J. Arthur Harris 
207 
of the first, although even here (see Diagrams I and II, lower lines. III and IV, 
less steep lines) the deviation from linearity is not great*. 
Returning now to the constants and considering them in detail, we note that for 
all localities the direct homotypic correlations are highest for the number of ovules 
per pod and lowest for the number of ovules failing f. The averages J are 
Ovules and Ovules -3430 
Seeds and Seeds -2164: 
Ovules failing and Ovules faihng -1198 
The cross relationships are highest for ovules formed and seeds developing per 
pod; much lower for ovules formed and ovules faihng to develop; lowest for 
seeds developing and seeds failing to develop — three of the four values for r^^f^ 
are negative in sign. 
The averages are: 
Ovules and Seeds ... + -2168 
Ovules and Ovules failing + -0712 
Seeds and Ovules failing — -0440 
From these data it appears that the individuahty of the trees is more marked 
for number of ovules per pod than for either number of seeds matured or number 
of ovules failing per pod. 
Number of ovules per pod may be regarded in comparison with number of 
seeds developing and number of ovules failing to develop as an independent 
variable§. 
The number of seeds per pod is absolutely limited by (i) the number of ovules 
formed in the pod, and (ii) the number of these which are prevented from developing 
by peculiarities inherent in ovum or sperm, by accidents of fertilization, inadequacy 
of food supply, or other unknown causes. Or conversely, the number of ovules 
failing per pod is dependent upon the number of ovules formed and the number 
of seeds developing. Thus the number of ovules per pod determines within certain 
limits the number of seeds developing, or the number of ovules failing, per pod ; 
but in the nature of things it cannot be influenced by either of these. 
Now since the three characters under consideration are correlated, i.e. since 
r„s, V) ^s/ have been found to have sensible values, it is clear that some homotypic 
relationship would arise for seeds because of the correlation for ovules, or for 
ovules failing because of the correlation for ovules and seeds. Tlius a statistically 
significant value of an intra-individual correlation for seeds per pod does not 
* In the case of the relationship for ovules and ovules in the 60 trees from Meramec Highlands 
I have found the raw value of t) to be -SSSSG, exceeding r by only •00064. 
f In the Ohio series r^^,,^ is very low; r,^,^ and rj-^ f^ are not sensibly different. 
X The constants for the 112 trees from Meramee Highlands are alone used in obtaining the general 
averages. 
§ This does not mean that in any given pod all three of these characters are not dependent upon 
the same ultimate causes, but merely that in a proximate sense number of ovules per pod is physio- 
logically independent while number of seeds and number of ovules failing are physiologically dependent. 
