The separation of liquid from crystals can be accomplished not only by zoning but also by the sinking of crystals from the upper layers of the liquid. The resulting liquid is enriched in albite and, under favourable circumstances, the final liquid may approach the albite-diopside eutectic of composition 97% Ab, 3% diopside. The analogy with the differentiation course of gab broic magma is clear. A mixture rich in diopside and anorthite rich plagioclase, corresponding to an anorthite-rich gabbro or eucrite may pass to an assemblage rich in more acid plagioclase with subordinate diopside, corresponding to an augite-diorite. We shall see by a study of another system how this differentiation from a basic magma may be extended to the granitic (diopside granite) stage.
The continuous reaction relation between crystals and liquid in the systems just studied is of fundamental importance in the crys tallization-history of magmas. This reaction is common to all solid solution series, and such solid solutions are called continuous reaction series. Very similar effects, however, are produced by another type of crystallization phenomenon revealed in synthetic melts. In this case an early precipitated phase reacts at a certain temperature with the surrounding liquid and a new phase is pre cipitated, a reaction process of a discontinuous kind. This reaction is illustrated in the system forsterite (olivine)-silica. Here there are three solid phases, forsterite, clinoenstatite and cristobalite (Si02). The monoclinic pyroxene, clinoenstatite, has an incongruent melting point and dissociates at 1,557° into for sterite and liquid. From all melts of composition between Mg2SiO4 and MgSiO3, the first phase to crystallize is forsterite, which continues to separate until the temperature of 1,557° is reached, when the olivine reacts with the liquid to form clinoen statite. When equilibrium is established the resulting crystalline mass consists wholly of olivine and pyroxene. If, however, the opportunity for reaction is inhibited, either by the sinking of olivine crystals or by their enclosure in the reaction product, the course of crystallization is extended and the third phase, silica, is precipitated giving a mass composed of residual olivine, pyrox ene and silica.
A reaction relation of this type may effect a series of compounds in a crystallizing melt. These, arranged in their order of succes sion, constitute a discontinuous reaction series. The combined effects of discontinuous and continuous reaction are revealed in the related ternary system diopside-forsterite-silica. Here cli
noenstatite forms a complete series of solid solutions (monoclinic pyroxenes) with diopside ; they form a continuous reaction series, the crystals in equilibrium with the liquid being relatively enriched in the magnesian end member The effect of hindering the reaction relations is now twofold, the residual liquid (or upper layers of the liquid if sinking of crystals obtains) is en riched in silica and the diopsidic pyroxene, owing to the offset in composition. The lowest point of formation of liquid in the system (there is no ternary eutectic) is the binary eutectic diopside-silica.
Evidence of these reaction processes in natural magmas is by no means wanting. It is seen in zonary banding of plagioclase felspars and pyroxenic solid solutions. One of the most charac teristic features of olivine-bearing basic rocks is the occurrence of olivine crystals with rounded outlines—indicative of resorp tion—or with mantles of pyroxene formed at their expense. In the synthetic melts the unidirectional reaction-relation existing among the mineral phases results in a differentiated liquid or chain of liquids enriched in silica, from early separation of oli vine, in albite from early formation of magnesian-rich pyroxene. Similar reactions proceeding in basic (gabbroic or basaltic) magma serve to explain in large measure the differentiation chain of the calc-alkaline rocks, gabbro-diorite-granodiorite, though here the presence of other other volatile con stituents is responsible for more complex reactions and new min eral phases not yet included in synthetic investigation.