Footprints in the sand: appraising the archaeology of the Willandra Lakes, western New South Wales, Australia.

Footprints in the sand: appraising the archaeology of the Willandra Lakes, western New South Wales, Australia. Introduction The discovery of human footprints dating to between 19 and 23ka inthe Willandra Lakes World Heritage Area (Figure 1) points again to thepre-eminence of the Willandra Lakes for the preservation of a humanrecord for Pleistocene Australia (Webb et al. 2006). The discovererscomment that '... This site offers a unique glimpse of human livingin the arid inland of Australia at the height of the last glacialperiod' (2006: 405). Here, we argue that, like the footprints site,the human activities which produced the archaeological record of thePleistocene Willandra Lakes consist almost entirely of'unique' events. The first part of this paper critiques the assumption thatAboriginal people during the Pleistocene acted in ways similar to thoseobserved by nineteenth-century ethnographers and the assumption thatarchaeological evidence can be analysed in a cumulative fashion, fittedtogether like a jigsaw, to provide a composite model of humanutilisation of the landscape. A corollary is the belief that the scaleof human and landscape events are somehow on a par with each other(Balme & Hope 1990). The second half of the paper advocates a revised methodology, basedon recent studies of semi-arid regions elsewhere in New South Wales, forsurvey and recording of archaeological data for the Willandra LakesWorld Heritage Area. It is argued that such a methodology is necessaryif Australian archaeologists wish to base their conclusions on reliableempirical data. In particular, we argue that archaeologists must be ableto obtain information on the movement of people and materials acrossthat landscape through time, including the duration, intensity andfrequency of resource and locality use. The scale of sample required togain reliable answers to questions of landscape use requires the studyof large surface exposures of thousands of artefacts and hundreds ofarchaeological features, and, as a consequence, archaeologists mustadopt a methodology capable of utilising samples of this size. [FIGURE 1 OMITTED] Archaeology of the Willandra Lakes World Heritage Area--a critique Archaeological work carried out in the Willandra Lakes area issummarised by Johnston and Clarke (Johnston & Clarke 1998; Johnstonet al. 1998) supplemented by studies carried out elsewhere in thecentral Murray-Darling basin (e.g. Balme 1995; Hope 1993) which allowthe Willandra Lakes archaeological record to be placed within a regionalcontext. Studies by Allen, Clark, Hope, Johnston, Jones, McBryde, McIntyre,Mulvaney, Shawcross, Thorne, Webb, a multitude of BA (Hons) and MAstudents from The Australian National University, and most recently, byWebb, Cupper and Westerway (Webb et al. 2006) contribute to our presentunderstanding and provide the basis for this reassessment. However, inspite of the number of studies carried out there remains a perception ofunfulfilled potential regarding the archaeology of the Willandra Lakes.Mulvaney and Kamminga (1999: 199) observe '... Despite the wealthof archaeological sites, the nature of settlement and subsistence ispoorly understood.' Difficulties apply in varying degrees to allprevious surveys and analyses. The use of a comparative ethnographic framework based onnineteenth-century observations of Aboriginal subsistence activitiesfrom the central Murray-Darling basin, or from contemporary observationsmade elsewhere, is problematical. Allen (1990; 1998) discounts specificparallels, but the problem is deeper than this. It rests at the centreof archaeological expectations whenever the record is organisedconceptually in terms of sites, quarries, middens, living floors, basecamps and dinner-time camps. It is exacerbated when such units arelinked into patterns of presumed behaviour and applied to models ofAboriginal use of the Pleistocene landscape (e.g. Balme 1983; 1995;Balme & Hope 1990; Bowler et al. 1970). Analyses of stone artefacts pay insufficient attention to the preand post-depositional processes that affect assemblage composition (suchas raw material form, supply, and reduction intensity). Some of theseissues are raised in more recent studies by Hiscock and Allen (2000),Muhlen-Schulte (1985), Shawcross (1998), and Williams (1991). Theproblem emerged most famously when the artefacts from the Mungo 1 sitewere used to define the 'Australian core tool and scrapertradition' without sufficient attention paid to the nature ofcollection, the site, its contents or the methods of recording (Bowleret al. 1970). Shawcross (1998: 198-9) raises the possibility ofconsiderable reworking of the sediments from which the Mungo 1 artefactcollection was derived. He, and others (e.g. Hiscock & Allen 2000),suggest that the caching of raw materials on the lunettes might be asvalid an explanation as artefact production and use. The existence of a well-defined environmental framework for theWillandra Lakes (Bowler 1998) has provided archaeologists with a falsesense of the scale of human-environment interactions. The landscapehistory of the Willandra Lakes has been reconstructed from multipleexposures of sediments and sequences (e.g. Bowler & Price 1998).However, archaeologists have failed to understand that the lunettes anddune systems of the Willandra Lakes system are the local expression ofcontinental and at times global processes. Even within the Willandra,lunette building and lake infilling involved the movement of millions oftonnes of sediment and water. Furthermore, sedimentary exposures maycover distances from hundreds of metres to kilometres, while processesof soil pedogenesis may extend over thousands of years (Bowler 1998:147-8). The difficulties lie not so much with the framework itself, aswith the failure of archaeologists to fully comprehend the extraordinarydifferences in temporal and spatial scale between the archaeologicalsignature of small middens, hearths and artefact scatters, and thegeomorphological processes that created the settings within which thesenow occur. Certainly, a major advance in archaeological methodology inthe Willandra Lakes area came with the realisation that each exposurerequired taphonomic and sedimentary analysis before the evidence couldbe considered reliable (Balme 1995; Balme & Hope 1990; Hope et al.1983; McIntyre & Hope 1978; Walshe 1998). To a degree, thesetaphonomic studies were designed to deal with the problems of open,surface sites and the possibility of mixing of materials (Allen 1990;Balme 1995; Hope et al. 1983). However, these problems are of lesssignificance than the failure to make full allowance for the small scaleof Aboriginal activities occurring at these localities compared with thetemporal and spatial scale at which geomorphic processes are measured. Much of the archaeological record in the Willandra Lakes consistsof small shell middens (Johnston 1993; McIntyre & Hope 1978). Oftenthese occur singly, but even when 10 to 20 or more of these heaps occurwithin close proximity, there is little evidence that the occupationswere contemporaneous (Johnston 1993: 200). This is also true of thehearths, fire-hardened remnants of earth ovens that are often associatedwith artefact deposits or middens. An example might be Jane Balme's comprehensive study of theLake Tandou middens. She surveyed 25 middens of which seven (28 percent) were poorly defined scatters and 11 (44 per cent) were small, welldefined, middens, consisting only of shell. Of the remaining seven withmore than a single dominant species represented, only two middens (8 percent) contained a broad range of fauna with some in situ materials(Balme 1995: 4-5). The dating of these middens showed two clusters, oneat c. 15ka, and between 22ka and 27ka for the others (Balme & Hope1990: 90). Both this, and the further evidence from the Darling basin,is a slim basis on which to frame a conclusion that there were largesocial groups in the Pleistocene with '... highly organised foodcollecting strategies which at times allowed large gatherings ofpeople" (Balme 1995:19). It is interesting in this context thatBalme's study did not include the numerous hearths and stone toolscatters that were also present (Balme & Hope 1990: 86). Conclusions regarding site size, frequency, duration and intensityof occupation require multiple lines of evidence to coincide. WhileBalme (1995: 13) makes use of ethnographic concepts, such as the'dinnertime camp', she is one of the few archaeologists toconclude that the Aboriginal occupants of the Pleistocene WillandraLakes demonstrated a subsistence pattern that was quite dissimilar tothat of their nineteenth-century counterparts, involving the absence of'base camps' and the opportunistic targeting of single species(Balme 1995: 15, 18). Balme's work has many things to recommend it, especially herdiscussion of the size of middens and the rapidity of theirdisintegration following exposure and erosion. She found that erosiondestroyed existing middens and uncovered new ones within a period asshort as 18 months (1991: 5; 1995: 3; see also Hope et al. 1983). Thusthe proximity of middens in space around the lakes is as much anartefact of contemporary erosion as it is of Aboriginal behaviour (seeFigure 2). Erosion differentially affects some parts of the Willandra systemmore than others. This is true of the Mulurulu and Mungo lunettes, andthe backshore areas between Lakes Garnpung and Leaghur and Mungo (Figure2), where erosion has produced wide exposures of archaeological material(Johnston & Clark 1998: 114-5). Elsewhere, largely on clay lunettes,exposures are restricted to chance finds through gully erosion. Thereare also differences between sandy and clay facies linked to differingsalinities in the Pleistocene lakes. The relative absence of middensfrom the clay lunettes might reflect this. However, the differentialfrequency of middens in clay and sandy lunette deposits might alsoreflect the mobility of sandy deposits and consequent higher visibilityof exposed archaeological materials compared to those located in claysediments (Bowler 1998: 134-6). The archaeological sites of the Willandra Lakes are small andephemeral, as are those of the wider Darling area. Balme and Hope (Balme& Hope 1990; Hope 1993) analysed the radiocarbon determinations formiddens across the region and found a series of pulses that reflectedthe presence of freshwater within the Willandra, Talyawalka, Darling,central Murray and Anabranch systems (see Figure 1). They