petrogenesis of orenditic and kamafugitic rocks from central italy 45

Canadian MineralogistYol.26, pp.4565 (1988)

PETROGENESIS OF ORENDITIC AND KAMAFUGITIC ROCKS FROM CENTRAL ITALY

ANGELO PECCERILLOIstituto di Scienze della Terra, Uniyersitti di Masina, Via dei Yerdi 75, 98100 Messina, Italy

GIAMPIERO POLIDipartimento di scienze della Terra, univenitd di Firenze, via La Pira 4' 501M Fbenze' kaly

GIANCARLO SERRIDipartimento di Scienze della Terro, Universitd di Pisa, Via S. Maria 53, 561A0 Pisa, Italy

ABSTRAcT

In central Italy, ultrapotassic rocks having an orenditicor lamproitic and kamafugitic affinity are found associatedwith potassic (KS) and highly potassic (HKS) volcanic rocksof the Roman Province. Orenditic rocks (OREI$ are silica-oversaturated, intermediate in composition and have a high(71-80) value of Mg# [00 Mg/(Mg + Fd+ )], high K,/Na,and high abundances of incompatible elements, Cr and Ni.Kamafugitic rocks (KAM) are ultrabasic, strongly silica-undersaturated, and show the highest Ca(90), Me# (75-81)and K/Na in the Roman province; incompatible elementsare highly enriched; Ni and Cr range from 79 to 153 andfrom 40 to 830 ppm, respectively. Orenditic rocks havehigher Sr-isotope ratio (0.71256-0,71715) and lower abun-dance of Sr (N2-8n ppm) with respect to KAM (15r/863r0.7 1037 4.11120; 1724-37M ppmSr). Derivation of ORENfrom KAM or HKS maglras by assimilation of crustalmaterial, although consistent with Sr isotopic variations,is ruled out by the high Mg#, Ni and Cr of OREN, as wellas by a large number of other geochemical and petrologi-cal data. The genetic model proposed suggests that KAMand OREN were generated by melting at different depthsof a residual, phlogopite-bearing upper mantle enriched inLILE and radiogenic Sr. Enrichment was provided by addi-tion of liquids derived by melting of sediments carried downby subduction processes that were active under the Appen-nines during Tertiary times. The strong degree of silicaundersaturation and the very high Mg# exclude significantinteraction of KAM magmas with crustal rocks. Contami-nation of OREN magmawith significant amounts of crustalmaterial is possible only if parental magma had extremelyhigh Mg, Ni, and Cr, as found in kimberlites and some high-Mg lamproites which, however, are not observed in theItalian peninsula.

Keywords: rltrapotassic rocks, orendite, lamproite, madu-pite, kamafugitic rocks, lamprophyres, RomanProvince, Italy.

SOT4I\4AIRE

On trouve, dans la partie centrale de I'Italie, des rochesultrapotassiques ayant une affinit6 soit orenditique ou lam-proitique, $oit kamafugitique; elles sont associ6es aux hvepotassiques (KS) et trbs potasslques QIKS) de la provinceromaine. Les roches orenditiquas sont sursatur€es en silice,

interm€diaires en composition, relativement mapn€sienneslrapport l@Me/(Mg + F*+)lou Mg# (entre 7l et 80), eK/Na €lev6, et enrichies en K, Cr, Ni et €l€ments incom-patibles. Les roches kamafugitiqus (KAM) sont ultraba-siques et fortement sous-satur6es en silice; elles possldentles valeurs de Ca(Vo), Mg# (75-81) et K/Na les plus 6le-v6es de la province romaine. De plus, ces roches sont for-tement enrichies en 6l€ments incompatibles; les teneurs enNi et en Cr se situent entre 79 et 153 et entre 40 et 830 ppm,respectivement. La suite orenditique posslde un rapportdes isotopes de strontium plus 6lev6 (0.712564.71715) etune concentration de Sr plus faible (402-847 ppm) que lasuite KAM lazsr,zrcSr entre 0.71037 et 0,711?I; eatxe l7?Aet 3704 ppm Sr). On ne pourrait d€river la suite OnEN ipartir des magmas KAM et HKS par assimilationde mat6-riaux crustaux, malgr€ les variations des isotopes de Sr, vule Mg# et la teneur 6lev6e en Ni et en Cr de la suite OREN,et compte tenu d'un gtrand nombre de donndes gdochimi-ques et p€trologiques. Selon le modble pdtrogdn6tique pr6-f6r6, les magmas KAM et OREN ont etd formes par fusioni des profondeurs differente$ d'un manteau sup€rieur r6si-duel, mais porteur de phlogopite e't enrichi en 6l6mentsincompatibles i large rayon et en strontium radiogdoique.L'enrichissement serait d0 i des liquides form€e par lafusion de sddiments arrivds i ces profondeurs par proces-sus de subduction sous la chalne des Appenins i l'6poqueTertiaire. La forte soug-saturation en silice et le Mg# trbs6lev6 font penser que les magmas KAM n'auraient pas pu€tre modifi6s par des roches crustales. Une contaminationdesmagmas ORENpar des quantit&important€s de rochescrustales serait permise dans ls seul cas oir le mrdgma aurait'au d6part, des teneurs en Mg, Ni et Cr e$remes, commedans les kimberlites et quelques lamproites magn6siennes'roches qui n'ont toutefois pas 6t6 repdr€es dans la p6nin-sule italienne.

Clraduit par la R€daction)

Mots-cl6s: roches ultrapotassiques, orendite, lamproite'madupite, roches kamafugitiques, lamprophyres, pro-vince romaine, Italie.

INTRODUCTION

Ultrapotassic rocks from the Roman province, inCentral Italy, have been the subject of many p€tro-

logical, geochemical and isotopic investigations that

45

46 TI{E CANADIAN MINERALOGIST

FIc. l. Location map of orenditic and kamafugitic rocks in central Italy (asterisks).Black dots: South Tuscany crustal anatec'tic rocks; ruled area: Roman Comag-matic Province.

have contributed to a clarification of many of theirpetrogenetic and geodynamic aspests; see Pecceril-lo & Manetti (1985) for a review and references.Detailed investigations canied out on different sec-tors of the Roman province (e.g., Civetta 1981,Rogers et al. 1985) have shown that volcanic rocksdisplay signifrcant compositional differences acrossthe Italian peninsula, as already stated some time agoby Turi & Taylor (1976). Whether these differencesreflect heterogeneity in the source of magmas orresult from different degree of interaction of mag-mas with upper crust is still a matter of debate Cluriet al. 1986, Holm & Munksgaard 1986).

In order 1o shed light on these problems we haveundertaken a petrological and geochemical investi-gation of some of the least-studied ultrapotassicrocks of central ltaly, i.e., the melilitites from SanYenanzo and Cupaello (venanzite and coppaellite offormer authors, see Sahama 1974) and minettes(mica-rich lamprophyres, locally known as selagiti)from Montecatini val di Cesina and Orsiatico,southern Ttrscany (FlC. l). Our investigation was alsoextended to the Sisco lamprophyre (Corsica) which,although older, is similar in composition to Tuscanminettes. Despite their rarity, these rocks are rele-vant to an understanding of the potassic magmatismbecause l) they have the highest Mg#lMe/(Mg+Fd*; atomic ratiol and abundance offerromagnesian elements among Italian potassicrocks, and, thus, they may be the most likely candi-

dates to represent primary magmas; and 2) ttrey areexposed near the northernmost portion of the Ro-man province, i.e., in a zcrne where, according tosome autlrors Cfuri & Taylor 1976, Fercara et ol-1986, Turi et al., l98Q the potassic rocks show thestrongest effects of crustal contaminationl some ofthe investigated rocks are closely associated with theTuscan anatectic province (Fig. 1).

The data obtained in the present work also willbe used to address a number of other topics of cur-rent interest in the study of Italian potassic volcan-ism, such as l) the nature of the mantle contamina-tion event; 2) the geochemical evolution of the mantlesource prior to generation of potassic magma; and3) the composition of the mantle prior to contami-nation.

GEoLocIcAL SSTTNC A}ID PETROGRAPHY

The rocks from Sisco (SIS), Montecatini val diCecina MVC), Orciatico (ORC), S. Venanzo(SVEN) and Cupaello (CUP) have been studied byseveral authors who report detailed information onthe geology and volcanology as well as data onpetrologf, and phase composition (e.g., Mttemperger1965, Barberi & Innocenti 1967, Gragnani l972,Polt1985, Wagner & Velde 1980. The following descrip-tions partly lean on these studies. The Sisco lampro-phyre crops out as a sill, 2 to 4 m thick, intrudedinto crystalline rocks of the so-called "Schistes Lus-

ORENDITIC AND KAMAFUGITIC ROCKS 47

tre's" Formation, which was affected by alpine fold-ing and metamorphism. The K/Ar ages reported byCivettaet al. (1978) are in the range 14-15 Ma. Thetexture of the sill is variable, from microgranular toporphyritic, and both these rock types have beenanalyzed in the present work. Porphyritic rocks occurat the border ofthe sill and contain phenocrysts ofbrown mica and altered olivinqset in a matrix of fine-gained brown mica, olivine, amphibole and sani-dine. The same phases occur in the microgranularlithologies. The brown mica is phlogopite, and theamphibole is a K richterite (Velde 1967, Wagrer &Velde 198Q. Accessory phases consist of apatite,

titanite, chromite, ilmsnils, priderite and rutile(Wagner & Velde 1980. Calcite is present as asecondary phase, together with alteration productsof olivine. Quartz crystals, more or less extensivelyresorbed, are found.

The rocks from Montecatini and Orciatico cropout as two associated subvolcanic bodies intrudedinto Lower Pliocene argillaceous sediments that showevidence of contact metamorphism. One single IVArage determination, performed by Borsi et al. (1967)on a MVC sample, gave a value of 4. I Ma. The MVCand ORC minettes are variable in texture and com-position @arberi & Innocenti 1967) and all the lithol-

TABLE t. llAJoR-ElElilEt{T C0IG0SITIoN (ht.S) AND CIPII NoRtlS 0F 0RSNDITES FR0l'l CEI1TRAL ITALY AllD CoRSICA

uont.catlnl val dl cacloa

sls-lo wc-loA

slo2rlo2A1203Fe203FeohougoCaONa20KzoP2o:

l 4 S t *

56.73 57.832 . 2 7 2 . 1 5

10.60 10.652 . 0 0 l . 8 t2 . 4 0 2 . 5 60 . 0 7 0 . 0 77 . 1 3 6 . 8 8

l . l o 1 . 3 010.46 ro. 150 . 6 4 0 , 6 62 . 9 6 2 . 6 9

7 8 . 1 7 7 . 5

57.89 54.292 . 3 7 2 . 3 r

ro.47 10,822 . t 7 t . 1 42 . 5 2 3 . 2 60 . 0 8 0 . 0 76 . 3 8 7 . O 53 . 6 4 2 . 9 3l . 0 6 t . o l

