FORESTS, RESERVOIRS, AND STREAM FLOW 247 
and comparative records are of such recent date, that precise demon- 
stration is scarcely possible. The popular belief is based upon a 
fact and an assumption forming together a basis for a conclusion. 
The fact is that forests in the eastern portion of the United States have 
disappeared to a large extent within the past century. ‘The assump- 
tion is that floods and low waters in the same region are more fre- 
quent and severe than before the forests were cleared away. The 
conclusion is that these assumed conditions must be due to the dis- 
appearance of the forests. Post hoc, ergo propter hoc is the argumenta- 
tive process relied upon, and little effort is made to consider whether 
there may not be some other and more satisfactory explanation. The 
writer will attempt to analyze the problem from a theoretical stand- 
point and will then cite existing records so far as these are sufficiently 
long-continued to be worth anything. He will consider, first, the effect 
of the forests where stream flow results from rain alone, and, next, 
where it results in part from melting snow. 
Effect of Forests upon the Run-Off from Rainfail.—The first of 
the above propositions—the retentive action of the forest bed—may 
be accepted at once as strictly true for average conditions. It is not 
true for extreme conditions—great floods and excessively low waters— 
the conditions that determine the character and cost of river control. 
Consider an inclined-plane surface, practically impervious to water, 
with a layer of sand covering some small portion of it, and let a uni- 
form spray of water be applied to the entire surface. Assume that 
the temperature and rate of evaporation are relatively low. As soon 
as the spray begins, water commences to flow from the uncovered 
surface, but not for a time from that covered by the sand. After a 
while it begins to trickle from the sand, increasing in volume until the 
sand is thoroughly saturated, after which it flows off in as great quan- 
tity per unit area as from the uncovered portion. If the spray is 
stopped, the water immediately ceases to flow from the uncovered area, 
but continues in diminishing quantity from the covered area until it 
finally ceases altogether; but not all the water that fell on this area 
has run away. The sand has retained some portion of it and given 
it off in evaporation, so that the total run-off per unit area is 
somewhat less than on the uncovered portion. Jf the shower be long- 
continued and the rate of evaporation very low, the difference of 
total run-off per unit area from the two surfaces will be very slight. 
. Suppose now that the temperature and rate of evaporation are high 
