LIBRARY OF THE UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN 025 C64p 1973 cop. 3 Latest Date stamped below. Theft, mutilation, and underlining of book, reason, for di.ciplinary action and may in di.mi.sal from the University. L161 0-1096 Proceedings of the 1973 Clinic on Library Applications of Data Processing: Networking and Other Forms of Cooperation PROCEEDINGS OF THE 1973 CLINIC ON LIBRARY APPLICATIONS OF DATA PROCESSING: NETWORKING AND OTHER FORMS OF COOPERATION Edited by F. WILFRID LANCASTER Papers Presented at the 1973 Clinic on Library Applications of Data Processing, April 29-May 2, 1973 University of Illinois Graduate School of Library Science Urbana-Champaign, Illinois 1973 by the Board of Trustees of the University of Illinois LC Card Number: 65-1841 ISBN: 0-87845-038-6 U.S. ISSN: 0069-4789 ?3 INTRODUCTION The tenth annual Clinic on Library Applications of Data Processing was devoted to the subject of cooperation between libraries in data processing activities. This topic is one of great current interest and one in which there is now considerable activity. While many library operations involve repetitive, routine tasks that lend themselves readily to automation, not all libraries have funds available to allow the design and implementation of automated pro- cedures. However, by inter-institutional cooperation, at national, state or local levels, certain automated procedures become both feasible and economical. The feasibility and the economics of such cooperation were covered by the speakers at this clinic. The papers presented are of broad scope, both from the viewpoint of approaches to cooperation and types of library involved. Major emphasis was placed on data processing within library networks and in cooperative process- ing centers. General overviews on library networking operations are presented by Becker and by McCarn, the latter with special reference to the role of MEDLINE in the National Biomedical Communications Network. The com- ponents and configurations of library networks are described by Evans, with emphasis on experience and accomplishments in New York State. Long, Hammer and Sokoloski, and Wright deal with three major programs in cooperative or centralized processing: the Ohio College Library Center, the Massachusetts Central Library Processing Service, and the College Bibliocentre of Ontario. Activities within the federal government are also represented. Cylke discusses cooperation in automation activities among the three national libraries, while Henderson presents the results of a survey of automation in libraries within the federal government and goes on to discuss the possibility of a cooperative processing center to serve these libraries. vi 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING The final group of papers is devoted to cooperation among libraries on a somewhat smaller scale or in more specialized environments. A program for coordinating automation activities within the libraries of the state universities of Iowa is described by Sage. Brodman discusses several data processing activities implemented at Washington University, School of Medicine, to produce various products of use to other medical libraries. An important aspect of her paper is the account of problems encountered in these activities, and lessons learned from them. Cooperative data processing within the industrial library environment is rare and, as Randall points out in his description of cooperative activities within IBM, difficult. Perhaps the most specialized paper in the volume is Jolliffe's description of Project LOG, a cooperative venture between the British Museum, the Bodleian Library at Oxford and the Cambridge University Library, which has as its objective the production of a catalog of the early books held by the individual colleges at Oxford and Cambridge. Despite an airline strike, which made travel to Urbana difficult and tiresome for many, the clinic speakers had an interested audience of over eighty registrants, drawn widely from the United States, Canada and further afield. F. WILFRID LANCASTER Chairman and Editor June 1973 TABLE OF CONTENTS INTRODUCTION v F. WILFRID LANCASTER NETWORK-OR ALL HANG SEPARATELY 1 DAVIS B. McCARN BACKING INTO NETWORK OPERATIONS 9 ESTELLE BRODMAN UTILIZATION OF THE MARC II FORMAT FOR SERIALS IN AN INTER-UNIVERSITY ENVIRONMENT 24 CHARLES R. SAGE U.S. NATIONAL LIBRARIES TASK FORCE: A REVIEW OF DATA PROCESSING INTERESTS 32 FRANK KURT CYLKE PROSPECTUS FOR A FEDERAL LIBRARY COOPERATIVE CENTER 56 MADELINE M. HENDERSON AN ONTARIO LIBRARIES' NETWORK, OR COOPERATIVE ENTANGLEMENT 68 GORDON H. WRIGHT PROJECT LOC: CENTRALIZED PROCESSING OF LOCAL COLLECTIONS 102 J. W. JOLLIFFE INTERLIBRARY COOPERATION IN AN INDUSTRIAL ENVIRONMENT 113 G. E. RANDALL THE MASSACHUSETTS CENTRAL LIBRARY PROCESSING SERVICE 124 DONALD P. HAMMER AND JAMES S. SOKOLOSKI BIBLIOGRAPHIC DATA CENTERS FOR NEW YORK STATE 150 GLYN T. EVANS OCLC. FROM CONCEPT TO FUNCTIONING NETWORK 165 PHILIP L. LONG LIBRARY NETWORKS: THE BEACON LIGHTS 171 JOSEPH BECKER LIST OF ACRONYMS 180 INDEX . 183 DAVIS B. McCARN Acting Associate Director of Science Communications and Computer Engineering Services The National Library of Medicine Bethesda, Maryland Network Or All Hang Separately "They that weave net-works shall be confounded." Isaiah XIX.9(1611) Over the past few years, there has been increasing talk of "networking." This word has meant two quite different things in library usage. First, it has meant resource sharing, efforts to reduce the cost of duplicating facilities and collections through primarily interlibrary loan agreements. Second, it has meant distribution through telecommunications of information services. The latter sense is the more recent, but the older sense is the more important because the telecommunications systems now emerging promise to allow a new age of library cooperation. Before describing this new promise, I would like to present my reasons for believing the fulfillment of the promise to be nearly ineluctable. The library is one of the remaining labor intensive areas of our social endeavor, which means that the labor costs of running a library make up the major cost of operations. For smaller libraries, labor represents a smaller cost than that of acquiring and housing a collection. In larger libraries, the cost of labor exceeds all other costs. The cost of books is rising rapidly. The Bowker Annual estimates the price of books is rising 15 percent per year, 1 twice as fast as the cost of living! Other prices, including subscriptions and housing costs are also rising. But labor costs are also rising, and so far, there are very few ways to offset these rising costs through labor-saving devices or proce- dures. There have, of course, been many efforts to apply automation. It is my impression that these efforts have largely failed to provide real cost benefits. 2 7975 CLINIC ON APPLICA TIONS OF DA TA PROCESSING Certainly, the efforts of the System Development Corporation to sell ALPS (Advanced Library Processing System) and IBM's effort on ELMS (Experimen- tal Library Management System) have not been resounding successes. I believe that the primary reason for the lack of success of these systems is that they really have not yet offered major cost savings. Neither of these nor others designed for operation in one library can make any real dent in the million dollars a day spent on cataloging in the United States. Automation has been successful, however, in another way; it has opened the door for a whole new series of information services the on-line retrieval services. I will not describe the on-line services of the National Library of Medicine nor their background; these have been covered in detail in another recent article. 2 I would like to discuss one aspect of the service: its confound- ing growth rate. Figure 1 shows the growth rate of the use of MEDLINE since June 1972 in terms of annual search rates. MEDLINE is now operating at a rate of 240,000 searches per year. Since June 1972, its use has grown at a rate of 13 percent per month, doubling every six months. February was up 30 percent from January, and March was up 30 percent from February. When searches were done in batches in 3-6 weeks, MEDLARS provided 16,700 searches in its peak year. MEDLINE is now operating at nearly fifteen times this rate. This increase has been cost-effective. MEDLINE service is provided with the same budget previously used for MEDLARS. But MEDLINE is only one such information service. When it was first offered over a nationwide network in February 1972, it was the first such nationwide on-line bibliographic service. It now has many complementary services. In this area, there are now three competing retrieval systems, ORBIT, BASIS-70, and DIALOG-RECON. The whole realm of discourse is changing from what would be "nice" (how can we seduce a few users?) to what is efficient (how can we serve all those beating on the door?). We at NLM regret some of the capabilities we have given the user, e.g., the ability to search all terms beginning with the same two or more letters, because they seriously reduce the response time to all users. BASIS-70 does not have this capability and may be more viable for the lack. The systems listed in table 1 now have or will have national audiences. Table 2 lists data bases now available for on-line searching. Thus, al- though I am not sure of the exact numbers, there appear to be nearly 3 million citations to the scientific literature available for on-line searching. Before July of this year, these will all be on the network of TYMSHARE, Inc. By fall 1973 TYMSHARE's network will be connected to the ARPA network to interconnect at least 63 computers, all technically available from any node NETWORK-OR ALL HANG SEPARA TEL Y Organization System Data Bases NLM ELHILL (ORBIT) MEDLINE, COMPFILE, SDILINE, etc. NLM (Informatics) RECON TOXICON, including CBAC, Toxicology Bibliography, Health Effects of Environ- mental Pollutants, etc. System Development Corporation ORBIT ERIC Chem. Abst.-Condensates Science Information Association BASIS-70 NTIS Chem. Abst.-Condensates Lockheed Information Sciences DIALOG ERIC NTIS PANDEX Table 1. Systems with National Audiences Data Base Estimated No. of Citations MEDLINE COMPFILE TOXICON ERIC NTIS Chem Abst.-Condensates PANDEX OCLC TOTAL 425,000 288,000 220,000 115,000 150,000 1,000,000 400,000 590,000 3,188,000 Table 2. Data Bases Available for On-line Searching on either network. Thus, on-line, networked information services have found a new group of user needs and begun a new breed of library services. But "on-line" has not yet reached the user. A recent evaluation of MEDLINE (March 26-30, 1973) shows results similar to those obtained in September 1972 with regard to the involvement 4 1973 CLINIC ON A PPL 1C A TIONS OF DA TA PROCESSING of the requester in his search; in both evaluations it was shown that over 75 percent of searches are run in the absence of the requester. This result is a matter of grave concern, because it means that the on-line system is not being used by those with the information need. The lack of MEDLINE use by or in the presence of requesters may be partly the inevitable result of the training. Training necessarily presents the complexity of the system in an attempt to provide the trainee with a competence to handle exhaustive searches and complex searching activities. The role of the librarian certainly includes assisting users in such activities, but his or her role is not limited to these functions. Librarians should consider their role as primarily educators. In this role, the goal is self-sufficiency in the student. Like a good parent, the ultimate goal of an educator must be to provide the learner with the skills and desire to be responsible for his own actions. The major cause for concern is that this nonuse is contrary to the interests of the user community. The medical literature is a vital part of the communication system in medicine. A recent study has shown that clinicians consult the library in connection with at least half their information needs. 3 This same study, however, shows that when they use the library, 75 percent of the time they do not consult the librarian but prefer to search the Index Medicus or the literature themselves. Contrary to common assertions, clini- cians do not appear to delegate information searching to either their assistants or to librarians. This is even more true in research than in patient care; for research information needs, 85 percent of users did not consult a librarian; for patient care information needs, 68 percent did not. This result of the study has a corollary which also deserves mention: since such a small fraction of users consult or delegate their searching to the librarian, the librarian's intui- tions and experience on the performance of users is an unreliable guide to user needs and competencies. Clinician needs and behavior patterns cannot be judged by experience with the small minority who ask for help. The inescapable conclusion of the evaluation and the above study is that MEDLINE is not reaching a major fraction of its intended audience. This conclusion is disappointing. Another study has shown that MEDLINE can be used effectively by medical professionals with little assistance. 4 This study showed clearly that subject matter expertise was valuable and that biomedical researchers with such expertise could search nearly as well as personnel who had had six months of training on MEDLARS and years of experience-much more training and experience than is provided in NLM's MEDLINE training courses. In part, the problem seems to be one of easy access to MEDLINE. The medical user will usually take the shortest route to the information he needs. NETWORK-OR ALL HANG SEPARA TEL Y 5 As Voigt says: "The user takes the most direct path he knows to the in- formation he wants. He uses the printed sources on the bookshelf in front of him if possible. Thus, his choice is governed to a greater degree by what is easily available than by which source is best." 5 But speed, according to Ames, is also important: Overworked as he tends to be and beset by the unpredictable demands of his patients, his teaching schedule, and the many conferences within the hospital, the physician can seldom plan his schedule in advance, and he must grasp such opportunities as may present themselves for reading literature. If thwarted by an article not being immediately available, the physician may well come to the conference or even to the bedside less than optimally prepared. In inserting the reading of scientific articles into a busy schedule, the time factor may well become critical and even something of a "threshold phenomenon." That is, it may be of relatively little benefit to reduce the time needed to obtain an article from 72 hours to 2 hours but make a great deal of difference if the time can be further reduced from 2 hours to 10 minutes and thus make the article available essentially at the time the physician recognizes his need for it. Most of these same considerations apply also to the investigator, who is also usually a physician. The planning of the next experiment, the pursuit of an exciting new concept, or the write-up of a paper must often be sandwiched into an hour or two of available free time and may be severely handicapped if the reference needed is not immediately available. The need for speed here depends largely on how much it will improve the efficiency of the performance of highly trained and often over-extended men, and this is clearly difficult to evaluate. It is my impression that this is a very important consideration and that having to wait for two days or even six hours for a reference, after its need is recognized, is both wasteful and expensive. 6 Both of these quotations illustrate that since the user often is forced by his own needs to adopt a cost/benefit approach to information, he may need the greatest payoff in the least time. He may be able to stop fighting Indians long enough to pan a few nuggets from a nearby stream, but may not be able to pan all the gold available. There are probably many reasons for this lack of direct or cooperative use of MEDLINE. There has, of course, been some difficulty with the system and some institutions pay line charges (some of these do allow users direct use of the system). However, MEDLINE is now available 13 hours a day during the week and 5 hours on Saturday, a total of 70 hours a week. Only six institutions use even close to 70 hours a month. Thus, most MEDLINE ; 9 73 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING 3oo,C Fig. 1. MEDLINE Growth Rates NETWORK-OR ALL HANG SEPARA TEL Y terminals are in use less than 25 percent of the time. There appears to be ample time available to allow users access to the service. In spite of this restriction, MEDLINE has shown the growth rates depicted in figure 1. One can surmise that the real need for rapid information services must be enormous and that, when on-line systems become as readily available as printed indexes and the card catalog, the demand will be very large indeed. But this information service networking still needs to be combined with the earlier sense of the word, library resource sharing. This combination is under development. OCLC has pioneered in this area. Through that center, cooperating libraries actually share the intellectual effort of cataloging. A book cataloged by any of the Ohio libraries need never be cataloged by another. The Federal Library Committee is trying to organize an effort to put OCLC on the nationwide network used by the other services mentioned above. When this is done, a prototype of a national library network will surely exist. This network will provide rapid information services, but, much more than that, it will begin the process of using telecommunications to permit the intellectual cooperation of libraries. Division of labor may really become possible. A variety of functions could be added to such a network: inter- library loans could be routed through it; a serial check-in service could be initiated to serve many libraries based perhaps on the National Serials Data Program; and if companies handling blanket orders were incorporated, much of the acquisition process could be handled by a resource in such a network. As stated in the beginning, I believe this development path almost inevitable. Caught between rising costs and user expectations and budget pressures, many libraries will adopt these new systems and services in order to remain viable. REFERENCES 1. The Bowker Annual of Library and Book Trade Information. New York, R. R. Bowker Co., 1972. 2. McCarn, Davis B., and Leiter, J. "On-line Information in Medicine and Beyond," 181:318-24, July 27, 1973. 3. Friedlander, Janet. "Clinician Search for Information," Journal of the American Society for Information Science, 24:65-69, Jan.-Feb. 1973. 4. Lancaster, F. Wilfrid. Evaluation of On-Line Searching in MEDLARS (AIM-TWX) by Biomedical Practitioners (University of Illinois Graduate School of Library Science, Occasional Paper, no. 101). Champaign, 111., 1972. 8 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING 5. Voigt, Melvin J. Scientists' Approaches to Information (ACRL Mono- graph, no. 24). Chicago, ALA, 1961, p. 26. 6. Ames, Adalbert, III. Unpublished letter to Ralph Esterquist, October 22, 1962. In Esterquest, Ralph. "Medical Library Service in the Hospital," Bulletin of the Medical Library Association, 52:260-61, Jan. 1964. ESTELLE BRODMAN Librarian and Professor of Medical History Washington University School of Medicine St. Louis, Missouri Backing Into Network Operations When asked to participate in this clinic, I gave the tentative title "Backing Into Network Operations" for my paper because I thought it might be useful to discuss some principles of planning for and consensus making within networks which I had derived the hard way from my experience directing two computer-based networks-networks which were entered into by chance rather than design. (In addition, the Washington University School of Medicine Library is a member of three other networks: the Regional Medical Library, the Regional Medical Program, and the Higher Education Coordi- nating Council, so that we have experience on both sides of networks.) It seemed that it might be helpful to examine the mistakes which we made and the generalized conclusions which we might draw from them as hypotheses to be tested in other networking operations, as well as a comparison with some other, more carefully planned, networks now in existence. I therefore devote the beginning of my paper to describing the two networks we have been involved in, merely as background to understanding, and not as another "How I do it good in my library" paper. PHILSOM The first and oldest network which the Wasington University School of Medicine Library runs is PHILSOM (Periodical Holdings in the Library of the School of Medicine). Started in more primitive form in 1963 from plans of Irwin Pizer, now director of the Health Sciences Library of the University of Illinois in Chicago, and Donald Franz, formerly of the Washington University Computing Facilities, it is now used as a serials control mechanism for seven 10 1973 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING medical libraries throughout the country. These include, besides our own library, the Library of the National Institutes of Health, the medical school libraries of the Universities of Missouri in Columbia, Utah in Salt Lake City, Texas at San Antonio, Illinois at Chicago, and Saint Louis University. In addition, the University of Missouri includes data for its Veterinary Library and its Veterans Administration Hospital Library, so we might claim that we are talking about nine rather than seven libraries. The system now has about 8,000 titles in it, approximately 40 percent of which are "dead" titles, with the rest still being received by someone within the network. As mentioned earlier, the system started in a more primitive form for one library in 1963; we thus have a decade of experience with it. Since its inception it has gone through two complete reworkings, each to give more and more sophisticated results, and innumerable small additions and changes. We are now planning a third basic revision to allow more libraries to enter the system, to provide certain things not in the present system, and to investigate the possibilities of going on-line. Obviously, ten years ago neither the available hardware nor the costs would have allowed us the option of an on-line system, so we use a batch system. Finally, the PHILSOM system has now been adopted by the Union Catalog of Medical Periodicals at the Medical Library Center of New York (the largest national data base for medical serials) to start a PHILSOM network in the eastern U.S. The Cornell University Medical Library and the Medical Library Center of New York itself are the first libraries to be in that network. The following describes the present system in brief terms: PHILSOM provides the libraries in the system with certain records on a monthly basis. (Anyone who wishes more information on the details of the system can write to Millard Johnson of the WUSM staff or buy the documen- tation, which we sell for $7.50.) The first of these records is the monthly list of holdings of the serials, meant for the reader and the reference staff. We provide exact holdings, one of the decisions we assumed a priori, rather than making it rigorously. Also noted is bibliographical history, shelving location, multiple sets, presence or absence in Index Medicus, and any informational notes the librarian wishes to include. For serials librarians we provide a much more elaborate and less easily read "work copy" which gives by codes for each library such items as the vendor of the serial, the price paid for it, when the subscription must be renewed, the number of bibliographical volumes per physical volume, whether the title page and index is bound in, whether the title is still being received, the subjects of the journal, and whether it is indexed in Index Medicus, among other things. BA CKING INTO NETWORK OPERA TIONS 1 1 Each month a library receives binding slips for journals reported in complete form, with the slips formated in the way the binder at our library wishes it. This is another problem I will discuss later when I discuss "backing into the system." The libraries also receive a list of journals which need to be renewed within the fiscal year, and a list of journal issues which were expected to be received but were not reported-a kind of "claims list" or "gaps list," where a later issue was checked in, but not an earlier one. Finally, the library receives a box of IBM cards, one card for each issue of each journal which is expected to be published the next month. The bill comes separately for our services. We do cost accounting and change our prices in relation to costs, of course. We also do "on demand" work with the data we have; a library can specify what it wishes, we cost it, make an estimate, then do the work for that price, if the requesting library agrees to it. So far, such special work has included: 1 . binding slip changes for the National Institutes of Health so that the set number of the title appears (i.e., Set 1, 2,...) on the binding slip, and an internal code for NIH use; and 2. a list of current titles only without holdings or other information, for the University of Illinois. In order to come into the PHILSOM network a library must send us the information about the journals it has. This information comes in two parts: bibliographic information and housekeeping information. The bibliographic information (title, bibliographic history, cross references, starting dates, and the like) is the same for each copy of the journal, of course, so that if the PHILSOM network already contains that title, the new library need not repeat the information. (Most medical libraries now find the large majority of their titles already in the system.) But each library has its own, unique, housekeep- ing information, and there is no way to know where the new library shelves its sets of the title, how it binds it, from whom it receives the serial, or how much it pays for it. These facts must be coded for us by the entering library, and any changes must be reported to us by them. For titles new to the PHILSOM network, both bibliographic data and housekeeping information must be supplied; I will discuss this problem later. Once the library has entered our network, it updates its records by sending us punched cards: for issues received in the ordinary course of time, the library merely returns to us the punched receipt card which the computer has produced and we have sent to them in anticipation of the receipt of that issue. For journals which are so irregular that there can be no anticipation 12 1973 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING cards, or for back issues suddenly received by the library, a punched card must be made and submitted. Since most medical libraries have machines to punch their cards, they usually do so for us, but we can handle coded sheets instead, if that is more convenient for the outlying library. We check all basic data input to make sure that it is consistent with what has been placed in the computer previously -we do not want to try to update a holding, for example, with a card for a twenty-fifth issue when the journal is listed as having only twenty-four issues per volume. There are probably many other details of the PHILSOM system which might be described, but I believe that these facts are sufficient to provide the background for the discussion later of generalized conclusions on networks. Problems With PHILSOM I will discuss some of the difficulties which we have had with the PHILSOM network. Later, I will do the same with our cataloging system; finally, I will try to derive some principles of networking operations which seem applicable in other similar situations. The most important difficulty encountered concerns the fact that the system was originally designed for one library and then adapted to many. This implies that we believed that the situations in medical libraries were sufficiently similar so that the same program could do for all. In general, this has turned out to be correct-most medical libraries stock the same journal titles, most place serials in their periodical records; most bind them in the same bibliographic way; most have the same problems of claiming missing issues, renewing subscriptions, paying their bills, and determining their budg- etary situation each month. On the other hand, there are many individual needs for which the PHILSOM system was not designed. For example, because we developed PHILSOM for ourselves only, we worked with our binder to produce binding slips formated in the way in which he wished them; but other binders use different record-keeping methods, and we have just begun to work on individ- ualizing this portion of our program. Secondly, we have come across the same problem of title entry which everyone else finds. When the Medical Library Center of N.Y. began a union list of serials of some eighty-eight medical libraries in the New York area fifteen years ago, Jacqueline Felter, then director of the Union Catalog of Medical Periodicals, said she thought there must be eighty-eight different ways to enter any medical serial in existence. Any cooperative scheme requires standardization, of course, and we have found some of the serials librarians in BA CKING INTO NETWORK OPERA TIONS 13 our network adamant about the correctness of their form of entry. In desperation we decided to adopt the entries right or wrong of an external group, and chose the same Union Catalog of Medical Periodicals at the Medical Library Center of N.Y. which, as noted before, is the largest reposi- tory of medical serials in the country. This has reduced the complaints to grumblings, but it has by no means solved the problem. Every time a new title is published which is likely to be subscribed to by more than one library, it is probable that we will get more than one entry for it. We ask the New York group to act as final arbiter, and we refuse to accept any variation in the main entry, though we add as many cross-references as are requested. The reason that serials librarians react in this way is one of the generalized conclusions I propose to draw later. A third difficulty we have found and one we were not prepared for is the need for continuing education of the people handling the serials in our member libraries. Serials librarians come and go, and serials clerks seem to have a half-life of about the same magnitude as deans of medical schools. As a result, to the difficulty of first instructing the staffs of new libraries on how to use PHILSOM is added the problem caused by the fact that outgoing staffs rarely transmit much information to new staff members. We have tried to handle this by a detailed manual of 130 pages, by sending a member of our staff to the new library when it enters the system and at any time thereafter when it requests aid, and by publishing a weekly news bulletin, PHILSOM Newsletter, which attempts to give instructions, bring member libraries up-to- date on new facets of the network, explain changes, etc. If only the person receiving this Newsletter would pass it on to all who handle PHILSOM, our problems would be substantially lessened! An increasingly large problem, now that more libraries with more titles are in the PHILSOM network, is the size of the printouts which we require our computing facilities to produce. It exceeds all the time and line jobs done for any other use of the facilities, printing over one-half million lines monthly; and even though it is run in the middle of the night, it causes scheduling difficulties. A time is reserved for us, and if we do not get onto the computer at that time, it may be a week later before there is an equivalent amount of time available. Explaining this to the other libraries, where serials librarians have traditionally not been bound by such tight time schedules, causes one difficulty. The vagaries of the postal service and the air freight which we use to send materials back and forth to the network also cause problems with the scheduling of our printouts. We have taken the firm stand that we will not wait beyond the deadline for input each month without very good reasons and a decision by me to change this rule. Generally, one such experience suffices until the library changes its staff again. 14 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING The size of the printouts are partially due, of course, to the fact that the system was "backed into"-developed for one library and then multiplied for the other libraries. Although no library receives the entire output of the computer, the total is too large for reason, and is one of the grounds for our planning to redo the entire system and (hopefully) go on-line. Cataloging In this section I will discuss our computer-based cataloging system and our attempt to work this into a network operation. Cataloging is one of the most difficult library procedures to automate because the quality of cataloging and bibliographic description is based on variables that by their nature do not lend themselves to programmed manipu- lation, but require unique judgments to determine. As a result, most systems of computer-based cataloging are more or less printing mechanisms, with the pure aspects of bibliographic description supplied by the human cataloging staff. This is in strong contrast to the various attempts to do computer indexing of natural text; that is, to use the computer to derive the very elements of the indexing. Even with this, however, computer cataloging is a real challenge because of the multiplicity of elements which must be handled and the difficulties of writing an algorithm which can describe the multiple variations within any element. About all such systems as the MARC cataloging can and have-done is to standardize the order in which the elements are stored and retrieved; but the elements themselves are still provided manually by a human cataloger. This is, of course, the purpose of MARC: to be a format for the transfer of bibliographic data. I do not say this to denigrate computer cataloging. I am not a believer in the all-or-nothing principle in intellectual work, no matter how efficient it is in allowing our bodies to perform physical movements. A requirement that the computer do everything, even those things better and more efficiently done manually, is as self-defeating as a requirement that everything be done manually. I wish merely to describe the situation, so that we can see how it fits into the question of networking operations. In spite of the difficulties with cataloging in libraries both manual and computer-based it is the sine qua non of librarianship. Without it the acquisi- tion of material is like the stacking up of volumes in huge bins; there is no access to their contents except through the tedious job of turning over each volume in turn. When the WUSM Library was able to persuade its dean to give an annual sum for experimentation in librarianship, without regard to the actual running of the Library, beginning in 1962, one of the decisions taken was that it would try to manipulate the most basic records of medical libraries BA CKING INTO NETWORK OPERA TIONS 15 (serials and the catalog) by computer means. I have already described here our experiences with serials: and Doris Bolef and I have also published explana- tions of some of our experiences with computer-based cataloging. 1 (In- cidentally, when Bolef and I described some of our failures, we were deluged with scurrilous correspondence and letters to the editors, as if we had broken faith with a religious concept! No wonder few librarians report such failures.) We decided to design a completely new cataloging system using MARC format because it looked, at the time, as if the three major libraries-NLM, LC and NAL-were going to resolve their differences and put all their biblio- graphic data on MARC tapes. Using the computer programming already in use and the experience we would have gained, we thought we could serve as a center for the libraries university, public and special-in the St. Louis area. We could pick off data from the MARC tapes for acquisition and cataloging purposes and then print lists or cards in accordance with individual library requirements. We thought we could serve as a pilot network for other communities in the nation. For reasons to which we are not privy, efforts to include NLM and NAL bibliographic data on the MARC tapes seem to have been abandoned, which took us out of the running to go on with this networking experiment. The death knell was sounded, however, when one of the largest libraries in the St. Louis area that was to serve as a keystone, that had initially announced its support and cooperation, began to drag its heels for financial and political reasons. We began, then, in 1963/64 to work on producing book catalogs from computer-based information, and we have continued to do so ever since. We thus have nine years of machine-readable cataloging in our files, representing both pre- and post-MARC format rules, and in 1968/69 we transformed our pre-MARC catalog into MARC format automatically by an excellent program worked out by Glyn Evans and the programmers at our computing facilities. We published annual catalogs from 1964 to 1968; these (like the early automobile which looked like a buggy) very much resembled older printed catalogs. When 1968/69 came, however, we wished to produce a five-year cumu- lated catalog, but found to our horror that the cost had gone up so greatly that our budget could not afford this traditional form of catalog. At this point we decided to experiment with Computer Output Microfilm. Feeling, however, that few of our readers would use microfilm directly, we decided to try what the salesman said was perfectly good technology: namely, to blow up the microfilm to readable size and print that. The salesman's story was far from reality. 16 1973 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING In addition to printing our cumulated catalog from COM, we tackled the problem of the cost of annual reprintings of the same data. In order to make it possible for a reader to find everything the library has in one place, the catalogs of libraries since the end of the nineteenth century have been infinitely accumulating files. New items are interspersed among old ones, which are never purged unless the book is lost, withdrawn, stolen, mutilated, or otherwise removed physically from the collection. Some questions have been raised recently about the value of this system in the sciences, where new material supersedes older material, but no library has had enough courage to try this on its catalog of holdings. All of this means that in a printed catalog and I use the word printed in a wide sense of "reproduced" one must interfile new material into the old and then reprint the entire list; otherwise, as our previous experiment when we hid the card catalog has shown, a reader may be required to search as many as five alphabets at one point in the cycle before he can determine conclusively that the library does not contain a particular work. What we did, therefore, was to divide the catalog into the bibliographic record and the indexes to that record. The full bibliographic record (the "register") was printed up in accession number order. Thus, to add to that list, one need only print accessions received since the last printing. The indexes, by name, by title, by subjects, by added entries, and the like, were shortened to act merely as pointers to the full bibliographic information in the register for the use of the person who wanted "corroborative detail for an otherwise bald and unconvincing narrative." For the person who merely wished to get the book in question, we supplied the call number in the index. It was this index which had to be added to each year by interfiling the new with the old, but since the index entries were generally only one line long, the cost was not as great as printing the whole record would have been. We have continued this scheme ever since 1968/69. We take pride in this work, even though it has not been an unqualified success in the only way in which I believe librarians ought to consider something successful namely, the usefulness to the library user of the products of his imagination and endeavors. What makes us proud of this development is that we tried to use the computer as a new technological methodology, not as a simulator of what had been done previously manually. I have said in another context that the unit record catalog card which we have inherited from the time the Library of Congress began to sell catalog cards at the turn of the century was based on the requirements of the printing press and the card catalog. In designing a system for a new methodology, I fear we have tended to fall into the trap of using the new system to produce the old results. We like to think that our catalog is a feeble effort to take a new look at things. BA CKING INTO NETWORK OPERA TIONS 1 7 (This has not kept us from producing traditional catalog cards from our computer record also; I will discuss this network below.) Under an agreement with the local Regional Medical Program's Library Project, we had agreed to sell them copies of our cumulated catalog, to be distributed to the 100 small hospitals in southern Illinois and eastern Missouri under their umbrella of services, so that interlibrary loans might be facilitated. We did so, but the poor legibility and lack of education of the librarians thoughout the Regional Medical Program's Library Project on how to use the catalog made it very rarely used. I cannot blame them, as legibility was not good. It certainly was simpler to call or write us about a book a hospital wanted than to try to find out from our catalog whether we had it. We were all ashamed of the physical appearance of the printed catalog, but proud of its contents and of our ability to use the computer for this purpose. In the next year, therefore, we decided not to blow up the microfilm in the COM, but to make microfiche for the libraries to use. The snag here, of course, was that a reading machine was required to use the fiche, and most of these hospitals were too poor to afford such a luxury, while the departments in our own medical center felt no need for one when they could lift up the receiver and call our reference staff. Consequently, we wrote a small grant request to the National Library of Medicine to allow us to purchase 125 DASA reading machines and distribute them free to the 100 hospitals in the Regional Medical Program and the 25 departments of our medical center. We also specified that a librarian must deliver the machines in person and explain their use and the use of the microfiche catalog, so that the educational message could get across. I wish I could say that these things resulted in a large-scale use of our microfiche catalog. The great American psychiatrist Adolf Meyer once said, "The trouble is not that people don't know things, but that they know things that aren't so." We knew that once the librarians had microfiched catalogs and a free microfiche reading machine, they would take them to their bosom. Only it is not so, no matter what the salesmen for microforms say! I have mentioned our experiment of hiding the card catalog and forcing readers to use the computer printouts of our cataloging and ordering proce- dures. Unless we bankrupted ourselves by continually updating and reprinting items, we found that we were making it necessary for our long-suffering readers to search many alphabets for their information. At that point we agreed that the printed catalog of books was not a satisfactory substitute for a card catalog, whether the printing was done by hot type, as in the 1890s, cold type as in the 1930s, photo-offset as in the NLM's catalogs for the 1950s or our own computer-produced catalog in the 1960s. We therefore returned to 18 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING our cataloging system with a new requirement: to produce catalog cards which we could place in our old-fashioned card catalog. When we were able to produce these cards, it seemed to us and to a few of our friends that the cards we were producing might very well be used by other medical school libraries, most of whom add the same titles to their collections. Admittedly, our selection was somewhat smaller than what some medical libraries encompassed, since we did not collect in dentistry, nursing, or pharmacy; but we felt if we were able to supply the common titles to libraries similar to ours, it would give those libraries more time to do the cataloging for the works we did not process. At that time, however, the NLM announced plans to sell its catalog cards through Bro-Dart, and again we waited before offering our cards for sale. Any medical library which can obtain NLM cards would be wiser to use theirs than ours, both for the subject and language coverage which we do not have, and for aesthetics, comparing a printed card with a computer-typed card. Unfortunately, the Bro-Dart attempt to sell the cards ran into diffi- culties. The firm received too few orders to make the scheme viable, and some of the technical problems in reproducing cards clearly were not overcome. After waiting six months beyond the due date for NLM cards, we offered ours. We now come to the reason that I have given this large spate of history so I can describe what happened and draw some conclusions from it. One of the ways we publicized our cards was to send samples to many medical school libraries who, we thought, might be interested in our scheme or had already shown interest in it. We modified our catalog program to omit call numbers, for those libraries which did not use the NLM classification scheme. We arranged for libraries to order cards without the tracings on them, for those who did not use MeSH subject headings. We offered the cards as unit cards or in sets, and we offered them arranged by the computer in one alphabetical array by main entry, or in the arrays we use: name, subject, and shelflist. As a bonus, we offered Se-lin tapes of the call numbers to affix to the spines of the books. To give potential purchasers an idea of what we had, we made available our cumulated catalog and our microfiche supplements to it, and we started a new free listing of recently acquired books RECAP. This gave the record number of the book, and allowed us to reduce the price to those libraries which gave us this number, thus removing from our staff the time and labor of searching for the record number to print the cards desired. For a long time we received almost no orders. Catalogers and our other friends in the medical library field often wrote us, congratulating us on our work, but then adding why they were not going to buy our cards. These BACKING INTO NETWORK OPERA TIONS 19 reasons are interesting, and the basis of my final conclusions. Some were: the call numbers were placed at the left top of the card and their catalog placed them at the right top or bottom or left bottom. The call numbers were too far to the left; they wished at least 3mm space from the margin of the cards. In their library tracings were on the back of the card and we printed them on the bottom front. They did not like the type font we used: it was too big or unaesthetic, or just not what their catalog had. Some did not like the numbering of our cards (e.g., "card 2 of 2"), and the slight ripple of the top perforations bothered others. Nobody, I am happy to say, complained about the level of cataloging, or said that it was not of the highest quality. I have mentioned already our work with the Regional Medical Program's Library Project of 100 small hospital libraries. One of the things which this project did was to offer workshops to the often untrained librarians in these hospitals, and one of the items which RMP wished to show them was the "Stearns List." This is a group of books selected by many physicians through a questionnaire as a minimum hospital library collection for the biomedical practitioners of a hospital. It seemed useful to us that a package deal of purchasing the books and getting an already-made catalog would make the sad little libraries in those hospitals 1000 percent ahead of where they had been. We therefore wrote a small program which allowed an order for the Stearns List and no further rules to produce a catalog of the approximately 250 volumes, already alphabetized and ready to go into the catalog trays, and we were able to offer this to the libraries for a small sum about $75. These instant catalogs have been a small seller. The recent floods in Pennsylvania and elsewhere, which wiped out hospitals and their libraries completely, have made it necessary for many libraries to start afresh. To do so, they have asked the NLM for the so-called small research grants of $3,000 each, and NLM in turn has allowed them to buy the Stearns List and our set of cards. When the second edition of the Stearns List came out and we received no requests for cards for the new package, we gave up our offer. We realized that hospital libraries preferred to choose their own books, using the Stearns List for reference. This alerted us to the needs of libraries just starting. We have now been in touch with all the developing medical schools to offer them catalog cards for the standard works (and anything else we have) which all new medical school libraries must have. This has been very satisfactory; the medical school at Duluth, Minnesota, for example, has come in with large orders several times, and must now have about 2,000 sets of our cards. The new medical school at the University of South Carolina has also been a good customer, so 20 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING that all in all in the last six months our sale of catalog cards has taken a slow but definite turn for the better. What pleases us is that most orders now are repeat ones. Apparently our customers are satisfied with our service. The problem is to add more customers. We are again in a waiting period, however. The NLM has stated that it would put its cataloging on the TYMSHARE computer lines from which all medical schools access MEDLINE and other data bases at a small connect charge and telephone hookup cost after July 1, 1973. If this is successful and cheap, the same reasons why medical libraries should use that cataloging rather than ours will hold here as I have previously enunciated when dis- cussing the Bro-Dart cards. We are a marginally successful venture economical- ly, and if an alternative manner of obtaining medical cataloging comes into being we should remove ourselves from the sales arena as gracefully as possible. Discussion I will attempt to draw some generalized observations from the descrip- tion of the two networks we run, and then set forth a series of hypotheses about networking in general. It is a cliche that successful networks depend upon the cooperation of all those involved in them. Yet, as Rose Vainstein pointed out at the ASIS meeting in Washington in October 1972, there are at least three different kinds of cooperation, based on three different kinds of standards of perfect- ability. First, there is the standard of the ideal: we will do only what is best. Second, there is the engineering standard we will try to reach the best commercially viable state. And, finally, there is the working standard, which is based on the concept of a tolerance level, the permissible variance from the best possible. I believe that only if the members of the network are willing to accept a working standard and strive for an engineering standard will a network really become viable. The question, then, is whether librarians generally are so constituted that they will accept someone else's working standard. A look at any serials record or large card catalog will show many variations from the ideal, but the errors are "our" errors, while the errors of a network are "their" errors. Even worse, when there is no standard, the concepts of "our way of doing things" must always prevail over "their way of doing things." I do not say this to throw brickbats at librarians, but to try to explain some of the effects which we see frequently throughout cooperative ventures in librarianship. Most librarians are people who are asserting their rights as professionals, without being backed up by the mandate of society. Now the BA CKING INTO NETWORK OPERA TIONS 21 sine qua non of professionalism, as any text on sociology will declare, is that the professional makes his own decisions on the basis of his expert knowledge, for the good of his client, without asking either the client or other profession- als about the wisdom of his course. To ask a professional to follow rigidly laid down institutional rules is to ask him to act not as a professional but as a bureaucrat. Few librarians wish to give up what freedom of action they have. Because so many of their actions are bureaucratic in nature, they must insist on their professional rights where they have them. This has led me, in an attempt to find examples which proved or disproved my point, to ask which had been the most successful networks in librarianship up to now? I think I would point to the sale of catalog cards by LC, the OCLC, and the NLM's MEDLINE network as being extremely success- ful networking operations. Next, I asked: What characteristics do these things have in common? It seemed that the one characteristic these three networks had in common was that they allowed the librarian to manipulate the store of knowledge out of the standardized bureaucratic mode into the personal-profes- sional mode. By this I mean that in each case the librarian could take the standardized material offered to him and modify it in any way he wished. I believe it was Kilgour who, when asked if his OCLC would allow for variation, replied that he already had as many variations in the systems as the libraries using it. The cataloger can call forth from the OCLC store the catalog card in it representing the book in hand, and then so modify what he receives that it then conforms to his style and contains his mark of identity. LC printed cards have been modified by so many catalogers for so many years that it has become a standing joke among librarians. The MEDLINE system requires not the restrained, elegant, lean Boolean searching required in the older MEDLARS searches, but the bumbling, hit-or-miss attack on a problem which is characteristic of most of us in looking for information; moreover, it allows for modification of the search strategy at any point in time. Uri Bloch of the Israeli Armament Development Authority is reported to have said that standardization is the universal problem and major stumbling block of all networks. "To expect continuous adaptations to a system ... is to invite certain disaster," 2 the Library Journal states he said. Once a standard has been accepted, he believes, it should be changed only for the most important reasons. I hold, on the other hand, that that way brings almost certain failure. Unless one were designing a network to do something that had never been done before, and for which, therefore, no one had made professional decisions about how and how much to do, it seems to me that insisting upon exact 1 9 73 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING obedience to inflexible rules will always result in dissatisfaction and eventual breakup of a network. It is necessary to take into account in the design of a network the facts of human action and reaction, and instead of trying to make people over, to make one's network responsive to deep-seated emotional needs. I admit that most people disagree with me. The present Constitution of the United States was adopted after the Articles of Confederation, with its decentralized power structure, proved a failure in the government of a vast country. Carolyn Landis in a recent article in the EDUCOM Bulletin on networks and disciplines, says that it is clear that computer center directors participating in networks are caught between conflicting and often irrecon- cilable demands for service from local users and from network users. 