The following are thoughts upon some of the problems which it is hoped the new "Office" in the Reference Division will assist in resolving. There are areas where I clearly need to have more information before I can make a useful contribution. But perhaps, as tentative as these preliminary thoughts are, I hope that there is enough here to make a start.
The first point I would make is that the new "Office" should have a name which signifies help rather than obstruction. Staff in the Reference Division are very sensitive to administrative nomenclature. The word "office" has resonances which will suggest that its activities are remote from the working departments, and there to hinder rather than help. The word "service" has (at the moment at least!) no such overtones, and you might consider one of the following: Catalogue and Collection Development Service; Systems and Resource Service. The problem is that the new service will be expected to assist in several areas, and finding a suitable name is difficult. As I understand its primary functions it is to assist in: (1) establishing acceptable (and affordable) standards for the cataloguing of existing materials in the following departments: Printed Books - Overseas English and foreign books and monographs; serials; Manuscripts; Music; Maps; Newspapers; materials in the Science Reference Library; (2) the conversion of existing catalogues (and revised versions) into machine-readable form; (3) establishing priorities (within available resources) for (2); (4) developing computer-literacy within the Reference Division by encouraging staff to become familiar with microcomputers and their application to specific housekeeping tasks; (5) establishing policies for collection development (in all departments) consistent with processing capabilities and financial resources.[i] One reason why Reference Division staff feel despondent about the way developments are taking place is a suspicion that management tries to implement too much too quickly. Nothing tangible ever seems to get finished. Another reason, I think, is that there is a widespread belief among curatorial staff that management is always looking for uniform solutions to inherently different problems. This belief (partially true) suggests that until global solutions have been found, any action is likely to be ill-considered. This is, of course, nonsense; and is one reason why seven years on we still do not have a clear idea of what preservation is all about.[ii] The point is, we never will! For the simple reason that circumstances are in a continuous state of flux. If only we could stop the universe's mutability for long enough to think clearly, all would be well. But the universe has precious little concern for what I call "procedural lapse". The institution is itself evolving, and we must take account of that evolution. We can do no better than tackle problems seriatim, much as doctors in the aftermath of Belsen had to do.
Implicit in the way the new service projects itself within the Reference Division should be a recognition that, since we live in times of quite extraordinary change, it is not going to be possible to tackle all problems simultaneously. Technology is developing at a rate faster than we can possibly accommodate it within the Library. Sensible choices must be made, effected quickly, and adhered to. To the solution of each problem there must be a clear understanding of:
The cardinal sin, it seems to me, is in considering too many options (and there never is just one solution to a problem) - decision is paralized, and by the time it is made we are overtaken by events. At a fairly primitive level of management theory must be the proposition that some solution is better than no solution!
The international cataloguing standards established in the wake of MARC are essentially constructs which, while recognizing the enormous new potential of the computer as it was perceived in the 1960's, never took account of the fact that resources would shrink; that the cost of input would soar; and that once libraries had access to machine-readable records for contemporary books they would inevitably want to convert their existing manual records. For major research libraries these records are a mosaic of changing practices. In the case of the Reference Division, GK3 records have an ancestry stretching back to 1841. There is some homogeneity in the printed GK3, but the Reading Room copy is now an intricate mixture of old and new, and much in-between.
Panizzi resisted a national plea for a printed catalogue, until he felt certain that the result would be unified in structure and detail. The GK2 experience dealt a decisive blow to the notion of a single unified catalogue, which incorporated both the existing stock and current acquisitions.[iii] The classic example of a catalogue overtaken by events is that of the Bibliothèque Nationale, which for the earlier letters of the alphabet is a century out of date with that library's holdings. It seems to me that in a library the size of the Reference Division we have no choice but to accept the fact that a single unified catalogue will never be realized.
This is not to say that the Library should abandon hope of eventually having records of its holdings in machine-readable form: but how that is to be done is clearly the most pressing problem which must be solved in the few years left before the first move to St. Pancras. There are, it seems to me, only two alternatives: (1) continue the GK OCR-conversion [total cost unknown]; (2) abandon the present scheme and adopt an alternative. I do not regard doing nothing as a serious alternative, since without control of the stock the Library will soon cease to function with efficiency. The arguments for achieving some sort of machine-readable file for the stock are: (1) shelf-mark control [the title-room is already inadequate]; (2) title access. An inventory, which can form the basis of automated book supply and the Preservation Service, simplifying the inevitable (and growing) need to alter shelf-marks.
