252 PROFESSOR JOHN GIBSON 



too small to permit of each set being treated separately, and as it is consequently 

 impossible to judge of their relative accuracy, equal weight was given to the several 

 values when calculating the constants a and b for these salts. 



The percentage differences for K 2 S0 4 are remarkably small, showing a close 

 agreement between the two sets of determinations. 



In the case of (NH 4 ) 2 S0 4 the two sets of determinations were treated separately, as 

 they each include four points within the range of concentration to which equation (1) 

 is in this case applicable, that is, from about 1*5 normal up to 5 '528 normal, this being 

 the highest concentration given. Within this range there is a satisfactory agreement 

 between the calculated and the experimental values for A M . The constants are not the 

 same, however, in the two sets, so that the straight lines obtained by plotting A M against 

 r do not coincide. Were the constants a and b calculated from Kohlrausch's more 

 concordant determination applied to the four determinations by Klein, three of his 

 determinations would appear too low. 



Now there are two points, one (r = 1*513) determined by Kohlraucsh, and the other 

 (r = 1*425) by Klein, which have very nearly the same concentration. Interpolating 

 for r= 1*425, and using the constants calculated from Kohlrausch's determinations, 

 A M is 67*3, while Klein's observed value is 66*0, the difference being 2 per cent. 

 Similarly, interpolating for V = T513 and using the constants calculated from Klein's 

 values, A>, is 65*6 while Kohlrausch's observed value is 66*8. Here the difference is 

 1 *8 per cent. The concentrations being nearly the same, there can be no doubt of the 

 discrepancy. It is therefore clear that the limit of error must be at least 0*9 per cent. 

 It is probably greater. 



Interpolated Points. 



Many of the values given by Kohlrausch and Holborn were obtained by graphic 

 interpolation and not directly from observation. A reference to Table A will show 

 that these interpolated values, when translated into mass units, agree in most cases 

 extremely well with those calculated by equation (l). Thus 6 out of the 9 values 

 given for AgN0 3 (20 per cent. — 60 per cent.) are interpolated values, and the differences 

 in no case exceed 0*3 per cent. This cannot be said, however, of the interpolated points 

 for KF and CdSG 4 . 



As A M = As and T = — , the relationship between equivalent conductivity and con- 

 centration may be expressed in volume units by the equation 



A = - + -2 m (2) 



s s 



Tins equation may be used in place of graphic interpolation for the ranges of 

 concentration to which equation (l) applies. 



Fig. 2 shows the graph for KF in volume units with Kohlrausch's interpolations, 