3 Yet it seems that from now on library networks must provide a way to meet these conflicting and often irreconcilable demands which are made on computer center operators and those who administer networks alike. The question we must ask, I believe, is: Can an inflexible computer allow us to have such a flexible network, or do we need to think of library cooperation in terms less like the Articles of Confederation and more like the Constitution of the United States, with vital "network" operations given to the central govern- ment, and all other powers reserved to the individual states? This is my main conclusion. I have a few less important, and perhaps less startling conclusions, which I will mention here briefly. The most obvious conclusion from our networks is that you cannot make everyone happy, so you have to make the results obtained through the network worth the loss of autonomy, standardization, and personal style which comes about. This is the cost/benefit ratio expressed in human terms, of course, and I will not belabor the point. Secondly, just as the administrative set-up for a small, closely knit Germanic tribe in the sixth century could not be extrapolated to rule the entire Roman Empire, so the methods used for a single library cannot be extrapolated into a network for a whole group of libraries without serious problems. When a problem changes in size it often changes in quality too; and when solutions worked out for small groups are applied to larger groups, stresses and strains are sure to result. A single library's system can be used as a working basis for discussion when a network is being contemplated, but it cannot be the only item considered. Thirdly, when methods are irreconcilable such as the true title for a journal it is wise not to use the method of any of the libraries in the network, but to go to an outside source for authority. In this way, all are equal, and no one stands out as the leader of the others. Everybody can join BA CKING INTO NETWORK OPERA TIONS 23 in complaining about how "they" catalog their titles, and a sense of camer- aderie may result. Finally, there is no more important place for continuing education than in the working of a library network. It cannot be assumed that everything is heard, read, or understood, even by the most intelligent with the best emotional stance to the network. Patience, continual work, and strong mar- tinis will always be needed by those running library cooperative networks. REFERENCES 1. Brodman, Estelle, and Bolef, Doris. "Printed Catalogs: Retrospect and Prospect," Special Libraries, 59:783-88, Dec. 1968; and Bolef, Doris, et al. "Mechanization of Library Procedures in the Medium-sized Medical Library: VIII. Suspension of Computer Catalog," Bulletin of the Medical Library Association, 57:264-66, July 1969. 2. Quoted in: Savage, Noel. "ASIS in Washington," Library Journal, 98:136, Jan. 15, 1973. 3. Landis, Carolyn. "Planning for National Networking: A Report on the Spring 1973 Conference," EDUCOM Bulletin, 8:4, Summer 1973. CHARLES R. SAGE Coordinator of Automated Library Services Libraries of the State Universities of Iowa Iowa State University Ames, Iowa Utilization of the MARC II Format for Serials in an Inter- University Environment This paper deals with a relatively small project undertaken by the libraries of the three state universities of Iowa that standardized their handling of machine-readable serial records using the MARC II format. The first section deals exclusively with programming techniques and conventions employed in the MARC II format. The second section describes procedures, generalization, and compromises that permitted the development of a generalized packaged program to serve three academic libraries. The third section outlines the early work environment with special emphasis placed on the library/data processing center relationship. Current environment is also described and projections are made about the next phase this project will enter. Conclusions drawn from this project bear on future handling of computer applications in libraries. Programmed System The initial impetus for developing a machine-readable central serials file incorporating the MARC II format came from a consideration of Serials: A MARC Format, 1 a working document by the Information Systems Office of the Library of Congress. At that time, August 1969, the three state university libraries of Iowa were exploring the feasibility of compiling and producing a union list of serials. Each university library was operating independently in recording active titles and corresponding holdings. The University of Iowa and Iowa State University maintained their files on magnetic tape. The University of Northern Iowa maintained a 3 by 5 inch card file. The three systems were 24 UTILIZA TION OF THE MARC II FORMA T FOR SERIALS 25 dissimilar except that the two computerized systems were written in 1401 Autocoder and operated on IBM 360s. We felt adoption of the MARC format would accomplish two objectives: (1) it would serve as an experiment and exercise to expose and familiarize our personnel with the MARC II communications format; and (2) it would provide adoption of a standard format that will conform to national standards and fit into a library computerized technical processing system of the future. Input to the current system is in two forms. Original data are punched into 80 column IBM cards and input in conventional batch mode. An input card format was designed specifically for this function. (See figure Ib.) The bulk of serials information at the University of Iowa and Iowa State Univer- sity was information already contained on magnetic tape. Special conversion programs were written, and tape-to-tape conversions were processed. The master files are maintained in the MARC II communications format. Minor modifications were made relative to variable and fixed length records; however, the working format is the communications format rather than an abbreviation or subset 2 of the communications format. At the present level of development three output formats are produced. They are: (1) card image of the input alphabetized by title and identified by a local system number (figure Ic); (2) an image of the masterfile in the MARC II format which does not include the "Record Directory" and the "Leader" (figure 2a); (3) selected variable fields in a formal printout for reduction and reproduction (figure 2b). To adopt the MARC II format in its entirety as an internal systems format did present programming difficulties. In spite of these minor hardships, we find the description adequate for our current needs, and anticipate very few problems when we further develop the capabilities of this system. Because of the nature and characteristics of our hardware and software, we actually use a fixed length record rather than a variable length record. The IBM 360 and 370 series machines and COBOL-F or USASI COBOL do not lend themselves readily to variable length records. We modified all records to 2048 characters with the variable differential padded with blanks. The variable MARC II record within the 2048-character record is terminated with an ASC II extended 8-bit "End of Record" (1D16) character for recognition purposes. It should also be pointed out that having programmed in USASI COBOL very little use is made of the "Record Directory." Manipulation of data is exclu- sively controlled by "Field Terminators" (8-bit, 1 El 6); however, the "Record Directory" is actively maintained and updated. The use of special characters unique to the MARC II format compounds the difficulty of maintaining data in the EBCDIC form. We allowed three 26 1 9 73 CLINIC ON APPLICA TIONS OF DA TA PROCESSING "1C SWIAIS WOK I0f AC Mil. IS W0 FOtM AtC SHIA1S WOK (OtM : 1 1 1 1 1)1 ii:mi))IIIII!l!l> """" ' ' 1J1IIJ. II It li """^'"^ III ^_' ut '- b. Hg. 1 . MARC II Serials Input UTILIZA TION OF THE MARC II FORMA T FOR SERIALS 27 uJl!IVE IHIOMl b. Fig. 2. MARC II Serials Output 28 1 9 73 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING exceptions to be maintained in 8-bit ASC II form. Those exceptions are: "End of Record," "Field Terminator," and "Delimeter." These characters are exempted from conversions either from EBCDIC to ASC II or conversely. All input data enter the system in card form. This form of input does limit the effective use of lower case characters to conform to MARC require- ments. To compensate, automatic conversion is made internally of upper case to lower case for "fixed length data elements" character groups. The lower case character, comprising half of the "Subfield Codes" of variable fields, are assigned internally when a variable field is being added or updated. Tags currently processed are given in table 1 . Tag Description Tag Description 001 Control number 269 Vendor name and address 008 022 Fixed length data Standard serial code 350 362 Subscription price Dates and volumes 035 050 Local system number LC call number 500 description General note 082 090 200 Dewey decimal classification Local call number Title 525 850 857-9 Supplement note Holdings Local holdings Table 1. Tags Currently Processed An exception was made to MARC II format with regard to local holdings. We added this tag so that we could report the holdings of peripheral libraries and reading rooms at each campus. The format is identical to the 850 entry except a local mnemonic is substituted for the NUC location symbol. Multiple peripheral libraries and reading rooms for the same titles are recorded in a linear string of characters separated by "Subfield Codes." Each subse- quent location is a prefixed with a "a" "Subfield Code." Any "Other" Tag entry can be input as long as it does not require more than one subfield code. Packaged Program Concept for Libraries The most difficult task in designing and implementing a general purpose programmed system is definition of data. Fortunately this task was provided in more detail than we required with the MARC II record format for serials. Adjustments and minor revisions were made to that format as described in the previous section. However, the ease and convenience of implementing a system cannot be overemphasized when this work is already available. The dialog between the systems analyst and serials librarians centered on the librarians interpreting the MARC II definitions. This dialog was extremely productive UTILIZA TION OF THE MARC II FORMA T FOR SERIALS 29 when the conversion programs (local format converted to MARC II format) were designed and written at the University of Iowa and Iowa State Univer- sity. Identifying and tagging data elements, particularly dates and volumes designations, general notes and various configurations of holdings statements, is a difficult task and would have been next to impossible to delineate without the aforementioned dialog. The MARC II format was also beneficial in attempting to standardize definitions and notations between our three institutions. Details such as incomplete holdings, ceased publication, "ordered on a selected basis," etc., are handled and noted in a uniform manner. Most conflicts or differences of interpretation of MARC II were handled by telephone conferences. In the period of development only two interinstitutional meetings were necessary to clarify differences. As with any cooperative programmed or manual system, we had our problems. Two major differences were not solved. The first revolved around the procedure for cataloging monographic series. One institution catalogs each entry separately, the other two do not. Individually this problem is not significant; however, in compiling a union list of serials this quantitative imbalance does distort the relative number or holding of the three universities. A second problem was the use of cross references Iowa State University and the University of Iowa use cross references quite selectively. The University of Northern Iowa, on the other hand, prefers extensive use of cross references. To compound the differences, the MARC II policies for handling cross references is so complex and difficult to maintain that relative benefits ob- tained are not warranted. A shortcut method was devised for handling cross references different from the MARC II method, but they are not to be included in union lists. We have been able to exchange bibliographic data between institutions with relative ease. Even though each institution utilizes an autonomous local systems number scheme, exchange, extraction, and reassignment of local systems numbers is easier and more economical than rekeying the desired bibliographic data. Collation and compilation for a union list of serials is available on request. A manual collation and matching is required to generate a union list. However, this is not a large effort (a full-time person for one-half month for a total serials file of 60,000) compared to manually compiling the same file or maintaining a centralized union file of serials. Program maintenance is handled centrally and does not impose any problems. All program changes are made to all programs and specialized changes for any institution are not done except in special cases such as systems timers, accounting routines, file identification and names, etc. Tables 30 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING (locations, symbols, etc.) are a composite of all three institutions and main- tained internally in the programs. Job control, tape labels, etc., are handled locally according to the prescribed procedures of the particular data processing center. Hardware, Software, and Administrative Critique of Project From the onset of this project, communication with the data processing centers was difficult and we were not really effective in convincing them that we wanted to shed the "Invented Here" syndrome. Our local data processing managers tend to disregard any programming effort that is not created within or supplied by the mainframe manufacturer. However, we mutually agreed this was an experiment and the implication was that there would be no repeat performances. When the project started we had IBM 360s at all three data processing centers. Two centers programmed exclusively in COBOL and the third pro- grammed in PL/1. Within the IBM repertoire of software, two centers oper- ated under OS (Operating System), one center operated under DOS (Disk Operating System). With those types of hardware and software configurations, fitting our system to each installation was quite easy. Basically the libraries were responsible for programs and the data processing centers were responsible for operations. We jointly documented the system at each data processing center to the local requirements of each center. Since we started operation of our systems, two data processing centers have replaced their IBM 360s with IBM 370s. This change has had no effect on ongoing operations. The development of the system to its present point has been technically quite easy. Originally our concept was to develop a simple but identical operational system at each institution and then expand applications by delega- ting modules of expansion to each data processing center. Upon completion and testing of each module, they were to be added to the system as if it were a single system rather than three independent systems. For a number of reasons which will not be detailed here this is neither logistically or administratively workable. The obvious alternate solution is to continue development at a central source. However, because of budget restraints, this programmed system is in limbo. Originally plans were to continue development of the serial system to the point that a central serials catalog at each institution would be replaced by a machine-readable file. The basic information to be contained in this file would include full bibliographic data, holdings, minimum of cross references, and added entries of previous changed titles. The system would be interactive UTILIZA TION OF THE MARC II FORMA T FOR SERIALS 31 for instantaneous updating and have full use of an expanded upper and lower case character set. With this systems enlargement no longer a reality, scrutiny of the current serials machine-readable file is now being made. The only purpose of this file now is to provide patrons at infrequent intervals (semi- annual and/or annual) with a consolidated book catalog of serials held by each library. The relative benefits of providing such a serials list for patrons versus cost is seriously under question. The identification and adoption of common standards was made ex- tremely easy through the use of MARC II. Well-selected standards impose a minimum of control and maximum of common advantage. There are even greater benefits to be obtained if we can use generalized packaged programs. Unfortunately our data processing centers have not reached a degree of maturity where developmental schedules and accurate budget projections are demanded or deemed necessary. Also, the application of computers is not widespread enough in libraries that libraries and data processing centers can compare notes with others in similar circumstances and benefit materially from these exchanges. Penetration of this market will come from two directions once a viable market is realized. Consortia, cooperatives, etc., will offer technical processing services utilizing high speed transmission lines forming regional and national networks. The software industry will offer for purchase or lease standardized packaged programs similar to the hundreds of programs now offered in other fields of endeavor such as Informatics's MARK IV, Applied Data Research's Autoflow, etc. 3 Viability of this market will come to pass only when it will simply be cheaper for libraries to do it this way than through their local data processing centers or by themselves. REFERENCES 1. U.S. Library of Congress. Information Systems Office. Serials: A MARC Format. Washington, D.C., Library of Congress, August 1969- 2. Avram Henriette D. "The MARC Project of the Library of Congress," Drexel Library Quarterly, 4:308-09, Oct. 1968. 3. International Computer Programs. ICP Quarterly. Indianapolis, 1973. FRANK KURT CYLKE Chairman, U.S. National Libraries Task Force on Cooperative Activities Library of Congress Washington, D.C. U.S. National Libraries Task Force: A Review of Data Processing Interests This paper has two specific purposes: (1) to relate what has been accomplished by the U.S. National Libraries Task Force on Cooperative Activities since its establishment in 1967 and (2) to identify areas which we are currently pursuing relative to the "brave new world" just around the corner. In June 1967 the Librarian of Congress and the directors of the National Library of Medicine and the National Agricultural Library announced their intention to work jointly toward the development of a centralized data base and the attainment of compatibility in technical and other procedures and services of the three institutions insofar as this could be achieved in terms of the national libraries' individual statutory obligations. At that time they announced the formation of the U.S. National Libraries Task Force on Automation and Other Cooperative Activities (the "on Automation and Other" part of the title has since been discarded) as the vehicle for guiding the cooperative program toward this broad goal. Areas and products of this endeavor, envisioned from the beginning, included: (1) increased cooperative acquisitions, (2) fuller integration of cata- loging policies and procedures, (3) centralized production and dissemination of catalog information in machine-readable form as well as, (4) production of catalog cards and book catalogs, and (5) the coordination of bibliographical and reference services. The Task Force, under the chairmanship of first Stephen Salmon, 32 U.S. NA TIONAL LIBRARIES TA SK FORCE 33 executive officer, processing department, Library of Congress (June 1967- November 1967), and then Samuel Lazerow, chief, serial record division, Library of Congress (December 1967-April 1972), identified a number of problem areas for detailed study by working groups established by the Task Force. Members of the original Task Force included: Samuel Lazerow, LC; James P. Riley, NLM; and Bella E. Shachtman, NAL. ACCOMPLISHMENTS During the first four and one-half years several recommendations were made by the Task Force to the directors of the NAL, NLM and LC. These recommendations summarized the work pursued and are discussed in brief below. (For the full text of each recommendation see Appendix 1 .) Recommendation No. 1: Issued on February 20, 1968, recommended adop- tion of the MARC II structure for the communication of all biblio- graphic information in digital form between LC, NLM and NAL. In addition it was recommended that the MARC II data elements for monographs be adopted, but the organization of individual records and files may be unique to each library. This recommendation was accepted by the directors. Recommendation No. 2: On the same date as the first recommendation the Task Force also recommended a six-part action be accepted in the area of descriptive cataloging practice. These parts included: 1. Conference publication when name of conference and title of work are identical NAL and LC supply a bracketed additional title following the name of the conference in the title paragraph, when such a title is found in the work. NLM uses the supplied title only. There are justifications for either practice. Recommendation: The Task Force recommends that NLM change its practice to conform to that of the other two libraries. 2. Titles beginning with numbers The Anglo-American Cataloging Rule provides for spelling out in words numerals that appear at. the beginning of a title in a foreign language; LC and NAL follow this rule. NLM uses a technique based on computer filing. Recommendation: The Task Force recommends that inasmuch as machine requirements cannot be met by the cataloging rule as it stands LC be asked to consider the problem in connection with 34 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING the MARC format requirements and that MARC submit a recom- mendation to the Working Group on Descriptive Cataloging Prac- tices. 3. Publisher as author of series LC follows the Anglo-American Cataloging Rule and gives the publisher; NAL and NLM omit it. The present MARC II format does not provide a mechanism for indicating that the author of the series is the publisher. Recommendation: The Task Force recommends that LC arrange for MARC II to provide a mechanism to identify multiple uses of the same data in the record, in this case a tag which indicates that the author of the series is also the publisher. 4. Date LC uses "[n.d.]." NAL and NLM do not use [n.d.]. They substi- tute a date, such as "[18--?] ." Recommendation: The Task Force recommends that LC follow the practice of NAL and NLM and that a change in the cataloging rule be implemented. 5. Notes: Bibliographies LC and NAL follow the Anglo-American Cataloging Rule and identify bibliographies. NLM rarely uses this rule because almost all scientific books have bibliography notes; therefore, it is con- sidered that there is no need to show such an inclusion. This omission was more easily justified when there were no machine- retrieval capabilities. Now that there will be the possibility of compiling subject bibliographies, this needs reconsideration. Recommendation: The Task Force recommends that NLM change its practice and follow the Anglo-American Cataloging Rule. 6. Notes: Contents The cataloging rule makes machine manipulation of items in con- tents notes impossible in many instances. NAL and NLM rarely provide contents notes; LC does in some instances. Recommendation: The Task Force recommends: a. That LC arrange for MARC II to provide a mechanism for tagging the individual data elements within the contents notes. When this is done the Task Force will ask the ALA Descriptive Cataloging Committee to change the rule to make it possible to present the contents notes in a format that will be machine manipula table. b. That all three national libraries be asked to change their policies in the direction of more liberal inclusion of contents notes. U.S. NA TIONAL LIBRARIES TASK FORCE 35 Recommendation No. 3: The Task Force recommended in July 1968 that the directors accept a standard calendar date code to cover any form of date representation in machine-readable form. The recommendation was accepted. Recommendation No. 4: The Task Force recommended a Standard Character Set for Roman alphabets and Romanized non-Roman alphabets. The ALA character set was in effect adopted in January 1969. Recommendation No. 5: In February 1969 the Task Force recom- mended that the national libraries agree to adopt the Standard Language Code, and any subsequent addition to and revisions in the code as they become necessary for use in the communication of bibliographic information in machine-readable form. Recommendation No. 6: In April 1969 the Task Force recom- mended that a National Serials System be established. The recom- mendation was accepted. Ramifications will be discussed below. Recommendation No. 7: In June 1969 the Task Force recom- mended that the national libraries copy materials for each other without charge. Recommendation No. 8: In October 1970 the Task Force recom- mended, based on their April 1969 action, that the national libraries accent funding responsibilities for the National Serials Data Project and that the Task Force retain responsibility for policy direction of the pilot project irrespective of location. The Task Force's development of the National Serials Data Project through the pilot stage made possible last year's announcement of the ongoing National Serials Data Program under the sponsorship of the three libraries. This step and the simultaneous placement of the Task Force in close associa- tion with the Federal Library Committee will extend the benefits of these cooperative programs-both technical and nontechnical to the widest possible library and information science community. One of the most significant undertakings of the Task Force during the early years was that of the National Serials Data Project. Phase 1 of the project resulted in the identification of data elements needed for the control of serials by machine methods, and the development of the MARC serials format; Phase 2 was a project directed by Donald W. Johnson, funded by the NAL, with additional support from the Council on Library Resources, Inc., NLM, LC, and administered by the Association of Research Libraries. Policy direction came from the U.S. National Libraries Task Force. The objective of this pilot project was to create a machine-readable file of live serial titles in science and technology, to produce a variety of listings, and to report on 36 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING conclusions and results. The conclusions included the finding that a national serials data bank is technically and economically feasible; that such a bank should have its own machine-readable authority file for corporate names; that input and output should be in upper case; and that the question of entry should be resolved. Following Phase 2 a separate effort was undertaken. The present Nation- al Serials Data Program that Paul Vassallo now heads is really Phase 3 of the original serials effort. First priority is to provide a data base for the use of the three national libraries which are supporting the program, although it is being kept separate from the internal serials operations of each of the three libraries. The needs of these libraries will be taken into consideration, but the program will not become directly involved with the specific serials problems of each institution. Each individual library will determine the point at which it wants compatibility with the national program. The second priority will be to meet the needs of the national user community. The program will thus have a wider impact if solutions take into consideration not only libraries but also information services, indexing and abstracting services, publishers and distributors of serials. The third major priority is international cooperation the assignment of international standard serial numbers to insure compatibility. This inter- national system is the result of the combined efforts of various cooperative groups, and the numbering plan is now in final draft. Copies have been circulated and when reactions to the draft are received, a final statement will be published. It is hoped that the complete standard will be ready by fall 1973. Paris will be the headquarters for the international center, which will have responsibility for issuance of ISSNs to national centers. The U.S. national center will be the Nationals Serials Data Program. Other national centers are currently planned for Germany, Great Britain, The Netherlands, France and the Soviet Union. They may be functioning in a year or so. The national center acts on behalf of the international center by assigning ISSNs to publications emanating from that country. The U.S. national center (that is, the National Serials Data Program) has negotiated for a block of numbers for all serials titles cumulative 1950-70. The tasks to which the National Serials Data Program will address itself in the near future are: (1) development of a corporate authority file to take into consideration the various authorities used by the three national libraries; (2) assignment of ISSNs to U.S. prospective titles published from 1971 on and listed in New Serial Titles; (3) pulling of titles in science and technology from Bowker cumulations and assignment of ISSNs to them; and (4) working with U.S. NA TIONAL LIBRARIES TASK FORCE 37 the international center in the development of various products in cooperation with other national centers. CURRENT ACTIVITIES Since my appointment as chairman of the Task Force in April 1972, a close look has been taken at the progress and program. The national libraries have now decided to place a greater emphasis on cooperative activities in all areas rather than concentrating somewhat exclusively on automation. The emphasis on cooperation generally is reflected in the name now revised to the U.S. National Libraries Task Force on Cooperative Activities. Placement of the Task Force near the Federal Library Committee has also helped in this area. The directors of the three national libraries have agreed that the Task Force will direct its attention to broad matters and function as a policy group with representation of two members at the highest policy level from each library. With this in mind, the directors named the following persons to the Task Force: from NAL, Joseph F. Caponio, currently acting director, and Samuel T. Waters; from NLM, Joseph Leiter and Elizabeth Sawyers; and from LC, Paul L. Berry and Emond L. Applebaum. The three national librarians serve as a board of directors, with the director of NLM as the first chairman of the board. The leadership will rotate among the three directors and the board will meet quarterly. Marlene Morrisey, who assisted the previous chairman in much of the substantial work of the Task Force, is even more involved in program planning and implementation of research studies. The action plan developed has two phases: Phase I Redefine objectives Develop program plans Determine financial and other support required Develop plan for obtaining support Report to directors Phase II Develop a detailed design of each activity Assign personnel/committees/work groups Establish implementation schedule Develop a reporting schedule Define an evaluation program Modify objectives based upon activity to date Report to directors 38 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING Each project, or effort, involves consideration of every noted point. Discussions are based upon Guidelines For Participation developed by the Task Force, and approved by the directors. (See Appendix 2.) Currently there are four projects, one directly related to automation, ongoing. A fifth project is approved and about to be implemented, and several additional projects are now under discussion. Automation Policy Statement The Task Force felt that appropriate staff working within each of the national libraries should be fully aware of automation activities which may affect their planning and operations. A knowledge of in-force, pilot, or planned operations in all three libraries will encourage concepts of compati- bility. Further, it was believed that the staff should be aware of the plans for coordinated/cooperative activity held by the directors. The directors agreed that all institutional automation efforts should develop in a cooperative and coordinated manner. Thus complementary programs should be identified and close relationships cultivated as appropriate. It has been agreed that a joint policy statement should be produced to serve as a guide for future work. Program Reviews were held at LC, NAL, and NLM to review work programs. Work toward the development of a policy statement is presently ongoing. The schedule looks like this: 1. Policy recommendations for the directors consideration will be developed by each library probably by late June 1973. 2. The directors and their designees will meet with the Task Force to develop policies for coordination and cooperation. 3. A combined policy statement would be issued. 4. A combined meeting of appropriate national library staffs would be held to consider the policy developed by the directors. 5. A continuing updating effort will be implemented. The future of the Task Force hangs on success or failure in this area. Work is not awaiting the policy statement, however. Concurrently strong effort is being expended on the development of a Standard Order Form and the identification of subscription dealer performance standards. Order Form A Standard Order Form offers an opportunity to reduce clerical effort and errors and speed book order procedures. The concept is of potential usefulness to the library community in general. Library order forms presently vary in size and kind. Agreement on U.S. NA TIONAL LIBRARIES TASK FORCE 39 standard size, format, and content of an order form would simplify ordering procedures, facilitate order work, and maximize return. Use of a Standard Order Form in the three national libraries will simplify ordering work and will assure publishers and dealers easy access to all the necessary data required for prompt filling of orders. A Standard Order Form would record data at the earliest possible point and insure compatibility of records among the three national libraries. Such a form will also be useful in internal training programs in each of the libraries. This is also an oppor- tunity for other federal libraries to benefit from such a form. In addition, a Standard Order Form could be utilized in all aspects of acquisitions, i.e., exchange or gift work. A study was undertaken which will lead to development of a form for use in the three libraries, with agreement sought on size, data elements, and placement of data. This study is being pursued in-house, with a group including appropriate representatives from LC, NAL, and NLM. Jennifer Magnus of the Library of Congress heads the group. At the same time a Standard Order Form is being developed, the possibilities of a Standard Book Order Format are being considered. The International Standard Book Number is being used by publishers and librarians as a means of identifying books, particularly in machine applica- tions. Both publishers and librarians may benefit from use of the ISBN as a third segment of an identification code for book orders. The first and second segments of such a code would identify the buyer and the book dealer, respectively. Thus, a Standard Book Order Form would contain three num- bersa buyer number, a dealer number, and an item number, the latter being the ISBN. Commercial interests have encouraged use of a Book Dealer Identifica- tion Number to speed computerized order transactions. The concurrent use of the ISBN with the BDIN might assist publishers, dealers, and librarians in identifying stock, save time, and increase accuracy and ease of handling orders. Similarly a buyer identification number might further facilitate the transaction and simplify the order process. Users might benefit from the convenience of using assigned numbers in place of tedious repetition of names and addresses in placing and filling book orders. The code might become an essential ingredient of a machine order system. 1 In the national libraries the availability of a SBOF might save time in ordering procedures and in billing. Precise billing and delivery address could be established at the time of order. A modified SBOF is of potential use in 40 1973 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING interlibrary loan, and in union lists. A current study concerned with the SBOF will: 1 . survey the publishing and library communities to define the dimen- sions of the task and to determine the interest in such a standard; 2. determine potential areas of cooperation for those holding respon- sibility for designating numbers for buyers and dealers; 3. explore with appropriate groups (American Book Publishers Association, Bowker Company, American Library Association, American Book Sellers Association) the degree of interest and potential support for, and acceptance of, a SBOF; and 4. develop a program for the SBOF with an appropriate implemen- tation program delineated, or make recommendations for an alter- native course of action. Progress of the SBOF has been phenomenal. A draft form has been developed and, at the request of Jerrold Orne, an American National Stan- dards Institute Committee was established (April 1973) to consider the devel- opment of a form for all types of libraries. Subscription Dealer Study William Katz, State University of New York at Albany, is working on a very important study related to subscription agents. Subscription agents are employed by the three national libraries to procure a high percentage of serials obtained through purchase. These agents are reimbursed for various services rendered such as, central placement of orders, handling of supple- mental charges, alerting regarding births and deaths, claiming, speed in deliver- ing, etc. Few librarians are fully aware of what services subscription agents can and should provide. 2 Further, it is believed that because of this, many services are not supplied. There is a need to: (1) review subscription agent claims; (2) review actual services rendered; (3) develop a checklist of appropriate services; and (4) develop a mechanism for insuring agency compliance. The checklist and mechanism will result in a more efficient procurement process with a resultant savings in cost to each library and an improvement in services to the public. Under contract William Katz will provide the noted documentation. Depository Alan Rees, also under contract, is developing a design for a study to consider the need for a central depository for little-used materials in the Washington, D.C. metropolitan area. Approved but not yet undertaken is a material procurement oriented project. U.S. NA TIONAL LIBRARIES TASK FORCE 41 Cooperation in Procurement The acquisition of library materials, including selection and procure- ment, constitutes for the three national libraries a cluster of very complex and costly operations individually carried out by each library. These various operations will be studied and considered for possible coordination and, where appropriate, for sharing of implementation responsibilities. Such coordination of the acquisitions activities of the three national libraries could lead, wherever possible, to the elimination of unnecessary duplication of effort and to a reduction of unit costs. An evolutionary program of cooperation in the procurement of library materials might be made the subject of a study which could lay foundations for parameters and methodology for implementation in specific areas. A feasibility review leading to an action plan will be undertaken to: (1) identify areas of possible cooperative activity; (2) recommend such areas in priority order; (3) gather sample data in all aspects of such areas; and (4) develop an Action Plan-if feasible. All work would be pursued under the direction of the Working Group on Acquisitions and Collections Development of the U.S. National Libraries Task Force on Cooperative Activities. The chairman of the work group and the chairman of the Task Force would share responsibility for directing the effort. How does the Task Force undertake work? Early in the life of the organization work groups were established and that mechanism exploited. Now we use the "contract concept" and the work group process. A working fund permits quick response to useful and potentially positive projects. I have not summarized all the efforts of the U.S. National Libraries Task Force. I have attempted to touch the high spots and illuminate areas perhaps previously not well known. For example, work relating to subject headings has not been discussed in this paper; concern for easily utilized on-line access to bibliographic data has not been mentioned; Ohio College Library Center interest has not been identified; and current discussions regarding the NLM and NAL relationship with the MARC program have been specifically avoided. REFERENCES 1. "Appropriate organizations should designate agencies to assign the standard identification numbers for the various forms of material. Such agencies will require the cooperation of the 'national bibliographic center' if 42 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING the broadest coverage is to be attained, or the coordination of standard numbering might be made a function of the national bibliographic center itself." Proceedings of the Federal Library Committee Meetings, December 16, 1970 and April 28, 1971 on Implications to the Federal Library Community of the Airlie Conference on Interlibrary Communications and Information Networks, September 28-October 2, 1970. Washington, D.C., Federal Library Committee, 1972, p. 22. 