Statistical Treatment of Stable Isotope Data

Abstract

In marble provenancing studies, a number of source quarries are characterised by their stable isotope composition. The problem is to assign an artefact of unknown origin to one or other of these sources, on the basis of its composition. For a review of the general problem see Herz (1). This paper is concerned with statistical assignment methods in this context. In line with most practical applications, it is assumed that only two variables (carbon and oxygen isotope ratios) are available. However, all remarks concerning computational procedures follow through to the case in which there are three or more variables, except that visual representation is more complex. The situation in which sources are not known a priori is not discussed here, as it comes under the much wider heading of cluster analysis. Everitt (2) and Aldenderfer and Blashfield (3) give reviews of clustering techniques.