. 418, vi. p. 574.

{434} _Origin_, Ed. i. pp. 419, 440, vi. pp. 575, 606.

Practically, naturalists seem to classify according to the resemblance
of those parts or organs which in related groups are most uniform, or
vary least{435}: thus the aestivation, or manner in which the petals etc.
are folded over each other, is found to afford an unvarying character in
most families of plants, and accordingly any difference in this respect
would be sufficient to cause the rejection of a species from many
families; but in the Rubiaceae the aestivation is a varying character, and
a botanist would not lay much stress on it, in deciding whether or not
to class a new species in this family. But this rule is obviously so
arbitrary a formula, that most naturalists seem to be convinced that
something ulterior is represented by the natural system; they appear to
think that we only discover by such similarities what the arrangement of
the system is, not that such similarities make the system. We can only
thus understand Linnaeus'{436} well-known saying, that the characters do
not make the genus; but that the genus gives the characters: for a
classification, independent of characters, is here presupposed. Hence
many naturalists have said that the natural system reveals the plan of
the Creator: but without it be specified whether order in time or place,
or what else is meant by the plan of the Creator, such expressions
appear to me to leave the question exactly where it was.

{435} _Origin_, Ed. i. pp. 418, 425, vi. pp. 574, 581.

{436} _Origin_, Ed. i. p. 413, vi. p. 569.

Some naturalists consider that the geographical position{437} of a
species may enter into the consideration of the group into which it
should be placed; and most naturalists (either tacitly or openly) give
value to the different groups, not solely by their relative differences
in structure, but by the number of forms included in them. Thus a genus
containing a few species might be, and has often been, raised into a
family on the discovery of several other species. Many natural families
are retained, although most closely related to other families, from
including a great number of closely similar species. The more logical
naturalist would perhaps, if he could, reject these two contingents in
classification. From these circumstances, and especially from the
undefined objects and criterions of the natural system, the number of
divisions, such as genera,