Gontemporaneous Basaltic and Rhyolitic Magmas'

Amefican Mineralogist, Volume 58, pages 153-171, 1973

GontemporaneousBasaltic and RhyoliticMagmas' H. S. Yoonn, Jn. GeophysicalLaboratory, Ca:rnegieIrctitution ol llashington, Washington,D. C. 20008

Abstract

Tho common interbeddingof basalt and rhyolite indicatesthat such magmas were available at the same place contemporaneously. Field observations support the view that two magmas of highly contrasting compositions coexisted at least for limited periods and were erupted simul- taneously throughout geologic time. The paucity of intermediate rocks casts doubt on the pro- duction of cohtrasting magmas by fractional crystallization. Previous explanations either have assumed the existence of separate basaltic and rhyolitic magmas or of remobilization of pre- existing granites,or have appealedto liquid immiscibility. The thermal properties of natural rhyolite and basalt do not support the concept of im- miscibility, and melts formed by the fusion of mixtures of common natural basaltic and rhyo- litic rocks as well as their glassesare completely miscible. Additional experimentsillustrate that Hro-saturated rhyolitic and basaltic melts may maintain an interface for short periods of time, and some hybridization appears to take place between such coexisting magmas. The similarity of concentrationgradients of elementsnear the interface, as determinedby electron microprobe analysis, suggeststhat element interactions play a dominant role in the diffusive transfer of material. Tho asymmetry of the concentration gradients relative to the optical interface indicates more extensive transfer in the basaltic liquid than in the rhyolitic liquid. A new hypothesis involving fractional melting illustrates one possible way in which two magmas of highly contrasting composition may be generated successively from the same parental material. Small batches (=10 km3) of parental material are partially melted as a result of adiabatic decompression.Each magma is extracted at separate invariant points, essen- tially isothermally, in an interval of several thousand years. No liquids of intermediate compo- sition are produced. The paucity of intermediate rocks, the "Daly gap," is attributed to the fractional melting process without regard to the time interval between contrasting magmas. Removal of small amounts of liquid from the parental material is presumed to take place with the same efficiency as the removal of residual liquids from monomineralic cumulate layers in layered intrusions. Alternatively, rhyolitic liquid may be removed in one batch even though removal of basaltic liquid would have to take place continuously. Variations imposed by the HnO content and its initial disposition are manifold; however, if a hydrous mineral is retained in the parental material during formation of both magmas, the products are rhyolite and an- desite. If the H,O of the parental material is consumed in the first stage of fractional melting, the products are rhyolite and tholeiite. Trace element contentsof each magma are dependenton ( 1) the kinds and proportions of major phasesin the parental material, (2) which major phase is consumedat the first stage of fractional melting, (3) the presenceof accessoryminerals, and (4) the distribution coeffi- cients effective under the conditions of magma formation. The extent of hybridization of con- tiguous contrasting magmas depends on the geometry of the interface, the degree of turbulence, and the storage period, and is not necessarily confined to producing liquids along normal lines of descent. A hydrous, olivine-rich parental material capable of producing quartz-normative liquids is assumed on the basis of experimental evidence from synthetic systems and natural rocks; however, the hypothesis is equally applicable under anhydrous conditions and to parental materials having other normative characteristics.

fntroduction membersabsent or of negligiblevolume. Bunsen in one of the oldest problemsin petrologyis the 1851 recognizedthehighly contrastingcompositions contemporaneousextrusion or intrusion of magmas of basalt and rhyolite'zin Iceland' He pro'posedtwo of greai contrast in cornpositionwith intermeiiate independentparental magmas'which when mixed in 'Bunsen (1851, p. 199) describedthe extreme,members l Presidential Address, Mineralogical Society of America. as a normal trachyte and a normal pyroxenite and indicated Delivered at the 53rd Annual Meeting of the Society, 14 in a footnote that these were to be taken as a silica-rich November 1972. trachyte and a silica-poor basalt or dolerite, respectively. 153 t54 H, S. YODER, IR.

minimum of intermediate compositions. The relative paucity of intermediate rocks has often been de- scribed as the "Daly gap" (Daly, 1,925). The bimodal frequency distribution in composition has been raised as a serious objection to the forma- tion of igneous rocks primarily as a consequenceof crystal fractionation. fn general, the volume of suc- cessiveliquids, and hence the rocks crystallizedthere-

(t) from, would be expectedto decreasein a petrogenetic process controlled primarily by crystal fractiona- h tion. Furthermore, no major gap or hiatus in liquid d n composition would be anticipated in an anhydrous d crystal fractionation process (see Chayes, 1963, Ap- rH o pendix);a however, sudden loss of volatiles in a hydrous fractionation process may produce major C) gaps in liquid composition.5 t4 F Because the question bears heavily on the pres- z ently accepted framework of igneous petrology, it is desirable to review the field evidence bearing on the contemporaneity of magmas outside the composi- tional gap, examine the relative thermal relations of the contrasting magmas, review previously proposed hypotheses of origin, present some new exploratory data on the coexistenceof two melts, and outline a new hypothesis for their generation from a single o parent. 45 55 65 75 85 Field Evidence of Contemporaneity si02 Most field geologistshave observed rhyolitic lava, Frc. 1. Distribution of SiO, (abscissa)in 551 analysesof pumice, tufis, or ash interbedded with basalt flows. rocks of the oceanic basalt-trachyte association (Chayes, Less commonly observed is an array of field asso- 1963, p. 1523, Fig. 1A). Note paucity of specimenshaving intermediatesilica contents. ciations which suggest that at times these magmas of highly contrasting composition have actually co- various proportions produced the wide variety of existed and were erupted contemporaneously. igneousrocks.s Although the concept of magma mix- l. Mixed welded tuffs containing glass shards with ing did not survive as a major process in petrogene- indices of refraction of approximately either 1.516 or sis, the contrast of magmas has since been recognized 1.545 corresponding to SiOz contents of about 67 as a world-wide phenomenon. percent and 56 percent, respectively, are known Chayes (1963) has illustrated the compositional contrast in a quantitative way by various plots of 4Attention is called to the compositionalgap generated oxides and normative minerals of associatedoceanic by fractionalcrystallization between the crystalresidua and basalts and trachytes. The time relationships of the the last liquid.Starting with a melt of Ab.Am composition in the Ab-An system,for example,the last liquid is of Ab associatedrocks are not known; however, the rele- compositionyet the aDeragecomposition of the zonedcrys- vance of the data to contemporaneous basalt and tal accumulatemay be Ab-Ano. The apparentanomalous trachyte will become evident. One example (Fig. 1) hiatus betweenphenocrysts and groundmasscompared to is sufficient to show that there is indeed a division of equilibriumcrystallization in some igneousrocks can be these igneous rocks into two major groups with a ascribedto the fractionalcrystallization process, or crystal accumulation.or both. sThe mixedmagmas were described by Durocher(1857) "See Yoder (1969,p. 83 and Fig. 4). Emphasiswas as hybridsin the strictestsense, although "hybrid" is some- placedon the changeof crystalcomposition with releaseof timesused to describethe liquid resultingfrom the assimila- volatiles;however, the abruptchange in liquid composition tion of crystallinematerials. is equallysignificant. BASALTIC AND RHYOLITIC MAGMAS 155

chambersand mixedin the duct duringsimultaneous eruption.Other similar localitiesare known in lce- land, and anothersuch "emulsion" has been found in the caldera of the volcano Fantale in Ethiopia (Dr. I. L. Gibson,personal communication, 1970). (Sometektites consist of two or more glassesof contrastingcomposition [E. C. T. Chao in O'Keefe, 1963, pp. 6A-621;however, they are not necessarily of volcanicorigin. ) 4. Compositedikes having basic margins and acid coresare of commonoccurrence in the Tertiary vol- canicregion of easternIceland and in Great Britain (Blake et al., 7965). The field relationshipssuggest Flc. 2. Photograph of sawed surface of hand specimen that a basicmagma first occupiedthe dike and while of banded pumice from base of Novarupta, Katmai area, still molten was evisceratedby the later inflow of Alaska. Rhyolite, white; andesite,black. SpecimenNo. 54 acid magma.The basic rock locally showschilled ACU 36, courtesy of Dr. G. H. Curtis. Longest dimension bordersagainst the acidcenter as well as the country of sawedsurface is 10 cm. rock. The converserelations are alsoobserved, although (CostaRica; Williams,1952). Someshards show more rarely, in central Scotland (Buist, 1952, p. "hair-fine"banding of both types.The implicationis 53), Skye (Harker, 1904,p. 208), Arran (Judd, that a singlevent was simultaneouslyproducing vol- canic products of distinctly different composition. Approximately15 percentphenocrysts were present, and,therefore, crystallization had commencedbefore eruption. 2. Particularly striking are the ban:dedpumice blocks consistingof light bandsof dacite and dark bands of andesiteejected from Mt. Lassen (Mac- donaldand Katsura,1965) andthose of rhyoliteand andesite(Fig.2) from Novaruptaat Katmai,Alaska (Curtis, 1968). The imperfectlyblended or hybrid- ized magmas,presumably gas-charged, reached the surface and erupted without entirely losing their identity. 3. A unique tufi-lauahas been found at the vol- cano Breiddalur in Iceland (Walker, 1963; Blake et al., 1965), which exhibitsan emulsion-liketexture on a scaleof millimetersof nearly equal proportions of a black basicglass and a clearacid glass (Fig. 3). (The compositionof the glassesand their thermal behaviorare describedbelow. ) The "emulsion"con- tainsphenocrysts of two kinds of feldspar;according to Blakeet al. (1965,p.43), the morebasic plagio- claseis mantledby basic glass,and the more sodic plagioclaseis enclosedin acid glass.One may con- fidently conclude from this occurrencethat the liq- uids with their respectivephenocrysts issued from of thin section of tufi-lava "emulsion" the samevent, althoughno firm conclusioncan be FIc. 3. Photograph rock from Breiddalur, Iceland. SpecimenNo. 05368, cour- drawn that they came from the same source.The tesy of Dr. G. P. L. Walker. Clear glass,SiO, = 73.A5Vo; two liquids could have been generatedin separate black glass,SiO, - 52.40Vo.Note phenocrystsin each glass. 156 H. S. YODER, IR.

