agrams, we must
remember, that, whatever kind and amount of variations are exhibited by
the few specimens here compared, would be greatly extended and brought
into symmetrical form if large numbers--thousands or millions--were
subjected to the same process of measurement and registration. We know,
from the general law which governs variations from a mean value, that
with increasing numbers the range of variation of each part would
increase also, at first rather rapidly and then more slowly; while gaps
and irregularities would be gradually filled up, and at length the
distribution of the dots would indicate a tolerably regular curve of
double curvature like those shown in Fig. 11. The great divergence of
the dots, when even a few specimens are compared, shows that the curve,
with high numbers, would be a flat one like the lower curve in the
illustration here given. This being the case it would follow that a very
large proportion of the total number of individuals constituting a
species would diverge considerably from its average condition as regards
each part or organ; and as we know from the previous diagrams of
variation (Figs. 1 to 7) that each part varies to a considerable extent,
_independently_, the materials constantly ready for natural selection to
act upon are abundant in quantity and very varied in kind. Almost any
combination of variations of distinct parts will be available, where
required; and this, as we shall see further on, obviates one of the most
weighty objections which have been urged against the efficiency of
natural selection in producing new species, genera, and higher groups.

[Illustration: FIG. 12.]

_Variation in the Mammalia._

Owing to the generally large size of this class of animals, and the
comparatively small number of naturalists who study them, large series
of specimens are only occasionally examined and compared, and thus the
materials for determining the question of their variability in a state
of nature are comparatively scanty. The fact that our domestic animals
belonging to this group, especially dogs, present extreme varieties not
surpassed even by pigeons and poultry among birds, renders it almost
certain that an equal amount of variability exists in the wild state;
and this is confirmed by the example of a species of squirrel (Sciurus
carolinensis), of which sixteen specimens, all males and all taken in
Florida, were measured and tabulated by Mr. Allen. The diagram here
giv