American Mineralogist, Volume 65, pages 26-36, 1980

Some metasomaticcalc-magnesian silicate rocks from Connemara'western lreland: mineralogicalcontrol of rock composition

BBnNeRn E. LsarP

Department of Geology, University of Glasgow Glasgow, Gl2 8QQ, Scotland

Abstract

Chemical analysesfor 33 major oxidesand minor elementsare reported for a seriesof 26 associatedmetasediments that range from staurolite-garnet through similar pelites with graphiteto schistswith varying contentsof actinolitic amphiboleand diopside-richcalc- magnesiansilicate rocks. The calc-magnesiansilicate rocks cannot be matched with any likely original sedimentand have probably formed by metasomatismof pelitesby processes similar to thosethat producerodingites. The presenceof graphite,gradation into pettes, and the positive correlationof Ti and Niggli al-alk support a part sedimentaryparentage for the calc-magnesianrocks. ca, Mg, cr, and Ni (maxima l3.7voCaO,24.2Vo MgO,247O ppm ct atd 620ppm Ni) were added from solutions derived from nearby ultramafic rocks serpentin- ized during static regional , which caused andalusite and cordierite growth in someof the associatedpelites. Although the original sedimentarycomposition and the compositionof the metasomattzitg fluids were important in determining the chemical composition of the metasomatic rocks, the stableamphibole and pyroxenemineralogy in the calc-magnesiansilicate rocks was no less important. Theserocks are an exampleof the partial mineralogicalcontrol of the composi- tion of metamorphicrocks. It is proposed that a characteristic feature found in many sedimentary rocks, but not in ig- neousrocks, is a marked positive correlation of Ti and Niggli al-alk.

Introduction in addition to calc-magnesianschists rich in acti- nolitic amphibole and diopsidic pyroxene.This suc' In the succession south of Cornamona, cession(Fig. l) lies north of a thin staurolite-garnet- Co. Galway, Ireland (Fig. l), is a series of actinolitic two mica pelite that itself lies north of a coarsepeb- amphibole-diopside-biotite or phlogopite-plagio- bly grit. The whole area has been mapped in detail clase-quartz-graphite schists that grade into graph- on the scaleof I : 10,560with every outcrop exam- itic pelites with staurolite and garnet. This paper tab- ined and recorded;Figure I is a simplified geological ulates the chemistry of these rocks and tries to map. The calc-magnesianrocks are interlayed with deduce the origin of the amphibole- and pyroxene- graphitic schists and have transitional boundaries bearing rocks. The study was initially undertaken to both acrossstrike and along strike. The schistosityis determine whether the present rocks were derived by poorer in the calc-magnesianrocks, due to late rather metamorphism of dolomitic pelites or whether their random growth of actinolitic amphibole and clinopy- origin involved basic or ultrabasic igneous material roxene.Some ofthe rocks have a bandedappearanoe mixed with sedimentary matter. The results obtained due to differential weathering,but the contactsbe- point to the operation of metasomatic as well as orig- tween the bands are almost always gradational (e.g. inal sedimentary processes,the rocks being produced samplesBL36l l-7). Clinopyroxene-richclots some- by the interaction of solutions, similar to those that times occur in amphibole-rich (e'9. BL36l7), produce rodingites, with graphitic pelites. and patchesof garnet-rich rock (e.9. BL3623) occur The samples come from a graphitic schist series occasionally. Although ultramafic lensesoccur in the that contains some serpentine and ultramafic lenses, area, these are always distinct with sharp contacts. rare thin calcite , and thin banded quartzites, There is no gradation into the studied rocks, nor do 0003-004x/80/0l 02-0026$02.00 26 LEAKE: CALC-MAGNESIAN SILICATE ROCKS n

AREA STUOIED

o"w

Fig' l' Geological map of part of the area south of Cornamona, Co. Galway, Ireland showing the sample locations. Mappcd by the author.

the calc-magnesianrocks occur in flattenedor unflat_ of pelites (defined by over 100 Connemara pelite tened tectonic lenses. analysesgiven in Senior and Leake, 1978),in which representativeConnemara pelites plot towards a composition that lies between magnesiteand dolo- mite. These plots show that neither pure limg5ge1. (Niggli c 100)nor pure magnesite(c zero,Niggfi mg 1.0) was involved as a component of the original kyanite pre-date a strong static metamorphism that sedimentary mixture. If carbonate-pelite mixtures producedcm-sized cordierite and andalusiteporphy_ were involved, then from the c-mg plot of Figure 2 roblastsin a patchy distribution in the pelites.It was they must have been roughly equal proportions of during this post-F. metamorphismthat the amphi_ dolomite and magnesite mixed with variable bole- and pyroxene-bearing rocks crystallized, for amountsof pelite. Such a sedimentwould have been theseminerals have random arrangement. a most curious one, in that the proportion of dolo- Geochemistry mite to magnesitewas rather invariable and also be- causeprimary magnesiteis Carbonate -pelite mix tures not at all usual in sedi- mentary rocks; indeed, Johannes (1969) supposes Figures 2 and 3 show that the analyzed rocks (Ta_ that magnesitealrnost always originates by metaso- bles I and 2) plot in a cluster extending from the field matism. 28 LEAKE: CALC-MAGNESIAN SILICATE ROCKS

Moreover, most of the samplescontain as much

SiOr, or even more, than typical Connemarapelites [!v ^ Published Connemoro which have less than 5Vototal CaO + MgO (Fig. a). v pelite dnolyses. Thus averageConnemara pelite tlas 55.39VoSiO, and + llo omphibole or Pyroxene -- Actrnolitic or tremolrtrc v omphibole wilh diopsrde . Actrnolile rich [-?v ^ Publishcd Connemoro " pelite omlyses. + No omphibote or pyroxere E ra Aciinotitic or ir.molilic dmphibole wilh dioDsidc . Actinolite tremotiie rich

a

6.p. ++ ? o a!) o 6)^-R -9 Ql -Tremolrte

0 c

Fig. 3. Plot of Niggli al-alk against c for the analyzed rocks plus published Connemara pelites.

