DISCUSSION: FORESTS, RESERVOIRS, AND STREAM FLOW 387 
of mud, sand, and gravel, and to fall quickly. In dry seasons they are 
very low, and the range from lowest to highest stages has increased, 
and has inflicted much damage in very recent years. 
Two other classes of conditions may be recognized in Professor 
Glenn’s descriptions: One, of valleys which are still timbered, but 
which are being damaged by the commencement of logging operations, 
and particularly by the erosion of logging chutes; the other, of water- 
sheds which are largely cleared, but grassed, and which the grass pro- 
tects from erosion, though not in the same degree from rapid run-off. 
Such evidence as this, collected by a qualified and impartial 
observer, with reference to individual water-sheds, establishes, for the 
region in question, comprising the mountains of North Carolina, 
Tennessee, and Georgia, the fact “that forests do exert a beneficial 
influence on stream flow by storing the waters from rain and melting 
snow, preventing their rapid rush to the streams, and paying them out 
gradually afterward, thus acting as true reservoirs in equalizing the 
run-off.” 
As compared with these direct observations on limited water-sheds, 
the statistical data of precipitation and run-off on large streams, as 
presented by Colonel Chittenden, have a minus value for consideration 
of the question at issue. The stations of the Hydrographic Service 
and of the Weather Bureau are too few in number and too widely 
scattered. The method of analysis does not take account of the 
great diversity of conditions in an extensive water-shed, to which 
Colonel Chittenden calls attention.* Engineers know that a circuit 
of approximate observations may be run and close fairly well if the 
number of observations be sufficiently large, the result being the mean 
of many balancing errors; this is also true in the observation of flood 
stages or low-water stages, on the Ohio, for instance. In any case 
the stage is a mean of balancing conditions on hundreds of small 
water-sheds, some of which are forested and some deforested, some of 
Mr. Willis, 
which are and some of which are not cultivated. In that mean, the ; 
effect of the individual water-shed is so obscured as to be quite in- 
distinguishable. Conclusions based on inferences from such data 
have not the value which they should have, in order to carry weight 
with the engineering profession. 
Colonel Chittenden, further, says: 
“The commonly accepted opinion is that forests have a beneficial 
influence on stream flow: 
“(2) By retarding the snow-melting in the spring and prolonging 
the run-off from that source.” 
In regard to the influence of forests on melting snow, Colonel 
Chittenden argues “that the effect of forests upon the run-off from 
* Page 249, 
