and then the books were placed upon the shelves, and numbered in regular numerical order until the capacity of each shelf was exhausted. At first, as we have already indicated, some attempt was made to keep different subjects separate by assigning certain alcoves or portions of the library to special topics. As the library grew, it usually expanded in unlooked for directions, and as a natural result the spaces assigned to some subjects were often filled to overflowing, long before others were at all crowded. As a result the subject, instead of being found in the place originally assigned to it, was frequently found in several different parts of the library, it might be at quite a distance from each other. The disadvantage of this system soon became apparent to every one using it. Not until within a few years, however, has a relative or movable location, or a system of assigning a number to a subject or topic, instead of to a certain location in the library, been generally adopted. This has been found so great an advantage over the old method, that it is almost entirely used in the arrangement of new libraries as well as in the re-arrangement of many old ones. Its great advantage lies in the fact that by moving the books along upon the shelves, or better still by leaving spaces at the end of each subject, all books upon that subject may be kept together by interpolation.

It has been said, that by the old methods the books were usually divided into broad classes. It was but natural, therefore, that with the adoption of a relative location, the old ideas as to the number of subjects employed should at first prevail. As a consequence we find the first libraries arranged by the relative location were divided into a small number of classes. It soon became apparent that more classes could be employed to advantage. As the relative location was used and better understood, it grew in favor, and the question of close classification has since come to the front as a natural outgrowth of its use. The questions that now most imperatively demand our attention are these: how far is it practicable and desirable to carry the division of the library by subjects and their

subdivisions; what advantages are to be derived from close classification; and what obstacles lie in the way of its general adoption?

Close Classification in the Catalogue vs. the Shelves.

It has long been the cataloguer's rule to enter each work under its specific subject. Without entering into any discussion as to the best form of the catalogue, whether classed or alphabetical, which is a question that needs. special time for its treatment, the question arises whether, as far as practicable, a corresponding minuteness shall not be employed in shelving the books of a library. Some librarians acknowledge the necessity of having the catalogue minutely subdivided, each topic appearing under its most specific head; but at the same time they are unwilling to admit that a corresponding minuteness in the arrangement of the books themselves is to be desired. Others see in such an arrangement, not only a great advantage in the ease and economy with which the library may be administered, but also that its usefulness as an educational power is increased and that a far-reaching and beneficent influence is exerted upon its patrons. Its advocates do not and never have claimed to accomplish the impossible or the impracticable; but they do claim, as far as possible, to transfer the advantages of close classification from the catalogue to the shelves; to make the shelyes their own catalogue. Close classification seeks to make the library more useful and available by arranging its resources in minute classes. This is of primary value to those who are in the library, as it enables them the more readily to ascertain the resources of the library upon a specific topic. It may be claimed that this is the function of the catalogue, which is true. The catalogue should give the resources of the library more fully than the shelves can possibly do. It is true that we cannot have the entire resources of the library standing side by side upon the shelves, unless we can take some of our books apart, and in some cases perform the impossible feat of making them. occupy two distinct places at the same time but all this does not and never can compen

would soon find himself supplying the deficiency by making an index of his own, thus showing not only that an index is a convenience but a necessity.

Class Notation.

We

I come now to speak of two great obstacles to be dealt with in the formation and application of the ideal system of classification; the system that is destined, because of its intrinsic merits, to take the precedence of all others and attain a general use. The first to be considered is class notation. The combined ingenuity, and the best talent of the library profession has, as yet, failed to devise a class notation that is entirely satisfactory. have systems that use letters alone, as Mr. Edmands'; letters and figures, as Mr. Cutter's; letters, figures, and symbols, as Mr. Smith's; and figures alone, as in Messrs. Dewey's, Perkins' and Schwartz's systems. Each of these notations has its peculiar advantages, and it is greatly to be regretted that some new notation cannot be devised that will combine all their best points, but this is clearly impossible. What is to be sought for, in our ideal notation is:

1. Naturalness and simplicity in its characters and their combinations;

2. Ease in reading, writing, and remembering them; and

3. Brevity, with a great capacity for subdivisions.

The most natural and available materials from which to construct a notation are letters and figures. There is little reason to suppose that any system, based upon the use of other characters, could be successfully or generally employed. Between the use of figures and letters, used alone, there are reasons to be deduced in favor of both, though personally I am inclined to prefer figures. Letters are open to two objections: 1. They are not as easily written or read as figures, besides, they require a greater number of strokes of the pen in their construction, and are therefore more liable to illegibility. 2. They are not capable of as great a variety of combinations, without producing unpleasant effects. Figures or letters used alone seem preferable

to their combined use, which can only be justified upon the ground of brevity and a greater capacity for subdivisions. The combinations are too complex to be fully understood by the usual frequenters of libraries.

