the opposite contingencies of increased and diminished fertility. It is a perfectly well ascertained fact[302] that hybrids will invariably breed more readily with either pure parent, and not rarely with a distinct species, than with each other. Herbert is inclined to explain even this fact by the advantage derived from a cross; but Gaertner more justly accounts for it by the pollen of the hybrid, and probably its ovules, being in some degree vitiated, whereas the pollen and ovules of both pure parents and of any third species are sound. Nevertheless there are some well-ascertained and remarkable facts, which, as we shall immediately see, show that the act of crossing in itself undoubtedly tends to increase or re-establish the fertility of hybrids. _On certain Hermaphrodite Plants which, either normally or abnormally, require to be fertilised by pollen from a distinct individual or species._ The facts now to be given differ from those hitherto detailed, as the self-sterility does not here result from long-continued, {132} close interbreeding. These facts are, however, connected with our present subject, because a cross with a distinct individual is shown to be either necessary or advantageous. Dimorphic and trimorphic plants, though they are hermaphrodites, must be reciprocally crossed, one set of forms by the other, in order to be fully fertile, and in some cases to be fertile in any degree. But I should not have noticed these plants, had it not been for the following cases given by Dr. Hildebrand:[303]-- _Primula sinensis_ is a reciprocally dimorphic species: Dr. Hildebrand fertilised twenty-eight flowers of both forms, each by pollen of the other form, and obtained the full number of capsules containing on an average 42.7 seed per capsule; here we have complete and normal fertility. He then fertilised forty-two flowers of both forms with pollen of the same form, but taken from a distinct plant, and all produced capsules containing on an average only 19.6 seed. Lastly, and here we come to our more immediate point, he fertilised forty-eight flowers of both forms with pollen of the same form, taken from the same flower, and now he obtained only thirty-two capsules, and these contained on an average 18.6 seed, or one less per capsule than in the former case. So that, with these illegitimate unions, the act of impregnation is less assured, and the fertility slightly less, when the pollen and ovules belong to the same flower, than when belonging to two distinct individuals of the same form. Dr. Hildebrand has recently made analogous experiments on the long-styled form of _Oxalis rosea_, with the same result.[304]