College and Research Libraries J E A N M . P E R R E A U L T Comparative Classification for Administrators: A Short Sermon It appears that in too many instances the administrators who must de- cide what classification schemes libraries should use lack understand- ing of comparative classification. As a result, these decisions are fre- quently based upon irrelevant considerations. The author points to the several weighty reasons for claiming that such systems as Bliss, the Colon Scheme, and the UDC are superior to those in common use in the United States today, and invites reconsideration of their merits. A - T X M E R I C A N L I B R A R I E S are growing more rapidly than was expected; perhaps even more rapidly than the libraries around the world. Among libraries that are not growing in absolute size, there is a more rapid rate of inclusion and exclusion, thus requiring an even faster means of making use of the material held there for so short a time. Librarians managing such pressurized institutions are aware of increasing needs for rapid and efficient access to their monumental and/or rapidly changing collections. There can be no sympathy for processing departments and their traditional (but growing) backlogs: the material must get out on the shelves and into the catalogs so it can be used! From all these pressures have come the move- ment toward automation, the use of simple computer-produced catalogs such as KWIC, size-storage, and centralized cataloging and classification. Such devices have their uses, when their limitations are understood. And their usefulness can be increased if ways Mr. Perreault is Lecturer in the School of Library and Information Services, Uni- versity of Maryland. can be found to overstep these limita- tions while retaining the speed-advan- tage of each basic technique. This paper (and two longer ones1 upon which it is based) is primarily concerned with the devices of automation and centralization. It is the headlong rush to reclassification with LC, as a supposedly invariable cor- ollary of acceptance of centralized Li- brary of Congress cataloging, that repre- sents to many the great danger today, particularly to libraries also in process of automation. What is needed, as prelimi- naries to that decision (or to alternatives to it), is the development of a body of insights into comparative classification. Such a title may suggest an austere and erudite discipline, and one cannot deny that, in its most developed forms, it is such. But it can perhaps be shown in a few fairly easy examples how it can be utilized, and what sort of conclusions can be drawn from it. The two longer papers aforementioned have been concerned to develop, as 1 "On Bibliography and Automation; or, How to Reinvent the Catalog" ( L i b r i , XV (No. 4 1 9 6 5 ) , 2 8 7 - 3 3 9 ) ; and "Re-Classification: Some Warnings and a Proposal" (Illinois. University. Graduate School of Library Science. Occasional Papers, no. 87—in press). 4 6 / Comparative Classification for Administrators / 47 the over-archingly guiding principle of all library service—whether conventional or automated, public or university/re- search, in a single institution or in a network, in public service per se or in "non-public" service such as acquisitions searching—the principle of search strate- gy, which can best (or most economical- ly ) be phrased "What, then, next?"—that is, what steps can be taken after the fail- ure of the first attempt to provide that which will meet the patron's need. These two key concepts, comparative classification and search strategy, are not often found among the armory of ad- ministrators, to whom falls the decision which can be based only upon them; ad- ministrators have their own species of reasons, which need not be recited here, all presupposing a state of "everything else being equal. . . ." Comparative clas- sification and the need for a search strat- egy together, though, can eliminate that only apparent state, and thus leave the administrator faced with issues other than "purely" administrative ones. In- deed, for administrators to have so long allowed themselves to be so little aware of the developing theory of library serv- ice as search strategy, even in such di- verse thinkers as Metcalfe and Rangana- than, bespeaks a need for a new invigo- ration of the profession—probably pos- sible only through the library schools. Why do we sometimes become biased against a particular classification? If we have only one document on twentieth- century Magyar lyrical poetry, and it is all we have on Magyar literature we may well rebel at Dewey, 17th edition, which yields a code like 894.51110409003. If the document just prior is coded 894.3 (Turkic literature), and that just posteri- or is 894.6 (Paleosiberian literature), we may well say that the middle number is over-developed, and unnecessarily so. Yet in our subject headings, where adjacent entries are not necessarily conceptually related, we do not object to one entry with a couple of subdivisions coming be- tween two unsubdivided entries, alpha- betically