Department of Education and Science

Automation Guidelines For Public Libraries

Project team P. H. Vickers B. C. Vickery

Jane Wainwright

Report written by P. H. Vickers

:·;.:·

London: Her Majesty's Stationery Office 1975

© Crown copyright 1975.

ISBN O 11 270269 4

FOREWORD

An increasing number of library authorities are planning to introduce computer-based systems for one or more of their operations. The recent re-organisation of local govemment is likely to accelerate this process as the new and larger authorities seek to solve the complex problems of merging the records and standardising the procedures inherited from their predecessor authorities. The need to provide guidance to library authorities on the application of automatic data processing has been under discussion for so me considerable time by the Local Authorities IIJianagement Services and Computer Committee and the Department, and in í 973 the Department

Department of Education and Science, Arts and Libraries Branch, 38, Belgrave Square, London SW1X 8NR.

commissioned Aslib to carry out a study of the problem, in conjunction with research already under way on modes of computer provision for libraries of all kinds. ft is hoped that the resultant manual of guidance, which draws on the lesson learned by the pioneer authorities and analyses the various factors which need to be taken into account, will facilitate the task of introducing automatic data processing and hefp to ensure that possible opportunities for cooperative development are not overlooked. The cost to the Department of this study, excluding the cost of printing and distributing this report, amounted to f2,800.

Contents

page

1 lntroduction Library visits Advisory Committee Organization of this Manual Present S cene in Public Library Automation Acknowledgements

11 Under What Conditions of Operation Should Computer Techniques be Considered? 111 How are Externa! Deveiopments Affecting the Situation? Local Government Reorganization The British Library BRIMARC- The BNB /Brighton Públic Libraries Catalogue Project Birmingham Libraries Cooperative Mechanization Project (BLCMP) London and South Eastern Regían (LASER) Commercial Computer-based Services for Libraries

8 8 8 8 9 9

1O 12 12 13 13 14 14 15

IV What is the Best Strategy for Automation?

16

V What Appiications Should be Considered? Ordering and Acquisition

18 18 19 19 19 19 20 20 20 20 21 22 22 22 22 22 23 23 23 24 24 24 25 25 25 26 26 27 27 27 27 28 28 28 28 28 28

Features and Options Input Techniques Computer Requirements

Cataloguing Union Catalogues Features and Options 'Bought-in' Services vs the Do-lt- Yoúrself Approach Coverage of the Cetalogue Points of Access for Retrieval Additional Outputs Form of Bibliographic Record Frequency of Updating Filmgruks Input Techniques (a) Paper Tape (b) Punched Cards (e) Magnetic Tape (d) Key-to-Disc (e) Optical Character Recognition Retrospective Catalogue Conversion Recapitulation on Choice of Input Technique Use of MARC and Other External/y Produced Records Forms of Output (a) COM (b) Line Printer Output (e) Phototypesetting

lssue Systems System Functions Charging and Discharging Renewals Overdues Reservations Limitations on Borrowing Location of Stock Borrower Records

28 29 29 29 29 29 30 31 31 31 32 32 33 33 33

lnterlibrary Lending On-line vs Batch Processing Numbering Systems Counter Design

Data Capture Systems (a) Automated Library Systems Ltd (ALS) (b) ALS Labef-Based System (LBS) (e) Plessey Telecommunications Ltd {d) British Olivetti (IDP Division) Computer Processing Requirements

lntegrated Systems Oxford City Libraries System Cheshire County Library System West Sussex County Library System

VI What Software Should be Used?

35 35 35 36 36 37

A Suggested Decision Path Use of Available Software Sources of lnformation on Available Software Deve/opment of New Programs Sources of Programming Ski//

VIl What Means of Computer Provision Should be Considerad? Modes of Computer Provision (a) The Library-dedicated Computer (b) The lnstitutional Computer (e) Commercial Computer Bureaux (d) Shared Computer Facilities (e) National Computer Facilities Types of Computer Use (a) Batch processing (b) On-line Access (e) Mini-computers (d) Remote Job Entry (RJE) Summing-up

VIII How Does One Select and Design a System? The feasibi/ity study Systems Analysis and Design (1) Specifying the System (a) What to Automate (b) What Incidental Services to Provide (e) Al/owances for Future Changes (2) Costing Various Plans (3) Making a Plan (4) Checking the Plan Who Should Design the System

IX What is lnvolved in lmplementation? Strong Man Wanted Programming Ordering New Equipment Reorganization of Manual Procedures Need for Accuracy Staffing Requirements Staff Retraining Para/le/ Running and Testing Transitional Prob/ems User Education How Long Does it Take?

X What Will lt Cost? Viewpoint ot Costing Workload Estímate

38 38 38 38 39 39 39 39 39 40 40 41 41

42 42 42 42 42 43 43 43 43 43 44 45 45 45 45 45 45 45 45 46 46 46 46 47 47 47

Operatíonal Cost Estímates (a) Staff (b) Equípment (e) Materia/s (d) Computer Processing (e) Externa! services Development and lmplementation Costs

47 47 48 48 48 48 48

Bibliography

49

Appendices 1 Computer Applications in Public Libraries 2 BNB MARC Tags and Fields

51 51 52

1. In March 1973, Aslib was commissioned by the Department of Education and Science (Arts and Libraries Branch) to undertake an investigation of computer use by public libraries. The aims of the project were (a) to identify those elements of library operations which are most susceptible to the application of automatic data processing (ADP), and the scale of operation necessary before ADP is economically justifiable, and (b) to examine the factors involved in the selection of the most satisfactory systems. The end product is this manual, which is designed to indicate the factors to be taken into account when considering the automation of any activities in public libraries. 2. The DES study was closely linked with another Aslib project, commissioned by OSTI, on modes of computer provision. The aim of this project was to examine the computer-generated services and products that are appropriate to libraries of different kinds (e.g. public, university, institutional, industrial), and to study in each case the possible ways in which the necessary computer facilities might be provided. 3. Much of the necessary data and background information for both projects was gathered from the same round of visits and interviews. 4. In addition, during 1973, the team responsible for these projects assisted the Aslib Computer Applications Group in carrying out a survey of operational computer applications in UK Libraries, the returns from which were used to compile a directory34 • Question naires were sent to about 190 libraries, and positive replies were received from over 130, including 35 from public libraries. The analysis of these returns provided much useful data for the present report, and also made it possible to select suitable libraries to visit. library Visits 5. The main round of library visits took place in July-August 1973. The sample of libraries for study was chosen to include examples with single and multiple applications of computers, and using both simple and advanced techniques. In the course of each visit, after general examination of the system, information was gathered on the following topics, which experience had shown to be the most significan!: Reasons for adoption of computertechniques; 2 Basis for selection of application(s) for computer processing; 3 Problems encountered in design, implementation and development of system(s); 4 Restraints imposed by available computer processing facilities; 5 Effects of local government reorganization; 6 Plans for extension of present application(s);

8

7 Attitude towards cooperation in the development and operation of computer-based systems (inter-library, regional, or national). 6. The public librarles visited were: Brighton P.L. Camden P.L. Cheshire County Dorset County Kingston-upon-Hull City Library Merton P.L. Oxford City Library West Sussex County Library 7. In addition, the study team visited the headquarters of LASER, and the Birmingham Libraries Cooperative Mechanization Project. lnformation from other libraries was obtained by telephone or correspondence. Advisory Committee 8. An informal Advisory Committee was set up to advise the Aslib tea m on the content and leve! of presentation of this guide to public library automation. lts membership comprised: Miss D M Hudson (North Riding County Library), R J Huse (West Sussex County Library), J W Lendon (Solihull Borough Library), W R Maidment (Camden Libraries and Arts Department), and E M Broome (DES Arts and Libraries Branch). The help provided by this Committee proved invaluable, and was greatly appreciated. Organization of this Manual 9. From the many discussions we have had with public librarians in the course of this project, we have built up what we believe to be a fa ir picture of the problems that confront a public library that wishes to consider using computer techniques. These problems are presented as the headings for each chapter, and we have endeavoured to present for each one a concise review of all the factors involved and the solutions available in the present state of the art. 1O. There have been two main areas of difficulty in compiling the manual- human and technical. Most of the librarians we have met have been those who have operational computerbased systems, and who had thus acquired sorne experience in the fie!d. We have had no way of estimating (apart from the help we were able to obtain from our Advisory Committee) the probable level of understanding of computer techniques among the librarians we have not metthat is, the people for whom this manual is intended. 11. The technical problem arises mainly from the rapid rate of development in this field. Systems that we visited in 1973 had undergone many (sometimes major) changes, by the time this manual was written in 1974. New equipment and new software packages are continually coming on the market.

12. Finally, the advice we are able to offer is derived largely frorn the experience of those who had introduced cornputer systerns before local governrnent reorganization. We have atternpted to make reasonable guesses as to sorne of the irnplications of the new, larger units of local government, and have been fortunate in being able to draw on the experience of sorne of the London Boroughs; but it is impossible to predict just what the outcome of the new regimes wil! be for public libraries. The Present Scene in Pub!ic library Automation 13. Reference wili be made throughout this manual to computer applications of various kinds in specific public

libraries. To give an overall picture of the position as it was in rnid-1973, however, we have listed in Appendix 1 the public library applications as revealed by the Aslib/CAG survey rnentioned above, with later additions frorn ref. 24.

Acknowledgements 14. Grateful acknowledgement is made to the public library staff who contributed their experience to this project; to my colleague Jane Wainwright, who in writing her report on modes of computer provision did m u eh of the preparatory work for this manual; and to Bill Arms ( of the Open University), who drafted a large part of Chapter 8.

9

be considered?

15. Decisions to adopt computer techniques in public libraries have up to now been prompted mainly either by actual or impending faHures of existing systems, or by a need to seek a common system when a number of previously separata libraries have been brought together under one authority. A third, but apparently rather rare situation, is for a library to make an overall evaluation of its present systems in order to decide whether there is a case for mechanization. 16. Of these, the first has been the most common starting point in the past, especially in the case of circu!ation systems. Libraries using photocharging systems have had to consider the replacement of ageing equipment, and those using automatic sorting of transaction records have found that manufacturers are withdrawing support for the equipment in the near future. In other cases traditional charging systems such as Browne have been strained by increasing workloads to a point where alternatives had to be considered. 17. With catalogues, the decision point tends to be less definite. The cataloguing system in a single library may become unacceptably inefficient, or incompatibility between the systems of a central !ibrary and its branches may reach a stage where drastic action is necessary. In either case, irritation and dissatisfaction with present methods may have built up over a period of years. 18. The second kind of motivation is demonstrated by the case of Camden. Here a number of !ibrary authorities were brought together by the London Government Act of 1963, and it was recognized not only that common systems would be necessary, but also that the opportunity should be taken to bring them all up to a high standard. Local government reorganization will provoke many similar situations throughout the country, and at the same time will bring into being many larger library authorities with much greater volumes of operation. Undoubtedly this factor, which is discussed in Chapter 3, will become the predominan! justification for the introduction of computer systems in the near future. 19. The third approach - deliberate appraisal of possibilities for future development- could well beco me a more common starting point for mechanization, as staff costs rise and computer costs decrease. Of the librarles consulted during the compi!ation of this guide, non e would claim that they adopted computer techniques primarily for reasons of economy, but this could well beco me a val id justification in the not-too-distant future. 20. Again looking into the future, new demands on public libraries could well make computar systems more attractive if not essential. One possible eventuality is that detailed records might need to be kept for the purposes of Public Lending Right.

10

21. Whatever the initial stimulus to consider mechanization, additionai benefits can be quick!y perceived to support the case. A general benefit worth mentioning here is that computer systems are space-saving, especially where they eliminate clerical tasks that wou!d normally demand working space.

22. The features and options possible with different applications will be discussed in detail in Chapter 5, but a few can be summarized here. With a circulation system, the primary requirement is usually that it should be able to cope with peak loads of charging and discharging, but a computer system will also offer the possibifity of generating overdue notices automatically, of improved reservation facilities, and of better management information.

23. A computer-based catalogue not only serves as a basis for unified practice, often replacing methods that have evolved in a haphazard way; it can also eliminate wasteful dup!ication in typing and filing; it offers the possibility of multiple copies at small extra cost; it may reduce the burden of local cataloguing by making use of bib!iographic records prepared centrally or in another library; and it can offer the possibility of producing reading lists or bibliographies with littie effort.

24. An advantage of automated systems that is not widely appreciated is that they can greatly facilitate record-keeping when moving stock from one location to another. This is particularly true where both circulation and cataloguing are computerized.

25. lt can be argued, of course, that many of the benefits offered by a computer system could equally we!l be provided by a well-designed manual system. But this wou!d almost certainly entail the use of more staff, and more space, both of which are becoming more and more expensive. lt is probably true to add that professional staff are becoming increasingly averse to clerical chores. 26. Of course, there are snags. First of all, there wil! be the cost of setting up the system, obtaining the necessary computar programs and possibly converting existing records into machine-readable form. Then the operating costs are unlikely to be any lower in real terms than those of the equivalen! manual system- although sometimes the additional costs can be disguised in various ways. A computer system may mean that a library has to accept a greater vu!nerability to externa! factors outside its control. For example, it becomes dependen! on the service it can obtain from the computer department or bureau, which in turn may be affected by power or equipment failures. Final!y, as with any change of system, sorne redeployment or retraining of staff will be necessary- an importan! factor that will be discussed in Chapter 9.

