( 146 ) 



/ vdp\ 1 



We mmlit also write a value for or — , but we will 



dilate the coefficient only indirectly from: 



T y^ dv\ / dp\ a 



or 



0,413 X 0000 = 27 pk 

 or with approximation : 



■9' 



1,6 = 



which agrees with c = 



l-f2-' 

 1 



The value of /? calculated according to our relations may there- 

 fore be considered to be at any rate approximately accurate. 



Yet it remains strange that for the liquid volume itself a calculation 

 according to our suppositions yields a value which is much too small. 



According to a table in Cont. I 2°^ p. 172 the liquid volume for 



temperatures which do not differ mucli from — Tf; is equal to 0,8 A„. 



Even if we take into account that /^ <^ //,, we cannot diuiiuish the 



factor 0,8 to less than 0,7. 



We have then the equation 



0,7 h, = h, (1 -f 2c) 



or 0,7 n = 1 + 2c. 



1 

 With n = 2, this yields c ^ -^ , which does not agree with the 



5 



value — , which we must assume for c, as we saw above. I haxe 



not yet been able to investigate, what modification must be made 

 in the relation assumed for h; e.g. to put « =i 1,8 or to suppose />„ 

 really to be smaller at low temperatures. If we sup})Ose b^ to be 

 a function of the temperature, then the calculations become very 

 intricate and difficulties of another kind arise. Therefore I prefer to 

 regard the above considerations as conducing to point out that 

 everything shows that b must really increase with v. 



Let us investigate what consequences of general nature follow from 

 this variability of b. In the first place we obser\'e that the three 

 real values of ?; for given temperature and given pressure cannot 

 be calculated any more by means of an equation of the third degree. 

 The equation of state namely may assume a very intricate form if 



