American Mineralogist, Volume 58,pages 721-726,1973

Al-RichPargasite

Tneonons E. BuNcn Sprce ScierrceDiuision, Ames ResearchCenter, NASA, Mofien Field, Calilornia 94035 MnnrtN OrnuscHl M i neralogisch-P et ro gr ap hische s I nstit ut der Uni p er sitiit zu Kiiln, D-5000 Kdln, GermanY

Abstract

Two Al-rich were found in a corundum- and -bearingcalcite marble together with minor phlogopite, margarite, anorthite, and sphenb or rutile. Microprobe analysesgive half their unit cell contents as: AmphiboleAA-5,Nao.uoKq.16ca1.eeMg3.rrFeo.orTio.ruAlvrr.4zAlrv2.26si5.z4oHl.4Fo.40clo.o?; AmphiboleHA-l/2,Nao.urKo.rrCar.nnMgr.urFeo.orTio2zAlvrl.r2Alrv2.2ssi5.z5oHr.o5F0.s8cl0.02' AA-5 is the most aluminous yet found in nature (22.6 wt percent AlrO.) and contains essentiallyno Fe (total iron as FeO : 0.05 wt percent). Both amphibolesare too rich in Alrv (uiz, high AlrO, combined with low SiOr) to consider them as mixed crystals strictly intermediate between pure pargasite and pure tschermakite' Al-rich par' gasiteappears to be a suitable name for amphiboles of this composition, owing to the fact that more than 50 percent of their structural formulae can be recalculated in terms of a hypothetical end-member molecule Al-pargasite, (Na,K)CarMgaAlvr2 Alrvs Si5O22(OH,F)2,which conforms to the limit- ing AlrY/Alvr ratio of Ca-amphibolesstatistically evaluatedby Leake (1965' 1971)' The water determination and X-ray powder data are given for one of the amphiboles but not for the other owing to insufficient material. Conditions for the crystallization of the Al-rich amphiboles are briefly discussed.

Introduction mineral separation.The same holds true for four from Leake'scompilation with Alvr > 1.30, The maximum Al-content possiblein calciferous analyses < 1.75 (Si > 6.25). In evaluatinghis and subcalciferousamphiboles is of considerablein- but Alrv Leake (1968, 1971) concludedthat terest for ascertainingcrystallization environments. compilation, amphiboles "the maximum possi- So far, no hornblendeapproaching the tschermakite in calciferous Si = 6.00,Alrv = 2.00,seems to be about (ferrotschermakite)end member, Ca2(Mg, Fe'*)u ble Alvr at A necessaryrequisite for the formationof Al- AlvI2AlrvzSioOzz(OH,F)2 as defined by Winchell 1.40." rich calcium amphibolesis a highly aluminous en- (1945), has beendetected in nature.Even amphib- vironment. In addition, Alvr substitution is favored olesintermediate between tschermakite (ferretscher- by increasingpressure (Leake, 1965,I971'; Kostyuk makite) and pargasite (hastingsite), NaCa2(Mg, Sobolev,1969). The reliably analyzedamphibole Fe2*)aAlvlAlrvpsi6o22(OH,F)r, arerare. From l2l7 and most closelyapproaching the tschermakiteend-mem- Ca-amphiboleanalyses compiled by Leake (1968), is from a kyanite-garnet-biotite-hornblendegneiss not morethan 31 revealedSi < 6.25 (AlrY > 1.75) ber from Frodalera (Lukmanier area,Switzerland) which andAlvr > 1.00perhalf-unitcell(I = )2.75 Al), indicates a high pressure,high-Al metamorphic en- and only four of them gaveAlvr > L.30.Moreover, vironment. The analysisof this amphibole by Frey tlese four analyseswere discreditedby Leake ( 1968) (1969), slightlymodified by Leake (1971,Table 1), who assumedthat the extremely high AlvI contents yieldsSi, 6.22; Alr!,1.78;Alvr, 1.35;Na * K, 0.49; were due either to analvtical errors or to insufficient and Ca * Na * K,2.15. Logically,it would seem find high-AlYr-amphibolesin corundum- lPresent address: Mineralogisch-Petrographisches Insti- promisingto tut der Technischen Universitet Braunschweig D-3300 hornblenderocks, provided the pressureof metamor- Braunschweig,Germany. phism was not too low. However, none of the 721 722 T. E. BUNCH. AND