Titanian Chondrodite- and Titanian Clinohumite-Bearing Metadunitefrom the 3800 Ma Isua Supracrustalbelt West Greenland: Chemistry, Petrology, and Origin

Ameican Mineralogist, Volume 73, pages 547-558, 1988

Titanian chondrodite- and titanian clinohumite-bearing metadunitefrom the 3800 Ma Isua supracrustalbelt West Greenland: Chemistry, petrology, and origin

Ronnnr F. Dvnrnr Department of Earth and Planetary Sciencesand McDonnell Center for the SpaceSciences, Washington University, St. Louis, Missouri 63130, U.S.A.

ARcooil andGas.",,o",,jr1llil: -t; ?i*erd, carirorniae3302, u.s.A. Slu C. Bnornnns Department of Geology, University of New Mexico, Albuquerque, New Mexico 87131, U.S.A.

Ansrnlcr Titanian chondrodite (Ti-Ch) and clinohumite (Ti-CD have been discovered in a serpentinite from the 3.8 Ga Isua supracrustalbelt, West Greenland, where they occur with olivine (= Fonr),magnesite (X*" = 0.99), magnetite,and Ni-Co-Pb sulfidesin a matrix of antigorite. This serpentinite has the bulk composition of a dunite (Niggli value mg x 0.89), with high Ni (-2300 ppm) and Cr (-950 ppm), and its REE pattern is broadly similar to those reported for cumulate dunites of igneous origin. The highly magnesian phase compositions indicate that the metadunite evolved from a serpentinite precursor, which experiencedadditional late serpentinization. Two different generationsof olivine and antrgorite, both with slightly different compositions, indicate an extremely complex history for the sample.However, the titanian humites locally occru as inclusions in olivine, indicating that they formed relatively early. Ti-Ch (-9 wto/oTiOr) occrus as separategrains and also forms lamellar intergrowths with Ti-Cl (-6 wto/oTiOr). These phasesdiffer significantly only in terms of TilSi, as X', (-0.98), X*u X-o, Xrt (-0.45), and X, (-0.05) are similar. The high content of Ti and low content of F indicate nearly complete TiOrMg ,F-, exchange,whereas the high X'n values make thesethe most magnesianTi-rich humites yet reported. The two-phaseintergrowths may representthe reaction Ti-Ch + forsterite : Ti-Cl. The sample lacks geikielite or any magnesian ilmenite, indicating that the terminal breakdown of titanian humites by the reaction titanian humites : forsterite + geikielite * HrO was not approached.Chemographic relations for the systemMgTiO,-MgrSiO4-HrO reveal that Ti-Ch should be the high-pressurehumite phase,on account of the experimentally determined location of the Ti-Cl breakdown reaction (Engi and Lindsley, 1980). The only other known occurrence of a Ti-Ch and Ti-Cl intergrowth is in kimberlite (Aoki et al., 1976),and the possibility that Ti-Ch is a mantle phaseshould be reconsidered. However, we do not regard the Isua sample to be of mantle origin, although the intergrowths may record decompressionduring the polymetamorphic history of the region.

INrnolucrroN the Ruby Range, Montana (e.g., cimmino et al., 1979; Natural occlurencesof titanian clinohumite (Ti-Cl) are Trommsdorffand Evans, 1980; Desmarais, 1981; Evans rare but have attracted considerableattention in recent and Trommsdortr, 1983).The work of Trommsdorffand years becausethey are found principally in ultramafic Evans (1980) is particularly important since it seemsto rocks, some of which are demonstrably of mantle origin. establish on petrographic grounds that the stability field McGetchin et al. (1970) noted the presenceof TiCl in for Ti-Cl is contained fully within that of antigorite. Ex- ultramafic nodules found in kimberlite from the Colo- perimental work (Engi and Lindsley, 1980) has shown rado Plateau and speculatedthat this phasecould repre- that Ti-Cl is unstable above 475 "C at 3.5 kbar and 675 sent an important upper-mantle reservoir for volatiles, "C at 2l kbar, which placessevere limitations on the im- especiallywater. A number of reports have also described portance of this mineral to mantle petrology, while at the occurrencesof Ti-Cl in peridotites and serpentinitesfrom same time confirming its stability within the P- Z regime orogenic belts that include the Alps, the Appenines, and relevant to "ordinary" crustal metamorphism. 0003-o04x/88/0506-0547$02.00 547 548 DYMEK ET AL.: CHONDRODITE AND CLINOHUMITE IN METADUNITE

