In 1906 I mated 2 of these white (+ red) bantam hybrid hens with a hybrid cock and obtained again red on the wing-coverts of some white hybrids, while some were without red. From one of the hens I got 4 offspring, or 20 per cent of all, with buff on hackle-lacing, breast, and wing-coverts. In 1907 I mated a prevailingly white male of the preceding year, that had red wing-bar, hackle, and breast, with the reddest females and obtained, along with pure whites and blacks and barred birds, these colors combined with red in various degrees, but not clearly in advance of the reddest of 1906. In 1908 I mated a white male, having red as in the Game, with my reddest hybrids. Again, white and white-and-buff birds appeared, but they showed no advance, except in one instance, among 138 young. This individual (No. 7950), derived exclusively from the Black-red Game and White Leghorn on one side and on the other from the White Leghorn-Game Bantam cross, had a _uniform buff_ down. Unfortunately the chick quickly died. The conclusion is that after three years of selection of the reddest offspring no appreciable increase of the red was observed--except for the remarkable case of one undeveloped chick with completely buff down. This, indeed, looks like a sport, or, perhaps, it is due to unsuspected factors. The experiment will be continued. F. NON-INHERITABLE CHARACTERS. So well-nigh universal is heredity that it is justifiable to entertain a doubt whether any character may fail of inheritance. So far as my experience goes, non-inheritable characters are such as are weak in ontogeny, so that they may readily fail of development even when conditions are propitious; or else they are so complex--so far removed from simple unit-characters--that their heritability in accordance with established canons is obscured. The first case is apparently illustrated by the rumpless cock (No. 117) and the wingless fowl; the second case by lop-comb and by right-and-left alternatives in general.