98 INHERITANCE IN GUINEA-PIGS. 



dominant. Heterozygous light-belly (with a non-agouti parent) by 

 heterozygous ticked-belly (also with a non-agouti parent) gave 16 

 light-bellied, 6 ticked-bellied, and 10 non-agouti young where expecta- 

 tion is 16 : 8 : 8. 



The results given so far show that light-belly is dominant or at least 

 epistatic over ticked-belly, that ticked-belly is a simple Mendelian 

 dominant over non-agouti, and that the difference between rufescens 

 and porcellus agouti is not a question of residual heredity. The fact 

 that crossing with guinea-pig non-agouti increases the difference be- 

 tween rufescens and porcellus agouti, instead of destroying it, shows 

 that rufescens agouti does not contain the same agouti factor as is found 

 in guinea-pig agoutis. Rufescens agouti must have an allelomorph of 

 guinea-pig agouti, recessive to the latter. This leaves two possibilities. 

 This allelomorph may be (I) the non-agouti factor or (II) a new allelo- 

 morph recessive to the porcellus agouti factor, dominant to non-agouti 

 (Detlef sen's hypothesis). Both of these explanations fit equally well 

 all of the data given so far. Under (I) a guinea-pig light-belly is 

 AAa'a', a non-agouti aaa'a', and a rufescens agouti aaA'A'. Under (II) 

 these three varieties are AA, aa, and A'A', respectively. The critical 

 test is whether it is possible to produce light-bellies which are double 

 heterozygotes AaA'a', capable of having both ticked-bellied and non- 

 agouti young, as well as light-bellies when crossed with non-agoutis. 

 Detlefsen obtained 5 light-bellied agoutis from the cross light-belly by 

 heterozygous ticked-belly which bear on this point. Each of these had 

 ticked-bellied young, but no non-agoutis. They, therefore, point 

 toward hypothesis (II), which is also more probable a priori. They 

 had, however, only from 3 to 6 young, 21 in all, so that it is not wholly 

 certain that they would have had no non-agouti young if tested further. 

 This point, therefore, seemed to the writer to be one on which additional 

 data would be desirable, and special attention has been paid to it. 



The cross heterozygous ticked-belly by heterozygous light-bellies 

 known from their parentage to be free from ticked-belly can be repre- 

 sented as follows under the two hypotheses : 



Agtb. Aglb. Aglb. Aglb, Agtb. Non-ag. 

 (I) aaA'a' X Aaa'a' = AaA'a' + Aaa'a' + aaA'a' + aaa'a'. 

 (II) A'a X Aa = AA' -f Aa + A'a + aa 



In both cases we expect 2 light-bellies to 1 ticked-belly to 1 non- 

 agouti. Under (I) the light-bellied young which can transmit ticked- 

 belly (AaA'a) must also have the power of transmitting non-agouti. 

 Under (II) such light-bellies (AA') should not transmit non-agouti. 

 Under (II) half of the light bellies should be of this type and the other 

 half should transmit non-agouti but not ticked-belly (Aa). Thus, if 

 a large number of young can be obtained from a light-belly from such 

 a cross, which has had ticked-bellied young in crosses with non-agoutis, 

 the presence or absence of non-agouti young is decisive between the 



