After two years of development, the Music Library at Harpur College, working in close co-operation with the College's Data Processing Center, has produced its first computer-generated catalog of scores and records. The results of this catalog were so positive that the Music Library has decided to continue its automation efforts.

There are several reasons why this type of catalog was undertaken at Harpur. First, the music collection was relatively small and no adequate system of cataloging had yet been devised. Second, the members of the Music Department staff were enthusiastic about the potential of the cataloging idea, and in this they received excellent co-operation from the Data Processing Center staff.

Because there are many colleges and universities throughout the country which are currently beginning to develop music collections, and because the university community across the nation is becoming the center of computer research, the writer would like to cite some of the advantages of an automated music catalog, in the hope that other institutions might benefit from the project. Available space will not permit a complete technical discussion of the system. However, for those who are interested, this information will be treated in much greater detail in the forthcoming book to be published by Cornell University Press: Computer Assisted Instruction in Music Theory, edited by Professor Harry B. Lincoln.

Before proceeding with a description of the cataloging system, it should be pointed out that the music librarian is still the person responsible for the accuracy and completeness of the basic catalog. Any machine will do only as it is directed, thus making it the librarian's responsibility to define each and every process involved in the cataloging operation.

To insure accuracy and for reasons of efficiency, the cataloging system makes extensive use of numerical coding. All attributes of a given score or record are given six-digit codes. (Attributes include composer, instrument, subject, performer, librettist, editor, compiler, arranger, transcriber and translator). The titles (conventional and distinct), publishing information, call number and any required annotations are written out. The basic process then, is one of decoding the attribute and associating it with the written out text portion of an entry. The attribute, then, is what places the entry in alphabetical order. The virtue of this approach is that once a name, subject or other attribute is assigned a code, it need never be written again. This arrangement insures that all attributes will be standardized and, by the same virtue, that all corrections or changes to an attribute may be executed from a single change. This system allows some 45 additional entries to be made from a single entry, as well as automatically generating cross references from distinct titles to conventional titles. This kind of control over the ordering of the catalog guarantees that the librarian's catalog will be in exact order.

To date, the Harpur Music Library has cataloged over 5,000 items. For every item cataloged there is an average of eight entries generated. The total cost of developing the system and printing a catalog of 40,000 entries may be itemized as follows:

Keypunching	$	2,000
Programming		4,000
Machine Time		6,000
Cataloging Time		12,250
	$	24,650

Note that the first 5,000 items cost $5.00 each to catalog. However, because much of the above cost was developmental, and because the programming cost will not recur, the projected cost of cataloging the next 5,000 items breaks down as follows:

Keypunching	$	1,800
Machine Time		1,200
Cataloging Time		8,000
	$	11,000

This will bring down the cost of cataloging to $2.20 per item. Since the average cost of traditional cataloging is around $5.50 per item, the cost of cataloging the first 5,000 items has really paid for the development of the entire system. The cost of cataloging in the future has been cut by more than 50%.

Other advantages offered to the music librarian by such a system are manifold. Once the catalog data is in machine readable form, a great flexibility is suddenly available. Specialized programming enables the librarian to request specialized catalogs. For example, a catalog of all cantatas on file written between 1700-1750 could be written, or a list of all chamber works including a part for violin edited by Galamian could be made, or a list of works by Beethoven for which a record and a score are on file could be generated. There is no end to the ways one might generate catalogs, be it a merged dictionary catalog, a card catalog or a separate listing. When the time arises that on-line inquiries can be made, then the user will be able to make his own specialized requests. In looking ahead, the same system could be enlarged to catalog books on music as well, which would pave the way for general information retrieval in the field of music.

Cataloging by computer enables the librarian to analyze the process by computer. Accurate logs can be kept on the amount of material cataloged, making cost analyses over any period of time possible. Additionally, a specialized machine checks on the data processed and can thus assure uniformity and correctness of all entries in the two files. Update programs also reject data incorrectly entered and make routine verifications of codes.

In sum, we have found that a greater efficiency in the cataloging process along with a greater diversity of service possibilities results from this system.