2. Doares, Juanita S. Report on Survey of Subscription Agents Used By Libraries in New York State (ERIC doc. no. ED061950). New York Library Association. (Report of a survey conducted by the Technical Committee in 1970.) APPENDIX 1 Task Force Recommendation No. 1 : Adoption of MARC II Format (February 20, 1968) The U.S. National Libraries Task Force on Automation and Other Cooperative Services recommends that the directors of the national libraries indicate below their agreement to adopt the MARC II structure for the communication of all bibliographic information in digital form between their libraries. In addition it is recommended that the MARC II data elements for monographs be adopted but the organization of individual records and files may be unique to each library. The MARC II format has been developed by the Library of Congress in consultation with a Working Group of the Task Force representing the three national libraries, and the format reflects the requirements of NAL, NLM, and LC. Task Force Recommendation No. 2: Compatibility in Descriptive Cataloging Practices (February 20, 1968) The Working Group on Descriptive Cataloging Practices has met several times since December 1 5 and has submitted a report in which it has identified all problem areas for which it has responsibility. The members of this Working Group are: Emilie Wiggins, Head, Cataloging Section, NLM, Chairman C. Sumner Spalding, Chief, Descriptive Cataloging Division, LC Paul Winkler, formerly Assistant Chief, Shared Cataloging Division, LC Mrs. Jeanne Holmes, Chief, Division of Catalog and Records, NAL This group has concluded, after thorough and detailed study of all descriptive cataloging rules, that out of 43 elements, 35 present no problem, either because of no variation, or because of a variation so slight as to be inconsequential. These 35 no-problem elements are: 1 . Short title 2. Title in two or more languages U.S. NA TIONAL LIBRARIES TASK FORCE 43 3. Translation of Far Eastern title 4. Subtitle 5. Alternative title 6. Author statement 7. Author statement (Transposition) 8. Omission from author statement 9. Additions to author statement 10. Edition 1 1 . Statement of the number of volumes 12. Illustration statement in title paragraph 13. Imprint (Order) 14. Imprint (More than one place and publisher) 15. Imprint (Printer as substitute for publisher) 16. Supplied imprint 17. Inflections in imprint 18. Fictitious and imaginary imprint 19. Place of publication 20. Collation (Pagination) 21. Collation (Extent of text in more than one volume) 22. Collation (Illustrations) 23. Collation (Size) 24. Atlases accompanying text 25. Phonorecords accompanying visual text 26. Price 27. Series statement 28. Works in more than one series 29. Notes: Dissertation 30. Notes: Bound with 31. Notes: Limited use 32. Notes: Others 33. Tracing of secondary entry 34. Special Rules 35. Atlases One data element (Title Romanized) is being studied at NAL and will be reported on later. One element (Added Entries for Joint Authors) is being referred to the Working Group on Name Entry and Authority File for further study. We are therefore left with six elements on which there are compatibility problems and on which the Working Group has made recommendations. The Task Force considered these recommendations in detail on February 13 and requests that the Directors take the action as indicated below. 1. Conference publication when name of conference and title of work are iden tical 44 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING NAL and LC supply a bracketed additional title following the name of the conference in the title paragraph, when such a title is found in the work. NLM uses the supplied title only. There are justifications for either practice. Recommendation: The Task Force recommends that NLM change its practice to conform to that of the other two libraries. 2. Titles beginning with numbers The Anglo-American Cataloging Rule provides for spelling out in words numerals that appear at the beginning of a title in a foreign language, for example "10 [i.e. dix] annees du travail." LC and NAL follow this rule. NLM uses a technique based on computer filing, i.e. "10 annees du travail.=/ Dix. . . ." Recommendation: The Task Force recommends that inasmuch as machine requirements cannot be met by the cataloging rule as it stands LC be asked to consider the problem in connection with the MARC format requirements and that MARC submit a recommendation to the Working Group on Descriptive Cataloging Practices. 3. Publisher as author of series LC follows the Anglo-American Cataloging Rule and gives the publisher; NAL and NLM omit it. The present MARC II format does not provide a mechanism for indicating that the author of the series is the publisher. Recommendation: The Task Force recommends that LC arrange for MARC II to provide a mechanism to identify multiple uses of the same data in the record, in this case a tag which indicates that the author of the series is also the publisher. 4. Date LC uses "[n.d.]." NAL and NLM do not use [n.d.]. They substitute a date, such as "[18-?]." Recommendation: The Task Force recommends that LC follow the practice of NAL and NLM and that Mr. Spalding be asked to request a change in the cataloging rule. 5. Notes: Bibl. LC and NAL follow the Anglo-American Cataloging Rule and identify bibliographies. NLM rarely uses this rule because almost all scientific books have bibliography notes; therefore, it is considered that there is no need to show such an inclusion. This omission was more easily justified when there were no machine retrieval capabilities. Now that there will be the possibility of compiling subject bibliographies, this needs reconsideration. Recommendation: The Task Force recommends that NLM change its prac- tice and follow the Anglo-American Cataloging Rule. 6. Notes: Contents The cataloging rule makes machine manipulation of items in contents notes impossible in many instances. NAL and NLM rarely provide contents notes; LC does in some instances. Recommendation: The Task Force recommends: a. That LC arrange for MARC II to provide a mechanism for tagging the individual data elements within the contents notes. When this U.S. NA TIONAL LIBRARIES TASK FORCE 45 is done, Mr. Spalding will ask the ALA Descriptive Cataloging Committee to change the rule to make it possible to present the contents notes in a format that will be machine manipulatable. b. That all three national libraries be asked to change their policies in the direction of more liberal inclusion of contents notes. The Task Force is convinced that if these recommendations are approved a substantial step forward will have been made in the compatibility of the three library systems. The Task Force, therefore, unanimously recommends approval and im- plementation of these recommendations. Task Force Recommendation No. 3: Standard Calendar Date Code (July 30, 1968) The Working Group on Bibliographic Codes (Language, Place, Date Representation) submitted to the Task Force in March a proposed standard for representation of calendar dates in the data processing systems of the three national libraries. The Task Force considered this recommendation in detail on March 26, 1968 and, after discussion with the Working Group Chairman, returned the proposal to the Group for further work. The Task Force asked that the recommendation be restudied to see if the standard code could be expanded to cover day, month, year; month, year; and year. Members of the Group are: Patricia Parker, Library of Congress, Chairman Constantine Gillespie, National Library of Medicine Jeanne Holmes, National Agricultural Library Lucia Rather, Library of Congress The Working Group restudied the matter and submitted a revised recom- mendation for a standard calendar date code to cover any form of date representation in machine-readable form. Purpose of the Code: This code provides a standard way of representing calendar dates in the data processing systems of the national libraries and may be particularly useful for application in data interchange among federal agencies. General use of this standard code will eliminate the confusion caused by many different representations of dates. Definition: Date in this standard code refers to any of the following: a single year; a single month and year; or a single day in the Gregorian Calendar. Specification: Because bibliographic dates are frequently pre-twentieth century, a date standard for use in libraries must allow for representation of century as well as year. Other types of dates used for administrative purposes, e.g., date of entry of a record onto a file, will necessarily be current and therefore limited to the twentieth century. The recommended standard code provides for four digits to be used in a computer field to represent the year when the date information is expected to be limited to the year only and when the field may contain pre-twentieth century dates. 46 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING Examples of 4-digit year codes are the following: 1865 is coded 1865 1729 is coded 1729 1928 is coded 1928 The standard provides for a six-digit code, based on proposed United States of America Standards Institute (USASI) and Bureau of the Budget standards, to be used to represent dates in a date field limited exclusively to twentieth century dates. The six digits of this code will represent, in order, year, month, day, with the first two digits (00 through 99) representing the year, the third and fourth digits (01 through 31) representing the day of the month. If day or both month and day are missing, then that portion of the field is filled with zeroes. Examples of 6-digit codes are: April 1, 1968 is coded 680401 April 1968 is coded 680400 1968 is coded 680000 This code provides also for eight digits to represent a date in a date field containing pre-twentieth century dates or both pre-twentieth century and twentieth century dates. The eight digits of this code represent, in order, year, month, day, with the first four digits (0000 through 9999) representing the year, the fifth and sixth digits (01 through 12) representing the month of the year, and the last two digits (01 through 31) representing the day of the month. In this eight-digit code twentieth century dates must use a four-digit code for the year. If day or both month and day are missing, then that portion of the field is filled with zeroes. Examples of 8-digit codes are: April 14, 1865 is coded 18650414 November 11, 1915 is coded 19151 1 1 1 December 1932 is coded 19321200 1923 is coded 19230000 The Task Force reviewed this recommendation on July 2, 1968 and submits to the directors the following: Recommendation: The Task Force recommends that the directors of the National Libraries indicate below their agreement to adopt the standard described above for representation of calendar dates in the communication of bibliographic information in machine-readable form. This code reflects the requirements of the National Agricultural Library, the National Library of Medicine, and the Library of Congress. Task Force Recommendation No. 4: Standard Character Sets for Roman Alphabets and Romanized Non-Roman Alphabets (January 29, 1969) The Working Group on Character Sets has studied the design of charac- ter sets with the basic assumption that determination of standards must involve the following aspects: (1) Consideration of all the characters any of the three national libraries might wish to use to represent bibliographic data in machine-readable form; (2) consideration of the characters that could actually be put into digital form; and (3) consideration of the ways in which these characters could be represented on output devices once they were put in U.S. NA TIONAL LIBRARIES TASK FORCE 47 digital form. The studies have been complicated by the fact that over 70 languages in 20 alphabets are used in at least one national library and that provision must be made for certain diacritical marks and certain scientific characters. On the basis of the Group's findings concerning the needs of the three national libraries, a standard set of 175 characters, including many diacritical marks, certain scientific characters, and other special characters, has been developed. The character set is represented in digital form in an extended version of the American Standard Code for Information Interchange (ASCII). ASCII was developed as a 7-bit code by the United States of America Standards Institute (USASI), and by executive order of the President it has become a standard for the executive agencies. For library uses, the 7-bit ASCII code has been expanded to 8-bits in accordance with suggestions from the chairman of the USASI committee on ASCII. Provision has also been made to contract to a 6-bit code for 7-level tape users. The ALA Machine- Readable Catalog Format Committee has approved the suggested character sets. The members of the Working Group on Character Sets are: Lucia Rather, Library of Congress, Chairman Irvin Weiss, Library of Congress Theodore Leach, Library of Congress Elizabeth Sawyers, National Library of Medicine Vern Van Dyke, National Agricultural Library Lillian Washington, National Library of Medicine The Task Force has reviewed the Group's findings and, after arranging for certain clarifications to be made, has voted unanimously to submit to the directors the following: Recommendation: The Task Force recommends that the directors of the three national libraries adopt the character sets described in the attached layouts as standards for use in the communication of bibliographic data in machine-readable form. Proposed Character Set for Roman Alphabet and Romanized Non-Roman Alphabets 1. Scope The character set relates solely to languages in the Roman alphabet and romanized forms of languages in other alphabets. (Non-Roman alphabets will be considered at a later time.) 2. Criteria Governing Selection of Characters 2-1. Frequency of occurrence of character 2-2. Degree of necessity in expressing character when it occurred 2-3. Possibility of substituting one character for another or of expressing a character by writing it out 3. Technical Specifications Governing Structure of Coded Character Set 3-1. Each code of a character set will contain the same number of binary digits 48 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING 3-2. The character set is structured to facilitate derivation of larger or smaller code patterns 3-3. The character set will be structured to facilitate ordering of its members in respect to usage 3-4. The composition of a character set is related to the state-of- the-art in the technology of input/output devices 4. Digital Codes The correlation of the character set to digital form code is based upon ASCII (American Standard Code for Information Interchange) standard. In conformance with the design considerations of ASCII (7-bit code), the character set is also correlated to an 8-bit code and a 6-bit code. The basic digital form code for the character set is the 8-bit code. 4-1. The 8-bit code is an extended form of the standard 7-bit ASCII. Some of the standard ASCII characters such as the braces or the backwards slash are not proposed for the Library's character set. However, no characters will be substi- tuted for these code positions. Other characters such as diacritical marks will be left in their standard position (un- used) and duplicated in another portion of the code set reserved for special characters and diacriticals. 4-2. The 7-bit code will be derived from the 8-bit code by removing the 8th bit. Those characters which previously had an in the 8th bit will be considered part of the standard 7-bit ASCII set. Those with a 1 in the 8th bit will be considered part of the non-standard set. A SO (shift out) control character will be used to go from the standard set to the non-standard. The code will stay in the non-standard mode until a SI (shift in) control character is reached. 7-bit 8-bit SI USASCII 8th bit = SO (special characters and diacriticals) 8th bit = 1 4-3. The 6-bit code will be derived by removing the 6th bit and the 8th bit. The 8-bit code set will be divided into 4 sets as follows: Columns 2,3,6 & 7 = Standard set Columns 0,1,4 & 5 = Non-standard set (1) Columns A,B,E,F = Non-standard set (2) Character 7B in the standard set will be used as a non-locking U.S. NA TIONAL LIBRARIES TASK FORCE 49 shift code to reach non-standard set (1); character 7D will shift to non-standard set (2). The presence of one of these codes will indicate that the next character is in one of the appropriate non-standard sets. The code will then be auto- matically shifted back to the standard set. 5. Escape Codes All subscript, superscript, and the 3 Greek alphabet characters will be in supplementary sets and will be reached by the use of the ESC (Escape) character followed by a specific character (or set of characters) to indicate which supplementary set is needed. The exact escape sequence has not yet been determined since the formula for devising escape sequence is still under consideration by the USASI committee on ASCII. Task Force Recommendation No. 5: Standard Language Code (February 18, 1969) The Working Group on Bibliographic Codes has submitted to the Task Force a proposed Language Code developed for use in the MARC II format. This language code is regarded as a provisional list because it will be reviewed and updated as records containing languages not previously included in the list are added to the MARC data base. The sources of the languages used in the development of this list are: 1. the languages processed by the National Agricultural Library; 2. the MARC I Pilot Project language code list and the languages processed by the Library of Congress; 3. the languages contained in the MEDLARS LANDS File (Languages and Subheading File) of the National Library of Medicine; 4. the languages studied by the Defense Languages Institute of the Department of Defense. A primary aim of the working group was to produce a language code that would be compatible with other codes that might become national standards. To this end both the Data Elements and Codes Office of the Bureau of the Budget and the USASI Z-39 Committee on Library Work and Documentation and Related Publishing Practices were consulted. Neither of these groups, however, is developing a language code. Characteristics of the list of languages: The languages represent the major body of published literature. The form of the language names used was based on examination of the various tools used at NAL and NLM, the language authority files of LC, consultation with various language specialists, and study of the language list of the Center for Applied Linguistics. Three- letter mnemonic codes using the first three letters of the English form of the name have been used in most cases. The members of the Working Group on Bibliographic Codes are: Patricia Parker, Library of Congress, Chairman Constantine Gillespie, National Library of Medicine Jeanne Holmes, National Agricultural Library Lucia Rather, Library of Congress The Task Force has studied the Working Group's proposal, consulted in detail with its chairman, reviewed revisions in the proposal, and voted unani- 50 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING mously to submit to the directors the following: Recommendation: The Task Force recommends that the directors of the National Libraries indicate below their agreement to adopt the Standard Language Code, and subsequent additions to and revisions in the Code as they become necessary, for use in the communication of bibliographic information in machine-readable form. Language Code List (developed for use in the MARC II format; the list will be reviewed and updated as records containing other languages are added to the MARC data base) Inclusions: Written languages only are included in the list. Where one spoken language is written in two different sets of characters, both written languages are included. (Example, Serbian and Croatian are the same spoken language but the former is written in the Cyrillic alphabet and the latter in the Roman alphabet.) Exclusions: Some languages have not yet been included in the list, e.g., certain East African languages, even though it is anticipated that literature in these languages will be received. These languages and others will be added to the list and coded as they are needed. Criteria for Assignment of Discrete Codes: Three-letter mnemonic codes using the first three letters of the English form of the name have been used in most cases. (Exceptions were necessary in some of the languages because of redundant initial letters for some languages, e.g., Arabic and Aramaic, Kannada and Kanuri, Malagasy and Malay, etc.) In the case of the modern and the older forms of some languages, the initial letters of each part of the language name were used to form the code, e.g., GMH for German (Middle High) and GOH for German (Old High). The code MUL for Multilingual was included for the purpose of providing the option of assigning a single code to represent works published in several languages within one physical piece. Cross-References: The following types of cross-references are provided: (1) for variant spellings of a language name, e.g., Biluchi see BALUCHI Denca see DINKA Kechua see QUECHUA Pashto see PUSHTO (2) from older forms to newer forms of a language name, e.g., Middle Persian see PAHLAVI Siamese see THAI (3) from lesser known and used forms of a language name to better known and more popularly used forms, e.g., Castilian see SPANISH U.S. NA TIONAL LIBRARIES TASK FORCE 51 Judaeo-German see YIDDISH (4) to indicate the alphabet in which one spoken language may be written when several written forms exist, e.g., Hindustani (Arabic) see URDU Hindustani (Nagari) see HINDI Serbo-Croatian (Cyrillic) see SERBIAN Serbo-Croatian (Roman) see CROATIAN For convenience in the use of the language list cross-references are given the same code as the accepted language name to which they are referred. (Example: both Erse and Scots Gaelic are coded as GAE since this is the code for GAELIC to which both of these alternate names are referred.) Under each accepted language name cross-references are provided to indicate all the variant and sub-ordinate names which are referred to the one given accepted name. Cross-references and variant and sub-ordinate names are printed in lower case, while codes and the established language names are printed in upper case. Format: The list appears in two forms: (1) in alphabetic sequence by the language name APR AFRIKAANS AKK AKKADIAN ALB ALBANIAN AMH AMARINYA See AMHARIC AMH AMHARIC x Amarinya (2) in alphabetic sequence by language code APR AFRIKAANS AKK AKKADIAN ALB ALBANIAN AMH AMHARIC x Amarinya ANG ANGLO-SAXON x English (Old) x Old English Task Force Recommendation No. 6: National Serials Systems (April 1 1, 1969) Because the national libraries of the United States share basic responsi- bility for the acquisition and servicing of serial literature and because national concern about the control of this voluminous quantity of significant research material is mounting, the U.S. National Libraries Task Force and its Serials Working Group have given priority consideration to alternative approaches to meeting the three directors stated objective of developing "a national data bank of machine-readable information relating to the location of hundreds of thousands of serial titles held by American research libraries." This task is complicated by the fact that the national libraries, and the library community in general, participate in an informal network of pub- 52 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING lishers, distributors, secondary and bibliographic services, individual sub- scribers, and funding agencies all of which are involved in the generation, processing, dissemination, or utilization of serials. This network is largely uncoordinated, is proving increasingly costly, and has been ineffective in meeting national needs. In arriving at its recommendations the Task Force has been concerned not only with over-all objectives and long-range plans for a national system, but also with actions that need to be taken now, in view of the completion of phase 1 of the National Serials Data Program, for which the Library of Congress served as executive agent. The recommendations that follow include the next immediate steps as well as the long-range and the specific as well as the general. The detailed background to the recommendations, and the alternatives to them, prepared by the Serials Working Group, is appended. The Task Force, after consultation with its Advisory Committee and later internal discussion, submits the following: Recommendation: 1. That the Directors of the National Libraries, acting on their own behalf, and jointly as the agent for the Joint Committee on the Union List of Serials, establish a National Serials System. 2. That the primary objective of the National Serials System be to provide to qualified requestors in the U.S. timely access to appropriate portions of the world's serial literature. 3. That a secondary objective of the National Serials System be to provide to the management of the serials system and its several subsystems (including the directors of the National Libraries and other major federal information systems) the information which they need in order to make decisions regarding the acquisition, processing, storage, and dissemination of the serial literature. 4. That a national serials center be established to carry out the objectives of the National Serials System. The center would develop formats and data conversion techniques as required for the developmnet of an appro- priate data base, and would assume responsibility for the production of appropriate bibliographic tools. 5. That the initial data base be limited to science and technical serials including new titles in the three national libraries or any other limitation defined by the three directors. 6. That the directors of the three national libraries seek funds immediately to implement a national serials center to work in close cooperation with the National Libraries Task Force. 7. That, upon availability of funds, the directors of the three national libraries give consideration to the direction, location, and other manage- ment aspects of a national serials center to assure adequate space, personnel, etc., and ease of coordination with the National Libraries Task Force. Task Force Recommendation No. 7: Photocopying Cooperation (June 18, 1969) U.S. NA TIONAL LIBRARIES TASK FORCE 53 Utilization of photocopies in lieu of loans of original library materials is a well established cooperative lending mechanism. The three national libraries currently request photocopies from each other when original materials are unavailable for loan. The number of photocopies requested by any one library is relatively small. One national library currently supplies such photocopies without charge; the other two libraries charge for the copies (their photodupli- cation work is performed through revolving funds). These charges require time-consuming approval and payment procedures and place an administrative burden on the requesting library. To simplify the procedure and assure an equitable photocopying arrange- ment, the Task Force believes that the photocopying charges should be absorbed by the library supplying the photocopy. The total cost to any one library would be small and would be more than compensated by the free copies received from the other libraries. In addition, savings in staff time would be realized through elimination of the present administrative approval and payment procedures. For these reasons the Task Force has voted unanimously to submit to the directors the following: Recommendation: Recognizing the needs of each of the three national libraries for photo- copies of research materials not available for loan in the original, the Task Force recommends that the directors adopt the following policy and arrange for its implementation in each of the libraries: The three national libraries will copy materials for each other without charge, subject to the regulations each observes for copyrighted materials. A suitable mechanism shall be worked out in each library to implement this policy. Task Force Recommendation No. 8: Continuation of National Libraries' Serials Pilot Project (October 15, 1970) It is the unanimous view of the U.S. National Libraries Task Force that it is imperative that the National Libraries' Serials Pilot Project be continued at least at its present level of staffing. It will require total support of $50,000, exclusive of computer time, to carry it at this level until July 1, 1971. The Task Force believes that it should be continued under the present Project Director at least until July 1, 1971 to enable the Project Director to docu- ment adequately the experience of the Project, after which another Director should be found. He could either be furnished by one of the libraries, or consideration could be given to requesting the CLR to permit the Task Force to designate one of its systems specialists as Project Director until permanent arrangements can be made. The Task Force therefore recommends: (1) that the directors of the three national libraries indicate not later than November 1, 1970 the funding arrangement they wish to share to provide the $50,000, exclusive of computer time, required to continue the Project until July 1, 1971; and (2) that the Project remain in its present location, under ARL sponsorship, until such time as the Project can be adequately supported at the 54 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING Library of Congress, with such transfer to be effected not later than July 1, 1971. The U.S. National Libraries Task Force should retain responsibility for policy direction of the Pilot Project irrespective of location. APPENDIX 2 GUIDELINES FOR PARTICIPATION BY U.S. NATIONAL LIBRARIES' STAFF IN THE COOPERATIVE PROGRAMS UNDER THE U.S. NATIONAL LIBRARIES TASK FORCE ON COOPERATIVE ACTIVITIES Background The U.S. National Libraries Task Force on Cooperative Activities was established in 1967 by the directors of the three National Libraries of the United States (Martin M. Cummings, M.D., Director of the National Library of Medicine; L. Quincy Mumford, Librarian of Congress; and John Sherrod, Director of the National Agricultural Library) to identify, recommend, and implement policies, procedures, and programs directed at strengthening and extending cooperation in activities and processes carried on in each of the three institutions. In creating the U.S. National Libraries Task Force, the directors indicated their expectation that intensified cooperation was essential in their determination to make certain that National Library resources are adequate to cope with problems created by the "information explosion" and to assure that immediate access to research materials and information is available to the nation's libraries. The over-all purpose of the U.S. National Libraries Task Force on Cooperative Activities, in the words of the directors, is to "improve access to the world's literature in all areas of human concern and scholarship, so that comprehensive access to the materials of learning can be afforded to all citizens of the United States." The directors of the three National Libraries have requested the full cooperation of their respective staff members in the identification of new areas where cooperation among the three National Libraries merits examina- tion and experimentation with a view to increasing productivity; accelerating, improving, or extending services; conserving manpower or other resources; developing and adopting uniform standards in National Library policies and practices where such standardization is appropriate, legal, and feasible; and building closer unity between the National Libraries and the clientele they collectively serve. Guidelines The following guidelines have been developed to assist in the operation of the U.S. National Libraries Task Force and its Working Groups: U.S. NA TIONAL LIBRARIES TASK FORCE 55 1. The six members of the U.S. National Libraries Task Force, appointed by the directors (two from each National Library), shall be officers at the policymaking level in their respective libraries. The members will represent their respective libraries and act as representatives of their libraries in all Task Force operations. They will have a full awareness of the policies of their respective libraries concerning matters that may come before the Task Force, with authority to speak as appropriate for their libraries. As representatives of their libraries invited by their direc- tors to lead this significant and extensive cooperative effort, they will explore without reservation opportunities for developing new, innovative cooperative endeavors; 2. Chairmen of Working Groups, named by the libraries, shall be know- ledgeable in the areas of concern to the respective Working Groups and shall have full understanding and cognizance of their respective libraries' programs in the individual areas of study. The recommendations and products resulting from the Working Groups' efforts shall be developed with consideration of existing policies and practices of the three institu- tions, but shall not be limited by differences in such policies and practices; 3. Working Group members, appointed by the libraries, shall reflect the interests, policies, and procedures of the institutions they represent, but shall not be limited in their recommendations or expressions of view by existing policies or attitudes in such institutions; 4. The full structure of the U.S. National Libraries Task Force and its Working Groups at all times shall reflect the directors' objective of working toward the furthering and extending of national library cooper- ation in all areas of concern to national library economy, efficiency, and service. MADELINE M. HENDERSON Institute for Computer Sciences and Technology National Bureau of Standards Washington, D.C. Prospectus for a Federal Library Cooperative Center* In order to discuss the activities within the federal community looking toward a federal library cooperative center, it may be useful to review the paths by which our group got started on the subject. For that purpose I will summarize the results of a survey of the federal library community which convinced us that the needs of that community would be well served by cooperative undertakings. The Federal Library Committee was established in 1965, with power to recommend policies to achieve better utilization of federal library resources and facilities, and to promote more effective planning and operation of federal libraries. To this end, the committee is authorized to examine and evaluate existing federal library programs, including study of the need for and potential of technological innovation in library practices. The parent committee devel- oped a mechanism of task forces, subcommittees and work groups to fulfill its functional responsibilities. The Task Force on Automation of Library Opera- tions, in particular, was established to review and report upon the status of automation activities in federal libraries, to encourage development of com- patible automated systems where feasible, to furnish guidance to federal admin- istrators and librarians on automation problems in libraries, and to provide liaison between federal libraries and other groups interested in the application of automatic data processing to information and document retrieval. As a voluntary group of people engaged full time in their own agencies, the Task Force has directed its efforts and limited resources largely to the collection of information about federal library automation. The group recog- *Contribution of the National Bureau of Standards. Not subject to copyright. 56 A FEDERAL LIBRAR Y COOPERA TIVE CENTER 5 7 nized that some libraries are already operating such systems, some are plan- ning systems, and many are seeking guidance in applications. It felt, however, that a great accumulation of experience in library automation was to be found in the federal government, and that study of this wealth of experience would benefit not only the federal agencies, but the library and information services community as a whole. I must note here also, that the Task Force on Automation has not included the three national libraries in its efforts and concerns. Recognizing the disparity between size and resources of most federal libraries and those of the national libraries-LC, NLM, and NAL as well as the existence and program of the National Libraries Task Force on Cooperative Activities de- scribed by Cylke in the preceding article, the Task Force determined to concen- trate on the activities and needs of the rest of the federal libraries. As one step in its program to collect information about federal library automation, the Task Force served in a technical advisory capacity for a broad survey of the current status of automated operations in federal libraries. A questionnaire survey was conducted in 197071 by System Development Corporation with the support of the U.S. Office of Education, Bureau of Library and Educational Technology. The survey was designed to accomplish three goals: (1) to define library operations susceptible to automation, whether such operations are now being automated or not; (2) to describe automation techniques of potential use in library operations, both those techniques now being applied and those of possible interest for library applications; and (3) to establish criteria for determining the feasibility of automation ("what to automate"), the types of hardware and software available for library automation, and the various factors to be taken into account in considering library automation possibili- ties. Survey Results A general picture of the federal library community was gleaned from the survey. It shows that the community is widely dispersed within the U.S. and around the world (figure 1); contrary to local belief, only 7 percent of all federal libraries are within the metropolitan Washington, D.C. area. Perhaps even more surprising, only 60 percent are located within the continental United States. The size of the libraries was determined from a number of factors. For example, the median size for a collection in a federal library is 16,500 total holdings (table 1). Books are predominant among those holdings, but there are also some less traditional materials such as audio recordings, maps, and films. 58 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING A FEDERAL LIBRARY COOPERATIVE CENTER 59 Number of Respondents Under 5,000- 20,001- 50,001- Over Type of Material AT 5,000 20,000 50,000 200,000 200,000 Books 929 280 454 137 46 12 Serials 877 827 34 12 4 Government documents 603 528 46 14 11 4 Pamphlets and reprints 591 546 36 6 2 1 Phonorecords, tapes, etc. 523 511 11 1 Maps and charts 520 502 10 5 3 Technical reports 491 375 57 21 21 17 Internal reports 376 341 22 7 4 2 Pictures 363 345 12 2 3 1 Films 354 344 6 1 2 1 Table 1. Types of Materials and Number of Titles Held Budget figures also indicate the size of a library; the median respondent spent less than $27,000 for materials, staff, and equipment in FY 1970. The total budget reported by all respondents is approximately $60 million; about two-thirds is devoted to personnel (table 2). Considering the heavy labor costs, and the preponderance of small libraries in the federal community, it is obvious that the most effective and efficient use of that labor is necessary to optimize operations and services. The Task Force believes that some amount of automation offers a potential here, either directly in the larger libraries or through cooperative centers and shared services for field libraries and other smaller libraries. Another indicator of size is staff; the majority of the responding libraries have fewer than three staff members (table 3). Typically, the federal library has one librarian who may or may not be a professional and who may or may not have supporting staff. The overall ratio of professional to nonpro- fessional staff is 1:1.3. These figures again suggest the potential of automation as a means to optimize operations and services and make the most effective use of those personnel. A series of questions in the survey was directed to the subject of cooperative networks, involving more than interlibrary loan and operating outside the parent agency. Only 10 percent of the respondents said that they were involved in such networks (table 4). These networks were rather small, involving ten or fewer libraries, but in some instances they covered extensive geographic areas and helped to augment the small staffs and collections already noted. 60 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING O o O O o * o" o" o" o" v>" V o 1- bo 00^ i in c r 1 ^2 V* 1 ^ 1 O \ < < o ^^ in (N O r- o wo t3 -~ Tf O o "S 0> ** in O O o ? ^ T3 C o a C/5 3 CN ]~l 5 fi O) */=> 1- C C O o r~ o m o r- \o 00 V C3 1 o" oo CO u -* o ^ CN ^ l ^^ w^ JJ CO 09 3 4) 2 CO "O ^J &9 00 r- ON o rf e 4* l 1 r~ m 3 oo" 0. X - CO '3 00 f"H CU a. 5 ^ ^1 5" ^ o r- CO T3 v= CO SN ." ^ . 'C c^ 0) V > 3 v> ON OO ^ 00 H 0) X) ^ c^ \o o f**. o^ fli 00 r^ r- (N 1 O) b .2 "o. 1 services C O M o ^ ^* M ^- "c CO 0) 4^ M -|.j j sj c G E g a CO l-l 2 u co g -| C 6 0) 5 iS * O O O O O O V r~ oo o o o oo bp CO OO CN ^" ^ "" a 1 1 1 1 1 1 in O n in co in O ^ O O O O 1M i! O O O to E? 5 (S - _',-]__'_ 3 1 U CM >n in in in in II oo o vo r^ o o en in vo'vo co oo JJ re H * ^ oo in in ON 00 vo co in vo vo 'C -C .0 ^ C C C O C co co g 3 O <=0 to 'g -H ^ 5" . ON O i^^ 5 lil A FEDERAL LIBRARY COOPERATIVE CENTER 61 Activity Percentage of Networks Performing Activity Union catalog or list Preparation of subject bibliographies Centralized cataloging Training courses, seminars Centralized acquisitions Centralized reference Other technical processing Indexes, other bibliographic aids (64 f^"} ) m(S 1 1 *f^1 1 : 5) ,(A\) M^ ) .f^n 10 20 30 40 50 60 70 Table 4. Activities Performed or Planned in Nonagency Networks in Which Respondents Participate Interlibrary activities other than formally established networks tend to involve other local libraries rather than more distant ones, even to the extent of more cooperation with local nongovernmental libraries than with parent agency libraries outside the immediate area. This pattern held true for all but exchange of materials (table 5). These activities, of course, also serve to augment the library's resources. Federal libraries, in spite of such constraints on their resources, do a creditable job of serving the needs and requests of their patrons. A relatively high proportion (27 percent) of the respondents said they use information retrieved from machine-readable data bases to answer some user inquires. Sixteen of these respondents have terminals on-line to the data bases, the rest submit written, formated search requests. Since these are not large libraries, they must be considered in the vanguard in library use of these tools. Most of the federal libraries with automation programs have emphasized systems related to user services (cataloging, reference services) as opposed to housekeeping operations. In addition, the computer is used for information retrieval, the publishing of bibliographies for SDI, and for abstracting and indexing. Most of the automation efforts have been in comparatively large and well-supported libraries. However, although the libraries with automation pro- 62 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING Number of Respondents Interacting with: Kind of Library Libraries within Local Area Libraries outside Local Area INTERLIBRARY LOAN Other libraries in own agency Other federal libraries Nonfederal governmental libraries (e.g., state libraries) Nongovernmental libraries 464 459 375 497 417 384 278 399 PHOTOCOPYING Other libraries in own agency Other federal libraries Nonfederal governmental libraries (e.g., state libraries) Nongovernmental libraries 186 194 134 205 190 173 115 180 REFERENCE ASSISTANCE Other libraries in own agency Other federal libraries Nonfederal governmental libraries (e.g., state libraries) Nongovernmental libraries 293 263 181 286 230 186 121 274 EXCHANGE OF MATERIALS Other libraries in own agency Other federal libraries Nonfederal governmental libraries (e.g., state libraries) Nongovernmental libraries 281 186 90 122 231 142 66 80 Table 5. Respondents' Interaction with Other Libraries by Activity grams are among the most advantaged federal libraries, none of them has resources comparable to large public and university libraries. At the lower end of the spectrum the libraries have total budgets of less than $75,000 a year, have one professional librarian, and fewer than 37,000 total holdings. This indicates that automation is being done even in small federal libraries (table 6). The overwhelming majority of respondents from all the libraries, how- ever, reported that local resources are inadequate to support automation, and they are very much in favor of the idea of centralized automation support and A FEDERAL LIBRARY COOPERATIVE CENTER 63 First Third Median Quartile Quartile Range BUDGET (Total for 56 libraries: $18,792,584) $176,500 $74,500 $464,000 $10,000- 1,765,000 STAFF Professional 1410 [N=53] 4.0 1.0 9.0 1-53 1412 [N=14] 2.0 1.0 3.5 1-4 Other [N=43] 2.0 1.5 5.5 1-57 Subprofessional [N=43] 6.0 3.0 10.6 1-46 Clerical [N=36] 4.0 2.0 7.0 .5-33 HOLDINGS Total Collections [N=56] 150,000 37,700 367,000 2500- 750,000 Estimated Percent in Microform [N=42] 5% 1% 25% 1%-80% Table 6. Resources in Federal Libraries services. In developing the survey questionnaires, a number of questions were included dealing with attitudes toward automation, centralized services, par- ticipation in networks, standard program packages, and cooperative arrange- ments. The results of the attitudes questions show, among other things, a strong tendency on the part of the librarians to be realistic about automation and a desire on their part for the Federal Library Committee to provide them with more support in the planning stages. Task Force Program The overall survey results, 1 including the examples of participation in cooperative and coordinated programs to help augment resources, influenced the Task Force on Automation of Library Operations in developing its program for current and future activities. The Task Force has set as one of its objectives, as noted earlier, to furnish guidance to federal librarians and administrators on problems of library automation. This function will take a number of forms including, at the present, cooperating in a study of the technical and administrative feasibility of the concept of a centralized service operation for federal libraries. 