418 BOTANICAL GAZETTE [MAY 
the laeta. Until now I have tried only one instance, O. grandiflora X 
O. Cockerelli and its reciprocal. They produced only one of the 
twins, namely, /aefa, and thus confirm our view. Other combina- 
tions should be studied for the same purpose. 
If Mendel’s law were applicable to the production of the twins, 
these must split after self-fertilization into three or more types. 
Our table shows that this is not, or at least not always, the case. 
The velutina never split, nor do the /aeta of O. biennis, O. syrticola, 
and O. biennis Chicago produce a splitting. Only those of O. Cock- 
erelli show this phenomenon, but here it is limited to the repetition 
of the mass mutation into velutina. From these facts we must con- 
clude that the hereditary factors involved are not in the condition 
required by Mendel’s laws. In Gruppenweise Artbildung I have 
called this deviating condition labile, leaving the question open 
whether it may be determined by means of linkage or otherwise. 
The cross of the pollen of O. grandiflora with the female gamete 
of O. biennis Chicago produces twins which are quite different from 
the laeta and velutina, and are therefore called densa and /axa. 
Formerly I assumed this difference to be due to the splitting of 
another factor, but my results with O. grandiflora, in connection 
with the appreciation of the complicated nature of so many muta- 
tions (7), open the prospect of considering it as due to the same 
unit, only under the influence of different linkage. In fact, if we 
assume the pollen of O. grandiflora to be dimorphous before fecun- 
dation, two types of hybrids must be expected in this cross as well 
as in others. This conception simplifies the problem, although it 
does not offer a direct proof against the presence of a special splitting 
factor for densa and laxa. These twins are constant in their pro- 
geny, even as the Jaefa and velutina just considered, and thereby 
indicate the same special condition of the factors involved. When- 
ever the differentiating characters of the twins are recessive to 
those of the other parent, the twins must show the same external 
marks. The hybrids will then be uniform instead of dimorphic, 
as, for example, in our experiment with O. grandiflora XO. syrticola 
(muricata). 
The cross O. grandiflora XO. biennis gave a dimorphic progeny, 
which may evidently be ascribed to the presence of mutated and 
