Habit of Igneous Rocks

Differentiation.

The process by which a magma is split into a variety of partial products is known as Differentiation; its importance from the standpoint of theoretical petrology is very great ; it is in point of fact, the fundamental problem of petro genesis. The variation expressed may appear in two ways ; either as a variation in a single rock body, or in an associated series of separate intrusions or extrusions. In the first case a single in trusion shows variation in different parts of its mass, the extreme varieties being usually connected by gradual transition. The con trasted parts are frequently arranged symmetrically with reference to the borders of the mass forming concentric zones in bosses or laccoliths suggestive of a differentiation in situ connected with the cooling of the mass. Most commonly the marginal zones are of more basic composition. Excellent examples of variation of this kind are provided by the gabbro mass of Carrock Fell (Cum berland) with its basic border rich in iron ore, and the shonkinite laccolith of Shonkin Sag, Montana, in which a transition from syenite to shonkinite becoming denser outwards, occurs, the shonkinite finally passing at the margin into a fine leucite-basalt porphyry. Variation in an associated series, is exemplified by the succession of lavas emitted from a volcanic focus. These may differ considerably from one another. Thus in the Berkeley Hills near San Francisco the volcanic succession is a repeated series of andesites, basalts and rhyolites. In other cases the lavas emitted are much more varied, and while no significant relations in the succession can be discerned, it is to be remembered that a significant order may be obscured by the overlapping of the flows from neighbouring volcanoes. In the simpler cases the order of eruption is one of increasing divergence from an initial type.

Plutonic Complexes.

A series of plutonic rocks intimately associated and localized at a centre constitutes a plutonic com plex. Such complexes often contain a great diversity of petro graphical types ranging from peridotite and gabbro through diorite to acid granite. The succession of these intrusions follows an order of decreasing basicity, the later and more acid rocks usually oc cupying the greater part of the complex, the earlier, and more basic being subordinate in amount and restricted to the borders of the mass. Excellent examples of such plutonic complexes are provided by the early Devonian intrusions of "newer granites" of the Scottish Highlands (e.g., Garabal hill, Loch Lomond). What then are the factors operating which led to differentiation? The possible processes leading to heterogeneity in a magma may be considered under (a) those occurring in the liquid prior to crystallization and (b) those occurring during or subsequent to crystallization. Of the first, we may consider (i.) differences of composition set up in a liquid due to a temperature gradient; (ii.) differences of composition due to a pressure gradient; and (iii.) differences of composition due to the separation of distinct liquid phases. Each of these processes has in the past been ap pealed to to explain the variation seen in an igneous rock mass intruded singly, or in a rock series arising as a result of succes sive eruptions or intrusions. The type of composition-variation due to a temperature gradient is commonly known as the Soret effect and has been especially applied to explain the variation seen in a single rock body, such as the concentration of the minerals of early crystallization towards a cooling boundary. In

a liquid with a temperature gradient, for dilute solutions the con centration varies inversely as the absolute temperature ; and it has been thought that by the operation of this process, substances near their saturation point might accumulate at the cooler sur faces. Both theoretical considerations and experimental results clearly indicate however that the actual Soret effect is in fact negligible and such effects as are actually observed are due in reality to departures from the laws of ideal solutions, for if both solvent and solute obeyed osmotic pressure laws there would be no relative concentration of one with respect to the other. The effects of a pressure gradient appear to be of the same order of magnitude and both these processes are now in effect abandoned as factors in differentiation.

The effects possible as a result of the separation of immiscible liquid phases stand on a different basis. Many liquids, homo geneous at high temperatures, separate with falling temperature into two or more non-consolute fractions, and the hypothesis that igneous magmas form such immiscible fractions has been favoured by some petrologists, notably Rosenbusch, Backstrom and Daly.

From his own studies Vogt concluded that the rock-forming silicates are freely miscible. The only known case among mag mas for which immiscibility can be claimed is that of sulphide containing silicate melts. Such a magma unmixes at a tempera ture above the region of crystallization of the usual sulphides, and liquid sulphides, especially those of iron, are separated as liquid drops. These, owing to their greater density, sink to the floor of the magma collecting as a distinct sulphide layer, or, in some cases, the liquid layer may be injected into the surrounding rocks. These sulphide aggregates form important ore deposits. It is among lavas which as a group show the various stages of quenching of liquid magmas that the evidence for immiscibility might be expected; the absence of lavas containing glassy glob ules of composition distinct from that of the main mass of lava, goes far to reassure us that immiscibility is not an operating factor in petrogenesis. Recent experimental evidence of immiscibility in silicate liquids has indeed been obtained, but the composition range of liquids which show it is unlike that exemplified by rock magmas. This limited miscibility is found in mixtures of silica with any of the oxides CaO, MgO, FeO, Fe203, but in a region of very high silica-content. The minimum temperature for the exist ence of two liquids is in each case only a little below the melting point of silica, or in the vicinity of 1,70o° C. The oxides show no such immiscible region in their silica mix tures, and moreover it requires only a small proportion of any of these miscible oxides to render miscible with silica the oxides which are themselves immiscible with silica. No natural magmas are known which approximate in composition to the region of immiscibility in these melts. It is true that natural magmas con tain in addition water, and that its effect is experimentally un known. Nevertheless if its addition changes the limits of immis cibility so that unmixing becomes a possible factor in differentia tion, its behaviour must differ from that of the oxides studied.