IRRIGATION IN NORTHERN COLORADO. 7 
The wide seasonal variation is due to the spring flood produced by 
rapid melting of snow in the hills. Its duration and intensity de- 
pend on the amount of snow to be melted and the temperature, a con- 
tinued high temperature producing a rapid rise, a high crest early in 
the season, and a subsequent rapid fall, and a low temperature pro- 
ducing a more gradual rise and fall with comparatively low and 
late crest. The average date of the crest is June 10, but it has come 
as early as May 17 and as late as June 28. The discharge at the 
crest has varied from a mean of 1,550 second-feet on June 19, 1888, 
to a mean of 5,800 second-feet on June 23, 1917. The lowest dis- 
charge is about 30 second-feet and occurs in winter after severe cold 
weather has frozen most of the stream. 
The daily rise and fall is pronounced only during the spring flood, 
when alternate freezing and thawing of snow on the high slopes of 
the basin produce a variation of several hundred second-feet at the 
gaging station, the maximum recorded being 1,500 second-feet. Sud- 
den floods, due to storms, are common and there are records showing 
a rise of more than 5 feet in less than 30 minutes, caused by cloud- 
bursts in narrow branch canyons with steep slopes. 
Records of discharge of the river from 1884 to 1917, inclusive, are 
given in Table 2. During the winter months ice conditions at the 
gaging station are such that automatic records are of little value, 
and such as were available were discarded. Estimates of the flow 
from November to March, inclusive, were furnished by John Arm- 
strong, water commissioner for the stream, who has handled the di- 
vision of the winter flow for over 25 years. To arrive at the annual 
discharge, Mr. Armstrong's estimates for the winter months were 
combined with available figures for other months, and then, if April 
or October records were partly missing, they were interpolated in 
the proportion of the percentages shown in the table. The winter 
flow is so small that this method of estimating could not produce an 
error of as much as 5 per cent. Data included in the table are from 
the original records of the Colorado Experiment Station and from 
reports of the State engineer. In considering this table it should be 
noted that the North Poudre and Poudre Valley Canals divert about 
40,000 acre-feet above the gaging station annually, and that part of 
the water passing the station is foreign water from other drainage 
basins. 
