Logic and Landscapes: Simulating Surface Archaeological Record Formation in Western New South Wales, Australia

Abstract

Surface archaeological deposits are ubiquitous in arid Australia, but using them to reconstruct human behaviour and chronological sequences is made difficult by stratigraphic mixing. Contrasting interpretations of late Holocene human activity in the region have emerged, alternatively suggesting high levels of mobility and frequent hiatuses in occupation, or greater sedentism and population growth, based on equivocal archaeological patterning. The latter feature in narratives of socioeconomic intensification in Australian prehistory, prompting two questions that guide this thesis: is directional change needed to explain patterns in the surface record in western New South Wales and how can we know? Two simplified models of formation processes, formalised as agent-based simulations, are used to resolve differences in interpretations by evaluating their logic in an explicit framework and generating tests. The first, called HMODEL, is based on the concept of the archaeological palimpsest and demonstrates that infrequent, high intensity sedimentation events can generate patterning qualitatively similar to that explained elsewhere in behavioural terms. The second model, called FMODEL, features agents reducing stone and distributing it across a simulated space to examine the influence of movement patterns on the ratio of cortical to non-cortical stone (the Cortex Ratio) in flaked stone assemblages. Outcomes suggest that movement can influence the variability of ratio values, while systematic shifts depend on the consistent import or export of stone. Patterns observed in a large-scale surface survey at Rutherfords Creek, western New South Wales, are considered in light of simulation outcomes. Alternative models of hearth formation are compared to establish how each explains patterns in radiocarbon chronologies. Predictions from these models are tested against a second chronometric proxy, showing that patterning in the data is more aligned with geomorphological change operating on a consistent record of occupation than any detectable changes in population dynamics. The results of FMODEL further show that patterning in at Rutherfords Creek can be explained in large part by the relative abundances of raw material in the region. Both finds are consistent with a model of regularly occurring, short-term occupations over the course of the late Holocene. These demonstrate the potential for exploratory modelling of formation processes to resolve issues concerning contrasting interpretations of the past.