prior and posterior.2 Nor, in a classification where the notation is non- structural and does not attempt to repre- sent lower classificatory orders by ex- tensions of the code, but simply numbers each node in the tree consecutively,3 would it be resented if a document bear- ing a simple code for a complex idea were preceded and succeeded by docu- ments bearing simple codes for simple ideas? DC is under serious attack, especially the 17th edition, and for serious reasons. Yet these reasons are not truly funda- mental; nor are they leading toward so- lutions which are fundamentally ameli- orative of a sticky situation. Since the first need in library service is for search strategy (an answer to "What, then, next?"), a structure must be provided to help patrons and reference personnel dis- cover the next most relevant documents. The two major types of such structure are syndesis (characteristic of subject headings) and juxtaposition (character- istic of notational classification). Syndesis, and subject headings along with it, might be perfectable, but surely a great effort would be required, and the present structure would need to be replaced at one blow by its successor. It seems better then to recommend a shift to a wholly new mode of search strategy, than to be dominated by juxtaposition: in a word, the classified catalog. 2 F o r instance, as chosen from Searso: ' H u n g a r y ' / 'Hungary—History—Revolution, 1956—Addresses and essays—Bibliography'/ 'Hunting.' 3 E.g. 0 _ J I 1 I I I I 2 3 6 9 1 0 I i _ I 1 I l 7 8 1 1 1 2 48 / College <6- Research Libraries • January 1968 What is classification? Most American librarians can think only of shelf ar- rangement as the answer, but this is far from all. And the resistance to classi- fication as search strategy is based on dissatisfaction with currently available "models," primarily DC and LC. Thus by a strange dialectic the majority of Ameri- can library administrators have come to distrust all classification and to place their whole search-strategic trust in sub- ject-headings—which, however, are nei- ther perfect for conventional libraries nor even remotely sensible for automated searching. Why is classification as search strategy resisted? What is one to think of a system that arranges but does not re- veal its mode of arrangement?4 What of a system where the same concept can be predicted to be in a large number of different places, depending on rela- tively minor connotational differences as interpreted by catalogers? What of a system which gives only one available search-strategic pathway, even from an initial point of attack that is complex,5 and thus must require several such path- ways? What we do with such a system is to cease to expect such a function from it; we call it a "shelf arrangement," and thus effectively cease to need to think about it seriously. But classification so characterized is not much of a representative of the fami- ly; where shall we find a better? In fact, several better ones are avaliable: BC (Bliss' Bibliographic Classification), CC (Ranganathan's Colon Classification), 4 It is generally assumed that this objection can- not touch alphabetically arranged catalogs; but I will show in a subsequent work (The Idea of Order: an Essay in Bibliographical Systematics) that this is not a really nonproblematical area at all. 5 In our earlier example, the next most relevant document is not necessarily that on twentieth century Magyar poetry in general, since we may have no such; it may instead be twentieth century Finno- Ugrian lyric poetry, but what do we have available in the given code to allow an economical and rapid transfer of our attention thence? and UDC (the Universal Decimal Classi- fication) would all do what is needed. A few reasons are given in the afore- mentioned paper on "Re-Classification" for possible option for the last of these —primarily in terms of its strong family resemblance to DC, and hence its greater familiarity—but they will not be repeated here at any length. The one thing to be absolutely clear about, however, is that the above-mentioned defects of DC and LC are not characteristic of UDC. It is nearly ideal as a search strategy6 in that it orders concepts hierarchically (but only after having separated out the elements of complex ones), its notation is structural (so that it can be ritually manipulated) and general-categoric (fol- lowing, that is, the separation of the elements of complex concepts). In other words, with it you do know what to try next, the first point of attack having proved unsuccessful; and you know so from the code itself, not from your grasp of its semantic contents. In the cited ex- ample, "twentieth-century Magyar lyrical poetry," the UDC code 894.511-14"19" uses a sub-code for "lyrical poetry" that is uniform in all uses under class 8 (literature); thus if the best available document is on "twentieth-century Finno- Ugrian lyrical poetry" the code is still recognizably relevant: 894.5-14"19." Sim- ilarly with "twentieth-centurv Magyar poetry [of all types]": 894.511-1"19," or "Magyar