8 BOTANICAL GAZETTE [January 



The result confirms the expectation and shows that the figures 

 given in the former table, although they give proof of the occurrence 

 of mass mutation among the offspring of every plant of rubrinervis, 

 are too low for the appreciation of the exact percentage of deserens. 

 This must be estimated at about 30 per cent, or almost one-third 

 of the whole progeny. In this race I self-fertilized three mutants 

 in 191 3 and two specimens of deserens among their offspring in 191 5. 

 I cultivated 150+84+180 specimens of the first group and left 

 about one-half of them to flower, and 70+89 plants of the second 

 group, all of which flowered in 1916. I had 573 plants in all, among 

 which 319 bore flowers and fruits. They were all uniformly 

 deserens, showing the marks of the type as previously described. 

 No specimens of rubrinervis and no new mutants were observed 

 among them. For one mutant of 19 13 and for two plants of the 

 second generation in 191 5 I determined the amount of germs in the 

 seeds and found 97-96 and 98 per cent, or an almost total absence 

 of empty grains. 



Besides the two described families of O. rubrinervis I have con- 

 trolled the seeds of some mutants in order to know whether all of 

 them contained specimens of O. deserens and in percentages point- 

 ing to mass mutation. I found the following figures: 



Percentage of deserens among the offspring of 



MUTANTS 



The strain of 0. Lamarckiana was derived from a rosette found 

 in the original station near Hilversum in 1905, and the pallescens 

 had been a mutant from this same strain. ^ Although the cultures 

 were small, they prove the existence of mass mutation. I sowed 

 the seeds of a specimen of deserens from the first culture in 1914, 

 cultivated 25 flowering plants, and found these uniform with the 

 type of their parent. 



' New dimorphic mutants of the Oenotheras. Bot. Gaz. 62:262. 1916. 



