above that the osmotic pressure and conductivity of aqueous solutions follow simple laws fairly easy to discover. The hypothesis of electrolytic dissociation in accordance with the general laws of chemistry is sufficient to explain most of the phenomena observed. These laws, however, are only very partially applicable to non-aqueous solutions.

It may be remarked at the outset that water, which, according to the evidence of surface tension, is a highly "associated" liquid-i.e. consists largely of double molecules HО,—has little tendency to admit of double molecules in substances dissolved in it. This is the rule: when the solvent is associated, the dissolved molecules are mostly simple, and vice versa. Ammonia, sulphur dioxide and pyridine, all good solvents, are unassociated, so that it is to be expected that salts dissolved in these media should tend to form molecular aggregates, and that these aggregates should give off ions of greater complexity than those occurring in water. It is not probable that a complete account of the behaviour of non-aqueous solutions can be derived from this principle, but it may at least indicate some of the difficulties in the way of an explanation.

Of the scanty data available, those about sulphur dioxide are the most extensive, thanks to the admirable work of Walden and Centnerschwer.1

Sulphur dioxide in the liquid form has a conductivity about equal to that of pure water. It therefore presumably suffers a small amount of dissociation itself. This can hardly take place otherwise than

SO2 = SO" + O" =S" + 20"

Either, or both, these pairs of ions may be formed, it is not known which; in favour of the existence of S it may be mentioned that ions of tellurium with four positive charges are known.

The univalent halogen salts mostly dissolve in sulphur dioxide. Those tested include chlorides, bromides, iodides, and thiocyanates of K, Na, Rb, NH1, mono-, di-, tri-, and tetramethyl ammonium, as well as ethyl and benzyl bases. The

1 Zeitschr. phys. Chem., 39. 513–596 (1902).

that a remarkable regularity exists in the behaviour of the various salts; but that the regularity is by no means identical with that of aqueous solutions.

results at o° are given in Fig. 27, in which the dilution (litres per gram-equivalent) is represented by the abscissæ, equivalent conductivity by the ordinates. It will be at once seen

=

For a dilution of 1024 litres, solutions of these salts in water (at o°) show A 50 to 80. In SO, solution they vary from 12 to 163. Yet this does not appear to be due to varying degree of dissociation, for the lower curves show just as much indication of a limiting value of A as the higher. Again, these very divergent conductivities cannot be expressed in Kohlrausch's manner as the sum of two parts, one for each ion; for if the values of ▲ be analysed in this way no constant values for and are obtained.

1

With regard to the law of dilution, the authors did not attempt to apply Jahn's theory of strong electrolytes; they tried empirical formulæ that had been found by Rudolphi 1 and van't Hoff' to hold for aqueous salts solutions; but neither formula is applicable to solutions in sulphur dioxide.

3

Walden and Centnerschwer next proceeded to make measurements on the temperature change of conductivity. These they extended over a far wider range than has been done for any other solvent; from near the freezing-point (-76°) to the critical point (+157°). The results are therefore of unusually great interest. It was found that in all cases the conductivity first increases with rise of temperature, and then falls off; 'following, roughly, a parabolic formula. The maximum conductivity lies at very varying temperatures-from-70° for benzylammonium chloride to +7° for tetra-ethyl-ammonium iodide; the temperature is higher for the better conducting salts.

At high temperatures the conductivity always falls off, and practically disappears at the critical point; this although some of the salts used remain dissolved in the gaseous SO, above the critical temperature.

Hagenbach has also measured the conductivity of solutions in SO, up to the critical temperature, and obtained on the whole similar results; he finds, however, that a trace of conductivity is shown by the gaseous solutions,

Liquid ammonia is another substance that possesses strong

1 Zeitschr. phys. Chem., 17. 385 (1895).

2 Ibid., 18. 301 (1895).

3 See also the important work of Kramers (p. 90, infra).

Ann. d. Phys., 5. 276-312 (1901).

ionising power, and whose solutions have been much studied. Franklin and Kraus, especially, have measured the conductivity of a large number of solutions in ammonia, mostly at temperature - 38°; and have also collected data as to boilingpoints and other physical properties of these solutions, as well as their facility for chemical reaction. The behaviour of ammoniacal solutions differs less from that of aqueous than do the SO, solutions studied by Walden; indeed, in one respect they have been found to follow even simpler laws than aqueous salt solutions. Alkaline salts, while freely dissolved by liquid ammonia, are less dissociated than in aqueous solutions of the same strength; but by pushing the dilution to an extreme (50,000 litres or more per equivalent) it was found possible to trace the equivalent conductivity till it approached the limit; A and then, on calculating the degree of dissociation from it Ac

was found that the values so obtained agree with Ostwald's law of dilution. Thus certain salt solutions in ammonia follow the very simple conditions that hold only for weak acids and bases in aqueous solution, giving an important confirmation of the principles on which the conductivity of solutions has been explained. The limiting equivalent conductivity in ammonia is extraordinarily high, some three times as great as that in water; so that the frictional resistance experienced by the ions must be unusually small.

Very many other detached experiments have been made on solutions in hydrocyanic acid, organic nitrites, amines, in alcohol, ether, mixtures of water and alcohol, and other liquids, but hardly anything in the way of generalisations has yet been arrived at. Mention may be made of the work of Kahlenberg 2 and his students, who give a large collection of experimental data, but too hastily dismiss the possibility of bringing them into accordance with the dissociation theory.

The examination of a large number of solvents has, however, at least given some insight into the physical and chemical properties that are associated with ionising power. The first 1Amer. Chem. Journ., 21. 1, 8; 28. 277; 24. 83.

suggestion was made by J. J. Thomson and by Nernst; they showed that it is liquids of high dielectric constant that afford good conducting solutions. The dielectric constant of water is about 80; other liquids with high values of it are hydrocyanic acid (95), methyl alcohol (32), formic acid (57), pyridine (20), ammonia (16), which all exercise considerable dissociating power; while the hydrocarbons, which do not form electrolytic solutions at all, have dielectric constants from 2 to 4. There is, however, only a rough parallelism between the two properties, many exceptions being found. It should be remarked that two electric charges separated by any medium possess an amount of energy inversely proportional to the dielectric constant of the medium. It will therefore require less work to separate a positive and negative ion in water or hydrocyanic acid than in other liquids. This fact, based on the principles of electrostatics, affords a certain theoretical justification for the Nernst-Thomson rule, since dissociation might be expected to occur most freely, cet. par., in a medium that allows the ions to be separated with little expenditure of work.

Brühl1 has given a widened meaning to the hypothesis, without, however, making it any more exact, by his conception of the medial energy. This is the energy of the medium on which its power of dielectric separation, tautomerisation, and ionisation depend, and which finds its simplest expression in the latent heat of evaporation, i.e. in the work required to break up the liquid entirely into a gas. It is true that large latent heat usually goes with high dielectric constant and considerable ionising power; but, again, the relation is an irregular one, evidently complicated by other factors. Brühl also points out that good dissociators have unsaturated chemical affinities (oxygen in water, nitrogen in ammonia, etc.); but, on the other hand, it is not all such unsaturated compounds that dissociate well.

Dutoit and others have attempted to bring dissociating power into connection with molecular complexity or association" in the liquid state; but for this there is little evidence, as sulphur dioxide and ammonia are both unassociated.

1 Zeitschr. phys. Chem., 30. 1-63 (1899).