A NEW METHOD OF ESTIMATING STREAM-FLOW 183 
response of the observed discharge to temperature was very small, less than 0.005 
c.f.s., and on the computed was about 0.030 c.f.s. This is in the first month of the 
freezing-melting period. There was over 1.0 inch (of water equivalent) of snow 
on the ground. In September and October the two responses corresponded well. 
In November to December 1914 (Plate 11) the observed discharge shows more 
response than the computed, though the two responses corresponded fairly well in 
October 1914. The long decided freeze, November 28 to December 18, with a 
minimum temperature of +3° F. on December 12, produced a decided drop in the 
observed discharge, down to a minimum of 0.041 c.f.s. on December 16. It pro- 
duced only a very slight drop in the computed discharge — perhaps one-fifth as 
large — down to a minimum of 0.087 c.f.s. on December 13. At this time, up to 
December 10, there was less than 0.5 inch of snow on the ground. A similar freeze 
in December- January 1915 (Plate 12) produced about equal moderate responses 
in the observed and computed discharge. 
In November-December 1913 (Plate 10), and January-February 1914 (Plate 
11) the response of the observed discharge is clearly less than that of the computed 
discharge. This also seems to be true, to a less extent possibly, of November 1912 
to January 1913, and January-February 1912. 
The proper conclusions from this evidence seems to be: (1) that in November 
to February, a steady freezing period, the observed discharge is somewhat less 
responsive in general than the computed discharge, and (2) that the observed dis- 
charge, very low, in December 15, 16, 17, 1914, is an abnormal, non-typical case, 
which has no counterpart on Stream B (Plate 16). 
This evidence is clearly in favor of a reduction in the derived value of M, +8.39. 
The decision to reduce M from +8.39 to +5.00 as based upon the evidence 
(a), (6) and (c) was made from a study of the computations and graphs on Stream 
A. Subsequent to that decision, a value of +5.04 was obtained in Solution X on 
Stream B in which the flood-coefficients referred to in (c) were used. This fourth 
piece of evidence, not connected in any way with that from Stream A, was a strong 
corroboration of the conclusions reached on Stream A. 
In Solution X on Stream B, the derived value of C was +3.97, which was much 
less than the value +6.40 used in Solution M. It was believed at that time that 
this reduction in C was caused by the introduction of the flood-coefficients into 
Solution X, which would theoretically make it nearer the truth. Hence in estimat- 
ing the dates to use in the computation of D f in Solution A A, Stream A, C was 
assumed equal to +4.0 and M, +5.0. Using these values and the best value of 
T", viz, 28° F., obtained up to that time, a prehminary computation of n, and from 
that, (wi — Gi), (rii — Gi), . . . , was made as shown on page 184. 
This computation was begun on March 29 because previous to that time 
(t—T") was largely negative, hence no flood-flow could be produced. The values 
in the second to fifth columns were computed in a manner analogous to that already 
described in detail in the illustration of the computation of net melting and reduced 
rate of melting, pages 159 and 160, except that F was not used, inasmuch as a flood- 
flow can not be initiated by melting when t < T" . The unit in all but the second 
and last columns is 0.01 inch of depth. The last column is in 0.1 inch of depth, and 
the second column is in °F. The values of n in the sixth column are the sums of the 
values of net melting in the fifth column and the values of Hi shown in the computa- 
tion of D„, page 180, 1 and G is obtained from the tabulation in Table 46 in con- 
'ByEq. (41), page 141. 
