(. 285 ) 



tlic other side of Ihc isobars, it would no longer be possii)le to draw 

 two tangents, and the loens tor whieli, willi regard to Z^^, tiie value 

 of /".,/ is 0, \vonld have but one branch. 



Now, howe\er, the point /\. is variable, first beeanse the volume 

 of the solid body (le[)ends on the pressure, and secondly when the 

 ('oneentration siionhl change. This enhances of course tlie ditKiculty, 

 if we wished to determine this locus. Ihit this w ill not detract from 

 the thesis that for the contact-curve, when it ascends from low 

 pressure to high {)i-essure, twice r.,/- is 0, when tlie solid body has 

 a smaller volume than it would have in tlnid form at the same 

 temperatui-e and luider the same pressure — and that oidy oiice 

 Vgf is in the opposite case. When Pg is variable, the locus for 

 which Vsf^zi), is construed by drawing from every special position 

 of Ps the tangents to the isobar of the pressure of 7*s, and by joining 

 the points of contact obtained in this way. 



If the contact-curve does not pass through the plait, the value of 

 Vg,f is negative for the [»oints outside the two branches of the locus 

 ?'.>,/■ = 0, and positive for the points inside. 



If however the contact-curve passes through the plait, the value 



of r.,/- is more complicated. In the tigure the two tangents have I)een 



drawn to the isol)ar BEDD' E' B' , Ps being supposed to be in the 



position that corresponds to the pressure of this line. In this case 



too the value of /'«ƒ is negative for the points lying outside the two 



[)oints of contact. For the points between the points of contact we 



cannot assume Vg/ to be positive, however. This holds only till 



the points D and D' are reached. Between D and 1)' , i\f is again 



negative, and the transition from positive to negatixe takes place 



in the points D and J)' through intinitelv great. 



f '^'^\ 

 In the same way the value of - — - \ is complicated for the 



l)oints of a coutact-cur\e, ])assing through the plait. I have stated ihis 

 already in "Ternary systems" I, Proceedings February 22"<'- 1902 

 footnote p. 45(]. For the points between the connodal aiul the spinodal 

 curxe this quantity is still positive ; for the points between the 



siiiuodal and the cui'\e for w hich - — is 0, it is negative ; whereas 



for the |t(»inls inside tins hisl curxe il is again positive. This last 

 transition from negative to positixe takes place through intini- 

 telv ii'reat. 



, , ^^/' 



Let us write the equation tor the determination of — in the fol- 



dxf 



lowing form : 



