The following paper is reproduced with the permission of the author and the journal. It was originally published as:

 Denis Byrne, 1980. Dynamics of Dispersion: The place of silcrete in archaeological assemblages from the Lower Murchison, Western Australia. Archaeology and Physical Anthropology in Oceania 15, 110-19. Copyright statement is provided at the bottom of the page.

RECENT studies of the behaviour of Aborigines using stone tools have stressed the importance of raw material, especially as a factor influencing interassemblage variation. Hayden (1977) suggests that availability of raw material may determine the extent to which primary flakes used as tools are retouched. Wright (1977:2) quotes Daisy Bates's observations that availability of raw material affected the size of artifacts on archaeological sites at Ooldea, Central Australia and Gould (1977:167), for the Western Desert, concludes that artifacts found in habitation contexts away from quarries will tend to be smaller and show a narrower size range compared to those found at quarries. O'Connell (1977) combines ethnographic information with ev.dence from stone tool assemblages from sites in the same area of Central Australia and discusses the effect of access to raw material on the relative proportion in assemblages of artifacts made from different types of stone, and also on the occurrence of various tool types, certain categories being more likely to be fabricated from certain types of stone than others. The present paper looks at archaeological sites surrounding a silcrete quarry in Western Australia and explores the effect of proxi~ity to the quarry on stone tool assemblages. I attempt to relate current thinking stemming from ethnography to a purely archaeological situation.

From 1976 to 1978 I surveyed an area of approximately 1200 sq km, from the mouth of the Murchison River to about 35 km inland. Collections of stone material were made from all sites found. Silcrete was used for flaked stone artifacts in 41 sites out of 45 recorded. The survey also located a large silcrete quarry central to the survey area. It also became apparent that the incidence of silcrete in the assemblages was related to distance from the quarry. Once this was noted a particular effort was made to locate sites at a range of distances from the quarry in order to explore the effect of the 'relative distance' variable.

As it approaches the Indian Ocean the Murchison River cuts a meandering course through deep gorges in the Silurian Tumblagooda sandstone from which it emerges, some 40 km from the coast, into a wide river valley. In the area surveyed the gorges and the valley are flanked by sandplain plateaux. The river flows intermittently, flooding once or twice a year. at other times a chain of pools retain water and fish. The sandplain supports a low scrub heath mostly of Acacia-Banksia; Eucalypts, Acacia and Melaleuca are found in the gorges and in the valley where in places they form thickets. The quarry is situated on the lower slopes of Pillawarra Hill which is an outlier of the steep scarp which forms the edge of the plateau (Fig. 1).

Survey coverage was restricted by the terrain. Coverage of the sandplain was limited to those sections traversed by station tracks, about 5%. Only one site was located in this zone which is not surprising in view of the lack of water or shade. In the river valley and gorges approximately 50% of the land surface was covered by vehicle and on foot. This country is very rugged, with the advantage of making obvious those spots suitable for habitation, such as rockshelters and more level areas. The survey was conducted over three drought years during which ground-cover vegetation was minimal and visibility of archaeological surface material correspondingly high.

Of the 45 sites recorded all, apart from the silcrete quarry, are judged to have been occupation-camps. They are thus unlikely to be representative of all places where lithic material might have been used and deposited. Omitted are butchery sites, places where weapons have been sharpened or repaired, places where wood has been obtained for implements, and spear manufacturing sites.

At 39 sites the assemblages were sufficiently small to enable complete collections to be made. The assemblages on the larger sites were sampled by total collection within blocks of ten metre grid squares - in each case the sample was not less than 40% of the total assemblage area.

The agencies of dispersion of Pillawarra silcrete were mobile individuals and/or groups of Aborigines, so that these assemblages must reflect a behavioural system, particularly its spatial dimension. In terms of direct distance from the quarry the assemblages are spread over 27 km, so they might be said to sample the dispersion process over a zone of radius 27 km.

I assume that the assemblages reflect the 'flow' of the resource through the prehistoric culture system. Models intended to depict such a flow have been advanced by Schiffer (1972), Gould (1977) Sheets (1975). Schiffer breaks the flow process down into five more or less discrete stages: Procurement: Manufacture: Use: Maintenance: Discard. This process has both temporal and spatial dimensions. In the case of flaked stone artifacts the time from procurement to discard may be an hour, a week, or several years. The space spanned may range from a single site, or area within a site, to a chain of locations stretching hundreds of kilometres. A complete reconstruction must incorporate both temporal and distributional studies.