27. There is no simple formula for determining whether or not a library should use a computer. Past decisions have been based on a complex mixture of technicat human, política!, and cost factors, and it is unlikely that decisionmaking will be very different in the future. The selection of computer applications is influenced strongly by location - the pressures affecting public libraries in London, for example,

are different from those affecting Cheshire. The style of management of public library authorities also varíes widely. In sorne cases, the librarían has to prepare a very watertight case to justify the use of computer facilities, while in others, there may be positive encouragement to ensure that the local computer is fully utilized.

11

28. Apart from pressures that may arise within a library to consider the use of computer techniques, various externa! factors are coming into play which will make the possibilities of these techniques more attractive, or make it easier to take advantage of them. 29. The biggest single factor from the public librarian's viewpoint is undoubtedly local government reorganization, and the regrouping of library authorities that this has entailed. Planning the future development of libraries is made all the more complex at the present time by the emergence of three other factors : The setting-up of the British Library; lncreasing acceptance of inter-library cooperation; The availability of commercial services for book selection, ordering and cataloguing. 30. All of these have the common characteristic that they offer to libraries the possibility of using computer techniques without running systems of their own. 31. lnter-library cooperation is used here to mean not merely interlending - for which effective systems are widely established - but cooperation in a wide range of activities including acquisition and storage, the provision of bibliographical services, and the development of common or 29 shared systems • Determination among !ibrarians to maintain or raise standards of service despite increasing financia! constraints has undoubtedly contributed to the growing acceptance of these forms of cooperation. The need to make effective use of increasingly costly human resources has encouraged this trend, but the computar has been a powerful unifying influence, succeeding in this respect where good intentions have failed in the past- a point brought out in the latest Annual Review of lnformation Science and Technologyro. The Birmingham Libraries Cooperativa Mechanization Project has demonstrated the feasibility of effective cooperation even between libraries of different typas. 32. In this chapter, the most important of these externa! developments of an organizational nature will be reviewed. There are also deve!opments of a technical kind which have improved the viability of computer-based systems. The production of specially designed data-capture equipment has paved the way for improved circulation systems. The introduction of computer-output microfilm has greatly reduced sorne of the problems of earlier computer-produced catalogues. The latest technical influence is the minicomputar, which will be discussed in Chapter 7. Loca! Government Reorganization 33. Much of what might otherwise have to be presentad here has conveniently been published elsewhere in two DES publications 11• 12• lt may be useful, however, to report sorne of the information they contain.

12

34. First, there are the basic statistics of the reorganization : the reduction from 314 to 75 library authorities in England, and from 37 to 12 in Wales. The median population to be served by the new authorities in England will be about 280,000 in the metropolitan districts, and 608,000 in the non-metropolitan counties. Prior to April 1974 in England outside greater London, there were only 1 O authorities with populations of more than 280,000, whereas there are now 56 authorities with more than 280,000, including 23 with populations of over 600,000. 35. In the 'Pub!ic Library Service: Reorganisation and after', it is pointed out that these larger authorities with their greater resources will be better equipped to meet the needs of their communities for comprehensive facilities. They will be able 'more effectively and economically to deploy specialist staff .... and they will be in a better position to make full use o.f computers ..... and other aids.' The same publication also draws attention to the importance of services to be provided by the British Library, and to the advantages of computer techniques for regional lending schemes, as exemplified by LASER. 36. To achieve effective coordination of library services within a new authority it will be necessary to work towards common systems, and common standards of operation. At the same time, the scale of operation in terms of selection, acquisition, cataloguing and lending will be greatly increased. The inducement to consider the use of computerbased systems will thus be much greater than in many of the smaller authorities that existed previously. The pool of technical resources in terms of computers and programming staff should be richer, but control over their use may beco me more rigorous. 37. In order to exp!oit fully the aggregate book collections of these larger authorities, it is likely that effective union catalogues will become all the more important. In many cases these will have to cover not only public libraries, but also educational, hospital and other libraries within the same area. A computer-based system will appear to be the only practicable solution to cope with such complex tasks. 38. Centralizad cataloguing offers further advantages where book selection is decentralized. Similarly, where centralizad book purchasing is undertaken, a computer-based system is likely to be attractive for control!ing multiple requests for the same books, and for handling complex accounting procedures. For accounting generally, computer systems can be very effective, and can result in useful staff savings. 39. !n sorne areas, special problems will arise where previously separate authorities have developed tlleir own computer-based systems on different lines. Examples of this kind cropped up at severa! of the places we have visited, and decisions were being made well in advance on how these prob!ems would be overcome. Dorset County, for example, had chosen to use the Olivetti RP50 equipment for its

circulation system partly because Bournemouth had already installed this equipment. As a trade-off, Bournemouth had agreed to change to a book numbering system based on SBNs. Brighton, using ALS charging equipment andan ICL 4/50 computer, was to be merged with East Sussex, using Plessey equipment and an ICL 1900-series computer. Here it was anticipated that the two systems would run in parallel for sorne time. The British Library 40. So far as public library automation is concerned, the importance of the British Ubrary lies in the centralized cataloguing and related services to be provided by the British National Bibliography. 41. These services are based largely on the use of MARC (Machine Readable Catalogue) tapes produced by the Library of Congress, BNB, and other national bibliographies. The British and American tapes contain all material deposited in fulfilment of copyright regulations, and are basically cataloguing records to the standards set by the AngloAmerican Cataloguing Rules, 1967. The bibliographic elements of a MARC record, showing the identifying 'tag' numbers are listed in Appendix 2, and it is worth noting here that the latest MARC format is hospitable to non-book materials. MARC records have been criticized by sorne librarians for their apparently needless complexity, but it should be understood that this is inevitable with a record designed for exchange between libraries with widely different requirements. The MARC record format is notan efficient one for machine processing, and for most library systems the tapes have to be reformatted, at which stage unwanted elements can be discarded. 42. The most frequent criticism of the MARC tapes is the time-lag of about six weeks between publication of a book and the appearance of the MARC record. Both BNB and the Ubrary of Congress are taking steps to arrange for prepublication cataloguing, in arder to reduce this delay. 43. In a recent circular 5 , BNB has announced its intention of maintaining a comprehensive MARC Data file which will contain as its first leve! the complete current UK/LC MARC file and, ata secondary leve!, an English language back file and foreign language materials as they become available through Library of Congress and foreign MARC services. 44. In addition, BNB intends to provide other services not exclusively based on MARC: A basic data file for catalogue conversion which will cover the highest possible proportion of library stocks. The BNB is converting its total back file, and this will be completed in the latter part of 1974. The convers10n programme has been planned in association with the LASER conversion project (see below). These programmes will together cover up to 80-90% of titles in local authority catalogues. 2 A selective tape service which will enable libraries to obtain only the records they require. BNB has been providing a prototype service to Liverpool University, which sends a paper tape listing the ISBNs of recently acquired books. These are matched against the comprehensive MARC file, and the records required are copied on to a magnetic tape which is sent to Liverpool for local processing. 3 A local catalogue service, whereby a library would commission BNB to produce its catalogue from lists of

ISBNs plus local housekeeping data (e.g. accession numbers) for recently acquired items. The library could also input locally prepared catalogue records for items not held in the MARC data base. This will be a 'production version' of the BRIMARC service, described be!ow. lf this local cataloguing service receives the support of a number of libraries, BNB anticipates that the central processing costs of a catalogue of approximately 300,000 titles with 10,000 annual intake will cost "not much more than f5,000 ayear for a regularly updated author/title and subject catalogue including local input" (unless local input exceeds more than about 20% of the total intake). This assumes that the catalogue is output in the form of COM. 4 An automatic subject index/subject heading system is being developed to provide for the automatic creation for any local library of a subject approach to its book stock. This will be produced primarily in the form of a subject index to a OC class number. The possibility of creating an alphabetical subject headings catalogue from the same data is also being considerad. The 'authority file' index will be constructed at the leve! of the schedules in OC 18th edition. lt is appreciated that this will not provide for totally specific indexing but experiments have shown that it would be capable of providing a reasonably adequate index to any classified catalogue based on the OC 18th edition. Because of its dependence on BNB's subject files, this facility would not be available for use with local computer systems. However, any library based on OC 18 could make arrangements with BNB to produce a subject index of its stock, even though its catalogue is produced elsewhere. 45. The success of these services remains to be seen, but it is clear that the new British Library will continue to experiment with new serví ces and options that will attempt to iessen the burden of computer provision in libraries, especial! y public libra ríes. lt is also worth mentioning that the British Library will provide a valuable pool of expertise on the use of computers in libraries. lt should also serve as a powerful co-ordinating influence on the development of local, regional and national systems. BRIMARC- The BNB/Brighton Public Libraries Catalogue Project 46. BRIMARC has provided a practica! illustration of how centralizad cataloguing services might be used by a library that did not want to maintain a local computer-based catalogue system, and it was to test this possibility that OSTI originally funded the project in 1972. A detailed account of BRIMARC is given in ref. 16 but it is worth summarizing here the essential features of the project. 47. Brighton Public Libraries, with an annual intake of 30,000 vol u mes (1 0-12,000 titles), had previously maintained individual card catalogues in the central library and eight branches. As from September 1972, no new titles were added to these catalogues, except for music and local collection books. New acquisition were notified to BNB, which produced a union catalogue in COM form*, cumulative updates being issued monthly. The new catalogue is in three sequences: classified, author/tit!e index, and subject index. The presentation of the author/title and subject indexes is similar to that of the BNB Weekly List, except that it is on microfilm. The subject index takes the form of PRECIS index 2 • Multiple copies of the film are suppiied for use in different locations.

* COM: Computer Output Microfílm-see Chapter 5. 13

..... ,.... .................. ~ .. .. .,e Luuv u• wuuuy ana HKely to be retained, it can be described in the present tense). All new acquisitions, with the exceptions mentioned above, are divided into 'standard' and 'local' input. Standard items are those for which records exist on the BNB/MARC files; local items are those for which no such records exist. For standard items (about 80% of the intake) a simple form is completad, showing ISBN, a local accession number, list of locations, and the author and title (for checking purposes only). The ISBNs are input to the BNB computar system to draw complete bibliographic records off the MARC file and to add these records to the BRIMARC catalogue.

49. 'local' items are catalogued and classified on local input forms which are sent to BNB for keypunching, these records then being also added to the BRlMARC catalogue. 50. The relativa costs of the manual and BRIMARC systems, expressed as index numbers, are shown in the following table:

Manual system

Direct costs/title (100=46p) Direct costs/entry point (100 = 24p)

BRIMARC system Brighton BNB costs costs

Total costs

100

83

54

137

100

25

17

42

51. Although the cost per title is apparently higher for BRIMARC, the cost per entry (in the author/title and classified catalogues, and the subject index) is much lower. As pointed out in ref. 16, however, these figures are slightly misleading, in that (a) they show no allowance for the fact that BRIMARC provides a union catalogue, which the manual system did not, and (b) they relate to a prototype system using available BNB software, with considerable scope for improved performance. 52. Average delay between completion of an input form and the incluslon of a title on the microfilm output was (in 1972) about 6 weeks, comparing favourably with 6-9 weeks for the manual system. This was under experimental conditions, and BNB are confident that the production delay could be cut by at least 1 week. 53. During the course of the original project, the Brighton computerized circulation system was unaffected, but it was intended that ata later stage, the BRIMARC system could be used to update the stock register used for printing out overdue notices. Birmingham Libraries Cooperativa Mechanization Project (BLCMP) 54. BLCMP, like BRIMARC, began asan experimental project but has now become an established system. lt is different, however, in that it is essentially a regional system, and it has involved cooperation between libraries of different types. 55. The project originally involved the libraries of Aston and Birmingham Universities and the Birmingham Public librarles (Birmingham Polytechnic was due to join the BLCMP scheme in April1974). The aim was to design and 14

records in local situations, and to assess the practicability of a regional data bank, accessible to a number of libraries, using these records and records produced locally. BLCMP was established in 1969 with the provision of an OSTI grant, and the cata!oguing system became operational early in 1972. 56. Early studies showed there would be advantages, in terms of running costs and hardware availability, in using the lB M 360 installation at the Birmingham lB M Data Centre, rather than one of the University computers. In 1974, however, dueto cost increases imposed by lB M, the system was transferred to Warwickshire County's IBM 370/145 computar. 57. The system maintains a backfile, or Potential Requirements File (PRF), of MARC records from BNB and LC which in 1973 covered about 100,000 items, and which is being added to regularly. This file is searched weekly"for catalogue records of new English language acquisitions by the libraries. These records are copied to the BLCMP data base, together with locally created records in MARC format for items not found in the PRF. The system provides different forms of output according to each library's requirements, sorne using COM, others using hard-copy book form, and one using cards printed by the computar. 58. Delay between receipt of a book and appearance of an entry in the catalogue is 5 weeks, provlding a MARC record exists. This is an improvement for Birmingham University Library, but not so good for the Public library. The latter, however, is the greatest beneficiary in terms of MARC record use. 59. Apart from the monographs cataloguing system, BLCMP has established a data base of serials records, and plans to extend its coverage to music. 60. Having established their system, BLCMP now aim to develop a range of services that will be attractive to public and academic libraries in the W. Midlands. These facilities might include participation at low cost in a large union catalogue; or machina processing for circulation systems, as alternativas to full participation in the cataloguing system. lt is possible that the software developed by BLCMP will be made available to other regions through the British library. 61. Severa! published accounts of BLCMP exist, but a good starting point for further reading is ref. 6. London and South Eastern Library Region (LASER) 62. LASER's history of computar use for providing facilities to member libraries goes back to 1970, when with BNB they p!anned and designad a system to produce a union findinglist, showing only ISBNs of BNB serial numbers and the relevant locations. This system is now well established, and the COM copies of the list, with updating supplements, are received on a subscription basis by most of the member libraries, which use them for direct interlending. At the time of writing, all but one of the Regional library Bureaux have now followed LASER's example. 63. The system is, however, limitad in scope to Englishlanguage books published since 1950, and much of the work of LASER relates to other material (amounting to about one million entries) that was recordad in its older manual catalogues. These records are now being edited to comply with Anglo-American Cataloguing Rules 1967 and to include

the fnternationaf Standard Bibliographic Description (ISBD), and are then being converted into machine-readable form. At the same time, BNB is converting afl its book records prior to 1970, so that an extensive machine-readable file will be available for English-language books published prior to the inception of the MARC cataloguing service. 64. The feasibility study for this conversion programme was supported by OSTI, but the conversion itself is being funded by U\SER members, who are contributing about [75,000, and by the British Library Lending Division, which is contributing a further [1 0,000. This figure of f:85,000 would have been much higher had BNB not decided to computerize its back files. Because of LASER's assistance in the editing of the entries in BNB, LASER libraries will have free use of the BNB file. On the completion of the operation, LASER libraries will receive a microform of the fufl LASER file (including BNB material) and in addition a sub-file sorted by their own library code number or numbers. The latter will be either in magnetic tape form, to which libraries can add their own local data, or on microform. Severa! libraries outside the Region have expressed interest in the LASER tapes, which will cost about f:3,500. As mentioned elsewhere in this report, this data base could be of great value to any public library wishing to computerize an existing catalogue. 65. A more detailed description of LASER's work is given in a recent paper by Bourne (ref. 4). Commercial Computer-Based Services for libraries 66. Various companies offer computer-based services to libraries. Book suppliers such as Bumpus and Maxwell,

