2i6 Journal of Agricultural Research voI.x.no. s 



there are differences in methods, that recommended for evaporation (17) 

 work has been followed. For example, the Denver branch of the Weather 

 Bureau makes observations regularly at 6 o'clock mountain time. 

 Weather Bureau regulations for evaporation observations recommend 7 

 o'clock, which hour has been used as far as possible, both morning and 

 night. Readings were taken once a day from the beginning of the work 

 until April i, 1916. From then until October 5, 1916, observations were 

 made twice a day, and then for the remaining portion of the year dis- 

 cussed herein regular readings at 7 a. m. only. Special and check read- 

 ings were taken at various times of the day and night. 



ACCURACY OF MEASUREMENTS 



No observer worked for a shorter time than one month, during which 

 period he was responsible for all the readings from a particular set of 

 instruments or tanks. He made from 60 to several hundred observations. 

 Thus, the error resulting from frequent change of men, that of personal 

 equation, has been eliminated as far as possible. 



The hook gage used for determining water losses by evaporation is 

 calibrated to o.ooi foot. Neither the gage used at the lake nor the one 

 at the laboratory was changed throughout the season. The datum on 

 each tank remained unchanged also. Some difficulty was expected and 

 found when measurements were made during winds. The still well was 

 of great benefit on the smaller pans, but on the large tanks a high-velocity 

 wind piles the water up at one side, so that a hook-gage reading might 

 not necessarily indicate the true water \q^^ for the pan as a whole. 

 Check observations were always made after a windstorm, and, even 

 though individual figures were in error, the weekly totals, which have 

 been used in the computations, have eliminated largely the errors due to 

 wind movement. Plate 34, B, shows another result of a windstorm. 

 The blown sand is caked at the side of the tank. The amount that 

 actually is blown into the tank is difficult to estimate, but is small for 

 the year. The catching of sand can not be prevented when the tank is 

 exposed where sandstorms occur. Attention is called to Plate 34, C, 

 which shows a typical winter condition. Regular observations of evapo- 

 ration were not attempted for the time during which the ice surface was 

 not readily broken. This period is indicated on the table of results. 

 The figures are more consistent than was expected, since the addition by 

 snow to an evaporation tank, having the surface of the water at about 

 the same level as the ground surface at that point, can not be accurately 

 determined from rain-gage results. Water in the tank may be slightly 

 warmer than freezing at the beginning of the snowstorm, and all the snow 

 may be retained by the tank. It then becomes colder and freezes; part 

 of the snow may then drift up against the rim of the tank and part be 



