204 
Journal of Agricultural Research 
Vol. VII, No. 4 
Barley and Burt oat increased in transpiration more rapidly and ripened 
more rapidly than wheat. The ratio graph for Swedish Select oat shows a 
steady increase from seedtime to heading, when there is a marked drop, 
followed at first by a gradual and finally by a rapid rise as the wheat 
ripened. The graph for rye shows a relatively greater transpiration than 
Kubanka wheat early in the season, decreasing as the season advances. 
No marked difference is shown in the flax varieties. C. I. 13 ripened 
nine days ahead of the other varieties. Except for a gradual increase 
in the ratio graph throughout the first crop, alfalfa E23 and 162-98Ai 
showed no marked differences. 
CORRELATION OF DAILY TRANSPIRATION WITH WEATHER FACTORS 
AND WITH EVAPORATION 
The correlation of the daily transpiration of a plant with the intensity 
of the weather factors presents difficulties, owing to the fact that the 
transpiration coefficient undergoes a gradual change from seed time to 
harvest. (See march of transpiration, p. 189.) It is consequently nec¬ 
essary in a correlation study to eliminate as far as possible effects due to 
changes in the size of the plant and to ripening processes. This can be 
accomplished by comparing the ratio of the transpiration on consecutive 
days with the ratio of the intensity of a given weather factor for the cor¬ 
responding days. Since the fluctuations in transpiration from day to 
day are large in comparison with the daily change in the transpiration 
coefficient, the effect of the latter is thus minimized. 
The use of direct ratios is not, however, wholly free from objection, 
owing to the fact that the departure of the ratio from unity is not the same 
in both directions. Transpiration ratios less than unity will be confined in 
the transpiration table to classes lying between o and 1, while transpira¬ 
tion ratios greater than unity have infinity as their upper limit. This 
can be avoided by correlating the logarithms of the ratios, instead of the 
ratios themselves, the departure of the logarithm being independent of 
the direction in which the ratio is taken. As an example, consider three 
successive days during which the transpiration is 2, *8, and 2 kgm., 
respectively. The ratio of the first to the second is 0.25, while that of 
the second to the third is 4. The departure of the ratio from unity is 0.75 
in the first case and 3 in the second, while the logarithms of the ratios, 
— 1.3979 (thatis, —0.6021) and 0.6021, respectively, show the same depar¬ 
ture from zero. 
Since it was not practicable to determine the daily transpiration with 
an accuracy greater than 0.1 kgm., the uncertainty of the ratio of the 
transpiration on consecutive days increases as the daily transpiration 
decreases. For this reason only pairs of terms in which at least one of 
the pair showed a transpiration of 0.6 kgm. or more have been used in 
the correlation tables. 
In all cases where the available transpiration measurements were 
sufficiently numerous to justify the procedure, the correlation has been 