TABLE1. Chemicalcomposition of lsua metadunite1W809-4 tiquity of the Isua belt and establishit as the oldest dated group ofrocks on Eanh. Cation % A variety of ultramafic rocks are found at Isua, includ- sio, 41.30 Si 34.0 ing (l) massive metaperidotite-metadunite (olivine + Tio, 0.09 Ti 0.1 + Al203 0.82 AI 0.8 chlorite + Cr-rich magnetite t tremolite cumming- FeO 9.92 Fe 6.8 tonite or anthophyllite + dolomite + magnesite, with MnO 0.16 Mn 0.1 relict greenspinel + orthopyroxene found at one locality; Mgo 47.07 Mg 57.8 CaO o.25 Ca 0.2 Brothers and Dymek, 1983);(2) ultramafic schist derived Naro 0.06 Na 0.1 from metaperidotite (talc + magnesite+ anthophyllite + K"O 0.07 K <0.1 P.o. <0.03 P <0.1 antigorite, with rare multiple chain silicates;O'Leary and Dymek, 1987); and, (3) antigorite serpentinite. Late li- LOI 10.77 zardite and/or chrysotile, either as pseudomorphs after v14 Rb <1 Ce 0.925 olivine or in small veins, occur in many samples. Cr 958 Sr <2 Nd 0.615 The sampleof interest here (IW809-4) is a serpentinite, Ni 2357 Y<1 Sm 0.143 having the bulk composition of a dunite (seebelow). It Zn 59 Zr <2 Eu 0.019 Ga <1 Nb <1 Gd 0.171 is fine-grained, dense, and very fresh, with the typical Ba <5 Dy 0.181 blue-graycolor of the other Isua ultramafic rocks.IW809-4 Pb 13 Er 0.145 red-weathering Th <1 Yb 0.196 was collected from one of a seriesof outcrops that locally contain bladed (metamorphic) olivine Niggfivafue mg: O.894 (Mg + Fe)/Si:1.903 crystals up to several centimeters long. These ultramafic outcrops form pods (on the order of l0 x 50 m) within (1984) Titanian chondrodite (Ti-Ch) is even more rare. a mafic schist that Nutman et al. termed the se- Trommsdorffand Evans(1980) remarked upon its pres- quence A-lower amphibolite formation. Out of about ence in their Alpine samples as a possible breakdown 50 samplesstudied from severallocalities throughout the product of Ti-Cl. The only published analysesthat we are Isua belt, Iw809-4 is the only one in which humite min- aware of are those from the Buell Park, Arizona, kim- erals were found, attesting to the relative rarity of these phases. berlite (Aoki et al., 1976;,cf. Smith, 1977; Fujino and Tak6uchi, 1978),where the Ti-Ch forms intergrowthswith Crmurcar, coMPosrrloN Ti-Cl. The exactparagenesis ofthis occurrenceis unclear, as the humites were not recognized "in situ" but were Major elements recovered through mineral separation on crushed rock The major-element composition of sample IW809-4 fragments. was determined by microprobe analysisof glassprepared In the course of investigating ultramafic metamorphic by fusion of 1.0 g each of sample powder and LirBoO' rocks from the 3.8 Ga Isua supracrustalbelt, West Green- flux using the same proceduresapplied to the analysis of land, we have discoveredTi-Ch in a serpentinizedmeta- the North American Shale Composite (NASC) by Gro- dunite, where it occurs as independent grains and also met et al. (1984).Results are listed in Table I on a vol- forms intergrowths with Ti-Cl. The purpose of this paper atile-free basis and representthe weighted averageof 15 is to document the chemistry of these humites and to replicate determinations. evaluate their petrogenesisin light of available data on The analysisshows the sample to be highly magnesian other occurrences.The present paper follows upon pre- (Niegli value mg = 0.9) with MgO, FeO, and SiO, being liminary resultsreported earlier (Dymek et al., 1983) and the only oxides present in amounts exceeding I wt0/0. is expanded here to include a thorough account on the [(Mg + Fe)/Si]","-,.is close to 2, and a cation-norm cal- phasepetrology of the host metadunite, togetherwith ad- culation yields an olivine content of -95o/o. Thus, the ditional information on its major- and trace-element bulk composition corresponds closely to a dunite. The composition. small quantities of NarO, CaO, KrO, and AlrO, perhaps representvestiges ofaccessory pyroxene, spinel, or mica Gnolocrc SETTTNGAND sAMpLE DESCRrprroN originally present in the protolith, or they were added The Isua supracrustalbelt (lat 65"15'N; long 50'00'W) during metamorphism or alteration. is located in the central part of the West Greenland Ar- Loss-on-ignition (LO! representsweight loss upon fu- chean gneisscomplex. The geology of the Isua area has sion for 15 min at 1000"C on 1.0-gsample powder pre- been studied by, among others, Bridgwater and Mc- viously dried overnight at I l0 'C. The value thus deter- Gregor(1974), Allaart (1976),and Nutman et al. (1984), mined (- I I wto/o)is quite high and indicative of the large who recognized a diverse suite of clastic and chemical amounts of serpentineand magnesitepresent in the sam- metasedimentary rocks, mafic metavolcanic rocks, and ple (seebelow). stratiform ultramafic rocks. Various radiometric dating techniqueshave yielded agesfor the supracrustalrocks in Trace elernents the 3700-3800Ma range[see Baadsgaard et al. (1984)for Abundancesof V, Cr, Ni, Zn, Ga, Rb, Sr, Y, Zr, Nb, recent summaryl, which leave little doubt as to the an- Ba, and Pb were determined by X-ray fluorescenceanal- DYMEK ET AL.: CHONDRODITE AND CLINOHUMITE IN METADUNITE 549 ysis of pressedpowder pellets at the University of Mas- sachusetts,Amherst, using methods describedin Rhodes Soopslore et al. (1978). Tabulated results represent the averageof (Finlond) analyseson two separatealiquants of the sample. Values for most elementsare very low, in severalcases \ below practical limits of detection for the method em- ISUA METAOUNIfE ployed. The high concentration of Ni (-2350 ppm) is I consistent with the inferred original olivine-rich nature of the sample, whereashigh Cr (-960 ppm) is probably due to spinel in the protolith. Of particular note is the high concentration of Pb (- 13 ppm), which is contained in the rare sulfide mineral shandite (Dymek, 1987).

Rare-earth elements Concentrationsof eight REEs were determined by iso- tope-dilution mass spectrometryat Brown University us- ,05 ing methodsdescribed in Gromet et al. (1984).A chondrite-normalized REE pattern (Fig. l) revealsan unusual .02 shape,with concave-downlight REE, concave-up hearry REE, and a prominent negative Eu anomaly (Eu/Eu* : .01 0.376). Ce lrrd SmEu Gd Dy Er Yb This pattern shapeis broadly similar to those reported Fig. l. Chondrite-normalizedrare-earth-elementpattern for for cumulate dunites of igneous origin, some of which Isuametadunite sample IW809-4. Also illustratedare data for have negative Eu anomalies (Fig. 1). The REE pattern a soapstoneof similarbulk composition(from Jahn et al., 1980), shape of IW809-4 is also similar to that reported for a and a field for ultramaficcumulate rocks-mostly of ophiolitic "soapstone" from the Tipasgarvi greenstonebelt, Fin- affinity(from Frey et al., 1971;Menzies, 1976; Montigny et al., land, which is also shown in Figure l. Jahn et al. (1980) I 973;Noiret et al., I 98I ; Pallisterand KniCht, I 98I ; Suenet al., attributed the REE characteristicsofthe "soapstone" to 1979).All REEare normalized to the chondriticabundances of (1974). alteration efects and suggestedfurthermore that such data Nakamura provide no constraintswhatsoever on the magmatic evo- droxyl bands; perhaps incipient serpentinization is the lution of their sample. causeofthe fuzziness. Finer-grained aggregatesof the clear olivine surround grains. Pnrnocnl.pHy AND MINERAL cHEMrsrRy many of the above-mentionedlarger relict olivine These crosscut the magnetite "dust" marking the relict In thin section, the metadunite has a "spotted" ap- grain boundaries,but are themselvesreplaced by (or sim- pearance presenceof large (several milli- owing to the ply intergrown with) elongatey-serpentine (Fig. 2B). This meters) subrounded areas that consist of relict olivine crosscuttingrelationship suggeststhat clear olivine post- grains (or, less commonly, magnesite), which are sur- dates the fuzzy olivine. rounded and replaced to variable degreesby serpentine Representativeolivine analysesare listed in Table 2 and magnetite. An example of the relict olivine is illus- (nos. I and 2). Both textural types are highly magnesian, in Figure 24, which consistsof several polygonal trated but the clear variety is slightly more Mg rich (Fo" vs. from fine magnetite cells that are separated each other by Fon,; Fig. 3). The high Fo content of these olivines is dust. consistent with an origin through prograde metamor- These featurescorrespond to what Wicks et al. (1977) phism ofserpentinite. This interpretation is supported by have termed meshtexture. Thus, eachpolygonal cell con- the fact that in all other analyzedIsua ultramafic rocks, tains an olivine core (mesh center) enclosed in platy, x9', - xh!'-, whereasin IW809-4, Xg! > x*nfu (Dymek, Iength-slow (meshrim). The individual mesh 7-serpentine unpub. results). of olivine tend to extinguish simultaneously, in- centers The abundance of minor elements are near or below dicating that they are segmentsof optically continuous limits of detection in both varieties of olivine, except for crystals.In some cases,an olivine mesh center is a single NiO and MnO, which range up to -0.5 wto/oeach (Fig. multicrystal aggregatein which the individual crystals 3). The fact that NiO exceedsMnO is somewhat unusual have rectangular polygonal cross-sections. to for metamorphic olivine, but most likely due to the absenceof phaseslike talc, which tend to concentrate Ni Olivine compared to olivine (Evans, 1977). Olivine rangesfrom fuzzy and pale-brown to clear and colorless.Most of the olivine mesh centersconsist of the Titanian humites fuzzy vaiery, and a spectrum of this material obtained The identification of humite-group minerals was facil- with the laser Raman microprobe (r-nru)reveals weak hy- itated by their intense yellow color and pleochroism ))u DYMEK ET AL.: CHONDRODITE AND CLINOHUMITE IN METADUNITE

ip \A {*,

, - i.. I

,q "; ;.n .j*!P'

