'915.] PURE SPECIES OF CENOTHERA. 7 



these plants frequently give extraordinarily low yields of fertile 

 seeds although seed-like structures may be formed in abundance. 

 These facts we are just beginning to appreciate as offering prob- 

 lems for study. They seem to the writer of vital importance to 

 the discussion of CEnothera genetics, facts which the Mutationists 

 cannot ignore and behind which the Mendelians can maintain at 

 present a very strong defence for their interpretations of the peculi- 

 arities of CEnothera behavior. 



With respect to pollen sterility it has for many years been 

 known that Lamarckiana and other species of CEnothera present 

 large proportions of abortive pollen grains. Bateson (1902) early 

 seized on the point and suggested that the high degree of pollen 

 abortion in Lamarckiana indicated a hybrid plant exhibiting partial 

 sterility. Geerts ('09) in an excellent account of the cytology of 

 Lamarckina showed that approximately one half of the pollen grains 

 fail to mature and that one half of the ovules fail to develop em- 

 bryo sacs. Geerts ('09, p. 89) also made an examination of more 

 than one hundred species of the Onagracese, giving us the condi- 

 tions of pollen and ovulue fertility represented in some fifteen 

 genera. He found generally in species of CEnothera and allied 

 genera a degree of sterility similar to that in CEnothera Lamarcki- 

 ana, about 50 per cent, for both pollen and ovules. On the other 

 hand certain species of Jussieua, Zanschneria, Epilobium, Boisdu- 

 valia and Lopezia are wholly or almost wholly fertile. 



My own examination of conditions in the material of CEnothera 

 with which in recent years I have worked has shown some remark- 

 able differences in the amount of pollen and seed sterility. Such 

 close pollinated types as the Dutch biennis, the Dutch muricata, 

 American muricata (from Woods Hole), Tracyi, and a number of 

 American small-flowered species ( for example biennis A and biennis 

 D of my cultures (Davis, '11, p. 197 and '12, p. 385)), have very 

 large amounts of sterile pollen. In the case of the Dutch muricata 

 much more than 50 per cent, of the pollen has been sterile. Yet 

 these are types which by virtue of their long history of close polli- 

 nation might be expected to be among the purest of the species. 

 On the other hand the race grandiflora B (Davis, '11, p. 203), and 

 the western species franciscana and venusta, all open pollinated 



