DILUTE SOLUTIONS AT THE FREEZING POINT. 353 



It seems, therefore, that for an accurate comparison with satisfactory freezing 

 point determinations, it is, as we supposed, necessary to use measurements of con- 

 ductivity made at 0. This result justifies the amount of trouble and time expended 

 on the present investigation, and the comparison with Mr. GRIFFITHS' promised 

 measurements seems likely to give important information. 



In comparing the diagrams for different substances, several results become evident. 

 The normal type of curve is apparently that shown in the cases of potassium chloride, 

 barium chloride, &c., in which the ionization increases as dilution proceeds and 

 eventually reaches a maximum. On further dilution, the curve remains a horizontal 

 straight line, and it therefore seems right to conclude that the ionization has become 

 complete. 



The abnormal shape of the curves for acids, observed when their conductivities are 

 measured at 18 or 25 in glass vessels, is seen to still represent the facts when the 

 conductivities are measured at in platinum. At extreme dilution the equivalent 

 conductivity falls off at a very rapid rate, and the cause of this fall, whatever it may 

 be, is not removed by avoiding the use of glass or by taking the observations at the 

 freezing point. 



It has been usual to explain this phenomenon by action between the acid and the 

 impurities present in the water ; the result of the action being to reduce the effective 

 quantity of acid in solution by an amount which becomes appreciable at great dilution, 

 the substances formed having a smaller conductivity than the acid. The present 

 experiments, however, seem to furnish a certain amount of evidence against this view. 



The quantity of such impurity must be very small, and it is probable therefore 

 that any action between it and the acid would be complete after the addition of 

 the first quantity of acid, the amount of which must be large, reckoned in chemical 

 equivalents, as compared with the total amount of impurity. Now, if the action 

 were complete at the first addition of acid, it would be possible to correct the 

 results of the other additions for the conductivity of that portion of the acid put 

 out of action by the influence of the impurities on the first instalment. We ought 

 to be able thus to get a curve agreeing in its general form with those of the normal 

 type, which reach a maximum as the dilution is increased, and show no signs of fulling 

 off on further dilution. The conductivity of the first solution, when reduced in the 

 usual way, gives a value for the ionization of '809. Let us calculate the conductivity 

 required to raise this figure to TOGO. It is k = 11 '29 X m, 11*29 being the maxi- 

 mum value of k/m, corresponding to complete ionization (see p. 340), m is 3*254 X 10~ s , 

 so that k should be 3*673 X 10~ 4 . It is, however, actually found to be 3*221 X 10~ 4 , 

 so that the destruction of conductivity amounts to '452 X 10" 4 . Adding this 

 value in the case of the second and following solutions, we get iouizations 1*006, 

 1*016, 1*017, &c., numbers much too high to give a horizontal curve. 



In this connection it is interesting to study the case of potassium permanganate. 

 This salt is one which is particularly likely to react with the residual impurities of the 



VOL. cxciv. A. 2 z 



