Given the occurrence of variations in certain individuals of a species, we have the alternative logical possibilities of their being inherited or their not being inherited. The latter alternative seems at first sight to be in contradiction to the law of persistence. Sir Henry Holland, seeing this, remarked that the real subject of surprise is, not that a character should be inherited, but that any should ever fail to be inherited.[DS] Intercrossing may diminish a character, and sooner or later practically obliterate it: annihilate it at once and in the first generation it cannot. This logical view, however, ceases to be binding if we admit, with Professor Weismann, that variations may be produced in the body without in any way affecting the germ. It is also vitally affected if we believe that the hen does not produce the egg, though she may, perhaps, modify the eggs inside her; for the modification of the hen (i.e. the variety in question) may not be of the right nature or of sufficient strength to impress itself upon the germinal matter of the egg. We may at once admit, then, that acquired variations need not be inherited. Passing to innate variations--variations, that is to say, which are the outcome of normal development from the fertilized ovum--must they be inherited, at any rate, in some degree? It seems to me that they must, on the hypothesis that sexual generation involves simply the blending or commingling of the characters handed on in the ovum or the sperm. The only cases where this would _apparently_ fail to hold good would be where the ovum and the sperm handed on exactly opposite tendencies--a variation in excess contributed by the male precisely counterbalancing a variation in the opposite direction contributed by the female parent. Even here the tendency is inherited, though it is counterbalanced. On the hypothesis of "organic combination" before alluded to (p. 150), variations might, however, in the union of ovum and sperm, be not only neutralized, but augmented. If the variation be, so to speak, a definite organic compound resulting from a fortunate combination of characters in ovum and sperm, it might either fail altogether, or be repeated in an enfeebled form, or augmented in the offspring, according as the new conditions of combination were unfavourable or favourable. Whether innate variations ever actually fail to be inherited, even in an enfeebled form, it is very difficult to say; for if this, that, or the other variation fail to be thus inherited, it is difficult to exclude the possibility of its being an acquired variation not truly innate. Certainly variations seem sometimes to appear in one generation, and not to be inherited at all. And, as we have seen, Mr. Romanes appeals to a gradual failure of heredity, apart from intercrossing, to explain the diminution of disused organs.