Full text of DOI:10.1007/s004100000215

Contrib Mineral Petrol -2001) 140: 701±719
DOI 10.1007/s004100000215
E. A. Dunworth á E.-R. Neumann á J. M. Rosenbaum
The Skien lavas, Oslo Rift: petrological disequilibrium
and geochemical evolution
Received: 30 September 1999 / Accepted: 12 October 2000 / Published online: 12 January 2001
Ó Springer-Verlag 2001
Abstract The Skien lavas, which form the earliest phase existed when the host basalts were erupted. All epsilon
of basaltic magmatism within the Permo-Carboniferous values lie within the range of data previously published
Oslo Rift, contain multiple generations of clinopyroxene for the Skien lavas. Detailed examination of the chem-
which exhibit strong petrological and geochemical dis- ical, isotopic and textural disequilibrium features seen in
equilibrium. Three principal core compositions have these lavas has enabled us to place constraints upon the
been identi®ed: -1) low-jadeite, high-Mg, Cr-diopside magmatic evolution of this basalt suite, ranging from
cores -CrMgDi) with strongly depleted trace-element xenocryst incorporation to cognate multistage pyroxene
signatures, which are believed to be xenocrystic in ori- growth, as well as identifying clear evidence of magma
gin; -2) Mg-rich, Cr-poor diopside cores -MgDi) with mixingand possible crustal contamination.
moderately depleted trace-element signatures which
probably represent early cognate growth; and -3) more
dominant, low-Mg, phenocrystic diopside cores -Phen-
Di). Several samples contain CrMgDi or MgDi cores
which have been subjected to resorption and partial re-
equilibration with their host melts, indicative of exten-
Introduction
Most geochemical studies of igneous rocks have as-
sive disequilibrium and magma mixing. These three core
sumed, often out of analytical necessity, that the samples
types are overgrown by trace-element-enriched Ti-aug-
beingstudied were formed under equilibrium conditions.
ite, which also forms megacrysts and late-stage lava
groundmass. Calculated Ti-augite/melt partition coe-
cients show clinopyroxene compatibility of the M-
Whereas petrographic descriptions have sometimes
shown evidence of disequilibrium crystallization -e.g.,
Duda and Schmincke 1985; Dunworth and Wilson 1998;
HREE, Zr, Hf and Y. The LILE, Sr, and Nb remain
Neumann et al. 1999), it has often been dicult to assess
incompatible. ꢀSr300 and ꢀNd300 of Ti-augite over-
growths, phenocrystic diopside, and MgDi diopside
cores show that intrasample isotopic disequilibrium
particularly between individual mineral phases, or even
between di€erent generations of growth of a single
the degree of corresponding geochemical and isotopic
disequilibrium preserved within individual samples,
phase. Such combined petrographic and geochemical
E. A. Dunworth -&)
characterizations are important for three primary
reasons: -1) to fully understand the relevance -and
limitations) of ``whole-rock'' analyses -i.e., isochrons,
trace-element data, calculated initial isotopic ratios,
etc.); -2) to identify the mantle -and crustal) sources that
have contributed to the magmas; and -3) to characterize
the magmatic evolution of the rocks being studied.
The Skien ``basalts'' -alkali basalts, tephrites and
nephelinites), which erupted in the earliest stages of
Permo-Carboniferous B1 basaltic magmatism within the
Oslo Rift -Sundvoll et al. 1990), o€er an excellent op-
portunity for the study of disequilibrium crystallization
since samples taken from many of the ¯ows in the
ꢀ1,500-m-thick sequence are clearly not in petrographic
or geochemical equilibrium. Our aim was to characterize
Mineralogisk-Geologisk Museum,
Sars'gate 1, 0562 Oslo, Norway
E-mail: lizzyann@magma.ca
E.-R. Neumann
Department of Geology,
Universitetet i Oslo, Postboks 1047 Blindern,
0316 Oslo, Norway
J. M. Rosenbaum
PRIS, University of Reading,
P.O. Box 227 Whiteknights, Reading,
Berks, UK
J. M. Rosenbaum
Present address: Geochemical Solutions Ltd.,
31 Harpsden Road, Henley-on-Thames,
Oxon RG9 1EE, UK
Editorial responsibility: I. Parsons

702
the type and degree of petrographic and geochemical
disequilbrium within the Skien basalts, focussingin
particular on the disequilibrium features exhibited by
the multiple types of clinopyroxene found within the
volcanic sequence, and to examine the problems asso-
ciated with integrating these features into a model of
magmatic evolution. It should be noted that, in the
context of this study, the term ``dissolution'' is de®ned
as the combination of chemical reaction or resorption
-kinetics) and partial melting-thermodynamics) that
occurs when a crystal encounters a melt with which it is
not in thermodynamic equilibrium.
Previous work
The Permo-Carboniferous Oslo Rift, alongwith its as-
sociated magmatic events, has been the subject of many
petrogenetic -Segalstad 1979; Neumann 1986; Schou-
Jensen and Neumann 1988; Neumann 1994), geochemical
-Anthony et al. 1989; Neumann et al. 1988, 1990), struc-
tural -Sundvoll and Larsen 1994; Heeremans et al. 1996)
and geophysical studies -Ramberg 1976; Ro et al. 1990).
The NNE-SSW-trendingrift contains three major
sections: the northern Akershus segment, the central
Vestfold segment, and the southern -submarine)
Skaggerak graben. Five stages of rifting, lasting from
about 315 to 240 Ma, are described in Rambergand
Larsen -1978). The earliest basaltic activity began
ꢀ300 Ma ago on the southwestern ¯ank of the Vestfold
segment, with the eruption of ꢀ1,500 m of nephelinitic-
basaltic ¯ows at Skien -Fig. 1), and it is this sequence
of ¯ows which is the focus of the present study. Earlier
petrogenetic and geochemical studies of the Skien bas-
alts by Segalstad -1979), Neumann et al. -1988) and
Anthony et al. -1989) have already shown preliminary
evidence of multiple generation of clinopyroxene
growth, intrasample Sr-Nd isotopic disequilibrium, and
the fact that at least one of the mantle sources tapped by
the Skien lavas was not involved in the genesis of any of
the other basaltic sequences found within the Oslo Rift.
Fig. 1 Map of the onshore section of the Oslo Rift -after Ramberg
and Larsen 1978). The Skien basalts represent the most south-
western basaltic activity within the rift. The main samplingsite is
located just NE of the ``n'' of Skien, while the Porsgrunn road-cut
is located just north of the main fault cuttingthrough the southern
part of the Skien basalt area
Analytical methods
Major-element analyses of pyroxenes were carried out
with a Cameca Camebax microprobe at the Geologisk
Museum, Oslo, usinga 15-kV, 20-nA beam, and 10 s
countingtimes for all elements except Cr and Ni -30 s). in the range 0.3±2 wt% -Neumann et al. 1999; this
Analyses were calibrated usingsilicate standards: albite study).
or omphacite -Na), K-feldspar -K), wollastonite -Ca,
Si), and oxides -Al, Mg, Fe, Mn, Ti, Cr, Ni). Duplicate termined at the University of Edinburgh Ion Microprobe
analyses usingforsterite -Mg) and fayalite -Fe) standard Facility with a modi®ed Cameca IMS 4f ion microprobe
Trace-element abundances of the pyroxenes were de-
calibrations produced similar results within the limits of usinga ꢀ15±20-lm-diameter, ꢀ5-nA O± primary beam.
analytical precision. Data reduction was carried out In order to reduce molecular interference, only positive
usingPAP Came ca software. Analytical precision is es- secondary ions with initial energies of 55±95 eV were
timated to be better than ‹1% for oxides >20 wt%, analyzed. Peak-strippingcorrections were made for
‹2% for oxides in the range 10±20 wt%, ‹5% for molecular interferences of BaO, BaAl, BaSi, rare earth
oxides in the range 2±10 wt%, and <10% for elements element -REE) oxides, and REEs-Al and REEs-Si where

