metasomatism and partial melting in upper-mantle …...journal of petrology volume 43 number 9 pages...
TRANSCRIPT
JOURNAL OF PETROLOGY VOLUME 43 NUMBER 9 PAGES 1749–1777 2002
Metasomatism and Partial Melting inUpper-Mantle Peridotite Xenoliths from theLashaine Volcano, Northern Tanzania
J. B. DAWSON∗DEPARTMENT OF GEOLOGY AND GEOPHYSICS, UNIVERSITY OF EDINBURGH, EDINBURGH EH9 3JW, UK
RECEIVED APRIL 10, 2001; REVISED TYPESCRIPT ACCEPTED MARCH 22, 2002
REE (LREE) into the glass with respect to clinopyroxene, but theA group of chrome-spinel peridotite upper-mantle xenoliths from thereverse for Sr and Y, in both mica-bearing and mica-free parageneses.Lashaine volcano, northern Tanzania, differs from other xenolithsIn a mica-bearing melt pocket, Rb and Ba partition preferentiallyat this locality in containing glassy melt pockets. Modal, mineralinto mica relative to both clinopyroxene and glass. Similar patternschemical and isotopic evidence indicates that, before the melting thatof partitioning exist in a similar xenolith from Labait (anotherwas coincident with the xenolith entrainment and eruption in theTanzanian xenolith locality) except that Sr is highly concentratedPleistocene, the sub-Tanzanian mantle lithosphere had a complexin the glass; and the glass and mica contain high Ba. Comparisonhistory. A major element depletion at >3·4 Ga gave rise to aof the chemistry of the melt pockets in the xenoliths from Lashainehigh-olivine restite protolith, and this was followed by an episodeand other northern Tanzanian volcanoes (Labait and Olmani)of K, Fe, Ca, Ti and Rb metasomatism at>2·0 Ga (Metasomaticindicates that the melting in each xenolith suite was accompaniedEvent I) resulting in the formation of well-equilibrated Cr-diopsideby a distinctive metasomatic influx.and phlogopite. A later episode of metasomatism (Metasomatic
Event II), first reported here, is recognized by texturally andchemically non-equilibrated titanian phlogopite, Ti–Cr-diopside,enstatite–bronzite, ilmenite and rutile, interpreted as due to an influx
KEY WORDS: Tanzania; mantle xenoliths; metasomatism; melting; elementof K, Fe, Ca, Ti, Nb and Ta. The recognition of two episodes ofpartitioningmetasomatism reported here provides an explanation for previously
recognized major differences in the isotopic chemistry of Lashaineperidotite diopsides. Immediately before entrainment and eruptionin the Pleistocene, some of the peridotites underwent partial melting
INTRODUCTIONwith the formation of melt pockets and veins containing vesiculatedglass, olivine, diopside, phlogopite, spinel (sensu lato), calcite, Although small-volume melts generated at the onset ofapatite and zeolite. Compared with most previously reported mantle mantle melting rarely reach the surface, incipient meltingglasses, the Lashaine glasses are potassic, rather than sodic, contain can be investigated by the study of glassy melt pocketsrelatively low amounts of SiO2, Al2O3 and total alkalis, and are in upper-mantle peridotites that, in effect, are small poolsnot in equilibrium with peridotite olivine. The melting event was of frozen embryonic magma. The purpose of this paperaccompanied by an influx of K, Ti, Ca, Rb, Sr, Ba, Zr, the rare is to document the phase chemistry of partially meltedearth elements (REE), H2O and (inferred) CO2, which may be upper-mantle peridotite xenoliths from the Lashaine vol-the culmination of Metasomatic Event II. Hence the melting cano, northern Tanzania (3°22′S, 36°26′E). Lashaineis interpreted as being metasomatically triggered combined with and the other xenolith localities mentioned in this paperdecompression melting. The computed composition of one of the (Fig. 1) are minor tuff cones that form part of the so-larger melt pockets is most similar to olivine lamproite. Ion microprobe called Younger Extrusives—an episode of Pleistoceneanalyses of glass, diopside and mica in two of the melt pockets volatile-rich, highly explosive alkaline volcanic activity in
northern Tanzania (Dawson, 1992).show preferential partitioning of Rb, Zr, Nb, Ba, Pb and the light
∗Telephone: 0131 650 7286. Fax: 0131 668 3184. E-mail:[email protected] Oxford University Press 2002
JOURNAL OF PETROLOGY VOLUME 43 NUMBER 9 SEPTEMBER 2002
particular paradox was the recognition of a bimodaldistribution in the Sr, Nd and Pb isotopic compositionsof clinopyroxenes from Lashaine lherzolites, possiblysignalling two mantle events widely separated in time(Cohen et al., 1984).
A further complexity in some Lashaine xenoliths isexplored here. First reported by Reid et al. (1975), anumber of the xenoliths exhibit textures indicative ofsmall amounts of partial melting in the form of smallfrozen melt pockets and inter-grain veinlets that containglass and precipitated phases. This paper provides phasedata on both the peridotite protolith and melt-pocketphases in seven of these partly melted xenoliths, anddiscusses the process(es) that may have triggered thepartial melting. In an assessment of possible regionaldifferences, the Lashaine xenoliths are compared withpartly melted peridotite xenoliths occurring at two othernorthern Tanzanian xenolith localities: Olmani (3°24′S,36°45′E),>40 km east of Lashaine, and Labait (4°35′S,35°26′E), some 150 km to the SW (Fig. 1).
Fig. 1. Map of northern Tanzania, showing xenolith localities (star PETROGRAPHYornament) mentioned in the text. TV, areas of Tertiary volcanics. The xenoliths are rounded to discoidal in shape, and
have maximum dimensions up to 25 cm. The range ofmineralogy is given in Tables 1 and 2. In the followingSince their discovery in 1961 (Dawson, 1964), thediscussion ‘primary’ refers to the well-equilibrated, un-lower-crustal and mantle xenoliths embedded in thezoned olivine, pyroxenes and spinel (sensu lato) that formankaramitic and carbonatitic tuffs of the Lashainethe bulk of the rocks, whereas ‘metasomatic’ refers tocratered tuff-cone have provided valuable insights into thephases replacing or rimming the primary phases; andmakeup of the continental lithosphere beneath northernwhich may be zoned or of variable composition. AsTanzania. An early xenolith collection has been ex-will be discussed below, the primary phases represent atensively studied by Dawson and others (Dawson &combination of restite phases—olivine, chromite andPowell, 1969; Dawson et al., 1970; Hutchison & Dawson,possibly enstatite, together with phlogopite and Cr-di-1970; Reid & Dawson, 1972; Dawson & Smith, 1973;opside that result from a later metasomatic overprint; asReid et al., 1975; Rhodes & Dawson, 1975; Ridley &a result of annealing the only extant evidence for theDawson, 1975; Cohen et al., 1984; Dawson, 1987; Burtonmetasomatism is the isotope chemistry of the equilibratedet al., 2000; Schiano et al., 2000), and other collectionsphlogopite and diopside.have been studied by Pike et al. (1980), Nielson (1989),
The modes were made by computer scanning of min-Henjes-Kunst & Altherr (1992) and Rudnick et al. (1994).eral distribution in traced overlays of A4-size photographsThese studies have provided abundant petrographic,of thin sections, using ‘Scion Image’ image analysismineral chemical and whole-rock geochemical data,software (this may be downloaded free of charge fromshowing that (1) the upper-mantle rocks, which comprisehttp://www.scioncorp.com). The modes show the dom-garnet-bearing and garnet-free dunites, harzburgites,inance of olivine but the volumes of minor phases andlherzolites, wehrlites and pyroxenites, are unserpentinizedmelt pockets vary widely even in thin sections prepared(unlike similar xenoliths from kimberlites) and hencefrom the same slab; for example, the modes of tworetain valuable intra-grain and grain-boundary texturalsections from the same slab cut from sample 747 varyinformation; (2) a depletion event, resulting in high Mg/from harzburgite containing 7·79% melt to mica-bearingFe and low Ca/Al ratios was overprinted by later meta-harzburgite containing only 1·1% melt (Table 2 and Fig.somatism, which may correlate with wall-rock–intrusive2). Specific difficulties in measuring the volumes of meltvein relationships seen in some specimens; (3) the suitepockets or veins with light optics arise from: (1) opticallyequilibrated along a 44 mW/m2 geothermal gradient.continuous overgrowth of melt-pocket phases on primaryIn summary, the present-day textures, mineralogy and
chemistry of the suite testify to a complex history. A phases at the margins of the melt pockets and on relict
1750
DAWSON NORTHERN TANZANIAN PERIDOTITE XENOLITHS
Tab
le1
:Su
mm
ary
ofph
ases
inL
asha
ine
peri
dotite
s
Sam
ple
Ro
ckty
pe
Oliv
ine
Op
xC
px
Mic
aO
xid
e
mg
mg
mg
cr
Pe
rid
oti
tep
ha
se
s
747
Ch
rom
ite
har
zbu
rgit
e92
93C
r-d
iop
sid
ere
pla
ces
chro
mit
e91
,re
pla
ces
chro
mit
eP
rim
ary
Mg
-ch
rom
ite
80
771
Ch
rom
ite
lher
zolit
e91
93P
rim
ary
Cr-
dio
psi
de
×P
rim
ary
Mg
-ch
rom
ite
84–9
0
542
Ch
rom
ite
har
zbu
rgit
e90
–92
(i)
95,
pri
mar
y(i
)p
rim
ary
Cr-
dio
psi
de
×P
rim
ary
Mg
-ch
rom
ite
65–6
9
(ii)
92,
hig
h-A
laf
ter
gar
net
(ii)
met
aso
mat
ic,
rep
lace
sen
stat
ite
1544
Mic
ah
arzb
urg
ite
88–9
1(i
)p
rim
ary
91P
rim
ary
Cr-
dio
psi
de
rep
lace
db
yP
rim
ary
89;
(i)
pri
mar
yM
g-c
hro
mit
e85
(ii)
met
aso
mat
ic,
rep
lace
scp
xo
px
hig
h-T
iri
ms
(ii)
lam
ella
rch
rom
ite
95ex
solv
ed
fro
milm
enit
e
(iii)
Mg
–Cr
ilmen
ite
(iv)
Nb
–Ta
ruti
le
3926
Ch
rom
ite
har
zbu
rgit
e90
–94
(i)
pri
mar
y92
Pri
mar
yC
r-d
iop
sid
e×
(i)
pri
mar
yM
g-c
hro
mit
e80
–87
(ii)
91,
rep
lace
scp
x×
(ii)
inte
rgro
wn
wit
h(i
i)o
px
Mo
reFe
-ric
hsa
mp
les
750
Ch
rom
ite
lher
zolit
e85
–90
88,
rep
lace
scp
x(i
)p
rim
ary
Cr-
dio
psi
de
×(i
)p
rim
ary
Mg
-ch
rom
ite
89
(ori
gin
ally
weh
rlit
e)(i
i)‘b
leac
hed
’C
r-d
iop
sid
e(i
i)M
g–C
rilm
enit
e
1546
Ch
rom
ite
har
zbu
rgit
e86
–91
(i)
83,
pri
mar
yP
rim
ary
Cr-
dio
psi
de
×(i
)p
rim
ary
Mg
-ch
rom
ite
(ii)
88,
rep
lace
scp
x(i
i)M
g–C
rilm
enit
e
Sam
ple
Oliv
ine
Cp
xM
ica
Oxi
des
Gla
ssZ
eolit
eO
ther
s
mg
mg
cr
Me
lt-p
ock
et
ph
ase
s
747
93D
iop
sid
e91
–92
Mg
-ch
rom
ite
75+
+A
pat
ite
lines
vesi
cles
and
isin
clu
ded
inze
olit
e
771
+D
iop
sid
e×
Mg
-ch
rom
ite
90+
+
1542
90D
iop
sid
e×
(i)
zon
edC
r-sp
inel
30–4
5+
+
Au
git
e(i
i)at
oll
spin
el40
1544
91D
iop
sid
e,86
,h
igh
-Ti
(i)
Mg
-ch
rom
ite
74–9
5(i
)‘n
orm
al’
+
En
dio
psi
de
(ii)
Nb
–Ta
ruti
le(i
i)v.
hig
h-F
e
(iii)
hig
h-M
g
3926
93D
iop
sid
e89
,lo
w-T
iM
g-c
hro
mit
e83
++
Sr-
calc
ite
incl
ud
edin
chro
mit
e
Cal
cite
and
wit
her
ite
inve
sicl
es
Mo
reFe
-ric
hsa
mp
les
750
90D
iop
sid
e85
–88,
hig
h–T
iM
g-c
hro
mit
e92
++
Ap
atit
ein
vesi
cles
1546
91D
iop
sid
e86
–89
Mg
-ch
rom
ite
70–9
4+
+
mg=
Mg
/(M
g+
Fe2+
)×
100,
exce
pt
inm
ica
wh
ere
mg=
Mg
/(M
g+
tota
lFe
)×
100;
cr=
Cr/
(Cr+
Al)×
100;×
,ab
sen
t;+
,p
rese
nt.
1751
JOURNAL OF PETROLOGY VOLUME 43 NUMBER 9 SEPTEMBER 2002
Table 2: Modes of Lashaine peridotites (vol. %)
Sample Olivine Opx∗ Cpx∗ Mica Opaques† Melt pocket‡
747A 89·4 2·8 0·0 0·0 0·0 7·8
747B 97·7 0·0 0·1 0·3 0·8 1·1
750 85·7 2·7 8·5 0·0 0·3 2·8
771 87·2 6·8 4·8 0·0 0·3 0·9
1542§ 85·5 5·8 2·1 0·0 0·2 2·2
1544 88·0 0·7 1·8 2·4 1·5 5·6
1546 87·4 2·2 3·8 0·0 0·3 6·3
3926 83·3 4·7 0·7 0·0 0·5 10·8
∗Primary and metasomatic pyroxenes not differentiated.†Mainly spinels except in 1544 (rutile and ilmenite included).‡Includes glass, precipitated phases, relict phases and vesicles.§Also contains 4·2% orthopyroxene–spinel symplectite (former garnet).
Fig. 2. Photomicrographs of two thin sections (a and b) of sample 747, showing differences in the volumes of melt pockets (arrowed). Note alsothe presence of orthopyroxene (‘o’) in (a) [absent in (b)] and the presence of chrome-spinel (‘s’) in (b). Scale bar represents 10 mm. (See alsoTable 2.)
primary phases in the melt pockets; (2) glass de- The mineral combinations in Table 1 classify most ofthe rocks as harzburgites; only one contains more thanvitrification.
The modes in Table 2, combined with the bulk chem- the 5% modal clinopyroxene required to be classifiedas lherzolite, according to the IUGS recommendation.ical data of Rhodes & Dawson (1975), show that the
peridotites studied here are more refractory than the Moreover, in this sample 750 (the author’s collection BDprefixes are dropped, for brevity), originally a wehrlite,Lashaine garnet peridotites. Modally they contain con-it is the occurrence of metasomatic bronzite that permitssiderably less enstatite (all <15%, most <8%) than theit to be classified as lherzolite.garnet lherzolites, which typically contain >10% and
often >20% modal orthopyroxene. Thus they are notdifferent mineralogical expressions of the same com-positions in response to equilibration at different levels
Peridotite petrographyin the upper mantle. Nor are they chemically equivalentto oceanic Al-spinel peridotites because, as shown below The rocks are dominated by olivine and, although petro-
graphically indistinguishable, on the basis of their phaseand earlier by Reid et al. (1975) and Rudnick et al. (1994),the spinels are chromites that contain substantially higher chemistry there are two chemically different groups of
peridotite. One group, comprising five of the sevenCr and lower Al.
