354 OENOTHERA RUBRINERV1S, A HALF MUTANT. 



gamete of 0. Lamarckiana, gave rise to a half mutant, 0. rubrinervis- 

 In other words, 0. rubrinervis arose as a half mutant between poten- 

 tial 0. deserens and normal 0. Lamarckiana. This half mutant, after 

 articial self-fe rtlization, must have produced a splitting into three 

 types, exactly in the same way as this splitting can be observed in 

 the half mutants of 0. gigas nanella. Of these types two must be 

 constant, but the third must repeat the splitting. 0. deserens is one 

 of the constant ones, whereas the other is assumed to be hidden 

 in the empty seeds, containing a lethal factor just as in 0. grandi- 

 flora and 0. Lamarckiana. The third type is the continuance of 

 0. rubrinervis, and repeats the splitting in every generation. 



According to my view 0. Lamarckiana produces yearly two 

 kinds of gametes in consequence of a secondary mutability into 

 velutina. These velutina are linked to a lethal factor, which kills 

 them in the young seeds. If we assume that the mutation into 

 deserens took place in the typical gametes, leaving the velutina un- 

 changed, we would conclude that 0. rubrinervis consists of two 

 types of gametes, even as 0. Lamarckiana, but that both of them 

 are in a mutated condition. One is the new deserens, without lethal 

 factor; the other is the old velutina, linked to a lethal factor. The 

 result of self-fertilization is now easily explained; the copulation 

 of deserens gametes among themselves must produce this form, that 

 of velutina must give empty seeds, and the combination of the two 

 types must repeat the rubrinervis with its splitting capacity. 



On the same basis the occurrence of twin hybrids may be ex- 

 plained, the deserens gametes giving the laeta hybrids; but here 

 we have a considerable advantage over other instances of twin 

 hybrids, since both the constituents are available in pure condition 

 for controlling crosses. Any cross which gives twins with rubrinervis 

 may be repeated with 0. deserens and with my 0. mut. velutina (0. 

 blandina). In the first case the result must be hybrids of the type 

 laeta, in the second case hybrids of the form velutina, and the addition 

 of these must simply duplicate the split progeny of the corresponding 

 cross of 0. rubrinervis. I have made these crosses in a number of 

 cases and found this deduction verified. 



Apart from the described secondary mutability into viable deserens 

 and dead velutina germs, 0. rubrinervis is not known to possess any 

 noticeable degree of mutability; is has, especially, never produced 

 those mutants which are of so common occurrence in allied mutating 

 forms. Thus we see that secondary mutability is not, in itself, to 

 be considered as a cause of further mutations, and this seems to me 