64 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING For several months in 1972 an unofficial study group, composed of a number of federal library directors from the Washington, D.C. metropolitan area and members of the Task Force on Automation, examined and discussed several aspects of possible cooperative effort among federal libraries, particu- larly for the purpose of determining the technical and administrative feasi- bility of establishing a Federal Library Cooperative Center. The initial func- tion of such a center is thought to be shared cataloging, with additional services and products desired by cooperating agencies to be added as quickly as possible. In order to proceed with its evaluation responsibility, the study group sought input on the technical characteristics of various cooperative or central- ized services. That is, we wanted to share the experiences of those operating on-line cataloging, circulation, search and retrieval, acquisition, serials, and other bibliographic systems. In addition, the group wanted input from those involved with central- ized and cooperative ventures of various types, i.e., interagency centralized processing, regional library cooperatives or subject-oriented cooperatives. From these operational facilities we sought to learn about administrative and mana- gerial aspects as well as to get some data on the impact of the cooperative or centralized operation on the services, products, and costs of the individual participants. Further, the group set out to examine the applicable federal bureau- cratic and organizational structure which can provide for cooperative activities. The legal or legislative characteristics of the federal establishment must be considered, as well as the language and intent of the authorizing or enabling charters of appropriate departments and agencies, to determine what is con- ducive to or restrictive of federal cooperative activities. Also the responsibili- ties of the OMB, GSA, and GAO need to be studied, and federal policies or programs in regard to computer utilization have to be examined. These studies, we felt, should point out ways to accomplish the necessary sympathet- ic consideration of our recommendation for the establishment of a coopera- tive center. The systems or services examined include those of Stanford University Libraries (the BALLOTS systems); the Shawnee Mission School System in Kansas; the systems of NLM, NAL, NASA, and ILO; and the activities of NELINET and SLICE; and, of course, the OCLC. In addition we were briefed on the characteristics of BIBNET-1000, the proprietary program of Informa- tion Dynamics Corporation. We were fortunate, furthermore, in being able to contract with Freder- ick Kilgour of OCLC for a study of the feasibility of a cooperative center A FEDERAL LIBRAR Y COOPERA TIVE CENTER 65 similar to the OCLC operation for the particular situation of federal libraries. Based on his extensive knowledge of the field and his broad understanding of the federal community, Kilgour reviewed for us the general library problem, the solution offered by a computerized cooperative, and the specific potential of that solution for federal library problems. He recommended a set of objectives for a Federal Library Cooperative Center, defined organization and funding arrangements, and described the implementation and operations of such a center. Kilgour's report added immeasurably to our study of the center concept. It is our conclusion that in spite of shortcomings or the limited develop- ment of a number of these programs, the evidences of success in key elements of their operation suggest that a cooperative effort among federal libraries would be successful in offering useful outputs, would be technically feasible, and can be made economically viable. Current Efforts We have, therefore, established a Work Group on the Federal Library Cooperative Center (FLCC) as a formally constituted unit of the Federal Library Committee. The work group is pursuing two objectives: one is to draft a proposal for a planning grant for the development of a FLCC. The purpose of the planning grant should be the specification of the organization and structure of the FLCC, its functions and management, the specification of a program of action (i.e., tasks in priority order) in the development of services and outputs, the specification of resources required (manpower, equip- ment, etc.) for reasonable operation; a calendar for their acquisition; and other elements that may be determined. As a first step in drafting this proposal, we developed the following statement of goals for the FLCC: 1. Facilitating the sharing of resources among federal libraries for the purpose of: (a) making those resources freely and widely available to users when and where they need them, and (b) enabling libraries to reduce their inventories while expanding their services through access to other collections. 2. Providing means for reducing the rate of rise in the operating costs of libraries by: (a) increasing the productivity of library personnel through effective use of automation, and (b) making available access to various information products and services to supplement the libraries' internal efforts. 66 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING We feel that this program of cooperative effort among federal libraries will be based initially on applications of computer, communications and information technologies and will use these technologies to enhance the missions of libraries to furnish their users with the information needed to conduct agency programs, arrive at managerial decisions, and contribute to policy-making in a timely and thorough fashion. Elements of the cooperative effort will include: 1. On-line access to computer-based files for shared cataloging of various collections: monographs, serials, technical reports, maps, audiovisual materials, etc.; 2. On-line access to computer-based files for retrieval of references in answer to specific questions, for preparation of current awareness alert- ing services, for compilation of special bibliographies, etc.; 3. On-line access to files and programs for technical processing, for book ordering and serials subscription efforts; 4. Generation and maintenance of statistics on these cooperative activities, to support the operations and plans of the participating libraries. Further elements of this program of cooperative effort will in general take the form of implementing projects developed by the Federal Library Committee and its various task forces dedicated to specific problems of the federal library community. The second objective being pursued by the work group is an experi- mental hook-up to OCLC for the purpose of providing hands-on experience with shared cataloging in an on-line environment. This is not meant as a test of the OCLC system; that has already proved its feasibility and effectiveness. Rather it is a test of the concept of shared cataloging for the federal community and a means for federal libraries to try out on-line access to a large data base consisting of LC MARC records and additional MARC-type records. The negotiations with OCLC to provide this experimental hook-up include adding the OCLC system to the TYMSHARE network. This means that the test would be available to selected federal libraries in cities through- out the U.S., by means of a local phone call. Also, the TYMSHARE system is compatible with a wide range of terminals which means that federal libraries having access to terminals already operating within their agencies have a good chance of participating in the experiment with a minimum initial investment. The work group and the parent Federal Library Committee are now seeking support for the startup costs in order to get this experiment under- A FEDERAL LIBRARY COOPERATIVE CENTER 67 way. These costs include modification of the OCLC system to connect it to TYMSHARE, particularly programming an interface for the front end of the OCLC system so that the multiplicity of terminals available through TYM- SHARE can be accepted; and costs of the TYMSHARE hook-up and equip- ment, i.e., the cost of installing and maintaining a TYMSHARE node at OCLC. The participating federal libraries, then, will bear the expense of the terminals plus the variable costs for system use: connect hours, characters transmitted, catalog cards requested, and titles matched and processed. An approximation based on these figures would suggest that the variable cost might be $2.10 per title. So the Work Group on the FLCC moves forward in its task of specifying a federal library program of cooperation. The task bears promise of success because the climate for such cooperative programs is more hospitable today than has usually been the case. Tight budgets and limited resources are facts of life for all libraries, but only recently has there been a real appreciation for what sharing of resources and cooperative programs can accomplish in making it possible to give good service within the limits of those resources. This appreciation has been coupled with a realization that increased productivity can be accomplished through automation, and increasing the productivity of library staff helps to reduce the rate of rise of library operating costs. So the chances of coupling the technical feasibility of library automation programs with the administrative feasibility of sharing and cooperating in establishing and operating those programs look much better than they used to. In our own case, the federal library community, we are further en- couraged by the fact that the GAO recently reviewed federal library opera- tions in the metropolitan Washington, D.C. area and made several helpful recommendations. One, for example, dealt with the need for strengthening OMB's role in the central management of libraries, in the sense of their developing efficient coordinating mechanisms to expand interagency coopera- tion and promoting improved plans of administrative management. Another suggested that OMB follow up on the work at OCLC for improving the cooperation of federal libraries in such activities as cataloging and control of periodicals. We feel, with the advances in technological capability available to us, the needs and desires of federal librarians for help in automating for more efficient operations so clearly expressed, and the encouragement on the part of GAO for cooperative programs by federal agencies, that the prospects for a Federal Library Cooperative Center are very bright indeed. REFERENCE 1. Contained in: Automation and the Federal Library Community. Re- port on a survey available from the Federal Library Committee, Library of Congress, Washington, D.C. 20540. GORDON H. WRIGHT Director, College Bibliocentre Ontario, Canada An Ontario Libraries' Network, or Cooperative Entanglement While I accepted the invitation to discuss the College Bibliocentre at this Clinic, I cannot say that I did so with equanimity. Quite apart from many organizational difficulties, the systems both in operation and in varying stages of development at the College Bibliocentre, have evolved from practical emersion without the benefit of the finite planning or initial test and research proce- dures from grant aids that many others have experienced. This is why I adopted the latter part of my title for this paper. I was asked particularly to discuss the techniques we are using to acquire the necessary input to the various systems. However, if I was asked to underline what I considered to be the major problems facing the development of a central technical service unit, the technicalities of how to input would be the least concern. The major problems are those beyond the technological requirements how to achieve the degree of coordination required and, in particular, how to overcome the financial hazards which face such an organiza- tion. Indeed, the prime motivation for establishing such a center as ours generally arises from financial restrictions imposed on educational establish- ments which, in turn, lead to less funds for libraries. However, in order to create such an organization and the systems it needs to implement, money is required. Meanwhile, those librarians who are already harassed by the depletion of their own budgets look to centralization of technical services for immediate cost savings. At the same time, the librarians impose on the center what they consider to be the significant system requirements, without due regard to their cost effectiveness to meet library user needs. Thus, the bibliographic descrip- 68 ONTARIO LIBRARIES ' NETWORK 69 tion of an item to distinguish it from any other item in the store has become the base of operation at international, national and local levels. Within an educational institution this is not the criteria of the majority of users of the library, because they are aware that the bibliographic description may define its physical identity but, as users, they are more concerned with intellectual identity. The result of this syndrome makes the operation of developing a central service, removed from direct contact with the ultimate user by a user librarian, rather like negotiating a greasy pole balanced on the snouts of two performing seals. Historically, the CB (Colleges of Applied Arts and Technology Biblio- graphical Centre) was developed when the Ontario government legislated the formation of twenty colleges in Ontario in 1967. It was created to purchase, catalog and process "packaged" libraries for each college, so that when they opened their doors in that year each would have a library even if it had no staff. Subsequently, the organization of this center, which began under the care of William Ready at McMaster University, was transferred to Toronto and established under a broadly based charter as a nonprofit-making organization. It is now partly financed by the Ontario Ministry of Colleges and Universities, with its policies approved by a users' advisory committee whose membership includes three college presidents, four directors of college learning resource centers, and four faculty members. The functions of the center are defined as follows: 1. To cooperate with publishers and other distributors in the establishment of a selective dissemination service for academic staff through college resource center directors to assist them in the selection of books, audiovisual and other resource materials. Such a service will incorporate a continuing central exhibition of print and audiovisual resources. 2. To facilitate the coordination of activities associated with the produc- tion of audiovisual aids in the colleges, paying particular attention to the suitability of programs for exchange purposes and the cataloging and retrieval of such material. 3. To negotiate with appropriate organizations on any matter which may facilitate the satisfactory application of usage of media to the benefit of the colleges. 4. To act as the central acquisition unit for the majority of the book and nonbook materials required by the college resource centers. 5. To act as the central processing and cataloging unit for all college resource book and nonbook material. 6. To establish systems which will facilitate the more efficient use of 70 1973 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING knowledge acquired by the colleges, so that others can retrieve informa- tion they need for their studies and problemsolving. 7. To establish in consultation with the colleges, central collections of more expensive material or material required for short periods of time which can be loaned as and when required. 8. To provide a professional center for the exchange of information and expertize in the development, administration and utilization of college resource centers. 9. To facilitate the use of any special reprographic or distribution facilities necessary for the center to the advantage of the colleges, with particular attention to microrecording. 10. To act in consultation with the appropriate staff at the Ministry of Colleges and Universities as a publisher/distributor of college-produced learning packages establishing copyright procedures wherever necessary. It may be of interest to specify the objectives which were used in compiling the 1973/74 budget. 1. All colleges participating in the operation will receive a computer tape, regularly updated, of their holdings cataloged by the College Biblio- centre for use in automated circulation systems and for general statisti- cal purposes. 2. A biannual book catalog with weekly cumulations will be produced for each college audiovisual collection cataloged by the College Bibliocentre. 3. A biannual author index with weekly cumulations to the print resources of each college which have been cataloged and processed by the College Bibliocentre. 4. A biannual title index with weekly cumulations to the print resources of each college which have been cataloged by the College Bibliocentre. 5. A biannual union catalog with weekly cumulations to material cataloged and processed by the center to meet college needs for Canadian studies. 6. In order to fulfill the requirement for book catalogs for those seeking guidance on resources to match subject requests, the College Bibli- ocentre will begin to implement a special automated indexing procedure utilizing the British PRECIS system. 7. To assist in the indexing of government documents and to provide college resource centers users with simple access tools, the College Bibliocentre will coordinate their records for government publications with the data base developed by the University of Guelph. 8. The Videotape Film Distribution Service, which was offering a limited ONTARIO LIBRARIES ' NETWORK 71 service for a few months during 1972/73, will offer a more extensive facility to faculty for a full year of operation. As these functions and objectives indicate, the center is responsible for creating a machine-readable data base for all materials required for individual learning in the colleges. As each college differs in the administrative organiza- tion of such materials, the center may be dealing directly with a number of people with different responsibilities in the same establishment librarians, audiovisual directors, educational resource directors to name but a few. Additionally, many colleges have several campuses widely dispersed and often autonomous in control and expenditure. There are now fifty-seven separate units involved, shortly to be doubled (again by the stroke of a pen) with the inclusion of all the schools of nursing which have been transferred to the same network. To provide some indication of the scale of activities in the year 1972-73 the center processed 160,133 items, of which 57,305 were new titles added to the college holdings file. Operating in 1969 with inadequate capital and premises more suited to the raising of mushrooms, the center was finally housed in a new building in January 1971, and has now reached a period of reasonable financial stability, though not financial adequacy. This year, 1973/74, for example, the govern- ment operating grant was slashed from $1,387,000 to $900,000. In order to obtain additional operating funds it has been necessary to charge $1.00 for all items ordered through the CB and $1.00 for all items cataloged and processed by the center. It is too early to judge the impact of these charges on our users, but they will undoubtedly alter the workload. It never costs the librarians anything to order or process within their own library. I stated earlier that systems grew more from practical necessity than planned efficiency although that should not be interpreted that we lacked an overall plan or, should I say, had no concept of the value of creative planning. Figure 1 (at the end of this article) is our goal: one input and a series of modules to meet each system requirement. Obviously, in order to achieve it, much more was required than our own system development. Simply stated, the chart points to: 1. Publishers and producers cooperating with national libraries (Why has Bowker produced yet another computer tape to U.S. resources repeating data appearing on LC MARC and adding nothing helpful to ultimate users of the bibliographic store?); 2. International coordination for machine-readable tape exchange (We pro- fess to the need, but the past year has shown how we beg to differ.); 3. National library networks with dedicated computer/s and computer 72 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING personnel wise in library record manipulation (more complex than com- mercial system requirements); and 4. Cooperative technical services created with common user profiles and interests. The CB did have control of the last: the potential in Ontario of a dedicated library computer at the University of Toronto. When we began, however, there was no indication of a Canadian National MARC Tape Service. In those early days the University of Toronto had problems of its own to resolve, so that it was even doubtful whether it could take on a major role for libraries in the province. Fortunately, the manager of the computer unit of the Ministry of Education agreed not only to assist with system development, but to accept the role of intermediary in the provincial network which we considered must emerge. At that time, while the major objective was still the production of a multimedia catalog to the resources of the colleges, it was deemed necessary to give greater priority to the development of an effective acquisition and accounting system. Apart from the fact that many colleges had inadequate bibliographic resources, and few, if any, qualified librarians, the complexity of the book trade in Canada, relying on imported titles from England and the United States, created considerable difficulty for college acquisition proce- dures. There was also a further complication; very few of the college require- ments were to be found on the available MARC tapes, so that the immediate use of these tapes for the order system was considered impracticable. The system which finally emerged has been documented, 1 but the following details may be helpful. Colleges may send their orders for any item, whatever the media, on a seven-part order form (figure 2). If the item has been purchased directly by the college, or received as a gift, the college may still send the item to the CB for processing and cataloging. If so, details are still input into the system for adequate control. College shelflist cards or a copy of the title page are acceptable alternatives although these are not input into the automated system. When the three top copies of the order form are received at the CB, copy 3 is immediately filed (by order number) to provide manual back-up. From the first two copies the information is verified, corrected or added to as necessary and then coded for the supplier. Copy 2 is sent to the supplier and copy 1 is used to keypunch the data into the system. At this stage the computer produces three punched cards coded with the order number. Origin- ally this was considered to be unnecessary, but it was discovered that most receiving update errors arose from incorrect recording of the order number. ONTARIO LIBRARIES' NETWORK 73 The three punched cards are filed with the order slip in an open-order file at the receiving department. If the item is clearly identified as one previously processed by the center and not likely to be a new edition when received, catalog cards, book card and spine label are produced and filed with the order slip and punched cards. Approximately 48 percent of the orders can be processed in this way. At the same time the computer compiles an order status report by college in author sequence, together with two indices by order number and by author. Copies of each college order status report are sent to each college fortnightly. These reports provide complete information on the status of any order in the system (figure 3). Initially the system is updated by either a status report from the supplier (figure 4), or the receipt of the item (figures 5 and 5a). If a supplier status report indicates the work is out of print, the record is held in the system for three weeks and then automatically deleted unless the college requests a search of the antiquarian market. If the order has been placed with a publisher or supplier who has no Canadian rights (a substantial problem in Canada), the order department searches for an appropriate supplier and issues a re-order update (figure 6) so that the system can then generate a reorder (figure 7). A separate subsystem maintains an agent/publisher file for this pur- pose. At present this file contains 3,500 addresses including North American and European suppliers. If neither item nor report is received within a specified period a follow-up request (figure 8) will enable the system to generate a follow-up chaser. In fact, the CB is also chased by means of a computer listing if the item has been received at the CB but has been held in a department too long (figure 9). The system is flexible in that: 1. Each order number represents an individual title, but volume or part numbers added to the order number can identify each volume or part uniquely. Thus, multivolume works may be received without prior knowledge that the original order was indeed a multivolume work, by the addition of the volume or part number on the receiving transaction card. 2. A series of books may be ordered on one order number. Individual titles in the series may be received separately and are assigned an "ID discriminator" or identification code. This code is one digit, and unique- ly identifies each title in the series, as the volume number uniquely identifies each volume in a multivolume work. 74 1973 CLINIC ON APPLICATIONS OF DA TA PROCESSING On receipt of an item which is part of the 48 percent already processed, the catalog cards, book card and spine label are matched; the book is jacketed and the item is ready for dispatch. As it is possible for this to be done within 48 hours, the system is capable of accepting a combined re- ceiving/shipping update. Of the 52 percent which have no catalog cards, once the items have been received they are checked once more against the union catalog. A further 16 percent are matched that is they have been processed by the CB on a previous occasion; catalog cards can be pulled from stock; spine labels and book cards produced; the item jacketed; and then prepared for shipping. Those items which have not previously entered the system meanwhile proceed to catalog for a search against the MARC records and against the various British, Canadian and LC catalogs. If a match is made, approximately 24 percent, the entry is photographed, verified, printed and the item proceeds through the rest of the system. The remaining 12 percent proceeds to original cataloging. When items reach the shipping department they are packed in a shipping box and a shipping update (figure 10) is entered. The system then generates invoices (figure 11). One invoice acts as the packing slip and one is sent separately to the college financing officer. This year it is hoped that the second invoice will not be necessary for reasons which will be apparent later. Whenever necessary, through faulty or incorrect billing, a credit note can be issued and invoice adjusted by a separate update (figure 12). In addition to the invoice, six colleges have availed themselves of an additional service. The CB can supply with each package an input to an automated circulation system in the form of tape records or punched cards. These records comprise data for author, title, accession number and LC call number. As the author and title are those input at the time of order input to the system it was considered probable that a large percentage would be incorrect. It certainly requires more care on the part of the colleges com- pleting the purchase order. However, the additional verification at the CB generally insures reasonable accuracy and very few records need to be cor- rected by the colleges. At present three formats are available as described in table 1. While the cataloging data base may eventually provide this data, at present the cost of providing it from the order system is so little that it may be the most economical technique. Thus for a box of fifty items, punched cards for all items in format 3 would be $0.23 and a tape in format 1 would be $2.20. Apart from these aspects of the system there are additional subsystems ONTARIO LIBRARIES' NETWORK 75 Position Field Length Description Format No. 1 1-7 7 Accession Number (Magnetic tape, 8-9 2 Copy Number 9 track, 800 b.p.i.) 10-39 30 Author 40-79 40 Title Total record length: 80-109 30 LC Call Number 1 1 2 characters 110-112 3 Media Code Format No. 2 1-7 7 Accession Number (Magnetic tape, 8-9 2 Copy Number 9 track, 1600 b.p.i.) 10-15 6 Filler 16-33 18 LC Call Number 34-40 7 Filler 41-100 60 Author 101-160 60 Title 161-164 4 Filler Total record length: 165-167 3 Media Code 171 characters 168-171 4 Filler Format No. 3 (Punched card) CARD 1 1-7 7 Accession Number 8-9 2 Copy Number 10 1 Card Number (always =1) 11-80 70 Author CARD 2 1-7 7 Accession Number (same as card 1) 8-9 2 Copy Number (same as card 1) 10 1 Card Number (always = 2) 11-80 70 Title CARD 3 1-7 7 Accession Number (same as cards 1 & 2) 8-9 2 Copy Number (same as cards 1 & 2) 10 1 Card Number (always = 3) 11-40 30 LC Call Number 41-42 2 Department Code 43-45 3 Media Code 46-80 35 Blank Table 1. Formats (There are always three punched cards for each record. We hope to standardize using formats 1 and 3.) for financial and management control. On receipt of the supplier's invoice a supplier reconciliation total card (figure 13) is completed and input into the system. A comparison is then made with the individual items received against that invoice number so that 76 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING discrepancies can be indicated in a supplier reconciliation summary (figure 14). Essentially the report pinpoints inaccuracies either at the CB or the supplier's end and insures that payment is not made for books listed on the invoice but not received. The budget subsystem is particularly important and could save addition- al costs to the colleges this year. Each college will now pay 10 percent of its anticipated expenditure for the year through the CB on the first day of each of the first ten months of the year. Its account will then be debitted as items are shipped to it, and the budget system will provide the necessary state- ments. The budget subsystem currently generates three reports: 1. The statement of commitment and expenditure (figure 15) which shows the funds committed (potential expenditures as represented by orders not yet received from the supplier nor shipped to the colleges) and actual expenditures (actual net costs including charges for binding, shipping, etc., for shipped items). This statement can be prepared for each department in a college, and is further broken down between print and nonprint materials. An average cost of print items for the campus is also given. 2. An analysis of item cost by subject (figure 16) showing the average cost of an item in a particular subject (LC class number groupings) and comparing it against the average cost for the previous six months. 3. An analysis of item cost by media (figure 17) which, again, illustrates the average cost of different nonbook media. A detailed media code has been devised not only for this purpose but as a technique for searching the catalog data base. Finally, a management information subsystem assesses the quality of supplier service to the CB. The report shows average time between order and receipt of items at the CB, the number of CB chasers generated and the level of discounts provided. Even though detailed costing of the operation is under constant study, it is not easy to isolate each system nor to provide data for accurate compari- son. There are no hidden costs at the CB as the unit is separately financed and all overheads, including computer costs, are distinguishable. However, the enormous variety in the colleges and their input costs preclude a factual costing of the total system from raising an order in the college to placing the item on a shelf. Nevertheless, there are some significant pointers. One college has never joined the system. Its library's operating costs are now three times those of any comparable college in the system. ONTARIO LIBRARIES' NETWORK However, before commenting on costs, a brief study of the main cataloging function might be appropriate. The CB has developed a machine- readable catalog of all the items it has processed. While LC MARC tagging was not utilized, the system was so devised that all significant elements could be compatible with the data appearing on MARC tapes to facilitate access in any network proposal. It should be realized that in the adoption of the system it was necessary to maintain the flow through the organization and the existing college card catalogs. However, the main objective would be the provision of more ade- quate user-orientated indices and retrieval facility. As a multimedia approach was necessary, it was decided that there could be only one format, but the media code would indicate the kind of data appearing within a field in the system. In other words, we did not choose to accept the LC pattern of creating separate MARC formats for each media. One other significant concern emerged. To meet the needs of users in an educational environment it was deemed necessary to consider more, rather than less, search codes. The CB is now applying the Intellectual Level Code recommended by the Canadian MARC Task Force, literacy and numeracy codes, course or program codes, a copyright code (for audiovisual material) and bilingual subject headings. In order to meet projected search and print requirements it was decided to adopt the British PRECIS system. This has only recently occurred and further development is still necessary. In order to create the machine-readable catalog three techniques were adopted: 1. A punch card is input with accession number and LC MARC number for all items located at the time the item is searched against the LC catalog data. This LC MARC record is then input into the CB data base together with any codes or additional data required. 2. Data for items originally cataloged are prepared on Flexowriters. A subsidiary tape is created from which catalog cards are prepared and printed. The master tape containing fully coded data is converted to magnetic tape and input into the CB catalog data base. 3. Each catalog card for those items forming the retrospective file is coded and then typed using an OCR font, optically scanned and so input into the data base. Pre-1968 items for which an LC catalog entry is photographed to prepare card sets, are also input into the system in this way. The data so obtained is now in the process of being structured and formated for subsequent manipulation. 78 1973 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING To facilitate this system and the proposals for future developments the CB has entered into a contract with the University of Toronto. Eventually this will mean full participation in a system of shared cataloging with other users, but that is still under development. In the meantime, the University of Toronto operates on our behalf a new books listing which is a combination of records appearing on LC and BNB MARC tapes. These are arranged according to the major programs offered by the colleges (figure 18). A copy is sent to each college in loose-leaf format, so that only the appropriate sheet need be given to a member of the faculty. It is hoped shortly to incorporate the Canadian MARC records and, possibly, the French MARC records. There are two points of special interest: a list for Canadian studies uses a search on the LC area code a very simple search device; and the input of C.I. P. records has led to us placing orders with publishers prior to their announcement of publication, a fact which has caused some astonishment from publishers. However, there is one major disadvantage of this service expressed by the faculty. Too many items are not relevant to their requirements, not by subject but by level. It is infuriating that national MARC services still refuse to apply an intellectual level, so that it is impos- sible to distinguish kindergarten from school, or college from postgraduate studies. As most publishers use simple codes in their catalogs, it is no longer true that publishers will not agree to the use of such codes. The first major manipulation of the CB machine-readable catalog has been the production of the union catalog of films (that is, film, videotape and filmloops). This will be regularly updated and cumulated. There are three sequences: (1) the subject catalog arranged under the main college course programs, (figure 19); (2) a name/title index compiled by the computer (figure 20); and (3) a subject index compiled manually to PRECIS configura- tion but which eventually will be machine generated (figure 21). Perhaps a simple statement at this stage will suffice on costs. The total cost of administering the CB with all that I have described, excluding $54,000 required for creating the retrospective machine-readable catalog (approxi- mately 180,000 records), but including all the functions as outlined at the beginning of this paper, divided by 160,133 (the total number of items we processed for the year 1972-73), creates a unit cost of $5.60. While this is a very basic costing, it can be compared with the Universities of California and Saskatchewan, and demonstrates that automation need not waste money, but to the contrary, can save while improving services. It need not have cost us as much if we could have controlled the information we needed to input into the system more effectively. I do not think the user requires much of the bibliographic data we input and I think he would prefer more valid information on intellectual ONTARIO LIBRARIES' NETWORK 79 content. However, the technical library service is the servant of librarians so, while we must wait patiently for the real user to comment, we can still hopefully cater to some of his unvoiced needs by using codes which no one but we can see at input. Sooner or later librarians will understand that cost effectiveness means providing services for the patrons at a price they can afford. REFERENCE 1. College Bibliocentre Acquisition and Accounting System Description Manual. Toronto, College Bibliocentre, 1970, p. 18; and College Bibliocentre Acquisition and Accounting System Operating Manual. Toronto, College Bibli- ocentre, 1971, p. 50. 80 1973 CLINIC ON APPIJCA TIONS OF DA TA PROCESSING I I ^ 300 C U 10 rH 01 i k2 en - ' x en o> W XI Z -H V) W H en o C a a to 01* 2 2 01 fi 1 0) ^ - < t XI y & J -H . / \ O rH Jl) s fT. W 41 A C 2 "O 10 -H NN 0> y ^ ' 0> C o O o l-l U g J2 4J 5 41 ^^ Q.4J M 4-1 553 H 3 w a U U H aa u c a io o o (U 00 I 5 vH X U / \ D 1 1 II "o U 1 \ M O \ / ^ n o 41 U H 10 4J > -H en X) H -H XI M-- 4J 0) H U 4) C >, H 0X1 -H O U b H U 10 01 ! M 4J c OS4) VO4JC(OIO O T> (0 4) 01 3 CO 6 ] 01 B> 41 *l ^s S "g ^ 5 S \ ^f * o. O XI 4J-~ M -H H"4) " U W"^" kJ ^ ij 0) 01 O w ^ M 01 10 01 *~ a-o 8^ ONTARIO LIBRARIES' NETWORK 81 Computer then automatically creates a new record with the same information as the first volume input. Fig. 2. Seven-Part Purchase Order Form (To be Completed by the Purchasing College) First 3 parts sent to CB. Information verified; corrected or added to as necessary. Supplier code applied. Copy 1 to Keypunch Copy 2 to Supplier Copy 3 to File (back-up only) 82 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING Information keypunched: Order number (9 digits) Volume Part number Supplement indicator Author Title Series Title Department Code Supplier Code mandatory) (3 digits) optional J ( ) ) (2 digits) optional ) ( ) ) ( ) ) (1 digit ) optional ) (120 digits) optional (variable) ) ( ) (1 of 2 (variable ) (mandatory Forms key to system (2 variable) optional (2 variable Canadian (7 variable Foreign Country of Publication Date of Publication Edition Number of copies ordered Date of Order Currency Code Set price indicator Estimated Price Media Code "Copy-previous-data" Code (1 digit) (4 digits) (6 digits) (3 digits) optional optional mandatory (6 digits) not keypunched, automatically applied by computer on addi- tion to master file (1 digit) Can., U.S., Brit., French, German (1 digit) For volume sets (6 digits) (3 digits) (1 digit) mandatory ($10.00 per item automatically assumed by computer if not input.) for multivolume sets where information is same for all volumes-one card for each additional volume or part is punched with order no., vol. no., pt. no., etc. and this code Computer then automatically creates news record with the same information as the first volume input. ONTARIO LIBRARIES' NETWORK 83 s r: S 2 IU VI 2 2 2 s u S i 4 i HI 3 S s c Ml IU s $ S m . 3 is z u i - U V 1 - " u u S z z 5 t * C I 3 % 3 i * " u >- . u U . c % o o c i o III O IU S ! c " 1 . - 5 - o I s ! A ? ^ *K tf U - ? B m f*l C P 2 ' c " i ~ S r- 3 ~ " ? 3 S 2 : a - -o I o K - a i g - - g * O 4 * *M i o IU Z a < IU - U z y o ^ " 3 c o 2 . 5 S 5 1 I * . 5 a - i W a. O J z S ? z' S < 01 Sj o z M o a . 5 - g 3 o 5 J 5 s s - * " ? " * - T ! 5 " S 8 g s R S! Si R S g o O S! a N ? a. H x 0- 1 * $ 1 Jj VI i u < z 1 o u c * o O VI IU X u. $ * J V. - i VI i s M s .J 2 5 5 "* o K ! 8 z z o a ~ 5 u o E s 2 O m J S 8 s 8 0. s i 2 > - 1 IK K * i 1 u 5 t 3 J 3 S w V) 2 i i r X g .s _ vi yj 5 | . OV, O A ! f i Is i I S u i zv, IS or ll ft n i i II x u 1 || U- 1- z o ll *"* " J H- Z 1 .$ x K a -< VI - I u. . t- 0- VI 22 > o W VI O a - ft oi o SVI O IU K U - - ;s a Ot -1 (- Sv, z ss 5 " $28 U a -j at II J s? s $ !"- 3t VI > VI 5 Fig. 3. College Order Status Report 84 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING STATUS UPDATE 1 5 6 7 8 9 ORDER NO tt 11 15 13 14 15 16 VOL PT < SUP D 69 60 61 62 79 *4 STAT WD DAYS AVAIL :ARD UPDATE Fig. 4. Status Update Input Form Report from publisher indicating delay of item translated to status code and input to system. Although normally a punched card is used, the information in- put is as shown above. ONTARIO LIBRARIES' NETWORK 85 ORDER NUMBER 10 11 12 *t i3 14 '' , *' It 15 1 "i If VOL PART SUP to 17 18 19 20 21 22 29 24 25 SUPPLIER CODE M 1 27 28 29 30 31 32 CURR. JNIT PRICE 93 34 35 96 37 38 41 42 43 44 45 46 47 48 NO. COPIES REC'O. SUPPLIER'S INVOICE NO 49 50 51 52 53 54 55 56 57 58 1 1 59 60 61 62 63 BINDING JACKETING U.S.B.E. 96 37 38 , ) ) MAILING 39 *9 41 42 43 44 45 46 47 ACCESSION NO. 48 49 50 51 52 53 54 55 86 57 CLASSIFICATION NO. *- 58 59 60 61 62 63 64 65 68 67 * 68 69 70 71 72 73 74 78 76 77 J. 2 .RECEIVING 1 [SHIPPING 3 i 78 79 80 REF. CARD NO. UPDAT ' - On receipt of item this form is filled out as far as indicated by the asterisk. Order number MUST be present. Vol., pt. sup. is normally identical to prepunched cards, although additional vols. or parts may be added at this stage if nec- essary. Supplier code is only input if different from sup- plier to which order was mailed. Currency code is input if necessary and the computer will convert to Canadian cur- rency to bill ordering Col- lege. Conversion rates are in- put monthly. Unit price Two of these must be Net price present. Computer calculates Discout missing field. Number of copies received must be present. Supplier's invoice number or date must be present. Status code pre- sent if item is to be bound, or, in the case of A/V material, the item is going to original cataloging. Set price indicator if the set price is charged against this colume. All subsequent vol- umes on this order number will be invoiced as no charge. If the number of copies re- ceived is less than the number of copies ordered, the computer will create a new record for the outstanding portion and identify it with an "ID dis- criminator" code. New cards will be produced for this to be replaced in the open order file. Fig. 5. Receiving Update 86 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING 1 1 1 9 DRDER NC ). n SUP. 10 . . . 16 /OL. PART ID 17 ?6 76 78 TI TLI i 79 80 CARD NO. UPDATE Fig. 5a. Series Title Receiving Update If the item received is one in a series or standing order, a second card will be punched with the specific title. R E Q R D F R A T F 1 9 order number vol . pt. sup H 1* 76 Ifo ID supplier code publisher's name (if different from supplier) -*- stat. no. days avail. Fig. 6. Reorder Update ONTARIO LIBRARIES' NETWORK 87 r OTY AUTHOR "'" " OAU TITLE ISBN Vl " PUBI (iwf lm . "" AGCNT/FUtUSMER - .AC, 1C CARD NO VARIATIONS MO CAHO . ^> COLLEGE BIBLIOCENTRE 20 HAILIIDf HO . DON MILLS. ONT . CANADA Fig. 7. Reorder Form New order form printed by computer on receipt of reorder update. 88 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING This is automatically produced by computer on time basis of country of publi- cation and location of supplier. Supplier's name and address are printed by com- puter. Perforated form can be torn off, checked by publisher and mailed back to the CB since the address is printed on the reverse. FOLLOW UP ONLY THIS IS NOT AN ORDER PLEASE ADVISE STATUS OF OUR ORDER NO. XXXXXXXXX ORDERED ON 'M v > H C YY FOR'"? COPY(IES) OF WHICH THE BALANCE OUTSTANDING IS 9-; COPY(IES) ***NO. OF CHASER. 001*** AUTHOR: TITLE: I I 5. NOT OUR PUBLICATION I I 6. OTHER I I 1. TEMPORARILY OUT OF STOCK I I 2. NOT YET PUBLISHED I I 3. OUT OF PRINT I I 4. NO CANADIAN RIGHTS IF YOU HAVE REPORTED OR SHIPPED ITEM WITHIN PAST FEW DAYS IGNORE THIS NOTICE COLLEGE BIBLIOCENTRE 20 RAILSIDE ROAD DON MILLS, ONTARIO, CANADA TELEX: 06-219837 ATTENTION: ORDER SERVICES MANAGER PRINTED MATTER Fig. 8. Follow-up Request ONTARIO LIBRARIES' NETWORK 89 ^v r ; 2 a a 3 00 3 39 a a a 1 ! 3 a N O C a 9 9 o a a o a o O o o o O 5 fht o ^ **- o c 1 Z UJ O O z ^ e x z ui O o S 2 "> z 3 "* E X v4 o o & c ^ a u X U o a O 1 2 i g o - 5 1 1 z c z o i a a 1 if s a: \ s S UJ I/ (J u O < j ' u i * K 1 UJ z UJ ac ee o a i 5! u J - z z f 3 z ct ! U i S SI u. a i JO ^ I i I L a. at i 1 oo ^ z z o z o z 5 o UJ >J **" a < at a 3 UJ Z o < z ui i X 3 u. - a. TO H^ 2 | S O aj Z _l i z *^> z 2 > - - a. o ! i 3 | *! 5 s 5 09 I a > z y I < o -i a z : a UJ v r -i UJ z < ( _. S o o < 1 X a o a : o - o - - oc S - a X C a 3 Z oc c O T ; d - X " 9 UJ 5 $ i 8 o X e z a z i 2 r 3 LU O 3 X 3 3 3 X z < Z Z z < o o O i- - U o /1 oc o Z a at z K 3 _ 3 Z 4 Z O >- O 9 _j Z z a CO C 1 O ff> O * < < < < < < < ' O o O - O - ^ 1 5 1 >- 3 C 3 (* a. f- 3 2 | E e 9 X n r U, a " a a u. a. _ a. ^ a. O j o _ 3 -J _J O J j - O o a 3 O | i | 5 5. i S S S . ? o o 9- O ? ? 5 ~. 2 2 2 S " sis ?: 2 000 " I 3 O ~ a, _ O uj o o r 8 2 S 2 S" S 221 g 5 S 2 ? S S 2- 222 , ( t , Fig. 9. Reminder Report 90 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING STATUS NO DAYS AVAIL ,o3 '"?ET 7 18 """19"' 2"0~f 21" j" 22 j 23~T~2"4' I L J... L. , :._i L L.. NO COPIES SHIPPED CB INVOICE NO. ~| 2 25 '26127^ 28; _> 29 ,30 131, _ T2 J_ ! i BINDING 33 ! 34 (35 36 ! 37 I 38 39 i 40 ! i 43 1 44 45 46 47 2 ACCESSION NO. ~4B~ r ~49 ' 50~"j~5T ; 52~| 53 j I ___ i i J i I ! 55 ' CLASSIFICATION NO i 5 * 59 60 61 62 63 64 65 1 66 67 68 ' 69 i 70 ! 71 j 72 j 73 i 74 J 75 ' 76 [ 77 J_L_L_L ("RECEIVING SHIPPING REF.. CARD NO. UPDATE Fig. 10. Shipping Update Information is filled in for a shipping update below the asterisked line. Number of copies shipped mandatory CB Invoice Number mandatory ( 1 invoice number is applied to each Box) Binding charge - if item was bound Jacketting charge if item jacketted United States Book Exchange - if applicable Mailing - automatic for CP items automatic 1 (ty for ordered items Mailing is filled in only if the mailing charge should obviously be more or less than the automatic charge. Accession and copy numbers mandatory Classification number mandatory Reference if item is to be used as a reference book ONTARIO LIBRARIES' NETWORK 91 1 K ifi I J o S9 ;! PACE NO < M z JS Si si sJ =< 52 85 - * -" - Q < s ill U 5 g i ^ I " s 3 S m ADDITI 4. IINOING 1 IOOK CHA & " 5 *^ 3 z U. O J 3 4 1 i to s- i li a _S)j II P 8 8 j 11 Fig. 11. Invoice Computer Printed from Shipping Data 92 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING INVOICE ADJUSTMENT CARD 1 3~ pr- L i i * Coll/Camp Consec.# n Sup I.D. Dept . Media Memo # 25 Invoice FT Quant . r 30 +7- 31 | 37 Amount Date Print Comment 49 | I I I 1 B Author/Title 67 78 f'T Fig. 12. Invoice Adjustment Card Used if credit or debit memo is necessary to adjust the invoice. Same form is used for printout as invoice. Stamped "CREDIT NOTE" if a credit. ONTARIO LIBRARIES' NETWORK 93 SUPPLIER REC ONCILIATION TOTAL CARD HI 19 35 Can. Foreign Code ' i * ' 1 40 "j Supplier r s Inv oice nur !LLU Invoic nber ZI T !Ti i :* Mailing Chrgs. :e total 53 1 | ]56 1 i 1 1 1 *7 SC. 13-6/71 No. Items on Invoice Fig. 13. Supplier Reconciliation Total Card Status updates as items move through manual procedures are written on one of the prepunched cards, and subsequently keypunched to update the system. This is filled in at the time of receipt of invoice. Comparison is made with this and the computer calculation of individual items received against that invoice number. Discrepancies are then indicated in a regularly produced report. 94 1973 CLINIC ONAPPLICA TIONS OF DA TA PROCESSING EH W Z O D I-H 2 o 1 w w > H < Q & U U r3 inovoooino o o mo M < ^3*roromcsJo*^ CN ^ o'tin O EH > O tTiininooin^Tit ^r r\j OI-H c Z EH M rH (N oo n n rH ro inoinoomo o in mo t 1 H ^roromrsjoio^ o o^ o*in < W EH O o\ inooinfrH oo o OrH U rH oo < Q g H inoinoinmo o in mo D EH W co o ^J*oroino%o^c^ in c^ o^in O W c r \inrsioo*^ i ^'( i v> o Oi i U K rH (NrHCTlrHrHrHrH rO (N rHCM rH M W H 0, . g fl. o D z W M a, a O W ou H (NrH^rHrHrHrH ^J 1 (N rHCN j-H Cn O > z o z (N W TJ- ro vo O P*i oo ro oo ro r** ro 1** ro H W vo r- oo r~ o\ r* Q r^ O Q aiP^rHrHavVOI 1 rH rH C^O* (N Z D vorMOOOrHO'ltf) O rH rHU roc^^o^rooocA *H ro rooi 00 ^ 2 OrHfMinTrrHOO rH i-H rHfK W D ft< Ct Q- J EH O O UEH ro 05 S O U D O H U >H SM SMZ r- U M W EH DM EH CQ rH Z W U W W -c . m * * 9 *" 5 s V O M O O o o o o o r\t IT O o o o -* 0- O O V s c. VN <-> - .1 *" * *vt ^ g8 o 03 < UJ K 00 m o h- o o o O CM O M CO O UN O O 0- m O - r- o O c n UN O oo * ^ 9. " S N 1" o> o M " I ' c >- * IT ir o O 2 i o o * 3 S S DUO m Z * J in s - z o a o g 7 a. or or IV' O O 5 S X li U U' o -e -r o o o * m - o a z o N - P- - 7 * OH a M M Z z a. J 1M u. (J o J O 1 LU C I X U - I o z * ^ * M ae *- 1- < 7 z S3 S i< |-> 1 | Fig. 15. Statement of Commitment and Expenditure 96 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING ONTARIO LIBRARIES' NETWORK 97 o *-> r- 98 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING GENERAL ARTS AND SCIENCE CANADIAN STUDIES CS90.B85 1971 Moorhouse. Eric Gelling. A brief record of In* family of Benjamin Lionel Burden end Wllhelnlne Margaret Gelling. (Toronto. Author, 15* Vlmbleton Roed. Islington. Ont . 1971) 6* p. S3. 50 1. Burden family. C71-t5fO 7219*690 E75 .P75 Price. John A.. 1933-. U.S. end Canadian Indian periodicals. Minneapolis. Training Canter for Community Program*, University of Minnesota, 1971. ?1 I. 1. Indians of North Americe - Periodicals. 