Fig. 2. Photomicrographs illustrating selectedtextural features of the metadunite. (A) Rounded, relict "fnzzy" olivine grain consistingofoptically continuous polygonal subareasseparated from eachother by zonesofplaty serpentineand magnetite;arcuate clear olivine surrounds this grain at left center (transmitted, plane-polarizedlight; bar scale: 0.5 mm). (B) Relict "fuzzy" olivine (right of center) surrounded by platy serpentine.Note blocky magnetite (lower center) and clear olivine (left and far right); clear olivines are cut by elongateserpentine (transmitted plane-polarizedlight; bar scale: 0.5 mm. (C) Titanian chondrodite (rounded darker grain in center of field) enclosedin clear olivine (transmitted, plane-polarized light; bar scale:0.1 mm). (D) Titanian chondrodite (high relieffrom left to center offield) intergrown with blocky magnetite,elongate serpentine and sulfide (largeopaque in lower right). Note platy serpentineabove sulfide and in lower left (transmitted, plane-polarizedlight; bar scale: 0.1 mm). (E) Lamellar structure in titanian chondrodite-clinohumite intergrowth (transmitted, cross-polarizedlight; bar scale: 0.1 mm). (F) Magnesiteporphyroblast surroundedby platy serpentineand cut by elongateserpentine (arrow near center);the latter also cuts platy serpentinehere (transmitted, plane-polarizedlight; bar scale: 0.1 mm). DYMEK ET AL.: CHONDRODITE AND CLINOHUMITE IN METADUNITE 551

TaeLe2. Representativeanalyses of olivine,serpentine, and magnetitefrom lsua metadunite1W809-4 0.6

o 0,4 Mgo 54.64 55.73 39.84 40.33 2.35 = Alr03 0.00 0.00 1.21 0.41 0.00 ;s ^^ sio, 42.13 42.20 40.70 42.42 0.05 CaO 0.00 0.00 0.01 0.01 0.03 = Tio, 0.00 0.00 0.06 0.02 000 0,0 Vro" 0.02 0.97 Cr"O" o.02 0.03 0j2 0.11 o.27 MnO o.28 0.29 0.00 0 01 011 FeO- 3.12 2.O1 1.48 0.98 89.52 0 Nio 0.40 0.39 0.04 0.08 1.04 0.08 o ZnO Sn 93.47 =' ,4 Total (wt%) 100.59 100.65 83.46 84.35 . "..1 Formula proportions ;e Cations 3 31010 + n, Si 0.997 0.993 3.955 4.079 0.002 Ti 0.000 0.000 0.004 0.001 0.000 0.000 0.000 0.138 0 046 0.000 0,0 AI 0.99 0.97 0'95 Cr 0.000 ':o ':o o:' 0.008 Fe3* 1.988 Xnc 0.001 Fig. 3. Wto/oNiO and MnO vs. X* in olivine from meta- 5.780 0.132 Mg 1.928 1.954 5.769 notethat "clear" olivineappears slightly more Fe2* 0.062 0.040 0.120 0.079 0.831 duniteIW809-4; Mn 0.006 0.006 0.000 0.001 0.003 Mg rich than "fuzzy" olivine. Ni 0.008 0.007 0.003 0.006 0.032 Zn 0.000 0.000 0.002 0.000 0.000 0.001 0 001 0.001 is probable that some of the Ti-Cl compositions are shift- Xus 0.969 0.980 0.980 0.987 0.137 ed slightly becauseof beam overlap with the host Ti-Ch phaseduring analysis. Note: Columns are (1) olivine (tuzzyl, (2) olivine (clear),(3) serpentine (platy),(4) serpentine(elongate), (5) magnetite(Calc. Fe,Os: 70.15 wto/.' Averagecompositions of the Ti-Ch and Ti-Cl are listed FeO : 26.39wt%, total : 100.50). in Table 3. Both are highly magnesian(X', = 0.98) with - proportions F€F* for Fe'?*apply Total Fe as FeO; formula listed for and amounts of minor elements,principally MnO only to magnetite. only small and NiO, and have low contents of F as well. Values for Fe/Mg, Ni/Mg, and Mn/Mg are essentiallyidentical, but TilMg and TilSi are clearly distinct (Table 4)' The fact (which resemblethose of staurolite). Humite crystalsare that Tilsi in Ti-Ch is approximately double that in Ti- all xenoblasticand range in size from -15 to 250 pm. Cl supports the preferential substitution of Ti into the They occur as inclusions in fuzzy olivine, as isolated Mg(OH,F)Olayer, as outlined by Ribbe (1979).The com- massessurrounded principally by platy serpentine,or as positional characteristicsof the Isua humites are consid- complex intergrowths with magnetite, sulfide, and elon- ered in more generalterms in the discussionsection. gate serpentine(Figs. 2C, 2D,2E). Most grains are optically homogeneouswith inclined extinction, but a few Serpentine show patchy coloration suggestiveof compositional zon- Serpentine forms platy masses, many of which are ing. Other grainsexhibit a distinct lamellar structure (Fig. chargedwith extremely fine-grained particles of magne- 2E); analysesreveal these to be intergrowths of titanian chondrodite (Ti-Ch) and titanian clinohumite (Ti-Cl). Ti- Ch is by far the more abundant of the two humite-group 'o ' a -a primarily in regions with ideol CHONDROOITE . . minerals, whereasTi-Cl occurs tl% ,r toa lamellar structure. In one case, however, a thin partial ia .- a rim of Ti-Cl occurs on Ti-Ch, where they are separated a -' by an optical discontinuity. As illustrated in Figure 4, the chondrodite phase contains variable TiO, (-8.5-10 wto/o),and has a metal/Si ratio close to the ideal value of 2.50. The clinohumite - phase also contains variable TiO, (-5.5-7.0 wto/o),but has M/Si slightly above the ideal value of 2.25. This apparent departure from stoichiometry is most likely caused by the mode of occurrenceof the Ti-Cl as thin lamellae. Backscattered-electronand X-ray imaging reveal streak- 0,0 2 0 4,0 6,0 8,0 10.0 wt % Ti02 ing and heterogeneousTi distribution parallel to lamellae orientation on a scale ranging from l0 pm down to the Fig. 4. Metal/Si vs. TiO, content in chondrodite and clino- limits of practical resolution (ca. <2 prm).Accordingly, it humite from metadunite Iw809-4. 552 DYMEK ET AL.: CHONDRODITE AND CLINOHUMITE IN METADUNITE

