406 BOTANICAL GAZETTE [MAY 
borne, or by the quantity of the plastic material which it receives 
from the shrub, in a way to bring about a correlation between the 
length of a pod and the seeds which it matures when the pods from 
an individual are used as a sample, there should be a correlation 
between the number of seeds developing in a fruit and the length of 
another fruit on the same inflorescence. In short, if there is some- 
thing inherent in the inflorescence which tends to influence both 
number of seeds developing and length of fruit in the same sense, 
so that a correlation arises between them, this influence should 
effect in some degree all pods with the result that the cross corre- 
lation between number of seeds developing in one pod and length 
of another pod should have a sensible positive value. 
The only disadvantage of this method is its extreme laborious- 
ness. It is necessary to draw up tables between the characters of 
the fruits of the same inflorescence, just as in a study of heredity 
one prepares tables showing the correlation between brothers in 
the same family. Each pod is used once in association with every 
other pod on the same inflorescence. The number of combinations 
thus secured will be 42(mz—1), and since for practical purposes we 
use each fruit once as a first and once as a second member of an 
associated pair,’ we have for each inflorescence m(m—1) combi- 
nations. 
All inflorescences have not the same number of pods, and an 
inflorescence with six locules will give relatively fewer entries in 
the symmetrical table than one with twelve. I regret the necessity 
of thus giving weight to the larger inflorescences, but since the 
8 For the intra-inflorescence relationship for the length of the fruit it is only neces- 
sary to draw up tables showing all possible combinations of the fruit lengths of the same 
inflorescence. But for ovules and seeds we are dealing with the individual locules, 
and there are three to each fruit. If we made every possible combination in the 
preparation of the tables, we would be correlating, in some cases, between the locules 
of the same fruit, and, in some cases, between the locules of different fruits on the 
same inflorescence. The point which we wish to get at is the relationship between 
the different fruits of the same inflorescence. The plan followed, therefore, has been 
to correlate the number of ovules formed or the number of seeds developing in each 
locule with the number in every other locule on the inflorescence, except those of the 
same fruit. In the same manner, in dealing with the relationships between length of 
pod and the fertility characters, the tables were so drawn as to show the relationship 
between the length of the pod and the fertility characters of all the locules on the 
inflorescence, except its own three. 