note theearliest pulse at the Willandra Lakes (28ka to 40ka), an intermediateone at Tandou (Anabranch 20ka to 28ka), an absence of evidence for theLGM period (16ka to 20ka), followed by widespread evidence along theDarling, Anabranch and central Murray 12ka to 16ka. Finally there areage estimates for 8ka to 12ka on the Talyawalka Lakes. These data arebrought together within a regional framework and interpreted as showingan almost continuous use of aquatic resources in the Murray-Darlingbasin since about 36ka BP (Balme & Hope 1990: 85, 97-9). Clearly, when lakes are dry, shellfish collection is impossible.Such behaviour is thus environmentally determined. On the other hand,given the smallness of the middens, their limited contents and theirephemeral nature, continuity of shellfish use is only of the mostgeneral nature, similar to the factual but general statement that peopleflaked silcrete throughout the Late Pleistocene and Holocene. Theevidence for non-linear, non-continuous use is at least as good as theevidence for continuity (see below). Finally, analysing the data inblocks of one or two thousand years each and locating this midden datawithin a regional-scale framework is problematic. The evidence forapparent continuity is as much a response to the way the data is broughttogether as it is to an archaeological reality. The archaeological record from the Willandra Lakes for the period15ka to 40ka BP consists of numbers of small archaeologicaloccurrences--middens, hearths, bone and artefact scatters--exposedthrough erosion. The accumulation of these data by previous workersshould provide the basis for understanding the human use of theWillandra Lakes in the past. However, locating these facts within avariety of larger scale frameworks, such as the Willandra Lakeshydrological system, the regional framework of the centralMurray-Darling basin, the sequence of stone tool typological changes,environmental cycles of lake filling and drying, or ofethnographically-derived models of seasonal site use and movement (Allen1972; Jones & Allen in Bowler et al. 1970; Mulvaney & Kamminga1999) overwhelms what is still a weak archaeological signature. Thismeans that the majority of the conclusions which have been drawn fromthe archaeology of the Willandra Lakes owe their origin to theframeworks of interpretation used to draw the data together rather thanto the archaeological data we have to hand. [FIGURE 2 OMITTED] We now set out a methodological approach to the archaeology of theWillandra Lakes which allows the archaeological signature to beamplified, but not overwhelmed by inappropriate interpretive frameworks.We explore ways of bringing the data together to generate reliablearchaeological conclusions. Finally, we note that most of the requiredinformation can be obtained without removing archaeological materialsfrom the Willandra area. Outside of the dating of hearths and middens,destructive archaeological techniques--excavations and collections--arelikely to be the exception rather than usual archaeological practice. The late Holocene archaeological record of western New South Wales Over the past 12 years, members of the Western New South WalesArchaeological Program (WNSWAP) have investigated the late Holocenearchaeology of western NSW (Holdaway et al. 2004). Although both thetime period dealt with and the specific geomorphological conditions ofthe areas investigated differ from the Willandra Lakes, the results ofthis research help to cast some light on the interpretative problemswith which researchers must grapple. It is therefore pertinent to reviewsome of the outcomes of the WNSWAP research in the light of thedifficulties with the Willandra archaeological research discussed above. Settlement systems? Much of the interpretation of late Holocene archaeology is based onthe definition of settlement systems because ethnographic work indicatesthat Aboriginal people were mobile and that critical resources weredispersed in both time and space. The archaeological analysis of spaceis therefore considered essential in understanding late Holoceneprehistory. In most cases, space is assessed archaeologically byidentifying a set of sites that can be broadly considered to becontemporary, allowing the definition of settlement systems. The resulting settlement systems should, in theory, differentiatethrough time and across space, permitting the creation of a regionalprehistory. In practice, the construction of settlement systems facestwo related problems. First, the specificity of the resulting systemscan be called into question. A comparison of settlement systems inAustralia and North America finds an alarming degree of similaritybetween the two continents, a similarity that is obviously not theresult of a common cultural heritage or a common set of environmentalconditions (Holdaway & Wandsnider 2006). The implication is thatthis similarity is therefore methodological in origin, probably theresult of combining sites where deposition is so different that theresulting patterns are trivial in their explanatory power. Second, theresult of research in western NSW shows that proximity in space does notmean closeness in time (Fanning et al. in press; Holdaway et al. 2005).Locations that occur within a few kilometres of each other can have verydifferent