PETROLOGY

since roro there has been a steady advance in all departments of petrology. for many years the description of the microscopic characters of minerals and rocks was held to be, if not the only, at least the most important, part of petrographical literature. text-books of petrography were in gencral a description of the recognised rock types, their composi- tion, structure and the stages of their decay, with notes on their geographical distribution and their geological age. chemical analyses were used principally as a means of identifying the classes to which individual rocks belonged and as a guide to the minerals of which the rocks consisted. more attention of late years has been directed to other prob- lems connected with rocks—such as the conditions of their origin, their chemical classification and the physical laws which deter- mine what minerals shall be formed, in what order they will crys- tallise and through what stages they will pass when subjected to cooling, pressure and metamorphism. there is no difficulty now in attaining temperatures such as occur in the deeper parts of the earth's crust and in the interior of volcanoes and in main- taining these temperatures quite steady for several days or weeks if necessary. the electric pyrometer has reached such precision that an error of one or two degrees is all that need be expected in measuring temperatures up to 1600° centigrade. very high pressures can be easily obtained provided the tem- perature is low and there is no necessity to study the action of compressed gases. it is less easy, however, to perform experi- ments by which the action of steam and other gases on molten rock magmas at temperatures about 1roco° and under pressures over 100 atmospheres can be exactly determined. more than one investigator has now been able to attain this, and a very correct reproduction of the conditions under which igneous rocks crystallise is consequently possible in the laboratory. descriptive petrology has been by no means in abeyance though the five years of war turned the activitics of many geol- ogists to other fields. exploring expeditions, such as those of scott, shackleton and bruce in the antarctic, have brought home large collections of rocks which have been examined and described, and the constant activities of geological surveys in all parts of the world, together with the researches of geological specialists, have added largely to our knowledge. igneous afagmas.—of the three great groups into which rocks are naturally subdivided, the sedimentary, igneous and meta- morphic, the first is on the whole best understood and presents the smallest number of unsolved difficulties. the manner in which sediments are laid down on the bottoms of seas and lakes, in river deltas and valleys and on land surfaces, at the present’ lime is open to investigation by simple means, and, except in the case of the deposits of the deeper parts of the oceans, is reasonably clear. igneous magmas, on the other hand, are, essentially obscure in their origin and history, and they have been the subject of much investigation in recent years. the origin of magmas is a problem belonging to geology rather than petrology. they have been regarded as unconsolidated rem- nants of the primeval molten globe, which by geological changes, such as the secular contraction arising from cooling and the pressures by which mountain chains have been upheaved, have ioo been afforded an outlet to the surface, where they appear as volcanoes, or have been forced between the rocks of the upper layers of the earth’s crust, where they may be laid bare by sub- sequent denudation as “ bosses ”’ of granite or gabbro or intrusive sills of porphyry or dolerite. others have held that magmas may arise in whole or in part by the fusion of solid rocks (of any of the above three classes); the agencies producing fusion being rise of temperature, either through crushing and movement or by de- pression into those regions where a high temperature naturally prevails, or through penetration of gases from the earth’s interior which are not only intensely hot but are capable of combining and setting free large quantities of energy. it is conceivable also that deep within the earth’s crust masses of rock occur at tem- peratures so high that, if pressure be relieved by the vaulting- up of the overlying crust or by fissures opening to the surface, they may become liquid and rise through any available channels. hardly less obscure than the origin of magmas is the question of their variety or differentiation. the outstanding fact in this connection is that no large developments of igneous rock are really homogeneous, and even in small masses a great number of varieties or rock-species occur frequently, differing in their chemical composition and their minerals. granite, diorite, gab- bro, norite and peridotite may all occur within a smal] outcrop not more than one or two square milesin area. the origin of differentiation has been much discussed. some have ascribed it to diffusion or to a principle by which the heavier atoms in the molten mass will be concentrated either towards the bottom or towards the cooler edges or surfaces of contact with the surround- ing rocks. along these lines no satisfactory explanation has been found. more recent speculations have