4.26VoCaO + MgO (Senior and Leake' 1978), o whereas the present samples (Table l) average +! E 6O.08VoSiOr, or if only thosesamples with more than l}Vo CaO + MgO are considered,then the SiO, aver- ages5495Vo and the CaO + MgO 22-34Vo.This is an insignificant silica change despite lSVomorc CaO + + 4, T pelite. Even if the SiOuhad tp oHJ MgO than in an average " cr, +,+ to complete decarbonation,no + been increaseddue mixture of carbonate with pelite can be as rich or richer in SiO, either before or after decarbonation than the same pelite alone, because of the residual CaO * MgO content in the mixture. the CaOlMgO ratio againstSiO, (Fig. 4) suggestsvariation between pelite, tremolite, and diopside,not pelite and carbon- ate. Complete decarbonation is not an attractive the- ory becausenearby stable calcite is common in the area (Fig. 1) and decarbonationis not signifi- Actinolite 9.'6 cant. with 70'l.Mg against the carbonate-pelite mix- 30'/. Fe #: lrremot,te Further evidence a the trace elements ..O a ture hypothesis is obtained from Ni and Cr, both of which are present in sedimentary + carbonatedeposits in very low amountswhereas pe- +, lites comnonly contain much more Ni and Cr. Mix- ,oB di -o g f progres- l+o tures of pelite and carbonate should have 0t sively lower contents of Cr and Ni as the carbonate 0 content increases.Figure 2 showsthat this does not m9 occur. The averageNi and Cr contentsof the eleven : + MnO Fig. 2. Plots of Niggli mB Mol MgO/(MgO * FeO samplesthat are amphibole- and pyroxene-freeare 2 FerOr) against Ni and Cr in ppm and Niggli c. Published + ppm comparedwith 166ppm Connemarapelite analysesare taken from Leake (1958), Evans 88 ppm Ni and 128 Cr, (1964),and Cruse and Leake (1968)in this and subsequentplots. Ni and 688 ppm Cr for the averageof the 16 amphi- LEAKE: CALC-MAGNESIAN SILICATE ROCKS 29

Table l. Chemical analyses of amphibole- and pyroxene-free Explanation to Tables I and 2 metasediments

