148 FLUIDITY AND PLASTICITY
ment of the kinetic theory of gases, it became evident that
viscous resistance arises from the diffusion of the particles
of high translational velocity into layers whose translational
velocity is lower, and vice versa. According to this explanation,
the loss of translational velocity must increase with the tempera-
ture, which accords with the fact that the fluidity of a gas
decreases as the temperature is raised.
But in liquids the fluidity increases with the temperature
and it is generally agreed that there is a second cause of viscous
resistance, which, without any very good reason in its favor,
has been repeatedly attributed to the attraction between the
molecules. According to Batschinski1" If we think of two parallel
layers of liquid as of two rows of men, the men moving in place
of molecules, we must assume that these men take hold of their
nearest neighbors and hold them for a time." This explanation is
however inadequate, for a particle A} coming within the range
of attraction of a particle B in an adjacent layer supposed to be
possessed of slightly less translational velocity, will be accel-
erated and only after passing B will the retardation take place.
Apparently the two actions exactly neutralize each other, or if
they do not there must result a destruction of energy in violation
of the first law of thermodynamics. No reasonable hypothesis
. has been proposed to extricate us from this dilemma, on the
basis of cohesion, hence, we are forced to look for some other
cause. Whatever -the explanation, it must show how transla-
tional or ordered motion is being continuously transformed
into heat or disordered motion.
To get a clearer idea of the nature of the two causes of viscous
resistance, we imagine two parallel planes A and B, the former
moving to the right parallel to itself in respect to the second
plane, which for convenience only may be assumed to be at rest.
We will first assume that between the planes there is a highly
rarefied gas. If the walls are smooth and unyielding and the
particles of gas perfectly elastic spheres, we will not have a
model of viscous flow; for as the particles collide with the sur-
faces, the angle of rebound will be equal to the angle of incidence,
there will be no translational velocity transmitted to or from the
walls and the so-called "slipping", would be perfect. In order
* (1913) p. 643.