288 FORESTS, RESERVOIRS, AND STREAM FLOW 
the perfect plan. It has always appealed to the imagination of lay- 
man and professional alike. It has often been resorted to, and the 
number of reservoirs in the world is very great and constantly 
increasing. Hitherto they have been mainly used for power, municipal 
supply, irrigation, and for navigation in canals. In very few instances 
have they been applied to improve the navigation of large natural 
watercourses, and in none, so far as the author is aware, for the 
exclusive purpose of preventing floods. 
The question arises, why are they not regularly applied to these 
last-mentioned purposes? The answer may at once be given that in 
the general case the cost is greater than the benefits to be received. 
This element of cost arises mainly from the absence of good sites 
(including dam sites as well as holding basins) and also, to consid- 
erable extent, from an interference with the purely artificial condi- 
tions growing out of the settlement of the country. 
The best reservoir site is a natural lake. Such a site is already 
-eovered with water, and original conditions are not materially changed. 
Evaporation is not much increased by the necessary enlargement. 
Smaller and safer dams accomplish a given storage than for the 
average dry site. The question of public health involved in uncov- 
ering large areas for reservoir beds in the heated portion of the year 
is less serious. Everything makes these sites the most advantageous 
that can be found, and it may be laid down as a rule that the public 
good requires the utilization of every such site to the fullest possible 
extent.* 
Except in a few cases, dry sites are deficient in these advantages. 
Greater areas of land have to be condemned, and larger and costlier 
dams are required, with vastly greater danger in case of accident. 
Really good sites are not as abundant as one might wish, and the 
*An interesting feature of these natural reservoirs may be noted. A natural 
lake, wholly uncontrolled at its outlet, may have a more effective control of the 
outflow than an artificial reservoir of equal superficial area when full, though of far 
greater capacity between high and low water. The outflow from a lake can be 
increased only by storing simultaneously a quantity of water measured by a rise in 
the surface equal to that in the outlet necessary to give the increased flow. But 
if the artificial reservoir has reached the limit of its allowable filling, the outflow 
must be made equal to the inflow. If this limit is reached before or at the time 
of maximum run-off, then a quantity equal to this run-off must be let out of the 
reservoir. This contingency can never happen in a natural lake. The turning point 
where outflow and inflow balance each other is always after the crest of the flood 
has passed—in fact at the time when the diminishing inflow and increasing outflow 
balance each other and the lake ceases to rise. In the case of the Yellowstone Lake 
(140 sq. miles), for example, this rise, in average seasons of snow-melting, con- 
tinues from 10 days to 3 weeks after the inflow has reached its maximum, and 
surrounding streams have subsided materially before the Yellowstone River (at the 
lake outlet) ceases to rise. 
