Material analysis is a basic task of archaeometry, but it is problematic in many ways: the analysis should be representative but should not destroy or damage the object: the analysis should be rapid and inexpensive in order to enable the study of large quantities of objects for statistical evaluation. Analyses based upon small subsamples tend to give random results, i.e. results not representative for the entire specimen, but analyses of the entire body should be truly non-destructive.

EDS-XFA (energy dispersive X-ray fluorescence analysis) is one of the few non-destructive analytical tools giving chemical information on the average composition of a surface. Nearly 90% of all natural chemical elements can be detected provided they are present as major or minor constituents. Like all instrumental techniques, EDS-XFA is based upon a comparison of unknown samples with standards of known composition (for details see Bertin, 1978); quantitative analysis is possible when surface conditions (size, morphology, texture, etc.) of unknowns and standards are identical. Regarding archaeological objects, the sample surface should not be treated, smoothed or flattened, but has to be left in its original morphology (which varies from specimen to specimen) and this may influence the analytical results. These are therefore considered to be qualitative and are essentially used to establish groups of similar chemical composition (Fig. 1). This chemical classification, when applied to cylinder seals, is closely associated with mineral/rock species since a clear interdependence exists between mineral species and their composition. The mineralogical classification, however, is, by tradition, based upon physical properties of minerals like optics, hardness, etc., and not primarily on their chemistry.