My views on the current scheme for converting GK3 are well-enough known, and every prediction I made six years ago has come true.[iv] Current estimates of the cost of completing the exercise seem to me unacceptable, given other urgent requirements within the Reference Division. The real problem is, I think, that what has to be done in Store Street (after Newcastle has done its bit) demands two absolutely conflicting capabilities of the staff: an expert understanding of the catalogue, and the objectives of the enterprise; and an ability to sustain mind-numbing procedures for years to come. The job is both curatorial and clerical. That seems to me a misuse of human (not to mention financial) resources. If it is continued, I predict a depressed morale in the staff (the room in question is already referred to as the British Library's "Siberia"); escalating costs; and, worse, a failure to achieve the principal objective within the six-or-so years left before the first moves to St. Pancras.[v] Whatever course is taken it seems to me imperative that we decide on one which is achievable within seven years.
What is needed is, in fact, quite simple to describe, though not so simple to achieve. It is nothing more (or less) than a machine-readable index to the catalogue. No-one is suggesting that existing copies of the printed catalogue (GK1/GK2/GK3/GK4/RR) should be discarded, since their use by researchers (in spite of historical idiosyncrasies) has shown them to be indispensable tools. Over the whole range of changes which the catalogue has undergone during the past 150 years, only a few elements remain constant:
This leaves three constant elements, with the possibility of regarding the heading as invariable (if changeable), and the iterative devices used in blocked entries as invariable (if unreliable in GK3). That does not give one a great deal to work with, but it might just be enough.
For all but works by personal authors subscribing their names to their publications and official publications, the headings in GK are a hopeless muddle. The total output of an author who published under pseudonyms can only be collected from main entries distributed throughout the catalogue (though the cross-references to be found under the main entry for his/her name provide the necessary clues). For a writer such as Swift there are hardly any main entries under his name! Anonymous works are distributed throughout the catalogue on principles which, to the average reader, seem arcane and can result in the work being effectively "lost". The rules for pseudonymous works are fairly precise, but require the reader to know that they were published pseudonymously.
So what we have are just the following elements retrievable on a global scale from GK (they are ranked heirarchically):
Of these, only the shelf-mark provides (in theory at least) a unique identifier (though it does not work efficiently with W.P., P.P., Ac., and a vast range of OPL shelf-marks). The title, the edition/issue, and the heading are interrelated and require intervention at some stage if the converted catalogue is to contain unit records.
It does not seem to me unduly difficult to conceive a strategy for getting such a minimal set of relatively stable and identifiable elements into machine-readable form. Because of the complexity of the catalogue's structure I have grave doubts that software could be developed to do it at one pass. There are two fundamental problems: character recognition; format recognition.
Even if the conversion of GK3 were allowed to proceed (and eventually completed), we would still be left with the daunting task of incorporating shelf-mark changes made during the past thirty years; additions since the last supplement; corrections incorporated in the Reading Room copy of the catalogue. The best available approximation to the Reading Room copy is GK4.
It must by now be clear that, in spite of my enthusiasm for computers and their capabilities, I am drifting in the direction of a manual conversion. Based on my experience in devising a way to get Goff's incunable census converted, I feel sure that it is possible to pre-edit a copy of GK4 (using colour coding) so that keyboarders used to bibliographical records (e.g. the work-force assembled at Irvine to do the REMARC conversion) could complete the task within the time-scale I have suggested. The pre-editing and inevitable post-editing could be effected with a small staff of about six trained Research Assistants under appropriate supervision. I realise that such an approach lacks the glamour of boldly embracing new technology; but I do not think that the problems which the catalogue poses for OCR were fully understood in the beginning. And during the past two years the planning of the project has, it seems, lost sight of the original objective: which was to capture the essential data - not to imitate in machine-readable form a catalogue which does not have a logic which lends itself to mechanical manipulation.
