26 BURTON EDWARD LIVINGSTON 



hastened that from the porcelain cylinder." This is no doubt 

 correct to a considerable extent, though temperature and humid- 

 ity changes may also have had an influence. Nevertheless, in 

 spite of the clear evidence that the ratio of hourly loss from cup 

 to hourly loss from pan actually fluctuated as much as from 84.7 

 to 114.4, an average of all the hourly ratios was taken as showing 

 the relation of the two rates for the entire day. Such a mean 

 value does, of course, represent the period for which it was ob- 

 tained, but the night hours were not included in that study. If 

 the different days of the general period were nearly similar, and 

 if the average ratio employed had been derived by use of night 

 as well as day rates, then the reductions of cup readings to units 

 of depth (given on p. 29 of the same publication) should be 

 approximately correct. As it is, these reductions are at least 

 highly questionable excepting in a merely approximate way. 



The mistake just mentioned befogged the writer's apprecia- 

 tion of the general atmometric problem for several years, and 

 the real condition of affairs in this connection came to conscious- 

 ness but slowly. In the summer of 1909 the method of standard- 

 izing porous cups was altered so as to make use of a water surface 

 as a basis. At this time it was clearly appreciated that the kind 

 of pan used must be highly important, but the possible difference 

 between the effects of a given change in the surroundings, upon 

 the rates of loss from cup and pan, was quite lost sight of — not- 

 withstanding the suggestion of this difference already made in 

 1906. Accordingly, the fullest account of the porous cup at- 

 mometer so far published (the writer's paper of 1910, already 

 cited) gives an erroneous method of reduction of evaporation 

 rates from this instrument to units of depth from free water. 

 The experimentation upon which this procedure was based lasted 

 for several summer months, and showed satisfactory agreement 

 between the reduction coefficients from day to day, no doubt 

 because no very great climatic changes were then encountered 

 and the effective conditions for each day of the series were much 

 like those for any other. 



During the following summer, however, a period of extraordi- 

 narily hot weather brought the coefficient of the standard cup (of 