M. OKRUSCH amphibolesfrom recently analyzedcorundum-bear- amountof pargasiteis estimatedat (5 vol percent. ing rock (Forestierand Lasnier,1969; Leake, 1971.) Exact data cannot be obtainedbecause of the irregu- revealedextraordinarily high alumina contents.The lar distribution of the amphibole and the small size most Al-rich amphibole within this group, from a of the specimens. corundum-zoisite-amphibolerock at Merkerstein(N. Physical Properties Arusha,Tanzania), is rather closeto the theoretical Optical and pargasitecomposition: Si, 6.05; Alrv, 1.95; AlvI, Megascopically,color of the amphiboleis some- 1.04; Na + K, 0.79, and Ca * Na + K, 2.60 what different in both specimens.Thus, using the (Leake,1971, Table 1). GSA-Rock-Color chart, amphibole AA-5 is mod- Amphiboles much richer in alumina were found erateyellow green(5 GY 6/4) andHA-l/2 is light as minor constituentsin a corundumand spinel-bear- olive brown (5 Y 6/6). For specimenHA-l/2, the ing marble, specimensof which were kindly provided hardnessequals 5* and thereis good {10O} to us by PrivatdozentDr. H. Bank of GebrtiderBank at 56o. In the latter specimen,chips of appropriate (Gem Cutting and Trade, Import and Export of thickness(about 0.5 mm) were obtainedthat show Gems and Minerals, at Idar-Oberstein,Germany). a faint but distinct pleochroism: y' (pale grayish The small rock sampleswere labeled"Pakistan," but orange: 10 YR 8/4) < c' (grayish yellow: 5 Y the exact locality of the material is unknown. Ac- 8/4).ln thin section,however, both amphibolesare cording to the microprobe data presentedin this nearly colorless.A summaryof optical and physical paper, these amphiboles are intermediate between propertiesis givenin Table 1. the pure pargasiteand tschermakitemembers; they The refractiveindices of amphiboleHA-l/2 were are distinguishedon a half-unit cell basis by low Si determined using mixtures of 1-mono-bromo-naph- (aborrt5.8), highAlrv (about2.2),high but variable thalene and butylcarbitol as immersion liquids and Alvr (rangingfrom l.l2 to 1.47),low to extremely filtered monochromaticlight (ur filter 589 nm). low Fe (from 0.07 to 0.006), and the absenceof For an exactorientation of a, B, andy, the amphibole minor elementsexcept for Ti. fragmentswere mountedonto a universal stage.The results agreewith a linear extrapolation of the best Description of Specimens fit curvesgiven by Deer et aI (1,963,Fig. 77) for The coarse-grainedmarble that containsthe am- tschermakitic amphiboles. In the pargasite-ferro- phibolesis white to light gray or even dark gray and hastingsitediagram of Deeret al (1963,Fig. 78), the composedmainly of xenoblasticcalcite grains. Red composition-refractive index curves are slightly corundum and pink, violet, or blue spinel are the straightenedby our data. most outstandingabundant accessory minerals. Mi- fn contrast,the optic axial angleZVr),whichvar- nor silicatephases include phlogopite and margarite, ies between50o and 55o, appearsto be rather low corundophyllite,Al-rich pargasite,anorthite, sphene, comparedto data of other known tschermakiticand and tourmaline. In addition, rutile and pyrite are pargasiticamphiboles (Deer et al, 1.963,Fig.77) al- also present.Detailed mineralogicalwork on this thoughsimilar to valuesgiven by Trd'ger(1971). So interestingparagenesis is in progressand the results far, the lowest2V, = 56o was reportedby Serdiu- will be publishedelsewhere. chenkoin 1954(cit.Leake,1968, analysis 538) from In the amphibole-bearingspecimens, the follow- ing mineral assemblageswere recorded: Specimen Data of Al-Rich Amphiboles A A-5, Calcite-corundum-phlogopite-margarite TAsre 1. Opticat and Physical -corundophyllite-Al-rich pargasite-anorthite- speclDen: EA-U2 AA-5 AA-5 AA-5 craLn: L 2 3 sphene-pyrite; SpecimenH A- I / 2, Calcite-Corun- 1.636 r0.001 dum-phlogopite-Al-rich pargasite-anorthite- g 1.639 10.001 Y r.652 r0.001 rutile-pyrite. y - a (diff.) 0.016 !0.002 The amphibole prisms are mostly idioblastic with y - d (ret.)