The new building will provide opportunities for several sets of GK4 to be publicly available. Readers who simply want to find books written by a known author will be able to find them; the old headings will still prove fruitful. A machine-readable index will, however, provide a vital tool for anonymous, pseudonymous, and corporate works; provide a basis for an automated delivery system; assist the Preservation Service; and help both staff and readers to identify books by place and date of publication. And access to keywords in the title field will certainly provide researchers with an invaluable aid.
One significant advantage to be gained from keyboarding entries as self-contained units will be that all cross-references (and many thousands of columns devoted to see- references can be ignored. In GK3 these probably occupy 40% of the total number of entries. This will create some problems at the pre-editing stage, since the cross-references will have to found and marked so that they are not keyboarded. But that is a minor exercise compared with what is now being undertaken.
I feel sure that it should be possible to establish with some degree of accuracy the production schedules and costs for the following:
And in determining what is to be keyboarded the aim should be to produce an accurate index: not the catalogue as it stands in another form. The present scheme seems (as I understand it) to be predicated on the assumption that all copies of the printed catalogue will be discarded!
Acquisitions of printed materials up to 1975 (when GK was closed) are unlikely to be of a magnitude which would make it impossible to maintain machine-readable entries in GK-converted form. This would have the distinct advantage of the Library providing readers with just two distinctive means of access to the collections: items published before 1975, and items published since that date. The present system is untidy and difficult to explain to those who use the various catalogues in the Reading Room.
Given that we need, and expect to have, a machine-readable index to the catalogue up to the point where it was officially closed, there exist numerous supplementary catalogues which range from the detail provided in BMC to the various foreign short title catalogues. For English books printed before 1641 the record provided by STC is virtually complete. For the period 1641-1700 Wing is adequate, and the University Microfilms volumes provide additional points of access. For 1701-1800 ESTC will soon be comprehensive and is available on-line. NSTC will provide access to authors, subjects, and place of imprint from 1801 to 1918.
The pre-1800 union catalogue has been divided into its logical constituent parts and the English portion is being dealt with by absorbing pre-1701 entries into the interleaved copies of STC and Wing in S8. The eighteenth century material will be absorbed in ESTC. That leaves the foreign material, for which I have suggested a solution currently being investigated.
I am doubtful about the future of the other in-progress foreign catalogues (French, Italian, German Dutch) under way. They should be machine-readable, but I cannot at the moment see where they fit into the pattern of urgent priorities.
It is my view that we already have a very complex structure of special catalogues, which presents no problem if they are in book form. But it is difficult to see how catalogues constructed according to different principles can be accommodated on a single Department of Printed Books short title database.
No further special catalogues should be undertaken without reference to the "Service"! We are already dangerously close to chaos.
The nineteenth century had its advocates for the principle of descriptive uniformity, with associated dogmatism, in dealing with library materials. This was, not surprisingly, connected with the ideal of a "universal catalogue" - a vision not far removed from the aspirations of the architects of AACR2! In the British Museum reason fortunately prevailed, and so emerged the diversity of descriptive inventories we have inherited. It was, perhaps, Bond who fully appreciated the radical difference between the needs of users in such different departments as Printed Books, Manuscripts, and Maps. Some materials need cataloguing; some need indexing. For users the index has always been of greater use in research than the catalogue. That is why MSS have always been unable to keep pace with acquisitions, and why BMSI has always been frustrated since T.H. Horne. His catalogue of Queen's College is a model of its kind: but could never form a model for a large encyclopedic library. In many ways the Göttingen Sachkatalog is the only one of its kind in the world. I doubt that it can be done for a library the size of the British Library, even though its entries are very brief.
Computers do, it is true, diminish the labour of double-cataloguing, but it seems to me self-evident that we must strive to develop a simple format for the Library's collections which permits a practicable transformation from print to electronics. And that format should be flexible enough to accommodate the various kinds of emphasis which materials as different as maps and newspapers require. Above all, the format must be simple enough to make that transformation practicable and affordable.
