No. 558] HEREDITY OF TRICOLOR IN GUINEA-PIGS 347 



areas that have the white factor. In pigeons the dark 

 wing-bar of some breeds may be white in other breeds, 

 although pigment is present, elsewhere. We can not as- 

 sume, of course, a pigment producer to be absent from 

 the germ. It seems more probable that there are special 

 color producers, which if present in the germ, and there- 

 fore in all the body cells, give a definite reaction in that 

 region where a white band is formed. In this case there 

 is no localization factor inherent as such, i. e., there is no 

 need to assume somatic segregation, but only germinal 

 segregation of a particular special factor that is realized 

 in a special part. The substitution of a white area for a 

 colored one in guinea-pigs might be looked at in the same 

 way. But the extent to which the spot develops is a more 

 difficult and perhaps a different problem. 



The most obvious objection to Castle's hypothesis of 

 overlapping areas is the excess of bicolors recorded both 

 in his own and in our results, assuming that no true bi- 

 colors were in the stock. An exact lap of the black 

 area over the red (yellow) could happen only when the 

 black spots were of the same size or larger, and occur 

 in exactly the same places as the red area left by the dis- 

 tribution of white producing factor. This would be ex- 

 pected to happen rarely, but, as stated above, tricolors 

 throw a considerable percentage of bicolors. 



Our matings show that the distributor for black is dom- 

 inant, as seen in tricolor by uniform and tortoise by uni- 

 form giving tortoise ; and tricolor by tricolor giving tri- 

 color black. On this basis our original race of tricolors 

 must have been heterozygous for the black distributor, 

 and hence could throw some bicolor blacks which are real 

 bicolors, not overlapped bicolors. This explains our ex- 

 cess of bicolor black which belonged to both types. 