* 0.017 10.001 0.014 i0,001 dominant{110} and are surroundedby calcite,some 2Y., (obs) 540 !20 I 11' of them being intergrown with phlogopite and/or (calc) 55' 11' 50' margarite.The grainsize parallel to the b-axisranges Dlspersion v> r v>! v>r Ync 23' 13' 240 !3" from about 0.3 to 1 mm in specimenAA-5, and OAP (0r0) (010) (010) (010) from 1 to 6 mm in specimen}IA-l/2. The total * Bll retatdsti,on measu?qents, AL-RICH PARGASITE a pargasiteoccurring in a spinel-bearingmarble. This were analyzedwith an ARL-EMxelectron microprobe. amphibole has a lower Me[Fe ratio (Niggli mg - Mineral standards (three wet-chemically analyzed Mg/(Fe3- * Fe * Mn * Mg) = 0.86) than the ) close in composition to preliminary amphibolesHA-l/2 (mg = 0.98, 2v"(ob"y54o) and analyseswere used for final analyses.Corrections AA-5 (mg = 1'0O,2Vrro*r 51-53"). Thus a nega- to the data were made for drift, background, and tive correlationbetween 2V, and the Mg/Fe ratio as deadtime; further corrections were made only for indicatedby Trtiger (1.971,Fig 186-2) seemsto be Al2O3 content as other elementswere very similar stipulated. For both amphiboles,the 2V values di- in content to the standards. Measurementswere rectly determined on the universal stage (using madeat 15 kV acceleratingvoltage and 0.02 - 0.04 conoscopiccontrol) and calculatedfrom retardation pA sample current. Analytical accuracy, expressed measurements(universal stage, Berek compensator, as weight percentagesof the amounts of oxide rar filter) agreevery well with eachother (Table 1). presentare: SiO2,TiO2, FeO, CaO :t 0.9 percent; The sameholds true for birefringenceas determined MgO -t 2.0 percent;Na2O and KgO -f 2.5 percent; by the calculateddifference y - a (HA-I/Z) and AlrOr -r 3.2percentandF:t l0percent. by retardation measurements.In the latter case, Analyses are given in Table 2, togetherwith the thicknessof the section was obtained by measuring structural formula, for the average of amphiboles retardation of corundum in the same specimen AA-5 and }{A-l/z from each of the two samples. (AA-5). Since none of the amphibolegrains are Sum of the cations,based on (O, OH, F) = 24.00, twinned, the extinction angle y/c had to be deter- rangebetween 15.80 and 15.96. Theseamphiboles mined by measuringone or a pair of cleavageplanes may be adequately described by the structural on the universal stage,a method which is not very formual WrXzYsZsOrr(OH,O, F), whereW - Na, accuratealthough the resultscompare well with other K; X - Ca, Y = Mg, Al, Ti, Fe, and Z = Si, Al. tschermakiticand pargasitichornblendes (Deer et al, Water Determination. Sufficient material was 1963,Table42'5. available in specimenHA-I /2 to separatean am- phibole fragment of approximately 150 mg. The X-Ray Data sample,after being crushedto a grain size of about X-ray film data were obtained from amphibole 0.5 mm, was purifiedby handpickingand dissolving HA-l/2by a 1.1.4.59mm camerawith Ni-filteredCu the remaining calcite in acetic acid. The combined radiation. The samplewas identified as amphiboleby H2O was determinedby the method of Lindner and comparison of the d-spacingswith those listed for pargasiteand hornblendeon JcpDS-pDFcards S11- 493 and respectively.The d-spacingswere #9-434, TlsI,e 2, Electron Microprobe Analyses (in wt percent) initially indexed by analogy to the data and Structural Formulae of Al-Rich Amphiboles sincepargasite d-spacings on JcpDS-pDrcard f 11-