72611897 E78.C2 B38 1973 Betty. Beetrice (Stebbing). Forty-two yeers amongst the Indians end Eskimo. Boston, Gregg Press, 1973. p. (History of minority education) 1. Horden. John. Bp. of Moosone*. 1828-1893. 2. Indians of North Americe - Canada - Missions. 3. Northwest Territories. Can. 0839801963 73001811 E92 .152 The Indient assimilation, integral Ion or separation Scarborough. Ont.. Prentice-Hell of Canada (1972) vl. 2H8 p. 1. Indians of North Americe - Canada - Government relations - Addresses, assays, lecture*. 013*5695*7 C** 7200*878 E99.E7 R5*13 1973 Rink. Hinrich Johannes. 1619-1893. Teles end traditions of the Eskimo. Boston. Gregg Press, 1973. p. (History of minority education) 1. Eskimos. 0839817770 73001790 E183.8.C2 1*8 The Influence of the United Steles on Canadian development i eleven case studies. Durham. N.C., Duke University Press. 1972. xll, 269 p. 1. United Stetes - Relations (general) with Canada - Addresses, essays, lectures. 06223027*8 72061337 GENERAL ARTS AND SCIENCE CANADIAN STUDIES FSS1 .837 Berry. James P. The fete of the lakes. Grand Rapids, Beker Book House 119721 192 p. *l*.95 1. Greet Lakes - Description end I revel. 0801005676 720877** F1059.V32 M69 1970 Mover. William G. Bill Meyer's Waterloo County diary. Kitchener (Ont.) CHYM (c!970) 1*5 p. SI. 75 I pt*. I 1. Waterloo Co.. Ont. - History. C71-3398 72169628 F1059.5.V38 RH2 Reamen. George Elmore. 1889-1969. A history of Veughen Township. (Toronto, Printed et the University of Toronto Press. c!971) 3*6 p. (5.00 1. Veughen Township. Ont. - History. C** 72170923 F1060 .R8 1971 Russell. Ralph Clifford. The Carlton Tretl. IZd eo. rev.) Saskatoon ISask.) Preirie Book*. 1971. xv. 158 1 . Car I ton Tra i I . 0919306306 C" 72197H33 F1089.7.A1 N6 Morris, John M. Strangers entertained. [Vancouver. British Colirfcia Centennial '71 Committee. 1971) 25* p. S6.95 1. Minorities - British Columbia. 2. British Coluifcla - History. C*** 72170963 6B2S .55773 Snoo. EoVard Row. Ghosts, geles end gold. New York. Dodd. Meed (19721 xil. 263 p. 85.95 1. See stories. 0396066585 72003936 61189. S2 P* 1969 Pearson. R. E., coep. At le* of St. John'*, Newfoundland. (St. John's, Newfoundland. 1969) 1 v. 1. St. John'*. Newfoundland - Mepa. C*** 72650133 /HAP Fig. 18. New Books Listing ONTARIO LIBRARIES' NETWORK 99 PHYSICAL SCIENCES INDEX NO. 01086 (CONTINUED) Tells the ttory ol men in the lor(ronl of the scientific revolution thai extends the power ol man's brain and charts the world of In* future. Points out that man. today, lives In an environment he has created and that ha is trying to understand. 02*0598 15.1 KIM POLITICAL SCIENCE BERTRAND RUSSELL DISCUSSES POWER .Chantarn Films. England. 1959. Ralaasad by Coronal Film. 1961.11 min. sd. b t v. 16 im. Lord Russall spaaks ol thra* kinds of power i tha power of armias and polio forces, economic power, and propaganda power. 0230 25 15.1 CMROMOPHOBIA .Belgium Ministry ol National Education and Culture. Brussels. 196B. Ralaasad by Intarnational Film Bureau. 11 mln. sd. col. 16 on. Presents tha dilamna ol a s< militaristic dictatorship, < eventual victory ol tha fraa spirit. aught in a MR. CHAIRMAN .Cinellects. Released by Encyclopaedia Britannica Fi lms.1959. 13 Kin. id. b I v. 16 ". With guide. Shon in a series of Dramatic episodes, presented in animation, hov any organized group can talk things over and reach decisions efficiently and fairly. 0218128 28.1 TAKE 30' DOCTOR BENJAMIN SPOCK .C8C.1971. 26 min. sd. b I w. 16 an. Special produced for the television program. Take 30. Filmed at Centennial College. Dr. Spock spaaks about dissent and civil POLITICAL SCIENCE 01090 (CONTINUED! disobedience, and answers questions from the audience. 01091 WOMEN ON THE MARCH .National Film Board ol Canada. 1958. 59 min. sd. b S w. 16 im. Records the struggle ol women lor equal rights the) characterized the suffragette movement. 'ided into 2 parts, the first h the fight for the Iranchisei the 0219763. p 0219761. p ing with the slatu today. JFW8 01092 2 1/2 .National Film Board ol Canada. 1965. 8 min. sd. b t w. 16 im. A camera study ol intent child behavior. Shows two toddlers, aged two-and-a-ha 1 1 , at play without adult surveillance or interference. 01093 ADOLESCENT DEVELOPMENT SERIES. Emotional maturity .HcGrawHi 1 1 .1958. 20 min. sd. b t w. 16 an. Correlated with the book Adolescent development, by Elizabeth B. Hur lock .Dramet ized incidents are used to explain a high school boy's innature behavior. Shows some ol tha consequences of an adolescent's failure to channel his emotions Into positive actions and feelings. Illustrates how important it is for adults to understand the increased emotional tempo of adolescence in order to help young people develop emotional maturity. 0210273 17.1 BF72H.H82 0109* ASSEMBLY LINE .Armenberg School of Cotmunications and the Institute of Cooperative Reseerch. University ol Pennsylvania. Releesed by Brandon Films. 1961. Fig. 19. Sample of Union Catalog of Films (Section Arranged by College Program) 100 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING TEACHER AND TECHNOLOGY. f oMnxjn i c a I i on theory and the new educational mod in 00667 UA(.MER EDUCA1ION IN MOOFRN MATHEMATICS. Conn-pi of functions 01116 Nuntwr fields 01117 TEACHER EDUCATION SERIES. Rending development in the secondary school ... 00729 Tt Ar.HFH 1WAINING FILM SERIES. Bit 790 TEACHING FILM CUSTODIANS. Art director 012V5 TEACHING JOHNNY TO SVIM 01613 TEACHING MACHINES AND PROGRAMMED LEARNING 00731 TEACHING THE WAY THEY LEARNI REMEDIATION OF LEARNING DISABILITIES. Learning ser ies 006% TEACHING THE 3'S. 'S AND 5'S. Guiding behavior 00732 Setting the stage lor learning 00733 TEAM BUILDING. Go Merman effective organize! ion. 5 00093 TECHNIQUES IN CHEMISTRY SERIES. F i I ter ing 01536 Titrating with phenolphthalein 01533 Ulinci a burot to 0153t Weighing procedure 01535 TELHNICM S ( NON-VtHBAI PSYCHO! CK.ICAI TESTING. 00511 TECTONIC MOVEMENTS. Geography OltJH TEE PEE PRODUCTIONS. Ja?20o 01 191 TEENAGE REVOLUTION 01665 TELEVISION AND THE WORLD 00571* TELEVISION CAMERA TUBE 01812 TELEVISION PICTURE TUBE 01813 TELEVISION TRAINING FILMS. Video tape editing 00575 TELL ME IF ANYTHING EVER WAS DONE 00365 TELL-TALE HEART 0133H TENSILE AND COMPRESSIVE STRUCTURES. Engineering film 01881 TEST YOUR TASTE 01777 TESTIMONY AND COURTROOM DEMEANOUR 01018 TESTING LAB. HP. s tudy 13 0006t TEVYE 01335 TEXTURE AND LIGHT. Elements of arti texture series. 5 00772 TEXTURE TECHNIQUES. Elements of ar 1 1 texture series. I* 00771 THAT WAR IN KOREA 00989 THAT'S ME 01336 THEORY Of THE LEAD-ACID STORAGE BATTERY 01537 THEORY X ANII THEORY Y. WORK OF DOUGLAS MCGREGOR. PT It DESCRIPTION. Fig. 20. Name/Title Index to Film Catalog ONTARIO LIBRARIES' NETWORK 101 ENGLISH LANGUAGE Paragraphs. Composition Spel 1 ing Syntax Use of ords ENGLISH LITERATURE See tttu DRAMA IN ENGLISH FICTION IN ENGLISH HUMOUR IN ENGLISH POETRY IN ENGLISH PROSE IN ENGLISH ENTERTAINMENTS Sec .nil' CIRCUSES PERFORMING ARTS RODEOS SHIMMING DISPLAYS ENTOMOLOGY See a/40 INSECTS ENTREPENEURSHIP ENVIRONMENT See a/40 ATMOSPHERE LAND ENVIRONMENT Pollution 184 Pollution. Ethics Pollution by rtfuse - Humo1ou4 ( ltd (({< ENVIRONMENT. Canada Conservation. Civilian Conservation Corps ENVIRONMENT ENGINEERING See a/r AIR CONDITIONING PRESSURISATION ENVIRONMENT PLANNING See 4/40 TOWN PLANNING ENVIRONMENTAL FACTORS See a/40 ECOSYSTEMS ENVIRONMENTAL FACTORS Psychology EPIDEMIC ENCEPHALITIS. United States 578 557 570 573 EPIDEMIOLOGY. Food poi Staphylococcl -1851 635 EPIDEMIOLOGY. Diseases Staphylococcl 488 298 332 EPEOEMIOLOGY. Salmonel losi s . Livestock EPIDEMIOLOGY. Salnonel los Is . Medicine 331 EPILEPSY See a/4o CHILDREN KITH EPILEPSY EQUATIONS Differential equations. Numer Iteration. Newton method EQUILIBRIA. Chemical reactions EROSION. Stream beds ESKIMO VISUAL ARTS. Cape Dorset. Northwest Territories ESKIMOS Legends Social change ESTRUS See OESTRUS ETHICS Environment. Pollution ETHIOPIAN-ITALIAN WAR 1935-1934 See ITALO-ETHIOPIAN WAR 1935-1936 ETHNIC TYPES See a/4u RACIAL TYPES EUGLENA GRACILIS EUROPE Foreign relations. Munich Conference, 1931 Hundred Years' War. Battle of Aglncourt, MIS EUROPEAN ARTS noo-noo EUROPEAN CAMPAIGNS See 4/40 EASTERN EUROPEAN CAMPAIGNS WESTERN EUROPEAN CAMPAIGNS EUROPEAN CULTURE 1700-1100 c. 1710-1900 EUROPEAN VISUAL ARTS mt- nn UOO-1700 9' 1168 331 1167 1083 1428 1437 895 1138 924 987 945 947 951 955 Fig. 21. Subject Index to Film Catalog J. W. JOLLIFFE Keeper of Catalogues Bodleian Library Oxford, England Project LOG* Centralized Processing of Local Collections Project LOG arose from the convergence of two factors: a growing awareness in the great libraries of Great Britain of the potential utility of computers in dealing with their large-scale processing problems, and a very long-standing need for the provision of information about the collections of early books in the college libraries of the Universities of Oxford and Cam- bridge. These college libraries are variously cataloged, both in physical form of the catalog and in degree of competence with which the records have been created. Some libraries have published their catalogs, others have barely listed their holdings. The central libraries, the Bodleian at Oxford and the University Library at Cambridge, have no responsibility for the college libraries and no authority in matters concerning them. In both universities, however, there have been movements, over the past 300 years, to produce union catalogs of the collections in the universities as a whole. One way of bringing this about, an old one in fact, is by the notion of borrowed cataloging; i.e., by using some existing catalog as a basis for cataloging style and arrangement, adding locations of duplicate copies and records in the same style for works not listed in the base catalog. The novelty, in the case of Project LOC, lies in the use of a computer version of the base catalog and in data processing techniques for updating and expanding this version. The base catalog chosen was that of the British Museum. Three principal reasons led to this choice: its availability in published form, the known richness of the Museum's collection of early books, and the desire to bring 702 CENTRALIZED PROCESSING OF LOCAL COLLECTIONS 103 into a practical working relationship the Museum Library and the two next largest libraries in the country. There was a fourth agent in this: the Andrew W. Mellon Foundation, or the Old Dominion Foundation as it was known when the project began. This foundation had already sponsored the conference at Brasenose College in 1966, at which plans and experiences were exchanged and discussed by a group of librarians and library data processing experts from both sides of the Atlantic. In the following year, it sponsored the visit of Foster Palmer of Harvard University and Lawrence Buckland of Inforonics Inc. to the three major libraries to discuss individually with them their plans for library auto- mation. These two reported privately to the foundation and to the three libraries their reactions to and their views on the progress envisaged. The chief problems were two: a lack of experience in library data processing, and a lack of personnel able to be involved in it. For this reason a joint approach, a joint venture, was encouraged. An investigation into factors affecting the creation of the long-desired union catalog of early books was to be carried out. Freed by the generosity of the foundation from the con- sequences of waste of funds, it was possible to try out, on a fairly large scale, different methods of data collection and matching of records; the project was able to make mistakes and explore blind alleys without committing any further expenditure to the erroneous procedures it might adopt. The use of a computer to manipulate a union list, to generate sub- catalogs for the libraries whose collections were to be included, to provide indexes of various sorts to the list, and to provide statistical information about the chronological, topographical and linguistic characteristics of the works contained in the list, seemed inevitable. The main list itself might be constructed in other forms and by other means, but only with a computer file would it be possible to provide the other indexes and information easily and relatively cheaply. So the project was concerned with two main problems: how to collect and process the information about the collections and what methods to adopt to avoid repetitious cataloging of duplicates. Two physical methods of record- ing and two levels of completeness of recording were tried out. I was appointed director of the project, working with a committee composed of representatives of the British Museum, the Bodleian Library and the Cambridge University Library, together with the Harvard University Librarian, Douglas Bryant. The employment of staff to collect, punch and proofread the information about the college and other libraries devolved upon the two university libraries; in each of these, the day-to-day organization and supervision of work was the responsibility of a senior permanent member 1 04 1973 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING of staff. The concern at the British Museum was with the conversion of the Museum records into machine -processable form, the provision of rules and procedures for the local teams, the correction of the machine files, the specifications for the computer programs and the evaluation of results. The immediate problem was one of selection: upon what data should the project work? The one restriction in scope which seemed permanently useful was that of attempting to cover only books published before the nineteenth century. Three catalogs had shown that unique items were hidden in the college libraries, and the history of the growth both of the large libraries and of the college libraries suggested that the proportion of unique items in the college libraries would decline rapidly if the nineteenth century were to be included. Certain libraries might have been excluded, but this would have removed the element of surveying the whole field, and might have led to false impressions about the scale and nature of the problem of producing a full union list. Similarly, restriction by language or further restriction by date of publication would have introduced a clear bias into any conclusions, since the problems associated with books of a particular age or in a particular language would not have been confronted. The only way in which a representative section of the collections could be studied without the considerable problems of devising and administering correct samples in each library was by choosing within a segment of the one type of file which was common to all libraries the alphabetical author catalog. Whatever strictures might be applied to the quality of some of these catalogs, their very existence gave the project an opportunity to choose books which would represent quite faithfully the distribution and overlapping of the collections as a whole. A single letter was chosen; all headings beginning with letter O were to be examined and the shelfmarks of pre-1801 books were to be noted so that the books themselves could be used. Other letters which, on the face of it, would have supplied about the same number of books to be dealt with, would have introduced problems of bias towards or against particular languages or particu- lar cataloging problems. The letter had the advantage of including a voluminous classical author, Ovid, a fair amount of Greek and German, and even some Russian and Irish, although the proportion of Irish surnames in the final list was by no means as large as had been expected. Because of the diverse standards of the college catalogs, we had in mind to bypass them in a full-scale operation; ironically, we had to work through them in the pilot project. For the comparison of two methods of recording, this selective approach was considered undesirable. A round-the-shelves operation would need to be as cheap and as quick as possible. The methods to be compared were a photo- CENTRALIZED PROCESSING OF LOCAL COLLECTIONS 105 graphic one, recording the title page together with a form giving details of library, classmark and supplementary details of imprint and authorship taken from the colophon and elsewhere in the book, and a manual one, in which the required information was filled in by hand on a form. Two levels of detail in the form were also to be compared: was it possible to carry out matching against the base file with a lesser amount of information? For this part of the project, in which estimates of the cost of continuous working for the methods and levels of detail were to be obtained, two complete college library collections were used. The early books of Hertford College, Oxford were at that time on deposit in the stacks of the Bodleian Library, while the college library building was undergoing redecoration and repair. This collection, in better surroundings than any college library, was used for comparisons between the costs of creating records to two levels of detail. The other library, at Cambridge, was that of Peterhouse, where the main part of the collection is housed in a single room. This library was used for comparing the costs of preparing machine records for the books using microfilm as a recording medium and of the similar operation using hand-writ- ten forms. There were approximately 3,500 books in the Hertford collection and approximately 4,400 in the Peterhouse one. For the investigation into overlapping in the whole range of libraries, some 22,000 records were ulti- mately used: approximately 7,000 from the British Museum, 2,600 from the Bodleian and 2,300 from the Cambridge University Library. All the other libraries together yielded just over 10,000 records for the books. The relevant section of the British Museum catalog was read, and papertape records made of all pre-1801 records. A similar scan of the Cam- bridge University Library's working catalog was made; the relevant entries were photocopied and half were sent to the British Museum for punching. The Bodleian Library's pre-1920 catalog was just beginning to be converted to machine-processable form; the O entries were taken out of sequence, and the pre-1801 records were then punched out on papertape for LOC processing. This processing was carried out on the Cambridge University's Titan computer. The fact that the director and the processing facility were in different places had certain consequences for the course of the project; chiefly in the delay in turnaround for some stages, but on the positive side it meant that the programmer was left to her own devices in many respects and was able to work more quickly and more confidently than if she had had to explain and perhaps justify all the details of her decisions. Since she had been chosen for her knowledge of the machine and the multi-access system, besides her manifest ability as a programmer, it was better that she worked on her own without the necessity of educating the director in details of the system 106 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING which were, properly, none of his concern. A terminal was provided for her so that she could work at her home in a village four miles outside Cambridge. In deciding what information should be recorded, it was necessary to look ahead to the ways in which matching between newly input records and the base file was going to be carried out. By choosing the Cambridge computer, the project was committed to a batch mode operation and, in consequence, algorithms were required which would yield a high probability of identity of records produced in different circumstances, without human intervention. The information on which these algorithms would work had, therefore, to be recorded. However, since the matching was between or within two classes of records, those included in the base file and the library catalog records converted directly and those created by the project itself on an examination of the books, any matching between the classes was constrained to the set of information already available in the base file. Matching within the class of records produced by the project could take place on information chosen to make such matching easier. A further consideration in deciding upon the information to be recorded was the possible use of all this information in a catalog entry. Information could be recorded for the catalog entry which would not be used in matching and vice versa. One of the aims of the project was to attempt to determine the likely differences in cost between a single-pass method of recording (in which all the information that might be required would be acquired at the shelves, whether or not all of this would be used to establish matches) and a two-pass method (in which only the information necessary to establish a match would be recorded on the first occasion, while a return to the shelves would be necessary for works found not to match against the growing base file). The work with the Hertford College books showed that recording all the informa- tion by hand took roughly twice as long as recording a set of information for matching alone (date, title, author). This, however, gives only one factor in the equation for determining the superiority of a one-pass or two-pass method; the other is the proportion of duplication. With the O books, all required information was set down to avoid the practical necessity of a second pass, but matching was tried on subsets of this information. The "cataloging sheet" on which this information was written approached most nearly, in details of description, the completeness of current cataloging codes: there was provision for a library symbol, a classmark, a title (split if necessary into three parts), an author name, date, language, edition, number of volumes, place of publication, and a publisher's or printer's name. In addition, there was a fingerprint. The labor force for this recording CENTRALIZED PROCESSING OF LOCAL COLLECTIONS 107 operation was composed of "intelligent but untrained" people. In a low-cost operation, the information to be recorded was reduced to that which people untrained in library work might most easily learn to recognize; thus collation statements, format statements, indications of editors or translator were omitted. Most of this information was contained in the records in the base file-language is the exception among the descriptive items. Matching between these records needed to be based on these items of information. The shorter records contained only the date as it stood on the book, a short title, the author's name and the fingerprint. It might seem naive, given the known difficulties of establishing an "author heading," for "author's name" to be included in the set of informa- tion which "untrained" people were judged capable of recognizing. The project compensated for this by paying as little attention as possible to what they had recorded as "author" when attempting to match records. This meant that for the shorter records only date and title were available for matching: the fingerprint was a separate exercise to be described later. It was plain that a straightforward literal comparison of the strings of characters representing the titles would lead in the majority of cases to a match not being found. Moreover, if one catalog had included an epithet or a name at the beginning of the title, while another had omitted it, the two records might be widely separated in a file sorted by title. A comparison of each record with every other might have been tolerable in the project, but was unthinkable in a full-scale operation. Much thought was therefore devoted to the problem of organizing the file so that there was a good probability that two records that should be compared would be, while as many futile compari- sons as possible would be avoided. The method adopted was to create a "keyed-title" record. Various normalizing procedures were carried out on the title before keys were gener- ated; all letters were changed to upper case, punctuation and diacritics were discarded. Up to three keys were created for each title; all words shorter than four characters were discarded; the first remaining word was taken as one key; and of the rest those which sorted first and last alphabetically. The effect of this was that with titles of average length, there was a better than 0.5 probability that two differing versions of the same title will produce at least one identical key. This virtually solved the problem of file organization. The matching problem itself remained. In the event, four matching procedures were used: a comparison of titles, which was the one for which the file organization just mentioned was intended; a comparison of search codes constructed by algo- 108 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING rithm from the records; a comparison of fingerprints recorded from the books; and, as a necessary step both to an evaluation of the effectiveness of these procedures and to the preparation of a specimen union list, visual comparison of records and human evaluation of the information they contained. I will describe the first three methods and the way in which their relative effectiveness was determined. I have already mentioned the way in which records for title comparison were prepared by using single words as keys. This word was not the sole element in the key the other was the date, and this leads to a little complication. When recording directly from books, the Project LOG staff was required to set down the title page date as it stood, whether in Roman or Arabic numerals; it is possible to derive an Arabic numeral from a Roman, but not in all cases of imprint dates is it possible to do the reverse. However, all the three large libraries, like most other libraries, had already normalized these dates as Arabic in their catalogs. The project used three kinds of date in its machine records: a text date taken directly from a book, a catalog date as given in a converted catalog entry, and a search date derived from either of those two. The search date was used as the second element in the keys for the title comparisons, thus permitting comparison of catalog records and of records created during the project. When two keys were found to be the same, the titles were then examined. As before, words of fewer than four letters were ignored. Now, it is not possible simply to count the number of words to be found in both titles: "Articles concerning the surrender of Oxford" and "Discussions at Oxford concerning the 39 Articles" both contain "Oxford," "Articles" and "concern- ing," but in reverse order. The order of words in titles was thus an important element. One other effect had to be allowed for: the differing truncations which might result from different catalogers. "A petition . . . presented to the ... House of Commons" and "A petition humbly presented to the hon- ourable House of Commons" are probably renderings of the same title. If we strip them and number the longer words with their original positions, we have PETITION(2) PRESENTED(S) HOUSE(6) COMMONS(7) and PETITION(2) PRESENTED(4) HOUSE(8) COMMONS(IO) for the lists of common words. Both the number of words in each list compared with the number in its present title, and the span of the longest string of common words were taken into account to produce an index number which could range from just over zero to 1.00. This number was printed out together with the full titles for subsequent evaluation by eye. The second comparison method used search codes. These consisted of alphanumeric strings of fourteen characters, taken from the date, title, an CENTRALIZED PROCESSING OF LOCAL COLLECTIONS 109 edition statement, if any, the place of publication and the author's name. These codes were sorted into a single sequence and the sorted list was scanned for identical items. The procedure of creating and comparing them was much faster than for the title word method, and no complexities of file handling were presented. These two methods were used for comparing catalog records with each other and with the records created by the project for college books. The third method was used only for comparisons between college book records, since it used information not available in the catalogs. This information was a "finger- print"; i.e., three groups of six characters, each group taken from a different page of a book: the recto after the title page (or the first recto if there was no title page), the third recto after this and the fifth after that. The characters were the last two on each of three lines: the last on the page, two lines up and two lines up again. It had been thought possible that such a character string drawn, as it were, at random from the text might function as a unique identifier of the book. The three "pages" were kept distinct and matching took place on the fingerprint and the text of the imprint date. The fingerprint pages were rotated to bring each to the head of the key in turn, and all of these three keys were sorted. The numbers 1, 2 and 3 were added into the sort key to avoid spurious matches on different pages. The sorted list was scanned, and any adjacent items with at least one page of the fingerprint in common were printed out. A full match was one in which the fingerprints were identical throughout and the text dates were the same. In each case, of course, the sort item included an identifier for the full parent record so that verification of matches by visual comparison of records could be carried out. All three types of matching were performed on the files containing the two complete college libraries, Hertford and Peterhouse. Although the overlap between the collections was small, each library contained some duplicate copies which had been recorded separately. The advantage of using these small closed sets was that it was not difficult to establish a list of duplicate items against which the lists produced by the different matching methods could be set. The first two types of matching were also performed on the combined files containing O books, while fingerprint matching was carried out on the combined files of college O books. In the case of the Hertford College and Peterhouse comparisons, the three systems gave the results seen in table 1 . These figures show clearly that the keyed title method was unsatisfac- tory because of the large volume of spurious matches. In a full-scale operation each such match would need visual verification, and, in the majority of cases, no 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING Matching Method Matches Spurious Percentage Matched Search code Fingerprint Keyed title 40 164 138 2 3* 297 22.2 95.2t 76.6 Table 1. Results of the Three Matching Methods *In each case, the items were variant issues of the same edition. fThis is calculated to a base of 172, since in 8 cases, one copy had not had its fingerprint recorded. rejection. Introducing a cut-off point into the computed index number would result in the loss of some genuine matches, since, in both the spurious and the genuine matches, the value ranged between 0.01 and 1.00. The search code seems equally fallible; in a batch-mode operation, the matches must be as definite as possible. No variations can be admitted which might remove the obstructions to a match of similar search codes, for each portion of the bibliographical information which contributes a character or two to the search code records details in which differences are significant. Nevertheless, in view of superior results with similar codes at OCLC, a careful review of all book search code matches was carried out. In these, the percentage of matches made rises above 50, although spurious matches also rise to 2 percent. This is still not adequate for a full-scale operation. Partial matching in the fingerprint was another matter. Miscounting the pages or the absence of a leaf in a given copy can easily lead to discrepancies in the characters recorded for different pages. Again, since the fingerprint has no meaning, it is liable to be degraded in recording or punching, and signifi- cant differences, e.g., comma for full stop, can easily be overlooked at the proofreading stage. The figure of 95.2 percent is for matching on one or more pages; that for two or more pages is 74 percent; matching on all three pages was effective in 46 percent of cases. In the majority of cases where a page did not match, the difference lay in a single character and this could be attributed very often to misreadings of handwriting, to shift errors or to mispunching of adjacent keys. The fingerprint therefore seems to be the best basis on which to perform batch-mode matching. The particular form adopted in Project LOG could be changed, say, to four groups of four characters or even five groups of three, while partial matching could be performed either on the basis of all groups except one, or of all characters except, say, two and those not adjacent in the same group. Shortening both the fingerprint itself and its component CENTRALIZED PROCESSING OF LOCAL COLLECTIONS 111 groups should lead to higher accuracy in transmission through the various stages of recording and entering into the machine file, with a resulting improvement in matching efficiency. In September 1972, a conference was held at Brasenose College, Oxford, to review the results of Project LOG. A provisional version of the report was circulated to participants, who were invited from libraries in the United Kingdom, the United States and Canada. At this conference, various aspects of the project were summarized by members of the LOC executive committee, and full discussions were held. I think it fair to say that no essential point of what had been done, what had not been done, and what might be done was left unexplored. After the conference, many of the participants responded to a request to submit observations in writing. The LOC committee was quite clear about what had been done and why; in the more doubtful area of future progress, it was greatly helped by the reactions of those who viewed the project from outside. These centered on two main issues and on several lesser ones. The first main issue raised was that of the scope of a full-scale opera- tion. Should it be to produce a union catalog of pre-1801 books in the libraries of Oxford and Cambridge? Should it be only an eighteenth-century English union catalog as a preliminary to an eighteenth-century short title catalog*! Should it include early books in other British libraries? Or in the major American research libraries? My view is that it should be confined to the original problem area, the libraries of Oxford and Cambridge, but an essential element in planning and implementing this scheme is its extensibility. The procedures, record and file structures must be designed both to permit acceptance of data from sources outside the range of libraries to be dealt with and to permit the transmission of the file or subsets of the file to other libraries and institutions. The LOC Project fulfilled neither of these objectives for two reasons: the first and simplest was the extra degree of planning and programming that would be required; the other was that nothing as tentative as the pilot project should be permitted to encroach by example and avail- ability on the question of standards. The other main issue raised by the participants of the review conference was that of the mode of computing activity that had been used and might be used again that is to say, off-line batch mode processing on a large central computer mainly employed on other tasks. It was suggested that various factors such as staff training, methods of matching, and process control would be changed, and for the better, by interactive computing. With this view I have considerable sympathy, especially because of the continuing fall in cost both of minicomputers and of random access mass storage devices. Indeed, the 772 processing facility for a full-scale union catalog must, I think, be a dedicated minicomputer system for on-line data entry and correction and for the first stage of matching, with the main full files being maintained on a bureau machine for batch access. Another point which needs attention is the quality of the staff to be used and the nature of their training. Project LOG used people without library training and gave them a minimum of training on the job. This want of training arose from the short time period allowed for working to collect and encode the data. The former feature, using people without library training, was a matter of policy, and insofar as such people are in greater supply than those with library training, it should still be a feature of a full-scale operation. Not that there should be no one engaged who has library training; the editorial and administrative posts need qualified and experienced librarians to fill them. Finally, let us consider the size of the problem. At the beginning of the project there was no firm knowledge of the number of early books to be found in the two universities. Adams's catalog of foreign sixteenth-century books in Cambridge gave some evidence both as to overlapping and as to the distibution of copies between the central university library and the other libraries. However, what was true for the sixteenth century might be less so for the latter centuries. By comparing the distributions within centuries and in Oxford and Cambridge both separately and together, it was possible to narrow the limits of error in our extrapolations. The final estimate is of some 1,600,000 copies in the libraries of Oxford, Cambridge and the British Museum, representing some 780,00 distinct editions. Of these, some 160,000 editions are to be found only in the college and departmental libraries of Oxford and Cambridge. So large a figure justifies our concern to find the means of recording and disseminating information about these collections. G. E. RANDALL Manager, IBM Research Center Library Yorktown Heights, New York Interlibrary Cooperation in an Industrial Environment Interlibrary cooperation in an industrial environment can be described in one short statement-"It ain't easy!" It takes two to cooperate and many industrial libraries are as individual and lonely as a male chauvinist at a NOW convention. The average special library (and industrial libraries belong to this genre) has a minimal collection and is fortunate when it has a professional member on its staff. When a special librarian talks about cooperation within the hearing range of university, governmental or large public library librarians, it is usually heard as a discourse on a one-way avenue of access to the resources and services of the larger institutions. I can understand this reaction. I was in Oak Ridge when we AEC librarians raided the resources of the University of Tennessee so heavily that the librarian, Bill Jesse, almost broke off diplomatic relations with us. In the twenty-five years since then, I, as an industrial librarian, have been aware of the impact a special library or a group of special libraries can make on the resources of any neighboring big brother and have sought ways of minimizing the onslaught. One obvious way for industrial librarians to ease the burden on the nearest large library is to increase the use of the resources and services of other special libraries. I do not want to raise unduly the hopes of the unviersity librarians that the future cooperation among industrial librarians will completely eradicate our demands on them. No industrial library serving 300400 chemists, mathematicians, engineers or physicists could justify the acquisition of a complete set of the Philosophical Transactions of the Royal Society dating back to the seventeenth century. Even if there is a plethora of industrial libraries in a given geographic area there is no possibility that one of 114 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING them will acquire such an esoteric set. And for the occasional request for an eighteenth- or nineteenth-century article, it will still be necessary to tax the resources of the larger libraries. Cooperation among industrial libraries "ain't easy" as I announced earlier. When North American Aviation is competing with Lockheed for an air force contract, it is not likely that their librarians will join forces to construct a union list of their combined periodical holdings, even if they are in the same geographic area. There are, however, a number of industrial organizations with geographically separated departments, divisions or laboratories, each of which is given literature support by its own library. These multilibrary corporations include Bell Laboratories, General Electric, Exxon, General Motors, Shell and IBM. The potential for cooperation exists among the libraries of these corpor- ations. In some cases it is mandated; in other cases it is ignored. But even among those libraries where cooperation is encouraged, the benefits are sel- dom exploited much beyond a minimal level. In this presentation I shall review in some detail the cooperative activi- ties pursued by the IBM libraries. It will not be a pitch on "how we run our libraries good," but rather a considered review of the problems we have encountered (and these are legion), the penalties we have had to pay (and these are not inconsiderable) and the benefits we have experienced (and these are numerous). The IBM Library Environment In the United States, IBM has at least twenty-five different locations- each with its own library. These range from almost miniscule collections with 200 or 300 journal subscriptions and a thousand or so books to the library of the T. J. Watson Research Center which is large enough to be socially acceptable in any library community. These libraries provide the literature support required by the laboratories and manufacturing plants dispersed from Vermont to California, from New York to Florida. Administratively, these libraries report to a number of divisions with such titles as Advanced Systems Development Division, General Products Division, Office Products Division, Research, and Systems Development Division. IBM's policy not only permits a generous degree of divisional indepen- dence but encourages local autonomy. Given this condition, it is not surprising that each library is unique and different. What is unexpected is the extensive cooperation and coordination which does exist among the libraries. I believe this is attributable to a number of factors. First, there is the professional bond among librarians, which probably explains the tighter cohesiveness and better awareness of who is who and who is doing what that exists among the INTERLIBRAR Y COOPERA TION IN AN INDUSTRIAL ENVIRONMENT 1 15 librarians than among the other support and service groups such as the maintenance people, the purchasing agents or the accounting groups. The professional bond is stregthened by the dependence of each on his or her counterparts at other locations. To do our job locally, we must call frequently on our counterparts at San Jose, Boca Raton, Endicott, Burlington, Owego or White Plains. Our lines of communication are facilitated by access to the company telephone tie-line, to a librarians' newsletter and to occasional company library conferences. The Tools of Cooperation The areas of cooperation in an industrial environment are identical to those in the public or academic fields. The cooperative tools we use in IBM include a union list of serials and a centralized purchasing and cataloging service, both provided by the Research Center Library, and a corporate-wide current awareness and retrospective searching service provided by ITIRC. As a backdrop for the subsequent discussion on the problems, penalties and benefits of cooperation, let us look in some detail at these three tools. IBM UNION LIST OF PERIODICALS The IBM union list is a by-product of the Research Center Library's computer record of periodical holdings and subscriptions. Like several of our computer-based activities, it is derived from a punched card record which could be printed out on the 407 or manipulated by a 1401 computer. Before being housed in our Saarinen-designed showcase in upper West- Chester County, the Research Division was dispersed among four locations ranging from Poughkeepsie to Ossining. The library collection was split among these locations and a punched card record was constructed to show which libraries held which journals. Although the library was less than ten years old in 1960, the information requirements of the research staff had resulted in the acquisition of the largest collection in the corporation. With a minimal amount of work it was possible to incorporate the serial holdings of the upstate New York IBM libraries in the research records. Acceptance of the union list has warranted revision and republication over the succeeding twelve years. It presently includes the holdings of 3,400 titles by 30 libraries. The information in the union list includes complete title information, an abbreviated title card which carries the CODEN, country of publication, the language of publication and frequency of publication, cross reference cards, title history cards and the holdings record for each partici- 776 7 9 73 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING pating library. The computer record from which the union list is compiled also includes subscription information for the Research Center Library. This information is excluded from the union list. The record is on tape and is run on the 360/91 with a SNOBOL program. The various subprograms enable us to se- lect various types of information from the basic record with output as either punched cards or printed lists. We are able, for example, to print the titles and holdings of any participating library to assist it in updating its record. We also produce our Kardex check-in cards from the tapes and the list of titles to be renewed. CENTRALIZED TECHNICAL PROCESSING In 1970 the Research Center Library undertook centralized processing functions including purchasing, receiving, cataloging and preparation of the books for use for four other IBM libraries. The products supplied include traditional catalog cards; circulation cards; a weekly status report of all books on order, in process and shipped; a weekly new addition list; a shipping list; and an abbreviated computer-produced shelflist with an author index which is updated on a quarterly basis. A 360 model 30 is used to produce the processing record which is termed PIL (Processing Information List). We have used the program for over ten years. Included in the PIL record are a 30-character call number; a 17-character author; and a 21 -character title, vendor identification, purchase order number, ordering library, and the date a book goes into each phase of processing. The processing phases recorded in PIL include ordering, claimed, received, cataloging, and completion of processing. By-products of the PIL program are the spine label, book pocket label, circulation card, a daily status report, claim notices to vendors and a list of items overdue in each phase. Although the program was written to trace books through the processing activity, we also include photocopy orders, journal subscriptions, as well as back issues and reports in the record. ITIRC ITIRC provides a centralized computer-based current awareness service and a retrospective searching service for all subscribing members of the corporation. The data base includes all IBM internal reports, IBM patent disclosures, IBM operating and systems manuals, the NTIS tapes for external reports, and the Engineering Societies COMPENDEX for the open literature. Each subscriber describes his area of interest. From this a profile is con- structed which not only includes keywords defining his subject interest but names of authors and institutions who produce work of interest to the INTERLIBRAR Y COOPERA TION IN AN INDUSTRIAL ENVIRONMENT 111 subscriber. The program also enables the requester to exclude certain topics. New tapes are searched each week and the subscribers are sent a printout of the bibliographic information and abstracts of the new items which match their profiles. ITIRC will provide microfiche or hard copy of the IBM material and the NTIS reports; copies are not provided of the articles in the open literature. The service is offered as an adjunct to that provided by the IBM library responsible for meeting the information requirements of the user rather than a service in competition with the local library. Each notice suggests that the recipient contact his library for the articles of interest to him. Our experience with these three examples of library cooperation has made us acutely aware of a number of the problems and penalties resulting from cooperation. It also has shown us, however, opportunities for benefit. Problems A major problem is the financing of library cooperation. In rare in- stances the library service of a multidivisional corporation is a part of the corporate headquarters; more