No. 532] GENETIC AL STUDIES ON (EN OTHER A 



197 



by capsule, which is the safest method to obtain quantita- 

 tive results. In the past I have sown from bulk collec- 

 tions of seed and my studies so far must be regarded 

 as primarily qualitative in character. This practise was 

 followed under the impression that probably only a small 

 proportion of the seed from a cross, apparently as ex- 

 treme as that between biennis and (jranriifiora, would 

 germinate. The results, however, have shown that the seed 

 of this cross is fertile to a very high degree. In conse- 

 quence my cultures of this season gave three or four 

 times more seedlings than it was practicable to bring to 

 maturity. A process of selection became at once neces- 

 sary, which was followed with the end in view of obtain- 

 ing a variety of types, but it soon became evident, with 

 the later development of the cultures, that it is impos- 

 sible to select with accuracy among the young seedlings. 

 Thus plants which showed certain tendencies as young 

 rosettes or seedlings of five to nine leaves often changed 

 very materially in later growth. For these reasons my 

 cultures of this year even as qualitative studies are un- 

 doubtedly not fully representative. 



(Enothera biennis L. 

 My first crosses between biennis and grandiflora were 

 made at Woods Hole, Mass., in 1908. This locality was 

 carefully searched for rosettes of biennis with broad 

 leaves and two plants of this character were transplanted 

 from waste ground to the garden. The rosettes were 

 similar and at maturity the plants proved to be the same 



of biennis? als\^tea A^n*^^ 



vated through two and three generations, respectively, 



It seems quite certain that under the name (Enothera 

 biennis is included a number of races with well-marked 

 peculiarities. These races are probably very pure, for 

 the reason that close pollination is certainly usual, if 