Richard Abel, Bro-Dart Corporation and Blackwells afl offer a notification service of books published in specified subject areas. In sorne cases the notification may be in the form of a computer-printed arder form or catalogue card, with the layout and classification to the customer's requirements. These services are usually free, on the understanding that if the library decides to acquire the books, it purchases them from the supplier concerned. Sometimes, such services are linked with an approval system. 67. Standard catalogue cards produced by computer from MARC records or other sources can be obtained from the Library of Congress, Catalog Card Corporation, and Richard Abel Ltd. In the USA, severa! organizations will produce catalogue cards from the customer's input. In the UK, one commercial company (Oriel Computing Co.) offers this service. 68. Oriel Computing Co. also offers a computer-based cataloguing service. Participating libraries provide keyboarded data, and Oriel undertake the computer processing, including output in the required form. 69. Services of this kind are likely to increase in number and scope, but as pointed out in the Wainwright report 30, sorne caution has to be exercised befare adopting such services for important library functions. In particular, there should be adequate guarantees for the long-term continuation of the service, and limitations on the extent of price increases. In the case of catalogue processing, it would be wise to ensure that the machine-readable file would be available for transfer to another system, if necessary.

15

70. There are basically three strategies that can be followed for the introduction of computer techniques into a library: it can adopt a piecemeal approach, converting individual operations one at a time; it can go directly for a fully integrated system; or it can work towards the latter progressively, using a 'planned instalments' approach. This assumes, of course, that the library is going to use the computer for more than one application, and there are sorne who would not accept this assumption - those who have been able to justify a computer-based circulation system, perhaps, but cannot support a case for any other application. In general, however, a kind of chain reaction tends to operate. A library may start with a minimal circulation system, but to enhance this needs machine-readable book records in order to produce meaningful overdue notices, so it goes in for a computerized stock record or catalogue. Having got this far, there is a temptation to consider computerized acquisition, so that book records can be created and used at an earlier stage. 71. There will also be a tendency to mechanize severa! applications, rather than a single one, when a number of library authorities are brought together by reorganization, and have to establish compatible systems. Another kind of motivation to consider full use of computer techniques arises where the library has a particularly well-integrated set of manual systems, and believes that to computerize a single operation will result in a less efficient organization as a whole. Taking all these possibilities into account, it seems likely that within a few years, libraries with only a single computerized application will be in a minority. 72. The piecemeal approach has severa! advantages: Effort is concentrated into the most urgent of costeffective areas; Greater availability of existing software; Ease of implementation; Greater efficiency in each application; Unnecessary mechanization of satisfactory manual systems is avoided. 73. There are also dangers in this approach, the chief one being lack of compatibility between systems for different applications. A particular problem here is the need for a book numbering system that will be compatible for all eventual applications. 74. The essential characteristic of an integrated system is that data files are interlinked in such a way that deietions, additions, or changes to one file automatically actívate changes in related files. Thus keyboarding effort is kept to a mínimum and better control over the files is maintained. 75. lntegrated systems in public libraries usually invo!ve linking ordering and acquisitions with cataloguing (as at Cheshire County), or linking cataloguing with circulation (as at Oxford City), or linking all three. lt is worth noting

16

that such systems can also take advantage of input from regional or national services, as in the case of Brighton. lntegrated systems can be operated either on-line or by batch processing.* 76. The extent to which certain data elements can be common to severa! applications is illustrated by Table 1. The potential berefit to be derived by sharing data between different operations can be illustrated by the following example. A circulation system needs only a small part of the cataloguing data to inform a reader which book is overdue, but the savings effected by not keyboarding the common 60 characters of a title twice for, say, 1 00,000 titles represents the effort of one keyboard operator for one year. TABLE 1: Bibliographic elements used in more than one activity Sefection (from MARCar other externa! tapes)

SBN Class number Author(s) Title Publisher Subject heading Accession number

Acquisítion Catalogue

+ + + + + + +

+ + + + +

1

I r

-¡-

+

1

-¡-

+

Cirufation

+ +

77. The important advantages of an integrated system are: lncreased overall efficiency, especially by reducing duplication of input and output; Assured compatibility between various parts of the system; lmproved service to users; Greater scope for achieving staff savings; Better scope for the provision of mangement information. 78. The main disadvantage is the size and complexity of the task of designing and implementing an integrated system. There are examples of such systems being designed for one library and then adopted by another (e.g. Wa!sall is to use the Oxford City system), and there are commercially produced general-purpose packages (such as the Dataskil one) ; but generally it is true to say that there is less likelihood of a library finding a package of this kind that suits its requirements than of finding acceptable

* These terms are explained in

Chapter VIl.

software for a single application. And even where an existing system can be used. the cost in terms of money and effort of such a majar change in the operation of a library is considerable. Beca use many of the advantages of an integrated system are intangible, it tends to be more difficult to present a convincing caseto justify the expenditure involved. 79. Individual operations such as circulation can now be mechanized with relatively little expert help, but an integrated system requires extensive technical support at al! stages. Most important of a!L it requires an extremely able

librarían to mastermind such a majar reorganization of procedures, and to make it work. 80. As we said earlier, the trend is almost certainly towards the mechanization of more than one operation, especially given the larger size of the new library authorities. The dangers of an unplanned piecemeal approach cannot be overstressed. The best strategy for many libraries is probably a compromise between the two extremes, that is, to design a system that allows for the integration of operations, but to implement it in planned instalments.

17

81. ln this chapter we shall discuss first the three main areas of application, ordering and acquisition, cataloguing, and circulation; and then the ways in which these can be combined in integrated systems. In each case we shall review the possib!e features and options that can be built into these systems, and where appropriate, describe the input and output techniques that can be used. Ordering and Acquisition 82. Under this heading we should first mention the possibility of using a computer system to aid book se/ection. This has not been adopted by any public library, to our knowledge, and has only been applied in academic and special librarles in association with other operations such as ordering or cataloguing. The basic idea is that the computer should search a file of newly published books and print out a list of those that match the library's requirements. In al! cases that we know of, this imp!ies selection according to specified subject interests.

Aldermaston's AMCOS system 9 , for example, lists books from a MARC tape which are of potential interest to the library; these are selected on the basis of a 'profile' of Dewey Decimal Numbers. Queen's University, Belfast, and Trinity College, Dublin, have similar programs. Aslib 3 1 investigateá book seiection from MARC tapes, searching other information fieids such as titles, Library of Congress Subject Headings and Precis indexing terms, but found that unless the processing was carried out for a large number of libraries, the unit costs would be very high. 83. Turning now to computer-based ordering systems, it can be seen from Tab!e 2 that the number of such systems in operation is relatively small, and that the majority are integrated with catalogulng systems. The reason for this is presumab!y that public libraries tend to have less prob!ems in this area than in others, especially where they have good working arrangements with booksellers. A case in point is Birmingham Public Library, which is unlikely to use the

TABLE 2: Ordering/acquisition systems

Library

Berkshire (modified Kent County System) Cheshire Dorset Huddersfield Kent Shropshire West Sussex Westminster, London Borough of,

Computer used

fntegrated with cata!oguing

lCL 1903A

no

IBM 370/145 ICL 1904A ICL 1902 lCL 1904A ICL 1902S IBM 360/30 IBM 370/155 ICL System 4/50

y es y es y es no y es y es y es

" 1971 figures on which decision was based.

automated acquisitions system to be set up by BLCMP because of the quick service it obtains from a local book supplier: there would barely be time for an order record to get on file before the book was receipted. Another factor is that the benefits of a good ordering system have less immediate impact on the quality of service than do cataloguing and circulation. 84. The conditions under which a computerized ordering system becomes most attractive are as follows: a Decentralized selection by branches with centralized purchasing and accounting, where it is desirable to merge orders, for multiple copies of the same title, orto avoid excessive duplicate ordering; 18

Numberof titles/volumes annual!y in computerized systems 64,000

Year operational

On-fine

1972

no

10,000-45,000** -48,000 26,000 149,000

1971 1973 1973 1971

y es no no no

20,000-69,000* 90,000

1970 1973 1970

no y es no

'"' adult non-fiction only.

b Complex accounting requirements;

e Commitment to an integrated acquisitions/cata!oguíng system. 85. Because the justification for a computerized ordering system has tended to arise from special local conditions, the few systems that exist differ widely in their method of operation, even though the outputs are similar. Two cases in point are the Kent and Shropshire County systems. 86. The Kent system 14 derived its initial ímpetus from the County Supplies Department, which buys books for use in schools and colleges, and had decided to adopt a computer system to overcome the problems of collatíng orders to

booksellers, and redistributing the books when received. lt was suggested that the two systems should use a common main file of a!l British books in print, from which records can be drawn (by SBN) to create orders. These records have a simple fixed-length structure, and contain the SBN, an author/title statement restricted to 40 characters, the published price, trade discount, and various indicators. The file is updated from MARC tapes, and data punched from books received. 87. To order a book, the library completes a requisition form, indicating branch, bookse!ler, and other control codes, together with the SBN and number of copies required. These details are input to the computer on paper tape. After validation of the data, orders are automatica!!y printed out on appropriate stationery, and sent to the bookse!lers. lf the orders are not fulfilled, reminders are issued at 6-weekly intervals.

88. When books are received, details are input to the system, which then prints the necessary payment cheques, makes a debit entry in the accounts, issues reminders for items outstanding, prints out figures of expenditure to which librarians are committed and which they have incurred, and credits the accounts for cancelled items. Facilities for integrating this with a cataloguing system have been incorporated. 89. The Shropshire ordering system 19 is very much a part of the cataloguing system. A basic catalogue entry is typed on a tape-typewriter, on which a conventional order form is then produced. A by-product paper tape from this operation is fed into the computer to up-date an 'on-order' file. The computer system wi!l subsequently generate reminder letters to suppliers for overdue items; a monthly report of expenditure analysed by area and book fund; and a quarterly analysis of expenditure and numbers of books added to stock arranged within specified classification headings. The system also deals with standing orders. 90. The on-line integrated systems operated by West Sussex and Cheshire both handle book ordering, taking advantage of their capability to maintain up-to-date records of books at various stages in the ordering-acquisitioncataloguing process stream. These systems will be described in a later part of this chapter. Features and Options 91. The records associated with ordering procedures and accounting are numerous and complex, so the list of potential features for a computer-based system is a long one: Ordering: Generate order form content, where possible (e.g. from a master file, as in the Kent system, or from a catalogue record of a previously ordered title). 2 Print purchase orders to suppliers. 3 Print bid lists or generate tria! title lists for the library selection committee.

4 List items on order.

5 Provide wants iist. Receiving: 6 Control items received. 7 Notify requesters that their book has arrived.

8 Control overdue items from suppliers (including standing orders). 9 Generate automatic follow-up notices. Payment and accounting:

1 O Fund allocation and accounting. 11 Print and control payment cheques to suppliers. 12 Determine optimum date to pay suppliers, in accordance with terms on the purchase order. 13 Supply appropriate information to accounts department. 14 Control expenditure in relation to budgets. Input to cataioguing:

15 Generate input data to the cataloguing system. 16 Provide new accessions list. Management: 17 Provide analyses of expenditure in relation to branches, areas, book classification, type of book etc.