Tnele 3. Compositionsof titanianhumites in lsua metadunite TneLe4. Selectedcation ratios. for coexisting titanian humites tw809-4 in lsua metadunite1W809-4

Chondrodite Clinohumite Mgo s0.55(0.46) 52.34(1.15) FelMg 0.023+ 0.003 0 021+ 0.003 Atros 0.00(0.01) 0.02(0.03) Ni/Mg 0.003+ 0.001 0.004+ 0.001 sio, 33.96(0.45) 37.06(0.68) Mn/Mg 0.004+ 0.000 0.004+ 0.000 CaO 0.00(0.00) 0.00(0.00) Ti/Mg 0 093+ 0.005 0.060+ 0.008 Tio, 9.31(0.43) 6.18(0.65) Ti/Si 0.206+ 0.011 0.125+ 0.015 Cr,O" 0.01(0.02) 0.02(0.02) MnO 0.37(0.04) 0.35(0.02) " Data from Table3. FeO 2.O9(O.241 1.99(0.21) Nio 0.28(0.06) 0.35(0.08) ZnO 0.01(0.01) 0.01(0.01) F 0.28(0.06) 0.15(0.05) ment substituents present in quantities significantly above cl 0.00 0.00 Hro* 2.84 1.34 detection limits are CrrO. and AlrOr, which range up to 99.70 99.81 0.3 wto/oand 2.0 wt0/0,respectively. These analysesalso -O=F 0.12 0.06 demonstrate that platy serpentineis slightly but consis- Total (wto/.) 99.58 99.74 tently more aluminous and lessmagnesian than the elon- Formulaproportions gatetype (Fig. 5). Thesedifferen@s perhaps could be caused 't3 Cations 7 by "contamination" of platy serpentineby the tiny mag- Si 2.O07 3.949 Ti 0.415 0.495 netite inclusions, thus making them appearmore Fe-rich. AI 0.000 0.002 However, it is not possible to account for differencesin Cr 0 001 0.001 Al content by this mechanism. Wicks and Plant (1979) Mg 4.441 8.312 also observedthat serpentineplates from mesh rims after Fe 0.103 0.177 Ni 0.013 0.030 olivine contained AlrO, (whereasprecursor olivine has Mn 0.018 0.031 none). Zn 0.000 0.001 portion Ca 0.000 0.000 Serpentinecompositions are shown on a of the MgO-FeO-SiO, ternary diagram in Figure 6. The data F 0.052 0.050 UI 0.000 0.000 points for platy serpentine cluster along the oH- 1.118 0.960 (Mg,Fe)rSirOr(OH)ojoin, whereasthose for elongateser- M/Si 2.495(0.026) 2.292(0.018) pentine plot on the SiOr-rich side of this join. Thesecom- Xlas 0.977(0.003) 0.978(0.003) positional features, together with differencasin Al con- Note: Columnsare (1) chondrodite(average r1 q for 19 analyses),(2) tents noted above,suggest that platy serpentineis lizardite clinohumite(average !1 6 tor 7 analyses). . whereaselongate serpentine is antigorite (Wicks and Plant, Calculatedfrom anion proportions -- Calculatedtuom (2 - 2Ti - F - Ct). 1979).Indeed, severalinvestigators have determined that lizardite with a platy or banded texture is typical of serpentine mesh rims (Wicks and Whittaker, 1977). The presenceof antigorite in sample IW809-4 was con- tite, locally concentratedin layers.These inclusions, along firmed by X-ray diffraction analysis (which also demon- with flexing and/or bending (causedby positive AV dur- strated the absenceof brucite and chrysotile); however, ing hydration of olivine?), impart a banded structure to the presenceof lizardite could not be established.Simi- the serpentine.This variety is found most commonly as larly, the LRM spectrum of elongate serpentine matches pseudomorphsafter olivine (Figs.2A, 2B). Serpentinealso almost perfectly that of an antigorite standard, whereas occurs as ragged,elongate sheaths that crosscutthe platy the nr'r spectmm of platy serpentineis quite unlike that variety, and are also intergrown with the humites (Fig. of a lizardite standard (but is virtually indistinguishable 2D). The two types of serpentine are commonly found from the IW809-4 antigorite). The ability of the rnrvrto together(Fig. 2D), and the diference betweenthem is not distinguish among the serpentine "polymorphs" is dis- always so conspicuous. cussedby Pasterisand Wopenka(1987). Representativeanalyses of the two serpentinetypes are These conflicting results, whereby Mg/Si and Al con- listed in Table 2 (nos. 3 and 4). Analytical totals are un- tent suggestthe presenceof both lizardite and antigorite fortunately low (by about 34 wtolofrom expectedvalues but xnp and rnrrr indicate only antigorite, can be recon- near 87 wto/o),a feature that appearsin many published ciled if the platy serpentinein IW809-4 crystallized orig- serpentineanalyses (e.g., Cogulu and Laurent, 1984).This inally as lizardite but "inverted" to antigorite during a problem is symptomatic of all our analysesof Isua ser- later event during which the elongateantigorite formed. pentines (Dymek, unpub. results), which we attribute to The fact that elongateserpentine cuts platy serpentine,as instability of serpentine under the electron beam even noted above, is consistentwith this interpretation. though a spot diameter of 20 pm was used routinely. The results, as such, yield very good formula propor- Magnesite tions, and show that both serpentine compositions ap- Magnesite forms large porphyroblasts, similar in size proach the pure Mg end member. The only minor ele- to the relict olivine grains, and is surrounded by platy DYMEK ET AL.: CHONDRODITE AND CLINOHUMITE IN METADUNITE 553