493 are not available above 3.35. Several dis- Structural Fomulae crepanciesin the comparisonled to the use of the Wt Percent (0, olr, F, CL = 24) EA-LIZ AA-5 HA-I AA-5 least-squaresunit cell refinementcomputer program slo2 40.7 41.1 si 5.75 1 5.741 z=8.o0 of Evans et aI (1963). This was done by keeping ,oIv z.z5 lz=8.0o 2,26/. A1203 2O.2 22.6 the indexed reflections fixed and allowing the pro- ^,vI 1.12 | r.471 gram to selectindices for the Tio2 2.54 1.53 Ti o'27 o'16 unindexedreflections. | r=s.oa | r=a.ss Feo 0,58 0.05 Fe 0.07 o.01 A total af 77 rcflections were distinguishedin the ',| I l4go L7.2 16,L 3.62 3,35 film. Refined X-ray data for amphibole HA-l/z CaO 13.1 13.3 Ca 1.99, 1 00 givecell parametersof a 9.849 i 0.002,b l7 .927 ! Na2O 2,44 2.OO Na o.6? -f | x=z.et ""* | x=2.69 0.005,c 5.308 0.004,B l}5"l2' I 3', o sin F K2O 0'95 0.91 K o.rrI o.to I 9.504, and cell volume 904.3 ! 0.6. Thesevalues + H2O * 1.11 (1.s) 1.05 1.40 correspond to those given for calcium monoclinic F 1.3 0.9 0.58 0.40 amphiboles(Deer et aI, 1963). c1 0.04 0.07 c1 o.o2 0.07 Subtotal 100.16 r00.06 0 = F,c1 0.55 0.40 Chemica,IAnalysis Total 99.60 99,66 Microprobe Analyses. A total of six amphibole * Hro- content in HA-L/z done by nethod of Lindner md Rudent grainsin sampleAA-5 and eight grains inHA-L/2 tTsas); uro* in AA-5 asswed to be N L,5 Dt ?ercent. T. E. BTJNCH, AND M. OKRUSCH

Rudert (1969) wherebythe sampleis thermallyde- 2.4 AA composedin a furnace and the vapor is transported Y PAROASITE a Vv TSCHERIA|(IIE gas automatic 2.O 'A by a stream of dry carrier into an .MAA i v E-p dead-stoptitration apparatus(Metrohm). Here, the tsva water is determinedby the iodometric method using -%t^a t.o v Av Karl Fischer's reagent. The standard deviation of ^/ this method is :t20 y.g, that is, in the present case, f"r14 o ",/ 1.2 / O i2 percentof the amountpresent. Whereas the bulk EDEilIIE O HOIIIBLEI{OE "/ of the water content was liberated at about 800- ,)/l 900oC, the samplehad to be heatedto 1300oCto A //l ,/ r trrsntrunr completely remove water. We have assumedHzO- ,' vTS0|{ERHA|(TT[-ftRRolsciltflrrltlE A PARGASITT- fTRROl|ASTIilGSIIT content of sample AA-5 to be higher compared to .4 o |{oflilBLiltDt view the dissimilar amounts of F in . EDEIIIIT HA-l/z in of + TREt0LtTE- f tRRotcTtr{0r.rft both amphiboles and lower Ti content in AA-5 (Table 2). Insufficientamounts of AA-5 material o .4 .8 1.2 1.6 2.O 2.4 2.4 were availableto perform a water analysis. tIal]o + Fe+3+ Ti) oloms Fro. 1. Plot of Alro us (AlvI * Fe* * Ti) in half unit Discussion cell of Al-rich Ca-amphiboles compared with different Ca- Leake (1968) has listed at least ten criteria that amphiboles from Deer et al (1963). must be fulfilled to sustain a "superior" analysis. Both amphiboleanalyses (Table 2) are consistent on M(2) site occupancypreference by concluding with those criteria (except that a water determina- that it is extremelydifficult to fill more than 50 per- tion was not possiblefor amphiboleAA-5 ). centof M(2) siteswith Alvr. The mostAl-rich calcic The distinction between amphibole AA-5 and amphiboleknown to Leake (1971), and stipulated HA-l /2 is small, and their proper classificationis a as being correctly analyzedaccording to his criteria, crucial question.In fact, both amphibolesare too containsAlvr in 67 percentof the M(2) sites;our high in Alrv (aiz, high AlzOs combinedwith SiOz) amphibole AA-5 would have an occupancy of 73 to considerthem as mixed crystalsstrictly intermedi- percentAl, if Alvr were restrictedentirely to M(2). ate betweenpure pargasiteand pure tschermakite This does not appear to be the case for pargasite (Figures l, 2). Thus, every