[MACGREGOR] DEPRESSION OF THE FREEZING-POINT 11 
ionization coefficients determined in this way for a series of mixtures 
containing two electrolytes, which have all the same concentration with 
respect to one and different concentrations with respect to the other, 
may be expected therefore, if plotted against the concentration with 
respect to the second electrolyte, to give a wavy curve. When the 
ionization coefficients given below for the chloride mixtures are treated in 
this way they are found to give such wavy curves, and it therefore seems 
probable that the method referred to is somewhat unsatisfactory. 
Instead of using Mr. Barnes's coefficients in calculating the depressions of 
his mixtures therefore, I have endeavoured to get better values by 
interpolation. In the case of each series I have plotted his coefficients 
against the concentrations of the mixtures with respect to the electrolyte 
present in variable amount, together with the values which the co- 
efficients would tend to have, if the concentration of the mixture with . 
respect to such electrolyte were diminished indefinitely. These values 
were obtained from the observations on simple solutions, directly in the 
case of the electrolyte present in constant amount, and in the case of the 
electrolyte present in variable amount, on the assumption that as its con- 
centration diminishes to zero its ionization coefficient will tend to the 
value it would have in a simple solution of concentration equal to the 
concentration with respect to the other. I then drew smooth curves 
through the points thus obtained, and read off from the curves what 
seemed to me more probable values of the coefficients. The corrections 
thus applied were in no case greater than about 4 per cent, and in most 
cases were smaller; but my own impression is that larger, in general 
positive, corrections would be justifiable. The corrections affect the 
calculated values only very slightly, usually improving the agreement 
between observed and calculated values. Larger positive corrections 
would still farther improve this agreement. In the tables given below 
the corrected values of the coefficients used in the calculations are given, 
as well as Mr. Barnes’s values, the former being inclosed in brackets, 
The values obtained by interpolation for the coefficients for solutions of 
zero concentration with respect to the electrolyte present in variable 
amount, are also given and are also inclosed in brackets. 
