Canadian MineralogistVol. 16, pp. 581-589 (1978)

GRVSTAL.GROWTH TEXTURES IN MAGNETITE FLOWS AND FEEDER DYKES,EL LACO, CHILE

FERNANDO HENRIQUEZ'E ANP ROBERT F. MARTIN

Depaftmerlt of Geological Sciences, McGilI University, 3450 University Street, Mont6al,Qu|bec H3A 2A7

AsstRAcr

The unigue Quaternary iron oxide melts atEl Laco, Chile, crystallized in intrusive and ex-trusive environments. These differ in morphological,structural and textural details that reflect the dif-ferent environments of crystallization. Massive mag-netite predominates in the intrusive bodies, whereasspherulitic, dendritic and idiomorphic magnetite,now locally oxidized, characterizes the oxide flowsand feeder dykes. A transition from spherulite fibreto platy dendrite to euhedral crystal may occurwithin a few centimetres as an open space is ap-proached. The spectacular rapid-growth featuresare attributed to sudden supersaturation due to de-gassing of the oxide melts. Much slower growthof idiomorphic magnetite or hematite (primary)then occurred directly from the gas phase: whichof these oxides formed probably depended on theability of hydrogen to diffuse rapidly out of certainsas pockets.

SoNarvrerns

Nous d6crivons les manifestations intrusives etextrusives d'un magmatisme unique i oxydes defer d'ige quaternaire b El Laco (Chili). Ces deuxgroupes de roches se distinguent par leurs formes,leurs structures et leurs textures qui diffdrent sui-vant les conditions diff6rentes de cristallisation. Lamagn6tite est massive dans les roches intrusives,mais elle est sph€rolitique, dendritique ou idiomor-phe (et localement oxyd€e) dans les coul6es et lesdykes alimenteurs. On peut voir, sur quelques cen-timdtres, une fibre de sph6rolite passer d unedendrite en plaques, et celle-ci b un cristal idio-morphe i fapproche d'une cavit6. Nous attribuonsces facids spectaculaires i croissance rapide i unesursaturation soudaine due au d6gazage des bainsd'oxydes de fer. Uno croissance beaucoup plus lentede magn6tite idiomorphe ou d'h6matite primairesuivit, directement de la phase gazeuse; lequel desdeux oxydes a cristallis6 d6pend probablement dela quantit6 d'hydrogdns qui s'est 6chapp6e despoches de gaz par diffusion.

INTRoDUCTIoN

The crystal-growth textures to be described

*On leave from Departamento de Minas, Univer-sidad T6cnica del Estado, Casilla 10233, Santiago,Chile.

here occur in the unusual El Laco iron oredeposits, located in the second Chilean region,La( 23" 48'S, Long. 67" 30\1, in the AndeanCordillera. 450 km east of Antofagasta. Thedeposits occur as extrusive and intrusive bodies

Frc. 1. Geologrcal sketch map showing lithologicalunits (1 to 5) and different localities (A to H)mentioned in the text. 1. Altos de Pica FormatronCfertiary igoimbrites); 2, El Laco rhyodacitedomel 3. El Laco andesitic flows; 4' Plio-Pleisto-eene andesitic flows; 5' Quaternary morainic andalluvial deposits. Hydrothermal alteration zones,thermal spring deposits and contact aureoles areshown in a stippled pattern. In black are the ironoxide flows, dykes Oeft white)" domes and boulderfields. Localities: (A) Pico El Laco, (B) LacoSur, (C) Laco Norte, (D) San Vicente Alto' (E)San Vicente Bajo, (F) Laquito, (G) RodadosNesros and (tI) Cristales Grandes.

581

582

on the flanks of a Quaternary rhyodacite'-andesite cone over an area of 7 X 5 km, at analtitude ranging from 4300 to 5470 rn.

