OENOTHERA RUBRINERV1S, A HALF MUTANT. 



373 



should be about 50, but in my experiments there was much flucta- 

 tion in this respect. I found as follows: 



Crosses with 0. rubrinervis also yield the expected result. This 

 would be in one direction (obl.+velu.)x(deserens+velu.)=obl.x 

 des.+obl.xvelu.+velu.xdeserens+velu.xve!u.:=25 per cent (obi. 

 xdes.) +25 per cent obionga +25 per cent rubrinervis +25 per cent 

 empty grains. I have not as yet tried the cross between 0. obionga 

 and 0. deserens, however, and thus must leave undecided the ques- 

 tion as to which characters will dominate in this hybrid. As a matter 

 of fact, I found 20 per cent obionga and 80 per cent rubrinervis and 

 no other types. The reciprocal cross must give (des.+velu.)x 

 velutina=des.xvelu.+velu.xvelu.=50 per cent rubrinervis +50 

 per cent empty grains. Only rubrinervis have been observed in this 

 culture. 



Crosses with the pollen of 0. biennis must give oblongax biennis + 

 velutinax biennis. The former is intermediate between the parents, 

 whereas the second is the same as the hybrid type Lamarckiana 

 x biennis. I found in one cross 65 per cent obionga (partly dwarfish) 

 and 35 per cent hybrids of the second type. In another instance, 

 however, the obionga failed from some unknown reason. 



With those species which ordinarily produce the twins densa 

 and laxa the pollen of 0. obionga must evidently give only the latter 

 type. This has been the case in three trials with 0. biennis Chicago 

 and in one with 0. atrovirens (cruciata). 



For further details and for the constancy or splitting in the 

 second generation I must refer the reader to the pages of my book 

 already quoted. These results, however, show clearly that all the 

 facts hitherto ascertained confirm the formula assumed for the self- 

 fertilization, and thereby the analogy with the phenomena observed 

 in 0. rubrinervis. 



Summing up this discussion we may say, therefore, that 0. obionga 



