402 Mr. R. A. Lehfeldt on the 



and (11) becomes 



/ Bq Bkq \1 1B^_ Q a3) 



Vbg + A Bfy + A/^pB? ' " ' ^ } 



the integral of which is 



l°g(? + g)- lo g(? + g^)+ log^ = const. 



We have, however, the two conditions that when q=0, 

 p = 7r B , and when ^ = 00 ,p = 7TA, hence we can evaluate both 

 k and the integration constant, and we get 



Bg7r A + Air B 



B? + A ~ P ' ^ j 



and *=- A (15) 



As the ratio between the partial pressures of the vapours is 



B7T 



in this case 5= -» — -q, it is clear that the two terms of (14) 



A7T B 



represent the two partial pressures respectively, or 



Bq A 



P^BqTA^ Ps= Bq~TA^' 



This result can be tested by Brown's table (loc.cit. p. 561), 

 in which are given the boiling-points of various mixtures of 

 benzene and carbon disulphide : using the vapour-pressures 

 of the pure substances at the temperature of this boiling- 

 point, the calculated value of p should be 760 millim. This 

 is not the case : thus at 50° C. p works out to 713 millim. 

 It is probable therefore that Brown's generalization is not 

 sufficiently close an approximation for our purpose. 



Guided by this result, we have, then, to choose some other 

 probable function for t and s. It must satisfy these condi- 

 tions : 1st, when only one substance is present in the liquid, 

 only the same can be present in the vapour, so that 



when 2 = £ = 0, 

 when q = cc t = co ; 



2nd, if t is any function of q then 1/t must be a similar function 

 of 1/q, for obviously it can make no difference which of the 

 two substances we choose for A and which for B. The 

 simplest function that satisfies these conditions is t = kq r , where 

 k and r are constants. 