geomorphic ages. Most archaeologists accept thatcontemporaneity is a relative concept but they do not take sufficientnote of this. This problem is exacerbated in older sites where archaeologicalchronologies are much less precise. As discussed above, sites that areclose together may appear to be contemporary but we must acknowledgethat, with the range of dating systems available, sites (or indeedcomponents within sites) could easily be separated by several centuries.The truth is that we have no way of assessing if sites are contemporarybut must learn to develop analyses that are not dependent on thisattribute. Otherwise, we run the risk of defining hunter-gatherersettlement systems that work in all places and at all times. Matching geomorphic process with archaeological interpretation One of the consequences of an imprecise chronology is the tendencyto see change in the past as slow and gradual. Punctuated events,perhaps represented by gaps in the record, can be hard to define, themore so since materials from which to obtain an age estimate are rare.It is too easy to think of age estimates ('radiocarbon dates')as though they represent sidereal dates and to maintain the fiction thatoccupation was continuous between these 'dates'. Our researchinto the western NSW Holocene archaeological record shows that'slow and gradual' masks a series of almost chaotic eventsthat cannot be interpreted as the outcome of equilibrium processes butmore likely represent processes operating at different spatial andtemporal scales (Fanning et al. 2007). In the rangelands of western NSW,environmental change is frequently episodic in nature. At a coarserchronological scale the cumulative effect of non-linear events isreflected in discontinuities in the occupation of individual locationsand regions (Holdaway et al. 2002; 2005). Aboriginal people also had todeal with environmental perturbations that, in at least some instances,led to localised abandonment. At Nundooka (Figure 1), for instance,Aboriginal occupation is not evidenced until several centuries after theestimated age of a 'once in a 1000 year' flood event. Thispresumably reflects the time needed for the location to becomeattractive enough for re-occupation to be countenanced (Fanning et al.2007). For the archaeologist, the discontinuous and non-equilibrium natureof the archaeological record poses a series of challenges. Individuallocations may be expected to provide markedly different geomorphichistories to one another and may also be expected to have complexoccupational histories as Aboriginal people coped with episodic changesthat affected local resources and even the physical nature of particularlocations. At Willandra Lakes, archaeologists face a similar series ofchallenges to match the nature of behavioural inference with the complexformational histories. Reasons for occupation and abandonment of sitesmay also vary, possibly between locations that are geographically quiteclose. As we have found for the Holocene, constructing meaningfulsettlement system models is likely to be a complex business. Patterns and the archaeological record In an episodic, non-equilibrium environment, pattern in thearchaeological record comes from repeated but variable use of locationsthrough time. Analytically, archaeologists are faced with the challengeof interpreting artefact assemblages that result from an unknown numberof potentially different occupations. From the Holocene record ofwestern NSW we now have artefact assemblages from a number of differentlocations that display varied depositional histories. Our approach toassessing the types of behavioural inferences that can be derived fromthese assemblages is comparative (e.g. Holdaway et al. 2006). Based onchronologies derived from the age of sediments on which artefacts restand the age of heat retainer hearths, we are able to define locationsthat were the subject of longer or shorter periods of deposition. Theartefact assemblages from these different locations show quite differentcharacteristics. If the archaeological record imitated the ethnographic record,assemblages from locations that have the shortest period of artefactaccumulation should produce patterns that are among the easiest tointerpret. In fact, our results show the opposite effect. Artefactsrecorded at sampling locations like Burkes Cave (in the Scopes Ranges)and Stud Creek (in Sturt National Park) both of which have age estimatesthat indicate artefact accumulation over a period of around 2000 years,show patterning related to raw material access, a pattern that is widelyidentified elsewhere in Australia (Holdaway et al. in press; Shiner etal. 2005; 2007). Artefacts found at such places, while treated togetheranalytically as assemblages, in fact come from a series of unrelatedoccupations. When combined they show patterns that reflect theirdistances from raw material sources. Indices show elevated degrees ofcore reduction and tool re-sharpening when raw material sources aredistant. The patterning is less distinct, however, in artefacts that haveaccumulated over relatively shorter periods of time, in the case of theND location at Fowlers Gap, for instance, over a period less than amillennium in duration. In these assemblages, the values of indices thatreflect the degree of core