This page needs to be proofread.
PETROLOGY
85


appear, and the last process will probably be the formation of a ternary eutectic at which the temperature remains steady till the liquid disappears and finally the completely solid mass cools down.

As a very simple case we may take the diagram in fig. 4. A, B and C represent the three pure substances. AB, BC and AC represent mix- tures of two components. Ear, Ebc and & represent the three binary eutectic mix- tures. The ternary eutectic is represented by Eat*, con- taining about equal quantities of C and B and a smaller amount of A. If a represent the composition of a certain liquid which is allowed to cool and crystallize, composition will change along the line A &. At a certain point the A component will crystallize out and the liquid will become poorer in A (richer in B and C), and the composition will change from a towards b (away from point A). After a time the liquid becomes satura- ted for b which will start to crystallize, and now the liquid -changes composition along the line bE a kc as the temperature continues to fall; finally C also begins to crystallize, and the ternary eutectic point is reached at which the three components crystallize simul- taneously in definite proportions (represented by the position of Eotc) until it is completely solidified.

It is probable that nothing quite so simple as this occurs in ordinary rock-forming minerals, at least the silicates, but some metallic alloys show this type. In considering silicates the following considerations must be kept in mind : (a) The liability to form com- pounds, which behave as new substances with their own fusion- points and eutectics. (b) The occurrence of isomorphous compounds is almost universal and these form mixed crystals unstable in the changing magma, and liable to resorption (this may upset the formation of a ternary eutectic altogether), (c) Compounds may appear at an early stage which subsequently become unstable and are replaced by different minerals (incongruent). (d) Many silicates refuse to crystallize in ordinary crucible experiments (except in presence of solvents which do not appear in the final product).

We must also keep in mind that in the crystallization of rocks certain conditions prevail which may modify the process to an un- known extent. Thus: (a) All magmas contain gases of various kinds which may have a very powerful influence in determining what minerals will form, (b) Intrusive magmas are under great pressure and the pressure diminishes as they rise to the surface; the pressure may act directly or by increasing the concentration of the gases dissolved in the magma, (c) Cooling in deep-seated mag- mas is extremely slow. This will tend to prevent supersaturation by undercooling and lessen the chance of the abnormality in the

sequence of crystallization which may appear in rapidly cooled melts. It will also favour the complete transformation of early un- stable minerals into stable permanent forms. Many varieties of minerals have already been obtained experimentally which are not known to occur in rocks. They are stable only at high temperatures (and possibly under low pressures).

As an example of the effect of isomorphous minerals on the se- quence of crystallization we may take a mixture consisting of 50 % diopside and 50% plagioclase (containing equal proportions of albite and anorthite). The composition diagram (fig. 5) is a triangle with each mineral at one of the corners and the mixture is repre- sented by a point (F). Crystallization begins with a separation of diopside (supposed to be a simple mineral and not an isomorphous mixture, as it would usually be in rocks) at about 1275. At 1245 the excess of diopside (G) has separated out, and felspar begins to crystallize. It has about 75% anorthite (H). Thereafter diopside and felspar both crystallize, but as the temperature travels along the line EGD from G to K the composition of the felspar changes from H to L (if we suppose that all the early felspar which is unduly rich in anorthite is stage by stage absorbed). The resulting rock has the mineral composition above stated; but if resorption of felspar is incomplete the last-formed felspar is richer in albite and has a composition T. The felspar crystals in that case are zonal with basic centres. If at any time crystallization is suddenly brought to an end, a glassy ground-mass will be formed, which is richer in soda and silica than the original magma and contains zoned felspar crystals. This is exceedingly like what takes place in many basaltic lavas. Again, if the original mixture had been richer in felspar, so that the composition point lay below the line DE, felspar would have crys- tallized out first. This seems to be in keeping with the structure of many dolerites, which contain felspar partly enclosed in augite


FIG. 6.

crystals of later formation (ophitic structure), while others show that the augite appears in porphyritic crystals and began crystallizing before felspar. Another interesting feature of this diagram is that there is no tertiary eutectic point, and the liquid residue continually changes in composition up to its final disappearance.

The phenomena of these component systems are extraordinarily varied. One of the best known is the system AUOj-CaO-SiOj which has been very fully tested at the Geophysical Laboratory in Wash- ington by Shepherd and Rankin. A copy of their diagram is given here (fig. 6). It is divided into fields, of which six are occupied by substances known to occur as minerals, cristobalite, tridymite, wollastonite, anorthite, sillimanite and corundum. In each of these fields the mineral named will crystallize if the temperature of the melt falls. The fields are separated by lines which show under what circumstances the two minerals whose fields adjoin will crystallize. Where three fields meet, the conditions exist at which three minerals will exist simultaneously (or, to express it otherwise, are in equi- librium with a liquid of the composition indicated). In no case do four fields meet in one point.

This system is also of much interest to technologists desiring to understand the chemistry of the manufacture of Portland cement. This is a mixture of lime, alumina and silica, with a fairly definite composition, and the compounds which form on fusing or sintering the mixture are indicated by the diagram. Similarly, the CaO corner of the figure shows what is the result of heating lime containing a little alumina and silica (rmpure limestone) to a very high tempera- ture. Silica is also a refractory mineral and is used in silica-bricks and ganisters for lining furnaces. A little lime and alumina are mixed with it (either naturally or expressly to obtain certain results), and the behaviour of such mixtures is indicated by the appropriate corner of the ternary scheme. These investigations accordingly are of the greatest value in many industries such as pottery, steel- making, glass-melting, brick-making, cement manufacture, lime- burning and the quartz-glass industry. During the war the Carnegie Geophysical Institute at Washington, which has earned great fame