1912] HARRIS—STAPHYLEA 409 
of the same inflorescence as compared with those of the tree in 
general. 
Finally, the results from the two correlations fundamental to 
our present purposes are unmistakable. The evenness of the 
division between positive and negative and the extremely low 
mean value of the coefficients 7y.»=0.014 and r."=0.025 prove 
that there is no material relationship between the length of one 
pod and the number of ovules or seeds in another pod of the same 
inflorescence. This seems to me to prove conclusively that there 
are no differences in the supply of plastic materials of the different 
inflorescences sufficient to account for both the number of seeds 
developing and the length of an individual fruit deviating from 
their means in the same way. 
For Cercis I have no data for the mature inflorescence com- 
parable with that for Staphylea. It is interesting, however, to 
compare the intra-inflorescence correlations for number of ovules in 
the ovaries collected at flowering time in the spring of 1907 with 
the results obtained for Staphylea. The data for roo” are given in 
table XXII. The correlations are 
Tree 1, r=0. 571#0.015 
Tree 2, r=0. 213+0.020 
Tree 3, r=0.378+0.021 
Mean, r=o.388 
These results indicate a differentiation among the inflorescences 
of an individual of Cercis somewhat higher than we have found in 
Staphylea, but three individuals are not sufficient for more than a 
suggestion. The finding of an intra-inflorescence correlation in 
another genus gives confidence in the results for Staphylea. Should 
correlation between the placentae of the fruit is due to ecological and physiological 
factors. There, however, we were dealing with individuals subjected to an externa 
environment, not with the several inflorescences of the same individual. Possibly 
(a) and (6) are optimum for all inflorescences, but only a small percentage of the seeds 
develop because of some internal limiting factor. The bladdery fruits are possibly 
adaptive, and too great a weight of seed would offset the advantage of the peculiar 
structural features. 
* Calculations by method described in Amer. Nat., 1910, The “ovules of first 
ovary” gives the grade, and the “number of associated ovaries” gives the frequencies 
for both variables, the tables being symmetrical. 
