198 A NEW METHOD OF ESTIMATING STREAM-FLOW 
tion formulas which reproduce fairly well the observed daily discharges of the period 
used in computing the constants, or, on the other hand, do these formulas represent 
the facts of nature and are the derived constants real physical constants such that 
for these streams they express the relations actually existing at all times between 
the meteorological facts and the discharge of the stream? 
Obviously in making these comparisons one can and should distinguish between 
the months of observations of stream-flow that were used in computing the con- 
stants on the one hand, and the other months of observations which did not enter 
into the computations at all on the other hand. 
For Stream A, the computed constants R\, R' 2 , R\, . . . R\ depend upon 
the 8 months, August to September 1912, June to September 1913. 
S c to .R'io depend upon the 19 months July to December 1911, January to 
March and August to December 1912, and January to February and August to 
December 1913. 
R' n, R' n, R' 'ft, . . . R' n depend on little more than two months, October 4 
to November 4, 1911, just after the greatest October rain that occurred within the 
period of observation, and April 8 to May 9, 1912, in the early part of a spring 
flood. 1 
C, F and M depend upon the six months, February to April 1912, and February 
to April 1913. In computing them, observations on Stream B were used still more 
than those on Stream A. 2 
G depends upon an inspection of all the observations on Stream A in the three 
years, 1911-12-13. 3 
Ei/Ey, depends upon the three years 1911-12-13. The value used on Stream 
A was 2.3, instead of the more correct value 2.6, derived later. 
For Stream B, the computed constants .R'i, R\, R\, . . . R' 7 depend upon 
193 selected observations in the 18 months March, April, June to October 1911, 
May, July, August 1912, and April to November 1913. 
S C) R't, R\ and R\ depend upon the 31 months March 1911 to December 
1913 except for May 1911 and May and June 1912. 
R' jx, R' fi , R' ft, . . . R' ti depend upon 166 days of observation in the following 
months: October to November 1911, June to August 1912, June to August 1913. 
C, F and M depend upon 151 days of observation in February to May 1912 
and February to April 1913. These were the values of Solution X, Stream B, 
1 The values of R'/i, R'fi, R'ft, ■ • • R'n here referred to are not those shown in Table 51. The values in 
Table 51 are believed to be much closer to the truth than the values here referred to. The values here referred to 
were the ones actually used in computing the flood-flow on Stream A, and are as follows: 
•R'/i = +0.82 i2'/4=+0.85 fl'/7=+l-87 
#'/2= + i.00 «' / 6=+6.17 
fl'/3 = + 1.04 i?' /6 =+3.83 
Compare these values with those for Stream A in Table 51. There is considerable difference, which is believed 
to be due to the use of the approximate and large value of the melting constant M= +8.39 instead of the latest and 
best value derived, +5.19. 
2 The values here referred to are those in equations (62) and (63). It is certain that the accuracy on Stream 
A would be much greater if the final values, equation (91), were used in connection with the R'/s of Table 51. 
3 The values here referred to are not the values of G in Table 51. Those values 'were obtained from an inspec- 
tion of all the observations on both Stream A and B in the three years 1911-12-13. The values referred to above, 
obtained on Stream A alone, are as follows: 
D' on preceding day G 
85 or less 60 
86 to 95 50 
96 to 105 40 
106 to 135 30 
136 and more 20 