recalculationof the amphiboles.The calculationsof Colville et al (1.966) present amphibolesin terms of end-membermole- yield,17.73 for b if M(2) is occupiedprincipally by culgsmust end up eitherwith excessfree aluminaor somehypothetical amphibole molecule oversaturated in A1. For instance,54.7 mole percent of amphibole z.a HYPOTHETICAL AA-5 and 57.1 mole percentof HA-l/2 could be AI- PARGASITE calculatedin terms of a hypotheticalAl-pargasite z - I\HYPoTHETTCAL ---_ molecule, (Na,K)CazMgAlvI2Alrv3SibO22(OH,F) 2, TSCHERMAKITE t nt16 which could be derived either by the substitution AA-5 \ MgSi = AlvIAlIv from pargasiteor by the substitu- tion Si <) yn4llv from tschermakite. HA-t/2^ I Despitethe highly aluminousnature of amphiboles PARGASITE HA-t/2 and AA-5, they still conformto the limiting AlIv/AlvI ratio of calcic amphibolesstatistically eval- uatedby Leake (1965, 1971) and expressedby the EDENITE. TREMOLITE\ diagonalsloping line in Figure2. A linear extrapola- tion leadsto the hypotheticalAl-pargasite molecule. 6.0 6.5 7.O 7.5 LO Si ?o ,4 z.o r.5 | .o .5 o hl]4 According to these data, an Alvr content )2 is Frc. 2. Plot of Alrv os AlYr in half unit cell of Al-rich Leake (1965) has even placed very unlikely and Ca-amphiboles and theoretical end-members. Diagonal line considerabledoubts as to the existenceof nearly pure indicates maximum possible Alvr content (Leake, 1965). tschermakitein nature. He has reemphasizedthis Linear extrapolation leads to the hypothetical Al-pargasite argumentrecently (1971) on the basis of studies molecule. AL.RICH PARGASITE 72s

Alvr (syntheticglaucophane II) and 18.O5if occu- lite) in which they crystallized.Total alkali content pied entirely by Mg (synthetic tremolite). If a strict in both rock samplesis probably low; no bulk analy- correlation between b and the mean radius of the ses have been made, but phlogopite is the only M(2) cations is stipulated and the influence of Na major alkali-bearing mineral. Nevertheless,amphi- and Ca on the length of the b axis can be neglected boles still took up sufficientalkali content to preclude (Colville et al, 1966, p. l74l), syntheticFe-free formationof pure tschermakite.Leake (1971) con- pargasite,with a theoreticaloccupancy of MgrAlr in cludes from amphibole analyses and theoretical M(2) shouldhave b between17.88 and 17.90.How- studiesthat maximumAlvr will occur in amphiboles ever, syntheticFe-free pargasitewas found to have that crystallized in highly aluminous environments b = 17.99by Colville et al (1966), probably indi- with moderateor low alkali contentsand under high catingpartial occupancyof M(2) by Alvr. Our sam- pressures,so that amphibolesin corundum-bearing ple has b = 17.93. If we plot this value on the assemblagesthat crystallizedunder lower pressure diagram of Colville et al (1966) that relateslength conditionswould contain less Alvr than amphiboles of b axisto radius of the M2 ion and if we usea linear in kyanite-garnet-bearingassemblages of higher pres- interpolation between synthetic glaucophaneII and sureconditions. In addition,metamorphic amphiboles synthetic tremolite, we arrive at a mean radius of tendto be more aluminousthan igneousamphiboles, M(2) ions of 0.60 A or 40 percent Al and 60 per- which indicatesthat lower temperaturesare also an cent Mg. This meansthat 0.80 AlvI per formula is important factor. This leads to a discussionof situatedin M(2) while the remainder(0.32) is dis- aluminousamphibole stability. tributed over the M(l) and M(3) sites of amphi- Pure pargasiteseems to be the amphibolewith the bole HA-1/2 (assumingno influence of possible highest thermal stability. According to the experi- small amounts of Fe or Ti in