Spin-". 



Fijt. 25. 



On the left of 



dp <f s i|? 



— = and at smaller volumes also — = has 



dx d.v* 



been drawn. Where this curve passes through 



dx 1 



:0 it has tangents 



// y-axis. The spinodal line has split up and the two parts have been 

 drawn far apart for the sake of lucidity. One part surrounds 



d*ip dp 



- — = 0, and the other part touches — = in the point in which 

 ax' dv 



dp 

 this curve is intersected by — =0. Further a o-line has been drawn 

 dx 



with two points of inflection. The right-hand point of inflection is of 



no importance for our case. And finally the detached branch of the 



locus of the points of inflection of the g-lines has been drawn. Now 



cPv 



<Pv 



and = intersect, is to be 



dx' 



too the point in which 



</.<-•> <~ g 



expected on the left side of the spinodal curve, which has got 

 detached. But for this case v. ^> r, for all nodal lines of the longi- 

 tudinal plait, and the second plaitpoint is really to be expected on 

 the left. I suspect these two ways of detaching to be connected with 



d*q 



the two series of values of x q , tor which = disappears in the 



dx* 



Proceedings Royal Acad. Amsterdam. Vol. X. 