reduction and the degree of toolre-sharpening do not make sense when distance to raw material source isconsidered. Based on our results, periods in excess of 1000 years arerequired if the accumulated pattern is to be easily interpreted.Assemblages that have accumulated over short periods are'messy' in the sense that patterns predicted by processes suchas differential raw material access are difficult to demonstrate. Pattern generation therefore appears to be a function of repeatedoccupation over time periods well in excess of those related to theethnographic experience. If our results hold for the Pleistocene recordat the Willandra Lakes, one might expect short duration occupations tobe the least patterned and therefore the hardest to interpretarchaeologically. This is certainly the case with the shell middens(Balme 1995). In contrast, those locations that have accumulated overlonger periods might be expected to show patterns that reflect moresystematic processes (e.g. Hiscock & Allen 2000). These results raise some interesting problems for archaeologists.While raw material access is a consistent and relatively easilyidentifiable pattern, its very ubiquity raises questions about itsinterpretive significance. Difficult-to-interpret assemblages (describedas 'messy' above) represent a quite different potential forassessing past behaviour from assemblages accumulated over longerperiods. The problem, of course, is how to interpret them if currentanalyses are unable to provide clearly identifiable patterns. Variation across space As with temporal variation, common sense might lead us to expectthat small-scale spatial distributions at the scale of the site or at anintra-site scale are more likely to show pattern compared todistributions over larger areas at a regional scale. Once again, ourresults question this assumption. At a geographic scale commonly thoughtof as 'intrasite', small scale geomorphic processes are adominant source of spatial pattern (Fanning & Holdaway 2004). Thepalimpsest nature of artefact assemblage accumulation together withpost-depositional changes obscures intrasite patterns so that there waslittle or no evidence of the type of spatial patterning suggested forsome of the Willandra and Darling middens (e.g. Balme 1995). Wherespatial pattern is apparent is in the distribution of assemblages alongthe length of drainage systems. For example, along Rutherfords Creeknear Peery Lake (Figure 1), assemblage composition is measured byindices related to core reduction and shows variability across a lineardistance of some 10km. Age estimations from hearth excavations indicatea discontinuous record accumulating over at least the last 2000 years.Explanations for this pattern will reflect local raw materialavailability and artefact types that were abandoned within the valley aswell as those that were abandoned elsewhere (Douglass et al. in press). Human remains As with analysis of the WNSWAP data, the analysis of human remainsfrom western NSW raises serious issues for archaeological analysis. Muchcontroversy concerns the dating and significance of individual humanremains (e.g. Lake Mungo Hominids 1 and 3, Bowler & Magee 2000;Thorne & Curnoe 2000: 598). When the totality of human remains fromthe Willandra Lakes, consisting of eroded scattered human remains, andthe occasional in situ burial, are discussed, however, differentconclusions are drawn (Webb 1989). In this sort of record, formal typesof both burials and morphology are arbitrary, imposing an external andoften synchronic ideal on a complex accumulation history. The analysisof a large sample of surface or recently uncovered human burialsprovides a more secure basis for generalisation than does a small sampleof excavated, in situ burials taken on their own. What appeared at firstto be an impediment to outcomes, the requirement by Traditional Ownersthat human remains be analysed in situ, has in fact directed attentionaway from the remains themselves in favour of the taphonomic, temporal,spatial and behavioural implications of the record (Littleton 2000;2002; 2007; Littleton & Allen 2007; Pardoe 1988). It is therepetition of patterning across a wide landscape through time thatreveals insights into the nature of Aboriginal occupation (Littleton1999; 2007; Pardoe 1995). Destructive and non-destructive techniques Erosion at Mungo is responsible for the exposure of quantities ofthe archaeological record unrivalled in many other world locations. Thisexposure is at once of benefit to archaeologists but also a conservationnightmare since the record is ultimately limited in its extent andexposure ensures destruction. It behoves us as archaeologists, togetherwith Traditional Owners, to gain as much as possible from a process thatwe cannot halt (Fanning 1994; 1999). Shawcross (1998) comments on thetwentieth-century archaeological techniques applied to the study ofburied and surface artefact exposures at Mungo. These involved bothsurface collection and subsurface excavation with objects returned toinstitutions for study. The wealth of artefacts recorded from someplaces combined with the complexity of the location of artefacts atothers meant that analysis took years or even decades to complete. Thisslowed the pace of research and made a reflective relationship betweenartefact