CONSTITUTION AND TEMPERATURE ON MAGNETIC SUSCEPTIBILITY. 273 



different forces of crystallization in different directions (which forces determine the 

 planes of cleavage) with the magnetic behaviour of the crystallized medium and lead 

 us to suspect that the forces of cohesion are probably of magnetic nature. The fine 

 points are so completely explained by the magnetic deportment that it is difficult to 

 dissociate the crystalline forces from a magnetic origin. If we assume that these 

 forces are of an electrostatic nature, then it must be admitted that the electrostatic 

 axis of the molecule must coincide with the magnetic axis if the action of a magnetic 

 field is to be decisive, as TYNDALL proved it to be, in isolating the planes of cleavage. 

 But if the electrostatic and magnetic symmetries of the molecules are coincident the 

 application of a field of either nature should induce a double refraction of the same 

 kind in a given liquid. This, however, is not true experimentally, the electric induced 

 double refraction in liquid carbon bisulphide being opposite in sign to the magnetic 

 induced double refraction.* Moreover, in crystalline media, the greatest axes of the 

 ellipsoids representing the magnetic and electric properties of the molecule do not in 

 general coincide. We may therefore say that the evidence points to the conclusion 

 that the forcive which holds the molecules together in a crystalline space lattice is 

 magnetic in nature and not electrostatic, t 



DRUDE,| in his experiments on the relation between valency and dispersion, 



* COTTON and MOUTON, 'Comptes Rendus,' vol. 155, p. 12:ft, December, 1912. 



t [Note culded April 26, 1919. After the present communication had passed out of my hands, an 

 important paper " On the Origin of Spectral Series" was published by Sir J. J. THOMSON (' Phil. Mag.,' 

 April, 1919). In this a new theory of atomic structure is suggested in which the atomic nucleus and the 

 revolving electrons play similar roles to those described on p. 274. AVithin the contour of the atom, 

 according to Prof. THOMSON, the electrostatic forcive due to the nucleus is of a periodic character and 

 determines a series of spherical or approximately spherical surfaces where the electric force vanishes and 

 over which the .periodic motion of the boundary electrons is determined solely by the magnetic field of the 

 atom. This magnetic field is supposed to be radial. If this is the case, these intra-atomic fields must be 



of the order of magnitude 10 s gauss / as a simple calculation shows, since v = ) to account for the 



frequencies of the visible spectrum. Still larger intra-atomic fields will exist nearer to the nucleus, of the 

 order 10 gauss. These will be sufficient to account for the frequencies of the K series. The infra-red 

 series will be accounted for by fields of the order 10 7 gauss. But this latter value is of the order of the 

 intermolecular magnetic field which has been deduced independently in various ways in the present 

 researches. Moreover, it is to this local field that we have ascribed the rigidity and other properties of 

 crystalline media in general. The frequencies of the infra-red series will, on this view, correspond with 

 the elastic vibrations of the rigid medium in conformity with the quantum theory of specific heats of 

 EINSTEIN and DEBYE as already stated (see Part III, p. 94, and supra, p. 259). Reasons have already been 

 given for assigning a magnetic nature to the intermolecular field in crystalline media (see Part III., 

 pp. 101-3, and supra,, pp. 270-276). This intermolecular magnetic field, which is of the order 10 7 gauss, is 

 suggestive in connexion with Prof. THOMSON'S theory, referred to above. On p. 274 (footnote) it was 

 suggested that the forces determining crystalline cohesion are magnetic in nature, the symmetry of the 

 magnetic forces being determined, however, by the electrostatic action of the nucleus. Therefore, in 

 this fundamental sense, the present theory and that of Sir J. J. THOMSON are identical.] 

 t ' Ann. der Phys.,' vol. 14, p. 677 and p. 936, 1904. 



2 P 2 



