Abstract
This paper considers two recently developed mathematical techniques in archaeological seriation. Both techniques take as given, a similarity matrix between the objects to be ordered. The first method is a "linkage" procedure, while the second (and preferred) is an "averaging" procedure. The second method, though a bit more mathematical, has been very successful in application to real data. Both techniques have several good features which will be considered. For both illustrative and comparative purposes, the methods have been applied to several sets of archaeological data considered by other researchers. As the comparisons will show, the results have been surprisingly good. More specifically, for such comparisons to be made, a criterion for comparing the "goodness" of different serial orders for a particular set of data is of great interest. A criterion, or "goodness of fit" index of this type will be developed and interpreted as well as used for the comparisons of interest. Lastly, a section is devoted to more general discussion of what sorts of data are suitable for reduction by seriation techniques.