Jan. 3,1916 
Hourly Transpiration Rate on Clear Days 
637 
the departure of the transpiration from evaporation should not be taken 
as proof of a change in the transpiration coefficient of the plant and that 
it is safer for the present not to base conclusions on this assumption but 
instead to consider directly the factors which influence both transpiration 
and evaporation. 
CORRELATION BETWEEN TRANSPIRATION AND ENVIRONMENTAL 
FACTORS 
Two methods have been employed by the writers in making a quanti¬ 
tative investigation of the relationships existing between the transpiration 
of the plant and the intensity of its environment: (1) The coefficient of 
correlation between the transpiration and a given environmental factor 
has been computed as a basis for the determination of the relative influ¬ 
ence of the various environmental factors and (2) the relationship between 
the mean hourly transpiration and the hourly values of the several envi¬ 
ronmental factors has been computed by the method of least squares, and 
the relative weights of the different environmental factors determined 
from the coefficients of the resulting equation. Such a reduction of the 
data appears highly desirable, for it affords a means of comparison inde¬ 
pendent of the personal element. The results of the correlation reductions 
will first be considered. 
In computing the correlation coefficients, 1 the individual hourly obser¬ 
vations as presented in Tables I to XXVI were used. The data in each 
instance embrace not less than three days’ observations with the transpi¬ 
ration measurements in duplicate, so that the number of pairs of terms— 
i. e., the “population” considered—approximated 144 for the 3-day 
periods in the transpiration correlations and in other cases exceeded this 
number. 
The correlation coefficients of the transpiration rate of alfalfa, amaran- 
thus, and rye, with the intensity of the several environmental factors, 
are presented in Table XXXVII, together with the probable error of the 
correlation coefficient in each case. 
1 For a presentation of the theory and the method of computing correlation coefficients, see Yule (19x2) 
and Davenport (1907). 
