I 
222 Journal of Agricultural Research voi. v, no. 6 
The results of this comparison are given in Table III, which shows that 
the second method of computation gives the more accurate results. In 
no class of canal section does either three-point method give as accurate 
average results as the 0.2 and 0.8 depth method alone. In the individual 
experiments in one-seventh of the total number the 2 Xa6 
4 
method gave more accurate results than the 0.2 and 0.8 depth alone. 
In one-fifth of the total number the 
0.2 -f- 0.84- 0.6 
method gave results 
more accurate than the 0.2 and 0.8 depth alone. These were for gagings 
in which the errors of the 0.2 and 0.8 depth method were of different 
sign from those of the 0.6 method, so that their combination reduced the 
actual error. These cases were generally for canals of irregular section 
and flow, and indicate that for unfavorable conditions of current-meter 
work the three-point method may be preferable to the two-point, but 
that for usual conditions the two-point alone is preferable. However, 
under unfavorable conditions of irregular velocity and cross section only 
detail multiple-point observations can be depended upon for accurate 
results. The 
0.24-0.8+0.6 
method is always preferable for computation 
0.2 + 0.8+2 X0.6 . , 
of the results to the-method. 
Table III .—Variation in discharge in percentage by the three-point method compared 
with the multiple-point method 
Type of canal cross section. 
Number of 
observa¬ 
tions. 
Average variation 
met 
Giving velocity at 
0.6 depth equal 
weight with mean 
of velocities at 0.2 
and 0.8 depths. 
Mean velocity** 
0.2+0.8+2X0.6 
4 
from multiple-point 
hod. 
Giving velocities 
at 0.2, 0.6 and 0.8 
depths, equal 
weight. Mean 
velocity — 
0.2+0.6+0.8 
3 
Rectangular flumes. 
21 
+ 2. 5 
+ 1.8 
Concrete-lined trapezoidal sections.... 
15 
+ 2. 7 
+2. 0 
Shallow earth canals, sloping sides. 
II 
+ 1. 7 
+ 1-3 
Shallow earth canals, steep sides. 
21 
+2. 5 
+2. 0 
Earth canals, relatively deep sections.. 
14 
+ 3-5 
+2. 7 
Mean of all. 
82 
+2. 6 
+2. 0 
MEASUREMENTS WITH SURFACE FLOATS 
In many experiments measurements with surface floats were made in 
order to secure data from which the proper coefficients for use with such 
measurements could be derived. It is often convenient to make such 
approximate measurements by timing floats over a known length of 
canal and applying some coefficient to the product of the velocity so 
