Multivariate and Spatial Visualisation of Archaeological Assemblages

• Main Author: Martin Sterry
• Format: Article
• Language: English
• Published: University of York 2018-05-01
• Series: Internet Archaeology
• Subjects: archaeology; Roman; pottery; Multivariate techniques; GIS; visualisation; Vetera 1; Roman Britain; terra sigillata
• Online Access: http://intarch.ac.uk/journal/issue50/15/index.html

Multivariate analyses, in particular correspondence analysis (CA), have become a standard exploratory tool for analysing and interpreting variance in archaeological assemblages. While they have greatly helped analysts, they unfortunately remain abstract to the viewer, all the more so if the viewer has little or no experience with multivariate statistics. A second issue with these analyses can arise from the detachment of archaeological material from its geo-referenced location and typically considered only in terms of arbitrary classifications (e.g. North Europe, Central Europe, South Europe) instead of the full range of local conditions (e.g. proximity to other assemblages, relationships with other spatial phenomena). This article addresses these issues by presenting a novel method for spatially visualising CA so that these analyses can be interpreted intuitively. The method works by transforming the resultant bi-plots of the CA into colour maps using the HSV colour model, in which the similarity and difference between assemblages directly corresponds to the similarity and difference of the colours used to display them. Utilising two datasets – ceramics from the excavations of the Roman fortress of Vetera I, and terra sigillata forms collected as part of 'The Samian Project' – the article demonstrates how the method is applied and how it can be used to draw out spatial and temporal trends.