analysis and field methodology difficult to achieve. There wasjust too great a separation of time between the time of excavation andanalysis for the one to inform on the other. As well, the later part ofthe twentieth century saw a change in the perception of archaeologistsas the sole custodians of the archaeological record. With the rise ofthe Indigenous voice, the right of archaeologists to remove culturalitems from their landscape context was challenged. Archaeologicalmethodologies must prove themselves capable of meeting these newcircumstances of research. The problems these conditions posed for twentieth-centuryarchaeology can be solved through the adoption of twentyfirst-centurytechnology. No longer must data recording be undertaken by removingobjects to institutions. Context is recordable in the field with a rangeof electronic Total Survey Stations and GPS technologies. Precisecontext, when combined with detailed geomorphic mapping at a variety ofscales, provides a degree of understanding about the effects of erosionon the preservation of the record (e.g. Fanning & Holdaway 2004;Holdaway et al. 2000). Artefacts exposed because of erosion can beanalysed and immediately returned to the place where they were found.This is also true for human remains, which can be analysed in situ andrecorded in the field (Littleton 1999). New imaging technologies basedon laser scanning together with high resolution casts means thatpermanent recording is possible in the field. Handheld computers makeartefact data acquisition rapid and reduce observer error. GIS softwarepermits the rapid integration of multiple datasets acquired with a rangeof instruments. The results are very rapid data acquisition--what oncetook months of recording can now be completed in days (Shiner et al.2005; 2007)--and the ability to deal with very large numbers ofartefacts either in single surface concentrations or in large numbers ofsmaller concentrations. This fundamentally changes the scale at which analysis can proceed.There is no longer the need to separate low resolution, spatiallyextensive 'survey' from high resolution, local intensivestudies of formation processes. Both sets of observation can be combinedand applied at the scale of drainage systems. The results of these fieldtechniques provide archaeologists with large numbers of observations onobjects (artefactual or otherwise). The quantity of data permits therecognition of patterns in lithics not previously identified like thosediscussed above (e.g. Shiner et al. 2005; 2007) as well as patterns inthe distribution of age estimates (e.g. Holdaway et al. 2002). Based onthe WNSWAP experience, the identification of pattern at the WillandraLakes will require that techniques for recording locally concentrated aswell as spatially dispersed distributions of artefacts be adopted. Allthis is achievable without the need to remove artefacts from where theyare found. Conclusion At Willandra Lakes, decades of research have revealed a wealth ofprocesses operating at a variety of different chronological and spatialscales, all of which affect the archaeological record. Archaeologistsmust develop methods that take the scale of these processes intoconsideration. Small shell middens, like single hearths or individualburials may be the product of single events but on their own theyprovide relatively little information. When they are grouped together,they cease to be interpretable as events and therefore cannot beinterpreted in ethnographic terms. Instead they form time averagedassemblages (Stern 1994) and so must be interpreted accordingly. Based on the results of research on the Holocene record, it islikely that pattern exists in the distribution of locations, theirchronology and the composition of artefact assemblages but thesepatterns may not be interpretable in intuitively obvious ways. Shortterm accumulations may be less patterned than those that haveaccumulated over longer periods of time. Spatial pattern may be clear atthe level of the drainage system but absent at the level of a camp. Whatappears to represent continuities in occupation through time may in factmask nonlinear, discontinuous patterns of habitation. There is unlikelyto be a simple relationship between the spatial location of artefactsand their age or a simple relationship between the location of artefactsand the time over which they have accumulated. The definition ofsettlement systems will have to deal with all the temporal complexity ofthe archaeological record if the result is to be other than trivial. This represents a formidable set of challenges for archaeologistsbut they are ones that are achievable, as the WNSWAP research shows. Theadoption of appropriate conceptual and methodological tools offers greatpotential for the continued analysis of the archaeology of the WillandraLakes World Heritage Area and what it can tell us about Pleistocene lifein Australia. Acknowledgements This paper follows our participation in the Legacy of an Ice AgeConference, a part of the Mungo Festival to celebrate the 25thAnniversary of the Willandra Lakes World Heritage Area. The conferencewas organised by Jim Bowler and Matt Cupper and held at Mungo Lodge,September 6-9, 2006. We would like to thank the representatives for thetraditional Aboriginal owners of