With undateable surface assemblages it is not possible to fill in the temporal context of the dispersion process. However, the spatial context of the process is more easily reconstructed. It begins with raw material available at one point in space; its useful role, however, occurs at a universe of points away from the source point. The importance of the material in the technology at these points will differ according to a number of variables but among the most basic will be its relative availability or potential supply. The actual supply is constrained by such factors as the existing means of transportation, lines of communication, cultural boundaries etc. In this study I focus upon relative linear distance from the source. The working hypothesis is that supply diminishes with increasing distance from the source.

From this hypothesis two questions are drawn for testing. First, how will the situation of diminishing supply be registered at particular sites? For example, will it be demonstrably apparent that the supply of raw material is different at 9 km than it is at 1 km from the quarry? It might be found that the size of cores available for flake production is smaller. Secondly, given a situation of diminishing supply, what behavioural responses may have been made in terms of the treatment of the stone? Humans appear to readily alter their behaviour to adjust to changing conditions in their environment. It is possible, for example, to soften the effect of a diminishing resource by using it more efficiently, or by using it only for special purposes, or by other means. The first question pre-supposes certain direct and irreversible effects of diminishing supply. The second aims at ways by which the pressure can be offset. The results of the first are called direct effects, and those of the second, mediating responses.

The assemblages are ordered in sequence according to the direct distance of each from the quarry, since for all recorded sites which have quarry-derived material the minimum assertion that can be made is that the stone must have covered at least this distance.

For the purposes of analysis I have ordered the assemblages into three concentric zones around the quarry, referred to as Zones I, II, and III respectively (compare McBryde 1978). The distance intervals are: Zone I, 0-2.5 km; II, 2.5-10 km; III, 10-27km. They are chosen to correspond to a tendency for the recorded sites to cluster into three concentric bands (Fig. 2). The assemblages within each zone are combined to form three agglomerate assemblages. This step is taken because of the great range in the numbers of artifacts per assemblage and hence the problem of making valid statistical comparison between assemblages. The problem is real, since assemblage size spans 1-904 flaked items. Combining individual assemblages into 'zonal assemblages' overcomes this problem and, though it means the effect of increasing distance from the quarry can only be observed in three 'steps' rather than from site to site, I hope it tempers the assumption that precise linear distance of sites from the quarry is a measure of the relative 'availability' of the resource at these points. It assumes a general correspondence between the distance of a site from the source and the availability of the stone. Additionally, sampling by zones allows any coastal/inland variation to be ignored, for Zones II and III at least include sites from both environmental regions.

Any dispersal study starts with the assumption that the source, or sources, of the dispersed material is known. John Glover (Department of Geology, University of Western Australia) examined specimens from Pillawarra quarry and a number of sites at a range of distances from it. The stone was found to be composed mainly of quartz grains in a very finely divided siliceous matrix. All were found to be of a lithological similarity consistent with their derivation from a common source. However, in view of the relatively simple texture of silcrete and the absence of any diagnostic minerals in the thin sections examined, the possibility of multiple sources must be left open. However, the survey failed to locate any source of silcrete other than the Pillawarra quarry although an effort was made to examine geologically equivalent locations.

As with collection, analysis of the flaked stone material attempted to be all inclusive: primary flakes and cores were included with the retouched items. The rationale for this is, first, I am describing a system of dispersion of a raw material. As far as I can tell, I am not dealing with the dispersion of 'finished' tools from a factory site, but with a raw material which appears to have left the quarry in the form of cores and primary flakes as well as, perhaps, 'finished' items. Second, given ethnographic evidence of significance of unretouched items as implements (Gould 1969:82-3; Gould et al. 1971; Hayden 1977; Horne and Aiston 1924:91), and Hayden's (1977:179) conclusion that much secondary retouch simply amounts to maintenance of cutting edges, it seems unwise to make a dichotomy between worked und unworked material, and better to think in terms of a continuum of increasing modification of stone material depending upon such variables as the intended function and the duration of use of implements. A large portion of the collected material could not be accommodated within any of the categories of flaked material even at the level of primary flakes. In the six largest assemblages the proportion of unclassifiable pieces averages 46%. Much of this can be accounted for by the propensity of Pillawarra silcrete to weather and break down in a manner which produces fissures and fractures superficially quite like those associated with percussion flaking but distinct in their lack of striking platforms or points of percussion. This 'natural' reduction process continues to operate 110W, producing hair-line cracks which widen till the stone fragments. I have found evidence of it on pieces of stone of all sizes, from large blocks to centimetre square debris, and numerous instances of percussion cores and flakes having fractured in this manner subsequent to their manufacture. Consequently, broken fragments of such artifacts (e.g., the distal Portion of broken flakes) are included in the unclassifiable category.