followed three lines: (a) subsidence or flotation of crystals; (8) absorption of sedi- mentary or other foreign rocks; (c) concentration of vapours. (a) subsidence or flotation of crystals —-when crystals form in a liquid they will, if heavier than the liquid, tend to subside and be collected near the base; and if lighter, they will tend to rise. if the crystals differ in composition from the liquid, as they usually do, consolidation will result in a mass which is not homogeneous. thus, for example, olivine crystallises early in a basic magma and, being heavy, will tend to sink; consequently magnesia and iron will be in excess (and silica will be iess abundant) at the base, while felspar will predominate towards the top of such a mass. instances occur showing this arrangement, but they are very exceptional; it is not the case that dolerite and gabbro masses as a rule have a pale- coloured felspathic top and a dark base rich in olivine and the oxides of iron. for such cases as do occur another explanation is often available. if now it were possible at an advanced stage in crystallisa- tion to drain away or force into another position the still liquid part of the magma a type of rock different from the original magma would be produced, because most of the heavy ferromagnesian minerals would have been abstracted. (b) absorption of foreign rocks.—a second method of differentia- tion which has received much attention of recent years 1s by absorp- tion of country rock. a gabbro mass, for example, may be sup- posed to dissolve a felspathic sandstone with which it is in contact and thus give rise to a more acid magma which might be represented by quartz-dolerite or even by granite. intrusive masses of igneous rock, as they ascend from beneath, break across the overlying strata and may shatter them into many small blocks. on these and on the surrounding walls a solvent action is likely to take place. if the invaded rock is heavier than the intrus've magma its fragments will tend to sink, and as they are warmed up they will slowly disappear. there can be no doubt that this action is by no means unusual, and many good instances of it are known, but there is little reason to believe that it is an important cause of differentiation. where igneous rocks have absorbed sediment in any quantity they present in general an abnormal facies. granites, for example, which have dissolved clay, slate or mica-schist usually contain andalusite, silli- manite, cordierite, garnet or corundum, minerals which do not normally occur in such rocks. the magmas are said to be ‘ con- taminated.’’ gabbros under similar conditions contain cordierite, garnet and an excess of hypersthene (forming cordierite norites) and are easily distinguished from normal gabbros. absorption of lime- stone by some nepheline syenites is indicated by the presence of crystalline calcite in the igneous rock, and peridotites may con- tain corundum. perhaps the diamond is an accessory of this type in olivine rocks which have dissolved graphitic matter. even when igneous rocks are absorbed the result will as a rule be an abnormality. this is fairly evident in the majority of cases merely by a study of the analyses of true igneous rock types and ordinary sediments. (c) concentration of vapours.—the theory that differentiation of magmas arises from the formation of partial magmas during cooling, pl trology which separated because they became insoluble in one another (as phenol docs with water), is favoured by many geologists, but an examination of the physical laws determining the production of such magmas has led to the conclusion that nothing is known that would make this process appear likely. many geologists, however, who have a wide knowledge of igneous rocks in the field, hold that there is evidence to show that differentiation took place before crys- tallisation began, and that the various types of rock were already distinct when they were injected in liquid form into the positions they now occupy. it has been suggested, though it has not been clearly explained, that the gases dissolved in magmas determine the sequence of crystallisation and may exert a powerful influence in differentiation. a magma rich in gases when it begins to crystallise yiclds crystals of anhydrous minerals. ‘lhe gases, if they do not escape, must in- crease in relative amount in the liquid residuum. the early minerals are those like olivine and augite, which can be crystallised without difficulty from anhydrous melts; the later minerals, such as the alkali felspars and quartz, crystallise readily only in presence of steam and other gases (or of solvents of a nature not usually present in rocks). in some respects the crystallisation of an igneous rock resembles the cooling and evaporation of a saline solution, the gases playing the part of solvent. the minerals appear in the order of their insolubility. {t is probable that the history of magmas will never be clearly understood till a very careful study is made of the consolida- tion of rock-making silicates under high pressures of steam and other gases such as are known to abound in natural volcanic magmas. classification of magmas—the igneous rocks of one geological period and province have often so many peculiarities in common that they can be regarded as having resulted from the consolida- tion