. //;-// 14, 188;] 



NA TURE 



5/1 



i;on.il ' behaviour of mercuric and beiyllium chlorides is lo be 

 explained by considerations such as these. 



To discuss such questions at all satisfactorily, however, we 

 require to know much mwe of the electrical behaviour of pure 

 fused salts. It is surprising how little accurate knowledge we 

 l>ossess on this subject. 



Ciitiiposite Electrolytes. — I assume it to be admitted that 

 neither water nor liquid hydrogen chloride, for example, is an 

 L'Iee;rolyte, although an aqueous solution of hydrogen chloride 

 conducts freely, and is electrolysed by an electromotive force of 

 l)ut luile more than a volt. 



The theory put forward by Clausius in 1857 in explanation of 

 electrolysis (cf. Clerk Maxwell's " lilementary Treatise on Elec- 

 tricity,'' p. 104), has been widely accepted by physicists ; but it 

 appears to me that, on careful consideration of the evidence, and 

 especially of recent exact observations on conditions of chemical 

 change, it must be admitted, as I have elsewhere contended 

 (B. .\. Address), that proof is altogether wanting of the exist- 

 ence of a condition such as is postulated by Clausius. Moreover, 

 Tt it has been shown by Hittorf that cuprous and silver sulphides, 

 and by V. Kohlrausch tliat silver iodide, all undergo electrolysis 

 in the solid state : the partisans of the dissociation hypothesis 

 would, I presume, scarcely contend that it is easily applicable to 

 such cases as these. It also does not appear to afford any 

 explanation of the abrupt change in conductivity which occurs 

 in solid silver iodide and sulphide as the temperature is raised ; 

 nor of the peculiar variation in conductivity which is observed 

 on diluting sulphuric acid with water. 



Again, I venture to think that the conductivity of a mixture 

 of compounds which themselves have little or no conducting 

 power is accounted for in but an unsatisfactory and insufficient 

 manner by the hypothesis put forward by F. Kohlrausch (Pogg. 

 Ann., 1876, cli.x., p. 2331 ; there appears to be y<(r too great a 

 difference in the behaviour of the ])ure compounds, water and 

 liquid hydrogen chloride for example, and of a mixture — no 

 decomposition apparently of either compound being effected by 

 any electromotive force short of that which produces disruptive 

 discharge, although the mixture of the two will not withstand 

 an electromotive force of little more than a volt. Influenced by 

 these considerations, I am led to conclude that there is no 

 salisfaclor)' evidence that the constituents of the electrolyte either 

 are free prior to the action of the electromotive force, or are 

 primarily set free by the eflect produced by the electromotive 

 force upon either metnber separately of the composite electrolyte ; 

 but that an additional influence comes into play, viz. that of the 

 one member of the composite electrolyte upon the other while 

 both are under the influence of the electromotive force. This 

 influence, I imagine, is e.xerted by the negative radicle of the 

 other member. Assuming, for example, that in a solution of 

 hydrogen chloride in water the oxygen atom of the water 

 molecule is straining at the chlorine atom of the hydrogen 

 chloride molecule, if when subjected to the influence of an 

 electromotive force the molecules are caused to flow p.ast each 

 other — the phenomena of electric endosmose may be held to 

 afford evidence that in composite electrolytes the molecules are 

 thus set in motion — it is conceivable that this influence, super- 

 added to that of the electromotive f jrce npou the electrolyte, 

 may bring about the disruption of the molecule and conduction : 

 in short, that a state may be induced such as Clausius considers 

 is the state prior to the action of the electromotive force. 



A large amount of most valuable information on the con- 

 nexion of dilution anJ electrical conduction in aqueous solutions 

 has been recently published by Arrhenius, Bouty, F. Kohlrausch, 

 and Ostwald. In his most recent paper, Ostwald (Journal 

 fiir praktische C/temic, 1885, xxxii. , p. 300) has given the results 

 of his determinations of the molecular conductivity in the case 

 of no less than about 120 diflercnt acids ; and it appears to me 

 that many — indeed all — of his obsenations afford most distinct 

 evidence in favour of the view I have expressed above. The 

 general result of his investigation is that the molecular conduc- 

 tivity increases with dilution : in other words, that the dissolved 

 substance exercises a greater specific effect, finally attaining a 

 maximum ; it then diminishes, but he believes this to be due to 

 impurities in the water, especially to neutralisation of the acid 

 by traces of ammonium carbonate. The maximum, he appears 

 to think, would be the same for all acids if the dilution could 

 only be pushed far enough : in the case of monobasic acids it is 

 about 90 (arbilrar)' units) ; it is twice this in the case of dibasic, 

 thrice in the case of tribasic, and so on. 



