20 



BOTANICAL GAZETTE 



[JANUARY- 



difference between the deserens and the lucida was very striking,. 

 the first reaching only half the height of the latter. I have broken 

 the stems of all the plants in August, at the time of the last counting, 

 and found all the deserens brittle and all the tall ones tough. The 

 first were evidently deserens and not rubrinervis, as seen by the 

 characters described for these two types. Among the tall ones, 

 however, I have not succeeded in finding any difference, the type 

 of hicida prevailing to the apparent exclusion of that of O. Lamarcki- 

 ana. For each of the crosses mentioned in the following table I 

 had 58-60 flowering specimens in August. 



Splitting progeny of hybrid lucida; culture of 191 7 



Oenothera Lamarckiana mut. oblonga and mut. nanella 



Our conception of Oenothera rubrinervis as a half mutant may 

 be applied to O. oblonga, and explain its behavior in crosses in an 

 analogous way. The main difference, as I have pointed out in 

 Gruppenweise Artbildung, is that some types of hybrids, as we might 

 expect, are constantly absent or suppressed, as I called it. If we 

 assume this suppression to take place in the pollen before fecunda- 

 tion, the remaining phenomena are easily explained on this basis. 

 It will be sufficient to review the facts given in my book, and to 

 combine them with the results of some determinations of the amount 

 of barren grains in the seeds of self-fertilized and crossed individuals. 



The amount of empty seeds is about the same in O. oblonga as 

 in 0. Lamarckiana. For the cultures of 191 1, mentioned in my 

 book, I found among the seeds of two self-fertilized individuals 

 25 and 33 per cent of germs. Seeds of biennial plants collected in 

 1 9 13 contained 30-18 and 17 per cent of germs; but seeds of annual 