of a single reservoir of molten matter. the chain of volca- noes that fringes the shores of the pacific ocean from tierra del fuego to alaska, and thence by japan and the thilippines to java and sumatra, is characterised by rocks which have so much similarity in many important characters that they are certainly of allied origin, even if they have not proceeded from the one source. these rocks are all of tertiary and recent age; their eruptions began in eocene or miocene time and have continued, with more or less frequent intermissions, up to the present day. for another example of this we may take the igneous rocks of the western and mid-atlantic area, from jan mayen, through iceland, the hebrides, canaries, cape verde islands, etc. all these volcanic centres have many rock types in common, and the whole assemblage is strikingly different from the pacific igneous rocks. each of these magmas has been taken as a type, and it has been found that in the older geological periods they are also represented; for example, the early devonian eruptions in scotland are distinctly of the pacific type, while the carbonifer- ous eruptions in the same district are of the atlantic type. if we seek for a precise definition of their respective characters, it is not easy to give a complete answer. it may be said, however, that the pacific suite has a great prevalence of hypersthene ande- sites, and andesites of all kinds. the atlantic lavas, on the other hand, are predominantly olivine basalts, with trachytes and phonolites. another feature which is especially striking 1s that practically all the rocks carrying nepheline and other felspathoids or ‘‘ alkali minerals ” are found in the atlantic suites. this has been regarded as proving that the atlantic magmas are richer in alkalis and the pacific in lime, but it is by no means certain that this is the explanation. in fact, a full chemical] discussion of the relations of these rock-series to one another has yet to be undertaken, but from the work of becke it seems that the pacific are essentially richer in silica, and in the “ light ’’ elements gen- erally, while the atlantic contain more of the “‘ heavy ” elements, such as magnesia, iron, chromium, titanium. several authors have pointed out that the rocks of the pacific group are associated with a folded mountain chain, and conse- quently have appeared in a region undergoing lateral compres- sion and upheaval; the atlantic, on the other hand, are associ- ated with a region of subsidence, with vertical dislocations along lines of fissure and faulting—in other words, a region subjected to lateral tension and depression. a third group of igneous rocks, very well characterised and distinct in many respects, is the pil- low lavas or spilites, which are perhaps the most abundant volcanic rocks of the earlier geological periods and are very widespread in the lake superior district, middle europe, wales and scotland. among these lavas types rich in soda are common petrowogy and albitisation is a frequent pneumatolytic change. these rocks seem to accompany depressions formed in consequence of folding. experiments on constitution of binary magmas.—a molten rock magma may be regarded as a liquid composed principally of oxides (mostly silicates). it crystallises from a variety of causes, of which cooling is the most important, though relief of pressure and escape of gases may also play a part. the laws followed in such a case have been very carefully in- vestigated, not only far metals, salts and organic compounds but also for many minerals. it is generally true that a mixture of two substances will have a lower consolidation point than either of the pure substances. thus salt and ice, if mixed in the solid state at temperatures a little below the freezing-point of water, will melt, forming a liquid which is colder than the ice originally taken; and aqueous solutions of salts have always a freezing-point lower than that of pure water. for each pair of substances there is a definite mixture which has the lowest temperature of consolidation, and this is known as the ezttectic mixture. in the diagram (fig. 1) the horizontal co-ordinate represents com- position, the vertical represents temperature. a mixture of any given composition is represented by a vertical line, and differ- ent points in that line repre- sent different temperatures of the mixture. thetwoslop- ing lines ae and eb divide thediagram intotworegions, of which the uppermost in- dicates substances in a pure- ly liquid state; below the line the substance is in part or wholly ina solid condition. a vertical (composition) line, accordingly, when it cuts these lines shows where 0] temperature loo a composition 100%b a substance begins to crys- fic. i. tallise. the point where the curves meet is the eutectic point e, and shows the composition of the mixture which has the lowest freezing temperature, and the temperature at which it consolidates. the horizontal line drawn through the eutectic point separates the diagram into two regions, a lower one in which the substance is entirely solid and an upper one in which liquid is present. ii we take any vertical line in the diagram, it will indicate a mixture of definite composition st and, followed downwards, it shows the changes tak- ing place in such a mixture as the temperature