Of all the processes suggested as important factors in the dif ferentiation of rocks, crystallization alone seems competent to produce important results. Indeed fractional crystallization is now believed to be the fundamental process in rock differentia tion. Either the localization of crystallization or the localized ag gregation of crystals is sufficient to produce heterogeneity in a rock magma. The first case, the localization of crystallization, is competent to explain those cases of differentiation in situ exem plified by the basic marginal phases of intrusive masses. The mechanism of the process has been much discussed. Examination of igneous rocks early showed that minerals crystallize in a more or less definite order. In general the first minerals to separate belong to a group known as the minor accessories : this includes zircon, apatite, titanite and iron-oxides; then follow in order olivine, augite, hornblende, biotite, plagioclase, orthoclase, micro cline and quartz. To this rule there are many exceptions, but the succession given above holds in the great majority of cases. Expressed in this way the more basic minerals precede the less basic : it is known as Rosenbusch's law of decreasing basicity. Accordingly it is supposed that at a cooling boundary there is a precipitation of minerals of early crystallization (basic minerals), the continued growth of many crystals being maintained by dif fusion. The completion of this process results in a crystalline mass with a border phase enriched in the minerals of early crystal lization.
The rate of diffusion in silicate melts has been the subject of laboratory study and the results obtained show that the possible effects are exceedingly small and incompetent to account for the formation of border phases about large intrusions. The marginal phase indeed appears to represent rather a chilled phase having a composition close to that of the original magma, the more acid phase which it encloses representing a differentiate formed by much slower cooling of the remainder of the magma. The rela tive movement of crystals with respect to the liquid from which they separated may be of two kinds, (a) movement under the influence of gravity—sinking or floating of crystals and (b) move ment by straining off or squeezing out of residual liquid by earth movement. Both of these processes can be applied to explain the more important case of differentiation exemplified by an as sociated series of intrusions or extrusions. In a crystallizing magma, provided viscosity does not inhibit the free movement of crystalline material, as the first formed crystals are denser than the molten material in which they form, they should sink under the influence of gravity in the liquid. The lower regions of the rock mass into which the crystals sink become enriched, the upper regions impoverished, in the constituents of which they are formed. The sinking of crystals in silicate melts has been both experimentally verified in the laboratory and observed in nature. Darwin long ago (1844) suggested that this process was a potent factor in producing diverse rock types. Turning to the evidence of consolidated rock masses themselves, we find intrusions in which such gravitational segregation is indubitably exemplified.
Thus the concentration of early formed olivine crystals near the base of sheet- or sill-like intrusions has been observed in the case of the quartz-dolerite sill of the Palisades (Hudson river) and the picrite sill of Lugar (Ayrshire). Examples of this nature
appear however to be exceptional, and indicate an extreme fluidity of the liquid magma. Most sheets or laccoliths show no hetero geneity throughout the vertical extent of their mass. In other cases where a variation is observed, it is to be ascribed to suc cessive injection of magmas of varying composition.
In the magma basins of the deeper parts of the crust, the case may be conjectured to be otherwise. The retarding influence of viscosity, which appears to prevent any notable gravitative segre gation in intrusive sheets cooled comparatively rapidly in those parts of the crust laid bare by denudation, is at deeper levels counteracted by the time factor. The extremely slow rate of cool ing in these deeper regions may permit significant settling of crystals, giving rise to a differentiated reservoir more or less stratified according to gravity, more basic at the bottom, more acid upwards. The mechanism by which it is conceived a varied group of liquids becomes drafted off from an intercrustal reser voir and injected to form the igneous rock bodies now made visible by denudation is, of necessity, one into which the element of speculation largely enters. Remelting of successive stratified layers by a gradual use of temperature from the base upwards, together with earth movement, the straining off of residual liquid from a crystalline meshwork under the influence of crustal stress, all are processes to which appeal is made to explain not only the variety among associated igneous rocks injected as plutonic complexes but also the chronological succession of the intrusions themselves.
Assimilation.—Fragments of foreign rock frequently become incorporated in rock magmas, and the mutual reaction between solid and liquid may lead to heterogeneity or the development of new varieties of rock. This process is known as assimilation and has been claimed by some petrologists as an important factor in petrogenesis.
It seems probable that the heat effect connected with the solution of solid rock by a liquid magma is always negative, solu tion being accompanied by absorption of heat usually of the order of the latent heat of melting. Simple solution demands large amounts of heat which are only available if the liquid possesses great super-heat. Intruding magmas, however, can scarcely be at a temperature much above their crystallization. range. Such effects of magma upon its enclosures, as are ob served, must therefore be attributed to reaction and precipitation, which in some cases leads simply to an adjustment in the compo sition and relative proportions of the existing phases, but is also productive of new minerals foreign to the magma, especially if reaction be incomplete. Where igneous rocks have absorbed sedimentary material in any quantity they present distinctive features. Granites which have assimilated shales or slates usually contain minerals such as andalusite, sillimanite or cordierite, which are foreign to normal igneous rocks. Gabbros under similar condi tions pass first into norite by the formation of rhombic pyroxene and anorthite at the expense of monoclinic pyroxene and eventu ally into cordierite norites. Rocks of this character are well de veloped in Aberdeenshire. In general it may be said that there is little reason to believe that foreign rock is essential to the pro duction of any particular type of differentiate, or that the process of assimilation is an important factor in petrogenesis.