BIJ511 eraphitic qurtz schist with albite dd Kf po4)hyroblasts ad - phlogopite, 3r t612L t61t J615 36j6 3621 1622 1621 1625 j612 1631 a little sulphide ed actiaolitic mphibole. 100 yeds SSW of Sench ylatk 14.4 that is on a bridge on the Co@orclerLusk bomdry. BL1612A E@ttz-tuscovite-biotite graphite schist uith rre soalI ganets md relic biotite crystalloids lying acloss the sclristosity. Iocalion ae 1511, BLt612B As 3612L b[t with ephibole ed pJEoxene clots, 8L1611 Albite, with q@rtz irclBions in a relict S., @covite- biotite-q@rtz-rde stauolite ed to@1in6 schist. Location as ,611. BLt61 4 Ewtz-actinolitic @phibole-diopeide-ph-loeopite-g?apllite- eufphide schi6t. As J611. BLr515 Biotite-nuscovlte schist with qrJatz folia, oligoclase over- growing puckeretl graphite trails, ecl rae gamet mal tomaline. l,ocation as ,611 . 81,1616 Phlogopite-qErtz-feldspd-gTaphite schist {ith a little prehlite in biotite ad rde K-Cveins. As ,511. 31,)61J Diopside-rich rock with coopletely sericj.tised feldspd, a clppo 117 110 224 176 1|14 fa pa 94 165 150 little act-inof itic @phibole, qMrtz, leucoxeLised sphene, 348 graphite md su-lphide. As J5'11 . 96 8L161 A qDil tu -tich-@phibole-sericiti sed f eldspar sclris t with a 41 little sulphide ed biotite. 70 tBrds SSWof the first o of Con@ona (tomlmd) as @itten on sheet ,9, Co. Galway 147 1 t1A,55O. BLJ619 Diopside-actlnolitic @phibole phlogopite-q@rtz-Kf-selicite schist with graplrite, sulpldde ed sphene. 140 yads SSV of 117 the last a Cofreom (tomlad) 12 of as ritten on sheet J9, Co. Galway 1,10,160. 91 Nb151a12i1161|71|7:|4 16 BI1620 Diopside-actinofitic @phibof e ph-logopite-Icf-sericitised cs 14 15 16 14 15 16 io 14 16 17 feldspar schist uith graphite, su-lphide, qrEtz dd sphene. Ba 1BB5 1850 940 B2B 1180 814 6A1 TOa 2175 As 1619. r,a 49 51 36 27 i2 4t 41 41 4t 42 BLJ621 Biotite-nGcovite-qwtz-p1ag'ioclase schist vith a little ce 110 123 89 76 129 105 150 102 1|o4 91 game!, su]ptdde ad eraphite. As J519. Pb 4t 37 41 17 37 46 69 18 57 a4 tL3622 Bioti,t e-qErtz-oligoclase-serj. c i tis ed. p1a€l oclas e-gmet rh211t1011171552Ai2 14 schist with grapbite, su-lphide, leucoxe[iBeal ihenite, green u8459127518 7 biotite ed ch-lorite. As ,619. 8L3623 qhltz-biotite-gamet-plagioclas e-sericite schis t with apatite ad ilneLite. As ,519. 81,1624 qaft,z-actinolitic ephibole-diopside-tr)h.1ogopite-Kf -sexicite with a little graphite ad sulphide. As J619. pyroxene-bearing BLl625 Siotite-nuscovite-qurtz-plagioclase-gamet schist ilith a bole- or samples,though there is a little stauolite. A.s 1619. wide rangeof values group. BLr626 ActinoTitic epldbole diopside scbist with d.ec@posing in this Thesevalues com- diopside. ,5 ydds east of the last h of Iarhitegoat Rock pare with the average ppm North as Eittm on sheet ,9, co. galway 1io'560, of 6 Ni and 15ppm Cr for 8L1627 A,ctinolitic ed @phjbole-diopside sctlist {ith a little 29 unsegregated siliceous dolomites chl-arite. As 1626. forming the 3Ir528A Actinolitic ephibole-rich scldst ilith sericitised feldspd green Connemaramarble (Leake et al., 1975). ad a little qErtz, na€netite @d graphite. As ,626. 8116283 sidilar to 362aA. The chemistry of the amphibole-pyroxene rocks is Bl,J629 Lc'Einofitic ephibole rock with lde phlogopite md sulphide. Ls 1619. therefore extremely difficult to explain in terms of 3lt6l0 Actinolj-tic @ptlibole altered clinopJEoxfle-Kf-sericite schist uith accessory sph4e, 180 yaals east of the last h possiblesedimentary parents. The possibility that the of lihitgoat Rock North as @lttd on sheet ,9' Co. Gal@y 1:10,560. failure of the silica to fall with increase of carbonate BL16t1 qp8tz-actinolitic ephibole sericite-gEaphite schist ili th was accessory su-lplLide' sphene, cl-loriter th.in Kf veins. As ,610. due to a coincidental increase in detrital quartz BL16t2 Bioti+, e-plagioclas e-q@rtz-Icf rich-graphite schis t with late with the increasein carbonate preh[ite dd some 6u-1phiale. 180 )Erds eaet of Sa1lo{ Bush is ruled out by the Cr isled. and Ni results. is extremelydeficient in these BL363t Blotit e-q@rtz-Kf-p1a€i oclase-graphite schist with accessoay Quartz sulphlde, prel4ite ed epidote. As ,612. elements,and very low Cr and Ni valuesshould then 3lr5r4 Dio?side-actinolltic @phibole seri ciie-(f-q@rlz-graptLi +'e schist with accessoly sufplide' sphene 4d calcite. ,r0 ,€rds be found in the amphibole-pyroxene locks, which is south of the last a of Com@o@ as rittq on sheet ,9' Co. Galway 1:10,560. not the case. BL1654L Ls t6t4 except rather noxe diopside. 3L16rB .ADphibole-qErtz-phlogopite-6 ericite-graphi te schist. r50 ydds south of the n of Co@ona as Eitt4 on sheet ,9' Co. Pelite-basic igneousmixtures ealilay 1 :10'560, BLt15, Actinolitic @phibole-diopside-q@tz-Kf-6ericite-eraphite An alternative possibility schist vith accessory sphene dd sulphide. By the tllird t of is that the amphibole- Whitegoat Rock North as Eitten on sheet l9r co. Galmy '1 and pyroxene-bearing rocks :10,560. were derived from a A11 ealj6es have been carieal out by n! eaLlEi's Eing the Eethod mixture of graphitic clay with basic volcanic of Lea,ke et a1, (1959) except fol Feo' E;0 dd C {hich were aletemineal ash or by wet nethods by A.f@p. Fe20J- i6 lotel iron as Fe2or. detrital fragmentsof a basic igneousrock. However, for this to be so, the original sediments should have containedeither somelayers or rock fragmentsmade tent would have been hsignifica1t. Such layers or almost entirely of augite with orthopyroxene (olivine fragments, whether tectonic or detrital, are absent. is excludedby the silica content of the rocks;Fig. 6), Such an ultramafic volcanic ashor detrital ultramafic becauseanalyses 8L3626 to 8L3629 (Table 2) have material is so unusual as to be implausible.The clus- very low Niggli al-alk values (Fig. 3), and very low tering of so many analyses so closely around the Al,O,(5.19to 2.l3%oALO,); i.e. the plagioclasecon- composition of actinolitic amphibole (Fig. 2) requires 30 LEAKE: CALC.MAGNESIAN SILICATE ROCKS

