% Copy #£ Glass. Book 1* t -y [Rep rinted from tl' e A. L . A. Curt fere nee, 1 906. ] SUBJECT HEADINGS IN DICTIONARY CATALOGS* By William Warner Bishop, Reference Librarian of Princeton University "M" O library worthy of the name fails to give its readers some sort of clue or guide to the contents of its collections. Its first purpose is, generally speaking, to pro- vide an inventory of its books as they stand on the shelves (the shelf-list), then to give an inventory by authors (the author catalog), and last, perhaps because most difficult, comes the index, or guide, or key to the subject mat- ter of the books. Most librarians are fairly well satisfied with their shelf-lists and author catalogs if they are reasonably up to date and accurate. But few librarians and fewer scholars who use libraries are thoroughly well satisfied with their subject catalogs. The principles of author entry are indeed not all determined. Few matters engage our in- terest more keenly than the long-expected agreement between our association and the British association in this particular. But the comparative simplicity of the rules now in force, and the substantial progress already made toward uniform and sane entries en- courage us to think that we are pretty well off on the side of author cataloging. Our methods of indicating to readers what the library possesses on the subjects of interest to them are by no means so simple or so uniform. It may not be out of place, then, to discuss some of the important problems of subject cataloging. It must be laid down as the prime essential of all subject catalog work that the end in view is the rapid and easy consultation of the catalog by the student who uses it. I say "student, - ' because no one spends much time on a subject catalog who is not inter- ested in some subject to the extent of want- ing to see what books the library has on that *In this paper I have limited myself strictly to my individual opinion on the matters treated, and have in no way endeavored to set forth the practice or theories of the Princeton University Library with regard to subject cataloging. I wish to make this statement, not because of any lack of sympathy with the practice of the library in which I am at work, but simply in order that my personal views may not be held to be an expression of Princeton's policy. topic. Now he must not be discouraged at the outset by any formidable and intricate machine which only an expert can use. The catalog must be so constructed that he can discover easily and quickly what he wants to know. This seems a simple requisite. Yet practice shows that it is one of the most dif- ficult ends to secure. No amount of ingenuity can make a subject catalog which shall be absolutely without flaw in the matter of uni- formity ; no one can always consult it without effort. The student who knows at least a little of his subject and related subjects must then be the normal "public" of a subject cat- alog. But his road must be made straight and the rough places must be made plain for him. Ease of consultation, then, may be laid down as a fundamental basis for work. Rapidity and ease of consultation will be secured only by most careful planning. There are certain decisions which must be made by every librarian beginning or revising a catalog of subjects. Once taken, these decisions must be adhered to, while a change once decided on must be carried out root and branch. Too many of our subject cata- logs of all sorts are medleys of opposing decisions of different catalogers, all made in good faith and with the best of motives. As compared with an author catalog there are few means of checking divergences. Careful planning, then, is half the battle. It matters little, from one point of view, what the decision is. The important thing is to have a conscious policy and to stick to it. The larger the library the greater is the need for uniformity in the matter of subject headings. The small library need not bother itself greatly about principles of subject entry. When its books are all easily accessible, its readers and the library staff alike will rely on classification and current bibliography rather than on catalogs. When you can go straight to the shelves and pull down in a few minutes all the books in the library hav- ing any possible bearing on the thing you want to know, you don't care much for a set WITH THE COMPLEMENTS OF THE AUTHOR of cards in a tray. But the library which confidently expects to become large must needs beware. The day when the librarian or reference librarian with his ordinary tools can answer all ordinary questions will pass suddenly, and then, if the subject cata- log work has been badly or inadequately done, comes confusion and trouble. Particularly is this true of the college libraries. Their cata- logs are likely to get out of hand easily, and they are liable to periods of sudden inflation by gift, and the most careful attention is needed lest the entries under subjects be- come the butt of students and faculty, the despair of the reference librarian, and the tor- ment of the cataloger. One of the greatest obstacles to successful work in the field we are considering is the unfortunate fact that fashions in nomen- clature change rapidly. Such headings as Mental Philosophy, Natural Philosophy, Fluxions, and scores of others current not so long since would hardly help the student of to-day. But more puzzling to him than these odd and old-fashioned forms will be the vague sort of "catch-all" headings that so frequently get into catalogs which do not have to be subjected to the test of cold print. "Practical Piety" in one card catalog I have seen was made to cover all modern sociological and economic works. The one essential for securing continuity and correct- ness in subject work is definition of the subject heading combined with sharp direc- tions as to its use in the library's practice. It is not