Full text of DOI:10.1007/s004100100287

Contrib Mineral Petrol &2001) 142: 127±146
DOI 10.1007/s004100100287
G. Demarchi á P. Antonini á E.M. Piccirillo
G. Orsi á L. Civetta á M. D'Antonio
Signi®cance of orthopyroxene and major element constraints
on the petrogenesis of Ferrar tholeiites from southern
Prince Albert Mountains, Victoria Land, Antarctica
Received: 12 July 2000 / Accepted: 23 May 2001 / Published online: 14 July 2001
Ó Springer-Verlag 2001
Abstract The least evolved Jurassic Ferrar tholeiites ing support the argument that theoretical magma com-
from southern Prince Albert Mountains &Antarctica) are positions suitable to be primary to the least evolved
characterized by the occurrence of orthopyroxene &opx), Ferrar tholeiites are compatible with hydrous
a mineralogical feature virtually absent in the tholeiites &H2O=0.3±0.5 wt%) and anhydrous melts obtained at
from the large igneous provinces of Karoo &South Af- 10±15 kbar by high melting degrees &>25%) of fertile
rica) and Parana &Brazil). Petrography suggests that opx and depleted spinel lherzolites, respectively, and later
is the early phase in the sequence of crystallization and contaminated by the high-grade metamorphic rocks
mineral chemistry indicates that it is in equilibrium with from the Victoria Land crystalline basement. The gen-
the host rock. In general, MELTS modeling predicts esis of primary Ferrar tholeiites does not necessarily
that opx is the liquidus phase in the Ferrar tholeiites re¯ect the generally assumed depleted source mantle
with MgO higher than ꢀ7 wt% at P=1.5±5 kbar, being also compatible with a fertile one.
H2O=0±1 wt% and fO2 ˆ QFM À 1 log unit condi-
tions. MELTS results also show that the early crystal-
lization of opx is primarily controlled by high SiO2 and
high SiO2/CaO, chemical characteristics typical of the
Ferrar tholeiites, but not shown by the Karoo and Pa-
rana analogs with similar MgO content. Major element
Introduction
Large-scale magmatic events produced large volumes of
geochemistry of the least evolved Ferrar tholeiite has
continental ¯ood basalts &CFB) associated with the
been modeled through fractional crystallization and
break-up of the Gondwana supercontinent during the
fractional crystallization coupled with crustal assimila-
Mesozoic &Macdougall 1988; Storey et al. 1992; Ma-
tion processes, starting from natural peridotite-derived
honey and Con 1997). The CFB of Parana &SE-Brazil,
experimental melts. Mass balance and MELTS model-
mainly 133±132 Ma; Renne et al. 1992; Turner et al.
1994; Stewart et al. 1996), Karoo &South Africa, about
180 Ma; Encarnacion et al. 1996; Duncan et al. 1997;
G. Demarchi &&) á E.M. Piccirillo
Marsh et al. 1997), and Ferrar &Antarctica, about
180 Ma; Heimann et al. 1994; Encarnacion et al. 1996;
Fleming et al. 1997; Elliot et al. 1999) constitute the
three major CFB provinces of the southern hemisphere.
The Ferrar magmatic province &e.g., Hergt et al. 1991)
extends in a linear belt from Antarctica to Tasmania and
New Zealand &Mortimer et al. 1995) and tends to be
parallel to the proto-Paci®c margin of Gondwana &e.g.,
Cox 1988). Ferrar magmatism belongs to the Gondwana
low-Ti province &Cox 1988) and di€ers from the other
two CFB provinces for its systematically high initial
87Sr/86Sr &>0.707), low 143Nd/144Nd &<0.5124), and
high SiO2 content &usually>52 wt%), even in the least
evolved rocks &MgO=9±10 wt%; e.g., Gunn 1966;
Hergt et al. 1989). These features, accompanied by high
contents of large ion lithophile and light rare earth ele-
ments, have been attributed to &1) mantle source
Dipartimento di Scienze della Terra, Universita di Trieste,
Via E. Weiss 8, 34127 Trieste, Italy
E-mail: demarchi@univ.trieste.it
P. Antonini
Dipartimento di Ingegneria Chimica,
dell' Ambiente e delle Materie Prime,
Universita di Trieste, P.le Europa 1, 34127 Trieste, Italy
G. Orsi
Osservatorio Vesuviano, Via Diocleziano 328,
80124 Napoli, Italy
L. Civetta
Dipartimento die Scienze Fisiche, Universita di Napoli,
Mte S. Angelo, Via Cinzia, 80126 Napoli, Italy
M. D'Antonio
Dipartimento di Scienze della Terra, Universita di Napoli,
Largo S. Marcellino 10, Napoli, Italy
Editorial responsibility: T.L. Grove

128

129

130
contaminated by crustal components during ancient
subduction processes &Hergt et al. 1989; Brewer et al.
1992; Molzahn et al. 1996; Hergt 2000), or by ``global
lithosphere recycling'' &Menzies and Kyle 1990), or &2)
basalt magmas crustally contaminated during their up-
rising to the surface &Faure et al. 1974, 1982; Fleming
et al. 1995; Antonini et al. 1999, 2000). The current
petrogenetic models, mainly based on trace element and
isotopic geochemistry, are &1) 3% addition of subducted
sediments &PAAS; post-Archean Australian shale; Tay-
lor and McLennan 1985) to a depleted MORB mantle
source. The primary tholeiitic melts would not be ap-
preciably modi®ed by low-pressure crustal contamina-
tion during their ascent to the surface &Hergt et al. 1989;
Molzahn et al. 1996); &2) assimilation combined with
fractional crystallization &AFC) processes, whereby the
primary Ferrar magmas, generated by high melting de-
grees of E-type Dupal MORB mantle, were contami-
nated by the granulitic rocks of the Victoria Land
crystalline basement &Antonini et al. 1999, 2000).
In general, little attention has been paid to the sig-
ni®cance of the Ferrar major element geochemistry and
the occurrence of orthopyroxene in the least evolved
Ferrar rocks. Hergt et al. &1991) suggested that the high
SiO2 of the Ferrar tholeiites cannot be totally ascribed to
fractional crystallization, representing instead a primary
feature caused by, in part, melting of the hydrous
mantle. Turner and Hawkesworth &1995), revisiting the
importance of the major element geochemistry to con-
strain the mantle source regions of many CFB provinces,
concluded that the observed major element composition
of many CFB, including Ferrar and Tasmania tholeiites,
would result from ``melting of depleted peridotite at
volatile-present solidus, under ¯uid absent conditions''.
However, a model that accounts for the whole major
element geochemistry of Ferrar is not known to the
authors of this paper.
In this paper we present new geochemical and
mineralogical data for Ferrar tholeiites from southern
Prince Albert Mountains and evaluate the signi®cance
of orthopyroxene as a liquidus phase. It will be shown
that the early crystallization of orthopyroxene in the
least evolved tholeiites is primarily controlled by high
SiO2/CaO, and particularly by high SiO2 content of
Ferrar magmas. Focusing on the origin of the SiO2-rich
signature of the least evolved Ferrar tholeiites, our
approach has been to use experimental melts of natural
peridotites as primary magmas for modeling the Ferrar
major element geochemistry through fractional crys-
tallization &FC) and assimilation coupled with frac-
tional crystallization &AFC). Anhydrous and hydrous
experimental picritic melts produced by fertile perido-
tite &Hirose and Kushiro 1993; Hirose and Kawamoto
1995) and anhydrous melts produced by a depleted
peridotite &Jaques and Green 1980) have been em-
ployed in order to take into account the role of hy-
drous melting and di€erent peridotite compositions on
the genesis of Ferrar tholeiites. We will show that,
starting from these picritic melts, the AFC process is

