376 S$. H. BARRACLOUGH AND T. P. STRICKLAND. 
It is directly comparable with the curve marked “II.” Evi- 
dently the formula indicates velocities considerably below the 
actual ones when applied to conditions such as obtain in the 
present series. By arbitrarily choosing a smaller coéfficient of 
roughness for the purpose, the discrepancy could be diminished, 
but the coéfficient here selected is justified by the results of 
experiments made on channels similar to the present one, as set 
forth in the tables published in “The Flow of Water in Rivers 
and other Channels,” by Kutter and Ganguillet. 
16. Experiments at small slopes.—In carrying out Series II, 
three experiments (Nos. 8, 10 and 9) were made with slopes of 
"0025, ‘0020, and -0012 respectively. Unfortunately the results 
are not reliable (and are therefore not reported) because the water 
dit not attain any steady condition of flow. At times the surface 
of the stream would be “glassy” smooth almost the whole length 
of the channel, but the slightest disturbance caused it to break, 
' the section increasing, so that the water came down in a series of 
waves making it impossible to obtain the area with any degree of 
accuracy. 
APPENDIX. 
Since the foregoing paper was written, a second series of experiments 
on the effect of change of temperature on the rate of flow, has been 
carried out and the following table summarizes the results : 
Temperature Velocity. 
3 ae em. per sec. 
21°1 1058 
22°8 107°1 
26°7 108-5 
34°4 110°2 
42°5 gH he 
It will be seen at once that the effect of change of temperature is very 
appreciable, but the rate of increase of velocity is considerably less than 
found in § 13 above, indicating that the correction there determined is 
large. The absolute correction is however so small a quantity as to 
make the error a matter of but little importance. 
