Mr. Willis. 
386 DISCUSSION: FORESTS, RESERVOIRS, AND STREAM FLOW 
evaporation in the shade. The amount thus retained is small as com- 
pared with that which is transmitted through wet litter to ground 
storage, but it is an amount which can be controlled by regulating the 
litter. The proportion of precipitation absorbed by forest litter, there- 
fore, need not be an injurious minus quantity if the water is wanted 
in the stream; but it is a beneficial minus quantity when there 
is more water than can be handled. The algebraic sum of these 
opposite results depends on climate, topography, and water service, and, 
under conditions of good forestry, determines what proportion of 
litter shall remain beneath the trees, and to what extent measures shall 
be taken to substitute an open soil for it at the surface. He reasons 
without thought who reasons without reference to forestry in regard to 
forest conditions. 
Thus considering the limiting case of great precipitation, it must 
be recognized that the forest on a slope holds back the most water, 
unless it can be shown that percolation into a bare slope exceeds 
loss from tree tops, loss in the forest litter, if any, and percolation 
into the ground under the forest. There are cases where percolation 
into bare ground on a slope reaches this surprising proportion, as 
on sand hills, and on some very coarse granite soils of equally open 
texture; but it does not reach this proportion in loams and clay soils; 
on the contrary, on these quick-spilling soils, percolation is a very 
small part of precipitation. 
Statistics of comparative run-off from bare and forested slopes, 
which are observed in such a way as to yield data for close comparison, 
are not known to the writer. It is probably safe to say that they have 
not been observed in the United States. To obtain them, one must 
observe the same precipitation on adjoining areas, of like slope, soil, 
and geological structure. Only in this way can the forest influence 
be isolated, and its effect ascertained statistically. The writer, however, 
has before him the results of three years’ investigation by Professor L. 
C. Glenn, of Nashville, Tenn., in the Southern Appalachians, the most 
critical region of the United States, and has gathered from Professor 
Glenn’s manuscript report to the Forest Service the following summa- 
tion of details which bear on this question. 
Forty-six creeks and small rivers, the water-sheds of which may be 
described as timbered rather than cleared, are known, according to 
observation and local repute, to rise gradually in times of flood, to 
continue high for several days, and to subside slowly. They carry 
but little sediment, and they maintain a good volume of water during 
dry seasons. 
Thirty-eight creeks and small rivers in the same water-sheds, having 
slopes which may be described as cleared rather than timbered, are 
known, according to observation and local repute, to rise rapidly in 
times of flood to extreme flood height, to carry excessive quantities 
