The question of optimal disk allocation for Cartesian product files is considered. Coding theory is used to analyze disk allocation methods and their optimality under certain conditions. The approach is strictly mathematical, and realistic workloads and performance measures are not considered. The conventional assumptions about the uniformity of query and attribute distributions are made. Since these assumptions do not match realistic applications, it is not clear how useful the results would be in practice.

This paper is a slightly expanded version of a conference paper [1]. Although this conference paper is included in the references, it is not acknowledged as an earlier publication of essentially the same material. It should have been. The utility of our archival literature is degraded by the all-too-common practices of duplicate publication and failure to acknowledge earlier versions.

The paper should interest theoretical researchers in the area of file organization.