[Vor. 11 
446 ANNALS OF THE MISSOURI BOTANICAL GARDEN 
in the number of seeds per fruit as the number of fruits per in- 
florescence increases. 
We may therefore conclude that the correlation coefficient fur- 
nishes as adequate & measure of the relationship between the 
number of flowers per inflorescence and seeds per fruit, and 
fruits per inflorescence and seeds per fruit, as we are able to 
command. This measure shows that the relationship is, prac- 
tically speaking, zero for the number of flowers and the number of 
seeds, but has a significant negative value for number of fruits 
matured per inflorescence and number of seeds ripened per locule. 
It will be quite evident on mature consideration that the very 
low correlation between number of flowers per inflorescence and 
number of seeds may be the resultant of physiological relation- 
ships between number of capsules and number of seeds on the 
one hand, and the number of capsules and number of flowers, on 
the other. Inflorescences producing larger numbers of flowers 
also produce larger numbers of fruits. As a resultant of the 
positive correlation between number of flowers and capsules per 
inflorescence, ту, and of the negative correlation between number 
of fruits and number of seeds per locule, 7, one might expect 
the correlation between number of flowers and number of seeds 
to be sensibly lowered.! 
It is necessary, therefore, to determine the partial correlation 
between number of flowers per inflorescence and number of 8 
for constant number of capsules per inflorescence. This is given 
by 
“л = Тү — Tfcl'es 
У1- revi = fe 
The values of „ту, are set forth in table xix. АП 4 values 
are now positive. Furthermore they are (as shown by the dif- 
ference column, „ту, — rys) larger than the uncorrected corre- 
lations. 
Thus they indicate that if correction be made for the influence 
of the number of fruits whch mature, the inflorescences which are | 
1 In inflorescences with larger numbers of fruits the plastic materials must be more 
extensively divided than in those with а small number of fruits. Number of fruits 
per inflorescence might, therefore, seem a more logical proximate cause of variation 
in the number of seeds matured than would number of flowers per inflorescence. , 
