Sorting and blending of unit characters. 229 



Davis (1910) has also shown that crosses of certain wild species 

 (grandiflora X american "■hiennis") give rise to twin hybrids in the first 

 generation, one of the twins resembling one parent, while the other 

 resembles more the other parent of the cross. But "the characters of 

 the parents, as presented in each cross, were so blended that as regards 

 measurements of parts, habit, texture, foliage, etc., the average for each 

 set of hybrids would probably present a fair mean between the two pa- 

 rents concerned". "No character of either parent was discovered which 

 appeared as dominant in these hybrids of the Fi generation"^). 



De Vries interprets this segregation in the first generation as 

 due to the mutating character of one of the parents {LamarcJciana or 

 certain of its mutants) with its consequent splitability of constitution, 

 combined with the splitting power of the constitution of another parent. 

 This proceeds from his hypothesis (112, 1913) that the segregations in 

 the first generation rest on the presence of labil pangens (in the muta- 

 ting species) and are brought forth through the meeting of these with 

 antagonistic pangens in the inactive state (in the other parent). 



In the case of the two species considered here (pycnocarpa and 

 nutans) it does not appear probable that, in the first generation from 

 crosses, the segregation into twin and triplet hybrids can be explained 

 on this interesting hypothesis formulated by de Vries for mutation 

 crosses. We have no evidence that either of the parents concerned is 

 in a mutating condition. For two generations they have been grown 

 in quite large numbers, have proven constant and given no evidences 

 of the production of new forms by mutation. The constitution of the 

 hybrids is quite different from that of the twin hj^brids described by 

 de Vries from mutation crosses. If we consider certain leaf characters 

 only, the two segregate hybrids resemble the laeta and velutina twin 

 hybrids of de Vries, since pycnella has broad leaves, while iortiiosa 

 has narrow and paler green leaves. But these characters along with 

 others are taken without modification from the parents. If pycnella 

 resembles nutans in the width of the leaves, it resembles pycnocarpa 

 in the furrowedness, repandness and somewhat darker color, it also 

 resembles pycnocarpa in the color of the stem, and nutans in the toothed 

 character of the (rosette) leaves, and in habit. If tortuosa resembles 

 pycnocarpa in the narrowness of the leaves, it resembles nutans in the 

 convexity and crinkledness of the (rosette) leaves, in the color of the 



^) The type of the american "biennis" is not given. 



