"The artefact survey protocols we have developed for the drylands of western NSW, Australia, target land surfaces with the greatest potential for exposure of the archaeological record. (...) In effect, the artefact scatters have become 'time-averaged' deposits in which the products of many individual behaviors are now lumped together."

• Background
• The study area
• Surface stone artefact scatters: Why can we see them?
• Applying geospatial technology to archaeological survey
• Artefact visibility
• Developing behavioral interpretations of the artefacts
• Putting place use into a temporal framework
• Summary: A new geoarchaeological framework
• References
• Author information
• Additional web resources

Background

(Back to top)
Typically, archaeologists call locations with concentrations of artefacts 'sites', and infer that these artefacts correlate with places where people lived in the past. The definition of a site requires that a boundary be drawn around a concentration of artefacts. However, in Australia, although Aboriginal hunter-gatherers frequently reused the same general location for activities such as tool-making, they often did not return to precisely the same spot within that general location. Hence the artefacts they left behind are frequently spatially dispersed and as a consequence, the boundaries of sites are hard to define. This is particularly the case in the drylands, which comprise nearly 70% of the Australian continent (Holdaway, Fanning and Witter 2000).

It has been very difficult for archaeologists in Australia to come to terms with the nature of this dispersed record. Archaeological work at Lake Mungo, in southwestern New South Wales (NSW), provides a good example. Nearly 30 years ago, Pleistocene dates from archaeological 'sites' in this region ensured that Lake Mungo became widely cited in archeology textbooks around the world. But a review of studies since then (Johnston and Clark 1998) shows that little published archaeological work has emerged. The chronology of the lakes has been progressively refined (e.g. Bowler and Price 1998, Gillespie 1998) but the archaeology has remained largely unknown because it was spread across areas measured in hectares rather than in a concentration buried at one small location (Shawcross 1998) and so was not easily definable in terms of sites.

Dealing with large amounts of artefactual material, distributed across wide areas with no clear boundaries, presents both technical and theoretical challenges to a range of stakeholders who must deal with the archaeological record in the drylands (for example, national park managers, graziers, archaeologists, indigenous owners, infrastructure developers, and the tourism and mining industries). Management works best when bounded entities like sites can be clearly defined and located. It is much harder to manage the heritage value of sites where boundaries are not clear and may, in fact, vary through time as the conditions that promote surface exposure change.

Here, we outline the ways in which geospatial technologies have been crucial in interpreting this record despite the difficulties imposed by the nature of the artefacts and their wide dispersal across the landscape. The approach we describe is based on fieldwork we have undertaken since 1995 in northwestern NSW, Australia, in Sturt National Park, and more recently at the Fowlers Gap Arid Zone Research Station north of the mining city of Broken Hill. In this project archaeological studies of artefact distribution are closely integrated with geomorphological studies of landsurface erosion and sedimentation, permitting interpretations to be developed for the many thousands of artefacts that lie exposed on the surface.

The study area

Link to Figure 1: Study area map (~30K)

(Back to top)
The rangelands of western NSW lie on the southeastern margin of arid central Australia, with annual rainfall averaging less than 200 mm and pan evaporation exceeding 2000 mm. Chenopod shrubland is the dominant vegetation type, with scattered trees (mostly Acacia spp. such as Mulga) along the ridges and ephemeral watercourses. Rock outcrops that provided abundant local sources of raw material for stone tool making include silicified duricrusts (silcrete), quartz, and quartzite. Silcrete and quartz gibber pavements mantling hillslopes were also utilized. Much rarer are fine-grained amorphous silcrete (or chert), hornfels, greenstone and petrified wood, the latter derived from the opalized Cretaceous marine sediments of the Great Artesian Basin.

Surface stone artefact scatters: Why can we see them?

