ON ELECTROLYSIS AND ELECTRO-CHEMISTRY. 199 
efficients of all the electrolytes are of the same sign and of the same order 
of magnitude. Probably all electrolytes have temperature coefficients of 
the same sign, and this may have to be explained, but it does not help 
towards classification, for some alloys! have a positive coefficient. More- 
over, according to Arrhenius, the sign of the temperature coefficient may 
be reversed at higher temperatures for a number of electrolytes of low 
conductivity (see p. 228). 
According to Kohlrausch,? electrolytes must be mixtures. This is 
supported by the observations upon the effect of mixing two non-con- 
ductors as H,O and HCl, which together form good conductors. And, 
perhaps, we should be justified in regarding whatever conducting power 
there may be in any pure sample of a single liquid as being due to the 
presence of impurity. The conductivity of mixtures of water and alcohol 
have been carefully investigated by Pfeiffer,’ and from his curve it 
is clear that certain percentages of mixture have higher conductivity 
than either water or alcohol. 
If this is to be regarded as a satisfactory definition of an electrolyte, 
the converse proposition, that a liquid will conduct if it be a mixture, 
should also hold. That is to say, in order to make one of the liquids 
‘given in Table III. conduct, all that is necessary is to mix it with some 
‘other substance. Mr. W. Coldridge* has examined from this point of 
view the effect of mixing various substances with SnCl,, and has found 
that whereas the absorption of a small quantity of dry HCl gas produces 
a liquid which has very slight conducting power and shows galvanic 
polarisation, platinum chloride or chloroform can be mixed with the 
tin chloride without producing any conducting power. Moreover, the 
tin chloride absorbs considerable quantity of dry H.S gas, which gives a 
yellow liquid insulating apparently as completely as the tin chloride 
itself, and at the same time no precipitation of SnS, occurs; but the 
addition of a minute quantity of water or alcohol to the mixture deter- 
mines at once the precipitation of the tin sulphide and at the same time 
the conduction through the liquid. There seems to be a wide field for 
useful experiments in this direction, with the primary object of determining 
what is the nature of the special kind of mixture which causes con- 
- ductivity and what are the ions when such a conducting mixture is pro- 
duced. The fact that mixture alone is not sufficient to account for 
electrolytic action may be to some extent inferred from the fact that no 
evidence of decomposition can be observed in the conduction of electricity 
through alloys.* 
Hittorf,® in his valuable survey of the history of electrolysis, maintains 
the proposition ‘Electrolytes are salts’; but, p. 401, he says, ‘As from 
chemical phenomena no sharp distinction can be drawn betwecn salts 
and non-salts, so it is with the distinction between electrolytes and insu- 
Jators.’ Hittorf’s definition of a salt’ is a compound which by double 
affinity exchanges its constituents with those of another recognised 
electrolyte, the ions of the respective compounds being those constituent 
parts which take part in the double exchange. Upon this definition 
Professor G. Wiedemann remarked at the B.A. Meeting, 1887 (‘ Report,’ 
1 Yon Aubel, Proc. Phil. Soc, vol. 9, p. 133. 
2 Gegenwartige Anschawug, pp. 10 and 17; Pogg. Ann. 159, p. 271, 1876. 
3 Wied. Ann. 25, p. 232, 1885. 4 Phil. Mag. vol. 29, p. 385, 1890. 
5 See B.A. Report, 1887, p. 341. 6 Wied. Ann. 4, p. 374, 1878. 
7 Pogg. Ann. 106, p. 561, § 65. 
