OENOTHERA GRANDIFLORA All. 219 



1914, and the culture of 1915 was a uniform set of 70 plants, most 

 of which were very weak and died off before flowering. Only 15 

 reached this stage. They grew up like the normal mutant ochracea 

 and had from the beginning its slender stems, broad and pale leaves. 

 No specimens of the type of biennis were seen. After self-fertilization 

 a splitting occurred. Some plants were green and stout; the majority, 

 however, were pale and weak. All of them had the broad leaves of 

 ochracea, but 9 among 80 made vigorous initial rosettes, whereas 

 the remainder grew up without this preparation. This gives a per- 

 centage of 89 ochracea and 1 1 biennis. The description for both 

 types is the same as in the cross between the two species. 



0. grand if lor a lorea x biennis. The narrow leaves of mut. 

 lorea are recessive to the broad form of the leaves of the species. 

 In other respects the mutant does not seem to differ from it, and 

 thus I could expect this cross to give almost the same results as the 

 first. I began the experiment in 1914, and in 1915 had the first 

 generation with 60 plants, most of which flowered. I counted 9 

 ochracea without rosettes, or 15 per cent; the others were of the 

 biennial type. Both types agreed in all respects with those of the 

 cross between the pure species. 



The second generation from the biennis plants was uniform, 

 repeating the type of the parent. I had 70 plants, half of which 

 flowered. They were very stout, and already so in the phase of 

 rosettes. When flowering, the bed looked almost like pure 0. biennis 

 L. No ochracea and no lorea were seen among them. 



The seeds of self-fertilized specimens of ochracea of the first gene- 

 ration produced in 1916, among 67 plants, a very striking splitting 

 into two types, 57 per cent being stout plants like biennis and 

 reaching 1.5 m. in height when they flowered. The others lacked 

 the initial rosettes but were not ochracea, evidently being lorea with 

 the narrow, dark green leaves of this type. They were far less stout 

 and reached only 1.20 m. in height and flowered some weeks later 

 than the biennis. The mut. lorea grows always without a preparatory 

 rosette and resembles in this respect the 0. grandiflora. From this 

 we may conclude that the splitting in our pedigree was exactly the 

 same as that between the two species, with the exception that 

 the lorea marks hid those of ochracea in the second generation. 

 I will now resume the results of the three crosses made with the 

 pollen of 0. biennis L. (Table IV). 



If we assume that in 0. grandiflora the mass mutation into och- 

 racea takes place at the time of synapsis, and that the egg cells 



