124 BOTANICAL GAZETTE [AuGUsT 
the correlation between the number of ovules formed and the size 
of the fruit may be of such a magnitude that it is impossible to tell 
without relatively refined statistical analysis whether the influence 
apparently due directly to the seeds may not be referred to the ovules. 
In the present case, for instance, the correlation for number of seeds 
developing exceeds only very slightly that for number of ovules formed. 
We may determine whether the number of seeds developing has 
per se any correlation with the length of the pod by the following 
simple process. We sort our material into classes according to the 
number of ovules per pod, and then into subgroups according to 
the number of seeds developing, and determine the mean pod length 
for each of these subgroups. If now the length of the pod and the 
number of seeds developing are in some measure interdependent, we 
should expect to find significantly different mean lengths for pods of 
the same number of ovules, but different numbers of matured seeds. 
TABLE II 
MEAN LENGTH OF PODS FOR DIFFERENT NUMBERS OF SEEDS PER POD IN THE FOUR 
HIEF OVULE CLASSES 
NumEeEr NUMBER OF OVULES PER FOD 
OF SEEDS 
PER POD ‘ 5 6 
fea: 55-46 61.75 ee ee 
Roe Ss 64.24 66.94 68.73 69 .00 
i aN ioe 68.37 .06 50.37 72.00 
PS RS a ian aan 72.41 73-85 73-97 
ee me hs ta 76.39 71-12 
Mey © ee 79-76 
Such an arrangement of the data is given in table II for the classes 
in which the number of observations is sufficiently large to give smooth 
results. Here the results are very clearly favorable to the hypothesis 
of a relationship between the number of seeds developing and the a 
length of the pod, independent of the relationship for number of 
ovules, for the mean increases as the number of seeds increases Whe? 
we work with constant numbers of ovules per pod. 
The same problem may be approached in another manner; i ; 
has the advantage of giving a terse quantitative statement of the a 
independent relationship between the number of seeds developins = 