1893, p. 561), and south Greenland (lJpton et d., 1971, p. 165), where the basic rock occupies the central portion and acid rock forms the margins. The contiguous borders of the members of the composite dike may exhibit mechanical mixing, with blobs of basaltic liquid trapped in rhyolitic liquid, or streaks of each intermingled in a flow structure. Hybridizatio'n may also be evident at the interface. A spectacular exhibit of mechanical mixing of hy- drous magmas was found by Dr. R. A. Wiebe (per- sonal communication, 1970) in a dike in northern Cape Breton Island, Nova Scotia. This body con- sists of pillow-like inclusions of hornblende gabbro in a granitic matrix (Fig. ). Ftc. 5. Thin section of mixed zone between quartz tra- 5. Some composite dikes have been traced into chyte and basalt, P' composite flow on Gran Canaria, Ca- composite laua flows (Gibson and Walker, 1963), nary Islands.Micropillows of basaltic glassin trachytic glass the basic unit usually underlying the more acid unit with large phenocrysts.Note vesiculationand diffusion au- SpecimenNo. 994, courtesy of of the flow. reole around micropillows. Dr. H. U. Schmincke. The converse relationship of units is found in the volcano Tejeda of Gran Canaria (Schmincke, 1967, p. 7), where a composite flow consists of a lower 5) in the quartz trachyte ash flow or ignimbrite, and ignimbrite of quartz-trachyte composition and an each component contains phenocrysts. upper columnar basalt. The quartz-trachyte com- Such flows are distinct ( 1) frorn "multiple" in- ponent is mixed with the basaltic component in a trusions in which the inner portions chill against the zone up to several meters thick and covers a vast outer portions, which in turn chill againstthe country area (300 km2). The clots of basaltic magma are rock, (2) from flows that have undergone in situ not uncommonly formed into "micropillows" (Fig. gravitative differentiation, and ( 3 ) from flows that have undergone gravitative differentiation in the conduit prior to eruption. The last two types of flow show no, chilling effects between the components and less compositional contrast, but often have por- tions distinguished by the presence or absence of phenocrysts.The basalt-mugeariteflows of Kennedy (1931) and the complex sill of Upton and Wads- worth (1967) appear to result primarily from a gravitative differentiation process. 6. "Emulsions" are known on a grander scale where meter-sized blobs or "pillows" of basalt with crenulate or cumulus borders are found within rhyo- lite. The mixed laoas of the Yellowstone Park region of Wyoming are familiar to most petrologists as a result of the Wilcox-Fenner debate (Fenner, 1938, 1944; Wilcox, 1944; Boyd, 1961). The mixing of contemporaneous basalt and rhyolite was consid- ered by Hawkes (1945) to have taken place before extrusion of basalt and rhyolite and was beliwed by Wilcox (1944) to have occurred after extrusion of the lavas. The critical observation,first made by Wil- Fto. 4. Photograph of mixed composite dike of horn- blende gabbro (black) and granite (light gray) at northern cox, is that the basaltic magma was quenched or Cape Breton Island, Nova Scotia. Courtesy of Dr. R. A. chilled against the rhyolitic magma, showing clearly Wiebe. that they were in contact as liquids. Later Wager BASALTIC AND RHYOLITIC MAGMAS 157

and Bailey (1953) suggestedthat pillow-likemasses SiO2, and then gradually changing to a lava with represent basic magma chilled against cooler but 54 percentSiOz. This sequenceof chemicalchanges liquid acidic magma at St. Kilda, Hebrides,and at in eruptions at Hekla appearsto reoccur in a per- SlieveGullion, Ireland. iodic way (Thorarinsson,1954, p. 43,Fig. 7 ) . 7. Other intrusiaelorms indicative of liquid-liquid There can be little doubt, therefore, that two relationshipsinclude net-veinedcomplexes and pipes magmasof greatlycontrasting composition-whether (Walker and Skelhorn, 1966, p. 95). Even some basalt (or andesite)versus rhyolite (or dacite) or large plutons in batholiths are believed to be the alkali basalt(or hawaiite) versustrachyte (or muge- result of magmamixing and hybridization (Piwinskii arite)-coexisted at the same locality and erupted and Wyllie, 1968,p.231). from the samevent (or wereintruded into the same 8. Further documentationof the contemporaneity dike) and at the same time. These magmashave of some basaltic and rhyolitic magmascomes from retained their identity with a minimal amount of the direct obseraationof eruption. The 1707 erup- mechanical mixing or chemical diffusion. The key tion of Fujiyama (Tsuya, 1955) is especiallynote- question is, "How are two magmas of such great worthy becauseof its almost simultaneousproduc- contrast in composition generated and maintained tion of an acid pumice (SiOp = 68Vo) and a basic at the sameplace at the sametime?" ash (SiOz = 57Vo) from closebut separatecraters. The 1947 Hekla eruption (Thorarinsson, 1954, pp. Relative Thermal Behavior of Basaltic and 33, 38) initially produceda pumicewith 63 p€rcent Rhyolitic Magmas SiO2,abrupf/y changingto ejecta having 56 percent The question becomeseven more puzzling when the thermal propertiesof the magmasof contrasting compositionare considered.Rhyolite and basalt will be used as examples;however, similar argrmentsare applicable to the alkali basalt-trachyteassociation. Rhyolites are usually explosive (e.3'', ash, tufts, pumice, ignimbrites), and gas, mainly H2O, is con- sidered to be the chief propellant. The minimum melting curve of granite in the presenceof an excess of water may be used as an approximation of the o Iowerlimit of stability of a potential rhyolitic magma U (Fig. 6). Similar lower limits of stability describe E (nl the behavior of trachytic and phonolitic magmas C) trj r (Millhollen, L971.;Barker,1965). In eachof these I r. casesthe melting behavior is of either the minimum F type or the eutectic type becausethe compositions of such magmaslie approximatelyat invariant points in their respective systems.This implies that the crystal * liquid interval betweenthe crystallinerock and the wholly liquid magma (H2O-saturated) is moderatelysmall ( < 100'C) or negligible(Piwinskii and Wyllie, l97O). In contrast,the eruption of basaltis usually quies- cent. The absenceof hydrous minerals, of course, 500 600 7@ 800 900 1000 llo0 1200 1300 does not necessarilyindicate the absenceof water. TEMFERATURE,"C Andesites, which may contain hydrous minerals, Frc. 6. Minimum melting curve of rhyolite composition often erupt explosively.The mixed welded tuffs of (- granite of Tuttle and Bowen, 1958; Luth et al.,1964) Costa Rica and the banded pumices of Novarupta, and the liquidus, solidus, and upper stability curve of am- Katmai area,Alaska, and of LassenPeak, California, phibole (Yoder and for the olivine tholeiite composition volatiles in both magmas. Tilley, 1962) in the presence of excess ItO. Conditions attest to the presenceof studied experimentally for other materials are marked with For thesereasons the lower limit of liquid in a basalt an X, above the liquidus of rhyolite and basalt, respectively. (olivine tholeiite) in the presenceof an excessof 158 H. S, YODER, IR,