1752
DAWSON NORTHERN TANZANIAN PERIDOTITE XENOLITHS
Fig. 3(a)–(d). See p. 1755 for explanation.
xenoliths studied, contains Mg-rich phases that are com- rounded, up to 2 mm and optically homogeneous, al-positionally similar to those in most upper-mantle peri- though electron probe micro-analysis (EPMA) has showndotites, whereas in the second group the phases are more small core–rim compositional differences in some grains.iron rich.
Metasomatic and unequilibrated phasesPrimary phasesMetasomatic orthopyroxene, forming replacement co-Primary olivine grains are up to 8 mm, have straightronas round primary Cr-diopside in samples 750 andmutual grain boundaries and are frequently strained and1544, contrasts with the primary enstatite by being inexhibit kink banding. The olivine margins often havesmaller (2 mm) light brown, turbid grains that formeuhedral projections into melt pockets (Fig. 3a–d), andaggregates with small (<0·5 mm) rounded grains ofrelict, partly resorbed olivine may occur as ‘islands’ inchromite and dark brown mica.the melt pockets. The primary orthopyroxene is also in
Metasomatic diopside rimming and replacing chromite8 mm grains but does not show the same degree ofin 747 is the same green colour as primary Cr-diopside,deformation as the olivines. Primary Cr-diopside is gen-whereas that replacing orthopyroxene in 1542 is col-erally of smaller grain size (up to 5 mm) than the olivineourless. Metasomatic mica forming rims of 1–2 mmand enstatite; it is green in colour but adjacent to replacingthickness on chromite grains is generally unzoned.metasomatic orthopyroxene, and where partly melted, it
Two specimens contain ilmenite. Replacement texturesis often colourless (‘bleached’), reflecting a change inare absent but, in view of the fact that such a high-chemistry (described below). Primary light brown micaFe–Ti mineral would not have survived the melting eventis up to 5 mm, and shows undulose extinction; it generally
has darker brown metasomatic rims. The spinels are inferred from the otherwise highly refractory mineralogy
1753
JOURNAL OF PETROLOGY VOLUME 43 NUMBER 9 SEPTEMBER 2002
Fig. 3(e)–(h).
of the samples, they are interpreted as being of meta- symplectites in sample 1542 are interpreted as completelyreplaced garnets.somatic origin. The ilmenite grains are rounded, up to
2 mm and mainly optically homogeneous; ilmenite in1544 contains exsolved lamellae of Mg–Ti chromite (Fig.3e), and the ilmenite itself is partly replaced by rutile. Petrography of the melt pockets and veinsAlso in this sample, rounded rutile grains are included The largest melt pocket found (in 3926) measures 7 mmin olivine and enstatite, and some larger (150 �m) grains × 3·5 mm; some are rounded but others are irregularhave exsolved ilmenite and fine lamellae of a NbTaCr with apophyses pinching out into intergranular veinlets.titanate. Ilmenite also occurs in enstatite–phlogopite– Most veinlets are of the order of 0·1–0·5 mm wide; thoseilmenite coronas replacing primary Cr-diopside in originating in melt pockets rarely persist for more thananother Lashaine specimen (Dawson, 1987). 5 mm but, exceptionally, persist for >25 mm (the width
Sample 1542 is unlike the others in containing rounded of a thin section). In detail, the margins of melt pocketsclusters up to 3 mm of dominant 100 �m high-Al enstatite are highly irregular (Fig. 3c), mainly as a result of crystalsgrains and rarer 30 �m high-Al diopsides that are sym- protruding into the melt pockets; these may be eitherplectitically intergrown with euhedral to vermiform grains the irregular ends of corroded primary phases or theup to 80 �m of high-Al spinel; all three phases are euhedral faces of new phases precipitating on the melt-compositionally distinct from the primary pyroxenes and pocket margins (Fig. 3d).Cr-spinel, and, as individual grains can vary widely in All the primary phases have been corroded at the melt-composition, they are regarded as unequilibrated. By pocket margins and occur as rounded or irregularlyanalogy with other Lashaine xenoliths containing pyrope shaped relicts in the melt pockets; however, comparedgarnet in varying stages of replacement by similar pyr- with other primary phases, enstatite is relatively un-
corroded. In the melt pockets the olivine occurs asoxene–spinel intergrowths (Reid & Dawson, 1972), the
1754
DAWSON NORTHERN TANZANIAN PERIDOTITE XENOLITHS
Fig. 3. Scanning electron microscope back-scattered electron images of textures and other features in Lashaine peridotites. (a) Melt pocket insample 1542. Note the ‘necklace’ of bright spinel grains apparently embedded in the primary olivines surrounding the melt pocket, but see (b).Note also their presence in the overgrowth margins of the large relict olivine grain (ROL) in the melt pocket. (b) Enlargement of left-hand partof (a) showing the location of bright spinel grains at the contact between primary olivine and precipitated olivine overgrowing the primaryolivine. Note the euhedral shape of the overgrowth olivine. A euhedral grain of precipitated melt-pocket olivine also has inclusions of spinel. (c)Contact between (left) a primary olivine grain (OL) and (right) a melt pocket containing, amongst others, grains of mica and diopside (CPX).Note the dark margin due to overgrowth of more forsteritic olivine, the darkness resulting from a lower atomic weight, in this case owing tolower Fe content. Sample 3926. (d) Enlargement of lower part of (c), illustrating the compositional difference between the primary olivine (Fo90)and the darker overgrowth (Fo94) (see Table 3, analyses 10 and 11). Ze, zeolite. (e) Ilmenite (IL) partially replaced by rutile (RUT); see analysesin Table 8. The ilmenite has exsolved bright lamellae of Mg–Ti chromite (Table 7, analysis 11). Sample 1544. (f ) Representative view of a meltpocket, showing euhedral diopside (CPX), bladed mica, small euhedral spinels (S), and vesiculated glass (G). Vesicles (black) are lined with zeolite(Z) and sometimes calcite (Cc). Peridotite olivine is at the left edge of the image. Sample 750. (g) Melt-pocket diopside overgrowing relict grainsof primary enstatite. Note also clusters of ‘atoll’ spinel. Sample 1542. (h) Relict primary chromite in a melt pocket, overgrown by melt-pocketspinel. Sample 1542. (i) Zoned melt-pocket spinel overgrown by ‘atoll’ spinel in sample 1542. Points A–D refer to analyses 3–6, Table 10. ( j)Bright euhedral grains of melt-pocket rutile (Table 10, analyses 16 and 17); in sample 1544. (k) Melt pocket in sample 1542, marginally linedby glass containing relatively low concentrations of Fe, in which are bright, rounded (?immiscible) blebs of very-high-Fe glass (see analyses 7and 8, Table 11). Zeolite (Z) lines vesicle.
epitaxial overgrowths on wall-rock olivine grains, or on phase and occurs as euhedral microphenocrysts up to300 �m (Fig. 3f ), occasionally adhering to the melt-rounded ‘islands’ of corroded primary olivine. Being
more magnesian than the primary olivine (see Mineral pocket walls, and sometimes overgrowing wall-rock orcorroded relict grains of Cr-diopside or enstatite (Fig.Chemistry, below), the overgrowths are slightly darker
in BSE images (Fig. 3a and b) and, in addition, there 3g). The mica occurs as euhedral 300 �m randomlyoriented platelets. The spinels generally are euhedral andare often chains of spinel at the primary grain–overgrowth
interface (Fig. 3a). Most microphenocrysts in the melt up to 100 �m, with the larger ones often having cores ofrelict primary Cr-spinel (Fig. 3h). Some spinels are zonedpockets have cores of relict primary olivine, but more
rarely have glass inclusions elongate parallel to their long (Fig. 3i) and may be overgrown by discontinuous chains(‘atolls’) of a compositionally different spinel, which mayaxes. Clinopyroxene is the most abundant precipitated
1755
JOURNAL OF PETROLOGY VOLUME 43 NUMBER 9 SEPTEMBER 2002
also occur within glass or overgrown by diopside (Fig. Geophysics at the University of Edinburgh. Meas-urements were made with an 8 nA 16O primary beam3g). In sample 1544, some melt pockets contain euhedral
30 �m grains of rutile (Fig. 3j). The glass itself is invariably with a net impact of 15 keV, and focused to a spot of>15 �m diameter. Molecular ion interferences werevesicular (e.g. Fig. 3g) and sometimes partly devitrified.
In most melt pockets there are no visible variations in discriminated using an energy filtering technique (Zinner& Crozaz, 1986). Intensities of all masses were measuredthe glass, but in one melt pocket in sample 1544 two
different glasses can be seen in back-scattered electron over 10 cycles for each analysis point, with an aquisitiontime of 10 s per cycle. The ion intensities were normalizedimages (Fig. 3k). The vugs are usually lined with zeolite
(Fig. 3g); in many instances, the vugs are surrounded by to Si. Corrections were made for overlaps of rare earthoxides (MO+); the BaO overlap on Eu was calculatedfresh glass, which, combined with an absence of alteration
in the other protolith and melt-pocket phases, strongly assuming that excess counts at mass 154 were 138Ba +16O.suggests that the zeolites are a late-stage magmatic phase,
rather than a precipitate from percolating groundwater.In a few cases, apatite and calcite occur within thefringing zeolite and, in one sample (3926), tiny 20 �m Peridotite phase chemistrygrains of witherite (BaCO3) are present.
Olivine
Olivines in the more abundant magnesian peridotites(Table 3) are highly magnesian (Fo89·2–92·8) and contain
MINERAL CHEMISTRY OF THE minor MnO (0·12–0·17 wt %) and NiO (0·12–0·42 wt %).Other detectable elements (Ti, Al, Cr and Ca) arePERIDOTITESgenerally in concentrations of <0·04 wt %. Olivines inAnalytical methodsthe more Fe-rich peridotites are Fo85–86 with slightlyThe minerals were analysed for major and minor ele-higher MnO (0·18–0·19 wt %); the other oxides are inments by wavelength-dispersal analytical techniques onsimilar concentrations.a Cameca Camebax electron microprobe at the Uni-
versity of Edinburgh. Details of operating conditions andOrthopyroxenestandards have been given by Dawson & Hill (1998).
Most phases were analysed with a spot beam of >1 �m Primary orthopyroxenes in the more magnesian peri-dotites (Table 4, analyses 1, 2, 3, 10 and 12) are enstatitesat 20 kV and 20 nA, but under these conditions glasses
and zeolites were unstable and so were analysed at 10 (mg 91·3–94·4) with <3 wt % diopside. Al2O3 is generally<1 wt %, though up to 1·42 wt % in the former garnetnA with the beam rastered over a 12·5 �m square, and
Na and K were analysed early in the routine to reduce harzburgite (sample 1542). Most enstatites contain CaOin the range 0·4–0·6 wt %; the enstatite in harzburgitealkali loss or migration. Compared with published ana-
lyses of other mantle glasses, the alkali contents of the 747 contains only 0·38 wt % CaO, which is similar tothat found in enstatites from refractory harzburgites inLashaine glasses were found to be low, so specific ex-
periments have been carried out to check for alkali loss. kimberlites (Hervig et al., 1980). Relict areas of enstatitebeing replaced by metasomatic diopside in 1542 (TableSpots in glasses in two specimens, 1546 and 3926, were
analysed over 25 3-s cycles at 20 nA (i.e. at more extreme 4, analyses 4 and 5) contain higher Ti, Cr, Al and Ca,but are more magnesian (mg 95 vs 94) than the primaryconditions for potential alkali loss than in the routine
analyses); over the analysis period, decrease in the number enstatite. In the more Fe-rich peridotite 1546, the primaryorthopyroxene (Table 4, analysis 15) is bronzite (mg 83)of counts for both Na and K was statistically insignificant.
As a further check, analyses were made on natural glasses that, compared with the other primary orthopyroxenes,contains lower Mg and Al but higher Fe, Ca and Mn.under similar conditions. Alkali loss from basaltic tephra
was again insignificant, whereas with hydrated rhyolite Metasomatic enstatites replacing Cr-diopside in 1544and 3926 (Table 4, analyses 11 and 13) fall within thetephra there was a 40% and 75% reduction in the
number of counts for K and Na, respectively, over compositional range of the primary enstatites, exceptfor higher CaO. In the more Fe-rich specimen 750,the analysis period. Thus, the low alkali contents in
the Lashaine glasses are considered to be real, and metasomatic bronzite replacing primary Cr-diopside(analysis 14), is less magnesian than the texturally similarnot the result of loss or migration caused by electron
beam excitation. metasomatic enstatites in the more magnesian peridotites(mg 86·5 vs >92).Analyses for Rb, Sr, Y, Zr, Nb, Ba, Pb and rare earth
elements (REE) in clinopyroxenes, mica and glasses were Enstatite in the pyroxene–spinel symplectites in 1542(interpreted as retrograded garnet) (Table 4, analysesmade in situ on polished, gold-coated thin sections by
secondary ion mass spectrometry using a Cameca imf- 6–9) as well as being a high-Al variety is also of variablecomposition (see analyses 6 and 7); spinel-free parts of4f ion microprobe in the Department of Geology and
1756
DAWSON NORTHERN TANZANIAN PERIDOTITE XENOLITHS
Tab
le3
:R
epre
sent
ativ
ean
alys
esof
oliv
ines
inL
asha
ine
peri
dotite
san
dm
elt
pock
ets
Sam
ple
:74
777
115
4215
4439
2675
015
46
Ph
ase:
po
mp
pp
om
pm
pg
cm
pg
rp
om
pp
om
pm
pg
crp
om
pp
om
pM
pg
cr
An
alys
is:
12
34
56
78
910
1112
1314
1516
17
SiO
240
·740
·940
·740
·340
·240
·140
·240
·140
·239
·941
·140
·839
·139
·839
·440
·240
·2
TiO
20·
030·
040·
010·
030·
040·
050·
050·
040·
040·
040·
040·
030·
040·
040·
030·
050·
04
Al 2
O3
0·01
0·05
0·01
0·05
0·05
0·06
0·05
0·02
0·14
0·02
0·05
0·03
0·03
0·02
0·01
0·03
0·04
Cr 2
O3
0·02
0·15
0·07
0·06
0·11
0·19
0·18
0·02
0·04
0·05
0·13
0·14
0·07
0·19
0·06
0·11
0·12
FeO
7·54
6·45
9·03
7·30
9·86
9·15
10·1
10·5
8·87
10·3
6·02
6·71
13·9
9·88
13·4
9·05
10·6
Mn
O0·
150·
090·
130·
120·
230·
210·
220·
150·
140·
150·
110·
140·
190·
170·
180·
150·
16
Mg
O51
·052
·049
·451
·049
·249
·449
·148
·749
·848
·952
·351
·345
·648
·046
·249
·348
·4
NiO
0·15
0·09
0·27
0·32
0·18
0·21
0·17
0·36
0·34
0·35
0·35
0·30
0·29
0·36
0·36
0·38
0·22
CaO
0·03
0·14
0·04
0·03
0·25
0·17
0·22
0·04
0·12
0·03
0·14
0·12
0·05
0·17
0·04
0·16
0·12
Tota
l99
·87
100·
2299
·66
99·2
110
0·12
99·5
410
0·29
99·9
399
·67
99·7
410
0·24
99·5
899
·27
99·4
399
·68
99·4
399
·90
mg
92·3
93·4
90·8
92·5
89·9
90·5
89·8
89·2
90·9
89·5
93·9
93·1
85·4
89·6
86·3
90·6
89·0
mg=
Mg
/(M
g+
Fe2+
)×
100
(in
this
and
sub
seq
uen
tta
ble
s,va
lues
init
alic
s,e.
g.m
g,c
r,al
,are
mo
lecu
lar)
.p,p
rim
ary;
om
p,o
verg
row
thin
tom
elt
po
cket
;mp
gc,
mel
t-p
ock
etg
rain
core
;m
pg
r,m
elt-
po
cket
gra
inri
m;
mp
gcr
,u
nzo
ned
mel
t-p
ock
etg
rain
.
