BRITTLE RACES OF OENOTHERA LAMARCKIANA. 697 



result. I counted the empty seeds, produced by the cross, on two 

 individual plants and found 48 and 65, or an average of 56 per cent. 

 The numerical proportions were found to be, besides the 56 per cent 

 barren grains, 32 per cent Lamarckiana, 9 per cent subrobusta, 2 

 per cent lucida, and 1 per cent mutants. These figures differ from 

 the expectation of four groups of equal size, but the deviations are in 

 the same sense as in other crosses, the velutina (barren grains) giving, 

 as a rule, too high, but the lucida too low figures (8). 



In order to compare 0. similis more closely with 0. Lamar ckiana, 

 1 have studied its crosses with 0. Lamarckiana mut. nanella, and its 

 mutability after self-fertilization. I failed to discover any differences. 

 After fertilizing 0. similis with the dwarfs of my race, I got in the 

 first generation partly tall plants and partly dwarfs, the number 

 of the latter being exceptionally high (90—94 per cent). In the 

 second generation both types gave a uniform progeny. After crossing 

 the tall hybrids with their dwarfish sisters, I found the same propor- 

 tion, the offspring consisting of 94 per cent dwarfs and 6 per cent 

 tall specimens. The types were the same as those of the crosses of 

 my typical race of 0. Lamarckiana. 



The mutability was studied in the seeds of the second generation, 

 cultivated as biennial plants in 1913—1914. From five specimens of 

 similis I had after self-fertilization 0.7—1.1 per cent nanella, 0.2—0.3 

 per cent lata, 0.3—0.5 per cent albida, 0.5—1.0 oblonga, a single 

 specimen of semigigas, but no rubrinervis. The seeds of four self- 

 fertilized plants of 1917 gave, among 936 seedlings, 0.5 per cent 

 nanella, 0.9 per cent lata, 0.1 per cent albida, 1.1 per cent oblonga, 

 1 .3 per cent scintillans, and no rubrinervis. It is evident that, apart 

 from the brittleness, the mutability of 0. similis is the same as that 

 of 0. Lamarckiana. We may therefore safely conclude that 0. La- 

 marckiana mut. similis corresponds in almost all respects with its 

 parent species, differing only in the factor for brittleness. This latter 

 is present in Lamarckiana but absent in similis. For this reason 

 0. similis may be considered to be an atavistic mutant. The an- 

 cestors of 0. Lamarckiana must have acquired the factor for brittle- 

 ness at some period; their descendants might lose it at any time. 



Returning to our formula, the difference between the two types 

 must be looked for in the velutina gametes, since the velutina x 

 deserens give subrobusta in the crosses of 0. similis, whereas the 

 corresponding combination gives rubrinervis in the crosses of pure 

 Lamarckiana. The three remaining combinations are the same in 

 both cases. In other words, 0. similis is a mutant which lacks in its 



