Frictional Resistance in Artificial Waterways 5 
check measurements being made by two methods. In some cases 
the discharge measurements were not made within the experimental 
section, but at some convenient support near the upper or lower end 
of the section. In every case, however, the flow was measured 
at a point near enough to make the seepage loss a negligible 
quantity. 
In some places there was no foot bridge across the channel and 
wading had to be resorted to, which introduced a slight error in 
the discharge measurements, but it will be shown later that where 
the coefficient of roughness alone is desired, a small c *ror in the 
determination of the discharge has an inappreciable effect upon the 
■alue of the coefficient. In all current meter measurements, cross- 
sections were chosen in which the filaments of flow apparently ap¬ 
proached and receded from the section in parallel lines. 
Values for the coefficient of roughness in the general formula 
of Ganguillet and Kutter, were secured from the field data by 
graphic methods for slopes up to .0621 feet per foot, and by com¬ 
putation for greater slopes. 
METALLIC FLUMES 
All of the metallic flumes upon which tests were made are of 
the semi-circular type, and may be divided into the following three 
classes, according to the characteristics of their interior surfaces: 
GROUP 1. Flumes whose connections at the joints are 
countersunk to the plane of the sheet metal, and which present a 
smooth, unobstructed water face. 
GROUP 2. Flumes whose joint connections protrude into 
the waterway beyond the plane of the sheet metal. 
GROUP 'L Flumes constructed of corrugated metal and 
whose corrugations at right angles to* the line of flow offer the 
only frictional resistance. 
In most of these flumes tie beams extended across the water¬ 
way, upon which bench marks were established at the upper and 
lower ends of the sections, and at intermediate points where neces¬ 
sary. Because of the fluctuations of the water surface, the vertical 
distance from these benches to the water was measured directlv 
with the rod. readings being taken at the highest and lowest levels. 
In this wav it was nossiblv bv averaging the two readings to mini¬ 
mize the error in obtaining the slone of the water surface. 
The weight of the water caused distortion of the original 
semi-circular cross-sections, and consenuentlv it was necessarv to 
make two sets of measurements to obtain the wetted cross-sectional 
area and wetted perimeter. The mean maximum dentil of watet 
