46 BOTANICAL GAZETTE [JANUARY 
The re-formation of a hybrid by the cross-pollination of the 
parents to which it may be ascribed is by no means simple in 
all instances, nor is it always easy of accomplishment. In the first 
place, the original cross-pollination may have taken place possibly 
under an exceedingly rare combination of favorable physiological 
conditions difficult to secure or duplicate in experimentation. Then 
one or both of the species as ordinarily recognized may in reality 
consist of two or more elementary species, which may not differ 
widely in external anatomical characteristics, but exhibit widely diver- 
gent physiological characters and behave quite differently in breeding. 
A hybrid with one of these forms may differ in very many important 
‘particulars from a hybrid with another constituent of the same species. 
Of course if we deal with elementary species only in our usage of the 
term, this difficulty does not exist; but it does appear as a serious 
matter with the customary practice, as has been found in a number 
of breeding experiments, and furthermore must be taken into account 
no matter by what method we consider a hybrid. 
Let us suppose, however, that we have actually in hand the two 
strains or elementary species by which the hybrid may have arisen, 
and we have still one more matter which may mislead us. This con- 
sists in the fact that reciprocal crosses are not always identical in 
their products. Thus the pollen of A and the egg of B do not neces- 
sarily make the same hybrid as the egg of A and the pollen of B. 
The pollination of Oenothera Lamarckiana by O. biennis gener- 
ally results in securing a progeny separable into seven types, some 
of which are stable and reproduce themselves exactly in succeeding 
generations, while others split into two or more forms in the second 
generation. On the other hand, the use of pollen of O. Lamarckiana 
on pistils of O. biennis results in a progeny embracing four types, 
none of which is identical with the components of the reciprocal cross. 
O. biennis shows a similar behavior in some other crosses in the 
limited observations recorded. In all such cases it must be under- 
stood that the number of types does not appear to be invariable, and 
that a progeny of a hundred thousand is likely to include more than 
one of a score. It is evident that in the determination of a hybrid 
by this method difficulties may be met with. Thus the failure of 
the operator to secure the supposed hybrid may not be considered 
| 
| 
| 
