800 b.p., the earlier trend towards intensive use of chert for manufacture of wood-working tools continued, and adzes of standardized burren type appeared. These composite tools with wooden handles and with small, easily resharpened, conveniently transported, readily replaced chert blades had a long use-life. Given the very high quality of cherts used in adzes and the disparity between number of cores and tools (Figure 8), testing of chert nodules and most steps in the production of suitable flakes for adze manufacture appear to have been undertaken off-site.

These developments in retouched tools closely follow the predictions made, both in direction and timing: there is greater selectivity on the part of knappers in raw material use, more efficient use of material, and increasing emphasis on extending the use-life of tools.

Differences between sites in the timing of technological changes seem to have occurred throughout the sequence. Pleistocene stone artefact assemblages are largely characterized by the expedient use of tools. At some sites the change to more selective and economical use of materials, especially chert, occurred as early as 15,000 b.p., at others it occurred in the past 1000 years. A very similar pattern is seen in the adoption of adze technology, which began about 6000 b.p. at sites used relatively intensively, such as Mushroom Rock, but did not appear until the most recent period of intensive occupation at Yam Camp and Magnificent Gallery.

One implication for the staggered introduction of technological changes is that the dynamics of stone procurement and knapping for artefact manufacture are determined by the context of site-use. There are certainly general trends, but their manifestation is site-specific. As differences in site context can be both diachronic and synchronic, expedient use of stone for artefact manufacture could have occurred at any time depending on circumstances. Expedient tools occur in the Recent Industry, but definitely decrease in the uppermost levels at sites, presumably as further population increases occurred and land-use patterns became more formalized.

The technological sequence identified in this analysis seems to have been a result of continuing processes rather than events. The direction and timing of turning-points in the sequence show a close fit with predictions made on the basis of regional population increases associated with a greater number of intensively used locales from the terminal Pleistocene. It is concluded therefore that chronological changes in retouched stone tools reflect continued population expansion in the region, with more intensive use of specific tracts and associated base-camps. This population expansion began c. 15,000 years ago.

Territorial bounding: the symbolic evidence Fragments of used pigment indicating some type of painting activity go back to 30,000 b.p. at Sandy Creek 1 and to the base of Mushroom Rock West. The earliest evidence for rock-paintings in southeast Cape York Peninsula dates to c. 25,000 b.p. at Sandy Creek 2, with other rock-painting events at this site occurring c. 16,000 b.p. and c. 6700 b.p. (Watchman 1993, but the motifs used in these Pleistocene and early-Holocene rock-paintings are now unrecognizable.

The earliest rock-engravings date to c. 13,200 b.p. They comprise pecked panels of bird tracks, lines and geometric motifs at Early Man and Sandy Creek 1 Figure 12; Rosenfeld et al. 1981; Cole et al. 1995). A similar pecked panel of non-figurative and track motifs at Green Ant shelter dates to the early Holocene (flood & Horsfall 1986). The evidence indicates a widespread and long-lived rock-engraving tradition. The relative degree of homogeneity of these panels suggests that one function of this art may have been the `linking, of local teffitorial groups in a relatively low-population-density system (e.g. Yellen & Harpending 1972; Smith 1992).

The mid Holocene may mark the appearance of the regionally distinctive Quinkan rock-painting tradition (Cole et al. 1995; Rosenfeld et al. 1981; Trezise 1971). If so, then ethnographic information on the operation of Aboriginal art systems strongly suggests a change in the general context of rock-art production from an early linking role to one more concerned with territorial bounding, higher population levels and more intensive use of resources (e.g. Morwood 1992; Smith 1991).

As with the development of adze technology, changes in rock-art function after 15,000 b.p. are likely to have been progressive and linked to both the general and specific contexts of production (see also Tacon & Brockwell, this volume). For instance, although figurative paintings predominated late in the sequence, direct dates for some non-figurative panels of engravings show that these continued to play a role in Aboriginal ceremonial life within the past 2000 years. In addition, there is some evidence for early figurative paintings at Magnificent Gallery, where two stylistically distinctive human figures, associated with crosses and hand stencils, may date to c. 10,000 b.p. (Morwood & Jung 1995).

