536 Harris: OVULES AND SEEDS IN CERCIS 
conspicuously, ranging from 13.12 to 22.12 for ovules and from 
20.37 to 34.87 for seeds. Unless the values of means and standard 
deviations are perfectly correlated—an assumption which we 
have no reason to make—one would expect the coefficients of 
variation to show considerable fluctuation in magnitude from 
the influence of the means alone. 
Since the constants showing the mean values and the vari- 
abilities of ovules and seeds per pod and perhaps those for the 
correlation of these two characters as well (see Table III) differ 
significantly from individual to individual it is clear that there 
may be serious disadvantages in lumping together the materials 
from different trees to form a large sample to be used in the 
investigation of delicate correlations. 
Turn now to the correlation coefficients. Table IV gives the 
results. Here as in the two random samples and their combi- 
nation there is a considerable correlation between the number of 
ovules per pod and the actual number of seeds developing. 
Trees I, 5, 7, 8, and 9 have been selected for the test of line- 
arity of regression because of the large number of pods available. 
The equations to the regression straight lines are given in the 
Equation to Regression 
Tree Straight Line Deviation* 
S$ = .284 + .705 0 
5 S$ = .653 + .6570 042 
i S = .652 + .658 0 7 .050 
8 S =.194 + .781 0 .030 
9 S = .100 +.7540 .048 
accompanying table. The average weighted deviation of obser- 
vation from theory of each tree is also given. Graphs have been 
made showing the empirical means and the fitted straight line 
for each tree. Only that for tree 9 need be given, as figure I. 
Perhaps the fit is slightly better than for some of the others, but 
the observations are numerous. All of the graphs show unusually 
good agreements of theoretical and empirical means. The 
theoretical and the empirical means differ on an average by only 
about five nuivedtte of a seed. 
* Weighted mean deviation, smiatdivw signs, of observed from theoretical 
mean number of seeds per pod. 
