131 
the tube with the windows now appears to be 549.1. In the other 
tube it is 558.0 em./sec. The mean value for the whole tube be- 
comes 553,6 (cocks B and D open). The result that in the two 
eases, when the cocks A and (' are open, and, when the cocks 2B 
and D are open, the velocities differ, may appear less startling when 
it is considered that there is a small dissymmetry in the supply 
tubes of the apparatus and that with A open the water before 
discharge undergoes a greater change of direction than with D open. 
A new proof for the change of velocity at the axis of the tube 
is given by ascertaining the velocity distribution over the cross section 
of the tube. In A, the velocity distribution is entirely different from 
that encountered for example in 5’, that is to say before and after 
the stream traversed the horizontal part. 
When the cocks A and C are open the curve traced in A, is 
of a parabolic character, in B’ the central part of the curve is of 
smaller curvature, corresponding to a smaller velocity at the axis. 
The velocity distribution in a vertical line passing through A, is 
represented in Fig. 4. The curve is of a parabolic character, and 
nearly, but not quite, symmetrical. 
DISTANCE To BOTTOM or TUBE 
Taking the mean abscissae for points at equal distances above 
and below the axis and constructing a curve with these points we 
may determine the volume enclosed by the surface of revolution, 
originating when the constructed curve revolves about the axis. 
This enables us to determine the mean velocity, which is found to 
be 468 em./sec., whereas on a former occasion (Comm. II) using 
the total quantity of water passing in a given time we obtained 
465 cm./sec. 
It is worth while to state that the distribution of velocity would 
according to Poisruinixn’s law give a parabola of far smaller width 
than the curve of Fig. 4, if the maximum velocity should be that 
in the figure. 
ge 
