TO THEIR STATE OF IONIZATION MACGREGOR. 241 



differences between the values of (P 1 P 2 ) /n will be somewhat 

 greater. As the (I k)'s in this case, however, are more than 

 twice as great as in the case of thermal expansion, the differ- 

 ences in the values of (P l -P 2 )/n, if expressed as percentages 

 of one of them, will be smaller than in the case of thermal 

 expansion. Accordingly we find from Bruckner's observa- 

 tions, 



For n = 0-5 1-0 1-5 2-0 2-5 



(P!-P a )/n - -0 2 116 2 122 -0 2 126 -0 2 128 '0 2 135 



For surface-tension I k for NaCl is 0'096 and for KCi 

 (H16. The approximation to constancy (judged by the per- 

 centage criterion) will thus not be so close as in the last case. 

 Bother's observations give, by graphical interpolation, 



For n = 1-0 1*5 2-0 



( p i-P 2 )/ = '016 -0113 -0105 



For refractive index k for NaCl is +0'0054 and for 

 KCI +0-0091. Thus the values of (l k)&a/&n for the two 

 salts are much more nearly equal than in the case of the other 

 properties and consequently the differences in the values of 

 (?! P a ) /n will be smaller than in the case of the other 

 properties. Bender's observations give for the D line, 



For n = 1-0 2'0 3'0 



(PjL-P^/n = -0 3 29 -0,17 '0 3 24 



If the value for n = 2 be omitted from consideration, as being 

 probably in error, (Pj P 2 ) /n is seen to be more nearly con- 

 stant so far as absolute magnitude is concerned than in the other 

 cases considered. As the values of (?! P 2 )/n however, are 

 small, their differences when expressed as percentages are com- 

 paratively large, and the approximation to constancy, viewed in 

 this way is less than, e. g., in the case of density. 



The above account of this phenomenon may be further 

 tested by the aid of Kohlrausch's observations of electrical 

 conductivity ; for in this case I k is the molecular con- 

 ductivity at infinite dilution (usually written fi.). The 



