4.2 Seriation

It is very difficult to assess a table visually (4.1) with artefact types per site, even with these relatively small numbers. The numerous empty cells make the table appear fragmented. In spite of the conscious arrangement of the artefact types, no clear picture of any site types emerges from the table. It is not immediately clear whether there are sites with more points (variables 3-5), tools (6-11), settlement indicators (12-14) or more flint working (15-17).

Another way of arranging this information and one that might provide a better picture of the site types, is seriation. Manual and automated techniques for this have been used for a long time for several archaeological questions (dating, typology, social status), at various spatial levels (regional, settlements and burial fields) (i.e. Herzog & Scollar 1987; Houtsma et al. 1996). Seriation techniques arrange both the variables (columns) and observations (rows) in such a way as to display the pattern/structure in the data as best as is possible. In order to illustrate the position of seriation among other multivariate methods, we may state in a more abstract sense that seriation arranges the sites (observations) on a line (one-dimensional space), with equal distances between observations. The archaeologist will interpret this sequence for example as a chronological or social arrangement.

Fig. 36 Seriation of the artefact composition of the Michelsberg sites. Artefact types which occur on one single site and sites that consist of one artefact type only had to be excluded from the analyses