131
necessary for ®tting the major element composition of Armienti et al. 1990). The BT and WT basement is
the least evolved Ferrar tholeiite, and that both de- overlain by the Permo-Triassic continental sedimentary
pleted and fertile peridotite are suitable Ferrar mantle sequence of the Beacon Supergroup and by the Middle
sources.
Jurassic magmatic rocks of the Ferrar Supergroup. The
latter consists of sills &Ferrar Dolerite) intruding the
Beacon Supergroup, lava ¯ows &Kirkpatrick Basalt),
and minor pyroclastic rocks &Mawson Formation and
Exposure Hill Formation) separating the Kirkpatrick
Basalt from the Beacon sediments &Elliot et al. 1986;
Elliot and Larson 1993; Roland and Worner 1996). Late
Cenozoic alkaline volcanic rocks, related to the West
Antarctic Rift System, outcrop in a linear belt parallel to
the Ross Sea coast in Victoria Land.
In this work we studied the sills of the Ferrar Dolerite
and the lavas of the Kirkpatrick Basalt from the
southern sector of Prince Albert Mountains, which ex-
tends south of David Glacier. In this area, the nunataks
most distal from the coast line are formed from the
Kirkpatrick Basalt &Worner 1992), whereas the Ferrar
Dolerite mainly outcrops towards the Ross Sea. Samples
were collected by G. Orsi during the 1994±1995 Ant-
arctic Italian Expedition. The location of samples is
shown in the sketch map of Fig. 1.
Geological setting
Three metamorphic ``Terrains'', separated by N±S to
NW±SE major faults, form the crystalline basement in
Victoria Land &Fig. 1). They are, from the Ross Sea
towards the East Antarctic craton, the Robertson Bay
&RBT), the Bowers &BT), and the Wilson &WT) Terrains
&Bradshaw 1987; Lombardo et al. 1991). RBT consists of
Cambrian±Ordovician
low-grade
metagraywackes
&Burrett and Findlay 1984). BT consists of low-grade
metavolcanics &Weaver et al. 1984) overlain by low-
grade quartzo-feldspathic metasediments of Middle
Cambrian to Early Ordovician age. WT is mainly
formed by medium-grade schists and gneisses, including
a metasedimentary sequence of Late Precambrian±
Cambrian age &Casnedi and Pertusati 1991), older
polymetamorphic migmatites with remnants of calc-al-
kaline meta-igneous rocks, ma®c granulites, and high-
grade metasedimentary rocks. RBT and BT are intruded
by the Devonian granitic plutons of the Admiralty In- Classi®cation and petrography
trusives suite &ꢀ400 Ma; Borg et al. 1987), whereas WT
is intruded by the Late-Cambrian±Ordovician Granite The studied Ferrar tholeiites mostly correspond to ba-
Harbor Intrusives &450±550 Ma; Borg et al. 1987; saltic andesites and subordinate andesites according to
Fig. 2 Ca±Mg±Fe* &Fe2+
+
Mn + Fe3+) pyroxene compo-
sitions &after Poldervaart and
Hess 1951, modi®ed) of the
investigated rocks from Prince
Albert Mts. a Coarse-grained
dolerites &samples VG1039,
VG1026, LZ57A, and LZ6B);
b chilled margin dolerite &trian-
gles:VG1048), lava ¯ow &®lled
circles:VG1028) and pillow la-
vas &half-®lled circles:VG1019,
VG1052). The arrow indicates
the ``quench trend'' of Muir and
Mattey &1982). The composi-
tional pyroxene trends of Dufek
intrusion &dashed lines; Him-
melberg and Ford 1976) and
Thern Promontory and Archa-
mbault Ridge &shaded ®elds;
Brotzu et al. 1988) are shown
for comparison

132
Table 4 Major &wt%) and trace &ppm) element abundances of wavelength dispersive X-ray ¯uorescence with PW1404 XRF
representative Ferrar tholeiites from southern Prince Albert spectrometer and the PhilipsÒ &1994) procedure for the correction
Mountains. mg# at. 100*Mg/&Mg+Fe2+), assuming Fe2O3/ of matrix e€ects; analytical error is less than 5% and 10% for
FeO=0.15. Analyses were carried out at the Dipartimento di Sci- major and trace elements, respectively
enze della Terra, University of Trieste, on pressed powder pellets by
Sample VG1039 VG1025 VG1026 VG1004 VG1006 VG1003 VG1048 VG1015 VG1016 VG1029 VG1028 VG1019 VG1052
Pillow Pillow
Rock
type
Sill
Sill
Sill
Sill
Sill
Sill c.m. Sill c.m. Sill c.m. Lava
Lava
Lava
SiO2
53.03
53.75
53.88 55.67
0.51 0.81
15.46 14.28
59.66
54.11
0.62
15.00
8.85
0.17
7.46
10.81
1.81
1.06
0.10
53.89
0.71
15.25
9.21
0.21
6.84
11.45
1.72
0.60
0.11
54.42
0.70
15.27
9.47
0.19
6.35
11.02
1.69
0.79
0.10
53.80
0.66
14.81
9.05
0.20
7.70
11.18
1.62
0.88
0.10
54.47 54.50
0.68 0.67
14.51 14.73
55.97
0.95
14.83
10.73
0.18
4.25
9.62
2.13
1.22
0.12
56.55
1.07
13.65
11.89
0.20
4.30
8.83
2.41
1.00
0.10
TiO2
Al2O3
FeOt
MnO
MgO
CaO
Na2O
K2O
0.44
14.44
8.57
0.17
9.68
11.42
1.60
0.58
0.07
0.46
16.76
7.93
0.16
7.25
11.14
1.81
0.65
0.08
1.05
14.25
10.33
0.15
2.61
7.05
2.61
2.05
0.23
8.84
0.17
7.68
10.69
1.91
0.77
0.09
9.93
0.17
6.04
8.76
2.79
1.43
0.12
9.64
0.19
6.97
9.70
0.19
6.78
11.09 11.09
1.79
0.56
0.10
1.84
0.40
0.09
P2O5
Sum
mg#
Cr
Ni
Rb
Sr
100.00
69.6
373
135
22
102
14
186
<10
111
10
99.99 100.00 100.00
99.99
33.8
14
16
78
99.99
63.0
126
78
49
126
21
121
<10
274
12
23
12
99.99 100.00 100.00 100.00 99.99 100.00 100.00
64.9
249
63.7
258
55.2
90
60.1
146
57.6
114
63.3
148
59.4
119
58.6
121
44.5
31
42.2
32
95
26
102
29
66
57
75
15
68
32
83
24
81
31
71
35
33
50
34
60
115
17
108
21
174
168
133
20
123
<10
174
11
141
23
135
<10
181
14
171
23
126
<10
437
12
131
22
122
<10
163
11
133
24
125
<10
153
11
147
149
Y
Zr
27
157
10
318
16
39
247
12
433
22
29
191
10
253
20
30
196
11
279
19
112
<10
131
<10
21
109
<10
165
<10
21
Nb
Ba
La
Ce
Nd
22
10
36
15
51
26
23
15
29
11
27
13
31
12
27
17
39
20
44
19
10
11
the TAS classi®cation of LeBas et al. &1986). Lava ¯ow tent of orthopyroxene phenocrysts is up to 15 vol%.
and pillow samples are ®ne-grained aphyric rocks with Augite commonly shows a pigeonite core or pigeonite
variable amount of holo-cryptocrystalline materials and intergrowths, and usually contains plagioclase laths.
red-brown glass. Textures are hyalo-ophitic, intersertal, Dolerites of andesite composition are orthopyroxene-
sub-ophitic, and intergranular. Lava ¯ow samples are free, coarse-grained rocks, characterized by abundant
basaltic andesites composed of pigeonite, augite, matrix and Fe±Ti oxides.
plagioclase, and minor Fe±Ti oxide microlites dissemi-
The textural relationships among pyroxenes suggest
nated in the matrix. Scarce orthopyroxene microphen- that, in the less evolved basaltic andesites, the orthopy-
ocrysts, commonly mantled by pigeonite, are present. roxene is the liquidus phase, followed by pigeonite, and
Pillow samples have more evolved basaltic andesite then by augite and plagioclase.
composition. Their mineralogical assemblage consists
of acicular crystals of pigeonite, augite, plagioclase, and
minor Fe±Ti oxide grains set in a red-brown quenched Mineral chemistry
glass.
Dolerites have medium-to coarse-grained sub-ophitic
The samples for microanalysis were selected on the basis
texture, ®ne-grained in the chilled margins, and a of texture and bulk-rock composition. Representative
groundmass made up of cryptocrystalline materials and analyses of pyroxenes and plagioclase are listed in
®ne-grained quartz±alkali feldspar intergrowths. Dole- Tables 1, 2, and 3. Pyroxene compositions are plotted in
rites range from basaltic andesites to andesites and are Fig. 2, where the samples of pillow lavas, lava ¯ows, and
composed of pigeonite, commonly zoned augite and chilled margin dolerites are distinct from those of the
plagioclase, and Fe±Ti oxide microlites disseminated in medium-and coarse-grained dolerites.
the interstitial matrix. Dolerite chilled margins and al-
most all the medium-to coarseg-rained dolerites with
Orthopyroxene &opx) has a rather uniform composi-
tion, ranging from Ca4Mg80Fe16 to Ca5Mg77Fe18 and
basaltic andesite composition are characterized by the only occurs in the least evolved basaltic andesites &mg#
presence of orthopyroxene microphenocrysts and phe- 70±55; see later). These opx compositions, very similar
nocrysts, respectively, which are commonly mantled by to those reported by Elliot et al. &1995) for the Ferrar
pigeonite and/or augite and lack plagioclase inclusions. rocks of Mesa Range, are higher in Mg than the early
In the medium-to coarse-grained dolerites with the least crystallized opx from the Dufek Intrusion &Himmelberg
evolved basaltic andesite compositions the modal con- and Ford 1976).