the Willandra Lakes area for theirwelcome, and the referees for their comments. Peter Quin producedFigures 1 and 2. Received: 8 March 2007; Accepted: 21 June 2007; Revised: 8 August2007 References ALLEN, H. 1972. Where the crow flies backwards: Man and land in theDarling Basin. Unpublished PhD dissertation, Australian NationalUniversity. --1990. Environmental history in southwestern New South Walesduring the Late Pleistocene, in C. Gamble & O. Softer (ed.) TheWorldat 18000 BP." Volume 2 Low Latitudes: 296-321. London: UnwinHyman. --1998. Reinterpreting the 1969-I972 Willandra Lakes archaeologicalsurveys, in H. Johnston, E Clark & J.P. White (ed.) Willandra Lakes:People and Palaeoenvironments (Archaeology in Oceania 33): 207-220.Sydney: University of Sydney, with National Parks and Wildlife Service,New South Wales, and Willandra Lakes World Heritage Area ManagementCouncil. BALME, J. 1983. Prehistoric fishing in the Lower Darling, westernNew South Wales, in C. Grigson & J. Clutton-Brock (ed.) Animals andArchaeology 2. Shell Middens, Fishes and Birds (British ArchaeologicalReports International Series 183): 19-32. Oxford: British ArchaeologicalReports. --1991. The antiquity of grinding stones in semi-arid western NewSouth Wales. Australian Archaeology 32: 3-9. --1995. 30,000 years of fishery in western New South Wales.Archaeology in Oceania 30: 1-21. BALME, J. & J. HOPE. 1990. Radiocarbon dates from midden sitesin the lower Darling River area of western New South Wales. Archaeologyin Oceania 25: 85-101. BOWLER, J.M. 1998. Willandra Lakes revisited: environmentalframework for human occupation, in H. Johnston, P. Clark & J.P.White (ed.) Willandra Lakes: People and Palaeoenvironments (Archaeologyin Oceania 33): 120-55. Sydney: University of Sydney, with NationalParks and Wildlife Service, New South Wales, and Willandra Lakes WorldHeritage Area Management Council. BOWLER, J.M. & J. MAGEE. 2000. Redating Australia's oldesthuman remains: a sceptic's view. Journal of Human Evolution 38:719-26. BOWLER, J.M. & D.M. PRICE. 1998. Luminescence dates andstratigraphic analyses at Lake Mungo: review and perspectives, in H.Johnston, P. Clark & J.E White (ed.) Willandra Lakes: People andPalaeoenvironments (Archaeology in Oceania 33): 156-68. Sydney:University of Sydney, with National Parks and Wildlife Service, NewSouth Wales, and Willandra Lakes World Heritage Area Management Council BOWLER, J.M., R. JONES, H. ALLEN & A.G. THORNE. 1970.Pleistocene human remains from Australia: a living site and humancremation from Lake Mungo, western New South Wales. World Archaeology 2:39-60. DOUGLASS, M.J., S.J. HOLDAWAY, P.C. FANNING & J.I. SHINER. Inpress. An assessment and archaeological application of cortexmeasurement in lithic assemblages. American Antiquity. FANNING, P.C. 1994. Long term contemporary erosion rates in an aridrangelands environment in western NSW, Australia. Journal of AridEnvironments 28: 173-87. --1999. Recent landscape history in arid western New South Wales,Australia: a model for regional change. Geomorphology 29: 191-210. FANNING, P.C. & S.J. HOLDAWAY. 2004. Artifact visibility atopen sites in western New South Wales, Australia. Journal of FieldArchaeology 29(3-4): 255-71. FANNING, P.C., S.J. HOLDAWAY & E. RHODES. 2007. A geomorphicframework for understanding the surface archaeological record in aridenvironments. Geodinamica Acta 20: 275-86. --In press. A new geoarchaeology of Aboriginal artefact deposits inwestern NSW, Australia: establishing spatial and temporal geomorphiccontrols on the surface archaeological record. Geomorphology. HISCOCK, P. & H. ALLEN. 2000. Assemblage variability in theWillandra Lakes. Archaeology in Oceania 35: 97-103. HOLDAWAY, S.J. & L.A. WANDSNIDER. 2006. Temporal scales andarchaeological landscapes from the eastern desert of Australia andintermontane North America, in G. Lock & B. Molyneaux (ed.)Confronting Scale in Archaeology: Issues of Theory and Practice:183-202. New York: Springer. HOLDAWAY, S.J., P.C. FANNING & D.C. WITTER. 2000. PrehistoricAboriginal occupation of the rangelands: interpreting the surfacearchaeological record of far western New South Wales Australia. TheRangelands Journal 22: 44-57. HOLDAWAY, S.J., P.C. FANNING, D.C. WITTER, M. JONES, G. NICHOLLS& J. SHINER. 2002. Variability in the chronology of Late HoloceneAboriginal occupation on the arid margin of southeastern Australia.Journal of Archaeological Science 29: 351-63. HOLDAWAY, S.J., J. SHINER & P.C. FANNING. 2004.Hunter-gatherers and the archaeology of the long term: an analysis ofsurface stone artefact scatters from Sturt National Park, New SouthWales, Australia. Asian Perspectives 43: 34-72. HOLDAWAY, S.J., P.C. FANNING & J. SHINER. 2005. Absence ofevidence or evidence of absence? Understanding the chronology ofIndigenous occupation of western New South Wales, Australia. Archaeologyin Oceania 40: 33-49. --2006. Geoarchaeological Investigation of Aboriginal LandscapeOccupation in Paroo-Darling National Park, western NSW, Australia (eRAL1). Auckland: Anthropology Department, University of Auckland. HOLDAWAY, S.J., P.C. FANNING & E. RHODES. In press. Assemblageaccumulation as a time dependent process in the arid zone of western NewSouth Wales, Australia, in S.J. Holdaway & L.A. Wandsnider (ed.)Time in Archaeology. Salt Lake City (UT): University of Utah Press. HOPE, J.H. 1993. Pleistocene archaeological sires in the centralMurray-Darling basin, in M.A. Smith, M. Spriggs & B. Fankhauser(ed.) Sahul in Review: Pleistocene Archaeology in Australia, New Guineaand Island Melanesia (Occasional Papers in Prehistory 24): 183-96.Canberra: Department of Prehistory, Research School of Pacific Studies,Australian National University. HOPE, J.H., A. DARE-EDWARDS & M.L. McINTYRE. 