( 139 ) 
§ 6. Starting from the first approximation for the virial coefficients 
given in § 5 we iooked now for small variations in the virial co- 
efficients so that the latter while keeping the agreement with the 
isothermal, which they have to represent, allow of a better deve- 
lopment in terms of the reduced temperature. And so the isothermals 
were mutually smoothed. The weak point in this method, when it 
is extended over more substances is the imperfect agreement of the 
reduced equations of state, further increased by uncertainty in the 
eritical data. But it is important to see how far we can advance under 
these circumstances. 
In eaiculating the variations A By, AC4, ADs, AEs, AF, which had 
each to satisfy each of the isothermals and which therefore had to 
be mostly sought out in the same way as By, C4, Da, Ey, Fa were 
calculated, it appeared desirable also to modify slightly the form of 
the development in series with regard to the reduced temperature. 
We put: 
1 2 
t ra i 
t 
\ 
| 
1 2 
t t 1 
C=, t+ te +¢,te “+c, — 
t 
1 
PS I) 1 = 
D=d,t +d, te} Eat yore ge ke (IV) 
ote 
1 
Bep t 1 1 
EH et + eo te Aas Faas 
| 
1 
ks ik DR | 
S=fit + fz te len Cn 
We find for the 20 coefficients, which cover the range of the 4 
substances the following 
