( 448 ) 



case, namely that iji Aviiicli the molecules of the admixture, both 

 with i-e^ard to vohime and to attraction, behave towards tlie mole- 

 cules of tlie principal component exactly as if tliej' were identical 

 witii these latter molecules. 



If at the same time a.^ = a^ , b.^ = h^ , wiiich is of course not 

 involved in the above suppositions, it is easy to see that at deci-ease 

 of temperature below the critical temperature the plait would suddenly' 

 appear all over the whole breadth of the tp-sui-face. 



Now it is true that every deviation from these equalities a^=n^, 

 h.^ = />! will prevent such a way of aj)pearance, but it is evident 

 that then the behaviour of j)laitpoint and corresponding plait will 

 depend on a.^ and A.^, i.e. the tirst approximation for vvhicli the 

 knowledge of x and y is sufticient and which everywhere else 

 suffices to make this behavioiu- known to us up to a certain distance 

 from the side of the i|7-surface, fails here. 



And also ali-eady in the neuilihourliood of the combination of the 

 values x=I,y = l, this first approximation will be resti'icled, to 

 the innnediate neigidwurhood of the point A' ami of the ci'itical 

 temperature 7^- of the principal compoiuMit. When we are not in 

 that immediate iieighbourhood the intluence of n.. and h.,, — of the 

 former of these quantities specially, — will soon be felt. On the 

 contrary for values of x or 7 sufficiently differing from nnify the 

 considerations derived from the first approximation will j)rol)ably be 

 of foi'ce within ])i'etty wide limits, at least in a (pialitative .sense. 



5. Before proceeding to a di.scussion of the border curves between 

 tho different regions, we will shortly point out that we camiot 

 attach an equally great importance to all the })arts of the graphical 

 representation. So all points lying left of the y-axis relate to negative 

 values of ^a^, i.e. to the case that the molecules of priiuM})al com- 

 ponent and admixture should repel each other, which is not likely 

 to occur. 



In the same way the negative values of y, so of /^j, of the points 

 below the >j-axis, should be cousidere<l as having exclusively mathe- 

 matical signification. If the relation, />,= — (/^j-j-Aj, should still be 



applied also for very unequal values of the //s, then y would even remain 



1 1 



always larger than — and so the part below^ the line y = — 



would lose its physical signification. 



6. With regard to the border curves between the different parts, 

 we have first to deal with the parabolic border curve separating the 



