Abstract

The chemical compositions of archaeological metals can be used as a proxy to discern technological choices relating to, e.g., cultural preferences over the use of specific alloy mixtures with different aesthetic and mechanical properties, as well as to infer raw material procurement practices. By analysing large numbers of compositional legacy data concurrently, we can, in turn, simultaneously examine local-scale and cross-regional differences in such material engagements in the past. This talk introduces a new set of tools that are argued to provide for more accurate archaeological reconstructions, than the less rigorous alternatives currently used to model cross-regional and compositional data within the field. These include beta regression, intended to account for the compositional nature of analytical chemical data; as well as multilevel and Gaussian process modelling, used to account for the sampling uncertainties inherent to archaeological legacy datasets, and to permit the explicit modelling of sampling interdependencies and compositional variation through space. The methods are discussed in relation to a new case study on the Muisca metalwork of pre-Hispanic Colombia (AD 600-1600), where the results shown to provide novel insights into the importance of different technological, environmental, and cultural drivers behind alloy choice, as well as into past shared communities of practice. The Bayesian framework is adopted throughout, for both its flexibility in incorporating multiple sources of information concurrently, as well as in providing for intuitive interpretation of model outputs.

All talks are hybrid, to assist online you will need to register for the event by sign-up here