falls. above the line ae the mixture is a cooling liquid. at s crystallisation begins. at t the last liquid portion disappears and consolidation is complete; below t the substance is a cooling solid. the diagram refers only to substances that crystallise on solidifying; glasses are solids which essentially resemble highly viscous liquids in their properties. when crystallisation begins the substance which is in excess of the eu- tectic mixture will crystallise out first, the residual liquid becom- ing poorer in that component until it has reached the eutectic composition, when the two components will go on crystallising simultaneously till all is solid. the composition of the liquid will travel along the line ae from s to the point e, where it will remain constant. such a diagram is based on a series of experiments in whicha known mixture of two substances (very carefully purified) is heated in a furnace (generally electric) to a temperature well above its mclting- a 15%b 30%b temp. 45%b 60/4b time fig,-2, point. the mixture is then allowed to cool slowly and steadily, and its temperature recorded at short intervals or continuously by some form of pyrometer or recording thermometer. the rate of cooling can be very accurately ascertained and plotted (fig. 2). crystal- lisation is attended by liberation of heat (the liquid losing its latent heat as it passes into a solid), and this involves a retardation of cooling. the simple liquid cools at a uniform rate; crystallisation begins and the cooling slows down; at a certain point the liquid is all joti crystallised and the mixture, now a solid mass of crystals, will go on cooling uniformly. the change of slope in the curve of cooling accordingly corresponds to the passage of the substance from s to e in temperature (or from the purely liquid to the purely solid areas). the physical condition of the substance at any given temperature can also be ascertained by the method of chilling. if the charge be taken from the furnace and plunged in water (or in some cases mer- cury) the mixture consolidates almost immediately, and any parts which were liquid will assume the form of a glass, or a very finely crystalline aggregate. this is especially the case with silicates, many of which crystallise with difficulty. microscopic investigations will enable us to determine the nature and relative proportions of the crystals which were present. the results obtained can be checked by experiments on mixtures having a different composition, and in this way a complete diagram built up on a sound experi- mental basis. the case outlined above is the simplest known. many complica- tions may appear, requiring special precautions and elaborate investigation. thus the liquid may not begin to crystallise at the prop- er point on the upper curve, as some substances crystallise with difficulty and the liquid becomes ‘ undercooled.”” a heating-up experiment may be tried to check the cooling experiments; the same phenomena should appear in the reverse order if no complications are present. many silicates crystallise with great difficulty in or- dinary crucible experiments (such as the felspars, albite ancl ortho- clase). again, it may be impossible to melt the mixture we desire to investigate in any furnace which is suitable for experiments of this kind. magnesia, alumina and lime are examples of substances which cannot be fused at temperatures such as 16007 to 1700°c., which are the limits of accurate work in our laboratories at present; part of the diagram accordingly will be incomplete when substances like these are studied, but an approximate solution can generally be made by extrapolation. another frequent complication is the ap- pearance of transformations in the solid state. the first mineral to crystallise becomes unstable as the temperature falls and changes spontaneously into another crystalline form of the same substance. the change will be attended by an alteration in the slope of the cool- ing curve, for in such cases heat is either liberated or absorbed, and successiul chilling tests can often be made by which the mineral transformation can be clearly demonstrated. silica, for example, appears in three minerals, cristobalite, tridymite and quartz, the transition temperatures being 1470° and 870°. each of these minerals occurs in two forms. below 1190° calclum mictasilicate crystallises as wollastonite; above that temperature it forms another mineral, pseudo-wollastonite. many of these “ high- temperature ”’ forms are not known as natural minerals, and as a rule they are very rare in rocks, probably because rock cooling is essentially a slow process, and the most stable forms at low tem- peratures are the only ones likely to be present when the mass has completely cooled. some very interesting results have been obtained in this field of research; for example, it is known that quartz has two modifications, one above 575°, the other below that temperature, and it has been proposed to use quartz as a geological thermometer and to show at what teniperature it crystallised in a rock mass. if above 575° it would appear as one kind of quartz and on coohng would pass into another; and by various indications, such as crys- talline form, cracks, etc., a record of this transformation may be obtained. very interesting modifications