( 589 ) 



curve in M" begins to get outside, i.e. outside the connodal line 

 proper of tlie principal plait, at which the phase 3 begins to split 

 up into 3 and 1 just as in iig. 4. This splitting- up itself is repre- 

 sented at ti in tig. 9. A three phase equilibrium has formed then 

 just as in tig. 5. The shape of the different connodal lines is still 

 quite the same as in the analogous case in tig. 5, only the plaitpoint 

 P of the principal plait had already disappeared there. This course 

 has alread}' been given by Kortf.wkg, as was mentioned above, and 

 VAN DER Wa.\ls, too, has accepted it in one of his last papers (loc. 

 cit.) on the transformation of a principal plait into a branch plait 

 and the reverse. 



The three phase equilibrium established is however not of long 

 duration as we shall see. At still somewhat lower temperature /, a 

 very interesting transformation takes place (see fig. 10), also men- 

 tioned by KoRTEWEG (loc. cit. p. 318, tig. 34), and later by van der 

 Waals (I.e.). The small letters a, b, c, d and a', b', c\d' placed in fig. 9 

 give a clear idea of the transformation. 



Still somewhat lower, at t^ (fig. 11), the plaits liave reversed liieir 

 functions; the branch plait of fig. 9 has become a principal plait, and 

 reversely the principal plait has been transformed into a branch 

 plait. We may notice that the "tail" at b is always turned to the 

 side of the principal plait, both in tig. 9 and in fig. 11. Also the 

 "ridge" has changed its place after the transition of fig. 10. 



And then the further transformation resumes its normal course. 

 There comes a moment, at /, (represented in fig. 8), that the isolated 

 connodal line of fig. 11 begins to retreat within the connodal line 

 proper of the principal plait. This takes place in J/', and the three 

 phase equilibrium, which accordingly has been of very short dura- 

 tion, finishes. The two phases 1 and 2 ha\'e again coincided, and 

 after this we have only coe.\istence of 3 and 2, as before, and as 

 with type II before ^1/ in the neighbourhood of R^. The plaitpoint 

 P of the principal plait continues to exist for some time more, but 

 will soon also disappear (at C\) ') Also the closed connodal line 

 remains past J/' still for a short time within the connodal line 

 proper, gets smaller and smaller, anil disappears at last at R^, where 

 the spiuodal line touclies the plaitpoint line once more (tig. 8 at t,). 

 The temperature t^„ is the lower critical temperature of the two 

 components, that of L\, and at still lower temperatures we begin 

 gradually to approach the second plaitpoint line L'„A. 



') Tlie temperature of R\ (and M') may also be lower than that of Co. This 

 really occurs for the above mentioned mixtures. The point P of the principal 

 plait has then already disappeared before 1 and 2 coincide at M'. 



