292 
Journal of Agricultural Research 
Vol. XXV, No. 7 
the same place. Although not typical of the region from which it came, 
this sample of Lone Star is in itself not atypical of cottonseed. Egyptian 
seed from Bakersfield, Calif., contained a slightly larger quantity of 
gossypol than the three other varieties from this place. This, as well as 
the high gossypol content of Egyptian seed obtained from Arizona, 
suggests that there may be slight differences between Gossypium her - 
baceum and Gossypium hirsutum, although the single analysis of Sea 
Island seed does not confirm this. 
The results of all the analyses indicate that the occurrence of an 
intoxication due to gossypol would not be influenced by the variety of 
seed from which the meal is made, but rather by the place from which it 
came and the season in which the seed is grown. If a varietal influence 
upon the gossypol content actually exists, practically it is concealed. 
The manner of cultivation (agronometric) also probably plays a rdle. 
29 
30 3 / 
32 33 34 33 36 37 33 39 40 4 / 
£ TP£P EXTRACT PER C£/VT 
Fig. i. —Relation between ether extract and nitrogen content of cottonseed. Since r is negative, the 
nitrogen content decreases as the ether extract content increases. The value of r (0.82), which is large 
as compared with its probable error (0.04), is significant. The partial correlation coefficient (0.62) indi¬ 
cates a significant relationship between the ether extract and nitrogen content. 
Although only presumptive evidence upon this point exists in these 
experiments, it is a logical supposition to make from the results of Bain 
and Anders reported by Cook (5). 
Figures 1, 2, and 3 give the mathematical expression and interpreta¬ 
tion of the writer's data. 7 In these computations each individual 
gossypol analysis has been used. The correlation coefficients 8 between 
ether extract (oil) and nitrogen (protein), and between gossypol and 
ether extract, respectively, show that relationships exist, but that they 
are not perfect. The reality of these relationships is further borne out 
by the determination of partial correlation coefficients. 
An apparent correlation exists between the nitrogen (protein) and the 
gossypol; in fact, the results of one analysis may be used to a certain 
extent to estimate the other. That this relationship may be false is 
shown by the fact that their partial correlation coefficient is very low. 
7 The calculations were made and the charts were plotted by J. C. Munch. 
# The correlation coefficient is a measure of the relationship between two variables. It is a relative value, 
zero, (o) indicating no relationship and unity (1) perfect relationship. The partial correlation coefficient 
indicates the relationship between two variables when other known variables are eliminated. 