of the process of crystallisation occur when two or more of the mincrals formed are members of an isomorphous series and can in consequence form mixed crystals. this is very common among the minerals of igneous rocks, of which the felspars, pyroxenes, olivine (and probably also hornblende, mica, nepheline and the felspathoids) all belong to isomorphous series. if we have, for example, two components such as albite and anorthite, they will tend to form mixed crystals (known generally as plagioclase felspars), anorthite has the higher melting-point (about 1550°c.); albite crystallises in crucibles only with great reluctance at a temperature about i100° centigrade. a mixture con- taining equal proportions of anorthite and albite will melt at about 1450°, and on cooling will begin to form crystals at that temperature. these crystals will contain about 60% anorthite—that is to say, they are enriched in the less fusible component. as crystallisation proceeds, the felspar that separates becomes gradually less rich in anorthite. the composition of the hquid also alters, because the crystallisation is abstracting anorthite molecules more rapidly than albite molecules. this process may go on till the felspar has all crystallised, the last deposited being near albite in composition, and the crystals, exam- ined microscopically, will show zones, of which the internal are rich in anorthite and the external are progressively richer and richer in albite. but if sufficient time is allowed, a reaction sets in between the crystals and the liquid; in other words, the felspars crystallised are not in equilibrium with the magma except at the moment of crystallisation, and as the magma becomes richer in albite it will attack the early plagioclase, replacing it by a variety containing more albite. these phenomena are we!l known to petrologists as zonal structure of plagioclase felspars; and corrosion of the cores and internal zones of the crystals is almost universal in such rocks as basalt and andesite. rocks of similar composition which have 102 cooled verv slowly, such as gabbro and norite, asa rule do not contain zoned plagioclase crystals, no doubt because equilibrium has been attained and homogeneous crystals formed by the process above described, this may be illustrated by fig. 3. the horizontal line represents composition, the vertical temperature. a is 100% albite, b is 100% anorthite. the upper curve is the liquidus above which there is only a v | b fic, 3. liquid; the lower is the solidus below which all is crystallised: be- tween these lines is a space representing stages in which crystallisa- tion is going on but still incomplete. each point on the solidus has a corresponding point on the liquidus, which is found by drawing a horizontal line across the intervening space. a mixture of any composition, say @ at @, is completely liquid. as the temperature falls it begins to crystallise at b. the crystals formed have the composition c. further cooling results in the formation of crystals at b’ the composition being c’. at b” all is crystallised, the last crys- tals being d. if resorption is completely accomplished the final crystals have the composition a, but they are usually more rich in albite: this will depend on the rate of cooling, the number and size of the crystals formed, and on a variety of other factors. the theoretical investigations of roozeboom and gibbs have shown that five types of crystallisation of isomorphous substances may occur, in some of which the mutual solubility of the two com- ponents is unlimited, while in others it is limited so that only mixed crystals of certain types may occur. several of these have been identified in rock-forming minerals, and others are suspected though not yet proved. ternary magmas.—magmas of three components (ternary) are much more-complicated than binary magmas. to represent their behaviour a triangular diagram is necessary. usually an equilateral triangle is employed, and the distance from any point to the three sides of the triangle is made to represent the three components of any mixture in their true proportions; the sum of these three per- pendiculars is constant and equal to the height of the triangle; if lines be drawn through the point, parallel to the sides of the triangle, they will cut the sides at distances which will represent the relative proportions of the components. any mixture of three components can be represented by one point in this triangle. to represent tem- perature another co-ordinate is required which is perpendicular to the plane of the triangular diagram: and a solid model must be made, resembling a triangular prism with flat base and an irregular surface representing the consolidation temperatures as the top of the prism. each of the three vertical surfaces of the prism represents the behaviour of the mixture of two of the components. to enable us to construct such a model a very large number of experiments must be made, first with binary mixtures and then with mixtures of the three sub- stances. their exact temper- ature of first crystallisation must be ascertained in each case, and also the nature of the mineral which crystallises. simultaneous crystallisation of two minerals will follow, and the temperature at which the second mineral appears is to be determined. three minerals will ultimately ap- pear, and the last process will probably be the forma- tion of a ternary eutectic at which the temperature re- mains steady till the liquid disappears and finally the fig. 4. 