94 BULLETIN 852, U. S. DEPART^IEXT OE AGFJCULTUEE. 
The writer's experience has been that curves should have a radius of 
not less than six times the diameter of the pipe. 
If these precautions are taken there is no reason why concrete pipes 
can not be so produced as to have a value of n = 0.01 1", or even 0.0105 
in Kutter's formula. It is therefore principally a question of engi- 
neering, which I am sorry to say has been greviously neglected in 
connection with concrete pipe work in the past. 
Little light can be thrown on the possible carrying capacity of 
concrete pipes by studying most of the old pipes of this class found in 
southern California, as they were nearly all laid out without any 
engineering supervision, or a proper survey to give uniform grade 
and proper alignment. A review of the author's experiments on 
these old pipes clearly prove this point. 
In practice, the carrying capacity of concrete pipes, in co mm on with 
other kinds, is often greatly reduced by the entrance of air at the 
intake. It is often customary to take the water into a pipe line 
from the overfall of a Francis weir for measuring it. The effect of 
this is to cause a large volume of air to be carried into the pipe with 
the water. The writer has often found this condition in connection 
with concrete pipes in southern California and elsewhere, and, after 
correcting it, has increased the carrying capacity of the line from 
10 to 20 per cent. 
The examples of concrete pipe given by the author, with which the 
writer was connected, are the 36-inch Boulder Creek concrete pipe 
line in Colorado, designated No. 52, experiment S 41, for which 
work he prepared the specifications and acted as consulting engineer. 
The experiment from which the author obtains a value of n= 0.012 
was near the upper end of this pipe. If the test had been made at a 
point farther down the line it is probable that a value of n = 0.0116, 
or better, would have been obtained, the same as for the Mill Creek 
Xo. 2 pipe (No. 50, experiment FF 1), because the two pipes were 
manufactured and laid in the same summer. 
It has been well demonstrated as a hydraulic principle that there 
is a slight but constant acceleration of the velocity in long gravity- 
flow pipe lines of this kind, and the writer has found a difference in 
the value of n near the upper and lower ends of such pipes, where all 
the conditions were exactly alike. This phenomenon has been referred 
to by some authorities and called by the name '"'Constant acceleration 
in gravity conduits.'' 
Both the Boulder pipe (Xo. 52, experiment S 41^ and the Mill 
Creek Xo. 2 pipe (Xo. 50, experiment FF 1) were made and laid by 
day work under careful engineering supervision, which explains the 
results obtained as to carrying capacity, although these pipes were 
both made in 2-foot sections by the dry-mix process. It is the opinion 
of the writer that, had these pipes been made with wet mix, they would 
have showed a value of n = 0.0105, or probably even a little better. 
As to the other two pipes, for which the "writer was engineer, 
namely Mill Creek Xo. 3 (Xo. 51, experiment FF 2) and Lytle Creek 
(No. 53, experiment FF 3 . it must be remembered that both of these 
were installed by contract and without proper precautions in regard 
to making the joints, which is wholly responsible for the higher value 
of n and less carrying capacity of these pipes. 
From the above and from all other experiences of the writer, his 
conclusion is that the values of n given by the author, as applicable 
to the best constructed pipes, which can be produced by the dry-mix 