enough to determine on a heading. It must in all doubtful cases be defined most carefully and the definition preserved. The manner of interpreting the definition in prac- tice must also be indicated. In other words, a (card) list of subject headings in use with all needed notes should be kept in every cat- aloging room. The extent to which these notes should appear in the public catalog is a matter for individual judgment. In this paper there will be no discussion of the relative merits of classed, partially classed and dictionary catalogs of subjects. These matters have been long before us, and their respective claims are well understood. The dictionary catalog has — for good or ill — been generally chosen in our American libraries. Hence our study will be directed toward certain typical difficulties which are met with in actual work. Before taking these matters up in detail, let me call attention to one source of assist- ance and guidance which we too often over- look. Since the seventeenth century the makers of encyclopaedias have been working at this problem. Scores of excellent encyclo- paedias have been in constant use in our reference rooms — and even in our catalog- ing rooms — but have they been studied dili- gently as models for headings? We may be very sure that they have been studied by their makers with exactly our chief problem in mind ; and that is how to choose a cap- tion which shall in a single easily understood word or phrase express the topic to be treated so clearly and definitely that it may be found and comprehended at once. The good ency- clopaedias do not show the fatuous entries and references found in even our good cata- logs. There is doubtless a reason. I sus- pect it lies partly in the excellence of the editorial supervision for which publishers can afford to pay, and partly in the undoubted fact that each encyclopaedia is based on half a dozen, or perhaps half a hundred, prede- cessors, and thus the headings as well as the articles are in a continual state of revision. The fact that the headings are all in print in convenient form, and are easily seen and found, is also a great aid in producing uni- formity of editorial treatment. Still the fact remains for us to ponder. Encyclopaedias seem to present fewer difficulties in consulta- tion than subject catalogs, and are familiarly and easily used by many people to whom a card catalog is a bugbear. Everybody is agreed on the fundamental principle that in dictionary cataloging the "specific'' subject must be our norm. We want to get exactly the caption which fits our book and no other. Especially do we wish to avoid general headings for treatises covering a limited field. A man looking for a book on trees does not want to be sent to look through all the cards on botany, nor does the inquirer for information about Nel- son want to see all the cards on British naval history and biography. He wants what we have about Nelson. As I have said, • everybody admits this. The smallest pos- sible unit must be sought out and made the basis for the subject heading. But the library has also books — many thousands of books, probably — which do not deal with one small, particular topic. It has treatises on Botany and British naval he- roes. Hence there arises of necessity a set of subjects of a general nature, which are in effect identical with the large divisions of the classifications. We have general treat- ises on Philosophy, on Religion, on Sociology, on Philology, and so forth. And, further, we have general works on such topics as Physics, Electricity, Mathematics, Latin literature, Hydraulics, Political Science, Psychology, side by side with works of equal bulk and im- portance on divisions of those subjects, such as Heat, Alternating currents, Differential in- variants, Latin pastoral poetry, Canal locks, Proportional representation, the Sense of touch. There must be general headings, class headings, if you please, in your diction- ary catalog. The difficulty is to use them wisely. These general headings must never be used for anything but general treatises of an inclusive sort. They will be the same in a classed and in a dictionary catalog, and should be treated alike in both. Moreover, a first-rate dictionary catalog will use under these class headings — or headings common to both sorts of catalogs — a few of the simple and large subdivisions of classification, such as History, Essays and addresses, Outlines, syl- labi, etc. In doing this it will not violate the dictionary principle. But we should stop right here. Let us use the class headings when needed, but let everybody understand that they are strictly limited in their scope. Put it on the guide card so that all may see that "General works only are listed under this caption. For special treatises consult the cards with the heading of the particular subject wanted." An ex- ample should be given in each instance, and more than one, if necessary. In the case of the guide card for Chemistry there should be a statement that works on particular chemi- cal products and compounds are to be sought under their own names. The illustration might perhaps take such a form as this — "for example, treatises on Chloroketodimethyltet- rahydrobenzene will be found under that word." It should be said, further, that caution is necessary at this point. Because some head- ings must be the same in any sort of catalog, and because some which are definitely group headings have to be used as a practical mat- ter of common sense in a dictionary catalog, you will find catalogers continually revert- ing to these class headings. It's vastly easier to label a book Sociology than to pin its gen- erally elusive contents down to one particular phase of social inquiry. We all tend to move unconsciously along the lines of least resistance. We shall never get our catalog of specific headings without constant vigil- ance, constant self-criticism, and