216 ON A DIAGRAM OF FREEZING-POINT 



the ratio mentioned really does become constant as dilution 

 increases, the method is likely to give coefficients with too low 

 or too high values, according as the ratio at moderate dilutions 

 diminishes or increases with dilution (it was found to increase 

 with KC1 and K 2 SO 4 .) For it will probably become constant 

 within the limit of error of observation, before it has really reached 

 constancy. And if it changes with dilution in a slightly wavy 

 manner, even though on the whole tending to constancy, it may 

 be regarded as having become constant, when really passing 

 through a maximum or a minimum point. 



Deguisne's* observations on the variation of conductivity 

 with temperature between 2C and 34 have enabled me, by the 

 method just mentioned, to make rough determinations of the 

 ionization coefficients at in some cases, on the assumption that 

 his empirical constants might be used down to 0. According to 

 Deguisne's observations, the ratio of the conductivities at and 

 18 usually changes gradually down to dilutions of 1,000 litres 

 per gramme-equivalent, and between that and 2.000, undergoes 

 rapid change. As observations at great dilution are attended by 

 considerable difficulty, I have assumed that these sudden changes 

 were probably due to errors of observation. If they were not, 

 my Deguisne coefficients (for which De'guisne himself is of course 

 not to be held responsible) may be considerably out. 



In some cases, 1 have obtained coefficients from the above 

 data by extrapolation, in order to make use of available depres- 

 sion data. In such cases I have plotted, side by side, ionization- 

 coefficient-concentration curves, for both and 18, using values 

 for 18 based on Kohlrausch's conductivities, and I have then 

 produced the curve beyond the limit of observation, under the 

 guidance of the 18 curve. 



I have used all the accessible observations of depression in the 

 case of the electrolytes for which data were available for deter- 

 ing the ionization coefficients at 0, including observations by 



* Temperatur-CoefHcienten des Leitvermogens sehr verdiinnter LSstingen, Dis- 

 sertation, Strassburg, 1895. See also Kohlrausch u. Holborn : Leitvermogen der Elek- 

 trolyte, Leipzig, 1898. 



