Reviews the information accrual processes in social science experiments. Stress is placed on the need for systematic, organized, and well documented accrual processes. These processes are illustrated by a hypothetical experiment that comprises four years of collected data and three different population samples. This illustration points out that retaining a maximal set of documentation causes the volume of this documentation to approximate that of the data collected. A model is presented that organizes documentation accrual by providing a glossary of key terms, a dictionary that defines all data elements, and codebooks that describe the data elements and transactions affecting them. Use of this model would facilitate data access by researchers, reduce ambiguity in terminology, and provide centralized control over a database and its documentation. 20 pp. Bibliog.