Towards a predictive model of early medieval settlement location: a case study from the Vale of Glamorgan.

Abstract

Early medieval (5th- to 11th-century) settlement sites are extremely rare in Wales, particularly in comparison to England and Ireland. The lack of sites is confounded by a bias within the existing dataset towards what appear to have been high-status defended settlements of the late 5th to 7th centuries such as Dinas Powys (Glamorgan), Coygan Camp (Carmarthenshire) and Dinas Emrys (Gwynedd). Current understanding of site types, morphology, settlement patterns and hierarchies is therefore very superficial. In this paper the author explores one possible way of addressing the problem by constructing a predictive model, using an area of twenty parishes in the eastern part of the Vale of Glamorgan corresponding roughly to the Dinas Powys hundred as a case study. By mapping the distribution of Romano-British and medieval settlement evidence in relation to landscape characteristics which are likely to have structured the nature of agricultural practices it was possible to divide the study area into zones reflecting past settlement intensity, and to identify which would be the most profitable areas for survey and targeted trial excavation aimed at identifying early medieval settlements. Large numbers of settlements of all periods are likely to be discovered through fieldwork in these zones, a problem compounded by the lack of a widespread diagnostic early medieval material culture to assist in the identification of sites. Thus the validity of areas identified as having high potential for early medieval settlement can only be tested through the implementation of large-scale survey methods such as fieldwalking and geophysical survey in conjunction with techniques which can identify ephemeral settlement evidence, such as metal detector survey and trial excavation with regular radiocarbon dating. The model may also have further applications in the examination of patterns of land-use.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 21 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.