1919] DeVRIES—OENOTHERA RUBRINERVIS 17 



described the pedigrees of two reciprocal crosses, both of which 

 produced as a second hybrid the suhrobusta. This was seen to 

 split off, after self-fertilization, brittle plants and dwarfs. In 191 5 

 I sowed some seeds of the suhrobusta plants of 1907 mentioned in 

 those tables, in order to compare their progeny with my newer 

 cultures. I found for two specimens of O. {nanellaXruhrinerois) 

 suhrobusta 38 and 45 per cent of dwarfs among 82 and 60 plants, 

 and for two parents O. {rubrinervisXnanella) suhrobusta 20 and 13 

 per cent of dwarfs among 60 and 46 individuals. The number of 

 brittle plants, however, was very small, being two specimens for 

 the first and one for the reciprocal group. It is possible that the 

 germs of this t>pe are weaker, and that some of them had died 

 during the 7 years of their preservation. I self-fertilized one 

 brittle specimen in each of the two main groups and had in 191 6 

 two lots of 45 and 60 flowering plants, all of which were brittle and 

 like their parents. They contained 9 and 8 per cent of dwarfs, the 

 stems of which were likewise brittle at the time of flowering.'" 



Other races of O. nanella or other conditions may produce in the 

 corresponding crosses brittle hybrids instead of suhrobusta. I made 

 the cross O. rubrinervisXnanella in 1905 with a dwarf mutant race 

 of 1895 and the reciprocal one with the progeny of a dwarf which 

 had arisen in 191 1 from Lamarckiana, using in both cases the same 

 family of rubrinervis as in all previous crosses. The first named 

 cross gave 35 per cent Lamarckiana, 3 per cent lucida, and 62 per 

 cent brittle rubrinervis among 68 specimens in 1913. The second 

 cross yielded the same three t^'pes, but the percentage figures devi- 

 ated widely. I had only 6 per cent Lamarckiana and 2 per cent 

 lucida, but 92 per cent brittle rubrinervis among 140 plants, most 

 of which flowered in August. The main result, however, is clear, 

 namely, that the crosses between O. rubrinervis and O. nanella give 

 three types of viable hybrids, one of which carries the visible marks 

 of O. rubrinervis, but may be either brittle or tough. 



I have made only one cross between O. deserens and a dwarf, 

 taking this latter from the first of the two last mentioned families. 



" By means of this the gap left in the second pedigree of p. 215 of my book is 

 filled up, and both pedigrees are completed by the production of dwarfs from the rubri- 

 nervis specimens. 