Figures alone, seem to answer most fully and satisfactorily the requirements we have named as essential to an ideal system of notation, being most easily written, read, and remembered, and being in their combinations the simplest forms known to the human mind. Figures in all their permutations are perfectly natural and simple, and are easily read or remembered, which cannot in all cases be said of letters. There are two methods of dealing with figures as a class notation that call for a moment's notice. We may first lay out our scheme of classification, and elaborate it to any desired extent, and then begin and number our classes, sub-classes, and sections in regular numerical order, leaving occasional gaps for new subjects that may arise in the future. While this may seem a more economical use of material than the other method which I shall presently mention, I am inclined to think that in the end nothing is gained, but that much confusion is liable to arise, especially if inadequate gaps are left for future contingencies. This is the very same objection which we saw rendered the fixed location objectionable, and led to its abandonment. The second method of using figures is that of using decimals for purposes of subdivision. This method, if we may judge by the favor with which it has been received and adopted, seems to be the best application of figures, when used alone. The great objection to the use of decimals lies in the fact that minute subdivision necessitates long class numbers; but I think I express the mind of many in saying that a class number of not more than four, and in exceptional cases of five figures, is preferable to one of mixed figures and letters, even if the latter were shorter by one or two characters. It does not follow because a system of classification has been carefully elaborated for the use of specialists in all its classes, that it must be adopted, with all its subdivisions and minute headings, in all libraries. The extent to which it is to be

adopted is purely a matter of judgment, to be determined by the circumstances of each particular case. For a library just starting, the wisest course seems to be to select some system that has been carefully coördinated and worked out, and decide how much of it shall be adopted, it constantly being borne in mind that the future growth and success of the library depend much upon its being laid out upon a broad and liberal basis, and that careful and even elaborate work will, in the end, prove most economical.

Alphabetical Book Numbers.

The second great obstacle to be dealt with in the application of our ideal system is in the too persistent use of alphabetical systems

of book numbers. Those best known and generally used are primarily designed to keep large classes of books, as fiction and biography, and even whole libraries, in strict alphabetical order. When, however, the library is broken up into minute classes, under a system of close classification, the necessity previously existing for a strict alphabetical order, either by authors or works, ceases; and instead of being a great convenience it becomes a disadvantage. This may be seen in those libraries that have attempted to use close classification, Mr. Dewey's for example, in connection with Mr. Cutter's system of book numbers. There are certain large form divisions, like fiction, drama, and poetry, as well as biography, which no system of close classification can satisfactorily break up. In these and other similar classes, an alphabetical order is highly desirable, and can in most cases be satisfactorily secured by abbreviating the usual class number; but where close classing is used and small divisions of books secured, some other system of book numbers should usually be employed. The alphabetical arrangement has perhaps more greater advantages than any other, but instead of employing an elaborately workedout system, the capacity of which is practically unlimited, I would use an approximate alphabetical arrangement. This can readily

and

AN issue of Science a few weeks ago con

tained the following in an editorial note: "The increasing number of series of monographs on special topics must have attracted the attention of all those who possess any acquaintance with current literature. We have an American science series, a set of science primers, several sets of literature primers, historical monographs, economic papers, and so on. The development in this direction is a perfectly natural one, and one which results from the increasing specialization in study. It is impossible any longer for even the well-informed man to follow methods and details he must rest content with results, and even those concisely stated." Prof. W. O. Atwater, in his article on Food in the June Century, says: "The experiments of the last twenty years are numbered by hundreds and even thousands, and the literature of the subject is so voluminous that few specialists even are able to handle it.”

In connection with this literature of specialization a difficulty has arisen to which it is the purpose of this note to call attention, and to suggest a possible remedy for the same. It is the common experience of the librarians of the older and the larger libraries to have specialists come to their collections for the purpose of consulting authorities not accessible elsewhere. Dr. Edward Eggleston, in the preparation of his valuable monographs for the Century on the early life of our ancestors in the colonies, had to go from his own fine library on the shore of Lake George (of pleasant memory) to the Astor Library and the Historical Society Library in New York, to the Boston Public and Athenæum, to the Massachusetts Historical Society Library, to Harvard College Library, to the State Library at Albany, to the National Library at Washington, and even to the British Museum. Prof. H. Carrington Bolton, of Trinity College, Hartford, finds one long looked for

authority on "musical sands" only at the Astor Library, and in the preparation of his catalogue of scientific and technical periodicals has the assistance of the Smithsonian Institution and of more than 125 librarians. "In a recent discussion, in the main quite an idle one," says the Boston Transcript, “as to the relative advantages of New York and Boston as places of residence for literary men, one significant statement was made by a New York man of letters, to the effect that he, in his work, not infrequently found it impossible to obtain the books he needed in New York, and was under the necessity of coming to Boston and Cambridge to get at them."

Much valuable aid has been rendered to specialists in ascertaining where the authorities they wish to consult may be found in the liberal exchange of catalogs, bulletins, and check lists between libraries, and in the publication of cooperative lists and special bibliog raphies. But all these have served to bring out more clearly the great but perhaps not wholly insurmountable difficulty; viz. that the authorities in any and all lines of research are widely scattered in libraries far apart from each other, and that the specialist is often put to an expenditure of time and money from 10 to 100 times the cost of the book he wishes to get at, in traveling to the place where it may be examined.

"When Huxley writes on science in general, Walker on political economy, Geikie on geology, Martin on biology, and Young on the sun, we may rest assured that the results will be beyond criticism."

Some specialists, like Ex-Pres. A. D. White, of Cornell University, Von Ranke, and Mr. Hubert H. Bancroft, have been able to gather to their own libraries needed authorities in such numbers as to make their collections unique and invaluable; but how very seldom during the life-time of the collector, as in the first instance, or at his death, as in the