drastic revis- ion. We must have class headings so long as our libraries are not. composed wholly of theses for the doctorate. And we must avoid them as much as possible. There is a special form of class heading which bobs up serenely with exasperating frequency. I refer to the so-called "forms of literature," such as poetry, ballads, essays,, orations and fiction. Shall we leave these out of our subject catalogs? Many libraries do. Shall we say to the student looking for German ballads, "You'll find them all classi- fied in number so and so"? But then, you know, he won't. There are dozens of vol- umes of them in collections of one sort and another, for one thing. Shall we let our novels go without subject cards and depend on a special finding list 'of fiction? Shall we lump them all under Fiction in the subject catalog? Shall we subdivide fiction and the "forms" by language, or perhaps by nation- ality? Or shall we classify fiction in our subject catalog, and put historical novels with the history divisions to which they suppos- edly belong? These are burning questions with many libraries. Probably every one of them represented here has a policy already decided on and in force in this matter. Here I will content myself with saying that it is my observation that the form divisions in a subject catalog when thoroughly made and kept up to date are a great help in reference work. (And the reference work should be in close touch with the catalog work for their mutual good.) It is, moreover, a con- siderable advantage to carry out the principle that ever)- author card, generally speaking, should have a subject card matching it. In- cidentally I may remark that I have found a mild form of the classification of fiction a great help. I refer to such headings as U. S. History, Civil war, Fiction, which have sat- isfied many a lazy body who wished to take his history diluted and disguised. There are few librarians who will not fol- low us up to this point. We all know that we cannot wholly escape headings which are the same as the major divisions of any classi- fication, and most libraries make some sort of subject lists of their works of so-called pure literature. But when we come to those large subjects which from their very nature suggest a geographical subdivision we leave uniformity behind. There is hardly any such thing, for example, as a treatment of Mathe- matics, or Logic, by countries, although we do find works on Greek Mathematics. These are, however, incidental to a certain period in the development of the science, and not a proper regional division such as may well be demanded in the case of Agriculture, or Geology, or Architecture. The pure sciences, then, do not enter very largely into this prob- lem. But a very large proportion of the sub- jects about which books are written offer a double interest. They may be considered from the view-point of the region or country described, or from that of the subject treated. A work on the geology of Texas, for instance, may seem to belong to Texas, and to require the subheading Geology; or it may appear to have its chief interest for the geologist, in which case it goes under Geology, with the inevitable subhead Texas. This is all familiar enough. Mr. Cutter (Sec. 164) insisted that the only satisfactory solution of this problem was that of double subject entry. With this view I cannot agree. A consistent policy with regard to this class of subject headings which will rigorously enter under either the topic or the country is demanded in the interests alike of economy and of common sense. Whatever decision is taken, a reference must be made from the opposite form. Thus, if the library decides to enter under Geology. Texas, there should be a subject reference from Texas.Geology. Such a subject ref- erence is much better than duplication of hundreds of subject cards. But what shall the policy be? The prac- tice of our leading printed catalogs is ex- tremely varied. On the one hand we have a tendency to provide long lists of subheads under each country. This is the practice at least impliedly recommended in the Ameri- can Library Association's "List of subject headings" by the printing of the long list of subheads to be used under country and state. On the other hand, to cite but a single in- stance, the Subject Index to the British Mu- seum Printed Books (1881-1900) restricts vigorously the entry under the country or region, and allows but few subheads. Be- tween the two plans there is a great gulf fixed. One assumes that a reader thinks along geographical lines when he wants a book, and looks under Greece for a book on Greek Architecture or Mythology, or for a treatise on the Geology or Agriculture or Education of that country. Perhaps he does. The other presumes that a reader considers his subject first, and then runs down its geographical ramifications later. Is there any principle on which this matter may be de- cided? Must we always make special de- cisions? There is at least one principle which favors grouping by countries rather than by topics. It is generally held that the diction- ary catalog should supplement rather than copy the classification. Now the books will doubtless be classified on the shelves by sub- jects rather than by -country in these topics which admit of double treatment. Therefore if books treating of such topics as Educa- tion, Missions, Agriculture, Slavery, Archi- tecture, Painting, etc., from a regional or national point of view — as Central African Missions — and not covering the whole field, are entered under the country or region, the subject catalog will show more about those regions than the classification will at any one point. This seems to me the sole argument for making use of this form of entry. Now, on the contrary, I believe that the British Museum practice and that of the Li- brary of Congress are more nearly in line with the habit of readers and