Table 2. Chemical analysesofamphibole- and pyroxene-bearingrocks

BL- J611 16128 1614 3617 16'tB 1519 3620 3624 1626 1627 1528A 152e8 1629 3630 3611 1614 36141 3118 3151 6t'55 56'76 Si02 80.18 51.55 59.55 'g.so'4.86 76,o1'i.ol 55.22 53.8e 58.11 '7.eo->.ig53.14 51.11 'l,gt51.45 55.81 52.35 52,o8 55'65 59''to 59'76 6.16 io.[o io.go ii.ts i1.os l,et 1.o2 2'1t 11'21 12'2, 1J'o5 11'9611'22 11'9'1 A12Oi 0.41 o'4't o'57 0"62 TiO2 o.t6 0.41 0.47 o.4o 0.60 o.4J o"49 o.44 ir.oz o.rr o.oo o.o2 0.16 o.44 o.53 Fe2O4 1.62 o.22 1.90 0.41 1.92 1.32 1.23 2.O3 1.07 tr'e0 1.oo 5.14 1.42 4.ro 1.45 1.51 3.97 2'9o 6 22 lleo 2.O7 11.27 7.22 11"79 4.O2 9.91 1O.2O 8,86 19"551 CaO 0.91 10.60 6,55 12.02 1.68 9'81 9.O1 7.AO 11.211 Na20 1.O1 1.18 4,tO 1.+1 1.04 O.41 O.57 O.55 0.1+ K2o 3.67 2.5+ 3.p 2,78 1.1o 4.9e 4.57 4.80 o.24 l.tlo o.o1 0.18 o.35 0.18 0.09 0.07 0.10 o.42 o.t1 PzOc 0.28 O.20 0.20 O.2O O.12 0.18 O.13 O.39 0.05 EZOr 1.16 j.9O 1.44 2JA 1,55 1.05 2,3+ 2.A6 2.34 o.27 O.06 O.4't O.12 O.02 O.26 0.l0 O.16 o.01 i.ie o:j6 0.06 o.1B o.oo 1.3+ o.8o a.24 o.oo o.oo 0.50 o.oo o'oo 0'00 o'14 o'51 o'40 c 1OO.49 1OO,501OO.OO 1OO.2O 1oo,3t 1OO.B' -i.i'e 1OO.11- 100.64 100.57 roo,15 99.75' too.io too.14 99.771o1.O1 100.21 1OO.2B 99.9798.47 Total 8.29 6'10 4'98 Fe2olo 2.73 j,93 5.7o >.ro i.zs 5.64 5.i6 a.oo 7.sa 8.62 7.67 5'55 5'o1 11' 116 166 169 85 14e 165 146 9t 17' 111 97 188 115 187 ClppB 275 25o v 215 409 274 74 198 278 108 184 14A 51 123 t14 90 115 e7 Cr 80 46 78 51 1420 1381 2+1A 22'12 2470 59 79 71 54 52 7t )69 50 Nt 20 4t 47 85 30 31 29 162 317 468 620 446 19 66 34 79 29 28 7o 71 6+ 20 61 52 21 55 17 97 109 1t2 ej 110 1O9 118 125 1O4 84 105 97 85 75 85 Zn 53 131 76 12 15 16 19 1+ 14 15 18 18 4 761 6 16 15 11 't15 lro 129 Rb 117 114 140 86 173 185 172 10 111 1 151 15'l 217 27O 278 Sr 170 270 259 2JO 22O 29 31 22 15 21 237 545 3+6 313 18 18 14 T 15 17 16 16 19 21 16162 124 1 20 Zt 22 49 21o 65 41260 152 77 46 44 55 +O 11 12 11 10 Nb 9 10 11 91211 11 B8 77 qg 2L Rq Cs 77 9 99 881 '1297 2710 2845 78 22244 17t+ 1840 1'15' 1527 866 1500 771 42O t2O4 24 21 621 20 27 23 24 5 21 24 52 55 58 18 61 55 55 75 52 49 44 16 19 48 41 41 56 tq 29 12 16 ?b 29 15 34 58 62 41 26 4t 52 17 62 47 17 14 99 Th 67 11 7911 5127 22t 7 9 5 14 )I U 178 11 10 7 7 14101 12 t

a regularity in the ratio of augite to orthopyroxene plagioclase was not important in the original sedi- that is beyond credibility in a tuffaceousor detrital ment, even for those samples with Niggli al-alk ex- rock. ceeding 5. If plots are made of many pelitic and There is, moreover, independent evidence that semipelitic series, characteristic features differentiate

To Colcite Key ^ Published Connemoro v pelrle onOlyses.

