ATMOMETRY AND THE ATMOMETER 29 



different time periods at the same place, it is quite essential that 

 the several atmometers employed shall be as nearly alike, in all 

 particulars, as is possible. Of course it is physically impossible 

 to construct several instruments exactly alike, and the slight 

 differences which occur should always be cared for by calibration, 

 as in the case of anemometers, thermometers and the like. Any 

 alteration of the instruments that may occur during the period 

 of operation should also be cared for by recalibrations carried 

 out from time to time. It is positively essential that some 

 single form of instrument be adopted throughout each series 

 of comparative measurements, and it is quite impracticable— 

 excepting in a roughly approximate way — to derive coefficients 

 by which readings from one form of instrument may be made 

 comparable with those from another form. 



Units of Measurement 



Connected with a discussion of atmometry in general, should 

 go some consideration of the kind of units to be used in com- 

 parative studies of evaporation rates. The misapprehension 

 mentioned above, that free water somehow furnishes the true 

 standard evaporating surface for such studies, is usually accom- 

 panied by another misapprehension, that units of depth are 

 somehow the truly fundamental units, in terms of which the 

 evaporating power of the air should always be measured. This 

 proposition is probably to be related to a desire, on the part 

 of meteorologists and climatologists, to compare rainfall with 

 evaporation. 



Rainfall is measured in depth units and its measurement is 

 not attended with serious difficulties; the size, shape and ma- 

 terial of the gauges exert relatively but negligible influence 

 upon the readings, and the variations in exposure commonly 

 allowed are not great enough to be considerable. Hence it may 

 have seemed to follow, that evaporation (which is obviously a 

 sort of negative rainfall) should be measurable in the same 

 kind of units. As has been indicated, however, many con- 

 ditions influence the rate of loss from an atmometer that do not 