(HrO-saturated)is just all liquid in the samepressure range, then the rhyolitic magma is superheated approximately250"C at I atm and 450oCat I kbar P11,6.Because both natural magmascontain pheno- crysts,the possibilityof such excessivesuperheating is most unlikely. Moreover, the natural basaltic magma is commonly chilled against the rhyolitic magma, but the reverse relationship has not been recordedin localities of mixed magmas.Only rarely is the remelting or resorption of the phenocrystsin Cumulotca contaminatedrhyolite observed(Wilcox, 1944, p. 1054).Herein lies the heart of the problem and the dilemma. The generationof the contrastingmagmas at or near their own liquidus temperatureimposes a seriousconstraint on their modesof origin. Frc. 7. Schematic representationof cupola-shapedmagma reservoir produced by melting of overlying granitic layer by convecting basic magma. Mechanical mixing develops hy- PreviouslyProposed Hypotheses brid magma and fractional crystallization yields cumulates. Somepreviously proposed solutions to the prob- From Wageret al. (1965,p. 303,Fig. 35b). lem will now be reviewedbriefly. water is alsogiven in Figure 6 alongwith the liquidus 1. The simplestsolution is obtainedby invoking two separatechambers or curve and the H2O-saturatedmelting curve of the reservoirs,one containing predominanthydrous phase, amphibole. The large rhyolitic magma and the other basalticmagma at regionof crystals* liquid for basalticcompositions their respectiveliquidus temperatures, perhaps at dif- ferent provide is evident.Quartz tholeiites,tholeiites, high-alumina depths.An earthquakemay horse- tail cracks vent lead basalts,and alkali basaltsexhibit similar stability that both chambersand to a vent. This fortuitous must limits in the samepressure range (Yoder and Tilley, common arrangement t962\. be repeatedworldwide, and the solutionignores the question It is of interest to note that if basalticmagma of how the two magmasarose in the first place. (HzO-saturated)is to crystallizeas basaltor gabbro and not hornblendegabbro, the water pressurecan- 2. Another solution to the problem was sug- not exceedapproximately 1.5 kbar.6The dike from gestedby Holmes( 1931), wherebya basalticmagma Cape Breton, Nova Scotia,mentioned above, con- invadesa region of granitic composition.The basaltic tains pillow-like inclusionsof hornblendegabbro in magma, through convectiveheat transfer, melts a granite, suggestingthat the mixing and crystallization portion of the granitic region, forming a cupola at of the magmas (assumingH2O saturation) took the upper region of contact. Fracturesmay occur place at water pressuresin excessof 1.5 kbar. that tap either layer of magma-the layers being The critical featuresillustrated in Figure 6 are that maintained by gravity and separateconvective sys- (1) under conditions where the rhyolitic magma tems-and feed into a common conduit. A single (HrO-saturated)is just all liquid below 1.5 kbar, fracture venting the cupola would yield first a rhy- that is, immediatelyabove its liquidus,the basaltis olitic flow, followed by a basaltic flow. completelycrystalline; and (2) if the basalticmagma The suggestionof Holmes was embellishedby o Wager et al. (1965) with the addition of a mixed At water pressuresless than the total pressure, basalt region, between the two con- liquid would crystallize at total pressureshigher than 1.5 zone, or hybridized kbar within certain limits of H,O content. No attempt will vectingcells and a cumulatezone at the baseof the be made here to include a discussionof all the many options invading basaltic member (Fig. 7). Schmincke introduced by insufficient H,O to saturate the magma or by (1969) appealedto a similararrangement to explain the presenceof gaseous other components.Some of the de- the compositeflows of soda rhyolite and basalt of termined effects are outlined by Holloway and Burnham (1972) and.Eggler (1972). Also see discussionin section Gran Canaria;however, he reversedthe convective below entitled "Water distribution and the role of hvdrous circulation in the soda-rhyoliticmagma chamber. phases." Blake et al. (1965) proposeda related mech- BASALTIC AND RHYOLITIC MAGMAS t59 anism whereby the granitic layer is remelted by the passageof basalt through a fracture (Fig. 8a). The mobilized granitic magma chokes off the flow of basalticmagma (Fig. 8b) and usesthe molten center of the basaltic dike as an easy accessroute to the surface (Fig. 8c). The interrupted flow of basaltic magma piles up as pillows or forms a net-veined complex in which pillows predominate. These suggestionsof Holmes, Wager and cowork- ers, Schmincke, and Blake and coworkers depend on a rather unusual crust or upper mantle in oceanic volcanic areas. Basicmagmo Mobile acid mogmo 3. Another simple solution is a chamber with two Bosicpillows Viscousacid magmo immiscible liquids. Of necessity the liquids at equi librium would have to be at the same temperature, FIG. 8. Schematic representation of the evolution of a at least along the interface. Both magmas could basic magma passingthrough a preexistingcrystalline gra- granite terminatesthe contain phenocrysts; however, the thermal and nitic layer (a), mobilizing the which flow of basic magma (b), and the granitic magma then evis- chemical properties of such a conjugate rhyolitic ceratesthe basic magma in the dike producing a composite magma would be most unusual compared with those dike and flow (c). The interrupted flow of basic magma of known rhyolitic magmas. The principle may be forms pillows in a net-veinedcomplex. From Blake el a/. illustrated by the simple systemsFeO-SiO: (Bowen (1965,p. 4,Fig.2) with permissionof the authorsand the and Schairer,7932,p.200, Fig.4) and KeO-FeO- Ceological Societyof London. Al2O3-SiO2(Roedder, 1951, p. 283, Fig. l). In the latter system two immiscible liquids in equilibrium inadequate to provide separate magmas from a with tridymite terminate in a conjugate line at homogeneous fluid phase. 1140"C; a separated"basic" liquid would crystallize Fenner (1948, p. 500) appealedto immiscibility completely at one eutectic, and the isolated "acidic" at the mixed magma localities of Katmai, Alaska, liquid would undergo reaction and with fractionation and Gardiner River, Wyoming, and suggestedthat proceed to another eutectic at a lower temperature the intermediate magmas were produced by the as- (<990'C). The final separateproducts would have similation of the basaltic magma by the rhyolitic different liquidus temperatures on remelting. Roed- magma. He believed that the assimilation was ef- der (1956) emphasizedthe dependenceof immisci- fected by exothermic gas reactions. One may recall bility on special compositions and noted that this Bowen's (1922) argument that rhyolite can assimi- property of liquids is suppressedin complex systems. late basalt exothermically dry, but by reacting and Nevertheless,two glassesof unusual composition freezing liquid, not by generating more liquid. Simi- have been identified in the residua of some lunar lar arguments hold when water is involved, except rocks. Roedder and Weiblen (1971) described co- that the accompanying release of HzO may be an existing glass globules as evidence of immiscibility endothermic process (see Khitarov et al., 1963). in the residual liquid. The normative compositions The net heat balance has not as yet been ascertained are approximately those of potassicgranite and iron- for a specific case; however, the writer believes that rich pyroxenite (o p. cit ., p. 5 19 . They reported sim- ) exothermic gas reactions, if any, are inadequate for ilar observations in the residua of six terrestrial ba- generatingmore magma. salts, and in one example the crystal content is about No doubt there are more complex schemes for 56 percent. The late-stagenature of the immiscibility producing two liquids. Ilowever, most suggestions, deduced from the normative compositions of glasses that involving immiscibility, presuppose and crystal content, therefore, precludes the simple excepting two contrasting compositions. application of this processto the generatio,nof early- the existence of the of stagebasalt and rhyolite. Preliminary data have led to the consideration The data available on hydrous systems with dis- a new hypothesis for the production of contempo- tinguishable liquid and gas phases indicate that the raneous rhyolite and basalt liquids, and these data relative volume relations of these two phases are will now be presented' 160 H, S, YODER, TR,