1 0 . 4 9 l o . 3 i0 . 8 2 0 . 7 62 . O 7 2 . 1 4

7 4 . 4 7 8 . 4

58.50 58.632 . 2 7 2 . 3 5

10.84 10.86o . a l 1 . 8 42 . 4 2 2 . 9 60.06 0.08

3 . r 2 3 . 1 5L . O 2 0 . 8 8

1 0 . 7 3 1 O . 7 0o . 6 7 0 . 7 12 . O 9 l , 9 t

7 7 . 8 7 7 , O

L . 6 2 t . 5 2r2.r2 12.273 . 6 6 3 . 1 93 . 2 0 3 . 2 00 . 0 7 0 . 0 77 . 2 r 7 . 7 13 . 4 8 3 . 5 51 . 5 0 1 . 3 07 . 8 1 7 . 8 4l . l 8 r . r 82 . 1 r t . 5 5

7 0 . 8 7 2 . 7

64.t4 64. 18o . 9 5 0 . 7 9

t4.69 t4.642 . 3 t 1 . 8 61 . 0 2 l . o 4o . 0 7 0 . o 53 . 3 4 1 . 8 21 . 3 3 1 . 8 61 . 7 8 2 . m9 . 9 9 9 . 8 3o . 3 7 0 . 3 20 . 8 1 t . 4 3

qzOrAb

NgKa

D1tlvtrtI I

1 . 9 9 3 . 7 25 7 . 8 0 5 7 . 9 1o . o 0 . 0o . 0 0 . 0r . 6 2 1 . 4 9t . o 9 0 . 5 42 . O 5 2 . O 5

1 0 . 9 9 1 0 . 4 41 5 . 4 3 1 5 . 2 4o . o 0 . o4 . 3 2 4 . O 9t , 5 2 1 . 5 6

4 . 7 3 4 . 2 657.09 59.m0 . o o . oo . o 0 . ot . 5 l 1 . 4 61 . 3 4 0 . 5 82 . 2 0 2 . m

1 0 . 0 6 7 . 5 414.32 16.60o . o o . o4 . 5 1 4 . 3 91 . 9 4 1 , 8 0

4 . 5 1 4 . 9 759. u 59.220 . o o . 00 . 0 0 . 01 . 4 9 l . 2 ll . 1 7 1 . 0 91 . 9 7 2 . @8 . 7 6 8 . 6 7

t 4 . 8 9 1 4 . 5 60 . 0 0 . 04 . 3 2 4 . 4 71 . 5 9 l . 6 8

o . 4 2 1 . 6 446.45 46.27r 2 . 6 8 1 0 . 9 93 . O 9 4 . 4 90 . o o . oo . o o . oo . o o . o5 . 1 8 4 . 3 3

2r.72 23.261 . 5 4 1 . 5 03 . O 8 2 . 4 92 . 7 9 2 . 7 9

9 . 2 7 9 . O 758.96 58.021 5 . 0 5 1 6 . 9 r2 . 5 9 t . 9 40 . 0 0 . 00 . 0 0 . oo . o o . 03 . 6 6 4 . 1 48 . l 0 4 . 1 3o . 7 7 0 . 6 7l . 8 l I . 5 0o . 8 8 0 . 7 6

lo!!e Alfloa

9i0z 56.74r lo2 r ,42A12O3 11.25Fe203 2 .00FeO 2 ,92Mno 0.08Mgo 8 .29Cao 4 .4ONa20 1 .36K20 7 .64Pz0s o .7oL . 0 . 1 . 3 . r 6Me i * 78 .7

5 7 . 1 8 5 7 . 6 rr , 4 2 t , 4 3

l l . 1 5 t 1 . 5 12,O9 2 .O52 . 8 8 2 . 9 6o.o8 0 .088 . 2 0 8 . 9 54 . 4 1 3 . 4 21 . 3 4 r , 3 27 . 4 4 7 . 8 20 . 6 9 0 . 6 92 . 6 6 2 . r 7

74,4 79 .4

5 7 , 6 9 5 8 . r 1 5 8 . 1 11 . 4 3 1 . 4 3 r . 4 2

l l . l 8 1 1 . l 8 1 1 . 8 22 . 6 t r . 7 4 r . 5 22 . s 2 3 . 3 2 3 . 3 80.08 0 .o8 0 .078 . 2 0 8 . 7 8 8 . 7 03 . 6 4 3 . 6 0 3 . 5 61 . 5 0 1 . 5 6 1 . 3 57 , 4 2 7 . 4 3 7 . 9 20 . 7 5 0 . 7 0 0 . 6 92 . 6 7 2 . O 7 1 . 4 6

7 8 . 0 7 9 . 1 7 9 . 4

5 8 . 1 1 5 6 . 4 3 5 7 , r 31 . 6 0 1 . 3 5 l . 3 l

10 .95 13 .63 14 .79r . 7 7 r , 7 3 2 . 7 53 . 3 2 4 . U 3 . 4 80.06 0 .09 0 . l06 . 9 6 8 . 7 2 1 . 5 63 . 8 4 5 . 0 3 4 . 7 2t . 4 9 0 , 9 7 1 . 2 07 . 3 5 7 . 0 8 5 , 7 70 . 7 6 0 . 4 7 0 . 3 52 . 2 6 0 . 4 7 0 . 8 3

74.9 75 .92 7r .95

qz

0t

Ab

h

Ns

K6

D 1

tly

Mt

1 l

AP

0.90 l .oo45.33 46.27t l . 5 0 1 1 . 3 31 . 9 1 1 . 2 80 . 0 0 . 0o . 0 0 . 00 . 0 0 . 0

t2 .16 12 .99t2 .29 18 .71l . 1 6 1 . 1 82 . 7 0 2 . 7 01 . 6 6 r . 6 3

1 . 3 6 3 , 2 4 1 . 8 5 1 . 4 846.16 43 .80 43 .85 46 .75l l . l . 6 t 2 . 6 8 1 3 . 1 9 1 1 . 4 12 . 3 8 1 . 8 6 1 . 5 6 2 . 8 00 . 0 0 . 0 0 . o o . o0 , o 0 . 0 0 . 0 0 . o0 . 0 0 . 0 0 . 0 0 . 07 . 9 A 8 . 9 7 9 . 5 2 8 . 2 0

23.03 20 .77 22 .74 22 ,21l . l 8 1 , 2 1 r . 3 2 I . 1 62 . 7 2 2 . 7 2 2 . 7 0 3 , 0 41 . 6 3 1 . 7 8 r . 6 6 1 . 6 3

7 . 0 9 1 . 6 1 7 . 4 043.38 4r .84 34 .1012.60 8 .21 10 . 151 . 4 8 1 1 . 9 3 1 7 . 9 3o . 0 0 . 0 0 . 00 . 0 0 . 0 0 . o0 . 0 0 . o o . 0

1 0 . 0 2 7 . 8 8 2 . 5 216.92 2r .96 19 .77l . 2 l 2 . 5 r 4 . 01 . 8 9 2 . 5 6 2 . 4 9i . 8 0 1 . 0 9 0 . 8 1

r uc/(rG + Fez+) atmtc ratlo caLculat€d asaulng Fe2o3/Feo value of o.2i lulrtplledby 100,

*t","+

"u" I 'F|s onc

t9 cun

T a Ars_nrs

tr8VEN

48 THE CANADIAN MINERALOGIST

2.5A 6

3 is"eo

AA''oN

-

*n|IF

640.5+

40 64 40Si02% SlO2%

o

AA

$sr+

o40 64

SiO2% SlO2%

64 40SlO2% SlO2%

ftc.2. Variation diagrams of some major and trace elements. Squaresl SanVenanzo; diamonds: Cupaello; asterisks:Ociatico; St. Andrea's crosses: Montecatini val di Cecina; crosses: Sisco. Differentiated rocks from MVC and SVENare not represented on trace-element diagrams. Open circle, open triangle aod full triangle represenl average valuesof mafic HKS rocks from Vulsini (data from Civetta et ol. 1984), of mafic KS and HKS rocks from Monti Ernici(data from Civetta e, aL l98l).

Bpo

No2

6=?ooo

6440

20

oo|Gz\

NY

bSo

Nv

2+40

A

xSxo iFte

S ;r',

&

o€o o 4

ol

+

+d

If*I

o A

ORENDITIC AAID IGMAIIUGITIC ROCKS

Crlv

SlO2%

b

e#*

T

do AA

SiOzo6

ogies have been sampled and analyzed. The ORCrocks range from porphyritic to microgranular. Theporphyritic rocks contain phenocrysts of olivine,brown mica and clinopyroxene set in a groundmass

*f

_ rit3

DIrD

oo

A

Sl02%

900

SiO2%

&fsx

,9

tp oa *F-A

49

o6|o

on|Y

64

Eoo.llE

EECI

C)

644064o-F40

Eoct,tF

Eco

N

o+40

si02%

containiqg the same phases plus sanidine, variableamounts of glass, and some K-richteritic amphibole(Wagner & Velde 198O. Apatite, ilmenite, chromiteand rutile are accessories. The brown mica is phlogo-

+o t r E D

o

, tq

IF***

*+ *' t

&il

O A

Ao5


pite (Poli 1985, Wagner & Velde 198O. The olivine(up to 90q0 Fo) is more or less extensively resorbedand altered. A few olivine phenocrysts have kinkbanding and may represent mantle-derivedxenocrysts. The clinopyroxene ranges from Fe-poordiopside to augite @oli 1985). Microeranular varie-ties consist of the same phases as the porphyriticones. MVC minettes have generally a gxanular tex-ture with a poorly porphyritic tendency. They arecomposed of sanidine, brown mica and minorclinopyroxene. Altered phenocrysts of partlyresorbed olivine, and a few intergranular quartzgrains are also present. Apatite and Ti-Fe oxidc arethe main accessories. The phlogopite shows color2sning (pale brown core, dark brown margin) whichreflests a rimward increase of Fe and Ti @oli 1985).The MVC minfies axe cutby felsicveins up to a fewcm across. These have a granular texture and con-sist of sanidine, minor quartz, and some brown mica.The last is present as thin elongate crystals that stillpreserve a pale brown core. Apatite is the main acces-sory phase.

The Cupaello (CUP) volcanic center consists ofone si4gle lava flow 500 m long tlat rests on Creta-ceous limestone and on Upper Pliocene lacustrinedeposits. The rock exhibits clinopyroxene and a fewphlogopite phenocrysts set in a matrix composed ofclinopyroxene, phlogopite, kalsilite, melilite, Fe-Tioxides, perovskite, monticellite and glass. Pyroxeneis diopsidic in composition and has very low Al con-tents typical of clinopyroxene from melilite-bearingrocks; phlogopite has a high Mg (MgO about 2lwt.9o), and is extensively transformed into opaqueoxides (Gragnafi 1972, Gallo et ol. 1984).

The San Venanzo (SVEI\! volcanic complex con-sists of two centers: San Venanzo and Pian di Celle(Mittempergher 1965). Radiometric dating indicatesan age of 0.5 Ma @ornaseri 1985). The Pian di Cellevolcano is a tephra cone with welded scoria and alava flow at its top. San Venanzo is a cinder cone.The lava flow consists of a weakly porphyritic rockcontaining olivine phenocrysts with chromite inclu-sions set in a holocrystalline groundmass of olivine,leucite, melilite, clinopyroxene and kalsilite withaccessory apatite, Fe-Ti oxides and perovskite. Zeo-lites and carbonate minerals are present as secondaryphases. The lava flow contains pegnatitic veins ofkalsilite, melilite, leucite and phlogopite, with abun-dant pneumatolitic minerals. These veins were con-sidered by Mittempergher (1965) as late-stagedifferentiates of the SVEN magma.

