278 TWIN HYBRIDS OF OENOTHERA HOOKERI T. AND G. 



induced me to control the absence of constant hybrids of the type 

 of rubiennis once more. With this view I have sown in 1915 some 

 pure seeds collected in 1909, partly on biennial plants of the first 

 generation, mentioned in my book for 1908, and partly on some 

 annual specimens of the second generation. Every plant was fecun- 

 dated purely by myself with its own pollen and their seeds were 

 sown in pans and transplanted into boxes in 1916. Every box con- 

 tained 60 seedlings. In April the difference between the rubiennis 

 and the Hooker /-like was evident but some doubtful specimens 

 remained. There was a large majority of the narrow-leaved form 

 and some few specimens of rubiennis. These, however, were seen in 

 every single box, and therefore among the progeny of each of the 

 parents of 1915. No uniform set was found. I cultivated the boxes 

 until the middle of June, when the rosettes became very stout with 

 about 15—20 leaves; the full-grown leaves reached 15 cm. Those of 

 the Hookeri-like were narrow with pointed tips. All in all I had self- 

 fertilized 22 plants, 1 1 of the second and 1 1 of the first generation. 

 The percentages varied between 2 and 12 and were on the average 

 6 for the first and 4 for the second group. These aie rather very 

 low figures, even if they are compared with those of previous years 

 (10—22 percent, see Gruppenweise Artbildung 1913, p. 103) but the 

 summer of 1915 had been in many respects unfavorable for my cul- 

 tures. The main point, however, was not weakened by this circum- 

 stance, since the question to be answered was, whether all specimens 

 of rubiennis would produce Hookeri-Wke hybrids among their progeny, 

 or whether there wouid be some without this splitting. The latter 

 was evidently not the case. 



We thus find the conclusion confirmed that in this case the splitting 

 of the hybrids is unilateral. The explanation given in my book is 

 thereby strengthened. It reposed on the heterogamy of Oe. biennis. 

 The hereditary qualities of the male gametes of this species, or at 

 least the majority of them, cannot be transmitted to the female 

 sexual organs, neither in self-fertilization nor in crosses. Thus the 

 hybrid Oe. HookerixOe. biennis carries in its egg cells only the 

 characteristics of the pistil parent, but in its pollen those of both 

 parents. If we assume that the latter are separated in synapsis into 

 two groups and equally distributed over the pollen grains, we have 

 one-half of the grains with the qualities of Oe. Hookeri and one-half 

 with those of Oe. biennis. The first must evidently produce, in fecun 

 dation, plants of the Hookeri-Wke type, the second renew the cross 

 and give hybrids with the qualities of rubiennis. It is easily seen, 