703
necessary. On the basis of the repeated testingof primary overall lava sequence. The main sequence of lavas at
and internal secondary silicate and pyroxene standards Skien consists of approximately 70 nephelinite, tephrite
with a similar matrix composition, all other molecular and alkali basalt ¯ows, numbered SKX 01±70 from base
interferences were assumed to be negligible. Accuracy for to top. A cross section and stratigraphic column are
all elements is ‹3±10% except for Hf -‹16%; 1r error). shown in Segalstad -1979). Further south, an additional,
Element intensity ratios were measured relative to Si smaller road-cut sequence, just north of Porsgrunn
whose content was veri®ed by electron microprobe -ac- -Fig. 1), contains pyroxene-phyric ¯ows which have been
curacy‹1%). Element contents were calculated on the numbered SKX 80±90. Their exact stratigraphic rela-
basis of ion yields from NIST SRM 610 silicate glass tionship with the main 70-¯ow Skien sequence is cur-
analyzed under the same conditions -Hinton et al. 1995) rently unclear but they appear to bear closest resemblance
and normalised to the NIST SRM 610 ``preferred aver- to ¯ows around SKX 10±15. North of the main Skien
age'' trace-element contents of Pearce et al. -1997). While basalts, a small block of pyroxene-phyric meta-basalt
we realise that the latter values are based on ICPMS ¯ows -SRX 1±6) are found near Skrim -Fig. 1) which are
rather than ion-probe determinations, the Pearce et al. again believed to be related to the Skien basalts.
-1997) values have been used to ensure internal consis-
The majority of the lower ¯ows in the main sequence,
tency with a LAM-ICPMS data set obtained from re- and all of the ¯ows in the additional sequence are clin-
lated samples from the Oslo Rift which will be published opyroxene-phyric, grading into clinopyroxene-plagio-
separately and which has been normalised usingthe
NIST SRM 610 and 612 values of Pearce et al. -1997).
clase-phyric towards the upper end of the main
sequence. Rare medium-coarse-grained, resorbed and
For geochemical whole-rock and pyroxene mineral- altered olivine -Fo 85±87) has been found in three
separate analyses, the samples were ®rst split usinga samples at the base of the main sequence. Partitioning
hydraulic splitter and jaw crusher. The gravels were calculations based on whole-rock data from these sam-
subsequently ground to a ®ne powder -<10 lm) in a ples -Mg# ˆ 100  Mg=ꢁMg ‡ Fe† ꢂ 69) as well as
steel shatterbox for whole-rock geochemical analysis, or experimental data from Roedder and Emslie -1970)
fed through a sieve crusher from which pyroxene sepa- suggest that these olivine crystals are early phenocrysts.
rates -500±250 lm) were obtained by magnetic separa- Flow SKX 40 -Mg# 56) contains Fo 89±90 olivine cores
tion and handpicking.
with 0.37±0.41% NiO which are overgrown by Fo 72±75
Major-element whole-rock analyses were obtained by rims. The latter composition also forms rare, late-stage
XRF -Universitetet i Oslo) and trace-element analyses olivine phenocrysts within the sample. The cores are
by ICP-MS -Actlabs, Ontario). Analytical precision for interpreted as xenocrysts whereas the rims and ground-
the former is given by Neumann et al. -1990), and for the mass phenocrysts are considered to be cognate. The ®rst
latter by Dunworth and Bell -2001).
appearance of plagioclase within the Skien sequence is in
Sr-Nd isotopic analyses were carried out at the the groundmass of ¯ow 40 -An 62), and it reappears in
Geologisk Musuem, Oslo using a VG Micromass 54 ¯ows 55 and above, with more sodic compositions an-
-Rb, Sr, Sm) and a Finegan-MAT 262 -Nd) mass spec- alyzed in ¯ow 55 -ꢀAn 40). It forms a phenocryst phase
trometer. Pyroxene separates were ground to ꢀ10 lm in in ¯ows in the upper 50s and early 60s where it is too
alcohol with a corundum pestle and mortar. Both the altered to yield reproducible analyses. Apatite, resorbed
pyroxene and whole-rock powders were washed in eth- titanite, melanite garnet, nepheline and iron-oxide min-
anol and water and subsequently leached in 6N HCl at erals are also found as groundmass phases in many of
100 °C for 4 h before beingdried, weighed, and dis-
solved. Dissolution was carried out in sealed 15-ml Sa-
villex beakers usinga HNO 3-HF mixture held at 120 °C
the Skien lavas -Segalstad 1979; this study).
for 48 h. Subsequent chemical and mass spectrometric Clinopyroxene petrography and
procedures were based on those described by Mearns major-element chemistry
-1986) and Neumann et al. -1990).
It should be noted that the sample numbering Clinopyroxene is abundant and petrographically hetero-
schemes of Segelstad -1979) and Anthony et al. -1989) geneous in the Skien lavas, and occurs in a variety of
are similar but not identical to that of the present study. textural and compositional varieties -Table 1, Fig. 2).
Due to extensive road alterations at the principal sam- The major- and trace-element composition of represen-
plingsite duringthe previous decade, not all the samples
tative samples from each of the di€erent compositional
in the lowermost part of the Skien sequence necessarily types are given in Table 2, and illustrated graphically in
correlate directly -e.g., S07 ± Segelstad -1979) ± and SK Figs. 3 and 4. They are listed and described below
07 ± Anthony et al. -1989) ± versus SKX 07 ± this study). accordingto their composition.
1. Cr-diopside cores -CrMgDi). These unzoned,
bright green cores often show evidence of dissolution
around the edges, and are always mantled with later
Results
generations of pyroxene -Fig. 2a). Such cores have been
The chemistry of the Skien pyroxenes should be studied found in seven ¯ows: SKX 07, 11, 31, 40, 44, 55 and 89A.
and interpreted in the light of the petrography of the Compositionally, these cores are distinct in Cr2O3-Mg

