THE FLOW OF WATER IN WOOD-STAVE PIPE. 93 
may involve a change in velocity head or be due to an incorrect mean area of the pipe 
being taken, and consequently an incorrect velocity used in a given test, if the indirect 
method of measuring the velocity is employed. The fact that the interior of most 
pipes tested in commercial service can not be examined means that the nominal 
diameter of a pipe must in many cases be accepted or that no experiments be made. 
The reverse of this problem confronts the designer. He must make computations on 
the basis of a given size of pipe, whereas the pipe when actually built may vary slightly 
from the designed size. 
No appreciable error was introduced by reading but one gauge glass of a U-tube 
manometer at a time. It must be remembered that before reading the pressure was so 
throttled that the mercury was barely "alive. " This means that in either gauge glass 
the "breathing" of the mercury did not extend over more than two or three thou- 
sandths of a foot. Any error of reading due to a difference in the size of the tubing, as 
questioned by Mr. Williams, would be so small a part of these two or three thousandths 
of a foot as to be negligible. Reading alternate legs of the manometer and computing 
the leg not read would tend to neutralize any error that did exist, remembering, of 
course, that both legs were "dead" columns of mercury, at intervals of approximately 
10 minutes. 
In answer to Mr. Williams's question as to comparing watches after the tests, the 
author would say that this was done and any necessary corrections of a few seconds 
was immediately noted in the field notes. Many experimenters are very careful to 
conduct all work with a stop watch. These watches may be read to a nicety but can 
still be inaccurate unless most carefully adjusted with a chronometer. This is quite 
evident if several of them be started together and compared at the end of an hour. 
Additional information on the weirs used has been written into the original text or 
shown in the views in conformity to Mr. Williams's suggestion. 
The accuracy of the meter calibration of the Altmar weir in connection with the 
tests on pipe No. 51 is questioned by both Mr. Noble and Mr. Williams, for the reason 
that the meter measurements must be made in the tailrace. As shown by figure 2, 
Plate V, the water downstream from the weir was not turbulent. When the gaging 
bridge was reached several hundred feet downstream from the weir, the water was 
flowing evenly and smoothly. The channel is excavated in flat, rather level strata 
of rock, and the four meter gaugings conform so closely with an even curve that the 
author does not believe any undue error was made. The verticals in which the meter 
was held were sufficient in number so that the section was well covered. The mean 
velocity of the channel was but 3.25 feet per second for the greatest discharge meas- 
ured. Such a velocity in a relatively smooth channel will not produce turbulence. 
Regarding the experiments on loss of head at entrance to pipes, Mr. Williams appar- 
ently misunderstands the author. The loss of head was not observed at a point three 
diameters down the pipe but at a point several diameters down the pipe, and the loss 
between the inlet and the 3-diameter point was computed by deducting the friction 
loss per foot for the number of feet bach from the point of observation to the 3-diameter 
point; thus finally the net loss from the intake to the end of a short tube 3 diameters 
long was deduced. 
Regarding the entrainment of air: A large proportion of the wood pipes used in 
irrigation practice are on inverted siphons between sections of open channel. When 
such a pipe is running at full capacity the inlet is usually well submerged and but a 
moderate amount of air is caught and carried into the pipe. When the pipe is carry- 
ing but a small part of its capacity the water rushes down the intake end and generates 
a violently turbulent condition just where the pipe is filled. The impact of the 
water rushing down the intake at high velocity causes many air bubbles to be carried 
into the pipe. Obviously, as suggested by Mr. Williams, a pipe at full capacity 