The emplacement of the El Laco complexof andesites, rhyodacites and intrusive and ex-trusive magnetite-rich bodies clearly postdatesthe regional outpouring of ash-flow tuffs thatconstitute the ignimbrites of the Tertiary Altosde Pica Formation. Its emplacement seemsmore or less contemporaneous with that ofPlio-Pleistocene andesites in the area. The com-plex defines a single volcanic edifice topped bya central crater, Pico El Laco" that is partly oc-cupied by a young rhyodacitic volcanic dome(Fig. 1). Iron ore deposits, evidently of mag-matic origin, occur in the periphery of PicoEl Laco (A, Fig. 1), where they are interlayeredwith andesitic and rhyodacitic flows. The entirecomplex shows signs of important and pervasivehydrothermal alteration: small-scale contactaureoles in andesitic wall rocks enclosing feed-ers to the iron-oxide flows show superimposedbroader-scale intense hydrothermal alterationthat is still going on. Fumarolic activity is nowconcentrated at half a dozen centres on El Laco.

Four medium- to large-sized high-grade oredeposits have been mapped: Laco Sur, LacoNorte, San Vicente Alto and San Vicente Bajo;there are also three smaller deposits: Laquito,Rodados Negros and Cristales Grandes. Theseare shown as localities B to H, Figure 1. Atleast one additional deposit, inferred by mag-

THE CANADIAN IIID{ERALOGIST

netic survey and sought by drilling, remains tobe discovered.

Park (1961) emphasized the unusual natureof these iron ore deposits when he called them"magnetite flows". Sanchez (in Ruiz et aI. L965)Eaye a short description of El Laco and pro-duced the first geological map of the complex.Rogers (1969) briefly described the differenttypes of ores, and Haggerty (1970a) providedthe first careful description of their mineralogy.Magnetite is certainly the more importantprimary iron oxide; primary hematite is scarce.Secondary hematite and maghemite develop ex-tensively as oxidation products of primary mag-netite. Unusual Ca-Fe and Fe phosphates(Hasgertv 1970b) have been identified.

Exrrusrve AND INTRUSTVE Dsposrrs

A two-fold classification of the deposits intoextrusive and intrusive types was first proposedby Henriquez & MacLean (1975). They con-sidered Laco Sur, Laco Norte, San Vicente Altoand part of Rodados Negros (Fig. 1) to beextrusive. Evidence for an effusive origin ismorphological, structural and textural, and in-cludes: development of blocky flows (Fig. 2),locally highly vesicular (Fig. 3); iron oxide-richpvroclastic material, largely lapilli and ash;slae-like material in flows characterized locallyby contorted flow layering, ropy upper surfacesand gas escape tubes (Figs. 4, 5). Many of thefield characteristics are those expceted of a low-

Frc. 2. Blocky flow of massive magnetite at I-aco Norte (C, Fig. 1). A part of the arcuate feeder dykecan be seen in the background.

Frc. 3. Top part of a feeder dyke at Laco Norte. Note the highly vesicular, bulbous aspect of theexposure. Typically, an individual bulboui mass has an interior open space lined with idiomorphicmagnetite. The white patches correspond to hydrothermal minerals that coal the walls of the largevesicles.

Frc. 4. Inclined gas escap€ tubes in one of the feeders at Laco Norte.

EL LACO MAGNETITE FLOWS 583

viscositv basaltic flow (e.g., Macdonald 196?.The effusive iron-oxide liquids have reached

the surface vla fissures, but these are not uni-form in their orientation. At Laco Sur" thefissures belong to sub-parallel swarms, but atLaco Norte and San Vicente Alto. the flowsoccupy arcuate fracfures that may reflect localcollapse owing to evacuation of magma reser-voirs. A drilling program in the central blockbetween the arcuate fissures at Laco Nortefailed to locate a central feeder dyke or anysign of an inwardlv dipping cone sheet. Whetherthe fissures are straight or arcuate naturallyleads to differences in outcrop patterns of thedeposits and also in the disposition of gasescape tubes: parallel and vertical in the first,inclined and semi-radiating in the second (Fig.4).

Evidence from surface exposures and fromdrill cores document the following typical strati-graphic succession: the first product to be ex-pelled from the fissure rflas a magnetite-richpyroclastic material deposited on El Laco ande-sitic lavas. In one hole. this unit is 30 m thick.Magnetite in this unit has been thoroughlyoxidized to hematite. On this apron of pyro-clastic material flowed a lava that crystallizedlargely to magnetite. The flows total 50 m inthe same drill hole. Apatite is an accessoryphase disseminated as fine grains. Where incontact with andesitic lavas, a narrow (1 m)contact aureole of actinolite * scapolite *quartz is developed. The outpouring of "oxide"Iava was followed by another pyroclastic unitthat contains fragments of massive ore, pre-sumably detached from the feeders during afinal , explosive event. It measures 20 m in thesection studied.