Conclusions

In southeast Cape York Peninsula, site-specific differences in context fe.g. proximity to geologically permanent springs, the nature of site depositional regimes) appear to have been major determining factors for the role of sites in times of climatic change - the Pleistocene patterns of occupation at Sandy Creek 1, Magnificent Gallery, Yam Camp and Mushroom Rock are different from each other. In the same way the archaeological correlates of different resource structures may be more complex than those used by O'Connor et al. (1993) in comparing changes in the archaeological record between three regions in Western Australia. In fact, sites within a region may show more pronounced differences in the pattern of Pleistocene-Holocene occupation than is apparent between sites in very different environmental regions.

The evidence from southeast Cape York Peninsula also sheds new light on the significance of the mid- to late-Holocene changes which occurred across the Australian mainland and relates these to earlier trends evident in the Pleistocene-Holocene transition period. Most of the changes in stone artefact technology in this region appear to have resulted from in situ processes, which commenced after 15,000 b.p. as a result of climatic amelioration and population increase. The dominant trend from the terminal Pleistocene through to the European, contact period in southeast Cape York Peninsula is one of increased efficiency in the manufacture and use of wood-working implements coupled with a shift from on-site to off-site primary reduction of chert cores. Hafted adze technology from 6000 b.p. was a local innovation, and continued the trend towards greater selectivity and more economical use of high-quality material by knappers. In contrast, burren adze technology in other regions to the south and west appears more recently with no obvious precursors in the sequence (e.g. western Arnhem Land). The evidence suggests that southeast Cape York Peninsula was a centre of development for hafted adze technology in Australia.

It is clear that stadial models for change in stone artefact assemblages are not appropriate for tackling some very basic issues in Australian archaeology. Terms, such as `the Australian Core Tool and Scraper Tradition' (Bowler et al. 1970) and `the Australian Small Tool Tradition' (Gould 1977) have served their purpose in drawing attention to shared trends in the archaeological records across Australia - but their underlying assumptions of diffusion do not fit the more complex picture now emerging (see Cundy 1990; Holdaway, this volume). There is no marked mid-Holocene period of transition in the Cape York Peninsula cultural sequence. Instead directional changes began about 15,000 b.p. and did not stop. Post-Pleistocene climatic change appears to have been the initiating factor, but when climatic changes diminished around 6000-7000 b.p., the process had its own impetus. In addition, inter-site differences in the timing of technological changes and the appearance of more intensive occupation at sites in southeast Cape York Peninsula are symptomatic of developing patterns of regional land-use, not the sudden infusion of new ideas. Such differences in timing provide crucial evidence for the nature of behavioural changes manifest in the archaeological record and should be similarly investigated in other regional sequences.

For the general Australian rock-art sequence, a number of pan@continental models have been advanced (e.g. Lommel 1961; McCarthy 1979; Maynard 1979; Mountford 1959). Gross differences in the character of rock-art in different regions were said to reflect its differential development in time and space. Absolute dates for rock-art in southeast Cape York Peninsula now indicate that the situation is far more complex. Rock-paintings have a Pleistocene antiquity in the region, but because of taphonomic factors the vast majority now evidently date to the past few thousand years. Changes in rock-art techniques and motifs used over time also seem to have been changes in emphasis rather than kind. Some figurative paintings may date to the terminal Pleistocene, while some geometric and track engravings are late Holocene in age.

