Oct. 1,1926 Sulphate Content of Leaf-Tissue Fluids*of Cotton 
659 
Except for a few outlying analyses, 6 the sulphate content of the 
Upland type is higher than that of the associated Egyptian plants. 
This is shown by the fact that practically without exception the 
sulphate contents of the Upland variety lie above the diagonal cells 
(marked by the short rules) about which they should be concen¬ 
trated if the content were the same in the two types. 
Determining the coefficients of correlation between the sulphate 
content of associated Egyptian and Upland plants, the values are 
set forth in Table V. 
Table V .—Correlation between sulphate content of associated plants or groups of 
plants of Egyptian and Upland cotton grown under irrigation at Sacaton, Ariz ., 
in 1922 and 1928 
Table 
Correlation co¬ 
efficient and 
probable error, 
r±E r 
Ratioof 
correla¬ 
tion to 
probable 
error, 
r/Er 
Comparison of Pima Egyptian and Meade Upland cotton, 1922: 
First series, July 25 to Aug. 9, correlation between Pima and Meade. 
II 
0. 6521±0.0720 
9.06 
First comparison of Pima Egyptian and Lone Star Upland cotton, 
1923: 
First series, July 29 to Aug. 14. Correlation between Pima and 
Lone Star__ ___ 
III 
. 1047± .0815 
1.2^ 
3 
V 
5.8^ 
Second comparison of Pima Egyptian and Lone Star Upland cotton, 
1923: 
Second series, Aug. 18 to Aug. 31. Correlation between Pima 
and Lone Star___ 
IV 
\ 3970± . 0679 
The three coefficients are all positive, ,and two of them may be< 
considered statistically significant in comparison with their probableJJ] 
errors. They are, however, more irregular in actual magnitude^ 
than those demonstrated for chloride content ( 5 ). © 
The positive correlation indicates that the sulphate contents ofg| 
associated Egyptian and Upland plants are influenced in a similar© 
manner by some common extrinsic factor. It also indicates thepj 
necessity for considering the correlation between associated de¬ 
terminations in the calculation of the probable error of the difference 
between them. 
The mean sulphate contents of the individual series and their 
probable errors, the differences in the mean sulphate content, of the 
Egyptian and Upland types, and the probable errors of these differ¬ 
ences appear in Table VI. The probable errors of the differences are 
calculated with due regard to the correlations between the variables 
by the formula 
G (p — u ) “ & u ^^pu Gp 
The differences are from 9 to 26 times as large as their probable 
errors, and hence are clearly significant. Absolute differences are, 
on the average, 3 to 4 grams of sulphate per liter. Percentage differ¬ 
ences, calculated by using the sulphate content of the Upland type 
as a base, range from about 18 to about 28 per cent. 
6 These exceptions are presumably due to experimental errors but are included in order to avoid any 
possibility of the selection of data in the formulation of the conclusion. 
LIBRARY 