I will quote first his results in the case of solutions of hydro- 



gen chloride, bromide, iodide, fluoride, and silicon fluoride, v is 

 the volume in litres which contains a weight in grammes corre- 

 sponding to the formula of the dissolved substance — 36 '4 

 grammes of hydrogen chloride, for example. 



It will be obser\'ed that hydrogen chloride, bromide, and 

 iodide practically behave alike ; the numbers for the chloride 

 are, however, slightly lower than those for the bromide and 

 iodide, and the maximum is not reached quite so soon in the 

 case of the chloride. Hydi-ogen fluoride is altogether diflerent ; 

 its molecular conductivity is exceedingly low to begin with, and 

 is considerably below the maximum even when v = 4096. But 

 I would call special attention to the numbers for hydrogen 

 silicon fluoride, which is commonly regarded as a dibasic acid : 

 at first, as Ostwald says, it behaves as a monobasic acid "of 

 moderate strength — iodic acid, for example ; but the maximum 

 for monobasic acids being exceeded, the molecular conductivity 

 increases more and more rapidly, ultimately exceeding the treble 

 value, 270. It must be supposed that it undergoes decomposi- 

 tion in accordance with the equation — 



HoSiFg -I- 2H2O = SiOo + 6HF. 



The noteworthy point is i/ie large excess of water required to 

 initiate this change : when v = \(> the solution contains less than 

 I per cent. H.^SiFj, and at this point, according to Ostwald, 

 decomposition probably begins ; but that it is far from complete 

 even when a very much larger excess is present is evident from 

 the fact that the maximum when v — 32,768 is 2S2 and not 

 above 400. 



Now it is well known that hydrogen chloride, bromide, and 

 iodide are, practically speaking, perfect gases under ordinary 

 circumstances : in other words, masses of these gases would 

 mainly consist of molecules such as are represented by the 

 formulas HCl, HBr, and HI. It has been proved, however, by 

 Mallet, that hydrogen fluoride at temperatures near to its boiling- 

 point mainly consists of molecules of the formula H„Fo. In the 

 aqueous solution the molecules would be brought more closely 

 together, and therefore it is probable that, even in the case of 

 hydrogen chloride, bromide, and iodide, a certain proportion of 

 more complex molecules would result : the relatively high boil- 

 ing-point of hydrogen fluoride (l9°-4) renders it probable that in 

 the liquid state this compound would at least partially consist of 

 molecules more complex even than is represented by the formula 

 HjFo. On the hypothesis put forward in this paper, the influence 

 exercised by the one member of the composite electrolyte upon 

 the other member during electrolysis is at all events mainly 

 exercised by their respective negative radicles, and the extent of 

 the influence thu; nmtually exerted by these radicles would 

 depend on the extent to which they are still possessed of " residual 

 affinity. " If the hydrogen chloride, bromide, and iodide are 

 present chiefly as simple molecules, they should exert, a/> initio, 

 almost the full eftect which they are capable of exerting ; and the 

 chief effect of dilution being to decompose the tnore complex 

 molecules, conductivity should increase to but a slight extent if 

 the extent to which simplification can take place be but small. 

 On the other hand, if owing to the formation of molecular aggre- 

 gates the residual aftinity be more or less exhausted, the initial 

 conductivity will be low, and it will increase on dilution only in 

 proportion as these aggregates become broken up. 



It appears to me that the behaviour of the four hydrides under 

 discussion is absolutely in accordance with these requirements of 

 the hypothesis. Evidence of the same kind is afforded by all of 

 Ostwald's results. 



The behaviour of solutions of neutral metallic salts on dilution 