M(2). In view mental data of Boyd (1959), the upper stability of the conclusionof Colville et al (1966) who con- limit is givenby the points 1040'C/1000 bars fluid sider a large contraction along b, as producedby the pressure,960"C/500 bars,and 850"C/250 bars.At presenceof small Al ions in M(2), to be extremely fluid pressureshigher than about 800 bars, pargasite difficult to attain, the definite preferencein HA-l /2 melts incongruentlyto yield Al-diopside * forsterite of Alvr for M(2) ratherthan M(1) or M(3) sites * spinel* liquid while,at pressuresbelow 880 bars, could indicate a high pressureenvironment for crys- the breakdownassemblage is Aldiopside * forsterite tallization. If low pressuresynthetic pargasite has * nepheline+ anorthite * spinel.Tentative experi- lessAlvr in M(2) than HA-1/2, then it seemsrea- mentsby Gilbert (1969) indicate a negativeslope sonableto assumethat AlvI was stabilizedin M(2) of the upper stability curve at high fluid pressures: through high pressure.This situation may be analo- pargasiteis still stable at 20 kilobars/800-900'C, gousto experimentalstudies on glaucophanewhere but becomesunstable at fluid pressuresclose to 30 the high pressureform containsmore Al\r in M(2) kilobarsgiving way to the breakdownassemblages of than the low pressureform (Ernst, 1963). In addi- "sheet silicate" * clinopyroxene * garnet * glass tion, the very high amount of AlzOo in HA-l /2 and at 800oCand forsterite * clinopyroxene* garnet* especiallyin AA-5 may be due, in part, to the glassat 900-950'C. As shownby Boyd (1959), extremelylow Fe contents.Amphibole AA-5 con- pure pargasitecannot exist in the presenceof quartz. tains the lowest amount of Fe and the highestNiggli The stabilityof pure tschermakiteis still open to value [Mgl(Fe3-*Fe*Mn*Mg) = 0.999] of any discussion.Boyd (1954) attemptedits synthesisal- known calciumamphibole. fn contrast,sample HA- though he was not sure that the amphibolecom- | /2 has a somewhatlower alumina content and is positionobtained was stoichiometric. By startingwith enrichedin Fe by a factorof 10. a glass of tschermakitebulk composition,Gilbert (1969) was able to synthesizean amphiboletenta- Conclusions tively identifiedas tschermakiteat 800'C/10 kilo- After considerationof the many parameters,we bars (uncorrectednominal fluid pressureof the chooseto refer to these two amphibolesas Al-rich piston cylinder apparatus) with a yield of about pargasite or Al-pargasite. The high Al-content of 85-90 percent together with minor amounts of theseamphiboles is not too surprisingin view of the garnet.Contrasting results were publishedby Jas- highly aluminousnature of the environment(corun- mund and Schiifer (1972) who worked on the dum-spinel-margarite-anorthite-corundophyl- tremolite-tschermakitejoin at fluid pressuresof 1, 2, 726 T. E, BUNCH, AND M. OKRUSCH

3, and 10 kilobars.The 10 kilobar runs were carried Relationship between all parameters and chemical com- out by H. S. Yoder in his internal heated hydro- positions of monoclinic amphiboles.Amer. Mineral. 51, r727-r754. thermal apparatus. A gel of tschermakite bulk com- Dr,sn, W. A., R. A. HowIE, eNo J. Zusslrm (1963) Rock- position was crystallizedat 800"C/10 kilobars to a forming Minerals. Vol. 2, Chain Silicates. Longmans, mixture of amphibole*, * chlorite * anorthite. Jas- London, 379 pp. mund and Schiifer (1972) report that, in the entire EnNsr, W. G. (1963) Polymorphism in alkali amphiboles. pressure range of 1 to 10 kilobars, the limiting A mer. Mineral. 