704
Table 1 Relationships between texture and composition for pyroxenes from the Skien basalts
Compositional type
Textural type
CrMgDi
MgDi
Cr-diopside
Xenocrystic core
Unzoned anhedral-subhedral distinct core of crystal, overgrown
with concentrically zoned clinopyroxene -Fig. 2a)
MgO>16 wt%
Cr2O3>0.9 wt%
Cr-rich PhenDi
MgO=14.7±15.1 wt%
Cr2O3=0.6±1.0 wt%
Mg-rich, Cr-poor diopside
MgO>16 wt%
Overgrowth
Mottled core
``Clusters''
Late-stage overgrowth rim in SKX 87 -Fig. 5) and rarely elsewhere
Resorbed mottled core, often with indistinct boundaries due to
reaction with host melt
Cr2O3<0.8 wt%
Glomeroporphyritic clusters of unzoned but sometimes twinned
crystals, often sub/euhedral. Rare, most common in SKX 07
Anhedral crystals showingstrained extinction -only SKX 07)
Less resorbed/mottled cores, with core edges in equilibrium with
rims. Concentric zones of diopside may periodically appear in
broad rim dominated by Ti-augite
``Strained''
Phenocryst
PhenDi
Moderately Mg-rich
-14±15.5 wt% MgO)
Ti-augite
Si-poor -45±48 wt% SiO2)
Ti-rich -2±4 wt% TiO2)
Mg-poor -10.5±14 wt% MgO)
Rims
Concentrically and sector-zoned rims found around xenocrystic,
mottled or phenocrystic cores -Fig. 2a, d)
``Strained''
Euhedral crystals -often megacrysts) which also show strained
extinction as well as strongsector zonation
Groundmass crystals
Groundmass
space -Fig. 3c). The most Mg-rich compositions white zone in Fig. 2a). Compositionally, these diopsides
-ꢃ18 wt% MgO) are found in sample SKX 11 -Fig. 4a± have lower Mgand higher Tschermak contents than the
c). The Cr-diopside cores are also characterized by low MgDi compositions -Fig. 3). Their aegirine contents
Na, Al and Ti contents, and hence contain low pro- remain low -1.5±4%; Fig. 3f) whereas their Cr contents
portions of jadeite and Tschermak components.
are very varied -0.0±1.1 wt% Cr2O3). The high-Cr
2. Mg-rich, Cr-poor diopside compositions -MgDi). compositions are found occasionally as late-stage re-
These are often bright green -depending on their Cr sorption rims, or overgrowth rims believed to be formed
contents), and have been found in a greater number of as a result of magma mixing -Fig. 5), prior to ®nal Ti-
¯ows than the Cr-diopside cores -SKX 04, 05, 07, 14, 20, augite rim crystallization. The diopsides form a com-
27, 37B, 38, 40, 41, 55, 87, 89A and 89B). Composi- positional continuum between the MgDi compositions
tionally, they have a wider range of Mg -16±17.5 wt% and the Ti-augite overgrowths with MgO contents
MgO) and lower Cr -0.1±0.8 wt% Cr2O3) contents but a ranging from 14 to 16 wt%. The boundaries between the
higher content of Tschermak components than the compositional groupings have been determined on the
Cr-diopside cores described above -Figs. 3, 4). Variation basis of both textural and compositional constraints.
in Al/Ti ratios is also common within the suite -Fig. 4d).
MgDi compositions are most commonly found as
parts of mottled crystal cores -e.g., Fig. 5), and most
likely represent an example of the earliest stage of py-
roxene crystallization within the magmas, or they are
xenocrysts which have, in some samples, been subjected
to varyingdegrees of dissolution by their host melts.
MgDi compositions are also found, particularly in
sample SKX 07, formingglomeroporphyritic clusters of
unzoned clinopyroxene crystals, some of which are
twinned, and which are rarely overgrown by later gen-
erations of pyroxene growth. The compositions of the
glomeroporphyritic crystals are plotted in Fig. 3 as
``cluster'' analyses, and have low Al/Ti ratios compared
with most of the other MgDi analyses. Sample SKX 07
is also host to rare anhedral crystals showingstrained
extinction and containingMgDi zones which are not
4. Ti-augite. This, the most common type of pyroxene
in these samples, is identi®ed easily by its dark brown
colour in hand specimen, and pale pink-brown colour in
plane-polarized-light. Although a few samples contain
Ti-augite phenocrysts -e.g., SKX 70; Fig. 2d), the com-
position is most commonly found as concentrically
zoned overgrowths on earlier-formed core compositions
-Fig. 2a, c), and as groundmass crystals. Ti-augite also
forms rare euhedral megacrysts in a few samples, some
of which show strained extinction, and are plotted as
part of the ``strained'' group in Fig. 3. Compositionally,
the Ti-augites have the lowest MgO -10±14 wt%) and
highest Tschermak contents of all the pyroxene types in
the Skien lavas -Fig. 3). Their aegirine contents range
from ꢀ1.5 to 6% -Fig. 3f).
mottled or resorbed -Fig. 2c). These analyses have been Geothermobarometry
plotted as part of the ``strained'' group in Fig. 3, and
again overlap the main MgDi compositions.
The major element data from the four pyroxene types
3. PhenDi. Yellow/green diopside forms either unre- described above were used to estimate pressures and
sorbed phenocryst cores or diopsidic zones within con- temperatures of crystallization, usingthe statistical
centrically zoned Ti-augite rim overgrowths -e.g., thin clinopyroxene geothermobarometer of Soesoo -1997).

705
Fig. 2 Photomicrographs -cross-polarized light) of clinopyroxenes
from the Skien lavas, showingthe four main compositional types.
The ®eld of view for all panels is 2 mm wide. a Clinopyroxene from
sample SKX 07. The unzoned -white) clinopyroxene core is Cr-
diopside, and is overgrown by concentrically and sector-zoned Ti-
augite. The thin white concentric zone appearing approximately a
third of the way out from the core -in the Ti-augite overgrowth) has
a PhenDi composition. b Clinopyroxene crystal from sample SKX
14, with a Mg-rich, Cr-poor core -MgDi) exhibiting mottled
zonation and evidence of dissolution. The core is overgrown by an
extensively remelted zone, essentially Ti-augite in composition, and
®nally by a smoother, homogeneous Ti-augite rim. c Anhedral
MgDi-rich pyroxene from sample SKX 07 which shows strained
clinopyroxenes, and was not used to calculate PT crys-
tallization conditions for any of the PhenDi or Ti-augite
compositions.
Clinopyroxene trace-element geochemistry
Five samples, containinga wide and representative va-
riety of the clinopyroxene types described above, were
selected for pyroxene trace-element ion-probe analysis
-SKX 07, 11, 14, 44, 70). Representative results are
provided in Table 2, and data from the ®rst four samples
are plotted as mantle-normalised spider-diagrams in
Fig. 6a±d, along with the whole-rock spidergram for
each sample.
extinction. d Concentrically and sector-zoned phenocryst of
Ti-augite containing no other type of core material -from sample
SKX 70)
The results are plotted in Fig. 3f, and suggest that the Ti-
It is clear from Fig. 6a±d that the di€erent genera-
augite compositions crystallized between approximately tions of clinopyroxene within each sample are not in
1,100 and 1,150 °C, and at relatively low pressures equilibrium with respect to trace-element composition,
-<3 kb) whereas the PhenDi compositions lie within the and that the large-ion-lithophile-elements -LILE) in
2±5 kbar and 1,150±1,200 °C ®eld. However, the particular show very inconsistent behaviour. The Ti-
CrMgDi and MgDi compositions lie outside the range augite trace-element compositions -overgrowths, and
of the ®eld of calibration of the Soesoo -1997) geother- ``strained'' in SKX 14) are commonly more trace-
mobarometer. The clinopyroxene geothermobarometer element-enriched than the related host rock.
from Putirka et al. -1996), which is based on the com-
position of clinopyroxene and co-existingliquid, was
A comparison of average trace-element compositions
for di€erent types of Skien pyroxene -listed in Table 2) is
used to estimate pressure and temperature of crystalli- plotted in Fig. 6e. Particularly noteworthy are the
zation of the MgDi compositions from sample SKX 14, extremely homogeneous trace-element compositions of
which is the most primitive lava found at Skien. the Ti-augite throughout the Skien sequence, the rela-
The calculations gave pressures of ꢀ12 kbar and tively close similarity in trace-element signatures be-
temperatures of ꢀ1,250 °C. The Putirka et al. -1996) tween the CrMgDi and MgDi compositions, the
geothermobarometer is not calibrated for Fe3+-rich similarity in composition between the MgDi cores and