1757
JOURNAL OF PETROLOGY VOLUME 43 NUMBER 9 SEPTEMBER 2002
Tab
le4
:R
epre
sent
ativ
ean
alys
esof
peri
dotite
orth
opyr
oxen
es
Sam
ple
:74
777
115
4215
4439
2675
015
46
An
alys
is:
12
34
56
78
910
1112
1314
1516
SiO
257
·557
·057
·055
·651
·554
·551
·849
·3→
52·6
56·4
56·4
57·0
57·2
55·9
56·3
57·0
56·3
TiO
20·
020·
200·
100·
140·
250·
150·
130·
07→
0·24
0·28
0·25
0·10
0·17
0·22
0·20
0·23
0·15
Cr 2
O3
0·39
0·60
0·52
1·38
1·12
0·87
1·84
1·53→
2·32
0·35
0·60
0·75
0·34
0·46
0·44
0·50
0·49
Al 2
O3
1·28
0·86
1·42
2·38
8·33
4·33
7·88
5·92→
10·6
1·02
0·96
0·35
0·38
0·31
0·22
0·30
0·44
Fe2O
30·
000·
760·
980·
991·
510·
901·
150·
76→
1·57
1·04
1·23
1·17
0·95
0·00
0·14
0·25
0·00
FeO
4·51∗
4·80
3·56
3·36
3·24
3·98
4·22
3·60→
4·77
5·75
5·18
4·39
5·42
8·82∗
11·1
8·95
8·18∗
Mn
On
.d.
0·17
0·12
0·19
0·21
0·24
0·31
0·27→
0·34
0·15
0·17
0·16
0·17
0·24
0·24
0·22
0·21
Mg
O35
·634
·635
·433
·331
·632
·930
·228
·6→
31·4
33·8
33·8
35·2
34·5
31·9
30·6
32·3
32·9
NiO
n.d
.0·
100·
080·
060·
050·
020·
020·
00→
0·03
0·19
0·11
0·08
0·09
0·10
0·05
0·08
0·05
CaO
0·38
0·65
0·42
2·07
1·16
1·44
1·95
1·72→
3·09
0·58
0·94
0·42
0·66
0·71
0·68
0·67
0·65
Na 2
O0·
120·
180·
170·
230·
190·
120·
110·
10→
0·08
0·15
0·13
0·14
0·07
0·13
0·19
0·19
0·20
Tota
l99
·80
99·9
299
·77
99·7
099
·16
99·4
499
·20
99·5
999
·82
99·6
910
0·01
98·9
510
0·17
100·
5999
·57
mg
93·3
93·0
94·4
94·7
97·6
96·6
93·0
91·3
92·1
93·3
91·7
86·5
83·0
86·5
87·7
fe1·
000·
880·
770·
770·
700·
780·
800·
860·
820·
800·
861·
000·
990·
981·
0
Wo
0·7
1·4
1·0
4·2
5·4
6·4
4·5
1·1
1·8
1·1
1·5
1·4
1·4
1·2
1·3
En
92·7
91·9
93·8
90·6
92·4
90·4
88·7
90·3
90·4
92·3
90·3
85·3
81·9
85·4
86·6
Fs6·
66·
75·
25·
22·
23·
26·
88·
67·
86·
68·
213
·316
·713
·212
·1
∗Sto
ich
iom
etry
req
uir
esn
oFe
3+.
fe=
Fe2+
/(Fe
2++
Fe3+
).A
nal
yses
:1–
3,9,
11,
14,
pri
mar
yo
rth
op
yro
xen
es(1
fro
mR
eid
etal
.,19
75);
15,
pri
mar
yg
rain
adja
cen
tto
mel
tve
in;
10,1
2,13
,16,
met
aso
mat
ico
rth
op
yro
xen
esre
pla
cin
gcl
ino
pyr
oxe
nes
;4,
relic
tp
rim
ary
gra
inin
mel
tp
ock
et(F
ig.3
g);
5,6,
dif
fere
nt
po
ints
insp
inel
-fr
eeen
stat
ite
insp
inel
–pyr
oxe
ne
sym
ple
ctit
e;7,
8,m
ean
and
ran
ge
ofe
igh
tan
alys
eso
fen
stat
ite
con
tain
ing
man
ysp
inel
incl
usi
on
s,in
spin
el–p
yro
xen
esy
mp
lect
ite.
1758
DAWSON NORTHERN TANZANIAN PERIDOTITE XENOLITHS
the grains contain less Cr, Al, Mn and Ca and are more 6·71 wt %) phlogopite replacing Cr-spinel in sample 1544(Table 6, analysis 4), and metasomatic mica replacing Cr-magnesian than enstatite in the spinel-rich areas of the
same grains; hence these pyroxenes are interpreted as diopside in 747; it is also higher in Ti than other primarymantle micas (Delaney et al., 1980). These overgrowthsunequilibrated.
An overgrowth on the bronzite in sample 1542, ad- or replacements of high-Ti mica on the primary phasestestify to the migration of Ti-rich fluids through specimensjacent to a melt vein, zones towards higher Mg and Al
(analyses 16 and 17); this more magnesian composition 747 and 1544.is analogous to the more magnesian melt-pocket over-growths on olivines. Spinel
Most primary spinels (Table 7) are Mg-chromites, Cr2O3
Clinopyroxene ranging from 50 to 60 wt % and MgO from 10 to 15 wt %.These compositions can be matched in chromites fromPrimary clinopyroxenes (Table 5, analyses 4, 5, 7 andother Lashaine samples (Reid et al., 1975), and from9) are Cr-diopsides with Cr2O3 in the range of 1·73–garnet lherzolite xenoliths from South African kimberlites2·84 wt %, Al2O3 1·7–2·4 wt % and Na2O 1·2–2·1 wt %;(Smith & Dawson, 1975), and the most chromian (62 wt %their compositions fall mainly within the ranges foundCr2O3; analysis 3) resembles inclusions in diamondin earlier studies on Lashaine peridotites (Reid et al.,(Meyer, 1987). Two different spinels are: (1) Mg–Al-rich1975; Dawson, 1987); although the Cr-diopsides in thespinel (analysis 7) in spinel–orthopyroxene symplectitesmore Fe-rich peridotites (Table 5, analyses 10–17) arein 1542, which, in its high MgO and Al2O3 contents, isslightly higher in iron (total FeO mainly >4 wt %)compositionally very similar to spinel in garnet-break-compared with those in the other peridotites (mainlydown coronas in Lashaine garnet lherzolite xenoliths<3·3 wt %). Ca/(Ca+Mg) ratios are 0·46–0·48, except(Reid & Dawson, 1972); (2) a high-iron (25 wt % FeO),for 0·42 in the diopside in (garnet) harzburgite 1542high-Ti (10·4 wt % TiO2), low-Al (2·23 wt % Al2O3)which implies a higher equilibration temperature. Alsochromite (analysis 11) that has exsolved from ilmenite inin 1542, non-equilibrated high-Al diopside in the spinel–1544.pyroxene symplectites (Table 5, analysis 6) contains less
Cr and Na than the primary Cr-diopsides. Cr-diopsidesbeing replaced by enstatite in 1544 and 1546 (Table 5, Ilmeniteanalyses 8 and 15) contain considerably less Al, Cr and Ilmenite, interpreted as being of metasomatic origin,Na, but more Mg, Fe and Ti than the primary diopside; occurs as rounded grains in 1544 (Table 8). Its highsimilar decreases in the jadeite–ureyite molecule have MgO content (14·1 wt %) is similar to other mantlebeen noted previously in partly melted Cr-diopsides in ilmenites but its very high Cr2O3 content (7·33 wt %)kimberlite xenoliths (Carswell, 1975) and in ‘bleached’ distinguishes it from ilmenite megacrysts from kimberliteCr-diopside being replaced by amphibole in veined peri- (Mitchell, 1986) and ilmenites in veined and meta-dotites from Pello Hill, another northern Tanzanian somatized peridotite and MARID xenoliths (e.g. Hartexenolith locality (Dawson & Smith, 1988). Metasomatic & Gurney, 1975; Dawson & Smith, 1977; Jones et al.,diopside replacing Cr-spinel in 747 (Table 5, analyses 1983b; Gregoire et al., 2000). Moreover, before the ex-1–3) differs compositionally from the primary Cr-di- solution of Cr-spinel, this ilmenite must have containedopsides in containing significant TiO2 (>1 wt %), and even higher Cr concentrations.lower Na2O (>0·8 wt % vs often >2 wt % in primary A large rounded grain of ilmenite in a melt patch inCr-diopsides); this diopside is zoned with respect to Cr, Fe-rich peridotite 1546 (see Table 9, analyses 18 and 19)which has its highest concentration immediately adjacent is around the same size (500 �m) and has the sameto the replaced spinel. high-Cr character as the ilmenite in 1544; hence it is
interpreted as a relict metasomatic grain, rather than aprecipitate. Like the other primary phases in the rock itMicais less magnesian than its counterpart in 1544.Primary mica has been found only in sample 1544. It is
a titaniferous phlogopite (mg 91, TiO2 2·18 wt %) withRutile1·19 wt % Cr2O3 (Table 6, analysis 2) and its composition
falls within the ranges found in other primary micas in Rutile has been found only in specimen 1544. Differentmantle xenoliths (Delaney et al., 1980). rutile grains have different compositions (Table 8, ana-
Metasomatic mica in 1544 forms overgrowths on prim- lyses 6–9), indicating that they are unequilibrated. Overallary mica of a highly titaniferous phlogopite (7·14 wt % the rutile is a Cr–Fe–Nb–Ta variety with 3·95–5·24 wt %TiO2, Table 6, analysis 3) that is less magnesian (mg 89) Cr2O3, and 0·29–0·81 wt % FeO. It has appreciablethan the primary mica. In its high TiO2 content, this concentrations of Nb2O5 (0·88–5·13 wt %) and Ta2O5
(0·04–1·05 wt %), with Nb/Ta ratios varying from 7·5overgrowth mica resembles metasomatic high-Ti (TiO2
1759
JOURNAL OF PETROLOGY VOLUME 43 NUMBER 9 SEPTEMBER 2002
Tab
le5
:R
epre
sent
ativ
ean
alys
esof
peri
dotite
clin
opyr
oxen
es
Sam
ple
:74
777
115
4215
4439
2675
015
46
An
alys
is:
12
34
56
78
910
1112
1314
1516
17
SiO
252
·051
·751
·954
·753
·146
·753
·453
·654
·054
·454
·654
·254
·053
·853
·454
·353
·2
TiO
21·
011·
081·
030·
250·
291·
100·
510·
720·
200·
450·
410·
620·
650·
390·
850·
460·
55
Al 2
O3
1·85
2·02
1·99
1·87
2·42
9·44
2·15
0·91
1·72
1·66
1·49
1·18
0·76
1·97
0·45
1·79
1·11
Cr 2
O3
2·12
2·36
2·48
2·84
2·72
0·27
2·41
1·51
1·73
2·76
2·52
1·26
0·27
3·08
1·34
2·90
2·57
Fe2O
31·
010·
551·
141·
780·
412·
672·
251·
131·
540·
000·
000·
000·
000·
000·
001·
991·
37
FeO
2·04
2·75
2·03
0·41
1·98
1·44
1·12
2·93
0·99
4·96∗
4·63∗
4·33∗
3·98∗
2·90∗
5·11∗
2·33
3·13
Mn
O0·
100·
090·
080·
050·
100·
230·
100·
110·
120·
150·
130·
140·
100·
120·
190·
100·
15
Mg
O16
·116
·116
·316
·018
·315
·415
·719
·016
·414
·915
·018
·817
·015
·518
·415
·315
·8
NiO
0·06
0·05
0·07
0·04
0·03
0·04
0·04
0·05
0·05
0·05
0·07
0·06
0·05
0·04
0·04
0·06
0·04
CaO
22·0
21·5
21·7
19·6
18·3
20·0
19·8
19·1
21·3
17·6
18·4
18·5
22·4
18·9
18·6
18·9
20·7
Na 2
O0·
860·
800·
852·
101·
190·
602·
120·
721·
492·
672·
490·
600·
382·
300·
752·
461·
33
Tota
l99
·15
99·1
999
·55
99·7
699
·04
98·9
899
·60
99·7
899
·54
99·5
899
·74
99·6
999
·59
99·0
099
·13
100·
5699
·95
Wo
47·9
46·6
47·3
46·4
41·0
45·3
46·7
40·0
47·4
41·7
42·8
38·5
45·5
38·5
44·2
44·9
45·9
En
48·7
48·7
49·2
52·8
55·6
50·4
51·2
55·2
50·9
49·0
48·6
54·4
48·1
53·2
50·4
47·8
49·2
Fs3·
64·
73·
51·
83·
44·
32·
14·
81·
79·
38·
67·
16·
48·
35·
44·
45·
4
Ca/
(Ca+
Mg
)49
·548
·849
·046
·742
·447
·347
·742
·048
·245
·946
·841
·448
·642
·046
·748
·448
·3
Tota
lFe
O2·
953·
243·
062·
012·
363·
843·
143·
952·
364·
964·
634·
333·
982·
905·
114·
124·
36
Tota
lir
on
asFe
Oas
sto
ich
iom
etry
req
uir
esn
oFe
3+.
An
alys
es:
1–3,
zon
edm
etas
om
atic
pyr
oxe
ne
rep
laci
ng
Cr-
spin
el(1
isfu
rth
est
fro
mco
nta
ctw
ith
spin
el,
3is
clo
sest
);4,
5,7,
9–11
,14,
16,p
rim
ary
Cr-
dio
psi
de;
6,h
igh
-Ald
iop
sid
ein
spin
el–p
yro
xen
esy
mp
lect
ite;
8,C
r-d
iop
sid
eb
ein
gre
pla
ced
by
ort
ho
pyr
oxe
ne;
12,1
5,C
r-d
iop
sid
e‘b
leac
hed
’at
con
tact
wit
hm
elt
po
cket
;13,
met
aso
mat
icd
iop
sid
ere
pla
cin
gp
rim
ary
Cr-
dio
psi
de;
17,p
artl
ym
elte
d,m
ott
led
Cr-
dio
psi
de
atm
arg
ino
fm
elt
po
cket
.
