Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
1034867 | Journal of Anthropological Archaeology | 2016 | 14 Pages |
•Regression-based predictive modelling is used to account for missing settlements.•Insights from a statistical analysis of seal assemblages are used to calibrate the model.•Evidence for the spatial distribution of elaborate seals conforms with model results.•The approach reveals underlying settlement patterns that would have otherwise been missed.•Links are highlighted between intensities of interaction, artefact variability and site hierarchy.
Simulations of spatial interaction in archaeology have been successful in predicting the emergence of central sites, and political and economic hierarchies that match observed long-term settlement patterns. It still remains unclear, however, to what degree such models can effectively allow for uncertainty in the archaeological record, especially when it comes to incomplete and unevenly distributed settlement data, and how best they might incorporate artefact-scale evidence. This paper aims to address these issues, while attempting to tackle widely debated aspects of socio-political organisation and cultural interaction in the prehistoric Cretan landscape at the period immediately before and after the foundation of the first palace of Phaistos, one of the less well documented Bronze Age phases. We employ a simulation of spatial interaction inspired by approaches first developed in urban geography and combine this with regression-based predictive modelling to address the uncertainty introduced by missing settlements. We use evidence from artefact analysis partly to calibrate and partly to validate our model. We conclude that such an approach can contribute to more convincing archaeological theories about socio-political organisation, cultural affinity and regional identity by providing new evidence even in the presence of very fragmented data.