HarRIs: OVULES AND SEEDS IN CERCIS 547 
III. Summary AND Discussion 
The investigations described in this and the preceding paper 
establish several points concerning fertility and fecundity. The 
following may now be cited. 
(a) The physical constants — type, variability, and corre- 
lation—of the number of ovules per pod and the number of seeds 
developing per pod in Cercis canadensis differ sensibly from 
individual to individual and from habitat to habitat. The data 
© not, however, justify the conclusion that the trees from the 
different habitats are to be distinguished taxonomically. 
(6) The correlations for number of ovules formed and number 
of seeds developing per pod, ros, have always been found positive 
and of a moderate, considerable or even high intensity. 
This is true for the pods of an individual tree as well as for a 
mixed sample from a considerable series of trees. The correlation 
coefficient is slightly raised by the combination of collections from 
different individuals. 
(c) Regression is sensibly linear, both within the series of 
pods from the same individual and in a population of pods from 
many individuals. Possibly, however, there is a departure from 
linearity in the pods with eight ovules, but in my largest series 
there are only 36 of these pods out of a total of 28,554; this number 
is too small to be given great importance. 
The significance of the linearity of regression is two-fold. 
Statistically, it justifies describing the interdependence between 
the number of ovules formed and the number of seeds maturing 
by the coefficient of correlation. Biologically, it shows that the 
rate of increase in number of seeds developing per pod remains 
the same as we pass from pods with the lowest to pods with the 
highest numbers of ovules. 
(d) Wherever large series of pods have been examined, the 
correlation between the number of ovules per pod and the capacity 
of the pods for maturing their seeds, 7,2, has a negative sign and a 
low, usually a very low, magnitude. When the number of pods 
is relatively small — say about 100 as in the case of the correlations 
from individual trees—the coefficient is sometimes positive. 
These results may well be due to the probable errors of random 
sampling which, with samples of this small size, may be quite 
large enough to screen such a slight relationship. 
