May 31, 1906] 



NA TURE 



proportional, upproximatoly at any rate, to the stiuare of 

 the- conii-ntralion. The conductivity, which is measured 

 bv the product of the number of ions and their mutual 

 vilocilv, will then vary approximately as the cube of the 

 loncenlration. But the facts are quite at variance with 

 these conclusions. The velocities of the ions are found 

 to be constant in dilute solutions, and to decrease slowly 

 with increasing concentration when the solutions are 

 stronger, while the conductivity of a dilute solution is pro- 

 portional at most to the first power of the concentration. 



We must, therefore, it seems to me, abandon the theory 

 of ionic interchanges at the instants of molecular collision, 

 and turn to the rival hypothesis. There are several pieces 

 of positive evidence in favour of the view that the ions 

 migrate independently of each other through the solution. 

 Of these pieces of evidence I may mention three : — (1) the 

 calculated velocity of an ion, as calculated from the con- 

 ductivity and the migration constant, is independent of the 

 nature of the other ion present when the solution is dilute — 

 the velocity of chlorine is the same whether the other 

 ion be potassium, or lithium, or sodium; (2) the difference 

 of potential observed between two solutions of different 

 concentrations can be calculated numerically on the assump- 

 tion that the ions migrate independently of each other until 

 the electrostatic forces of attraction prevent further separ- 

 ation ; (3) it is possible to explain satisfactorily the curious 

 relation between the valency of the metallic ion of a salt 

 and its specific coagulative power on certain colloid solu- 

 tions on the assumption that coagulation is due to the 

 effect on the surface conditions produced by a certain 

 minimum electric charge brought together by the chance 

 conjunctions of isolated ions (Phil. Mag., 1899). 



The evidence sketched above has been obtained by observ- 

 ations on aqueous solutions. It remains a further question 

 whether similar relations hold for solutions in other solvents 

 and for such electrolytes as fused salts. In the latter case, 

 at any rate, the conditions may be fundamentally different, 

 and it seems necessary to wait for further investigation. 



Such, it seems to me, is the present case for the theory 

 of ionic dissociation when put in brief form and restricted 

 to direct evidence. The indirect evidence, based on the 

 success of similar ideas in coordinating and explaining the 

 conduction of electricity through gases, however, must 

 now be taken into account in any full discussion of the 

 subject. I confess that to me the cumulative evidence 

 seems overwhelming. But I hold no brief for the theory, 

 as Prof. Armstrong seems to imagine, and if ever it ceases 

 to be the best hypothesis in the field I shall willingly 

 abandon it. Consistency always seems to me to be the 

 meanest of the virtues, and in matters scientific it may 

 become the most deadly of the vices. 



I have endeavoured, as well as the exigencies of space 

 admit, to comply with Prof. .Armstrong's request, jlay 

 I ask him in return to state his own view of the mechanism 

 of electrolysis, to explain, in effect, what is the alternative 

 he proposes to the theory of ionic dissociation? 



I am glad to have given an opportunity for Mr. Xormim 

 Campbell to explain in public his interesting views on the 

 subject of thermodynamics — views which we have dis- 

 cussed more than once in private without coming to an 

 agreement, though, for my part, I have learnt much by 

 Mr. Campbell's clear-sighted criticism. He will know what 

 my answer to his letter must be, though I have no hope 

 that he will be convinced thereby. 



I do not wish to discuss the whole question of the theory 

 of thermodynamics; I am not entitled to do so. But Mr. 

 Campbell holds that, even if general thermodynamics be 

 sound, its application to osmotics is not justified. Now 

 it is here that we do not agree, and I should like to 

 explain the grounds of my belief that not onlv is the 

 application of thermodynamics to osmotics successful, but 

 that the success is one of the strongest arguments we 

 possess for the general validity of reversible thermo- 

 dynamics as an ideal towards which practical arrange- 

 ments may approximate. 



Referring to the concordance between the theoretical 

 value of the osmotic pressure of dilute solutions and that 

 observed experimentally, .Mr. Campbell says : — " It must 

 be remembered that there is not perfect agreement. . . . 

 The errors are larger than those involved in the direct 



NO. 1909, VOL. 74] 



measurement of the pressure and the other quantities- 

 involved ; there is a systematic error. But this is due, say 

 the thermodynamicists, to the imperfection of the mem- 

 brane." I am not sure that I like to be called such a bad 

 name, even by Mr. Canipbell, but may I point out that 

 .Mr. Campbell has not put forward the whole of the 

 evidence? 



The direct measurement of osmotic pressure may not be 

 a very accurate performance, though experience shows that 

 as our membranes become more and more perfect the 

 observed values more nearly approach that indicated by 

 theory as the ideal case. But let us pass that by for a 

 moment, and consider the relation between the osmotic 

 pressure of a solution as defined theoretically and its 

 freezing point. To do so we must use again thermo- 

 dynamical reasoning, and introduce its necessary assump- 

 tions. But most of the assumptions will be the same as we 

 made in deducing the theoretical value for the osmotic pres- 

 sure, and they, at any rate, do not really add to the total, 

 number. Having obtained the well-known relation between 

 osmotic pressure and freezing point, let us calculate 

 numerically what the freezing point of the solution would 

 be if the osmotic pressure possessed the theoretical value. 

 With the most probable number for the latent heat of 

 fusion, the molecular depression of the freezing point of 

 water bv the solution in it of a small quantity of any 

 substance should be i°-S57 C. Mr. Griffith's experiments, 

 made with all the refinements of platinum thermometry, 

 gave for the molecular depression of solutions of cane- 

 sugar (from a concentration of 0-0005 to 0-02 gram- 

 molecules per litre) the number i°-858. The limiting 

 number for solutions of potassium chloride was 3°-720, 

 half of which is i°-86o. Now these numbers agree with 

 astonishing closeness ; the opposite errors on Mr. Camp- 

 bell's hypothesis must balance each other to within very 

 little more than one part in a thousand. I cannot disprove 

 his contention, but I think the balance of probability is 

 against him, and, after all, any evidence which involves 

 experiment must be merely a question of less or greater 

 probability. 



Now it seems to me that in this concordance we have a 

 very strong piece of evidence in favour of the existence of 

 an efTectively reversible operation in the process of freezing 

 — an operation much more nearly truly reversible than any 

 it is possible to obtain with the rough and crude arrange- 

 ments of even the best practical heat engines. Hence my 

 present opinion is that, not only is the application of 

 thermodynamics to osmotic phenomena in their theoretical 

 aspect justified, but that it gives one of the most valuable 

 confirmations we possess of the general validity of thermo- 

 dynamic reasoning as a means of discovering an ideal limit 

 towards which practical processes may tend. 



In conclusion, may I point out the great interest of Mr. 

 Campbell's argument that it may be "the swifter mole- 

 cules which escape into the vapour and the slow-er which 

 escape into the solid." The contention seems reasonable; 

 and, though I feel that the strength of the case I have 

 stated above is so great as to indicate that such a sifting 

 of molecules cannot affect appreciably the thermodynamic 

 result, it is clear that Mr. Campbell's point should be 

 considered. Has Mr. Campbell really caught a Maxwellian 

 demon? If so, all the world will crowd to study the speci- 

 men. But this letter is too long already, and I must leave 

 the discussion to some one more learned than I am in 

 this particular branch of demonologv. 



W. C. D. WllETH.^M. 



Trinity College, Cambridge, May 26. 



A New Vesuvian Mineral. 



Among the ejected masses that strew the flanks of 

 Vesuvius many minerals of known species are to be 

 recognised, and will be referred to in a publication now in 

 preparation. 



Perhaps the most interesting among a large number of 

 blocks that I broke open after the eruption were two con- 

 taining fine crystals of halite associated in spots with a 

 canary-yellow mineral. This latter occurs in crystals up 

 to several millimetres in diameter, and, so far as I have 

 yet examined it, appears to be in rather flat rhombohedra. 

 It is somew-hat deliquescent, and therefore requires special 



