544 HARRIS: OVULES AND SEEDS IN CERCIS 
B. The Individuals from the Vicinity of Lawrence, Kansas 
The relationship for the whole material has already been 
discussed. For. the individual trees the results as shown in 
Table IX are most interesting. Of the 22 values of r.:, only 1 has 
TABLE IX 
CORRELATIONS FOR INDIVIDUALS 
Tree Yo, and Ly, Wo and Ey, 
I O71 + .037 — .134 + .066 
2 692 + .035 + .099 + .099 
3 -595 + .044 — .052 + .067 
4 604 + .043 — .204 065 
S 580 + .045 — .172 + .065 
6 549 + .047 — .131 + .066 
7 400 + .057 — .182 + .065 
8 380 + .058 — .189 + .065 
9 365 + .058 — .34I + .060 
to 365 + .058 — .319 + .o61 
Ir 554 + .047 — 27H te .065 
12 394 = .057 — .189 + .065 
13 609 + .042 — .030 + .067 
14 302.057 — .263 + .063 
15 381 + .058 — .274 + .062 
16 543 + .048 — .117 + .067 
17 537 + .048 — .253 + .063 
18 181 + .065 — .322 + .060 
19 572 + .045 — .022 + .067 
20 , 644 + .039 — .212 + .064 
21 .514 + .050 — .085 + .067 
22 ‘ -635 + .040 — .288 + .062 
the positive sign. Taking the ratios of the constants to their 
probable errors to test their significance I find that the single 
positive constant deviates from o by an amount equal to its 
probable error; hence no significance can be attached to it. Of 
the 21 negative coefficients 14 differ from o by more than 2.5 
times their probable error. The whole situation is summed up 
graphically in diagram 3 in which these ratios are plotted out 
with the signs of the correlations. If there were no real biological 
correlation between the number of ovules per pod and their ca- 
pacity for maturing their seeds, the distribution of these ratios 
would be centered at 0, marked by the “theoretical mode and 
mean”’ line, with the deviations about equally distributed above 
