1903.] Thermal Relations of Energy of Magnetisation. 



235 



derived from purely thermal sources; and the following thermo- 

 dynamic argument* will strengthen this conclusion. 



If the value of the magnetic susceptibility k for any material is a 

 function of the temperature, we can perform a Carnot reversible 

 cycle by moving a small portion (say a sphere) of the substance in the 

 permanent field of a system of magnets supposed held rigidly 

 magnetised by constraints. We can move it into a stronger region H 2 of 

 this field, of varying strength H, maintaining it at the temperature 6 

 by a supply of heat from outside bodies at that temperature ; we can 

 then move it on further, having stopped the supply of heat, until its 

 temperature becomes 6 - 86 • we can move it back again isothermally 

 by aid of a sink of heat at this temperature until the stage Hi is reached, 

 when further progress back adiabatically will restore it to its 

 original condition. If k is a function only of the strength of field and 

 of the temperature, this cycle will be reversible. If E is the heat- 

 energy supplied at temperature 6, and W is the work done on the 

 sphere by external bodies in the cycle, the principle of Carnot gives the 

 relation 



E = W 

 6 86 ' 



Now W = - ^ (JIoH, - JliHi) 86 



(Ik 



= -\-te (Ho 2 - Hi' 2 ) 86, if k is small,! 



Cut/ 



when the cycle is taken such that the change of H along the adiabatic 



* This theoretical deduction of Curie's law lias been already given substantially 

 in ' Phil. Trans.,' A, vol. 190, 1897, p. 287. 



The theory of diamagnetism, which assigns it to modification of conformation 

 in the individual molecule by the inducing field rather than to average spaciai 

 orientation of the crowd of molecules, leads to a non-thermal origin as regards 

 that part. The analogous question {Joe. cit.) as to whether dielectric polarisation 

 is mainly an affair of orientation of unaltered molecules like paramagnetism, or one 

 of polarity due to internal deformation of the molecule like diamagnctism, is now 

 answered by the experiments of J. Curie and Compan (' Comptes Eendus,' June 2, 

 1902). It appears that the dielectric coefficient of glass, for rapid changes, 

 diminishes, but not very quickly, w T ith fall of temperature, and that at temperatures 

 below —70° C. duration of charge ceases to have influence on its value. The 

 electric excitation is thus analogous to diamaguetism and has no thermal bearing, 

 its energy being self-contained in the molecule ; the signs of the susceptibilities in 

 the two cases are different, because the one is of static, the other of kinetic 

 character. The sharpness of the Zeeman magneto-optic effect lias already led 

 (' Aether and Matter,' 1900, p. 351) in this direction, for it requires that the 

 electric polarisation in the molecule shall be of isotropic type, so that there may 

 be no axis of maximum susceptibility. 



t This restriction is not necessary for the final result ; if k is not small, W and 

 E have both to be multiplied by the same factor. 



