THE DERIVATION AND APPLICATION OF THE BRADFORD‐ZIPF DISTRIBUTION

Abstract

Any statistical regularities found in documentation should be fully exploited to produce estimates or predictions and to save documentalists work. But present formulations of the Bradford distribution demand penetrating search for peripheral papers and tedious computation in application. The present paper shows that the Bradford distribution is closely related to the Zipf distribution. It requires data on only the most productive journals, is mathematically simple and amenable to graphical methods if a proposed idea of the ‘completeness’ of a search is accepted. For comparability of results, certain conditions, which include a specified minimum level of productivity of journals, need to be standardized. A standard form is suggested. It is found, however, that a modified form of the Bradford distribution is required when Bradford‐type collections of journals are merged into larger collections, when ‘saturation’ of the most productive journals occurs.