Shallow intrusive bodies have also beenrecognized: Laquito, Rodados Negros, CristalesGrandes and San Vicente Bajo (Fig. 1). Theycontrast with the extrusive type by their form,total absence of any signs of effusive activity,abundance of large, well-developed apatite crys-ta,ls in amygdules and predominance of massive-textured magnetite. In all these deposits. thesize and relative abundance of apatite increasetowards the top.

The San Vicente Bajo intrusive bodv is dome-like, having the form of a bowler hat, 500X300m. The other intrusive bodies are dyke-like,and attain 3 to 15 m thickness. The emplace-ment of these has led to a contact aureole ofaptinolite * scapolite f quartz in the hostrbcks within 2 m of the contact. Amygdules upto 15 cm across in massive magnetite near thesecontacts are commonly filled with actinolite *

quartz * apatite; these must represent gas-richbubbles such as observed in the extrusive de-posits, but here the gas phase has evidently beenunable to escape.

Cnvsrer-Gnownr Tnxtunss

From the previous discussion, we infer thatat El Laco, gas-charged iron-oxide-rich liquidsreached near-surface environments to form in-trusive domes and dyke systems; locally, somedykes also served as feeders for thin flows. Wenow turn to the signs of such an origin in thehabits of crystals that grew from these unusualmelts or from associated gases.

Crystal habits in intusive bod"ies

The intrusive bodies consist essentially ofmassive magnetite, locally oxidized incipientlyto hematite. In textureo the magnetite resemblesthe occurrences of massive magnetite alreadydescribed by Park (1972) and Bookstrom (1977)from a number of iron ore deposits along theChilean Coast Range. These deposits have beenvariously interpreted to be hydrothermal, con-tact-metasomatic or magmatic (Geijer 1931,Ruiz el al. 1965, Park 1972, Bookstrom 1977).An important difference with these, however, isthe occurrence of spectacular apatite prismsthat attain 7 cm in length in amygdules in

Flc. 5. Close-up view of one of the feeders at LacoSur (B, Fig. 1). To the left, more or lgss parallelto the hammer, a vertical gas-escape tube can beseen with euhedral magnetite developed alongits walls. At the centre a contorted magnetitelayer shows the upward direction of movement.

584 TIIE CANADIAN MINERAI'GIST

massive magnetite, typically near the roof zonesof intrusive bodies.

Even though these deposits undoubtedlyformed from melts, we believe tlat crystalliea-tion was too slow to give rise to distinctivecrystal habits.

Crystal habits in extrusive bodies

Our aim here is to document the spectacularand unique growth forms developed in the mag-netite flows and feeder dykes of the Laco Sur,Laco Norte and San Vicente Alto occurrences.Dendritic magnetite is the most common ex-pression of rapid growth. The dendrites formin situ at high levels in the feeders, particularlyin walls adjacent to gas escape tubes andvesicles, and near the uppermost parts of slabsof magnetite flows (Fig. 6). These dendritesappear in uniform arrays of columnar oopris-

matic" magnetite; each prism typically measures2 cm across by 1O cm in length, though minia-ture columnar individuals L x 4 mm also arefound. The columns are generally distributedapproximately perpendicular to a nearby openspace (e..9., vesicle, gas escape tube, top offlow) and attain a strikingly parallel disposition(Figs. 6, 7). The branching forms in thedendrites may look like fibres in two dimen-sions, but are really thin plates in parallelarrays, consistently 0.01 to 0.03 mm apart.The angle between plate and dendrite axis is 45 "(Fig. 8), suggesting that the plates define the

FIG. 6. Part of single maenetite flow now highlyoxidized to hematiten showing tle general dis-position of "prismatid' dendrites, Laco Norte(C, Fig. 1).

( 111) growth face of magnetite; ttre com-plementary set of plates gives the appearanceof an arrow pointing towards the open space(Fig. 7). No evidence of twinning has beenfound.

