all to mind that the amount of external
difference between two species will not safely guide us in foretelling
whether or not they will breed together,--some closely allied species
when crossed being utterly sterile, and others which are extremely
unlike being moderately fertile. I have said that no case of sterility
in crossed races rests on satisfactory evidence; but here is one which
at first seems trustworthy. Mr. Youatt,[222] and a better authority
cannot be quoted, states, that formerly in Lancashire crosses were
frequently made between longhorn and shorthorn cattle; the first cross
was excellent, but the produce was uncertain; in the third or fourth
generation the cows were bad milkers; "in addition to which, there was
much uncertainty whether the cows would conceive; and full one-third of
the cows among some of these half-breds failed to be in calf." This at
first seems a good case; but Mr. Wilkinson states,[223] that a breed
derived from this same cross was actually established in another part
of England; and if it had failed in fertility, the fact would surely
have been noticed. Moreover, supposing that Mr. Youatt had proved his
case, it might be argued that the sterility was wholly due to the two
parent-breeds being descended from primordially distinct species.

I will give a case with plants, to show how difficult it is to get
sufficient evidence. Mr. Sheriff, who has been so successful in the
formation of new races of wheat, fertilised the Hopetoun with the
Talavera; in the first and second generations the produce was
intermediate in character, but in the fourth generation "it was found
to consist of many varieties; nine-tenths of the florets proved barren,
and many of the seeds seemed shrivelled abortions, void of vitality,
and the whole race was evidently verging to extinction."[224] Now,
considering how little these {105} varieties of wheat differ in any
important character, it seems to me very improbable that the sterility
resulted, as Mr. Sheriff thought, from the cross, but from some quite
distinct cause. Until such experiments are many times repeated, it
would be rash to trust them; but unfortunately they have been rarely
tried even once with sufficient care.

Gaertner has recorded a more remarkable and trustworthy case: he
fertilised thirteen panicles (and subsequent