wherry: compositions of minerals 113 



come more or less crystalline, for this has evidently been the his- 

 tory of most of this class of minerals. 



If the original colloids have compositions approaching those of 

 definite crystallizable compounds, then the change to a meta- 

 colloid may be regarded as a simple molecular rearrangement. 

 But if adsorbed constituents are present in excess over the 

 requirements for such compounds, what becomes of them? Ob- 

 servations bearing directly on this point are lacking, but some 

 analyses of meta-colloid minerals are certainly on record in which 

 the presence of foreign material is shown. It is only necessary 

 to turn to garnierite (Dana, System, p. 677), chloropal (ibid., p. 

 701) and dufrenite (ibid., p. 797), all of which often look crystal- 

 line (cryptocrystalline) and homogeneous under the microscope, 

 to find examples of this. It would therefore seem that in some 

 cases, at least, adsorption may occur even in meta-colloids, so 

 that it can not be regarded as safe to consider crystallinity alone of 

 minerals, whose features are those of meta-colloids, as a certain 

 proof of their definiteness of composition. 



Solid solution. Especially in substances appearing in distinct 

 crystals it is quite generally assumed that, except for isomorphous 

 replacement,^ the composition is fixed, constant, and definite. 

 Yet even here there is evidence that some variation may occur. 

 Two cases have recently been discussed elaborately: nephelite, 

 which often contains 2 per cent more SiOo than is required by its 

 simple formula without evident effect on its crystallographic 

 constants, optical properties, or any other features;^ and pyrrho- 

 tite, in which the amount of sulphur can vary from 36.3 per cent 

 corresponding to FeS, to over 40 per cent, with only slight changes 

 in axial ratio and specific gravity, and without loss of crystallinity 

 or characteristic properties.^ 



But other well crystallized minerals have yielded analytical 

 results exhibiting even greater variability in the amounts of con- 



^ The subject of isomorphous replacement is fully treated in the text-books, 

 and need not be discussed here. In one sense it is, of course, only a special case 

 of solid solution, but this term is here used only for non-isomorphous combina- 

 tions. 



^ Foote and Bradley, Amer. Jour. Sci., 31: 25, 1911; 33: 439, 1912. 



* Allen, Crenshaw, Johnston and Larsen, Amer. Jour. Sci., 33: 169, 1912. 