frequently it is an organizational part of the unit to which it provides the information service. The Bell Laboratory Library structure reports to a corporate officer; IBM's ITIRC is a corporate head- quarters function. The IBM Research Center Library, more traditionally, reports to local management as does each of the IBM location libraries. When the cooperative information service is part of the corporate head- quarters, the cost of the service is charged, at least at IBM, to the receiving organizational unit. Although there is justification for this procedure, it does provoke problems. Charging a user for a service provides a built-in control limiting the amount of the service requested to the recognized value of that service. During its introductory years, when ITIRC was funded from the corporate headquarters budget, it was used without regard for cost. When ITIRC became a cost center there was a lot of second thinking, and there was a change in the demands placed on it. Judging from the subsequent year-by- year decrease in unit costs of ITIRC service, the valid uses have increased. When the ITIRC costs were charged to the using locations, the financial crunch on these local library budgets provoked an emotional reaction from many of the IBM libraries and librarians. The new ITIRC charges, in many cases, were paid out of that library's budget. Some of us found that these costs amounted to a very appreciable proportion of our book and subscription budget while the services accounted for a relatively small fraction of the total demands on the library. We reacted emotionally and intuitively because we 7 18 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING lacked the information to make a judgment based on fact. We did not know our costs. If academic and public libraries have a dearth of cost information, the industrial library sector has an absolute void. Even worse and here we are in good company with our big brothers we have no concept of the value of information. The adverse reaction of individual librarians receiving centralized services and then being charged for them will continue until, and probably long after, good cost data are available and some methods are devised for realistically assessing the value of information services. When a cooperative service is provided by a library responsible for servicing a discrete audience, other problems of a financial or budgetary nature arise, even when a charge is made for the service. We have experienced these problems with both our union list and our centralized processing service. The IBM union list is a by-product of a computer-based periodicals record. We believe that the benefits we realize from the record of our subscriptions and holdings justify the computer application. The addition of the holdings records of other IBM libraries gives us access to a much larger resource which we tap on a daily basis. The other IBM libraries claim it is one of their more valuable tools in providing service to their clientele. Creating and maintaining the records for other IBM libraries does take personnel time. The printing of a hundred copies of the resulting union list requires both computer and printing services over and above that required to meet our local needs. These requirements have been met without additional personnel, equipment or space. The library processing center does require additional personnel, and more people mean more space. The library is reimbursed by the recipients of the service for the actual cost of the services. But this does not alleviate the budgetary problem from management's view- point. To put the problem in perspective, look at it from management's point of view. The IBM Research Center has a finite amount of space and corporate headquarters has given us a total head count which must not be exceeded. The mission of the center is to provide the research necessary for the manufacturing activities of the corporation. The library's objective is to provide the literature-information support required by the professional staff at the center. Any dilution of the resources or services to the personnel of the division by providing services outside the division minimizes the amount available for the division. For example, the periodicals clerk works an honest forty hours per week. The time she spends updating the holdings of the other libraries is time she is not able to spend building a want list. We do have a backlog in binding because, in part, there are a number of volumes with missing issues. INTERLIBRAR Y COOPERA TION IN AN INDUSTRIAL ENVIRONMENT 119 The acceptance of the Library Processing Center's activities indicates we could extend the services to other libraries in the corporation and let them underwrite the cost. To do so, however, would require additional personnel, and added staff members mean more floor space. If either space or added personnel were available they would be used to accommodate additional research projects for which there is an acknowledged need. Services in a research establishment are necessary to enable the research staff member to do his job. They are not built and supported, however, to increase the nonresearch activities of the division when there are other organi- zations in the corporation which are staffed and funded to provide inter- divisional services. Penalties I have limited the discussion of the problems of cooperative services to those of financing and staffing. There are other problems and some of them can be described as penalties which provoke problems. Among these are the loss of autonomy, the requirement to meet or accept standards, and the building of increased demands on a limited resource of collection and staffing. When library cooperation results in the transfer of a function to a centralized activity, the donor library loses some of its autonomy as well as a degree of control over the services it is expected to provide. Discuss technical processing with an industrial librarian and you get a dire discourse on trouble which sounds like a contemporary revision of the Book of Job: vendors' services have deteriorated; turn-around time for book orders has nosedived from an unacceptable three weeks to an impossible six weeks; and as for cataloging, forget it. The books bought frequently are not covered by LC copy; the catalog cards, when they are available, are inconsistent and frequent- ly in error. Cataloging costs are at an unprecedented high. You may weep in compassion, but do not offer to solve his problems by offering to do his ordering and cataloging for him in your technical processing center! You are immediately suspect as a scheming, monetary-minded knave threatening his autonomy, independence and ability to provide the level of service to which his clientele has become accustomed! The threat of a centralized service to local autonomy should not be underrated. Special librarians are a proud lot and like being in control of their functions and services. As long as we are the big frog we do not mind the size of the pond. Subject us to interaction with other big frogs and no matter how much our pond is enlarged, our ego and job satisfaction are diminished. Let someone else give us a union list of serials and his decision on title arrangement governs the order by which we arrange our periodicals. If we do 120 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING our own cataloging, we can understand and accept the vagaries of the resulting catalog record; insist that we accept external cataloging service and we become so critical of departures from LC practices that we sound like an applicant for a position as catalog reviser at the Library of Congress. Direct us to use an external, mechanized searching service and we complain that filters are being imposed between us and our clientele. Standardization is another threat to the local autonomy of the industrial librarian. We make a fetish of claiming that the information requirements of each library's clientele are unique and that the organization of the resources and services rendered must be tailored to meet these requirements. This claim is based more on emotional reaction than on an evaluation of the facts. Frederick Kilgour, at the Ohio College Library Center, is able to accommodate the local desires for cataloging differences, as we are at IBM. A centrally produced current awareness service, such as that provided by ITIRC, provokes problems. If the data base from which service is provided is larger than the collection of the host library, the service is going to build requests for items which are not in the collection. Give a library a fixed budget, engender requests which cannot be supplied from the collection supported by that budget and you promote ulcers in the responsible librarian. More specifically, ITIRC uses the Engineering Societies COMPENDEX tapes as a source for the current awareness service on the open literature. The editors of Engineering Index (the basis of COMPENDEX) review several thousand American and European periodicals, many of which can charitably be described as esoteric from the viewpoint of the librarian who can subscribe to only 300-400 domestic journals. In addition, Engineering Index faithfully reports the proceedings of engineering conferences held in Yugoslavia, Poland, East Germany and Latin America. Enough "hits" are found between these announcements and the interests of the clientele of the underprivileged library so that the photocopy bill from the Engineering Societies Library threatens the library budget with magnificent overspending. If one is the librarian in the situation, what does one do? Deny the user his request? Discontinue the announcement service? An easy answer is to get the budget changed to accommodate the additional demand. But as an industrial librarian with a couple of years experience let me suggest that it might be easier to clean the Augean stables than it is to get a divisional controller to augment the library budget allowance. Benefits Now that I have reviewed some of the problems and penalties of library cooperation in an industrial environment, let us get from under the cloud that INTERLIBRAR Y COOPERA TION IN AN INDUSTRIAL ENVIRONMENT 121 besets Al Capp's miserable character and consider some of the benefits which accrue special libraries which participate in cooperative activities. Many of these advantages or benefits are identical to those experienced by academic or public libraries. The most obvious benefit is the expanded literature resource which is available to the participating libraries. No library can afford to be completely self-sufficient and the industrial library, with its smaller collection, depends heavily on other libraries to fill many of its literature requests. The IBM union list of periodicals gives each of the thirty contributing libraries access to the 3,400 titles which are listed in it. The union list is used extensively by the IBM librarians. At the IBM Research Library Center we do more than 2,000 pages of photocopy each month for other librarians, and we assume that this is a fraction of the total photocopying workload resulting from the use of this tool. The union list does more than provide an in-company source of photo- copy service. During the past three or four years the IBM libraries experienced the same economic belt tightening that has aged library administrators of public and university libraries. Subscription lists were chopped and binding programs curtailed. The installation librarian used the records of the union list as a guide to which journals he could drop, i.e., which titles he could discontinue retaining and binding without seriously endangering the services he could give his clientele. Although we all work for the same company, each installation has its own area of activity and the library at each installation has built its collection to support the projects and interest of its personnel. Our occasional meetings, the visits to other locations, the tie-line telephone conversations have given each of us some comprehension of what the specialities of each of the libraries are. We look to the ASDD library for market forecasting, the corporate library for economics, the San Jose Research Library for organic chemistry. This informally developed area of cooperation is buttressed by the book holding record, produced by our library processing center. Nearly 100,000 volumes are recorded in this combined shelflist and the interlibrary loan traffic amongst the IBM libraries has been facilitated by its publication. In addition to encouraging interlibrary use of resources, the centralized processing activity solves some problems which are not experienced by univer- sity libraries. One of the characteristics of an industrial library is that it is small. A number of our libraries buy only 300400 books a year. This rate of acquisition does not justify, and the limited head count authorized the small libraries does not permit, the employment of a full-time cataloger. In these circumstances, the one professional dilutes his services by accepting the cata- 722 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING loging activity as an additional responsibility or by assigning the job to a clerical assistant. Cataloging is a specialized activity and is best done when it can be made the complete and full-time assignment of a professional cataloger. Catalogers are people like the rest of us. To make effective use of their time the rate of acquisition should be a little larger than the number they can easily catalog. There are many clerical aspects of the cataloging routine and in a small activity it is customary for the cataloger to search for cataloging copy and do the filing of catalog cards. It is only when the volume warrants it that clerical support can be provided. Centralization of cataloging provides the volume necessary to warrant the use of a professional cataloger and to provide the clerical support necessary to enable the cataloger to do only the professional aspects of the function. This specialization results in or should result in a higher quality of cataloging than could be achieved in a small industrial library. A centralized computer-based current announcement and retrospective searching service is a natural in an industrial environment. Providing this type of service includes the creation or purchase of the necessary data bases, access to a computer system and the necessary programming support to institute and maintain the service. This facility is beyond the capability of a single indus- trial library but it is an economic possibility for a consortium of libraries. The consortium can provide the necessary volume and the financing required. The services provided by a centralized SDI activity such as ITIRC supplement and complement the information service the installation library can provide. The value and benefits vary from one location to another. It is most valuable at an engineering installation where the personnel are housed in buildings at a distance from the library. In such an enviornment it is not convenient for the engineers to spend a part of each day visiting the library, scanning the recently received journal issues or browsing the stacks to enable serendipity to lead them to all the information they always wanted to have about vacuum deposition but did not have the time to locate. Neither is it part of their tradition to use library resources in this fashion. My experience with engineers is that you either put the literature on their desk, preferably with pertinent passages underscored, or they cannot be bothered to use it. For the personnel of such locations the current awareness service which provides citations to the contemporary literature which matches their interest profile builds library use. It is a necessary and viable method of keeping them current in their information resources. Even more important, it provides a filtering function which protects the engineer from exposure to more than he is able or willing to evaluate and assimilate. INTERLIBRAR Y COOPERA TION IN AN INDUSTRIAL ENVIRONMENT 123 Some time ago librarians believed it necessary to provide clientele with all of the information pertinent to their areas of interest. Some librarians are still trying to do this. In today's market there is a requirement for the librarian to protect his reader from the plethora of available information and limit it to the pertinent and appropriate. The computer in an ITIRC type service can assist in this. I have mentioned that there are at least twenty-five IBM locations in the U.S. which support their personnel with libraries. There are a great many more which, because of their size or by the nature of their activities, do not require or cannot economically justify a separate library. The IBM employees at these locations do have access to the ITIRC services and can be kept advised of the literature which meets their information requirements. Interlibrary cooperation in an industrial environment "ain't easy." There are a number of factors working against it. Parochial interests of local management and the desire for installation autonomy are limiting factors. The probabilities of success for interlibrary cooperation are enhanced when the service is provided from corporate headquarters rather than being offered on a voluntary, cooperative basis by the divisional or location libraries. Resistance to interdivisional cooperation is frequently based on emotional rather than factual considerations. What the entire library profession needs is a method of measuring the value of information. We pay lip service to the value, we believe in it, but we have not found a technique for assessing it. In the areas where we have tried cooperation we have only superficially exploited it. The IBM union list of periodicals promotes the use of the serials collection at each location for the benefit of the personnel at other locations. The advantages and savings we have realized are only a small fraction of that which we could obtain if we had a mandated centralized control of our periodical resources. A centralized storage collection of serials providing over- night photocopy service could reduce local installation storage and copying costs. There are problems and penalties of interlibrary cooperation in an industrial environment, but the benefits of voluntary cooperation, even though less than could be realized from a strong, mandated coordination, far out- weigh the problems and penalties. DONALD P. HAMMER Associate Director and JAMES S. SOKOLOSKI Systems Manager Library and Information Systems University of Massachusetts Library Amherst, Massachusetts The Massachusetts Central Library Processing Service In the world of libraries and their related institutions, the Massachusetts Central Library Processing Service is an unusual and unique organization. It is probably the only automated processing center that performs the full spec- trum of monographic technical processing routines from the production of selection materials through the ordering, fiscal accounting, cataloging, and bibliographic display functions. It provides to the participant libraries many products such as catalog cards, book labels, pockets, book catalogs, special listings, and financial statements. The service has been in actual production since July 1970, and its success has strongly aided the development and effectiveness of the libraries in the higher education system in Massachusetts. Before the project was inaugurated, the then twenty-seven libraries involved ranged in collection size from about 3,000 to 120,000 volumes, excluding the University of Massachusetts Library at Amherst. The average size of the libraries, again excluding the university, was 41,000 volumes. The student bodies of the institutions ranged from 250 to 5,500 students, the average being about 2,000. At this time, the university had a student body of 19,000 and a library of about 900,000 volumes. The need for a policy -setting body for the project was recognized early in the planning stages. To meet that need an advisory council called the Massachusetts Conference of Chief Librarians of Public Higher Education 124 MASSACHUSETTS CENTRAL LIBRAR Y PROCESSING SER VICE 125 Institutions (MCCLPHEI) was organized. This group not only deals with policy decisions concerning the processing service, but also deals with other cooperative efforts among the participant libraries. It is interested, for example, in establishing uniform job descriptions and classifications for aca- demic librarians in the state institutions, with the acquisition of audiovisual materials, and in establishing a formal consortium in order to obtain govern- ment funding for special projects. A venture like this processing center can only be successful if certain basic tenets are agreed upon and adhered to by all of the participants. Obviously, the members must cooperate in all things individually, as well as through their advisory body, but beyond that they must agree to accept a standard product and to coordinate their needs with the needs of the other members. The center was at first a "catch up" program operated on a basis of massive funding and rapid development. After the initial phase of growth was passed, the operation became less restricted and it was possible to allow more diversification. The basic principle of cooperation has been maintained, how- ever. For the center to be a success it was also necessary for its members to agree to some practical standards. A mass production activity such as this requires acceptance of standard products whenever available. To this end, the members agreed to accept LC classification even though some of the libraries were classified in Dewey. Further, they agreed on the use of the LC catalog card format. Also, they accepted common selection sources for the first two years. These were Melvin Voigt's Books for College Libraries list for the New Campus Libraries Program at the University of California 1 (from which the center's informal name of BCL derives), and, in the second and successive years, the MARC tapes. In addition, the participants had to agree to accept a quota system as the means of fund distribution. In short, the principles of cooperation, standardization, and mass production have been the secrets of success in this processing service. In order to establish and maintain an economical operation and to gain the advantage of such benefits as large discounts and inexpensive cataloging, it was necessary for the group to develop a planned program of acquisitions. The greatest economies are achieved by a volume operation. If an average of at least seven copies of a book ordered at the same time can be maintained, good discounts will be had and volume cataloging can be produced. If a system gets away from the coordinated acquisitions concept, the benefits of a volume operation are lost and costs rise rapidly. For example, in the first year of BCL, the average number of copies was 12 and the discount average was about 20 percent. By the middle of the fourth operational year the average 726 7975 CLINIC ON APPLICA TIONS OF DA TA PROCESSING number of copies had dropped to 1 .6 and discounts were down to 11.7 percent. The concept of planned coordinated acquisitions was no longer in force and the center was operating much like a single library as far as acquisitions were concerned. Operational and Administrative Problems The problems of the processing service have been on two levels: (1) the level of day-to-day operations and (2) on the level of administration. On the day-to-day status, the center has the usual problems of maintaining an even flow of work in order to avoid the wasteful costs of slack time, or conversely, the inefficiencies of backlogs. Unfortunately, the center is subject to the same vicissitudes of any library processing agency or department it cannot control the rate of arrival of either new orders or new books. It is, therefore, extremely difficult to maintain a constantly flowing pipeline. Another operational problem concerns the book selection cards that are provided biweekly from MARC tapes. These cards will be discussed in detail later. This tool has served well in the past as a common selection aid, but now that the participant libraries have progressed to a point of diversity they want something more than just MARC data. In addition, they no longer care to be inundated with cards every two weeks. A committee has been established to study the problem of selection aids and suggest other approaches to a standardized but broader procedure. On the administrative level the problems are perhaps more subtle, but just as difficult or even more difficult to solve. One of these problems concerns the use of the various services provided by the center. As is obvious, certain services, such as original cataloging, cost the processing center more to perform than others cataloging with LC copy, for instance. If a library, therefore, consistently orders books not included on the MARC tapes, it is receiving custom cataloging while another library that orders only materials from MARC data is using the less expensive cataloging. There is, therefore, an imbalance of values received and costs incurred. The question then arises as to how this unfair imbalance can be cor- rected so that no library can use only the most expensive services available. One solution to this problem, and one that the advisory group is considering, is that of allocating the processing money to the participants on the same percentage basis as the book allocations are made. This basis will be discussed later. To date, only the book funds have been allocated. Another high level problem is that of the lack of continuous funding. The processing center is funded by the state legislature on a year-to-year basis which, in addition, provides no funds specifically earmarked for salaries and MASSACHUSETTS CENTRAL LIBRAR Y PROCESSING SER VICE 127 wages. The employees, therefore, must be hired on an hourly basis and have no job security beyond the fiscal year at hand. If they are to be available for another year, they must be transferred onto the more stable university payroll at the juncture of fiscal years if BCL funds for the new year have not been appropriated on time. After the funds are appropriated, the hourly personnel are transferred back to the BCL payroll. In the beginning of the project, it was felt that this problem would lead to a great deal of personnel turnover on the clerical level. The method used to get around this problem was previously described in print as follows: "a careful study was made to divide the unavoidable manual jobs into assem- bly-line steps. The goal was for each staff member to become an instant expert in one task, to learn everything about his job in three to five days. Exceptions to any routine were turned over to the exception expert." 2 This procedure tended to keep the jobs more or less simple and personnel training to a minimum. One of the problems borne by any centralized service is the acceptance of its products or services by those for whom the products and services are intended. Libraries are notorious for their individuality. This independent attitude sometimes continues to exist even after a library joins a cooperative effort. While BCL has been fortunate in this regard, as by far most of the participants have accepted it with enthusiasm, a few have been lukewarm. These people feel that they would like to have the funds independent of the center and be able to act in "freedom." Some would like to use the funds for other than book acquisitions; i.e., to engage in special projects or acquire special materials. This diversity of interest has its value in that the advocates of change are always present, but it can defeat the purpose of a cooperative effort. As has been pointed out, "Altogether too often a library cooperative is thought of by its members as supplying each member with a service to further its own goals. Such service centers have enjoyed only limited successes. A truly cooperative center establishes goals that are not achievable by individual libraries." 3 Project History Because of the relatively limited funds available for the processing of the books, as well as the short time in which all of the material had to be handled, it was decided that the University of Massachusetts Library in Amherst would serve as the processing center. The university had readily available the necessary technical knowledge in both the library and the data processing areas, and it also had the hardware required to accomplish the proposed objectives. In addition, it had the largest library of any of the state 128 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING institutions and also had some automated library systems operational. The members of the library staff who were responsible for the design and development of the BCL system were Merle Boylan, director; James H. Kennedy, associate director; and James S. Sokoloski, systems manager. In late 1969, the Massachusetts state legislature appropriated $2 million to the project's sponsoring body, the Board of Higher Education, as the initial allocation for the program. The appropriation was divided among the institu- tions in proportion to the collection deficiencies of the various libraries. These deficiencies were determined by the use of the U.S. Office of Education's collection deficiency formula. To develop the system and process the books, $250,000 was also appropriated. By late spring of 1970, the BCL project had placed orders for all of the material to be purchased and processed with the initial $2 million allocation. During this time, a second appropriation of $2 million for books and $250,000 for processing was granted by the legislature. By the middle of the summer of 1971, fourteen months after the project had started processing material, approximately 486,000 volumes, representing the total allocation of $4 million, had been purchased for the then twenty-eight institutions of higher learning in Massachusetts. The total processing cost of $500,000 repre- sents not only the actual cost involved in processing the books, but also all of the computer system development costs and all related costs incurred in implementing and operating the system. The third year of the BCL project was funded in the fall of 1971 in the amount of $1.5 million for books and $175,000 for processing. At this time, there were two significant changes made in the acquisition policy of the project. For the first time, the participant libraries were allowed to order any books they chose not just those included on MARC tapes and they could order audiovisual materials in an amount up to 15% of their BCL budgets. Both of these changes were without doubt inevitable, but they were also responsible for rising costs and less efficiency because of the increase in cataloging effort. The 1972/73 budget year has been funded in the amount of $1,425,000 for books and $123,000 for processing. The selection source used for the first year of the program was the book list compiled by Melvin Voigt mentioned earlier. 1 The basic selection aid for the second and successive BCL years was MARC tapes. Production System When a MARC tape is received in the processing center, certain items of bibliographical data describing each title included on the tape are listed on a MASSACHUSETTS CENTRAL LIBRARY PROCESSING SERVICE 129 printout and a punched card corresponding to each book on the list is produced. Some of the entries that appear on the MARC tapes represent analytics, ephemeral pamphlets, and books that will again appear on later tapes. In order to eliminate these items from the book selection process, the lists are ex- amined and the corresponding punched cards representing such items are discarded. The remaining punched cards are segregated according to the vendor from whom the books will be ordered if selected by any of the institutions. These punched cards are later submitted to the computer to trigger the printing of the two-part selection cards which are sent biweekly to each of the institutions participating in the project. In figure 1, the righthand card is to be returned to the processing center if the institution desires to order that particular book. The left-hand card is retained by the institution as a record of items ordered. The dates in the lower right-hand corner show when the selection card was issued and when it should be returned to the processing center for ordering. her, John R. Mew pr.p< t i .. In Job .nrlchwent. Edited b7 John B. Maher. Pew York, Yen Ho* trend Relnhold Co. 1971 ] III, 22 p. llluw. 24 r.. (Frontier* In o.nt *er lee ) Include* blbllorphl. Order New perspective* 1 enrichment. Edited b Mew York, Van Nontr. [ 1971 ] ill, 226 p. lllue 7 \ b hn R. M elnhold 4 <-.. Include* blbllor hi RTUIN BY: 02/07/72 ASM. 31/4 Fig. 1. Two-part Selection Card 1 30 1973 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING When the selection cards are received at the processing center from a participating library, the identification number assigned to the institution and the LC card number are keypunched. After all requests from all institutions for a given batch of selections are keypunched, the LC card numbers on the punched cards are then matched by computer with the LC numbers in the data file, selections are posted to the file, order numbers are automatically assigned to each title selected, and the purchase order forms are produced. The purchase order is a two-part form of which the left half contains instructions to the vendor as well as the necessary bibliographical data to order the item. The right side of the order form is the packing slip which is intended to be returned to the processing center by the vendor along with the copies of the book ordered. At this time various edit checks are made on the data and some statistics are compiled. Among the statistics produced are the number of titles selected by each participant and the percentage that number represents of the titles on a particular weekly MARC tape. One of the most useful reports received from the system is the financial summary (figure 2). It enables the administrators to monitor the progress of the processing center in fiscal terms as well as in quantity of materials processed. This same information is also produced for individual institutions (figure 3) so that the progress of each library can be assessed. This report is generated after each batch of orders is placed to allow the center to review the financial status of each institution on a continuing basis. When the order forms are produced, catalog cards for all books ordered are also printed, if the data for the orders was obtained from the MARC tape file. These cards are printed on continuous card stock which must later be cut to the usual 3 by 5 inch card size for filing in card catalogs. The equipment used to trim the cards is a blade cutter that trims only the sides of the cards. The top and bottom edges are precut. A complete set of cards includes two main entry cards, a shelflist card, a title card, a series entry, if appropriate, and multiple subject cards. These cards are computer produced using the ALA print train at eight lines per inch. In addition to the production of catalog cards, the computer prints the spine labels using a special print train of large type font. The call number is printed on pressure sensitive labels affixed to a peel-away backing. Two labels are produced for each book, one to be placed on the spine of the book and the second to be attached to the book pocket or used in any way preferred by the receiving library. After the catalog cards and the spine labels have been produced, it is necessary to arrange them in some temporary storage until the books on order MASSACHUSETTS CENTRAL LIBRARY PROCESSING SERVICE 131 BCL FINANCIAL REPORT AS OF 03/23/73 SUMMARY TITLES ORDERED CANCELLED RECEIVED SHIPPED IN-PROCESS OUTSTANDING COPIES ORDERED CANCELLED RECElVtO SHIPPED IN-PROCESS OUTSTANDING VOLUMES ORDERED CANCELLED RECEIVED SHIPPED IN-PROCESS OUTSTANDING DOLLAR VALUE ORDERED CANCELLED RECEIVED SHIPPED IN-PROCESS OUTSTANDING EXPENDITURES INVOICES FREIGHT MISC. CHGS ADJUSTMENTS TOTAL ENCUMBERANCES ON ORDER IN PROCESS SHIPPED TOTAL BCL QUOTA EXPENDITURES ENCUMBERANCES QUOTA BALANCE MONOGRAPHS SERIALS TOTAL NON-PRINT 41,562 1,220 15,699 12,683 3,016 24,643 2,094 37 1,464 1.167 297 593 43,656 1,257 17,163 13,850 3,313 25.236 3.718 13 941 909 32 2,764 72,743 2,192 26,428 22,138 4,290 44,123 2,260 45 1,603 1.282 321 612 75,003 2,237 28,031 23.420 4,611 44,735 3.759 14 958 921 37 2,787 77,975 2,434 28,985 24,188 4,797 46,556 26,051 420 22,391 18,917 3,474 3,240 104,026 2.854 51,376 43,105 8,271 49,796 18,727 50 7,896 7.561 335 10,781 MONOGRAPHS SERIALS NON-PRINT TOTAL S778.713.82 19,806.97 298,601.26 257,944.23 40,657.03 460,305.59 $442,513.56 10,743.85 J21.455.61 273,118.96 48,336.65 110,314.10 $202,314.26 902.70 50,431.01 48,046.43 2,384.58 150,980.55 $1,423, 541.64 31,453.52 670,487.88 579,109.62 91,378.26 721,600.24 $288,588.18 0.00 0.00 49,395.00 337,983.18 $272,169.86 0.00 0.00 0.00 272,169.86 $48,273.99 0.00 0.00 0.00 48,273.99 $609,032.03 1,728.90 0.00 49,395.00 660.155.93 $453.510.07 tt.290.00 8,518.60 470, 318.67 $109,650.85 16,485.50 33,463.50 159,599.85 $149.827.71 12.00 3,297.86 153,137.57 $712.988.63 24, 787.50 45,279.96 783,056.09 I $356,250.00 )$1,425,000.00 337,983.18 272, 169.86 48,273.99 660,155.93 470,318.67 159,599.85 153,137.57 783,056.09 1 $154,838.00 ) $16.212.00- Fig. 2. Financial Summary 132 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING FINANCIAL REPORT AS OF 03/23/73 UNIV. OF MASS. /BOSTON MONOGRAPHS SERIALS TOTAL NON-PRINT TITLES ORDERED 3,162 27 3,189 CANCELLED 77 1 78 SHIPPED 1,275 21 1,296 OUTSTANDING 1,810 5 1,815 COPIES ORDERED 3,164 27 3,191 CANCELLED 77 1 78 SHIPPED 1.275 21 1.296 OUTSTANDING 1.812 5 1,817 VOLUMES ORDERED 3,451 272 3,723 CANCELLED 81 1 82 SHIPPED 1,416 241 1,657 OUTSTANDING 1,954 30 1,984 DOLLAR VALUE MONOGRAPHS SERIALS NON-PRINT TOTAL ORDERED S36.211.06 t9.674.26 SO. 00 S45.885.32 CANCELLED 853.40 149.25 0.00 1.002.65 SHIPPED 14,680.86 4.567.81 0.00 19,248.67 OUTSTANDING 20,676.80 4,957.20 0.00 25,634.00 EXPENDITURES INVOICES *15, 502.96 $4,992.41 SO. 00 S20.55S.37 FREIGHT 0.00 0.00 0.00 55.32 BCL BALANCE 12,703.00 0.00 0.00 12,703.00 TOTAL 28,265.96 4,992.41 0.00 33,313.69 BCL QUOTA 1 S14.040.00 ) S56.161.00 EXPENDITURES 28,265.96 4,992.41 0.00 33.313.69 ENCUMBERED 19,794.70 4,532.60 0.00 24.327.30 QUOTA BALANCE S14.040.00 ) SI. 479. 00- Fig. 3. Individual Library Financial Report MASSA CHUSETTS CENTRAL LIBRAR Y PROCESSING SER VICE 133 are received. The cards and labels for each book are inserted into a book pocket thereby making a packet of materials (figure 4) that will be filed by order number to await the arrival of the corresponding book. At the time the computer produces the book order and the catalog cards, it also generates a "receiving packet" of punched cards which when later read back to the computer serve various fiscal and other record control functions. A packet of these cards consists of a red striped card, a yellow striped card, and one or more plain cards. The red striped card, the "receiving master," is used to notify the computer that a particular book has been received. The yellow striped card is used for fund accounting, and the plain cards indicate which institutions have ordered a particular title. In figure 5, for order number 004-02 16-E, the first (or red striped) card indicates that five copies of the title were ordered from vendor number 5236. The remaining numbers on the card indicate the number of copies represented by the order as well as the number of volumes per copy. The second (or yellow striped) card contains the order number and the identification number of the vendor and the remainder of the (plain) cards contain the identification number of the library that ordered the book, the order number, and a short title. These "receiving cards" are filed by order number to await the arrival of their corresponding books. At a later time, when the books are received from the vendor, they are unpacked and all copies of a specific title are placed together. After the books are arranged in that manner, they are taken to tab files that contain the receiving packets. The packing slip is then removed from the book that is to be checked in, and the corresponding receiving packet of punched cards is pulled from the file. Each card in the packet is used to indicate that certain transactions have occurred concerning a given title. For example, the red striped card indicates the number of copies that were ordered and the number of volumes expected per copy. Additional fields in these cards are used to indicate the number of copies actually received and the actual number of volumes per each copy received. If all volumes of all copies are received as expected, the card in the packet is processed without change. If this is not the case, a new card is punched to indicate the number that was received and a second card is punched to indicate the items outstanding. This second card is returned to the tab file to await the arrival of the remainder of the order. The yellow striped card is attached to the invoice that was received with the book. If the invoice did not arrive with the book, the yellow striped card is held until the invoice does arrive. The function of these yellow cards will later be explained in detail. Each of the plain cards is placed inside a copy of 134 1973 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING MUNICIPAL (;i.'Vl !, - ! -- I ( \l I - ' ' ! ' 1 I ST ' by J Al p. xi-xvi, JS7762.A3B6H Fig. 4. Materials Packet MASSACHUSETTS CENTRAL LIBRAR Y PROCESSING SER VICE 135 THE CAVE orniRinwwBrnjRr VO x -(ie!lb-l 1 THL CftSL* nr roKii HAHJX ni HI i ii i ii i i 90^ '4-lm-L iHf. CAit Of FCIfcb HatltlX f'Ofcft 9U j (l(l4-M.-'1i--f 1TE t^st or Tt*ii"fwifax Ftftti /^ - (li I *u 1 lit t* OIMOC K o*c i ii i tet( nioioecnooo MMMiMMliMMMilit l i 11 i i 1 1 i 1111111111111111111 t I I i 1 1 1 1 i i i i i i i i i i i ' i i : 1 1 i i i i i i i i i i i i i i i i i ti t i i i 11)111)3)111)31)1111 1111111111111 111 ) 3 1 1 3 3 3 3 J 3 3 3 3 1 1 1 1 3 1 3 3 3 3 3 3 3 ) 3 3 3 > 3 J .' 1 1 J311J 44444 44> 4 44444 444444 4 144 444 44 4444 4 4 44 144 4 (4444 4< 4 4 444444 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 ilSIitSSiSSS S $ IJ SJJ55535iSSSS155555 5 S 5 1 S 5 i 5 i :. J 5 5 S 5 i 5 5 1 i ! b '. 5 5 5 5 : i 1 S S 5 5 ', r , tiiiiiitti -1111111111 ttiiiiitittiiitiiiiti c nieit K iiiiiftiicttiiitKdi >I?TIIIfl?Ilf?I777tlTIIIIIITTTIl??IT77TIIf777TIT)77}fMl1Ilfr7T)77?7T>77.;777] tltttttltlllttttttttl(ltltl*ll>IIIM1llltt1t>l tttl(t*ll llttllttllltliitll i * * *^^ ^* "*'i t l | "**l*******'lf li4VMC'Vr>>4't !1 t Fig. 5. Receiving Packet the book. There should be a plain card for each copy of the book received. If this is not the case, an error situation exists which must be investigated. As mentioned earlier, catalog cards and spine labels are usually produced when the order form is produced. These cards and labels are now united with their respective books and turned over to a cataloger for further processing. If no LC cataloging data is available on the MARC tapes for a book, no catalog cards will be available for that book when it arrives in the processing center. In such cases, the original bibliographical data used for ordering is forwarded with the book to the catalogers for further searching and perhaps original cataloging. The books for which cards have been produced are compared with the cards to determine if any differences exist or if any changes are necessary in 136 1973 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING the bibliographical data. Those books for which no bibliographical changes are necessary are forwarded to the shipping room along with their respective cards, labels, and pockets for final processing. Those books for which biblio- graphical changes must be made on their catalog cards require extra pro- cessing. They are, therefore, retained in the processing center for usually no more than an additional two to three days before they are forwarded to the shipping room for final processing. The process of changing bibliographical data is handled on computer printed edit sheets (figure 6). These edit sheets are used by the catalogers to indicate changes or additions to be made to the bibliographical data. Each change or addition results in a card or set of cards being keypunched. Each card contains the title, ID number, a code to indicate the type of change to be made, a code to indicate the bibliographic field to be changed, plus the changed data that are to appear in the specified field. The changed data are processed by the computer at night and, at this time, a listing is produced if any errors exist in the update cards. In addition, a set of statistics concerning that particular update run is produced. The statistics include, among other things, the number of master records read, the number of edit sheets produced, and the number of catalog card records prepared for printing. The fact that the required changes have actually been made by the computer can be verified if necessary by a second printed edit sheet. If the data are correct, the system can be signaled through a cathode ray tube to produce the required catalog cards. The computer runs necessary to produce a set of catalog cards are performed each night. After the catalog cards are produced, the books, cards, labels, and pockets are forwarded to the shipping room where they are placed in wooden bins according to their library destination prior to packing for shipment. When the books are packed for shipping, the plain cards that were inserted in the books when they were originally received from the vendor are pulled out of the books and placed behind a punched (green striped) "box card." These box cards are punched by the computer as a separate process and provide a sequential number for the shipping boxes. When they and the plain cards are read by the computer, a bill of lading is printed. The bill of lading provides such information as the order number for each book, a truncated title, the box number, and the shipping date. The receiving library can determine, therefore, that the books intended for a specific shipment were actually received by them. As was mentioned earlier, yellow striped cards are attached to the invoices as the latter are received. An additional card, a blue striped one that MASSACHUSETTS CENTRAL LIBRAR Y PROCESSING SER VICE 137 B C L EDIT SHEET CALL MUUUBR: OBD-MO : 006-S443-B BT VOL-ISS: 418-0049 701.2 95 >BT701.2.(7 1972< / 7 ' 197^ 96 > 7J1S6240< COBHBNT DATK 02/22/73 97 >233< OUOEB DATE 10/19/72 01 >ri H ht, John Stafford. < 04 >Mlnd t -n, and tbo pirlt;< 05 >n ' daprata rch for nin la i nt I lc t ul I , mr< >*tlcimm, and th occult, by J. Stafford wriht.< 06 >Crand Maplda, Mich., Zondorvan Pub. Mou*o< 08 >[ IH7J, c!968]< 09 >19O p. 18 cm. ( Zondorvan books >< 13 >Prlovily publlhod uador tltlo: What 1. aan?