Also, in table 28, matings are varied. Some are probably matings of two heterozygotes, others of two recessives, and others still of a recessive with a heterozygote. On our hypothesis we should expect some of the families of the mated hybrids to show evidence of the inhibiting factor and others to show no such evidence. In those families in which small tail appears it is found in about 19 per cent of the cases. On account of this weakness of the inhibitor in the germ-plasm of No. 117 that inhibitor is rarely fully activated. Only in one case out of the 250 or more in which that germ-plasm is used is the development of the tail completely stopped. In this case a hybrid cock derived from pen 526 (series 2, table 26) was crossed with various birds of tailed races (probable RR's), and produced in addition to 20 tailed offspring 1 devoid of uropygium and oil-gland. In this case we may conceive that an unusually potent condition of the inhibitor wholly stopped the development of the tail. The behavior of No. 116 is that of a pure dominant. Mated to DR (and some RR?) females he produces pure dominants and heterozygotes. His inhibiting factor is potent enough to be active in the DD offspring at least; as a matter of fact 47 per cent of his get have their tails inhibited. Even in the DR's the inhibitor may sometimes work itself out. Thus No. 116 crossed on No. 508, without tailless ancestry, had 56 per cent of the progeny without tail. Since tailless birds may be either pure dominants or DR's, we may expect families of two sorts when two such are bred together--those containing only tailless offspring and those containing only 75 per cent or less of such. Both sorts of families are to be expected in a table with the composition of table 30, and both appear there. The case of the rumpless fowl that arose _de novo_ will be explained, then, as follows: Even in normal RR matings the inhibiting factor may arise by mutation. But even when two of these inhibiting factors are paired they show themselves so weak as not to appear in 25 per cent, much less the typical 75 per cent of cases, but, as in our case, merely 4 per cent. The strain takes on, indeed, the essential features of the "eversporting varieties" of De Vries (1905). It seems probable, therefore, that even in eversporting varieties inheritance may be Mendelian, modified by variations in "potency" as shown by irregularities in dominance. CHAPTER V. WINGLESSNESS.