706
Table 2 Average clinopyroxene compositions of representative Trace-element data -ppm) determined by Cameca ion probe -see
samples of the pyroxenes shown in Figs. 3±6. Major-element data Analytical methods for further details). ``Avg. -7/3)'' indicates the
-wt%) obtained usinga Camebax microprobe. Analyses norma-
average of the number of major-element and trace-element ana-
lised with 4 cations:6 oxygens; Fe3+ calculated by charge-balance. lyses, respectively
Pyx-type
Sample
Cr-diopside -CrDi)
Mg-rich, Cr-poor -MgDi)
No. analyses
-major/trace)
SKX 07
Avg. -1/1)
SKX 11
Avg. -3/2)
SKX 44
Avg. -3/3)
SKX 07 SKX 07
Avg. -14/5) Avg. -4/3)
SKX 14
Avg. -1/2)
-cluster)
-strained)
SiO2
53.18
0.61
1.67
1.07
0.59
2.84
0.06
16.67
0.10
23.37
0.32
b.d.
53.61
0.45
1.21
0.99
0.43
2.83
0.06
17.47
0.07
22.41
0.37
0.02
52.55
0.60
2.36
1.07
1.40
3.01
0.08
16.27
0.04
22.57
0.50
0.02
52.96
0.73
1.67
0.57
0.60
3.61
0.09
16.21
0.05
23.25
0.32
0.03
52.53
0.81
2.10
0.41
1.53
3.29
0.09
16.00
0.10
23.04
0.44
0.02
52.99
0.58
2.13
0.51
0.76
3.13
0.05
16.39
b.d.
TiO2
Al2O3
Cr2O3
Fe2O3
FeO
MnO
MgO
NiO
CaO
Na2O
K2O
23.01
0.42
0.02
Total
100.48
99.91
100.47
100.08
100.36
99.99
Si
1.935
0.065
1.955
0.045
0.001
0.011
0.011
0.012
0.028
0.086
0.002
0.884
0.002
0.876
0.026
1.916
0.084
1.941
0.059
1.921
0.078
0.005
0.017
0.041
0.022
0.012
0.101
0.003
0.893
0.003
0.903
0.031
0.001
1.936
0.064
ivAl
ivFe3+
viAl
viFe3+
Ti
0.007
0.016
0.017
0.031
0.086
0.002
0.950
0.003
0.911
0.023
0.018
0.039
0.016
0.031
0.092
0.002
0.885
0.002
0.882
0.035
0.013
0.018
0.020
0.017
0.111
0.003
0.872
0.001
0.913
0.023
0.002
0.028
0.021
0.016
0.015
0.096
0.002
Cr
Fe2+
Mn
Mg0.904
Ni
Ca
Na
K
0.901
0.030
Total
4.000
4.000
4.000
4.000
4.000
4.000
Ba
Th
U
Nb
Sr
Zr
Hf
Ti
0.15
0.038
0.051
0.21
94.2
10.07
0.35
3148
2.65
2.23
0.37
0.010
0.032
0.19
100.2
6.2
0.33
0.021
0.028
0.28
78.2
8.6
0.56
0.016
0.015
0.20
114.6
13.3
1.16
0.020
0.055
0.16
117.3
15.8
0.42
0.024
0.011
0.16
96.5
13.2
0.72
3745
3.01
2.19
7.16
1.30
7.37
1.81
0.56
1.48
0.19
0.76
0.12
0.37
0.045
0.33
0.032
0.22
2168
1.63
2.04
6.16
0.99
4.94
1.08
0.33
0.66
0.08
0.52
0.06
0.13
0.026
0.30
0.009
0.33
2919
3.30
1.73
5.44
0.90
6.01
1.55
0.50
1.12
0.14
1.00
0.12
0.32
0.049
0.35
0.025
0.69
0.91
3732
4.60
3.21
10.43
1.82
9.36
4003
5.18
3.27
10.74
1.95
10.23
2.65
0.69
1.36
0.23
Y
La
Ce
Pr
7.63
1.17
6.35
Nd
Sm
Eu
Gd
Tb
Dy
Ho
Er
2.07
0.68
1.43
0.23
1.13
0.19
0.45
0.053
0.67
0.038
1.40
0.11
0.85
0.09
0.23
0.04
0.39
0.03
0.94
0.16
0.49
0.074
0.86
0.100
Tm
Yb
Lu
Pyx-type
Sample
Phenocryst diopside -PhenDi)
Ti-augite
No. analyses
-major/trace)
SKX 01
Avg. -8)
SKX 07
Avg. -3)
SKX 11
Avg. -1/1)
SKX 07
Avg. -7/1)
SKX 11
Avg. -3/2)
SKX 14
Avg. -9/6)
SKX 44
Avg. -7/4)
SKX 70
Avg. -27/5)
SiO2
TiO2
Al2O3
Cr2O3
50.72
1.27
3.67
0.40
51.37
1.36
2.65
0.31
51.48
0.94
2.93
0.28
48.54
2.35
4.68
0.07
46.92
2.24
6.24
0.09
46.24
2.81
6.74
0.08
47.80
2.35
5.89
0.03
44.19
3.45
8.91
0.06

707
Table 2 -Contd.)
Pyx-type
Sample
Phenocryst diopside -PhenDi)
Ti-augite
No. analyses
-major/trace)
SKX 01
Avg. -8)
SKX 07
Avg. -3)
SKX 11
Avg. -1/1)
SKX 07
Avg. -7/1)
SKX 11
Avg. -3/2)
SKX 14
Avg. -9/6)
SKX 44
Avg. -7/4)
SKX 70
Avg. -27/5)
Fe2O3
FeO
1.55
2.63
1.98
3.87
1.57
3.84
3.54
4.16
4.22
3.21
3.98
4.12
3.16
4.47
4.27
4.37
MnO
MgO
NiO
0.06
14.93
0.05
0.10
15.26
0.07
0.11
15.45
0.08
0.16
13.60
0.06
0.09
12.94
0.03
0.14
12.00
0.03
0.17
12.72
0.03
0.15
11.10
0.03
CaO
Na2O
K2O
24.41
0.25
0.02
23.22
0.36
0.01
22.52
0.42
0.01
22.63
0.55
0.02
22.82
0.53
0.02
23.14
0.53
0.02
22.83
0.57
0.02
22.43
0.62
0.01
Total
99.96
100.55
99.63
100.34
99.35
99.83
100.03
99.58
Si
1.867
0.133
1.886
0.113
0.002
0.002
0.054
0.038
0.009
0.119
0.003
0.850
0.002
0.914
0.026
1.900
0.100
1.802
0.196
0.006
0.012
0.097
0.066
0.002
0.129
0.005
0.723
0.002
0.900
0.040
1.758
0.242
1.735
0.265
1.781
0.219
1.666
0.334
ivAl
ivFe3+
viAl
viFe3+
Ti
0.026
0.043
0.035
0.012
0.081
0.002
0.835
0.002
0.963
0.018
0.027
0.044
0.026
0.008
0.119
0.003
0.753
0.002
0.890
0.030
0.034
0.119
0.063
0.003
0.101
0.003
0.671
0.001
0.916
0.039
0.033
0.112
0.079
0.003
0.129
0.004
0.707
0.002
0.930
0.039
0.040
0.089
0.066
0.001
0.140
0.005
0.062
0.121
0.098
0.002
0.138
0.005
Cr
Fe2+
Mn
Mg0.819
Ni
0.624
0.001
0.912
0.041
0.001
0.906
0.045
Ca
Na
K
Total
4.000
4.000
4.000
4.000
4.000
4.000
4.000
4.000
Ba
Th
U
Nb
Sr
Zr
Hf
Ti
0.28
0.05
0.01
0.37
119.5
46.52
2.07
8,088
6.45
5.16
17.35
2.95
16.04
3.80
1.23
1.82
0.32
1.57
0.22
0.51
0.07
0.46
0.03
0.59
0.06
0.05
1.50
230.9
156.27
6.53
12,974
19.28
18.35
58.04
9.31
45.61
9.33
2.47
7.00
0.85
4.80
0.71
1.84
0.99
0.168
0.030
1.86
252.9
184.7
6.70
14,464
17.04
18.45
57.33
9.53
46.04
10.35
3.15
1.23
0.176
0.033
2.45
264.8
231.7
7.75
16,166
19.95
21.28
66.04
10.86
55.02
11.22
3.15
0.63
0.71
0.196
0.021
2.57
226.9
216.7
7.33
15,975
18.17
19.64
63.00
10.56
52.95
11.33
3.24
0.180
0.032
2.29
251.7
242.3
8.75
17,322
22.60
22.11
68.00
11.03
57.42
11.95
3.72
Y
La
Ce
Pr
Nd
Sm
Eu
Gd
Tb
Dy
Ho
Er
4.53
0.78
3.95
0.68
7.70
1.02
5.07
0.79
7.64
1.25
5.43
0.86
6.54
0.93
4.32
0.68
1.50
1.78
1.94
1.63
Tm
Yb
Lu
0.282
1.52
0.121
0.160
1.59
0.123
0.229
1.57
0.170
0.162
1.34
0.182
0.190
1.38
0.147
twinned glomeroporphyritic MgDi compositions from
Also shown in Fig. 6e is the calculated composition
sample SKX 07, and the clear di€erences between the Ti- of clinopyroxene in equilibrium with the whole-rock
augite and Mg-rich diopsidic trace-element patterns, composition of sample SKX 14 at 3 GPa and 1,380 °C,
especially with respect to REE, Zr and Hf contents. As usingpartition coecients taken from Hart and Dunn
expected, the single PhenDi analysis from sample SKX -1993; Table 3). SKX 14 is one of the most primitive
11 lies between the Ti-augite and MgDi analyses. De- lavas found at Skien -Mg#ꢀ61; Al2O3=10.8 wt%;
spite their general similarity, there are some consistent Cr= 270 ppm; Mg#=100 ´ Mg/-Mg+FeT)), with the
di€erences between the CrMgDi and MgDi composi- lowest REE contents. The calculated pyroxene com-
tions, particularly with respect to Hf*N -where Hf ÃN ˆ position lies between that of the PhenDi and MgDi
HfN À ꢁZrN ‡ TiN †=22) which is consistently positive in analyses and is believed to be representative of the
the MgDi compositions, and negative in the Cr-diopside earliest cognate pyroxenes, although earlier fractiona-
cores.
tion may have enriched the trace-element composition

