THE FLOW OF WATER IN CONCRETE PIPE. 73 
the mean velocity of four batches of fluorescein for each observation. 
The color was injected at the. standpipe marking the upper end of 
the reach and observed at a similar standpipe marking the lower end 
of the reach. The slope of the water surface was determined by 
piezometer tubes of type A connected with gauge glasses outside the 
pipe. The piezometer tubes were under identical dynamic conditions 
and examination of column 13, Table 11, shows that the corrected 
observed slopes practically agree with the nominal slope of the con- 
duit. It was necessary to correct the observed slope for changes 
in velocity head between the upper and lower ends of the reach, the 
flow not being uniform throughout the reach tested. It is the 
writer's opinion, based upon his experience, that " uniform flow" is 
an ideal that is assumed in design but seldom attained in practice. 
The pipe was designed under an assumed value of n in the Kutter 
formula of 0.012 and observations made by the writer prove this 
assumption to have been correct, even after a period of 6 years 
without cleaning. So far as examination of the interior could be 
made from the various manholes the conduit is clean and practically 
free from slime. As the water comes from a large reservoir located 
on a mountain stream, it is clear and cold at all seasons of the year. 
No. 56, Experiment S-49. — 42-inch jointed reinforced concrete 
pipe, Victoria Aqueduct, Vancouver Island, British Columbia, 
Canada. — As mentioned under the descriptions of Nos. 30, 54a, and 
55a, the Victoria Aqueduct consists of about 27 miles of flow line, 
broken by six inverted siphons (PI. V, fig. 3, is typical of this flow 
line). Simultaneously with the experiments conducted on siphon 
No. 1 (pipe No. 30, p. 39) readings were also taken on gauges at 
manholes 3 and 4. This reach of pipe, 1,986 feet long, is downstream 
from the reach 800 feet long tested by Ehle in 1915 (No. 54a). 
Piezometer tubes of type A, under identical dynamic conditions, were 
held upstream against the current at the two manholes. True siphon 
tubes were carried over the edge of the manholes and connected the 
piezometers with graduated gauge glasses. The gauge glasses were 
then considered in a scheme of levels and the fall of the water surface 
was thus determined. This method is probably more accurate than 
to accept the nominal slope of the pipe. The areas of the water 
sections at the ends of the reach for the various runs were determined 
by careful measurements in the manholes. The discharge of the pipe 
was taken as the velocity in feet per second (determined by color 
tests in siphon No. 1), multiplied by the mean area of the siphon 
interior (p. 41). This discharge, divided by the mean area of the 
water section in the flow line, gave the velocity within the flow line. 
The reach tested was typical of the whole aqueduct, being about 
half curve and half tangent. The friction factors confirm those 
found by Ehle and show the same decrease in the values of n as the 
depth of water (consequently the velocity and the hydraulic radius 
for the depths considered) is increased. It was not feasible at the 
time these tests were made to turn sufficient water into the pipe line 
to fill completely the flow section, as a repair, due to a hillside slip 
several miles downstream, was in progress. The values of the 
retardation factors show that 0.011 is probably as low a value of n 
as is feasible to obtain in a commercial pipe and should only be used 
for pipes under practically ideal conditions, such as hold on this 
aqueduct. 
