OENOTHERA RUBRINERVIS, A HALF MUTANT. 



365 



In the first place, I determined the amount of empty grains, 

 using the same method as in previous cases. 



The presence of empty grains is thereby proven, although the 

 percentages of germs are much smaller than would be expected; but 

 this may be due to quite different causes, as has been shown elsewhere. 



In the second place, I studied the living progeny for the fourth 

 cross of this table and for as cross made in 1907 with another strain 

 of 0. Lamarckiana. Each of these cultures was trimorphous, con- 

 taining the types Lamarckiana and lucida, and besides these either 

 subrobusta or rubrinervis (that is, tough or brittle). 



The expectation would be for equal parts, but for some unknown 

 reason the lucida almost always fall short of this. 1 ) Apart from this 

 difficulty the results of these cultures coincide with the theoretical 

 deductions from our formula. I have made quite a number of further 

 crosses between these two forms, partly in 1905 and partly in 1913, 

 using always the same family of rubrinervis, and taking the com- 

 binations in both reciprocal directions. Six of them have given for 

 the third hybrid rubrinervis and three of them subrobusta; but since 

 I have not determined the amount of ludica among them, it is of 

 no use to give the percentage figures. 



The exactness of the identification of the types in the formula 

 can be controlled by direct crosses with the constituents, mut. 



x ) See below: Preferential Fertilization. 