Link to Fig. 2: Artefact scatter (~23K)

(Back to top)
The drylands of Australia offer a unique opportunity for archaeologists to study stone artefacts scattered over extensive areas. They are characterized by high levels of artefact visibility, related in part to naturally discontinuous vegetation cover under the prevailing dry climatic conditions. In addition, artefact exposure has been enhanced by accelerated erosion of topsoils and surficial sediments over the last 100 years or so (Fanning 1999). The presence of large numbers of introduced herbivores (mostly sheep and rabbits), coinciding with a prolonged period of drought conditions in the 1890s, reduced the vegetation cover. This, in turn, affected the hydrogeomorphic balance by reducing infiltration and increasing surface runoff and erosion when the drought-breaking rains finally came. Topsoil material eroded off the slopes was deposited over the valley floors and flood plains, forming a distinctive sedimentary unit commonly referred to as 'post-European material' (PEM). Incision of valley floors, channel enlargement and knickpoint retreat, especially in the upland catchments, subsequently destabilized the valley floors, leading to partial stripping of the PEM and the formation of scalded and lagged surfaces over extensive areas. The fine sandy and silty sediments were washed and blown away, leaving behind the coarse clasts - the pebbles and cobbles and stone artefacts - as a lag. These processes continue today, in spite of more conservative stocking rates over most of the region (Fanning 1994).

Rill and gully erosion has disturbed the integrity of part of this lag by moving some artefacts downslope, away from their original resting places (Holdaway et al. 1998). However, statistical analysis of relationships between artefact size and topographic factors demonstrates that, outside of the rills and gullies, little lateral disturbance can be detected (Fanning and Holdaway 2001a). The vertical integrity of the artefacts dropped by different groups of Aboriginal people occupying particular places in the landscape at different times has been lost, but rather than damaging the record, the process of erosion is a boon for archaeologists. Firstly, the erosion has 'excavated' areas two to three orders of magnitude larger than those commonly tackled by archaeologists who have to deal with buried deposits. The costs of excavation by conventional means preclude those deposits covering thousands of square meters, yet we know from studies of contemporary Aboriginal people who live in the arid zone that this is the size of the area over which camps sometimes extend (e.g. O'Connell 1987). There is every reason to expect that prehistoric camps were of a similar size, and reoccupation is likely to have extended the spatial distribution of artefacts beyond the limits of the original campsite. Secondly, the lagging process has conflated the discard products of many events together in one place. While this means that individual occupation events cannot be identified, it also means that patterns created through repeated occupations are much easier to study. Far from destroying the scientific potential of the archaeological record, the erosion-induced lagging process has greatly enhanced the ease with which this record can be observed.

Applying geospatial technology to archaeological survey

(Back to top)
Our survey strategy reflects this temporal and spatial variability in geomorphic dynamics, and characterizes the geomorphic landscape at three distinct spatial scales. At the regional level, the sampling and survey methodology makes use of available Land Systems mapping, a form of regional reconnaissance mapping developed by Christian and Stewart (1953) and applied to extensive tracts of Australia. Land Systems are defined as "an area or group of areas throughout which there is a recurring pattern of topography, soils and vegetation" (Christian and Stewart 1953). Land Systems form a convenient synthesis of the natural environment, and provide, amongst other things, a basis for assessment of land capability and for the study of land use problems. Land Systems are distinguished on the basis of aerial photograph patterns, and their characteristics are identified by ground observations at sample locations chosen from aerial photographs. Their spatial scale is usually of the order of tens to hundreds of square kilometers.

Since Land Systems are primarily defined by their topographic signature, using remotely sensed data (i.e. airborne and satellite imagery), those geomorphic and other environmental processes that affect artefact exposure and visibility are expected to vary more between Land Systems in a particular region than within any one of them. Therefore Land Systems maps provide a convenient means of assessing which parts of a region are most likely to exhibit maximum artefact exposure.

At the second or meso-scale level, our survey method focuses on the smaller landform units and landform elements that make up Land Systems, as these reflect the operation of geomorphic processes over the contemporary to historic timescales. Standard aerial photograph interpretation and field survey methods are used to map landform elements and classify them on the basis of dominant geomorphic environment i.e. residual, transportational/eroding, fully lagged or depositional. In this way, the physical landscape is subdivided on the basis of both landform and dominant processes.

The third or micro-scale level survey concentrates on documenting local variability in landsurface condition that reflects the operation of processes with a short time scale of perhaps hours to days. These landsurface conditions affect the archaeological record at the moment of survey. Such processes include local erosion and deposition of sediments, bioturbation, and vegetation growth.