The British Library has undertaken a leading role in supporting international standards for the cataloguing of printed materials, and this is going to make it very difficult to adopt - within the Reference Division - standards which could be regarded as degressive. The question we must ask is: whom do we, and should we, serve? I note that Donald Urquhart has recently renewed the argument about "simple" records [Assistant Librarian, 77:4, April 1984], and his view is doubtless shared by many a librarian facing the costs of maintaining a catalogue according to the standard implicit in full AACR2. Historically, those who occupied positions of influence in libraries also occupied positions of influence in scholarship. The membership lists of the Library Association and the Bibliographical Society during the early years had many names in common. The working life of an Edwards, a Garnett, a Bradshaw, was a mixture of librarian and scholar. They actually used libraries, and had a rich understanding of the problems on both sides. We are today, in an environment increasingly bureaucratic and technocratic, in danger of losing sight of the fact that libraries are not ecclesiastical bodies!
Subject cataloguing, and the benefits it might bestow on "curious" enquiry is as old as cataloguing itself. The difficulty is that, as society changes, and our intellectual view of the universe changes, so too does the framework within which we classify books. The cataloguer's distemper with "headings" is equalled only by the philosopher's distemper with subject "headings". The simple "mechanistic" philosophy of a Dewey, with his decimal universe of knowledge, seems to me as unhelpful as Library of Congress Subject Headings, BMSI or PRECIS. In machine-readable form the description of a book, given adequate indexing, seems to me more than enough. And it is precisely where subject classification approaches the "exact" (chemistry, technology, medicine) that keyword-indexing is sufficient. Where the former drifts into imprecise interpretation of "matter" (Gulliver's Travels, Erewhon, Culture and Anarchy, The Water Babies, &c.) it is unlikely that indexing of any sort is going to be helpful. I have yet to meet a scholar who admitted that his research was materially assisted by a subject catalogue more refined than that of Gessner. The old "BM HEADINGS" are, in spite of their intellectual crudity, as revealing as any, and are key-word derived. But the "XCI Rules" predisposed the cataloguer to fix upon the substantive title-phrase, author statement or name. With title-indexing in a computer file the enquirer has more options, and therefore a greater likelihood of finding what he wants. It has always seemed to me obvious that the physical description of a book presents far fewer problems than describing what a book is "about". ESTC is not subject-coded: yet I have seldom encountered complaints from scholars wishing that it was.
It is my feeling that subject classification is accurate when it is least required, and inaccurate when it is most required. One has only to look at the history of classification to realise that one person's "religion" is another's "philosophy"; that "chemistry" can be "magic", that "fiction" can be "satire". There ought to be a permanence about the physical description of a book; that is what bibliography is about. I am doubtful that we can ever hope to devise an intellectual framework which places a book in a permanent contextual position. Which is why Hunter on the gravid uterus is describable and forever obsolete, whereas Joyce on Dublin is describable and forever relevant.
It has always seemed to me a naive expression of faith that the transformation from massive duplicate cataloguing in the card-index era to machine-readable cataloguing would necessarily bring the benefits foreseen by the Indiana Network Conference of 1971. Cataloguers are as reluctant to accept another's description in machine-readable form as they are in manual form. Within even ESTC - as a world-wide project - there are fundamental disagreements about detail between libraries that are participating. London University derives a basic record from ESTC for the Porteus Collection project quite cheaply: but then spends as much on tinkering with the derivative as it might have done if it had started from scratch! This is nonsense, and results in a situation not very different from that which computers were supposed to liberate libraries. On a scale which has to be seen to be fully appreciated, the National Library of Australia (using the WLN system) devotes resources to producing an acceptable AUSMARC record which far exceeds the expense which would have been incurred in continuing in the old way. Unless libraries face squarely the fact that the flood of world printing is such that only shared resources can balance the difference between the rate of accession needed to keep "league status" and the capability to make acquisitions accessible using the new technology, we shall be forced to adopt a policy which reduces acquisition to the level of what can be processed. For a great research library this would be an admission of defeat.