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 mixtures of two components. ea-, ese and eas represent the three binary eutectic mixtures. the ternary eutectic is repre- petrology sented by eare, containing about equal quantities of c and band a smaller amount of a. if @ represent the composition of a cer- tain liquid which is allowed to cool and crystallise, composition will change along the line aw. ata certain point the a component will crystallise out and the liquid will become poorer in a (richer in b and c), and the composition will change from @ towards b (away from point a). after a time, the liquid becomes satu- rated for 6 which will start to crystallise, and now the liquid changes composition along the line 6 e.s, as the temperature continues to fall; finally c also begins to crystallise, and the ternary eutectic point is reached at which the three components crystallise simul- taneously in definite proportions (represented by the position of ese) 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 matters must be kept in mind: (a) the liability to form compounds, which behave as new substances with their own fusion-points and eutectics. (6) 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 dit- erent minerals (incongruent). (d) many silicates refuse to crystallise 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 crystallisation 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. (6) intrusive magmas are under grcat 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 crystallisation which may appear in rapidly cooled melts. it will also favour the complete transformation of early un- stable minerals into stable permanent forms. many varictics 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 crystallisation we may take a mixture consisting of 50% diopside and 50% plagioclase (containing equal proportions of qiop shoe f albeit. = cf ab, an, l fig; 5. 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). crystallisation 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 degrees. at 1245° the excess of diopside (g) has separated out, and felspar begins to crystallise. it has about 75% anorthite (h). thereafter diopside and felspar both crystallise, but as the temperature travels along the line egd from g to k the composition of the felspar changes from ii to z (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 ts 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 crystallisation is suddenly brought to an end, pe lrology 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- tallised out first. this seems to be in keeping with the structure of many dolerites, which contain felspar partly enclosed in augite crystals of later formation (ophitic structure), while others show that the augite appears in porphyritic crystals and began crystallising before telspar. 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 alo;-cao-sio, 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 e . b gags ayre 2 : « ' « 3000 hd, side \s - * . a * nag ° * e = mae’, oe eo; a j 9£a0 ja, 8ca0. baiz0y alzoy fic. 6. tea aigdy substances known to occur as minerals, cristobalite, tridymite, wollastonite, anorthite, sillimanite and corundum. in each of these fields the minera! named will crystallise if the tempcrature of the melt falls. the fields are separated by lines which show under what circumstances the two minerals whose fields adjoin will crystallise. where three fields meet, the conditions exist at which three minerals will exist simultaneously (or, to express it otherwise, are in cqui- 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 (impure limestone) to a very high tempcra- 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 industrics such as pottery, stecl- making, glass-melting, brick-making, cement manufacture, lime- burning and the quartz-glass industry. it should be noted that the compound al, sio; which corresponds to the three common minerals of rocks, andalusite, sillimanite and kyanite should not be represent- ed in this diagram. it has not yet been made in crucible melts. another mineral, viz.: mullite, which has the composition 3al.0s, 2lio, is produced under these conditions. theory of ternary afixtures.—the theory of ternary mixtures has been very fully worked out mathematically by willard gibbs, backhuis roozeboom, schreinemakers, smits, kuenen, tammann and others, and may be said to be well understood in its main ap- plications. experimental investigation has also made great progress. it is a laborious matter, requiring great skill and very elaborate appliances. the diagram we have given showing the ternary system cao, alsos, sioe, required over 7,000 experiments, in some of which the heating had to be continued for many hours and even for days; and the thermal results had usually to be checked by microscopic examination of the product. other systems equally important have received very careful study. these are not so simple as the instance quoted, because mineral transformations, either during or after crystallisation of the melts, often give rise to many complications. 