J. A. Harris 
469 
mean values of the number of ovules in the matured fruits might not differ sensibly 
from that for the original ovaries, but the variability in number would be reduced. 
On the other hand it is not impossible that a selective elimination might effect 
chiefly the modal region of the curve of distribution, materially flattening it off 
and so increasing the variability as measured by the standard deviation. To the 
biologist this seems theoretically less probable than the elimination of either the 
largest or the smallest ovaries, or the extreme variates of both types, but of course 
all possibilities must be tested out on the basis of actual data. 
Ovules per Ovary. 
DiAGKAii 3. Percentile curve showing frequency of different numbers of 
ovules per ovary in the eliminated series and the series developing 
to maturity. Dots and broken line = Eliminated series; dots and 
firm line = Matured series. 
In comparing the means of the eliminated and uneliminated series, it was only 
necessary to determine the values for either ovules per locule or ovules per fruit, 
since one is necessarily three times the other. For the variabilities, however, it 
seems best to examine the standard deviation for both ovules per locule and total 
ovules per ovary, since one cannot readily be got from the other. 
We are dealing with integral variates, and hence Sheppard's modification for 
the second moment was not applied. 
Biometrika vii 60 
