Of the 7 cock-feathered males, one was like the F₁ _castrated_ males; another had a similar back, but a darker and differently marked breast; 2 were game-cock type; 3 were odd birds much like the game cock above except for absence of black, with reddish heads without any black. The males may be approximately classified as follows: +----------+-----------------------------------------------+ | | Back-cross F₁ ♀ by game ♂. | | +---------------+--------------+----------------+ | | Hen-feathered | Intermediate | Cock-feathered | | | ♂. | ♂. | ♂. | +----------+---------------+--------------+----------------+ | F (or K) | | 1 | 1 | | (C) | | 1 | 2 | | B | | | 2 | | A | | | 2 | | | | +----------------+ | | | | 7 | +----------+---------------+--------------+----------------+ Four or five types may then be recognized in this rough grouping. None of the groups seem uniform and probably might be split again. D. THE NUMBER OF COLOR FACTORS INVOLVED. The theoretical expectation for two pairs of factors calls for 4 classes in the back-cross, but this assumes that the parent type used for back-crossing contains all (here 2) recessives. But this simple assumption can not be true in this case, for the F₁ bird would have been like the Sebright. On the 3-factor assumption the expectation for the back-cross is 8 classes, but this would apply only if the game were the triple-recessive form, which, again, it is not, as shown by the F₁ cross. But if the dominance of one or more of the Sebright color factors is incomplete, then either a 2 or 3 factor assumption might apply to the back-cross. If only 2 pairs of factors are present we should expect to recover the game type once in 16 cases in F₂. But, as will be shown, only 1 game was recovered out of the 49 F₂ females. This result fits better with a 3-factor assumption, for even with the small number in the back-cross the indications are that more than 4 classes are present. In the F₂ birds at least 11 classes may be distinguished, and some of these appear composite. For 3 factors the maximum number of possible classes (including heterozygotes) is 27. We can recognize at least 11 F₂ classes amongst the females alone, and a few others are doubtfully present in the males. In favor of the view that the heterozygous classes are here different from the homozygous, the following evidence may be utilized: