NO. II STRUCTURE OF THE ATOM PARSON 43 



resemble P and S. The elements Fe, Co, Ni have been given the 

 constitution 3y + 8, and, although they themselves are not known to 

 be octovalent, the analogous elements Ru (5y + 8) and Os (o,y + 8) 

 give the oxides Ru0 4 and Os0 4 . 



Again, Cu, Zn, and Ga, the last three of the series, give compounds 

 in which they are mono-, di-, and tri-valent respectively, thus resem- 

 bling the typical elements of the groups I, II, and III, in which they 

 lie. Now we have seen (§7) that the atoms of these elements may 

 be supposed to undergo an intra-atomic tautomerism similar to the 

 intra-molecular tautomerism described for CH 4 , NH 3 , OH 2 , HF, in 

 §9 ; and, as in that case, brackets have been used to represent roughly 

 the proportions we may expect of the two phases. What we should 

 predict from this is true in fact, for the first long period at all events : 

 compounds of monovalent Cu (from the 4y+i phase) are less stable 

 than those of divalent Cu (from the 3y + 9 phase, as explained below), 

 while compounds of trivalent Ga (from the 4y + 3 phase) are stabler 

 than the other compounds of Ga. In the case of Zn, it will be shown 

 that both phases make for divalency. For the other two long periods 

 the agreement is not so good. 



In addition to the individual properties already referred to, all the 

 elements of this series give basic oxides in which the atom of the 

 metal tends to be trivalent towards the left of the series, and divalent 

 towards the right. The very regularity of this series of oxides 

 indicates that they are in some way due to the 3y phases of these 

 atoms : (3y+ ) 5, 6, 7, 8, 9, 10, 11 ; but the connection seems at first 

 sight to be remote. It appears as if the successive additions of 

 magnetons have a very slight effect in these transitional series, as far 

 as the basic oxides are concerned (compare with this the even greater 

 monotony in the series of rare-earth elements : see the table of the 

 Periodic Scheme at the end of §7). 



Of course Cu, for example, could not be expected to give stable 

 salts in which it is 9-valent, such as CuCl 9 , not only because of the 

 mechanical hindrance to this, but also because the extraction of so 

 many magnetons from the Cu atom would have to be effected against 

 comparatively great electrostatic forces : towards negative groups the 

 atom must remain unsaturated, and some of its nine magnetons will 

 be free. This brings us to yet another application of the principle of 

 the disturbing action of free magnetons, which was discussed in §12. 

 We get the result, paradoxical at first sight, that the more free 

 magnetons an atom possesses, the fewer it can use to combine with 

 negative radicles — unless it succeeds in using them all, which is not 