133
Fig. 3 Selected major &wt%) and trace &ppm) elements versus mg#
&at. 100*Mg/[Mg+Fe2+] assuming Fe2O3/FeO=0.15) for the
investigated Prince Albert Mts. rocks. Open triangles coarse-
grained dolerites; ®lled triangles chilled margin dolerites; ®lled
circles ¯ows &mg# ꢀ60) and pillow lavas &mg# ꢀ40±50)
Augite &aug) generally exhibits a wide range of com-
positions from Ca39Mg51Fe10 to Ca33Mg33Fe34. Aug of
the dolerites de®ne a Ca±Mg depletion and Fe-enrich-
ment trend, which ®ts that shown by Brotzu et al. &1988,
1992) for the Ferrar tholeiites from Thern Promontory
Pigeonite &Pgn) has variable composition and is and Archambault Ridge &Fig. 2a). The Mg-rich compo-
characterized by a marked Fe-enrichment trend, from sitions correspond to crystal cores from the least evolved
Ca8Mg75Fe17 to Ca13Mg32Fe55. The Pgn highest in Mg, basaltic andesites, whereas the Fe-rich compositions
which have the same Mg/Fe ratios of the coexisting opx represent crystal rims and microlites of the most evolved
&Fig. 2b), are the cores of discrete grains from the least andesites. By contrast, aug from pillow lavas, lava ¯ows,
evolved lava ¯ow &VG1028) and chilled margin dolerite and chilled margin dolerites &Fig. 2b) have compositions
&VG1048) samples. Pgn mantling opx and needle crystals trending to the sub-calcic augite ®eld &Ca29Mg48Fe23±
in the quenched glass of the pillow lavas usually are Ca23Mg49Fe28) similar to those reported by Elliot et al.
highest in Fe &Ca10Mg60Fe30±Ca12Mg52Fe36).
&1995). These compositions follow the ``quench trend'' of

134
29) and are characterized by a mg# gap between 54 and
37, where the pillow lava samples &mg# 47±42) plot.
Note that both the dolerites and lava ¯ows with mg# of
70±55 correspond to opx-bearing basaltic andesites. In
general, with mg# decreasing, the studied samples show
SiO2, TiO2, FeOt, P2O5 and alkali increasing, and CaO
and Al2O3 decreasing. Incompatible trace elements ex-
hibit negative correlations with mg#, whereas positive
correlations are shown by Cr and Ni. Sr shows a slightly
incompatible behavior in the mg# range 70±55 &opx-
bearing basaltic andesites), whereas it remains virtually
constant for mg# less than 47. Similar Sr behavior was
observed in the Ferrar tholeiites from the Mesa Range
&Elliot et al. 1995).
Signi®cance of orthopyroxene occurrence
in the Ferrar tholeiites
Petrography and mg# variation diagrams suggest that
the evolution from basaltic andesites to the most evolved
andesites was dominated by fractional crystallization of
pyroxenes and plagioclase. The geochemical behavior of
Sr, the mineral crystallization sequences and textural
features indicate that the early stages of the di€erentia-
tion were dominated by the fractionation of pyroxenes
over that of the plagioclase, whose fractionation became
essential in the later di€erentiation stages. As previously
noted, the least evolved tholeiites studied here are
Fig. 4 FeO/MgO&opx) of opx core compositions versus whole-rock
FeO/MgO&w.r) of the investigated chilled margin dolerite &VG1048)
and lava ¯ow &VG1028) from Prince Albert Mts. Whole rock
Fe2O3/FeO computed by MELTS &Ghiorso and Sack 1995) as a
function of composition, liquidus temperature, pressure and fO
2
&after Kress and Carmichael 1991). fO ˆ QFM À 1 log unit bu€er,
2
open symbols: P=1 kbar and dry condition; ®lled symbols:
P=5 kbar and 1.0 H2O wt%. FeO of opx calculated according
to Papike et al. &1974). The equilibrium KD lines after Hunter
&1998) are shown for comparison. Ferrar lava sample 81-2-16
&diamond) from Elliot et al. &1995)
characterized by early-crystallized opx, which are in
equilibrium with their host rocks for a KD
opx/melt
&FeO/
Muir and Mattey &1982), which is consistent with rapid
cooling of the host rocks &Smith and Lindsley 1971).
Both the Ca-poor and the Ca-rich pyroxenes of
Ferrar rocks are characterized by low Al2O3 &and AlVI
atoms per formula unit; Brotzu et al. 1988, 1992; Hornig
1993; Elliot et al. 1995 and present study). The Al2O3
content of pyroxenes in coarse-grained dolerites is sim-
ilar to that in pyroxenes from ®ne-grained chilled margin
dolerite and lava samples. Because the latter display
quench textures and a quench compositional trend, the
Al2O3 of pyroxene likely re¯ects low-pressure crystalli-
zation rather than low-pressure re-equilibration.
Plagioclase &pl) ranges from An87 to An56 in dolerites,
and from An82 to An65 in the lava samples. As previously
observed by Elliot et al. &1995) for the pl from the Mesa
Range tholeiites, the pl compositional range shown by
cores and rims within a single sample approaches that of
the whole sample set &e.g., sample VG1026: An83±An56).
MgO) ranging from 0.25 to 0.29 &Fig. 4). The presence of
early-crystallized opx is a common feature of the least
evolved Ferrar rocks &Gunn 1966; Himmelberg and
Ford 1976; Elliot et al. 1995), but is absent in the simi-
larly evolved tholeiites of the other Gondwana Mesozoic
igneous provinces &Parana and Karoo).
MELTS modeling
It is known that the crystallization of opx as the liquidus
phase, given a magma composition, is favored by high
pressure &P) and low values of oxygen fugacity &f ) and
O2
H2O. This is predicted &see below) also by the MELTS
program &Ghiorso and Sack 1995). We have used the
fractional crystallization &FC) modeling of MELTS for
evaluating which of these factors may have played a
major role on the early crystallization of opx in the
Ferrar magmas. MELTS calculations have been carried
out at low P &1±5 kbar) under dry and H2O-undersatu-
rated conditions &H2O up to 1 wt%). Oxygen fugacity
Geochemistry
Whole rock samples have been analyzed by wavelength was assumed to vary according to the QFM-1 log unit
dispersive X-ray ¯uorescence. Representative analyses curve, as suggested by fO values obtained by Brotzu
are listed in Table 4. All the investigated samples belong et al. &1988) for Ferrar tholeiites.
2
to the low Fe &FeOt<13 wt%) MFCT group of Fleming
The role of P and H2O on the early opx crystalliza-
et al. &1992). The contents of selected major and trace tion is illustrated in Fig. 5 where the least evolved &mg#
elements versus mg# [100*Mg/&Mg+Fe2+)] are shown 71) Ferrar tholeiite &TPS31, Table 5; Wilhelm and
in Fig. 3. The dolerites show a wide range of mg# &70± Worner 1996) from the investigated area &Thumb Point

135
lava &Gunn 1966; Hergt et al. 1989; Elliot et al. 1995;
Wilhelm and Worner 1996; present study) as starting
compositions. FC MELTS results are consistent with the
petrographic features observed in our samples and in-
dicate that Ca-poor pyroxene is the liquidus phase for
pressure more than 1 kbar &Fig. 6). Under dry condi-
tions, opx crystallizes as liquidus phase only from Ferrar
magmas with MgO>8 wt% &TPS31, DG49), whereas
pigeonite crystallizes from liquids with lower MgO
&Fig. 6a). The concomitant increase of P and H2O does
not promote the early crystallization of pigeonite, sug-
gesting that under H2O-undersaturated conditions opx
may also crystallize from Ferrar magmas with MgO
<8 wt% &Figs. 6b, c). The early opx crystallization also
depends on SiO2/CaO of magmas, i.e., for similar MgO,
high SiO2/CaO values favor opx appearance on the
liquidus &Figs. 6b, c). The importance of the SiO2±CaO±
MgO relationships to obtain opx as the liquidus phase is
apparent from Fig. 7, where the MELTS liquid lines of
descent obtained by fractional crystallization at
P=1.5 kbar and dry conditions are compared, starting
from the volatile-free compositions of the least evolved
low-Ti tholeiites from Ferrar &TPS31, DG49), Karoo
&B69, ON35), and Parana &1032) CFB provinces
&Table 5). It is apparent that opx is the liquidus phase
Fig. 5 Magma temperature &°C) versus % crystallized mass &c.m.)
per 10 °C decrease, showing the relative proportions of fraction-
ating minerals calculated by MELTS program &Ghiorso and Sack
1995) for di€erent pressure &P) and H2O content, starting from
TPS31 composition &Thumb Point Sill; Wilhelm and Worner 1996).
fO ˆ QFM À 1 log unit bu€er. aug augite; ol olivine; opx
2
orthopyroxene; pgn pigeonite; pl plagioclase
sill) was used as the starting composition. MELTS
modeling shows that pressure increase favors the early
crystallization of opx, whereas the increase of H2O
promotes that of olivine. It follows that opx crystalli-
zation as the liquidus phase requires higher pressure for
increasing H2O content. Note that increase of both P
and H2O does not favor an early crystallization of
pigeonite and delays that of plagioclase. Similar results
&not shown) have also been obtained for the Ferrar sill-
chilled margin from Darwin Glacier &DG49: mg# 68,
Table 5; Hergt et al. 1989). It should be noted that for
P>1 kbar the crystallization of opx as the liquidus
phase is not very sensitive to pressure at near anhydrous
&H2O=0±0.5 wt%) conditions. The latter data suggest a
primary role of magma composition on early opx crys-
tallization in the least evolved Ferrar tholeiites.
The in¯uence of magma composition has been eval-
uated using the least evolved Ferrar chilled dolerite and