708
Fig. 3a±f Major-element and pyroxene-component plots showing
the compositions of the di€erent types of clinopyroxene found in
the Skien lavas, as de®ned in this study on the basis of petrographic
and compositional data. ``Tschermak'' includes the combined
calculated presence of the components CaAl2SiO6, CaFe3+AlSiO6,
CaFe3+2SiO6, and CaTiAl2O6. Na has been assumed to be present
in the aegirine component. The ``cluster'' analyses refer to
glomeroporphyritic associations of mostly MgDi crystals in sample
SKX 07. ``Strained'' crystals are those which exhibit obvious
strained extinction. They include an anhedral MgDi crystal from
sample SKX 07, and euhedral Ti-augite crystals from samples SKX
14 and 44. Figure 3f shows calculated pressure-temperature
crystallization conditions for the pyroxenes based on the statistical
multivariate calculations of Soesoo -1997). †XPT ˆ 0:446SiO2‡
0:187TiO2 À 0:404Al2O3 ‡ 0:346FeOꢁtot† À 0:052MnO ‡ 0:309MgO‡
of sample SKX 14 relative to the original, primitive
magma. However, given the whole-rock Mg# of 61, we
do not believe that the whole-rock composition has
undergone substantial fractionation, and thus the
``calculated pyroxene'' composition cannot be substan-
tially more enriched than the true values of the earliest
cognate pyroxenes. It must be noted that the CrMgDi
compositions are substantially more depleted in most
elements, particularly in Zr-Hf, than the ``calculated
pyroxene'' composition.
The consistency of the Ti-augite trace-element
compositions supports the assertion that these pyrox-
enes were in equilibrium with their host melts during
late-stage magma crystallization. In turn, this obser-
vation has allowed us to calculate Ti-augite/melt par-
tition coecients -D) for each sample usingboth
whole-rock compositions as well as groundmass com-
positions estimated by point-countingand stripping
the earlier diopside growth compositions from the
0:431CaO À 0:466Na2OO.
OYPT ˆ À0:369SiO2 ‡ 0:535TiO2À
0:317Al2O3‡ 0:323FeOꢁtot† ‡ 0:235MnO À 0:516MgO À 0:167CaOÀ
0:153Na2OO. This geothermobarometer was calibrated for 338
experimental analyses ranging from 1 atm to 34 kbar and 1,000±
1,470° C, and subsequently tested usingseveral hundred natural
analyses from ma®c rocks, mostly from the oceanic environment,
whose petrographic histories are well constrained

709
Fig. 4 a±c Compositions of
Cr-diopside cores -CrMgDi) as
used in Fig. 3, plotted according
to sample number. Sample SKX
11 contains the most depleted
-high-Mg, low-Tschermak) xe-
nocryst compositions of all
samples analyzed. The low-Mg,
high-Cr analysis from sample
SKX 11 comes from the rim of
a xenocryst core, and may rep-
resent the result of xenocryst
reaction with the host melt.
d±f Compositions of Mg-rich,
Cr-poor diopside cores -MgDi)
as used in Fig. 3, plotted ac-
cordingto sample number. The
graphs do not include the
``cluster'' and ``strained'' ana-
lyses from sample SKX 07
which have compositions simi-
lar to the MgDi analyses from
the same sample
whole-rock analyses. These D values are listed in enes crystallizingunder high temperature and pressure
Table 3 and plotted in Fig. 6f along with clinopy- conditions. The samples analyzed in this study for
roxene partition coecients taken from the literature. pyroxene trace-element compositions are solely py-
Our results show that, duringthe late stages of crys-
roxene-phyric, and contain no plagioclase or signi®-
tallization at approximately 1,100±1,160 °C and <2 kb cant quantities of other minerals that might a€ect
-Fig. 3f), most elements more ``compatible'' than Nd trace-element partitioning, with the exception of Ti
are compatible in Ti-augite. Our calculated values partioning into minor amounts of groundmass mag-
show closest resemblance to those of Francalanci et al. netite. Crystallization of minor amounts of magnetite
-1989), which were calculated for similar compositions prior to the formation of the Ti-augite compositions
under similar crystallization conditions and which are would mean that our calculated DTi values are mini-
substantially higher than those calculated for pyrox- mum values.