Frc. 7. Detail of an array of parallel platy dendritespointine upward as arrows towards an openspace. The dendrite to the left of centre showsa transition from abundant to few branchingplates, then to a euhedral crystal at the top' LacoSur (B, Fie. l). Bar : 1 cm.

The number of plates decreases and theirindividual thiokness increases as the crystalapproaches the open space. This decrease isgradual, occurring over 2 to 3 cm in some ofthe larger dendritic individuals. The columnardendrites are typically terminated by well-devel-oped octahedral faces (111) , (111) , and (111) .

Fig. 7). Depending on the proximity ofthese clusters of columns to circulating post-magmatic fluids, the dendritic magnetite mayor' may not show appreciable oxidation tohematite.

Spherulitic textures oe*ut in ritl at evenhigher levels in the feeders than dendrites.They are generally found near large vesicles,and'smaller open spaces occur between indi-vidual spherulites (Fig. 9). In outcrop, theresulting rock looks knobbly and pitted (Fig. 4).The spherulites vary from truly spherical arraysof fi6ros to fan spherulites, in the usage ofLofgren (1974). They range from 1.8 to 4.5 cmacross, and thus could also be called macrodphe-rulites owing to their size (Donaldson et al.1973). The individual fibre varies from 0.05 to0.21 cm across; approximately 50 such fibres

EL LACO MAGNETTTE FLOWS

Frc. 8. Platy dendrite showing a 45' angle between plates and dendriteaxis. Section from tle same hand specimen as in Figure 7. Bar = I cm.

Frc. 9. Fan spberulite with small vesicle between individual spherulites.Euhedral hydrotlermal minerals can be seeo among the clusters ofeuhedral magnetite crystals in the vesicle. Laco Sur (8, Fig. 1). Bar =I cm.

Frc. 10. Fan and- almost spherical spherulites. Upper surface correspondsto an op€n space. Dendrites develop from spherulite fibres in thisspecimen. The pits here represent gas cavities and axial holes in fibres.Iaco Sur (8, Fig. 1). Bar - 1 cm.

Ftc. 11. Octahedral magnetite developed along wall of a gas-escape tube.Laco Norte (C, Fig. 1). Bar - I cm.

FIc. 12. Primary hematite crystals ("nuts") with promrnent {0006} andt1120) fases and less commor and smaller {10T0} faces. The crystalsinterpeletrate but are not twinned. Laco Sur (B, Fig. 1). Bar - 1 cm.

Ftc. 13. Primary hematite plates grouped into a rosette pattern. Note thecrystal of alunite (right of centre) deposited from late hydrothermalfluids. Laco Norte (C, Fig. 1). Bar = I cm.

585

586 TIIE CANADIAN MINERALOGIST

can be counted in a typical array in two dimen-sions @igs. 9, 10). Fibres of adjacent fanspherulites come in contact at angles attaining100" (Fig. 10). Individual fibres may be partlyhollow or filled with accessory co-precipitatedphases, thus illustrating the intrafasciculatetexture of Drever et al. (L972).

The larger spherulite fibres commonly showa transition over a short distance to a typicaldendritic growth habit wherever the fibres areoriented towards an open space. There, theyare invariably terminated by octahedral faces.Within centimetres, therefore, a fibre in aspherulitic array thickens and widens into aplaty dendrite, then this growth form gives wayto well-formed, idiomorphic magnetite crystals.Later transformation to hematite is sporadic,increasing near the open space.

The idiomorphic magnetite crystals occuralong walls of gas-escape tubeso open spacesformed by the contortions of flow layering,and vesicles up to 7O x 20 x 15 cm (Figs. 3,5, 7, 11). The crystals have an octahedralhabit, with faces that are pitted and not trulyplanar; a single crystal may reach 6 cm across,but most are much smallero lining the wallslike quartz in a geode. Set among these clustersare occasional euhedral single crystals of min-erals deposited from late circulating aqueousfluids (Fig. 9). Detailed X-ray diffraction studies(to be described separately) have led to theidentification of intergrown iron-bearing sanidineand plagioclase. Rutile and quartz have alsobeen identified by microprobe analyses. Thefluids from whish these feldspar crystals precipi-tated may have been responsible for the mildetching and oxidation of the magnetites.

