OF TWO CONFINED STREAMS OF WATER. 801 
opening through the wall (,),-inch diameter) was bored with a hot wire, and all 
roughness at its junction with the interior surface of the passage was afterwards 
carefully removed. Brass nozzles gradually changing in section from a circular to 
rectangular form were fitted to the two inlets and to the outlet. The primary 
stream (1) was supplied through a Venturi meter having a 3-inch body and a 1-inch 
throat. This was calibrated immediately before, and checked at intervals during, 
the experiments. The total discharge was caught and measured in a calibrated 
tank, and the volume and velocity of the impinging stream (2) were then com- 
puted by differences. ‘The velocities adopted in the experiments ranged from 6 fis. 
to 22 fis. 
Friction Loss.—The friction loss in each pipe was measured after carrying out the 
impact tests, by blocking up the side opening with plasticine carefully trimmed off to 
shape, and by measuring the drop in pressure between points (1) and (8) at different 
velocities. As the surface area of this plasticine is comparatively small, and as its 
roughness does not differ appreciably from that of the varnished surface of the wood, 
no special allowance has been made for its presence. From these results a friction- 
velocity curve was plotted showing the friction loss per foot run of the pipe, and the 
loss between each piezometer opening and the junction was then calculated for each 
experiment and deducted from the total loss to give the impact loss. 
The friction loss varied slightly in different channels of the same cross-section, but 
the mean of the experiments gives the result, for all sizes of channel :— 
ils 
h feet of water, 
2gqm\ 
where /, is the friction loss. 
/ ,, length of channel in feet. 
m ,, hydraulic mean depth in feet. 
v,, velocity in feet per second. 
f= 0036. 
3. Loss at ELBows. 
A general expression for the loss at impact in terms of the two velocities must 
be true when v, is zero. ‘This corresponds to flow round an elbow having a dead 
end in the line of the final direction of flow, and preliminary cxperiments were 
earried out to determine the losses in such cases. Within the range of velocities 
adopted (6 fis. to 22 fis.), the excess loss of head over that in the same length 
of straight pipe is sensibly proportional to the square of the velocity of the initial 
V2" 
oe feet of water, where F has the mean values given in the 
g . 
jet (2), and is given by F 
following table :— 
