204 A NEW METHOD OF ESTIMATING STREAM-FLOW 
date of full net melting by method (d), page 157, only. Thus, on Stream B, the 
maximum R' f was R'/ 4 = +1.64, hence, S}4 days preceding the day of maximum 
flow in the spring was the day of maximum addition to storage by melting, which 
was normally taken as the last day of full net melting. 
The systematic change noted in comment (6) above is a possible indication of a 
divergence of the facts from the simple assumptions stated as Assumptions Nos. 6 
and 7. These are indications that equation (34) does not represent the facts as 
closely as they might be represented. For example, consider equation (34) with 
reference to Assumptions Nos. 6 and 7, and comment (6). From the equation 
r= (rainfall) - (E .) (^-J - (run-off) (93) 
it appears that if the abstract number — - is made larger in July and August than 
E m 
the assumed constant value 2.6, r will be smaller, the computed stream-flow will be 
smaller and C — O will be a smaller +. That is, the evidence of a systematic distri- 
jp 
bution of the residuals with the summer season suggests that — - is not a constant 
E m 
but a variable whose value is larger in summer than in winter. On some streams 
■p 
a further refinement may be necessary. It may be proved that — varies with 
E v 
time between rains, being a maximum at the time of a rain, and diminishing to a 
minimum just before the next rain. This last variation would simply mean that 
the rate of evaporation is a maximum when the land surface is thoroughly wet, and 
diminishes as the water disappears under the ground surface. The increase of 
Ei 
— in the spring and summer would mean that the loss of water by evaporation 
E w 
may be enhanced because (a) of the larger exposed evaporation surfaces produced 
by the growing vegetation, and (b) because of the loss of water by that utilized in 
the plants themselves, and by transpiration. The increase in loss due to (a) will 
be offset partially or wholly by the decrease in loss from the land surface which 
will be partially shaded by the growing vegetation. 
Consider the distribution of the residuals in Tables 52 and 53 in connection 
with Assumption No. 8, page 137. If the temperature of the surface of the snow 
and ice is generally warmer than the surrounding air, the evaporation from land 
will be actually larger than as figured on the basis of equality of temperature, and 
vice versa. Hence the general presence of positive residuals in the winter months 
November to February, inclusive, is an indication that the surface of the snow and 
ice is, in general, warmer than the surrounding air. Similarly, the general presence 
of negative residuals in those months is an indication that the surface of the snow 
and ice is generally colder than the surrounding air. The residuals on Stream A 
appear to be generally positive in the winter months mentioned, whereas, on Stream 
B they appear to be generally negative. This evidence is, therefore, inconclusive 
and contradictory. The proper conclusion seems to be that, insofar as can be 
determined from this evidence, the temperature of the snow and ice surfaces and 
that of the air are about the same, and Assumption No. 8, insofar as it depends upon 
equality of these temperatures, is as near the truth as can be at present established. 
The evidence of a systematic change in the sign of the residuals in the summer 
season is too weak to be given much weight. It is evidently masked by other 