18 Provide statistical analyses of bookseller performance. Input Techniques 92. Input techniques are described and compared in the section on cataloguing, later in this chapter. As was indicated by the two sample systems mentioned earlier, the book details required for an order can be keypunched as part of the ordering procedure, or taken from an existing machine-readable file, such as a MARC file. 93. Beca use of the need to keep files current, reliable data preparation facilities are particularly importan! in an ordering system. For this kind of application, there may be a strong case for having the data preparation equipment operated by the library to ensure speed of turn-round. Computer Requirements 94. Again, because of the need for prompt processing of orders, and up-to-date reporting on the status of ordered items and expenditure, it is important that the library should have guaranteed regular access (weekly, or preferably more frequent) to the computer. Processing should be carried out within 24 hours of receipt of data. 95. The complexity of the processing is such that the computer installation should have at least four tape drives or one disc drive, so that at least four separata files can be handled concurrentiy in the same program. Cataloguing 96. Under the heading of cataloguing it is necessary to consider not only the main catalogue to which readers will have access but also various subsets or extracts such as bibliographies and reading lists and other types of list or record that are needed for administrativa purposes, such as an accession register. Also, we have to take into account un ion catalogues as well as the catalogues of individual libraries. 97. With a conventional manual system, all these different files and records have to be produced and updated separately, which can entail a considerable burden of clerical effort. One of the main attractions of a computerbased system is that once the initial catalogue record has been prepared, all these necessary outputs can be produced automatica!ly - providing that the system is designed to do so. 19

------.--~--~-- u• ........................ ~uu1~ .;:>yo:>L1:0'JH, tt.llldY ut: useTUI

IO

review the purposes of a catalogue. Apart from its inventory function it may be required to show whether a specific book* (identified by author, subject title, ISBN, or other means) is in the library, or whether any unspecified book or books on a given subject or by a given author, are available. In either case, the catalogue should also indicate the location of the book(s) within the library. In a mechanized system, the catalogue may be linked with other functions. For example, it may be integrated with the ordering and acquisition procedures, or it may be used in conjunction with a circulation system for the production of overdue notices.

99. Because of these multiple requirements and the many optional extras that can be specified, there is a temptation to make a computer-based cataloguing system very complex. The librarían may see the introduction of a mechanized system as a once-in-a-lifetime opportunity to revolutionize his cataloguing procedures and incorporate every possible attractive feature into the final products. This, of course, can be dangerous un!ess unlimited technical {and financia!) support is available in terms of systems design expertise, programming effort, and computer availability. A more healthy attitude is to regard a change of system as an opportunity for re-appraisal of requirements and ruthless qiscarding of non-essential features of the previous system. Union Catalogues

1 OO. One of the key attractions of computerized catalogues in the public library field has been the possibility of producing and maintaining a union catalogue, with copies in every branch, without the problems of multiple filing and updating that such a facility would have entailed in the past. Among existing applications, the trend has certainly been towards union catalogues rather than catalogues for single libraries.

1 01. Virtually ali of what follows in this chapter is applicable to either approach. The main differences between the two kinds of system is of course that the union catalogue has to show multiple locations of the same title and for reasons of economy more abbreviated records are generally used. There is no great technical difficu!ty in producing both a full catalogue for an individual library or librarles, and a union catalogue with mínima! details of author, title and classification. Features and Options 102. Two major factors to be considered in designing a cataloguing system are the method of input and the form of output_ and these will be discussed separately later in this chapter.

103. Another key decision to be made is whether the system should operate on-line or in batch mode. This theme wi!l recur in later sections on integrated systems and modes of computer provision. in the specific context of cataloguing, the use of an on-line system for this application alone, in a single library authority, would be very hard to justify. The real cataloguing requirements of a public library can be effectively met by a batch system. An on-iine system offers additional convenience to staff, better control, perhaps better service to readers (better than necessary ?), but at very high cost - at least, if the real costs of computer processing are taken into account.

* Or music, gramophone record, cassette, etc. 20

1 U4. 1he question considered here is whether the library should run its own catalogue production system, using a local computer and its own, or purchased software, living with its own problems and hopefully fulfilling its own performance criteria; or should it use an externa! service, maybe not ideal for its purposes (in terms of speed of updating, for example), but dispensing with most problems of implementation and operation.

105. The services available now and in the future have been described in Chapter 3. The most attractive of these at presentare BRIMARC, which has been successfully used by Brighton, and will probably become a regular service of the British Library; and future regional cooperativa schemes based on the example of BLCMP. In either case, a library authority on!y has to notify the central agency of the SBNs of the books it acquires, or submit a locally-prepared catalogue record in cases where an SBN is not available; the agency then provides the catalogue in whatever form is required. 106. This way of taking advantage of computer techniques appears so attractive that one might wonder why any library should undergo the trials and tribulations of running its own system. Schemes of this kind also seem to be in keeping with the trends of the times- cooperativa working, centralizad processing, and contracting for specialized services. There are, however, for the present at least, a few p;oblems. The first of these has already been mentioned: timeliness. 8oth the BRIMARC and BLCMP systems depend on the use of MARC records, the production of which imposes delays in the appearance of a catalogue. Secondly, these systems have not yet been developed to cover non-book materials such as gramophone records. Both of these problems are likely to be relieved in the near future to meet customer demands - certainly, vigorous attempts are being made to speed up the generation of MARC records. 107. The timeliness factor would beco me particularly critica! if one tried to use a central agency service as a component of a larger, integrated system. For example, if a circulatíon system requires an up-to-date stock register in order to produce meaningful overdue notices, delays in getting machine-readable records from the central service might be very inconvenient.

108. Lastly, there is the problem of cost. Many public libraries with computer-based systems at present enjoy the benefit of computer processing and systems support at bargain prices from their local authority. In at least one case, the library is positively encouraged to use its local authority computer by a procedure whereby all departments are debíted for a share of the running costs, on the basis of the number of staff employed. For such libraries, the attractions of an externally-produced catalogue are marred by the fact that they will be faced with a real, and possibly quite significant, bill for services rendered! Coverage of the Catalogue

109. As can be seen from Table 3, the majority of existing cataloguing systems cover non-fiction material only, although in severa! cases there are plans to extend the catalogues to inc!ude fiCtion, end possibly gramophone records. The tendency to limit coverage is obviously a function of cost, coupled with user needs. Short-entry catalogues or stock registers of fiction as we!l as nonfiction are needed, as mentioned befare, to produce meaningful overdue notices from círculation systems.

11 O. The piecemeal approach to catalogue conversion is eminently reasonable- costs, and administrative and technical difficulties can be spread out over a period of time. On the other hand, it is importan! that the system should be designed from the outset to cope with all the types of bibliographic records and output formats it will be called upon to handle.

111. Thought must also be given to the need for producing separate catalogues to parts of the collection. lf necessary, separate printouts could be arranged tb cover special subject collections, local history collections, or a music collection, for example. Another option would be to generate, from a union catalogue, separate catalogues for each branch.

TABLE 3: Cataloguing applications in public libra ríes (operational in June 1973) Local authority

Computer

lntegrated

Form of output

Barnet, London Borough

ICL 19025

with circulation

COM

Bedfordshire

Honeywell 125

no

COM

y es

Adult non-fiction

Berkshire

ICL 1903A

with acquisitions

COM

y es

Adult non-fiction published since 1960

Birmingham

IBM 360/40

no

Book

y es

All post-1971 material (part of BLCMP)

Bournemouth

Honeywell 200

with circulation

COM

y es

Brighton

(IBM)

will be

COM

BRIMARC system

Camden, London Borough Cheshire Cornwall Coventry

ICL 1903

no

Xerox reproduced

IBM 370/145 ICL 1904A ICL 4/72

with acquisitions no no

Dorset Flintshire

ICL 1904A IBM 370/145

y es no

Greenwich, London Borough Harrow, London Borough Hounslow, London Borough Huddersfield Kingston-upon- Hull Luton

ICL 4/70

no

COM COM la beis line printer COM COM and line printer subsets line printer

Present system being rewritten On-line input Non-fiction

IBM 360/30

no

COM

IBM 360/30

no

line printer

ICL 1902 ICL 1904E ICL 1900

y es with circulation no

COM

ICL 1900 IBM 360/30 ICL 4/70

with circulation with ordering no

IBM 370/155

y es

ICL 4/50

with acquisitions

Oxford Shropshire Southwark, London Borough West Sussex Westminster, London Borough

Points of access for retrieval 112. A computer-produced catalogue offers great advantages over its manually-produced counterpart in the ease with which different types of listing may be produced and multiple entries incorporated. In this respect it can be justly claimed that the computer makes it easier to achieve the purposes of a catalogue, as we enumerated them at

Software available

Comments

Non-fiction

y es y es

y es

Adult non-fiction (11 0,000 titles) Non-fiction

y es

Adult non-fiction

y es

Non-fiction

Non-fiction pilot project line printer COM photocopy reproduction computer typesetting COM

y es Adult non-fiction On-line Non-fiction

the beginning of this chapter. Once a catalogue entry has been put into machine-readable form, it is possible to generate author, subject, classified, and title listings with no additional human effort. The greatest part of the cost will derive from the initial record creation and keyboarding. The cost of additional computer processing required for generating more than one output will be relatively small. 21

----·-v...,•••~ ""1..,.'-"""'IIV UIC'IC'

111usuy rouowea traaitional practice: author and classified catalogues, with a subject index to the latter. lt is worth noting that W. Sussex have taken the computerproduced subject index of another library (Barnet) and adapted this for their own local needs, rather than compile their own from scratch. Many alternative approaches to subject retrieval have been tried in academic and special libraries, and there may well be a case for considering sorne of these in the public library context. Campey8 has reported on the many forms of computer-produced index for which software is available.

"'""" 11avt:

Additionai Outputs 114. In the same way that multiple listings can be produced for retrieval purposes, other outputs can be generated for management purposes, such as an accessions register. Again, the need must be recognized, and the system designed accordingly. Similarly, provision can be made in the system for preparing selective print-outs of items by a given author, or on a given subject.

Form of Bibliographic Record 115. An absolutely minimal record, for a simple finding list. might comprise only the main author, title (possibly limited in length) and shelf mark, amounting to perhaps 180 characters. At the other end of the scale would be a full MARC record, comprising anything up to 60 elements, and an average of 780 characters.

116. The tendency in public libraries using computerproduced catalogues has been to adopt as simple a record structure as possible, often at the expense of elegant traditional refinements. 117. Obviously, this question of record format is fundamental to the design of the whole system. lt involves questions of what forms of output are required, costs, and compatibility (both within the system itself. and with regard to externa! systems).

1:, d

time-lag between receipt of a book and its full incorporation into the catalogue. With a computer system there is an additional delay (which may be greater or lesser than the other) between computer runs to produce the latest update of the catalogue.

122. Frequency of updating is usually determined by a combination of factors: cost, computer availability, and the librarian's judgement asto the need for a completely up-to-date catalogue. At one time, a large part of the cost related to the large volumes of line-printer output needed and their reproduction, but the advent of COM has alleviated this problem greatly. 123. The most commonly accepted arrangement seems to be one where the main catalogue is only updated and output once a year. This is usually supplemented by monthly updates, sometimes with a six-monthly cumulation of the monthly outputs. Filing rules 124. The way in which computers sort words into alphabetical arder is basically very simple. A numerical value is assigned to each character and punctuation mark, and a standard program is used to arrange records into ascending or descending arder. The sort program usually works on a limited number of characters at the beginning of a specified part of the record - say, the first ten characters of the author field.

125. There are variations in the character va!ues from one make of computer to another, so that the arder of precedence of punctuation marks is not standard. But so far as letters and numbers are concerned, the end result of a computer sort operation using utility software is a straightforward letter-by-letter filing order. This is a far cry from the logical arrangements that traditional library filing rules require. 126. To overcome this problem, it is necessary to provide or

118. The forms of output and their purpose determine what information is needed in the record. This may include author, corporate author, title, imprint, classification, ISBN, accession number, and location (within the main library, or within a number of branches). The more complex the record, the greater the cost of creating it, and of manipulating it to produce different outputs. A special need for compatibility arises where the catalogue is linked to a mechanized circulation system, as the equipment used will determine the kind of book numbering system that can be u sed.

119. Sorne public libraries have introduced special features into their numbering systems, in arder to obtain management information, and these affect their catalogue record format. Examples are mentioned in the section on Management lnformation later in this chapter.

Frequency of Updating 120. Although the computer can perform a task equivalen! to filing catalogue cards very quickly and efficiently it is not usually so immediately accessible to do this work as a library assistant- excluding for the moment the expensive possibility of an on-line system. And whereas cards can be added one at a time to a card file to keep it up-to-date, the computar can only show what it has done by printing out ail or part of the catalogue. Which brings us to the problem of updating frequency.

22

generate a sort key for each record, i.e. a string of characters which when sorted into computer filing arder will give the desired arrangement of the original records. At the very simplest leve!. this might be the first significan! word of a title, omitting punctuation marks and definite or indefinite articles. This sort key can be added manually to the record at the input stage, or it can be generated automatically by using rather complex programs. In a system using upper and lower case characters, the sort key will be all in one or the other.

127. A Working Party on Computer Filing Rules was formed in 1966 and reported in 1972 (ref. 35). lts report includes a draft specification of processing facilities for mechanized filing rules, and software for applying these principies has been developed by BNB and BLCMP. The feasibility of using computers to achieve filing arrangements of a high standard has been established, but the necessary software is not readily available for all types of computer. The processing required adds to the cost and complexity of the system.

Input Techniques 128. The choice of input technique is critica! for a cata!oguing system, as it is for any other; the computer is quite unforgiving if the data fed into it is not of the quality it demands, and if a wrong decision is made in this area the whole system can be wrecked.

