OWLEDGE OP THE MUTATI]S"G OENOTHEEAS. 57 



their parents as the latter are from each other, and the flowers remain intermediate. 



However, there is the difference that (0. riibrinervis X LamarcMana) X {Lamar chiama 

 X rubrinervis) presumably gives both 0. ruhrinervis and O. ZamarcManay while 

 (O. LamarcMana X biennis) x (biennis X LamarcMana) gives, if the result of DeVries 

 is invariable (1911, p. 100), only 0. LamarcMana. But, as ah'eady stated, there is some 

 reason to doubt the constancy of the latter result. 



I have already expressed my agreement with Honing that 0. LamarcMana and 

 O. rubrinervis differ from each other in the same way that the laefa and veluHna hybrids 

 differ — i. e., in width of leaves^ crinkling, hairiness, etc. This point has now been 

 adequately proved by the careful studies of Honing. But, I have further shown in the 

 present paper (p. 40) that O. muricata in the wild condition also frequently contains 



p 



two types which differ from each other in similar ways. The appearance of both (twin) 

 types in the Pi of many crosses is probably a result of this dimorphic condition, which 

 is therefore a fundamental one in this section of the genus. It remains for further 

 studies to disclose the nature of the relation between the two types as regards origin, 

 but it seems not impossible that they represent alternative conditions, either of which 

 may come into expression under certain circumstances. 



1 



Certain results of Miss Saunders with Stocks (1908) and Petunias (1910) also 

 indicate that races may breed true and yet be heterozygous, and it is probable that this 

 is the explanation of the behaviour in such crosses as 0. rubrinervis X 0. LamarcMana, 

 although that is not yet proven. 



Prom one point of view, especially regarding the origin of new characters, one of the 

 most important features of heredity in Oenothera is that both blended and sharply 

 alternative inheritance occur. The cases in Oenothera of blends which breed true in 

 specific crosses are now well established. My cultures of Lancashire Oenotheras have 

 also yielded several true-breeding races whose characters are evidently a blend of 

 previously existing races. Certain crosses between these and other forms give the 

 impression of an absolute blending of characters between races in the Pi. The behaviour 

 of later generations of these hybrids is not yet known. It remains to determine whether, 

 with repeated crossing back, blending continues to occur until the differences are 

 obliterated. The reason for intermediate or blended races breeding true is not yet 

 explained, but from the behaviour of certain of the laeta and velntina types (DeVries, 

 1908) it is possible that the failure of segregation may here also be due to a difference 

 between the characters borne by the male and female germ-cells. On the other hand, it 

 is not improbable that there is in such cases a real blend without segregation of the 

 materials representing the germinal characters. It seems probable that when real 

 blends occur, they will be found chiefly in characters which have a quantitative basis, 

 yet certain apparent cases of this kind, such as O, rubricalyx, show sharp segregation 

 phenomena. 



Through the maze of these complicated types of behaviour, it is beginning to appear 

 evident that there are certain connections between the systematic relationships of a 

 form and its hereditary behaviour, and crossing therefore is one of the most fundamental 



SECON^D SEEIES. — BOTANf, VOL. VHI. 



I 