48, 241-260. EvlNs, H. T., D. E. AIILEMAN, eNo D. S. HlNowrmrn amphibole composition is tremolitels tscher- (1963) The least-squaresrefinement of crystal unit cells makite,;5:L 4 mole percent. Bulk compositions richer with powder diffraction data by an automatic computer in Al than this limit yield amphibole ** f anorthite * indexing method (abstr.). Amer. Crystallogr. Ass. Annu. chlorite or amphibole", * anorthite + forsterite * Meet.Program,4243, March, 1963. enstatite(depending on 7 and P11). Fonrsrren, F. H., lNo B. LesNIpn (1969) Ddcouverte de niveaux d'amphibolites d pargasite, anorthite, corindon, No experimental data are available for the solid et saphirine dans les schistescristallins de la va1l6eHaut- solution series pargasite-tschermakite which, of Alliers. Contrib. Mineral. Petrology, 23, 19'4-235. course, would be of special interest for our am- Fnr.v, MenrrN (1969) Die Metamorphosedes Keupersvom phiboles. Moreover, an experimental investigation of Tafeljura bis zum Lukmanier-Gebiet. Beitr. zur Geol. (N.F.), the stability of pargasite-tschermakiteamphiboles in Karte der Schweiz 137,1-161. Gnosr, Susnlre (1965) A schemeof cation distribution in the presence of a CO:-H9O fluid phase would be the amphiboles.Mineral. Mag. 35,46-54. highly desirable. GILBERT,M. C. (1969) Reconnaissancestudy of the stabil- In conclusion, the present data concur mostly ity of amphiboles at high pressure.Carnegie Inst. Wash- with Leake's contention that highly aluminous am- ington Y ear Book, 67, 167-170. (1972) phiboles must form in metamorphic crystallization Jesuuwo, K., eNo R. Scnirnn ExperimentelleBes- timmung P-T-Stabilitiitsbereichein der Mischkristallreihe environments of high aluminum and low con- alkali Tremolit-Tschermakit. Contrib. Mineral. Petrology, 34, tents, and moderate to high pressures;temperatures 101-115. are probably moderate to high, although a complete Kosrvur, E. A., rNn V. S. Soeouv (19'69) Paragenetic study of the paragenesisof more samplesis necessary types of calciferous amphiboles of metamorphic rocks. to ascertain if margarite is primary or secondary Lithos, 2, 67-81. Lrexr, B. E. (1965) The relationship between tetrahedral and if calcite anorthite corundum is the t * maior aluminum and the maximum possible octohedral alumi- limiting assemblage. num in natural calciferousand subcalciferousamphiboles. Amer. Mineral. 50, 843-851. Acknowledgments (1968) A catalog of analyzed calciferous and sub- We thank H. Bank (Idar-Oberstein)for providingthe calciferous amphiboles together with their nomenclature samples,H. Seck (Kciln) for stimulating discussions, and associatedminerals. Geol. Soc. Anter. Spec.Pap.98, B. Leakefor suggestionsand criticalcomments, V. Rudert 210 p. and Mrs. U. Meisenburgfor advice and help with the (1971) On aluminous and edenitic hornblendes. water determination,and Mrs. J. Etheridgefor assistance Mineral. Mag. 38, 389407. in the microprobedata reduction.One of us (T. E. B.) LrNoNEn, B.,,rNo V. Ruornr (1969) Eine verbesserte thanksthe Mineralogisch-PetrographischesInstitut der Uni- Methodo zur Bestimmung des gebundenen Wassers in versitdtzu Kciln for use of their facilitiesdurine a Guest Gesteinen,Mineralen und anderen Festktirpern.Z. Anal. Professortenure. Chem. 24E.21-24. Tniicrn, W. E. (1971) Optische Bestimmung der gesteins- bildenden Minerale. Band 1, Bestimmungstabellen.4. References Aufl . Stuttgart (Schweizerbart). Bovn,F. R. (1954)Amphiboles. Carnegie Inst. Washington Wrxcur,rr-, A. N. (1945) Variations in composition and YearBook,53, 108-111. properties of the calciferous amphibo es. Amer. Mineral. (1959) Hydrothermalinvestigation of amphiboles. !0,27-50. In, Ph. H. Abelson,Ed., Researchesin Geochemistry. NewYork (Wiley),pp.377-196. Manuscript receiued,September 6, 1972; CorvIlre, P. A., W. G. EnNsr,eNl M. C. Grlnrnr (1966) acceptedlor publication, March 29', 1973.