the view-point of the makers of books. If we leave out the historical sciences, the main interest is the topic and not the region. In the pure sci- ences we have already eliminated the regional or national principle. In the applied sciences and the arts, both useful and fine, we may safely do the same thing. These divisions are very extensive. I advocate, then, a de- liberate policy of restricting the entries under the country or region to those topics which have a strictly local interest, i.e., the field of the historical sciences, and such of the social sciences as depend for their value on local conditions. To be specific, I would not put a book on the geology of Texas under Texas, but under Geology with the subhead- ing Texas. I would limit the subheads under a country to those which seem absolutely necessary. For everything else which might be expected under country I would make a subject reference card. This may be begging the question. It may be abandoning the search for a guiding principle. But it seems to me that the habit of most readers and authors is a fair guide for us. After all it is for them that the catalog is made. One word before leaving this topic. At no other point of subject catalog work is definite adherence to a fixed rule more neces- sary than here. A decision once taken in this matter should be rigidly executed. If this is done, the people who use the catalog will quickly learn to follow the principle adopted and will in consequence consult the catalog with ease. If the practice of restricting the entries under subheads of countries or locality be followed, we at once encounter the difficulty of the so-called "national adjective." Having eliminated France.^rf, are we going to cut out French Art, Greek Mythology, Roman Roads? Certainly we must. We must say Art.France, Mytho]ogy,Greece, Roads.Rome, or we shall soon find ourselves in a maze of confusion. It will, however, be necessary, in ^ my opinion, to use the national or lin- guistic adjective with the literature or lan- guage of a country or region. We shall probably be obliged to say French language and French literature, since France.Law- guage and France.Literature do not necessa- rily express the same ideas. As in the case of France, so also in many other instances the national and linguistic areas are not identical. German language and German lit- erature, for example, are wider in their scope than the political boundaries of the present German Empire, and the same is true of the English language. The linguistic and na- tional areas are different in Switzerland, in India, and in many other regions. Another objection to the use of the national adjective is found in the fact that we have all sorts of corporations and institutions whose names begin with American, British, French, etc. Read the headings beginning with either "American" or "British" in the published cat- alogs made on the dictionary principle of any of our libraries, and see what a medley is produced by the mingling of names and topics. I hold that the national adjective should be eliminated from subject headings, save for the two linguistic usages mentioned, This will cause some trouble, for a great many people are accustomed to think of American Indians, British commerce, French porcelain, etc. But the practice will save trouble, too. It will reduce the number of places in which one must look for a topic (the chief drawback of Poole's Index), it will obviate much apparent confusion in the arrangement of headings, and it will intro- duce some system into alphabetical subject catalogs at a point where system is much needed. The practice of the encyclopaedias is against the extensive use of the national adjective.* It may be objected to this that we merely transfer our excessive use of subheadings from the country heading to the subject or topic heading. It may be urged that by this plan the subdivisions under topics become very unwieldy. In answer I would say that the subheads undoubtedly become more nu- merous under the topic, but that they belong there rationally, and there will be plenty left under the country. The person consulting the catalog is obliged, it is true, to run his eye over many guide cards, and perhaps over several trays to find his particular books. But that is far easier for him than going from one part of the catalog to another, look- ink now under France and now under Spain for a work on the mineralogy of the Pyrenees, *There are some exceptions, notably the most recent edition of Brockhaus. for instance. Again he remains certain, after looking at the subdivisions under Mineralogy, that he will not have to look also at the cards headed Pyrenees Mts. — he has all the cards before him for Mineralogy. We can't elim- inate subheadings from our alphabetical sub- ject catalog. At least, if we can, no one has arisen to show us how. If a separate guide card is used for each heading and subheading, we shall find the difficulty of consultation very greatly diminished. And with all the admit- ted difficulty of finding a small subdivision of a big topic, we still get it more quickly, I think, by this method than by the classed catalog with its alphabetical index. It will have occurred to those who have followed this discussion thus far that a good many subheadings under both country and subject might be avoided by the use of in- version. We might say, "Roads, Roman," "Architecture, Gothic," "Psychology, Social," etc., and everybody would understand what we mean. The use of inversion has its chief defense, it seems to me, in the fact that it keeps together related topics. It is certainly convenient to have "Psychology, Animal," "Psychology, Comparative," "Psychology, Morbid,'' "Psychology, Social'' in orderly se- quence and close together. But despite this convenience, as a matter of form of heading, the practice of inversion is to be regarded as fully as pernicious in the subject