240 DAVIS— THE TEST OF A [April 23, 



iiig true; flies homozygous for confluent wings are apparently not 

 viable. Is it not possible that parallel or related phenomena are 

 extensively present among the Oenotheras ? The mortality as shown 

 by sterile seeds may indicate the elimination of large groups of 

 forms divergent from the parent types, and some of the curious 

 dwarfs and aberrant plants which again and again have been re- 

 ported in CEnothera lines may be from zygotes barely able to sur- 

 vive the death-producing conditions that eliminate so many of their 

 companions. 



So' far we have considered evidence chiefly of a negative charac- 

 ter for the contention that many of the species of CEnothera are 

 impure or hybrid species. We have tried to show that pollen, 

 ovule, and seed sterility must all be reckoned with as conditions 

 which may eliminate Mendelian classes of gametes and hold a line 

 to a history of relatively true breeding even though the stream of 

 germ plasm remain heterozygous or impure in character. The nat- 

 ural corollary of such behavior, if proven, might be the interpreta- 

 tion of so-called " mutants " as segregates from a hybrid stock that 

 were able to survive the destruction meted out by conditions that 

 produce sterility. To what extent the causes of sterility may lie 

 in the history of gametogenesis or may be due to unfortunate com- 

 binations of gametes, or to what extent sterility is the result of 

 physiological factors, these are problems that lie before us. 



Let us now examine some positive evidence that certain species 

 of Oenothera do form distinct classes of gametes and in consequence 

 seem likely to be heterozygous in their constitution. That which 

 first demands attention is the situation discovered by De Vries in 

 certain first generation hybrids and by him named " twin hybrids." 

 We have already referred to this phenomenon first described by De 

 Vries ('07) for the behavior of Lamarckiana which as a pollen 

 parent in crosses with other species of CEnothera gives not uniform 

 F-L generations but the two types Iceta and velutina (twin hybrids),, 

 produced in about equal numbers. Certain " mutants " of La- 

 marckiana also give twin hybrids under the same conditions as 

 those produced by Lamarckiana. The behavior is so exact that the 

 simplest hypothesis must suppose that Lamarckiana and these "mu- 

 tants " form two classes of gametes which are fertile in these par- 



