( 245 ) 



a. When tlie hilcjit liojit of iiiixiiiü,' in the solid |)h<ise a' =^ <j^ ,i' 

 is (jn'dt, the solid pliase eoiitaiiis but \ory little of the second eoni- 

 ponent. The portion of the nieltin<i,p()int-cui've which may he realized, 

 has a course as in fig. 1 (see the phxte). The curves 2^=zf{x'), viz. 

 Aa and Bb show maxima at m and //, which maxima descend 

 gradually for smaller values of {^' till they are below a aiul Ji, the 

 maximum at n sooner than that at in. (fig. 2). [We leave for the moment 

 out of consideration what happens below the horizontal line through 

 the ])oint C\ the entectic i)oint : for this see my preceding communicati(»u]. 



I). For smaller values of /•?' we get the case of fig. 3, where the 

 branch BC shows a miniinwn, no longer below the temperatui-e of 

 C, but exactly at i\ Immediately after (i.e. when /?' is still somewhat 

 smaller), the meltingi)oint-curve assumes a shape as in fig. 4. C 

 remains the eutectic point, where the two branches of the melting- 

 point-curve meet with a break. As appears from the figure, we 

 have now got parts of the meltingpoint-curve, which may be realized, 

 also below the point C (see also fig. 14 and 14^r of the communi- 

 cation referred to). 



It is however very well possible, that in the meantime the minimum 

 at D has already disappeared, and then we get a course as is 

 represented in fig. 5 (observed i. a. by Hissink for mixtures of 

 AgNO, and NaNO,. (see also fig. 146 I.e.). 



c. For still smaller values of /?' the curve T = f {x') becomes 

 continuously realizable. The points b and a coincide in a point of 

 inflection b,a with horizontal tangent (fig. 6j, Avhicli point of inflection 

 soon passes into one with an oblique tangent L (fig. 7), while in most 

 cases it disappears afterwards altogether for still smaller values of 

 i?'. (fig. 8). 



The break at C has disappeared in the case of fig. 6 and from 

 this moment there is no longer question of a eutectic ])oint, and 

 the meltingpoint-curve assumes the perfectly continuous shape of 

 fig. 7 and 8. 



d. As has already been observed in b, also the minimum at D 

 will sooner or later disappear. For very small \alues of ^' we get 

 then alivays a course as in fig. 9. 



Observation. As has been elaborately demonstrated in the preceding 

 paper, a maximum at A for normal components can 7ierer occur 

 with positive values of the different absorbed latent heats of licpie- 

 faction and mixing (see p. 156 I.e.). When such a maximum is 

 observed, as was done c. g. by F. M. Jakgkr M for two isomeric 



1) Akademisc.h Proefschrift (19()B), p. 173-174. 



17 



Proceedings Royal Acad. Amsterdam. Vol VI. 