si02

/ \F" ,t \4 54 t ,' ,' ,/ ,' vv. d! r ll-" ? ---. d elongote _N i o.o1 ra a -31 -s aa 50 l- MgrSi,0r(0H), 0,99 0,98 09? 0.96 095 .f 49 xug pl0ty %Ms) Fig.5. Al/Si vs.Xr, in serpentinefrom metaduniteIW809-4 showingslight chemical differences between the platyand elon- Fig. 6. A plot of normalizedwto/oMgO-FeO-SiO, in serpen- gatevarieties. The fact that the data points for eachtype are tine from metaduniteIW809-4. arrangedin approximatelyparallel arrays ofincreasing Al/Si and decreasingMg,/Fe suggests a simpleAlrMg-rSi-r Tschermak- (Dy- typesubstitution. (Ni3PbrSr),which are discussedin detail elsewhere mek, 1987). DrscussroN serpentine (Fig. 2F). Magnesite also occurs as smaller Metamorphic history grains disseminated throughout the sample. It is highly A multistage history for the metadunite is required by likely that these smaller magnesitesrepresent a second its complex petrography.We infer a magmatic stage,dur- generationof carbonate. ing which the primary dunite formed, followed by an Analysesshow that the magnesiteis nearly pure MgCO, episodeofearly serpentinizationand carbonation, which (X., = 0.99), with the following averagecomposition (in is needed to account for the subsequentgrowth of very wto/o):MgO, 47.0;FeO, 0.80; MnO, 0.43;CaO, 0.08. Mg-rich olivine during prograde metamorphism. This prograde event yielded fuzzy olivine, humites, blocky Magnetite magnetite,magnesite, and sulfides.This was followed se- Magnetite occurs not only as the micrometer-sized quentially by formation of platy antigorite (i.e., the mesh "dust" associatedintimately with serpentine, but also texture), formation of clear olivine, and finally by later forms blocky grains (up to 50 pm across)that are found overgrowth of bladed antigorite (and additional magne- as inclusions within, or intergrown with, olivine, serpen- site?)on the other phases. tine, humites, and sulfide (Figs. 2C, 2D,2E). A represen- The bulk composition of the metadunite is approxi- tative analysisof blocky magnetite (Table 2, no. 5) shows mated very closelyby the model systemMgO-SiOr-HrO- that it contains little CrrOr, has essentiallyno detectable COr. In this system,Evans (1977) noted that the assem- Al2O3,V2O3 or TiOr, but has unusually high contents of blage of forsterite + magnesite* antigorite lies along the MgO (2-3 wto/o)and NiO (0.8-1.2 wto/o).Such values are univariant curve among the highestreported for a relatively pure Fe3+spi- antigorite + magnesite: forsterite, (1) nel phase (cf., Haggerty, 1976). The high Xru is in line with the overall highly magnesiancharacter of the sili- at a pressureof 2000 bars and a temperature of -450- cates,whereas the elevated Ni content is most probably 480'C for X.o. between0.0 and 0.2 (Fig. 7). This tem- a result of the fractionation of that element into the mag- peraturerange is quite closeto the estimateof -470'C netite during serpentization of an originally Ni-bearing for retrogrademetamorphism of the Isua belt, as inferred olivine [althoughit may also relate to reactionswith nick- from studies of pelitic assemblages(Boak and Dymek, el sulfides,as discussedin Dymek (1987)1.It was not 1982).The severalstages ofserpentization and recrystal- feasible to obtain reliable analvses of the fine-erained lization could conceivably representfluctuations in fluid magnetrte. composition and temperature, such that Reaction I has been crossedat least twice, to form two separategener- Sulfides ations of olivine, magnesite,and antigorite. Sulfides occur as somewhat amoeboid-shapedgrains, As shown in Figure 7, Reaction I intersectsan invari- up to 250 pm across,that are intergrown with clear oli- ant point involving talc. Throughout the Isua belt, talc + vine, titanian humites, magnetiteandlor elongateserpen- magnesite + antigorite schist is ubiquitous as the retro- tine (Figs. 2D , 2E). In addition, they are found as isolated grade product of metaperidotite. However, talc is not masses(with magnetite)enclosed in platy serpentine.Each found in sample IW809-4, which is problematic: either grain comprises composite aggregatesof heazlewoodite it never formed or else it was eliminated completely by (Ni.Sr), cobalt pentlandite (Co?Ni*Feo rSr), and shandite later reactions. 554 DYMEK ET AL.: CHONDRODITE AND CLINOHUMITE IN METADLTNITE

= a Too ideol 'tfaa L ^ a . ,tf cLtit0Hur rTE I A AA 2.20 00 2.0 4,0 8,0 wt% Ti02

0.1 X^^ 0,3 Fig. 8. Metal/Sivs. TiO, contentin Isua titanian humites vv2 comparedto other occurrences(sources listed in Fig. l0); tie lines compositionsof intergrownchondrodite-clino- Fig. 7. T-X.o" diagramshowing the relationshipamong the connect humitereported by Aoki et al. (1976). phasesforsterite (Fo), magnesite (Mg), antigorite (Ant), and talc (Tc).[Adapted from Evans,1977; see text for discussion.]The arrowshows the bulk compositionof the Isuametadunite along thejoin magnesite-talc. intergrowths that have yet to be recognized.The few other analysesof Ti-Ch have M/Si very close to the ideal value of 2.50. Xr, in the other samplesspans a wide range The latter interpretation seems more consistent with of values(-0.84-0.96), which in all casesare more Fe- regional geologicrelationships. Early progrademetamor- rich than the Isua humites (Fig. 9). The Ti-Ch and Ti-Cl phism of the ultramafic rocks produced assemblageswith pair from Aoki et al. (1976) show similar values for X'r, magnesiocummingtoniteand/or anthophyllite (Brothers as do the Isua data. In terms of TiO, content, the analyses and Dymek, 1983), which are above the stability of an- of Isua Ti-Cl are slightly but consistently more Ti-rich tigorite. For the bulk composition of IW809-4 (shown by than the other occurrences,whereas the Isua Ti-Ch are the arrows in Fig. 7), talc would have appeared above also slightly more Ti-rich (on average)than the others. reaction (Mag), albeit in small amounts. Subsequentret- Collectively,these data and comparisonsreveal how close rograde metamorphism probably followed a T-X"o,path the Isua humites are to the ideal end-member composi- such as shown by the dashed line in Figure 7, which would tions for Ti-Cl and Ti-Ch, MgrSioO"(OH)'Mgo'TiorO' have eliminated talc and produced the observed assem- and MgoSirOr(OH)'Mgb'Tio rOr, respectively. blage of forsterite + magnesite + antigorite. Trommsdorf and Evans (1980) have remarked upon Regardlessof such speculations,the fact that titanian the general absenceof Fe-Mg fractionation between co- chondrodite occursas inclusions infuzzy olivine suggests existing titanian humites and olivine, which is supported that it formed relatively early in the evolution of this by our results. Figure l0 summarizes data on olivine- sample. Thus, the "late" history of IW809-4 places no humite pairs; with the exception of two occurrences(one limitations on the P-Zregime in which the humite phase of which is Mn-rich), the data indeed fall closeto the line originally crystallized-a regime that was most likely (K")Fe-Mg : l.0. For the Isua humites, equal Fe-Mg par- above the stability ofantigorite basedon the overall char- titioning is best for the clear olivine. acter of the Isua ultramafic rocks, as noted above. Hence, even though Trommsdorf and Evans (1980) otreredcon- Some commentson Ti in the humites vincing evidence that the stability of hydroxyl Ti-Cl is The humite minerals constitute a homologous or poly- restricted to antigorite metaperidotite, the same conclu- somatic seriesthat has been considered traditionally as sion need not apply to Ti-Ch. We consider this point in combinations of olivine and brucite (or sellaite) groups more detail in a later section. having the generalformula n(MgrSiOo)'Mg(OH,D, (n: l, -- norbergite; n : 2, chondrodite; fr : 3, humite; ft 4, Comparisonwith other titanian humites clinohumite). Ribbe et al. (1968) pointed out that such Compositions of titanian humites, compiled from a va- groups, in fact, do not exist and have no structural sig- riety of sources,are summarized on Figures 8 and 9. In nificanceand that it would be more appropriate to regard, both diagrams, the tie lines connect the compositions of for example, the norbergite formula as Mg'SiOr(OH,F)' intergrown Ti-Cl and Ti-Ch reportedby Aoki etal. (1976; Mg(OH,F)O. The various members of the humite series cf. Fujino and Tak6uchi, 1978). can thus be viewed as stacked,ordered sequencesofnor- The other analysesof Ti-Cl yield M/Si ratios that scat- bergite and forsterite modules, as discussedby Thomp- ter about the ideal value of 2.25 (Fig. 8); perhaps those son(1978). three compositions with the highest M/Si contain humite Each member of the humite series has a correspond- DYMEK ET AL.: CHONDRODITEAND CLINOHUMITE IN METADUNITE 555

