•740 



under its own vapour-pressure touolies BC in H, the tangent in 

 n and the tangent in h to the curve, going thi-ongh these points, 

 Avill yet be almost parallel BC. 



Because the equilibria T'^-j- ^'Quid « -f- vapour ??i and i^'+ liquid 

 A + '^'^pourAj ditfer but \ery little from one another, the perspective 

 concentrations aS' and S^ (see the previous communication) will be, 

 on addition of a third substance, also approximately equal. Therefore, 

 when in the one equilibrium Sy> S^, this is also the case in the 

 other. Of couise the same applies to *S<>Si. Now we distinguish, 

 according as the substance expands or contracts on melting, two 

 principal cases. 



1. F expands on melting {V^v). The point H is theji situ- 

 ated with respect to F as in fig. 4 — 6 (XI) viz. between F and C, 

 but close to F , A V is negative between F and H, positive in the 

 other points of BC From the situation of n and // with respect to 

 F, it follows that .S and S^ are both positive. We distinguish /S> >Si 

 and >S <[ S^ . 



a. S^S^. As hV is positive in h and negative in n, it follows 

 from our previous communication that the pressure decreases from 

 h along the saturationcurve under its own vapour-pressure and it 

 increases from n. In which direction shall ihis curve now proceed 

 from h? As the tangent in h coincides almost with BC, the curve 

 must go from h either almost in the direction towards 7i or almost 

 in opposite direction. We tind the first iji fig. 5, the second in 

 fig. 6 [XI). Ill order to determine this direction, it is to be consi- 

 dered that the region L — G shifts on decrease of pressure from h 

 towards 7i, so that the pressure decreases in this direction. As 

 the pressure along the saturationcurve under its own vapour-pressure 

 must also decrease from h, this curve must therefore, also go from 

 h almost in the direction towards n. It has, therefore, from h a 

 direction as in fig. 5 (XI). As the tangent in n coincides almost 

 with BC, the curve must go from n either almost in the direction 

 towards h or almost in opposite direction. Considering that the region 

 L — G shifts on increase of pressure from n towards h, so that the 

 pressure increases from n towards h and further that the pressure 

 along the saturationcurve under its own vapour-pressure must also 

 increase from n, we see that this curve must go, therefore, also 

 from n almost in the direction towards //. 



The saturationcurve under its own vapour-pressure has, therefore, 

 a form as curve h n in fig. 5 (XI) ; it is situated, therefore, close to 

 the side BC and it disappears at 7// in the point H. 



b. S <^ Sy In a similar way as above we find that the pressure 



