lO HEREDITY OF HAIR-LENGTH IN GUINEA-PIGS. 



from an experiment like the foregoing is that there will be produced three 

 times as many animals of type B as of type D, or in a total of fifty-one 

 voung (the number whose character was definitely ascertained) thirty-eight 

 B to thirteen D. Comparing with these numbers the observed ones (twenty- 

 nine B and ten D), we notice that there are fewer individuals than expected 

 in each of the classes B and D, from which we may conclude that each has 

 contributed to the formation of the intermediate class C. The hypothesis 

 on which the IMendelian expectation rests is this, that each cross-bred animal 

 will form in equal numbers gametes bearing the pure B character and the 

 pure D character respectively. Evidently from this experiment it follows 

 either that some of the expected B and D gametes are not pure, or else that 

 in this generation the result of a union between a B and a D gamete is not 

 the same as in the original cross. In either case we are forced to admit 

 modification of gametes from their original pure condition. 



In two other families of guinea-pigs, cross-breds mated inter se pro- 

 duced a result similar to that already described. The young in these families 

 were classified as nine B, five C, two D and four undetermined. These cases 

 differ from the foregoing in that, in producing the cross-breds used, two 

 crosses with the B race were made to one with the D race, whereas in pro- 

 ducing the first mentioned lot of cross-breds the B and D stocks were 

 equally represented in the ancestry. It is perhaps significant that the D 

 young are relatively fewer in these families, though the total is too small 

 to allow one to attach much importance to the proportions observed among 



the young. 



(c) M.\TiNG B(D) XD. 



On the Mendelian hypothesis of pure gametes, this cross should yield 

 classes B and D in equal numbers. In matings in which three different D 

 males were employed, it has produced fourteen B. seventeen C and nineteen 

 D offspring, together with five of undetermined character. Since D parents 

 are known to form only D gametes, it seems clear from this experiment that 

 the hybrid, or B(D) parent, formed gametes a considerable number of 

 which bore the intermediate or C character. If a single cross of D with B 

 has such an effect in modifying gametes, a repetition of the cross should 

 have a still more marked influence, producing a still larger proportion of 

 intermediate or C gametes. The mating next to be described bears on 

 this question. 



(rf) Mating B{D) [from Two Successh-e Crosses wrrn B] X D. 

 Three diflFerent D males were used in making this cross. They pro- 

 duced thirty -two B, thirty-seven C, and twenty-nine D offspring, together 

 with seven of undetermined character. Here the C class is actually larger 



