244

FLUIDITY AND PLASTICITY

TABLE LXVI.—EVIDENCE FKOM MAXWELL (1866) THAT THE VISCOSITY
OF AIR is INDEPENDENT OF THE PEESSURE

Temperature, degrees Centigrade
Pressure in mercurial inches
Logarithmic decrement of oscillating disks

12.8 12.8 13.3
0.50 5.52 29.00
0.15378 0.15379 0.15398

Warburg and Babo (1882) were the first to prove that the
viscosity of a gas fluctuates widely with the pressure in the
neighborhood of the critical temperature, using carbon dioxide
as their experimental substance. We have already commented
upon the data for this substance recently obtained by Phillips.
Kundt and Warburg (1875) measured the viscosity of carbon
dioxide by the disk method at pressures as low as 0.1 mm of
mercury and they found that the logarithmic decrement of the
amplitude of vibrations became noticeably smaller when the
pressure became less than about 1.5 mm, the distance between
the disks being from 1 to 3 mm. At atmospheric pressure the
molecular mean free path of carbon dioxide at 0° is 0.0000065
cm, and at 2 mm the mean free path is therefore approximately
0.02 mm. Since a considerable portion of the molecules depart
widely from the mean velocity, we should expect the viscosity to
decrease long before the molecular mean free path became equal
to the distance between the boundary surfaces. Kundt and
Warburg believed that the decrease in viscosity due to the in-
creasing length of the mean free path should not occur so long as
the thickness of gas was 14 times the mean free path and they
therefore assumed that at high exhaustions there is "slipping"
at the boundary. No one has yet explained why a molecule of a
rarefied gas is any less likely to give up its translational velocity
than a molecule of gas at ordinary pressures. Whether the
decrease in the viscosity is due to the increase in the free path or
not, the hypothesis of slipping seems improbable, and there may
be some other explanation for the results observed. For example,
in the case of the experiments of Kundt and Warburg with hydro-
gen, the decrease in viscosity at moderately low pressures is,
according to Crookes, " probably due to the presence of a trace of