1760
DAWSON NORTHERN TANZANIAN PERIDOTITE XENOLITHS
Tab
le6
:A
naly
ses
ofm
icas
inL
asha
ine
peri
dotite
san
dm
elt
pock
ets
Pri
mar
yan
dm
etas
om
atic
mic
asM
elt-
po
cket
mic
as
Sam
ple
:74
715
4474
715
4439
2675
015
46
An
alys
is:
12
34
56
78
910
1112
1314
1516
1718
1920
2122
SiO
238
·839
·836
·538
·337
·439
·439
·339
·639
·538
·138
·638
·739
·838
·739
·639
·439
·038
·438
·539
·439
·639
·5
TiO
23·
502·
187·
146·
717·
482·
973·
143·
694·
956·
738·
537·
992·
141·
903·
034·
154·
224·
574·
864·
684·
304·
46
Al 2
O3
13·0
13·0
15·1
13·0
13·4
12·8
12·8
12·6
12·6
12·1
12·1
12·3
12·5
13·4
12·6
12·8
12·7
12·4
12·2
12·4
12·6
12·7
Cr 2
O3
1·87
1·19
2·23
1·78
0·52
1·53
1·37
0·22
0·14
0·22
0·29
0·35
1·24
1·15
1·15
0·37
0·43
0·78
1·16
0·25
0·50
0·81
FeO
4·12
3·72
4·00
4·27
5·91
3·94
3·59
4·10
4·59
5·62
6·06
5·36
4·28
5·16
4·82
5·06
5·39
6·04
6·40
5·77
5·77
5·97
Mn
O0·
070·
020·
010·
050·
060·
030·
040·
030·
030·
070·
040·
020·
080·
050·
030·
040·
050·
050·
060·
070·
030·
06
Mg
O22
·623
·619
·320
·620
·723
·123
·823
·522
·820
·719
·420
·023
·223
·123
·222
·221
·821
·020
·421
·722
·420
·9
NiO
n.a
.0·
210·
200·
170·
17n
.a.
n.a
.n
.a.
0·15
0·15
0·13
0·17
n.a
.n
.a.
n.a
.n
.a.
n.a
.n
.a.
n.a
.0·
100·
100·
10
CaO
0·05
n.a
.n
.a.
n.a
.n
.a.
0·02
0·03
0·06
n.a
.n
.a.
n.a
.n
.a.
n.a
.n
.a.
n.a
.n
.a.
n.a
.n
.a.
n.a
.0·
010·
030·
01
BaO
0·12
n.a
.n
.a.
n.a
.0·
180·
110·
220·
15n
.a.
n.a
.n
.a.
n.a
.0·
150·
230·
210·
050·
110·
040·
000·
380·
200·
12
Na 2
O0·
530·
430·
270·
190·
180·
550·
430·
320·
140·
260·
290·
220·
310·
310·
440·
120·
200·
290·
340·
270·
310·
31
K2O
9·65
9·32
9·70
10·1
10·2
9·72
9·80
9·80
9·54
9·93
9·62
9·83
9·71
9·68
9·63
9·37
9·05
9·55
9·63
9·55
9·55
9·65
F0·
79n
.a.
n.a
.n
.a.
n.a
.1·
040·
881·
04n
.a.
n.a
.n
.a.
n.a
.0·
920·
660·
94n
.a.
n.a
.n
.a.
n.a
.n
.a.
n.a
.n
.a.
Cl
0·01
n.a
.n
.a.
n.a
.n
.a.
0·01
0·01
0·00
n.a
.n
.a.
n.a
.n
.a.
0·02
0·01
0·01
n.a
.n
.a.
n.a
.n
.a.
n.a
.n
.a.
n.a
.
Tota
l95
·10
93·4
794
·45
95·1
794
·39
95·2
595
·41
95·1
094
·44
93·7
895
·06
94·9
494
·35
94·3
595
·66
93·5
692
·95
93·1
293
·73
94·8
995
·39
94·5
9
Oo
F0·
330·
410·
370·
430·
390·
270·
39
Tota
l94
·77
93·4
794
·45
95·1
794
·39
94·8
495
·04
94·6
794
·44
93·7
895
·06
94·9
493
·96
94·0
895
·27
93·5
692
·95
93·1
293
·73
94·8
995
·39
94·5
9
mg
0·91
0·92
0·87
0·90
0·87
0·91
0·92
0·91
0·90
0·87
0·85
0·87
0·91
0·89
0·90
0·89
0·88
0·86
0·85
0·87
0·87
0·86
mg=
Mg
/(M
g+
FeT).
An
alys
es:
1,ri
mo
np
rim
ary
chro
me
dio
psi
de;
2,3,
ligh
tb
row
np
rim
ary
core
and
dar
kb
row
nm
etas
om
atic
rim
of
larg
eg
rain
;4,
rim
min
gp
rim
ary
chro
mit
eg
rain
;5,i
nte
rgro
wn
wit
hm
etas
om
atic
enst
atit
ere
pla
cin
gC
r-d
iop
sid
e;6–
8,11
–14,
15–1
9,d
iffe
ren
tg
rain
sin
dif
fere
nt
mel
tp
ock
ets;
9,10
,20–
22,
dif
fere
nt
gra
ins
insa
me
mel
tp
ock
ets.
n.a
.,n
ot
anal
ysed
.
1761
JOURNAL OF PETROLOGY VOLUME 43 NUMBER 9 SEPTEMBER 2002
Tab
le7
:A
naly
ses
ofpe
rido
tite
spin
els
Sam
ple
:74
777
115
4215
4439
2675
015
46
An
alys
is:
12
34
56
78
910
1112
1314
1516
SiO
20·
230·
060·
130·
070·
220·
060·
200·
050·
060·
040·
090·
070·
050·
050·
070·
03
TiO
21·
933·
422·
991·
570·
711·
670·
223·
613·
763·
8610
·42·
981·
302·
973·
043·
12
Al 2
O3
9·78
7·17
4·82
16·0
16·5
17·6
49·8
6·40
6·20
11·8
2·23
5·29
9·81
5·13
5·24
4·81
Cr 2
O3
57·2
56·3
62·0
50·7
51·5
50·0
18·3
54·2
53·9
49·4
51·5
57·0
56·0
58·0
54·8
57·1
Fe2O
33·
264·
324·
543·
413·
963·
142·
506·
056·
035·
030·
004·
834·
656·
687·
795·
37
FeO
12·2
16·6
6·62
12·9
13·2
10·8
6·74
19·2
19·8
16·0
25·6
18·4
16·6
13·1
17·9
20·1
Mn
O0·
250·
320·
370·
230·
310·
300·
220·
330·
350·
220·
260·
370·
270·
310·
300·
32
Mg
O13
·812
·818
·614
·813
·816
·421
·311
·010
·713
·710
·710
·811
·614
·410
·89·
89
NiO
0·17
0·16
0·12
0·13
0·09
0·20
0·06
0·20
0·17
0·16
0·18
0·14
0·14
0·18
0·17
0·11
CaO
0·04
0·01
0·02
n.a
.n
.a.
n.a
.n
.a.
0·02
0·02
0·01
0·00
0·07
0·02
0·04
n.a
.0·
00
Tota
l10
0·77
101·
1610
0·21
99·8
199
·36
100·
2899
·33
101·
0810
0·99
100·
2210
0·96
99·9
599
·94
100·
8610
0·11
101·
17
mg
0·63
0·58
0·83
0·66
0·66
0·73
0·85
0·52
0·49
0·60
0·43
0·51
0·56
0·67
0·52
0·46
fe0·
830·
760·
620·
810·
780·
800·
750·
770·
790·
781·
000·
810·
790·
690·
720·
81
cr0·
800·
840·
900·
690·
680·
650·
200·
850·
850·
740·
950·
870·
800·
880·
870·
89
mg=
Mg
/(M
g+
Fe2+
);fe=
Fe2+
/(Fe
2++
Fe3+
);cr=
Cr/
(Cr+
Al)
.A
nal
yses
:1,
4–6,
8,12
,13
,15
,16
,p
rim
ary
chro
mit
es;
2,3,
core
and
rim
of
larg
ero
un
ded
chro
mit
ein
clu
ded
inen
stat
ite;
7,re
pre
sen
tati
vean
alys
iso
fsp
inel
inp
yro
xen
e–sp
inel
sym
ple
ctit
e;ve
rylit
tle
com
po
siti
on
alva
riat
ion
wit
hg
rain
size
,o
rw
ith
loca
tio
nin
the
sym
ple
ctit
e;9,
10,c
ore
and
rim
of
pri
mar
yg
rain
,bei
ng
rep
lace
db
ym
ica
(an
alys
is4,
Tab
le6)
;11,
lam
ella
rM
g–T
ich
rom
ite
exso
lved
fro
milm
enit
e(F
ig.
3e);
14,
spin
elin
clu
ded
inm
etas
om
atic
ort
ho
pyr
oxe
ne
rep
laci
ng
chro
me-
dio
psi
de.
1762
DAWSON NORTHERN TANZANIAN PERIDOTITE XENOLITHS
Table 8: Analyses of metasomatic ilmenites and rutiles in peridotites
Phase: Ilmenite Rutile
Sample: 3929 1546 1544
Analysis: 1 2 3 4 5 6 7 8 9
SiO2 n.a. 0·01 0·01 0·02 0·04 0·10 0·02 0·08 0·00
TiO2 54·0 52·4 50·8 55·6 54·3 95·5 91·6 94·3 88·6
Al2O3 0·02 0·15 0·14 0·10 0·19 0·04 0·04 0·04 0·06
Cr2O3 3·56 7·30 8·70 7·33 8·22 3·95 4·34 4·14 5·24
FeO∗ 27·7 30·7 31·2 23·1 21·5 0·56 0·81 0·29 0·49
MnO 0·34 0·35 0·34 0·29 0·26 0·04 0·00 0·01 0·02
MgO 12·9 10·4 9·98 14·1 15·4 0·10 0·05 0·05 0·07
NiO n.a. 0·14 0·16 0·23 0·29 0·03 0·02 0·03 0·04
Nb2O5 n.a. n.a. n.a. 0·04 0·20 0·88 2·94 1·79 4·81
Ta2O5 n.a. n.a. n.a. 0·01 0·07 0·04 0·42 0·22 1·05
Total 98·71 101·45 101·33 100·68 100·47 101·24 100·24 100·95 100·39
∗All iron expressed as FeO.Analyses: 1, ilmenite in enstatite-dominated reaction corona, replacing primary Cr-diopside (from Dawson, 1987); 2 and 3,darkest and brightest areas (in BSE image) in large (500 �m) rounded relict grain in melt pocket; 4, large (400 �m) primaryilmenite grain containing lamellae of exsolved Mg–Ti chromite (Table 7, analysis 11), being replaced by rutile (see Fig. 3e);5, ilmenite exsolved from rutile grain (analysis 9); 6, rutile replacing primary ilmenite (analysis 4); 7, rounded rutile grainincluded in olivine; 8, rounded rutile grain included in enstatite; 9, large rounded rutile grain exsolving ilmenite (analysis5). n.a., not analysed.
to 8. Their high Cr content resembles that in the very Clinopyroxenesfew other analysed rutiles from peridotite and MARID On the basis of their Ca–Mg–Fe contents, most MPxenoliths (Smith & Dawson, 1975; Dawson & Smith, clinopyroxenes (Table 10) are diopsides; some have1977), and distinguishes them from eclogite rutiles enough alumina to be classified as augites, and othersin which Cr2O3 is usually <1%. Of the upper-mantle (e.g. analyses 10, 19 and 22) are sufficiently magnesianrutiles listed by Haggerty (1983, 1987), ones from to be endiopsides. Overall, TiO2 and Cr2O3 con-the Jagersfontein kimberlite are the closest match for centrations are <1·5 wt %, Al2O3 <3 wt % and Na2Othe Lashaine rutiles. <1 wt %. However, within these generalizations, there
are many variations in the minor element concentrationsbetween pyroxenes from different specimens, betweenpyroxenes from different melt pockets within the sameMINERAL CHEMISTRY OF THEsample, and between different pyroxenes (some of which
MELT POCKETS may be zoned) within individual melt pockets. For ex-Olivine ample, pyroxenes from three different melt pockets inOlivines in the melt pockets occur as discrete grains, 771 (analyses 5 and 6) show only small variations in Tias overgrowths on wall-rock grains and on partly and Na, and only moderate variation in Cr and Al (e.g.resorbed relict grains in the melt pockets. Compared 0·27–0·79 wt % Al2O3), whereas the core and rim of awith the primary olivine in the host peridotite, most single grain in a melt pocket in 1544 (analyses 11 andmelt-pocket (MP) olivine is more magnesian, e.g. Fo93·4 12) show wide variations in TiO2 (0·89–1·64 wt %), Al2O3
vs Fo92·1 (Fig. 4; Table 3, analyses 3 and 1); a very (2·84–0·55 wt %) and Cr2O3 (0·14–0·40 wt %). Again,few are more iron rich than the primary olivine (some pyroxenes from certain specimens or melt pockets canonly by <1% Fo) and others zone towards more iron- have distinctive chemistry; for example, those over-rich rims (Fig. 4). MP olivines consistently contain growing relict enstatite in 1542 are relatively aluminoushigher CaO (frequently ×4) and Cr2O3 (×2 to ×3), (Al2O3 >2·4 wt %), whereas sectors in a sector-zonedbut lower NiO than primary olivines; relative MnO grain in 1546 (analyses 19 and 20) have higher TiO2
(1·11–1·55 wt %) than other MP pyroxenes. One ofconcentrations are erratic.
1763
JOURNAL OF PETROLOGY VOLUME 43 NUMBER 9 SEPTEMBER 2002
Tab
le9
:A
naly
ses
ofm
elt
pock
etsp
inel
and
rutile
Ph
ase:
Sp
inel
Ru
tile
Sam
ple
:74
777
115
4215
4439
2675
015
4615
44
An
alys
is:
12
34
56
78
910
1112
1314
1516
17
SiO
20·
070·
160·
190·
240·
260·
110·
110·
130·
150·
090·
210·
140·
150·
290·
100·
060·
05
TiO
23·
121·
381·
320·
770·
620·
690·
187·
751·
238·
420·
811·
411·
982·
3410
·890
·989
·4
Al 2
O3
12·1
4·83
29·7
38·1
41·0
33·9
50·0
7·03
13·1
2·06
7·86
3·73
10·3
13·6
1·89
0·08
0·09
Cr 2
O3
53·0
63·4
36·8
28·6
25·7
32·2
18·2
49·0
56·4
54·9
58·9
65·8
55·0
46·5
43·7
3·49
3·56
Fe2O
33·
732·
593·
123·
363·
123·
842·
763·
732·
500·
003·
003·
453·
136·
507·
330·
000·
00
FeO
12·0
16·6
9·33
8·30
8·55
10·8
6·14
21·8
11·5
23·6
16·8
14·2
19·2
19·4
27·6
0·67∗
0·71∗
Mn
O0·
230·
390·
210·
180·
190·
300·
160·
390·
200·
390·
500·
320·
320·
280·
400·
000·
00
Mg
O12
·511
·318
·419
·319
·517
·121
·811
·815
·511
·710
·810
·910
·610
·99·
220·
030·
03
NiO
0·26
0·06
0·17
0·15
0·15
0·16
0·06
0·19
0·16
0·24
0·07
0·10
0·17
0·16
0·19
0·03
0·04
CaO
0·06
0·11
n.a
.n
.a.
n.a
.n
.a.
n.a
.0·
010·
020·
000·
05n
.a.
0·01
0·02
0·02
n.a
.n
.a.