ESTC has, perhaps unfortunately, demonstrated a fact which scholarship is just beginning to grasp: that bibliographical information in machine-readable form represents an advance in the handling of information about books as profound as the transformation from manuscript to print. The generation which will patronize Somers Town will be as impatient with traditional methods of access to the literature of the past as Bembo must have been when he made his journey on mule-back from Italy to St. Gallen to examine a text of Aristotle. But seizing opportunities has never been a characteristic of the dogmatic mind, and I fear that already opportunities are being lost because of the ever-increasing gulf that separates those who administer libraries from those who use them. When libraries become victims of orthodoxy (especially ecumenical orthodoxy) they threaten the very purposes they are supposed to serve.
No two research libraries have identical collection development policies, and it would be surprising if they did. Some of the largest ones have similar policies, of course, even if their foundation collections have different origins. All large libraries have collections which were gained by accident rather than design. The "royal" collections of several national libraries are a case in point. But collection development, even on as grandiose a scale as envisaged by Panizzi, is inevitably governed by factors of an unforeseen nature. There will always be, therefore, a proportion of any research library's collections which is unduplicated in other institutions. Derived cataloguing can only ever be a partial solution to the problem of converting manual into machine-readable records. But it is a vitally important partial solution. And it is fortunate that, notwithstanding the orthodox desire for uniformity, the computer in fact liberates us from the tyranny of the "right card in the right place". I daily demonstrate to visitors in S8 the fact that one can find a book on the ESTC file even if the crucial data concerning author, place of publication and date are wrong! Nowhere is the failure to perceive the difference between a card catalogue and a machine catalogue more evident than in OCLC which produces a daily torrent of computer-generated cards. It reminds me of those early printers who, possessed of a resource of enormous power, sought only to imitate manuscripts.
Of the many machine-readable resources which exist today as aids to the conversion of bibliographical records (RLIN, LCMARC, OCLC) there is one which deserves investigation: the REMARC file. From awkward beginnings it has developed into a tool which could have considerable impact on Reference Division cataloguing. As originally conceived by Buchanan of Carollton Press it was no more than a machine-readable Library of Congress shelf-list, with the unique identifier of the LC card number. The project now offers libraries the capability of inexpensive retrospective conversion using Apple microcomputers and a simple search-key which consists of: Title: 29 characters; place of publication: 3 characters; date: 3 characters; local data: as needed (shelf-marks, &c.). Field-tests have established that search-keys can be entered at the rate of 100 per hour (in-house or contracted-out). Matching records are then supplied in MARC communications format at a cost to the library of between 25p and 30p depending upon the matching rate. Unmatched records must, of course, be dealt with in the accepted manner (current cost at about $5 per record). OCLC is more expensive. RLIN records can never be derived unless the proposed link project makes the use of intelligent terminals possible. I would recommend that a test sample of about 1000 titlepages (currently being catalogued in B3) be submitted to REMARC in order to establish the matching rate for the material currently being catalogued by that section. I enclose as an appendix the relevant documentation on the REMARC project.
There is no doubt that the library has a difficult task ahead in making staff familiar with, and unafraid of, the new technology. I have found, in ESTC, that just having machines in the room helps a great deal. Staff get curious about what micros can do, and providing that curiosity is sympathetically handled, soon become quite fascinated by their capabilities, and quickly realise the ways in which they can make life both simpler and more productive.
We must endeavour to ensure that each principal section within the Reference Division has access to a Sirius, in the first instance associated with a specific task to be performed. Thus, in the Photographic Section there is a clear need for making schedules and routine housekeeping machine-readable, and I have written a program for Doug Webb which will simplify considerably the control over conservation microfilming. It is simple to use, and will reduce a great deal of manual searching currently required when microfilms are ordered. In the Department of Manuscripts I have written a simple program for Hilton Kelliher to enable the first line index of English verse to be made machine-readable. Ilse Sternberg and Peter Hogg have recently completed their catalogue of Onitsha books on the S8 Sirius using a program I developed for them. Similar minor tasks could be identified throughout the library I feel sure. The more staff we can encourage to perceive the ways in which microcomputers can add to the efficiency and quality of essentially clerical operations the more likely we are to avoid union obstruction to their introduction on a scale larger than that now possible.
As many libraries and universities in other countries have discovered, one of the best ways of breaking down the distrust of computers is to encourage staff to buy their own micros for use at home by making available interest-free loans for their purchase. This may not be possible in the British Library, but the idea should at least be explored.
