THE VISCOSITY OF GASES

245

foreign gas rnost likely water/'which seems to have been sus-
pected by jCxindt an(j Warburg themselves.
Crookes (1881) measured the logarithmic decrement of
a mica disk swinging in a glass bulb and supported by a glass
fiber, using pressures as low as could be measured, by means of a
McLeod gage. The gases employed were air, oxygen, nitrogen,
carbon dioxide, carbon monoxide, and hydrogen at 15°C. In
the case of hydrogen the logarithmic decrement was found to be
almost perfectly constant from atmospheric pressure down to
0.25 mm. A-t about this pressure the viscosity of all gases
decreases rather suddenly. With other gases there is a slow
decrease with the pressure even from atmospheric pressure,
except in a sample of air which contained some water vapor,
in which case the logarithmic decrement was at first that of
air, but at about 50 mm it decreased rapidly to that of pure
hydrogen. In an absolute vacuum we must assume that the
fluidity is infinite, hence Maxwell's law must break down at very
low pressures.
According "to the data of Phillips, Fig. 54, we should expect
that Maxwell's law would break down at low temperatures
or at very high temperatures. There is a curious dearth of data
with which to test out this point. However, a hydrocarbon
vapor, "kerosoline," was measured by Crookes and the viscosity
was found to decrease rapidly from the highest pressure obtained
of 82.5 mm down to 8 mm. Lothar Meyer found in experiment-
ing with benzene that the viscosity of the saturated vapor was
smaller the higher the back pressure at the exit end of the capil-
lary tube. At- high temperatures we are led to expect that just
the opposite conduct will be observed, viz., that the viscosity will
decrease as the pressure is increased, see Fig. 54, but there is so
far as known to the author no data to support this conclusion.
VISCOSITY OF GASES AND TEMPERATURE
From the formula
17.= 1/3PFL
it is evident tttat the effect of an increase in temperature will -be
to increase the mean velocity, but it is not known what effect
the tempera.ttrre may have upon the mean free path, although it