Tota
l10
0·57
100·
8299
·24
99·0
799
·09
99·1
899
·41
100·
8310
0·75
101·
4099
·00
100·
0510
0·85
100·
0010
1·28
101·
12†
101·
19†
mg
0·71
0·55
0·80
0·80
0·80
0·74
0·86
0·49
0·71
0·47
0·54
0·57
0·49
0·50
0·37
fe0·
760·
880·
770·
740·
740·
760·
700·
870·
841·
00·
850·
850·
870·
770·
81
cr0·
750·
900·
450·
340·
300·
400·
200·
820·
740·
950·
830·
920·
780·
700·
94
∗All
iro
nas
FeO
.†T
ota
lsin
clu
de
resp
ecti
vely
4·55
and
5·49
wt
%N
b2O
5an
d1·
31an
d1·
82w
t%
Ta2O
5,g
ivin
gN
b/T
a5·
6an
d4·
8.m
g=
Mg
/(M
g+
Fe2+
);fe=
Fe2 +
/(Fe
2++
Fe3+
);cr=
Cr/
(Cr+
Al)
.A
nal
yses
:1,
2,12
,eu
hed
ral
spin
elg
rain
sin
gla
ss;
3,4
and
5,co
re,
inte
rmed
iate
zon
ean
dri
mo
fzo
ned
spin
el(p
oin
tsA
–C,
Fig
3i);
6,at
oll
spin
elo
verg
row
ing
zon
edsp
inel
(po
int
D,
Fig
3i);
7,sp
inel
ove
rgro
win
gp
rim
ary
Cr-
spin
el;
8–10
,sm
all
euh
edra
lsp
inel
sin
dif
fere
nt
mel
tp
ock
ets;
11,e
uh
edra
lsp
inel
incl
ud
edin
ove
rgro
wth
oliv
ine
atm
arg
ino
fm
elt
po
cket
;13,
14,c
ore
and
rim
of
100
�meu
hed
rals
pin
elg
rain
emb
edd
edin
mel
t-p
ock
etm
ica;
15,e
uh
edra
l80
�mg
rain
incl
ud
edin
mel
t-p
ock
etp
yro
xen
e;16
and
17,e
xtre
mes
of
com
po
siti
on
so
fsm
all(
up
to30
�m)
euh
edra
lru
tile
gra
ins
inth
ree
mel
tp
ock
ets
(Fig
.3j
).
1764
DAWSON NORTHERN TANZANIAN PERIDOTITE XENOLITHS
SpinelsThe MP spinels (Table 9) can be grouped into two mainsets. The first set comprises Mg-chromites, with Cr2O3
>50 wt % and MgO>12–15%; it includes samples (e.g.analyses 2 and 14) in which the high Cr2O3 concentrations(>63 wt %) are similar to those in spinels from highlyrefractory peridotites and spinel inclusions in diamond.A zoned grain in sample 1546 (analyses 15 and 16) hasrims relatively enriched in Ti, Al, Fe and Mg and, onthe basis of the stoichiometric calculation of Fe3+, theiron is more highly oxidized. This Mg-chromite set,although chemically similar to many of the Cr-spinels in
Fig. 4. Plot of CaO vs Fo content (mg) for primary and melt-pocket the host peridotites, is distinguished on textural grounds;olivines. c and r indicate cores and rims of melt-pocket olivines. that is, the MP spinels are considerably smaller and are
euhedral. (Relict peridotite Cr-spinel grains undoubtedlyoccur within some melt pockets but they are relatively
the most magnesian clinopyroxenes (19·5 wt % MgO; large and are rounded.) A subset of the Mg chromitesanalysis 22) occurs in a melt pocket within a partly melted occurs in some melt pockets in 1544, in which somebronzite grain. chromites (analyses 8 and 10) contain unusually high
The MP pyroxenes are generally distinguishable from amount of TiO2 (7–9 wt %), a feature in keeping withthe peridotite Cr-diopsides by their lower Cr2O3 and the overall high-Ti mineralogy of this rock.Na2O concentrations (Fig. 5). However, they can be The second set of spinels, found only in the Mg-chemically similar to some ‘bleached’ rims on some peridotites and represented by analyses 3–7, is moreprimary Cr-diopsides (e.g. Table 5, analyses 8 and 12) aluminous, with cr (Cr/[Cr+ Al]) ranging from 0·20 tothat have lost Cr and Na during melting. 0·45, compared with cr >0·74 in the Mg-chromites. In
addition, they are more magnesian, with MgO in therange 17–22 wt %. One grain in 1542 zones to a rimricher in Al and Mg, but poorer in Ti, Cr and Fe, thanthe core (analyses 3–5).
MicasThe MP micas are Ti-phogopites (Table 6). There areonly minor variations in the major element concentrations
Rutileof SiO2 (38–39·6 wt %), Al2O3 (12·1–12·8 wt %) andK2O (9·05–9·83 wt %) but there are, within limits, ranges Rutile occurs in the melt pockets of specimen 1544, thein the concentrations of the other oxides. As in the MP only specimen to contain peridotite metasomatic rutile.clinopyroxenes, these variations occur between micas The MP rutile (Table 9, analyses 16 and 17) contains lessfrom different specimens, in micas from different melt Cr than the metasomatic rutiles. It has Nb concentrationspockets within the same sample, and in different micas (>5 wt % Nb2O5) similar to the metasomatic grains (seewithin an individual melt pocket. For example, micas in Table 9, analyses 5 and 6), but has exceptionally high747 (Table 6, analyses 5–11) contain relatively low TiO2 Ta2O5 (>1 wt %) resulting in a low Nb/Ta ratio of >5,(3–3·67 wt %) and are more magnesian (mg 91–92) compared with 7·5–11 for the peridotite rutiles. The MPcompared with those in 1544 (TiO2 4·95–8·53 wt %; mg rutiles themselves vary little in composition except for85–90). In the MP micas in the more Fe-rich peridotites Nb2O5 (range 4·55–5·49 wt %) and Ta2O5 (range 1·31–(analyses 12–19), TiO2 variation is small (4·15–4·86 wt %), 1·82 wt %).but mg varies from 85 to 89, with the higher valuesoverlapping with those for micas in the other peridotites.Where analysed, BaO concentrations are low (mainly
Glass>0·2 wt %) compared with up to 1·38 wt % in MP micasAnalyses presented in Table 11 show that, in general,in peridotites from Labait (Dawson, 1999). For mostthe glasses are low in SiO2 (<50 wt %), Al2O3 (<8 wt %),micas, the lower Cr2O3 concentrations (0·3–0·8 wt %)CaO (mainly <2 wt %) and Na2O (<0·20 wt %). K2Odistinguish them from the peridotite primary micasis always in greater concentrations than Na2O, reflected(>1·2 wt %). However, it is worth noting that in sample
1544 the darker brown metasomatic rim on primary in K/(K + Na) ratios generally >0·9. Although Alconcentrations are low, the glasses are not peralkaline,mica (Table 6, analysis 3) is chemically very similar to
one of the MP micas (analysis 10). (K + Na)/Al ratios being <0·7. These low Si, Al, Ca
1765
JOURNAL OF PETROLOGY VOLUME 43 NUMBER 9 SEPTEMBER 2002
Tab
le1
0:
Ana
lyse
sof
mel
t-po
cket
clin
opyr
oxen
es
Sam
ple
:74
777
115
4215
4439
2675
015
46
An
alys
is:
12
34
56
78
910
1112
1314
1516
1718
1920
2122
SiO
253
·253
·253
·253
·053
·453
·852
·151
·950
·853
·051
·353
·053
·453
·852
·654
·151
·952
·953
·252
·453
·453
·7
TiO
20·
660·
760·
710·
890·
680·
570·
820·
832·
181·
121·
640·
890·
510·
360·
610·
581·
500·
741·
111·
550·
710·
93
Al 2
O3
1·32
1·47
0·97
1·16
0·79
0·27
2·78
2·43
2·16
0·95
2·84
0·55
1·34
0·86
1·27
0·60
1·68
0·47
0·37
0·61
1·08
0·37
Cr 2
O3
1·36
1·10
1·48
0·20
1·35
1·85
0·97
1·26
1·53
1·11
0·40
0·14
0·58
0·34
1·64
0·68
0·67
0·61
0·74
1·25
1·13
0·68
Fe2O
31·
201·
380·
601·
411·
501·
111·
621·
411·
160·
761·
271·
301·
201·
031·
930·
001·
741·
931·
321·
720·
991·
08
FeO
1·63
1·60
2·01
2·11
2·39
2·04
2·37
2·21
2·85
3·31
2·28
2·42
2·20
2·62
1·60
5·64∗
2·97
2·92
4·99
3·38
3·08
5·34
Mn
O0·
080·
090·
060·
100·
100·
120·
220·
120·
120·
130·
090·
150·
100·
110·
090·
220·
140·
160·
160·
140·
120·
25
Mg
O17
·217
·216
·717
·717
·317
·517
·817
·317
·720
·416
·417
·417
·818
·216
·518
·216
·617
·219
·616
·816
·119
·5
NiO
0·03
0·05
0·08
0·04
0·03
0·03
0·05
0·07
0·09
0·07
0·05
0·06
0·06
0·04
0·06
0·05
0·01
0·06
0·03
0·04
0·12
0·05
CaO
21·9
22·0
22·5
22·0
20·8
21·1
19·6
20·8
20·1
18·0
22·9
22·9
21·7
21·4
22·3
18·9
21·5
22·0
17·3
20·8
22·4
17·3
Na 2
O0·
810·
790·
740·
490·
960·
990·
760·
640·
500·
350·
380·
260·
540·
460·
790·
460·
680·
450·
520·
840·
820·
52
Tota
l99
·39
99·5
499
·05
100·
0099
·30
99·3
899
·09
98·9
799
·19
99·1
999
·55
99·0
799
·43
99·2
299
·39
99·4
399
·39
99·4
499
·34
99·5
399
·88
99·7
0
Wo
46·5
46·6
47·6
45·7
44·4
45·0
42·5
44·6
42·8
36·7
48·2
46·9
45·0
44·0
48·1
38·8
45·9
45·7
35·6
44·5
47·1
35·6
En
50·8
50·7
49·0
50·9
51·6
51·6
53·6
51·6
52·4
58·0
48·0
49·4
51·3
51·8
49·2
52·1
56·2
49·9
49·9
49·8
49·6
55·8
Fs2·
72·
73·
43·
44·
03·
43·
93·
84·
85·
33·
83·
93·
74·
22·
79·
14·
94·
78·
15·
65·
18·
6
∗All
iro
nas
FeO
.A
nal
yses
:1,
un
zon
edm
elt-
po
cket
pyr
oxe
ne;
2an
d3,
core
and
rim
of
zon
edcr
ysta
l;4,
inm
elt
vein
let,
wit
hm
ica
and
zeo
lite;
5an
d6,
extr
emes
of
com
po
siti
on
so
fsev
enh
om
og
eneo
us
dio
psi
des
inth
ree
mel
tpo
cket
s;7
and
8,m
elt-
po
cket
dio
psi
de
ove
rgro
win
gre
lictc
ore
so
fpar
tly
mel
ted
pri
mar
yen
stat
ite
[7is
imm
edia
tely
adja
cen
tto
enst
atit
eco
re,
8is
euh
edra
lri
mp
rotr
ud
ing
into
vein
gla
ss–z
eolit
e(F
ig.
3g)]
;9
and
10,
extr
emes
of
com
po
siti
on
sin
sin
gle
gra
in;
11an
d12
,co
rean
dri
mo
fg
rain
ind
iffe
ren
tm
elt
po
cket
;13
–15,
un
zon
edd
iop
sid
esin
dif
fere
nt
mel
tp
ock
ets;
16,
euh
edra
lu
nzo
ned
dio
psi
de;
17an
d18
,co
rean
dri
mo
feu
hed
ral
gra
in;
19,
dar
kest
(in
BS
Eim
age)
sect
or;
20,
bri
gh
test
sect
or
inco
mp
lexl
yse
cto
r-zo
ned
euh
edra
lg
rain
ind
iffe
ren
tm
elt
po
cket
;21
,co
lou
rles
so
verg
row
tho
nC
r-d
iop
sid
ep
roje
ctin
gin
tom
elt
po
cket
;22
,h
igh
-Mg
gra
inin
mel
tp
ock
etw
ith
inb
ron
zite
.
1766
DAWSON NORTHERN TANZANIAN PERIDOTITE XENOLITHS
Fig. 5. Plot of Cr2O3 vs Na2O for primary and metasomatic Cr-diopsides, and melt-pocket diopsides. Note that the metasomatic Cr-diopside,although ‘inheriting’ high Cr2O3 from replaced Cr-spinel, has lower Na2O concentrations than primary Cr-diopside.
Table 11: Representative analyses of glasses in melt pockets in Lashaine peridotites
Sample: 771 1542 1544 3926 750 1546
Analysis: 1 2 3 4 5 6 7 8 9 10 11 12 13 14
SiO2 49·6 48·7 47·7 50·93 49·0 47·1 49·8 22·3 50·5 49·3 48·7 49·0 50·4 49·7
TiO2 0·26 0·35 2·25 0·53 0·93 1·12 0·56 2·76 0·79 0·72 0·75 0·96 2·12 1·81
Al2O3 6·49 8·06 16·3 17·95 6·74 6·66 4·45 2·12 6·36 6·40 7·49 7·08 5·16 5·40
FeO∗ 15·5 16·2 5·11 6·35 13·1 14·5 9·94 54·9 12·2 14·0 18·0 17·0 15·8 15·9
MnO 0·07 0·09 0·11 0·18 0·21 0·16 0·03 0·75 0·12 0·16 0·25 0·18 0·08 0·10
MgO 6·21 11·1 4·15 13·58 6·57 6·93 14·3 2·97 6·92 6·24 6·79 6·60 8·40 8·27
CaO 0·86 1·49 5·74 7·94 2·01 1·71 0·57 1·77 2·23 1·85 1·55 1·47 0·99 0·71
BaO 0·44 0·00 0·00 n.a. n.a. n.a. 0·08 0·32 0·00 0·00 0·00 0·00 0·00 0·00
Na2O 0·04 0·09 0·08 0·08 0·08 0·11 0·10 0·15 0·17 0·15 0·14 0·07 0·12 0·13
K2O 1·71 1·00 1·61 0·15 3·98 4·38 1·25 0·22 3·60 3·85 5·04 4·69 2·57 2·76
P2O5 0·05 0·08 0·16 0·07 n.a. n.a. 0·21 0·05 0·07 0·08 0·06 0·06 0·14 0·11
Cl 0·07 0·04 n.a. n.a. n.a. n.a. 0·09 0·02 0·11 0·07 n.a. n.a. 0·07 0·05
SO3 0·06 0·06 n.a. n.a. n.a. n.a. 0·13 0·05 n.a. n.a. n.a. n.a. 0·07 0·05
Total 81·36 87·26 83·21 82·62 82·67 81·30 88·42 83·32 82·91 88·77 87·17 85·92 84·99
mg 0·24 0·35 0·39 0·28 0·27 0·53 0·04 0·31 0·26 0·23 0·23 0·29 0·29
k 0·97 0·88 0·96 0·97 0·96 0·89 0·49 0·93 0·95 0·96 0·96 0·93 0·93
alk 0·29 0·15 0·16 0·66 0·71 0·34 0·23 0·66 0·69 0·75 0·74 0·58 0·59
mg = Mg/(Mg + FeT), k = K/(K + Na), alk = (K + Na)/Al∗. Total iron expressed as FeO. Cr2O3 and F not analysed. n.a., notanalysed. Analyses: 1, 2, glasses in different melt pockets; 3, glass in melt pocket; 4, interstitial glass in garnet reactioncorona, garnet lherzolite, Lashaine (Schiano et al., 2000); 5–8, glasses from different melt pockets [7 and 8 are from thesame melt pocket and exhibit an emulsion texture (Fig. 2k) (?immiscible relationship)]; 9, 10, glasses in different meltpockets; 11, 12, glasses in different melt pockets; 13, 14, glasses in different melt pockets.
