Lead (Pb) isotope signatures for silcrete sources from the Willandra Lakes region, Australia: A pilot study of a new method for provenancing silcrete artefacts

Highlights

• Patterned variation in Pb isotope composition characterizes these silcrete sources.

• The variation is randomly distributed within sources but signatures differ between sources.

• The sources with the greatest geographic separation exhibit the greatest differences.

Abstract

Silcrete was often used to make stone tools and the ubiquity of this material in the archaeological record has sparked considerable interest in developing techniques that can be used to trace its geographic origin. However, the highly variable physical and chemical properties of silcrete means that artefacts made from this raw material have proved difficult to provenance. This paper describes the use of Pb isotope analysis to characterize and differentiate silcrete sources in the Willandra Lakes region, a UNESCO World Heritage listed site in southeastern Australia. The sample collection strategy employed in the field has allowed Pb isotope variation both within and between the silcrete sources to be described. Pb isotope variation within each silcrete source does not exhibit spatial patterning, but Pb isotope signatures differ between silcrete sources in the Willandra Lakes region, and clear separation between more distant sources, is demonstrated. This represents a first step in being able to use isotope analysis to investigate how silcrete from different sources was used and how it was moved around the landscape.

Introduction

Archaeologists have long been interested in identifying the rock outcrops that provided the raw materials used to make stone tools. The distribution and characteristics of raw material sources are fundamental to understanding the strategies people employed to ensure that they had raw materials and/or tools available where and when they needed them (Andrefsky, 1994). The types of stone artefact assemblages found in different parts of the landscape can provide information about the extent of people's home range, the form in which raw material was carried around the landscape and the stone working activities that were undertaken in different locations (Isaac, 1986). Stone artefacts discarded at greater distances from raw material sources tend to be used more intensively, and are therefore smaller, than stone artefacts discarded near their source (Newman, 1994). Deviations from this pattern provide the opportunity to investigate the influence of other social and ecological factors, such as access to other resources and the presence of trade and exchange networks (e.g. Blumenschine et al., 2008; McBryde, 1984). The ability to determine the origin of stone raw materials provides the basis for interpreting patterns of stone artefact discard in these terms.

However, some types of rock are more amenable to sourcing than are others. Silcrete, which is a sedimentary rock with a high silica content and low fracture toughness, was used often as a raw material for making stone tools and is particularly common in Australia, southern Africa and western Europe (Nash and Ullyott 2007: 95). Silcrete forms via the silicification of regolith (Thiry and Milnes, 2017) and as a result, retains many of the sedimentological and geochemical characteristics of its parent material (e.g. Webb and Golding, 1998). Regolith tends to be highly variable within and between regions, which explains the high degree of textural and geochemical variation that characterizes silcrete. Individual hand specimens sometimes exhibit substantial variation in both color and texture, making it difficult to distinguish silcrete from different sources on the basis of visual characteristics alone.

Studies attempting to distinguish material originating from different sources of silcrete have relied primarily on variations in trace element compositions. Nash and others (Nash et al., 2013a, Nash et al., 2013b; Nash et al., 2016) explored the potential for using patterns in major and trace element abundance to source silcrete in sub-Saharan Africa, focusing initially on a study of silcrete outcrops in the South African Cape region (Nash et al., 2013a). They showed that variation in major and trace elements is sometimes linked to the bedrock on which the silcrete formed. Although artefacts were not included in the initial study, the results suggested that regions where silcrete has formed in association with a variety of different bedrock lithologies are likely to be better candidates for sourcing studies than regions where the bedrock is homogenous. Subsequently, Nash et al., 2013b, Nash et al., 2016 investigated silcrete sources in the Tsodilo Hills region, which encompasses adjacent regions in Botswana and Namibia, and showed that discriminant function analysis of bulk-sample major and trace element abundances can distinguish outcrops from one another. However, of the 82 artefacts they studied, only 35.4% were assigned to one of the documented sources (although a conservative confidence level of 99% was used). Another 24.4% of the artefacts plot just beyond specific source clusters, leading the authors to suggest that sampling may have been insufficient to capture intra-outcrop variation and that these artefacts were likely to have been procured from the sources to which they plot most closely. They suggest that the remaining 40.2% of artefacts, which are clear outliers on the discriminant function plots, were likely to have been procured from sources that lie beyond the study area. Given that the efficacy of using major and trace element abundance data to source silcrete artefacts has not yet been demonstrated in other studies, this explanation requires empirical confirmation. So, although Nash et al., 2013a, Nash et al., 2013b, Nash et al., 2016 research has demonstrated significant potential for using major and trace element abundance data to source artefacts made from silcrete, the relatively high proportion of artefacts that do not fit within source clusters does need further investigation.

