OENOTHERA LAMARCKIANA. 549 



we are only concerned with the number of germs produced by these 

 two forms in different crosses. 



The cross 0. deserensx Lamarckiana has given 18 per cent of 0. 

 lucida and 82 per cent of 0. subrobusta, the latter representing the 

 combination 0. deserensx velutina. This obviously gives a confir- 

 mation of the preferential fertilization, among the uniform ovules 

 of 0. deserens, by the pollen tubes of the type of 0. velutina. The 

 specimens of 0. lucida obtained in this experiment had 14—15 

 per cent empty seeds. Among their living progeny I counted in 

 1917 39—55 per cent of the parental type, and 45—61 per cent 0. 

 deserens. If we compare these figures with the hybrid formula, 

 which would lead us to expect 25 per cent barren 0. amphilaeta 

 seeds, 25 per cent pure 0. deserens, and 50 per cent of the hybrid 

 combination, giving two thirds 0. lucida and one-third 0. deserens 

 on the beds, we see that the 0. deserensx deserens have a marked 

 advantage over the 0. deserensx laeta, which may be ascribed to 

 the greater weakness of the 0. laeta gametes, female as well 

 as male. 



In the first place, I have studied the ovules of the hybrid 0. 

 lucida, fertilizing them by uniform pollen of different constitution. 

 In 0. albida the pollen is 0. velutina, with its usual lethal factor. 

 Combined with 0. lucida it gives hybrids of the types of 0. Lamarcki- 

 ana (= laeta x velutina) and 0. rubrinervis (= d eser ens' x velutina). 0. 

 elongata, which is a mutant from 0. simplex, has only 0. laeta gametes 

 among its active pollen grains. It must produce barren grains of 0. 

 amphilaeta besides the maternal type (= deserens' x laeta). 0. La- 

 marckiana mut. velutina=0. blandina has uniform pollen without 

 a lethal factor. It should give partly an 0. Lamarckiana-Yike. hybrid 

 (laetax blandina), as previously described (3), and partly 0. deserens' x 

 blandina or 0. subrobusta. Lastly, 0. deserens itself must reproduce 

 from 0. lucida partly the maternal and partly the paternal type. 

 In every case the expectation from the formula would be for equal 

 parts of both types. 



The crosses were made in 1921 and sixty offspring were cultivated 

 for each of them, comparing them during their whole lifetime (es- 

 pecially at the period of flowering and ripening their capsules) with 

 analogous hybrids of known extraction. The types on the beds 

 were those previously given, but the numerical results differed 

 widely from the expectation. As is seen from table I, the culture 

 on the bed of the second cross was a uniform one, whereas the 

 others sharply showed their two types. 



