This edited volume aims to demonstrate that there is indeed something special about this method, something that makes it curiously attractive to diverse and sometimes conflicting interests and approaches in GIScience. Those interested in clustering and classification will recognize in it elements of k-means clustering, but with an explicit representation of topological relationships between clusters. Anyone accustomed to dealing with ndimensional data through a transformation and reduction of variables, as in principal components analysis (PCA) or multidimensional scaling, will tend to interpret the SOM method in that light