is, however, erroneous, owing to an arithmetical error in the last equation (loc. cit. p. 57). The correct number deduced from his data is 0.725 × 10-6.

A more accurate value can be calculated from the dilution formula,

as follows. At 18°, the dilution of a solution saturated with carbon dioxide at a pressure of 760 mm. is, according to Bunsen's absorption data, equal to 24 litres. If the partial pressure of carbon dioxide in air is 0.0003 atmosphere, the dilution of a solution saturated at this pressure will be 24/0·0003 = 80000. If, then, in the above equation we substitute 0.000000304 for k and 80000 for v, we obtain m= = 0·144, that is, 144 per cent. of the carbonic acid dissolved from normal air by pure water is dissociated into the ions H and HCO3. From the degree of dissociation m we obtain the molecular conductivity μ by multiplication with 336, the maximum molecular conductivity of carbonic acid at 18°. From this value, namely, 48.4, we obtain the specific conductivity on dividing by the dilution in cubic centimetres, so that we have the conductivity 48-4/80,000,000=0.605 x 10-6 for water which has been in contact with the atmosphere at 18°.

It is highly improbable that Henry's law in an unmodified form can be applied with propriety to such a case as that discussed above. From the study of analogous cases, it appears much more likely that the concentration of the gas in the air bears a constant ratio to that of the undissociated portion of the dissolved gas, rather than to the concentration of the total dissolved gas. The dilution of the undissociated portion 1 - m thus becomes 80000, and the dilution of the whole gas dissolved 69000. Calculating in the same manner as above described, we obtain the value 0.65 × 10-6 for the conductivity of water which has been in contact with air. If we use the constant derived from Knox's numbers, the values become 0.67 x 10-6 and 0.71 x 10-6 for the unmodified and modified forms of application of Henry's law respectively. Kohlrausch (Zeit. physikal. Chem., 1894, 14, 321) found that water prepared in a vacuum and of conductivity 0·11 × 10-6, gained in conductivity on being left in contact with the air until the value 0.60 × 10-6 was reached. It is also stated by Kohlrausch and Holborn (loc. cit., p. 111) that the lowest conductivity obtainable for water distilled in air is 0.65 × 10-6. It will be seen that these values are in excellent agreement with those calculated from our experiments, so that we may assume with confidence that carbon dioxide is the only substance in the atmosphere which confers conductivity on water.

State of Carbon Dioxide in Aqueous Solution.

In what has been said above, it is assumed that all the dissolved carbon dioxide exists in the aqueous solution as carbonic acid, H2CO. This is by no means necessarily the case, for a large proportion might exist in the solution as carbon dioxide without entailing any alteration in the apparent dissociation constant. We may suppose, for example, that only half of the dissolved carbon dioxide exists in the solution as H.CO, and its dissociation products H. and HCO'. If v, as before, represents the volume in which 1 gram-molecule of the carbon dioxide is dissolved, irrespective of the condition it assumes in the dissolved state, the dilution formula becomes

since the quantity of H,CO,, which was formerly 1, is now only. Now, in solutions of this strength which we investigated, m does not amount to more than 0.006, so that we can write the dilution formula in the form

3

=

We have therefore k' 2k. The real dissociation constant of the acid H.CO, would therefore, in this case, be twice the apparent dissociation constant, namely, equal to 0.0,608. In general, if 1/n represents the fraction of the total dissolved carbon dioxide which exists in the solution as HCO3, the dissociation constant for the acid will be nk, where k is the apparent dissociation constant calculated from our experiments. What is here stated holds good, however, only for moderate degrees of dilution and for moderate values of n, for as soon as m becomes of dimensions approaching those of 1/n, the simple formula can no longer be applied.

We have assumed above that the proportion of the dissolved gas which remains as CO, is constant and independent of the dilution of the solution. This assumption is justifiable, since the active mass of the solvent water must remain sensibly constant for dilute solutions, and the quantity converted into H,CO, will therefore be proportional to the quantity dissolved. It is possible, however, that the equilibrium is between the CO, in solution and the undissociated HCO3, not

the whole amount of H,CO, and its dissociation products. For moderate dilutions and moderate values of n, this latter assumption in no way alters the deductions given above.

Since we obtain a constant value of k for dilutions up to 125 litres, the value of n cannot be very great-cannot, for instance, well be more than 5, for otherwise Ostwald's dissociation formula would not be applicable in its simple form. The agreement, too, between the actual and calculated values of the conductivity of water which has absorbed carbon dioxide from normal air points to the value of n being small, probably not greater than 2. We may take it, then, as fairly certain that when carbon dioxide dissolves in water, at least half of the dissolved substance exists in the form of the acid H2CO3.

