178 A NEW METHOD OF ESTIMATING STREAM-FLOW 
accounts for 197(1.23) +2(8.50) =0.259 c.f.s., leaving a residual of 0.044 c.f.s. 
That is, of the flood peak on October 9 of 0.586 c.f.s., 0.127 c.f.s. reached the stream 
by percolation, and 0.415 c.f.s. reached it by surface travel, making the total com- 
puted flow for that day 0.542 c.f.s., which is 0.044 c.f.s. or 7.5 per cent smaller than 
the observed flow. 
Referring to the October flood, Plate 13, it is seen that the curve of total com- 
puted flow does not rise, and particularly does not fall, as smoothly as the curve of 
observed flow. Instead of falling along a smooth curve, the computed-flow curve 
falls by steps. This is also seen in the residuals of the preceding tabulation. It 
should be evident that the cause of this is peculiar to the nature of the form of 
equation used, in which a single constant is made to account for the flow over an 
interval of several days. Obviously if the equation were lengthened out so as to 
account for each day's flow by one constant, it would require an equation of 30-odd 
terms to represent the flood-flows on this stream alone, which would be practically 
prohibitive, if not theoretically so. 
FLOOD-FLOW EQUATION OF STREAM A 
By the same process explained in detail on Stream B, resulting in equation 
(86), the following flood-flow equation for Stream A was derived: 
+ 1.6r /1 +1.7r /2 +1.9r /3 +1.6r /4 +8.9r /6 +4.5r /6 +1.3r /7 = Flood-now of Stream 4.... (87) 
in which the r/s are defined the same as in equation (86) ; that is, the same values of 
G may be used. 
The above constants, converted into percentages, give a total of 19.0 per cent 
as that part of the change in storage above the ground surface on a day which is 
delivered to Stream A by surface travel during a period of 33 days thereafter, and 
including the current day. 1 This is much smaller than the 29.6 per cent obtained 
for Stream B. 
RUN-OFF CURVES OF STREAMS A AND B 
The constants in equations (86) and (87) converted to percentages are shown 
plotted on Plate 8 above the normal run-off curves plotted from the constants of 
equations (76) and (77) converted to percentages and to a different scale on the 
left. In plotting the constants of equations (86) and (87), the last value plotted, 
R' / 7 , falls in the center of the 16-day interval covered by R' /7 . This is the end in 
each case of the known part of the flood-flow curve, as determined from the least- 
square computations. Evidently if these curves could be extended they would 
ultimately join the normal-flow curves at some point, which, for these two streams, 
is not less than 32 days after the current day. The curves are shown extended 
beyond R' fl as dotted lines. Similarly, the normal-flow run-off curves are shown 
extended beyond R' i0 as dotted lines to the end of the 256-day interval. 
The characteristics of the two streams as shown by the run-off curves are 
considerably different. Stream A responds to both normal and flood-flows more 
quickly and to a greater extent at first than Stream B. This might be expected 
from a comparison of the shapes and gradients of the watersheds (Plate 7) and from 
the dimensions (Table 40). Stream A is long and narrow in comparison with 
Stream B. The mean gradient of Stream A is slightly less than that of Stream B. 
In the case of normal-flows, after about the thirtieth day the rate of delivery, ex- 
1 Assuming the water table to be at ground surface level on each day of the 32-day period after the current day. 
