284 A. FRANKLIN SHULL 
derived from their union. These various lines reacted to the 
same environment by yielding different proportions of male- 
producers. Such differences could only be explained by assum- 
ing that each line possessed an internal nature different from 
that of the others. 
The nature of the internal agent or agents was not discussed in 
connection with the earlier experiments, since what.I hoped 
might prove a crucial test of their nature, namely, the results of 
inbreeding, was not available because the fertilized eggs obtained 
by inbreeding did not hatch. I have now obtained offspring by 
inbreeding, and the evidence from this source, together with that 
from the former crosses, serves at least to eliminate some possible 
conceptions of the internal agents. The inbreeding experiments 
are described and their bearing discussed in the following pages. 
The question whether the internal factor, whatever its nature, 
is constant, or whether it may undergo progressive change, is also 
taken up. And finally, experiments which seem to determine 
the time at which external agents may affect the cycle are 
described. 
OBSERVATIONS AND EXPERIMENTS 
Decrease in the proportion of male-producers with long-continued 
parthenogenesis 
From observations incidentally made on several lines of roti- 
fers bred through a considerable number of generations, I gained 
the impression that there was a progressive decrease in the pro- 
portion of male-producers with long-continued parthenogenesis. 
Such a condition, if proven to exist, besides being of considerable 
theoretical interest, would be of importance in judging the results 
of breeding experiments in which two distinct or unrelated lines 
are compared with one another. Inasmuch as a number of my 
experiments involved such a comparison, I have compiled data 
from a number of lines which were reared through a number of 
generations large enough to give results of value. There are 
eight of these lines, each including 46 or more generations, which 
are given in table 1. 
