( 232 ) 



both with type I in the line CoG, and with type II in the line Co^, 

 and that tiierefore with decrease of temperature a separate plait 

 begins to detach itself starting from C^ at a definite temperature 

 Tq (the plaitpoint temperature in Co), which plait will merge into 

 the main plait (or its branch plait) later on in an homogeneous 

 double point. The consequence of this is, that with type I e. g. at 

 lower temperatures the main plait will always be open towards 

 the side of the small volumes, so that increase of pressure will never 

 cause the two split phases to coincide. 



Let us , however specially consider the case of type II. Here the 

 usual course, inter alia described in the last cited paper in the 

 Proceedings of Dec. 30, 1905, is this. At a certain temperature, 

 passing from higher to lower temperatures, a spinodal curve touches 

 the branch of the plaitpoint line AC^ in R^. In the well-known way 

 a closed connodal curve begins to form within the connodal line 

 proper, wiiich closed curve gets outside the original connodal curve 

 at lower temperatures, and gives rise to a new (branch) plait, and at 

 the same time to a three phase equilibrium (figs. 3'' and 3^*). In many 

 cases this branch plait has already appeared before the plait which 

 starts from Co, begins to develop at somewhat lower temperature. 

 Later on the two branches coincide (at the minimum temperature 

 in D), and then form again a continued branch plait (fig. 3'). ^). 



Now for the special case h^ = ^^ the point D lies always very 

 near Co (see the paper in these Proceedings referred to under b. 

 in § 1). If then e.g. Tj7\ = T/„ then r,„/r„ = 0,96, when Tm 

 represents the temperature in the minimum at D. The real longi- 

 tudinal plait round Co exists then only at very high pressures- 

 (fig. 3^), while the ojien plait of fig. 3'- can hardly be called a 

 longitudinal plait, but is much sooner to be considered as the 

 branch plait of the transverse plait which has joined the original 

 longitudinal plait. Increase of pressure makes here always the two 

 coexisting liquid phases approach each other, unless with very high 

 pressures, when these phases diverge again. 



The calculation proves that in the quite general case b^ ^ b, the 



point D may get much nearer in the neighbourhood of R^, and also 

 that the temperature in the plaitpoint Cq may be comparatively high, 

 so that in opposition to what has been represented in fig. 3" the 

 longitudinal plait has already long existed round Co before a three 

 phase equilibrium has formed at M (fig. 4" and 4^). The meeting 



') In this and some other figures the spinodal curves seem to touch in the 

 homogeneous double point D, instead of to intersect, as they should. 



