226 A NEW METHOD OF ESTIMATING STREAM-FLOW 
By this method, for any given stream, the procedure would be as follows: 
First, compute its constants from the series of stream-flow observations which 
are available; 
Second, compute the stream-flow (by use of the derived constants) for other 
months and years for which no direct observations of stream-flow are available. 
Third, from the total of stream-flow data, both observed and computed, thus 
made available, make the estimates for the future by use of the methods for that 
purpose which now exist, and also by using extensions of those methods and new 
methods based upon the more complete understanding of stream-flow which has 
been gained. 
The new method has the inherent advantage permeating every part of it of a 
truer and more complete understanding of stream-flow. The gain from such 
improved understanding is difficult to estimate but it is certainly great. 
In applying the second part of the process to compute the stream-flow for 
months and years from which no stream-flow observations are available, one is free 
to choose any month or year for which the necessary meteorological observations 
are available. The meteorological record of the necessary completeness in any 
locality is ordinarily much longer than the stream-flow record. Therefore there 
will ordinarily be more possible extension of the computed stream-flow than is 
feasible or necessary to use. From the many months or years of meteorological 
record one may, with the constants of the stream before him, select by examination 
the months and years of extremely large flow, extremely small flow, or average 
flow, whichever is desired, and so secure from a given amount of computation a 
much more reliable estimate of the maximum flow, minimum flow, or mean flow, 
than would otherwise be feasible. 
RULES FOR APPLYING THIS METHOD 
In deriving the constants of the stream from the stream-flow observations 
which are available, the order of procedure would be as follows: 
(1) Estimate the mean, median and mode discharges by fitting a Pearson 
Type III frequency curve to the available data. Then by use of the values of K h 
K 2 and K s (100) page 217, estimate the S c of the stream by equation (97) page 212. 
(2) The estimated S e of step (1) becomes the R" of the general equation (59) 
page 148. Estimate R'\, R" 2 , R">, ... of equation (59) by comparison with 
other streams already studied, in connection with the general ideas set forth on 
pages 222 to 225. 
Ei 
(3) Estimate the ratio ~^r for the stream by use of equation (35a) page 137, 
III io 
in which the net melting is to be computed by equation (92) page 193, and E«, by 
equation (23) in Part I, page 82. 
(4) By use of the ratio -i established in step (3), and equations (92) and (23), 
JU tf 
compute r h r 2 , r,, . . . r„ for the stream for the years of available discharge obser- 
vations. In the case of a watershed on which the distance from the point of stream 
measurement to the farthest rain-gage is greater than 44 miles, the values of r, 
should be set forward according to the rules stated on page 225. The beginning 
of the period of decreased rate of melting is to be estimated by method (d), page 
157, assuming that the maximum R' t for the stream will be R' /3 , and during the 