Link to Fig. 3: GIS system image (~21K)

We use a vector GIS (ARCINFO/ARCVIEW) to integrate these three survey scales (Holdaway, Fanning and Witter 1997; Fanning and Holdaway submitted). Each of the three survey scales is mapped as a separate coverage in the GIS independent of the distribution of artefacts, which is also mapped as a separate coverage. Data tables hold descriptive information on the nature of each of the coverages. This format permits comparison of the interaction of the three survey scales with the distribution of artefacts. In addition, because of the relational format, data acquisition in the field at all three scales can progress independently.

Three different approaches to artefact recording were also developed. In the first, aimed at relating artefact density and variability to landsurface type, the characteristics of every artefact larger than 20mm maximum dimension were recorded, together with their location in three-dimensional space. Once the raw data were converted to coverages in the GIS, the relationships between a variety of artefact characteristics and landsurface type could then be analyzed. In the second, within-unit artefact variability was examined by surveying all of the artefacts within just one landsurface type with high artefact exposure. Having determined the likely range of artefact types and densities from this intensive piece of proveniencing, systematic sampling techniques were applied to a much larger area. A grid of 1m x 1m squares oriented north-south and east-west was laid out, in which all the artefacts in every fifth square were recorded. This resulted in a 7% sample of the total area being recorded.

Link to Fig. 4: Artefact survey (~22K)

Artefacts were recorded individually in the field and then replaced where they were found. A series of technological and typological variables were described for each piece. These included the nature of the evidence for flaking (flake, tool or core), the raw material type, the presence of cortex, the overall shape of the flake or core, the form of the tool and the number of retouched edges, and measurements of flake, tool and core size. These attributes were recorded into palmtop computers using data entry software (McPherron and Holdaway 1996) and transferred each day to the relational database and GIS. This system permits the rapid recording of large numbers of stone artefacts with only minimal disturbance of the surface record, honoring the wishes of indigenous custodians for their material cultural heritage to be left in place.

Artefact visibility

(Back to top)
Linking different data sets within a GIS permits the rapid assessment of differential artefact visibility as a function of the presence of erosional and depositional surfaces, as well as the presence of a range of shorter-term biological features (e.g. vegetation clumps, ants nests, animal tracks) (Fanning and Holdaway submitted). Predictably, artefact exposure is greatest on erosional surfaces and least on depositional surfaces, but attempts to precisely quantify these relationships reveal that at a local level, there is considerable variation, so much so that at times the predicted relationships between surface condition and visibility are obscured. This is particularly true when small spatial samples are considered.

This has important implications for the definition of archaeological 'sites'. Despite the management utility of identifying small bounded entities, concentrations of artefacts (i.e. 'sites') when considered over small areas (tens or hundreds of square meters) occur for a variety of reasons. Our experience is that differentiating between the effects of surface condition, vegetation patterns and human behavior in the past can be very difficult in these situations. Only when areas of several thousands of square meters are considered in some detail does it become possible to understand what is causing artefacts to be concentrated. Spatial data collection technology has been essential to developing the methodology required to deal with this complexity.

Developing behavioral interpretations of the artefacts

(Back to top)
The artefact attributes recorded and the variety of artefacts present in any one place can be used as indicators of the frequency and/or duration of human occupation of that place. Three sets of stone artefact analyses have been developed as a means of assessing this place use history (Holdaway and Shiner submitted). In the first, the proportion of raw materials from different sources is used to provide an indication of the way that local versus imported material was deposited. Secondly, a series of technological measures is used to assess the composition of the assemblages in ways that reflect the relationship between mobility, occupation intensity, raw material choice, and discard. Thirdly, assemblage composition is considered to determine the nature of differential discard across space.

Using these techniques to compare assemblages from different places within single field survey areas, we find that we are able to distinguish those places that were visited more often and perhaps for longer periods from those where visitation was less frequent and less prolonged. When extended to comparing assemblages from the broader geographic region of western NSW, these techniques can be used to differentiate and characterize variability in place use history between places with similar environmental and chronologic settings (Shiner et al. in preparation).

