value of v' and elevation above the bed, and (b) between the root- 
mean-square value of v' at a fixed elevation and the flow velocity, Up. 
The assumptions also allow an approximation of the absolute magnitude of 
the root-mean-square value of v'. The horizontal component of turbulent 
velocity fluctuation is probably on the same order of magnitude as the 
vertical component and therefore, for qualitative analysis, the value 
of K"' in equation (20) can be approximated as equal to unity. 
The procedures used in measuring velocity distributions in the flume 
were as follows. A period and an amplitude of flume oscillation were 
selected and the flume linkage adjusted to give a symmetric motion. 
After the flume was filled with deaerated water to the elevation of the 
wave suppressent board, the asymmetric roughness elements were adjusted 
to eliminate secondary currents in the central part of the flume. The 
sensor was then placed in the flume as near the bottom as possible and 
its elevation recorded. The flume was started and the motion allowed to 
continue until equilibrium flow conditions were established. A record 
of the hot-film bridge output voltage was made on magnetic tape, the 
length of which was an integer multiple of the flume oscillation period. 
The flume was stopped and the sensor elevation raised for a new measure- 
ment. The procedure was repeated until an elevation was reached at which 
the velocity fluctuations were too small to be accurately measured with 
the anemometer. The sensor was then lowered in a stepwise manner to 
obtain velocity measurements at intermediate elevations. In this manner, 
10 to 13 velocity-elevation measurements were obtained to give a velocity 
distribution for the flow condition used. The period of the flume was 
changed and the measurements repeated to give a second velocity distribu- 
tion. In all, four velocity distributions were obtained for four differ- 
ent flow conditions. 
3. Results. 
The purpose of the velocity measurements was to obtain the following 
three relationships needed for an analysis of the suspended-load equation: 
(a) An approximation of the magnitude of the root-mean-square value of v' 
versus flow velocity, Up; (b) the distribution of the root-mean-square 
value of v' versus elevation above the bed; and (c) the distribution 
with time of v' at a constant location in space. Results pertaining 
to the third unknown listed above will be discussed first. 
Two sets of data were analyzed to determine the distribution with 
time of v'. The period and amplitude of flume oscillation for both sets 
of data were 10.48 seconds and 0.925 foot, respectively. In both cases, 
the length of record analyzed was 10.48 seconds (1,224-voltage samples). 
One set of data was taken at an elevation of 0.168 foot above the crest 
of the artificial dunes, the other 0.209 foot above. 
The data were analyzed in the following manner. For each voltage 
sample recorded, the effective heat transfer velocity, U,, was calcu- 
lated from equation (14). The velocities were ordered and percentages 
595 
