OENOTHERA GRANDIFLORA AIT. 



231 



be analogous to the one deduced from our previous table (24 per 

 cent). Apart from this splitting, the types T and L had a uniform 

 progeny. The third generation of ovata repeated the constitution 

 of the second in its most essential points. 



The offspring of self-fertilized lutea consisted of this form and 

 type R, but none of the other secondary types appeared among 

 them. Table X is made up in the same way as that for ovata. In 

 one instance a third generation, derived from two successive genera- 

 tions of lutea, was also studied. 



Table X 

 Second and third generation of lutea cultivated in 1916 



The percentage for R is 38 per cent, coinciding sufficiently with 

 those in the two previous cases (24 per cent and 42 per cent). 



As shown in table VII, I have self -fertilized specimens of brunnea 

 in 7 different cultures of the first generation. I cultivated 60—70 

 offspring from each of them and studied them all during their whole 

 lifetime until the first fruits began to ripen in August. One of them 

 flowered in 1915, and one plant was fertilized so as to have a second 

 generation of brunnea in 1916. It embraced 70 flowering plants. 

 All these cultures were uniform; they produced no R and no others; 

 and not even a lorea in the beds derived from crosses of this mutant. 

 The same table shows three self-fertilized specimens of contraria, 

 the offspring of which was studied, in 60, 70, and 72 flowering in- 

 dividuals in 1916. In one case there were three doubtful specimens 

 like type R, but apart from these the cultures were uniform and 

 like their parents. 



We now come to the crosses of the mutant lorea. The special mark 

 consisting in the almost linear leaves, was latent in the first gene- 

 ration, but was seen to return in the second, whenever specimens 

 of ovata or lutea are self-fertilized. The brunnea, however, did not 

 split them off, as we have just seen. I made the following cultures 