136
Table 5 Recalculated as 100% anhydrous major element compo-
sitions of the least evolved tholeiites of the low-Ti Gondwana
provinces used in the MELTS-FC modeling of Fig. 7. Data source:
Ferrar, TPS31 base of Thumb Point Sill, Wilhelm and Worner
&1996); Ferrar, DG49 dolerite chilled margin of Darwin Glacier Sill,
Â
Hergt et al. &1989); Parana, 1032 lava, Guata-Bom Jardin, Piccirillo
and Mel® &1988); Karoo, B69 lava, Leribe, Basutoland Basalt, Cox
and Hornung &1966); Karoo, ON35 lava, Ma®ka Lisiu, Lesotho
Formation, Marsh et al. &1997). mg# as in Table 4
Sample
CFB province
TPS31
Ferrar
DG49
Ferrar
1032
Parana
B69 ON-35
Karoo Karoo
SiO2
TiO2
Al2O3
FeOt
MnO
MgO
CaO
Na2O
K2O
P2O5
53.43
0.48
14.65
8.41
0.16
10.15
10.55
1.52
0.58
0.07
71.0
5.1
53.00
0.50
16.42
8.42
0.17
8.68
10.64
1.64
0.47
0.06
67.6
5.0
52.05
0.79
15.77
8.26
0.15
8.62
11.79
1.92
0.51
0.12
67.9
4.4
52.46
1.01
14.66
10.02
0.17
8.19
10.50
2.20
0.61
0.17
62.3
5.0
51.11
0.9
15.3
9.95
0.17
8.73
10.99
2.25
0.45
0.15
64.0
4.7
mg#
SiO2/CaO
only if the melts have MgO ³8 wt% and SiO2/CaO
>4.7, and that high SiO2 content, rather than low CaO
of the magmas, is essential for the early opx crystalli-
zation &cf. the curves of sample ON35 in the insets of
Fig. 7). In summary, the occurrence of early-crystallized
opx in the least evolved Ferrar tholeiites is primarily
related to their high SiO2 content.
The origin of high SiO2 in the least evolved
Ferrar tholeiites
Major element modeling of Ferrar tholeiites needs to
take into account the results of petrogenetic models
based on trace element and isotopic compositions. Ac-
cording to these models the high SiO2 of Ferrar tholeiites
could represent &1) a crustal signature caused by AFC
processes of primary magmas generated by a E-MORB
Dupal-type mantle source &Antonini et al. 1999, 2000);
&2) a feature of primary magmas derived from hydrous
melting of depleted mantle source contaminated by
PAAS-crustal material &Hergt et al. 1989; Molzhan et al.
1996). Because the PAAS only has a limited in¯uence on
major element composition of a depleted mantle, hy-
drous melting has been invoked because FC alone can-
not account for the SiO2-rich compositions of Ferrar
tholeiites &Hergt et al. 1991). Although these models
postulate di€erent mantle sources and di€erentiation
processes, both agree in the requirement for high melting
degrees &>15%) for the origin of Ferrar magmas.
Therefore, it is important to evaluate the extent to which
Fig. 6 Whole-rock SiO2/CaO versus MgO wt% of the least evolved
Ferrar chilled margin dolerite and lava samples illustrating, for
increasing pressure &P) and H2O content, the role of these
parameters on the crystallization of opx as liquidus phase by
MELTS &Ghiorso and Sack 1995) assuming fO ˆ QFM À 1 log
2
unit bu€er. Arbitrary dotted lines separate the compositions suitable
for opx crystallization &®lled symbols, star and cross) from the others
&open symbols). Circles opx-bearing rocks from the present study,
numbers refer to the last digits of specimen label VG1000; squares
aphyric and hypersthene-bearing &En82) tholeiites from Gunn
&1966); diamond Tasmania dolerite 84-111 sample from Hergt et
al. &1989); triangles lava samples from Elliot et al. &1995); cross
DG49 sample of Darwin Glacier from Hergt et al. &1989); star
TPS31 sample of Thumb Point Sill from Wilhelm and Worner &1996)
a fertile and depleted peridotite that has undergone high Peridotite melting experiments
degrees of dry and hydrous partial melting may con-
tribute to SiO2-rich primary magmas, and then estimate Numerous experiments have been carried out to deter-
the additional e€ects of fractional crystallization and mine the composition of anhydrous peridotite derived
crustal assimilation on the SiO2-rich signature of the melts &Ito and Kennedy 1967; Kushiro 1968; Presnall
least evolved Ferrar tholeiites.
et al. 1979; Jaques and Green 1980; Stolper 1980;

137
Fig. 7 SiO2/CaO versus MgO &wt%) of liquids evolving by
fractional crystallization calculated by MELTS &Ghiorso and Sack
1995). Liquid lines of descent &steps for 10 °C decrease) computed
with less then 3 wt% H2O in the melts are reported by
Hirose and Kawamoto &1995). These hydrous melts,
compared with the anhydrous melts produced at the
same pressure and from the same peridotite &Hirose and
Kushiro 1993), allow us to evaluate the e€ects of H2O
on the compositions of melts generated from a natural
fertile spinel±lherzolite &KLB-1). The in¯uence of the
source peridotite on partial melt compositions may be
evaluated by comparing the anhydrous experimental
melts obtained from natural fertile &KLB-1) and de-
pleted spinel lherzolites &Tinaquillo peridotite; Jaques
and Green 1980).
at dry conditions, P=1.5 kbar and fO ˆ QFM À 1 log unit bu€er.
2
Starting compositions &see Table 5) are the least evolved tholeiites
of Ferrar &circles TPS31, Wilhelm and Worner 1996; DG49, Hergt
et al. 1989), Parana &squares 1032, Piccirillo and Mel® 1988) and
Karoo &triangles B69, Basutoland Basalt, Cox and Hornung 1966;
ON35, Lesotho Formation, Marsh et al. 1997). Filled symbols
MELTS liquid compositions in equilibrium with orthopyroxene.
Numbers in brackets are the starting liquidus temperatures. The
insets show that, for similar MgO &wt%), high SiO2 &wt%) rather
than low CaO &wt%) promotes early orthopyroxene crystallization.
Opx liquidus ®elds &shaded areas) are arbitrarily drawn
In Fig. 8, the least evolved Ferrar tholeiites &MgO
>6 wt%) are compared with the experimental melts
produced at 10 and 15 kbar from KLB-1 and Tina-
quillo spinel lherzolites believed to be representative of
natural fertile and depleted peridotites, respectively.
Note that these pressures are in agreement with the
pressure range estimated for low-Ti Parana basalts &at
the lower end of a 10±22-kbar pressure range for
Gramado unit; Garland et al. 1996) and for low Ti±Zr
picrites of Karoo &13±15 kbar; Sweeney et al. 1991).
Following the results given by previous models on
Ferrar trace element composition &Hergt et al. 1989;
Molzhan et al. 1996; Antonini et al. 1999, 2000) only
melts generated by melting degrees higher than ꢀ15%
have been considered. In Fig. 8a the hydrous
&P=10 kbar; Hirose and Kawamoto 1995) and anhy-
drous &P=10±15 kbar; Hirose and Kushiro 1993) ex-
perimental melts produced from KLB-1 spinel
lherzolite are shown while those produced by anhy-
drous partial melting of Tinaquillo spinel lherzolite
&P=10±15 kbar; Jaques and Green 1980) are shown in
Fig. 8b. The hydrous melts &Fig. 8a) are broadly similar
to the anhydrous ones formed at the same pressure and
similar degree of melting, as reported by Hirose and
Falloon and Green 1987; Kinzler and Grove 1992a,
1992b; Hirose and Kushiro 1993; Baker and Stolper
1994; Walter and Presnall 1994; Baker et al. 1995;
Kinzler 1997) whereas experimental data on hydrous
peridotite melting remain scarce. The ®rst experimental
studies on simpli®ed systems &Kushiro et al. 1968;
Kushiro 1969, 1972) and subsequently those on the
origin of andesite &Kushiro et al. 1972; Nichols and
Ringwood 1972, 1973; Green 1973, 1976; Nichols 1974;
Mysen and Boettcher 1975a, 1975b ; Nehru and Wyllie
1975) suggest that partial melting of hydrous peridotite
produce SiO2-rich melts. Recently, the experimental
study of Gaetani and Grove &1998) has, however,
documented that the higher SiO2 content of the hy-
drous melts, compared with the anhydrous melts, is
only apparent when the comparison is made on a vol-
atile-free basis. Hydrous magmas with 3±6 wt% of H2O
have SiO2ꢁ48±49 wt% &on anhydrous basis), which is
only ꢁ1 wt% higher than that of anhydrous melts
&Gaetani and Grove 1998). Note that these hydrous
melts have SiO2 that is considerably lower &ꢀ3±4 wt%)
than those of the least evolved Ferrar rocks. Experi-
mental melting studies on natural peridotite for systems