+ No omphibole or pyroxsc

6 Actinolitic tr. lremolilic. - omDhtbole wrth drops|oe o 15 . Actinolite tremolite rich o 10 .s CoO% a 090 5 a ++ "w- i 0'80 CoO E.o +0.7o a Mgo o @. % o'eo 30

050 25 CoO 0.10 +20 Mso O%li

r0 ,+ ,a f 0r0 d oit'+f MogEite OMinerols Quortz o 0 /.0 60 70 r00 60 Si O2wt% %Si02

Fig. 4. Plots (weight 7o) of SiO2 against CaOlMgO and CaO. LEAKE: CALC.MAGNESIAN SILICATE ROCKS

A c

o 50

40 of; aa l.' o %++e" ' .l +o a n ^ 30 ^a IcP o 20 +'. 69v t0 + . AngtGs.y + o BuinhWclts fo + Llondcilo 0

50 &v ^ Publihd conn.mro B v patil. omly*3 D o +bohphibot.d O

/..0 6)&tinolriic o. li.molitc vomohiDola wilh dioD!i&. -R O a Acitnolitc rcmotrt. O ^ + ojoo 3.0 30 tl ++o^oo ov o o o . ,t+o oo 2.0 aaT . r+o* o 20 P.. 10 f oro- oo ) og@ f>o 0 l0 20 40 50 5 l0 15 20 25 ol-otk30 or-olk Fig' 5' Plots of Niggli ti against al-alk. (A) Three seriesof ordovician sedimentsfrom Wales (Bjdrlykke, lgzl). (B) Cornamona and other Connemara analyses.(C) Karroo dolerites (Walker and Poldervaart, 1949).(D) Western part of tni Galway Granite; analysesfrom wright (1964)' Claxton (1971),Aucott (1966), and Leake (1974).Note scalechange of al-alk from A and B isediments)to C and D (igneous rocks).

these rocks in general from rocks of igneous origin, lykke, l97l). Clearly, the dominant sourceof the Ti becausethe sedimentshave a negativecorrelation of and the excessal-alk lies in the clay mineral content al-alk with c (Leake, 1969)and a positive correlation of the sediments. Consequently plots of al-alk of al-aft with ti. The latter indicates that the Ti is against ti for igneous series,both acid and basic as mainly in the clay minerals (e.g.Lange,1970; Senior exemplified by the Galway Granite and Karroo and Leake, 1978),of which al-alk is a good quan- Dolerites (Fig. 5C, D), contrast with typical sedi- titative measure.If plagioclasebecomes a major con- ments in showing very low degreesof correlation of ti tributor to the sedinent, al-alk increasesespecially if and al-alk. Increase in al-alk in igneous rocks is calcic plagioclaseis involved, but Ti should not in- mainly controlled by the contents of calcic plagio- crease.Thus Figure 5A showsthat for three seriesof clase, modified by the presence and abundance of Welsh Ordovician sedimentsvarying from grits, ashy biotite. shales,and mudstonesto calcareousshales, a marked The presentsamples (Fig. 58) give a marked posi- correlation of ti and al-alk exists even thoush some tive correlation of ti and al-alk, revealing a quite un- 'sedimentary of the sedimentscontain somedetrital feldspir (Bj6r- mistakable trend confirmedby the very LEAKE: CALC-MAGNESIAN SILICATE ROCKS high al-alk values which pelitic sediments display, values far in excessof those found in typical igneous rocks. Accordingly, probably the presentrocks were orig- 25 inally sediments,as is also indicated by their graphite content and by the field evidence of continuous gra- 20 dation into graphitic pelites.The view that the diop- and actinolitic amphibole-bearingrocks were side "no'u originally mixtures of graphitic pelite and volcanic ash or detrital or tectonically emplaced basic igneous 10 material is rejected. To Colciic (CoO=56%) Becauseof the incompatibility of the geochemistry 0 5r01520253035 with isochemicalmetamorphism, the compositionof I %Co0 the samplesmight be the result of metasomatismof [9v original carbonaceouspelites. In the calc-magne- o Published Connemoro pelite onolys.s sium-rich rocks pyroxene and amphibole dominate + No omphrbole d Pyrorene ^-r Actirclrtic or treNlitrc the compositionand it is possiblethat their metaso- omphibolc wrth diopside rrch matic growth controlled or at least greatly influenced . Aclinolite tremolite the rock composition. fm Figure 6 supports this interpretation, because many of the pyroxene-bearingsamples plot nearly on a line that if extendedwould passthrough either pure diopside or dolomite. But from Figure 4 it has al- ready been deducedthat dolomite cannot have been 200 250 300 350 400 a major constituentof the variation, so that Figures4 sl and 5 are best explained by the rock compositions Fig. 6. Plots (weight Vo) of CaO against MgO and Niggli fm fsing controlled by three constituents:pelite, diop- against si for the analyzed samples and published Connemara side, and an amphibole whosecomposition is a little pelites. less magnesian than tremolite, l.e. actinolite. Note that mixtures of dolomite and magnesitewould form bole- and pyroxene-bearingsamples show a very a broad band acrossthe periphery ofthe plot, and yet wide and erratic variation, with much higher values the analyzed samplesfall in a narrow segnent whose for someof the samples(BL36288has2272 ppm Cr extremesare defined by pyroxeneand amphibole. and 620 ppm Ni) than any normal sediment-a typi- This interpretation is also supportedby Figure 6, cal pelite has about 100 ppm Cr and 70 ppm Ni in which the analyzed rocks show decreasing Niggli (Taylor, 1965,p. 170),an order of magnitudeless. It si down to that found in actinolite and diopside but is not known what controls the Cr and Ni contents, not extending to lower values approaching the car- but Figure 2 shows that the sampleswith the highest bonateson the left-hand side of the figure. If it is ac- Cr values also have the highestNi values,and both cepted that the occuffence of pelites with layers of elementscould plausibly have been derived from ul- d€trital actinolite and diopside is an unacceptable tramafic rock during serpentinization. No other possible sedirnent,then clearly these rocks have their sourceis evident. compositions controlled by the stable actinolitic am- Discussion phibole-pyroxeneassemblage. The most probable original rock, in view of the If the metasomaticinterpretation is sound then at bedded nature of the series, was a graphitic pelite least some elements can be expected to show inter- that has been soakedby Ca- and Mg-bearing solu- relationships that are the result of two quite di-fferent tions, presumably derived from the serpentinization processes-sedimentary on the one hand and me- of the associatedultramafic rock now present as ser- tasomatic on the other. This may be the explanation pentine- rocks. Figure 2 strongly supports this in- for Figures 7 and 8, in which some elementsshow terpretation, as the Ni and Cr values of the amphi- two rather different trends. The low Cr and Ni values LEAKE: CALC-MAGNESIAN SILICATE ROCKS 33

