OENOTHERA RUBRINERVIS, A HALF MUTANT. 



361 



cultures were only small. Thus we see that the expectation from 

 our formula is confirmed by the experiment. 



Twin hybrids of 0. rubrinervis. — The twin hybrids of 0. La- 

 marckiana and 0. grandiflora are now explained as the result of the 

 mass mutation of these species, but the experimental proof is not 

 complete as yet because neither of these species is known to occur 

 without that form of mutation. In this respect the case of 0. rubri- 

 nervis is far stronger, since its two constituents are both represented 

 in my cultures. This fact makes a complete analysis possible, as I 

 have already pointed out. If 0. rubrinervis is split by some cross 

 into laeta and velutina on account of its composition of gametes 

 of deserens and velutina, then the corresponding cross with 0. deserens 

 must evidently give the same laeta and that with 0. Lamar ckiana 

 mut. velutina the same velutina. Thus the split progeny can be du- 

 plicated by the addition of its components. 



I have described the splitting crosses in Gruppenweise Artbildung 

 (pp. 122, 190—200, 1913) and repeated some of them so as to have 

 the dimorphic progeny together with the cultures of the presumed 

 constituents, in order to be able to identify their characters during 

 the whole time of their development. The percentage figures given 

 in my book are as follows: 



Twin hybrids of O. rubrinervis. 



Cross 



Percentage 

 of laeta 



O. biennis x rubrinervis 



O. rubrinervis xO. biennis Chicago 



O. rubrinervis x O. Cockerelli 



Mean 



30-49 

 39-44 



49 



42 



Percentage 

 of velutina 



51-70 

 56-61 



51 



58 



In the second and third generation of the two latter crosses the 

 laeta have split off brittle rubrinervis in about one-third of the cul- 

 tures, whereas the velutina remained constant. 



I repeated the two first named crosses in 1915, but not the third 



