354 BELL SYSTEM TECHNICAL JOURNAL 



and o-Q, the susceptibility of the material, should vary inversely as 

 the absolute temperature. This, as Curie found, is true for the 

 paramagnetic gases. It is true also for a number of salts in dilute 

 solutions, and even for a certain number of solid substances, although 

 for these the underlying assumptions would scarcely be expected to 

 remain valid; one has the feeling that the data are left floating in 

 the air by the withdrawal of the logical basis for the theory with which 

 they agree. 



Suppose nevertheless that the theory remains valid; then, for any 

 substance of which the susceptibility o-q varies inversely as T, one 

 can calculate the moment M of its molecular magnets from (13); 

 for ^ is a known constant, and TV is knowable at least when one is 

 dealing with a gas of known density or a solution of known concen- 

 tration (with solids there may be doubt as to the number of atoms 

 grouped together to form an "elementary magnet"). Multitudes of 

 such values have been computed; their orders-of-magnitude are 10~^^ 

 to 10~^**. Commonly they are expressed as multiples of a certain 

 unit, the "Weiss magneton," which is equal to 1126/A7^o or about 

 1.858- 10~2^ Many of them are nearly integer multiples of this unit.* 



On taking any observed value of M, and multiplying it by the 

 corresponding value of N to obtain the "theoretical" value of /max. 

 for the substance in question, we find that as a rule this last is so 

 much larger than the highest value of / attained with practicable 

 fields that there is no contradiction between the theory and the fact 

 that I is sensibly proportional to H all through the feasible range of 

 fieldstrengths. There is only one substance (gadolinium sulphate) 

 for which /max. can be approached and this only at extremely low 

 temperatures, below 5° absolute; in Fig. 12 the data are displayed; 

 it is evident that the Langevin curve, drawn with the initial slope 

 best suiting the points near the origin, fits fairly well to all the other 

 points. 



I pass now to the assumption whereby Weiss so extended Langevin's 

 theory that it became competent to describe not only these simplest 

 cases of paramagnetism in which l/o" is proportional to T, but also 

 the much more numerous cases of paramagnetic substances conforming 

 to a more general law, and certain aspects of ferromagnetism also. 



Formally the extension amounts to this, that in the expression for 

 the parameter a which figures in equation (9), the fieldstrength H is 

 replaced by a linear function of // and /: 



* To enter into the long and fiercely debated questions about the meaning and 

 even the reality of the Weiss magneton would lead me too far afield; but it is so 

 frequently used as a unit in stating data of experiment that one must know at 

 least its value. 



