ROUGH FUR. Ill 



rough. This hypothesis explains very satisfactorily all of the crosses 

 given except those under (7), repeated under (9) and (10). It can not 

 explain the case of male 4003 rough E, who was undoubtedly homo- 

 zygous for the rough factor (RfRf) and yet was the lowest grade of 

 partial-rough. Further, it can not explain the occurrence of partial- 

 rough young coming from the cross partial-rough by 4- toe smooth. 

 The latter are necessarily rfrf under the above hypothesis. 



Rfrf X rfrf = Rfrf + rfrf'. 



Rough C, D, E X smooth = rough A + smooth. 



Under this hypothesis, the rough factor Rf and the factor which 

 reduces the grade of rough rf ' can not be present in the same gamete. 

 But this cross actually gave 70 partial-rough young, 56 from tricolor 

 partial-roughs, 12 from partial-rough cutleri hybrids, and at least 1, 

 probably 2, where the partial-rough parent owed its low grade to Cavia 

 rufescens. Female A606 rough C was | rufescens. Her parents were 

 2193, a full-rough guinea-pig, and A63, a smooth rufescens hybrid. 

 The hypothetical factor rf' could only have come from the latter. 

 The parents of A63 were A55, a pure Cavia rufescens, and 9586 of 

 BW stock, a stock which has shown no tendency to reduce the grade 

 of full-roughs on crossing with them. It thus seems clear that A606 

 owes her low grade of rough to her Cavia rufescens grandfather. She 

 was crossed with a 4-toe smooth male, 166, and had two rough young, 

 whose grades unfortunately were not recorded at birth. One, however, 

 A1687, is still living (August 1915), and is a typical rough C. There is 

 reason for believing the other to have been of the same grade. Thus in 

 tricolors, Cavia cutleri and Cavia rufescens hybrids, the same gamete can 

 transmit the rough factor and the factor or factors which limit the full 

 development of the rough character. The formula Rfrf can not, 

 therefore, be used for partial-roughs. 



There remains only one line of explanation. Partial-roughs must 

 differ from full-roughs by possessing an independently inherited modi- 

 fying factor (or factors). An incompletely dominant unit modifying 

 factor will explain all of the results satisfactorily. Let us represent the 

 wild condition (aperea, rufescens, cutleri) by rrSS. Let us suppose that 

 the dominant mutation R is necessary for any roughness [of series I] 

 and produces with rare exceptions at least reversal of hair direction 

 on the hind toes (rough E). The second mutation, s, when heterozy- 

 gous, may permit roughness to extend to grades D or C ; when homozy- 

 gous it permits roughness to reach grades B or A. 



rrSS smooth Wild species, lea stock. 



rrSs smooth Most tricolor smooths. 



rrss smooth 4-toe, BW, BB, dilute stock, most Lima smooths. 



RrSS rough E (rarely smooth) R 140, etc. 



RRSS rough E (rarely smooth) 4003. 



RrSs rough C or D (rarely E or B) . . . Most tricolor partial-roughs. 

 RRSs rough C or D. 



RrSS rough A (less frequently B) . . . . 4-toe, most Lima roughs. 

 RRss rough A. 



