College and Research Libraries The Costs of Data Processing in University Libraries ON JuLY 18 the membership of ACRL's University Libraries Section heard a panel discuss the costs of data processing in university libraries. Don S. Culbert- son, head of data processing at the Uni- versity of Illinois library, Chicago Un- dergraduate Division, discussed data processing costs in book acquisitions and cataloging; Melvin J. Voigt, director of libraries at the University of California at San Diego, compared the costs of man- ual serial records with those of computer- based serial records; James R. Cox, head of the circulation department at UCLA, examined the costs of operating circu- lation act.ivities by manual, semi-auto- mated, and fully automated systems_ Since this panel developed some new cost data, not previously published, it was deemed desirable to print its papers here in full.-D. K. In Book Acquisition and Cataloging A STORE WINDOW on Michigan Avenue in Chicago recently displayed several pen- guins and a small pool. These birds have developed their strange, unique shape through living for generations on end in an isolated place with few natural ene- mies . . When enemy finally appeared in the form of man, they were unable to adapt to this new danger and were easily slaughtered. In a new environment these birds were obsolete. John G. Kemeny of Dartmouth is one of the latest in a long line of individuals who have followed the lead set by Van- nevar Bush nearly twenty years ago in predicting that in their present form libraries are as obsolete as penguins, and that librarians cannot cope with infor- mation being produced at its present rate. They also predict that unless librar- ians mend their ways and adapt to mod- ern conditions they and their libraries may soon be as extinct as the dodo bird. On what do the prophets of the magic libraries base their predictions of the NOVEMBER 1963 obsolescence of libraries? While an im- portant recognized factor has been the obvious and much discussed increase in publication, another factor that has been virtually ignored for many years is only now coming under close scrutiny. This is the high cost of processing. We all have arrearages in acquisitions and cat- aloging; we have book collections parked on ~torage shelves or packed into boxes awaiting-frequently for years--some kind of action; we have collections of technical reports that we do not know how to handle; thousands of government documents are rarely used becaus~ access to them is made difficult by skimpy in- dexing. Why do such conditions exist? The answer is readily known. They exist be- cause we need more staff members to process more i terns. We need more ac- quisitions librarians, more catalogers~ more clerks, more of everything. But where will this end? Eighty years ago three out of every four library dollars 487 were spent for books; today three-fourths of our expenditures are for staff. Admit- tedly public services have been much expanded during this period, but they cannot account for such a complete re- versal of the ratio. How much out of our total budgets should we be spending for books? Who has the magic formula to which we. can compare our library to see if we are good, bad, or indifferent? I checked to see if answers had been published in the literature. My findings were that little really helpful information has been writ- ten on library costs. Much of what exists has appeared in periodicals which are not widely circulated; methodology is uneven; data are totally incompatible from one study to another; and informa- tion more than ten years old is of rela- tively little value. I would like, however, to mention four studies which have been made fairly recently and which are of four different types. In July 1956 Bella Shachtman re- ported a study made at the USDA library on the collection of cataloging statistics. This study, while it does not contain cost information, is the raw material which must be first collected and sum- marized before a cost analysis can be at- tempted. Quantitative data, such as the percentage of titles requiring original cataloging (99 per cent) compared with those for which Library of Congress copy is available; the number of titles which can be searched in an hour (fifteen); the number of titles which can be cataloged in an hour (two); and the number of cards a professional cataloger can file in an hour (eighty), are needed to complete the picture. More information is avail- able in this article including the check- ing form each cataloger filled in at the end of the day. Mrs. Catherine MacQuarrie was chair- man of a committee which made a cost survey of southern California libraries. Its report, made in the spring of 1961, covered public and college libraries in the area. Of six college libraries report- ing, cataloging costs rangec! from $1.08 to $2.88 per title, and processing costs ranged from $.11 to $.91 per title. Among twenty-three public libraries, cataloging costs varied from $.76 per title to $7.411 to this may be added book preparation costs from $.09 to $.54. A detailed cost computation form is included in this re- port which might be an excellent start- ing point for a library considering a sim- ilar study. Fred Heinritz recently compared the cost of a book catalog to the cost of a card catalog as his doctoral dissertation at Rutgers University. His comparison was made between the production of the Engineering Index and its cumulation in the bound form as the book catalog, and the cost of producing the Engineering Index on cards. Whether or not these two catalogs are actually comparable to a library book catalog and a library card catalog only additional studies of this nature will tell. His .conclusions are that book catalogs are much cheaper to make and more difficult to use. Louis Schultheiss and I made a cost study in early 1961 of the University of Illinois library of the Chicago Under- graduate Division. Our cost of cataloging at this time was $8 .07 per title, and our cost of acquisition was $4.55 per title. We did not include the cost of maintain- ing our building in these figures . What do these studies indicate? First, they indicate that we need some kind of standard method for making reports of this nature; standard report forms, stan- dard procedures for collecting data, and standard ways of cumulating and dissem- inating this information. Second, they seem to indicate that we need a central agency which can be responsible for prodding librarians who do not turn in their statistics. Third, they show that we need norms to which each librarian can compare his own performance so that he knows whether or not changes in tech- nique or personnel will improve or harm 488 COLLEGE AND RESEARCH LIBRARIES the library. The ALA now has several clearinghouses for information of this kind under consideration. We must en- courage the expansion of these efforts into the establishment of one large inter- divisional agency which can perform all the functions which might go into the several smaller ones. THE usE of data processing methods for various aspects of library operation is be- ing considered by many libraries today. In most areas-acquisitions, cataloging, circulation, for example-the activities are interrelated to such a degree that if advanced data processing methods are to be used efficiently, it appears that they must be applied in a coordinated system. The processing ·of serials and the main- tenance of serial records are sufficiently independent areas of library activity to make a system feasible which is not nec- essarily tied to other aspects of library operation. Also, the repetitive nature of much of serials work lends itself to mech- anization. The difficulty large libraries have in providing holdings and receipt information for their users suggests that by the use of data processing, needed services can be provided at little or no additional cost. At the outset, I would like to make clear that this discussion. will be limited to serials processing using computers. It is also possible to handle serials with IBM cards and standard sorting and printing equipment. Such procedures may be satisfactory for small · libraries but mechanical card equipment does not have the capacity, speed, or efficiency necessary for . serials control involving the thousands of titles found in most NOVEMBER 1963 James Skipper, now president of R TSD, stated when he accepted the gavel, that the future belongs to tech- nical services. He is right. Unless we get our costs under control and become able to keep them there, the whole library risks belonging to technical services.- Don S. Culbertson. In Serials Handling university libraries. To maintain up-to- date holdings records, produce receipt and holdings lists, and to create records for the receipt of current issues, for bind- ing, and for subscriptions, requires a, computer with a large memory and which, for efficiency, operates at a high speed. This panel has been asked to discuss the cost of data processing operations and to compare them with those of man- ual systems. For serials, as in the other areas, this is a difficult task for a number of reasons. First, only a few libraries have started serial computer operations and none have applied them thus far to· more than a few thousand titles; second,. no thorough studies of costs have been completed; and third, as in most areas. of library work, there are no accepted standards or even any experimental stud- ies of work output or unit costs for any of the parts of traditional serials process- ing. Thus what can be reported here must be considered 1 as tentative. Cost data are based primarily on evidence from a single operation, which is just emerging from the pilot plant stage. In this brief report I do not intend to describe serials operation through com- puters, . but I would refer those who ~ould like information on its possibili- ties and development to the article pub-· 489 lished in Library Resources and Tech- nical Services~ in the Winter 1963 issue. 1 A further report on this . project will be issued in about six months. The costs involved may be divided into three categories. The first is the labor cost of serials processing and main- tenance. The second is computer cost in terms of computer time and cost per unit of time. The third is the cost of com- puter program development and main- tenance. Looking first at labor costs, there are a number of processes for which it should be possible to make comparisons. The cataloging of a new serial title or the recataloging of an old title is virtu- ally the same process in a manual system as in one utilizing computers. Entry must be established and holdings determined. If reference cards are to be maintained in the catalog for main and added en- tries, these must be produced. Instead of a holdings card for a central serials rec- ord, information must be prepared for the key punch operator. We call the form which results the "intermediate serials record." Producing this form from an already completed serials record card takes a serials cataloger about three and one-half minutes. At three dollars per hour this would be about $.175 per title. This includes the time required to as- sign a mnemonic title and a serial num- ber. The only additional task requiring labor prior to converting the record to tape is the key-punching time . In our operation, entries average 2Y2 cards, punching time for an operator whether