129. The first decision to be made is whether input should be prepared in-house (within the library), in the computer department or through a commercial bureau. Each of these alternatives has its advantages and drawbacks. 130. lf the library gets its own equipment it can keep close control over the work and ensure quick processing, but it has to face the problems of staffing and supervision, equipment reliability, and in sorne cases, noise ( card and paper tape punches are worst in this respect). lt may not have a sufficient workload to ensure full utilization of staff and equipment; yet its punching unit is likely to be small, so that if one operator is away sick, a backlog soon bui!ds u p. 131. Using the facilities of the computer department is often the most attractive option, but the library's work may be competing for priority with other kinds of work; the equipment may not be ideally suited to the preparation of catalogue input; and the keypunching staff will almost certainly require proper punching sheets, whereas it may be possible for operators within the library to work directly from the books themselves or handwritten slips. 132. There is much to recommend the use of commercial bureau services, in terms of quality, accuracy and speed of turnround. The main disadvantage is that the cost may be apparently higher than that of the other options. lf OCR* is to be used, the necessary equipment is unlikely to be available anywhere but in a bureau. We shall return to the merits of bureaux when discussing retrospectiva conversion. 133. The next decision to be made is whether input is to be prepared on-line or off-line. In the UK at present, off-line techniques seem to be a natural choice except in the case of integrated systems, and the use of on-line methods would certainly imply an in-house operation - that is, the terminal would be linked to the local authority computer. lt is worth noting, however, that in the USA there are shared cataloguing systems such as the Ohio College Library Center (OCLC), where many wideiy separated libraries have on-line access to the computer. There is also the possibility of on-line systems using a mini-computer dedicated to library operations, to which we shall return in Chapter 7. 134. Now let us examine the various possibilities for off-line data preparation, bearing in mind that before considering their intrinsic qualities, it is essential to know what constraints are imposed by the computer installation. lB M instaliations, for example, tend to be punched-card oriented, while ICL installations often prefer paper tape input. Whatever input medium is used, the coding used (7 or 8 track paper tape, ICL, lB M or ISO etc.) will have to conform to computer requirements.

a Paper Tape 135. Paper tape input can be prepared either on a tapetypewriter, which will give a hard-copy record of what has been keyboarded, or on a 'blind' punch which generates paper tape only. The hard-copy output from the tapetypewriter can be used simply for visual checking, or it can be a functional record such asan order form (as in the Dorset County system). A blind punch is a simpler piece of equipment (and hence more reliable), and is cheaper than a tape-typewriter. lt is common practice to train typists to operate tape-typewriters, but a blind punch is more akin to a card punch, and will usually need a specialist operator.

* OCR: Optical Character Recognition.

136. Paper tape is attractive for cataloguing applications where variable-length records are employed. Features can be incorporated in the equipment to prompt the operator to key the elements of each record in the correct sequence and in the correct form - a valuable aid when keyboarding a complex record format. The character set available is usually larger than that on a card punch. The worst feature of paper tape is the difficulty of making corrections - al! of the original tape must be copied (often ata fairly slow speed) to prepare a new error-free tape, whereas with punched cards, individual cards can be removed from a file for re-punching. 137. Paper tape is prone to various faults when it is fed into the high-speed reader on a computer- faults which may not be apparent when it is read on a slow device such as a tape-typewriter. Parity failures, incorrect positioning of codes along or across the tape, even paper-tape of the wrong colour, all can lead to rejection by the computer of large volumes of input unless adequate checks are carried out beforehand. Oespite these problems, paper tape has up till now been the most commonly used input medium for library cata!oguing applications in the UK.

b Punched Cards 138. The great advantage of the punched card is its durability, which makes it less susceptible to problems when being read into a computer. There is no visual record of what has been punched, unless an interpreting punch is used, in which case each character is printed out above its corresponding cede in the card. The card punch has been improved great!y in recent years, thus maintaining the competitiveness of punched cards in the face of other input media. With modern equipment characters are keyed into a magnetic memory ( called a buffer) large enough to hold the contents of an 80-column card. Keying errors can be corrected befare the contents of the store are punched into a card, thus overcoming the disadvantage of older equipment where a single error meant rejecting an entire card. 139. The nature of an 80-co!umn card and the design of the punch make this medium more suitable for fixed-!ength records, but punched cards can also be used for variablelength records, although format control facilities for this purpose are minimal. 140. The usual procedure with punched cards is to punch the data and then re-punch it on a verifier in order to ensure accuracy. As with other input media, however, there is the alternativa method of keying the data once and then feeding it into the computer to make certain checks on the validity of the records, and at the same time to obtain a print-out which can be checked visually.

e Magnetic Tape 141. There are two basic ways of recording data on magnetic tape. One is to use a device which encodes direct!y on a computer-compatible magnetic tape, while the other is to make an intermediate record on a small tape cassette which can subsequently be transcribed onto computer tape by means of an off-line converter. 142. Typical of the first type is the M DS Mohawk magnetic tape encoder, which was used for conversion of the existing catalogue for the W. Sussex system. Another example is the Burroughs N9800 Encoder used by Flintshire County Library. Equipment of this kind has been developed to replace the traditional card punch, and there are obvious

23

_ -~ --·- ·--._.. .......... ....-. up l.V VV \..oiiOid\.;Lt:::f:) tlne ourrougns equipment has a 160-character memory) can be stored in a buffer memory before being transferred to the tape. Format control facilities for variable-length records are limited, except on sorne special machines. No visual record is produced, unless a tyepwriter or other slow-speed output device is attached to the encoder. The great attraction of this kind of equipment ís that ít records on a medium which is immediately acceptable to the computer, and can be read and processed at high speed.

143. The latest and most attractive type ot cassette encoder is the Olivetti DE 523, which is being used by Staffordshire County Library, and evaluated by BLCMP. This machine has a small screen on which the record format can be displayed, and the characters input are superimposed on this 'mask' lf the maximum length of a field is exceeded or other pre-set conditions are infringed both audible and visual warnings are given. As each part of the record (i.e. a 'screenful' of data) is completed, it is automatically transferred to the cassette, which can hold a total of 250,000 characters. 144. Correction facilities are excellent. So long as the input is displayed on the screen, the operator can go back to any point and correct it. lf necessary, the contents of a cassette can be played back through the machine and corrected where necessary, before it is converted to computer tape. 145. These machines are comparatively expensive, and in addition there is the cost of the cassette-te-tape converter (one of which could handle the output of severa! encoding units), but they are very well suited to the requirements of library catalogue input, especially with MARC-type records. 146. In selecting magnetic tape encoding equipment. care must be taken to ensure that the tape produced will be compatible with the computer used, both in terms of coding and density (bits per inch).

d Key-to-Disc 147. This technique is mentioned here for the sake of completeness, but we know of no current use in a iibrary application. The principie is that a number of keyboards are connected to a magnetic disc store via a small computer which identifies the data from each keyboard and controls the location of data on the disc. This has advantages for applications involving large vol u mes of input, in that the complex electronic equipment required to record and organize the data is shared by severa! inexpensive keyboard units.

so this part of the operation is usua!ly carried out by a bureau. The typing can, however, be carried out ín-house, providíng that a high standard of quality can be maintained, in terms of cleanliness, alignment, and clarity of typing. Procedures for correction involve the use of special characters which ínstruct the scanner to ignore the previous character, word, or fine; the correct data can then be inserted. 151. OCR is technically attractive, and has been found in severa! instances to be competitive in terms of cost. Success in using it seems to depend largely on choosing the right bureau and in laying down clearly-defined requirements. More detailed practica! advíce on the use of OCR is gíven in a paper by Robinson 28 • Retrospective Catalogue Conversion 152. Whatever technique is adopted for current input to a catalogue, a different approach may be needed where an existing catalogue or stock register has to be transferred to a computer system. This usually entails a very large vol u me of keypunching which if done by a small in-house unit can impose long delays in implementation, and generally cause probiems through over-straining the system. Among the successful systems we visited, most had engaged a bureau service to do this work. Both Brighton and Oxford, incidentally, used an OCR bureau for this purpose, and it is interesting to compare the different approaches used. Brighton sent their stock register in its original form (on 5" x 3" cards) to be typed for OCR. lt was then scanned and written to magnetic tape, and printed out. The print-out was edited, and the tape was corrected as necessary. Oxford took their old stock record (in sheaf catalogue form) and amended this by hand before sending it to the bureau for OCR processing. 153. A very low error rate (0·3%) was achíeved on test runs, and no checking was carried out on the records later supplied. Oxford claim that their success in using an OCR bureau was largely dueto the detailed advance discussions between their computer manager and the bureau, and to careful preparation which included the use of special stamps to show field limits on the input records. 154. The cost of an operation of this magnitud e can be considerable, and may be a significant factor in deciding on the record format to be used in the catalogue. The more characters per record and the more complex the record structure, the greater will be the cost; and even small unit differences are going to be mu!tiplied many thousands of times.

e Optical Character Recognition 148. Optical character recognition (OCR) has been successfu!ly used as an input technique in severa! library app!ications in the UK, including those at Brighton and Oxford City libraries, and the retrospectiva conversion work by LASER. 149. The ultimate development of this technique may one day make it possible to read any printed or written records directly into a computer memory, but for practica! purposes at the present it is necessary to type the data, using a typewriter with special!y designed characters, on an input sheet of pre-determinad size. The input is scanned at high speed by a device which can recognize each character and then encode it on magnetíc tape. Either upper-case, or upper and lower-case alphabets can be used. 24

155. Quite apart from the problems of keypunching, preparation and editing of catalogue records for conversion represents a daunting task. !n most cases, it seems, sorne modification to previous cataloguing rules and standards is introduced as part of the conversion operation. So me public libraries have found it impossible to undertake a task of this magnitude using only their normal staff resources, and have engaged teams of temporary helpers. Oxford, for example, took on five qualified staff (wives of the library staff) for 6 months to recatalogue 250,000 books. 156. The problems of retrospective conversion can be alleviated by using machine-readable records prepared elsewhere, such as the BNB-LASER data base, which is discussed in Chapter 3.

Recapitulation on Choice of Input Technique 157. Many factors have to be taken into account in choosing an appropriate method of data preparation, and the arder of importance of sorne of these factors was indicated at the beginning of this section. The logical sequen ce of questions to be considered should be as follows: 1 Who should do the work- the library, the computer department, or a service bureau ? 2 Should conventional off-line techniques be used, or is there a case for using on-line techniques? 3 Which technique can ensure adequate accuracy and reliability? 4 Which technical characteristics are most important for this application - character set. format control, facilities for visual checking, ease of correction, etc? 5 Which technique is cheapest?

158. Cost is deliberately shown as the last factor to be considered, because differences in cost between one technique and another are likely to be small in relation to the cost of trying to make a system work with the wrong input technique.

159. There is a certain amount of literature on the comparative technical merits and costs of different data preparation techniques, including the extensive tests carried out for the Library of Congress RE CON project 1 . In practice, however, decisions in this area must be strongly influenced by what is practicable and reliable in a given situation.

160. Finally, a warning. Whatever input technique is used, tests should be carried out at the earliest possib!e stage to ensure that the computer wil! accept the records produced, and sorne form of routine quality control on the product of any data preparation equipment is recommended.

Use of MARC and Other Externally Produced Records 161. Attitudes towards the use of MARC records for current input to public library catalogues tend to extremes. There is one school, of which Brighton is a prominent member, that believes a MARC-based system can be made to work, and regards the potential benefits in terms of cost, efficiency and standardization as too good to miss. The other school declares that the de!ay in production of BNB-MARC records (4-6 weeks) is unacceptable, and that it is essential for their catalogues to be up-to-date; sorne also feel that MARC records are needlessly long and complex for their purposes.

162. Cataloguing is an expensive operation, especially in its use of professional staff, and it does seem absurd that such work should be duplicated by many different libraries with quite similar collections. There is a strong probability that delays in preparation of MARC records will be significantly reduced, as both the Library of Congress and BNB are working towards cataloguing in advance of publication. Objections to the complexity of MARC records need not be taken too seriously; there is no great technical difficulty involved in extracting selected fields or reformatting to produce whatever kind of record is required by a library's own system. Sorne degree of pre-processing is necessary, anyway, in arder to add appropriate accession numbers and locations to the records as supplied. Problems would arise, however, if a library were to use a computer for which a compatible MARC tape was not available, but this situation wi!l be fairly rare.