1767
JOURNAL OF PETROLOGY VOLUME 43 NUMBER 9 SEPTEMBER 2002
and alkali concentrations reflect the refractory, olivine- (1994), in containing relatively low concentrations ofSiO2, Al2O3 and total alkalis, and also being potassicdominated protolith, but the glasses are high in Fe
(analysis 7 being exceptionally so) mostly in the range rather than sodic (Fig. 6a and b).The Lashaine glasses also show certain differences from10–18 wt % total FeO; this combined with relatively low
MgO concentrations (mostly <10 wt %) gives mg mainly those in other Tanzanian xenoliths. Glasses in xenolithsfrom Labait show considerable compositional variation<40. Importantly, the low MgO (4–14 wt %, mostly
<10 wt %) concentrations show that these glasses were between and within different individual melt pockets,and many contain significant amounts of Ba (Dawson,not in equilibrium with upper-mantle olivines of >Fo90
1999). Glasses in Olmani xenoliths are similarly verycomposition (Roeder & Emslie, 1970).variable in composition ( Jones et al., 1983a), and com-Whereas the above general observations pertain toparison with the Lashaine glasses shows that most aremost of the glasses, there are exceptions. First, glass inrelatively high in Al2O3 (12–20 wt %) and P2O5 (>1 wt %),1542 (analysis 3), associated with the high-Al pyroxeneand some are highly sodic. However, like the Lashaineand spinel formed by breakdown of former garnet, con-glasses, many of the Labait and Olmani glasses havetains more Al2O3 (16 wt %) and CaO (5·74 wt %) butanalytical total shortfalls, attributed to unanalysed vol-lower FeO (5·11 wt %) and MgO (4·15 wt %) than theatiles.other glasses; in its high alumina content, this glass is
similar to glass reported in a reaction corona in anotherLashaine garnet lherzolite (Schiano et al., 2000) (Table11, analysis 4). Second, in specimen 1544 one of the
Zeolitesglasses is a very high-Fe variety (FeO 54·9 wt %),Like the glasses, the zeolites within the vesicles arecombined with very low SiO2 (22·3 wt %), Al2O3
unstable under the electron beam. Analyses (Table 12)(2·12 wt %) and MgO (2·97 wt %) and relatively highshow most to be a Ca–K phase with Si:Al >2:1. ThereTiO2 (2·76 wt %).are variations between samples and between zeolites inIn most samples, glass compositions in different meltdifferent vugs in single samples. In particular, the zeolitespockets have roughly the same characteristics (compare,in the more iron-rich peridotites contain a small amountfor example, the pairs of analyses 1 and 2, and 11 andof BaO (up to 2·18 wt % in sample 750), which is12, Table 11). Exceptionally, in 1544, glass compositionsapparently absent in zeolites in the magnesian peridotites.in two melt pockets (analyses 4 and 5) are reasonably
similar with respect to Al, Fe, Ca, Mg and K, but thesecontrast with two glasses (analyses 6 and 7) in anothermelt pocket, which are also very different from each
Trace element concentrations in peridotiteother—the one being a low-Al, high-Mg variant, thediopside and melt-pocket diopside, micaother the exceptionally high-Fe glass noted above.and glassAll the glasses have low analytical totals and, with the
possible exception of the high-Fe glass, the totals would Ion microprobe analyses (Table 13) have been made onremain low even if total iron was calculated as Fe2O3. phases in two Lashaine specimens (1546 and 3926) and,The vesicularity of the glasses testifies to the volatile-rich for comparison, in peridotite 4213 from Labait, majornature of the melts, so unanalysed (or fugitive) dissolved element data for which have been given by Dawsonvolatiles probably account for the shortfalls. CO2 is the (1999). Analyses were made on glass and MP clino-most likely, as ion microprobe analyses of glasses in two pyroxene in all specimens, MP micas in two specimensspecimens (not reported in detail here) give low H2O (one from Lashaine and one from Labait) and primaryconcentrations ranging from 1·54 to 3·88 wt % in glass Cr-diopside in 1546. For the MP micas, in which thein specimen 1546, and from 0·42 to 3·85 wt % in 3926; REE concentrations were <1 ppm, only the light REEeven when these values are integrated, there is still a (LREE) were analysed. An earlier electron microprobeshortfall in the analyses of these glasses. analysis of primary mica in 4213 showed it to contain
Schiano & Clochiatti (1994) reported that glasses in 0·15 wt % BaO (1340 ppm Ba) (Dawson, 1999).mantle xenoliths from a large number of worldwide Comparing primary Cr-diopside with MP diopside in
1546, the former contains only slightly higher con-localities have certain common characteristics, with highSiO2 (55–65 wt %), Al2O3 (15–25 wt %) and total alkalis centrations of Sr, Y, Zr, Nb, Pb and the REE (Fig. 7),
and the REE pattern (Fig. 8) is almost identical. Both(4–11 wt %) relative to their peridotite protoliths; inaddition, most glasses are sodic with K2O/Na2O rarely clinopyroxenes contain much lower concentrations of
most of these elements than the MP glass, which alsoapproaching unity. Subsequent studies on mantle glasseshave revealed variations but have, in general, confirmed contains relatively high amounts of Rb, Ba and the LREE
(La to Nd). However, the glass contains less Sr, Y andthe findings of Schiano & Clochiatti (1994). The Lashaineglasses differ from those reported by Schiano & Clochiatti the heavy REE (HREE; Dy to Lu) than the pyroxenes.
1768
DAWSON NORTHERN TANZANIAN PERIDOTITE XENOLITHS
Fig. 6. (a) SiO2 vs total alkalis plot for Lashaine glasses compared with ‘Schiano’ glasses from other worldwide localities [Schiano data fromSchiano & Clocchiatti (1994)]. (b) Al2O3 vs K2O/Na2O plot for Lashaine glasses compared with Schiano glasses from other worldwide localities[Schiano data from Schiano & Clocchiatti (1994)].
Table 12: Analyses and structural formulae of zeolites lining vesicles in melt pockets in Lashaine peridotites
Sample: 747 1542 1544 3926 750 1546 1542 1544
Analysis: 1 2 3 4 5 6 7 8 SF SF
SiO2 49·0 46·4 45·8 52·7 44·8 48·9 46·7 51·9 22·837 22·071
TiO2 0·56 0·18 0·00 0·47 0·30 0·63 0·35 1·59 0·427 0·000
Al2O3 23·2 23·2 24·1 22·4 25·6 24·3 24·0 21·8 13·404 13·702
FeO∗ 0·78 0·58 0·24 0·65 0·62 0·34 0·45 0·90 0·672 0·395
MnO 0·01 0·00 0·00 0·00 0·00 0·00 0·03 0·00 0·000 0·000
MgO 0·25 0·09 0·05 0·19 0·10 0·15 0·07 0·36 0·065 0·033
CaO 9·30 8·64 9·56 8·85 9·59 8·76 8·98 8·63 4·555 4·940
BaO 0·49 0·72 na 0·27 2·18 1·13 0·98 1·15 0·139 0·000
Na2O 0·16 0·19 0·14 0·09 0·04 0·09 0·00 0·17 0·180 0·133
K2O 1·50 2·72 6·09 5·36 5·55 6·24 6·18 5·46 1·700 3·747
P2O5 n.a. 0·11 0·17 0·15 0·12 0·10 0·16 0·19 0·047 0·070
Total 85·25 82·83 86·23 91·11 88·9 90·64 87·90 92·15 44·026 45·091
∗All iron expressed as FeO. Cl, S and Cr sought but below detection limits. Analyses: 1, 2, 3, 8, lining vugs in individualmelt-pockets; 4, total includes 0·16 wt % F and −0·07 wt % O ‘congruent to’ F; 5, 6, lining vugs in same melt pocket; 7, indifferent melt pocket. Structural formulae (SF) from analyses 2 and 3, on basis of 72 oxygens. It should be noted that thezeolites degrade rapidly even under a rastered beam at 10 nA.
Zr, Nb and Pb partitioning between glass and pyroxeneMP diopside in 3926 is similar to that in 1546, withis not as high as in 3926; moreover, the Ba content oflow concentrations of most incompatible elements andthe MP mica (2970 ppm) is double that in the primarydifferences between them only of the order of about×2.mica from this specimen (1340 ppm; Dawson, 1999).In contrast to 1546, 3926 also contains MP mica, whichOverall the MP phases in the Labait specimen containcontains high Rb (340 ppm) and Ba (1490 ppm) relativehigher concentrations of Rb, Sr and Ba, but lower Zr,to both glass and cpx, and high Nb and Pb relative toNb and Pb, than those in the Lashaine melt pockets.diopside. Diopside has the highest concentrations of Sr
In summary, in all specimens Zr, Nb and Pb areand Y and glass has the highest concentrations of Zr,preferentially concentrated in the glass; in which Ba isNb and Pb.also concentrated in mica-absent 1546. In mica-bearingIn Labait specimen 4213, element partitioning betweenMPs, micas concentrate Rb and Ba relative to glassMP phases is in most respects similar to that in 3926;
except that Sr is high in glass relative to pyroxene, and and pyroxene, and contain more Nb than coexisting
1769
JOURNAL OF PETROLOGY VOLUME 43 NUMBER 9 SEPTEMBER 2002
Table 13: Ion microprobe analyses of trace element concentrations (in ppm) in phases and glasses in northern
Tanzanian spinel peridotites
Locality: Lashaine Labait
Sample: 1546 3926 4213
Cr-diopside Cpx Glass Cpx Mica Glass Cpx Mica Glass
Analysis: 1 2 3 4 5 6 7 8 9
Rb 0·0 0·12 154 0·50 342 189 0·71 510 286
Sr 103 71·9 61·8 133 28·7 27·9 74·7 31·3 322
Y 4·81 2·61 1·90 4·82 0·19 3·22 4·19 0·103 4·01
Zr 95·6 7·77 573 16·1 6·1 531 22·5 3·63 81·9
Nb 0·49 0·32 97·1 0·31 14·7 148 1·28 4·93 31·4
Ba 0·20 0·45 295 0·0 1489 57·3 3·00 2966 793
Pb 5·49 2·60 101 1·36 16·7 17·7 7·32 8·52 13·4
La 2·62 2·19 49·2 5·30 1·53 23·0 5·09 0·55 15·9
Ce 10·3 7·59 37·4 17·8 1·54 22·6 15·5 0·045 28·3
Pr 1·84 1·28 6·55 2·88 0·28 2·93 2·73 0·004 3·18
Nd 10·6 6·66 17·4 15·9 1·32 12·0 13·8 0·055 12·0
Sm 2·89 1·73 2·14 3·78 0·030 1·41 3·32 0·077 2·06
Eu 0·897 0·558 0·323 0·970 0·316 0·904 0·653
Gd 2·69 1·73 3·64 3·15 1·15 2·19 3·15
Tb 0·384 0·196 0·279 0·390 0·160 0·271 0·294
Dy 1·99 1·13 1·38 1·75 0·660 1·62 1·50
Ho 0·267 0·154 0·139 0·263 0·050 0·216 0·306
Er 0·599 0·633 0·435 0·630 0·496 0·394 0·748
Yb 0·529 0·448 0·037 0·573 0·083 0·694 0·190
Lu 0·073 0·038 0·027 0·036 0·010 0·027 0·040
All analyses are of melt-pocket phases, with the exception of the Cr-diopside in sample 1546 which is a primary Cr-diopside.
pyroxenes. However, as noted for other hydrous mantle found in the melt pockets. The disparity would be evengreater if the MP apatites (not analysed) contain REE.assemblages (McDonough & Frey, 1989), the mica con-
tains only relatively small amounts of the REE. An REEplot (Fig. 8) shows that the clinopyroxenes, both primaryand melt pocket, and from both Lashaine and Labait,have similar patterns with a gradual slope from the LREE
DISCUSSIONto the HREE (La/YbCN 3–6) although slightly convexupwards from Ce to Sm. The glasses have higher con- Equilibration conditionscentrations of the LREE (up to ×200 chondrite) and Five specimens contain both primary ortho- and clino-lower HREE, giving a steeper slope (La/YbCN 195–835). pyroxene, and temperatures of equilibration calculatedIn addition, all three glass patterns tend to have a weak using the two-pyroxene thermometer of Brey et al. (1990)negative Eu anomaly. The LREE preferential par- are given in Table 14. Three equilibrated in the rangetitioning into glass relative to clinopyroxene, combined 950–990°C similar to other garnet-free Lashaine peri-with the steeper REE patterns for the glasses, results in dotites, whereas two (including 1542, in which sym-a crossover in the middle REE (MREE) area (Fig. 8). plectites are interpreted as former garnets) equilibratedAn important point is that, on a volume-to-volume basis, at 1100–1150°C, within the range found for Lashainethe main REE-bearing phase in the peridotite (Cr-di- garnet lherzolites (Rudnick et al., 1994). Assuming the 44
mW/m2 geotherm established for the Lashaine garnetopside) is not able to provide the LREE concentrations
1770
DAWSON NORTHERN TANZANIAN PERIDOTITE XENOLITHS
Fig. 7. Chondrite-normalized trace element distribution in Lashaine and Labait clinopyroxenes and glasses. Data from Table 13. Normalizedagainst chondrite values of McDonough & Sun (1995).
lherzolites (Rudnick et al., 1994), depths of equilibration An Archaean age of not less than 3·4 Ga for this initialmelting event has been provided by Os isotopic dating ofrange from >115 to 140 km.a sulphide concentrate from a Lashaine garnet lherzolite(Burton et al., 2000). This date is comparable with theage of melt extraction and stabilization of the sub-
Ancient melting and Metasomatic Event I continental lithospheric mantle beneath the Kaapvaal,An early model for the evolution of the northern Tanzania Zimbabwe and Wyoming cratons (Carlson & Irving,upper mantle was based on the whole-rock chemistry of 1994; Pearson et al., 1995; Nagler et al., 1997). By contrast,Lashaine peridotites. It proposed a melting event during however, chromites from Labait xenoliths have yieldedwhich the peridotites were variably depleted in Fe, Ca, a relatively young Os age of 2·5–2·9 Ga (Chesley et al.,Al and alkalis relative to estimates of the composition 1999).of primitive upper mantle; this refractory residue was The earlier suggestion of a metasomatic overprint wassubsequently overprinted by K, Ti, Fe, Ca and REE based on whole-rock chemistry (high concentrations ofmetasomatism (Rhodes & Dawson, 1975; Ridley & Daw- incompatible trace elements combined with a refractoryson, 1975). The model did not attempt to give a time major-element chemistry) and the presence of phlogopiteframe for these events, nor could it predict whether the in Lashaine garnet peridotites (Dawson & Powell, 1969).metasomatic effects result from one or more events. Recent isotope work on Cr-diopside in a Lashaine phlo-
This early model is refined here in the light of more gopite-bearing garnet peridotite has provided a Nd andrecent data on Lashaine peridotite xenoliths, including Pb isotope model age of >2·0 Ga (Burton et al., 2000),those from the present study. The high amount of modal confirming earlier isotopic work (including data on phlo-olivine reported in Table 2 confirms the highly refractory gopite) for an event at around this time (Cohen et al.,nature of the peridotites, and the very low Ir/Pd ratios 1984). This age also coincides with a >2·0 Ga modelin some Lashaine peridotites reported by Rehkamper et age for the mafic granulites that are part of the Lashaine
xenolith suite (Cohen et al., 1984), which together suggestal. (1997) are further indications of extensive melt loss.