ANar,vrrcar, MsrHoos

The major elements have been determined by X-ray fluorescence, Mg and Na by atomic absorption,Fd+ by titrimetic technique, and LOI by classicalgravimetric methods. The concentrations of Ni, Cr,

V, Th, Rb, Sr, Zr, Nb, Y and Ba have been deter-mined !y X-ray fluormcence @ranzini et al. 1972)and Th, Hf, Ta, Co, Sc and REEby INAA follow-ing the method of P oh et al. (1977). Precision is bet-ter than 590 for all the elements determined by X-ray fluorescence and for Th, Ta, Co, Sc, Sm andEu, better than 10Vo for La, Ce and Hf, and betterthan l5qo for Lu, Tb and Nd.

The isotopic composition of Sr has been deter-mined on one sample from each of the Sisco, Mon-tecatini, and Orciatico outcrops. Analyse were per-formed at the Department of Earth Sciences,University of Rome, by a VG Micromass 54E spec-trometer. Values were recalculated to the SrCO3standard value of 0.71011. For San Venanzo andCupaello we have used the data already available inthe literature (Holm & Munksgaard 1982, Holn e/al. 1982, Taylor et al. 1984).In view of the smallvolume of the rock masses and of their rather cons-tant composition, denronstrated by major and trace-element data, we believe that the vSr,/E6Sr damavailable are representative oftle rocks under inves-tigation.

Rrsulrs: rnn Srsco-MoNrecATrNr Var, nrCrcna-OncrATrco Rocrs

Major, trace-element and Sr-isotopis data arereported in Tables I and 3 together with the CIPWnorms. Except for the MVC felsic veins, the singlemagnatic bodies show a rather homogeneous com-position for major and most trace elements.

Major elements and clossiftcation

The rocks from the three localities have manypetrological characteristics in common; they haveintermediate silica content, high Mg# (71-80), K,K,/Na and K,/Al. All are silica-oversaturated; onlythe SIS minette is peralkaline. Compared with typi-cal Roman intermediate volcanics, these rocks havehigher Mg, Ti, K./Na, K/Al, and lower Al and Na(Fig. 2). All these characteristics make tle inves-tigated mineffss distinct from typical rocks of theRoman province and result in their classification asultrapotassic rocks with an ormditic or lamproiticafJinity (Sahama 1974, Mitchell 1985, Foley et al.1986, 1987). These rocks also differ from theminettes found by Van Bergenet al. (1983) as inclu-sions in the Monte Amiata lavas. Comparison withrocks from other ultrapotassic provinces indicatesa close compositional similaxity with orendites fromSpain (Venturelli et al. 1984) and Leucite Hills (Car-michael 1967, Kuehner et al. l98L). Preliminarymajor- and trace-element data indicate that otherrocks with a similar composition also crop out atTorre Alfina (Fig. l), on the northern edge of theVulsinian complex. Detailed investigations of Torre

ORENDITIC AND KAMAFUGITIC ROCKS

0.001

Alfina volcanics are still in progress and only tworepresentative analyses are reported in Tables 1 and3. The acidic veins that cut the MVC minette havehigher Si, Al, Na, K, and lower Fe, Mg, Ti and Pthan the host rock.

Trace elements

The transition-group elements all have high values,especially if one considers tle SiO, level of the ana-lyzed rocks. Ni and Cr are particularly enriched andshow values that fall within the range of primarybasalts from different tectonic environments (Fig.3). Co, Sc and Y are less abundant than in basaltsand increase from SIS to MVC and ORC minettes;on the contrary, Cr and to a lesser degree Ni, arealmost constant. Consequently, CrlY (Fig. 2),CrlSc, Ni/V, and NVSc are higher in SIS than inTuscan minettes. The incompatible elernents displaya high enrichment. The REE patterns are stronglyfractionated for both ZREE and HREE (Fie. 4a,b).Th, Rb and Cs increase from SIS to MVC and ORC,whereas Ta, Hf, Nb, Zr. LREE and Sr have an oppo-

51

MnSc Co Ni

Frc. 3. Chondrite-normalized patterns of transition-group elements for avera_geoREN and KAM rocks. rne oottea are{! represents ranges of values of basaltsfrom different environments (from Langmuir et al' 1977)'

site trend. Accordingly, LILEIHFSE is higher inTuscan minettes; note, however, that LILE/HFSEvalues are invariably higher than those ofultrapotas-sic rocks from intracratonic areas (Mitchell & BellL97A andare within the field or arc-related volcanicrocks (Pearce 1932). The same has been found truefor all volcanic rocks from Central Italy @eccerillo1985), indicating a close genetic relationship ofOREN with typical Roman volcanics. The patternsof hygromagmatophile elements for the atalyzedorendites (Fig. 5a) all show strong fractionation, withnegative anomalies of Ba, Ta, Nb and Ti, 1s foundin a[ noman volcanic rocks (e.g., Peccerillo el a/'1984). Tuscan minettes, however, also have astrongly negative anomaly in Sr content tlat islotseen in mafic rocks from the Roman province. TheMVC felsic veins have lower concentrations of theferromagnesian trace elements,Ba, and LKEE, ?nLdhave higher Rb, Zr, Th, Ta and Nb than the hostminette. These variations are qualitatively consistentwith a derivation from the host minette by crystalfractionation, as suggested by Barberi & Innocenti(rw).


La Ce NdSmEu Tb La Ce NdSmEu Tb Yb Lu

+ oR-10x oR-12*MVC-10AI MVC-24

LaCe NdSmEu Tb yb La Ce Nd SmEu Tb Yb Lu

Flc.4. Chondrite-normalized REEpatterns for the analyzed rocks, for average KS and HKS mafic rocks from MontiEroici and Vulsini (A,B,C), and for three selected siliceous rocks from Northern Apennines @).

og

!co

Eo

too

o

Itco

gct:oo

Yb

100

+ G - 3x G-11* L - 3 4 7

D

o

ttr0&{atR 1 0

o

ttgos-(,

I(,o

Sr-isotope ratios

The initial rSr/sSr value ranges from 0.71256 toO.7l7l5 and increases from SIS to ORC and MVC,in which they reach values that are witlin the rangeof crustal anatectic acidic rocks from the s4me area(Vollner 1977).

RESULTS: THe SAN VeNANzo ANDCUPAELLo RocKs

Major- and trace-element data, together withCIPW norms, are reported in Tables 2 and 4. Exceptfor'sample Ven-7Q, the San Venanzo and Cupaellovolcanics have a homogeneous composition both for

A

+ vEN -3x vEN -2xCUP-2Arf iKS-VulslnlI KS - Ern lc lOHKS-Ern lc lA VEN-7O

c

ORENDITIC AND KAMAFUCITIC ROCKS

+sts-4roR-12*MVC-104ocuP-24AVEN-2

x HKS Emlclx KS Emlcl+ HKS Vulslnl

+ G - 3* L-347x G-11

C o lRb Ta P

Sm TbTI Rb Th Ta

B a K CoS r P Z r f l

TABLE 2. MJoR-ELEMENT CoMPoSITIoN (!t.S) AND CIPH NORI'1s 0F I4ADUPTTES FRoM CENTRAL ITALY AND TESIERII| usA

53

10

.EE6E

5EoEso

Bar T hZr Tl

Cel s r

Frc. 5. Patterns of hygomagmatophile elements (Wood 1979) for some of the analyzed rock$ (A), for average KS andHKS mafic rocks fiom Monti Einici and Vulsini (B), and for three Northern Apennine siliceous sedimentary rocks (C).

San Vetranzo cupaello Leuc l te H l l l s (USA)

VEN-7Q VEN-3 CUP.2A CUP-IA CUP-18 UAD-7 MID-I

s1.02

T102

41203

re203

FeO

h0

llS0

C€0

Na20

Kzo

P zo5

L . O . r .

