As said already, early catalogues of libraries were mostly either classified or simply author catalogues. The classification was, often enough, very poor, the sub-division not being carried very far, and this led to the invention or evolution of the dictionary catalogue and brought the classified, such as it was, into disrepute.
The cataloguing of a library is one of the most troublesome and expensive departments of its administration. The cost of printing is greater than ordinary printing, and the expense to a library with its limited income is always serious, because people will not buy a catalogue even at half the cost price of printing but prefer to make use of the copies provided at the desks. Moreover, at the end of six or even fewer months after publication the public usually regard it as out of date and decides to wait for the next edition. In this respect the classified catalogue has the advantage, as it costs less to print, and for this reason, as well as owing to the custom of admitting readers to the shelves of public libraries, there has been a revival of this style of catalogue in late years, especially as it serves as a key or guide to the arrangement of the books upon the shelves of "open access" libraries. It can moreover be printed and issued in sections without affecting its completeness in the end.
The dictionary form, as distinguished from a mere alphabetical list of authors, consists of entries of books under their specific subjects, instead of their classes. To quote Cutter again: "Thus if a book treats of Natural History it is put under that heading; if it treats of Zoology alone that word is the rubric; if it is on Mammals it will be found under mammals; and, finally, if one is looking for a treatise on the elephant, he need not know if that animal is a mammal; he need not even be sure that it is an animal; he has merely to be sufficiently acquainted with his alphabet to find the word Elephant, under which will appear all the separate works that the library contains on that subject. Nothing, one would think, can be more simple, easy to explain, easy and expeditious to use than this. No matter what he wants he will find it at once provided that the library has a book on just that subject and that it has been entered under the very word which he is thinking of. If these conditions are not fulfilled, however, there is more trouble. If the library has no book or article sufficiently important to be catalogued on that topic he must look in some more comprehensive work in which he will find it treated (as the history of Assyrian art is related in the histories of Art), in which case he will get no help whatever from any dictionary catalogue yet made, in finding the general work, he must trust to his own knowledge of the subject and of ordinary classification to guide him to the including class, or there may be something to his purpose in less general works (as books on Iron bridges or Suspension bridges might be better than nothing to a man who was studying the larger subject Bridges), but in this case also he will very seldom get any assistance from dictionary catalogues, and must rely entirely upon his previous knowledge of the possible branches of his subject. Even in those catalogues which relieve him of this trouble by giving cross-references, he must look twice, first for his own word and then for the word to which he is referred from that."
A judicial statement of the merits of both these styles of catalogue is contained in a paper by Mr. F. T. Barrett, of Glasgow, entitled "The Alphabetical and Classified Forms of Catalogues compared," printed in the Transactions of the Second International Library Conference, 1897. In the Library Association Record, 1901 (pt. 1), pp. 514-531, there is a verbal and friendly duel between Mr. W. E. Doubleday and the author upon the matter, mainly from the practical point of view.
The Alphabetico-Classed catalogue, as its name denotes, is an attempt at a classified catalogue in alphabetical order of subjects or classes, and is a mixture of the two systems already spoken of, and about as satisfactory as hybrids usually are. The late Prof. Justin Winsor characterised it as "the mongrel alphabetico-classed system, a primarily classed system with an alphabetical graft upon it is a case of confusion worse confounded." The great difficulty both to compiler and user is to know where the subjects leave off and the classes begin—in other words, whether a subject or a class entry is likely to be the one wanted. One of the best examples of this kind of catalogue is the late Mr. Fortescue's "Subject Index to the British Museum Catalogue," and he apparently experienced the difficulty of deciding, as for instance a book on the elephant appears under Elephant, but a work upon the Elk must be looked for under "Deer." The usefulness of this particular catalogue cannot be gainsaid as its value is too well known, mainly because there is no other form of subject-catalogue for the library of the British Museum. Besides it has such a comprehensive series of cross-references that difficulty is largely obviated, and then again it is only meant as a subject supplement to the principal catalogue. Admirable as it is, we may see how it works out in practice. Suppose we are interested in Law. Under the heading "Law" we find a large number of entries divided into particular kinds of law as "Commercial," "Criminal," "Ecclesiastical," &c., and these are further sub-divided under the names of countries. One would suppose that the subject would be here treated in a most exhaustive manner. But that is not so, as if we require books on the Laws of England we must turn to the word "England." Thus we have books on English criminal law under "Law"; a book upon English general law under "England"; and a book say upon English election law under "Elections, Law of." If it is right to put books on the law of elections under Elections it might be assumed that books on criminal law would go under "Criminal law," but there is not even a reference to say where they are to be found. Admittedly "Law" is a large and complex subject, and would fill many pages if the books upon it were brought together. As it is the searcher must take a long time to ascertain in any exhaustive manner what books upon the subject are contained in Mr. Fortescue's Indexes. Even if the inquiry is narrowed down to say Italian law, searches must be made in many places without touching special Italian law at all. However there is no system but has its drawbacks, though probably the alphabetico-classed has the most.
There is such a thing as a dictionary system that combines an unseen but systematically classified system. Its root method would be to adopt some thorough scheme of classification permitting of the finest possible detail in topic and adjust thereto any necessary cross-references to cover synonymous names and double subjects. The cataloguer would keep the complete scheme in all its details before him and, by means of an alphabetical index to every adopted name, he would have a list of the subject-headings in dictionary order and to these he would adhere. There would still be specific entry. This is the method that should be pursued in the compilation of dictionary catalogues. The classification may exist only in the mind of the cataloguer and be formulated in no other way unless he relies upon headings already fixed in his catalogue. By trying to adjust headings in such catalogues to any logical classification one can soon ascertain whether they are systematic or haphazard.
The following definitions should be noted before proceeding to the next chapter:—
Author-Catalogue is one in which the entries are arranged alphabetically according to the names of the authors (a dictionary of authors).
Title-Catalogue is one in which the entries are arranged alphabetically according to some word of the title, especially the first (a dictionary of titles).
Subject-Catalogue is one in which the entries are arranged according to the subjects of the books, alphabetically by the words selected to denote those subjects (dictionary arrangement). If these subject entries are not arranged alphabetically, but are formed into classes philosophically according to the scientific relations of the subjects, then it is a classed or classified catalogue.