juiyso, 1917 Evaporation from Water and River-Bed Materials 255 



to control the water level ; by the tank arrangement it was necessary to 

 entirely cut off the supply at night. During a part of the time a mid- 

 night observation was made, but not through all the season. Occasion- 

 ally a greater depth than was required would be applied to the tanks, 

 resulting in an excess of perhaps }i to % inch, never greater. The 

 net result for the end of the season and for any day of the season was 

 saturation. As all of the work done was performed with more care than 

 is ordinarily used, the only explanation for the apparent discrepancy has 

 been given. 



In the final curves the results of this saturated condition are not incor- 

 porated. Examination of streams shows that the bed is very rarely in 

 that condition. Water may be flowing in measurable depths, from a 

 very small part of an inch up, or the sand will be moist, with the water 

 table below the surface at varying depths. The saturated condition is 

 rare in the extreme. 



Tanks with water at depths of 3 and 6 inches were maintained through- 

 out the run of the other tanks. The results are not given, since in weigh- 

 ing water was spilled at different times before the weight had been taken. 

 Figures for losses from these tanks, when known to be correct, cor- 

 responded almost exactly with those for tanks in the series of varying 

 depths, tanks 18 to 22, inclusive. 



With the detailed data and curves of figure 11, together with the 

 analysis curves for the river-bed material to work with, it was necessary 

 to devise a means of showing the evaporation results based upon the sand- 

 analysis figures in order to make the data of maximum value. There 

 are two indices in common use representing the type or classification of 

 a particular sand: Effective size and uniformity coefficient. The ad- 

 vantage of basing evaporation losses from wet sands upon one of these 

 was obvious. A determination of both effective size and uniformity of 

 coefficient was made for the materials used in the tanks. The effective 

 size apparently offered the greatest possibilities. However, the water 

 losses did not follow this consistently; uniformity coefficient was not at 

 all applicable. The 60 per cent size, used in arriving at the uniformity 

 coefficient from the effective size, was found to be much better. It is 

 a term and classification in quite common use, defined as: That size of 

 grain of material such that in mechanical screen analysis 40 per cent of 

 the material by weight is larger and 60 per cent is smaller than that size, 

 or based upon screen opening, that size of opening such that 60 per cent 

 of the material passes and 40 per cent is retained. 



When evaporation amounts were plotted as ordinates upon this 60 per 

 cent size as abscissae, fair curves resulted. To place the data in shape for 

 use. Table XXV was prepared. 



