ases of dimorphic and trimorphic plants the
different crosses between the same pair of individuals may be fertile or
sterile at the same time. It appears as if fertility depended on such a
delicate adjustment of the male and female elements to each other, that,
unless constantly kept up by the preservation of the most fertile
individuals, sterility is always liable to arise. This preservation
always occurs within the limits of each species, both because fertility
is of the highest importance to the continuance of the race, and also
because sterility (and to a less extent infertility) is self-destructive
as well as injurious to the species.

So long therefore as a species remains undivided, and in occupation of a
continuous area, its fertility is kept up by natural selection; but the
moment it becomes separated, either by geographical or selective
isolation, or by diversity of station or of habits, then, while each
portion must be kept fertile _inter se_, there is nothing to prevent
infertility arising between the two separated portions. As the two
portions will necessarily exist under somewhat different conditions of
life, and will usually have acquired some diversity of form and
colour--both which circumstances we know to be either the cause of
infertility or to be correlated with it,--the fact of some degree of
infertility usually appearing between closely allied but locally or
physiologically segregated species is exactly what we should expect.

The reason why varieties do not usually exhibit a similar amount of
infertility is not difficult to explain. The popular conclusions on this
matter have been drawn chiefly from what occurs among domestic animals,
and we have seen that the very first essential to their becoming
domesticated was that they should continue fertile under changed
conditions of life. During the slow process of the formation of new
varieties by conscious or unconscious selection, fertility has always
been an essential character, and has thus been invariably preserved or
increased; while there is some evidence to show that domestication
itself tends to increase fertility.

Among plants, wild species and varieties have been more frequently
experimented on than among animals, and we accordingly find numerous
cases in which distinct species of plants are perfectly fertile when
crossed, their hybrid offspring being also fertile _inter se_. We also
find some few examples of the converse fact--varieties o