138
Fig. 8 SiO2 versus MgO &wt%) of the least evolved Ferrar
tholeiites &®lled circles present study; ®eld literature data, source as
in Fig. 6; stars base and uppermost samples of Thumb Point Sill;
Wilhelm and Worner 1996; cross DG49; Hergt et al. 1989)
compared with experimental primary melts produced by melting
degree &f)³15%. a: Hydrous &®lled squares, P=10 kbar; from
Hirose and Kawamoto 1995) and anhydrous melts &open squares,
P=10 kbar; gray squares, P=15 kbar; from Hirose and Kushiro
1993) produced from the same fertile spinel lherzolite KLB-1.
Numbers refer to literature run # &r) experiments. b Anhydrous
melts &open triangles, P=10 kbar; gray triangles, P=15 kbar)
produced from depleted spinel lherzolite &Tinaquillo peridotite;
Jaques and Green 1980; number in brackets refer to % melting).
Anhydrous melts from a are shown for comparison &open and gray
squares). The calculated primary Tasmania tholeiite &triangle TAS;
Hergt et al. 1989), the calculated parental Ferrar picrite &inverted
triangle OG; Ortez and Green, unpublished work, in Sweeney et
al. 1991) and the Ferrar olivine±tholeiite &diamond sample 26903,
lower contact of Painted Cli€ sill; Gunn 1966) are also reported.
The dotted arrows show the e€ect of olivine &mg#=89 from picrite
basalt KP111, Lebombo; Bristow 1984) and orthopyroxene
&sample VG1048, Table 1) addition &bar=10% increment) to
TPS31 composition. The ®eld between the arrows constrains the
experimental melts which may have generated the TPS31
composition through orthopyroxene and olivine fractional crys-
tallization
Fractional crystallization
Major element mass-balance calculations &Stormer and
Nicholls 1978) and fractional crystallization±MELTS
&Ghiorso and Sack 1995) modeling were made to test
whether the least evolved Ferrar tholeiite from Victoria
Land &TPS31; Wilhelm and Worner 1996) could be de-
rived by fractional crystallization &FC) from the exper-
imental primary melts shown in Fig. 8. The following
FC modeling results refer to the experimental melts that
plot between orthopyroxene and olivine fractional
crystallization trends, which constrain the melts suitable
to generate the TPS31 composition.
Mass-balance calculations yielded high values for the
sum of the squares of residuals &Sres2=2.3±4.1) starting
from depleted Tinaquillo peridotite-derived melts, and
lower values &Sres2=0.8±1.4) starting from fertile KLB-
1 peridotite-derived melts. The best results have been
obtained by the KLB-1-derived hydrous melts that have
the highest SiO2, i.e., run #44 &r44; H2O=0.5 wt%) and
run #42 &r42; H2O=1.7 wt%). TPS31 composition was
matched through 26 and 19% of olivine &ol) + ortho-
pyroxene &opx) fractionation, in the proportion ol=8%
Kawamoto &1995). As expected &Hirose and Kushiro opx=18%, and ol=11% opx=8%, starting from r44
1993; Gaetani and Grove 1998; Hirschmann et al. &Sres2=1.0) and r42 &Sres2=0.8), respectively. FC-
1998), SiO2 of the anhydrous melts decreases with in- MELTS liquid lines of descent &lld) are reported in Fig. 9.
creasing melting pressure and refractory nature of the The lld that more closely approach the TPS31 compo-
peridotite source, as well as with the decrease of melt- sition have been obtained starting from the KLB-1-de-
ing degree &Fig. 8b). It is apparent that the least rived hydrous magmas r44 &2 kbar, Sres2=2.6) and r42
evolved Ferrar tholeiites do not resemble any of the &4 kbar, Sres2=1.0) through 24 and 19% of ol + opx
reported picritic experimental melts in terms of MgO fractionation, in the proportions of ol=9% opx=15%,
and SiO2. Even if direct evidence of picritic magmas in and ol=11% opx=8%, respectively. Higher values of
the Ferrar magmatism is lacking, picritic melts occur in Sres2 have been obtained for lld relative to both KLB-1-
the low-Ti Karoo magmatism that Elliot and Fleming and Tinaquillo-derived anhydrous magmas &5.6 and 6.8±
&2000) suggest is generated from the same mantle 7.6, respectively, P=1±1.5 kbar). Note that, in spite of
source in the Weddell Sea region. Moreover, Ortez and the lower Sres2 value, the r42 composition is not favored
Green &unpublished work, in Sweeney et al. 1991) as a primary melt for TPS31 because the di€erentiated
proposed a picrite composition as parental Ferrar liquid matching the TPS31 composition would have
magma, which is very close to run #43 of the hydrous ꢀ2 wt% of H2O. This relatively high H2O delays the
experimental melt &Hirose and Kawamoto 1995) re- crystallization of plagioclase in the subsequent stages of
ported in Fig. 8. This supports a picritic nature for FC. It follows that the di€erentiated tholeiites &e.g.,
primary Ferrar tholeiites.
MgO =6 wt%) at Thumb Point Sill would have Al2O3

139
Fig. 9 FC-MELTS &Ghiorso
and Sack 1995) liquid lines of
descent versus mg# starting
from experimental primary
melts produced by a fertile
KLB1 &Hirose and Kushiro
1993; Hirose and Kawamoto
1995) and b depleted Tinaquillo
spinel lherzolites &Jaques and
Green 1980). Step of tempera-
ture decrease and fO as in
Fig. 7. Symbols and labels as in
Fig. 8
2

140

141

142
Fig. 10 AFC-MELTS &Ghiorso
and Sack 1995) liquid lines of
descent versus mg# computed
using the meta-igneous granu-
lite 21BD6 &Talarico et al. 1995)
as contaminant. Starting mag-
mas are experimental melts
produced by fertile KLB1 &r43,
r44; P=10 kbar; Hirose and
Kawamoto 1995) and depleted
Tinaquillo spinel lherzolites &at
P=15 kbar for 34% melting
degree, TQ; Jaques and Green
1980). Step of temperature de-
crease and fO as in Fig. 7. The
2
assimilation is con®ned to the
achievement of TPS31 compo-
sition &continuous lines AFC
process; dashed lines FC pro-
cess). The total assimilated
mass is: 24 g for r43 &3 g per
10 °C decrease; mean of assim-
ilated mass/fractionated
mass=1.3); 12 g for r44 &1 g per
10 °C decrease; mean of assim-
ilated mass/fractionated
mass=0.5); 22 g for TQ &1 g per
10 °C decrease; mean of assim-
ilated mass/fractionated
mass=0.6). The uppermost di-
agram on the right shows the
comparison between the
amounts of fractionated miner-
als &opx ‹ ol) and assimilated
mass &21BD6) required by mass
balance calculations &mass%
XLFRAC) and those required
by MELTS &mass% MELTS).
Sres2 refer to the values ob-
tained by AFC±MELTS mod-
eling, while those relative to
XLFRAC are listed in Tables 6
and 7. Other symbols as in Fig. 9
as high as 18 wt% instead of ꢀ13 wt% &cf. Thumb Point component to a depleted mantle source that has un-
Sill tholeiites, Wilhelm and Worner 1996). Therefore, dergone hydrous melting would account for the origin of
r42 melt will not be considered in further modeling.
the major element signature of Ferrar tholeiites. Un-
Figure 9a shows that the KLB-1-derived melts fail to fortunately, melting experiments on a mantle source
approach K2O &and partly CaO) values of TPS31. On such as that predicted by Hergt et al. &1989) are lacking.
the other hand, Fig. 9b reveals that Tinaquillo-derived Nevertheless, the following considerations can be made:
melts do not match TPS31 composition not only for &1) the major contributions to the Sres2 values relative to
K2O but also for CaO, TiO2, and Al2O3. This suggests Tinaquillo-derived melts are due to CaO and Al2O3
that additional K-rich components have to be consid- &MELTS res2=1.7±3.3 and 1.4±2.6, respectively). These
ered in the genesis of the Ferrar tholeiites. According to results would not be improved by the addition of a
Hergt et al. &1989), the addition of 3% PAAS crustal PAAS component to a depleted source, because such a