In Australia, silcrete commonly occurs in one of two contexts: in association with basalt (e.g. Webb et al., 2013; Webb and Golding, 1998) or in association with other substrates in arid, inland settings (e.g. Doelman, 2005, Doelman, 2008; Doelman et al., 2001; Thiry and Milnes, 1991). When silcrete forms in association with basalt, the silica is derived from the weathering of the basalt (Webb and Golding, 1998), but in arid, inland contexts like the Willandra Lakes region, the silica is either dissolved out of components of the regolith itself or introduced by fluvial processes from further afield (Thiry and Milnes, 2017). These formation processes are likely to impact on the potential for identifying the source of the material used to make artefacts. Basalt-associated silcretes tend to form in localized settings, deriving silica from basalt deposits that often have distinct genetic histories, and thus distinct geochemical characteristics, which may create silcretes with distinct geochemical characteristics. However, arid, inland silcrete outcrops often represent ground surface exposures of expansive deposits that have a shared genetic history (i.e. they may be part of the same sand-sheet, some or all of which has become silicified by broad-scale processes such as ground water activity). As a result, different arid, inland outcrops may be more difficult to differentiate from one another than different basalt-associated silcrete outcrops.

Cochrane et al. (2017) investigated the potential for sourcing artefacts made from basalt-associated silcrete from the Arcadia Valley in the central highlands of Queensland, Australia. Portable XRF analysis was applied to 86 silcrete cobbles collected from three different sources, and to 100 artefacts from nearby surface accumulations. The samples from the three sources plotted in clusters on the basis of their median iron (Fe) and zirconium (Zr) concentrations. Although these clusters are not discrete, they are sufficient to demonstrate patterned variation, which provides the rationale for employing Fe and Zr element concentration data as a basis for sourcing the artefacts. The Fe and Zr concentrations of 75% of the artefacts plot within the 95% confidence ellipses identified for two of the three sources. None of the artefacts plot within the 95% confidence ellipse identified for the third source, indicating that this source was probably not exploited despite being the same geographical distance from the surface artefact accumulations as one of the other sources. With respect to the two utilised sources, it is worth noting that the artefacts plot in a continuous distribution, rather than producing clusters that correlate with the geological sample data for these two sources. If the sources are distinct from one another, and each artefact originated from one of these two sources, it follows that they should cluster with the geological sample data, even if the clusters are quite loose. The continuous distribution indicates that further investigation is needed. Nevertheless, the data highlight the possibility of identifying the sources of artefacts made from basalt-associated silcrete. Furthermore, by ruling out one of the potential sources, Cochrane et al. (2017) were able to determine that geographical proximity was not the most important factor governing raw material source preference in the Arcadia Valley.

