50 president's address — SECTION A. 



perfection of the thermal measurements involved, seeing that the changes 

 of thermal capacity afiect onl}^ the fom'th and fifth significant figures of 

 that quantity — never the third by more than a single unit. Consider- 

 ing the state of exact thermometry 35 years or more ago, when Marignac 

 and Thomsen determined these thermal capacities — to say nothing of 

 the other difficulties involved in calorimetry — it is not unreasonable to 

 suppose, with Berthelot, that the fourth figures may be uncertain to 

 the extent of several units, and the fifth of little or no value — an 

 assumption which would deprive the discrepancy, and, with it, 

 Richards' whole discussion, of all significance. 



The point at issue is important, inasmuch as it raises once more 

 the question of the measurement of chemical affinity, and even of its 

 nature. To ask whether the free energy represents the energy of 

 chemical affinity, or whether it needs to be supplemented by energy 

 obtained from atomic contraction, is pretty much the same as to ask 

 whether chemical combination is a wholly inter-atomic or a partly 

 intra-atomic process. I need hardly remind you that the first of these 

 alternatives has hitherto been the accepted one, but may content 

 myself with remarking that the data under discussion give us no ade- 

 quate reason for changing our opinion, though it may be well to keep 

 an open mind on the subject. 



By far the most numerous advances in voltaic theory during the 

 last 17 years are due to the work of the advocates and opponents of 

 Nernst's theory. To this theory, therefore, we will now turn our atten- 

 tion, and, as the language of both parties is adopted from that of the 

 ionic hypothesis, we will, to save circumlocution, follow their example. 

 The success ef Nernst's theory, as applied to concentration cells, 

 was immediate and complete. This is not in itself surprising, as his 

 formula3 — stripped, as I said before, of their ionic garb — are, to all 

 intents and purposes, identical with those of Helmholtz's general 

 theory. Nernst, however, was enabled by his mode of presenting the 

 subject to test the theory in a vastly larger number of cases than his 

 predecessor could do, and to prove that his formulae were generally 

 applicable. The application of the theory to liquid cells — which differ 

 from concentration cells merely in having indifferent electrodes — was 

 not so successful ; the observed and calculated values of the electro- 

 motive force always differed, the latter being invariably in excess — 

 sometimes by as much as 25 per cent. So far as I can learn, the dis- 

 crepancy has never been cleared up, though Ostwald called attention 

 to its importance more than 13 years ago. 



But the real battleground of the theory has lain in the region of 

 those cells whose electromotive forces depend on the nature of the solid 

 metals present in the voltaic chain. Here, for the first time, it is neces- 

 sary to introduce hypothesis in order to describe the way in which a 

 metal goes into solution. Nernst effected this by his idea of electrolytic 

 solution pressure, or, what comes to the same thing, by explaining the 

 potential difference at a metal-liquid boundary as depending only on 

 the ionic pressure in the metal and the partial pressure of that metal's 

 own ions in the liquid. Writing as he did in the early days of the ionic 

 hypothesis, it was natural that he should take the same view as others 



