The rural settlements of the Roman Somme (Northern France) are poorly understood in terms of site location. Although the physical landscape of the area is rather smooth, local variations influence...Show moreThe rural settlements of the Roman Somme (Northern France) are poorly understood in terms of site location. Although the physical landscape of the area is rather smooth, local variations influence the distribution of sites. Furthermore, the socio-economic context around Roman farms plays an important part in human behaviours of settlement creation. Predictive modelling constitutes an effective tool for dissecting settlement patterns and understanding their locational parameters through the quantification and evaluation of formal hypotheses. A specific methodology was tailored for the subject and inspired by theory-driven and cognitive predictive modelling approaches. It involves the creation of multivariate models through weighted map algebra, which are then confronted with the distribution of archaeological settlements in four micro-regions along the Somme River. The correlation of the variables with archaeological location indicates that slopes, landforms and the relative distance to rivers are the main influential factors of the physical environment. Socio-economic parameters such as the relative distance to cities, secondary agglomerations and Roman roads are even more influential. Notwithstanding the lack of representation of settlements in the Late Roman period, site location follows similar trends from the 1st century AD to the end of the 4th century AD. Villas prefer economically well connected locations, as do stone-built and post-built settlements. Nevertheless, no parameter can be considered as deterministic in site location. This demonstrates the diversity of choices and influences which favoured the creation of Roman sites in the landscape.Show less
