26 FLUIDITY AND PLASTICITY

but since from Eq. (9a) we have that

and therefore

87T

nR\

•fT

This is the equation of a straight line. If values of ~nrrT

5.—Finding the value of
n for the "end correction."

plotted as ordinates and those of R/l as abscissas, the intercept
on the axis of ordinates will give the corrected viscosity, i.e., the
value of the viscosity when I = co or R = 0; and the tangent of
the angle made by the line with the axis of abscissas when divided
by the viscosity will give the factor
n required. Figure 5, taken from
Knibbs' work, illustrates the method
as applied to the tubes used by Poi-
seuille B to Bv and F to FIV. The
values of n are found to be —5.2 and
+ 11.2 respectively. According to
Knibbs "these results challenge the
propriety of Couette's statement that
A may be always regarded as positive
and taken as nearly three times the
diameter of the tube." In order to adequately test the question
Knibbs took the whole series of Poiseuille's experiments at 10°
and reduced them rigorously on the basis of Eq. (8) taking into
account the peculiarities of the bore of the tubes used by Poiseuille
as indicated in his data. Whenever possible the value of pt (qf.
Eq. (9)) was obtained by extrapolation since then the correction
term vanishes; in the other cases marked with a star, the value
of m was taken as 1.12. The results are arranged according to
increasing values of R/l, since if n has a positive value there
should be a progressive increase in the values of the viscosity.
Rejecting the last four values as uncertain, the general mean is
0,013107 which is almost identical with the mean for each group
of eight, whereas if n had a constant value there should be a
steady progression. On the other hand the values for the vis-
cosity for the B series of tubes increase while those for the F series
decrease as we go down the Table. It appears therefore that no
general value can be assigned to n unless it be zero.