ABSTRACT
The typological structure that we developed for the Niederwil pottery collection makes it evident that a taxonomic—and not a paradigmatic—typology is the likely outcome when forming a typology based on a data set composed of not well-defined variables and not well-defined data. For data of this kind, the typological task is geared toward determining a subset of the variables relevant to forming categories on the basis of the data brought forward for analysis. There is no reason to assume that types can be defined using a fixed set of variables applied equally to all entities in the data set. More likely, we will find that only some of the measures are relevant to defining types that make up just a portion of the data set.
