22 KAINFALL AND FLOODS. 



earth and exposed to the rain, provision being made at 

 the bottom of the cylinder or tank for letting off and col- 

 lecting in a separate vessel the water which runs through. 

 The quantity of water which under such an arrangement 

 percolates through the soil is found to be very much less 

 than the quantity that falls. In some experiments the pro- 

 portion has been no more than 20 or 25 per cent, in the 

 whole year. In other experiments it has been about 50 per 

 cent. In the summer months it may be only 10 per cent., 

 or even less than that. These facts, which have been known 

 from the time of Dalton, are justly considered of great 

 importance in their bearing upon questions of water supply. 

 What becomes of the rain that does not percolate ? There 

 can, I think, be but one answer. It is lost by evaporation. 

 One sometimes reads in this connection of "absorption and 

 evaporation," as if these two processes were two distinct 

 causes of loss. This, I think, is not an accurate mode of 

 expression as applied to average results or results of long 

 period. Much — probably most — of that which is lost by 

 evaporation is first absorbed, the water subsequently rising 

 by capillary action, in every dry interval of w^eather, from 

 the deeper and moister soil, and evaporating from the sur- 

 face, while (in nature's experiments) a still larger portion, 

 perhaps, is taken up by the roots of plants and evaporated 

 from their leaves. But absorption itself cannot be the 

 cause of any permanent loss for the simple reason that it is 

 not a cumulative process. The amount of water held in 

 the surface soil and the subjacent strata will be the same at 

 the end of a period as it was at the beginning, or if there 

 should in any case be such a difference as materially to affect 

 the result, it will only show that the period taken for ob- 

 servation has not been long enough to eliminate accidental 

 error arising from variation of seasons. 