Primary hematite crystals are very scarce,and have not been found lre slfa. These "nuto'-looking tabular crystals attain 3.2 x 0.7 cm(Fig. L2); they have prominent {0O06} and{1120} faces, and less common and muchsmaller t1011) faces. Interpenetrating crystalsgive a first impression of being twinned, withtheir {0006} faces roughly perpendicular (Fig.l2).ln fact, the crystals are not related by a twinlaw. Also found are thinner plates of hematitegrouped into a rosette pattern (Fig. 13). Low-temperature quartz, rutile and alunite have beenidentified as accessories deposited hydrother-mally in op€n spaces between plates.

DtscusstoN

So far as is knowno the occurrence of den-dritic and spherulitic magnetite in quenchediron-oxide liquids at El Laco is unique. A care-ful search of the geological and metallurgical

literature on rapid-crystal-growth phenomenahas provided only one pertinent reference(Hirano & Somiya L976), a description of den-dritic magnetite though from a hydrothermaln nota magmatic environment. Nor has any mentionbeen made of the transition from spherulitefibre to dendrite to euhedral crystal in one singlecrystal. Comments that follow on interpretationof the textural developments are based onanalogous habit modifications in different min-eral groups and in metallurgical systems.

Tiller (1964, L977) and Jackson et al. (1967)have described hopper dendrites of bismuth thatresemble the El Laco magnetite dendrites; Billig(1955) has studied lamellar dendrites in Gecrystals, and Donaldson (1974, 1976) discussedplate dendrites in spinifex olivine crystals. Theplate dendrites of El I-aco are stacked like adeck of cards as are olivine plates in spinifextextures, but differ in the absence of otherplates that cut across the main set of plates.This may be due in part to be high symmetryof magnetite compared to olivine. Furthermore,the interplate spaces at El Laco are empty.

The habit of a crystal is strongly dependenton such interrelated variables as diffusion rate(D), growth ftte (G) and temperature. Thetransition in crystal habits from spherulitic todendritic to hollow or skeletal to well-formed isrelated to variation in D/G ratio (Jackson1958; Lofgren 1971, 19'74; Donaldson 1974,Kirkpatrick 1975). Unfortunately, experimen-tal studies are lacking and thermodynamic dataof relevance are rare, such that semi-quantitativeinferences of diffusion and growth rates herewould be hazardous.

We do know the melting point of FeaOa at1 atm, 1870 t 2 K, and its enthalpy of f'usionAI/o-, 33.0 -F 2.0 kcal/mole (JANAF tables1971). The entropy of fusion of FegOa we getfrom these data, A,S". - 17.3 e.u., is relativelyhigh (Uhl,mann t972) and suggests that thestructural reorganization in crystallizing a liquidof composition FesOe is rather important. Inturn, this implies that nucleation will requirehigh supersaturations and that growth will beanisotropic (Jackson et al. 1967, Uhlmann 1972)favoring the development of faceted dendrites.This presumably accounts for the rigorouslyparallel platelet$ in Figure 6.

But what is the effect of adding volatiles tothe system? Gases (e.g., HrO, HCI) certainlyplayed a major role in lowering the temperatureof the liquidus at El Laco, as no evidence ofunusually high temperatures can be found inthe contact aureoles. The fluxing effect ofgaseous species has been documented already:in the system Fe-C-O, Weidner (1968) found

EL LACO MAGNETITE FLOWS 587

that liquids rich in FegOe could be expected attemperatures as low as 815'C at 35O bars;Gibbon & Tuttle (1967) have also described anFeaOa-rich liquid at 1060"C in the systemFeG-FegOs-SiOa-Hz0; Philpotts (1967) foundan iron oxide liquid at l42O"C in the "system"apatitdiorite-mapetite at low pressures.

