( 63 ) 



considered as the most important ones, because they can be the 

 subject of experimental investigation. Though it is necessary for a 

 clear insight that for a simple substance we know that below certain 

 temperature the isotherm possesses unstable pails, and that we can 

 indicate the limits of these unstable parts, yet the determination of 

 the points of coexisting equilibrium is of the greatest importance 

 tor the experiment. In the same way it is, indeed necessary for a 

 clear insight into a binary mixture that the existence of the unstable 

 phases and their limits are known, mi the spinodal curve: but 

 the knowledge of the binodal line is of still more importance, and 

 to determine the latter must lie taken as the final end of all con- 

 siderations, because it can constitute the subject of experimental 

 investigation, and the results derived from our considerations can 

 only he tested by experience in so far as they refer to the binodal 

 line. If we are to admit an exception to this rule, this applies to 

 the plaitpoints to whose existence could be concluded without an 

 examination of the binodal curve being necessary. But moreover, it 

 deserves attention that not even the whole of the binodal line can 

 be realised by the experiment. The binodal line can possess portions 

 lying in the unstable region, and others which are metastable. This 

 has already been observed in the Theorie Moleculaire (Cont. p. 14\ 

 but appears in an ampler and more complete measure from the 

 diagrams occurring in These Proc. .March and June 10(15. At the 

 same time it appears there how very complicated the binodal line 

 can be, when the spinodal curve hardly deviates from the usual 

 shape. Hence if the more or less complexity of a plait is to be judged 

 according to its spinodal curve or according to its binodal curve, a 

 very different opinion will be arrived at. 



Thus paying attention to the properties of the binodal curve I 

 have been able to speak of a main plait and a branch plait in 

 the last cited paper. In the same way, regarding only the binodal 

 line and its nodal lines, we may speak of a transverse plait and a 

 longitudinal plait, whereas, regarding only the spinodal curve, we 

 shall have to consider these two as one single plait. However, to 

 prevent confusion, it is desirable to follow one and the same termi- 

 nology. At the moment it seems most desirable to me to consider 

 particularly the spinodal curve when choosing the name, leaving 

 that part out of account that may also sometimes exist, but which 

 then encloses the concave-concave part of the ip-surface. If no plait- 

 point exists on the spinodal curve, or only one and then a reali- 

 sable one, such a plait might be called a normal one. If besides 

 there are a couple of heterogeneo u s plaitpoints found, we 



