CONSTITUTION AND TEMPERATURE ON MAGNETIC SUSCEPTIBILITY. 281 



Thus iron-carbonyl (Fe (CO) 5 ) and nickel-carbonyl (Ni (CO) 4 ) are diamagnetic* ; 

 potassium ferri-cyauide is paramagnetic, while potassium ferro-cyanide is diamag- 

 netic. t It would appear that in these cases the loss of magnetic property of the iron 

 and nickel atoms is due to a transfer of valency electrons, i.e., it involves an electric 

 charging up of the atoms. The behaviour of the oxygen atom in organic compounds, 

 in compounds with chlorine and in metallic oxides, where it acts always as an 

 electronegative element, may possibly be accounted for in the same way. Free oxygen 

 and ozone are strongly paramagnetic, but no semblance of an additive nature of the 

 magnetic property is found in any of the oxygen compounds. 



The appearance of strong magnetism in the Heusler alloys and its disappearance in 

 manganese steels, are similar effects, dependent on the formation of intermetallic 

 chemical compounds accompanied by an electron transfer. 



There are many paramagnetic substances which possess molecular magnetic 

 moments comparable with, and in some cases much superior to, those shown by ferro- 

 magnetic substances. The apparent feeble susceptibility they possess is due to the 

 fact that, with the largest magnetic field which can be applied, we can never produce 

 anything like a saturation effect. In fact, according to WEISS, the molecule of cobalt 

 chloride, Co.CLj contains 25 magnetons, while an atom of cobalt below the critical 

 temperature contains 9 magnetons. We may well ask ourselves what is the nature 

 of the process by which the addition of a diamagnetic substance, H.C1 to cobalt, 

 produces such a large increase in the number of magnetons per molecule.J Assuming 

 the work of WEISS holds good, and there is certainly a very considerable amount of 

 evidence in favour of his theory, we can interpret this result either by supposing that 

 the atom of cobalt really contains more than 9 magnetons, or else that the 

 diamagnetic acid supplies the additional magnetons when it acts on the cobalt to form 

 the chloride. In either case, it seems that we must admit that a molecule may 

 possess systems of magnetons which, in certain circumstances, are so arranged to 

 counterbalance one another, producing no additional moment of the molecule. || These 

 magnetons would contribute nothing to the paramagnetic or ferro-magnetic property 

 of a substance and could not be included in WEISS'S theory. The grouping of these 

 " latent " magnetons, according to our extended view, would be perturbed by the 

 union of the cobalt atom with the Cl ion, in a manner similar to that by which an 

 external field reveals the spontaneous magnetization in iron (as interpreted on EWING'S 

 theory) by orientating groups which were formerly so constituted as to show no 

 magnetic effect externally. If this is so, then we are only justified in assuming that 

 purely diamagnetic molecules contain groups of magnetons so arranged that the 



* A. E. OXLEY, ' Proc. Camb. Phil. Soc.,' vol. 16, p. 102, 1911. 



t J. S. TOWNSEND, 'Phil. Trans. Roy. Soc.,' A, vol. 187, p. 547, 1896. 



I The specific susceptibility of cobalt chloride is 90 x 10~ 6 , that of hydrochloric acid - 0'80 x 10~ 8 . 



Some of which are self-compensated, 



|| I.e., of the molecule as a whole. 



2 Q 2 