lyrical poetry [of all periods]": 894.511-14; or, varying more than one facet at a time, 894.5-l"19"; or, adding in additional facets, 894.511-2-14"19" (-2 means "drama"); or, both adding in and 6 1 cannot claim, in fairness, that any available general classification of subject-heading system is really perfect; all that it is fair to do is to make a compari- son in which w e choose the best of the candidates in terms of the criteria recognized. It has been sug- gested that a thoroughly presuppositionless attempt to establish goals and criteria of performance would be advisable, a n d I must admit to a certain sympathy for such an undertaking; but there is a more imme- diate need, for which more immediate solutions are required. Comparative Classification for Administrators / 49 varying, 894.5-2-14"15/19." In each such code, simple program recognition would indicate the degree of distance of the examined document-surrogate from the initial search-specification. What more do we expect from a classification, whether it be for shelf-arrangement, as the basis for a classified catalog, or as the basis for electronic searching? LC cannot do any of these things for us; it has, as mentioned, probably been a large factor in the general disaffection with classification in the minds of Ameri- can librarians and documentalists. Why then change to it? Why indeed! For the sake of monetary advantage, that's why! What other service offers us as large a proportion7 of classificatory work ready-done? None. What other service offers us descriptive cataloging along with this ready-made shelflist and shelf arrangement information? What other offers us a catalog-arrang- ing and search-strategic device in ad- dition to these other advantages? None! But what good are these advantages in light of what we want to accomplish? None, if we can see significance differ- ences between available classifications, some better and some worse (in terms of purpose and its achievement); a great deal, it would seem, if we cannot see such differences, since in that case we should look for a way to save money for purposes which can be effected by excellence. This paper argues that there are such significant differences, and that our pri- mary purpose is the provision of docu- mentary relevances; hence we must choose the means for the achievement of this purpose, doing as well as we can within the financial constraints that such a choice imposes. And library admini- strators must do so too; they must be, 7 E v e n though small enough for a really large a n d / o r rapidly growing library. Also see P. A. Richmond, " S w i t c h Without Deliberation," Library Journal, X C I ( O c t o b e r 1 5 , 1 9 6 6 ) , 4 8 7 0 . in the fullest sense, librarians. This does not just mean possessors of library de- grees, but rather persons oriented to the true purpose of libraries. As admini- strators in the narrow sense they may need to take refuge with the wise coun- sel of their technical personnel, but they must not rest content if these are un- able to outline to them the relations between input and output, cataloging (and classification) and reference, in- formation storage and information re- trieval. If they cannot find reference li- brarians who know the details of classi- fication theory nor catalogers who know the details and needs of reference work, they must become librarians on their own and find out for themselves. The classified catalog, then, arranged by UDC, is in the thinking of some peo- ple a far better solution than would be reclassification to LC, which does not really attack the central problem at all. But even if none can be persuaded to adopt the classified catalog, a search strategy such as UDC can be extremely helpful in the search of electronically stored catalogs which are the by-prod- ucts of library automation. Only, how- ever, if libraries either do their own tape-stored cataloging by UDC (which many would feel is not such a terrible problem), or if they can get such infor- mation externally (and centrally) ready- made. Therefore, a widespread agita- tion appears warranted that such a cen- tralization of service comes about by the establishment (at the Library of Con- gress perhaps, or cooperatively by the Library of Congress and the British Na- tional Bibliography) of an agency to do what is now being done in terms of LC and DC codes—the assignment of UDC codes to a large proportion of the mono- graphic literature. Indeed, this could be made an even more helpful project if such companies as Bowker and Wilson were to index by UDC, so that the card- 50 / College <6- Research Libraries • January 1968 or page-catalog, as well as external bib- liographies contributing to the same searches, were to utilize the same rather than a pointless variety of strategies. The shelves, then, could continue to be arranged by DC or by partial UDC codes, or even by LC (though the es- sential browsing function would be lost thereby). The Library of Congress has always said that its classification was a private system; let's let them have it back. • •