710
reported in the radiogenic isotope systems -Anthony
et al. 1989), Rb-Sr and Sm-Nd isotopic analyses were
conducted on pyroxene separates and related whole-
rock powders. The pyroxenes were handpicked accord-
ing to colour -bright green=Cr-rich -CrMgDi or
MgDi), yellow-green=PhenDi, and brown=Ti-augite)
and subsequently powdered. All powders were leached
accordingto the procedures described above -cf. Ana-
lytical methods). The results are given in Table 4. Data
for the leached whole-rock powders and pyroxenes,
alongwith nonleached whole-rock and pyroxene ana-
lyses from Anthony et al. -1989), are shown in Fig. 7.
The di€erences in the Rb/Sr -and Sm/Nd) ratios between
leached and nonleached whole-rock and pyroxene ana-
lyses, and the subsequent e€ects on the age-corrected
isotope ratios -Table 4) are discussed in detail below.
The combined pyroxene and whole-rock initial-ratio
isotope data -Fig. 7a) demonstrate clear intrasample dis-
equilibrium between the di€erent generations of pyroxene
and their host magmas, particularly in the leached sam-
ples. The reasons for pyroxene-whole-rock disequilibrium
may be due to alteration of parent-daughter ratios
c
Fig. 6 a±d Mantle-normalised spidergrams of trace-element data
from di€erent types of pyroxenes in samples SKX 07, 11, 14 and
44. Also shown for comparison is the whole-rock composition for
each of the samples -WR). Mantle normalisation factors were
taken from McDonough and Sun -1995). a The ``cluster'' analyses
in sample SKX 07 are from the same crystals as those whose major-
element data are plotted in Fig. 3. There is one Cr-diopside analysis
-CrMgDi; core in Fig. 2a), two analyses of a MgDi crystal showing
strained extinction -strained; Fig. 2b) whose major-element data is
also plotted as ``strained'' samples in Fig. 3, and one analysis of a
Ti-augite overgrowth rim. b Analyses from sample SKX 11 include
two from the most Mg-rich Cr-diopside core plotted in Fig. 3
-CrMgDi), and three from the overgrowth rim around the core.
The ®rst of these is from a diopside zone, while the other two are
taken from Ti-augite zones. c Sample SKX 14 data include two
MgDi core analyses and two Ti-augite overgrowths taken from the
crystal shown in Fig. 2b, and ``strain'' analyses from a sheared,
euhedral Ti-augite megacryst. d The three trace-element-depleted
xenocryst-core analyses -CrMgDi) and two overgrowth analyses
-Ti-augite) shown in this plot are from a single crystal in sample
SKX 44. The ``megacryst'' analyses are taken from a nonsheared,
sector-zoned Ti-augite megacryst, approximately 2 cm in diameter.
e Average trace-element data for Cr-diopside cores -CrMgDi),
MgDi -including ``strained'' and ``cluster'' analyses from samples
SKX 07), and Ti-augites. Note the consistency of the Ti-augite
compositions, the depleted nature of the Cr-diopside and MgDi
compositions, and the di€erences in HfN* between the Cr-diopside
and MgDi averages -HfN* ˆ HfN À ꢁꢁZrN ‡ TiN†=2†). Also plotted
is the calculated composition -calc pyx) of clinopyroxene crystal-
lizingin equilibrium with a melt of the whole-rock composition of
sample SKX 14 -one of the most primitive lavas found at Skien) at
3 GPa and 1,380° C, calculated usingpyroxene/melt coecients
from Hart and Dunn -1993). f Calculated Ti-augite/melt partition
coecients -D) for the Skien samples -cf. Table 3), based on the
average Ti-augite compositions from the ®ve samples analyzed -cf.
Table 2). The D values for the middle and heavy rare earth
elements, Zr, and Hf are greater than unity. Other relevant
partition coecients from the literature are shown for comparison
-cf. Table 3), alongwith their pressure and temperature of
measurement -where known). The calculated error bars are
normally smaller than the symbols used for each data point
Fig. 5 Clinopyroxene crystal from sample SKX 87 exhibitingan
MgDi core overgrown by Ti-augite, and subjected to subsequent
sector-selective dissolution processes. The partially melted [ab0]
and relatively intact [111] sectors are overgrown by a concentric
zone of Cr-rich, Mg-poor diopside, and ®nally Ti-augite. a Crossed-
polars view of the 2-mm-diameter crystal. b Close-up of one of the
crystal apices shown in a above. c Compositional demonstration of
the relationships between texture and chemistry. Symbols as shown
in a and b above
Sr-Nd isotope geochemistry
In order to test if the chemical and petrographic dis-
equilibrium corresponds with disequilibrium previously

711
through the leaching process. However, experimentation
Work is currently in progress to fully characterize the
with leachingprocedures duringthis study has shown that Sr-Nd-Pb isotopic signatures of the di€erent pyroxene
leachingof the pyroxene separates merely removes sec- types within the Skien basalts and other B1-sequence
ondary alteration and does not a€ect the primary parent- basalts elsewhere in the Oslo Rift.
daughter element ratios and the isotopic ratios. Thus,
intrasample pyroxene disequilbrium is almost certainly a
real characteristic of these samples. When plotted ac- Discussion
cordingto pyroxene or whole-rock types, there are sys-
tematic variations in the isotope data -Fig. 7b). The Ti-augite partition coecients
``green'' pyroxenes -mostly PhenDi compositions) exhibit
very similar 87Sr/86Sri but varied 143Nd/144Ndi signatures, Intrasample disequilibrium can cause diculties in
whereas the brown -Ti-augite) and whole-rock analyses interpretingand modellingwhole-rock data. In the case
show similar Nd but varied Sr signatures. The lone MgDi of the Skien basalts, the results above demonstrate that
analysis comes from sample SKX 37B.
the characterization of such intrasample textural and

712
Fig. 6 -Contd.)
geochemical disequilibrium is necessary in order to clinopyroxene. Speci®c intrasample equilibrium is not
unravel the complex magmatic history and evolution of always seen in isotope space, but the isotope composi-
these lavas. The data clearly show that the Ti-augite is the tions of the whole-rock samples and Ti-augites overlap
only generation of pyroxene growth in the Skien which in a general sense, particularly with respect to ꢀNd.
may be considered to be in equilibrium with the host Alteration e€ects and problems in accurately estimating
lava. This is consistent with their mode of occurrence the primary Rb/Sr parent-daughter ratios may account
asovergrowths around various types of diopsidic for the remainingdiscrepancies in the Sr isotopic system.

713
Table 3 Comparison of high-temperature magmatic partition co- solidus; Blundy et al. 1998). No. 2 1,380 °C, 3 GPa, alkali basalt
ecients for cpx/melt from this study and the literature. WR -Hart and Dunn 1993). No. 3 T and P not available, shoshonite-
D=cpx/whole-rock. GM D=cpx/-estimated) groundmass. The six basalt -Francalanci 1989). No. 4 T and P not available, shoshonite
datasets extracted from the literature have the followingT, P, and
``melt'' composition. No. 1 1,269 °C, 1.0 GPa, cpx-melt -sp. lherz. 1994). No. 6 1,250 °C, 1.0 GPa, alk ol basalt -Skulski et al. 1994)
-Francalanci1989).No.51,270 °C,2.5GPa,nephelinite-Skulskietal.
Sample
Average
-WR)
SD
-%)
Average
-GM)
SD
-%)
No. 1
No. 2
No. 3
No. 4
No. 5
No. 6
Ba
Th
U
Nb
Sr
Zr
Hf
Ti
0.0011
0.031
0.020
0.028
0.27
1.0
1.2
0.78
1.1
0.31
0.44
0.68
0.9
1.0
1.0
1.0
1.0
1.1
1.1
10
22
11
18
10
9
22
13
7
12
14
15
14
13
10
13
14
10
12
12
13
14
14
0.0008
0.020
0.014
0.018
0.20
1.1
1.3
0.80
1.1
0.23
0.34
0.59
0.9
1.1
1.0
1.1
1.1
1.3
1.3
10
22
11
18
10
9
22
13
7
12
14
15
14
13
10
13
14
10
12
12
13
14
14
0.00068
0.01
0.05
0.04
0.03
0.02
0.062
0.27
0.55
0.71
1.27
0.089
0.16
0.0077
0.1283
0.1234
0.256
0.384
0.467
0.15
0.11
0.26
0.35
0.10
0.28
0.014
0.087
0.238
0.595
0.658
0.426
0.079
0.127
0.016
0.076
0.145
0.153
0.340
0.442
0.072
0.122
Y
0.90
0.11
0.22
1.20
0.24
0.39
La
Ce
Pr
0.0536
0.0858
Nd
Sm
Eu
Gd
Tb
Dy
Ho
Er
0.36
0.67
0.38
0.99
0.1873
0.291
0.37
0.62
0.59
0.73
1.12
1.22
0.255
0.359
0.260
0.421
0.58
1.38
0.442
0.387
0.418
0.405
0.571
0.494
1.1
1.0
1.3
1.0
1.3
1.2
1.4
1.2
1.44
Tm
Yb
Lu
1.43
1.48
0.43
0.433
0.58
0.80
1.24
1.10
The consistency of the Ti-augite trace-element com- although the Zr and Hf values calculated for the Skien
positions, and the Nd isotopic equilibrium between the Ti-augites are still signi®cantly higher than those found
whole-rock and Ti-augite data suggest that the calcu- in the other datasets, both in terms of absolute values
lated pyroxene/melt partition coecients are geochemi- and relative to elements with similar degrees of com-
cally reasonable for these lavas, bearingin mind that the patibility. The reasons for this are currently unclear,
middle and heavy rare earth elements -REE) as well as although the high Tschermak-component content of the
Zr, Hf and Y are compatible in Ti-augite formed under Skien Ti-augites may explain the strong compatibility of
low pressures and basaltic magmatic temperatures these quatro-valent cations -Rocholl et al. 1996).
-Neumann 1994). The large ion lithophile elements
-LILE), Nb and Sr are incompatible, and Ti is only
mildly incompatible.
D values were calculated usingboth whole-rock
-WR) and estimated groundmass -GM) -assumed melt)
Origin of the Cr-diopside -CrMgDi) and
MgDi compositions
compositions. The true Ti-augite D values should lie There are currently three main hypotheses to explain
somewhere between these two extremes. DTi is particu- the origins of the CrMgDi and MgDi compositions,
larly dicult to estimate accurately due to uncertainty in these being-1) mantle xenocrysts, -2) partially melted
the relative timing of Ti-augite and groundmass mag- mantle xenocrysts, or -3) early congnate phenocrysts.
netite crystallization. Early crystallization of the latter
will a€ect the content of Ti in the calculated groundmass
compositions used for the D-value calculations. The Mantle xenocrysts
calculated DYb values are believed to be unrealistically
high due to the large analytical uncertainty associated It is common to ®nd Cr-diopsides in mantle lherzolite
with low Yb contents.
xenoliths brought to the surface within Si-undersatu-
Table 3 includes clinopyroxene/melt partition coe- rated basalts -e.g., Nixon 1987). However, most lherz-
cients from Francalanci -1989), Hart and Dunn -1993), olite Cr-diopsides contain ³1.0 wt% Na2O alongwith
Skulski et al. -1994), and Blundy et al. -1998), chosen to signi®cant amounts of viAl to form the jadeite compo-
represent diopsidic clinopyroxenes which have crystal- nent, and have very low Ti contents -e.g., Frey and Prinz
lized under a variety of PT conditions from ma®c 1978; Nixon 1987). Such jadeite-rich Cr-diopsides are
magmas. The D values from groundmass pyroxenes found in mantle xenoliths exhumed by silicate rocks in
from shoshonitic lavas at Stromboli -Francalanci 1989) the nearby 580-Ma Fen carbonatite complex -Grin
are particularly similar to the Skien Ti-augite data, 1973). The results of the present study show that neither