I F= F= \ ogo oo I

F: .lo ^ Smilh(1979) ts= . Cinminoel ol,(1979) 0 Mitchell(1979) v Desmorois(l98ll (1980) ^1 -g r TrommsdorffI Evons il ; . EvonsI Trommsdorff(1983)

100 0.0 2.o 40 60 80 100 l oo 095 oto"o""" o'tt o8o wt % Ti02 ^Ng Fig. 9. Xr, vs. TiO, content in Isua titanian humites com- Fig. 10. X* in coexistingolivine and titanian humite. As pared to other occurrences(sources listed in Fig. l0). Note that notedby TrommsdorffandEvans (1980), most pairs lie nearthe the Isua analysesare more magnesianthan all others and that Kp : 1.0line, althoughthere is a slighttendency for olivine to there is no significant Fe/Mg fractionation between the Ti-Ch be moreFe-rich. and Ti-Cl. ing Ti-rich variety described by the general formu- tions of the chondrodite and clinohumite structures(i.e., la I(n l)Mg,SiO,l[Mg,SiO,(OH,F).Mg, exclusively within the norbergite module), it is difrcult "Ti"- (OH,F), ,rOr*r*f, in which X cannotexceed 0.5 (Ribbe, to account for these findings solely in terms of crystal 1979). Compositional variations in titanian humites can chemistry. As discussedin the next section, the differ- be portrayed by means of a simplified model chemical encesbetween the two Isua humites may reflect variable system (Evans and Trommsdorfl 1983) defined by three P- Z conditions of their formation. linearly independent exchangevectors: (l) FeMg ,, (2) F(OH) ,, and (3) TiO,Mg-,(OH)-,; an additional ex- Petrogenesisof the titanian humites change,(4) TiOrMg_,F_r, is an alternativeto exchange3. Trommsdorff and Evans (1980) observed that the This model system, condensedalong FeMg-,, is shown breakdown of Ti-Cl in the Alpine peridotites is marked in Figure I l. In order to examine relationshipsamong Ti, by the appearanceof geikielite (MgTiOr). One might ex- F, and (OH) substitutions, it is useful to transform this pect that the breakdown of Ti-Ch would likewise involve ternary to a binary diagram in which the axes,&, (: Ti a geikielite-forming reaction. We have searchedfor this cations/formula) and X. (: 0.5 F anions/formula), are the mineral at great length but have failed to find it. The measurableparameters of interest (seeFig. l2). On such apparent absenceof geikielite suggeststhat the thermal a diagram, exchange2 operatesvertically, exchange3 op- stability of the titanian humites has not been approached erateshorizontally, and exchange4, which is actually a and additional possibilities should be explored. For ex- linear combination of 2 and 3, is representedby a 45' line. t4l Ti1zvg_tF-z Joneset al. (1969) first evaluated Ti and F contents of the humite minerals. Their analyses(not plotted) scatter Mq.SiaO,s(F) MqsSiaOts(F) MgeSiaOl5 . . . in the area below and to the left of the 45" line in Figure M9(F)O M9.5Ti.5O2 TiOz 12, indicating no generalrelationship betweenTi and F. Evansand Trommsdorff(I983), however, drew attention to the fact that their analysesof Alpine Ti-Cl formed a occtsstble linear array that plotted very close to the 45" line (Fig. \ P MgaSioOls(OH) MgsSiaO15 (OH) .$- ^ . l2), thus indicating a strong coupling of Ti with F. They -^ .Mg(F)0 Mg,5Ti,5O2 .; -tr argued that such evidence for internal (crystal-chemical) \q s'v control on exchanges2 and3 was apparentonly in closely -€ \ related specimens. A Analyses of Isua Ti-Cl extend this array to still lower values of F. Analysesof Isua Ti-Ch, however, plot below Mg.SiaO15 (OH). Mg (OH)O the 45'line and form an approximate horizontal array. Fig. ll. Model chemical system for the humite minerals This suggests(assuming negligible contributions from an- (condensedon MgFe-,). Although this diagram is drawn for alytical error) that the two humites behave somewhatdif- clinohumite, identical relationshipsapply for other members of ferently; becauseTi substitution occurs on the same por- the series(adapted from Evans and Trommsdorfl 1983). 556 DYMEK ET AL.: CHONDRODITE AND CLINOHUMITE IN METADUNITE

