Oenothera Nanclla. 373 



they arise from another new species, exhibit in the first 

 generation not only the same characters but are as true 

 to seed, as those which arise directly from Lamarckiana. 



For testing the constancy of this form under self- 

 fertilization in subsequent generations I used the second 

 of the above mentioned experiments as a starting-point 

 (p. 371 ). Some of the 2463 plants mentioned there were 

 chosen as seed parents and self-fertilized. The seeds 

 gathered from 4 of them were sown in 1897 ; they gave 

 respectively 94, 135, 154 and 164 seedlings — 547 in all — 

 which proved without exception to be dwarfs when they 

 were recorded as large rosettes m July. I allowed about 

 100 of these to flower and fertilized some of them with 

 their own pollen. In 1898 I raised from the seed thus 

 produced the fourth nanella generation which again was 

 perfectly constant, and allowed about 100 specimens to 

 flower. The fifth and sixth generations (1899 and 1900) 

 also came perfectly true to seed. Of the total number 

 of seedlings — about 400 in 1900 — I allowed about 70 to 

 flower and used about 30 of these as seed parents. 



Thus from the third generation to the sixth, embra- 

 cing in all over a thousand plants, there occurred no single 

 instance of atavism. The new species must therefore be 

 regarded as perfectly constant. 



The constancy of nanella is however incomplete in 

 the sense that it has inherited the capacity of mutating, 

 from the parent species. For it gives rise to individuals 

 which though obviously nanellas also betray the charac- 

 ters of some of the other new species. 



And, conversely, it occasionally happens that dwarfs 

 arise from other new species and then bear the characters 

 of both types together. In this way we get species of the 

 second order, which correspond to the cultivated varie- 



