NO. II STRUCTURE OF THE ATOM PARSON 73 



In comparing these values two thing's should be borne in mind : 

 first, that susceptibilities of paramagnetic substances in solution are 

 in general greater than for the undissolved substances (Pascal's 

 values under (P) are for solutions) ; and second, the values obtained 

 for diamagnetic salts have little significance for the metal atoms that 

 are in them except to show that they are not decidedly paramagnetic, 

 for the acid radicles themselves are diamagnetic. 



In the table I have given space to the compounds of the transition 

 metals of groups V, VI, VII, VIII, I B, II B only, because there the 

 relations are much more complex. We may notice first that the 

 free elements V, Cr, Mn, are much less paramagnetic than their 

 salts. This, I think, is due to the same cause as the general rule that 

 these salts are more paramagnetic when hydrated than when anhy- 

 drous (F), and still more so in solution. If, as this theory would 

 indicate, paramagnetism is an effect of free magnetons in the metal 

 atom, then the farther removed from one another these atoms are, 

 the more free from constraint must be their magnetons (as many 

 as remain free). In metallic Chromium, for example, the six val- 

 ence magnetons are used to a considerable extent, as the high 

 melting point indicates, in binding the atoms together by positive 

 bonds (not so diamagnetic an arrangement, however, as if it involved 



groups of eight). In ( Cn - Cr y^ it is true that only three free 



magnetons are left, but these are likely to be more free from the 

 influence of other Cr atoms than can be the case in metallic Chro- 

 mium. In hydrated CrCl 3 they are still more free, not being used up 

 in combining with H 2 molecules as might at first be suspected, for 

 since H 2 is most of the time in the " saturated " phase (see §9) it 

 would not have much attraction for free magnetons. For other facts 

 illustrating this general principle see §23. 



Turning to the relations between the compounds of these metals 

 of groups V-VIII, we see that, if the uncombined metals are excepted, 

 the paramagnetism runs parallel with the number of free valences or 

 magnetons, until in the saturated compounds it vanishes (NaVO s , 

 CrO a ), or becomes very small (KMn0 4 ). This relation has been 

 roughly indicated by Pascal (loc. cit.). 



In the transition metals of groups I B and II B, however, we find a 

 different set of relations. We have seen that a great deal is explained 

 by the tautomerism which naturally falls to the lot of these elements 

 (see table of Periodic Scheme, §7). The very striking fact that 



