AMPLIFICATION OF THE LAW OF POISEUILLE 47

calculated velocities and, with very few exceptions, the agree-
ment is within 2 or 3 per cent, and it does not appear that there
is any systematic deviation.

Couette (1890) has strongly confirmed the work of Reynolds
by his measurements with coaxial cylinders. The external
appearance of the apparatus used is shown in Fig. 15 where V
is the outer cylinder of brass which can be rotated at a constant
velocity by means of an electric motor around its axis of figure T.
The inner cylinder is supported by a wire attached at n.
A section through a part of the apparatus in Fig. 16, shows the
inner cylinder s while g and g1 are guard rings to eliminate the
effect of the ends of the cylinder. The torque may be measured
by the forces exerted on the pulley r which are necessary to
hold the cylinder in its zero position. Plotting viscosities as
ordinates and the mean velocities as abscissas, he obtained
Fig. 17. Curve I represents the results for the coaxial cylinders,
curve II represents the same results on five times as large a
scale in order to show better the point where the regime changes.
Curves III and IV are for two different capillary tubes. It is
clear from the figure that the viscosity is quite constant up to
the point where the regime changes. The apparent viscosity
then increases very rapidly, and finally becomes a linear func-
tion of the velocity. The dotted parts of the curves where the
viscosity increases most rapidly, represents the region of the
mixed regime, and the measurements were very difficult to ob-
tain with precision.

He proved that pRItp — a constant by a series of experiments.

(1) The mean velocity at the lower limit of the oscillations is
independent of the length of the tube. He used a glass tube
R = 0.1778 and obtained the efflux per minute V, thus:

TABLE XV.—LAW OF LENGTHS

Length, centimeters
V mean

86.5
388

71.5
367

57.9
365

41.8
376

25.7
394