90 



E. M. HARVEY 



that water could be discharged upon the cups, maintaining either 

 a fihn over the entire evaporating surface or only over a portion 

 of it {i.e., any kind of "shower" desired). Rain drops striking 

 the cups should exert a certain amount of pressure not reproduced 

 by the simple discharge of water above, but the effect of this 

 extra force, from the impact of the rain drops, would undoubtedly 

 be only to increase, somewhat, the rate of absorption. 



When a film of water is made to cover the entire evaporating 

 surface of the cup (as in a very hard shower), the mercury valves 

 reverse immediately, but after that there is neither loss nor gain 

 of water in the burette. If the water film covers only a portion 

 of the cup (as is often the case during moderate rains), it was 

 found that the area of the film necessary to suppress loss from the 

 burettes depended upon the evaporating power of the air (the 

 greater this evaporating power the larger must be the area of the 

 water film) and upon the height of the cup above the surface of 

 the water in the burette. The height of the cup influences the 

 rate of absorption as shown below. In an equilibrium of this 

 sort, where there is neither a loss from the burette nor a reversal 

 of the mercury valves, the evaporation from the free surface of 

 the cup exactly equals the absorption of water from the portion 

 covered by the film. 



AVhen the film intermittently covers the cup (as in frequent 

 showers with rainless intervals) the backward flow of water into 

 the burette was again only the amount required to operate the 

 mercury valves, this amount of back flow (error of the instru- 

 ment) occurring as often as the valves were forced to reverse. 

 This jcsLse differs from that of moderate rains probably only in 

 giving more complete reversals of the valves, but less often. 



The amount of the error in the reading, for each complete 

 reversal of the valves, depends always upon the size of the tubing 

 that dips into the mercury of the valve between the cup and 

 the water supply, and upon the height of the cup above the supply. 

 The tubing used in this experiment had an inside diameter of 

 approximately 0.8 mm. The volume per linear centimeter was 

 about 0.005 cc. The length of the mercury column when the 

 valves are reversed (i.e., the back flow of water involved in the 



