576 Prof. 0. W. Bichardson on a Theory of the 



B in its neighbourhood. We can obtain an upper limit to 

 the disturbance, which will probably be very much greater 

 than the actual effect, by considering the atoms B as perfectly 

 conducting spheres of atomic dimensions. In this case, when- 

 ever an atom B moves to a place where the strength of the 

 magnetic field is H, it will have circular electric currents 

 induced in it which will make the normal magnetic force 

 vanish at the surface. If the radius of B is b, it can be shown 

 to behave like a magnet of moment M=— ^H6 3 . If the 

 atom of magnetic moment A is placed at the origin and the 

 coordinates of B are a, y, z, the components of the induced 

 doublet M z M y M r at B are 



2 -das^r)' 2 B^W' 2 &^W' 



These give rise to a potential at the origin of amount 



The change in the magnetic force at the origin due to the 



atom B is therefore = — -^A and the effect H / due to the 



0% 



whole of the atoms which occur 



-m^-'^ 



The displacement of the spectral lines will be that due to 

 the Zeeman effect produced by a field PF. If H / is measured 

 in electromagnetic units it is easy to show that 



b\= -j. -Ao = — — 3 — A ~A, 



4.itc m oc a m 



where hX is the displacement, and X the original wave- 

 length. Putting v (the number of atoms per c.c. at atmo- 

 spheric pressure at the temperature of the arc) =4xl0 18 , 

 h/a=£ 9 'e/ma=l'& X 10 7 , X =4 x 10 ~ 5 cms., and A = 2 x 10 - 20 , 

 we find B\'X = S x 10 ~ u per atmosphere. As all the observed 

 displacements lie between 5000 and 25,000 times this value, 

 it seems very unlikely that magnetic perturbations are a 

 factor requiring serious consideration. It might be thought 

 that a larger magnetic effect might be obtained by treating 

 the atoms as permanent magnets, the moment of each being 

 of the same order as that for an atom of iron. A comparison 

 of the magnetic potential energy with the translational 



