( 233 ) 



of this longitudinal plait, which has then already greatly extended, 

 with the branch plait takes place much more in the neighbourhood 

 of the line 1, 2 of the three phase triangle, so that after the meeting 

 the plait assumes the shape drawn in tig. 4'', which makes it for 

 the greater part retain its proper character of longitudinal plait. So 

 at first increase of pressure makes the phases approach each other 

 (this portion may be exceedingly small, but as a rule it will exist); 

 then further increase of pressure makes the phases 1 and 2 again 

 diverge, till .I'l and a', approach to limiting values at ^^ ^ x, without 

 the longitudinal plait ever closing again — as was formerly considered 

 possible [cf. inter alia van der Waals, Coiit. II, p. 190 (1900)]. 

 For in consequence of the minimum at D the longitudinal plait 

 always encloses the point C^. Only at temperatures higher than T^y 

 at which the longitudinal plait does not yet exist, there can be 

 question of homogeneity till the highest pressures. But then the 

 plaitpoint P belongs to the branch plait of the transverse plait, and 

 not to the longitudinal plait. This is indicated among others by 

 fig. 3'', after the closed connodal curve in M has broken through 

 the connodal curve proper of the transverse plait ; or by fig. 3*, 

 before a longitudinal plait has developed round Cq. 



Of course we may also meet with the case, that the plait round 

 Co coincides with the branch plait at the moment that the latter 

 with its plaitpoint just leaves tlie transverse plait, as shown in fig. 5^, 

 but this involves necessarily a relation between S and -t, and is 

 therefore always a very special case. Then the branch plait happens 

 to leave the transverse plait exactly in the minimum at D. After 

 the meeting the plait shows the shape as traced in fig. 5'^ Now 

 increase of pressure causes the two phases 1 and 2 to diverge /Vö»i 

 the beginning. 



But the longitudinal plait round C^ may also meet the connodal 

 line of the transverse plait, before the closed connodal line has got 

 outside the transverse plait (fig. 6"). Then the three phase equilibrium 

 does not develop, as in fig. 4^^, at the transverse plait (from which 

 a branch plait issues), but at the longitudinal plait round C^. The 

 latter penetrates then further into the transverse plait, till its meets 

 the isolated closed connodal curve in D (fig. 6''), after which the 

 confluence with it takes place in the unrealizable region (fig. 6^). 



This plait is then the longitudinal pjlait proper, of which there is 

 generally question with mixtures of substances which are not miscible 

 in all proportions. But we should bear in mind that just as well 

 the above treated case of fig. 4 may present itself, with that of fig. 5 

 as transition case. 



