374 OENOTHERA RUBRINERVIS, A HALF MUTANT. 



arises through a mutation of the typical sexual cells of 0. Lamarcki- 

 ana, leaving the velutina gametes and also the lethal factors unchanged 

 but producing, besides the externally visible marks of the mutant, 

 a suppression of the mutated pollen grains. 



On the other hand, 0. mut. nanella seems to arise through muta- 

 tions in the velutina gametes of 0. Lamarckiana, as is shown by the 

 fact that the laeta do not split off dwarfs, whereas the velutina regularly 

 do so. The figures given in my book for the crosses with 0. nanella 

 may be calculated in the same way, and will be found to comply 

 with the views proposed in this article. It would lead us too far, 

 however, to reproduce these calculations here. 



In all these cases the conception that mass mutation is the chief 

 cause of the production of twin hybrids evidently makes the sup- 

 position of a labile condition of the factor for laeta superfluous. It 

 seems desirable, therefore, to lay stress on the fact that this sup- 

 position does not rest on the phenomena observed in the produc- 

 tion of these twins. It is mainly derived from other observations, 

 and some of them may be briefly repeated here in order to make 

 this point clear. They refer to the brittleness of 0. rubrinervis and 

 0. deserens and to the dwarfish stature of 0. nanella. 



In crosses brittleness behaves in three different ways. With 0. 

 biennis Chicago and 0. Cockerelli it is recessive to the tough structure 

 of the fibers, since it fails in the first generation and reappears in 

 the second in ratios corresponding to Mendel's law. In crosses with 

 0. Lamarckiana it is sometimes dominant and sometimes recessive 

 as has been shown. In 0. rubrinervis and 0. deserens the toughness 

 is wholly absent. From these and other facts it is clear that at least 

 three conditions of this factor are possible. I call them active, labile, 

 and inactive. Whether the labile condition is due to linkage or to 

 some other cause is as yet an open question, which, however, has 

 no influence upon the main contention. The combination "active x 

 inactive" is assumed to be responsible for Mendelian crosses, but 

 the combination "labile x inactive" may cause a splitting in the first 

 generation and produces, as a rule, constant hybrids. The two types 

 of first generation hybrids appear in variable numerical propor- 

 tions according to different circumstances. If one of the groups is 

 so small as not be represented in every 100 specimens, the splitting 

 may seem to fail, and such extremes are of common occurrence. 

 This would explain the dominance of an evidently recessive character. 



The case is exactly the same for the dwarfish stature. The factor 

 for tallness must be in the inactive condition in the dwarfs, but 