catalog as in the author catalog. The objections to it are patent and well known. There is one catalog which regularly and always inverts, which enters under an adjective form only in the rarest instances. I refer, as most of you will surmise, to the magnificent Index Catalogue of the Surgeon General's Library. No one will dispute the high authority of this catalog as a scientific product. It is the most remarkable thing of the kind ever done in this country. But I imagine that despite its example we may be more truly scientific if we set our faces squarely against inversion. The worst thing about inversion is the utter lack of certainty as to which several forms may be used. If in our author catalogs we have come to the point where we can write "Michigan. University," why should we not write "Psychology. Animals"? There is not time to elaborate in this paper the argument against inversion. We must be content to dismiss it with the single pro- viso that well established phrases beginning with an adjective such as Republican Party, Political Science, etc., need not be called in question either by those who would always in- vert to serve their convenience, or those who are steadfastly against the practice of inver- sion. The larger question whether the ordinary phrase, e.g., Comparative anatomy, Animal psychology, should not always be employed instead of some device where- by the noun remains in the first po- sition is well discussed by Mr. Cutter in his Rules. My own opinion is for the regu- lar use of the phrase in current use in the form in which it habitually occurs in titles, save in the numerous cases in which a caption with proper subhead better expresses the idea. There is one class of subjects which gives trouble alike to classifiers and catalogers. Wherever a classification or a catalog is sub- divided on a geographical basis, or wherever geographical headings are given, the fact stares us in the face that "geographical ex- pressions," to use Prince Mettemich's phrase, are by no means permanent or dependable. The map of the world has suf- fered startling changes since books began to be made. Certain difficulties which confront us in geographical headings deserve attention. Even the continents give trouble. The terms America and Asia are used very loosely in popular speech, and even in indexes of subjects. Does North America include Mex- ico and Central America ? Where does West- ern Asia leave off and Central Asia begin? Does the term America as a heading or sub- heading include both North and South Amer- ica? Shall we write America, North or North America? What do we mean by Cen- tral Africa? These are questions which have but to be asked to raise sharply the point that definition and consistent adherence to definition are essential in the geographical terms to be used. I say nothing of the for- midable adjective American, for I hope we may largely banish initial geographical ad- jectives from the catalog. But the official catalog of subjects should certainly contain very carefully planned directions as to the use of continental designations, as well as of the smaller divisions of geography. But troublesome as ill-defined geographical concepts may be, they are nothing in the way of difficulty compared to the name of regions which have ceased to represent present po- litical conditions. There are a number of •countries which no longer exist as states, whose political life as separate entities has ceased. A region such as Poland, for ex- ample, which has been absorbed by one or more countries offers a most perplexing prob- lem. The word Poland corresponds to noth- ing on the map or in official gazetteers, but it is still in everybody's mouth. Travellers ■still use the old national name on title pages of descriptive works ; historians and others write on former or even present-day condi- tions. And yet in our larger libraries we have official documents and other works treating of this once independent state from the stand- point of Prussian, Austrian and Russian provinces. I do not believe that we can get around the difficulty by lumping everything tinder the popular name. Neither do I be- lieve that we can ignore it in the case of travel and descriptive works. (Of course I am not referring to books on Poland before the partition.) There are plenty of similar cases, although few with such complications. It seems to me that the common name must still be used where it is employed on title pages, and that the official regional designa- tion of the present day must be employed where needed because of either the title or the contents of the work. This will neces- sitate a lengthy "See also'' reference, a thing to be avoided wherever possible. Ancient and mediaeval states and countries with no continuing name or precise modern geographical equivalent give less difficulty. Their ancient names may safely be used. The trouble is, however, that both descriptive and historical works dealing wholly with present- day (or at least modern) conditions frequent- ly employ the ancient name in titles. In such cases the modern form of name should be regularly used as a heading. Such ancient regions as Pontus, Epirus, Dacia, Africa, Gaul, Granada (Kingdom) may well receive separate subject entry, but it will instantly be seen how much confusion would arise from using these headings for modern works deal- ing with present conditions. Take "Africa," for example. Properly used it means in an- tiquity the single Roman province erected on the ruins of the Carthaginian city-state, lim- ited in its extent to about the boundaries of modern Tunis. So used the term has a dis- tinct value. But a modern work on Tunis, or even a discussion of archaeological problems occurring in the limits of the ancient province should not receive the heading of Africa. There is, then, great need for care and a well- defined policy in these matters of ancient geo- graphical