143
component has a limited in¯uence on the major element calculated amounts of assimilated mass, as well as as-
composition of the mantle source &Hergt et al. 1991); &2) similated/fractionated mass are, in general, negatively
if the addition of a small amount of H2O to the dry correlated with SiO2 from the contaminants.
Tinaquillo peridotite had e€ects comparable to those
The crustal contamination process has been further
reported for the experimental melting of KLB-1-fertile tested by AFC±MELTS modeling, starting from melts
peridotite &0.1±0.5 wt%; Hirose and Kawamoto 1995), that yielded the most mass-balance solutions &r44, r43,
the hydrous melts would be expected to be close in and TQ15-f34) and using, as contaminant, the meta-
composition to the anhydrous melts for the same degree igneous granulite 21BD6 &SiO2=65.4 wt%), already
of melting, and, therefore, the results of the above FC employed by Antonini et al. &1999) for the AFC
modeling would not change signi®cantly; &3) at constant modeling of the Ferrar trace element patterns. MELTS
pressure, the increase of H2O would produce, at con- modeling indicates that the TPS31 composition can be
stant MgO, melts with higher CaO and Al2O3, and lower matched &Sres2<1) at pressures of 5 and 2 kbar,
FeO &Gaetani and Grove 1998). The above FC-model- starting from KLB-1 and Tinaquillo-derived melts, re-
ing results might improve for Al2O3, but the contrary spectively, through amounts of fractionated minerals
would occur for CaO. Again, the results of the above FC and assimilated mass similar to those obtained by mass
modeling would not change signi®cantly.
balance calculations &Fig. 10). The high value of as-
Therefore, we would anticipate that, even assuming similated/fractionated mass &mean 1.3) required for the
hydrous partial melting of a depleted mantle source r43 melt is consistent with the thermal limits and extent
contaminated by PAAS, mass-balance and FC±MELTS of crustal contamination modeled by Reiners et al.
results would not be substantially modi®ed.
&1995).
In summary, the AFC model explains quite well the
major element composition of the least evolved Ferrar
tholeiite starting from experimental melts derived by
high melting degrees of both fertile and depleted peri-
Crustal assimilation
The possibility that the major elements of the least dotite at pressures of 10±15 kbar. The fact that the AFC
evolved Ferrar tholeiites could be the result of crustal model is compatible with many types of contaminants
assimilation during fractional crystallization of primary makes it a plausible process because it is unlikely that
basalts was tested by mass-balance calculations. We only a speci®c contaminant may be responsible for the
modeled the transitions from KLB-1- and Tinaquillo- crustal signature shown by Ferrar magmas whose
derived melts to the TPS31 composition, assuming the emplacement occurred over thousand of kilometers.
granulites of Victoria Land basement &Talarico et al.
1995) to be the contaminants, as proposed by Antonini
et al. &1999) to explain the trace element content of the
Ferrar rocks.
Concluding remarks
Mass-balance results yielding Sres2 values <1 and
assimilated amounts <40% are listed in Tables 6 and 7. 1. The least evolved investigated Ferrar Dolerite and
As regards the KLB-1-derived melts, the transitions
from hydrous magmas r43 and r44 to TPS31 are com-
patible with the assimilation of almost all the di€erent
types of granulites found in Victoria Land &SiO2=52±
74 wt%; Talarico et al. 1995). In comparison, only one
transition is possible from anhydrous melts &i.e., r17) to
TPS31 through the assimilation of the more ma®c
granulite &SiO2=52.6 wt%). Also the transitions from
Tinaquillo-derived melts appear to be compatible with
Kirkpatrick Basalt from the southern Prince Albert
Mountains are characterized by the occurrence of
early-crystallized orthopyroxene. This feature petro-
graphically distinguishes these tholeiites from those of
Parana and Karoo. Textural relationships indicate
that opx was the liquidus phase, followed by pigeonite,
and then by augite and plagioclase in the crystalliza-
tion sequence. The mineral chemistry indicates that
opx is in compositional equilibrium with the host rock.
the assimilation of many granulites, particularly for the 2. MELTS modeling tested on the least evolved Ferrar
melt generated by 34% partial melting at 15 kbar
&TQ15-f34). Note that the transitions from Tinaquillo-
derived melts require olivine fractionation in addition to
orthopyroxene, whereas those from KLB-1-derived
hydrous melts are also possible with orthopyroxene
fractionation only.
The results of Tables 6 and 7 show that relatively low
amounts of contaminant &mean=14‹5%; mean of as-
similated/fractionated mass: ass/fr=0.5‹0.2) are in-
volved in the transition from r44, whereas higher
amounts are required for the transition from r43 &25‹8;
tholeiites gives opx as the liquidus phase. Compari-
son with the analogs from Parana and Karoo sug-
gests that the primary control on opx crystallization
is played by the Ferrar magma composition, and in
particular by its high SiO2.
3. Mass-balance calculations and MELTS modeling
indicate that the SiO2-rich signature associated with
early opx crystallization, and, in general, the major
element geochemistry of the least evolved Ferrar
tholeiite, can be reproduced by AFC process. Ex-
perimental picritic melts derived at 10±15 kbar by
high melting degrees &>25%) of natural fertile
&KLB-1) and depleted &Tinaquillo) peridotites, under
mean
&mean=22‹7; mean ass/fr=0.6‹0.2). In any case, the
ass/fr=1.3‹0.4)
and
from
TQ15-f34

