A NEW METHOD OF ESTIMATING STREAM-FLOW 199 
shown in Table 49. Note the close correspondence of these values with the 
final values (91). 
G depends upon all the observations on both streams in 1911-12-13. 
jp 
— = 2.6, depends upon all of the months from March 1911 to December 1913 
E w 
inclusive. 
Thus in each case, the constants depend upon the observations of 1911-12-13. 
The observations of 1914 and 1915 were not used in determining them. 
ACCURACY AS TESTED BY GRAPHS. EXPLANATION OF PLATES 9 TO 19. INCLUSIVE 
The comparison of observed stream-flow and computed stream-flow on Stream 
A for each day of 4 years is shown on Plates 9 to 12 inclusive. On Stream B the 
comparison is shown for 5 years on Plates 13 to 17 inclusive. 
The heavy solid line is the observed stream-flow. The parts of the high peaks 
which would fall beyond the upper edge of the sheet are plotted below on a reduced 
scale one-tenth as large as the scale of the main drawing. 
The horizontal, light, solid line is the constant part of the stream-flow, S c , 
0.108 cubic foot per second for Stream A, and 0.096 cubic foot per second for 
Stream B. 
The dotted fine is plotted from the S c line as a zero and represents the change 
in storage in the drainage area, both above and below ground, in each day. 
The dash-dot-dash line is the computed normal stream-flow, without flood- 
flow. The dash-dot-dash line (with dots enclosed in small triangles), where it 
occurs, is the total computed flow — the sum of the computed normal flow and 
the computed flood-flow. Where no dash-dot-dash line (with dots enclosed in 
triangles) occurs, there was no computed flood-flow. The highest dash-dot-dash 
line, with dots not enclosed in triangles in ordinary times and with dots enclosed 
in triangles in times of flood-flow, is everywhere the total computed flow. 
The line made up of long dashes and extending from the upper left-hand 
corner of the lower right-hand corner is the duration curve of observed stream-flow. 
The dash-two-dot-dash line is the duration curve of total computed stream-flow. 
If the theory were perfect, if the formulas as written represented accurately 
all of the relations between the meteorological elements and the flows of these 
streams, if the adopted constants were the exact physical constants, and if no 
errors or approximations were made in the computations, the computed stream- 
flow would agree exactly with the observed stream-flow and the highest dash-dot- 
dash line would everywhere coincide with the solid black line, and the duration 
curves would coincide. To the extent that there is an approach to perfection in 
the various respects indicated the highest dash-dot-dash curve of the hydrographs 
should be near the heavy solid curve and should be of the same shape. Similarly 
the dash-two-dot-dash duration curve should be near the dashed duration curve, 
and of the same shape. Detailed comparison of the curves is invited. They fur- 
nish convincing indication of a close approximation to a satisfactory degree of ac- 
curacy. The agreement is not poorer in the last two years, 1914-15, than in the 
first two or three. It is doubtful if anyone could tell simply from the examinations 
of the curves which three years out of the five were used in determining the con- 
stants. 
On Plates 18 and 19, respectively, are shown the mean duration curves for 
Streams A and B for the period 1912-1915 inclusive. The mean duration curve of 
