1912] HARRIS—STAPHYLEA 209 
For the 16 plants of the 1907 series, the correlation for length 
and ovules is positive in 13 cases, and negative in 3; significantly 
negative in one instance, and significantly positive in 11. The 
values are low, but, as in 1906, there seems to be a slight positive 
relationship between the length of the pod and the number of ovules 
formed. 
For length and number of seeds all the correlations are positive, 
and all are significantly positive with regard to their probable 
errors. Furthermore, they are of a substantial order of magnitude, 
ranging from 0.150 to 0. 500. 
Comparing, as in 1906, by taking the differences y,—m., with 
their probable errors as shown in table IV, we find that in one case 
the difference is negative’; of the 15 positive differences, 13 are 
over 2.5 times their probable error and hence trustworthy. 
Diagrams for 7, 7s, and 7;—?. make the relationship very 
clear. Graphs for only one year (1906) need be published. In 
these diagrams (fig. 1) the vertical line represents the zero line on 
either side of which the constants would fail if they were due 
merely to the chance errors of random sampling. The magnitude 
of the constants for the individuals, or the difference in their corre- 
lations, is shown by the length, and the sign by the direction of the 
bars. The amount of the constant which might be due to the 
probable errors of random sampling (2.5 times its probable error) 
is shown by the unshaded area, while the shaded portion gives some 
idea of the biological trustworthiness of the constant. These 
diagrams make very clear to the eye that 
1. The length of the fruit and both the number of ovules formed 
and the number of seeds developing are interdependent, and often 
moderately closely interdependent. 
2. The correlation for length and number of seeds per locule is 
significantly higher than that for length and number of ovules per 
locule. 
3. These two facts taken in conjunction indicate that there is 
some physiological relationship between the length of the fruit and 
the number of seeds developing. 
7 It is significantly negative in comparison with its probable error. I find no slip 
in the arithmetic. The constant is based on only 53 pods, and so too much signifi- 
cance must not be attached to it. 