designations which have no precise modern equivalents. Somewhere a very care- ful working out of the proper limits of the subject heading adopted for such countries and regions must be accessible to the catalog- ing staff, and perhaps to the public. It will not do, for instance, to say merely, "Tunis — See also Africa (Roman province)"; "Africa (Roman province) — See also Tunis." These loose "See also" references are the refuge of careless catalogers. In their stead must be a careful explanatory note giving the dates and boundaries within which the heading is ap- plicable. It may be worth while to stop at this point for a word as to these "See also" references. It was a rule at some time in the dim and dis- tant past of cataloging to make "See also" references from each subject named on a title page to every other subject so named. All students of cataloging methods well know some of the ludicrous results of this rule. It is creditably reported that as a result of this rule rigidly applied such references as these were made and printed. "Brain, See also Cheek, Tumors of the ;" "Cheek, Tumors of the, See also Brain," because forsooth both subjects got into one of the long-winded titles of earlier days. Probably these "See also" references cannot be wholly eliminated from catalogs. It is a very good thing at times to have a student reminded of allied topics and similar headings. But the tendency to their abuse is so great that it would seem a better course to make carefully worded ex- planations rather than to multiply these refer- ences. And I believe we should not suffer greatly were they excluded entirely from the subject catalog. They frequently give the im- 8 pression — unjustly, of course — that the cat- aloger is either trying to show off his knowl- edge of subjects, or considers that the user of the catalog has none. To return to matters geographical. Few problems are more difficult as matters of ac- tual practice than the making of a perfectly clear arrangement in a card catalog of easily understood and intelligible headings for coun- tries or regions which have had a continuous written history from ancient to modern days. The most conspicuous of these are Egypt, Greece, Rome, and Syria. The boundaries of Egypt have been practically the same from antiquity to the present day. Hence we are not so much troubled by the question of the physical extent of the heading. But we are directly "up against" the question whether we shall say Egypt (Ancient), Egypt (Graeco- Roman), Egypt (Saracenic), Egypt (Turk- ish), Egypt (Modern), or something of this sort, or whether these headings should be used as second subheads following the recog- nized subdivisions under the country. For example, Taxation is a frequently employed subheading under country, and we happen to have a great mass of material on tax- ation in Egypt in many ages. Shall we write Egypt ( Ancient). Taxation, Egypt (Graeco-Roman). Taxation, etc., or Egypt. Taxation. (Ancient), Egypt.Taxation (Graeco- Roman) period, etc.? The second method keeps the country as the main heading and places the period last, and is therefore pref- erable, in my opinion. But in neither case can we get away from three alphabets in ar- rangement. The method advocated, namely, of keeping the period division last and consid- ering the topic as the more important matter, falls in with our ordinarily received method for modern states. Thus we generally find such headings as this: United States.Ta.ra- tion (Colonial period), rather than United States (Colonial period). Taxation. Which- ever method is adopted, whether we break up the country's history into certain well-defined periods and treat these as if they were sep- arate wholes, or whether we regard the coun- try in all its history as one and arrange topics under it with chronological divisions, the dates of the different periods will have to be worked out with care and recorded in the official list of headings. When this is done it will probably be found that the books seldom fit the dates previously arranged. What to do with overlapping books — books which fit into- no general scheme — is a sore problem in cataloging as in classification. We must either go on forever making new and more minute subdivisions and arranging the sub- ject cards chronologically by the first date in the heading, or else we must assign the sub- ject by the preponderance of interest of the book itself, placing it in that division of the subject where most of the narrative or dis- cussion falls. The majority of catalogers will doubtless prefer the latter method. The spe- cific dates may be put in the heading as a matter of guidance to the person consulting the catalog, but in this case they will be ig- nored in filing. Rome presents worse difficulties than Egypt. In the first place we have to encounter the fact that both the city and the state — origin- ally one — have a voluminous literature. Con- fusion here is disastrous, and yet it is found in many catalogs. The city of the seven hills must be a subject by itself, reserved for sep- arate treatment. Its municipal history is- to be kept separate — where possible — from the march of the mighty empire,, and its monuments must receive treat- ment distinct from that of Roman re- mains in general. It would seem a very good plan in arranging cards to put the coun- try heading first, then the city heading, and finally the heading for its numerous monu- ments and regions. Thus I would have such classes of headings as Rome. History. Em- pire, Rome (city) .History. Middle Ages, Rome- (city). Forum Romanum. If this distinction between the city and the state is not made in this and other cases, we shall have a confusion which will make our catalogs un- usable. Moreover, in treating the Roman state it will be as necessary to define dates and boundaries as in the case of Egypt. I will not go on to speak in detail of Greece and other countries