1771
JOURNAL OF PETROLOGY VOLUME 43 NUMBER 9 SEPTEMBER 2002
Fig. 8. Rare earth element plots for clinopyroxenes (filled symbols and×, continuous lines) and glasses (open symbols, broken lines) in Lashaineand Labait peridotites. Data and analysis numbers from Table 13. Normalized against chondrite values of McDonough & Sun (1995).
is that it may have formed by silica metasomatism of theTable 14: P–T conditions ofhigh-olivine restite during a subduction event, as argued
Lashaine peridotites by Rudnick et al. (1994) to explain the high amounts ofenstatite in Lashaine garnet peridotites. There is neither
Sample T (°C)∗ P (kbar)† textural nor isotopic evidence to resolve these alternatives.
771 990 3·9
1542 1130 4·25 Metasomatic Event II1544 950 3·75
The present study provides textural evidence for re-1546 1150 4·5
placement of Cr-spinel by both phlogopite and Cr-3926 950 3·75 diopside, replacement of primary enstatite by diopside,
replacement of Cr-diopside by metasomatic Ti-bronzite–∗Two-pyroxene thermometer of Brey et al. (1990). ilmenite–phlogopite aggregates, formation of high-Ti†Assuming 44 mW/m2 geotherm (see text).
rims on primary phlogopite and, in 1544, formation ofilmenite and its subsequent replacement by com-positionally variable Nb–Ta-rutile. These phenomenaa major chemical event in the northern Tanzania mantleindicate a metasomatic event during which there wasand deep crust at >2 Ga. This age is that of majoraddition of Si, K, Ti, Ca and Fe and, in 1544, Nb andsubduction around the eastern edge of the TanzaniaTa. Moreover, variations in the chemistry of differentCraton, manifest further south in Tanzania by the Us-grains of the same phase (e.g. rutile in 1544), togetheragaran orogeny (Moller et al., 1995); the effects of thiswith chemical zoning (e.g. in diopside replacing spinelorogeny in northern Tanzania have been largely erasedin 747) and overgrowth of compositionally different rimsby the later Pan-African orogeny (Moller et al., 1998).on primary phases (e.g. phlogopite in 1544) indicate thatThe status of enstatite in the primary assemblage isequilibrium has not been achieved. The diopside andambiguous. It might have been part of the restite as-
semblage after the 3·4 Ga melting event. An alternative Ti-phlogopite are similar to those formed during the
1772
DAWSON NORTHERN TANZANIAN PERIDOTITE XENOLITHS
earlier 2 Ga event, but this second set of unequilibrated An important point is that although peridotite mica mightprovide a small amount of water, the abundance of vugsphases could not have remained in the high thermalindicates that volatiles played an essential role in theregime of the upper mantle for >2 Gyr without an-formation of the melts.nealing. The inference is that the Lashaine peridotites
In the two Lashaine specimens for which there arecontain a second set of metasomatic phases that still hadphase trace element data (Table 13 and Fig. 7), althoughnot achieved equilibrium at the time of entrainment andthe protolith Cr-diopside could theoretically provide theeruption and which, by implication, must be youngerSr, Zr and Y concentrations found in the MP phases,than those formed at> 2 Ga. This conclusion provides anunrealistically large amounts would need to be meltedexplanation for the two isotopically and chronologicallyto account for the Nb, Pb and REE concentrations,distinct types of diopside identified in Lashaine peridotitesparticularly in the glasses; further, in the apparent absenceby Cohen et al. (1984).of mica (although the rapid modal variations referred toin the petrographic section must be borne in mind), thereis no protolith source for Rb and Ba. Even where protolith
Melting and its causes mica is present, an additional source is probably requiredVarious causes have been proposed for the formation of to account for the high concentrations of Rb and Ba inglasses in mantle xenoliths [see Shaw (1999) and ref- the MP phases, although the mica is a plausible sourceerences therein]. These include: (1) contact anatexis for Nb. In sample 1544, assuming rutile to be the mainaround a mantle magma chamber; (2) breakdown of source of Nb and Ta, the relatively high Ta in the MPhydrous phases; (3) decompression melting; (4) partial rutiles (Table 9, analyses 16 and 17) indicates preferentialmelting by a subsequent heat flux of a part of the mantle partitioning of Ta into the melt, but no need for anwhose solidus had been lowered by an earlier phase of additional external source for the Ta. In short, duringmetasomatism (Dawson, 1984; Hauri et al., 1993; Yaxley the partial melting in the Lashaine peridotites, it appears& Kamenetsky, 1999) and metasomatically triggered that there must have been an additional influx of Rb,melting (Ionov et al., 1994; Dawson, 1999). Ba, Nb, Pb and REE. Taken in conjunction with the
In the case of the Lashaine xenoliths, there is no petrography, which shows a relatively high amount ofevidence from the geothermometry studies (Rudnick et MP mica, diopside, zeolite and vug (former volatiles)al., 1994) for a thermal perturbation that might reflect a in the melt pockets (even in samples lacking primarymantle magma chamber, so possibility (1) appears un- phlogopite and Cr-diopside), it can be inferred that thelikely. A recognizable primary hydrous phase (mica) is trace element influxes accompanied an addition of K,present in only one xenolith (Table 1); however, in other Ti, Fe, Ca and, particularly, volatiles, both water andsamples there is the possibility that it may have been (inferred) carbon dioxide. This combination of elements,destroyed in the melting process. From several lines of although not necessarily relative concentrations, isevidence—the fact that the glasses are not in equilibrium strongly reminiscent of those added to the peridotitewith upper-mantle olivine; incomplete melting of relict protoliths during Metasomatic Event II, and it is im-phases within melt pockets; the presence of glasses of possible not to speculate that the elements and volatilesdifferent compositions in the same xenolith (implying found in the MP phases, although deriving in part fromlack of time to permit migration, mixing and ho- melted protolith phases, have also been supplementedmogenization)—it is apparent that the length of time by the metasomatic fluxes percolating the protolith duringbetween the formation of the melt pockets and their Metasomatic Event II. In summary, decompression, vol-subsequent quenching must have been very short. The atile release from pre-existing mica, and a volatile-richpreservation of glass itself testifies to rapid transport to metasomatic influx may all have contributed to thethe surface, which could be consistent with decompression melting.melting. In contrast, the Labait specimen trace element data
The protolith at Lashaine had undoubtedly undergone suggest certain differences from Lashaine, assuming prim-light-element metasomatism before the melting event, ary Cr-diopsides at Labait contain similar trace elementbut the pertinent question is whether there is evidence concentrations to that analysed from Lashaine. Cr-di-for a metasomatic flux at the time of melting and, if so, opside could provide sufficient Y, Zr and possibly butis it possibly tied to the younger Metasomatic Event II? not enough Nb, Sr, Pb and LREE for the melt-pocketOn the assumption that diopside and mica might be assemblage. Protolith mica, for which there are no ana-expected to be the main contributors to melts generated lytical data for Rb, might have been a sufficient sourceat the onset of melting of the peridotite protolith, it is for Rb but the Labait protolith mica (1340 ppm Ba;possible to assess qualitatively whether these two phases Dawson, 1999) could not provide the high Ba con-alone are capable of satisfying the melt-pocket phase centrations found in the Labait MP phases. Thus, at
Labait a Nb, Sr, LREE, Ba and ?Rb influx is required;chemistry, or whether an additional influx is required.
1773
JOURNAL OF PETROLOGY VOLUME 43 NUMBER 9 SEPTEMBER 2002
the presence of Sr-bearing apatite and the Ba-zeolite concluded that only glass inclusions armoured withinmantle minerals have not been strongly modified by aharmotome in most Labait melt pockets particularly
emphasizes the necessity for high Sr and Ba influxes decrease in pressure; hence, they suggested that glassesfound as intergranular films and pockets do not preserve(Dawson, 1999).
Following the melt formation, precipitation of new the composition of the melts formed at depth. Finally,there are differences of opinion as to whether ex-phases of distinctive composition took place. The diopside
contains less jadeite–ureyite molecule than the peridotite perimental glasses can be used to interpret the formationconditions of mantle glasses; the debate concerns whetherdiopsides and the micas are in general more titaniferous
than the primary micas. An unusual aspect is the crys- the diamond-aggregate approach or the sandwich tech-nique is the better experimental method for the achieve-tallization of olivine that is higher in Mg than in the
peridotite protolith, and its intimate association with ment of equilibrium (Draper & Green, 1999).There is, moreover, a preoccupation with the glassprecipitated spinel (see Fig. 3b). Similar Mg-rich olivines
in melt pockets in peridotite xenoliths from Mongolia chemistry; and in some studies glass compositions areequated with magma types ranging from basanite toare attributed to incongruent melting of primary clino-
pyroxene and spinel to give residual chromite, pheno- trachyte, the preoccupation extending to the glasses beingplotted on TAS (total alkalis vs silica) diagrams of Lecrysts of olivine and clinopyroxene of a new composition
and melt; subsequent fractional crystallization of the melt Maitre (see Draper & Green, 1997, fig. 1; Shaw, 1999,fig. 1). In some studies the compositions and amounts ofwas limited to new rims on the olivine and clinopyroxene
phenocrysts (Ionov et al., 1994). This model could be the precipitated phases (if any) are largely ignored (e.g.Varela et al., 1999) even when these have been extensivelyapplied in part to the Lashaine rocks, but the Mongolian
case is unlike the Lashaine examples in which mica is analysed (e.g. Chazot et al., 1996). In these studies, thefundamental principle is ignored that, unless there is aalso involved. The close association of the high-Mg olivine
with inclusions of euhedral chromite is significant, and a total absence of phenocrysts or microlites, glass com-positions do not equate with the original melts.possible explanation is that the highly magnesian olivine
(mg 0·92–0·94) formed when the melt was briefly depleted In the case of Lashaine, the original melts have beenmodified by precipitation of variable amounts of olivine,in Fe as a result of preferential Fe partitioning into the
earlier-precipitated chromite (mg 0·47–0·80). diopside, mica and spinel, so an attempt has been madeto compute the chemistry of the melt in one of the largerLashaine melt pockets by integrating the mode of themelt pocket and the chemistry of the phases (Table 15).
Glass compositions—comparisons with In the computation zeolite has been included because,magmas as has been argued above, it is a late-crystallizing fluidOver the past decade there has been intense interest and phase (rather than a deposit from groundwater). Broadly,debate on the occurrence and chemistry of the glassy the computed composition is silica undersaturated, lowcomponents of upper-mantle xenoliths, together with the in alumina and has K > Na, parameters that, withreasons for the melting. In particular, much attention differences in detail, would group it with low-volumehas been given to the worldwide occurrence of glasses potassic magma types such as kimberlite, katungite andhigh in Na, Al and Si (e.g. Schiano & Clocchiatti, 1994; olivine lamproite. Because of problems with measuringDraper & Green, 1997). However, the chemistry of the mode, specific comparisons should be treated withmantle glasses is variable and, not surprisingly, opinions caution, but the computed composition can be mostfor the variation are diverse. Distinctions have been made closely matched with olivine–madupitic lamproite (Tablebetween the formation of glasses in which the chemistry 15, analysis 3). In the more limited northern Tanzaniahas been influenced by melted hydrous phases (amphibole context, it has some affinities with katungite from Pelloor phlogopite), and those in anhydrous peridotites. In Hill, another xenolith locality (analysis 4). Importantly,the context of metasomatic triggering, Coltorti et al. (2000) its high K/Na ratio precludes any link with the sodicappealed to different metasomatic fluxes accompanying ankaramite that hosts the Lashaine peridotites (analysisthe melting to explain chemical variations in glasses 5).in anhydrous xenoliths; glasses related to carbonatitemetasomatism are stated to be characterized by highCaO and Na2O, and low SiO2, and have high Na2O/
Comparison with other Tanzanian xenolithK2O >2, whereas glasses related to K-alkali silicatesuitesmetasomatism are characterized by high SiO2 and K2O
and Na2O/K2O of <1. Another area for debate is that, The data presented above indicate that, at the time ofthe glass pocket formation in the Lashaine peridotites,whereas most researchers have accepted the chemistry
of the glasses to be real, Schiano & Bourdon (1999) there was an influx of K, Ca, Rb, Sr, Ba, Zr, Nb, Pb,
1774
DAWSON NORTHERN TANZANIAN PERIDOTITE XENOLITHS
of the metasomatism accompanying the melting is vari-Table 15: Calculated composition of Lashaineable, and that the metasomatism of each xenolith suite
melt-pocket magma, and comparative low- has a distinctive signature. The inference is that involume mantle-derived magmas relatively recent geological times, and coinciding with
volatile-rich nephelinitic–carbonatitic regional mag-matism, the sub-Tanzanian mantle lithosphere has been1 2 3 4 5percolated by metasomatizing fluids that are hetero-geneous over a small geographical area.SiO2 44·86 49·20 48·86 38·47 40·4
TiO2 0·91 1·00 1·46 4·35 2·43
Cr2O3 2·85 3·13 n.a. n.a. 0·11
Al2O3 7·45 8·17 8·16 7·55 5·98ACKNOWLEDGEMENTSFeO∗ 5·76 6·32 6·19 11·4 12·87
I am grateful to Peter Hill and Paula McDade for helpMnO 0·10 0·11 — 0·26 0·17with the electron microprobe analyses and SEM images,MgO 16·58 18·19 16·45 14·72 18·10and to John Craven for assistance with the ion-micro-NiO 0·02 0·02 — n.a. n.a.probe analyses. The analytical work was funded byCaO 8·72 9·56 7·01 6·40 12·53NERC Grant GR9/02883. I also acknowledge perceptive
Na2O 0·28 0·31 1·62 1·84 2·02reviews by Leonid Danyushevsky, Massimo Coltorti and
K2O 3·64 3·99 5·28 2·68 1·01a third anonymous referee, which have much improved
Total 91·15 100·00 — 95·42† 97·14†the original manuscript. The samples were collectedwhen I was a member of the Tanganyika Geological
∗All iron as FeO. Survey, and during later field-work supported by the†Low totals reflect high Fe2O3 in original analyses (nowCarnegie Trust for the Universities of Scotland. Therecalculated to FeO).
1. Calculated composition of magma in melt pocket in Mg- analytical facilities at Edinburgh are partly supported byperidotite 3926. Based on chemistry and volumetric abund- NERC.ances of the following melt-pocket phases: olivine 7% (Table3, analysis 18), diopside 31% (Table 9, analysis 13, total FeO3·28 wt %), mica 24% (Table 6, analysis 14), glass 20% (Table11, analysis 9), zeolite 11% (Table 12, analysis 4), spinel4% (Table 10, analysis 11, total FeO 19·50%); calcite 1% of
REFERENCESstandard calcite (56% CaO). Total ignores small amounts ofBaO, P2O5 and CO2. It should be noted that no allowances Brey, G. P., Kohler, T. & Nickel, K. G. (1990). Geothermometry inhave been made for density differences between the phases. four-phase lherzolites. I. Experimental results from 10 to 60 kb.2. Analysis 1 recalculated to 100%. Journal of Petrology 31, 1313–1352.3. Average olivine–madupitic lamproite, Murcia–Almeira (Mit- Burton, K. W., Schiano, P., Birck, J.-L., Allegre, C. J., Rehkamper,chell & Bergman, 1991, table 7.2).
