COBB AND BARTLETT: INHERITANCE IN OENOTHERA 48 1 



strain E, however, is not explainable by any such hypothesis. 

 We have shown that the reciprocal crosses between strains C and 

 E have a different factorial composition, and do not behave at 

 all alike. To view as due to metacliny the occurrence of mut. 

 nummularia in the cross E X C (see table 3) is inconsistent Avith 

 our explanation of the Mendelian behavior of the character-pair 

 flatness vs. revoluteness in the same crosses. 



A logical explanation would appear to follow if we assume that 

 the round leaves of mut. nummularia, although brought about 

 by a change in the a group of chromosomes (or chromosome) of 

 the female gamete, can come to expression only in the presence 

 of the Mendelian factor F which can be introduced by the 

 male /S gamete of strain C. In other words, the round leaves 

 of nummularia are due to the modification of a, but can 

 reach expression only in the presence of the mendelian factor 

 F. Just as mut. formosa can attain expression only in 

 the absence of the factor F, and consequently never 

 occurs in pure strain C, of which the f. typica has the com- 

 position o;/3FF, so mut. nummularia attains expression only in 

 the presence of the same factor, and is consequently never thrown 

 by strain E, in which the f. typica has the composition o;/3ff. On 

 crossing the two strains, the a portion of the E gamete shows its 

 potentiality for producing mut. nummularia when factor F is 

 present; conversely the a portion of the C gamete shows its 

 potentiality for the production of revolute-leaved mutations when 

 the factor F is removed. 



^ SUMMARY. 



I. In Oenothera pratincola normal fertilization is assumed to 

 take place by the fusion of two kinds of gametes, which are 

 designated as a. and /3, respectively. Certain chromosomes of 

 the a. gametes carry character determiners which have no coun- 

 terparts in the /3 gametes. On account of this unbalanced con- 

 dition, mutations affecting uncompensated determiners appear 

 at once, and neither in origin nor subsequently in heredity do 

 they display Mendelian behavior. 