144
Cox KG &1988) The Karoo Province. In: Macdougall JD &ed)
Continental ¯ood basalts. Kluwer, Boston, pp 239±272
Cox KG, Hornung G &1966) The petrology of the Karroo basalts
of Basutoland. Am Mineral 51:1414±1432
Duncan RA, Hooper PR, Rehacek J, Marsh JS, Duncan AR
&1997) The timing and duration of the Karoo igneous event,
southern Gondwana. J Geophys Res 102:18127±18138
Elliot DH, Fleming TH &2000) Weddell triple junction: the prin-
cipal focus of Ferrar and Karoo magmatism during initial
breakup of Gondwana. Geology 28:539±542
hydrous and dry conditions, respectively, have the
appropriate compositions to be the Ferrar primary
magmas, assuming the Victoria Land granulites to be
the crustal contaminants. Therefore, the Ferrar tho-
leiites do not necessarily re¯ect the generally assumed
depleted source mantle as they are also compatible
with a fertile mantle.
Acknowledgements We are grateful to A. Cundari, R. Petrini and
R. Varne for their useful comments. The authors greatly appreci-
ated the suggestions and the constructive criticism of G. Moore and
another anonymous reviewer. Thanks are due to L. Furlan &Trie-
ste) and R. Carampin &Padova) for their technical and analytical
collaboration. This work has been carried out with the ®nancial
support of the Italian ``Programma Nazionale di Ricerche in
Antartide'' and MURST.
Elliot DH, Larsen D &1993) Mesozoic volcanism in the Transant-
arctic Mountains depositional environment and tectonic set-
ting, In: Findlay RH, Unrug R, Banks HR, Veevers JJ &eds)
Gondwana eight: assembly, evolution and dispersal. AA Balk-
ema, Rotterdam, pp 397±410
Elliot DH, Siders MA, Haban MA &1986) Jurassic tholeiites in the
region of the upper Rennick Glacier, north Victoria Land. In:
Stump E &ed) Geological investigations in northern Victoria
Land, Antarctica. Am Geophys Union, Washington, DC,
Antarct Res Ser 46:249±265
Elliot DH, Fleming TH, Haban MA, Siders MA &1995) Petrology
and mineralogy of the Kirkpatrik Basalt and Ferrar Dolerite,
Mesa Range Region, North Victoria Land, Antarctica. In: El-
liot DH, Blaisdell GL &eds). Contributions to Antarctic research
IV. Am Geophys Union, Washington, DC, Antarct Res Ser
67:103±141
Elliot DH, Fleming TH, Kyle PR, Foland KA &1999) Long-dis-
tance transport of magmas in the Jurassic Ferrar Large Igneous
Province, Antarctica. Earth Planet Sci Lett 167:89±104
Encarnacion J, Fleming TH, Elliot DH, Eales HV &1996) Syn-
chronous emplacement of Ferrar and Karoo dolerites and the
early breakup of Gondwana. Geology 24:535±538
Falloon TJ, Green DH &1987) Anhydrous partial melting of
MORB pyrolite and other peridotite compositions at 10 kbar:
implications for the origin of primitive MORB glasses. Mineral
Petrol 37:181±219
Faure G, Bowman JR, Elliot DH, Jones LM &1974) Strontium
isotope composition and petrogenesis of the Kirkpatrick Basalt,
Queen Alexandra Range, Antarctica. Contrib Mineral Petrol
48:153±169
References
Antonini P, Piccirillo EM, Petrini R, Civetta L, D'Antonio M, Orsi
G &1999) Enriched mantle ± Dupal signature in the genesis of
the Jurassic Ferrar tholeiites from Prince Albert Mountains
&Victoria Land, Antarctica). Contrib Mineral Petrol 136:1±19
Antonini P, Piccirillo EM, Petrini R, Civetta L, D'Antonio M, Orsi
G &2000) Reply to the comment by J. Hergt on the paper
``Enriched mantle ± Dupal signature in the genesis of the Jur-
assic Ferrar tholeiites from Prince Albert Mountains &Victoria
Land, Antarctica)'' by Antonini et al. &Contributions to Min-
eralogy and Petrology 136:1±19, 1999). Contrib Mineral Petrol
139:245±249
Armienti P, Ghezzo C, Innocenti F, Manetti P, Rocchi S, Tonarini
S &1990) Isotope geochemistry and petrology of granitoid suites
from Granite Harbour Intrusive of Wilson Terrane, North
Victoria Land, Antarctica. Eur J Mineral 2:103±123
Baker MB, Stolper EM &1994) Determining the composition of
high-pressure mantle melts using diamond aggregates. Geochim
Cosmochim Acta 58:2811±2827
Baker MB, Hirschmann MM, Ghiorso MS, Stolper EM &1995)
Compositions of near-solidus peridotite melts from experiments
and thermodynamic calculations. Nature 375:308±311
Borg SG, Stump E, Chappell BW, McCulloch MT, Wyborn D,
Armstrong RL, Holloway JR &1987) Granitoids of northern
Victoria Land: implications of chemical and isotopic variations
to regional structure and tectonics. Am J Sci 287:127±169
Bradshaw JD &1987) Terrane boundaries in North Victoria Land.
Mem Soc Geol It 33:9±15
Brewer TS, Hergt JM, Hawkesworth CJ, Rex D, Storey BC &1992)
Coats Land dolerites and the generation of Antarctic conti-
nental ¯ood basalts. In: Storey BC, Alabaster T, Pankhurst RJ
&eds) Magmatism and the causes of continental break-up. Geol
Soc Lond Spec Publ 68:185±208
BristowJW&1984)PicriticrocksinthenorthLebomboandsouth-east
Zimbabwe. In: Erlank AJ &ed) Petrogenesis of the volcanic rocks
of the Karoo Province. Geol Soc S Afr Spec Publ 13:105±123
Brotzu P, Capaldi G, Civetta L, Melluso L, Orsi G &1988) Jurassic
Ferrar Dolerites and Kirkpatrick Basalts in northern Victoria
Land &Antarctica): stratigraphy, geochronology and petrology.
Mem Soc Geol It 43:97±116
Brotzu P, Capaldi G, Civetta L, Orsi G, Gallo G, Melluso L &1992)
Geochronology and geochemistry of Ferrar rocks from North
Victoria Land, Antarctica. Eur J Mineral 4:605±617
Faure G, Pace KK, Elliot DH &1982) Systematic variations of
87Sr/86Sr ratios and major element concentrations in the Kirk-
patrick Basalt of Mount Falla, Queen Alexandra Range, Tran-
santarctic Mountains. In: Craddock
C
&ed) Antarctic
geosciences, University of Wisconsin Press, Madison, pp 715±723
Fleming TH, Elliot DH, Jones LM, Bowman JR &1992) Chemical
and isotopic variations in an iron-rich lava ¯ow from the
Kirkpatrick Basalt, north Victoria Land. Contrib Mineral
Petrol 111:440±457
Fleming TH, Foland KA, Elliot DH &1995) Isotopic and chemical
constraints on the crustal evolution and source signature of
Ferrar magmas, north Victoria Land, Antarctica. Contrib
Mineral Petrol 121:217±236
Fleming TH, Heimann A, Foland KA, Elliot DH &1997) 40Ar/39Ar
geochronology of Ferrar Dolerite sills from the Transantarctic
Mountains, Antarctica: implications for the age and origin of
the Ferrar magmatic province, Geol Soc Am Bull 109:533±546
Gaetani GA, Grove TL &1998) The in¯uence of water on melting of
mantle peridotite. Contrib Mineral Petrol 131:323±346
Garland F, Turner S, Hawkesworth C &1996) Shifts in the source of
the Parana basalts through time. Lithos 37:223±243
Ghiorso MS, Sack RO &1995) Chemical mass transfer in magmatic
processes. IV. A revised and internally consistent thermody-
namic model for the interpolation and extrapolation of liquid±
solid equilibria in magmatic systems at elevated temperatures
and pressures. Contrib Mineral Petrol 119:197±212
Green DH &1973) Experimental melting studies on a model upper
mantle composition at high pressure under water-saturated and
water-undersaturated conditions. Earth Planet Sci Lett 19:37±53
Green DH &1976) Experimental testing of ``equilibrium'' partial
melting of peridotite under water-saturated, high-pressure
conditions. Can Mineral 14:255±268
Burrett CF, Findlay RH &1984) Cambrian and Ordovician con-
odonts from the Robertson Bay Group, Antarctica, and their
tectonic signi®cance. Nature 307:723±725
Casnedi R, Pertusati PC &1991) Geology of the Priestly Formation
&northern Victoria Land, Antarctica) and its signi®cance in the
context of early Paleozoic continental reconstruction. Mem Soc
Geol It 46:145±156