103 thus, for example, casios crystallises as nseudo-wollastonite at high temperatures, which may change to woilastonite at low tem- peratures. mgsiog; has four forms—kupferite, magnesia amphibole, enstatite and clinoenstatite. where both these substances separate from the melt, none of the above-named mincrals appears, but a diopside clinoenstatite solid solution or isomorphous mixture, which belongs to the common group of minerals known as the pyroxenes. these facts are of the greatest interest to both mincralogists and petrologists. they have introduced many new minerals (artificial) to our knowledge, and taught us their relationships at atmospheric pressures in certain dry melts. they have also enabled us to under- stand many of the peculiarities of the minerals that occur in rocks, and the method of their origin, as another instance we may quote the relations of forsterite (mg:sio4) and enstatite (ai[gsio;). forsterite may crystallise from melts of suitable composition and after a time it may become unstable in the residual magma; then it is dissolved up or “ cor- roded "' and clinoenstatite replaces it. again, if an enstatite (mgsio;) mixture be fused and allowed to cool, forsterite begins to separate out; it will tend to be reabsorbed and converted at a lower temperature to clinoenstatite, but if this be prevented the mixture solidifies as forsterite, clinoenstatite and quartz. to the petrologist this is of great interest, because forsterite is one of the constituents of olivine, a very common mincral of the basic igneous rocks. now olivine generally occurs along with some member of the pyroxene group, and the olivine crystals show rounded outlines, which have been taken to indicate corrosion or resorption by the magma after crystallisation with concurrent formation of pyroxene. in some rocks there is clear evidence that olivine crystallised early and sub- sequently was entircly dissolved. in other rocks olivine is found enveloped in clusters of enstatite crystals which have evidently formed at its expense; and it is a very frequent characteristic of the olivines of gabbro and norite that they are surrounded by “ reac- tion rims " or “ corrosion borders ’’ consisting of enstatite, tremolite and other magnesian silicates that have a higher percentage of silica than olivine itself. the meaning of these phenomena was fairly well apprehended by petrologists, and now thcir conclusions have been confirmed by experimental evidence. another feature of some interest is that, in the ternary system under consideration, it has been shown that forsterite may crystallise early, then dis- appear by corrosion, and a second generation of forsterite may sub- sequently crystallise. whether this is a common phenomenon is not yet known, and its exact relation to the frequent appearance of minerals in two generations in rocks remains to be demonstrated. the number of ternary systems that have as yet been fully in- vestigated is small, though the results are of the highest importance. those which contain the alkalis potash and soda present certain special difficulties, such as the volatility of part of the mixtures and the difficulty of crystallising some of the minerals. rapid progress is being made and it 1s unlikely that experimental difficulties will retard the advance of knowledge. a special case, of the greatest possible interest to the petrologist, arises where one or more of the com- ponents are volatile. natural magmas are probably always richly charged with gases. the theory of such systems has been explored by schreinemakers, smits and others. experiments with mixtures enclosed in steel bombs which can withstand great pressures at high temperatures have been very successful in certain cases, but as yet only the borders of this field have been explored. a good deal of work has been done on the dissociation pressure of sulphides of the. mctals, a subject of great importance as regards the paragenesis of natural sulphides and the conditions under which mineral veins have been formed. ‘the effect of steam in the formation of silicates is under investigation and during the next few years will probably be the subject of much research. . recent developments ——the most recent developments of petro- logical investigation have followed certain well marked lines such as the behaviour of systems of more than three components, the effects of volatile ingredients at high pressures and temperatures, the theoretical basis of metamorphic processes and the genesis of min- eral ore deposits (including the changes that effect secondary en- richment). systems with many components can be handled only by means of mathematical developments. for example, a tetrahedron will represent a four-component system, but for higher orders of complexity special methods must be used which as yet have only received a preliminary discussion. the complicated series of solid solutions which many common minerals represent (such as tour- maline, mica, pyroxene) have been discussed on these lines, but only the simplest rocks are suitable for this method. another line of advance has been the treatment of rocks as systems of three or four minerals, each having a complex composition, and with certain precautions this gives useful results. constantly increasing attention is being given to the effects of volatile constituents in determining