In an overview of the Australian rock-art sequence, Rosenfeld (1993; 77) has argued for two basic categories of rock-markings. The first category comprises a visual system of gesture, as characterized by mechanically produced finger-markings and hand stencils, which can be thought of as having individual significance and which lack cultural constraints on style. The second comprises referential symbols imbued with style (i.e. art), which marked the landscape in a way determined by corporate land-owning groups. The earliest evidence for referential rock-art in southeast Cape York, at this stage, dates to the terminal Pleistocene, possibly reflecting the emergence of tightened social and teffitorial organization. This fits well with the scenario based on other associated archaeological evidence in the region. If the local stone artefact sequence is any guide, then changes in the corporate rock-art system since 15,000 b.p. are also likely to have been gradual, to have resulted from in situ processes rather than diffusion and to reflect both the regional and site-specific contexts of production.

On the other hand, the evidence of Lor-blanchet (1992) in the Dampier region, as well as the speculations of other researchers in Arnhem Land and the Kimberley, suggest that rock-art with a stylistic component may extend back to the Glacial Maximum in those regions of northwest Australia with Complex Figurative rock-art traditions. Multi-disciplined, regional research projects, which include comparative analyses of well-dated rock-art sequences in Cape York Peninsula, Arnhem Land, the Kimberley and other parts of (tropical) Australia should revolutionize our understanding of Aboriginal prehistory.

Acknowledgments. Our research in southeast Cape York Peninsula was funded by the Australian Research Council, the Australian Institute of Aboriginal and Torres Strait Islander Studies, the Queensland Department of Conservation and Heritage and the University ofnew England. We also wish to thank Kathy Morwood for the stone artefact drawings, and Robyn McDougall, Peter Veth, Harry Allen and Jim Allen for their comments on earlier drafts.

