A NEW METHOD OF ESTIMATING STREAM-FLOW 
209 
The evaporation from land was computed from equation (34) in which the 
value of Ei/E„ used was 2.3, which was computed from the totals in Table 596. 
This value is smaller than the value, 2.6, computed from the calendar years 1911-13 
in the illustration of computation of Ei/E„ on pages 160 and 161. Also this ratio 
computed from the totals for the calendar years 1911-15 is 2.6. 
The principal facts brought out in Table 596 are that the computed net melting 
is possibly too large, and the computed evaporation from water is probably too 
small. The latter resulted in a ratio Ei/E„ = 2.6, which is undoubtedly too large; 
that is, the true evaporation from land is probably not 2.6 times as great as that 
from a water surface. 
In Table 36, Monthly Weather Review Supplement No. 17, the precipitation 
and run-off for the years beginning October 1 are as shown in the accompanying 
tabulation in inches of depth on Watershed B: 
Water year 
Precipitation 
Run-off 
1911-12 
1912-13 
1913-14 
1914-15 
Total 
Mean Annual . . 
21.49 
19.66 
21.84 
19.85 
8.38 
5.21 
5.55 
5.40 
82.84 
20.71 
24.53 
6.13 
The precipitation is said to be (in Supplement No. 17) 52 per cent in the form 
of rain and 48 per cent in the form of snow. Applying these averages to the mean 
annual values for the four years, the average precipitation as rain appears to be 
10.77 inches (20.71X0.52), and as snow 9.94 inches (20.71X0.48). 
From this it appears that the mean annual rainfall shown in Table 596 is 
[ — '- — =1.08) 8 per cent larger than that given in Supplement 17, the mean 
annual estimated net melting is ( — : — =1.66] 66 per cent larger, and the total 
precipitation estimated in this investigation is 36 per cent larger ( — '- — — — : — = 
1.36) than the total precipitation shown in Table No. 36, Supplement 17. 
The estimate of rainfall in this investigation was made from the original record 
of observations at Wagon Wheel Gap, and the distinction between snow and rain 
was made according to notations shown in that record. The above 8 per cent 
discrepancy in rainfall can be attributable to different interpretations of those 
records. 
The computed net melting is possibly too high and casts suspicion upon the 
derived freezing-melting constants, C, F, M and T". This external evidence shows 
that the small probable error of M (page 195), especially, is probably fictitious to 
some extent, due probably to the rejection of too many observations in the spring 
months used in the observation equations of Solution X. Since the values from 
Solution X are very close to the best values obtainable to date in this investigation 
(91), the latter possibly give a net melting which is too large. 
