ARTICLES 589 



investigated over a temperature interval extending from 

 — 1 8o° C. to 200 C. For details concerning the method of experi- 

 ment and the elimination of errors, due to disturbing sources, 

 reference is made to Part 1. In 1895, Curie investigated the 

 magnetic properties of a large number of substances. He 

 found that the specific susceptibility (i.e. the susceptibility 

 per gram) of paramagnetic substances varied inversely as the 

 absolute temperature, while that of diamagnetic substances 

 (with the exception of bismuth) remained constant as the tem- 

 perature varied. In 1905, Langevin developed an electron 

 theory of magnetism for gases, which accounted theoretically 

 for the results observed by Curie. On this theory, the sum of 

 the magnetic moments of the electron orbits in a molecule is 

 zero, when the substance is diamagnetic, but is different from 

 zero in a paramagnetic substance. Weiss extended Langevin's 

 theory to account for ferro-magnetism, where the molecules 

 exert intense mutual action on one another. In the present 

 experiments it was found that the Curie law of constancy of 

 diamagnetism held very approximately for most of the sub- 

 stances, to within the limit of experimental error, providing no 

 appreciable change of molecular aggregation with variation of 

 temperature took place. A sudden change of molecular con- 

 stitution, however, such as occurs on crystallisation, might be 

 expected to affect the magnetic property appreciably, the 

 molecules becoming distorted by mutual forces. In confirma- 

 tion of this view, it was found that a definite change of specific 

 susceptibility actually did take place on crystallisation, the 

 change amounting to a few per cent, in the case of aromatic 

 compounds. Aliphatic substances showed a smaller effect, 

 and in some cases the change was inappreciable. The small- 

 ness of the effect presented by such substances is considered 

 below (p. 592). If the substance did not crystallise, but passed 

 into a supercooled liquid or jelly state, the change of specific 

 susceptibility was too small to be detected. In the case of 

 benzophenone, by cooling the liquid carefully, the substance 

 passed into a vitreous mass with no change of susceptibility. 

 After cooling to — 20°C. (i.e. 70 degrees below the m.p.), crys- 

 tallisation set in and the crystals were found to be less dia- 

 magnetic than the liquid or vitreous mass. On heating the 

 crystals, their susceptibility remained practically constant, 

 until the fusion-point was reached, where the susceptibility 

 increased to the normal value possessed by the liquid. Thus 

 a definite hysteresis loop with respect to temperature was 

 obtained with this diamagnetic substance. The loop was 

 similar to those presented by nickel steels, 1 the critical tem- 



1 J. H. Hopkinson, Proc. Roy. Soc, A, vol. xlviii, p. 1, 1890. Other 

 curves than the one reproduced are given. 



