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



Each value of ^r can De calculated four times : — (1) we may 



regard it as corresponding to the pressure p^ followed by three 

 other pressures, p 2 , p S) and p 4 , and therefore calculate it from 



the first of equations (11) ; (2) we may suppose that -j- corre- 

 sponds to the pressure p 2 which is preceded by one value p li 

 and followed by two, p s and p 4 , and therefore the differential 



coefficient must be calculated as -^ •> i. e. from the second of 



at 



the equations (11) ; (3) when the pressure^) is preceded by 



two and followed by one pressure, then -j- must be calculated 



from the third of the equations ( 1 1 ) ; and, lastly, (4) the differential 



coefficient is calculated from the equation for ^p> If the 



at 



figures forming a table follow some definite law, then the values 

 of the differential coefficient, although determined by different 

 methods, should be similar and differ only in the inevitable 

 errors of observation ; or, more strictly speaking, of calcu- 

 lation, because all tables of vapour-pressures are found by 

 means of interpolation formulas. If the differences are great 

 it shows that in one figure, at least, there is a misprint or 

 error exceeding the possible one. The tables of vapour- 

 pressures which I investigated nearly always gave four very 



clj) 

 nearlv similar values of ~ Considerable discordances en- 



dt . 



abled me to discover misprints or errors in the tables. I may 

 here remark that irregularities in the values of a, calculated 

 by the first method, also sometimes revealed an errov in 

 observation. 



In all the calculations the results of which are given below, I 

 did not calculate the differential coefficient four times, but only 



twice by means of the formulas -~ and -~ , and took the 



J at dt 



arithmetical mean of the results. 



Knowing how to find the differential coefficient -|-,it is 



dt 



possible to calculate the heat of vaporization r for each tem- 

 perature according to equation (10). Let us endeavour to 

 determine the temperature with whose uniform rise or fall 

 the value of r also uniformly varies, so that if the tempera- 



