248 Thomas and Ferguson . — On the 
(approximately i*6), and that for a vessel filled to about 3 centimetres from 
the brim, under similar circumstances, the value of n is practically equal 
to 2. 
Unfortunately the importance of filling the calibration vessel to a 
definite distance from the rim has not, so far as we are aware, received due 
recognition. The vessel is usually spoken of as being ‘ filled ' with water ; 
and, where any quantitative details are given, this appears to mean ‘ filled 
to within 5 or 6 millimetres of the brim \ 1 Darwin and Acton, indeed, in 
discussing a very similar problem, recommend the use of a ‘ shallow vessel ’ ; 2 
as their results, however, are to be reduced by the area law, it is the very 
opposite that is required — a deep vessel, filled to about 3 centimetres from 
the brim. Otherwise, the calculated results cannot have any quantitative 
meaning. 
The serious difference that may exist between the true value of n and 
that usually assigned to it may, as we shall show, introduce errors in the 
calculated value of the equivalent water surface of an atmometer that may 
be as great as 30 or 40 per cent. This being so, it is hardly too much to say 
that any figures obtained from two different atmometers, and any figures — 
such as those of ‘ relative transpiration ’ — based on readings of atmometers 
calibrated by comparison with water surfaces assuming the area law, cannot 
by any means be compared one with another. Not only are the numerical 
magnitudes of the figures involved probably in error, but the degree of the 
error, depending as it does on the true value of n, is also variable, and is 
variable by an amount that cannot be estimated so long as the exact 
conditions of the experiment remain unspecified. 
It is evident then, that to remove this unfortunate uncertainty, it is 
necessary to calibrate an atmometer by means of some quantity that shall 
possess those qualities which are usually demanded of a standard or unit— 
that is, it must be constant under constant and easily specified conditions, 
and must be capable of reproduction at any place or time. The figures 
cited above seem to show that a free water surface contained in a cylindrical 
basin fulfils these conditions. The surface is one which can easily be 
reproduced and the evaporation from such a surface follows a regular law. 
Further, the value of the constants in 
E — koP 1 
can readily be calculated. Secondly, after the depth of the free surface of 
the water below the rim of the containing vessel has reached a certain value, 
evaporation is proportional to the area, and therefore a water surface of this 
kind may, without further difficulty, be used for standardizing purposes. 
1 Yapp filled the crystallizing dish used to calibrate his atmometers to ‘ about 3 mm. below its 
upper edge’. Loc. cit., p. 311. 
2 Darwin and Acton : Practical Physiology of Plants, 1S94, p. 89. 
