Finding Roman brickyards in Germania Superior by model-based cluster analysis of archaeometric data

In: Posluschny, A. ; Lambers, K. ; Herzog, I. (Hrsgg.): Layers of Perception. Proceedings of the 35th International Conference on Computer Applications and Quantitative Methods in Archaeology (CAA), Berlin, 2.-6. April 2007. Koll. Vor- u. Frühgesch. 10. Bonn 2008, pp. 360-365

PDF, English Download (2MB) | Terms of use

Abstract

Chemical analysis of ancient ceramics and of other archaeologically important materials has been used frequently to support archaeological research. Often the dimensionality of the measurements has been high. Therefore, multivariate statistical techniques such as cluster analysis have to be applied. The aim of the present paper is to give a review of the research on bricks and tiles from Roman military brickyards in Germania Superior and to present the main results obtained by multivariate statistical analysis. In particular, new adaptive cluster analysis methods and modified model-based clustering are applied on archaeometric data (Mucha / Bartel / D olata 2002; 2003a; 2005b; in press; Bartel / D olata / M ucha 2000; 2003). The main result was the discovery of military brickyards that were not known when the project began about ten years ago. Recently, they have been discovered by the application of these multivariate statistical analysis models. Newly developed visualization methods support and facilitate the interpretation of both the data set and the results of grouping. This means archaeologists can easily identify a new finding of a Roman brick or tile by comparing its chemical fingerprint with those from the detected provenances.