r any
particular matter can not always be limited to one group without such
extensive cross-referencing as would in some cases defeat the purpose of
classification. Forming the subdivisions of a class according to the
total similarities of the inventions, rather than according to some
selected more or less important characteristic, and arranging them in
the correct order of superiority and inferiority, with care to maintain
throughout the schedule the relative positional values of the several
selected bases of division, will ordinarily in a closely bonded class
limit the search for any single invention to the subclass particularly
suited to receive it and some subclasses preceding that one, excluding
from the necessity of search the subclasses succeeding.

Example: In Class 80, Metal-Rolling, it would not be
expected to find any tube-rolling mill lower in the
schedule than the tube-rolling subclasses, but a tube-mill
might be found higher up in "Heating and rolling," "Drawing
and rolling," etc. No concave and roll combination should
be found succeeding the subclass of "Concave and roll," but
it may be found under subclasses above, such as "Tubes,
Screw-threads," etc. No rolls should be found lower than
the subclass of "Rolls," but they may be found in many
subclasses above.

DEFINITION.

(13) Having some knowledge of the nature of the materials about to be
classified, a tentative definition of a class to be formed may be
framed, which may be either written down or merely carried in mind, to
serve as a tentative guide. This tentative definition must be considered
as subject to change to any extent by the fuller knowledge obtained by
careful consideration of the material. After a full knowledge of the
materials to be classified has been acquired, it will be necessary to
frame a careful definition of the class, and also of each subclass whose
title does not unequivocally indicate what is contained in it.

(14) A definition of any class should state the "qualities and
circumstances possessed by all the objects that are intended to be
included in the class and not possessed completely by any other
objects." A proper definition should not ordinarily contain the name of
the thing defined. "Definitions in a circle" are, of course, worthless.
A definition should be exactly equivalent to the species defined and
should not be expressed in obscure or ambiguous language, but should
employ terms