Multivariate cluster analysis in geology

The article discusses the role of cluster analysis in the sciences of classiology, defines clusters using text as well as logical and mathematical formulas. For the first time, the ideas and methods of cluster analysis are discussed side by side with the graph theory. The algorithm is defined for the automatic recognition of multivariate clusters by joining string clusters - clusters with have a common point (a row in the table of point-to-point distances). Separately, the following special issues are discussed: the choice of critical path, clustergrams, transformation of input data for cluster analysis, and interpretation of plain (two-dimensional) images of multivariate cluster analysis.