t i { t

3 8 . 9 6 4 1 . O 3

2 . 1 4 0 . 5 9

10.44 12.39

3 . 1 4 2 . O 9

7 . l o 4 . 0 8

0 . 1 8 0 . r o

7 . 3 0 1 3 . 0 9

1 8 . 8 i l 5 . s o

1 . 6 5 t . 0 4

7 . O 3 8 . 8 6

1 . 4 9 0 . 3 6

) . . 7 2 0 . 8 6

60.7 82.2

4r,20 41.46

0 . 6 7 0 . 7 2

1 1 . 5 1 1 2 . O O

r . 3 9 L . 6 9

4 . 8 8 4 . 7 6

o . l 1 0 . l l

12.76 12.93

1 . O 7 l . l 3

8 . 4 3 8 . 5 0

0 . 3 9 0 . 4 1

r . 8 6 0 , 6 8

8 1 . 4 8 1 . 2

4 r . 5 2 4 1 . 5 5 4 r . 4 6

0 , 7 6 0 . 7 1 0 . 7 8

12.05 12.17 11.62

1 . 6 8 1 . 8 4 1 . 8 2

4 . 8 8 4 . 5 6 4 . 8 0

0 . l l o . l l o . l l

t2.76 12.76 12.70

1 5 . 6 7 1 5 . 6 2 1 5 . 6 4

1 . 2 0 1 . 2 0 t . l 3

8 . 3 6 8 . 5 3 8 . 3 1

0 . 4 3 0 . 4 2 0 . 4 2

0 . 5 7 0 . 5 5 l . l l

80.8 8t,2 80.6

43,68 43.43 43.98

l . r 5 l . t 9 l . 1 6

7 . 5 3 8 . 4 0 8 . 5 5

5 . 4 7 5 . 3 3 5 . 3 0

2 . 2 0 2 . 5 6 2 . 5 5

o . L 2 0 , 1 2 0 . L 2

1 0 . 5 2 1 0 . 5 0 1 0 . 8 6

1 4 . 9 7 i g . s g t 4 . 9 r

0 . 4 5 0 . 4 0 0 . 4 0

9 . 8 8 8 . 5 6 8 ' 0 1

r . 1 4 1 . 2 6 L . 2 4

2 . 9 0 3 . 2 7 2 . 7 4

7 5 - 6 7 5 . 0 ' 7 5 ' 7

43. l0 44.65

2 . 3 2 2 . 2 5

8 . 5 8 8 . 1 6

5 . 3 4 5 . 3 8

0 . 8 0

0 . 1 3 0 . 1 2

1i.60 12.30

1 0 . 7 1 1 4 . 0 6

0 . 9 3 0 . 9 3

8 . 5 3 7 . 1 8

2 . 1 3 3 . O 4

8 1 . 3 8 1 . 8

Ne

KP

Na(s

c€

Dt

o l

o . 3 2 2 . 9 7 t . 7 1 2 . 5 7 2 . A 0 2 . 6 3 2 . O 9

7 . 5 6 4 . ? 7 4 . 9 0 5 . 1 8 s . 5 0 5 . 5 0 5 . r 8

32.54 26.14 30.43 29.34 29. ' t r 29.48 31.62

o . o 1 0 . 7 8 6 . 2 2 7 . 2 5 6 . 5 1 7 . 2 4 4 . 9 6

0 . 0 0 . 0 0 . o 0 . o o . o 0 . 0 0 . 0

0 . 0 0 . 0 0 . 0 0 . o 0 . o o . 0 0 . o

0 . 0 0 . o 0 . 0 0 . 0 0 . 0 0 . 0 0 . o

2 4 . 1 1 2 2 . 1 2 2 2 . 8 1 2 2 . 3 2 2 2 . 3 0 2 2 . 2 9 2 2 . 4 9

2 . 8 7 0 , O 0 , 0 0 . 0 0 . 0 0 . o 0 . o

20.05 28.64 28.18 2A.56 28,33 28.22 28.27

2 . 4 4 r . 4 7 1 . 5 1 1 . 5 4 r . 5 7 1 . 5 3 1 . 5 8

4 . O 7 t . t 2 r . 2 7 1 , 3 ? 1 . 4 5 1 . 3 5 1 . 4 8

3 . 5 3 0 . 8 5 0 . 9 2 0 . r 7 l . o 2 0 . 9 9 0 . 9 9

o . o 0 . 0 0 . 0 0 . 0 0 . o

o . o o . o 0 . 0 0 . 0 1 . 0 8

32,20 35.92 36.56 36.69 33.23

o . o o , 0 0 . 0 0 . o 0 . 0

o . o 0 . 0 0 . 0 I ' t o 0 . 6 2

4 . 7 8 1 . 3 1 0 . 1 8 0 . 9 9 0 . O

3 . 3 5 2 . 9 8 2 . 9 8 2 , 7 5 2 . E l

r 3 . 3 r r 4 . o 7 1 3 . 4 7 2 . 8 7 3 . 5 3

t 9 . o 4 1 6 , 3 9 1 7 . 9 0 2 4 . 6 7 3 0 . 4 I

18.95 19,80 20.05 16.16 15.36

o . o 7 0 . 3 2 0 . 3 0 0 . 0 0 . 0

2 , r 9 2 . 2 6 2 . 2 1 4 . 4 1 4 ' 2 8

2 . 7 0 2 . 9 8 2 . 9 4 5 . 0 4 7 . 2 0

utI1Ap

. carculated stth Fa2O3/feO latlo of 0.2.1of 7 and wl, Milupltes frm kuclte Hillg reapectlvely frm Kueh.e! e! 41. (l9Bl) and

froo Thonpson et 41. (1984).

major and trace elements.

Major elements and classificatio:n

Lavas from the two volcanoes axe ultrabasic andstrongly silica-undersaturated; they have very highMg# and Ca. The CUP rocks are peralkaline and

have higher Ti, Si, P, K and K,/Na than SVEN lavas'whereas Na and Al are lower and reach the lowestvalues among potassic rocks from central-southernItaty (Fig. 2). The LOI is relatively high in CUP' tes-tifying to some degree of alteration. Compared withlitiraiure data for the same rocks (Gallo et al. 1984'Taylor et al. 1984), however, our samples have lower


IABLE 3. TMCE-ELEMENT CoNCENTMTI0NS (ppn) IN oRENoITIC RoC$ FR0l'l CENTRAL ITALY AND CoRSICA

s Is -10 s rs - t t sIs-4 srs-2 vtc-22 wc-10A uvc-24 u9c-204v 8 9cr 425ScNl 239Conb 310Sr 839Ba 1235k 188Ce 337NdSEEuftYbr 2 8lh 43HfZr 1257TaNb 6I87sr/86sr

89 904 2 5 4 l l

9 . 8 1 327t 272

t 9 2 7343 327812 738

1261 rO22183 176302 400t l l t 6 5t 9 2 l

0 . 8 l . 3l . l 1 . 4

26 254t 46

r35l t274

66 63

37 40t22 118

I 4536 502

2 . 232 3l

0.71.7 t5

58 4899 68

7 . 850 34

9977 94845t 436693 63562 a6

1 3 5 1 5 tt4025 13

3 . 8 2 . 21 . 4 0 . 8r . 7 1 . 4

39 37139 153

708 1062 . 338 31

85400

243

u1

188333

53

1384

420l t . 5

264

380803

1 3 1 0t72347t39l 93 . 2r . 2t . l

2 7433 3

I 3094 . 3

63o.7 1256

42r

254

376441

193344

2 743

t302

64

t23 t24404 435

20148 156

32781 770469 454

1167 t20276 1a

ta2 206

torre ALflna

oR-28 OR-2I 0R-7 0R-4 0R-10 0R-12 OR-I TA-1 TA.2

v 9 8cr 439ScNl 292CoRb 59tsr 638Ba l l54La 143ce 310Nd

EuIbYbY 3 0'lh

l l5Hfzr 764TaNb 4087s!/86sr

96 91449 442

300 297

536 455624 649

1198 I r77r45 l4l3 t9 309

97 100 103455 449 364

1 5 1 7307 274 24728 31

449 609 637646 613 594

1175 1224 91rr49 141 r54315 338 332155 18429 283 . 3 3 . 81 . 1 1 . 21 . 9 1 . 4

29 32 32r09 I2 I 116

18 20787 797 830

z . o J . t

41. 41 420 . 7 1 5 8 8

t25 r42842 552

32t 259

458 33473'' 606

1292 1145100

257 2r2

J )

655 581

35 370 . 7 1 6 5 - 0 . 7 1 6 8 *

463

299

59764r

r43321

30l l 8

777 796 747

40 4I 40

30 29t19 r20

* data f , !@ Felrara et af . ( f986).

LOI, Si, and Ca, and much higher K; we concludethat the lower degree of alteration in our samplesmay account for tle observed discrepancy. TheCIPW norms show that CtlP and SVEN lavas aremore strongly silica-undersaturated than otherRoman rocks and contain larnite. Large amounts ofkalsilite app€ar in tle norm of CUp. Many of theabove petrological sharacteristics (especially thedegree of silica undersaturation, abundance of Naand Al, alkali/silica, K,/Na and FelMg values) arepeculiar to the SYEN and CUP rocks and are notfound in the typical Roman volcanic rocks. Becauseof t}te extreme degree of silica undersaturation, lead-ing to the occurence of modal melilite, leucite andkalsilite, the SVEN and CUP rocks have been con-sidered by Salama (1974) and Gatlo et at. (1984) tobelong to ultrapotassic rocks with a kamafugitic

afJinity. The same affinity has been recognized byFoley el al. (1987). Accordingly, the olivine melili-tite of SVEN is mineralogically comparable to katun-gites (Sahama 1974), whereas the Toro-Ankole bio-tite mafurite would be the closest equivalent to thephlogopite melilitite of CUP. As stated by Sahama(1974), however, the latter rock "is chemically nottoo different from the madupits of Leusite Hills andin this sense is intermediate between the rocks oforenditic and kamafugitic affinity" possessing thestrong undersaturation of kamafugites and the Feand Mg abundance of orendites. In our opinion theCUP and SVEN significantly differ from typicalkamafugitic rocks from central Africa and shouldsimply be called madupites in view of their chemi-cal equivalence with Leucite Hills madupites as wellas of their association both in the western United

ORENDITIC AND KAMAzuGITIC ROCKS

TABTE 4. TMCE.ELEIIENT CONCENTMTIOIIS (PPN) Iil I'/|ADUPITTS FRON CENTML ITALY

) )

cupsel10

VEN-7Q VN-3 VN-8 VEN-5 l'!N-2 VEN-6 MP-2A CUP-IA CUP.IB

v 430Cr 75sc 58Nl 40co 38kb 271sr 2330Ba 2044k 236ce 482Nd 214sn 52

T b 3 . 9Y b 6 . 1L u 1 . Or 9 5fr rl2

zt 952T a 2 , 3lrb 39

87g.7863"

1 1 2830

l 9r 5 340

45817216t765

7 LL 42 . 1l . l2 . O0 . 3 3

3530

7 . 4264

t 3

0 . 7 1 0 3 7 i

126 t35280 793

139 L49

483 4751124 1118642 67086 87

t77 186

141146

2 313639

462l 8 l 97209 5

t 9 18820

3 . 3t . 4

0 . 4 34 l389 . 2

3 4 1o . 8 4

150 139

8350t 88036

5963704.4390

254578232

4 l

2 . 60 . 3 6

591 3 725

8253 , 9

0 . 7 1 1 2 0 r

49 4r5 2 5 l

79 80

563 5813649 36l l4388 4557

295 296483 481

60 59t34 r35

835 842

46 47

39 4r36 38

3t7 335

1 5 1 5

36 38

283 3t2

1 3 1 5

134 146787 812

480 4771781 1784690 698

90 93185 198

t Data froB b10 ! uunksaaard (1982)

States and Italy with ultrapotassic rocks of orenditicaffinity. In order to avoid confusion with previousnomenclature, however, we will continue to considerthe SYEN and CI-IP volcanics as ultrapotassic rockswith a kamafugitic affinitY.

Sample Ven-7Q has lower Si, Al, Mg, and fer-romagnesian trace elements, and has higher Fe' Pand incompatible trace elements as the main faciesof the SYEN lavas. This composition, which will notbe discussed further, may represent a typical exam-ple of differentiation of a madupitic melt.

Trace elements

Both SVEN and CUP show high enrichment inincompatibls elements, but this feature is extremein CUP. The hygromagrnatophile element patterns(Fig. 5a) are similar to those of typical KS aud HKSRoman rocks (Fig. 5b) and exhibit a strong fractio-nation, negative anomalies of Ba, Ta, Nb, and Ti,and a small or negligible Sr anomaly. This clearlyindicates a close genetic relationship of KAM withtypical Roman potassic volcanics. The CUP samplesare distinctive in the higher absolute abundance ofincompatible elements and in the small negativeanomaly of Sr and Ba. The REE are strongly frac-tionated (both ZREE and HREE) and have a smallEu anomaly (Flg. 4c). The ferromagnesian elements,especially Cr, are very abundant in SYEN, but notso much in CUP (Fig. 3).

Sr ond orygen isotope rotios

The 875r/865r and 6180 in SVEN and CUP lavashave been reported by Holn & Munksgaard(1982,1986), Holm e/ al. (1982) andTaylor et al.

(1984). CUP rocks have a higher Sr-isotopic ratio

io.ltiz) and 6180 (+12.1, Holm & MunksgaardisgO tiruo the SVEN lavas, which have 87sr/86sr

0.71037 and 6180 + ll to + 12.

DISCUSSION

Peccerillo & Manetti (1985) reviewed the petrolog-ical and geochemical characteristics of Italian potas-sic rocks and concluded that'at least four differenttypes of ultrapotassic rocks occur in the Italianpeninsula: the potassic series (KS), high-potassiumseries (f{KS), orenditic (ORE}9, and kamafugitic(KAM) rocks. The distinct degree of silica undersatu-ration, K/Na ratio, K, Na, Al, and trace-elementabundances (see also Fig. 2) led Peccerillo & Manettito conclude that these magmas cannot be related toeach other by any conrmon evolutionary processes.Accordingly, it was suggested that the magrnas wereformed independently within an anomalous uppermantle heterogeneously effiched in incompatible ele-ments. On the basis of evidence emerging fromexperimental petrology (e.9., Arima & Edgar 1983,Wendlandt & Eggler 1980a,b), Peccerillo & Manetti(1985) suggested that the strongly undersaturated tooversaturated potassic rocks from central Italyformed by a low degree of partial melting of aphlogopite-bearing olivine pyroxenite or phlogopiteperidotite at upper-mantle depth. The pressure ofmelting, however, was thought to be different forthe various potassic magmas, namely, the silica-oversaturated OREN were formed at relatively lowpressures (< 15 kbar) in the uppermost mantle, where

bhlogopite melts incongruently to give olivine andoversaturated potdssic liquids as reaction products


(Wendlandt & Eggler l980a,b). A genesis at relativelylow P has been hypothesized also by Foley et al.(198Q for high-silica lamproites on rhe basis ofexperimental data. On the contrary, KS, HKS andKAM were attributed to melting at progressivelygreater depths. Variable H2O/CO2 in the source isthought to have played an important role in deter-mining the petrological variations between CUP andSVEN. A similar petrogenetic model was put for-ward by Kuehner et al, (1981) to explain the genesisof associated madupites and orendites at LeuciteHills. A deeper source for HKS than for the KS wasalso suggested by Civetta et al. (1981) on the basisof trace-element evidence.