163. Sorne proponents of the anti-MARC school place a high value on autonomy of operation, and no doubt feel that even if the use of MARC could show so me financia! economy, they would prefer to prepare their input in their own way. This attitude may be justifiable from a local viewpoint, in optimizing the system of one library or one authority, but is likely to inhibit the possibility of achieving effective interlibrary cooperation. 164. The various services available for providing MARC records were described in Chapter 3, but it may be helpful to review them again briefly here: 1 A library can subscribe to the complete MARC tapes and extract from them the catalogue entries it requires. This means maintaining a large file against which searches will have to be run to locate catalogue records for newly acquired stock- a very expensive operation for even the largest library authority. 2 A library can subscribe to the British Library's Selective Tape Service. This would entail sending to the BL a list of ISBNs relating to new acquisitions. The BL would match this list against their complete MARC file and transfer the matched items to a magnetic tape, which would be sent to the subscribing library together with a list of non-matches. The latter can be checked against later additions to the MARC file. This service relieves the library of the burden of maintaining a complete MARC file at the cost of a slight delay. 3 A library can join in a cooperative scheme like BLCMP, or subscribe toa central service like BRIMARC. In either case, it would submit ISBNs for new acquisitions, or locally-prepared catalogue records, and the agency would provide the finished catalogue. In the case of BRIMARC, on!y COM output has been offered so far, while the BLCMP scheme offers output in any desired form. Again, sorne delays are inevitable, but the library is relieved of all the technical problems associated with running its own computer system for cataloguing purposes. 165. The above facilities are designed primarily to deal with current processing, but could also be used in retrospectiva conversion, with the limitation that BNB-MARC records are available only for books published since 1969. Machinereadable catalogue records (AACR 1967, and Dewey 18 classification) for material back to 1950 are available in the LASER data base. Costing outlines for libraries planning a total retrospective conversion are available from the MARC office, British Library, Central Administration, Africa House, Kingsway, London WC2B 608. Forms of Output 166. Three basic forms of output for computer-generated catalogues are in use in public libraries in the UK : Computer-output microfilm (COM); Line printer output; Phototypesetting. 167. In sorne other libraries, notably academic libraries in the USA, catalogue cards printed out by computer are used. Such a practice wouid seem to have limited attractions in a public library, however, sin ce it offers no relief from the chore of card filing -a factor made all the more adverse by the high stock turnover. For the future, there is the further possibility of an on-line catalogue, that could be consulted by means of a visual display unit (VDU), but the cost of such a system in a form that the public could use would at present seem prohibitive. 25

______

~~··

................ .., ,._,, uvu• vvtuvu

a IJrt:tt:rence ror C:UJVI IS

apparent. Undoubtedly, the advent of this technique has played an important part in encouraging the use of computers for catalogue production.

aCOM 169 A COM camera accepts signals from a computer or from a magnetic tape and generates character images which are recorded direct on microfilm. The way in which the character images are generated, and the complexity of the equipment, can vary, which is why the facilities offered in terms of typographical quality can differ. Sorne machines can only produce an output equivalent to a computer fine printer- all upper-case characters, with no automatic control of line length. Others can offer a wide range of type fonts and sizes, variations in boldness, proportional spacing, automatic right-hand justification, and other graphic refinements. The first kind can accept a magnetic tape that would otherwise be used to drive a fine printer, the second may need a tape incorporating complex setting instructions or it may have sorne built-in facilities for enhancing output quafity. 170. The three great merits of COM are compactness, cheapness and speed. Local authority computer departments are unlikely to have their own COM equipment, so that work is usually put out to a bureau - often to the relief of the computer manager, who thus avoids the frequent and sometimes lengthy printing-out operations required for catalogues. 171. Bureau facilities for CO M output are widely avai!able, * but care must be taken to ensure that the bureau can produce film of high quality, so that optimum legibi!ity is achieved. 172. Apart from Shropshire, which uses fiche, all known public library COM catalogues are on 16 mm film stored in twin-spool cassettes, the most popular of these being the VSM F cassette shown in Fig. 1. A cassette of this kind, holding up to 2,000 frames, can very easily be mounted on or removed from the reader. The number of entries per frame depends on the length of record used and the layout, but sorne idea of acceptable density can be gained from the experience of Cornwall, which has about 30,000 records stored on 2,300 frames; Cheshire, which has 37,000 titles on 2,600 frames (including both author and classified sequences), and Flintshire, with about 30 titles per frame.

typographic qua!ity demanded - upper and lower-case output costs more than ai! upper-case, for example - and from one bureau to another. Cost figures quoted by CornwalF can, however, be regarded as fairiy typical : cost of their master silverfilm is f:7.50 per 1,000 trames; copies on Kalvar film cost 75p per 1,000 frames. 174. The biggest doubt so far as COM catalogues are concerned is, will the public accept and use them? We know of no pubiic librarían who has abandonad COM as an output medium because the public objected to it, but this may not be regarded by sceptics as positive evidence of acceptance. Certainly, in many libraries we have visited we have seen members of the pub!ic using the microfilm reader without protest or apparent difficulty. lt is important, however, when introducing this novel medium to the pub!ic to ensure that conditions are as propitious as possib!e. Good image quality is essentia!, with a larger size of type than would be acceptable for hard copy. The reader should be located in an area with subdued light, so that maximum contras! is achieved. Choice of reader is also important. The cost of readers is, of course, an item of expenditure which does not arise with traditional forms of output, but severa! COM users have claimed that savings achieved through using COM (as against line printer output) have been more than enough to pay for the readers. A review of experience in the use of COM has been published by NRCd 23, and more recently a report on the use of COM for library catalogues has been written by Spencer (also of NRCd).* 175. An acknowledged disadvantage of COM catalogues is that, unless multiple readers are provided, only one person can use the catalogue ata time. 176. The financia! advantages of COM are especially significant in the context of catalogue updating. The cheapness and ease with which this form of output can be produced makes it possible to have more frequent updates than would be practicable with other media.

b Line Printer output 177. Oxford City Libraries deliberately chose to use conventional computer print out for their catalogue beca use of the very high level of catalogue consultation by their readers 32, and for most other libraries where it is felt that COM would not be accepted, line printer output is likely to be the most practica! alternative. 178. The two chief drawbacks of line printer output are its typographic limitations and its bulk. Most line printers are limited to upper-case characters only (although there are exceptions), and the quality of the printlng in terms of alignment and clarity can be very poor if the equipment is not properly maintained.

Fig. 1 - VSMF Cassette

*A directory of microfilm services including COM bureaux has been published by the National Reprographic Centre for Documentation, Hatfieid.

26

179. The number of copies that can be produced in one pass is limited, so that Oxford, to produce 8 copies, print out their catalogue twice on 4-part stationery. (They indicate, however, that if as a resu!t of local government reorganization, further copies are found to be necessary, they will modify their system to produce both print-out and COM).

.,. Soencer, J. R. "An appraisal of computar output microfilm for Ubrary Catalogues'" National Reprographic Centre for documentation, 1974.

180. lt is worth noting that sorne libraries which began with paper output have subsequently switched to COM because of the prob!ems of bulk. Cornwa!l, for example, started with line-printer output on offset-litho plates, from which 30 copies were printed and bound in post binders. They found, however, that ''this method of reproduction was both cumbersome and cost!y". Flintshire also began with !ineprinter output, but after 3 years of operation, the cumulative catalogue, of which 24 copies were produced, presentad a formidable spectacle of a stack of paper 25 feet high. 181. Bu!k of output can be reduced - at a cost. One possibility is to photoreduce the line-printer output. Another (used by Camden) is to use a Xerox Computer Forms Printer, which reduces a page of output down to A4 size and makes multiple photocopies as a semi-automatic process. 182. Having referred twice already in this section to the Oxford catalogue, we feel it only fair to add that the end product is exceptional!y attractive in form and presentation. The print-out is kept in Series 1 Twinlock Binders, which are displayed (with boldly printed spine iabeis) in suspended filing racks on trolleys. By proper systems design, with the needs of the user in mind, many of the disadvantages of line-printer output can be overcome- except the difficulty of reproduction.

e Phototypesetting 183. The only current example we have seen of a public library catalogue produced by phototypesetting is that of West Sussex, although Birmingham Pub!ic Library was planning to use the same technique. The reason for this choice in the case of W. Sussex has been explained in a paper by R J Huse 21 • West Sussex has a high proportion (22 per cent) of borrowers over retirement age and many would be unable or unwilling to use microfilm readers. Local government reorganization means that the new West Sussex will have within its boundaries 41 public and 6 college libraries, each of which will need ready access to the catalogues - sorne major !ibraries requiring two or more copies. Eva!uation of computer typesetting produced figures which indicated that economies of scale could bring down the price to a leve! competitiva with other forms of catalogue, without heavy initial capital outlay. The master fHes are now sent to Computaprint Ltd where an RCA Videocomp is used to set and compase the pages, which are then transferred to !itho plates. Charges in 1973 amounted to 59p per page for p!ate-making (page capacity is about 5,500 characters), and :E1 ,800 for 30 copies of 5,200 pages. 184. The complete catalogues are printed out annually, with updating supplements at intervals varying from 2 weeks to 3 months for different parts of the catalogue. 185. The finished product is very attractive, easy to read and reasonably compact. The West Sussex catalogue is printed in upper-case only, but the Birmingham catalogue will be in upper and lower case. lssue Systems 186. The As!ib/CAG survey in 1973 revealed more or less similar numbers of public librarles with operational cataloguing and issue systems, but there is little doubt that in the near future the number of issue systems will show the more rapid growth. Motivation to use computer techniques for this activity is generally stronger, and such systems are simpler and hence easier to introduce so far as computer processing is concerned. Also there is an active sales effort on the part of firms that make computer book-charging equipment, to encourage more libraries to buy their wares.

187. A question frequently asked is at what point does it beco me more economic to use a computerized issue system rather than Browne? As with al! app!ications, it is well-nigh impossible to pin down the real truth so far as operational costs are concerned, but sorne concensus did emerge from this study. Brighton managed to cope with Browne for up to 900,000 loans a year, but they and others feel that if the decision were made on economic grounds, the !evei for change would be lower. Various public libraries have quoted figures between 300,000 and 600,000 per service point as the mínimum at which a computer system is justified. Figures below 400,000 depend for justification on factors other than pressure on service points. This justification is likely to be based on the need to 'trap' a title at the first branch to which a copy is returned. Thus a system !inking severa! or all branches is needed, although only the central library is large enough to have problems with Browne. The change to a computer system may be justified more easily if all circulation procedures within a library, covering all types of loanable material, are taken into account. We found little firm evidence that staff savings had been achieved through the adoption of computer charging. Oxford relinquished 2 staff out of a total of 18, but theirs is an integrated system. A more common claim was that staff had been released for other work. 188. All of the systems we shall describe have similar basic principies of operation. At each service point there is a device that can automatically read a borrower number and one or more book numbers, and thus create a series of transaction records on magnetic or paper tape. The same equipment is used to record discharges. The machinereadable records are fed into the computer which maintains a complete up-to-date transactions file, and produces overdue notices when the loan period is exceeded. System Functions 189. The requirements of an issue system vary greatly from one library authority to another, and even within an authority, between the centrallibrary and its branches. When setting up a computer-based system, the various possible features that can be incorporated should be carefully considered. All of the facilities discussed below can be achieved with a manual system, but often only at large expense in terms of clerical effort. Sorne can be added to a computer system at very little extra cost, but the temptation to add features for which there is no positive need has to be resisted; over-complexity can be fatal. Charging and Discharging 190. Essential requirements are that the system should be quick, error-free and reliable in operation, and the libraries we have visited all seemed satisfied on this score. Both the ALS and Plessey equipment (the two main contenders) are claimed to handle 600-700 transactions per hour, and one busy library (Merton) had found that the ALS equipment could deal with up to 450 transactions per hour. West Sussex (who were pioneers in this fie!d) find that speed of charging is about the same as with their previous manual system, but that discharging is quicker. 191. Validity of the book and borrower numbers can be assured by safety measures such as parity checks and the use of check digits, which can be built into the equipment at the service point. 192. Experience with Plessey, ALS and Olivetti equipment has shown them all to be reliab!e in operation. Problems can arise, however, due to overheating and to voltage surges.

27

•

----·..,···~ .... 1 ~-. .... •u.:J vall lldVt: li:ICIJI"[leS

for recording renewals, a small keyboard being used to register book and borrower numbers, if the renewal is requested by telephone. Where trapping stores are used, the system gives an automatic warning if an attempt is made to extend the loan of a book which has been reserved.

Overdues 194. A computar system offers clear advantages in its ability to prepare prompt and regular overdue notices - an otherwise fabour-intensive operation. For a mínima! system, al! the computer needs in addition to the transactions file is an up-to-date file of borrower names and addresses finked to the appropriate borrower numbers. lt can then generate notices showing the identity numbers of overdue books, and this is the way that sorne libraries (such as Merton) operate. Others feel that a more meaningful notice is essential, showing authors and titles. Todo this, either the complete catalogue must be in machine-readable form ora simpler kind of file, such as a stock record, can be used. Sometimes a special file is extractad from the main file for this purpose; in the case of Oxford, for exampfe, this file contains the first 12 characters in the author fieíd and the first 50 characters in the title fiefd from each catalogue record. lt is worth noting that sorne libraries (e.g. Brighton and West Sussex) started off their issue system without the faciiity for printing author and title information or overdues, but added it fater. 195. A computer system makes it possibfe to introduce other refinements into the procedure for sending overdues. Camden, for example, empfoy a points system so that notices are onfy sent if, say, three books are overdue for a long period, but not for a single book.

Reservations 196. Reservations are usually handled not by the computar but by a special device called a 'trapping store' which is connected to the charge/discharge recording units. This consists of a memory, to store the numbers of reservad books, and a logic unit which compares these against each number input at the serví ce point, and activates a warning signa! when a match occurs. Numbers are entered into the store, and erased when necessary, by means of a small keyboard. 197. The trapping store is a fairly expensive device, but it is possible for one to be shared by severa! branches, the necessary !inks being provided via low-cost data transmission facilities (J Unes) provided by the Post Office. Brighton uses this technique. 198. Libraries which have adopted the trapping store technique have been well pleased with the results, especially in that they are relieved of the clerical chores involved in the equiva!ent manual systems. Sorne, however, find this feature too expensive and retain their visible index systems alongside the computer charging equipment. 199. West Sussex, with their integrated system, can have on-line access to their transactions file (for branches using computar charging) but use this main!y as a means of location, not as a trapping store. They use a conventional visible index for reservations, but if this has not been effective after 5 weeks, they track down a copy vi a their terminal, notify the branch con cerned, and then advise the borrower.

limitations on Borrowing 200. The Browne system and token charging impose a !imit on the number of books borrowed, but photocharging 28

rncorporated in a computer-based system, if necessary, and Camden for example do this. A fairly simple program is required to sort the transactions file by borrower, count the number of items charged to each, and list those that exceed a specified !imit. Because of the size of the file, however, this can be a time-consuming and fairly costly operation. 201. A trapping store can also be used to detect specific borrowers, such as those who have been guilty of overborrowing, or who have left property in the iibrary. The identity number of a borrower can be stored in the same way as a book number, and the equipment will give a warning signa! if that borrower's card is read by the charge/discharge unit.

location of Stock 202. We have mentioned the West Sussex on-line facility which enables them to ascertain the location of any book. The same can be achieved, though not so quickly, with any batch-processing system. This would simply require a program to search the transaction file for specific book numbers, and to print out their location: In Camden, a daily print-out is made showing the locations of books reservad on the previous day.

