748 



own vapour-pressure, going through H, but also those which are 

 situated in the vicinity of H are parabolas. 



In tiie point H of figs. 5 — 6 (XT) L W is negative, when // is 

 situated on the other side of F, A W is positive. From (28) it now 

 follows : 



when the curve, touching in H is situated outside the triangle 

 [fig. 5 (XI)], it shifts on decrease of T, within the triangle [curve 

 lin in fig. 5 (XT)] 



when the curve, touching in H is situated within the triangle 

 [fig. 6 (XT)], it shifts on increase of T" within the triangle [the 

 closed curve in fig. 6 (XT)] and on decrease of T partly outside 

 the triangle. Therefore, curve hn of fig. 6 (XI) must be imagined 

 to l)e closed by a part hn situated outside the triangle; this part, 

 however, has no physical meaning. 



Ill fig. 1 three curves are drawn through F ; Fl is the liquid- 

 curvi o'" the region L — G at the temperature Tp and under the 

 press... V. Pf, therefore at the minimummeliingpoint of F\ FK is 

 the boilingpointcurve and Fs the saturationcurve under its own 

 vapourpressure. The two first curves are but partly drawn. We 

 now construe in F a tangent to each of these curves. With the 

 aid of the formulas from the ]n-evious communication, we find : 



for the tangent to the liquidcurve (7^/) of the region LG: 



^'1=-^^:^^ I- . . . . (29) 



for the tangent {FZ^) to the boilingpointcurve (I^K) : 



(-l-l\RT + {y,-^)s-RT- 



'1) =-^ ^ -- ^^r^U^.-— (30) 



dxjk~ (i/i— /5)^ \dxji B {y^—^)t 



and for the tangent {FZ^) to the saturationcurve under its own 

 vapourpressure {Fs) : 



C 



(^-l-l^RT + {y-^)s-RT- 



+ ^-7 -1^. (31) 



dxjs (.Vi— ^)* \dxji A iy,—l^)t 



Now we take again the most probable case that BC — AD is 

 positive (communication IT). That there may be agreement with the 

 figs. 5 and 6 (XI) and fig. 1, we take V^v therefore A positive. 

 As further y^ — /? is negative, we can deduce : 



n>(i)>(ji- *-) 