References

Allen. H. & G. Barton. 1989. Ngarradj Warde Djobkeng: White Cockatoo Dreaming and the prehistory of Kakadu. Sydney: University of Sydney. Oceania Monograph 37. Allen, J. 1994. Radiocarbon determinations, luminescence dating and Australian archaeology, Antiquity 68: 339-43. Allen, J. & S. Holdaway. 1995. The contamination of Pleistocene radiocarbon determinations in Australia, Antiquity 69: 101-12. Anderson, J.C. 1984. The political and economic basis of Kuku-Yalanji social history. Unpublished Ph.D thesis, Department of Anthropology and Sociology, University of Queensland. Beaton J.M. 1982. Fire and water: aspects of Australian management of cycads, Archaeology in Oceania 17: 51-8. 1985. Evidence for a coastal occupation time-lag at Princess Charlotte Bay (North Queensland) and implications for coastal colonisation and population growth theories for Aboriginal Australia, Archaeology in Oceania 20: 1-20. Birdsell, J.B. 1977. The recalibration of a paradigm for the first peopling of Greater Australia, in J. Allen et al. (ed.), Sunda and Sahul: prehistoric studies in southeast Asia, Melanesia and Australia: 113-67. London: Academic Press. Bowler, J. 1976. Recent developments in reconstructing late Quaternary environments in Australia, in R.L. Kirk & A.G. Thorne (ed.). The origin of the Australians: 55-77. Canberra: Australian Institute of Aboriginal Studies. Bowler, J.M. et al. 1970. Pleistocene human remains from Australia: a living site and human cremation from Lake Mungo, western New South Wales, World Archaeology 2: 39-60. Brandl, E. 1973. Australian Aboriginal paintings in western and central Arnhem Land. Canberra: Australian Institute of Aboriginal Studies. Chaloupka, G. 1984. From palaeoart to casual paintings. Darwin: Northern Territory Museum of Arts and Sciences. Monograph Series 1. 1994. Journey in time: the world's longest continuing art tradition. Port Melbourne: Reed. Chappell, J. 1983. Sea-level changes, 0 to 40 ka, in Chappell & Grinrod (ed.): 121-2. Chappell, J.& A. Grinrod (ed.). 1983. Climanz: proceedings of the First Climanz Conferenc. Canberra: RSPacS, Australian National University. Cole, N.A., A. Watchman & M.J. Morwood. 1995. Chronology of Laura rock-art, in Morwood & Hobbs (ed.): 147-60. Cundy, B.J. 1990. An analysis of the Ingaladdi assemblage: a critique of the understanding of lithic technology. Unpublished Ph.D thesis, Department of Prehistory and Anthropology, Australian National University. David, B. 1990. Fern Cave, rock-art and social formations: rock-art regionalisation and demographic models in southeastern Cape York Peninsula, Archaeology in Oceania 26: 41-57. 1993. Nurrabullgin Cave: preliminary results from a pre-37,000 year old rock-shelter, Archaeology in Oceania 28: 50-4. 1994. A space-time odyssey: rock-art and regionalisation in North Queensland prehistory. Unpublished Ph.D thesis, Department of Anthropology and Sociology, University of Queensland. Dortch, C. E. 1977. Early and late industrial phases in Western Australia, in R.V.S. Wright (ed.), Stone tools as cultural markers: 104-32. Canberra: Australian Institute of Aboriginal Studies. Flenniken, J.J. & J.P. White. 1985. Australian flaked stone tools: a technological perspective, Records of the Australian Museum 36: 131-51. Flood, J.M. & N. Horsfall. 1986. Excavations of Green Ant and Echidna shelters, Queensland Archaeological Research 3: 4-64. Gollan, K. 1984. The Australian dingo: in the shadow of man, in M. Archer & G. Clayton (ed.), Vertebrate zoogeography and evolution in Australasia: 021-7. Hesperian Press. Gould, R. 1977. Puntutjarpa Rock-shelter and the Australian Desert Culture. New York: American Museum of Natural History. Anthropological Papers 54. Hayden, B. 1989. From chopper to celt: the evolution of the resharpening techniques, in R. Torrence (ed.), Time, energy and stone tools: 7-16. Cambridge: Cambridge University Press. Hiscock, P. 1984. Preliminary report on the stone artefacts from Colless Creek Cave, Queensland Archaeological Research 1: 120-51. 1985. The need for a taphonomic perspective in stone artefact analysis, Queensland Archaeological Research 2: 82-97. 