< 14 >Hlbl lo^raphy: p. 188. < 60 >Man I Thooloir K Fig. 6. Edit Sheet provides summary data to the computer, is keypunched from the invoice data. These data include such information as gross total of the invoice, net total, freight charges, etc. With each blue invoice summary card, a yellow striped card representing each title on the invoice is returned to the computer. The yellow cards have had additional data such as gross price, net price, number of copies received, discount, etc., keypunched into them. When all of these various cards are processed by the computer, fiscal and statistical data are produced. Among other things, invoice totals are cumulated and summary statistics are created for management purposes. A few of the statistical items of data produced for a given day include the number of books received from the vendors, the number of boxes of books shipped to the BCL participants, and the number of invoices processed that day. In addition, a summary of the price of materials acquired can be obtained, on demand, from these data. Among other items of information, the price per item, per copy, and the average discount can be retrieved. If, during the course of day-to-day operations, it becomes necessary to determine the status of any given orders, this information can be acquired by means of a CRT display terminal. By keying in the order number, a request is made to display an order. As a result of this inquiry, such information appears 138 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING on the display screen (figure 7) as the number of copies of a specific book that were ordered, received, invoiced, and shipped to BCL libraries. Invoice numbers and dates can also be retrieved, along with fiscal data such as estimated price, net price, etc. In addition to determining the current status of a certain order, it is also possible using CRT displays to change many of the data elements in the file. The system also produces many kinds of manage- ment data in the form of statistics and production information that help to control and refine the operation. Although MARC tapes have been the basic source of book selection for the participant libraries, other sources have been used. Such things as pub- lishers' and vendors' sales catalogs, special subject lists, and bibliographies specifically directed at college libraries, have been frequent sources for book acquisition. System Characteristics As has been mentioned, the Library and Information Systems Staff furnished a nucleus of skilled data processing and library professionals to develop the specifics of the project. They sought to automate the routine portions of each job, and fully to exploit new bibliographic resources just becoming available in machine-readable form in 1969. Eventually, a combina- tion of batch processing and on-line techniques was implemented, supported by the hardware and software of the University Administrative Data Process- ing Center, which in 1969 included IBM/360, Models 30 and 40. The System/360, Model 40 has since been replaced with a System/370, Model 145. This latter system, running under DOS is the backbone of the Administrative Data Processing Center on the Amherst campus. With 393K bytes of core available, the system has three partitions: one being used for teleprocessing and two being used for batch processing. In the two batch partitions, all input/output functions are performed by an IBM spooling package called POWER, which greatly improves the utilization of the periph- eral components of the system. For its on-line applications, the Administrative Data Processing Center utilizes a programming system called FASTER BASIC (Filing And Source data entry Techniques for Easier Retrieval), an IBM type III program. Together with the System/360, Model 30, one 2314 and two 2319 direct access storage facilities, 22 visual display terminals on the Amherst campus and 4 on the Boston campus, plus 7 magnetic tape drives, 3 line printers, and other periph- eral equipment, the university has an advanced data processing and communi- cations facility to serve the Amherst, Boston and Worcester campuses. MASSACHUSETTS CENTRAL LIBRARY PROCESSING SERVICE 139 OR5ER-NUH CO CR CS CI V-0 VKR y-I JOJ 882-3553-a 04 OT W 89 661 801 801 fc STO*T RTC&-OT LC-O-KJM NMC-Nff *9->132 00 00000 78-143243 33V-8734 IFL INVOICE INV-H INV-GR IHHET 1 *44788 841*72 89005.15 i OR>-T RECV-DT VENMR an 0*-HUCE W 821872 8fe1372 84282 1 68885.V 1 NOTES.* 82-1-eb132-92^ ri-l-ebl 32-6484 11-1-eb1b2-8b4 14-1-%152-874j Fig. 7. Display Screen Showing Order Information Most administrative functional areas of the Amherst and Boston cam- puses are connected by visual display terminals to the System/370, Model 145. Applications such as the university personnel records, payroll, student records, housing, grades, and class scheduling are already functioning in highly refined forms. By the time the processing service was established, the library had already implemented an on-line acquisitions and a precataloging system using six of the visual display terminals. With this staff and computing facilities, the design of the system to be used for the processing service was initiated. Because of the expected short life of the service, it was decided at the onset that only "off the shelf hardware and software would be utilized for any data processing techniques used. It was realized that taking this approach might sacrifice efficiency for availability, but realizing that time was of the essence this sacrifice had to be made. 140 1973 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING In spite of the wide range of functions being performed by the Adminis- trative Data Processing Center, a number of constraints are placed upon any user of the facility. First, although a teleprocessing capability was provided, it was not a highly sophisticated facility, and as such did not support the use of interactive display terminals. Furthermore, in considering the application of on-line techniques, the scarcity of file space was always a primary concern. Also, the personnel at the center did almost all of their programming in COBOL. Although other languages, such as BAL, PL/1 and RPG were provided and could be used, few, if any, personnel at the center were knowledgeable about the use of these languages. Since the library and information systems staff was required to rely on this personnel for any problems that might arise, the decision to utilize COBOL was obviously influenced by the center. In addition, the facilities available did not allow the library to fully investigate and explore various methods of data entry. As a result, it was decided that keypunches would perform all of the functions required. Al- though they are not considered by many to be efficient data entry devices, the equipment itself was readily available, and was in fact capable of perform- ing the required functions. Finally, and probably most important, the most significant constraint imposed by the Administrative Data Processing Center was that it was operating under the disk operating system, and as such did not allow for any on-line storage and access of variable length bibliographic records. The use of the indexed sequential access method meant that only fixed length records could be used for any on-line applications. Any one of these above constaints can significantly influence the man- ner in which a system can function. Taken as a whole, they played a very significant factor in the manner in which the processing service functions were designed, and the manner in which data processing techniques were utilized to perform them. Two basic types of file structures are used in performing the functions of the processing service. All bibliographic data, which by nature are variable in length, are retained on magnetic tape files. As has just been mentioned, one of the primary reasons for this is that DOS does not support variable length records on disk files. As a result, all bibliographic data used by the processing service, which consists of MARC as well as non-MARC data, are stored on magnetic tape. To date, the MARC file is segregated from the local file. The MARC records are retained in their communications format with the addition of a 50-character leader that contains information used for retrieval purposes. The local file is retained in order number sequence, while the MARC file is retained in LC card number sequence. Whenever a MARC record is used for bibliographic information, the first step taken is to convert the MARC record to the local format. The primary MASSACHUSETTS CENTRAL LIBRARY PROCESSING SER VICE 141 reason for doing this is two-fold: (1) it was deemed unnecessary to convert or process any MARC record until the data were actually requested; and (2) all programs to maintain the local bibliographic files and produce such products as catalog cards and miscellaneous listings were written to work with records in the local format. The local bibliographic record contains all information necessary to describe a given item. The general format for local bibliographic records is as follows: Fixed Length Data Variable Field Pointers Variable Length Data < j 1. Record Code 4. LC Call Number (25 char, max.) 2. Local ID Number 5. ISBN 3. LC Card Number 6. Misc. Local Data As with any variable length record, the beginning portion of this record contains certain fixed length information such as the local ID and LC card number. Following the fixed length data, a series of 99 pointers exist in the record which refer to all of the possible data elements that can exist. For example, field 01 would be used if there were an author main entry, while field 95 would be used to contain the complete LC call number. Also, there are groups of fields assigned for other types of bibliographic data, primarily the notes and tracings. As an example, notes may occupy fields 13 through 29, while the series tracings may occupy fields 90 through 92. Although this type of format differs considerably from the MARC format, it was established at a time when MARC was just coming into use. Furthermore, this type of format is easier to work with on a continuing basis. If a pointer value is non-zero, it indicates where the data for the field start in the variable length portion of the record. After going to this particular location in the record, the first two bytes of information indicate the actual length of the data. Although this may not be the most desirable format to be used for variable length data, it should be said that this is the format that some of the initial BCL records were in, and therefore played a determining factor in the format to be used for all local records. On a daily basis, records in the local file may be created or changed. For example, if there were an error in the author main entry, a punched card indicating a change for field 01 would be created and processed that night. This change to field 01 would be made to the record and an edit sheet would be produced so that the appropriate personnel could verify the correction. It 142 1973 CLINIC ON APPLICATIONS OF DA TA PROCESSING might also be pointed out that the tracings are carried in the record as they will appear as headings. Prior to printing catalog cards, this form of entry is expanded to reflect the format used for printing the tracings in the body of the card. In this manner, all tracings need only be carried in one field and, if updated, will produce both the proper headings and tracings, tracings in the body of the card to be produced. The two types of bibliographic records (i.e., MARC and local) main- tained by the processing service serve considerably different functions. The MARC records are primarily used to retrieve bibliographic data for either ordering or cataloging purposes. The local bibliographic record, on the other hand, is used to contain ordering data in addition to the bibliographic data. Once a MARC record is retrieved, it is transferred to the local file, where all of the necessary updating and processing programs can work with the record. In addition to producing bibliographic data for ordering and cataloging purposes, records in the local file perform a much more significant function. Since the data did not come from MARC, it must be verified and corrected before cards can be printed. This requires processing in a daily update run in order to complete the bibliographic entry. Each time a change is made to a bibliographic record, a new edit sheet is generated so that the change may be verified. Further, these records are used for such procedures as claims, cancel- lations, order notifications, as well as various line entry listings, and even for the production of book catalogs. In effect, although the MARC file is a primary source of bibliographic data, the manner in which the MARC data are utilized centers around the local bibliographic file and its contents. Since DOS does not support variable-length ISAM records, the only manner in which on-line activities could be performed was to utilize fixed length records. As a result, it was decided to maintain a processing file on-line; this consists primarily of bookkeeping or accounting type information for each item in process at any point in time. The processing file is structured as shown below. Primary Record Order Number Fixed Length Data Shipping Information x 1. 2. 3. 4. Order Information 5. Vendor Number 6. Receiving Information 7. LC Card Number 8. "~N Status Codes Report Codes Invoice Information Notes MASSACHUSETTS CENTRAL LIBRAR Y PROCESSING SER VICE 143 Secondary Record Order Number Shipping Information This record not only records such information as vendor number, LC card number, and invoice information, but it is also used to control the final disposi- tion of each volume of any given order. This is done primarily by the information that appears in the shipping area of the record. All data preceding the shipping information refers to all copies of all volumes of an order. Again, because of DOS ISAM limitations, it was necessary to develop a chain-type fixed length record to contain all of the information that might possibly exist for a given order. This is accomplished as shown by the primary and secon- dary records. If six or less copies of the title are ordered, only the primary record is needed, since the shipping portion of the primary record can accommodate six copies shipped. If, however, more than six copies of an item are ordered, provision has been made to allow for these extra copies to be recorded when shipped. This is accomplished by setting an indicator in the first, or primary, record to indicate that a secondary record exists. Since the primary record contains all of the fixed length data necessary for the order, the secondary record now has room to record shipping information for an additional twenty-three copies of an item. By utilizing this approach, it has been possible to accommodate essentially variable length data within fixed length records. Although this is hardly a novel approach, it is usually done by the system, and not by the programmer. The processing file is normally maintained via batch programs. Some of these programs are used to indicate shipping and receiving processing, and are run on a daily basis. Others, for example, the programs necessary to produce a new batch of orders, are run weekly. In either case, records on the processing file are updated or created, and all of the appropriate information is recorded for each record. The processing file itself is used primarily as an inquiry answering mechanism. Since rapid turnaround and response to any queries are extremely important to the processing service, it is desirable to determine the status of a given order as quickly as possible. In addition, it is extremely convenient to be able to alter any bookkeeping or control type information as soon as possible and while looking at all of the information, rather than batching this type of change. One final significant functional characteristic of the processing service is the manner in which bibliographic records are retrieved from either the local or the MARC bibliographic files. Since these records are stored on magnetic tape, on-line retrieval is of no consequence. Furthermore, since these files 144 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING take a relatively large amount of magnetic tape for storage, processing on a daily basis is also not feasible. As a result, a procedure was established by which all areas included in the processing service may search the available bibliographic data on a weekly basis using punched cards as an input request form. The data on this punched card are as follows: Identifying Information Search Information Misc. Information Output Codes Institution LC Card Number Vendor Copies ISBN Date of Request Type of Search Control Number Type of Output Since the bibliographic search request cards can be used to request records for ordering purposes as well as bibliographic or catalog card purposes, a variety of information can appear on the card, as is shown above. However, depending on the type of output requested and the type of search being performed, various data elements are entered on the punched card. For example, if the type of search being performed is based upon an ISBN request, and the output required is an order for the processing service, the proper codes would be entered in the "type of search" and "type of output" fields. This would then trigger the requirement to enter other information such as institution and number of copies, as well as the vendor. Although each search request does not result in the retrieval of a bibliographic record, entering all of the required information at the search stage provides rather close control over the entire search procedure. If the request is only for the production of an edit sheet, the information necessary for ordering could be omitted. The date of request and control number may be entered to provide this information on the edit sheet, as well as other listings which are produced as a result of the search. Although the final result of the search is to retrieve a full bibliographic record, the search itself is actually performed against an index. This index contains ISBN and LC card numbers for all items that are accessible on the MARC and local bibliographic files. The search first verifies the existence of an item by searching the index. If an item is in the index, the location of the complete bibliographic record is indicated, and it is only after all items for all areas have been processed against the index that the actual retrieval of the bibliographic data occurs. Although this search procedure does not provide on-line searching, it has proven extremely beneficial over previous procedures employed. Because MA SSA CHUSETTS CENTRA L LIBRA RY PR OCESSING SER VICE 145 of the flexibility of the data that may be entered on these search request cards, the various areas initiating searches are able to have their requests answered and listings produced for only their area, and are not required to extract what they require from what is done for the entire search. Undoubted- ly, one of the most significant benefits of the above procedure over previous ones is the amount of computer time saved. On occasion it has taken as little as 25 percent of the previous time needed to perform a search against the bibliographic files. Currently, as many as 9,000 search requests may be processed against a total bibliographic data base in excess of 440,000 records in approximately 40 minutes of computer time. This time includes not only the amount needed to perform the actual search, but also includes the production of all output required as a result of a normal search. In spite of the constraints placed upon the processing service by the facility it is utilizing, the processing service has demonstrated that it is possible to perform the necessary functions without extremely large machines or sophisticated systems. The entire range of processing service functions, centered around the existence and maintenance of the processing and biblio- graphic files discussed earlier, are performed in a straightforward and efficient manner. Furthermore, it is only by the search procedure discussed that an efficient and inexpensive approach towards retrieving data from the biblio- graphic files has been achieved. Improvement Efforts As is well known to anyone who has been involved in library networks or processing centers, it is very difficult to determine how the products and services provided by the network or center are being accepted by the recipi- ents. Are the results of the network or processing effort being used to advantage? In fact, are the results being used at all? What alterations are being made to the products? How necessary are the alterations? To what extent, if any, do these alterations negate the value of the network or processing center's work? How do the users rate the overall quality of the products and services? These are a few of the problems that all cooperative efforts face. In order to find the answers to such questions, a survey was carried out with the BCL participants. A questionnaire of 25 queries was distributed in an effort to determine the way in which the processing services output was being used, how the participants felt toward the processing center, what effect the center's operations were having on the participant libraries, and to elicit suggestions and ideas for improvement. We are not here concerned with the survey itself, and, therefore, we will discuss only those results that are of more than local interest. 146 1973 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING Out of the 28 libraries queried, 25 returned questionnaires. Nine were community colleges, 12 were state colleges, and the remaining 4 were 1 of each, a medical, art, maritime, and technical school. It was pleasing to determine that all 25 were using the BCL computer-produced catalog cards, and that all but one rated the quality of the cataloging as good, very good or excellent. Most of them (17) proved consistent by stating that they made alterations to 5 percent or less of the cards received. Of the remainder, one made changes to 55 percent of the cards, another made 50 percent, two made 30 percent changes, and the others were in the 6-15 percent bracket. These changes were primarily to the added entries, subject headings or call numbers. All but 6 of the libraries use the spine labels provided, and those 6 prefer Se-Lin labels. Among the more significant questions, two were asked that it was felt would indicate the degree of success the BCL effort has had in the area of personnel savings. These questions were as follows: 1. What impact has the acquisition of materials through BCL had on the size of your processing staff? 2. If your library had received its share of the BCL funds without the processing center being established, how many additional personnel would you estimate would have had to be added to your present staff in order to handle the workload? It was felt by those designing the questionnnaire that if BCL had had no appreciable effect on processing staff sizes or if such staffs had to be largely increased in order to handle the BCL products, then the project was not successful in one of its major purposes. In addition to building good collec- tions in the participant libraries, BCL was expected to accomplish its goals with little or no increase in processing staffs in the libraries or, even better, with a decrease in staff. To the first question above, 17 libraries answered no effect on the size of their processing staffs, 2 claimed a decrease, and 6 stated that there had been an increase. Other questions and comments revealed that many of the staffs were so small that they could not decrease -hence, no effect. In the case of the six instances of increase, clerical level personnel was needed to cope with the increase of that type of work occasioned by the continuous flow of new materials from BCL. The second question was the really significant one. The answers to that question indicated that 19.5 new professionals and 48 new clerks would have had to be added to the staffs of the 25 answering libraries in order to handle MASSACHUSETTS CENTRAL LIBRAR Y PROCESSING SER VICE 147 the workload of processing the new materials if the processing center had not existed. This result in itself seems good justification for centralized processing. Unfortunately, it would require almost a scientific investigation to determine the actual dollar savings in a really credible form. In terms of generalization, however, the total annual salaries of 19.5 beginning professionals would amount to $171,600 and the total annual salaries of 48 medium level clerks would amount to $302,400. The total amount of processing funds provided by the legislature for all 29 libraries (not just the 25 anwering the question- naire) was $123,000 for 1972/73. Even on this very rough basis, it appears that $351,000 was saved in fiscal year 1972/73 in estimated personnel costs. The area of acquisitions, including selection, is fraught with problems in the individual library, and many of the same problems are extant in the centralized processing center as well. Two of those problems that the ques- tionnaire brought out are the time lag between order and receipt and the lack of claiming. Of the former item, it seems that no time is ever fast enough. Of the latter, if the jobber or publisher does not respond to claims, it is the processing center that looks bad to the network participants. There is, how- ever, no library or processing center that cannot improve in these areas. Many of the survey respondents commented on the contribution BCL has made to developing their collections. Some felt that their research poten- tial had been greatly enhanced, others indicated that the project had enabled them to build good supportive collections. One commented that his institution was now "literally accreditable." In regard to how well BCL was meeting the acquisition needs of their libraries, twenty respondents rated the center as "satisfactory" or better. Three of the remaining five felt that they needed "more latitude in choice of materials." In all probability, these libraries would have liked to acquire more nonprint materials than the agreed upon 15 percent of their BCL budgets allowed. One of the more interesting questions in the survey asked the partici- pants to describe the impact the BCL project has had on their libraries. Many of the statements indicated an appreciation for the greatly enlarged and improved collections that the project provided. In addition, however, a few commented on the "hardship" the great influx of new materials caused their small staffs. Most, however, were glad for the opportunity to build their collections and made comments such as, "created an entirely new library" or "dramatic improvement in quantity and quality." Interestingly, 18 libraries made special comments indicating their satisfaction and belief in the coopera- tive effort. Many of them stated they felt BCL demonstrated that cooperative efforts worked, and expressed a wish to develop other such ventures. 148 1973 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING Finally, to the request, "Briefly list any changes you would like to see made in the BCL project," 16 participants mentioned the selection process. The biweekly distribution of selection cards produced from the MARC tapes is no longer held in great favor by some of the participants. Those librarians made it clear that they want alternate tools for selection and more qualitative information about books than MARC provides. Since MARC is the only ongoing source for machine-readable bibliographical data, this request will be a very difficult one to satisfy. Costs and Statistics Due to the fact that BCL-73, the present year, has not yet ended and, therefore, there is no way to know what the statistics will be for this year, the following costs and statistics are for the 1969 through 1972 fiscal years, unless otherwise specifically indicated. They are generalized estimates rather than precise cost figures because of the continual improvements made in both the manual and the machine systems during the course of the year. Further, it is impossible to determine exact machine costs for the year because all of the equipment used for BCL is also used for other purposes. These factors, therefore, make it impossible to arrive at precise data. The figures in table 1 should be regarded as informed estimates. In spite of the changes in the economics of the operation that have evolved because of the loss of coordinated acquisitions and the increase in cataloging effort (which are looked upon as natural outgrowths of a maturing library processing network), we believe that the center has established itself as one of the most successful users of library data processing techniques in the country and had proven that a well-designed automated system is the key to modern, efficient, rapid and inexpensive processing of library materials. REFERENCES 1. Voigt, Melvin, J., and Treyz, Joseph H. Books for College Libraries; A Selected List of Approximately 53,400 Titles Based on the Initial Selection Made for the University of California's New Campuses Program. . . . Chicago, ALA, 1967. 2. International Business Machines Corp. Massachusetts Central Library Processing Service. Application Brief GK20-0647, White Plains, N.Y., IBM Corp., 1973. 3. Kilgour, Frederick G. "Computer-Based Systems, A New Dimension to Library Cooperation, College & Research Libraries, 34: 137, March 1973. MASSACHUSETTS CENTRAL LIBRARY PROCESSING SERVICE 149 Fiscal Year Coverage Book Funds Processing Funds 69/70 BCL-70 Titles from the bibliography Books for College Libraries (Chicago, ALA, 1967) $2,000,000 $250,000 70/71 BCL-71 Titles from MARC tapes, backfiles of periodicals, special bibliographies $2,000,000 $250,000 71/72 BCL-72 Titles from MARC tapes, some nonprint materials, any in print title from the world trade $1,500,000 $175,000 Total $5,500,000 $675,000 Volumes purchased, cataloged and distributed**: 650,000 Volumes purchased, cataloged and distributed**: (1969/72) Periodical backfiles purchased, cataloged and distributed: (1969/72) Catalog cards produced: (1969/72) 650,000 volumes 1,500 titles 3,200,000 cards In addition to the data above, the per volume costs for each completed year of BCL operation were as follows: BCL-70 $ .83 BCL-71 .90 BCL-72 with MARC data .97 without MARCdata 1.86 In order to present a record as up-to-date as possible, the following are data for the 1972/73 year. BCL-73 Book Funds Processing Funds $1,425,000 $123,000 Estimated volumes purchased, cataloged and distributed * * : 118 ,000 volumes Estimated Catalog cards pro- duced: 600,000 cards **Excluding microtext units, backfiles, and nonprint materials. GLYN T. EVANS Director for Library Services State University of New York Albany, New York Bibliographic Data Centers for New York State In the years which have elapsed since librarians began to use the computer to attack some of the problems they face, massive technological change has altered computer design and performance. Technical development has resulted in larger, faster machines, with storage capacities and processing speed factors higher than a decade ago, and a consequent dramatic reduction in the unit cost of storing and processing data. Simultaneously, teleprocessing has also developed rapidly, with programmable CRT terminals relatively com- monplace, where little more than a decade ago, in 1962, a computer manufac- turer announced the first linkage between a Telex and a computer. As a result, the concept of a large computer file of bibliographic data constantly maintained and usable at random by a number of users simultane- ously (or rather concurrently) for a variety of purposes (much like a card catalog in a large library) has become not only technically feasible but fiscally desirable. In this paper I want to describe the work being undertaken in New York State to realize this concept. Environment I will provide some sense of the size of the problem. Firstly, the state of New York is rich in library resources. One figure commonly quoted and certainly inaccurate but nonetheless an indicator-is that 10 percent of the bibliothecal wealth of the country is located in New York State. There are some 217 academic libraries in the state, including the great collections at Cornell, Columbia and New York University, for example. Seventy-two of the libraries are on campuses of the State University of New York; there are BIBLIOGRAPHIC DA TA CENTERS 151 nineteen campuses of the City University of New York, some jointly adminis- tered with state university; and the remainder are private institutions. There are extensive public library systems, including such major resources as New York Public Library, and Buffalo and Erie County Public Library. The State Library in Albany is a fine library; and there are good school library services. Finally, there are magnificent special and private libraries, including such organizations as the Research Libraries of New York Public Library, the Engineering Societies Library, and libraries of the large industrial concerns. I hope I will be forgiven if I speak of the libraries of the State University as a rich and diverse collection in themselves. They include the four university centers at Albany, Binghamton, Buffalo and Stony Brook; fourteen four-year colleges of arts and science; four medical schools; six specialized colleges including the agriculture, veterinary and industrial labor relations schools at Cornell; forestry and environmental science at Syracuse; ceramics, maritime and optometry colleges; and six agricultural and technical colleges. Finally, there are thirty-five community colleges administered jointly by the State University and the local communities. Secondly, there is already within New York State a strong tradition of library cooperation. The public libraries are grouped into twenty -one cooper- ating systems. The State Education Department Division of Library Develop- ment designed and supports nine regional reference and research councils (3Rs councils), including METRO in New York City, which completely cover the state and are a local mechanism which bring together all types of libraries for their mutual benefit. State University is itself grouped into four regions and the librarians in those regions are beginning to explore local modes for cooperation. Finally, there are other groupings, such as the Five Associated University Libraries (PAUL) which comprises Binghamton, Buffalo, Cornell, Syracuse and the University of Rochester, and has now added as affiliate members Albany and the State Library, and the Associated Colleges of the St. Lawrence Valley. Thirdly, the state has demonstrated considerable support for libraries, such support manifesting itself in the NYSILL interlibrary loan system, the work on the ANHLTS processing center project, and aid for public libraries. Bibliographic Data Centers Having described, albeit too briefly, the libraries in New York State, I should now concentrate on the latter half of my title and define bibliographic data centers. A bibliographic data center is a central computer system which provides bibliographic data, primarily to libraries but also to their users, and whose prime operational mode is by on-line interaction supported by batch 752 1973 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING services. The data stored and supplied will range from item location data through standard bibliographic data to index and abstract data. A biblio- graphic data center will not involve itself with handling and physically process- ing library materials. I prefer to visualize and discuss the role of such a center from the point of view of the librarian who will use the services offered by it. The librarian needs, close at hand, a range or first circle of files (see figure 1) to control circulation, acquisitions and in-process items, serials check-in, and probably union location lists of serials and monographs. These specifically job-oriented files are the volatile areas of library record keeping where there is the greatest change of state in the files, and where currency of information is vital. While these files need authoritative, standard data, they do not need a full standard bibliographic description. Instead, each record will be composed of a selection of data elements, the selection being based on the absolute necessity of the presence of each element in order to be able to perform each task. For example, a subject heading is not necessary in a circulation record which would be accessed by name, title, call number, etc. Just as the files are shown to be job-oriented, so will it be clear that the librarians must be able to work with each file and change it (and the records in it) by adding, changing or deleting data. This statement, obvious enough, raises interesting questions of file security. Each file is shown to be linked to all others since data will clearly move between some of the files at this level, although perhaps not among all the files. In the second circle of files (figure 2), the librarian requires access to autho- ritative bibliographic data primarily in order to describe the collection in hand, but also for verification and searching purposes. The data will clearly be MARC- formated data for monographs, serials, maps, and so on. In addition to relying on the Library of Congress to supply data, each library will accept the responsibility of contributing data to the common data base in a shared cataloging mode, as do members of the Ohio College Library Center. The records in this set are not as volatile as in the first group. In fact, once a record is established on the file there should be no change other than that which emerges from normal upgrade through usage of the record, and the free proofreading and scrutiny which such usage engenders. It is true, how- ever, that records may be supplanted in toto by other versions, particularly if a system of preferred cataloging source (or sources) is established. This does not mean that standard bibliographic data will be forced down users throats, but merely that a user will adopt and adapt them for local usage while the master file is protected. While the file will be used heavily, it will BIBLIOGRAPHIC DATA CENTERS 153 X Fig. 1. The First Circle of Files: Volatile Working Files 154 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING MARC Mono- graphs MARC A/V o X MARC Serials 1 O O etc. Fig. 2. The Second Circle of Files: Standard Bibliographic Data BIBLIOGRAPHIC DATA CENTERS 155 Fig. 3. The Third Circle of Files: Abstracts and Indexes not be volatile (in the sense that there is no activity against the records, only retrieval), but it will grow rapidly as more and more libraries contribute to the master file. From the third and outer circle of files (figure 3), the librarian, on behalf of the user, and indeed the user directly, will acquire the ability to search data bases comprised of abstracting and index data, such as MEDLARS, CBAC, BA Previews, etc. These files will be used for retrospective searching in both interactive and batch mode, and will also be used for selective dissemination of information services. The files will be very large, but at the same time the most stable, inasmuch as the data themselves will not change (indeed some may be 156 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING Level 3 Level 2 Fig. 4. Linkage Among Levels: A Subject Search protected from change by copyright agreements). The size of the on-line files, and the proportion available for retrospective search, will be a reflection of the volatility of the subject matter; files of scientific information will probably cover a shorter time span than social science or humanities files. The reader should note that although in the three figures only four files are shown at each level, this number is only used to illustrate the concept. In the BIBLIOGRAPHIC DATA CENTERS 157 seventh Annual Review of Information Science and Technology, the review chapter on machine-readable bibliographic data bases lists many data bases. 1 While the existence and availability of these files is essential, even more important is the creation of the necessary linkages among the levels and among files at the samel level. Two examples will suffice to indicate what the linkages would do in an on-line system and why they are essential. In figure 4 the librarian (or user) is searching a data base, say, ERIC, and retrieving a list of citations. These citations are really almost useless unless they carry the user to the document. (Giving the user citations has not solved his problem; it has merely changed it.) In the figure the serial citations are passed through the serials data file to impose bibliographic consistency, al- though this may not be a necessary step in all cases. From there, the citations are passed to a Union List of Serials data base to discern locations. From there they go to a circulation file to assess availability and then back to the librarian or user, some data perhaps being forwarded into an interlibrary loan module either directly or following user decision. I suggest, of course, that this be automatic and that the results of all searches be passed through this route, such that the user gets a complete report in response to his search request. Figure 5 is an example of a different kind of linkage, one devoted primarily to internal housekeeping, which shows the path of events following the decision to order a new serial. The librarian goes out to the file of standard bibliographic data to acquire authoritative data prior to placing the item in the on-order acquisition in-process file. The subsequent receipt of the first of the serials would trigger a range of actions. The in-process file would be amended; full serials control of data would be needed to update the serials catalog; an addition would be made to the Union List of Serials; and a serials check-in record created, again based on data already extant in the serials data base and in- process files; and finally data from the serials check-in record would be needed both for the serials catalog and the circulation system. Figure 5 differs from figure 4 in that there is much delay between some steps rather than immediate transfer of information in order to complete the transaction. Thus far these files have been depicted in a flat time-dependent schema in the relationships of one to the other, but there is another set of relation- ships which are equally important-their relationship in space. I do not believe we yet know which is the most appropriate level at which all of these files should be maintained. Should circulation be done at each local campus, by subregions within state, or statewide? At the other end 158 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING Level 3 Fig. 5. Linkage Among Levels: Order a New Serial BIBLIOGRAPHIC DA TA CENTERS 159 NATIONAL Fig. 6. Levels of Appropriateness of the scale, how many centers should be carrying, for example, engineering or geographical data one national center, a set of regional centers, or every campus? The answers to these questions will lie in the usage to which each data base is put (measured quantitatively) and whether or not linkages can be forged between the present disparate systems which exist on a national basis. There must be a level at which each system can achieve close to optimal cost benefit for its users. When figures 3 and 6 are imposed one on the other, a three-dimensional picture, more like a contour or relief map, emerges where files will be located at varying levels and distances from the user, and from each other. The important, indeed vital, fact to remember is that where the file is held is of no concern to the librarian. What is important is access to the file. It is possible to set up performance criteria that insure good response times (measured in seconds) for the functions described above. That is what large on-line interactive bibliographic computer systems are all about. Furthermore, let me reiterate that these systems will be cost-beneficial to the participating library. 160 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING O Fig. 7. Modules Which Already Exist Work in New York State With this bibliographic data center in mind, what progress is being made toward the realization of such a system in New York State? First let me admit that the description of the center has that all-too-familiar, all-too-de- pressing ring of blue sky about it. We have been reading and dreaming about similar systems for a decade. I suggest, however, that we are a lot closer to realizing such a center than is generally appreciated. In figure 7 I have filled in examples from each level to indicate that systems do exist for each. BIBLIOGRAPHIC DATA CENTERS 161 In the first circle one can point, for example, to the Syracuse University on-line in-process and shelflist system and its circulation module. The Ohio State University Libraries Control System, described by Hugh Atkinson, 2 is another example. The Ohio College Library Center is performing the function of the second circle as it provides on-line access to shared cataloging data, presently for monographs, but hopefully for all other media. It should also be remem- bered that a link does already exist between these first two levels when OCLC provides OSU with a weekly cataloging tape to maintain their own (OSU) on-line shelflist for the circulation system. Examples of the third level are the State University of New York's Biomedical Communication Network and the National Library of Medicine's MEDLINE system, through both of which the MEDLARS data base may be interrogated interactively by librarians or users. Thus in New York (on the SUNY central computer) we already have a segment of the outer level. And clearly if the system can interrogate MED- LARS it can also work against other data bases. We are, therefore, actively examining other data bases and working on a schedule with which to bring up other data bases, ERIC and BA Previews, for example. At the inner level, State University has taken the OSU libraries Control System and is adapting it for use on the SUNY central computer to serve in the first instance the Albany University Center Campus, by September 1973, but we plan to amend the programs to handle multiple files and gradually extend the use of the system throughout the university. Several campuses of the university have already expressed keen interest in participating in such an expansion. This is a very attractive mode of action for another entirely different reason, namely that the result of such action would be a start on the development of a University-wide shelflist, a tool which is absolutely essential for the best utilization of the library resources of the university. In addition, also at the inner level, the NYSILL interlibrary loan system runs partially in an interactive mode. What of the second level? Work in this direction is taking very positive form within the state. State University and the State Education Department have within the last few months created a Task Force on Library Data Centers, which has eighteen members. Six of these members are from State University; the State Library and the Division of Library Development are each represented, as are three of the 3Rs Regions. Other members are from Five Associated University Libraries, New York University, New York Public Library, Columbia University, City University of New York, ANYLTS, and the Commission of Independent Colleges and Universities of the State Univer- sity of New York. 162 1 9 73 CLINIC ON APPLICA TIONS OF DA TA PROCESSING This group has met approximately each month since its first meeting on October 17, 1972, and has prepared the following set of guidelines, objectives, and recommendations. OBJECTIVES The mission of the Task Force on Library Data Centers is to promote the development and to advise in the implementation and operation of a statewide bibliographic computer network which will improve services to users of libraries within the state through on-line access to various data bases for the purposes of shared cataloging, circulation control and interlibrary lending, serials control, union lists, information retrieval, acquisitions and other similar services. GUIDELINES 1. The Task Force will be guided by such technological advances and develop- ments as have already been made, or are being made, in the fields of library data processing, recognizing that the use of existing technology will generally be more cost beneficial than parallel or competitive development. 2. The above statement, however, shall not inhibit independent development where it is considered that such development will be clearly cost beneficial to the libraries of the state. 3. The Task Force will seek to make available to libraries within the state such data bases as are available where access to such a data base will clearly be of cost benefit to users. These data bases may be of cataloging data, holdings data, indexes and abstracts. 4. The Task Force will encourage the acceptance of and adherence to appro- priate national standards for data input, format, and description. Such systems as are developed will interface with other bibliographic data networks, such that a national network can be realized. Developments made by the system should be made available to other networks on a reciprocal basis for the purpose of contributing to the total network by interlocking with other developments. RECOMMEND A TIONS Within the above guidelines, the Task Force makes the following recom- mendations: 1. That capabilities and services such as those provided by the Ohio College Library Center be made available in New York State, as a first step toward BIBLIOGRAPHIC DATA CENTERS 163 providing the above bibliographic services. It is recognized that the needs of libraries may require more than one center. 2. Planning and development of such capability and service should be initiated immediately, and should proceed as far as possible pending the availability of funding. Concurrently, a continuing evaluation of best methods for total fulfilment of the Task Force objectives should commence. 3. That the State Education Department and the State University jointly seek state and other funding for planning, development and implementation, but that operating costs for the provision of services are expected to be supported by participating institutions. 4. That such services will be available to private and public academic libraries, and to such other users as feasible. 5. That the planning and execution of all studies and development work will be the joint responsibility of the State Education Department and the State University with the advice of the Task Force. 