The new data do not fully agree with the abovemodel. The strong HREE fractionation in ORENindicates a genesis by melting of a garnet-bearingrock, with garnet left in the residue. Because garnetis stable at high P (> 15-20 kbar) in the manrle, thehypothesis that OREN were formed in the uppermostmantle conflicts with geochemical evidence. As willbe discussed later, the derivation of OREN magmafrom a source with fractionated HREE, resultingfrom enrichment processes, overcomes this problem.

If there is a consensus on the mantle origin ofItalian potassic magmas, there is still debate on therelative role played by crustal contamination andenrichment in the mantle in determining thegeochemical peculiarities of Italian potassic rocks.

Interaction with crustal materiol

Italian potassic rocks are characterized by higheurichment in incompatible elements and higheSr/sSr (typically in the range 0.705-0.711) andhigh gltQ values. These observations led severalauthors to suggest that primary potassic magmaswere generated in an anomalous upper mantle (e.g.,Holm el al. 1982). Others have stressed the role ofinteraction with continental srust in determininganomalous Sr and O isotopic compositions (e.g.,Turi & Taylor 1976).

Recently, integrated geochemical and isotopicinvestigations (e,9., Civeltaet al. 1981,1984, Rogerset al. 19t35) haveshown that the most primitive mem-bers of KS and HKS have an abundanse of fer-romagnesian trace elements that are close to thosethouCht to be typical of poorly evolved mantle-derived magmas. This characteristic has been inter-preted as evidence that the most primitive magmasof the potassic series have undergone moderate crys-tal fractionation with littleinteraction with continen-tal crust (a9., Peccerillo 1985, Rogers et ql. 1985).

We believe that the same conclusion holds true forSYEN rocks as well. In fast, these have similarrsr/88r ratios and comparable or higher Ni and Crabundances than majic HKS rocks from Alban Hills,Vico, Vulsini and Ernici. In addition, the Mg and

Ca abundances are the highest in the Romanprovince. In spite of this, Taylor et al, (1984) sug-gested, chiefly on the basis of orygen-isotopic evi-dence, that the SYEN magma may have assimilatedsignificant amounts of crustal material (about 2090of the initial magma's volume). Such a process,together with the coucomitant crystal fractionation,should have raised the silica content of the magma.Taylor et al. (1984) suggested that the original com-position of the magma had a carbonatitic affinity.It must be pointed out, however, that the similaritybetween the patterns of hygromagmatophile elementsof SYEN and HKS mafic rocks (Fig. 5) contrastswith a hypothetical carbonatitic affinity of SYENvolcanic suite. Accordingly, we believe that the SanVenanzo rocks represent barely evolved or primarymantle melts that have not suffered important frac-tionation and interaction with crustal material. If so,fts high enrichment in incompatible elements andthe high rSr/ElSr and, possibly d18O values, areinherited from the source, as also suggested by Hotn& Munksgaard (1982, 198Q, Civetta et ol. (1981),Holm el al. (1982), Rogers et al. (1985), and Pec-cerillo (1985) for mafic rocks from several Romanvolcanoes. An alternative process that may explainthe high 6180 could be that referred to by Taylor etol. (1984), when they state that in narrow conduitsstrong t8O,/1O interaction between magma andcountry rocks may occur without siCnificant changesin the chemical composition of the magma. In thiscase, however, the relevance for petrogeneticinterpretation of oxygen-isotope values is dubiousand may lead to invalid conclusions.

Turning to CUP rocks, these have higher abun-dances of Si, incompatible elements, Sr, and orygen-isotope ratios, together with lower ferromagnesian-element abundance with respect to the SYEN lavas.Therefore, it is tempting to suggest that the CUProcks were derived from SYEN magmaby processesof crustal assimilation. However, the CUP rockshave higher Sr abundance than SVEN rocls, and thisis the opposite to what is expected from assimilationprocesses. Combined assimilation and crystal frac-tionation (AFC) increase both Sr and 875r/863r

values if partition soefficients (r(,r) of Sr for thecrystals that separated are lowerthan unity @e Paolo1981). Olivine phenocrysts in the SVEN volcanicshave /Gt much less than l; their separation, con-comitantly with crustal contamination, couldproduce an increase of both Sr-isotope ratio and Srabundance. In order to model the Sr increase,however, we must assume a large amount of olivineseparation. In addition, we also need a high ratiobetween mass of assimilated material versus that ofcrystallized magma (Ma/Mc) in order to obtain thenecessary increase in Sr-isotope ratio. Assuming aMa/Mc val'oe of 0.8, we need a separation of about40Vo solid in order to get the Sr and rSr,/8€r values


o.724

0.708

of CUP. Because olivine has KN between 15 and 40in Italian potassic rocks (Holm 1982, Francalanci elal. 1987), such extensive $eparation of olivine shouldproduce a decrease of Ni to less ttran one ppm, evenwith the conservative assumption of i(Ni in therange 10-15. The element Ni does decrease, but notas sharply as expected. More complex evolutionaryprocesses, such as repeated injection of new magmacontemporaneously with AFC, could account for thelack of a sharp decrease in the abundance offerromagnesian-elements abundance (O'Hara 1977).ffowever, these processes are likely to occur in vol-canic systems with well-developed magma chambers.On the contrary, Cupaello volcano consists of onesingle lava flow. Accordingly, we beliwe that SYENand CUP magmas cannot be related by evolution-

SrFrc.6. A) Sr-isotope ratio versus Sr abundance for the analyzed rocks, foraverage

HKS mafic volcanics from Vulsini and Ernici, Torre Alfina lavas, a Sneiss fromTuscany Oalf-filled square), and a South Tuscany rhyolite (filled square). Ilfetaiagram G'ig. 68) represents 15r/865r versns Sr relationship of a mantle perido-

tite contaminated with 2090 of a liquid generated by 4090 melting of a sedimen-tary rock containing different proportions of siliceous and carbonate components.St. Andrea,s cross represents a peridotite contaminated with a liquid derived froman almost pure siliceous sediment and mimics the sourc+rock composilionof Tuscan orendites; asterisk and open squaxe represent a mantle contaminatd with marlysediments containing increasing amounts of carbonate fraction and mimic thesource composilion of SIS and HKS-KAM rocks, respectively. For furthe'r expla-nation see text.

1600

ary processes; rather, they indicate distinct types ofmagmas generated independently in the source. Thesame conclusion has been drawn by Holm &Munks-gaard (1986) on the basis of O and Sr isotopic evi-dence.

Data on OREN rocks are even more intriguing.These have lower Sr contents than KAM or HKSrocks and higher values of Sr-isotope ratio. Theseobservations hold true even for Torre Alfina lavaswhich, as previously mentioned, have similar petro-loeical characteristics as MVC and ORC rocks (seeTable 1).

Figure 6a illustrates 875r/86sr verslas Sr relation-ships for all the rocks invatigated and for some HKSlavas from Monti Ernici, Mts. Vulsini and TorreAlfina (Ferrara et al. 198Q. A Tuscan metamorphic

L

a(Do

a6

o

o.722

0.708o


L

o(oo

tt,o

0.708

gneiss @errara et al. 1986) and a Plio-Quaternaryfelsic volcanic rock from the same area (Vollmer1977) are also plotted. Ir is evident tlat the ORENsuite plots along a hyperbolic curve having at itsextremes KAM-HKS and Tuscan rocks. ThetrSr/8$r verszrs Sr relationship becomes linear on aSr-isotope ratio versus L/Sr plot. These relationshipsare strongly indicative of mixing and suggest thatOREN may have beeir formed by interaction of HKSor KAM with the metamorphic basement or withacidic anatectic rocls of Plio-Quaternary age that.occur in several different places in southern Tuscany(Fig. l). The positive relationship of rSr,/eSr yersrltRb/Sr @ig. 7) also points to the same conclusion.This hypothesis was proposed byFenanet al. (1986)and Turi et al. (1986) for Torre Alfina lavas.However, if we consider the volumes of end-membermaterial involved in such a process, at least 10Vo ofcrustal end-member must be assimilated by HKS orKAM magma to reach the esr,/8qsr and Sr ppmvalues of T\rscan OREN. The abundance of crustalend-member becomes even higher if one invokesAFC instead of simple two-end-member mixing witheither acidic volcanic rocks or metamorphic base-ment. These values are in strong disagreement witha large number of geochemical data. For instance,Rb, Th, REE atd other incompatible elements are

oRblSr

Frc.7. Rb/Sr verors Sr-isotope ratios. Symbols as in.Figure 64.

much higher in OREN than in KAM, HKS, meta-morphic basement, and Tuscaa rhyolites and rhyo-dacites (Gianelli & Puxeddu 1979, Ciraad et al.1987). More significantly, Cr and Ni abundances inOREN ars similar to or higher than those of KAMand HKS mafic rocks. The same holds true for TorreAlfina lavas, which have Mg#, Cr and Ni similar tothe analfzed orendites (see Tabls 1,3). One may sus-pect that the high Ni and Cr abundance ofthe inves-tigated OREN results from mafic-mineral accumu-lation. This possibility, however, is ruled out by thefact that the analyzed rocks have a variable texturewhich ranges from porphyritic to microgranular, buthave a constant major and trace-element composi-tion within each si"gle magmatic body. On the basisof this evidence, we conclude that the OREN rocks(Torre Alfina insluded) are not produced by inter-action of HKS or KAM magmawith srustal material.

If this conclusion is accepted, one must explainthe high values of 875r/865r of OREN as well as themixing relationship (Figs. 6,7). In general tenns twopossibilities exist: 1) OREN rocks are produced byinteraction of subcrustal magma with upper crust;in this case the mantle end-member is not akin toHKS or KAM but must have a very highferromagnesian-element abundance, rcembling kim-berlites or some high-Mg lamproites; 2) OREN are


primary or poorly modified mantle-derived melts,uod 15" trigh Sr-isotope values are inherited from tlesource. Hypothesis (1) is not unlikely because high-Mg, -Ni, {r, ultraba$ic to basic lamproites are foundassociated with intermediate, Mg-, Cr-, Ni-rich rocksin several arerars, e.9., in western Australia (Jaqueset al. 1984); however, there is no evidence for theoccurrence of ultrabasic lamproitic rocks in Italy.Hypothesis (2), which implies that the mantle source-rocks of OREN were more strongly contaminatedwith crustal material than KAM, is explored in detailbelow.