Exploratory Experiments horn (1962, p. 139). Central acid component "closely allied to granophyre." Some exploratoryexperiments were carried out B. Dolerite (SiOs - 48.25Vo). Complete analysis made to ascertainthe behaviorof rhyoliteand basalt under (R. R. Skelhorn, unpublished data, 1970). Lower the liquidus conditionsmost likely to obtain for basic member. each. The intent was to obtain some notion as to Scallastle Bay, Mull: Composite cone sheet (Skelhorn er a/., the relevantphysicochemical parameters and to test 1969;see Fig. 10). A. Craignurite (SiO, - 71.50%). Completeanalysis made the miscibilityof basaltand rhyolite. (R. R. Skelhorn, unpublished data, 1970). Central acid member. Experimental Conditions B. Dolerite (SiO, = 49.85Vo). Complete analysis made All experiments were performed at P;g,"o: I kbar and at (R. R. Skelhorn, unpublished data, 1970). Lower temperaturesof 950'C for 2 hours and 1200"C for t hour to basicmargin. avoid the formation of amphibole, as indicated in Figure 6. One Breiddalur, Iceland: Tufflava "emulsion" (Walker, 1963; set of conditions was chosento be above the liquidus of rhyolite Blake er al.,1965,p. 33; seeFig. 3), No. 05368. and below the solidus of basalt, and the other set ofconditions, A. Colorless acid glass (SiO, = 73.O5Vo). Partial elec- above the liquidus temperatures of both rock lypes. Two tron microprobe analysisgiven in Table l. physical arrangementsof the specimenswere tested at each of B. Black basic glass (SiO, - 52.40%). Partial electron the two setsof conditions. One arrangementconsisted of equal microprobe analysisgiven in Table 1. proportions by weight of the powders of contrasting com- Nooarupta, Katmai area, Alaska: Banded pumice (Curtis, positions, which were intimately ground together for t hour and 1968;see Fie. 2), SpecimenNo. 54 ACU 36. placed in a platinum tube with HzO in excessof that required A. Rhyolite, unanalyzed. to saturate the liquid. The second arrangement consisted of B. Andesite, unanalyzed. cylinders of the rock or glass,or of the two powders of contrast- Socorro Island, Mexico: Oxidized hawaiite-pantellerite as- ing composition, packed end-to-end in a platinum tube with sociation(Bryan, 1970). excess H:O. The tubes lay horizontal in the furnace under A. Pantellerite (SiOa - 68.91%; Bryan, 1970, p. 199, pressure with the interface of end-to-end specimensvertical. Table13, No. 1, S138). Thermal quenching was accomplishedin the internally heated, B. Oxidized hawaiite (SiO, - 46.96Vo;Bryan, 1970, p. gas-media pressure apparatus (Yoder, 1950) in less than 30 195,Table 10, No.2,Sl4l). secondsat constant pressure. Pantelleria Island, Italy: Titaniferous basalt-pantellerite as- Materials sociation (Zies, 1960, 1962). A. Hyalopantellerite (SiO, - 69.567o;Zies, 1960,p. 307, Natural materials were chosen because of their close re- Table 1. pnc 2007). lationship in the field or because of their special chemical B. Titaniferous basalt (SiO, ,= 46.31%; Zies, 1962, p. physical properties or even though unrelated in the field. 178,Table 1, rnc 2006). The samples used were unadulterated except for the addi- Gran Canaria Island, Spain: Composite flow P' (Schmincke, present tion of water, a constituent known to be in the ana- 1967,1968,1969). lyzed materials. A. Quartz trachyte. No. 979; upper half of lower mem- - I. Field-related materials used. ber of flow (SiO, 675% ) . B. Oxidized basalt. No. 983; "least contaminatedsample" (Skelhorn Craignure, MuIl: Composite cone sheet et ol., of uppermember (SiO, = 49.5%). 1969). - A. Craignurite (SiO, 71.80%). Analysis given by Skel- II. Materials not field-related but with special properties. Analytical relerencesarnples: (Fairbairn et al., l95l)' A. Granite, Westerly, R.I. U. S. Geological Survey Stan- TesrE, 1. Electron Microprobe Analyses of Clear and Black dard Sample G-l (consensusmean SiO, : 72.45%). Glass in Tuff-Lava Emulsion Rock. Breiddalur. Iceland* B. Diabase, Centerville, Va. U.S.G.S. Standard Sample W-l (consensusmean SiOn= 52.50%). Glass cylinders and powder: sio2 73.05 52.40 A. Rhyolitic obsidian, Lake Naivasha, Kenya (Bowen, Tlo2 0.19 L.62 1935,p. 493,Table 1, No. 1). (SiOn= 75.55Vo). Izog LL.23 16.10 B. Basaltic tachylite, Kilauea, Hawaii (Adams and Gib- "Feo"** 1.30 9,27 son, 1926,p. 276). (SiO, = 49.7Vo). VoO 0.18 6,L4 CaO 0.61 9.O4 Results Na2O 10n 3.77 Becauseof the similarity of results,only the re- K20 3.78 o.54 sponseto the conditionsapplied to four pairs of 94.24 98.86 sampleswill be describedin detail on this occasion. * F"qn llaLker (L963) md Blake et aL. (L965), p. 33, The exploratory nature of the experimentsis ob- Auerage of thnee ueas each: 45 W beon, 0.02 W4. '*.. lotal Ee conerted to Ee1. vious; however,the behaviorof the samplesis clear BASALTIC AND RHYOLITIC MAGMAS 161 and the directions for further systematicresearch are with the basalt-rhyolite contact as observed in the indicated. The results for the pair of analytical ref- field. The similarity of the cuspate, crenulate inter- erence sampleswill be describedfirst. face as well as its sharpness is readily discerned. Based on the sharp contact observed optically and L Mixtures ol the analytical reference samples. its assumedcorrelation with compositional contrast, Equal proportions by weight of the G-l granite the diffusion is appa:rentlyslow even when the gra- and W-1 diabase were ground together for t hour and placed in a platinum tube with HsO in excess of that required for saturatingthe liquid. After hydro- thermal treatment at |2OO"C for t hour a homo- geneous brown glass was obtained. There was no optical evidence (resolution no greater than 1 pm) of immiscibility under these conditions. There is need, however, to examine the glass quenched from liquids formed at temperaturesgreater than 1200"C to ascertain whether a possible immiscibility loop exists and is closed before reaching the liquidus. Al- though similar results were obtained in the absence of water by Ginsbergand Nikogosyan (1924), who heated mixtures of granite and iron-rich diabase at 1400'C, an electron microscope search for incipient globule formation such as that observed by Ohlberg et al. (1962) would be desirable, even though opal- escencewas not observed. In another hydrothermal run at 950'C for 2 hours the products from a mixture of the analytical ref- erence samples were clinopyroxene, plagioclase, opaques, and a glass of relatively uniform index of refraction. It was evident that with intimate mixing intermediate compositions could be obtained. Hy- bridization under these conditions and for these ma- terials is, therefore, a demonstrable process if me- chanical mixing is adequate. In another set of experimentspowdered G-l gran- ite and powdered W-1 diabaseand H2O were packed in a platinum tube so that the separate but equal proportions by weight of the powders lay end-to-end with a single common interface. After hydrothermal treatment at 1200'C and t hour the products con- sistedof a colorlessglass in the position of the granite and a dark brown glassin the position of the diabase, the two being separated by an exceedingly sharp cuspate interface convex toward the granitic glass. A photograph of the undisturbed charge after the Frc. 9. (A) Photograph of products resulting from the platinum container has been peeled back is given in hydrothermal treatment of G-l granite (top) and W-l diabase Figure 9,A'. Figure 98 shows the basalt-rhyolite con- (bottom) at 12ffi'C for I hour at Pt.s : I kbar. Turned 90" tact of the Orn6lfsfjall composite flow in eastern from run position for comparison below. Width : 3 mm. (B) (topFbasalt (bottom) of the Iceland as photographed from a polished hand speci- Photograph of rhyolite contact Orn6lfsfjall composite flow in easternIceland reproduced from men by Gibson and Walker (1963) and reproduced Gibson and Walker (1963, after p.315, Plate 8A) with per- with their kind permission. The photograph of the mission of the authors and Benham and Co., Ltd. Width : charge in Figure 9A was rotated 90o for comparison 5 cm. 162 H. S, YODER, ]R. nitic magma is highly superheatedand H2O-saturated. prevailing at 1200"C. Dr. S. Bhattacharji (personal It is concluded that liquids of greatly contrasting communication, 1972) suggestedthat the cusps are composition, even when saturated with H2O, can be related to differential volumetric thermal expansion maintained in contact at least for a limited amount due to volatiles. The cuspate interface is indeed a of time. Additional observations on this point will phenomenon worthy of detailed theoretical analysis. be presentedbelow. Samples similarly packed and held at 950'C for Interpretation of the cuspateinterface is especially 2 hours yielded a colorless glass in the position of intriguing. The initial temptation is to conclude that the granite and a recrystallizedhard cake in the posi- the hydrous diabase liquid is more viscous than the tion of the diabase. The interface was sharp, though hydrous granitic liquid.? One, however, would draw some of the granitic liquid had flowed around the the opposite conclusion if the cusps are traces of the end of the basaltic cake. Based solely on the lack of surface of convective cells. On the other hand, the coloration of the granitic glass,no reaction appeared interface possibly records the sequenceof quenching, to take place during the short time of the run. diabasicliquid quenchingto a glass (-10" poises) Further experiments were conducted at the same before the granitic liquid, thereby recording relative temperatures and time periods using powdered obsi- viscosities at difierent temperatures and not those dian from Lake Naivasha and powdered tachylite from Kilauea mixed intimately and packed in a plati- num tube with excess H2O. In addition, cylinders of the two natural glasseswere cut and placed end- to-end in the platinum tube with excessH2O. The re- sults for both physical conditions were comparable to thoso obtained from the G-1 granite and W-1 diabase. An experiment analogous to that with the cylinders of glass at l2O0"C was performed under anhydrous conditions by Bowen (192I). He heated a crucible containing alayer of plagioclaseglass over a layer of diopsideglass at about 1500'C for periods of 17 to 48 hours and measuredthe extent of diffusion by the changes of index of refraction from top to bottom of the resulting partially homogenized glass. The ex- Lower Bosic Morgin tensive diftusion observed by Bowen is not directly comparable to the results of short runs at lower tem- Frc. 10. Diagrammatic sectionof compositeintrusion at peratures using obsidian and tachylite cylinders. Altcrich Cottage, ScallastleBay, Mull, illustrating relation- ship of craignurite to basic margins of dolerite. From Skel- II. Mixtures of field-related materials. horn e, al. (1969, p. 33, Fig. 12A) with permission of the The experiments described above involve materials authors and Benham and Co.. Ltd. unrelated in the field, and experiments on materials " The conclusion is drawn intuitively from the interfa- that are intimately related in the field will now be cial tension where the pressure on the concave side is described. A well documented pair of samples was greater than that on the convex side. Viscosity usually in- kindly provided by Dr. R. R. Skelhorn from the creases with pressure; therefore, the viscosity, discounting Craignure composite intrusion of Mull (cl Fig. 10). important compositional effects, is presumed greater to be The most interesting results of the experiments run on the concave side of the interface. To the best of the writer's knowledge, there is no known dlrect relationship at the two temperatures and two physical arrays were between interfacial tension and viscosity. A comparison those from the powders packed end-to-end at Prr^o: of viscosity data obtained on rhyolitic liquids with 1 kbar and 1200oC.One cusp of the cuspateinterface 4.3-6.2 p€rcent HeO extrapolatedto 1200"C (Shaw, 1963, between the resulting glassesis shown in polished 1972) with that measured on anhydrous basalts at the same section (Fig. ll). The gray portion at the bottom of temperature (Murase and McBirney, 1970) suggeststhe vis- figure (dolerite) glass cosities may be about the same. If those basaltic liquids the is basaltic and the black were hydrous, the viscositywould probably be lessthan that portion, containing highlights from vesicles and of the hydrous rhyolitic liquids-a conclusion opposite to inclusions, is the rhyolitic (craignurite) glass. The that drawn from the interfacial tension argument. rhyolitic glass is conspicuously free of vesicles and BASALTIC AND RHYOLITIC MAGMAS 163

inclusions near the interface. The basaltic glass (gray) count for this deviation as well as the large drop in has a bright border of 20 pm and then a broad band KzO content at about 400 p.m on the arbitrary scale' of darker material about 150 pm wide. A continuous It is now realized that what appearedto be a sharp compositional profile at an angle to the trace of the contact on visual inspection is in fact a diffusion zone interface(vertical line in Fig. I 1), in a plane essentially involving at least 300 pm, although most of the com- normal to the interface, was obtained by electron positional changes take place within 60 pm. (The microprobe. The horizontal portions of the curve electron beam, as adjusted, averages over a 5-pm (Fig. 12) extending to the edge of the figure were region.) The concentration gradientswere most likely assignedthe valuesobtained by wet chemical analysis, due to diffusion in the liquid; however, other mecha- and the electron microprobe results were scaled nisms, such as vapor transport, might have been op- linearly to these oxide values. The change in index erative. The general similarity of the concentration of refraction of the two glassessuggests, however, gradients for each element measured indicates the that some hybridization had already been achieved. great influence of element interactions. The concen- Attention is called to the slight but obvious rise tration gradients appear to be consistent with the in values for TiO2, MgO, CaO, and "FeO" (total concept that multicomponent diffusion operates to Fe reported as FeO) above the initial amount in the maintain constant volume and to preservelocal elec- basalt some 100-150 pm from the observed inter- trical charge neutrality. These factors are presumed face. A similar rise appears to take place for K2O to be accommodatedby the diffusion of units of struc- in the rhyolitic glass. However, inclusions may ac- ture in the liquid related to speciesor subspeciesof minerals.8 The asymmetry of the gtadients relative to the optical interface suggeststhat the rate of dif- fusion is greater in the basalt than in the rhyolite. When time dependence and accurate compositions are obtained, it will be possible to determine the effective difiusion coefficients. Experiments evaluat- ing the diffusion characteristicsof these contrasting liquids at higher pressures,particularly where basalt and rhyolite are themselvesno longer stable assem- blages,would also be of interest. For glass in other fragments exhibiting the same type of interface, partial electron microprobe analy- sis at some distancefrom the interface indicated some

" The units of structure in the liquid no doubt changewith temperature.It is common practice at the GeophysicalLab- oratory to "acclimate" a glass prior to crystallization by quenching liquids from successivelylower temperatures with intermediate crushing of the glass. After such treatment' crystallizationbelow the liquidus takesplace rapidly' usually without intermediate precursors. Crystallization of a glass formed by quenching a liquid from the high temperatures required to achievesolution of the constituentsand homog- enization usually yields few crystals or metastablephases' Relatively abrupt changesin viscosityof silicateliquids with temperature are also known, and the structure of the liquid is assumed to have changed accordingly. Diffusion rates of the structurally related elements would be expected to the Ftc. 11. Polished section of part of the interface formed change as well. Additional support for the view that found in betweenmelts of basalt (bottom, gray) and rhyolite (top, black) multicomponent difiusion is structurally related is from the Craignure, Mull, composite intrusion. Vertical line the vapor transport experimentsof Morey and Hesselgesser in marks the approximate position of the electron microprobe (1951). They found that componentswere transported known traverse(see Fig. 12).Conditions of the experimentwere 1200'C the vapor for the most part in the proportions of as a and t hour at Ps"o : I kbar, with powdered samplesof each mineral species.A structural study of silicate liquids rock type placed end-to-end in a platinum tube. function of temperature would be of considerable interest' 164 H. S. YODER, ]R.

averagingseveral points measured with a broad beam (45 pm) and low current(0.02 pA). The sharpnessof 72 the interface was determinedby traversing for the 68 elementsTi, Fe, and Si simultaneouslywith a 5-pm 64 beamand wasequal to or lessthan the beamdiameter.

