538 Harris: OVULES AND SEEDS IN CERCIS 
correlation for the number of ovules formed and the number of 
seeds developing per pod, and (b) the coefficient of correlation 
between the number of ovules per pod and the deviation of the 
number of seeds per pod from the probable number on the assump- 
tion that the number of seeds per pod is proportional to the 
number of ovules. These constants are presented in Table V in 
a form which will be clear without further comment. 
The values for all the trees (large and small collections) are 
seriated in Table VI. 
At the outset of the work I had thought that perhaps by the 
collection of a sufficiently wide series of individuals I might find 
some in which the coefficient of correlation between the number 
of ovules formed and the number of seeds developing would be 
very low. The results for the 110 trees show, however, that 
not only are all the values positive but that in every case they 
are of a substantial order. The lowest value entered in Table VI 
is .325 and the highest .850. 
The constants of these correlation coefficients calculated with 
Sheppard’s correction, are: 
ean, .5994 = .0075 
Standard Deviation, -1165 + .0053 
Coefficient of Variation, 19.431 
It seems of considerable interest for our present problem to 
try to ascertain whether this observed variation in the magnitudes 
of the correlations between the number of ovules and seeds 
represents a real biological difference in the individual trees 
examined, or whether it is merely a statistical result, attributable 
to the fact that constants for each tree are based upon a small, 
not immensely large, sample of its pods. I believe we may 
make some progress as follows. at 
The standard deviation of random sampling of ris (1 — 7°)/ Vn, 
where is the number of individuals included in the sample on 
which 7 is calculated. Let us assume now that the true value of 7 
is the same for each individual tree and that it has the value 
found as our mean, say r = .600. This seems reasonable. 
If » for each individual were 100, a condition holding for 60 
of the trees but only with rough approximation for the remaining 
50, one would expect 
i a 
a eee eT 