Borrower Records 203. As mentioned above, a file of borrower names and addresses has to be maintained for preparing overdues. This can be updated as often as necessary, and printed out from time to time, thus avoiding the need to maintain a card file. 204. In designing the system, it is worth considering what useful information could be derived from the borrower file. One possibility would be to incorporate a simple location code (based on a grid) in each borrower record, so that reader distribution in relation to branches could be plotted. 205. Developments toward a Corporate Data Base in local governments are relevant to the organization of borrower records. The intention is that the population element of this data base wi!l be fed by all of the departmental applications which are concerned with population, and although sorne of this information will be protected, a significant part wili be available for a!l corporate activities. In this way, the library could have access to more popluation information than is avai!able through its own borrower files.

Management information 206. Computers are very good at counting things, and can be used to generate unlimited quantities of statistics. Sorne operational statistics are regarded as necessary in most if not all public libraries, and those with even the simplest computer charging systems can obtain useful information very easily. Merton, for example, obtain weekly and 6-monthly statistics of issues, and an annual check on the number of books on loan; they can also make an automatic count of issues per hour. By transferring tasks of this kind to the computar system, staff can be re!ieved of sorne particularly boring duties. 207. Sorne libraries have far more ambitious schemes, though none of these was sufficiently advanced to see results and determine their usefulness in practice. Dorset, for example, plan to interrogate their transactions and catalogue files and to obtain analyses of location, class number, intellectual leve! and issue data, in order to provide management data on the performance of their bookstock. Kingston-upon-Hull have been experimenting with the use of meaningful book numbers that wou!d enable them to achieve similar analyses direct from their transactions file.

lnterlibrary Lending 208. The improved stock control that can be achieved with computer charging results in additional benefits, an importan! one being the better procedures for interbranch or inter!ibrary lending that can be established. On-line vs Batch Processing 209. Existing public library issue systems aíl operate in batch mode; the computer is used only once a day or at longer intervals to update the transactions fiie and to perform other processing. An ideal system would be one where the data capture devices were on-line to the computar, so that the file was always up to date, and where the files could be accessed at any time via a terminal in order to determine the status of any book or reader. Such an ideal would be exorbitantly expensive to achieve using a mainframe computer, but is quite feasible with a minicomputar. Systems of this type are being developed in at least two universities (Lancaster and Surrey). 21 O. In the public library field, the trapping store is likely to be superseded in the not-too-distant future by a purposebuiit mini-computer that will perform the same tasks, and in addition will permit on-line file interrogation by means of a teletype or VDU terminal. Plessey are already marketing such equipment, and the London Borough of Havering are dueto have the first installation. 211. Developments of this kind indicate future trends, and shouid be watched with interest. They need not, however. inhibit decisions on the adoption of computar charging, because it will in most cases be practicable to convert a batch-processing system toan on-line (mini-computer) one, without changing the data capture equipment.

Numbering Systems 212. The choice of equipment for an issue system will be influenced by many factors, importan! among which are the numbering systems to be used for books and readers. Different data capture devices impose different limitations on the length and form of numbers u sed, for example :ALS equipment can read 1 alpha characters;

+ 6 or 7 numeric

Plessey equipment can read 9-14 alpha-numerics; Olivetti equipment can read 12 numerics. 213. ThÍs obviously has implications in two ways. First, is it worth rejecting an existing numbering system, and renumbering the stock, in arder to use a preferred make of equipment? Secondly, will the new issue system cal! for a different kind of numbering? 214. There are three schools of thought on book numbering systems. The use of a simple accession number assigned by the library is probably the most widespread beca use many libraries initially followed the example of West Sussex in their use of the ALS equipment, which was first on the market. Brighton use a number of the following form : -

C 056474 O in which the first character indicates the type of number (in this case a book), the next 6 digits are the book number, and the last is an area code. There are severa! examples of libraries using a 9-digit number of which the last 2 digits represent the copy number. Any copy of the book can then be 'trapped' on the first 7 digits. 215. The limitation on number length imposed by the ALS equipment precluded the use of numbers based on ISBN's or pseudo-ISBN's which were later adopted by sorne users

of Plessey and Olivetti equipment. Dorset, for example, uses a 12-digit number, of which the first 9 digits are the SBN (omitting the prefix), and the remaining 3 are the copy number. 216. Finally, there is the possibility which has been explored by Kingston-upon-Hull, of using a meaningful number, as mentioned earlier. 217. There are potential advantages, in terms of simplicity and economy, in using a single book identity number like an ISBN as a common factor in alllibrary systems. Against this, it must be borne in mind that ISBNs require more time to sort by computer, a factor which can be significan!. Simple numbering systems are the easiest to implement. Another factor to be taken into account in considering numbering systems is the limitation imposed by the capacity of trapping stores, if used.

Counter Design 218. Computer charging equipment makes it possible to achieve a clean, uncluttered issue counter, but obviously has to be used in a different way from other kinds of equipment. The design of the counter installation is therefore importan!, and the investment of effort here can contribute greatly to efficiency and hence savings in staff time. Sorne libraries have taken great pains to optimize counter design. Oxford, for example, experimentad backstage with a model counter, using different arrangements and procedures, before installing the system 'live'. Data Capture Systems 219. We know of only three types of equipment that are in use in public libraries at the time of writing: ALS and Olivetti, both of which use forms of punched card to carry book and reader numbers; and Plessey, which uses bar-coded labels, read by a 'light pen'. ALS have recently introduced a third method which uses a non-magnetic metallic label. Other data capture techniques have been used in academic and special libraries, but are unlikely to be fast enough for public library use. Wilson 33 has interpretad the Aslib/CAG study as indicating that the ALS and Plessey equipment "are easily the strongest candidatas for future computer-based circulation systems in public and academic libraries in the United Kingdom". This study also showed that, in 1973, about the same number of installations of each type were in use, as indicated in Table 4.

TABLE 4: Data capture equipment used in operational issue systems

ALS

Plessey

Birmingham Barnet Brighton Camden (Kentish Bromiey Town) Camden (St Paneras) East Sussex Huddersfield Grimsby Kingston-upon- Hull Lancashire Merton Luton Motherwell & Wishaw Oldham Oxford Reading West Sussex Paisley Sutton (8 branches) Torbay Worthing West Sussex (2 branches)

0/ivetti Bournemouth Dorset Salop Staffordshire

29

____ ... - ................................. .......uu

~·wouauv LYJJ"'" Ul

t;qUipment, mcluding the various optional extras. We shall, however, briefly describe their most significant characterisitcs.

a Automated Library Systems Ltd (ALS) 221. The prototype equipment for this system was developed by Elliott Automation Ltd, and was first installed by West Sussex at Chichester in 1967. The system uses 2" x 3" punched cards, as shown in Fig. 2. Book cards are made from card, but borrower cards are usually made of durable plastic (Mylar). A variety of punch coding systems can be employed. The format of these cards is similar to that used in traditional systems, which means that they can easily be used during the transition from manual to automatic operation.

CT 597896 CAMDEH PUBUC liBRARlES This ticket must: not be removed from the book.

o

o 00

o •

e

S

o• o G

e

• • •

00000

.

e

00

Fig. 2. ALS Book Card

1

stop borrower

¡

make

1

1 reservation

1

1

free borrower

cancel reservation

1

1

SET BORROWERS NUMBER CONFIRM

SET RESERVER'S NUMBER SET ACTION CODE

1í 2

3' 4< s¿

TRAP

6~

223. A charge is recorded by inserting the borrower's card in the slot of the appropriate reader, and then inserting the book card or cards in the other reader. The latter will not work unless there is a valid borrower's card in the associated reader. When all the book cards have been read a button is pressed, causing the borrower data to be punched. The output tape thus contains blocks of data comprising a book number or numbers followed by a borrower number.

224. Discharges are recorded simply by reading the cards from returned books. The card readers are connected via a logic unit (which controls the system and carries out the necessary validation checks on the data) to a punch unit, where the data is recorded on paper tape, industrycompatible magnetic tape, or magnetic-tape cassettes. A reel of paper tape will hold about 10,000 transaction records. The tape punch is an inherently noisy machine, so it is usually installed remate from the quiet areas of the iibrary. Fault conditions on the punch unit are signalled to the charge and discharge stations and further use of the equipment is inhibited until the fault is cleared. A maximum of twelve readers can be connected to one recording unit.

1 1

borrower action

book action

_l

1

1

check borrower

1

1reservations test ~

G

7

D Fig. 3. ALS Supervisory Panel

30

LWU

225. The ALS Trapping Store can hold up to 5,000 book and/or borrower numbers, and is usually loaded each day by reading in a paper tape containing the numbers of reserved books and borrowers that for one reason or another need to be 'trapped'. This tape is best prepared by the computer, using the reservation list, transaction file, and other data such as the availability of multiple copies. Once the store is loaded the trapping list can be altered or added to by means of the Supervisory Panel, which is normally mounted on the readers adviser's desk (see Fig. 3).

o

00000000 000

n::;;au~:;;l~,

of which are coupled to form a charge station, one being used to read book cards and the other, borrower cards. The third reader is used for the discharge station.

r

issue

1 discharge

SET BOOK NUMBER SET REASON CODE SET ACTION CODE

1 1

m••-¡ TRAP

226. The Supervisory Panel carries ten expiicitly labelled function buttons, two instruction panels, and two sets of wheel switches. To use it, the operator touches the appropriate function buttons, and the appropriate instructions are illuminated, indicating what is to be set on the wheelswitches. When the numbers have been set up and checked, the operator touches the 'confirm' buttons and the data is recorded. 227. A recent ALS development has been the introduction of System J, in which severa! remota branches are connected on-line to a shared trapping store and recording unit. 228. A Real Time Clock can also be incorporated in the system. This registers one-minute intervals on the transactions tape, so that time-based statistical analyses can be carried out.

b ALS Labei-Based System (LBS) 229. LBS is a recently-developed alternativa to the ALS card-based system. Book identity information is recorded on non-magnetic metallic iabels which are fixed inside book covers. The la beis are read by a sensing device built into the library counter, which can penetrate a book cover as books are slid along the surface. 230. Borrower identification is still encoded on a card, and read on a card reader as with the standard system. 231. This system is to be installed in a library in Lancashire County.

e P/essey Telecommunications Ltd 232. In the Plessey system books and borrowers are identified by means of bar-coded labels 2!" long (see Fig. 4). Book labels are stuck inside the cover of each book, and the borrower label is mounted on a reader's card. The labels are read by a 'Library-Pen' or light-pen which when drawn across the label (a motion similar to crossing-out) scans the bar code and transmits appropriate signals to the recording unit.

card causes an end-of-record marker to be registered on the output tape.

235. To record discharges, a dummy borrower's card is inserted to activate the terminal, but only the book la beis have to be read. 236. The terminals are connected via a Multiplexer to a Data Capture Unit in which the transactions are recorded on magnetic tape cassettes, each of which will hold about 10,000 transaction records. These cassettes can either be transported physically to the computer centre, oran automatic data transmission facility can be used (as at Camden), so that the data is automatically sent by telephone line to the computer overnight. A converter can also be supplied to transfer data from the cassettes to computarcompatible magnetic tape. 237. Another type of Data Capture Unit is also available which allows data to be recorded directly on to a computer compatible magnetic tape. This greatly increases the recording capacity and also eliminates the need for a converter. 238. The Multiplexer scans the terminals continuously and routes data to the Data Capture Unit, adding a code which identifies the originating terminal and indicates the function performed (issue, discharge, renew, reserve, or erase). The Multiplexer can handle up to 8 pen terminals and 2 auxiliary keyboards. 239. The Plessey Trapping Store can be supplied in a range of capacities from 2,000 to 6,000 records. lt is loaded by entering book or borrower numbers from a Composite Terminal which incorporates a keyboard. Unlike the ALS Trapping Store, its memory retains records even when the power is disconnected. 240. The Composite Terminal can be used both to record issues and discharges with the Library Pen, and to contra! reservations and renewals. 241. The Plessey System also allows remote branches to be connected to the central data capture equipment via GPO telephone lines. This allows eco no mies in equipment costs to be made as well as providing centralized trapping facilities for a number of branch libraries.