It is only the apparent dissociation constant which is of interest to us, however, for it is that which enables us to calculate the strength of carbonic acid in solution as an accelerator, as a conductor of electricity, or as competing for a base against other acids. A knowledge of the real constant, and of the constant regulating the equilibrium,

H,O+CO,=H,COg,

would be of undoubted theoretical interest, but for practical purposes and ordinary solutions, the apparent constant supplies us with all the information necessary for the treatment of problems likely to occur.

Hydrogen Sulphide.

In 1885, Ostwald determined the conductivity of hydrogen sulphide, and found that it was very small. No constant can be calculated from his numbers, however, as at that date the influence of the quality of the water employed in making the solutions was insufficiently understood. We therefore made several determinations with the best water we could obtain, and with hydrogen sulphide as free as possible from foreign conducting matter.

The hydrogen sulphide was prepared by the action of hydrochloric acid on a very concentrated solution of pure sodium sulphide, and was subjected to no other purification than thorough washing with water, the final washing taking place through water contained in a Geissler potash apparatus. If the hydrochloric acid is added at such a rate that the disengagement of hydrogen sulphide is slow and steady, the method gives a product of constant conductivity. The strengths of the solutions thus prepared were estimated by adding a measured quantity of the solution to a known excess of silver nitrate solution, filtering, and determining the amount of silver in the filtrate by Volhard's method.

The maximum conductivity of hydrogen sulphide, treated as the

monobasic acid H.HS', was fixed by means of measurements of the conductivity of sodium hydrosulphide, NaHS (Walker, Proc. Roy. Soc. Edin., 1893-4, 255). These measurements give 58 as the ionic rate for HS', and therefore 356 as the maximum conductivity for hydrogen sulphide at 18°.

Another similar set of experiments at somewhat smaller dilutions gave the mean value 0·0,574, which is practically identical with the former result.

Since the impurity in the water used for dilution is carbonic acid, the dissociation constant of which is known, it is possible to correct the individual values of the conductivity by using the dissociation equations for the separate acids. As the correction in this case, however, is of very small dimensions, it may be neglected without sensible error. An example of the method of calculation employed in the correction will be given when the conductivity of phenol is under consideration.

Hydrocyanic Acid.

We are again indebted to Ostwald for measurements of the electric conductivity of hydrocyanic acid. He found the conductivity to be considerably smaller than that of hydrogen sulphide, but, as before, his numbers are not sufficiently accurate to permit of the calculation of a dissociation constant, owing to the uncertain correction for the conductivity of the water employed as solvent.

In our experiments we used water of conductivity not exceeding 0.65 x 10-6, and even with water of this quality experienced much difficulty in obtaining satisfactory solutions. The method we finally adopted for preparing solutions of hydrocyanic acid was first to prepare a liquid acid very nearly free from water, and then allow the vapour of this to pass slowly into the water of minimum conductivity. The liquid hydrocyanic acid was obtained by gently heating a mixture of potassium ferrocyanide and glacial phosphoric acid with an equal bulk of water, and condensing the vapour in a cooled distilling flask. When a sufficient quantity had been collected, the

distilling flask was disconnected from the generating apparatus and attached to a delivery tube which dipped beneath the surface of the water used as solvent. As the conductivity of the hydrocyanic acid was very small, the solutions were made as strong as was consistent with the theoretical possibility of obtaining a constant value for the expression k. It was found that non-platinised electrodes gave better results than those which had been platinised.

Kohlrausch has shown that the molecular conductivity of potassium cyanide in concentrated solutions (normal and semi-normal) is intermediate between the molecular conductivities of equivalent solutions of potassium chloride and potassium iodide. As these two salts have practically the same molecular conductivity for infinite dilution, it was assumed that the conductivity of potassium cyanide would have a maximum value equal to 1216, which gives 60 for the ion CN' and 358 for the conductivity of HCN at infinite dilution.

The concentration of the original solution was determined with silver nitrate solution according to Liebig's method.

Another set of experiments gave a mean value of 0.014, and several preliminary experiments gave still higher values. We have chosen the smallest value as being the most probable, owing to the fact that any possible impurity would increase the conductivity and thus the

constant.

The constant as it stands is probably still too high, for even the presence of the carbonic acid in air-saturated water would effect an increase of about 2 per cent. on the mean value.

Boric Acid.

Kahlenberg and Schreiner (Zeit. physikal. Chem., 1896, 20, 547) have shown that in all probability only one boric acid exists in solution, namely, HBO, and that in dilute solutions the only stable salt is of the type NaH,BO,. From their conductivity numbers, it would appear that the maximum conductivity of the salt NaH,BO, is about 75.0 at 18°. Now this number is certainly too great, as Shields has proved