486 Harris: FACTORS INFLUENCING THE WEIGHT OF BEAN SEED 
varieties. These cultures were made under a wide range of en- 
vironmental conditions. They are here designated by the key 
letters employed in other publications. It is, therefore, un- 
necessary to burden this paper with information which the reader 
who cares to do so may obtain elsewhere.* 
The biometric methods employed in the analysis of the data 
should now be familiar in a general way to all serious biological 
workers.f 
Ill. PRESENTATION AND DISCUSSION OF DATA 
The physical constants for the unweighted frequency distri- 
butions of both number of pods per plant and weight of seed have 
been given elsewhere.{ 
The weighted frequency constants have little independent 
biological interest. We may, therefore, limit our attention to the 
degree of interdependence between the two variables. 
The correlation coefficients for number of pods and seed weight 
with their probable errors and their ratios to their probable 
errors appear in TABLE I. 
The regression straight line equations giving the numerical 
equivalents of 
w= ( — re? *7) + row! — >, 
op 
where the bars denote means and the sigmas represent the standard 
deviations of the characters indicated by the subscripts are given 
in TaBLE II. These show the absolute change in seed weight 
associated with unit change in number of pods per plant. If the 
regression straight lines and the empirical mean seed weights for 
plants with various sain be pods for a number of series be 
* See a tetbamrerascpti in Amer. Jour. Bot. 1: 410-411. 1915. 
r of pods per plant is an integral variate which has been treated without 
0.175-0.200, 0.200-0.225, 0.225-0.250. 
second moment for seed weight but not to that for number of pods per plant. 
Number of pods per plant was of course weighted with number of seeds weighed 
per plant. 
are sometimes based upon slightly different numbers than these here 
employed, but the constants are sensibly the same 
