494 Selective UUrnination in Staphijlea 
tion of ovaries and serves for the determination. The coeflBcient calculated from 
the surface given as Table XLIV is 
r = 0221 ± -0078. 
Possibly the coefficient is statistically significant, being about thrice its probable 
error, but it is clearly of so low an order that it can have no appreciable biological 
significance in the problem now in hand. 
Apparently, therefore, both number of ovules and radial asymmetry in the 
distribution of the ovules in the three locules are independent and physio- 
logically significant factors in determining whether an ovary shall or shall not 
reach maturity. 
3. The Correlation between the Number of " odd" Locules per Ovary 
and the Number of Ovides -per Locule. 
Again, series B may serve as material. The correlation surface is shown as 
Table XLV. The constant is 
r = --0368 ±-0078. 
With a correlation so low as this it seems reasonable to think that, in their rela- 
tion to selective elimination, these two characters are practically independent. 
VII. Recapitulation. 
1. The foregoing pages contain an account of a quantitative study of the selec- 
tive elimination occurring in the ovaries of Staphylea trifolia between the time of 
flowering and the matui'ing of the fruit. The scope of the paper is limited strictly 
to the presentation of observed data and their description by means of biometric 
constants. No biological theories to explain why one ovary should be less fit than 
another are suggested, and no hypotheses concerning the influence of the selective 
elimination demonstrated upon evolutionary progress are brought forward ; there 
will be ample time for this after the much more extensive data already collected are 
published. A comparison of these results for the selective elimination of organs 
with those secured by others for a selective elimination of individuals may also be 
profitably postponed. 
2. The ovaries with relatively low numbers of ovules are more extensively 
eliminated than those with high numbers. The mean of the population re- 
maining after selection is about 7 or 8 per cent, higher than that of the 
eliminated individuals. 
3. The variability of both the eliminated ovaries and the series remaining 
after elimination is less than that of the original population. This condition 
