324 MR. W. C. D. W1IETHAM ON THE IONIZATION OF 



complete, the temperature coefficient of conductivity is determined by the change in 

 the velocities only. Since the temperature coefficient decreases as the concentration 

 increases, it must follow that, in the case of these salts, the eifect of temperature on 

 the ionization is of opposite sign to its effect on velocity, so that heating the solutions 

 decreases their ionization. The curves giving the variation of ionization with con- 

 centration at 18 will, therefore, be more steeply inclined to the axis of concentration 

 than the curves for the same quantity at 0, and, in order to get satisfactory com- 

 parisons with freezing point measurements, it is necessary to make conductivity 

 measurements at 0, and to carry the dilution to such an extent that the limiting value 

 of the equivalent conductivity can be directly estimated. This point will be again 

 considered later, and ionization curves given for 18 as well as for the freezing point. 



The measurements of the freezing points of dilute solutions which have been 

 hitherto made show great divergences when the values obtained by different observers 

 are compared, though the results of any one observer often agree well among them- 

 selves. The experimental knowledge of this side of the question must be regarded as 

 even more unsatisfactory than that of the electrical conductivity. 



In the course of the year 1897, Mr. E. H. GRIFFITHS and the present writer under- 

 took the further experimental investigation of the subject. Mr. GRIFFITHS arranged 

 to make the freezing point determinations by the method of platinum resistance 

 thermometry. The following paper contains an account of the corresponding measure- 

 ments of the electrical conductivities. Mr. GRIFFITHS proposes to publish a separate 

 account of his experiments when finished, and this will be followed by a joint com- 

 parison of the two lines of research. 



SECTION 1. On the Preparation of the Solutions. 



In investigations of the variation of electrical conductivity with concentration, it 

 has been usual to begin with a solution of the maximum concentration required, and 

 to gradually dilute it till the desired limit was reached. This method involves the 

 use of a very large volume of water, the quality of which must be constant. It also 

 involves the pouring of the solutions into and out of the electrolytic cell at each 

 dilution. Impurities are always liable to enter the liquids during this operation, 

 especially when, as in this investigation, the temperature of the liquids is so low that 

 aqueous vapour from the air must be condensed on their surfaces. 



The alternative method of beginning with the pure solvent and gradually adding 

 weighed quantities of a stock solution, seemed, on the whole, better suited to the 

 case. The conductivity of the actual sample of solvent used for the solutions could 

 then be determined for each series of observations, and, since the addition of the 

 successive amounts of stock solution could be made in the electrolytic cell itself, the 

 necessity of pouring the liquids backwards and forwards would be obviated. This 

 not only eliminated errors due to the taking up of impurities, but also effected a 



