102 



Studies on Solution. 



TABLE 64. KI in Abso- 

 lute A IcoholConduclances 

 in mhos at 25. 



original equations. By reference to the tables quoted above we can 

 observe how such values are derived. It is to be noticed that alter- 

 nate A values are equated. This is done so that the difference may 

 be of sufficient degree of magnitude and that any inaccuracy in an 

 individual measurement may not affect two successive derivations. 



A glance at the tables and calculations will show that the calculated 

 values of A are by no means concordant. The higher the value of A 

 used in the equation, the lower becomes the calculated A . His final 

 conclusions are vague and inconclusive. The value chosen for A must 

 be regarded as only approximate; it was usually the highest possible. 



Goldschmidt seems to have overlooked the very exact and admirable 

 piece of work done on the subject of the limiting conductance and 

 degree of ionization of alcoholic solutions by B. B. Turner 1 in the 

 Johns Hopkins laboratory. Turner carried his 

 dilutions to far greater limits, as table 64 illus- 

 trates. We have repeated this work and have 

 every reason to believe that it is unquestioned 

 and is remarkably accurate, especially when one 

 considers that it was done without the more re- 

 cent conductivity apparatus now at our disposal. 



Turner showed that up to 5,000 liters dilution 

 it is easy to obtain concordant results; but the 

 values for A as calculated according to the 

 Kohlrausch method are not constant for these 

 higher dilutions. Like those of Goldschmidt, 

 they decrease the higher the values of A used 

 in the equation. Turner also showed that plot- 

 ting A against the reciprocal of the cube root of 

 the volume does not give a straight line as in 

 aqueous solutions of equal dilutions, but rather 



a smooth curve slightly convex towards the dilution axis. He there- 

 fore assumed that the Kohlrausch method fails to answer the require- 

 ments of absolute alcoholic solutions. Extrapolation of his results with 

 the formula would give us a value of 56 for A instead of the experi- 

 mental value of 48.5 obtained. He thought that accidental introduc- 

 tion of water into his solutions might affect the readings, and to test 

 this he added as much as 0.2 to 0.3 per cent of water by weight to his 

 alcoholic solutions, with a variation in conductivity of only 0.01 X 10~ ( 

 units, showing that no accidental experimental error of this nature 

 had crept in. 



Furthermore, Dutoit and Rappeport 2 showed identically the same 

 phenomenon with a number of inorganic salts in work to which reference 

 has already been made (page 99). This work, like that of Turner, 



'Amer. Chem. Journ. 40, 558 (1908). 



2 Journ. chem. Phys. 6, 545 (1908). 



