CONCLUSIONS. 63 



experimental fact. Neither can it be objected that Krister's solutions 

 were not sufficiently dilute to reveal the true relation, for the observa- 

 tions upon naphthaline and ,5-naphthol have been repeated by 

 Bruni* with very dilute solutions of one of the components in the 

 other, and completely verified. 



Now, the laws of solutions hold for solid solutions even for moder- 

 ately high concentrations (Bodlander) when the components are not 

 isomorphous, and on the other hand, even liquid crystals, when iso- 

 morphous, follow Kuster's rule more nearly than the law of solutions. 



An extended discussion of existing data from this standpoint would 

 involve us in unnecessary detail ; but there can be no question that 

 Kuster's rule represents the data which have been gathered upon 

 isomorphous mixtures at least approximately while the laws of 

 dilute solutions appear to fail of application there. On the other 

 side, the rule admits of no independent theoretical derivation. Van't 

 Hofff suggests that judgment be suspended pending the accumula- 

 tion of further data and intimates that the close similarity of chemi- 

 cal composition and molecular structure in compounds which form 

 isomorphous mixtures gives them an unusually close inter-relation, 

 and their influence one upon the other may render a simple theo- 

 retical treatment very difficult. 



Our case is especially interesting when considered from this stand- 

 point, but it distinctly emphasizes the difficulty rather than helps 

 toward its solution: (1) Although the chemical reactions of albite 

 and anorthite are not of such a character as to prove or disprove a 

 close analogy between them, a comparison of their formulas certainly 

 does not suggest an isomorphous relation. If their formula weights 

 represent true molecules, they possess the same number of atoms to 

 the molecule (NaAl S13O8, CaAl 2 Si20 8 ) and the group Si20 8 in com- 

 mon, but the remaining atoms taken separately are not mutually re- 

 placeable. (2) The melting points of the components in the feldspar 

 series are very far apart more than 300 while Kuster's organic 

 mixtures were all included within a narrow temperature interval 

 (2 to 56 ). For reasons which will appear presently, both GarelliJ 

 and Roozeboom have pointed out that the farther apart the melting 

 points of the components the less probable is the linear relation. 

 (3) The homogeneity of the solid phase is established within 1 per 

 cent by the optical examination of the slides. Moreover, separate 

 chemical analyses of the solid and liquid phases of the mixture Abi An 5 



*G. Bruni, Atti della reale Accademia dei Lined, 5, vn, p. 138, 1898. 

 f Van't Hoff, Vorlesungen ub. Theoret. u. Phys. Chem. (Braunschweig, 1901). 

 Part II, p. 64. 



\ F. Garelli, loc. cit. 



