2 53 
Reduction of Transpiration Observations. 
discrepancies of this magnitude forces us to the conclusion that very little 
reliance can be placed on figures for evaporation in linear measure obtained 
on the assumption of the area law. 
7. Effect of Observational Errors and Deductions 
THEREFROM. 
It remains now to discuss the effect of given errors in the observed 
quantities on the resulting errors in the computed quantities. The point of 
most importance so far as we are concerned is the relation between the 
percentage error introduced into the value of the radius (or area) of the 
equivalent water surface of an atmometer by a given percentage error in 
the determination of n. 
Let E l and E 2 represent the evaporation per unit time from the atmo- 
meter and calibrating dish respectively ; let r 2 represent the radius of this 
latter vessel, and r that of the surface equivalent to the atmometer. Then 
and, remembering that E v E 2 , and r 2 are given constants, we obtain at once 
by differentiating logarithmically 
Sr Sn . Vo 
— = — log. 
r n r 
This result shows, as is otherwise obvious, that, when the radius of the 
calibrating dish is very nearly equal to the equivalent radius of the atmo- 
meter, a large percentage error in n may have very little effect in producing 
errors in r, and it may be very usefully and easily applied further to reduce 
any uncertainty that may exist as to the true value of the equivalent radius. 
The procedure, then, that appears to us to be safest to follow in 
calibrating an atmometer is — first perform the calibration with two or three 
vessels filled to about $ centimetres from the brim ; reduce the observations , 
assuming the area law , for each basin separately , and thus obtain a mean value 
for the equivalent area, in the manner previously demonstrated. This value 
may be slightly in error , owing to a slight error in the assumed value ofn; 
but if we now repeat the experiment , using as a calibrating vessel a dish whose 
area is as near to this mean value as is conveniently possible , the value of the 
equivalent area deduced from this last experiment will be accurate enough for 
all practical purposes . 
8. Concluding Remarks. 
During the course of his experiments on atmometry, Livingston first 
used, then discarded, the method of calibration by comparison with free 
water surfaces 2 and in the experiments carried out during the year 191a 
1 Livingston (1906), loc. cit. 
2 Ibid. : Operation of the Porous Cup Atmometer. Plant World, 13, 1910. 