Diffusion rates depend rather importantly onviscosity. No data are available for liquids evenclose to FeaOa composition, nor for liquids con-taining appreciable FezOs. The closest approachto an answer comes from rare investigations ofslag systems, in which the component FeO is byfar more important than Feroa. Extrapolation ofviscosity determinations in the system FeG.SiOzbetween 1100" and 1450"C (Jrbain 1951) andin the system CaG-FeO-SiOe between 120O'and 1350'C (Riintgen et al. t960) to pure FeOgives a value of the order of 0.2 poises. It couldbe argued that Fes0g does not affect viscositydifferently than an equivalent amount of FeO(Bottinga & Weill L9'72), but the bulk composi-tion effect here would weaken this inferencebased on silicate systems. The El Laco liquidswere undoubtedly more viscous than 0.2 poises,but by how much? Lower temperatures innature would lead to an increase in viscosity,but complicating the system by adding HzO, HCIand P would likely induce a compensating effect;Ca and Si impurities will raise viscosity atconstant T @iintgen et al, 1960). We tenta-tively envision fluidities comparable to thoseof ultrabasic liquids that led to rapid-growthhabits of olivine (e.9., Arndt et al. 1977).

Degassing of the El Laco FegOa-rich liquidsprobably played a crucial role in determiningrates of crystal growth. Jackson et al. (1967)have shown experimentally that degassing canincrease growth rate in certain organic systems;Donaldson (1974) has postulated that spinifexolivine clearly forms in response to sudden su-persaturation. Lofgren (1974) and Donaldson(1976) have catalogued the growth habits ofplagioclase, pyroxene and olivine as a functionof degree of supercooling in isobaric experi-ments. We contend that the gas escape tubes,the vesicles and the pyroclastic unit confirm therole of degassing, locally violent, in the caseof the extrusive flows and related feeder dykes.This is the best mechanism to induce suddensupersaturation. In contrast, the intrusive bodiesat El I-aco, though very shallow, failed todegas completely or as rapidly. The presenceof amygdules with large apatite crystals suggeststhat volatiles were more efficiently trappedwhere tley did separate; degassing may alsohave been more gtadual.

At Bl Laco, spherulitic magnetite grew at the

top of the feeder dykes, where supersaturationdue to degassing or to thermal quenching (orboth) might be most extreme. Spherulites de-velop wherever G >> D (Lofgren 1974). Plalydendrites grew lower down in the feeder dykeswhere cooling rate might have been sloweror where degree of supersaturation due to lesscomplete or rapid gas loss might have beenlower. The change in morphology towards aplaty dendrite involves a decrease in G/D (Lof'gren L974, Kirkpatrick 1975, Donaldson 1976).With decreasing degree of supercooling, [T, inisobaric experiments, Lofgren (1974) found thediameter of spherulite fibres to increase. At ElLaco the diameter of these fibres is relativelyIarge compared to those in other mineralogicalsystems, such that they may have formed at adegree of supersaturation already close to thetransition to dendritic growth forms. The de-crease in rate of growth with time may resultfrom local buildup of dissolved gaseous speciesowing to initially very rapid growth of an anhy-drous mineral from a liquid that was nottotally degassed.

One may argue qualitatively that spheruliticmagnetite would not have been possible at ElLaco if the melt viscosities were as low asthoso obtained in the svstems FeO-SiOg andCaG-FeO-SiOr. Higher viscosities would be re-quired to ensure that G > > D. From an ana-lysis of the factors that affect the transitionspherulite-dendrite, Tiller (1977) proposes thatviscosities exceeding 1 poise are necessary toobtain either habit.

Whereas massiye, spherulitic and dendriticmagnetite undoubtedly grew from melts ofvariable degrees of supersaturation, we visualizea different mode of formation for the idiomor-phic crystals. These grew frorn a gas phase; theentropy change during this process of crystalliza-tion is considerably larger than when FesOacrystals grow from oxide melts. Consequently,using the reasoning of Jackson (1958) andUhlmann (1972), we expect fewer but well-formed and euhedral crystals. A question arisesconcerning the possible dissociation of water inthese high-temperature gas pockets, and subse-quent loss of hydrogen. This phenomenon mayhave caused localized unusually oxidizing con-ditions in some pockets; this would explain theformation of the "nut"-shaped hematite crystals.Kolb el al. (1973) have shown that he,matitemay form in this way experimentally, confirm-ing Eugster's (1959) prediction. The persistentnucleation and growth of hematite from anaqueous fluid in the experiments of Martin &Piwinskii (1969) with granitic rocks may alsofind an explanation in the same phenomenon.

