In the former case where _A_ and _a_ are characters which can be denoted by reference to a common scale, the law assumes of course that the inheritance will be, to use Galton’s term, _blended_, namely that the zygote resulting from the union of _A_ with _a_ will on the average be more like _a_ than if _A_ had been united with _A_; and conversely that an _Aa_ zygote will on the average _be more like A than an aa zygote would be_. But in the case of _A_’s and _B_’s, which are assumed to be mutually exclusive characters, we cannot speak of blending, but rather, to use Galton’s term, of _alternative_ inheritance. Pearson, finding that the law whether formulated thus, or in the modified form in which he restated it[16], did not express the phenomena of alternative inheritance known to him with sufficient accuracy to justify its strict application to them, and also on general grounds, proposed that the phenomena of blended and alternative inheritance should be treated apart--a suggestion[17] the wisdom of which can scarcely be questioned. [16] In Pearson’s modification the parents contribute 0·3, the grandparents 0·15, the great-grandparents ·075. [17] See the works referred to above. Now the law thus imperfectly set forth and every modification of it is incomplete in one respect. It deals only with the characters of the resulting zygotes and predicates nothing in regard to the gametes which go to form them. A good prediction may be made as to any given group of zygotes, but the various possible constitutions of the gametes are not explicitly treated.