( 630 ) 



with the smallest volume has now become double point. I have, 



however, omitted the drawing of this ease 1. because most likely 



the case does not really occur, and 2. because the drawing may 



easily be found by reversing the preceding one. There are e.g. with 



the solution of salts in water cases which on a cursory examination 



d?a 

 present some resemblance with the assumption — — negative, but 



dx 



which yet are brought about by influences perfectly different from 



the fact of a negative value for — — . 



dx 



d % a 

 Such a diagram for the case negative, though, would quite 



fall in with the right side of fig. 1. As in the given figure Tk 



increases with x on the right side, and there is a maximum value 



d*a 

 of Tk oil the supposition — , fig. 1 might be still extended to the right 



dx 1 



till such a maximum Tk was reached. But then we should also have 



to suppose that a value of x could exist or rather a mixture for 



d*a 

 which at a certain value of x the quantitv — reverses its sign. 



' dx* 



Every region of fig. 1 of certain width which is taken parallel to 



the r-axis can now be cut out for «,-{-«, — 2a„ positive, to denote 



the course of the isobars. Regions on the left side indicate the course 



of the isobars for mixtures for which with increasing value of b the 



critical temperature decreases — regions on the right side for mixtures 



for which with increasing value of £ the critical temperature increases — 



the middle region with the complicated course of the isobars when 



there is a minimum Tk. The left region would be compressed to an 



da 

 exceedingly small one if we wished to exclude the case — negative 



dx 



da 

 or — = 0. We do so when putting a ls = \/a x a % . On such a suppo- 

 sition a minimum Tk is still possible, but the left region must then 

 have an exceedingly narrow width. There is, however, no reasonable 

 ground for the supposition a x a t = a ri \ There would be, if the quantity 

 a for the different substances depended only on the molecular weights, 

 and so a = sm 2 held for constant value of 8. If the attraction, just 

 as with Newton's attraction, is made to depend on the mass of the 

 molecules, and so if we put a, = e^ 8 , and also a 3 = *= 2 m 2 2 , it appears 

 that t x and f a are not equal. If we now put a 13 = |/a x a a , we put 

 a i 3 = wijWjj/fjf,. What reasonable ground is there now for the sup- 