having a continuous recorded history of many centuries. Enough has been said to show the need for careful planning in giving subjects to works on such countries. Still another cause of confusion is closely allied to these we have just been considering. We have numerous cases in which ancient and modern geographical terms do not mean the same thing. I have already cited Africa as an example. The loose habit of catalogers of projecting modern geographical terms into the past is most discouraging to students. Take, for example, such designations as Ger- many and Austria, to cite large regions. Their boundaries are not to-day what they were even fifty years since, and books describing particular regions not formerly in their limits and referring wholly to former times should not be listed under the modern caption, if suit- able ones can be found in the older names. This is merely the principle of the specific heading applied to geographical problems. Again in certain particulars the modern geo- graphical term may represent a much smaller area than the same term at an earlier date. Venice and Genoa are instances in point, and many more might easily be cited. A book on the Venetian remains in the Greek islands hardly deserves a subject, Venlce.Description ■and travel, although one on the Venetian su- premacy in the Levant might well have a sub- ject for Venice. Separate geographical enti- ties such as islands and peninsulas are more easily treated as a rule than other regions, as •confusion is less likely to arise in their cases. Finally a word should be said in protest against subject headings of an indefinite sort for frontier or partially settled regions. "The West" in American history is one such. The phrase "Old Southwest" is another. The ob- jection lies rather against the indefinite nature of the heading than against its use, if once it be well defined. The various regions in Cen- tral Africa offer similar difficulties. If countries having a continuous recorded history present difficult problems to the cata- loged so also do subjects of inquiry which have given occupation to generations of schol- ars. Such studies as political science, eco- nomic:, philosophy, mathematics, chemistry, botany, medicine, theology, rhetoric, etc., had their beginning for our Western world in Greece and are live topics to-day. History and description of countries show the same long line of writers. Now it is obvious that seme discrimination is needed in cataloging the authors who for twenty odd centuries have discussed such important subjects as the theory of the state, the art of healing, or the science 'of mathematics. The distinctions which a printed catalog can show by varieties of type and the rapid view of many pages with their headings are of course impossible in a card catalog. If it is manifestly im- proper to compel the student seeking the li- brary's best treatise on agriculture to turn over numerous cards for editions of Cato and the other Scriptores de re rustiea, so also is it un- wise to neglect the fact that agriculture and all other sciences have their historical side. If we are going to give subjects to all our books, then Cato must have a subject card somewhere under agriculture. Here is where the average dictionary catalog breaks down. It furnishes under such topics as those we have mentioned a dreary array of cards, fre- quently many trays of them, through which the discouraged student must work to find his modern books. Every hundred thousand vol- umes added to the library but increases the task of consultation. The cards thus become what no one wants, an alphabetical list of all the writers who have ever treated of a given topic. The catalog must either distinguish books whose value for the subject is purely historical, or it must arrange its cards chro- nologically (by author), putting the latest works first. In other words, the alphabetical principle of sub-arrangement must be aban- doned under subjects, or else we must intro- duce another division under these subjects having a continuous history, i.e., a class of books having an historical value only. But when does a book begin to have a merely historical value ? There's the rub ! It is not possible to determine this by chronology alone. Can we consider Aristotle of merely historical importance in the discussion of poe- try or drama, of political science or ethics? Most assuredly not. But yet his works on physics and natural history are absolutely without profit to the average student of to- day. No one will say that Kant's writings are out of date, and yet his psychology would IO hardly benefit the modem student in our col- lege classes. It is plain that discrimination of the highest order must be employed in this matter, or else we must adopt some mechan- ical arrangement such as the filing of cards in chronological order, which after all works a sort of rough justice in the matter of rela- tive values. Let us be severely honest with ourselves here. Who of us can say that the trays headed Theology or Law in most of our catalogs of libraries of over one hundred thousand volumes are practically useful as they stand to-day? Who would not rather consult a good bibliography and then the au- thor catalog for books on those topics than attack the direful array of cards in the hope by some means of at length securing an inter- esting and valuable set of references? In formal political history and in economic history as well the sources should certainly be distinguished from the recent treatises. The Germania of Tacitus, for instance, is an excellent source for the early history of the German Empire, but it is positively foolish to list it side by side with the works of Von Sybel and Ranke under Germany.History. The subhead of "Sources" under history is a convenient and valuable limbo for bygone works and for collections of documents. There is opposition, and sensible opposition, however, to