40 BULLETIN 852, U. S. DEPARTMENT OF AGRICULTURE. 
too noticeable such a surface was wiped with thin mortar. Unless 
pitted no wash coat was applied. 
The great amount of curvature was cared for by making three 
types of bevel-end sections and making the degree of curvature for 
each curve on the line such that one of the three types could be used 
with a maximum bend of 3 C at each joint. This limit could be 
attained without inducing appreciable roughness within the pipe. 
As shown in Plate VIII. figure 1. the various sections of pipe were 
fitted together on the ground without the use of cement, a "soup 
hole" being broken out of the top of each section just before fitting 
the following section. The outside of each joint was banked up with 
clay or stiff gravel to a point well above the spring line, in order to 
prevent loss of grout. In cementing the joints the invert was 
troweled from the inside. Then a band of spring steel 6 inches wide 
was tightened against the interior of the joint and grout poured into 
the soup hole from a coal bucket. When the grout had set the band 
was removed and any necessary smoothing done with a trowel. The 
writer examined the interior of this line and found the joints as 
smooth to the touch as the rest of the pipe. Although this line is 
not covered with earth, no expansion joints were inserted at the time 
of construction, it being thought that the great amount of curvaure 
would automatically care for temperature changes. Experience with 
this line during the winter of 1916-17 shows that it would have 
been better to insert an expansion joint every few lengths of pipe. 
Siphon Xo. 1, between manholes 4 and 5, was chosen for the major 
tests upon this line. It is longer than any of the other siphons and 
near enough to the gatehouse so that alterations in discharge may 
comparatively quickly reach the section under test. 
Holes were drilled through the top of the pipe near each end of 
the siphon at such elevations that the pipe was completely filled at 
these holes regardless of discharge. Piezometer tubes of type A were 
thrust into the siphon through these holes. Water was turned out of 
the pipe while the piezometers were carefully placed on the inside. 
These tubes were connected with gauge glasses by means of which 
the pressure head was read directly in water columns. 
Velocities were determined by fluorescein injected into the hole 
in the concrete pipe made for piezometer Xo. 1 and observed through 
the manhole in the flow line at the end of the siphon, beyond the 
hole for piezometer Xo. 2. A correction was necessary for the 
reason that the water did not fill the pipe for a few feet between gauge 
Xo. 2 and the manhole. Data for this correction were carefully 
'taken with the level at the time the levels were run for the determi- 
nation of the loss of head. 
As stated in the discussion under Xo. 54a, page S6, there is a weir 
just above the intake to the pipe line. The crest is about one-fourth 
inch thick, rounded over on a radius equal to about one-half the 
thickness of the plate. Water entered the chamber above the weir 
from one side, making an indeterminate condition of approach 
velocity. However, a float gauge was read for each run of water 
and tlie discharge computed by the Francis formula. The com- 
parison of the velocity in the siphon pipe as computed from this weir 
discharge and as determined by the direct timing of color is given 
in Table 2. page IS. It is to be noted that, for all runs, the velocity 
by direct measurement is greater than as computed by weir discharge. 
