86 
Proceedings of the Royal Society 
would be the angle rotated whilst the light traverses the glass rod. 
This is on the assumption that the ether within the rod is likewise 
rotated. In the magnetic experiment it is easy to produce a 
rotation of 1° in a piece of glass three inches long. Light takes 
T.Tnrff, trmr , 777777 a seco,n 6 to traverse this distance. Hence, to 
produce an effect equal to the magnetic effect, the glass rod would 
require to be rotated 10,000,000 times in a second. We cannot 
determine with great precision the plane of polarisation of a ray of 
light, hence we cannot measure any rotation of the plane of polari- 
sation which might be thus produced. 
In the same manner, if we suppose the molecules of glass and 
the accompanying ether to be rotated round the lines of magnetic 
force, and in the direction of the positive current producing the 
given magnetic field, then the phenomena observed by Faraday 
would be explained ; and we should be able to determine the 
number of rotations per second induced in any specimen of glass 
wiih a given intensity of magnetic field. 
So soon as the electro-magnet is demagnetised, the rotation of 
the molecules ceases. It seems, then, that there is a friction 
among the molecules tending to stop the rotation. Hence we 
should be led to conclude that the energy of the magnet is gradu- 
ally used up by the presence of the piece of glass. 
Assuming, then, that there is a friction among the molecules 
of glass, it follows that when the electro-magnet is magnetised the 
rod of glass has a tendency to turn bodily round an axis through 
its centre of gravity parallel to the lines of force ; and if the rod 
of glass were free it would turn round this axis. 
In the winters 1872-3 and 1873-4, I made a number of 
experiments to put this hypothesis to the test. The general idea 
of the experiments was this. A rod of glass was suspended by a 
fine skein of silk fibres between two poles of an electro-magnet, 
one pole above, the other below. A small mirror was fixed on the 
rod, and a lamp and scale arrangement was mounted for measuring 
rotations. Readings were taken when there was no current, and 
also in the two positions of the commutator. 
The result of these experiments seems to be that there is an 
effect of the kind anticipated. Sometimes, it is true, a deflection 
was produced by a diamagnetic repulsion, owing to a want of 
