598 PETROLOGY AND MINERALOGY 



as something novel, had already long been accepted by petrographers 

 as a matter of course. We have long known that while the lava 

 magma is a molten solution of varying composition, and that the 

 chemically identical, homogeneous, firm, amorphous product of its 

 solidification, which forms as soon as the molecular mobility of the 

 magma is lost, and before crystallization sets in, i. e., the natural glass 

 corresponding to the magma, cannot be other than an under-cooled 

 solidified solution. 



We no longer assume the natural silicate fusions (Silicatschmelz- 

 fliisse) to consist of substances dissolved in a solvent of definite 

 stoichiometric composition, but regard them as being, probably, 

 mutually dissociated solutions. Speculations upon the nature of 

 solvents, which have always been of most problematical origin, are 

 thus rendered futile. 



Bunsen already emphasized the fact that the same laws control 

 crystallization from molten solutions as control crystallization from 

 aqueous solutions. There can be no doubt that Gibb's Phase Rule for 

 aqueous salt solutions also controls solidification from molten solu- 

 tions. However, the presence of many compounds dissolved in the 

 magma introduces complications which will make it difficult to apply 

 the rule to the order of crystallization. 



The order in which the individual constituents of a uniformly 

 granular eruptive have solidified, or, to speak more accurately, the 

 order in which they have begun to crystallize, is an old petrographic 

 problem of the first rank. To.-day no one may deny that this suc- 

 cession is normally controlled, as Lagorio has shown, by the character 

 of the bases, and is not, as Rosenbusch believed, according to increas- 

 ing acidity. The least soluble substances separate out first and the 

 most soluble separate last. It has been shown experimentally that 

 the descending order of solubility in molten silicate solutions is as 

 follows: Iron oxide, magnesia, lime, soda, potash, and alumina, the 

 last entering relatively late into the molecule of the various constitu- 

 ents, and finally silica itself. Yet there are hundreds of well-verified 

 instances where the corresponding mineral series iron ores, olivine, 

 and rhombic pyroxene, monoclinic -pyroxene, amphibole and biotite, 

 anorthite, lime-soda feldspars, nepheline, albite, and segirine, ortho- 

 clase, quartz - - has not been adhered to, either through an inversion 

 of order or by the contemporaneous crystallization of minerals which 

 should have separated successively. 



Two facts alone seem to be well established. First, in those silica- 

 rich rocks which contain quartz, the latter mineral as a rule is among 

 the last to solidify. Second, the minerals -- apatite, zircon, rutile, 

 titanite, ilmenite, and perofskite -- containing those compounds, 

 such as phosphoric, zirconic, and titanic acids, present in the magma 

 in the smallest and even scarce traceable quantities, are the very 