0,4 fi0, Mg.,F_, h

c0 be

o0. FoA \W

0,1 Ge

r..ttb. 00 0t 02 0,3 0,4 05 Xr,= Form. Prop. Ti

Fig. 12. A plot of X. (: 0.5 F anions/formula) vs. X'' (: Ti cations/formula) for titanian humites. Data for Ti-Cl fall near T the 45' line, closely approximating TiOrMg-,F-, exchange, Fig. 14. SchematicP-7 orientationfor the four univariant whereasdata for Ti-Ch plot below this line (see text for addi- reactionsin theternary subsystem Fo-Ge-W (HrO). Intergrowths tional discussion).Note that Evans and Trommsdorff(1983, p. of Ti-Cl and Ti-Ch and the total absenceof geikelitemay indi- 36 1) statedthat their Fig. 3, from which the presentone is adapt- cateconditions between the reactions(Ge,W) and (Fo) (see text ed, usesXF: lF/(F + OH) as a plotting parameter.When Xo is for discussion). defined this way, the dashedline in Fig. 1l maps as a curve in the above figure;however, data points in their Fig. 3 were plotted (correctly)using X. as definedherein (B. W. Evans,pers. comm.). Ti-Cl: Ti-Ch + 2Fo (Ge,W) 2 Ti-Ch: Ti-Cl * t/zGe-r lzW (Fo) Ti-Cl: 4Fo 'r t/zGe-r lzW (Ch) t/zW. ample, Hinz and Kunth (1960) have determined that, in Ti-Ch: 2Fo + VzGet (CD the system MgrSiOo-MgFr, clinohumite breaks down to The P-T orientation of these reactions is shown sche- chondrodite + forsterite. The lack ofF and the presence matically in Figure 14. Becausereaction (Ch), which rep- of high Ti contents in the Isua humites preclude direct resentsthe terminal breakdown of Ti-Cl, has been shown application of these experimental results. Nevertheless, through experiment to have a very steep positive slope the general idea of a humite series "reaction relationship" (Engi and Lindsley, 1980), the other reactions are re- may provide insight into the interpretation ofour obser- quired topologically to have Ti-Ch on the high-pressure vations. side. Of particular significanceto our interpretations is The compositions of the Isua humites are contained the vapor-absent reaction (Ge,W), which representsthe almost completely in the ternary system Mg'SiOo-Mg- formation of Ti-Cl from Ti-Ch + Fo. TiO3-HrO (see Fig. 13) in which the following four uni- Becauseof the generallack of thermodynamic data for variant reactions characterize relevant phase relations: the titanian humites, calculatedslopes for thesereactions are uncertain. Nevertheless,applying the Clapeyron relationship to the data compiled by DuB and Greenwood (1979)indicates dP/dT = 2.5 kbar/100 K for (Ge,W) in MgTi 03 the Ti-free system,with AZ* : 9.6 cal'kbar-'. Engi and Lindsley (1980) reported AZ* : 12 + 3 cal'kbar-' for (Ge,W) in the Ti system. Unless there are drastic, unex- pecteddifferences in AS* for the Ti-rich systemcompared to the Ti-free system,dP/dT for (Ge,'W)should be similar to the value listed above. Where (Ge,W) intersects (Ch) is unknown, but it should be at high pressure(Z = 600 oC,P x 25 kbar), as Engi and Lindsley (1980) have syn- sio2 MqrSi0. thesizedTi-Ch from Ti-Cl + Ge at T: 600 "C, P : 23 Fo w kbar (this synthesisshould lie betweenthe reactions (Fo) M92Si 0a Hzo and (Ge,W) shown in Fig. l4). in Fig. 13. The quaternarysystem MgO-SiOr-TiOr-HrO, show- Although there are clearly large uncertainties this ing the plane Mg.SiOo-MgTiO3-H,O (Fo-Ge-W) that describes analysis of the titanian humites, the important point re- the compositions of end member Ti-Ch and Ti-Cl; the four uni- garding the reaction topology of Figure I 4 is that the two- variant reactionsin the ternary subsystemare shown in Fig. 14. phase intergrowths of Ti-Ch and Ti-Cl found here (and DYMEK ET AL.: CHONDRODITE AND CLINOHUMITE IN METADUNITE 557 in the kimberlite by Aoki et al., 1976) may representan Desmarais,N.R (1981) Metamorphosed Precambnan ultramafic rocks arrested "decompression" reaction. Accordingly, Ti-Ch in the Ruby Range,Montana PrecambrianResearch, 16, 67-101. H.J. (1979) Phaseequilibria in the system further possible Duffy, C.L., and Greenwood, ought to be evaluated as a upper-mantle MgO-MgFr-SiO'-H'O. American Mineralogist, 64, | 156-l l7 2. mineral, as it is the expectedhigh-pressure phase. Infor- Dymek, R.F. (1987) An occurrenceofshandite, NiiPbrsr, in a serpentin- mation on other occunencesof titanian humites and ex- ized metadunite from the Isua supracrustalbelt, West Greenland Ca- perimental data on the stability of Ti-Ch are needed to nadian Mineralo glsr, 25, 24 5-249 (1983) (Ti-Ch) resolve this issue. Dymek, R.F., Boak,J.L, and Brothers,S. Ti-chondrodite and Ti-clinohumite (Ti-Cl) in ca. 3800 Ma metadunitefrom Isua, West In regard to the Isua sample, we do not consider this Greenland(abs.). EOS, 64, 327. metadunite to be of mantle origin, as there is no petro- Engi, M., and Lindsley, DH. (1980) Stability of titanian clinohumite: logic or geochemicalevidence among the suite of ultra- Experimentsand thermodynamic analysis.Contributions to Mineral- mafic rocks to warrant such a conclusion. It does appear, ogy and Petrology,72,415424. ( Alpine peridotite and serpentinite. passed Evans,B.W I 977) Metamorphism of however, that its metamorphic P-T-t lrajectory Annual Reviews ofEarth and PlanetaryScience, 5, 397447. through a "higher pressure" region where Ti-Ch was sta- Evans,B.W., and Trommsdorfl V. (1983) Fluorine hydroxyl titanian cli- ble and that later retrogressioncaused the Ti-Ch to react nohumite in Alpine recrystallized garnet peridotite: Compositional partly to Ti-Cl. controlsand petrologicsignificance. American JournalofScience, 283-'4 (Orville volume),355-369. Acxxowr,nrcMENTS Frey,F A., Haskin,L.A., and Haskin,M.A. (1971)Rare earth abundances in some ultramafic rocks. Journal ofGeophysical Research,76,2057- Funding for this project was provided by grants ftom the National 2010. ScienceFoundation (EAR78-234I 2 and EARS I -09466) and the National Fujino, K., and Tak6uchi, Y. (1978) Crystal chemistry of titanian chon- Aeronauticsand SpaceAdministration (NAGW-484 and NAG9-98; Ear- drodite and titanian clinohumite of high-pressureorigin. American ly Crustal GenesisProgram). Additional support for field work at Isua Mineralogist, 63, 535-543. was provided by a grant from the National GeographicSociety. Gromet,L P., Dymek, R.F., Haskin,L.A., and Korotev,R.L. (1984)The We thank J. Sparks,University of Massachusetts,Amherst, for assis- "North American Shale Composite" (NASC): Its compilation, major tance in obtaining the X-ray fluorescenceanalyses, and L. P. Gromet, and trace element composition. Geochimica et Cosmochimica Acta, Brown University, for use of his laboratory for the rare-earth-element 48,2419-2482. determinations. We also thank the Department of Geological Sciences, Haggerty,S E. (1976) Opaque minerals in terrestrial igneousrocks. Min- Harvard University, for permission to use their microprobe laboratory, eralogicalSociety of Amenca Reviewsin Mineralogy,3, H6101-H6300. while we were in residencethere Hinz, W , and Kunth, P.O. (1960) Phaseequilibrium data for the system Comments on an earlier version of this manuscript by H. W. Day, C. MgO-MgFlSiO.. American Mineralogist, 45, 11 98-1 2 I 0. A. Francis,and L. P. Gromet helped to clarify and focus our presentation, Jahn,B.M., Suvray,B., Blais,S., Capdevila,R., Comichet,J., Vidal, F., and their eforts are most appreciated.Finally, joumal reviews by B W and Hameurt, J. (1980) Trace element geochemistryand petrogenesis Evans and M. Engi resulted in considerableimprovement of the paper. of Finnish greenstonebelts. Joumal of Petrology, 21, 201-244. J D. Pasteris,Washington University, kindly obtained the laser Raman Jones,N.W., Ribbe, P.H., and Gibbs, G.V (1969)Crystal chemistry of microprobe spectraon the serpentinesand olivine. We thank Freda Sofian the humite minerals.American Mineralogist, 54,391411. for typing the manuscript. Mccetchin, T.R., Silver,L.T., and Chodos,A.A. (1970)Titanoclinohum- ite: A possiblemineralogical site for water in the upper mantle. Joumal Rnrnnouces crrno of GeophysicalResearch, 7 5, 25 5-259. Menzies,M. (1976)Rare earth geochemistryof fusedophiolitic and alpine Allaart,J.H (1976)The pre-3760m.y. old supracrustalrocks ofthe Isua lherzolites,I. Ofhris, Lanzo and Troodos. Geochimica et Cosmochim- area,central West Greenland,and the associatedoccurrence ofquartz- ica Acta. 40.645-656. bandedironstone. In B.F. Windley, Ed., The early history ofthe Earth, Mitchell, R.H. (1979) Manganoan magnesianilmenite and titanian cli- p. 177-189.Wiley, New York. nohumite from the Jacupirangacarbonatite, Sao Paulo, Brazil. Amer- Aoki, K, Fujino, K., and Akaogi, M (1976) Titanochondrodite and ti- ican Mineralogisl, 63, 544-547. tanoclinohumite derived from the upper mantle in the Buell Park kim- Montigny, R., Bougault, H., Bottinga, Y., and Alldgre, C.J. (1973) Trace berlite, Anzona, USA Contributions to Mineralogy and Petrology,56, element geochemistryand genesisof the Pindos ophiolite suite. Geo- 243-253. chimica et Cosmochimica Acta, 37, 2135-2147. Baadsgaard,H , Nutman, A P, Bridgwater, D, Rosing, M., McGregor, Nakamura, N. (1974) Determination of REE, Ba, Fe, Mg, Na and K in V.R , and Allaart, J.H. (1984) The zircon geochronologyofthe Akilia carbonaceousand ordinary chondrites. Geochimica et Cosmochimica associationand the Isua supracrustalbelt. Earth and PlanetaryScience Acta. 38. 757-775 Irtters. 68. 221-228 Noiret, G , Montigny, R., and Alldgre, C.J. (1981) Is the Vourinos Com- Boak,J.L., and Dymek, R F. (1982)Metamorphism of the ca. 3800 Ma plex an island arc ophiolite? Earth and Planetary Sciencel-etters, 56, supracrustalrocks at Isua, West Greenland:Implications for early Ar- 3'15-386. chaeancrustal evolution. Earth and PlanetaryScience Letters, 59, 155- Nutman, A.P., Allaart, J.H., Bridgwater, D., Dimroth, E., and Rosing, t76. M. (1984) Stratigraphicand geochemicalevidence for the depositional Bndgwater, D., and McGregor, V.R. (1974) Field work on the very early environment of the early Archaean Isua supracrustalbelt, southern Precambrian rocks of the Isua area. Rapports Gronlands geologiske West Greenland PrecambrianResearch, 25, 365-396. Undersogelse,65, 49-54. O'Leary,EF., and Dymek, R.F. (1987)BioptTibole in Archaeanultra- Brothers,S., and Dymek, R.F. (1983)Ultramafic metamorphicrocks from mafic rocks, West Greenland (abs.).EOS, 68, 452. the 3800 Ma Isua supracrustalbelt, West Greenland.Geological Soci- Pallister, J.S., and Knight, R J. (1981) Rare-earth element geochemistry ety of America Abstracts with Programs, 15, 534. of the Samail ophiolite near Ibra, Oman. Journal of GeophysicalRe- Cimmino, F., Messiga,8., Piccardo,G.B., and Znda,O. (1979)Titanian search.86. 2673-2697. clinohumite-bearing assemblageswithin antigoritic serpentiniteof the Pasteris,J.D., and Wopenka,B (1987) Use of a laserRaman microprobe Voltri massif (westernLiguria): Inferenceson the geodynamic evolu- to trace geologicalreactions. In R.H. Geiss,Ed., Microbeam analysis- tion ofPiemonteseultramafic sections. Ofioliti. 4. 97-120. t987,205-209. Cogulu, E , and Laurent, P (1984) Mineralogical and chemicalvariations Rhodes,J.M., Blanchard,D.P., Dungan, M.A., Rodgers,K V., and Bran- in chrysotile veins and peridotite host-rocks from the asbestosbelt of non, J.C. (1978) Chemistry of Leg 45 basalts.In W.G Melson et al., southern Quebec.Canadian Mineralogist, 22,173-183. Eds., Initial reports ofthe Deep SeaDrilling Project, 45,447459. 558 DYMEK ET AL.: CHONDRODITE AND CLINOHUMITE IN METADIINITE

