j6 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 65 



5. Diamagnetic compounds, such as NaCl, HC1, H 2 0, etc., do not 

 show noticeable magnetic changes as the temperature or physical 

 state is changed. This also would be expected from the magneton 

 theory, .because the simplest possible molecules of these substances 

 contain no free magnetons, and are essentially diamagnetic: hence 

 polymerization or solidification, which it should be observed is 

 brought about, in these cases by the electrostatic forces mentioned in 

 §12 rather than by magnetic forces, cannot appreciably affect the 

 magnetic susceptibility. 



To summarize : As the changes, 



Complex molecule — > Simple molecule -> Atom, 

 take place, from whatever cause, we may expect, with the qualifica- 

 tions already noted, that diamagnetism will give way to paramagnet- 

 ism. Gaseous dissociations are the cases where new evidence is 

 most urgently needed — and where it is most difficult to get. 



This collected evidence seems conclusive for para- and dia- 

 magnetic substances, but it is important to observe that we are driven 

 to exactly the opposite conclusion in the case of ferromagnetism. 

 Here it seems that it is easier to obtain a system with a large magnetic 

 moment that is made up of constituents drawn from two or more 

 atoms than to obtain such a system within a single atom. The 

 conclusive evidence on this point is the behavior of the Heusler and 

 similar alloys : in these, as has frequently been pointed out, the 

 ferromagnetic units must be groups of several atoms ; it is very 

 likely, then, that the same is true for ferromagnetic elements like 

 Iron. The way in which these complexes are built up is not at all 

 indicated by the magneton theory up to the present ; but see §19. 



§23. Weiss' Magneton, and Quantitative Relations 



With regard to a comparison of the results of the theory here 

 described with Weiss' work on " the magneton," I will first quote a 

 few sentences (translated) from the conclusion of a summary of his 

 work that appeared in the Journal de Physique, [5] 1, 900, 965, 191 1. 

 These should be compared with the passages already quoted from 

 Langevin (§2). 



" What is the role of magnetic phenomena in chemical combina- 

 tion? Are chemical forces magnetic in nature? Are the valences, 

 indeed, referable in some way to magnetons ? " In the same paper he 

 mentions the possibility that his magneton is the same as the unit 

 magnet postulated by Ritz in the latter's theory of spectrum series. 



Notwithstanding these suggestive passages, Weiss' magneton is 

 not in any way identified with the electron, but is an empirical quan- 



