17 
that for L895, much of the water escaped and was not available for irri- 
gation. Neither the reservoir nor the feeder-ditch capacity was suffi- 
cient to store more than a fraction of the Mow for May and June, nor 
would it he economy to increase them to an extent sufficient to store 
such floods, occurring as they do at rare intervals; in fact, only once 
in twelve years is such a Hood recorded. Therefore a uniformly dis- 
tributed rainfall, as in L895, even though it give a much smaller 
percentage of run-off as compared with precipitation, is better than 
that of a year like L900, when most of it occurs in a short period. 
The variations in the percentages of run-off and its causes will now 
be considered. The run-off for L895 was 52 per cent, while that of 
L900 was !>7 per cent of the precipitation. The great loss of 1895 was 
caused by the long exposure to the dry winds, a large amount of even 
the early winter snows passing directly from the solid to the vaporous 
state. The remarkable run-oil' of L900 is due principally to the rapidity 
with which the water reached the stream and to the short exposure. 
Two other causes, which might be classed as errors in the data used, 
contributed to swell the percentage of run-off: First, the precipitation 
on the lower reaches of the river, as shown by the Arkins station rec- 
ords, was greater than that shown by Moraine station records; second, 
almost no old snow remained after the summer of 1900. Snow banks 
which had been considered wi perpetual" had by the action of the rains 
been almost entirely melted. This snow, deposited in previous years, 
of course does not appear in the records of precipitation for the year 
1900. Considering the long exposure of 1895, the percentage of loss 
by evaporation is not remarkable. In the year 1899 there was a large 
precipitation in the early spring, probably in the form of light snow, 
which, melting rapidly as the weather warmed, caused high water in 
the early summer. The bountiful snowfall of the preceding Novem- 
ber and December furnished a good late flow. 
The record of Longs Peak station failed to establish a relation 
between a late flow and the snowfall in the early winter; but here we 
have, when considering the precipitation for lower altitudes, appar- 
ently positive proof that early winter snows do augment the flow. 
What, then, is the reason for the apparent slower run-off at the lower 
altitudes \ As already mentioned, the timber near the headwaters is 
still well preserved and new timber is coming up where fire passed 
some years ago. In these low r er altitudes the soil is deeper, the many 
small depressions form lakes and marshes, and numerous small grassy 
parks are found; all these aid in retarding the escape of the water. A 
good fall of snow in the spring will lie in the timber for a long time; 
such as falls on the little open parks is absorbed by the ground as it 
melts; the melting in the timber is slow" and almost all the water sinks 
into the ground, and, while the snow itself may disappear early, con- 
2817— No. 118—02 2 
