The data stored in the similarity matrix (N times N) can also be used to generate tree graphs (genealogical trees). The methods of hierarchical agglomerative classification can be used for this purpose. There are many useful algorithms, of which only the methods Ward and Complete Linkage are used here. The calculated trees always contain N binary branches, which are generated starting from the N "leaves" of the "tree" by pairwise mergers. Finally, from the structure of the tree the ramifications located near the root (or trunk) are used for purposes of geolinguistic classification.