The difficulties emphasized by the pessimist, the overlapping of studies and the rival claims, arise chiefly from improper subordination. The material is common to the several sciences because these are portions differentiated from larger fields. Child-psychology is part of Psychology. The science and art of education are mainly concerned with the mental. They are related to Physiology and to Sociology as Psychology is related. But to place Education under Sociology, as is done by the D. C. and the E. C. is to answer the relation of second, not of first dependence, and is as false as it were to put psychology under sociology, to put the cart before the horse. Education and Psychology are working together, and their books should be contiguous. How shall we arrange these practically? Well, scientifically, in the order of generality, thus:
I Anthropology.
ID to IG Human physiology.
J Psychology.
JN Social psychology.
JO Child-psychology.
JP Education.
JQ Educational psychology.
K Sociology and Ethnology.
KA Sociology.
KE Ethnology.
L History.
The principles of consistent subordination and practical collocation should guide the maker of a system, and his notes should guide the classifier of books. Here indeed should be a "code for classifiers" more intimately articulated than in a separate book. But herein lies the practical art of classification, so to dispose classes, divisions, and subdivisions, that they shall produce a relative minimum of inconvenience under the average conditions of demand and a relative maximum of collocation not only of special classes but of general, as well as a degree of consistency as high as practical conditions permit, and ultimately, as an ideal, a consistency not only with the pedagogic but with the philosophic organization of knowledge. This ideal, I believe, is not beyond approximate realization.
This critical but optimistic view ascribes the failure of library classifications to the dispersion of related material under subject, or close classification, without proper subordination and collocation. The subject-index, however useful to classifiers, is of little value to students. I approve close classification, but find it the more unsatisfactory and baffling as it is the less consistently adapted to good broad classification, with good articulation of related subjects according to predominating interests, and with alternative locations for flexibility to changes and for durability in the progress of science.
III
Having answered the main question of feasibility, we may now take up some minor practical questions, first Notation. It is not likely that reason shall soon remove all traces of prejudice and controversy in this matter. A few propositions, however, are so reasonable that I think they will be accepted. Notation should be brief and simple. Its simplicity depends upon its brevity, though also upon the familiarity and homogeneity of its elements. Letters give brevity. The capacity of three-letter notation, allowing for omission of all objectionable combinations, is about 15,000. Using letters and figures together increases this capacity to about 25,000, omitting confusing mixtures such as K7G and 8B4. Since somewhat more than 10,000 subdivisions seem requisite, the question reduces to this form: "Which is simpler, notation of three letters, or of five figures?" But figures, it is argued, are more familiar. They may be so to bookkeepers, but to the keepers of books! Familiar here means familiar with the numbers of the D. C. Then, are unmeaning combinations like DAL or GWK really more meaningless than numbers like 13859? On the other hand, isn't RAG easier to see and to remember? But the argument, so far as it is not merely prejudiced, is childish. Such combinations as A1, 3B, C42, and CF6, are hardly objectionable, and may prove convenient and economical in class-notation as they do in the author numbers, with which librarians are so friendly. Since they are come to stay, what is the use of arguing for homogeneous notation?
Notation is the more systematic and economical where it reduces in part to schedules applicable to the subdivision of many classes or divisions. This feature appeared to a minor extent in the "form signs" of the D. C., but was carried out extensively and complexly in the E. C. It is apparent also in the L. C., but there is more conspicuous by its absence through hundreds of pages of names of countries, places, and persons. Time does not permit me to describe here the six schedules that economize the system I have worked out: Schedule 1, Mnemonic numerals, constant throughout; Schedule 2, for subdivision by countries, applicable under subjects, where-ever desired; Schedule 3, for subdivisions under countries and localities; Schedule 4, for subjects under any language, except the chief literary languages; Schedule 5, for the chief literary languages; and Schedule 6, for arranging the material under any prominent author.
Some who admit the feasibility of better classification object that a classification modern for the present will be out of date in a generation. This in new guise is the familiar argument that it is useless to clean the house today, for it will need again to be cleaned next week—which all good housewives say is an unreasonable argument. It would be a pity to have fair librarianship called a slouch.
Is it conceivable that your books shall remain forever classified as they are at present? Are there to be no changes, merely additions of new captions? Conservatism is not strange, considering the cost of changing notation; but that cost is small compared with the cost of new building or new collections, and is justified by the service to be rendered. The longer postponed, the larger the cost, the larger the burden. Some libraries are changing now—to what? That change may indeed have to be changed again in a decade or two. But how long, then, should a classification endure—or rather, be endurable? One who would not prophesy may nevertheless give an opinion. I believe that a good classification should last a century—with some minor alterations. I believe that a good library should be willing to reclassify, if necessary, at least some of its collections two or three times in a century. I think that library economy should have been developed with better regard to this problem. It is not practical to arrange books inconsistently with the scientific and pedagogic organization of knowledge. Organization based on consensus is one of the marked tendencies of modern thought and purpose, and is not likely to be overcome by dissenting or disintegrating philosophical counter-tendencies. This organization is more stable than the theories on which it rests, and these are more stable than the popular press would lead us to suppose. New theories, new statements, are assimilated to the established body of knowledge without much dislocation of members. Durability in a system would depend not only upon present consistency with the organization of knowledge, but upon flexibility through reserved and alternative locations, judiciously chosen with regard to tendencies in science. There might be flaws and errors, but all practice, in whatever profession is thus imperfect and tentative.