Fig. 4. Plessey bar-coded label

233. The Library-Pen is about 8" long and is connected toa terminal which has a card holder, 4 warning lights, and a reset button. The terminal is designad for use at both issuing and discharging positions. 234. A charge is recorded by inserting the borrower's card in the card holder, thus activating the terminal; the pen is then used first to read the borrower's number, and then the labels on the book(s). A self-inking date stamp may be mounted on the pen, so that the return date can be marked in the book at the same time. Withdrawing the borrower's

242. As mentioned on p 29, Plessey has now developed a mini-computer library system which is currently being installed in a number of librarles. Features include ability to control from a central location up to sixty-four remote branch librarles, centralized reservation control, production of daily statistics, and on-line enquiry facilities allowing details of books, borrowers and other information to be accessed either by ISBN or by accession number.

d British 0/ivetti (IDP Division) 243. The Olivetti system uses book and borrower cards measuring 3!" x 3t'' (see Fig. 5), borrower cards being usually made in durable plastic material. The cards are read by an RP50 data collector, being fed through a slot and then delivered into a hopper. The same unit is used for recording both issuses and returns. The data from the cards is transferred on to paper tape, the tape punch being incorporated into the card-read unit. 31

----- - .................. " ~....... uiii'C' •cca;::.L aur;,.t::~y

ro cause not1ces to arrive

after the book has been returned - i.e. over the week-end, after closing on Saturday, or early on Monday.

MRS J K SMITHSON 72 SUNNEYLANOS AVENUE 80Uf1NEMOUTH OH6 3AL 7654321

248. Regular computer availability must be assured before committing the library to a computerized circulation system.

A

249. Sorne indication of computer processing time required is shown by the examples in Table 5. TABLE 5: Computer processing time for issue systems

Library

=

=

= c:::J=

DO []

o o

o D

==

1234567

Computer ICL 1903 Loans per 350,000 year (approx) 340,000 Computertime: input and update 60-75 min/day overdues 90 min/week borrowers· register update and cost 5 hoursjmonth

Brighton

ICL 4/50 900,000

85 min/week 30 min/week

60 min. as required.

891.709

HARE.!l RUSSIAN LITERATUAE

o o

o o

Camden Camden (St. Paneras) (Kentish Town)

1234567

o

Fig. 5. Borrower card (top) and book card for use with Olivetti system. 244. Details of each transaction are simultaneously printed on a tally-roll. The RP50 was oringinally designed for factory use, and is robust, cheap and easily transportable. 245. One minor problem with this equipment is that the paper tape output is punched with square holes, and few computers have facilities for reading this. This can be overcome, however, by using an off-line converter which can read this form of coding and transfer the data direct to computer-readable magnetic tape.

lntegrated Systems 250. We have already discussed the main characteristics of integrated systems in Chapter 4, when comparing alternative strategies for the introduction of computer techniques, but it may be useful to recapitulate the main points that we made then, and to expand sorne of them. 251. An integrated system in the present context is essentiafly one which combines two or more applications in the ordering-cataloguing-circulation sequence. By integrating multiple applications it becomes possible to use interlinked files, so that deletions, additions or amendments to the data automatically become effective for all parts of the system. This saves a great deal of the clerical effort that can be spent in updating separate files for different components of a non-integrated system. lt also avoids errors, for with manually processed files and with computer files used in a single application, it can often happen that deletions, additions and changes are not consistently incorporated. At the simplest leve!, if a book is withdrawn from a library, it is difficult to ensure that all the relevant cards are withdrawn from the author and classified catalogues. 252. The potential advantages of interlinked files are evident from the number of data elements that are common to ordering, cataloguing and circulation systems, as shown in Table 1.

Computer Processing Requirements 246. A mínima! circulation system can be run on a small computer having only 3 or 4 magnetic tape drives, a line printer, and paper tape reader. Additional tape drives or disc storage will be needed if author and title information is to be printed out on overdue notices

253. The main advantages of an integrated system are increased efficiency; compatibility between applications; improved service to users; greater potential for economy; and better scope for providing management information. The disadvantages lie in the greater difficulties of planning, design and implementation - though the latter can be made easier if done in planned stages; and in the greater commitment to change, and hence larger initial investment of resources.

247. ldeally the transactions file should be updated daily, but the system could be operated with a mínimum of three computer runs each week. The processing of overdue notices

254. Beca use of the variety of possibilities offered by integrated systems, it is not practicable to analyse them exhaustively here. Virtually any of the features and options

32

that have been described for individual applications can be combined in such systems. lt is also important to appreciate that integrated systems can be operated either on-line or by batch processing. Our approach will be to illustrate the concept by describing three operational systems: Oxford City; Cheshire; and West Sussex. Oxford City libraries System 255. The Oxford City system 32 integrates the functions of cataloguing, stock control and circulation. Book ordering and accounting were deliberately not computerized, because it is considered that there would be no advantage in doing so. The system is a batch processing one, and runs on an !CL 1902A computer. 256. The system covers all holdings (except Local History and Special Collections), amounting to 110,000 titles representing over a quarter of a mi Ilion books. About 10,000 titles are added annually. About 650,000 loans are made annually from the City library alone. The system became operational in May 1973, when the library moved into its new building. 257. Input forms are prepared by the cataloguers, and keypunched by the Computer Department. From this input, three book files are generated. The main file is held in author order and a second file is held in class order. These files employ a fixed-length record format, allowing 54 characters for the author and 1 02 characters for the title field. From the main file is extracted a Mini Book File, containing the first 12 characters of the author field and the first 50 characters of the title field. This file is held in title number order, and its primary use is for generating overdue notices. 258. The catalogues are output on a line printer, and their form and method of storage has been referred to in the section on cataloguing. Locations for all tit!es held by the central and three main branch libraries are indicated in the catalogues, which are completely updated and printed out every 6 months. These are supplemented by monthly cumu!ations. A subject index is also produced (bi-monthly). 259. Book lists can be produced from the computer files, using author and class number sort keys, or special extraction codes for different types of reader etc. 260. The circulation system operates in all lending departments of the Central Library, and uses Plessey equipment, including a 2,000-word trapping store. This will be progressively extended to the branches. Cheshire County Library System 261 . The Cheshire County system 3 at present performs the functions of ordering and cataloguing. Circu!ation control may be added later. The system runs on an IBM 370/145 computer, and data entry is on-line. 262. At the time of our visit, the catalogue held 37,000 non-fiction titles, out of a total of sorne 90,000. Annual intake is about 10,000 titles. lt was intended that fiction titles should be included later, when the non-fiction catalogue was up-to-date. The system was introduced in December 1971, and was being used to order al! adult non-fiction material for headquarters and branches. 263. To order books, individual !ibrarians enter details on requisition forms, and forward these to Library Headquarters at week!y intervals. Details of each title, together with relevant order housekeeping data, are entered at one of the two visual display units and transmitted to the computer,

where they are stored on a magnetic disc file. About 60% of orders are for titles already in stock, and in these cases the terminal operator is able to call up the existing bibliographical details from the file and use these to prepare the new order. At the end of each week, individual orders to booksellers are printed on to special stationery in the form of an order together with associated counterfoil. At the same time, reminder notices are also printed for books which have been outstanding for more than a specified period. Lists of new orders and new titles entered each day are printed out overnight for checking. 264. Management information statistics are produced at weekly intervals, showing an analysis of new orders for the current financia! year to date, and of outstanding orders (unrelated to the financia! year). 265. The bookse!ler retains the orders for his own files and returns the counterfoils with each consignment of books. On receipt at Library Headquarters, the book number on the counterfoil is used to retrieve the terminal. The operator 'receipts' the computer record to show that the order has been completed, and at the same time adds cataloguing details to the record. 266. At four-weekly intervals, catalogued titles are added to a stock tape file which in turn is used to produce a magnetic tape file containing catalogue entries of all titles added to stock since the last annual cumulative catalogue. This tape is processed by a COM service bureau to produce 35 microfilm copies of the author and classified catalogues, which are then posted direct to the individual branch libraries. 267. At yeariy intervals, the master stock file is transcribed into catalogue entry format to produce fully cumulative author and classified catalogues. West Sussex County library System 268. The West Sussex system 21 integrates the functions of ordering, cataloguing and circulation. The County's lBM 370/155 computer is used, and on-line visual display terminals are employed for data entry, and for book location; purchase orders, lists, catalogues, and overdue notices are prepared by batch processing. 269. The computer files (in 1973) held records of 190,000 titles (fiction and non-fiction), representing about 700,000 books. Net intake is about 40,000 books per year. West Sussex pioneered the use of computer book charging in 1967, and about 2?,- million issues a year are currently handled by the libraries using this part of the system. The integrated system described be!ow became operational in 1973. 270. A major source of bibliographical data for input to the system is a file of records taken from weekly BNB/MARC tapes and converted to the West Sussex format. The size of this file is limited, records being deleted when transferred to the ordering/cata!oguing system, or after a maximum of 6 months. 271. Books can be ordered by calling up records from the MARC-based file for display at the terminal. The operator then keys in the number of copies required, details of the bookseiler, and the library !ocations to receive the book. When a MARC record is not available, details of the book have to be entered via the terminal. The computer prints out orders weekly, together with reminders for orders outstandingr and a list of books on order for use in the library. 33

¡;uup1ea wnn tne 1ocat1on code of the branch to receive it. The computer-held book information is then transferred from the order file to the catalogue file. - · · - ···~ uvvv""'v" uuurut:r r:s

273. The catalogues are in loose-leaf book form, and are produced by computer typesetting, which is carried out by a commercial agency. 274. A complete new edition of all parts of the catalogue is produced annually, and cumulative supplements are issued at different frequencies for different sequen ces, ranging from fortnightly for the author sequence of the adult non-fiction catalogue, to quarterly for the adult and junior fiction catalogues. 275. Computerized book charging is used in 1 O libraries (mid-1974), 8 having ALS equipment and 2 Plessey.

34

by a conventional visible index. lf reservations are not fulfilled after 5 weeks, however, copies of the books required are located by using the terminal to interrogate the computer files. 276. The system includes facilities for generating statistics of additions to and withdrawals from stock; transfers from one library to another; book expenditure; and loan traffic. 277. lntegrated systems are undoubtedly attractive for the great variety of useful features they can offer, and their general efficiency. lt seems highly probable that in the long term they will become the accepted way of lite in most public libraries. But in concluding this section we would re-emphasize that such systems cal! for a prodigious amount of careful thought and planning if they are to be effectively implemented.

VI What software should be used?

278. Befare we attempt to answer this question, it may be useful to explain sorne of the terminology relating to software, as this is a subject that gives rise to a certain amount of confusion among many librarians. 279. Software is a general term covering not only programs for specific applications (which include 'utility' programs for common operations such as sorting and printing), but also the programs which convert 'program language' to 'machine language' (known as 'compilers'), and the programs which control the operation of the computer (known as the operating system). In this chapter, however, we shall be dealing almost exclusively with the first categoryapplications software. 280. An application program is a set of instructions for the computer, telling it what to do with the data. The machine only understands strings of binary numbers, but programs are written in programming languages, employing meaningful words and symbols which are then translated by a compiler into binary number instructions. 281. There are various levels of programming language, the lowest (i.e. nearest to machine language) being an assembly language. This requires that the programmer expresses himself at a very low leve! of detail, but its use results in efficient programs. Examples of assembly languages are PLAN for ICL 1900 users, Assembler for lB M 360 and 370 users, and Easycode for Honeywell users. Assembly language programs are machine-dependent, that is, they can only be processed on the computer type for which they were written. 282. With high-level languages, severa! machine instructions can be expressed by one program instruction or statement. These are easier to use, and programs can be written more quickly than with assembly languages. But they require more computer time to compile and to run, and are less efficient. Also they take up more computer storage which sometimes makes them unsuitable for use on very small computers. High-levellanguages tend to be oriented towards certain types of applications. For example, FORTRAN and ALGOL were designed for scientific and engineering work, and COBOL for business processing. 283. In theory, programs written in high-levellanguages can be run on any computer for which an appropriate compiler is available. In practice, however, a program written in, say COBOL for an ICL 1900 will need sorne alterations befare it can be run on an lB M 360. To complicate matters further, there are often severa! versions of the same language which can be used on the same computer. 284. In sorne situations a useful compromise is possible in that programs can be written in a high-level language, with appropriate parts (sub-routines) written in assembly language. In this way, programs can be written quick!y, but still achieve high operating efficiency.

285. Finally, although we shall be referring mainly in this chapter to software and programming, it is worth drawing a distinction here between systems analysis and programming. The systems analyst determines the system requirements (quantities and characteristics of input and output) and designs the necessary systems and procedures to meet these requirements. The programmer then writes the programs to perform these procedures. Sometimes both skills are found in one individual, who can do the entire job, but more often different staff are used for each kind of work. A Suggested Decision Path 286. Once a decision has been made to use a computer for a given application, severa! options are open to the librarían as to the software to be used. We suggest that these possibilities should be explored in the following logical arder:1 First, consider the possibility of using available software from: a another library; b a computer manufacturer; e a software house or specialist agency. 2 lf no suitable software is available, consider alternatives for funding the development of new programs: a working independently; b in co-operation with other libraries. 3 Whether working independently or in co-operation with others, consider alternative sources of programming skill: a local authority systems analysts and programmers; b employ a systems analyst programmer(s) in the library; e train library staff in programming; d employ an outside firm (software house). 287. In the remainder of this chapter, we shall discuss the pros and cons of these various possibilities. Use of Available Software 288. Programs are expensive to write, test, develop and maintain, so the possibilities of using available software deserve very thorough examination. A great deal of effort has already been invested by public libraries in the development of software, much of which is now in operational use, and every effort should be made to exploit this reservoir of experience. Unfortunately, although many public libraries are willing to make their software available, very few are in a position to offer a true package, which should include : The programs themselves; Full documentation (including detailed used and operator instructions); lnitial training of the purchaser's staff; A guarantee of maintenance (assistance in correcting program errors, and notification of available improvements). 35

•