OENOTHERA GRANDIFLORA AIT. 233 



dwarf mutants (8). These spring from the self-fertilized strain of 

 0. gigas in about 1 — 2 per cent of the offspring of every generation, 

 and have done so since the very origin of the parent form. Arti- 

 ficially crossed with their parent, they produce hybrids of high 

 stature, which are not externally distinguishable from 0. gigas itself, 

 and which split, in the next generation, into three types, according 

 to the formula of Mendel for the monohybrids. Assuming, as is now 

 generally conceded, that mutations take place before fecundation, 

 we can easily see that the gametes of 0. gigas which mutated into 

 nanella must for some part be united in fecundation with normal 

 sexual cells. Such combinations must produce half mutants, as I 

 called them in my book Gruppemveise Artbildung 1 ), or mutant 

 hybrids, as they have since been called (8, p. 345), and these will 

 split in the next generation into about one fourth dwarfs, one- 

 fourth high and normal gigas, and one-half new mutant hybrids. 

 The latter will continue to reproduce the splitting in the succeeding 

 generations, and this may obviously be repeated during an un- 

 limited number of successive years. 



If we now suppose that, by means of some contrivance, the 

 dwarfs and the constant high specimens were yearly eliminated 

 before flowering, we should have a race which would produce in 

 every generation about one-fourth dwarfs. The phenomenon would, 

 then be an instance of mass mutation, and we may choose it as the 

 prototype from which to explain our observations on 0. grandiflora. 

 From our point of view the splitting would be a repeated appearance 

 of the dwarf mutants, due to the original mutation of one gamete. 

 For this reason we shall call it a secondary mutation. 



Let us now consider the strain of 0. grandiflora, found in 1912 

 near Castleberry, Alabama, as such a mutant hybrid, originally 

 produced by the mutation of a sexual cell into ochracea and its 

 copulation with a normal gamete of the strain. Whether this initial 

 mutation took place a few or many years before 1 9 1 2 is of course without 

 interest for this discussion. It may even be older than the species 

 itself. We further assume a close analogy with the mutant hybrids 

 of 0. gigas nanella. This conduces to the expectation of three types 

 in every generation, namely, constant ochracea, constant grandi- 

 flora, and hybrid mutants. 



The ochracea are our secondary mutants; they were seen to arise 



*) The case is especially clear in the instance of 0. Lamarckiana semigigas r 

 where the half mutants with their 21 chromosomes are obviously the result 

 of the copulation of normal gametes with others mutated into gigas. 