145
Gunn BM &1966) Modal and element variation in Antarctic tho- Macdougall JD &ed) &1988) Continental ¯ood basalts. Kluwer,
leiites. Geochim Cosmochim Acta 30:881±920
Heimann A, Fleming TH, Elliot DH, Foland KA &1994) A short Mahoney JJ, Con MF &eds) &1997) Large igneous provinces. Am
Dordrecht
interval of Jurassic continental ¯ood basalt volcanism in Ant-
arctica as demonstrated by 40Ar/39Ar geochronology. Earth
Planet Sci Lett 121:19±41
Geophys Union, Geophysical Monograph 100, Washington,
DC
Marsh JS, Hooper PR, Rehacek J, Duncan RA, Duncan AR
&1997) Stratigraphy and age of Karoo basalts of Lesotho and
implications for correlations within the Karoo igneous prov-
ince. In Mahoney JJ, Con MF &eds) Large igneous provinces.
Am Geophys Union, Geophysical Monograph 100 Washing-
ton, DC, pp 247±272
Hergt JM &2000) Comment on the paper ``Enriched mantle ± Dupal
signature in the genesis of the Jurassic Ferrar tholeiites from
Prince Albert Mountains &Victoria Land, Antarctica)'' by An-
tonini et al. &Contributions to Mineralogy and Petrology 136:1±
19, 1999). Contrib Mineral Petrol 139:240±244
Hergt JM, Chappel BW, McCulloch MT, McDougall I, Chivas AR Menzies MA, Kyle PR &1990) Continental volcanism: a crust±
&1989) Geochemical and isotopic constraints on the origin of mantle probe. In: Menzies MA &ed) Continental mantle.
the Jurassic dolerites of Tasmania. J Petrol 30:841±883 Clarendon Press, Oxford, pp 157±177
Hergt JM, Peate DW, Hawkesworth CJ &1991) The petrogenesis of Molzahn M, Reisberg L, Worner G &1996) Os, Sr, Nd, Pb, O
Mesozoic Gondwana low-Ti ¯ood basalts. Earth Planet Sci Lett
105:134±148
Himmelberg GR, Ford AB &1976) Pyroxenes of Dufek Intrusion,
Antarctica. J Petrol 17:219±243
Hirose K, Kawamoto T &1995) Hydrous partial melting of lherz-
olite at 1 GPa: the e€ect of H2O on the genesis of basaltic
magmas. Earth Planet Sci Lett 133:463±473
isotope and trace element data from the Ferrar ¯ood basalts,
Antarctica: evidence for an enriched subcontinental lithospheric
source. Earth Planet Sci Lett 144:529±546
Mortimer N, Parkinson D, Raine JI, Adams CJ, Graham IJ, Oliver
PJ, Palmer K &1995) Ferrar magmatic province rocks discov-
ered in New Zealand: implications for Mesozoic Gondwana
geology. Geology 23:185±188
Hirose K, Kushiro I &1993) Partial melting of dry peridotites at Mysen BO, Boettcher AL &1975a) Melting of hydrous upper
high pressures: determination of compositions of melts segre-
gated from peridotite using aggregates of diamond. Earth
Planet Sci Lett 114:477±489
mantle: phase equilibria of a natural peridotite at high pressures
and high temperatures as a function of controlled activities of
water, hydrogen and carbon dioxide. J Petrol 16:520±548
Hirschmann MM, Ghiorso MS, Wasylenki LE, Asimow PD, Mysen BO, Boettcher AL &1975b) Melting of a hydrous upper
Stolper EM &1998) Calculation of peridotite partial melting
from thermodynamic model of mineral and melts. I. Review of
methods and comparison with experiments. J Petrol 39:1091±
1115
mantle. II. Geochemistry of crystals and liquid formed by
anatexis of mantle peridotite at high pressures and high tem-
peratures as a function of water, hydrogen and carbon dioxide.
J Petrol 16:549±593
Hornig I &1993) High-Ti and low-Ti tholeiites in the Jurassic Ferrar Muir ID, Mattey DA &1982) Pyroxene fractionation in ferrobasalts
Group, Antarctica. Geol Jahrb Reihe E47:335±369 from the Galapagos spreading centre. Mineral Mag 45:193±200
Hunter AG &1998) Intracrustal controls on the coexistence of Nehru CE, Wyllie PJ &1975) Compositions of glasses from St. Paul's
tholeiitic and calc-alkaline magma series at Aso Volcano, SW
Japan. J Petrol 39:1255±1284
Ito K, Kennedy GC &1967) Melting and phase relations in a natural
peridotite up to 40 kilobars. Am J Sci 265:519±538
peridotite partially melted at 20 kilobars. J Geol 83:455±471
Nichols IA &1974) Liquids in equilibrium with peridotite mineral
assemblages at high water pressures. Contrib Mineral Petrol
45:289±316
Jaques AL, Green DH &1980) Anhydrous melting of peridotite at Nichols IA, Ringwood AE &1972) Production of silica-saturated
0±15 kbar pressure and the genesis of tholeiitic basalts. Contrib tholeiitic magmas in island arcs. Earth Planet Sci Lett 17:243±
Mineral Petrol 73:287±310 246
Kinzler RJ &1997) Melting of mantle peridotite at pressures ap- Nichols IA, Ringwood AE &1973) E€ect of water on olivine sta-
proaching the spinel to garnet transitions: application to mid- bility in tholeiites and the production of silica-saturated mag-
ocean ridge basalt petrogenesis. J Geophys Res 102:853±874 mas in the island-arc environment. J Geol 81:285±300
Kinzler RJ, Grove TL &1992a) Primary magmas of mid-ocean ridge Papike JJ, Cameron K, Baldwin K &1974) Amphiboles and py-
basalts. 1. Experiments and methods. J Geophys Res 97:6885±
6906
Kinzler RJ, Grove TL &1992b) Primary magmas of mid-ocean ridge
basalts. 2. Applications. J Geophys Res 97:6907±6926
Kress VC, Carmichael ISE &1991) The compressibility of silicate
roxenes: characterization of other than quadrilateral compo-
nents and estimate of ferric iron from microprobe data. Bull
Geol Soc Am 6:1053±1054
Philips &1994) X40 Software for XRF analysis, Software Operation
Manual
liquids containing Fe2O3 and the e€ect of composition, tem- Piccirillo EM, Mel® AJ &eds) &1988) The Mesozoic ¯ood volcanism
perature, oxygen fugacity and pressure on their redox states.
Contrib Mineral Petrol 108:82±92
of the Parana Basin &Brazil). Instituto Astronomico e Geofõsico
Publishers, Sao Paulo, Brazil
Ä
Kushiro I &1968) Compositions of magmas formed by partial Poldervaart A, Hess HH &1951) Pyroxenes in the crystallization of
melting of the Earth's upper mantle. J Geophys Res 73:619±634 basaltic magmas. J Geol 59:472±489
Kushiro I &1969) The system forsterite±diopside±silica with and Presnall DC, Dixon TH, O'Donell TH, Dixon SA &1979) Genera-
without water at high pressures. Am J Sci 267A:269±294 tion of mid-ocean ridge tholeiites. J Petrol 20:3±35
Kushiro I &1972) E€ect of water on the composition of magmas Reiners PW, Nelson BK, Ghiorso MS &1995) Assimilation of felsic
formed at high pressures. J Petrol 13:311±334
Kushiro I, Yoder HS Jr, Nishikawa M &1968) E€ect of water on the
melting of enstatite. Geol Soc Am Bull 79:1685±1692
Kushiro I, Shimizu N, Nakamura Y, Akimoto S &1972) Compo-
sitions of coexisting liquid and solid phases formed upon
melting natural garnet and spinel lherzolites at high pressures: a
preliminary report. Earth Planet Sci Lett 14:19±25
Le Bas MJ, Le Maitre RW, Streckeisen A, Zanettin B &1986)
Chemical classi®cation of volcanic rocks based on the total
alkali±silica diagram. J Petrol 27:745±750
Lombardo B, Pertusati PC, Ricci CA &1991) Some geological and
petrological problems in the Wilson Terrane of Borchgrevink
crust by basaltic magma: thermal limits and extents of crustal
contamination of mantle-derived magmas. Geology 23:563±566
Renne PR, Ernesto M, Pacca IG, Coe RS, Glen JM, Prevot M,
Perrin M &1992) The age of Parana ¯ood volcanism, rifting
Gondwanaland, and the Jurassic±Cretaceous boundary. Sci-
ence 258:975±979
Roland NW, Worner G &1996) Kirkpatrick Flows and associated
pyroclastics: new occurrences, de®nition and aspects of a Jur-
assic Transantarctic Rift. Geol Jahrb B89:97±121
Smith D, Lindsley DH &1971) Stable and metastable augite crys-
tallization trends in a single basalt ¯ow. Am Mineral 56:225±
233
Coast &Northern Victoria Land, Antarctica). Mem Soc Geol It Stewart K, Turner S, Kelley S, Hawkesworth C, Kirstein L,
46:135±143
Mantovani M &1996) 3-D 40Ar/39Ar geochronology in the Pa-

146
rana continental ¯ood basalt province. Earth Planet Sci Lett Turner S, Regelous M, Kelley S, Hawkesworth CJ, Mantovani S
143:95±109
&1994) Magmatism and continental break-up in the South At-
lantic: high precision 40Ar/39Ar geochronology. Earth Planet
Sci Lett 121:333±348
Stolper EM &1980) A phase diagram for mid-ocean ridge basalts:
preliminary results and implications for petrogenesis. Contrib
Mineral Petrol 74:13±27
Storey BC, Alabaster T, Pankhurst RJ &eds) &1992) Magmatism
and the causes of continental break-up. Geol Soc Lond Spec
Publ 68
Stormer JC, Nicholls J &1978) XLFRAC: a program for the in-
teractive testing of magmatic di€erentiation model. Computer
Geosci 4:143±159
Sweeney RJ, Falloon TJ, Geen DH, Tatsumi Y &1991) The mantle
origins of Karoo picrites. Earth Planet Sci Lett 107:256±271
Talarico F, Borsi L, Lombardo B &1995) Relict granulites in the
Ross Orogen of Northern Victoria Land &Antarctica). II ±
geochemistry and paleo-tectonic implications. Precambrian Res
75:157±174
Walter MJ, Presnall DC &1994) Melting behavior of simpli®ed
lherzolite in the system CaO±MgO±Al2O3±SiO2±Na2O from 7
to 35 kbar. J Petrol 35:329±359
Weaver SD, Bradshaw JD, Laird MG &1984) Geochemistry of
Cambrian volcanics of the Bowers Supergroup and implica-
tions for the Early Paleozoic tectonic evolution of the north-
ern Victoria land, Antarctica. Earth Planet Sci Lett 69:129±
140
Wilhelm S, Worner G &1996) Crystal size distribution in Jurassic
Ferrar ¯ows and sills &Victoria Land, Antarctica): evidence for
processes of cooling, nucleation, and crystallization. Contrib
Mineral Petrol 125:1±15
Worner G &1992) Kirkpatrick lavas, exposure hill formation and
Ferrar sills in the Prince Albert Mountains, Victoria land,
Antarctica. Polarforschung 60&2):87±90
Taylor SR, McLennan SM &1985) The continental crust: its com-
position and evolution. Hallam A &ed) Blackwell Scienti®c
Publications, Oxford
Turner S, Hawkesworth C &1995) The nature of the sub-continental
mantle: constraints from the major-element composition of
continental ¯ood basalts. Chem Geol 120:295±314