using it for merely obsolete treat- ises. We might adopt some such scheme as this : Political Science. Modern works (since 1850) and important earlier works. Works between 1500 and 1850. Mediaeval works. Ancient works. The divisions suggested here might perhaps be the same in all cases, or they might better be made to conform to well-recognized di- visions in the history of each topic. The al- ternative plan is the arrangement of cards by date of publication, or by first date of the author (to keep editions together). I confess I prefer the latter, although I am far from wishing to put myself in the position of as- suming that the most recent work is necessar- ily the best. Still the chances are that it rep- resents the most recent stage of investigation. Almost every librarian is willing to concede this in the matter of bibliographies, acknowl- edging that the last to appear should first meet the eye of the person consulting the catalog. Why not adopt the same principle for every topic, as is done in some of our libraries? We have, be it remembered, the author catalog at hnnd for every one who already knows the authors he wants. Why compel the seeker after information to wade through another author list under each topic? It may be ob- served that an annotated catalog would be al- most forced to put first its cards for the books most highly recommended. There are a few practical points which I wish to take up before closing this paper. First, shall we definitely limit the number of subject cards to a given book? In view of the immense size to which our card catalogs are growing is it wise to say that when the library reaches a certain size — say 500,000 volumes — it will henceforth assume that the necessity for making cards for any other than the subject of prime interest in a book has passed? Shall we take it for granted that there will always be other works which cover the topics of secondary interest? This view is maintained in some libraries whose author- ity we all acknowledge. I venture, however, in opposition to this idea, to call attention to the statistics of our Princeton work published in the Library Journal for June, 1906. It was there shown that the number of subject cards per main entry was 1.47, and per title 1.2, al- though no restriction was placed on the cat- alogers other than a rigid insistence on the specific heading in all cases. This is so nearly the result aimed at in the rule that I submit that it is a better way of attaining the desired restriction of the unduly rapid growth of the card catalog than the strict limitation to one subject per book. It permits the liberal han- dling of a book which treats definitely of sev- eral topics, and yet it does not too greatly burden the subject catalog. The device of using but one subject entry for the various editions of a work whose value is chiefly his- torical would diminish the per cent, of subject to author cards to less than one in our library. Again, it may not be amiss to urge that the II revision and co-ordination of subject head- ings should be definitely assigned to one per- son. Only thus can continuity and uniform- ity of the work be secured. Particularly is this provision needed in our largest libraries. I urge also as a most vital matter of practice that the chief reference librarian should be in constant touch with the cataloger who passes finally on subject headings. They will work together to great mutual advantage. Moreover I wish once more to set forth the imperative necessity for an official list of headings in use in the library. This should be kept up to date with the utmost care. Each cataloger should have in convenient form a list of all subheads previously authorized un- der each class of topics, together with defini- tions of all these subheads. The list without definitions and interpretative notes will be of some small value, but with them will be vast- ly more useful. It should be kept where every cataloger can consult it, preferably in a case of trays made to swing on a pivot so that it may be consulted without disturbing the one at whose elbow it must be placed. The American Library Association list and the Sydney list, admirable as both sre in their own way, will not suffice for any large library. An up-to-date list of subjects with adequate definitions kept on cards, is an absolute ne- cessity in a well-ordered catalog department. Is all this worth while? Is the card catalog of subjects alphabetically arranged a real ser- vice to an institution? Most assuredly it is. When once it is made on consistent principles, when the student no longer has to fumble long trays of cards without headings or guides, filled with all the contradictory accu- mulations of generations of catalogers, when the specific topic stands out prominently, when each subject capable of two interpretations is sharply defined on a guide card, when con- sistency in geographical matters and uniform- ity of entry and sub-entry in topics of de- batable form have been reached, there is no reason why a student should not find the card catalog of subjects self-interpreting, inclu- sive, useful. It has the all-important merit of definiteness and point. It tells any one who knows his topic what he can get directly on it. It lists both the obsolete book and the dead and gone state by themselves. It opens up to the reader the contents of the library. It is, in short ,an alphabetical subject index to the books. If this is not worth while, what library effort is? If this be formal, dry-as- dust work, why are we working with books at all? Our aim as librarians is not merely to accumulate books. It is to help the reader to the books he wants — or ought to want. In a large library the only tool which accomplishes this result is the catalog, and of this the sub- ject catalog is the part most difficult to make, most useful when well made. LIBRARY OF CONGRESS I ll ll 1 1 m 1 1 il :i