EQUATION AND THE NATURE OF COHESION. 9 



when relieved from pressure, or b g , is 26 . This assumption gives, 

 when combined with the assumption that the density at absolute 

 zero is S times the critical density, 8 being the critical coefficient, 

 m\.jP r I ;, the result : a = ( (S 2 — S -f 2)/(S — 2) ) P c F c 2 . Van 

 Laaii criticises this method, and no doubt correctly as I have al- 

 ready pointed out elsewhere. The assumptions are not exactly true 

 either of them. The values it gives for a are too high, in the 

 simpler substances especially, but in reality they are not so far wrong 

 for many substances. b c instead of being %b is, in reality, about 

 1.8 — 1.9 b and in a great many cases d is actually very nearly 

 8d c . By using these various methods of computing a for these 

 simple gases 1 found the law discovered in more complex substances 

 would hold if we assigned 2 valences to the molecule in 1L 2 , N 2 , 

 2 and CO, and 4 or 5 valences to C0 2 . That H 2 and N 2 should 

 have each 2 valences was expected, but that 2 and CO should 

 have 2 was surprising. I concluded that, on the whole, it seemed 

 to me probable that such a simple rule as that correlating cohesion 

 and gravitation would hold everywhere, if it held at all, and that 

 while in these two or three simple gases the number of valences 

 necessary to ascribe to the molecule was surprising, yet if the law 

 were true we must accept them. These exceptional substances, it 

 will be noticed, all contain oxygen, a very' significant fact. There 

 was no juggling with the valences at all. I can see no reason why 

 carbon should not be univalent and oxygen also. Carbon is known 

 to be di, tri, and tetra-valent, so why not univalent? Oxygen is 

 di and tetravalent; and is oxygen not univalent in the peroxides? 

 A juggler is one who conceals how tricks are accomplished. There 

 was no concealment of the fact that these substances were excep- 

 tional ; in truth I discussed their apparent exceptional position more 

 than its importance deserved. As regards the hexavalence of sulphur, 

 even in H 2 S, it may be pointed out that so eminent a physical 

 chemist as G. N. Lewis has recently proposed that sulphur is 

 everywhere hexavalent, and if so good a chemist as Lewis makes 

 such a suggestion we should not be surprised if we actually find 

 that this suggestion is correct, and that sulphur is hexavalent in some 

 locations in which, as in H 2 S, one would not at first expect it to be. 

 The greatest interest attached to the argon group of gases. If it 

 is true that cohesion depends on valence, then these gases since 

 they have cohesion must have, sometimes at any rate, valence 

 electrons attached to their atoms. This seems perhaps, at first thought, 

 improbable, and I fear for the effect of this information on van 

 Laar when I tell him that Lewis has recently suggested that these 



