From empirical data on the authorship of scientific papers, Alfred Lotka deduced on inverse-square law relating the number of authors of scientific papers to the number of papers written by each author. A basic assumption underlying Lotka's law is that the number of papers published by a scientist is a measure of his contribution to science. This assumption is debatable. In this paper Lotka's law is applied to the literature of computer science. The inconsistent results of earlier attempts to apply Lotka's law to the literature of various scientific disciplines, including computer science, are ascribed to the differences in sampling procedure and treatment of multiple authorship.

This article reports an investigation of the technical and economic aspects of the multiple use of bibliographic data and abstracts of articles in machine readable format. Various techniques such as OCR, word processing, and photocomposition are available for data acquisition and transmission. There is also sufficient economic justification to encourage sharing of such data among publishers. But questions of cost sharing, standards copyright, and procedural adjustments in management and accounting are powerful deterrents to the multiple use of publication data both within an organization and among different publishing concerns. The advantages of sharing bibliographic data can be achieved only if publishing concerns in both the public and private sectors are willing to cooperate and adjust.

Attempts to identify the `core' primary journals in the newly emerging discipline of computer science. Two measures of significance have been used to draw up a core list of journals: Productivity in terms of the number of source items drawn from each journal for review in a secondary journal (Computer Reviews); and extent of use, as reflected by the frequency with which the journals are cited in a leading primary journal (IEEE Transactions on Computers) over a period of two years. This study has shown that the primary journal is the most favored medium for scholarly communication, and that computer science journal literature has consolidated into a relatively few very productive and highly cited primary journals. These core journals are indispensable in any specialized collection of computer science literature.