The very large quantity of material limited classification to three stages: an initial sorting into stone types isolated the quarry-derived fraction from chert and quartz, the only other types present; the material was then sorted into 'core', 'flake', and 'other' categories ('other' include flake material which would be allocated to either of the previous categories, plus material which may or may not be flaked e.g., fragrnents, nodules etc.); finally, the 'cores' and 'flakes' were separated into retouched and unretouched categories.

Some variables were then measured. The flakes were placed into four size classes: <2, 2-4, 4-6, >6cm, according to the longest dimension of each item. The total weight and mean flake weight for each class was obtained. Individual weights were obtained for cores and retouched flakes. The thickness of retouched flakes was measured, along with the edge angles on all retouched edges.

 The analysis deals with silcrete artifacts only. The presence of flaked artifacts of other types of stone in assemblages was noted and their relative importance shown in Fig. 3. Many sources of chert occur close to the coast and along the river valley as far as the gorges. My superficial impression is that chert artifacts are, on average, much smaller than those of silcrete and have a higher incidence of retouch. There is no obvious sign of distance-related variation (falloff patterns or distance-decay) pointing to one or a limited number of chert sources as there' is in the case of silcrete. The same may be said of the quartz artifacts: quartz pebbles are relatively common in the river bed and are often found in the Tumblagooda sandstone.

 I discuss first the weight of cores and flakes and the ratio of retouched to non-retouched items at the quarry and in each of the three zones. In addition the flakes are divided into four size classes and figures are presented so as to show the variation occurring from the quarry out through each of the three zones.

 1. Weight. The large number of cores in the sample from the quarry and Zones I and II allows valid comparison of weights. This is not so for Zone III which has only eight cores-a striking drop from 103 in Zone II (Table 1). Further, Zone III cores are drawn from 25 assemblages whereas those from Zone 11 are drawn from only nine. Thus cores almost drop out of the archaeological record by Zone III i.e., 10 km from the quarry. Cores from the quarry itself come from a surface collection along a transect covering 37 square metres, less than 1% of the total area.

Variation in the weight of cores is striking. At the quarry there is a fairly even spread over a broad size range whereas in the first two zones cores are mosfly small. Although a small sample, this seems also to be true in Zone III. There also seems to be an increasing number of smaller cores further away from the quarry, a tendency shown by the falloff in the mean weight of cores between the quarry and Zone II (Table 1).

 The mean weight of flakes (Table 2), shows a similar fall-off pattern to that of the cores. For both retouched and unretouched flakes there is a sharp fall-off in weight between the quarry and Zone I. For retouched flakes, this trend continues to Zone III, but for Unretouched flakes is slightly reversed there, as was the case with cores.

Table 1.

Number and weight of cores by zones. Figures for Zone 111 exclude one core weighing 436 gm.

Table 2.

Mean weight in grn. and (number) of flakes in each size class and as a total. Unretouched flakes above, retouched flakes below.

 With flakes it is possible to compare the observed variation in mean weight with the variation in the significance of each size class from the quarry to Zone III (Table 3). There is a tendency for the two smaller size classes to become proportionally more important over distance while the two larger classes become less so.

Table 3.

Percentage of each size class of flakes by zones.

 2. Retouched/Unretouched. The terms 'tool' and 'non-tool' are used here to distinguish between cores which exhibit retouch modification of cutting edges and/or edgewear and those which do not. Modification resulting from use as a cutting tool may of course disappear entirely or partly as a result of subsequent use of the core as a source of flakes. Also, this modification could, in some cases, be damage caused to striking platforms by hammerstones hitting too far in from the edge resulting in crushing (R. Gould: pers. comm.).