714
Table 4 Sr-Nd isotope results for selected Skien pyroxenes and and Nd concentrations, measured Sr and Nd isotope ratios, and
whole-rock samples -leached whole-rock and pyroxene data plotted the resultant calculated initial isotope ratios. Average of standard
in Fig. 7; see Analytical methods section for details). The ``leached/ analyses carried out duringthis study: NBS 987
ICP-MS'' whole-rock results use ICP-MS Rb, Sr, Sm and Nd 87Sr/86Sr=0.71024‹1 -VG 54) and JM 143Nd/144Nd=0.51111‹1
concentration data, and the leached measured Sr and Nd isotope -Finnegan-MAT 262). Procedure blanks of ꢀ0.18 ngNd and
ratios to calculate the resultant initial ratios. Samples SKX 01, ꢀ1.5 ngSr are considered to have neglibible e€ect on the isotope
49 and 55 also list nonleached isotope-dilution -ID) Rb, Sr, Sm measurements
Sample
Whole rock
Rb
Sr
Rb/Sr 87Rb/86Sr 87Sr/86Srm 87Sr/86Sri ꢀSr
Sm
Nd
Sm/Nd 147Sm/ 143Nd/ 143Nd/ ꢀNd
144Nd 144Ndm 144Ndi
SKX O1
Leached
Nonleached -ID)
Leached/ICP-MS 82
0.56 507.8
73.53 739.1
0.0011 0.0032
0.0995 0.2877
0.1059
0.70359
0.70491
0.70358 ±8.37 4.66 30.35 0.1536 0.0929 0.51248 0.51230 0.91
0.70368 ±6.93 12.40 76.02 0.1631 0.0986 0.51249 0.51230 0.85
774
0.98 702.4
27 667
1.20 78.1
52 1251
0.70228 ±26.74 13
0.70344 ±10.28 5.89 36.24 0.1625 0.0983 0.51248 0.51229 0.73
0.70296 ±17.12 10 55 0.1818 0.51227 0.29
0.70335 ±11.60 3.97 21.15 0.1876 0.1134 0.51259 0.51237 2.24
0.70303 ±16.19 11 60 0.1833 0.51237 2.35
0.70320 ±13.71 4.05 20.77 0.1948 0.1178 0.51256 0.51233 1.54
0.70270 ±20.85 14 67 0.2090 0.51231 1.22
68
0.1912
0.51225 0.05
SKX 14
SKX 34
Leached
0.0014 0.0040
0.0405
0.70346
0.70354
0.70320
Leached/ICP-MS
Leached
Leached/ICP-MS
0.0154 0.0445
0.0416
SKX 37B Leached
Leached/ICP-MS
Leached
0.25 1,073.1 0.0002 0.0007
0.0410
49
1,196
SKX 49
SKX 55
SKX 70
Sample
14.93 898.0
69.00 571.0
0.0166 0.0481
0.1208 0.3494
0.0674
0.70386
0.70483
0.70365 ±7.30 2.89 14.94 0.1936 0.1170 0.51255 0.51232 1.25
0.70334 ±11.76 8.20 42.40 0.1934 0.1169 0.51255 0.51232 1.28
Nonleached -ID)a
Leached/ICP-MS
Leached
Nonleached -ID)a
Leached/ICP-MS
Leached
57
846
0.70303 ±16.19
9
41
0.2098
0.51230 0.88
5.71 1,085.2 0.0053 0.0152
0.70370
0.70376
0.70364 ±7.54 3.81 16.48 0.2312 0.1398 0.51268 0.51240 2.96
0.70331 ±12.19 10.30 59.90 0.1720 0.1040 0.51263 0.51243 3.42
34.30 940.0
22 816
5.50 1,937.4 0.0028 0.0082
0.0365 0.1056
0.0270
0.70337 ±11.34 10
0.70338 ±11.18 7.89 29.66 0.2660 0.1608 0.51251 0.51219 ±1.13
51
0.1961
0.51245 3.78
0.70341
Leached/ICP-MS
61
1,739
0.0351
0.70298 ±16.83 11
59
0.1864
0.51229 0.72c
Pyroxene type
Rb
Sr
Rb/Sr 87Rb/86Sr 87Sr/86Srm 87Sr/86Sri ꢀSr
Sm
Nd
Sm/Nd 147Sm/ 143Nd/ 143Nd/ ꢀNd
144Nd 144Ndm 144Ndi
SKX 01 Green -PhenDi)
SKX 14 Green -PhenDi)
SKX 34 Brown -Ti-augite)
SKX 37B MgDi
SKX 37B Green -PhenDi)
SKX 37B Brown -Ti-augite)
SKX 49 Brown -Ti-augite)
SKX 55 Green -PhenDi)
SKX 63 Brown -Ti-augite)
SKX 70 Brown -Ti-augite)
0.12 158.7
0.37 84.2
0.11 364.6
0.07 91.7
0.13 106.4
0.05 113.6
0.37 125.0
0.70 76.6
2.29 71.7
0.43 282.2
0.0008 0.0022
0.0044 0.0128
0.0003 0.0008
0.0007 0.0021
0.0012 0.0036
0.0005 0.0013
0.0029 0.0085
0.0092 0.0266
0.0320 0.0925
0.0015 0.0044
0.70344
0.70336
0.70343
0.70328
0.70336
0.70339
0.70366
0.70365
0.70402
0.70323
0.70343 ±10.50 2.82 13.03 0.2166 0.1309 0.51255 0.51230 0.87
0.70330 ±12.30 1.17 5.96 0.1960 0.1185 0.51260 0.51236 2.20
0.70343 ±10.52 12.75 63.13 0.2019 0.1221 0.51257 0.51233 1.56
0.70327 ±12.78 1.16 4.76 0.2438 0.1474 0.51268 0.51239 2.73
b
0.70334 ±11.74 n.d. 6.13
0.70339 ±11.09 3.15 13.44 0.2344 0.1417 0.51264 0.51236 2.15
0.51266 0.51240 2.93
0.70362 ±7.76 5.42 20.94 0.2589 0.1565 0.51265 0.51234 1.77
b
0.70354 ±8.89 n.d. 5.21
0.51274 0.51242 3.26
0.70362 ±7.70 5.57 27.80 0.2004 0.1212 0.51257 0.51233 1.51
0.70321 ±13.63 12.43 58.94 0.2109 0.1275 0.51256 0.51231 1.04
a Data extracted from Anthony et al. -1989)
b Data from Dunworth et al. -in preparation)
c Used in Fig. 7
the Cr-diopsides nor the MgDi compositions contain the nearby Aleutian arc -Conrad and Kay 1983; Swan-
signi®cant amounts of the jadeite component, whereas son et al. 1987). Ti contents remain relatively low in
TiO2 contents of ꢀ0.6 wt% are common, and thus many of these examples -0.1±0.5 wt%).
higher than expected for mantle lherzolites. Jadeite-
A graphical comparison of the Skien pyroxene com-
poor, Ti-rich Cr-diopsides appear to be uncommon in positions with those from worldwide mantle xenoliths
mantle material exhumed in alkaline and kimberlitic and cognate phenocryst growth in volcanic magmas is
magmas.
shown in Fig. 8, based on the multivariate statistical
Jadeite-poor MgDi compositions reported in mantle comparisons of Koloskov and Zharinov -1993). This
xenoliths in the literature include diopside found in a shows clearly that the CrMgDi composiions are most
garnet pyroxenite vein from Pipe 200, Lesotho -PH 210, unlikely to be cognate, and in fact bear closest resem-
Carswell et al. 1979), a spinel websterite xenolith from blance to Kamchatka xenoliths. Figure 8 provides ad-
the Hyblean Plateau, Sicily -Nimis and Vannucci 1995), ditional con®rmation that, if the CrMgDi compositions
a spinel garnet lherzolite xenolith in alnoite from Mal- are mantle xenocrysts, they are almost certainly not
aita, Solomon Islands -PH 4069, Nixon and Neal 1987), derived from a typical intraplate lherzolite. This may
a discreet diopside nodule found in the type-locality also be linked with the fact that xenocrystic olivine is so
Alno alnoite -Kreston and Persson 1975), spinel lherz- rarely found in the lavas, a feature hard to explain if the
olite xenoliths from Nunivak, Alaska -Francis 1976, CrMgDi cores have been ripped up from an olivine-rich
1978), and deformed porphyroclastic±adcumulate lherzolite by the host magmas ± such xenocryst incor-
wehrlitic xenoliths from the Adak and Kanga islands in poration is rarely so phase-selective.