1983. Middens andmegafauna: stratigraphy and dating of Lake Tandou lunette, western NewSouth Wales. Archaeology in Oceania 18: 38-45. JOHNSTON, H. 1993. Pleistocene shell middens of the WillandraLakes, in M.A. Smith, M.Spriggs & B. Fankhauser (ed.) Sahul inreview: Pleistocene Archaeology in Australia, New Guinea and islandMelanesia (Occasional Papers in Prehistory 24): 197-203. Canberra:Department of Prehistory, Research School of Pacific Studies, AustralianNational University. JOHNSTON, H. & P. CLARK. 1998. Willandra Lakes archaeologicalinvestigations, in H. Johnston, P. Clark & J.P. White fed.)Willandra Lakes: People and Palaeoenvironments (Archaeology in Oceania33): 105-19. Sydney: University of Sydney, with National Parks andWildlife Service, New South Wales, and Willandra Lakes World HeritageArea Management Council. JOHNSTON, H., P. CLARK & J.P. WHITE fed.). 1998. WillandraLakes: People and Palaeoenvironments (Archaeology in Oceania 33).Sydney: University of Sydney, with National Parks and Wildlife Service,New South Wales, and Willandra Lakes World Heritage Area ManagementCouncil. LITTLETON. J. 1999. East and West: burial practices along theMurray River. Archaeology in Oceania 34(1): 1-14. --2000. Taphonomic effects of erosion on deliberately buriedbodies. Journal of Archaeological Science 27: 5-18. --2002. Do cemeteries exist? Aboriginal burials on the Hay Plain.Archaeology in Oceania 37:105-21. --2007. From the perspective of time: hunter-gatherer burials insouth-eastern Australia. Antiquity 81: 1013-28. LITTLETON, J. & H. ALLEN. 2007. Hunter-gatherer burials and thecreation of persistent places in southeastern Australia. Journal ofAnthropological Archaeology 26: 283-98. McINTYRE, M.L. & J.H. HOPE. 1978. Procoptodon fossils from theWillandra Lakes, western New South Wales. The Artefact 3:117-32. MUHLEN-SCHULTE, R. 1985. Mungo rocks: a technological analysis ofstone assemblages from Lake Mungo. Unpublished BA (Hons) dissertation,Australian National University. MULVANEY, D.J. & J. KAMMINGA. 1999. Prehistory of Australia. StLeonards (NSW): Allen & Unwin. PARDOE, C. 1988. The cemetery as symbol: the distribution ofAboriginal burial grounds in southeastern Australia. Archaeology inOceania 23: 1-16. --1995. Riverine, biological and cultural evolution in southeasternAustralia. Antiquity 69: 696-714. SHAWCROSS, W. 1998. Archaeological excavations at Mungo, in H.Johnston, P. Clark & J.P. White fed.) Willandra Lakes: People andPalaeoenvironments (Archaeology in Oceania 33): 183-200. Sydney:University of Sydney, with National Parks and Wildlife Service, NewSouth Wales, and Willandra Lakes World Heritage Area Management Council. SHINER, J., S.J. HOLDAWAY, H. ALLEN & P.C. FANNING. 2005. Stoneartefact assemblage variability in Late Holocene contexts in western NewSouth Wales: Burkes Cave, Stud Creek and Fowlers Gap, in C. Clarkson& L. Lamb fed.) Lithics 'Down Under': AustralianPerspectives on Lithic Reduction, Use and Classification (BritishArchaeological Reports International Series 1408): 67-80. Oxford:Archaeopress. SHINER, J., S.J. HOLDAWAY, H. ALLEN & P.C FANNING. 2007. BurkesCave and flaked stone assemblage variability in western New South Wales,Australia. Australian Archaeology 64:35-45. STERN, N. 1994. The implications of time-averaging forreconstructing the land-use patterns of early tool-using hominids.Journal of Human Evolution 27: 89-105. THORNE, A. & D. CURNOE. 2000. Sex and significance of LakeMungo 3: reply to Brown 'Australian Pleistocene variation and thesex of Lake Mungo 3'. Journal of Human Evolution 39: 587-600. WALSHE, K. 1998. Taphonomy of Mungo B assemblage: indicators forsubsistence and occupation of Lake Mungo, in H. Johnston, P. Clark &J.P. White fed.) Willandra Lakes: People and Palaeoenvironments(Archaeology in Oceania 33): 210-16. Sydney: University of Sydney, withNational Parks and Wildlife Service, New South Wales, and WillandraLakes World Heritage Area Management Council. WEBB, S.G. 1989. The Willandra Lakes Hominids. Canberra: Departmentof Prehistory, Research School of Pacific Studies, Australian NationalUniversity. WEBB, S.G., M.L. CUPPER & R. ROBINS. 2006. Pleistocene humanfootprints from the Willandra Lakes, southeastern Australia. Journal ofHuman Evolution 50: 405-13. WILLIAMS, D. 1991. The case of the shattered stones: an analysis ofthree Aboriginal quarry/reduction sites from the Willandra Lakes WorldHeritage Area, south western New South Wales. Unpublished BA (Hons)dissertation, Australian National University. Harry Allen (1), Simon Holdaway (1), Patricia Fanning (2) &Judith Littleton (1) (1) Department of Anthropology, The University of Auckland,Auckland, New Zealand (Email for correspondence: h.allen@auckland.ac.nz) (2) Graduate School of the Environment, Division of Environmentaland Life Sciences, Macquarie University, Sydney, Australia