100 Ca, Mg, Ni, and Cr that readily enter the amphibole and pyroxene structures.The role of Sr is lessclear, 300 B as this elementcan enter amphibolesand pyroxenes o+ Sr in moderateamounts. The paucity of Sr may be a re- ++ pp.200 o flection of impoverishment in the metasomatizingso- lutions derived from the ultramafic rock. a factor that t00 could also be important for Rb, K and Ba, elements that are not abundantin ultramafic rocks.Rb, K, and 0 [!v Ba showa closecoherence and are richestin the sam- 3000 ^ Publishad Connemro " pclile onolyses. ples containing significant biotite and (or) K-feld- + No omphibole or py.orenc spar.These minerals provide sitesfor theseelements, 2500 (.) Actinolittc or tremolilic - omnrFre wlh dtopsrde which were presumablyoriginally presentlargely in . Aclinotite tremolite rich the pelites, or have been added by solutions which ?000 8o ow obtained them from nearby pelites. Some evidence ppm o oo supportsenrichment of Ba in somesamples, because ooo excluding the five samplesdepleted in Ba mentioned + €to above,the remaining 22 samplesaverage 1568 ppm + Ba, whereas Senior and Leake (1978) have shown that typical pelites in Connemaraaverage 781 ppm Ba. The whole processis an instance of the partial 0 mineralogicalcontrol of the chemicalcomposition of a metamorphicrock (Leake,1972). Fig. 7. Plot of tttt%oCaO against ppm Sr and Ba. Very low Sr and Ba in the most metasomatized samples suggest removal of If the aboveinterpretation is correct,material rich thes€ elements. Published analyses of Connemara pelites from in Ca, Mg, Ni, and Cr must have left the ultramafic Leake (1958) and Evans (1964). rocks, presumablyduring serpentinization,while the metamorphic grade was high enough for amphibole are similar to sediments,in which Cr and Ni are of- and pyroxene crystallizationelsewhere, so that low- ten containedin clay mineral structures;but the high grade minerals sometimesfound in rodingites, such Cr and Ni values are widely separatedand suggest as prehnite,did not form. The origin of the Ca is par- the operation of a different process iayslying the ticularly interesting. During the serpentinizationof massive addition of these elements. Likewise the ultrabasic and ultramafic rock, rodingites can form relationshipof Rb and K with al-alk is suggestiveof by loss of Ca from the ultramafic rock, even though one clustering dominated by presumably sedimen- the ultramafic rock may contain extremelylittle Ca, tary processesand another controlled by depletion. mostly in diopsidic augite (e.9.Challis, 1965).This is In Figure 7 the normal positiveassociation of Sr with becauseCa does not enter the structuresof serpen- Ca is evident in the normal pelites which plot from tine minerals. Ca is therefore removed in solution 04Vo CaO; the metasomaticrocks scatterand do not (Coleman and Keith, l97l) during the serpentiniza- follow the sametrend. tion of the pyroxene,resulting in serpentinitethat is In eachplot (Figs. 6,7, and 8), the samefive sam- commonly essentiallyCa-free. In total, the amounts ples, BL3626-9, show either depletion (Rb, K, Ba, of Ca removedduring serpentinizationshould gener- and Sr) or the most pronouncedenrichment (Ca, Mg, ally be lessthan the amount of Mg, as diopsideis not Ni, and Cr), indicating that the most pronounced usually abundant and Mg is also derived from the metasomatismis recorded in these samples,which breakdown of olivine as well as pyroxenes.The gen- are almost entirely actinolitic amphibole (82-99%) erally higher Mg than Ca contents in the present with up to l6Vopyroxene. These are the extremede- samplesagree with this. velopment of the putative metasomaticcalc-magne- Although it could be postulatedthat the Ca came sian silicaterocks whosecompositions are dominated from the calcitemarbles in the succession,this seems by a stablemineralogy of amphibole and pyroxenes. most untkely, as they show no evidenceof instabil- Theseminerals have rejectedthose elementssuch as ity. Calcite marbles are common in the Connemara Rb, K, and Ba which do not easily enter amphibole Schistand are not associatedwith calc-magnesiansil- and pyroxene, and acceptedthose elementssuch as icate rocks ofthe presenttype. The significantassoci- 34 LEAKE: CALC.MAGNESIAN SILICATE ROCKS