588 THE CANADIAN MINERAI,OGIST

Also, Hirano & Somiya (1976) found that hydro-gen plays a very important role in producingeuhedral magnetite crystals grown from anaqueous fluid, whereas Tiller (1977) emphasizesthe importance of hydrogen content of the gasphase in producing euhedral crystals in general.

Another important variable in the gfowthof these idiomorphic crystals slight be thepartial pressure of Fe at high subsolidus tem-peratures. These partial pressures have beenshown to be appreciable above 900"C byDarken & Gurry (1946) and Eugster (1959).Charasterizdtion of the post-magnetite and-hematite minerals deposited from the gas phasewill allow a more complete reconstruction oftle evolution of the post-magmatic hydrother-mal fluids at El Laco.

We have presented here the striking evidencefor grofih of magnetite from iron-oxide-richliquids, and of magnetite, hematite and otherminerals from related gas-phases. Petrological,mineralogical and geochemical studies, now inprogress, will hopefully resolve the meaty issueof the ultimate derivation of these unusual, gas-charged liquids.

AcrNowr,nocsMENTs

Fernando Henriquez thanks the UniversidadT6nica del Estado (Chile) for continuing supporrand the National Research Council of Canadafor a CIDA/NRC grant which made this studypossible. Robert F. Martin also thanks N.R.C.for support through its grant N72I. We ac-knowledge helpful discussions with Drs. L.S.Darken, J.D.H. Donnay, H.,P. Eugster, W.H.Macl-ean, G.C. Ulmer and J.R. Weidner andthe help of Prof. D. Francis on the electronmicroprobe:

RgrrRBNcps

ARNDT, N.T., Nu,pnprr, A.J. & Pvrr, D.R. (1977):Komatiitic and iron-rich tholeiitic lavas of Munrotownship, northeast Ontario. I. Petology 18, 319-369.

Bnr,rc, E. (1955): Growth of monocrystals ofgermanium from an undercooled melt. Roy. ,Soc.London Proc. 4229, 346-363.

Boorsmort, A,A. (1977): The magnetite depositsof El Romeral, Chile. Econ. Geol. 72, lL0L -I 130.

Borrncr, Y. & Wsrr.r, D.F. (1972): The viscosityof magmatic silicate liquids: a model for cal-culation. Amer. J. Sci. 272, 438-475.

DARKEN, L.S. & Gunnv, R.W. (1946): The systemiron--oxygen. II. Equilibrium and thermodyna-

mics of liquid oxide and other phases. l. Anter.Chem, Soc.68, 798-816.

DoNALDSoN, C.H. (1974): Olivine crystal typesin harrisitic rocks of the Rhum pluton and inArchean spinifex rocl*g. Geol. Soc. Amer. Bull.85, t72t-t7?5.

-(1976): An experimental investigation ofolivine morphology. Conn. Mineral. Petrology57, t87-213.

-, Drevrn, H.t. & JoENsroN, R. (1923):Crystallization of poikilo-macro-spherulitic feld-spar in a Rhum peridotite. Nature 243, 69-70.

DREVER, H.I., JonNsroN, R., BurLER, P. & Gtns,F.G.F. (L972) t Some textures in Apollo 12lunar igneous rocks and terrestrial analogs. Proc.9rd Lunar Sci. Conf. t, l7l-t84.

EucsrER, H.P. (1959): Reduction and oxidation inmetamorphism. fz Researches in Geochemistry(P.H. Abelson, ed.), Iohn Wiley & Sons, NewYork.

GEUER, P. (1931): The iron ores of the Kirunatygn. Sver. Geol. Unders. (ser. C.), Avhandlingeroch uppsater, No. 367, Arsbok 24(4).

Gnror*, D.L. & Trrrnn, O.F. (1967): A note onthe system FeO-FerOr-9iO2-H2O. Amer. Min-eral, 52,886-889.

HAccERTy, S.E. (1970a): The Laco magnetite lavaflow, Chile. Carnegie lwt. Wash. Year Book 68,329-330.

-(1970b): A new iron-phosphate mineral.Carnegie Inst. Wash. Year Book 68, 330-332.

