442 Stout: The Origin of Dwarf Plants 



two types of "Cupid". Bateson attempts to explain this case as he 

 does all cases of increased variation in terms of presence or absence 

 of factors, giving different interactions in different combinations. 

 The point is clear that greater variability does develop and that not 

 only is height of sweet peas evidently a compound character, but 

 that the assumed factors take on different values. For example, 

 the factor T ("tallness") when combined with P (erect: branch- 

 ing) gives the bush variety. The two sets of allelomorphs assumed 

 in this case are not independent units in expression. It may be 

 emphasized that the original rigid unit-factor hypothesis is being 

 given up in favor of the admission, as we see in this case, of a marked 

 modifiability of the effect of the assumed factors. The integrity 

 of the assumed units in these cases can only be maintained by 

 assuming a permanence in them, which is entirely at variance with 

 the visible results w^hich they are assumed to produce. 



The behavior of the dwarf characteristics of Oenothera nanella 

 in hybridization is of special interest in this connection. The 

 group of characters appearing in Oe. nanella go together in crosses 

 with Oe. Lamarckiana, not as a unit giving alternative expression 

 but in the splitting that develops in the Fi generation giving the 

 parent forms as twin hybrids. De Vries assumes (1913) that the 

 alta pangen is inactive in Oe. nanella and labile in Oe. Lamarckiana. 

 In this case inactive X labile gives both inactive and labile. 



When Oe. ruhrinervis is crossed with Oe. nanella splitting also 

 occurs in the Fi giving Oe. Lamarckiana and a new type called 

 Oe. subrobusta. The alta pangen in Oe. rubrinervis is assumed to 

 be active, so in this case active X inactive = labile and inactive + 

 active. The Oe. subrobusta form splits in the next generation, 

 giving dwarf forms which, however, are evidently not Oe. nanella. 

 None of the types derived are like either of the two parent forms. 



Again when Oe. muricata is crossed with Oe. nanella the Fi is of 

 two new types called Oe. laeta and Oe. velutina. Oe. velutina splits 

 up in the F2 giving a dwarf Oe. murinella that is not like Oe. nanella. 

 None of the types are like the parent forms. The assumed condi- 

 tion of the Oe. alta pangen in this case is active X inactive = labile, 

 active and inactive. 



The appearance of twin hybrids not like the parents and of sec- 



