But domestic varieties are not strictly adapted by Natural Selection to definite conditions, and thus have less specialised sexual constitutions.) This elimination, though admitted by many authors, rests on very slight evidence, yet I think is very probably true, as may be inferred from the case of dogs.

Under nature it seems improbable that the differences in the reproductive constitution, on which the sterility of any two species when crossed depends, can be acquired directly by Natural Selection; for it is of no advantage to the species.

Such differences in reproductive constitution must stand in correlation with some other differences; but how impossible to conjecture what these are! Reflect on the case of the variations of Verbascum, which differ in no other respect whatever besides the fluctuating element of the colour of the flower, and yet it is impossible to resist Gartner's evidence, that this difference in the colour does affect the mutual fertility of the varieties.
The whole case seems to me far too mysterious to rest (92/4.

The word "rest" seems to be used in place of "to serve as a foundation for.") a valid attack on the theory of modification of species, though, as you say, it offers excellent ground for a mere advocate.
I am surprised, considering how ignorant we are on very many points, [that] more weak parts in my book have not as yet been pointed out to me.

No doubt many will be.