Key

^ Published Connemoro " pelitr onolys?s. + No mphibole or pyroxene or tremolitic e^ Actinotitic omphibole with diopside Zr . Actinoliic tremoliie rich + + ppm + +

o+ o* . o @@ .^ o

++

oo aoo o .t 20m |I 500 5 Oo-+o^oo-s - @ r-^ o r000 Kzo 4 t o o ? t* o J I -+ Cr 2 + ppm a I o o8oo%6ooo 0 + r00 o **t* I $t* t 30 ', +i "c * oo Go 20 pPm @.o .o6 o q l0 a a'+ .aO .o % 0 Ni ppm 250 + 't + o o og^ o 200 ^^ j +o m@ . + .*tgoo "; tto"^ %o Rb ,.^ a ***** ppm'-' Oo_Gr+o ) o*q * arf- o 100 ser a'O $

0510152025 350 5 1015202530354045 ol-olk ol-olk Fig. 8. Plots of al-alk against Cr, Ni, Ga, Rb and Zt ppm for the analyzed samplesand published Connemara pelites (from Eva$, 1964;Leake, 1958).

ation seemsto be with the serpentinizedultramafics. was less abundant. Water deficiency seems an un- SampleBL36l6 has 560 ppm Cr Lnd 425 ppm Ni likely explanationfor the presenceofpyroxene. and has clearly been enriched in these elements as The random arrangement of the amphibole and well as Mg, but has not been greatly enriched in Ca pyroxene in many of the rocks, and their relatively or depleted in K. This specimen is unusual in being late crystallization after the F, folds, indicates that the only analyzed sample that has been metasoma- these minerals probably grew during a late static tized rather strongly but without the formation of metamorphic phase that produced abundant andalu- amphibole or pyroxene, the key stable mineral in this site and some cordierite in patches in the associated sample being phlogopite. This may have been due to pelites. This late metamorphic phase was of lower insufficient Ca being available compared with the grade than the main regional metamorphism and was Mg. Speculatively, it is possible that the amount of characterized by considerable metasomatic move- Ca available might have determined the relative pro- ment, including substantialnovement of Al and Si portions of the pyroxene and amphibole, as diopside to give quartz-andalusite veins with much sericite contains about 25VoCaO but tremolite only about and some chlorite. These veins were accompaniedby ll%o CaO. so that tremolite would be favored if Ca massive movement of water through the metamor- LEAKE: CALC-MAGNESIAN SILICATE ROCKS phic rocks.Because serpentinization is unlikely to oc- the solutions. The rock compositionsmust be the cur above500'C (Scarfeand Wyllie, 1967),while the product of the interaction of the original composition co-existenceof andalusite and cordierite suggestsa (possibly with some ratios such as TilAl being little pressurebelow 4 kbar (Holdaway, l97l; Bird and changed), the metasomatizing solutions, and the Fawcett, 1973),these values give maximum limlls fs1 composition of the stable minerals in that environ- the conditions during the metasomatic formation of ment: an interacting trinity. Equally interesting is the the calc-magnesiansilicate rocks. deduction that the crystallization of the amphiboles Although the occurrence of the in and pyroxeneswas a consequenceof the instability lensesand pods meansthat the metasomatizedrocks of olivine and pyroxene elsewhereand that the com- could be, in the third dimension,even closer spatially position of the metasomatizing solutions was strongly to serpentinethan they appearto be on the map, nev- influgnssd by this instability: a partial mineralogical ertheless it is particularly interesting that the most control of the compositionof the solutions.The pin- metasomatizedrocks do not appear to envelop the pointing of the origin of the solutions has been facili- serpentinitesbut occur in the same zone at varying tated by the associationwith particular trace ele- distancesfrom them. This may explain some of the ments,and this feature may occur in other situations numerousinstances of serpentinitehaving apparently such as when cordierite, a mineral known to concen- no metasomatic effects on the contact rocks (e.g. trate Be, is pinitized. Whether the Be leavesthe rock Turner and Verhoogen,1960, p. 319), that has often will be determined largely by whether pinite or any led to the deductionthat no lossof Mg or Si occurred of the other alteration products forming at that time from the during its formation. If the op- can accorrmodateBe. If not, then the intergranular timum conditions for fixation of the rerroved mate- fluid will carry the Be away to deposit it eventually rial are not immediately adjoining the ultramafic elsewhere, perhaps in a concentrated form. Thus rock, a more favored site further away may be meta- mineral stabilitiesand instabilities in metamorphism somatized.This would not necessarilybe apparentif can be a partial control over the chemical composi- the examination doesnot range well outside the ser- tion of metamorphicrocks, whereas the reverseview pentinite envelopeor if calc-magnesiansilicate rocks has traditionally dominated,namely that the minsl- ofthe presenttype are not regardedas possiblybeing als in a metamorphic rock are entirely controlled by formed by metasomatism,a deduction only made by the conditions of metamorphism and the original the presentauthor when the chemicalevidence made compositionof the rock. a simple sedimentaryorigin untenable. Recent oxygen and hydrogen isotopestudies have Acknowledgments shown that much of the water involved in serpentini- I thank Mr. A. Kemp, the University of Bristol, for determining zation and other hydrothermal processes is often the FeO, H2O, and C contentsof tbe rocks.The X-ray spectrome- was purchased with a grant from the former DSIR, heated meteoric ground water (e.g. Taylor and For- ter now NERC; the experimental and field work were completed while the ester, l97l; Taylor, 1977), and that serpentinization author was in the Universitv of Bristol. is often a near-surfaceeffect. The presentresults with serpentinizationoccurring during the formation of References andalusite,cordierite, amphibole, and pyroxene in- dicatea metamorphicsituation with a moderatepres- Aucott, J. W. (1966)The Petrology,Structure and Geochemistryof sure.It would be extremelyvaluable to know if mete- the Galway Granite in the Crook Moithan Area, Connemara,Co. Galway, Eire. Ph.D. Thesis, University of Bristol. oric water penetratesto the depth at which these Bird, G. W. and J. J. Fawcett(1973) Stability relationsof Mg chlo- common metamorphic mins14l5form. According to rite-muscovite and quartz between 5 and l0 kb water pressure. Taylor and Forrester (1971,Fig. 5) meteoric ground J. Petfol.,14,415428. waterspenetrate at least3 km in depth, equivalentto Bjdrlykke, K. (1971) Petrology of Ordovician sediments from more than I kbar pressure. Wales..l[ors& geol. Tidsskr.,51, 123-139. Challis, G. A. (1965)High-temperature contact metamorphism at It is extremely unlikely that the solutions carrying the Red Hills ultramafic intrusion-Wairau Valley, New Zea- the Ca, Mg, and probable Si contained these ele- land. J. Petrol.,6,395419. ments in the proportions in which they occur in am- Claxton, C. W. (1971)The petrology, chemistryand structure of phibole or pyroxene(even allowing for the contribu- the Galway Granite in the Rosmuc area,Co. Galway. Proc. R Irish Acad., 7I B, 155-169. tion of the original composition of the rock), so that Coleman, R. G. and T. E. Keith (1971)A chemicalstudy of ser- the amphibole- or pyroxene-richcalc-magnesian sili- pentinization-Burro Mountain, California. J. Petrol., 12, 245- cate rocks do not preciselyreflect the compositionsof 310. 36 LEAKE: CALC-MAGNESIAN SILICATE ROCKS

