348 Prof. Trouton and Mr. Andrews on the 
applying the constant torque (see fig. 1), The shaft carried 
a square socket for the purpose of gripping the squared end 
of the cylinder of the substance, which was made to fit 
exactly. A similar but fixed socket prevented the other end 
from turning. The rate of rotation was observed by means 
of a divided circular disk carried on the shaft. With this 
apparatus the experiments described below were undertaken 
Bie. 1. 

to test whether (1) the rate of rotation was proportional to the 
torque; (2) the rate of rotation of cylinders of the same 
material was inversely as the fourth power of the radius. 
Incidentally two unsuspected effects were at once disclosed 
by the use of this apparatus. One is that the coefficient of 
viscosity of bodies such as pitch is a function of the time, 
observations showing that the velocity of flow for a given 
stress diminishes with time from its initial value down to 
a constant quantity. The second is that on removing the 
stress there is a flow back in the opposite direction, which 
gradually diminishes to zero with time. 
The method lends itself also to the determination of the 
coefficient of viscosity at different temperatures, as the cylinder 
can be conveniently surrounded by a jacket and kept at any 
required Feuer ih This mainly arises from its not being 
necessary to have access to the bar while under observation. 
In this way the coefficient for soda-glass was determined at 
temperatures ranging between 500° C. and 700° C., and that 
of pitch from 0° C, to 15° C, 