 Table 4 shows core 'tools' to be consistently heavier than 'non-tool' cores. While at the quarry the difference in weight between the two categories is small, in Zones I and II it is considerable. For Zone III the sample is too small for comparison. The increasing frequency of 'tool' cores relative to 'non-tool' cores can be seen also. The significance of this will be discussed later.

Table 4.

Mean weight in gm of cores in 'tool' and 'non-tool' categories and 'tools' as a percentage of total cores. Number of specimens in brackets. From Zone III one 'tool' core (436 gm) is excluded.

Table 5.

Number of retouched flakes as a percentage of the total number of flakes in each size class within each zone. Only flakes >6 cm at the quarry are retouched.

Table 5 shows that the frequency of retouched flakes increases over distance in three out of the four size classes-sufficient to establish a definite overall trend. With few exceptions retouch takes the form of alterations to one or more edges and does not entail major changes in form.  There is no readily discernible pattern of deliberate formal types. While it has thus not been possible to test for any typological changes over distance, I have attempted to judge whether the nature of the retouched edges changes. The angles of all retouched 'edges has been measured (304 edges on 205 flakes), and the distribution of these is shown in Fig. 4. It can be seen that there is a close similarity in the range of plotted angles for each zone and no indication of any distance-related change.  Thickness measurements have also been taken on all retouched flakes: there is a steady falloff in thickness over distance (Fig. 5).  To summarize, among both cores and flakes the retouched items tend to be heavier than the unretouched items. The percentage of both 'tool' cores and retouched flakes tends to increase with increasing distance from the quarry.

3. Raw material. Changes in size and intensity of use are complemented by raw material changes. The proportions of silcrete, chert, and quartz are shown in Fig. 3. Silcrete decreases from nearly 100% at the quarry to 34% in Zone III. If the observed fall-off over distance were to continue at the same rate silcrete would be very rare more than 40 km from the quarry.

 I turn now to the nature of the process of dispersion, and will discuss it with reference to the flow model referred to earlier. While it is reasonable to assume that a fall-off in raw material supply will have direct effects upon the stone tool technology of a group at any one site, it does not follow that these effects will be directly reflected in the archaeological assemblage. An attempt must be made to frame any interpretation of the assemblage analysis presented above in terms of the intervening stages of the model, since archaeological assemblages are a direct reflection only of the discard process-they are that fraction of a technology which has been abandoned at certain points in space.

1. Procurement. The process of dispersion obviously begins at the source of the raw material-the quarry. A detailed study of the quarry and its assemblage material is still being made, but a brief description can be given.

 The silcrete occurs in outcrops on the lower slopes of Pillawarra Hill. Quarry material extends over about 27 hectares. The relatively steep terrain suggests the quarry itself cannot have served as a camp site; what flat land there is available produced very little artifactual material, nothing in any way comparable to that from the closest adjacent occupation site 0.6 km away. I cannot as yet say how the stone was 'obtained'-rendered into units of handlable size-or to what extent manufacturing was carried out before it was removed from the quarry.

Table 3 shows all size classes of flakes are present at the quarry in near equal proportions. The two smaller size classes may indicate some preliminary working of cores. Eleven of the 66 cores collected show signs of retouch/utilization on what appear to be working edges. Since there is no obvious use to which core 'tools' could have been put at the quarry this may also suggest the 'working' of at least some implements at the quarry.

There are two possible ways material was moved from the quarry to individual sites. Either direct procurement occurred involving transportation of the material from the quarry to individual sites presumably by members of groups occupying the sites at particular times or, alternatively, there was indirect procurement, whereby the material arrives with a group when they first occupy a site and is part of a 'curated' stock of material in the form of cores and implements. In the second case the material may have passed along a chain of sites since leaving the quarry and have been an integral, though archaeologically invisible, component of the technology at these sites. Utilization of previously discarded material found on a site would be included as direct procurement. While we cannot conclusively decide between these alternatives the evidence for decreasing weight of cores and flakes with increasing distance from the quarry seems to argue against direct procurement. If direct procurement operated at all sites the input of raw material into the technology and the output into the archaeological record might be expected to be equal. However, it is quite conceivable that the relative distance to be travelled to procure material directly might determine the weight of the material transported-larger material being taken to the closer sites and smaller to the more distant.

