VISCOSITY AND FLUIDITY 101
appearance of the drops of an emulsion as they pass through a
capillary tube. Due to the friction against the walls, the rear
end of each drop is flattened and the front end is unusually convex.
It is to be especially noted that when the drops are small in
diameter as compared with the diameter of the tube and yet
large enough to occupy the whole cross-section of the tube,
the motion of the liquid is by no means entirely linear, being
transverse as well as horizontal as indicated by the arrows. The
effect of this transverse motion is to increase the apparent
viscosity of the liquid. If, however, the drops are very large in
comparison to the diameter of the tube, the importance of this
transverse motion may become vanishingly small. Thus if the
drops of an emulsion are large enough to fill the cross-section of a
tube, the viscosity, as measured by the rate of efflux, will be at
least as great as the sum of the component viscosities, but it
may be greater due to the transverse motions. We grant that
below the critical-solution temperature a part of the increase in
viscosity may be due to these transverse motions, but Bose
would seem to account for all of the abnormal increase in the
viscosity in this way. This however is not warranted, for the
reason that at the center of the capillary the liquid has normally
a high velocity while at the boundary the velocity is zero, so that
there is a considerable tendency for any drops to become dis-
rupted and drawn out into long threads. It is impossible to
believe that above the critical-solution temperature the surface
tension of the "drops" is sufficient to prevent disruption, for
we are accustomed to think that the surface tension at the critical
temperature is zero, and the abnormality in the fluidity is a
maximum at this temperature. We conclude therefore that
neither the explanation of Scarpa nor of Bose is sufficient, but
that the explanation based upon the nature of viscous flow in
a heterogeneous mixture is both necessary and sufficient.
The theory requires that if the fluidities of the two components
of the mixture are identical, it makes no difference whether
fluidities or viscosities be considered additive; hence there should
be no irregularity in the fluidity curves of such a pair of sub-
stances even in the vicinity of the critical-solution temperature.
No case has been examined, so far as we know, in which
the components have approximately the same fluidity and