696 BRITTLE RACES OF OENOTHERA LAMARCKIANA. 



three individuals in the third generation in 1917. From one of the 

 two first ones my family of erythrina has been derived. 



Those initial rosettes of 1905 which produced erythrina mutants 

 instead of brittle ones, might be considered as constituting a special 

 mutant race, sprung in the field from the ordinary Lamarckiana. 

 Of one of them seeds were no longer available, but in the third 

 generation derived from the other in 1917, I have made some crosses, 

 besides self-fertilizing some flowers in order to continue the race. 

 I shall call this race 0. Lamarckiana mut. similis, since it is exter- 

 nally an exact copy of its parent species, differing from this in no 

 visible character and at no single period of its development. Now 

 the question arose, whether the production of tough or brittle 

 hybrids is connected with that of throwing off tough or brittle 

 mutants. If this were so, we should expect the normal Lamarckiana 

 to give, as it does, brittle specimens in both cases, but the production 

 of tough mutants and tough hybrids to be a character of our new 

 type similis. In order to try this it was, of course, only necessary 

 to cross this new race with 0. rubrinervis. 



I made the cross in 1917, using my typical strain of 0. rubrinervis 

 as a pollen parent. Next year I had a culture of 180 plants, about 

 one-half of which have flowered, since 1 1 1 specimens of the type of 

 Lamarckiana had to be removed before the flowering period in order 

 to save space. They constituted three main types: Lamarckiana 

 and lucida, which were to be expected, and subrobusta. No brittle 

 specimen was found. Two mutant individuals occurred, belonging 

 to the types albida and oblonga. From this we may conclude that 

 the new mutant strain similis, which produces erythrina instead of 

 rubrinervis as mutants, also gives subrobusta instead of brittle 

 hybrids in the corresponding crosses. 



The seeds of 0. similis contain the same amount of barren grains 

 as those of 0. Lamarckiana. In the harvest of the third generation 

 of 1917, I counted 39, 39, 50, and 53 per cent, giving an average of 

 45 per cent; and in that of the fourth generation in 1921 I counted 

 60 per cent of empty seeds. This leads to the conception that similis 

 has essentially the same constitution as its parent, and consists of 

 laeta and velutina gametes, but without the factor for brittleness. 

 We should therefore expect the result of the cross to comply with 

 the following interpretation: similis (= laeta -\- velutina) x rubri- 

 nervis (= deserens -f- velutina), giving laeta x deserens (^ lucida) + 

 laeta x velutina (= Lamarckiana) -f velutina x velutina (= barren 

 grains) -f velutina x deserens (= subrobusta). Such was exactly the 



