160 CANNON: STUDIES IN PLANT HYBRIDS 
the reversion to one or to another of the pure parents might not 
under certain conditions take place. It thus is necessary in order 
to effect the splitting after the Prswm-type that on the average, 
taking many plants together, an equal number of each sort of 
germ-cells of the same descent shall degenerate, and that a like 
number shall become functional. The variability of the higher 
plants as regards the development of the macrospores may thus 
in a broad way account for certain conceivable modifications or 
apparent exceptions to the Mendelian law, so that the morphology 
of the hybrids must needs be studied if the structural basis for 
hybrid variation is to be understood. 
We have then sketched what supposedly ee place when 
mosses, ferns and also when seed-plants of the /yswm-type are 
hybridized, and the general and particular conclusions may be 
summed up thus: 
1. In monohybrids of the Piswm-type the spores and the 1m- 
mediate spore derivatives are pure as regards the chromatin content 
of their nuclei. The application of this principle to the Zverworts 
accounts for the apparent lack of hybrids among them, either in 
the first or in the later generations ; applied to the mosses it ac- 
counts for the failure to recognize hybrids in the second and later 
generations. In both liverworts and mosses the “ plant” (gameto- 
phyte) is of pure descent in any generation, and the sporogonia 
only are hybrid. This last consideration leads to the generaliza- 
tion that in plants of whatever class where hybridity occurs tt ts per- 
haps always the sporophyte only which ts hybrid.* 
2. In the ferns the chromatin may be transmitted from one 
sexual generation to another in such a manner as to retain its indi- 
viduality and thus it may tend to reproduce the sort of gameto- 
phyte from which it descended, and in dioecious forms the sex 
also, so that the purity of the spores forms the background that may 
make the genesis and the evolution of heterospory a morphological 
possibility. 
It is likely that the study of such hybrids as have distinct 
sexual and asexual generations would be of advantage since in 
them a morphological analysis of the nature of a hybrid might be 
made, and thus a clearer understanding of the nature of hybridity 
* The possible exceptions to this have already been referred to (page 15 5). 
