324 Mo7'p1iolofjif of the Fruit of Sanguinaria Canadensis 
is essential to our understanding of fertility. This is a knowledge of the influence 
of the number of ovules per fruit upon the chance of an individual ovule 
developing into a seed. This is quite a different problem from that of the 
relationship between the total number of ovules per pod and the total number of 
seeds developing. 
As pointed out in an earlier paper*, the constant which we need is the corre- 
lation between the number of ovules and the deviation of the number of seeds 
from their probable number. The coefficients calculated by the method described 
in that paper are given in Table 14. 
TABLE 14. 
Correlation hetiueen Number of Ovules and Deviation of Number of Seeds from 
the Probable Number on the Assumption of the same Proportional Fecundity 
throughout the Population. 
Eelatiouship between Ovules 
and Seeds 
Correlation and Probable Error 
1906 
1907 
Difference 
For both Placentae 
For the Fruit 
-•178+ -015 
-•191 ±-021 
•038 + ^024 
•051 ±-034 
+ •216+ •028 
+ -241 ± •039 
These results show how important it is to have more than one series of material 
upon which to base conclusions when the correlations are of a low order of 
magnitude. Considering only the 190G series we see constants with a negative 
sign and about ten times their probable error. It would seem from these results 
alone that as tlie number of ovules per placenta or per fruit increased the capacity 
of the fruit for maturing its ovules into seeds decreased. But when the 1907 
series is taken into consideration also we note that the correlations are hardly 
twice their probable errors but with the positive sign. Of course it may be 
possiljle that a larger sampling in the 1907 plants would have given a sensible 
negative correlation. But all that we can conclude is that while there may be 
a relationship between the number of ovules per fruit and the capacity of the fruit 
for maturing its seeds there is no very satisfactory argument for such a theory in 
our present collections. 
These conclusions agree with the results announced for one sample of 10,000 
pods of Gercis given as an illustration in the paper on the method of determining 
the relationship between ovules and seeds, but do not agree with the results 
for the general sample of Robinia. The factors underlying differences of this kind 
deserve investigation. 
* Harris, J. Arthur, Biometrika, Vol. vi. pp. 438—443, 1909. 