Nature of the enrichment process

Both petrological and geochemical data indicatethat the mantle source of OREN and KAM wasaffected by contamination. The process increased theabundance of incompatible elements and the875r/865r value of peridotite, and stabilized phlogo-pite which, according to experimental petrology, isa key mineral in the source rocks of potassicriagmas. Such a hypothesis has been proposed formany K-rich rocks from different areas (e.9., Bachin-ski & Scott 1979, 1980, Barton & Hamilton 199,Kuehner et al. 1981, Cullers et al. L985).

For Italian potassic magmatism, however, thereis no agreement on the nature of the enrichmentprocess. According to some authors (e.9., Cundari1980), the enrichment was provided by fluids fromthe lower mantle, as suggested for other potassicmagmatic provinces @ifel and East Africa). On theotler hand, other authors (e.9., Thompson 197, DiGirolamo 1978, Edgar 1980, Peccerillo et al. 1984,Peccerillo 1985) used discriminant-element abun-dances to suggest tlat enrichment was provided byintroduction intir the mautle of crustal material car-ried down by subduction proce$ses that were activeunder the Apennines in Tefiiary times (Cletta et al.1978). A suitable crustal contaminant could besedimentary rocks or their metamorphic equivalentswhich make up tle Tuscan basement @eccerillo1985). Trace elements of Jurassic siliceous sedimen-tary rocks from various zones of Northern Apen-nines were determined by is'|'an*ti et al. (199). Theserocks have variable trace-element contents; data forthree representative samples are shown in Table 5.The patterns of incompatible elements and REE ofthese rocks are shown in Figures 4d and 5c. The fewavailable data (Gianelli & Puxeddu 199) indicatethat Tuscan gneiss has a similar distribution patternof incompatible elements as that of Northern Apen-nines sedimentary rocls. Inspection of incompatible-element patterns (Fig. 5c) reveals a number'ofinteresting features. The absolute abundance of theelements in the sedimentary rocks considered herehave a fractionated incompatible-element patternwith'high values of LILE/HFSE, and negative

5. CONCENTRATIONS OF SOI.IE I.IAJOR AND TRACE ELEMENTS INREPRESENTAT1VE SILICEOUS - PELITIC SEOIUENTS FIOI'I'THE

NOR'THERN APENNINES

c-11 L-347

K 2 0 r . z

T102

P 205

Rb ppn

Sr

Be

k

C€

Nd

SD

Eu

fr

Yb

lu

Th

nf

Ta

t . 1 2

0 . 1 7

0 . 0 6

43

44

60

1 8 . 5

2 . 6

0 . 3 9

0 . 3 5

1 . 1 0

0 . l 8

o . 6 9

o . 2 9

3 . 6 5

o . 7 2

o . 2 4

28

390

45

36.4

4 . 7

t , 4 ?

0 . 9 3

o . 6 7

12.7

o . 4 2

o , 0 E

0 . 0 7

83

1 5 . I

4 . 3

o . 7 4

0 . 5 7

0 . 8 9

0 . 1 6

t . 4

0 . 2 8

o . 5

All the data except for (20, Ti02, P20s, Rb and Sr are frcntlanetti e, aL. (1979).

anomalies of Ta, M, Ba, Sr and Ti which mimicthose observed in Roman potassic rocks. The R.EEpatterns are fractionated and show small but signifi-cant negative Eu anomalies. This latter characteris-tic is common in many Roman potassic rocks (seeFig. 4c), even the most primitive ones, in whichplagioclase is absent as a liquidus phase.

The role of these sediments as possible subduction-related mantle contaminants is also consistent witha Sr-isotope determination on one siliceous rocksample from the same outcrop as the samplesreported in Table 5, which give a present-dayeSr,/sSr value of 0.7235 (M. Barbieri, pers. comm.1987) and with preliminary data of a systematic Sr-isotope investigation of Northern Apennine sedi-ments, which gave values ranging between0.7 14-{..144 (authors' unpublished data). This rangeencomp:rsses tle values of metamorphic rocks of theTuscan basement (0.7254.133) reported by Ferraraet al. (1986). We adopt a value of 0.724 for averagesedimentary rock in the following calculations. Forthe absolute abundance of Sr we will use a value of50 ppm, which is the mear of the three samplesreported in Table 5. Under the assumption of a sim-ple two-end-membgr mixing, Sr-isotope ratios asthose of KAM can be obtained by interaction of aprimitive mantle having 875r/8651 0.7M and Srabsolute abundance at chondritic levels (say 10-15ppm) with about 5-1590 of sediment. In contrast,to model the rSr/eSr ratio of OREN one must mixprimordial mantle material with l59o to 3090sediment.

59

TABLENRTE

60 TT{E CANADIAN MINERALOGIST

o102o 0.4 0.8

CelSrFtc.8. CelSr versas 8751165r diagram for tle analyzed rocks and for HKS from Vul-

sini and Ernici. Large asterisks represent values of mantle, limestono$, KS-KAM and OREN source rocks, and siliceous sediments. For other symbols seeFigures 2 and 6A. OREN and HK$-KAM sources have been modelled assumingthat primordial mantle interacted with 2090 of a liquid formed by 4090 meltingof sedimentary rocks containing variable amounts of carbonate component. Thedotted line represents compositional variation of sediments containing siliceousand carbonate components in different proportions. For further explanation seetext.

l-

@(o@

:-a

Fo

Some geochemical characteristics axe not explainedby a process of simple two-end-member mixingbetween mantle and bulk sediments. These include:1) the very high Th/Ta values of OREN and KAM(10-50), which are higher than those of the analyzedsediments (3-11); 2) the strong fractionation ofHhEE in OREN and KAM (lbn/Ybn 3-5) whichis less strong in sediments (IbnlYbn 1.3-2.O; and3) the absence of Sr negative anomalies in KAM(CelSr 0.1-0.15), which are present in OREN (CelSr0.3-0.5) and in sediments (CelSr 0.25-0.4) Gee Fig.8).

Additional problems arise frorn the difficulty inmechanical mixing between solid sediments withmantle, and from the rather high amount of crustalmaterial that should be involved in mantle contami-nation to attain rSr/trSr values of MVC orendites.Many of these problems become less severe if oneconsiders more complex but also more realisticmodels for sediment involvement in mantle contami-

nation. It is unlikely that sediments carried down bysubduction can mix with upper mantle witloutundergoing any modification prior to mixing. Morerealistically, sediments are first dehydrated and thenmetamorphosed; successively they melt partially,producing acidic liquids. These rise through themantle wedge and, because of the compositional con-trast, react with the peridotite (Nicholls & Ringwood1973). Such reaction leads to the formation ofphlogopite (Wyllie & Sekine 1982) and generates anenrichment in incompatible elements and radiogenicSr. Later, melting of phlogopite-bearing rocksproduces potassic liquids (Wyllie & Sekine 1982,Wendlandt &Eggler 1980a,b). Accordinely, tlte con-centration of the incompatible elements of potassictiquids depends on several fastors that include: l)the elemental abundances in the original sediments,2) the elenental abundances in sedimentderivedmelts which, in turn, depend on the degree of sedi-ment melting and the residual mineralogy, 3) the

slLlcEous IsEDf MENrs /'

/

x/

F(*ril souRcE./

. /+

-&4uu.ffil"LIMESTONE

S unrurle


elemental abundance in the upper mantle prior toreaction, and 4) the degree of melting of phlogopitepyroxenite or peridotite and the residual mineralogy.Obviously, in order to model satisfactorily the con-centration of incompatible-element contents in potas-sic rocks, one must reconstruct the possible residualmineralogy and degree of partial melting of sedi-ments along tle subduction zone and of theanomalous mantle. Problems connected witl uncer-tainty about the degree of partial melting can bepartly overcome by using evidence from experimen-tal petrology, and by cousidering ratios of incom-patible elements with a similar degree of incompati-bility for the main rock-forming minerals, insteadof absolute abundances of elements.

The melting temperatures of sedimentary rocksinvolved in subduction are variable and depend onthe amouut of water present. Accordingly, thedegrees of melting can be variable also. We assumea degree of melting of 4090 in the following calcula-tions. As for the residual phases, garnet is a liquidusmineral during melting of siliceous sediments 3f highP (Green 1970. Accessory minerals also can bepresent and can lavs important effests on thebehavior of some trace elements. The occurrense ofTi-, Nb-, Ta-bearing minerals as residual phases dur-ing ma€maformatlon is often invokedto explainthehigh Tb,/Ta ratios of arc-related rocks. This hypothe-sis may even apply to our rocks. The presence, inthe residue, of magnetite and ilmsnils is by itselfsufficient to produce alarge increase in ThlTa andLREE/Tavalues in the liquids. In fast, during sedi-ment melting, residual minsmlg are in equilibriumwith acidic liquids. Nash & Crecraft (1985) reportedKvr in magnetite and ilnenite coexisting_with acidicmagmas. These data indicate that Kra is muchhigher than Kfr and I(w. h is obvious that asmall but significant amount of these minerals leftin the residue after sediment melting can producefelsic liquids with much higher Th/Ta than the origi-nal solid. The presence in ttre residue of 490 magne-tite and2t/o ilmenite is able to produce a liquid withTMTa -40, close to the highest values found in therocks investigated. Higher Th/Ta values can beobtained by decteasing the degree of melting. Finally,the presence of l09o garnet in the residue bringsHREE fractionation of melted sediments to valueshigher than those of the most -ElREE-fractionatedrocks under investigation. Reastion of these meltswith the upper mantle generates a rock which hasa strong HREE frastionation. The melting of thisrock is able to produce potassic liquids with highTblYb values as that of OREN even if garnet is notleft in the residue.

Consider now the absence of negative Sr anomaliesin KAM and all HKS rocks. It is impossible to workout a melting model that gives liquids with no Srdepletion, starting with an average sediment. Any

liquid formed from these sediments should show anegative Sr anomaly, independently of the amountand type of phases left in the residue. Accordingly'the only possible alternative is that a second com-ponent took part in mantle contamination. This com-ponent should have a high aluadance of Sr, but alow concentration of other incompatible elements,so that the final sedimentary end-member retains thesame pattern as the analyzed siliceous rocks' but con-tains more Sr. This additional component could bea rather pure carbonate rock @eccaluva et al. 1985)-In this view the differences in CelSr among ORENand KAM can be explained simply by differences inthe amount of carbonate material mixed with sili-seous sedimentary rocks involved in subduction'melting and contamination. Assuming a carbonatecomponent with 500 ppm Sr (values co4mo^nlyfound in Apennine carbonate rocks), nsr/89sr

0.709 (similar to current seawater values) withnegligible abundances of other incompatible ele-ments, we should mix about 4090 of carbonate with6090 ofaverage siliceous rock in order to reach CelSrvalues equivalent to those of KAM and HKS rocks.Such a mixture contains about 230 ppm Sr and hasa eSr/86Sr value of 0.711. A liquid formed by Ntlomelting of this mixture has Sr = 5fi) ppm, if.fd'= 0.1 for residual minerals. A mixture of Xsloof this liquid with 80Vo primitive mantle (two timeschondritic abundance ofCe and Sr) gives a Sr-isotoperatio of 0.7103 and CelSr 0.15 (Fig. 8), similar tothose in SYEN and HKS rosks.

As for OREN, which have a sharp negative Sranomaly, the amount of carbonate compouentinvolved in source contamination must have beeneither much lower or absent. In order to model theSr isotope and CelSr values of the OREN sourcerock, we shall again mix 8090 mantle atd 20v/omelted sediment; however, in this case the originalsedimentary rock must contain less than 1G'1590 ofcarbonate component (Fig. 8).

