414 ‘ BOTANICAL GAZETTE [NOVEMBER 
flowers at first, and ovulate flowers almost exclusively later in the. 
season. He found numerous instances of degenerating embryo 
sacs in R. Acetosa. I have examined a great many sections of 
R. Acetosella, and found no instance of such behavior. It would 
seem that degenerations in the ovary are much less frequent in the 
dioecious species than in the so-called hermaphrodite, and that 
when the ovulate plants occasionally produce stamens, all the 
pollen is functionless. 
The conclusion is that Rumex formerly produced only hermaph- 
rodite flowers, and that by degenerations in the stamens and in the 
carpels the condition has been attained such as is found in R. 
crispus, and the species of section Lapatuum in general, where the 
inflorescence contains a mixture of physiologically staminate, 
physiologically ovulate, a few bisporangiate, and many com- 
pletely sterile flowers. This is physiological dicliny. The process 
has been carried farther in some forms, resulting in segregation of 
the staminate and ovulate flowers to separate plants, as is now 
the case in the species in section AceTosA. That these latter 
forms have been derived from bisporangiate or monoecious ancestors 
is indicated by the occasional production of stamens on “ovulate” 
plants. That this derivation has been due to degenerations is 
indicated by the sterility of the staminate structures when they 
are formed. 
It seems clear that the stamens are more readily eliminated 
from the flowers than are the ovules. They start degenerating 
earlier in their development, and it is very common for all trace of 
them to be lost by the time the flower opens, while the ovules 
invariably persist in physiologically staminate flowers, and very 
frequently are defective only in the embryo sac. 
It is probable that the degeneration processes favor the occur- 
rence of apogamy. Not only does degeneration result in elimina- 
tion of stamens from many flowers, but it results also in sterility of 
the pollen that is produced by normal anthers. Dioecism would 
render pollination by the small amount of pollen remaining normal 
a very uncertain process. Finally, it is altogether possible that 
when the degeneration process in the ovary is but weakly mani- 
fested, it may interfere with reduction in the megaspore mother 
