MENDELIAX EXPERIMENT OX FOWLS. 183 



amount of skin-pigment in some cases increased. Conversely, it 

 was noticed in some cases that the amount of pigment in 

 heterozygous individuals was diminished — for example, F. VII 1 (j" 

 was pigmented like all the rest of that brood, but when killed 

 and phicked in August the following year the breast and back 

 were scarcely darker than in a recessive. Both the parents of 

 F. were pigmented, but if one of them were a heterozygous 

 dominant, half the offspring would also be heterozygous, and this 

 was probably the condition of F. YII 1 S . 



Structv.ral Characters. 



The three characters hitherto considei-ed are colour-characters 

 due to pigment in the plumage or in the skin and connective 

 tissue. I have now briefly to review the structui-al characters, of 

 which there are five — namely, silky plumage versus normal, comb, 

 crest, feathering of legs, double hallux. 



In the 10 birds of F., generation tliere were tiiree with silky 

 plumage and seven normal, in the first brood the numbers were 

 two silky to six normal, the proper proportion of one recessive to 

 three dominant. In F^, 23 birds altogether, all were normal, 

 although one parent of some of the broods was silky — the other 

 must therefore have been pure dominant. In later generations 

 recessive silkies were again obtained. In two silkies of F^Y, I 

 tliought the silky charactei- was not perfect, the proximal parts 

 of the large wing-feathers being as in normal plumage, and only 

 the ends loose-barbed. But I was not able to make a careful 

 comparison with the pure silky, so that I have no distinct 

 evidence of imperfect segregation in this character, and in any 

 case it would be difficult to be certain about a slio-ht des'i'ee of 

 the normal character in silky plumage. 



In my previous paper I suggested that the form of the 

 posterior end of the rose-comb was connected with the presence 

 or absence of the crest, and accordingly these two characters may 

 here be considered together. I stated in the previous paper that 

 when the crest is present and large the posterior encl of the 

 rose-comb is truncated and trifid. F^ III 11 (^ seemed to con- 

 tradict this rule, for it had a rose-comb pointed behind together 

 with a crest. But in this case the crest was slight, and the point 

 behind was not so perfect as in cases where the crest is absent. 

 F3 ly 1 2 was another similar case, the rose-comb being pointed 

 behind and the crest very slight. In subsequent generations 

 many cases occurred in which the crest was absent, and these 

 must, of course, be regarded as pure recessives. 



But the crest, when present, showed great variation in size 

 and development, noted by me as very sliglit, moderate, and full. 

 It may be assumed by Mendelians that the full crests are pure 

 dominants and the lower degi'ees ai'e valuations in heterozygotes. 

 I was unable to investigate this by breeding specially the indi- 

 viduals with minimum amount of crest, either with one another 

 or Avith those without crest (recessives). The great variation in 