In conslusion, an interaction between mantle andfelsic liquids generated by partial melting of sedi-ments of variable composition, i.e., containing avariable amount of carbonate fraction' is considereda possible explanation for the high and variablersr/8$r and CelSr values of OREN and KAM. Inother words, the source rocks of minettes were prob-ably affected by contamination with partial meltsderived from a siliceous sediment containing lowamounts of a carbonate component. In contrast, theHKS andKAM sourcerocks were contaminatedwithfelsic liquids derived by melting of a marly sedimentcontaining comparable amounts of a carbonate andsiliceous components. This model is also consistentwith the variation of Rb/Sr versus nSr/KSr. Thelamprophyres sfudied have higher Rb/Sr values andSr-isotope ratios than KAM in accordance with thehigher Rb/Sr and eSr/86Sr of siliceous sediments

6l

62 TIIE CANADIAN MINERALOGIST

with respect to marly rocks.The proposed model also explains the puzde of

the hyperbolic mixing line from KAM and HKS toTuscan metamorphic rocks tirough OP.EN (Fig. 6a).In fact, primordial mantle-rocks contaminated witha fixed amount of melt derived from sediment ofdifferent composition show a hyperbolic relationshipbetween esr/8$r and Sr absolute abundance, thatmimiss that displayed by OREN, KAM and HKS(Fig. 6b). Accordingly, the observed hyperbolic rela-tionship could simply reflect the different propor-tions of siliceous sediments and limestones added tothe mautle source-zone of OREN, KAM and HKS.

Constraints on mantle composition

Both OREN and KAM rocks show some p€tro-logical and geochemical characteristics that are notfound in otler potassic Roman volcanic rocks. Theseinclude the low to very low Na, very high K/Na andMg#, high Ni, and Cr, and variable but moderateSc, V and Co. CrlV (Fig. 3), Ni,/V, and Ti/Caarealso high in OREN, whereas KAM volcanics havehigh CalAl. Some of these characteristics are foundin other ultrapotassic rocks (e.9., Jaques el al, 1984)and have been attributed to several proc€sses thatinclude: l) preferential melting of some phases (e.9.,phlogoprte) tlat produce liquids with a compositionsimilar to that of the phases being melted; 2) somesort of metasomatism or gaseous transfer in themagma after formation that may lead to depletionin mobile elements; and 3) melting of residualperidotite.

Possibilities (l) and (2) can explain only some ofthe above chaxacteristics. For instance, preferentialmelting of phlogopite can give liquids with low Na,high K/Na and high Mg#. The preferential meltingof phlogopite has been invoked by several authors(e.g., Van Kooten 1980, Cullers et al. 1985, Wagner& Velde 1980 to explain the major-element chemis-try of ultrapotassic rocks from different areas. Thishypothesis, howwer, does not explain the high CrlYand Ni/V of OREN. In fact, if the process occurredunder equilibrium conditions, the trace-element con-tents in the liquid were not affected by the phasesthat entered the melt but only by tle rmidual miner-alogy (Consobnagno & Drake 1976). Residual gar-net during mantle melting may be responsible for tlelow Sc in the liquid, but not the low V. Residualspinel would preferentially retain Cr instead of V.Gaseous transfer is able to change the contents ofmobile elements such as alkali elements, but shouldnot affect other elements such as Ti. Accordingly,the possibility that the mantle material successivelyaffected by contamination was reprcented byresidual peridotite seems a suitable explanation formany of the above characteristics. Experimentalinvestigations (e.9., Mysen & Kushiro 1977) indicatethat silicate phases of a peridotite that suffered par-

tial melting with extraction of basaltic liquid becomeenriched in Mg, Cr and Ni, and depleted in Al, Naand Ca with respect to unmodified lherzolite. Ifresidual puidotite undergoc a furtler melting event,the newly formed liquids have higher Mg#, Ni andCr, and lower Na, Al and Ca than normal basalts.If residual peridotite is affected by contaminationbefore melting, additional changes may occur as afunction of the composition of the added material.We have shown that contamination with sedimentsof variable gomposition affected the source rocks ofOREN and KAM, the contaminant in the case ofOREN being less carbonate-rich than that added tothe source rocks of the KAM suite. Accordingly, wesuggest that the presence of a residual mantle in thesource zone of both OREN and KAM is responsiblefor the high lVtgf, Ni, Cr and Na contents of theinvestigated rocks, whereas the high Ca of KAMreflects the higher amount of this element presentin the sedimentary contaminant. In tlis view, ttre var-iation of CrA, Ni/V, and TTAI observed in theinvestigated OREN may indicate that the sourcerocks of these magmas underwent different degreesof partial melting prior to sontamination, and thatthe source peridotite of SIS minettes had a strongerresidual character than that of MVC and ORC rocks.

The hypothesis of a residual mantle does notchange significantly the results of previous calcula-tions, but only leads to a lowering of the amountsof crustal material that must be added to the perido-tite in order to model the Sr-isotope ratio.

ColqclusroNs

Ultrapotassic rocks having a kamafugitic andorenditic affinity are minor but important compo-nents of the Roman comagmatic province. Theserepresent distinct magma types with respect to theprevailing HKS and KS rocks. The OREN types aresilica-oversaturated, intermediate in composition,and have high Mg#, Ni and Cr together with extremeenrichment in LILE. KAM also have a high Mg#,Cr, Ni, and LILE, but are ultrabasic in compositionand are strongly silica-undersaturated. Ceochemical,isotopic, and experimental widence suggest that bothOREN and KAM can be derived by melting of ananomalous mantle that was affected by contamina-tion with crustal (probably sedimentary) materialdragged down by subduction processes which wereactive under the Italian peninsula during Late Ter-tiary times (Civetta et al. 1978).6sco1ding to themodel proposed, the sediments involved in subduc-tion were first dehydrated and then partially melted.Partial melting generated acidic liquids with higherHREE fractionation and higher LILEIHFSE thanthe parent rocks owing to the presence inthe residueofgarnet and some ascessory phases such as magne-tite and ilnenite, which have Kys much lower for

ORENDITIC AND KAMAFUCITIC ROCKS 63

LILE than for HFSE when they coexist with acidicmelts. Melted sediments rose through the uppermantle, reacted with peridotite, and stabilizedphlogopite, as suggested by Wyllie & Sekine (1982).Successive partial melting of phlogopite-peridotiteor phlogopite-olivine-pyroxenite at different pres-sures gave potassic liquids whose petological charac-teristics in terms of silica saturation \ryere determinedby the reaction products of phlogopite which changewithpressure (Wendlandt & Eggler 1980a,b). Somecompositional peculiarities, such as high Mg#, Ni,Cr and low Al in KAM, and high Ni, Cr, Mg# andlow Ca andNa in OREN are probably connected tothe residual nature of the mantle peridotite whichcould have suffered a previous melting event beforecontamination with sediments. The different Sr-isotope, CelSr and Rb/Srvalues betwe€n OREN andKAM are explained by the different relative amountsof silicate and carbonate sediments involved in sub-duction and mixing with the upper mantle.

The proposed model assumes that the OREN andKAM samples studied represent primary or poorlyevolved mantle-derived melts. This assumption issupported by the strong silica undersaturation ofKAM and by the high ferromagnesian-element abun-dance and Mg# in both KAM and OREN. This lat-ter evidence positively exsludes the possibility thatOREN are formed by interaction of KAM or HKSwith crustal rocks.

REITERENCES

Anu',n, M. & Epcen, A.D. (1983): High pressureexperimental studies on a katungite and their bear-ing on the genesis of some potassium-rich magmasof the western branch of the African Rift. "L Petrol-oB! A,l6Gl87.

Bacnnwn, S.W. & Scorr, R.B. (199): Rare earth andother trace element conents of minettes (mica lam-prophyres). Geochim. Costnochim. Acto 43, 93-100.

(1980): Reply to a critical commentby N.M.S. Rock. Geochim. Cosmochim. Acta &,t389-t392.

Benneru, F. & ItNoceutt, F. (1967): Le rocceselagitiche di Orciatico e Montecatini val di Cecina.Atti Soc. Toscana Sci. Not.74, 139-180.

BanroN, M. & Hauu-roN, D.L. (1979): The melringrelationships of a madupite from Leucite Hills,Wyoming, to 3OKb. Contr. Mineral. Petrolog1t69,133-t42.

Brccanwe, L., Dr Gnor-evo, P. & Snnru, G. (1985):High-K calc-alkaline, shoshonitic and leucitic vol-canism of Campania @oman Province, SouthernItaly): trace element constraints on the genesis ofan orogenic volcanism in a post-collisional exten-sional setting. 1985 IAVCEI ScientiJic Assembl!,Giardini Naxos, Italy (abstr.).

Bon$, S., FERRARA, G. & ToNaroncr, E. (1967): Deter-minazioni con il metodo K/Ar delle eta' delle roccemagmatiche della Toscana. Boll. Soc. Geol. Itol. M,40.3410.

CARMIcHAEI,, I.S.E. (1967): The mineralogy andpetrology of volcanic rocks from Leucite HiUs,Wyoming. Contr. Mineral, PetrologSt 15, A-66.

Cwerte, L., Dnr. CanunvE, P., MANrr:t, P., Pnc-cBRrLLo, A. & Polr, G. (1984): Petrology,geochemistry and Sr-isotope characteristics of thelavas from the area of Commenda. Bull. Volc. {1 ,581-595.

CwLeRs, R.L., Rauenr${.IAN, S., Brnrxosmv, P. &Gnrnn, T. (1985): Geochemistry and petrogenesisof lamproites, Late Cretaceous age, Woodson Coun-try, Kansas, USA. Geochim, Cosmochim Acta 49,r383-M42.

Cuxoenr, A. (1980): Role of subduction in the genesisof leucite-bearing rocks: facts or fashion? Reply toA.D. Edgar's discussion paper. Contr. Mineral.Petrologlt 73,432434,

Dr Peor.o, D.J. (t981): Trace element and isotopiceffests of combined wall-rock assimilaiion and frac-tional crystallizatiou Earth Planet. Sci. Lett. 53't89-202.

ACrotowr,sDcEtffiNTs -, INNo(ENn, F., MANEI'II, P., Pncrcnnuo' A. &Por.r, G. (1981): Geochemisal characteristics of

The authors thank Mrs. Lia Todaro, Massimo potassic-volcanics from Mt. Ernici (Southern

Menichini, and Giulio sbrana for their help durini ['atium' ltaly)' contr' Minerql' Petrologt1E'3141 '

ffiTrq,!$dff**#T'itro'ffi 'l.lilff :-_id"3ft F;f ,iffi ig,h j?lhfff fr :ing the Sr-isotope composition in some orenditic ;'ffi;d.1;;fn ire ,otatioo of the-Apennines.rocks. The sritical reading of Prof. A. Mottana, Dr. yoiun 4C,-AAIAA.R.F. Martin and of two anonymous referees greatlyimproved the style and the contents of the CoNsor.uacNo, G.J. & DnAxe, M.J. (1970: Equiva-manusctipt. The research was financed by the Italian lence of equations describing trace element distri-Ministry for Education (grant MPI 4090 on Tertiary bution during equilibrium partial melting . Geochim.and Quaternary volcanism). Cosmochim. Acta 4O' 142l-1422.