Putting place use into a temporal framework

(Back to top)
Surface artefact scatters are notoriously difficult to date because they lack the stratigraphy found in sequences excavated beneath rock shelters or in caves. However, we overcome this limitation by taking a two-stage approach to determining the chronology of occupation at each location. First, we date the charcoal preserved in the remains of heat-retainer ovens associated with the artefact scatters (Holdaway et al. 2002). Second, we date the landsurface upon which the artefacts have been discarded, using standard geomorphological and stratigraphic techniques coupled with absolute dating of the sedimentary sequence below the discard surface (Fanning and Holdaway 2001b). The radiocarbon determinations from the heat-retainer ovens provide a range of time during which the ovens were constructed and used. The stratigraphic analysis provides a maximum age for the artefacts lying on the surface. Together, they provide an envelope of time during which the stone artefacts were discarded. Equally important, the stratigraphic analysis allows a history of geomorphic change to be developed, with implications for the preservation of the archaeological record.

The results of two sets of radiocarbon determinations obtained so far from ovens in western NSW challenge some commonly held views about Late Holocene Aboriginal occupation patterns. At Stud Creek near Mt Wood in Sturt National Park, 28 determinations returned ages between 220 and 1630 y BP. However, the spread of dates is discontinuous, with a 300 year gap between about 800 and 1100 y BP (Holdaway et al. 2002, Table 1), suggesting that this location may have been abandoned for a considerable period of time in prehistory. Further south, at Fowlers Gap, the age range is greater, extending to about 5000 y BP. However, it is not the time span of these dates that is of most interest, but the pattern of dates in relation to the inferred ages of the land surfaces on which the hearths and associated artifacts are found. On land surfaces where geomorphic processes of erosion and deposition are relatively active, such as the margins of modern stream channels, the dates are relatively recent and the record relatively short, reflecting the dynamic geomorphic environments at these locations. Sediments accumulate over short periods of time, perhaps a few hundred years, and are then removed by erosion during moderate flood events. In contrast, fine-grained alluvial drapes on low terrace surfaces are only geomorphically reworked during less frequent, high magnitude floods, and thus preserve a longer archaeological record.

Furthermore, at each of the sampling locations at Fowlers Gap, the determinations form multiple clusters rather than a continuous sequence, a similar pattern to that found at Stud Creek. This suggests that Aboriginal people moved into and out of the region throughout the mid to late Holocene, occupying areas that had not been occupied for several centuries. This contrasts with the commonly held view from elsewhere in Australia that Aboriginal occupation of some regions was more or less continuous from the late Holocene onwards, reflecting increasing population and a more sedentary lifestyle. On the contrary, our research shows that Aboriginal hunter-gatherer groups remained relatively mobile, moving away from then back into previously occupied areas on a fairly regular basis throughout the mid to late Holocene and right up to the time of European contact around 150 years ago.

Link to Fig. 5: Hearth image (~35K)

These findings have implications for the application of conventional settlement pattern analyses based on site survey data. Surveys that do not take into account the geochronology of the surfaces on which 'sites' are located run the risk of combining archaeological materials with markedly different ages into a single settlement system. Geomorphological studies in other areas of Australia indicate that our findings are not likely to be unique.

Summary: A new geoarchaeological framework for interpreting surface artefact scatters in the drylands

(Back to top)
The record of Aboriginal hunter-gatherer activity, i.e. the stone tools they manufactured, used and then discarded, needs to be studied in spatially extensive sets. However, surface scatters of stone artefacts had previously been largely dismissed by archaeologists because they were unbounded, appeared to be seriously disturbed by post-depositional formation processes, and lacked the stratigraphy considered essential for establishing a chronology of occupation. Moreover, until the advent of Global Positioning Systems (GPS), electronic surveying equipment, digital data capture and Geographic Information Systems (GIS), the technology necessary to record, store and analyze large, spatially extensive data sets was not available. The artefact survey protocols we have developed for the drylands of western NSW, Australia, target land surfaces with the greatest potential for exposure of the archaeological record. The protocols accommodate differential artefact visibility at the time of the survey by including assessment of surface cover and quantifying its effects on artefact density.

Erosion of the fine sediments surrounding and supporting the stone artefacts has meant that the discard products of Aboriginal activity have been conflated. In effect, the artefact scatters have become 'time-averaged' deposits in which the products of many individual behaviors are now lumped together. By looking at the nature of the artefacts accumulated over time in different places in the landscape, archaeologists are able to distinguish those places that were used frequently from those that were used less often and those used for common activities from those used more sparingly. But, rather than seeking to date individual events, the goal is to seek pattern in both the spatial and temporal record of events that will allow a detailed prehistory of Aboriginal 'use of place' to be developed.

References

(Back to top)
Bowler, J. M. and D.M. Price. 1998. Luminescence dates and stratigraphic analyses at Lake Mungo: review and new perspectives. Archaeology in Oceania 33:156-168.

Christian, C.S. and G.A. Stewart.1953. General report on survey of Katherine-Darwin Region, 1946. Commonwealth Scientific and Industrial Research Organisation, Australian Land Research Service 1. Canberra: C.S.I.R.O.

Fanning, P.C. 1994. Long-term contemporary erosion rates in an arid rangelands environment in western New South Wales, Australia. Journal of Arid Environments 28:173-187.

Fanning, P. C. 1999. Recent landscape history in arid western New South Wales, Australia: a model for regional change. Geomorphology 29(3/4):191-209.

Fanning, P.C. and S.J. Holdaway. 2001a. Stone artifact scatters in western NSW, Australia: Geomorphic controls on artifact size and distribution. Geoarchaeology: an International Journal 16(6):667-686.

_____. 2001b. Temporal Limits to the archaeological record in arid western NSW, Australia: Lessons from OSL and radiocarbon dating of hearths and sediments. In Australasian Connections and New Directions: Proceedings of the 7th Australasian Archaeometry Conference, eds. M. Jones and P. Sheppard. Research in Anthropology and Linguistics 5:91-111.

_____. Submitted but not published. Stone artifact exposure and visibility at open sites in western New South Wales, Australia: a geomorphic framework for survey and analysis. Paper submitted for publication 2002.

Gillespie, R. 1998. Alternative timescales: A critical review of Willandra Lakes dating. Archaeology in Oceania 33:169-182.

Holdaway, S., P. Fanning and D. Witter. 1997. GIS analysis of artefact distributions in an eroding landscape: the Western New South Wales Archaeological Project. In Archaeological Applications of GIS: Proceedings of Colloquium II, UISPP XIIIth Congress, Forli, Italy, September 1996. Johnson, I. and M. North, eds. Sydney University Archaeological Methods Series 5.

Holdaway, S., D. Witter, P. Fanning, R. Musgrave, G. Cochrane, T. Doleman, S. Greenwood, D. Pigdon, and J. Reeves. 1998. New approaches to open site spatial archaeology in Sturt National Park, New South Wales, Australia. Archaeology in Oceania 33:1-19.

Holdaway, S.J., P.C. Fanning, and D.C. Witter. 2000. Prehistoric Aboriginal occupation of the rangelands: Iinterpreting the surface archaeological record of far western New South Wales Australia. The Rangelands Journal 22:44-57.

Holdaway S.J., P.C. Fanning, D.C. Witter, M. Jones, G. Nicholls and J. Shiner. 2002. Variability in the chronology of Late Holocene Aboriginal occupation on the arid margin of southeastern Australia. Journal of Archaeological Science 29:351-363.

Holdaway, S.J. and J. Shiner. Submitted but not published. Hunter-gatherers and the archaeology of discard behaviour: an analysis of surface stone artefact scatters from Sturt National Park, western New South Wales, Australia. Paper submitted for publication 2002.

Johnston, H. and P. Clark. 1998. Willandra Lakes archaeological investigations 1968-98. Archaeology in Oceania 33:105-119.

McPherron, S. and S. Holdaway. 1996. Entrer Trois. In A Multimedia Companion to the Middle Palaeolithic Site of Combe-Cappelle Bas (France). H. Dibble and S. McPherron, eds. Philadelphia: University of Pennsylvania Museum.

O'Connell, J. F. 198). Alyawara site structure and its archaeological implications. American Antiquity 52:74-108.

Shawcross, W. 1998. Archaeological excavations at Mungo. Archaeology in Oceania 33:183-200.

Shiner, J., S. J. Holdaway, H. Allen and P.C. Fanning. In preparation. Understanding stone artefact assemblage variation in late Holocene contexts in western New South Wales, Australia: Burkes Cave, Stud Creek and Fowlers Gap.