1992. Settlement patterns in the Kakadu wetlands: initial data on site size and shape, Australian Aboriginal Studies 1992: 84-9. Hiscock P. & A.P. Kershaw. 1992. Palaeoenvironments and prehistory of Australia's tropical top end, in J. Dodson (ed.), The naive lands: 43-75. Melbourne: Longman Cheshire. Hope, G., P.J. Hughes & J. Russell-Smuth. 1985. Geomorphic fieldwork and the evolution of the landscape of Kakadu National Park, in Jones (ed.): 220-40. Jones, R. (ed.). 1985. Archaeological research in Kakadu National Park. Canberra: Australian National Parks and Wildlife Department. Special Publication 13. Jones, R. & I. Johnson. 1985. Deaf Adder Gorge: Lindner Site, Nauwalabila I, in Jones (ed.): 165-218. Kamminga, J. 1978. Journey into microcosms: a functional analysis of some classes of prehistoric Australian stone tools. Unpublished Ph.D thesis, Department of Anthropology, University of Sydney. Kamminga, J. & H. Allen. 1973. Alligator Rivers Environmental Fact-finding Study: reporl of the archaeological survey. Canberra: Government Printer. Lewis, D. 1988. The rock paintings of Arnhem Land, Australia. British Archaeological Reports. BAR International Series 415. Lommel, A. 1961. Rock-art of Australia, in H.G. Bandi et al. (ed.), The art of the Stone Age: 205-31. London: Crown. Lorblanchet, M. 1992. The rock engravings of Gum Tree Valley and Skew Valley, Dampier, Western Australia: chronology and function of the sites, in J. McDonald & I.P. Haskovic (ed.), State of the art: regional rock-art studies in Australia and Oceania: 39-59. Melbourne: AURA. Occasional Publication 6. Lourandos, H. 1983. Intensification: a late Pleistocene-Holocene archaeological sequence from southwestern Victoria, Archaeologv in Oceania 18: 81-97. McCarthy, F.D. 1939. `Trade' in Aboriginal Australia and `trade' relationships with Torres Strait, New Guinea and Malays, Oceania 10: 171-95. 1979. Australian Aboriginal rock-art. 4th edition. Sydney: The Australian Museum. McConvell, P. 1990. The linguistic prehistory of Australia: opportunities for dialogue with archaeology, Australian Archaeology 31: 3-28. Macknight, C. C. 1976. The voyage to Marege: Macassan traders to northern Australia. Melbourne: Melbourne University Press. Magee, J.W. & P.J. Hughes. 1982. Thin-section analysis and the geomorphic history of the Colless Creek archaeological site in northwestern Queensland, in W. Ambrose & P. Duerden (ed.), Archaeometry: an Australian perspective: 120-28. Canberra: Department of Prehistory, RSPacS, Australian National University. Maynard L. 1979. The archaeology of Australian Aboriginal art, in S.M. Mead (ed.), Exploring the visual art of Oceania: 83-110. Honolulu (HI): University of Hawaii Press. Morgan, G., et al. 1995. The biophysical environment, in Morwood & Hobbs (ed.): 5-17. Morwood, M.J. 1987. The archaeology of social complexity in SE Queensland, Proceedings of the Prehistoric Society 53: 337-50. 1989. The archaeology of Aboriginal art in SE Cape York Peninsula: a research proposal, Rock-art Research 6: 71-2. 1990. The prehistory of Aboriginal land-use on the upper Flinders River, North Queensland Highlands, Queensland Archaeological Research 7: 3-57. 1992. Introductory essay, in Morwood & Hobbs (ed.): 1-6. Moewood M.J. & L. Dagg. 1995. Excavations at Yam Camp, in Morwood & Hobbs (ed.): 107-15. Morwood, M.J. & D.R. Hobbs (ed.). 1992. The ethnogrophy of rock-art. Melbourne: AURA. Occasional Publication 5. (ed.). 1995. Quinkan prehistory: the archaeology of Aboriginal art in SE Cape York Peninsula, Australia. Brisbane: Anthropology Museum, University of Queensland. Tempus 3. Morwood, M.J., D.R. Hobbs & D.M. Price 1995a. Excavations at Sandy Creek 1 and 2, in Morwood & Hobbs (ed.): 71-92. Morwood, M.J. & S. Jung. 1995. Excavations at Magnificent Gallery, in Morwood & Hobbs (ed.): 93-100. Morwood, M.J., S. L'Oste-Brown. 1995a. Excavations at Red Horse, in Morwood & Hobbs (ed.): 116-77. 1995b. Chronological changes in stone artefact technology, in Morwood & Hobbs (ed.): 161-77. Morwood, M.J., S. L'Oste-Brown & D.M. Price. 1995b. Excavations at Mushroom Rock, in Morwood & Hobbs (ed.): 133-46. Morwood, M.J. & P.J. Trezise. 1989. Edge-ground axes in Pleistocene Greater Australia: new evidence from SE Cape York Peninsula, Queensland Archaeological Research 6: 77-90. Morwood, M.J., G.L. Walsh & A. Watchman. 