Ribbe, P.H. (1979) Titanium, fluorine and hydroxyl in the humite min- Thompson, J.B., Jr. (1978) Biopyriboles and polysomatic series.Ameri- erals.American Mineralogist, 64, 1027-1035. can Mineralogist, 63, 239-249. Ribbe, PH, Gibbs, G.V., and Jones,N.W. (1968) Cation and anion Trommsdorff V., and Evans,B.W. (1980) Titanian hydroxylclilohumite: substitutionsin the humite minerals.Mineralogical Magazine,37 ,966- Formation and breakdown in antigorite rocks (Malenco, Ital/. Con- 975. tributions to Mineralogy and Petrology,72,229-242. Smith, D. (1977) Titanochondrodite and titanoclinohumite derived from Wicks, F.J., and Plant, A.G. (1979) Electron-microprobeand X-ray-mi- the upper mantle in the Buell Park kimberlite, Arizona, USA: Discus- crobeam studies ofserpentine textures. Canadian Mineralogist, 17, 785- sion. Contributions to Mineralogy and Petrology, 61,213-215. 830. -(1979) Hydrous minerals and carbonatesin peridotite inclusions Wicks, FJ., and Whittaker, E.J.W. (1977) Serpentinetextures and ser- from the Green Klobs and Buell Park kimberlitic diatremes on the pentinization. Canadian Mineralogist, I 5, 459-488. Colorado Plateau.In F R. Boyd and H.O.A. Meyer, Eds., The mantle Wicks, F.J., Whittaker, E.J.W.,and Zussman,J. (1977)An idealizedmod- sample:Inclusions in kimberlitesand other volcanics,vol.2, p.345- el for serpentinetextures after olivine. CanadianMineralogist, 15,446- 356. American GeophysicalUnion, Washington,D.C. 458. Suen, C.T, Frey, F.A, and Malpas, J. (1979) Bay of Islands ophiolite suite, Newfoundland: Petrologic and geochemicalcharacteristics with emphasison rare earth elementgeochemistry. Earth and PlanetarySci- M,C,NUSCRTPTRECEIvED AUCUST 8, 1986 enceLetters. 45. 337-348. M,c.NuscRrprAccEprED JeNuenv 20, 1988