A method is explained for analysing square matrices of statistics giving transactions between each member of a set of nations, papers, journals, etc. In general self-transactions are different in kind to other exchanges of money, citations, etc., and a special method is given to compute row and column coefficients without relying on the diagonal elements. It is shown that this method yields very satisfactory analyses for journal and national citation data, enabling the members of the set to be assigned measures of size, quality and self-interest and a fuzzy set of clustered members from which all data may be derived.