6. When operational, the bibliographic data center will have governance that provides for appropriate participation by public and private institutions, and SUNY and the State Education Department. The Task Force is now seriously examining the opportunity for devel- oping a system at the second level, either by totally replicating the SIGMA V System used at OCLC or, much more likely, performing a functional replica- tion of the OCLC system on hardware which exists within the systems, and using existing (and expensively trained and experienced) personnel. In particu- lar, State University, State Education Department and New York Public Library have available enough experienced people to build a very powerful team. Whatever path is chosen, there is clear intent to develop a system and, more important, clear fiscal support for such work. It would have been pleasant and very gratifying to report that the work is done, rather than to present an account of work in progress. However, we are sufficiently along the road on which our thinking is taking us as to allow hope that some worth can be gleaned from this report. New York cannot do the work on its own, of course. We are already drawing heavily on work performed elsewhere, and will continue to do so. We do hope that our work is a contribution to a national on-line bibliographic 164 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING data network such as must be realized in the near future; indeed we are committed to interface with and share in such a development, even though at present we may barely be able to enunciate the problems which will be encountered in such a network development. If we believe anything concerning our work in New York, however, it is that the job can, should, and will be done. REFERENCES 1. Gechman, Marvin C. "Machine-Readable Bibliographic Data Bases." In American Society for Information Science. Annual Review of Information Science and Technology. Vol. 1. Chicago, Encyclopaedia Brittanica, 1972, pp. 323-78. 2. Atkinson, Hugh C. "The Ohio State On-Line Circulation System." In F. Wilfrid Lancaster, ed. Proceedings of the 1972 Clinic on Library Applications of Data Processing: Applications of On-Line Computers to Library Problems. Urbana, 111., University of Illinois Graduate School of Library Science, 1972. PHILIP L. LONG Associate Director Ohio College Library Center Columbus, Ohio OCLC: From Concept to Functioning Network The Ohio College Library Center (OCLC) is a not-for-profit corporation chartered by the state of Ohio to provide a means for greater cooperation among libraries of all types in Ohio and in regional library systems outside Ohio. The members of the Center are approximately fifty academic institutions and public libraries in Ohio. The origins of the Center go back to the 1950s when various efforts were sponsored by members of the Ohio College Association toward bringing about greater cooperation in the field of library service. These various efforts came to fruition when, in the summer of 1967, the Ohio College Association "spun off the Ohio College Library Center as a separately chartered and independent corporation, the purposes of which were to increase the availability of resources and to diminish the rate of rise of per-unit cost of library service in its member institutions. The latter part of 1967 and the majority of 1968 was, for the Center, a time of introspection and self-definition. During that period, Frederick Kilgour, the Center's director, traveled extensively, speaking with members in order to obtain the sense of the original membership regarding what services the Center would provide and what were to be the initial priorities for the establishment of these services. The result of this effort was the determination that the Center would concern itself initially with implementing, through the use of advanced on-line computer technology, services which would augment and aid library personnel in classical areas of library service . Programming for information services which are new to librarianship was to be deferred until after programming had been activated for already existing areas. The initial definition of areas of activity encompasses the concept of a tightly integrated, nonredundant system based on central usage of bibliographic data in an on-line catalog. These initial systems were to be: 165 766 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING 1 . shared cataloging, 2. remote catalog access and circulation control, 3. serials control, 4. a technical processing system, and 5. retrieval of information by subject approach. The Center's staff was augmented in early 1969 by the addition of programming personnel. At that time the decision was made to provide, for an interim period, off-line catalog production to be based on a program previously developed at the Yale University Medical Library. Several man -years of effort had been expended in production of this system, and a considerable saving in time and effort could be realized by the transferral of this programming and its associated operating system from the Yale Direct Coupled 7094-7044 system to a stand-alone IBM 7094 system which was operational at the Ohio State University. Concurrently with programming to adapt the Yale bibliographic system for usage in OCLC member libraries, an effort was begun to define and select necessary computer equipment for the Center's long-range use. Ten major manufacturers of large computing equipment submitted proposals to the Center for equipment. However, because successful operation of a computer-based on-line bibliographic network had not yet taken place, neither the Center's personnel nor the computer manfacturers' personnel had a clear understanding of processor requirements for such tasks. Hence the Center could not rely upon others' experience as an aid in selection of equipment. Since the construction of a realistic bench mark test was effectively impossible under the circumstances, the Center sought another means for evaluation of proposals. The evaluation technique which was selected was simulation. The Center choose the services of COMRESS, Inc. of Rockville, Maryland. COMRESS has, since 1962, been successful in the simulation of computers in various environments, including that of real-time interactive processing. The SCERT simulation program used by COMPRESS offers the advantage that file definition, record access, and data flows are defined in a fashion which is independent of the computer and operating system configuration on which this processing is to take place. These latter items are included as two small definition decks. Thus by the expedient of changing these two small decks one can quickly compare the relative merits of computers from various manufacturers employing different operating systems. The central question to be answered by the simulation was: What machines, if any, are available to service an extremely heavy load of communications requests made against an extremely large data base, for a 767 population of users capable of supporting computer expenditures in the area of $20,000 per month? Each manufacturer was explicitly invited to view and review the results of his own machinery in simulation. The intent of this was to obtain each manufacturer's best thinking, and to afford each manufacturer an opportunity to challange the results of the SCERT simulator, as applied to its machinery. Several manufacturers availed themselves of this opportunity, based on simulation results, and modified their proposals. One manufacturer challenged COMRESS's ability to properly simulate its equipment. This challenge was adequately rebutted by COMRESS analysts in joint meetings with the manufacturer's representatives and Center personnel. An unfortunate initial result of these simulations was the determination that none of the proposed equipment could, using manufacturer-supplied operating systems, handle the projected system load. This came as a distinct and unpleasant surprise to Center personnel, since we had imagined that the central processing load in such a system would be relatively light, and that the principal loading would be found on communications controllers and disk files. Careful examination of simulation results revealed that a considerable proportion of the central processing unit load would arise from overhead functions in operating systems. This gave rise to the hope that improved operating systems for these machines could be devised-operating systems which would reflect efficient programming for the specific tasks which the Center's activities encompass, rather than generalized operating systems capable of responding to a virtually infinite variety of unforseeable processing requirements. The Center notified all manufacturers of these results and stated that an effort would be made, by modification of the so-called factor library of the SCERT modeling system, to devise nonexistent operating systems which did not exhibit half a dozen or so almost universal inefficiencies. After the key inefficiencies were identified and removed from the models, simulations were again run for all the machines. The result of this was an almost across-the-board reduction by a factor of four in the central processing power required to sustain the peak traffic loading anticipated for the system. Three computers were identified as requiring a sufficiently low percentage of available central processing power to make the project feasible. At this point other features of the machinery and the manufacturers' support and financing arrangements were examined. The Xerox Sigma-5 computer emerged from this latter evaluation as being the machine of overall choice as a vehicle for solving the problem stated above. A Xerox Sigma-5 was placed on order in February 1970. Thereafter, programming to interface the Yale bibliographic system catalog card formating 168 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING progarm with Ohio State University computing environment, and to provide a conversion interface between the internal format of the Yale program and Library of Congress MARC distribution service tapes proceeded rapidly. By summer 1970, over one-third of the OCLC membership was activated in the off-line catalog production system. In that system, MARC tapes were first preprocessed on an IBM 360/Model 75 computer and then passed to a conversion and file maintenance program which operated on that machine. This latter program (CONVERT) updated the master bibliographic file according to function codes contained in MARC records, and, at the same time, selected records to be processed for catalog production from the data stream. These selections were based on requests mailed to the Center by user libraries on 80-column punched cards. The output of this program was a seven-track tape containing bibliographic records in Yale internal format, and catalog card format instructions contained in tabular form. Using this data, the reprogrammed Yale formating system, which now ran about ten times faster than it did upon its arrival at Columbus, produced catalog cards in one of 3,000 possible combinations of options. The formated card images which were produced by this program were taken from the IBM 7094 computer back to the IBM 360/Model 75 to be sorted in filing order for specific catalogs. A final print tape was then produced on the IBM 360 before printing the card images on special forms, in "two-up" fashion, on the IBM 1403 printer, for which the Center had obtained a modified TN print train. Approximately one-half million cards were ultimately produced by this system, at an average cost of approximately 6.8^ per card. Users of the system were able to begin a smooth transition from wholly manually oriented processing to the on-line interactive processing system which was to come. At the same time Center personnel were able to gain invaluable experience in the exceedingly complicated programming necessary to produce individually custom-printed catalog cards for the catalogs of participating libraries. As the initial off-line catalog production system was coming into being, Center personnel completed a survey of available CRT plus keyboard interactive terminals for use in the on-line shared catalog system. After considerable deliberation, it was decided that no terminal was sufficiently close to the requirement of library processing, and that the Center should participate in the design of an interactive terminal which had extensive text editing capability embodied in the processor of the terminal and which had a sufficiently large character-set to encompass the vast majority of processing (in romanized form) of foreign-language materials. Spiras Systems Inc. of Waltham, Massachusetts presented the most attractive opportunity for the production of such a terminal. 169 Close interaction between senior personnel at the Center and engineers of Spiras was begun in late 1970. Prototype terminals were made available to the Center for testing in the Spring of 1971. The Sigma-5 computer system was delivered to the Center in September 1970 and accepted for operations in October of that year. Programming was then begun to modify the operating system of the Sigma along the lines suggested by previous simulation as being necessary to accomplish the objectives of a network. In parallel with this activity, Center programmers began work on the on-line catalog interaction system, and on the production of third generation versions of programs for the production of catalog cards. The flexibility of the catalog production system was greatly enhanced during this period, such that some 8,000 combinations of printing options came to be available for user libraries' catalogs. The on-line system was made available in training mode to catalogers in member libraries during July 1971. In October 1971, input of new bibliographic data from terminals in member libraries began. By February 1972, a system which permitted the smooth interaction of both member-created records and those from the Library of Congress was active for all but a very small percentage of Library of Congress records. This latter percentage represented records which very probably were for titles for which bibliographic data had already been input from a terminal, but which were such that an automatic program verification of duplication could not then be made. During 1972, advisory committees on technical processing and serials control met regularly and produced detailed requirement definitions for these two subsystems. In February 1973, the Center activated programming to provide an extended search capability for those titles for which the normal truncated search key produces more than two screens of main entry-title-publication date personalized catalogs. (It is from these personalized miniature catalogs that the cataloger finally selects the bibliographic record on which to base her institution's cataloging.) This technique provides a very rapid indication of whether or not material the cataloger seeks is, in fact, contained in the on-line file of reference bibliographic information. At this moment the monograph shared cataloging system is functioning smoothly and is, for all intents and purposes, nearly completed. It currently services more than ninety terminals, of which approximately seventy-five are located in member libraries in Ohio. To date the Center has concluded cooperative agreements with a number of like-minded groups outside of Ohio, including NELINET, PAUL, PRLC, PALINET, and CCLC. When terminals in these regions have been activated, the number of on-line terminals will have been approximately doubled. 170 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING On March 26, 1973, the membership of OCLC agreed to extend the services of the Center to regional groups outside the state on one of three regular bases: 1. that a regional group already operates a computing facility; 2. that a regional group does not now operate a computing facility, but intends to do so; or 3. that a regional group simply wishes the Center to serve their constituency, using a computer located at Columbus. Additionally, the membership instructed the Board of Trustees and the Center management to prepare a plan whereby the Center could conclude such agreements with individual large libraries. Currently the Center is actively negotiating contracts with a number of regional groups throughout the nation. Programming to activate the serials control system will commence in June of 1973. It is anticipated that entry of bibliographic data for serials, check-in of journals, claiming for journals, and production of catalog cards for serials will all have been activated before the summer of 1974. The Center will begin adding personnel for the purpose of programming the acquisitions component of the on-line technical processing system, starting in late July 1973. It is currently anticipated that system definition will begin in 1974, for activation to start in 1975 of the remote catalog access and circulation control component of the Center's integrated on-line bibliographic system. Subsequent to that will come subject approach retrieval of information, interlibrary loan, and other information services which participants in the Center may require. JOSEPH BECKER President, Becker and Hayes, Inc. Los Angeles, California In early 1950, Ralph R. Shaw, the father of library mechanization, persuaded Louis N. Ridenour, a science advisor to the President, to come to Urbana and give one of the Windsor Lecture Series in librarianship at the University of Illinois. The program was entitled Bibliography in an Age of Science, and the resulting collection of papers became a classic in library literature. Ridenour was a dynamic, energetic man. He was barely into his thirties then, a technologist, not a librarian, intimately acquainted with the research and development activities of World War II and keenly perceptive about the future role of technology in the evolution of the nation's libraries. One particular passage in the lecture caught my attention and stayed with me. Ridenour said: What is needed is a re-evaluation of the whole fabric of the library, from the bottom to the top. In making this analysis, no prejudice should enter, nothing should be taken for granted. To mention only one thing, a [research] library should no longer necessarily be regarded as a place where books are stored. Perhaps it is entirely something else. Possibly a library is a combination of study rooms, seminars, and a first-rate communication center of a specialized sort. 1 This was Ridenour's way of challenging librarians to rethink their basic goals. He was urging them, because of technological developments, to shuck their traditional role as passive keepers of books and aim at operating dynamic information mechanisms. His phrase "a ... communication center of a ... sort" implied reaching out to users with library and information service, sharing resources, switching information messages, and exchanging knowledge. He was talking about networks without mentioning the word, and he was dramatizing the need for an organizational and technological upgrading of libraries. I had the good fortune to know Louis Ridenour personally, and he was more than just a visionary. He was a practical, hard-hitting engineer who, in 1951 , was already aware of the potential information power of digital computers and high speed communications. Regrettably, Ridenour never saw his 777 ; 72 7975 CLINIC ON APPLICA TIONS OF DA TA PROCESSING technological predictions come to pass; he died unexpectedly a few short years after delivering his Urbana address. Ridenour was right in saying we need to examine the whole fabric of the library from the bottom up. The "whole fabric" means looking into the way libraries are organized in this country, and viewing them not as separate pieces but as one total system. Changes in libraries will not come about by technology alone. Library growth at the turn of the century was in the direction of decentralization. Public libraries spawned branches, academic libraries created departmental collections, and so forth. But in recent years, we have seen this trend reverse. During the last decade, libraries across the country began to develop new organizational relationships to facilitate the sharing of resources. These cooperative programs are variously referred to as regional library systems or library consortia, but they are the germs of networks that one day soon will link one library to another through some national system of interlibrary communication. Libraries are regrouping both by type of library and by geography striving to serve wider jurisdictions. The concept of serving wider jurisdictional units is attractive for a number of reasons. First, it has political appeal because sharing implies better utilization of existing resources. Second, sharing has professional appeal because it makes a larger base of knowledge available to serve local needs. And third, it has administrative appeal because sharing across jurisdictional boundaries implies greater economy and efficiency of operations. From the standpoint of computer application, for example, libraries with similar problems and responsibilities stand a better chance of automating as a group than as individual institutions. Regrouping of libraries by type and by geography must inevitably lead to regional hierarchies and then to a national intertype library network. Unmistak- able signs of this trend are already evident. Intrastate intertype networks are being planned or are in partial operation in Washington, Illinois, New York, Wisconsin, Maryland, and California; and regional networks such as SLICE in the Southwest, NELINET in the Northeast, and a new one forming among the states of the Southeast are examples of emerging networks that will cross state lines. Interlibrary cooperation has been practiced by enlightened libraries for many years on the principle of the golden rule it is every bit as important to give as to receive. Libraries have been organized essentially as separate entities and, for the most part, they cooperate with one another to the extent that it does not interfere with local obligations. Library networks, on the other hand, are something new and something else. They imply interdependent rather than independent, organization; they imply intermembral rather than individual decision-making; and they imply having extrajurisdictional responsibilities rather than merely local ones. LIBRAR Y NETWOR KS 1 73 Before considering how a national intertype library network would function as a total system, let us examine the consequences of segmented organization by looking at some of the major problems confronting libraries today: We are in a period of "quiet crisis. " Most libraries are crowded and understaffed, unable to keep pace with service demands, critically short of money and uncertain about their future goals, objectives, and sources of funding. We are not up to standard. The level of library and information service is below ALA standards in most parts of the country. Certain segments of the population are better served than others, and some are not served at all. In thirteen states there are no state programs for providing aid for libraries. We have a public relations problem. Few people fully realize the extent of the services we can and do provide. The public associates us with books. They do not perceive us as information specialists nor do they recognize the national significance of our information efforts. We have reached the limits of local self-sufficiency. Most libraries are unable to afford the cost of acquiring all the books and other materials they feel they need for their constituents. As Paul Wasserman points out, "the ever mounting spiral of acquisition costs (has led to) the concommitant realization that comprehensive collections in any but the greatest libraries of the land are not realistic." 2 We are unsure of federal funding . The traditional federal funding structure for libraries has collapsed. Washington is discontinuing categorical programs like LSCA in favor of revenue sharing allocations at the local level. Since libraries must compete with other local agencies for such funds, the amount they will receive from revenue sharing is still uncertain. However, it is clear that revenue sharing funds are unlikely to be allocated to projects involving extrajurisdictional services and facilities. We are drifting toward incompatible systems. Those libraries that have formed consortia have no national standards to follow which will assure compatible systems development of the technical components of networks. Without technical standards for interstate and regional network development, we are in danger of developing a series of systems that may never connect. We are not teaching enough librarians about the new technology. Most of our library education facilities are turning out professionals who are not technically equipped to deal with nonprint materials or with the new computer and communications technology. Carlos Cuadra says, "the hardest problem about networks. . . [and] the most serious ... is the lack of training and preparation of people so that they can contribute to progress in the individual libraries which will be the nodes of networks." 3 174 1973 CLINIC ON APPLICA TIONS OF DA TA PR OCESSING We are without a master plan. Libraries and information centers are not developing according to any national plan, and consequently, from a systems viewpoint, their growth is uneven and uncohesive. If ever there was a point in library history when we needed a beacon light to guide us, now is the time. Will a national network be the goal that mobilizes fresh initiatives, cures ills, and stimulates library progress? Some librarians believe its achievement can breathe new life and spirit into present-day librarianship. Others feel that the benefits of an organizational and technological upgrading of librarianship are more conjectural than real. I ally with the first group, but no one yet has a clear enough picture of how a national network will function to know whether it will represent a true information breakthrough. The main reasons why libraries must seek greater communication with sister institutions are very clear. First, they recognize the economic impracticality of massive duplicate collections proliferating in different geographic locations. The logical alternative is to consider interconnecting libraries so that the combined information utility is available to each of them. Second, libraries believe in the right of each citizen to the information he needs. A mobile, expanding, diverse population requires equal access to available knowledge no matter where people reside, or where the information resides, unless, of course, there are valid legal or proprietary restrictions. With the advent of new communications technology, the constraints of geography need no longer impose barriers to the free flow of information. And finally, libraries see a change in their own role as a social institution. Society requires improvement of the amount, kind, and quality of information services it receives, and libraries must recognize that they serve their clientele not just with print, but with information in all forms. At this juncture, it may be helpful to consider a hypothetical national intertype library network so that we begin to develop some common understanding of its principal components and capabilities. The focal point for any national network will of course be the Library of Congress. Although the Library of Congress is not officially designated as a national library, it does in fact perform many common processing services and provide many user services for the libraries of the country. Its latest national processing activity is MARC, which led to the establishment of a host of commercial and nonprofit processing centers, serving well over 1 ,000 satellite libraries. The national libraries of other countries have also adopted the format. Today, LC exchanges tapes with the United Kingdom by 747 jet, but the time is near when this will happen by satellite, and the data bases of several countries LIBRARY NETWORKS 175 will be integrated into a single electronic network. R. M. Duchesne of the United Kingdom has already proposed an idea he calls SUPERMARC-that is, a superset of national MARC fields for truly international communication as one way of proceeding toward an international network. 4 The point is that LC is crucial to the organization of a national network because it has the capacity and the materials to perform many common services in both the areas of technical processing and reference, and because it can set bibliographic standards for the network. A national plan would define these new national services and authorize and support LC to perform them for the common good. If LC is the apex of the national network hierarchy, the next level must certainly consist of other national libraries like NLM and NAL, plus a number of other institutions in the country, in both the public and private sectors, whose collections constitute unique national assets. It seems that a national network should protect and nourish these national resources whether they be research libraries, indexing and abstracting services, special libraries, or data bases, so that, in their respective specialized domains, they can offer user services to all libraries in the country that are affiliated with the national network. No systematic program currently exists to permanently safeguard or develop these resources so that their use can be extended nationally. A national network would have the leverage, for example, to assure standardization and the orderly development of machine-readable data bases in every important subject field represented by these collections, and thus speed the eventual integration of the information they contain. Computer processing installations are at the third level of the hierarchy, and I perceive them to be of two types: type I dedicated to bibliographic production, and type II dedicated to service uses. Many computer centers will be needed to help the network transform the machine-readable bibliographic records produced by LC and other national libraries into by-products for local distribution such as cards, book catalogs, special bibliographies, SDI services, etc. Each center would also be responsible for assisting local institutions with the conversion of unique holdings. For each library to own its own type I computer installation would be prohibitively expensive, so the cooperative, multi- institutional approach, successfully demonstrated by Kilgour at OCLC, seems to be a most economical and efficient solution. Type II computer centers would be devoted to service uses; first as electronic holdings directories, later as automatic sources to on-line data bases in different fields. While we have no library example to point to today, the theory of use is similar to that of ARPANET. ARPA is the acronym for Advanced Research Projects Agency of the Department of Defense, which currently 176 operates an experimental, interdependent computer network among a set of far-flung university computer centers. ARPANET enables users on one campus to interrogate and manipulate data files that are under the control of a computer on another campus. Commercial timesharing networks likewise have this capability. e.g.. NLM's MEDLINE mentioned in Davis McCara's article in this volume. Goser to home is a system of interlibrary communication under investigation by the ARL. ARL is considering using a computer as an "electronic mailbox," which would store interlibrary loan request messages for particular institutions and transmit them automatically according to predetermined schedules or on demand. A national system of interlibrary communication for interlibrary loan would not only route messages more effectively, but it could also utilize companion computer programs to manage and administer the operation of the total system. Thus, a computer could handle billing, maintain the statistics, do accounting, keep track of copyright royalties, etc. In time, with heuristic programming and a directory of holdings, a computer might even learn to switch incoming requests automatically to those institutions in the network that have the highest response potential. This is only one example of an application which a type II center could perform for a family of libraries; I suspect there are many more. A national plan would designate the number and the location of these type I and type II centers, and support them with research, software, technical guid- ance, and perhaps even funds for equipment. A good way to picture the compu- ters at this third level of the national network is to think of them as a set of fast, large, timeshared information computers with many receiving sets in libraries. Just as large generators distribute electrical energy directly to homes, so time- shared computers in a national network will probably operate as information utilities. Computer usage usually implies economies of scale, and this suggests that type I and type II installations will be massed to serve the processing and ser- vice needs of many institutions on an intrastate, multistate. or regional basis. A major resource library in each state win represent the next level of the national network hierarchy. These fifty libraries will be the backbone of the network because they will be responsible for establishing a compatible intrastate network, and for switching referrals in and out of the state. Each state would try to mirror the national network structure within its own borders, or band together with other states to achieve the same pattern regionally. By affiliating with the national network, each state receives the products and services offered by all of the national libraries and by the type I and II computer centers. A national plan would spell out each state's obligations to the network and specify the standards to be followed for guaranteeing technical and operating compatibility. LIBRARY NETWORKS 177 Telecommunications is the final major component in a hypothetical national network. A communications grid which ties all of our libraries and information centers together will do more for the democratization of infor- mation in the US. than anything else. Until now, the U.S. mails, the telephone, and the teletype were the principal arteries of interlibrary communication. Federal postal regulations provide special, low mailing rates for books, states provide WATS tine telephone facilities for libraries, and most of the major libraries in the country have teletype equipment for processing interlibrary loans. It is obvious, however, that the future of intertibrary communications ties well beyond the use of the mafls, the telephone, and the teletype. Communi- cations are needed to bridge the physical distances between library and library, and between library and user. Communications are needed which can mix the variety of digital and analog signals that are destined to flow back and forth over tomorrow's library and information networks. And, communications are needed with channel capacities great enough to handk the message switching loads, and the enormous volume of data traffic implied by a national library network. Although the advantages of telecommunications have been known to libraries for many years, operational use has been hindered by vexing problems of cost and systems planning. However, as libraries begin to make greater use of computers they are discovering the opportunities which an effective communi- cations interface makes possible. Thus, with the continued application of computer technology to libraries, we can expect to see the increased use of advanced forms of telecommunications. Communication requirements of a national network may be grouped into three divisions: intertibrary communications, internal communications, and user communications. These divisions form a communications hierarchy which requires a different combination of communication capability at each level, as well as communication compatibility between levels. Intertibnry communi- cation enables one library to communicate with another. Only in this way can a library develop information exchanges with networks in adjacent states or regions, with the national libraries, or with type I and D computer centers. Telecommunications at this level must be broadband, bidirectional, and capable of accommodating and switching a mix of printed, digital, and video formats. Internal communication refers to the use of library communications by the professional staff in a library to speed up the local processing and retrieval of information. Telecommunications at this level can be narrowband for both data and voice. User communication concerns the use of communications for delivery of information or material by the library directly to the user's home or office. Because user needs for information vary considerably from person to person, the / 78 1973 CLINIC ON APPLICA TIONS OF DA TA PROCESSING communication requirements at this level will vary, too, and very likely require broadband facilities. A national network must, therefore, incorporate the means for communi- cating among the nodes of the network. While it is true that AT&T, Western Union, Microwave Corporation of America, and other companies are in the process of upgrading their commercial lines for domestic use, it seems that a library network exception to the federal telecommunications regulations will be needed to guarantee low telecommunication rates. If the main purpose of a national network is to place the user in contact with his material, then rapid, inexpensive telecommunications among libraries are absolutely essential. Achieve- ment of open telecommunications among libraries will be the greatest boon ever to national distribution of knowledge for education and progress. Although I have described a hypothetical national network in terms of organization, a group of communication and computer specialists met in 1970 at Airlie House in Warrenton, Virginia, under the leadership of John Meany of Notre Dame to study the technical aspects. The group was part of a national conference on interlibrary communications and information networks sponsored jointly by the U.S. Office of Education and the ALA to explore the telecommunications domain for library and information purposes. 5 Meany 's working group formulated several basic technical assumptions about a national network. As far as I know, nothing more useful has emerged since they were formulated. I would like, therefore, to repeat his technical assumptions about a hypothetical network because they complement the ones I have made about organization and, in a way, summarize the concept: 1 . The network would be national, regional, and local in scope. 2. It will include all types of libraries, data, information analysis centers, instructional media centers, etc. 3. It will facilitate the exchange of bibliographic data, mediation of reference inquiries, and the distribution of library and audiovisual instructional materials. 4. It will have no geographic restraints. 5. It will make maximum use of computer and communications technology. 6. It will provide timely access and response rates consistent with the urgency of a user's need for information. 7. It will adopt a standard format for bibliographic interchange and establish other protocols and common practices. 8. It will supply incentives and evolve a financial structure to stimulate network use. 9. It will consist of a formal set of major nodes at the national and regional LIBRARY NETWORKS 179 levels, and individual access points within a reasonable radius of local nodes. 10. It will incorporate switching stations and directories for request and response referrals. 11. It will enable users connected to one node to have access to any other node. Will we see the beginning of a national intertype library network in the decade ahead? I think we definitely will. There is a marked trend in the profession today to search for new operating responsibilities and a new role for the library. While uncertainty may be our constant companion, that search must go on. A national network is appealing because it represents a unifying idea that is in tune with the times and professionally meaningful. The alternative is to continue compartmented development and to redress old problems. Despite the social engineering headaches that will surely accompany network building, we begin with the foreknowledge that the technology will work it has already been proven in other fields and that the organizational implications are well within the practical limits of management science. Perhaps I read more into Ridenour's words than he intended, but I do not think so. Now is the time to re-examine the whole fabric of the library in the context of technological developments. We must move forward not working against machines, but working with them; not growing apart, but growing together; not thinking in local terms, but thinking nationally. A national intertype library network is our beacon light and it's just around the bend. REFERENCES 1. Ridenour, Louis N. "Bibliography in an Age of Science." In Louis Ridenour, et al. Bibliography in an Age of Science (Phineas L. Windsor Lectures in Librarianship). Urbana, University of Illinois Press, 1951, pp. 28-29. 2. Wasserman, Paul. "Engineering Change in Librarianship: From Revised Paradigm to Prototypes for the Future." In Library Lectures. Baton Rouge, Louisiana State University Library, 1972. 3. Cuadra, Carlos. "Ethos," Five Associated University Libraries News- letter, 4:1, Jan. 1973. 4. International Seminar on the MARC Format and the Exchange of Bibliographic Data in Machine Readable Form. The Exchange of Bibliographic Data and the MARC Format (Bibliothekspraxis Band 6). Berlin, Verlag Dokumentation, 1972. 5. Becker, Joseph, ed. Interlibrary Communications and Library Net-works (Proceedings of the Conference held at Airlie House, Warrenton, Virginia, Sept. 28, 1970-Oct. 2, 1970). Chicago, ALA, 1971. LIST OF ACRONYMS AEC Atomic Energy Commission ALA American Library Association ALPS Advanced Library Processing System ANYLTS Association of New York Libraries for Technical Services ARL Association for Research Libraries ARPA Advanced Research Projects Agency ASC II American National Standard X3 .4- 1 968 ASDD Advanced Systems Development Division (of IBM) ASIS American Society for Information Science BASIS-70 Battelle Automated Search Information System BCL Books for College Libraries BDIN Book Dealer Identification Number BNB British National Bibliography CBAC Chemical Biological Activities (a publication of Chemical Abstracts Service) CCLC Cooperative College Library Center CIP Cataloging in Publication COM Computer Output Microfilm COMPENDEX Computerized Engineering Index CRT Cathode Ray Tube DOS Disk Operating System EBCDIC Extended Binary Coded Decimal Interchange Code ELMS Experimental Library Management System FASTERBASIC Filing and Source Data Entry Technique for Easier Retrieval PAUL Five Associated University Libraries FLCC Federal Library Cooperative Center GAO Government Accounting Office GSA General Services Administration IBM International Business Machines ILO International Labor Organization ISAM Indexed Sequential Access Method ISBN International Standard Book Number ISSN International Standard Serial Number ITIRC IBM Technical Information Retrieval Center LC Library of Congress LSCA Library Services and Construction Act MARC Machine -Readable Cataloging MEDLARS Medical Literature Analysis and Retrieval System MEDLINE Medical Literature Analysis and Retrieval System On-Line MeSH Medical Subject Headings 180 LIST OF ACRONYMS 181 NAL NELINET NIH NLM NOW NTIS NUC NYPL NYSILL OCLC OCR OMB ORBIT PALINET PHILSOM PIL PRLC SBOF SCERT SDI SLICE SUNY WUSM National Agricultural Library New England Library Network National Institutes of Health National Library of Medicine National Organization for Women National Technical Information Service National Union Catalog New York Public Library New York State Inter-Library Loan Ohio College Library Center Optical Character Reader Office of Management and Budget On-Line Retrieval of Bibliographic Information Time-shared Pennsylvania Automated Library Network Periodical Holdings of the School of Medicine Processing Information List Pittsburgh Regional Library Center Standard Book Order Form Systems and Computers Evaluation Review Technique Selective Dissemination of Information Southwest Library Interstate Cooperative Endeavor State University of New York Washington University School of Medicine INDEX Acronyms, list of, 180-81. "An Ontario Libraries' Network, or Cooperative Entanglement," Gor- don H. Wright, 68-101. "Backing Into Network Operations," Estelle Brodman, 9-23. Becker, Joseph, "Library Networks: The Beacon Lights," 171-79. Bibliographic data centers, definition, 151; files, 152-57; uses, 157. "Bibliographic Data Centers for New York State," Glyn T. Evans, 150-64. Book Dealer Identification Number, 39. Brodman, Estelle, "Backing Into Net- work Operations," 9-23. Cataloging, computer-based, 14-20. Centralized processing: in Great Bri- tain, 102-12; in industrial libraries, 116; in Massachusetts, 124-49; fi- nances of, 131-48; problems of, 126-27. College Bibliocentre, 69-70; costs of, 78; flow chart of functions, 80; functions of, 69-70; objectives of, 70-71. Computer hardware and software, 30. Cooperation: among British libraries, 102-12; definition of, 20; in indus- trial libraries, 113-23; in procure- ment, 41. Cooperatively programmed systems, difficulties, 29. Cylke, Frank Kurt, "U.S. National Libraries Task Force; A Review of Data Processing Interests," 32-55. Data bases available for on-line searching, 3. Evans, Glyn T., "Bibliographic Data Centers for New York State," 150-64. Federal libraries, automation efforts, 61-62; elements of cooperation, 66; service to patrons, 61. Hammer, Donald P., and Sokoloski, James S., "The Massachusetts Cen- tral Library Processing Service," 124-49. Henderson, Madeline M., "Prospectus for a Federal Library Cooperative Center," 56-67. 183 184 1973 CLINIC ON APPLICATIONS OF DATA PROCESSING ITIRC, 116-19. Industrial libraries, 113-23; benefits of cooperation, 120-23; financing cooperation in, 117-19. "InterUbrary Cooperation in an In- dustrial Environment," G.E. Ran- dall, 113-23. International Standard Book Number, 39, 144. International Standard Serial Number, 36. Jolliffe, J.W., "Project LOG: Central- ized Processing of Local Collec- tions," 102-12. Lancaster, F. Wilfrid, "Introduction," v-vi. Libraries, problems of, 173-74. "Library Networks: The Beacon Lights," Joseph Becker, 171-79. Long, Philip L. "OCLC: From Con- cept to Functioning Network," 165-70. McCarn, Davis B., "Network-Or All Hang Separately," 1-8. MARC, 14, 15, 25-31, 49, 66, 71, 74, 77, 78, 126, 140, 141, 142, 144, 154; definitions in, 28-29; tapes, 127, 128, 129, 130, 135, 138, 168. MARC II format, 24, 28, 29, 31, 33, 42, 44, 50; as an internal systems format, 25. "The Massachusetts Central Library Processing Service," Donald P. Hammer and James S. Sokoloski, 124-49. MEDLARS, 2, 4, 49, 155, 161. MEDLINE, 2, 3-4, 20, 21; access to, 4-7; lack of use, 4, 5; use growth rates, 6. National intertype library network, communications requirements, 177, 178; organization, 175, 176, 178; potential, 174, 178, 197. National Serials Data Program, 36-37. National Serials Data Project, 35-36. "Network-Or All Hang Separately," Davis B. McCarn, 1-8. Networks, defined, 1. "OCLC: From Concept to Function- ing Network," Philip L. Long, 165-170. Ohio College Library Center, 7, 21, 64, 65, 67, 110, 120, 152, 161, 162, 163, 165-170, 175; coopera- tion with federal libraries, 66-67; extension of services outside Ohio, 170. PHILSOM, 9-14; description of, 9-12; problems of, 12-14. Processing center, acceptance of pro- ducts by participants, 145-147; costs, 148-149; problems, 126-27; standards, 125. "Project LOG: Centralized Processing of Local Collections," J.W. Jol- liffe, 102-12. "Prospectus for a Federal Library Co- operative Center," Madeline M. Henderson, 56-67. Randall, G.E. "Interlibrary Coopera- tion in an Industrial Environ- ment," 113-23. Regional Medical Program's Library Project, 19. Ridenour, Louis N., 171-72. Sage, Charles R., "Utilization of the MARC II Format for Serials in an Inter-University Environment," 24- 31. Sokoloski, James S., and Hammer, Donald P., "The Massachusetts Central Library Processing Ser- vice," 124-29. Standard Book Order Form, 39-40. Standard Order Form, 38-39. Stearns List, 19. Subscription agents: study of, 40. Systems standards, identification and adoption, 31. Task Force on Automation of Lib- rary Operations, 56-57, 63-65. Task Force on Library Data Centers, 161-64. U.S. National Libraries Task Force, 32-55; current activities, 37-41; formation of, 32-33; recommenda- tions of, 33-35; 42-54. INDEX 185 'U.S. National Libraries Task Force: Work Group on the Federal Library A Review of Data Processing Inter- Cooperative Center, 65-67; goals, ests," Frank Kurt Cylke, 32-55. 65. 'Utilization of the MARC II Format Wright, Gordon H., "An Ontario Lib- for Serials in an Inter-University En- raries' Network, or Cooperative vironment," Charles R. Sage, Entanglement," 68-101. 24-31.