234 MASS MUTATIONS AND TWIN HYBRIDS OF 



in my cultures constantly during the three generations of my ex- 

 periments, and every time in large numbers. The mutant hybrids 

 are the apparently normal specimens of grandiflora of my strain; 

 they repeat the splitting in every generation, but no constant 

 grandiflora have been found, since all the specimens tried reproduced 

 the mass mutation. Here, therefore, we have to introduce another 

 principle. This is the assumption of a lethal factor. Morgan and his 

 students have discovered the presence of four such units in their 

 experiments with Drosophila, and from their studies we know 

 exactly what to expect from them (11). I have already proposed 

 this principle for the explanation of the empty seeds of 0. Lamarcki- 

 ana (4), and we may apply it here in the same way. I determined 

 the amount of barren grains among the seeds of my strain of 0. 

 grandiflora and found 12—41 per cent, with an average of 25 per 

 cent for the harvest of 8 self-fertilized plants of different generations 

 (Opera VII, table on p. 115). Now our argument led us to expect 25 per 

 cent of constant specimens, and the hypothesis that these are killed 

 within the seeds by some lethal factor would at once explain their 

 absence and the presence of the barren grains. 



By means of this hypothesis the conception of our strain of 0. 

 grandiflora as a hybrid mutant now becomes complete. It starts 

 from two succeeding initial mutations in sexual cells, which copu- 

 lated with normal ones. One of these was the mutation into a 

 weak, yellowish ochracea, the other was the production of a lethal 

 factor, linked to the non-mutated gametes. This linkage must be 

 assumed to be so complete as not to interfere with the applicability 

 of Mendel's formula for monohybrids. 



The presence of the lethal factor in both the gametes of a copu- 

 lation kills the germ after some time, but the presence of the same 

 factor in only one of the two gametes leaves them viable. This 

 latter proposition is proved by numerous crosses between species 

 witrrbarren grains and those without the factor in question. Such 

 seeds! are always capable of normal development. 



The supposed initial mutations of our strain, therefore, must 

 have produced half mutants, the gametes of which split in every 

 generation into about equal parts of potential grandiflora with the 

 lethal factor, and into viable ochracea. The fecundation must then 

 produce one-fourth of germs of grandiflora with the double lethal 

 factor and thereby be doomed to die off within the seed; one-fourth 

 of viable but weak ochracea, which will be constant in their progeny; 

 and one-half of hybrids between the two mutants, in which the 



