512 
MESSES. ELEEMING- JEN KIN AND J. A. EWING- ON FKICTION 
The breadth of the notch is considerably greater than the diameter of the axle. Thus 
when the disk is caused to revolve in the direction of the arrow in fig. 1 the axle rolls 
along the bottom of the notch until it reaches the vertical side on the right hand, and 
then any subsequent motion can only occur by the sliding of the circumference of the 
axle upon the bottom and side of each bearing. To prevent the disk from sliding 
laterally in the event of the bearings ceasing to be exactly level, steel end-plates ( c c) 
are provided which can be adjusted by means of screws so as just to touch the ends of 
the spindle without sensible pressure, and these are slightly rounded at the ends so that 
their points of contact with the end-plates may lie as nearly as possible-in the geometrical 
axis of the axle. The end-plate is omitted for the sake of clearness in fig. 1, but it is 
shown in fig. 3, displaced, however, from its proper vertical position in order to allow 
the end of the axle and the bearing to be fully seen. There was no difficulty in adjusting 
the end-plates so that their pressure on the axle should cause an infinitesimally small 
retardation of the motion of the disk, in comparison with that due to the friction of the 
axle upon the bottom and side of its bearings. Owing to the great weight of the disk 
this friction was so considerable that the resistance of the air could safely be neglected 
so long as the velocity of rotation of the disk did not greatly exceed the greatest velo- 
city which occurred in our experiments. When, therefore, the disk was caused to revolve 
by the temporary application of any force, and was then left to itself, it gradually came 
to rest in virtue of one cause only — the friction of the axle upon its bearings. By 
observing the rate of (negative) acceleration of the disk throughout its motion the value 
of the friction could be determined for all velocities of the rubbing surfaces from the 
greatest or initial velocity down to the lowest velocity at which the acceleration could 
be measured. 
In order to determine the rate of retardation of the disk at all times throughout its 
revolution it was necessary to devise some means of recording with great exactness the 
angular distances moved through by it during successive short intervals of time. This 
was effected by recording the linear distances moved through by its circumference 
during successive semiperiods of oscillation of a short pendulum. To obtain a perma- 
nent record of these spaces without introducing any new source of retardation whatever, 
such as would be introduced if a pencil or brush were caused to press either continuously 
or intermittently against any part of the moving disk, we adopted the method of recording 
which Sir William Thomson invented for the purpose of registering the arrival of elec- 
trical impulses through long submarine cables, and which has found practical application 
in his siphon recorder. The recording apparatus is shown on the right-hand side in 
fig. 1. A pendulum (C) is supported on a horizontal knife-edge ( d ) attached to a fixed 
stand (D), and is capable of oscillating in a plane perpendicular to that of the disk, its 
position of rest being directly opposite the middle of the circumference of the disk. 
The knife-edge ( d ) enters a hollow cylinder on the top of the pendulum-rod, and this 
hollow cylinder is of greater internal diameter at the centre than at the ends, so that 
the knife-edge bears against it only at the ends. The pendulum is therefore free to 