1994. The dating potential of rock-art in the Kimberley, NW Australia, Rock Art Research 11: 79-87. Mountford, C.P. 1959. The rock-art of Australia. Unpublished thesis, Department ofanthropology, Cambridge University. Nelson, D.E. et al. 1995. Radiocarbon dates for beeswax figures in the prehistoric rock art of northern Australia, Archaeometry 37: 151-6. O'Connor, S. 1990. 30,000 years in the Kimberley: a prehistory of the islands of the Buccaneer Archipelago and adjacent mainland, West Kimberley, Western Australia. Unpublished Ph.D thesis, Centre for Prehistory, University of Western Australia. O'Connor, S., P. Veth & N. Hubbard. 1993. Changing interpretations of postglacial human subsistence and demography in Sahul, in Smith et al. (ed.): 95-105. Odell, G.H. 1994. Prehistoric hafting and mobility in the North American Midcontinent: examples from Illinois, Journal of Anthropological Archaeology 13: 51-73. Pearson, W. 1989. A technological analysis of stone artefacts from Yam Camp Surface Scatter and Rock-shelter, SE Cape York Peninsula, Queensland Archaeological Research 6: 91-102. 1990. Prehistoric Aboriginal land and resource use in south east Cape York Peninsula: a technological view. Unpublished BA (Hons.) thesis, Department of Archaeology and Palaeoanthropology, University of New England. Roberts, R.G.. R. Jones & M.A. Smith. 1990. Thermo-luminesence dating of a 50,000-year-old human occupation site in northern Australia, Nature 345: 153-6. 1994a. Beyond the radiocarbon barrier in Australian pre-history. Antiquity 68: 611-16. Roberts, R., et al. 1994b. The human colonisation of Australia: optical dates of 53,000 and 60,000 years bracket human arrival at Deaf Adder Gorge, Northern Territory, Quaternary Science Review 13: 575-83. Rosenfeld, A. 1993. A review of the evidence for the emergence of rock-art in Australia, in Smith et al. (ed.): 71-80. Rosenfeld, A., D. Horton & J. Winter. 1981. Early man in north Queensland. Canberra: Department of Prehistory, RSPacS, Australian National University. Terra Australis 6. Schire, C. 1982. The Alligator Rivers: prehistory and ecology in western Arnhem Land. Canberra: Department of Pre-history, RSPacS, Australian National University. Terra Australis 7. Smith, C.E. 1991. Colonising with style: reviewing the nexus between rock-art, territoriality, and the colonisation and occupation of Sahul, Australian Archaeology 34: 34-42. 1992. Testing the information theory of style: an Australian case study, in Morwood & Hobbs (ed.): 39-45. Smith, M.A., M. Spriggs & B. Fankhauser (ed.). 1993. Sahul in review. Canberra, Department of Prehistory, RSPacS, Australian National University. Stephens, K. & L. Head. 1995. Palaeo-ecology of archaeological and swamp sites in SE Cape York Peninsula, in Morwood & Hobbs (ed.): 18-32. Tacon, P.S.C. 1988. Identifying fish species in the recent rock-art paintings of Western Arnhem Land, Rock-art Research 5: 3-15. 1989. From Rainbow Snakes to `X-Ray' Fish: the nature of the recent rock painting tradition of Western Arnhem Land. Unpublished Ph.D thesis, Department of Prehistory and Anthropology, Australian National University. Thomson, D.F. 1939. The seasonal factor in human culture, Proceedings of the Prehistoric Society 5: 209-21. Tindale, N.B. 1979. Aboriginal tribes of Australia - their terrain, environmental controls, distribution, limits and proper names. Berkeley (CA): University of California Press. Trezisr, P.J. 1971. Rock-art of South-east Cape York. Canberra: Australian Institute of Aboriginal Studies. von Sturmer, J.R. 1978. The Wik region: economy, territoriality and totemism in western Cape York Peninsula, North Queensland. Unpublished Ph.D thesis, Department of Anthropology and Sociology, University of Queensland. Watchman, A. 1993. Evidence of a 25,000 year-old pictograph from northern Australia, Geoarchaeology 6: 465-73. Woodroffe, C.D., J. Chappell & B.G. Thom. 1988. Shell middens in the context of estuarine development, South Alligator River, Northern Territory, Archaeology in Oceania 23: 95-103. Yellen, J. & H. Harpening. 1972. Hunter-gatherer populations and archaeological inference, World Archaeology 4: 244-53.