J. A. Harris 
327 
The agreement is remarkably close. The condition r >rj for the ovules is due 
to the use of Sheppard's correction in the calculation of r. Without Sheppard's 
correction for the standard deviations we find r = '9203, where the theoretical 
condition 7]> r is fulfilled. For the direct correlation for both ovules and seeds 
developing the differences in the constants for the two seasons are about four times 
their probable errors. But considering the possibilities of error in the sampling of 
the habitat in the two years I would not attach much significance to them as 
indicating real biological differences in the two years. The other two correlations 
differ in the two years by an amount of about the magnitude of its probable error. 
The significance of these constants seems to me to be twofold : (a) descriptive, 
(b) as a means of gaining a somewhat deeper insight into the physiology of seed 
development. 
The constant showing the degree of interdependence between the number of 
ovules on the two placentae taken in connection with the means and standard 
deviations, is a terse description of these features in the fruits of a given popula- 
tion. As it stands it is interesting as showing the high degree of perfection in the 
morphogenetic process giving rise to the ovary*. After studying a considerable 
mass of reduced data, some of which I hope to publish before long, it seems to me 
that a comparative study of these interplacental or inteilocular correlation coeffi- 
cients presents some problems of considerable interest. 
These organic correlation coefficients seem to me to illustrate very well the way 
in which statistical methods may be applied to certain physiological problems. 
The direct correlation for number of ovules per placenta is higher than that 
for either number of seeds developing or number of abortive ovules. Since the 
ovules are uninfluenced by the accidents of nutrition or fertilization which affect 
the number of seeds or the number of aborted ovules, the correlation between the 
number of ovules on the two placentae is quite independent of that for the other 
fertility characters ; but the converse is not at all true. 
We may ask to what extent the correlation between the number of ovules per 
placenta influences the correlation for number of seeds per placenta. The number 
of seeds developing per placenta is closely correlated with the number of ovules, 
hence a correlation between the number of seeds on the first and on the second 
placenta would necessarily be found even though there were no physiological 
relationship whatever between the development of the seed on the two sides of the 
fruit. Our object now is to remove the influence of the correlation for number of 
ovules on first and on second placenta from the correlation for seeds developing. 
If there is a physiological interdependence between the two placentae of the fruit 
depending on some factor other than the number of ovules present, we should 
expect to find a significant correlation between the number of seeds developing 
after the contribution of the ovules has been removed. 
* Since the tables are symmetrical the slope of the regression line is the same as the coefficient of 
correlation. 