the minerals which form in rocks and the transformations which take place during and after solidification. the solubility of water and other gases in silicate glasses at various temperatures has been determined, and it is shown that some of them may contain from 8 °% to 25% of water. the calcite series has been very completely investigated as a mix- ture of cao and cos, and its relations and properties are well known. such series as copper and oxygen; silica, potash and water, and the 104 nature of the bond by which the water in zeolites is held to the other components have received very detailed investigation. the pneu- matolytic or post-consolidation effects of vapours set free by crys- tallisation have attracted more and more attention, especially the formation of zeolites, albitisation, analcitisation, scapolitisation and the production of mica at the expense of original felspar. in the study of metamorphism, new and fruitful advances have followed, especially the consideration of rocks as systems of mineral components in more or less complete equilibrium according to their composition and the conditions to which they are subjected. this was determined eventually by the phase rule, but the application is difficult, as rocks contain many components and no rock remains for an indefinite period in exactly the same conditions. an igneous rock, for example, has a gradually falling temperature, a meta- morphic rock may suffer great changes in pressure or in tempcrature in the course of its history. hence a simplified or modified phase rule must be devised for the case in hand. still the broad fact remains that certain minerals may be expected to form in a rock of known composition according to the physical conditions to which it is exposed.’ the hydrous silicates of alumina, such as clay, are stable only at low temperatures; and are replaced by andalusite or silli- manite when the rocks are greatly heated, and generally by mica when there has been great pressure or movement. the recognition of distinct ‘' facies ’’ of rocks, each characterising special zones of pres- sure and temperature, has been carried to a great length by swiss, austrian and german petrologists, and is undoubtedly very helpful to the study of the science. along with this there has been an effort to classify mineral associa- tions as groups of phases which may or may not occur together in stable equilibrium. for example, silica forms three minerals, each having more than one form (the commonest in rocks being quartz); alumina forms corundum; the combination of silica and alumina forms sillimanite, andalusite and kyanite (and also mullite). ina hypothetical rock consisting of silica and alumina, quartz and corun- dum should not occur together if the temperature has been raised above the point of combination. if we add magnesia to the two oxides mentioned above we get a further group of minerals such as enstatite, cordierite and spinel. rules have been laid down which determine the associations of minerals which should occur together under various conditions in mixtures with the composition given above. in all this work the sciences of metallurgy, physical chemis- try, and the crystallisation of saline solutions have exerted a very strong influence on petrological theory, and have stimulated research along new lines. bibliography.—for the recent advances in physico-chemical and experimental petrology no english text-bookisavailable. a. harker’s natural history of igneous rocks (1909) gives a sketch of the subject from the standpoint of its time, but is now out of date. a more adequate work is h. e. bocke’s phystkalische-chemische petrographie (1915), which gives full references and is the best general account. j. 1. l. vogt’s silikatschmelsiesungen (1904) is interesting but ver speculative. his sulphid silikatschimelziesungen (1925), deals with some special aspects of the problem. the fundamental works on the theory of equilibrium are h. w. b. roozeboom's jteterogenen gleichgewishte, vol. 1 (1901), vol. 2 (1904) and its continuation (vol. 3) by f. a. h. schreinemakers (1911). see a. findlay, the phase rule (1923); g. tammann, arysialliziren und schmelzen (1903). for english and american students the best authorities are the papers published by the workers of the geophysical institution in wash- ington. see also p. niggh, lehrbuch der mineralogie (1920), die leichifliichtigen bestandteile im magma (1920); r. a. daly, igneous rocks and their origin (1914); j. p. iddings, igneous rocks (1913); a. d. holmes, nomenclature of petrology (1920); j. stark, ‘petro- graphische provinzen,” fortschriite der mineralogie (1914); eh. rosen- busch, elemente der gesteinslehre, 4th ed. by prof. osann (1923); jacques de lapparent, lecons de petrographie (1923); i. rinne, gesteinskunde, gth ed. (1923); a. harker, petrology for students, 6th ed. (1923); h. milner, sedimeniary petrograplhy (1922); a second edition of h. e. boeke’s phystkalische-chemische petrographie (ed. w. eitel) has appeared (1923); o. h. erdmannsdorffer, grundlagen der petrographie (1924); u. grubenmann and p. niggli, die gesteins- metamorphose, 3rd ed., pt. 1 (1924); w. eitel, physikalisch- chemische mineralogie und petrologie (1925); h. rosenbusch and e. a. wulfing, afikroskopische physiographie, ist vol. of 5th ed., (1925). (j.s. f.)