Natural History, or under Brazil. The decision is still more difficult as regards books that include widely different topics. Journals and other periodicals are specially difficult to class; while publications of many learned societies are almost as troublesome. Such books as that first referred to will probably be found under different heads in different libraries, according to the views of the librarians; while periodicals will receive special locations; not being assigned to the separate subjects, save when devoted to single departments of knowledge. But the specialist desires to have all works that treat on his special department brought together, and, as far as practicable, subjected to close classification under this great division, otherwise he is apt to overlook information of importance. How can these different standpoints best be brought into harmony? Should it be effected by full cross references? If so, these would require to include references to at least the more important papers contained in the several books or periodicals placed under heads apart from the one in question.
In the several sciences there are often parallelisms that are helpful to the memory, and that it is well to indicate by corresponding series in the library. This is peculiarly the case in the more closely related sciences, e.g., Botany and Zoology. This is recognised to some extent in most systems; but it might be more fully recognised. In all classifications it is desirable that, so far as possible, the various grades should be of nearly equivalent importance, apart from the mere number of subordinates in each.
The succession of groups should also be, in so far as possible, a natural one, with no serious dislocations; nor should groups be associated when they have little in common, nor should they include sub-divisions that are not akin. The several groups should be so well defined that each heading should appear but once in the scheme, nor should it be too closely approached by some other heading elsewhere. The neglect of these points leads to difficulty in assigning books to their several positions, and to equal difficulty in securing that all the information on each subject is accessible. Under the several sciences it is not advisable to assign fixed numbers or symbols very minutely in the sections devoted to the groups in classifications. It is probable that the larger divisions, and many of the smaller ones also, that are now recognised by specialists will continue to be so; but even these must be subject to revision in the light of wider knowledge, and there is almost the certainty of more or less extensive changes