56 The remarkable sharpnessof the interface, even 52 without correcting for possible edge effects in the E +'FeO' 48 electron microprobe analysis, signifies that unique 2 +Co0 5 conditionsexisted under which the liquids wereunable I K2O+ to diffuseor mix. Theseconditions cannot be defined; however,they might include(1) immiscibility,(2) short time in contact, and (3) high viscosityoccasioned by + MgO volatile loss. Immiscibility is not likely becausethe powders of each glass phase, with phenocrystsre' moved, interdiffusedand becamehomogeneous when mixed and held at 1200'Cfor I hour under P",o : Frc. 12. Schematic presentation of compositional changes I kbar. If the time of contact was short, the mixing across the melted basalt-rhyolite interface near the path presumablytook place immediatelybefore eruption shown in Figure 11. Oxide changes calculated from electron and the quenchclosely followed. The electronmicro- microprobe traverses are correlated with optical observa- probe analyseshave low totals, and it is presumedthat scale is arbitrarily chosen. tions. Origin of micrometric mostof the missingmaterial is HrO. SomeHrO would havebeen lost during the eruption; on the other hand, heterogeneityor hybridization.In general,the analy- in view of the absenceof vesicles,HrO may havebeen sesof the elementsmeasured lay on or near straight gained after quenching.nIt is consideredunlikely, lineson a SiOz-variationdiagram, with the exception therefore,that the coexistenceof the two glassesof of AlzOs, which is higher than expected,and KzO, contrasting composition can be attributed solely to which is lower than expected,from a straight-line high viscosityattending loss of volatiles. extrapolation.The lack of discolorationof the rhyo- It may be concludedfrom the experimentsusing litic glassmay, in fact, be deceiving.It is suspected pairs of natural materials,related and unrelatedin that transfer of material takes place through the in- the field, and from the publishedfield interpretations terface (not necessarilythe original contact of the that the identity of two contiguousmagmas of diver- two materials)and that the optical demarcationbe- gent compositionmay be maintainedfor at least a comesdiffuse only after much of the compositional limited amountof time. contrasthas been reduced. It is alsopossible that iron TentativeHypothesis in one electronicstate (i.e., Fe3-) is diffusedprefer- entially into rhyolitic liquid and no discolorationis It is of interestto inquire whethersuch contiguous produced (cf. Fig. 5). Miissbauerstudies of the magmasof contrastingcomposition can actually be glasseswould, no doubt, resolvethat issue. generatedfrom the sameparent without intermediate Of considerableinterest was the distribution of members rather than from separateindependent HzO betweenthe two liquids.An attemptwas made sources.The mechanismto be presentedstems from to map the H- with an ion probe. Qualitativelythe the belief that the heterogeneityof the mantle may rhyolitic glass contained more H* than the basaltic nThe (0.7089 -+ 0.0010) is glass. However, dewateringof the glassesalong Sf?/Sre of Breiddalur basalt anomalous and about that of sea water (Moorbath and cracks after quenchingwas evident, giving rise to Walker, 1965). Isotopic exchangewith sea water may have highly variable distributions. accompanied the absorption of water' The interface of a natural "emulsion" of basaltic The identification of juvenile water and determination of and rhyolitic glasswas studied for comparison.The the extent of mixing with meteoritic water are formidable facing the geochemist. Geothermal systems inves- specimen from Breiddalur, Iceland (courtesy of tasks still tigated with present methods indicate little or no magmatic Dr. G. P. L. Walker), is illustrated in thin sectionin component. In addition, primary hydrous minerals of ig- Figure 3. The bulk composition of each glassphase neous rocks may have exchanged their OH with ground- was obtained by electron microprobe (Table 1) by water. BASALTIC AND RHYOLITIC MAGMAS 165 be largewith respectto the amountsof phases;how- ever, as long as the kinds of phasesdo not change within a mantle province,the liquids producedby fractionalmelting will be relativelyuniform in com- position(see Yoder and Tilley, 1962,p.518). The uniformity of magma types produced around the world results primarily from the singularityof the physicochemical"invariant" conditionsthat deter- minethe compositionof the firstliquids in suchheter- ogeneoussystems. The greatfloods of basaltclearly require some physicochemicalcontrol to accountfor their uniformity.The closerelationship of major rock typeswith the assemblageof phasesat the invariant pointsin a pertinentfive-component system was con- sideredstrong support for this conceptby Schairer and Yoder (1964, pp.7O-74). For thesereasons a dominantrole is assignedto fractionalfusion in the generationof magmasof contrastingcomposition. A careful geometricalanalysis of fractionalfusion has Frc. 13. rrt" aiopr;oJ'rorsterite-silicasystem at Ps,o : 20 been made by Presnall (1969), and some of the kbar (Kushiro, 1969). The X marks an assumed parental its principlesoutlined by him are appliedhere. composition, and the arrows from it indicate the change of composition resulting from continuous removal of Iiquids I and The data of Kushiro (1969) on the diopside- B with heating. The dotted lines are construction lines. Dashed forsterite-silicasystem at Pn"o : 2Okbar serveas a lines indicate limits of three-phase assemblagesstable im- suitable exampleto illustrate a possiblemechanism mediately below invariant point temperaturesof A and B. Di' Opx, ortho- for the generationof both magmasfrom a common diopside; Cpx, clinopyroxene; En, enstatite; pyroxene; Fo, forsterite; Ol, olivine; Q4 quartz. parent. The pressure of 20 kbar may be slightly excessiverelative to the depthsof origin presumedfor such magmas;however, the principles involved are may migrate up a fracture away from the site of informative,and the relationsin the simplesystem will senerationor find a suitablereservoir. " the remaining not changedrastically at somewhatlower pressures. Beforethe resumptionof meltingof C, Consider a parental material of quartz-normative parentalmaterial, which now hasthe composition compositionanalogous to X in Figure 13. If the it is necessaryto raise the temperatureto 1'220"C liquid B temperatureis raisedto 960oC,melting of composition (point B in Fig. 13). Continuousremoval of X in the presenceof excessHrO beginsand yieldsan in a similar way drivesthe bulk compositionof the initial liquid of the compositionat point I in Figure remaining parental material toward D, where de- 13.Removal of that initial melt as soonas it is formed pletion of another major phase, diopside solid causesthe bulk composition of the parent to move solution (Cpx), occurs. About 22 percent of away from the liquid composition. Continuous parentalmaterial with the compositionC has been isothermal removal of liquid of the composition I removedfrom the parentalmaterial as basic liquid eventuallycauses the bulk compositionto move to C. having the compositionB. The basicliquid B may Alternatively, the liquid may be left in equilibrium follow up the fracturebehind the siliceousliquid ,4, with the parental material and removed after all occupythe samereservoir as the siliceousliquid ,'4, possibleliquid is producedat 960'C. The residual or seek a separatereservoir. Further meltingof the crystalswould also have the bulk composition of C. residual parental material D produces a series of In either eventthe critical observationis that the bulk liquids along the olivine-orthopyroxeneboundary composition of the remaining parental material has curve until the temperatureof 1295"C is reached changedto a point on the Di-En join. Herethe phase on the En-Qz join. qtJartz of the parental material is exhausted,and In brief, two homogeneousliquids have been re- melting ceasesafter about 20 percent of siliceous movedfrom a commonparent without the production liquid has been removed from the original bulk of liquids of intermediatecomposition. The analogy to liquid composition.The liquid, occupyinga largervolume, of rhyolitic magmato liquid A andandesite 166 H. S. YODER, ]R.

