Harris, Öbservations on the Physiology of Seed Development in Staphylea. 3 
and the number of seeds matured per fruit. Further data along 
these lines are much needed. 1 ) 
We now take up the data available for Staphylea. As a 
basis of comparison for the correlations for number of fruits per 
infiorescence and number of seeds per locule I have determined 
the same relationship for number of ovules per locule. 
To avoid the criticism so often urged against biometricians, 
that they obscure biological relationships by massing data, we again 
work with the individual shrubs for the 1906 collection. The 
data for the correlation between number of fruits per infiorescence 
and number of ovules per locule are given in Table I and those 
for the correlation between number of pods per infiorescence and 
number of seeds per locule in Table II. These are combination 
tables, comprising twenty individual correlation tables each. Data 
for r nf are unnecessary, for this correlation can be calculated from 
the data for r no and r ns 2 ) provided o f be known. 3 ) 
Table 1 gives the correlation coefficients. 
These constants are of so low an Order that it has seemed 
idle to calculate the sixty individual probable errors. As in another 
paper on Staphylea , I have used the approximation E r = . 04. If 
we demand that a correlation constant shall deviate from 0 by at 
least 2.5 times its probable error to be considered trustworthy, it is 
clear that all values between +. 100 and —. 100 must be looked 
upon as of questionable signifiance. 
Applying this criterion, we note that for r no thirteen of the 
twenty values fall within the limits of + . 100, and four exceed it 
so slightly that no significance is to be attached to them. Two 
constants are of a magnitude about four times our approximation 
to the probable error, and one, namely that for plant 28, is about 
eight to ten times its probable error. Surely no one would venture 
1 ) Unfortunately, in Staphylea we cannot determine the relationship 
between the number of flowers produced per infiorescence and the fertilty 
characters. Morphogenetically, it would be interesting to know whether the 
inflorescences which produce a high number of flowers have fruits with re- 
latively large (or small) numbers of ovules. From unpublished work, we know 
that there is sometimes a correlation between the number of fruits which an 
annual produces and the characteristics of these fruits, but as yet our knowledge 
of these matters is limited. 
*) It is hardly necessary at this time to note that the biometrician de¬ 
signateis the coefficient of correlation by r and the two characters between 
which the correlation is to be determined by the subscript letters. Thus, r no 
is to be read “the correlation between the number of pods and the number of 
ovules per locule“. 
The characters considered in this paper are: 
number of fruits per infiorescence = n 1 
Position of fruits on the infiorescence axis, numbered 
from the proximal to the distal end = p, 
number of ovules per locule = o, 
number of seeds per locule = s, 
number of ovules failing to develop per locule = f. 
3 ) Data for of are not published here since the correlations between the 
number of ovules failing to develop are not considered in great detail. They 
may be published in another connection later. 
1* 
