q 
: 
: 
| 
TRANSACTIONS OF SECTION A. 391 
Independently of all hypothesis we may define what I call the ‘ specific tempera- 
ture’ of a radiating body for the wave-length A to be the temperature T,, at which the 
emissivity ¢,r, of the black body equals the fraction = for the radiating body. 
» 
(Mr. E. Bauer defined the same quantity and called it ‘ temperature of emission.’) The 
fraction ee would give us a scale for the degree of luminescence. 
2. Resonance Spectra under High Dispersion. 
By Professor R. W. Woop. 
3. A Theory of Magnets. By Professor S. B. McLaren. 
It is my object to recall some difficulties of magnetic theory and to suggest 
how they may be escaped. 
The history of magnetic science divides into an ancient and a modern period, 
the times before and after Ampére. In the earlier period the fundamental 
idea used is that of a magnetic substance; after Ampére this idea disappears. 
The magnet is now regarded as a whirl of electric particles or electric fluid. 
In modern electro-magnetic theory all substance is electric; with Lorentz, for 
example, matter is the electric fluid. 
Thus Ampére may claim to have given what previously did not exist, a 
theory of magnets. Before him the existence of molecular magnets was the 
starting-point; any explanation of magnetic phenomena, as, for example, 
Poisson’s account of magnetic induction, has to begin with matter whose 
elements are already magnetised. Poisson, I may remind the reader, can only 
account for paramagnetism; of diamagnetic phenomena there is no obvious 
explanation. I wish to point out that the modern electro-magnetic theory has 
its own difficulties. It cannot take over unmodified Poisson’s way of viewing 
the phenomena of magnetic induction. 
It is true that the magnetic field due to a magnet is the same as that due 
to a rotating electric charge; it is true also that the resultant force exerted by 
the field on the charge is the same as that which it exerts upon the magnet. 
But it by no means follows that this force produces the same results in the two 
cases. And one fact sufficiently establishes a difference. A magnetic field does 
work upon the magnet in changing the direction of its axis, or the magnet in 
that field has potential energy; on the other hand, the same field can do no 
work by changing the axis of rotation of the electric charge, because the force 
on each element is at right angles to its motion. Take in particular a spherical 
distribution of charge rotating about any diameter. Set up a steady magnetic 
field, the only effect is to superimpose upon the original rotation about an axis 
fixed in the charge another rotation about the magnetic lines of force, with an 
angular velocity simply proportional to thaf force. This second motion accounts 
for diamagnetism, but there is no tendency for the axis of rotation fixed in the 
charge to approach the magnetic lines. The diamagnetic field is independent, as 
it ought to be, of the temperature, but no paramagnetic field is created. 
The difficulty has been remarked by Lorentz, Voigt, J. J. Thomson, N. Bohr, 
and, I doubt not, by others. It is not, however, generally recognised that in 
such a theory of magnetic induction as Langevin’s we are back at the ancient 
postulate of magnetic substance. 
The difficulty may, I think, be evaded by moving still further from the old 
point of view as well as by returning to it. We may give up not only magnetic 
substance, but electric substance as well. 
I assume the electro-magnetic field defined by two vectors E and H. This 
field is not all space; it is bounded by closed surfaces within which E and H 
do not exist. The space within these is ‘matter,’ without them ‘ther.’ All 
formule are to be deduced from the principle of least action. Following Larmor, 
suppose that the action, in so far as it involves E and H, is identified with the 
expression 
(Sm)! fff f (H? — E%) dv dt, 