B is evidentand has alreadybeen drawn by Yoder the block of parental material relative to the (1969).10The parentalresiduum D canbe described isotherms,to meet the requirementsof observed as a pyroxeniteor saxonite. eruptive cycles. D. K. Bailey (in Newall and Some of the critical aspectsof this hypothesis Rast, 1970, pp. 1.77-186)suggested that melting require examinationwith regard to ( 1) volume of is effectedby heat focusing and fluxing by vola- parentalmaterial involved, (2) heat requirements, tiles in zonesof lower pressureor tension. (3) time involved,(4) extractionand storageof 3. Time inaolued.The time requiredfor raising magmas,(5) water distributionand role of hydrous the temperaturebetween the respectiveliquidus tem- phases,(6) trace elementpartition and isotopes, peraturesof the associatedmagmas and supplying (7) hybridizationpotential, (8) eruptivesequence, the latent heat of melting would probably exceed and (9) normativecharacter of parentalmaterial. severalthousand years.11 The temperatureinterval 1. Volume ol porental material inuolaed. A sin- betweenthe liquidustemperatures at relevantdepths gle eruptiveevent rarely exceeds2 km3 in volume. may be much greaterthan that in the simplesystem, On the basisof the simplesystem described and the dependingon the water contentof the parentalma- choice of bulk composition,the fractional melting terial. The time requiredto achievethe liquidustem- processneed involve only L0 kms, for example. peratureof the basic melt is, however,not particu- Volumes of an order of magnitudelarger would larly relevantbecause the hybridization process would probably be required if the parental material is not becomesignificant until the basicmagma began peridotite. The small volume of residual parental to evolve and the magmaswere brought into con- material, eventually reduced to pyroxenite or sax- tact. The Daly gap is not, in fact, time dependent. onite, could not contribute significantlyto the pro. The time interval does become important in the duction of rhyolite or basalt. Small volumes of compositeintrusions and extrusionsbecause of the parental material would be readily accommodated storageproblem discussed immediately below. in cyclical pr@essesinvolving conduction of heat, 4. Extraction and storage ol magmas.The diffi- diftusion of volatiles, or mechanicaltransport. culty of removal of small amountsof magmacon- 2. Heat requirements.The ever-presentproblem tinuously from parental material has been of con- of providing essentiallya local sourceof heat for siderableconcern to many igneouspetrologists. The production of successivesmall batchesof magmacan removal of the residual liquid from monomineralic possiblybe resolvedby Successiveadiabatic decom- cumulatesin layeredintrusions as well asin the cruci- pressionsof relativelyflat plates (Yoder, 1952, pp. ble would indicatethat the processis not only pos- 370-373). Alternatively, a convectivecycle may siblebut efficient.In any case,the batchremoval of carry new parentalmaterial within rangeof a fixed the first-stageliquid, in part or in its entirety, an hot spot. Basedon present-dayearth plate motions alternativealready mentionedabove, would alleviate (1 to 10 cm per year),the rateof supplymay, how- the problem. Brace et al. (1968, p. 2l) indicate ever, be too slow, depending on the geometry of in their experimentson the onsetof melting of granite under an effectiveconfining pressure that a few per- loThe formation of "rhyolite" at eutectic A in the syn- cent melting is sufficientto form interconnectingpas- thetic system is compatible with the known thermal behavior sagewaysbetween grain boundaries.Movement of of natural rhyolite; however, the formation of ,,andesite" at reaction point B may not at first glance appear to be the liquid no doubttakes place in responseto thermal compatible with the thermal behavior of natural andesite in expansion on melting, density difference,compres- the presence of excessH,O. This apparent anomaly is read- sionalstresses, and the pressuredrop accompanying ily explained by considering tbe crystallization of liquid B. fracture. The course of crystallization follows the Opx-Cpx boundary Storageof the generatedmagmas for frequent curve, and final crystallizationoccurs at the eutecticA. The large temperature range of crystallization of .,andesite"B eruptivecycles introduces other problems.The HzO- is in accord with that observed in natural andesitesin the saturatedsiliceous liquid has essentiallya minimum- laboratory and in the field. On the basis of this simple sys- like or eutectic-likebehavior. Generallv the meltine tem, the temperatureof the beginning of melting of ,.rhyo- lite" and its associated"andesite" should be the same. A " An exact calculation of the time required is not possible comparison of the hydrous melting curyes of unrelated gran- because of the many variables involved. Some of the pa- ite and basalt (Yoder and Tilley, 1962,p.463, Fig. 33) rametersare discussedby Jaeger (1964), and estimatesmay suggeststhat this relationship is closely approached at high be made for specific geometries of parental material having water pressures. assignedthermal properties. BASALTIC AND RHYOLITIC MAGMAS 167 and crystallizationinterval may be of the order of H2O, the parental material would then have to 30"-100"C in naturalsiliceous systems, depending be raised to a much higher temperatureto pro- on the composition(see Piwinskii and Wyllie, l97O). duce a liquid analogousto hydrousliquid B (Fig. Thus, cooling,of necessity,must be limited to pre- 13). Kushiro (1969, p. 282) has studieda por- vent freezingin the conduit. Although the time in- tion of the Di-Fo-SiO2 system at P = 20 kbar terval required for the parental residua to reach the under anhydrousconditions, and the relevant in- liquidus of the basalticor andesiticmagma is rela- variant point on the Di-En join for the bulk com- tively small, a small drop in temperaturebelow the positionC has a temperatureclose to 1650'C. The liquidusof the saparatedsiliceous liquid would pro- larger temperatureinterval betweenmelting stages ducea largeamount of crystallization.lzOn the other increasesthe heat requirements and the time involved. hand, the cooling problem is not critical in the asso- Vesiculation of the rhyolitic magma on pressure ciatedbasaltic or andesiticliquid (e.9.,liquid B, Fig. reduction(Yoder, 1965)as it risesto higherlevels in 13) because of the large temperature interval the crust probably initiates the explosive eruptive betweenliquidus and solidus (Yoder and Tilley, phase.Preservation of phenocrystsunique to each 1962). Heat transferfrom the basalticmagma may magma(e.g., "emulsion" tuff lava from Breiddalur, maintain the thermal state of the siliceousmagma, Iceland)indicates that the mixing took placeprimarily as suggestedby Holmes (1931), but not without duringthe turbulentperiod of simultaneouseruption. thermal effectson the basaltic magma. Release of volatiles may also be initiated as the 5. Water distribution and the role ol hydrous hydrous rhyolitic magma reacts with phenocrystsof phases.Variations imposed by the H2O contentand the basaltic magma. The schematicalbite-anorthite- its initial disposition(free phaseor in hydrousmin- HrO systemat Ps"o : 5 kbar and 800oC,in Figure14, erals) are manifold. It is perhapsuseful to record is particularly instructive in this connection. If one set of conditionsas a basisfor discussion.It is 20 percentof a basic plagioclase(An'o) or a basic assumedthat water is not a free phasein the mantle hydrous mineral (e.g., lawsonite) is added to a but is containedin a commonhydrous mineral (e.9., relativelyalbite-rich, HrO-saturated magma (l), the phlogopite)present in rocks believedto come from bulk compositionshifts into the crystal{ liquid * gas the mantle (see Yoder and Kushiro, 1969). The region (X) and gas is released.A thorough analysis hydrous mineral is probably of sufficientquantity to producerhyolitic magmawithout beingconsumed in Hzo the first stageof the fractionalmelting process.'3 If, however,the hydrous mineral is the phasethat is kb P""=5 consumedin the first stageof fractional melting, sub- t\ir =BOO'C sequentliquids produced at the next thermallevel of T meltingwill be anhydrous.That is, basaltwould be expectedrather than andesite(Yoder, 1969). If the secondstage of melting proceedswithout G+L u In addition, the lowering of pressureon adiabatic rise of the separated,HrO-saturated magma to shallower depths would produce a large amount of crystallization. Magmas undersaturated with regard to tLO would, on the other hand, melt sufficiently on pressure reduction to offset cooling. -r'---\,---Lowsonite '3 Kushiro et aI. (1972) melted a mixture of natural garnet i i lherzolite minerals and 2 percent synthetic fluorophlogopite in which the initial IGO - 0.25 percent. All of the mica Ab An.. Anzo An was consumed in the liquid at 1500"C and 30 kbar when the liquid content reachedabout 10 percent (see their Table Weightper cent 2, p.21). It would appear,therefore, that the mica content Frc. 14. Estimate of the albite (Ab)-anorthite (An)-HzO of the parental material would have to exceedseveral per- system at Ps.o : 5 kbar and Z : 800"C based on data of cent to satisfy this requirement. This amount of mica and Yoder et al. (1957, p. 207, File. 36). Bulk composition obtained its KeO content may not be acceptable in some models of by addition of 20 percent basic plagioclase,Anzo, or a basic the mantle because of excessiveradiogenic heat production hydrous mineral (e.g., lawsonite) to a saturated albite-rich from K*. magma, l, marked with an X. G, Eas;I, liquid; Pl, plagioclase' 168 H. S. YODER, IR,

