( 608 ) 



have heen doloniiiiUMl. Ncillicr docs SCI, iiiidcr.tio d<>c()iii|M»si(lon ;d 

 these h)\\ leuiperaliii-es. If, lheretoi-e, it were [(ossililc lo isolnic \)\\\v 

 SCI4, all iiHa<i,iiial)le niixliii-es ihereof with ('1, and S.,CI., oi- willi 

 botii inighl he ohlained and the fiirlhei- porlioiis of tjo'. 2 c(\}\\(\ then 

 be eoiistnu'led. This, h()\ve\ei-, is not \\\o ease. 



On aecount of the extreme minnteness of its er\ stals and the low 

 temperature at w liieh crystallisation lakes |»lace, the eompoMnd S(/l^ 

 canjiot he isolated in a pure condition and caiinot, therefore, be 

 addetl in detinite proportions to (^l.^ or S.^Cl.^. \]\ ex[)osin<i' various 

 mixtures of these c(nnpouu<ls to temperatures of 0^ and hiiihei- it is 

 ])0ssible to cause the voluntary foi'uiation of those (piajdities ofSCl, 

 Avliich are in ecpiilibrium with ('1., and S,('l., at the temperature 

 chosen. As these do not cliange much with the t(Mnpej'atui-e, the 

 same series (tf nuxtui-es is fairly \vell i-etained on cooli]]"' to the 

 (uystallisation temperatures. The numbei* of ternai-y mixtui-es whose 

 crystallisation may be in\esti<iated is, therefore, restricted to one for 

 each gross comj)osition. 



We, therefore, always obtain for the first solidifyiuü' poijits a series 

 sueh as AH'P]M\'I) in fig. '2, just as if we were dealing with a 

 binary mixtnre in which a dissociable com[)ound occurred. If, ho\vevei', 

 the equilibrium in the liquid does not change Avhen CI3, SCl^ or 

 S.^C1.^ crystallises out, the system oii solidilication must behave as a 

 ternary one. ' 



This shonld have appeared in practice by the fact that the curve 

 of the first solidifying |)oints did not also represent the path of 

 crystallisation, and, therefore, the mother liquoi- left after cooling 

 each mixtnre to a lower temperature differed in composition from 

 that given by the point on the curve corresponding with the 

 tenq)erature in (piestion. Owing to the nature of the ci-ystals, 

 however, it was inqiossible to remove the mother-licpior with a 

 pi])ette. 



It was, however, possible to prove in the case of a li(pnd of the 

 gross composition SCl^ that this does not conqdetely solidify to SCl^ 

 at E', Init only commences to deposit SOI., at that temperature 

 ( — 30°), which quantity increases on further cooling, but in such a 

 manner that no comi)lete solidilication takes place below — 80°. In 

 the points on both sides of E' the continuity is still more pronounced 

 and this makes the accurate detei-mination of even the first solidifying 

 points very difticult. 



In the second place the solidilication of the dilFerent mixtures 

 ought not to be conqdete in H' or in K', but only in I after traversing 

 the crystallisation paths II'l and Iv'I. 