HENRreuEz, F. & MIcLEAN, W.H. (1975): Ante-cedentes preliminares sobre los yacimientos vol-canicos de fierro El Laco, Antofagasta, Chile.Resum. II Congreso lbero-Americano Geol. Econ,,52, (Abstr.).

HneNo, S. & Somva, S. (1976): Hydrothermalcrystal growth of magnetite in the presence ofhydrogen. J. Crystal Growth 85, 273-278.

JAcKso\ K.A. (1958): Mechanism of growth. InLiquid Metals and Solidification. Amer. Soc.Metals, Cleveland, Ohio.

Unr,ueNr, D.R. & HuNt, I.D. (1967): Onthe nature of crystal growth from the melt. J.Crystal Growth 1, 1-36.

Jeren (1971): Thermochemical tables, 2nd ed.(D.R. Stull & H. Prophet, project directors).Nat. Bur. Standards NSRDS - N8537.

KrRKpATRrcK, R.J. (1975): Crystal growth from themelt: a review. Amer. Mineral.60,798-814.

Kor.n, E.D., CeroReso, A.J. & Leuorsn, R.A.(1973): Hydrothermal growth of hematite andmagnetite. J. crystal Growth L9,242-246,

LoncnrN, G. (1971): Spherulitic textures in glassyand crystalline rocks. J. Geophys. Res. 76, 5635-5648.

-(1974): An experimental study of plagio-clase crystal morphologyl isothermal crystalliza'tion. Amer. I. Sci. 274, U3-273,

EL LACO MAGNETITE FLOWS 589

VIAcDoNALD, G.A. (1967): Forms and structures ofoxtrusive basaltic rocks. /n Basalts, the Polder"vaart Treatise on Rocks of Balsaltic Composition1. Interscience, New York.

Mlnrr.r, R. F. & PrwtNsK[, A. J. (1969): Experi-mental data bearing on the movement of ironin an aqueous vapr. Econ. Geol. 64,798-803.

Pem, C. F. (1961): A magnetite 'oflow" in northernChile. Bcon. Geol. 66, 431-441.

(1972): The iron ore deposits of the PacificBasin. Ecoz. Geol, 67, 339-349:

Pnnrorrs, A. R. (1967): Origin of certain iron-titanium oxide and apatite rocks. Econ. Geol,62,303:315.

RocERs, D. P. (1969): The extrusive iron oxidedeposits, "El Laco," Chile. Geol. Soc, Amer.Spec, Pap. I.zL, 252-253.

RdNrcBN, P., Wvrrr,secnn, H. & KAMMEL, R.(1960): Struktur und Eigenschaften von Schlackender Metallhiittenprozesse. II. Viskositiitsmessun-gen an Schmelzen der Systeme Eisenoxydul-Kalk-Kieselsiiure und Eisenoxydul-Kalk-Koner-de-Kieselsiiure, Z. Erzberbau Metallhiittenwesen(Erzmetall) Lg, 363-414,

Ruz. C.. Actnne, L,, Conver,en, J., Kr-ottN, C.,KionN, E. & Livl,'B. (1965): Geologia v Yaci-rnientos Metalileros de Chile. Inst. Invest. Geol.,Santiago, Chile.

Tn-r.rn, W. A. (1964): Dendrites. Science t46' 871'879.

(1977): On the cross-pollenation of crys-tallization ideas between metallurgy and geology.Phys. Chem. Minerals 2, 125-15t.

Unr.rvrANN, D. R. (1972): Crystal erowth in glass-forming systems - a review. la Advances inNucleation and Crystallization in Glasses (L. L.Hench & S. W. Freiman, eds.), Amer. Ceram.Soc. Spec. Publ. 5, 91-115.

UnnNN, G. (1951): Mesures des viscositds des lai-tiers synth6tiques du systbme FeO-SiOz-TiO:.CJl. Acad. Sci. Paris 202,330'332,

WeroNrn, J. R. (1968): Phase Equilibria in a Por'tion of the System Fe-C-O trom 250 to 10,000bars and 400"C to 1200'C and its PetrologicSignificance, Ph.D. thesis, Pennsylvania StateUniv., University, Parkn Pa.

Manuscript received UaV (970.