 In what follows I will assume that the procurement process was indirect rather than direct. Since this stage of the flow process is basic to those which succeed it, the interpretation offered below is but one of the two alternative schemes, depending upon which of the procurement alternatives is chosen.

2. Manufacture (flake production). The most obvious direct result of diminishing supply over distance is the decrease in core weight. Since cores are essentially portable sources of raw material they constitute the 'supply' of raw material. In this role cores serve as the basic units in flake production, the very nature of which reduces them in size. The extent to which they are reduced depends upon how often they have served in this role. Decreasing weight with distance suggests more frequent use as a source of flakes, and thus greater reduction. The case is strengthened by evidence suggesting flake production actually ceased somewhere in Zone II: there is the virtual absence of 'non-tool' cores in Zone III and there is also a site in Zone II, 8 km from the quarry, which may represent the tail end of the flake production process. The assemblage from this site (about 50% of the site was covered by the collection grid) yielded 38 'non-tool' cores with a mean weight of 5.5 gm, the lowest value for any assemblage. Beyond this distance no sites had significant numbers of cores of any type.

 If cores were being reduced in size with distance from the quarry it follows that flakes struck from them will also tend to be smaller - that is, their maximum size will be diminished. Table 3 suggests this is the case.

 This gives us the essence of a 'supply fall-off' situation: increasing distance from the source makes it increasingly unlikely that existing supply can be replenished, while the existing supply is continually being diminished.

 Given that the technology has a constant. need for flaked stone implements a fall-off. in supply will call for certain responses in order to maintain balance. One such response might be an increasing reliance upon other types of stone, and positive evidence of this is shown in Fig. 3. Another might be to maximize the usefulness of the existing, diminished, supply. This could be done in two ways: first, at the maintenance level, by carrying out maintenance retouch on working edges and, secondly, at the discard level, by tightening the criteria governing the selection of items to be discarded.

 1. Among Western Desert Aborigines, Hayden (1977) found the aim of retouching primary flakes was to maintain sharp cutting edges. Retouch was uncommon where 'fresh' primary flakes were readily available; in such conditions of favourable supply primary flakes would be discarded once the cutting edge had been dulled. However, 'In traditional situations where raw material might be scarce, one could reasonably expect more primary flakes to exhibit retouch modifications' (1977:180). My field data should be able to test for such a supply-sensitive response using purely archaeological means. Table 5, however, does not strongly demonstrate that the number of retouched flakes increases relative to primary flakes as distance from the quarry increases, although it suggests this may be so.

 2. One very obvious way of maximizing a resource is to minimize waste. As a group moves from one occupation site to another, it takes flaked stone material with it. Schiffer (1976:56) uses the term 'curate set' for this material, which includes tools and non-tool items such as cores. The curate set is continually being depleted, principally as a result of the discard of items along the way. The surface assemblages discussed here derive from that part of the curate set discarded at the various sites, and such assemblages tell us as much about people's attitudes to 'waste' as anything else. The discard processes leading to the formation of archaeological assemblages are unlikely to produce a random sample of the curate set, since certain criteria will select items for discard. For instance, the most obvious criterion governing selection of cores for discard would be their potential for further flake production-only those cores which are 'exhausted' being discarded. But if all assemblages contain only 'exhausted' cores why does their weight tend to decrease in assemblages increasingly remote from the quarry? I suggest that usefulness for further flake production is a concept that is directly related to the supply of raw material. Where supply is great, such as at sites close to the quarry, the criterion permits cores to be discarded which, technically, are not by any means 'exhausted'. As supply diminishes with distance the discard criterion moves down the weight scale to embrace an increasingly smaller and lighter range of cores. Thus I propose that the decrease in core weight is produced at the behavioural level by the combined effect of the flake production process and the criterion governing core discard.

 Concerning flakes, if we could assume that all flakes manufactured at a site were also discarded there then the reduction in the size of cores by flake production would be sufficient to explain the decrease in flake length and weight. But if, as maintained above, the discard of cores is a non-random process the same might be the case with flakes: primary as well as secondary flakes may have been curated, and carried from site to site. If only some flakes were discarded at their place of manufacture, what criteria operated to deter-mine which to discard and which to curate? An obvious one might be their potential for further use. If one follows Hayden (1977), larger flakes might be curated because their cutting edges can be 'maintained' by retouch for a longer duration. Thus, as with cores, selection for discard might increasingly favour smaller, lighter flakes.

