60 K. D. Kraevitch on an Approximate Law of the 



temperatures the divergence increases, especially on the side 

 of a rise in temperature. This signifies that aqueous vapour 

 in a state of saturation follows Boyle's and Gay-Lussac's 

 laws near 70°, and diverges from these laws at a certain 

 distance from 70°. Other causes still further increase the 

 divergence. Thus c— c ± should remain constant according 

 to our assumption, instead of which both specific heats, 

 especially c, vary with the temperature. The volume of water 

 (w) is taken as equal to zero, which in reality is never the 

 case. These facts have their significance, especially for high 

 pressures, because with water c at low temperatures varies 

 very little, and the volume of liquid (w) is then minute com- 

 pared to the volume of vapour (y) . For this reason, probably, 

 the theory and experiments agree better for low pressures 

 than for high. 



8. The determination of x and y from experimental data 

 may lead to the idea that these quantities are arbitrary, and 

 that therefore formula (5) or, which is the same, (7), can be 

 numbered among empirical formulae. Such a conclusion 

 would, however, not be right ; x and y have each a perfectly 

 definite physical value, and calculated by the above methods 

 prove, at least y does, in the case of water and of other 

 liquids subsequently considered by me, to be near those 

 values given by experiment. If there be not an entire 

 agreement, it may be ascribed to the insufficient accuracy 

 of the methods by means of which the heat of vaporization 

 and specific heats are determined, while the calculation of 

 these quantities, or, which is the same, of x and ?/, depends 

 according to our theory upon the vapour-pressures, which 

 may be measured with incomparably greater accuracy. This 

 even affords the possibility, although for one temperature T 

 only, of obtaining these quantities for substances which have 

 not been investigated in this aspect. The inexactitude of 

 x and y determined by experiments from the quantities c—Ci 

 and r, moreover, proceeds from the fact that we do not know 

 accurately the values of A, D, and the temperature of absolute 

 zero. We may add that if, instead of the calculated values 

 of c — C\ and r, we take those which are given by experiment, 

 then the tensions calculated after equation (5) approximate 

 to those given by observation, although not so closely as the 

 same values determined by the above mentioned methods. 

 Thus, although the quantities x and y are calculated like 

 arbitrary constants in empirical formulae, they are not arbi- 

 trary, because they have a definite physical meaning and 

 approximate to those values which are given by experiment, 

 and can even be replaced by the latter in equation (7). It 



