602 



THE AMERICAN NATURALIST [Vol. XLV 



which I may touch upon very briefly. That is, the exact 

 method of origin of the various mutants, from the cyto- 

 logical standpoint. It is now certain that the nature of 

 the change involved is not the same in all cases. In the 

 case of 0. gigas, the most striking change, which brought 

 with it many size changes, is the doubling in the chromo- 

 some number. This most probably occurred either in 

 the fertilized egg or in the megaspore mother cell, which 

 then developed apogamously (Gates, '11c). On the 

 other hand it seems most probable that several other 

 mutants are the results of changes occurring during the 

 reduction divisions. All the retrogressive mutants may 

 be accounted for in this way, as I have shown (Gates, 

 '08), as the results of occasional irregularities in the dis- 

 tribution of members of the chromosome pairs, if we as- 

 sume the chromosomes to differ in their chemical activ- 

 ities. 0. rubricalyx again is a mutant from 0. rubri- 

 nervis in which a marked quantitative change in one 

 character (namely, capacity for anthocyanin produc- 

 tion) has taken place. We have here the mutational ap- 

 pearance of a new dominant character. A change of 

 this kind is not likely to be concerned with a new chro- 

 mosome distribution, but is perhaps due to a cyto- 

 plasmic difference. As far as can be determined, the 

 external conditions under which this mutant appeared 

 and developed differed in no way from those of the rest 

 of the culture, although it is impossible to deny that 

 some local soil-difference might possibly have actuated 

 this germinal change. This would, however, necessitate 

 the assumption that the change took place in the germi- 

 nating seed. 



Mutation appears, therefore, to be not a simple uni- 

 tary process of splitting, but to be the result of a condi- 

 tion of instability in the germinal material, which is 

 again probably a result of previous crossing, and which 

 leads to various types of departure from the parental 

 race. That this process will account for much species- 

 formation, and the polymorphism of many genera, can 