Dr Gnolauo, P. (1978): Geotectonic settings ofMiocene-Quaternary volcanism in and around theEastern Tyrrhenian sea border (Italy) as deducedfrom major element geochemistry. Bull. Yolc, 41,229-250.

Eocar, A.D. (1980): Role of subduction in the gene-sis of leucite-bearing rocks: discussio\. Contr,Mineral. Petrology 73, 429431,

Fennana, G., Pnerte-MenrrNEz, M., levlon, H.P.,ToNer.ru, S. & Tunr, B. (1980: Evidens€ forcrustal assimilation, mixing of magmas and a 87Sr-rich upper mantle. An oxygen and strontium iso-tope study of the M. Vulsini volcanic area, CentralItaly. Contr. Minerql Petrology 92,269-280,

Forrv, S.F., Tarron, W.R. & GnmN, D.H. (1980:The role of fluorine and orygen fugacity in the gen-esis of the ultrapotassic rocks. Contr. Mineral.Petrclogy 94, 183-192.

-, VErvruneur, G., Gnenx, D.H. & Tosceu, L.(1987): The ultrapotassic rocks: characteristics, clas-sification and constraints for petrogenetic models.Earth Sci. Reviews U,8l-134,

FonNesenr, M. (1985): Geochemistry of volcanic rocksfrom Umbria. Rend. Soc. Ital. Mineral. Petrologia40, 107-110.

Fnalcelarcr, L., Pnccrmr,r,o, A. & Por.r, G. (1987):Partition coefficients for mineral of potassic-alkalinerocks: data from the Roman province. Geochem.J. 2t, l-10.

Fnaxarr, M., Lrou, L. & Saura, M. (1972): A sim-ple method to evaluate the matrix effect in X-rayfluorescence analysis. X-ray Spectrom. l, l5l-154,

Gallo, F., Gnvmrr, F., Vmrunulr, L. &Vnnxn, L.(1984): The kamafugitic rocks of San Venanzo andCupaello. Neues Jahrb. Mineral. Monatsh. l9M,198-210.

Gnxnt.t. , G. & Pur<soou, M. (1979): An attempt atclassifying Tuscan Paleozoic: geochemical data.Mem. Soc. Geol. Italy ?fr,435446,

Gneuo, A., Dueuy, C. & Dosrer, J. (1987):Behaviour of trace elements during magmaticprocesses in the crust; application to acidic volcanicrocks of Tuscany (Italyr. Chem. Geo|.57,268-288,

GnacNaru, R. (1972): Le vulcaniti melilititiche diCupaello @ieti). Rend. Soc. Ital. Minerol. Petrolo-sia 7,8, 165-189.

Gnnnu, T.H. (1976): Experimental generation ofcordierite- or garnet-bearing granitic liquids froma pelitic composition. Geology 4, 85-88.

Holr"t, P.M. (1982): Mineral chemistry of perpotassiclavas of the Vulsinian district, the Roman province,Italy. Mineral. Mag. 6,379-386.

-, Lou, S. & NrnrseN, A. (1982): The geochemis-try and petrogenesis of the lavas of the Vulsiniandistrict, Roman province, Central Italy. Contr.Mineral. Petrologlt 80, 367-378.

& Muru<soe.mo, N.C. (1982): Evidence formantle metasomatism: an orygen and strontium iso-tope study ofthe Vulsitrian drstrict, Central ltaly.Earth Planet. Sci. Lett.60, 376-388.

& - (1986): Reply to: a qriticism of theHolm-Munksgaard oxygen and strontium isotopestudy of the Vulsinian distrist, Central ltaly. EarthPlanet. Sci. Lett. 78, 454y'.59.

Jeeuss, A.L., Lewrs, J.D., Srutrn, C.8., Gnnconv,G.P., Fpncusor, J., CHappgtt, B.W. & McCut-LocH, M.T. (1984): The diamond-bearingultrapotassic 0amproitic) rocks of the West Kimb€r-ley region, Western Australia. In Kimberlites (J.Kornprobst, ed.). Elsevier, Amsterdam, Vol. l,225-254.

KuEHr.En, S.M., Eocan, A.D. & ARIMA, M. (1981):Petrogenesis of the ultrapotassic rocks from tle Leu-cite Hills, Wyoming. Amer. Mineral. 66,663-677.

Lalrcruun, C.H., BBNoBn, J.F., BnNct, A.E., HeNsoN,G.N. & Tenon, S.R. (1977): Petrogenesis ofbasalts from Famous area: Mid-Atlanticidee. furthPlanet. Sci. Lett.36, 133-156.

MaNerrr, P., Pnccnnrr"lo, A. & Port, G. (1979): Rareearth element distribution in Jurassic siliceous rocksfrom Northern Apennines. Mineral. PetrologicaActa 23,87-98.

Mrrcnell, R.H. (1985): A review of the mineralogy oflamproites. Traw. Geol. Soc. S. Af,rica$,4ll437,

& BSLL, K. (1976): Rare earth elementgeochemistry of potassic lavas from Birunga andToro-Ankole regions of Uganda, Afuica. Contr.Mineral. Petrologlt 58, 293-303.

Mrr:sr"rpsRcnEn, M. (1965): Volcanism and petrogen-esis in the S. Venanzo area (Italy). Bull Volc, 28,85-94.

MysEn, B.O. & Kusnnro,I. (1977)t Compositional var-iations of coexisting phases with degree of partialmelting of peridotite in the upper mattle. Amer.Mineral. 62,843-865.

Nesn, W.P. & Cnrcnarr, H.R. (1985): Partitioncoefficients for trace elements in silicic magmas.Geochim. Cosmochim. Acta 49' 23094322.

Nrcrror.rs, I.A. & Rrxcwooo, A.E. (1973): Effect ofwater on olivine stabili8intholeiites and theproduc-tion of silica-saturated magmas in the island-arcenvironment. J. Geol.81, 285-300.

O'HARA, M.J. (197): Geochemical evolution duringfractional crystallization of a periodically refilledmagma chamber. Nature 266, 503-508.


Prancn,.J.A. (1982): Trace element characteristics oflavas from de,structive plate boundaries. Iz Ande-sites (R..S. Thorpe, ed.). John Wiley, Chichester,England, 525-548.

Fnccnrur.lo, A. (1985): Roman comagmatic province(Central Italy): evidence for subduction-relatedmagma genesis. Geology 13, 103-106.

& MeNrrrr, P. (1985): The potassium alkalinevolcanism of Central-Southern ltaly: a review of thedata relevant to petrogenesis and geodynarnic sig-nificance. Trans. Geol. Soc. S. AfricaE8,379-394.

Por.r, G. & Torovro, L. (198Q: Genesis, evo-lution and tectonic significance of K-rich volcanicsfrom the Alban Hills (Roman Comagmatic Region)as inferred from trace element geochemistry. Contr.Mineral. Petrologlt 86, 230-240.

Por-r, G. (1985): Phase composition in volcanic rocksfrom South Tuscany magmatic area (Central Italy):Inferences on condition of magma crystallization.Neues Jahrb. Mineral. Abh.75l, 14l-161.

-, Mamru, P., Pnccrr.ruo, A. & Ceccm, A.(1977): Determinazione di alcuni elementi del gruppodelle terre rare in rocce silicatiche per attivrzioneneutronica strumentale. Rend. Soc, Ital. Mineral.Petrologia 33, 7 55 -763.

Rocnns, N.W., HawrmwoRTH, C.J., Panren, R.J. &Marsu, J.S. (1985): The geochemistry of potassiclavas from Vulsini, Central Italy and implicationsfor the mantle enrichment processes beneath theRoman province. Contr. Mineral. Petrology 90,244-2s7.

Sauaue, T.G. (1974): Potassium-rich alkaline rocks.In The Alkaline Rocks. (H. Sorensen, ed.). JohnWiley, London.

Tenon, H.P., Turu, B. & CuNpam, A. (1984):180/160 and shemical relationship in K-rich vol-canic rosks from Australia, East Africa, Antarctica,and San Venanzo-Cupaello, ltaly. Eorth Planet, Sci,Lett. 69,2$416.

THorvnsoN, R.N. (1977): Primary basalts and magmagenesis. III. Alban Hills, Roman ComagmaticProvince, Central ltaly. Contr. Miner. Petrolog1t50,91-108.

Monnrsor.r, M.A., Hexonv, G.L. & PARRv,S.J. (1984): An assessment of the relative roles ofcrust and mantle in magrna genesis: an elementalapproach. Phil, Trans. Roy. Soc. London 310,549-590.

Tunr, B. & Tavr.on, H.P. (1976): Oxygen isotope

studies of potassic volcanic rocks of the RomanProvince, Central ltaly. Contr. Mineral. Petrology55 ,4 -31 .

-, & Fnnnana, G. (1986): A criticism ofthe Holm-Munksgaard oxygen and strontiumisotope study ofthe Vulsinian district, Central ltaly.Earth Planet, Sci, Lett. 78,447453.

VaN Bencnr, M.J., Grnzzq C. & Rrccr, C.A. (1983):Minette inclusions in the rhyodacitic lavas of Mt.Amiata (Central Italy): mineralogical and chemicalevidence of mixing between Tuscan and Roman typemagma. J. Volc. Geotherm, Res. 19, l-35.

Var KoorrN, G.K. (1980): Mineralogy, petrology andgeochemistry of an ultrapotassic basaltic suite, cen-tral Sierra Nevada, California, USA. ,f. Petrologlt2t,65t-684.

Veloe, D. (1967): Sur un lamprophyre hyperalcalinpotassique: la minette de Sisco (Corse). Bull, Soc.frang. Mineral. Crist, 90,21+223,

Verrunrl-u, G., CapsoRr, S., Dr Berrnrwr, G., Cnaw-FoRD, A., Kocarco, L.N. & Crlesrnu, S. (1984):The ultrapotassic rocks from southeastern Spain.Lithos 17.37-54.

VoLLr,rER, R. (1977): Isotopic evidence for genetic rela-tions between acid and alkaline rocks inltaly. Contr,Mineral. Petrology 60, 109-118.

WacNrn, C. & Velor, D. (1980: The mineralogy ofK-richterite-bearing lamproites. Amer. Mineral. 71,17-37.

WnrolaNor, R.F. & Eccmn, D.H. (1980a): The ori-gin of potassic magmas: l. Melting relationsin the system KAlSiOa-Mg2SiOo-SiO2 andKAlSiO.a-MgO-SiO2-CO2 to 30 kb. Amer. J. Sci.2l,0,385420.

& - (1980b): The origin of Potassicmagma. 2. Stability of phlogopite in natural spinellherzolite and in the system KAlSiOa-MgO-SiO2-H2O-CO2 at high P and T. Amer. J. Sci. 280'421458.

Wool, A.D. (1979): A variably veined suboceanicupper mantle. Genetic significance for mid-oceanridge basalts from geochemical evidence. Geolog!7,499-503,

Wmrs, P.J. & SnKNE, T. (1982): The formation ofmantle phlogopite in subduction zone hybridization.Contr. Mineral. Petrology 79, 375-380.

Received December 23, 1986; revised manuscriptaccepted May II, 1987.

petrogenesis of orenditic and kamafugitic rocks from central italy 45

Documents