813 



s and r, from which all the otlior quantities can then be calculated. 

 For r we could then have assumed 2,12, and we had then found 

 X^ : ^^ = 1. But it is better to start from two quantities, which are 

 experimentally determinable. And for the present s and ƒ' seem the 

 most suitable, even though s depends on the accurate determination 

 of the critical density (which is often very difficult, and generally 

 takes place through the strictly speaking unpermissible prolongation 

 of the so called "straight diameter" — unpermissible, because this 

 straight diameter exhibits a perceptible curvature close to the critical 

 point), and though the determination of f' (supposing we may put 

 ƒ' ^/) is from the nature of the thing always connected with not 

 very accurate calculations of pressure and temperature differences 

 close to the critical point. 



We might e.g. have assumed .s^ := 3,77 and ƒ = 7, and we had 

 then also found ^^ = a„ r= 0,98. And the equality of A^ and X^ 

 would also have become clear for other substances, as Oxygen, 

 Argon etc., where another set of values for s and ƒ had been 

 started from.^ 



We shall, however, immediately see thai in consequence of the 

 new relation found by me instead of f' and .s- another quantity can 

 be introduced, which in contrast with the two mentioned ones is 

 experimentally pretty sharply definable. We mean the direction of 

 the straight diameter, which can be determined very accurately from 

 observations even far below the critical temperature, and is at any 

 rate not affected by any uncertainty in the observations close to the 

 critical temperature. T^ and Vk will, indeed, of course be of influence 

 in the determination of the "reduced" coefficient of direction. 



If, as said, we take, however, for the present f' and s, we find 

 easily from the above relations : 



Further from ƒ — 1 = 27 : A, r' and ƒ = {X, -. x,) X (8 : (r — 1)) : 

 X,= 

 through which 



27 s / s\ , 27 / s\ 



Aj s 



(y) 



Finally we shall find : 

 l—b'k — — -^ 



r- 



In this particularly the last relation, viz. for /3'V-, is remarkably 



2s(f'—l) vkh"k /'— 4 



