14 



THE AMERICAN NATURALIST 



[VOL.LH 



zygotes are expected on the assumption of a three-factor 



ABC ABc Abe aBC abC aBc abc 

 abc abc abc abc abc abc abc 

 Of these eight kinds of individuals, some of only one class, 

 might be expected to be wild type (viz., of class abcABC) 

 in the sense that individuals of this class correspond in 

 formula to the F t offspring, and, of these F 1 offspring, 2 

 out of 41 have tails with 12 feathers, or 1 in 20. Amongst 

 the 24 back crossed individuals, there were none with 12 

 feathers only and at most one is expected. If we assign 

 to the group of abcABC also the four individuals of the 

 back cross in the 14 and 16 groups, and assign the 3 indi- 

 viduals of the 30 and 31 groups to the pure fantails, there 

 remain 17 individuals in the middle range that belong to 

 the six intermediate groups that are homozygous in one 

 or in two fantail modifiers. There are six intermediate 

 classes between the end classes just spoken of. If we 

 are right in the limits assigned to the end classes, the ex- 

 pectation would be 18 individuals for the intermediate 

 classes, where 17 are so classified, which is also not a bad 

 fit. 



Four factors fit the data about as well as three, 6 but if 

 three will suffice the smaller number is perhaps preferable. 

 It is evident that the data do not allow close analysis, but 

 only because they are not sufficiently large, especially in 

 the back cross. Nevertheless, it is important to find out 

 that, so far as the results go, they are not unconformable 

 with the Mendelian assumption of segregation of a few 

 pairs of factors. 



Linkage 



When all F 2 tails that are blue are classified they fall 

 into the groups shown in Fig. 6 ; similarly, the white tails 



8 On this assumption relatively fewer fantails are expected in F„ which 

 is a better fit, but fewer also in the back cross, which apparently is not so 

 good a fit. The proportion would also depend, however, on the relative 

 efficiency and the completeness of the dominance of each factor. The 