 As mentioned earlier, the physical evidence that a core has been used as a tool may be obliterated by subsequent use of the core as a source of flakes. A dual role for cores might be particularly true of the present material as there are very few which appear to have been shaped into actual 'tool' forms (e.g., 'horse-hoofs'); rather, the 'working edges' appear to be a fortuitous by-product of the flake-production process. This makes it difficult to assess the incidence of the use of cores as tools at particular places. Nevertheless, Table 4 shows an increasing incidence of 'tool' cores with in-creasing distance from the quarry and this requires explanation.

 The flake production process means that over time there are smaller and fewer cores in the curate set. At the same time there is likely to have been a constant demand for cores as 'tools'. This would mean that the further one moves from the quarry the higher the proportion of cores in the curate set exhibiting evidence of use as tools and, since discarded cores are drawn from this curate set, an increasing number of 'tool' cores will find their way into the assemblages. In interpreting the tendency for cores in the 'tool' category to be consistently heavier than those in the 'non-tool' category, two alternatives need to be considered. Does the difference stern from non-random discard favouring abandonment of the heavier cores in the 'tool' category or does it reflect a real difference at the level of 'use'? The first alternative seems sufficiently unlikely to be rejected without further consideration. In support of the second alternative we might look to a functional variable: while cores probably served a dual role, there may have been a functional reason for choosing heavier cores for use as tools.

 On the basis of the surface assemblages an attempt has been made to describe and explain the flow of a raw material away from its point of origin, and how its position in the technology as a whole might be expected to alter as availability and, specifically, supply diminish. Three responses to diminishing supply are suggested:

· increasing maintenance of objects at hand

· increased selectivity in regard to that fraction of the curate set discarded

· increased procurement of alternative raw materials.

In terms of the flow model, these responses will be apparent respectively at the maintenance, discard, and procurement stages.

I am grateful to Dr John Glover for his interest and advice on silcrete specimens and to Mrs Sylvia Hallam, Prof Jack Golson, and Prof Richard Gould for criticism of earlier drafts. I also thank the people of Murchison House Station for advice and hospitality.

Gould, R. A. 1969. Yiwarra. Foragers of the Australian Desert. Collins, London.

Gould, R. A. 1977. Ethno-archaeology; or, where do models come from? In R.V.S.Wright (ed.), Stone Too/s as Culture Markers, pp. 162-8. Canberra, A.I.A.S.

Gould, R. A., Koster, D. A. and Sontz, A. H. L. 1971. The lithic assemblage of the Western Desert Aborigines of Australia. Amer. Antiq. 36, 143-69.

Hayden, B. 1977. Stone tool functions in the Western Desert. In R. V. S. Wright, (ed.), Stone Tools as Cultural Markers, pp. 178-88. Canberra, A.I.A.S.

Horne, G. and Aiston, G. 1924. Savage Life in Central Australia. Macmillan, London.

McBryde, I. 1978. Wil-im-ee Moor-ring: or, where do axes come from? Mankind 11:354-82.

O'Connell, J. F. 1977. Aspects of variation in Central Australian lithic assemblages. In R. V. S.Wright (ed.), Stone Tools as Culture Markers. pp. 269-81. Canberra, AlAS.

Schiffer, M. B. 1972. Archaeological context and systematic context. Amer. Antiq. 37, 156-65.

Schiffer, M. B. 1976. Behavioral Archaeology. Academic Press, New York.

Sheets, P. D. 1975. Behavioural analysis and the structure of a prehistoric industry. Curr. Anthropol. 16, 369-91.

Wright, R. V. S. 1971. Archaeology of the Gallus Site, Koonalda Cave. A.I.A.S., Canberra.

Wright, R. V. S. 1977. Introduction and two studies. In R. V. S. Wright, (ed.), Stone Tools as Culture Markers, pp. 1-4. Canberra, A.I.A.S.

COPYRIGHT

You are visitor to this page.

Author: Peter Hiscock, Dept. Archaeology and Anthropology
Feedback: peter.hiscock@anu.edu.au .
Date Last Modified: 10-August-97
URL: http://artalpha.anu.edu.au/web/arc/resources/papers/paapapers/byrne.htm