Cruse, M. J. B. and B. E. Leake (1968) The geology of Renvyle, Scarfe, C. M. and P. J. Wyllie (1967) Serpentinedehydration Irishbofin and Irishshark, NW Connemar4 Co. Galway. Proc. curves and their bearing on serpentine deformation in oroge- R. Irish Acad.,678, l-36. nesis.Nature, 21 5, 945-946. Evans, B. W. (1964) Fracticnation of elements in the pelite horn- Senior,A. and B. E. Leake (1978)Regional metasomatism and the fels of the Cashel-Lough Wheelaun intrusioq Connemara, geochemistryof the Dalradian metasedimentsof Conn€mara, Efte. Geochim. Cosmochim. Acta,28, 127-156. westernIreland. J. Petrol., 19,585-625. Holdaway, M. J. (1971) Stability of andalusite and the aluminium Taylor, H. P., Jr. (1977)Water/rock interactionsand the origin of silicate phase diagram. Am. J. Sci., 271,97-131. H2O in granitic batholiths. J. Geol. Soc.Lond., 133,509-558. Johannes, W. (1969) An experimental investigation of the system - and R. W. Forester(1971) Low Ot8 igneousrocks from the MgO-SiO2-HrO-CO2. Am. J. Sci., 267, 1083-1104. intrusive complexesof Skye,Mull and Ardnamurchan,western Lange, J. (1970) Geochemische Untersuchungen an Sedimenten Scotland.L Petrol., 12, 465497. des Persischen Golfes. Contrib. Mineral. Petrol., 28,288-395. Taylor, S. R. (1965)The applicationof traceelement data to prob- Leake, B. E. (1958) Composition of pelites from Connemara, Co. lems in petrology. Phys. Chem. Earth, 6, 133-216. Galway, Ireland. Geol. Mag., 95,281-296. Turner, F. J. and J. Verhoogen(196O) Igneous and Metamorphic - (1969) The discrimination of ortho and para charnockitic Petrology.McGraw-Hill, New York. rocks, anorthosites and amphibolites. Indian Minerals, 10,89- Walker, F. and A. Poldervaart (1949) Karroo dolerites of the t0/.. Union of South Africa. Bull. Geol. Soc.Am., 60,591-706. - (1972) The mineralogical modification of the chemistry of Wright, P. C. (1964)The petrology,chemistry and structureof the metamorphic rocks.Geol. Mag., 109, 331-337. Galway Granite of the Carna area, County Galway. Proc. R. - (1974) The crystallization history and mechanism of em- Irish Acad., 638, 239-264. placement of the western part of the Galway Granite, Con- Yardley, B. W. D., B. E. Leake and C. M. Fanow (1980) The nemara,Western lreland. Mineral. Mag., 39,498-513. metamorphismof Fe-rich pelites from Connemara,Ireland. .L Petrol., 20, in press. and origin of the Connemaradolomitic marbles and ophical- cites.J. Petrol., 16,237-2'17. G. L. Hendry, A. Kemp, A. G. Plant, P. K. Harvey, J. R. Wilson, J. S. Coats.J. W. Aucott, T. Liinel and R. J. Howarth (1969)The chemicalanalysis of rock powdersby automatic X- Manuscript received, January 8, 1979; ray fluorescence.Chem. Geol., 5,7-86. acceptedfor publication, July 10, 1979.