able to take note
of in subdivision titles. A moment's thought shows the impossibility of
taking care of any large number of combined characteristics so as to
provide exactly for each combination, for the reason that the
limitations of space and of the perceptive faculties forbid. For a
simple illustration, the imaginary classification of books for use by a
bookseller may be recurred to. The dealer, it may be assumed, has books
on (1) four different subjects, history, science, art, and fiction, (2)
each printed in four languages, English, German, French, Spanish, (3) in
four different sizes of page, folio, quarto, octavo, duodecimo, (4)
bound in four materials, leather, rawhide, cloth, paper. Here are four
main characteristics, each in four varieties. A customer is likely to
ask for Ivanhoe in English, octavo, bound in leather. Now if the
bookseller had sought to arrange the books into one class according to
subject matter, into another according to language, another according to
size, another according to binding, he would have fallen into confusion,
because his classes would be formed on different principles or bases and
overlap. Some histories will be in French, some will have octavo pages,
and some cloth bindings. But if he divides first on the basis of subject
matter, then each subject matter into language, each language book into
sizes, each size into material of binding, he can immediately place his
hand on a class wherein the book will be if he has it; but this
classification, based on four different characteristics and four
varieties of each, has necessitated the formation of 256 classes or
divisions, and if five characteristics were provided for, 1,024
divisions would be required.

Adapting the illustration of the books to a patent office
classification: If it were possible to view these characteristics as
patentable in combinations of all or in any combinations less than all,
and also as separate characteristics, 16 divisions additional to the 256
for each independent characteristic would have to be provided, as well
as other divisions for combinations of less than the whole, in order to
make the classification absolutely indicative of every feature, and the
number of divisions would be enormous. In such a classification, after
the proper division had been located, the search would be nothing, the
difficulty would be to find the appropriate class.

_Expedients to reduce the number of subdivisions._--Fortunat