Silcrete was the raw material most often used to make stone tools in the Willandra Lakes region (Stern et al., 2013), a relict overflow system located in southwest New South Wales, Australia (Fig. 1). The sedimentary sequence in this area preserves archaeological sites with an antiquity of at least 45 ka as well as the world's oldest-known ritual ochre burial and the oldest-known cremation (Bowler et al., 2003). In 1981, these dry lakes and their immediate surroundings were placed on the United Nations Educational Scientific and Cultural Organization (UNESCO) World Heritage List because of their outstanding natural, and exceptional cultural, features. The archaeological research conducted in this region has focused primarily on the archaeological traces incorporated into the Mungo lunette, a large, crescentic dune that built up on the leeward side of Lake Mungo, a large, terminal lake within the overflow system (e.g. Bowler, 1998; Bowler et al., 1970; Shawcross, 1998; Spry, 2014; Stern, 2014, Stern, 2015; Stern et al., 2013; Tumney, 2011). During survey work conducted as part of a larger research program, numerous outcrops of silcrete were identified in the area, several of which exhibit signs of having been used as sources of raw material for making stone tools (Kurpiel, 2017). An understanding of the geographic origin of the silcrete used to make tools has the potential to provide crucial insights into the way that people moved around this landscape. For example, it could provide detailed insight into the ways mobility strategies changed in response to local environmental and hydrological conditions. Visually, there is a high degree of variation both within and between sources. Geochemical techniques, such as Pb isotope analysis, offer a more rigorous approach to the problem of provenancing silcrete than do physical characteristics alone.

Pb isotope studies have a long history in the earth sciences and are used routinely to characterize magmatic rocks and base metal ore deposits (e.g. Gulson, 1986; Sun, 1980) and to provenance sediments (McLennan et al., 1993). Variations in Pb isotope ratios of rocks, minerals and fluids result from the long-term accumulation of radiogenic ²⁰⁶Pb, ²⁰⁷Pb and ²⁰⁸Pb from radioactive decay of ²³⁸U, ²³⁵U and ²³²Th, respectively, with half-lives of 0.7, 4.4 and 14.1 billion years. Pb isotopic variations exist on all scales and depend on the age, U-Th-Pb concentrations and origin of a given Pb-bearing material (e.g. Faure and Mensing, 2005), providing an excellent isotopic tracer for metals and other Pb-bearing materials. In archaeology, Pb isotopic variations are used to provenance metal artefacts (e.g. Baron et al., 2011; Ling et al., 2013; Shortland, 2006; Stos-Gale et al., 1997; Ponting et al., 2003), pottery (e.g. Renson et al., 2011; Wolf et al., 2003) and glass (e.g. Henderson et al., 2005).

Pb isotope studies of rocks and minerals used in tool-making are much less common. Weisler and Woodhead (1995) compared Pb isotope compositions of six basalt artefacts from Henderson Island, a limestone island in the Pitcairn group, with 18 basalt samples collected from deposits scattered across southeastern Polynesia. This comparison placed the origin of five of the Henderson Island artefacts at nearby Pitcairn Island, while the sixth could be traced to the Gambiers, some 400 km to the west. This provided evidence that the Pitcairn islands were colonized by people from the Gambiers. Differences in Pb isotope composition were also noted between ‘shield’ and ‘post-shield’ basalt deposits from the same island, illustrating the power of Pb isotopes as a provenancing tool. In another study of this type, Collerson and Weisler (2007) combined isotopic data with major and trace element data to compare Polynesian adzes with possible source locations. Three of the nineteen adzes could be traced precisely, and others could be assigned to likely source regions. The results were consistent with Hawaiian oral histories that describe prehistoric voyages from Hawai'i to Tahiti and back via the Tuamotus.

More recently, ten Bruggencate and others (ten Bruggencate et al., 2013, ten Bruggencate et al., 2014) carried out a provenance study of quartz in the Churchill River basin, in central Canada. In this area, quartz was used to make tools during the pre-contact period and was procured from several different sources. In order to explore the geochemical variability of these sources, ten Bruggencate et al. (2013) measured Pb isotopes and a small selection of trace elements from seven sources of quartz within a 25 km² area. A combination of color, Pb isotope compositions and Ti-Ge concentrations were found to distinguish most of the quartz source locations from one another.

In this contribution, we investigate the Pb isotope composition of six silcrete outcrops that were used as sources of raw material to establish the variation that exists within and between silcrete sources in the Willandra Lakes region. Such work must be undertaken before inferences can be made regarding the origin of unprovenanced materials (Oulhote et al., 2011: 303), such as stone tools. To the best of our knowledge the work presented here is the first application of Pb isotopes to the problem of sourcing silcretes used to make stone tools.