Abstract

Library of Congress MARC II records provide partial conversion of a library's holdings to machine readable form (i.e. English language monographs cataloged at Library of Congress starting in 1968). But a library wanting a machine readable file of its entire holdings must decide how to supplement its MARC II records. Basically the choice is whether to use the library's catalog cards to create these supplementary records or to select records from a machine readable data base other than MARC, one example being the University of California Union Catalog Supplement (UCUCS) data base.

606 sample computer printed records from the 800,000 unique titles acquired by the University of California 1962–1968 which constitute the UCUCS magnetic tape file were used to estimate collection overlap between UCUCS (source file) and three academic libraries (target files) comprising respectively 70,000 titles, 280,000 titles and 1,000,000 titles. General cost equations were developed for retrospective conversion of a target library's card catalog to a machine readable data base using an already existing source data base. Estimated collection overlap values and reported costs for creation and selection of UCUCS records were inserted in these equations. Costs from various groups presently using other machine readable data bases are cited in support of using the UCUCS data.

Several simple inequalities involving selection costs, number of records in the target file, and amount of collection overlap between source and target were developed, both for selecting from the entire source file, or alternately, selecting from a limited part of the source file. They provide the librarian with a criterion for deciding whether it is more economical to use an existing file than to create the target library's entire retrospective data base without using an existing file.

Article Outline

• References