of the assimilativeprocess in various types of hydrous tals) only tends to reinforce the normal lines of liquid systems is needed but will not be undertaken here. descent.However, an examination of the SiOz varia- 6. Trace element partition and isotopes.The trace tion diagram of a magma series produced by frac- element partitions will depend first on whether the tional crystallization indicates that magmas obtained elementsare saturated or undersaturated.Exhaustion by liquid-liquid mixing, hybridization in the strict of phasesin the parental material on fractional melt- senseof Durocher (1857), are not initially confined ing will produce subsequentdiscontinuous changes in to the normal lines of descent (see Harker, 1904, trace element composition in the liquid. Initially, the p. 232, Figs. 50 and 51). A chemical study of the first liquid formed will concentrate those elements glassy rocks may reveal whether magma mixing or which preferentially partition into the liquid. The fractional crystallization (and its corollative reactive parental mineralogy in kind and proportions will assimilation) is more applicable to the evolution of obviously have great influence on the distributions. the calc-alkalineseries. Magma produced in successivestages will be greatly 8. Eruptiue sequence.In generalthe compositional reduced in those preferred elements. sequenceof eruption is rhyolitic followed by basaltic These deductions appear to account in general for magma, and the fractional melting hypothesissatisfies the trace element distributions in depleted oceanic this observation. On the other hand, the composite tholeiites. In addition, the production of alkali basalts dikes and composite lava flows more often indicate in the oceansimplies that a mica, or equivalent phase, the reverse sequence of magma emplacement. The was retained in the parental material after the first conduits may no doubt feed the magmas in the latter stage of melting. There are, however, obvious diffi- sequence,but such fortuitous plumbing doesnot have culties in correlating these deductions with depth of worldwide application. Buist ( 1952) attributed the origin in the mantle, degree of melting, and deter- sequence to viscosity differences, the less viscous mined partition coefficients in crystal-liquid equi- basic magma progressingmore rapidly to the surface. libria. Future tests of this hypothesis will no doubt Similarly, Gibson and Walker (1963, p. 3I7 ) believe rely heavily on the trace elements. that viscoussiliceous magma is "rendered more fluid" Difterences in $1sz/$1s0should be small between by the passageand incorporation of the hotter basic rhyolitic and basaltic magmas derived by fractional material that precedes the siliceous magma to the melting from the same parent. The similarity of this surface. The volumetric predominance of acid intru- ratio for some of the associatedrhyolites and basalts sives over basic intrusives would appear to support (and andesites) of Central America (McBirney and their view; however, a wholly satisfactory solution to Weill, 1966) and Iceland (Moorbath and Walker, the sequenceproblem is not at hand. 1965) lends support to this hypothesis.H. P. Taylor 9. Parental material. The assumption of a paren- (in McBirney and Weill, 1966, pp. 11-12) has tal material capable of producing quartz-normative determined the O's7O'e for rhyolite obsidians and liquids is not without some foundation. Studies of basalts in Central America and concluded that the simple hydrous systemsinvolving olivine, clinopyrox- rhyolite obsidians could not have been derived by ene, orthopyroxene, and spinel, a possible mantle remelting of the basement rocks and are more di- assemblage, indicate that the first liquids (HrO- rectly related to the associatedbasalts. The sensitivity saturated) arc quartz-\ormative (Yoder and Chin- of these isotopesto exchangemay temper interpreta- ner, 1960, p.79; Yoder, 7971,p. 178). Direct de- tion in other areas. termination of the composition of the liquids derived 7. Hybridization. The contiguous magmas (Fig. from the hydrous partial melting of natural perido- 7) may hybridize if the storageperiod is long or the tite demonstrated that the liquids are quartz-norma- interfacelarge as outlined by Wager et al. (1965). tive (Kushiro et al., 1972).In the latter experiments Contiguous magmas confined to a dike system would the liquids were describedas andesiticor dacitic after have limited opportunity to hybridize until flow be- 20 percent melting had occurred at25 kbar. At some- gan in the eruptive stage.The hybrid rocks observed what lower pressures and with a smaller degree of at Gardiner River, Wyoming, Gran Canaria Island, melting, it is believed that the H2O-saturated liquid and in the Katmai area suggestthat mechanical mix- would be analogousto liquid I in trigure 13, that is, ing and diffusion between magmas had taken place. rhyolitic. Olivine would of necessity be present in Bowen (1928, p. l2l) argued that hybridization of that assemblageat high pressurein the natural multi- the reactive type (magma consuming foreign crys- component system. BASALTIC AND RHYOLITIC MAGMAS 169

The basic principle of the hypothesis is equally applicable to a nepheline-normative parental mate- rial. For example, the relations in Figure 15 illus- trate a eutectic at C, analogous in assemblageto phonolite, and a reaction point at B, analogous in assemblageto basalt. Two liquids of the composi- tions B and C could be removed by fractional melt- ing from a nepheline-normative ultrabasic parent without intermediate members. The contrast in SiOz content of the anhydrous liquids at B and C is ob- viously small in the simple system, but the principle remains clear in its application to nepheline-norma- tive natural materials. It is probably evident that a simple ternary eutec- tic system is equally capable of yielding two liquids of contrastingcomposition. Presnall (1969, p. 1180, Fig. 1) demonstrated that the ternary eutectic com- position and any one of the three possible binary Werghl per cenl eutecticswould yield liquids of considerablecontrast Frc. 15. Hypothetical representation of the nepheline on fractional melting of any composition in the sys- (Ne)-forsterite (Fo)-silica system at P = 1.5 kbar, based tem. The exhaustion of a phase in the parental ma- in part on findings of Kushiro (1965, p. 106, Fig. 24), il- terial is the critical event that brings about the lustrating possible existence of invariant points B and C, hiatus in extracted liquid compositions. representing compositions of contrasting magmas which could be extracted from a nepheline-normative parental ma- terial by fractional melting. En, enstatite; quartz; Ab, Concluding Remarks Qz, albite; Sp, spinel; Jd, jadeite. From Yoder (196a, p. 100' There is little doubt that the principal framework Fig.25). of igneouspetrology is the piezochemical"flow sheet" for liquids that occur in the natural multicornponent Washington, D. C.), B. G. J. Upton (University of Edin- (Imperial system.Fractional crystallization provides the mech- burgh, Edinburgh, Scotland), G. P. L. Walker College of Science and Technology, London, England)' anism for extending the liquid line of descent. The R. A. Wiebe (Frariklin and Marshall College, Lancaster, paucity of some members of the magma series has Pennsylvania),R. E. Wilcox (U. S. GeologicalSurvey, Den- been attributed to the imperfect way in which the ver, Colorado), and colleaguesat the GeophysicalLabora- mechanism operates. The tentative hypothesis ad- tory. M. Bell' vanced above outlines one possible way in which The manuscript was kindly reviewed by Drs. P. F. Chayes, D. H. Eggler, F. N. Hodges, T. N. Irvine, liquids of high contrast in composition originate. I. Kushiro, and G. T. Stone of the GeophysicalLaboratory; Fractional melting also operatesimperfectly, and the D. C. Presnall of the University of Texas at Dallas; extent to which each process obtains from province S. R. Hart of the Department of Terrestrial Magnetism; to province and from time to time remains to be and R. E. Wilcox and D. B. Stewart of the U. S. Geological analysis of the seen.The criteria for identifying theseprocesses from Survey. Their prompt review and critical exploratory experiments and tentative hypothesis are greatly the products alone have yet to be enumerated. appreciated. Acknowledgments References The writer is grateful to many field geologistsfor their Anevs, L. H., eNo R. E. GrssoN (1926) The compressi- samples,photographs, discussions, and kindnessin conduct- bilities of dunite and of basalt glass and their bearing on ing field excursions.Among theseare Drs. W. B. Bryan the composition of the earth. Proc. Nat. Acad. Sci. USA, (WoodsHole OceanographicInstitution, Woods Hole, Mas- 12,275-283. sachusetts),G. H. Curtis (Universityof California,Berke- Blnrpn, D. S. (1965) Alkalic rocks at Litchfield, Maine' ley, California),I. L. Gibson(Bedford College, University I. Petrology,6, 127. of London, England), G. Kullerud (Purdue University, Brere. D. H., R. W. D. Erwr,lr, I. L. GnsoN, R. R. Srer- Lafayette,Indiana), H. U. Schmincke(Ruhr University, HoRN,AND G. P. L. Wexr,n (1965) Some relationships Bochum,Germany), R. R. Skelhorn(Sir JohnCass College, resulting from the intimate association of acid and basic London,England), D. B. Stewart(U. S. GeologicalSurvey, magmas. Quart. J. GeoI. Soc. London, l2lr 3l-50. t70 H, S. YODER, IR.

BowrN, N. L. (1921) Diffusion in silicate melts. I. GeoI. Mem. Geol. Suru. U.K., J. Hedderwich and Sons, Glas- 29, 295-317. gow,481pp. (1922) The behavior of iriclusionsin igneous mag- Hewxrs, L. (1945) The Gardiner River rhyolite-basaltcom- mas.I. Geol.30, 513-570. plex. Geol. Mag. E2, 182-184. (1928) The Euolution of the IgneousRocls. Prince- Hor-rowev, J. R., ,cNo C. W. BunNHeu (1972) Melting ton University Press,Princeton, N. J., 332 pp. relations of basalt with equilibrium water pressure less (1935) "Ferrosilite" as a natural mineral. Amer. J. than total pressure.l. Petrology, 13, l-29. ,Sci.30, 481-494. Hollms, A. (1931) The problem of the associationof acid eNo J. F. Scnernsn (1932) The system FeO-SiO,. and basic rocks in central complexes. Geol. Mag. 68, Amer. L Sci.24, 178-213. 241-255. Bovn, F. R. (1961) Welded tuffs and flows in the rhyolite JeEcrn, I. C. (1964) Thermal effects of intrusions. Reu. plateau of Yellowstone Park, Wyoming. Geol. Soc. Amer. Geophys.2, 443-466. Bull. 72, 387-426. Juon, J. W. (1893) On compositedykes in Arran. Quart.I. Bucr, W. F., W. C. LurH, ANDJ. D. UNcEn (1968) Melt- Geol. Soc. London, 49, 536-564. ing of graniteunder an effectiveconfining pressure (abstr.). KENNEnv,W. Q. (1931) On compositelava flows. Geol. Geol.Soc. Amer. Spec.Pap.115,21. Mag. 68, 166-181. BnveN, W. B. (1970) Alkaline and peralkaline rocks of Ksrr,rnov, N. I., A. S. Kerrr, lNo E. B. Lerrosv (1963) Socorro Island, Mexico. CarnegieInst. Washington Year Estimate of the thermal effect of the separationof water Book, 68, 194-200. from felsic melts based on data for the system albite- Bursr, D. S. (1952) A contribution to the petrochemistry water. Geochemistry(USSR), 7, 637-649. and petrogenesisof the composite sill of South Bute. Kusnrno, I. (1965) The system forsterite-nepheline-silica. Trans. Edinburgh Geol. Soc., Campbell VoI., 15, 52-68. CarnegieInst. Ilashington Year Book 64, 106-109. BuNsEN,R. W. (1851) Uber die Processeder vulkanischen (1969) The system forsterite-diopside-silicawith GesteinsbildungenIslands. Ann. Phys. Chem. 83, 797- and without water at high pressures.Amer. l. Sci., 272. Schairer Vol., 267-A, 269-294. CnevEs, F. (1963) Relative abundance of intermediate N. SHrvrrzu,Y. Nerevunl, eNo S. Axnrroro (1972) members of the oceanic basalt-trachyte association."I. Compositionsof coexistingliquid and solid phasesformed Geophys.Res. 68, 1519-1534. upon melting of natural garnet and spinel lherzolite at Cunrrs, G. H. (1968) Stratigraphy of the ejecta from the high pressures:A preliminary report. Earth Planet. Sci. 1912 eruption of Mount Katmai and Novarupta, Alaska. Lett.14, 79-25. Geol. Soc. Amer. Mem. 116, 153-210. Lurn, W. C., R. H. JeHxs, eNo O. F. Turrr-e (1964) The Derv, R. A. (1925) The geologyof AscensionIsland. Proc. granite system at pressuresof 4 to 10 kilobars. I. Geo- Amer. Acad. Arts Sci.60, 1-80. phys.Res.69,759-773. DunocHen, J. (1857) Essaide p6trologiecompar6e. Annales MecooNero, G. A., eNo T. Kersune (1965) Eruption of des Mines, Memoires (Paris),Ser. 5, 11,217-259. Lassen Peak, Cascade Range, California, in 1915: Ex- Eccrrn, D. H. (1972) Amphibole stability in H,O-under- ample of mixed magmas.Geol. Soc. Amer. Bull. 16, 475- saturatedcalc-alkaline melts. Earth Planet. Sci. Lett. lS, 482. 28-34. McBrnNEv, A. R., .qxo D. F. Wsrrr (1966) Rhyolite mag- FrnneRN, H. W., et al. (1951) A cooperativeinvestigation mas of Central America. Bull. Volcanol. 29, 435-448. of precision and accuracy in chemical, spectrochemical, MrLruoLrEN, G. L. (1971) Melting of nepheline syenite and modal analysis of silicate rocks. U. S. GeoL Suro. with ILO and H,O + CO", and the efiect of dilution of Bull. 980, 71 pp. the aqueousphase on the beginning of melting. Amer. I. FrNNrn, C. N. (1938) Contact relations between rhyolite Sci. 270, 244-254. and basalt on Gardiner River, Yellowstone Park. Gaol. Moonnern. S.. eNo G. P. L. Wer-rrn (1965) Strontium Soc.Amer. Bull.49, 144l-1484. isotope investigationof igneous rocks from Iceland. Na- (1944) Rhyolite-basaltcomplex on Gardiner River, ture (London), 207, 837-840. Yellowstone Park, Wyoming: A discussion.Geol. Soc. MonEv, G. W., lNo J. M. Hrssnrcnssrn (1951) The solu- Amer. BulL 55, 1081-1096. bility of some minerals in superheatedsteam at high pres- (1948) Immiscibility of igneous magmas.Amer. I. stres. Econ. G eol. 46, 82l-835. Sci. 246, 465-502. MURASE,T., eNo A. R. McBInNev (1970) Viscosity of lunar Gesr, P. W. (1968) Trace element fractionation and the lavas.Science,16T' 149I-1493. origin of tholeiitic and alkaline magma types. Geochim. NEwALL, G.,,rNr N. RAsr (1970) Mechanism of igneous Cosmochim.Acta, 32, 1057-1086. intrusion. Geol. L, Spec.Issue 2. GrBsoN, I. L., eNn G. P. L. WerrEn (1963) Some com- Onrnrnc, S. M., H. R. Gor-os, eNo D. W. Srnrcrrrn (1962) posite rhyolite/basalt lavas and related composite dykes Crystal nucleation by glass in glass separation.ln, Sym- in easternIceland. Proc. Geol. Ass.74,301-318. posium on Nucleation and Crystallization in Glassesand GrNsnnnc,A. S., rNo C. S. NrrocosveN (1924) The fusion Melts, American Ceramic Society, Inc., Columbus, Ohio. of diabase and granite. Bull. Geol. Comm., Leningrad, O'Kseru, I. A. (1963) Tektites. University of Chicago 43r 735-765 [in Russian;French summary]. press,Chicago, Illinois, 228 pp. Hln'rln, A. (19M) The Tertiary igneous rocks of Skye. Prwrwsrrr, A. J., lNo P. J. Wynrs (1968) Experimental BASALTIC AND RHYOLITIC MAGMAS l7r