The now-absorbed Cataloguing Systems Branch (CSB) did, for a while, make a significant contribution to computer-awareness with its short familiarisation courses. What is needed is, as the business community has discovered is needed, a regular and widely available series of courses on microcomputers. If necessary these should be contracted-out to one of the many companies currently offering such courses. On a modest scale I have been involved with the R&D project at Leicester, and I have found that even a two-day course produces quite gratifying results.
Just two years ago the average professional (as opposed to home) micro was capable of performing a very limited number of tasks, and with the exception of word-processing. Since then, a micro such as the Sirius has demonstrated that these machines can be put to quite remarkable uses. Thus, the Sirius is now the preferred machine for Cortex, a highly sophisticated software package. The availability of 1-2-3 (May) and dBASE III (July) makes possible applications in accountancy and database management on a scale undreamed of in 1982. For example, dBASE III enables a database file of a billion records, with 128 fields, and a maximum record-length of 5000 characters. The new generation of hard disks have capacities of 180 megabytes. The hardware and software now exists to enable an institution such as the British Library to plan for independent workstations capable of performing adequately local tasks, without the necessity of establishing an expensive central Data Processing department. This is what is happening in the business world, and if we really do wish to model our activities on what is commercially understood to be best, then the opportunity is there for the taking.
My own definition of management is a simple one: it is understanding what one wants to do, and motivating those for whom one is responsible to get it done. Pseudomanagement consists, for the most part, in the empty rhetoric of jargon and half-understood principles. If the new Office is really to help in restoring the confidence of staff in the objectives of the Library then it is imperative that its own objectives are clearly understood and stated. That cannot be achieved by the mere promulgation of a Staff Notice. It is a pity that the announcement of the new Office should have been couched in the directive rhetoric of Staff Notice 22/84. Apart from the fact that the implications of the notice will hardly be grasped by the staff involved, it has served only to reinforce the widely-held view that the library's destiny is governed by considerations of structure rather than clearly articulated objectives; that the library is an ecclesiastical body not a collection of individuals working towards a common goal. Does anyone seriously believe that ESTC would have happened had it been the subject of a Staff Notice, directing the combined resources of the various parts of the British Library (in three Divisions) to start work?
[ii] The Preservation Service, which was inaugurated in 1975, carried out numerous surveys – all superfluous – into the state of the Library’s books. In 1973 Nigel Seeley had drawn the attention of the British Library Board to the disastrous conditions prevailing in the Bloomsbury stacks. This report was never published.
[iii] GK2 refers to the revision of the catalogue begun in 1932? and abandoned in ???? The task of revising the earlier entries and keeping up with current acquisitions was an impossible one. A similar problem beset the printed catalogue of the Bibliotheque Nationale which started in 18?? and was not completed until 19??, and unlike most other catalogues of national libraries did not include works published anonymously or pseudonymously.
[iv] Proposals to convert into machine-readable form the printed catalogue of the British Museum Library go back to the 1950s. The key document is the ADP Report by John Jolliffe prepared for the Trustees in 195?. Jolliffe was instrumental in getting the British Library to consider the computer developed by I.I.I. for the U.S. Navy, in the early 1970s being used to convert Department of Health records at Newcastle. The computer used a sophisticated Optical Character Recognition program, which worked tolerably well with simple records, but was quite unable to handle the subtle layout of the BM’s printed catalogue. The Newcastle scheme was eventually abandoned – as I predicted it would have to be – in favour of a keyboarding approach suggested by the company (Carrolton Press) that had converted the shelf-list of the Library of Congress – and which became known as REMARC. That, too, was abandoned, and the eventual keyboarding was effected by Saztech International. In spite of my urging that the every entry in the printed catalogue should be treated as a unit record, this was not done. When it came to designing the OPAC for the new building at St. Pancras a huge amount of effort had to be spent to simulate a unit record on the screen.
[v] Fortunately, delays in completing the building meant that the move was postponed until 1998. Had it been completed in 1990 users would have had a distinctly primitive system to enable them to find and order books.