M., Halliday, A. N. & Dawson, J. B. (2000). The distribution and4. Katungite, Pello Hill, Tanzania (Dawson & Smith, 1988).behaviour of rhenium and osmium amongst mantle minerals andTotal includes 4·90% H2O, 0·37% P2O5 and 2·48% CO2.the age of the lithospheric mantle beneath Tanzania. Earth and5. Lashaine ankaramite (Dawson et al., 1970). Total includesPlanetary Science Letters 183, 93–106.0·83% H2O+, 0·83% P2O5 and 1·44% CO2.
Carlson, R. W. & Irving, A. J. (1994). Depletion and enrichment historyof subcontinental lithospheric mantle: an Os, Sr, Nd and Pb isotopicstudy of ultramafic xenoliths from the northwestern Wyoming craton.
REE, CO2 and H2O. In the case of granulite xenoliths Earth and Planetary Science Letters 126, 457–472.erupted at Lashaine at the same time as the peridotites, Carswell, D. A. (1975). Primary and secondary phlogopites and clino-
pyroxenes in garnet lherzolite xenoliths. Physics and Chemistry of themelting and K-metasomatism has also occurred ( JonesEarth 9, 417–430.et al., 1983c). By comparison, glass formation in the
Chazot, G., Menzies, M. & Harte, B. (1996). Silicate glasses in spinelOlmani xenoliths was accompanied by addition of K,lherzolites from Yemen: origin and chemical composition. ChemicalCa, Na, Ti, P, REE, Cl, F and CO2 ( Jones et al., 1983a;Geology 134, 159–179.
Rudnick et al., 1993, 1994); Ca, P, REE and CO2 addition Chesley, J. T., Rudnick, R. L. & Lee, C.-T. (1999). Re–Os systematicswas particularly high, resulting in the formation of new of mantle xenoliths from the East African Rift: age, structure anddiopside and monazite and the Olmani glasses are more history of the Tanzanian craton. Geochimica et Cosmochimica Acta 63,
1203–1217.aluminous (mainly 11–14 wt %) than those in the LahaineCohen, R. S., O’Nions, R. K. & Dawson, J. B. (1984). Isotopemelt pockets, and most are sodic. The melting at Labait
geochemistry of xenoliths from East Africa: implications for thewas characterized by addition of K, Ti, Ba, Sr, H2O, Pdevelopment of mantle reservoirs and their interaction. Earth andand CO2 (Dawson, 1999), Ba being particularly enhancedPlanetary Science Letters 68, 209–220.
and resulting in the formation of harmotome and high- Coltorti, M., Beccaluva, L., Bonadiman, C., Salvini, L. & Siena, F.Ba mica and glass. Taken together, the evidence from (2000). Glasses in mantle xenoliths as geochemical indicators of
metasomatic agents. Earth and Planetary Science Letters 183, 303–320.these three xenolith localities is that the detailed chemistry
1775
JOURNAL OF PETROLOGY VOLUME 43 NUMBER 9 SEPTEMBER 2002
Dawson, J. B. (1964). Carbonate tuff cones in northern Tanganyika. Henjes-Kunst, F. & Altherr, R. (1992). Metamorphic petrology ofxenoliths from Kenya and northern Tanzania and implicationsGeological Magazine 101, 129–137.
Dawson, J. B. (1984). Contrasting types of mantle metasomatism? for geotherms and lithospheric structures. Journal of Petrology 33,1125–1156.In: Kornprobst, J. (ed.) Kimberlites II; the Mantle and Crust–Mantle
Relationships. Amsterdam: Elsevier, pp. 289–294. Hervig, R. L., Smith, J. V., Steele, I. M. & Dawson, J. B. (1980).Fertile and barren Al–Cr-spinel harzburgites from the upper mantle:Dawson, J. B. (1987). Metasomatised harzburgites in kimberlite and
alkaline magmas; enriched restites and ‘flushed’ lherzolites. In: ion and electron probe analyses of trace elements in olivine andorthopyroxene: relationship to lherzolites. Earth and Planetary ScienceMenzies, M. A. & Hawkesworth, C. J. (eds) Mantle Metasomatism.
London: Academic Press, pp. 125–144. Letters 50, 41–58.Hutchison, R. & Dawson, J. B. (1970). Rb, Sr and 87Sr/86Sr in ultrabasicDawson, J. B. (1992). Neogene tectonics and volcanicity in the north
Tanzania sector of the Gregory Rift Valley: contrasts with the Kenya xenoliths and host rocks, Lashaine volcano, Tanzania. Earth and
Planetary Science Letters 9, 87–92.sector. Tectonophysics 204, 81–92.Dawson, J. B. (1999). Metasomatism and melting in spinel peridotite Ionov, D. A., Hofmann, A. W. & Shimizu, N. (1994). Metasomatism-
induced melting in mantle xenoliths from Mongolia. Journal of Petrologyxenoliths from Labait, Tanzania. In: Gurney, J. J., Gurney, J. L.,Pascoe, M. D. & Richardson, S. H. (eds) Proceedings Volume of 7th 35, 753–785.
Jones, A. P., Smith, J. V. & Dawson, J. B. (1983a). Glasses in mantleInternational Kimberlite Conference, 1. Cape Town: Red Roof Design,pp. 164–173. xenoliths from Olmani, Tanzania. Journal of Geology 91, 167–178.
Jones, A. P., Smith, J. V. & Dawson, J. B. (1983b). Mantle metasomatismDawson, J. B. & Hill, P. G. (1998). Mineral chemistry of a peralkalinecombeite–lamprophyllite nephelinite from Oldoinyo Lengai, Tan- in 14 veined peridotites from Bultfontein Mine, South Africa. Journal
of Geology 90, 435–453.zania. Mineralogical Magazine 62, 179–196.Dawson, J. B. & Powell, D. G. (1969). Mica in the upper mantle. Jones, A. P., Smith, J. V., Dawson, J. B. & Hansen, E. C. (1983c).
Metamorphism, partial melting and K-metasomatism of garnet–Contributions to Mineralogy and Petrology 22, 233–237.Dawson, J. B. & Smith, J. V. (1973). Alkalic pyroxenite xenoliths from scapolite–kyanite granulites from Lashaine, Tanzania. Journal of
Geology 91, 143–165.the Lashaine volcano, northern Tanzania. Journal of Petrology 14,113–131. McDonough, W. F. & Frey, F. A. (1989). Rare earth elements in upper
mantle rocks. In: Lipin, B. R. & McKay, G. A. (eds) Geochemistry andDawson, J. B. & Smith, J. V. (1977). The MARID (mica–amphibole–rutile–ilmenite–diopside) suite of xenoliths in kimberlite. Mineralogy of Rare Earth Elements. Mineralogical Society of America, Reviews
in Mineralogy 16, 100–145.Geochimica et Cosmochimica Acta 41, 309–323.Dawson, J. B. & Smith, J. V. (1988). Metasomatised and veined McDonough, W. F. & Sun, S. S. (1995). The composition of the Earth.
Chemical Geology 120, 223–253.upper mantle xenoliths from Pello Hill, Tanzania: evidence foranomalously-light mantle beneath the Tanzanian sector of the East Meyer, H. O. A. (1987). Inclusions in diamond. In: Nixon, P. H. (ed.)
Mantle Xenoliths. Chichester: Wiley, pp. 501–522.African Rift Valley. Contributions to Mineralogy and Petrology 100, 510–527. Mitchell, R. H. (1986). Kimberlites: Mineralogy, Geochemistry and Petrology.
New York: Plenum.Dawson, J. B., Powell, D. G. & Reid, A. M. (1970). Ultrabasic xenolithsand lava from the Lashaine volcano, northern Tanzania. Journal of Mitchell, R. H. & Bergman, S. C. (1991). Petrology of Lamproites. New
York: Plenum.Petrology 11, 519–548.Delaney, J. S., Smith, J. V., Carswell, D. A. & Dawson, J. B. (1980). Moller, A., Appel, P., Mezger, K. & Schenk, V. (1995). Evidence for
a 2 Ga subduction zone: eclogites in the Usagaran belt of Tanzania.Chemistry of micas from kimberlites and xenoliths—II. Primary-and secondary-textured micas from peridotite xenoliths. Geochimica Geology 23, 1067–1070.
Moller, A., Mezger, K. & Schenk, V. (1998). Crustal age domains andet Cosmochimica Acta 44, 857–872.Draper, D. S. & Green, T. H. (1997). P–T phase relations of silicic, the evolution of the continental crust in the Mozambique Belt of
Tanzania: combined Sm–Nd, Rb–Sr, and Pb–Pb isotopic evidence.alkaline, aluminous mantle-xenolith glasses under anhydrous andC–O–H fluid-saturated conditions. Journal of Petrology 38, 1187–1224. Journal of Petrology 39, 749–783.
Nagler, Th. F., Kramers, J. D., Kamber, B. S., Frei, R. & Prendergast,Draper, D. S. & Green, T. H. (1999). P–T relations of silicic, alkaline,aluminous liquids: new results and applications to mantle melting M. D. A. (1997). Growth of the subcontinental lithospheric mantle
beneath Zimbabwe started at or before 3·8 Ga: Re–Os study onand metasomatism. Earth and Planetary Science Letters 170, 255–268.Gregoire, M., Lorand, J. P., O’Reilly, S. Y. & Cottin, J. Y. (2000). chromites. Geology 25, 983–986.
Nielson, J. E. (1989). Nature and sequence of mantle magmatic eventsArmalocolite-bearing Ti-rich metasomatic assemblages in harz-burgite xenoliths from the Kerguelen Islands: implications for the indicated by compositional variations in a composite xenolith from
Lashaine Volcano, Tanzania. In: Ross, J. (ed.) Kimberlite and Relatedoceanic budget of high-field strength elements. Geochimica et Cosmo-
chimica Acta 64, 673–694. Rocks. Geological Society of Australia, Special Volume 14, 700–708.Pearson, D. G., Carlson, R. W., Shirey, S. B., Boyd, F. R. & Nixon,Haggerty, S. H. (1983). The mineral chemistry of new titanates from
the Jagersfontein kimberlite: implications for metasomatism in the P. H. (1995). Stabilisation of the Archaean lithospheric mantle: aRe–Os isotope study of peridotite xenoliths from the Kaapvaalupper mantle. Geochimica et Cosmochimica Acta 47, 1833–1854.
Haggerty, S. E. (1987). Metasomatic mineral titanates in upper mantle craton. Earth and Planetary Science Letters 134, 341–357.Pike, J. E. N., Meyer, C. E. & Wilshire, H. G. (1980). Petrographyxenoliths. In: Nixon, P. H. (ed.) Mantle Xenoliths. Chichester: Wiley,
pp. 671–690. and chemical composition of a suite of ultramafic xenoliths fromLashaine, Tanzania. Journal of Geology 88, 343–352.Harte, B. & Gurney, J. J. (1975). Ore mineral and phlogopite min-
eralisation within ultramafic nodules from the Matsoku kimberlite Rehkamper, M., Halliday, A. N., Barfod, D., Fitton, J. G. & Dawson,J. B. (1997). Platinum group element abundance patterns in differentpipe, Lesotho. Carnegie Institution of Washington Yearbook 74, 528–536.
Hauri, E. H., Shimizu, N., Dieu, J. J. & Hart, S. R. (1993). Evidence mantle environments. Science 278, 1595–1598.Reid, A. M. & Dawson, J. B. (1972). Olivine–garnet reaction infor hotspot-related metasomatism in the oceanic upper mantle. Nature
365, 221–227. peridotites from Lashaine, Tanzania. Lithos 5, 115–124.
1776
DAWSON NORTHERN TANZANIAN PERIDOTITE XENOLITHS
Reid, A. M., Donaldson, C. H., Brown, R. W., Ridley, W. I. & versus interstitial glasses. Earth and Planetary Science Letters 169, 173–188.Dawson, J. B. (1975). Mineral chemistry of peridotite xenoliths from
Schiano, P. & Clocchiatti, R. (1994). Worldwide occurrence of silica-the Lashaine volcano, Tanzania. Physics and Chemistry of the Earth 9,rich melts in sub-continental and sub-oceanic mantle minerals. Nature545–557.368, 621–624.Rhodes, J. M. & Dawson, J. B. (1975). Major and trace element
Schiano, P., Clocchiatti, R., Bourdon, B., Burton, K. W. & Thellier,chemistry of peridotite inclusions from the Lashaine volcano, Tan-B. (2000). The composition of melt inclusions in minerals at thezania. Physics and Chemistry of the Earth 9, 545–557.garnet–spinel transition zone. Earth and Planetary Science Letters 174,Ridley, W. I. & Dawson, J. B. (1975). Lithophile trace element data375–383.bearing on the origin of peridotite xenoliths, ankaramite and car-
Shaw, C. S. J. (1999). Dissolution of orthopyroxene in basanitic magmabonatite from the Lashaine volcano, northern Tanzania. Physics andbetween 0·4 and 2 GPa: further implications for the origin of Si-
Chemistry of the Earth 9, 559–569.rich alkaline glass inclusions in mantle xenoliths. Contributions toRoeder, P. L. & Emslie, R. F. (1970). Olivine–liquid equilibrium.Mineralogy and Petrology 135, 114–132.
Contributions to Mineralogy and Petrology 29, 275–289.Smith, J. V. & Dawson, J. B. (1975). Chemistry of Ti-poor spinels,
Rudnick, R. L., McDonough, W. F. & Chappell, B. W. (1993).ilmenites and rutiles from peridotite and eclogite xenoliths. Physics
Carbonatite metasomatism in the northern Tanzanian mantle: petro-and Chemistry of the Earth 9, 309–322.
graphic and chemical characteristics. Earth and Planetary Science Letters Varela, M. E., Clocchiatti, R., Kurat, G. & Schiano, P. (1999). Silicic114, 463–475. glasses in hydrous and anhydrous mantle xenoliths from Western
Rudnick, R. L., McDonough, W. F. & Orpin, A. (1994). Northern Victoria, Australia: at least two different sources. Chemical GeologyTanzania peridotite xenoliths: a comparison with Kaapvaal 153, 151–169.xenoliths and inferences of metasomatic reactions. In: Meyer, H. Yaxley, G. M. & Kamenetsky, V. (1999). In situ origin for glass inO. A. & Leonardos, O. H. (eds) Kimberlites, Related Rocks and mantle xenoliths from southeastern Australia: insights into traceMantle Xenoliths. Proceedings Volumes, 5th International Kimberlite Conference, element compositions of glasses and metasomatic phases. Earth andAraxa, Brazil 1. Brasilia: Companhia de Pesquisa de Recursos Planetary Science Letters 172, 97–109.Minerais, pp. 336–353. Zinner, E. & Crozaz, G. (1986). A method for the quantitative measure-
Schiano, P. & Bourdon, B. (1999). On the preservation of mantle ment of rare earth elements in the ion microprobe. International
Journal of Mass Spectrometry and Ion Processes 69, 17–38.information in ultramafic nodules: glass inclusions within minerals
1777