studiesof igneous rock series:A zoned pluton in the UrroN, B. G. J., J. E. Tnouls, ANDR. MAcDoNt-o (1971) Wallowa batholith, Oregon. I. Geol. 76, 205-234. Chemical variation within three alkaline complexes in AND-- (1970) Experimentalstudies of igneous south Greenland.Lithos, 4, 163-184. rock series: Felsic body suite from the Needle Point plu- eNo W. J. Waoswonrn (1967) A complex basalt- ton, Wallowa Batholith,Oregon. J. Geol.78, 52J6. mugearite sill in Piton des Neiges volcano, Reunion. PnrsNnr, D. C. (1969) The geometricalanalysis of partial Amer. Mineral. 52, 1475-1492. fusion.Amer. l. lci.267, 1178-1194. W.rctn, L. R., ,rNo E. B. B,rrlsv (1953) Basic magma Rotnoen, E. (1951) Low temperatureliquid immiscibility chilled againstacid magma. Nature (London), 172, 68-69. in the system K,O-FeO-AlgO,-SiO".Amer. Mineral. 36, E. A. VrNcrNr, G. M. BnowN, ANDJ. D. BELL 282-286. (1965) Marscoite and related rocks of the Western Red (1956) The role of liquid immiscibilityin igneous Hills complex, Isle of Skye. Pfii1. Trans. Roy. Soc. Lon- petrogenesis:A discussion.I. Geol.64, 84-88. don, 257,273-307. ANDP. W. Wersr-rN (1971) Petrologyof silicate Wllren, G. P. L. (1963) The Breiddalurcentral volcano, melt inclusions,Apollo 11 and Apollo 12 and terrestrial easternlceland. Quart. l. Geol. 9rc. London, 119,29-63. equivalents. Proc. Second Lunar Sci. Conl., Geochim. ANDR. R. SxsLEonN (1956) Some associationsof Cosmochim, Acta, Suppl. 2, l, 507-528. acid and basic igneous rocks. Earlft-Sci. Reu. 2,93-109. ScHlrnEn,J. F., eNo H. S. Yoorn, Jn. (1964) Crystal and Wrlcox, R. E. (1944) Rhyolite-basaltcomplex on Gardiner liquid trends in simplified alkali basalts. Carnegie Inst. River, Yellowstone Park, Wyoming. Geol. Soc. Amer' WashinptonYear Book 63. 65-74. Bull. 55,1047-1080. ScnurNcxt, H. U. (1957) Cone sheet swarm, resurgence Wrnrevs, Howrr (1952) Volcanic history of the Meseta of Tejeda caldera, and the early geologichistory of Gran Central Occidental, Costa Rica. Calil. Uniu. PubI. Geol. Canaria.Bull. Volcanol. 31, l5t-162. lci.29,145-180. (1968) The geologicframework and origin of alkali Yoorn, H. S., Jn. (1950) Highlow quartz inversion up to trachytic to alkali rhyolitic ignimbrites on Gran Canaria, 10,000bars. Trans. Amer. Geophys. Union, 31, 827-835' Canary Islands. In, Abstracts, International Symposium (1952) Change of melting point of diopside with on Volcanology, Spain-Canary Islands, Int. Ass. Vol- pressure.I . Geol. 60, 364-314. canol. Chem. Eattlt's Interior. (1964) Genesis of principal basalt magmas. Car- (1969) Petrologie der phonolytischen bis rhyolitis- negie Inst. Washington Year Book, 63' 97-l0l. chen Vul.kanite aul Gran Canaria, Kanarische Inseln. (1965) Diopside-anorthite-waterat five and ten Habilitation Schrift, Ruprecht Karl Universitiit zu Heidel- kilobars and its bearing on explosivevolcanism. Carnegie berg, Heidelberg,Germany. Inst. WashingtonYear Book, 64,82-89. (1969) Calcalkalic andesites: Experimental data Sslw, H. R. (1963) Obsidian-H,O viscositiesat 1000 and on the origin of their assumedcharacteristics. In, 2000 bars in the temperature range 700-900"C. l. Geo- bearing Andesite Conference,Ed. A. R. Mc- phys. Res. 6E, 6337-6343. Proceedingsof the Birney, Oregon Dep. Geol. Mineral. Ind. Bull. 65,77-89. (1972) Viscositiesof magmatic silicate liquids: an (1971) The join diopside-pyrope-ItO at 10 kb: Its empirical method of prediction.Amer. I. Sci.272,810- bearing on the melting of peridotite, the ecr metamorphic 893. facies, and the gedrite-hornblende miscibility gap. 'Carne' SrernonN, R. R. (1962) after tridymite in an acid Quartz gie Inst. llashington Year Book, 69, 176-181. intrusionfrom Mu1l. Mineral. Mag.33, 1.38-144. AND G. A. CruNxen (1960) Grossularite-pyrope- J. D. S. MrcDoucen, exo P. J. N. LoNcuNo water system at 10,000 bars. Carnegie Inst' Washington (1959) The Tertiary igneousgeology of the Isle of Mull. YearBook,59,78-81. Geol. Ass. Guides, No. 20, Geologists'Association, Lon- ANDI. KusHIRo (1969) Melting of a hydrous phase: don, Benhamand Co., Ltd., 35 pp. phlogopite.Amer. l. Sci.,Schairer Vo|.,267-A,558-582. TnotrnrNssow, S. (1954) The tephra-fall from Hekla on D. B. Srewenr, ANDI. R. Slrnn (1957) Ternary March 29th 7947. In, The Eruption ol Hekla 1947-1948, feidspars.Carnegie Inst. WashingtonYear Book, 56,206- Vol. II, No. 3, Soc.Sci. Island, 68 pp. 214. Tsuve, H. (1955) Geological and petrological studies of ANDC. E. TtlLEv (1962) Origin of basaltmagmas: Volcano Fuji. V. On the 1707 eruption of Volcano Fuji. An experimentalstudy of natural and synthetic rock sys- Bull. Earthquake Res. Inst., Tokyo Uniu., 33, 341-384. tems."1. Petrology, 3, 342-532. Tur'rI-t, O. F., eNo N. L. BowrN (1958) Origin of granite Zrcs, E. G. (1960) Chemical analysesof two pantellerites. in the light of experimental studieson the system NaAl l. Petrology, 1, 304-308. Si.O"-KAISLO"-SiOz-H,O.GeoI. Soc. Amer. Mem. 74, (1962) A titaniferous basalt from the Island of 153pp. Pantelleria.I. Petrology,.1, 177-18O.