715
Fig. 8 Statistical multivariate plot of pyroxene compositions from
Koloskov and Zharinov -1993). P1 ˆ À0:02SiO2 ‡ 0:09TiO2‡
0:64Al2O3 ‡ 0:59Cr2O3 ‡ 1:35FeO ‡ 9:65MnO À 0:5MgO ‡ 0:29
CaO À 2:26Na2O ‡ 8:0;
P2 ˆ À0:93TiO2 ‡ 0:07Al2O3 ‡ 1:23
Cr2O3À 0:46FeO ‡ 1:74MnO ‡ 0:36MgO‡ 0:10CaOÀ 1:66Na2O ‡
Fig. 7 ꢀSr±ꢀNd diagram at 300 Ma for leached pyroxene separates
and whole-rock powders from Skien -cf. Table 4). a Pyroxene and
whole-rock data plotted by sample number. b The same isotope
data as in a plotted accordingto type of pyroxene, or whole-rock.
Pyroxene separates were picked on the basis of colour, assuminga
reasonable correlation between colour and composition as deter-
mined by microprobe analyses from the same samples
8:0. Areas I-VII represent mantle xenolith compositions from
particular tectonic settings.
I Periodotites in kimberlites; II
Eclogites in kimberlites; III Spinel lherzolites -continental alkali
basalts); IV Spinel pyroxenites -continental alkali basalts); V Island
arc peridotites; VI Island arc pyroxenites; VII Plagioclase-bearing
xenoliths. Fields for crystalline pyroxenes -grey shaded area) and
megacrysts -M) in volcanic rocks are also shown, alongwith the
compositions of alpine-like ultrabasic rocks from Kamchatka -K).
Additional data from the Skien CrMgDi and MgDi compositions
-black area) are from the present study. Whereas many of the MgDi
compositions overlap with the grey shaded area, the CrMgDi
compositions do not
Partially melted and/or re-equilibrated mantle xenocrysts
A possible explanation for the low jadeite content of the
CrMgDi compositions is that they are lherzolitic mantle
xenocrysts which have undergone dissolution, either pressures -Shaw et al. 1998; Shaw 1999), although it is
duringmagma ascent to the surface or prior to inclusion uncertain whether this could provide the quantity and
within the host magma. Fractional melting normally range of MgDi compositions and textures seen in the
depletes the original pyroxene composition in Na and Skien lavas, given the fact that these experiments only
Al, enriches it in Ti, and leaves Cr and Mgcontents
virtually unchanged while creating a variety of melting
textures in the process -Shaw and Edgar 1997). Similar
produced grains a few 100 lm in diameter.
chemical and textural e€ects are also seen in a partially Early cognate phenocrysts
melted clinopyroxene in an olivine-clinopyroxene xeno-
lith from the Rhinegraben -Dunworth 1991). Equilibri- Given the general lack of olivine in the Skien lavas, we
um meltingincreases the Mgand Cr contents and Cr/
cannot exclude the possibility that the CrMgDi and/or
-Cr+Al) ratios in the residual pyroxene, while decreas- the MgDi compositions represent the earliest liquidus
ingthe Na/-Na+Ca) ratio -Mysen and Kushiro 1977). phase in these lavas. Similar low-jadeite compositions,
Thus, it is possible that the CrMgDi cores could repre- described as cognate, have been found in an alkali basalt
sent lherzolitic mantle xenocrysts which have undergone from Loihi Volcano, Hawaii -Garcia et al. 1995), ul-
fractional melting.
There appears to be no means of creatingthe MgDi
trama®c volcanics from Kamchatka -Kamenetsky et al.
1995), and shoshonitic lavas from Stromboli -Franca-
compositions as a residue from the CrMgDi cores via lanci 1989). The similarity between the calculated trace-
partial melting. However, MgDi compositions have been element composition of the early cognate pyroxenes and
produced experimentally by disequilibrium meltingof the MgDi compositions, combined with the formation
orthopyroxene in basanitic magmas at a variety of of euhedral, glomeroporphyritic MgDi pyroxenes in

716
sample SKX 07, suggests that the MgDi compositions depleted compositions would necessitate crystallization
may indeed be representative of early cognate growth. If from an unusually depleted mantle melt, the like of which
the MgDi compositions are indeed cognate, then their is not seen in the trace-element-rich, ma®c Skien lavas,
highest Cr contents may be explained by using a higher and thus we provisionally rule out a cognate origin.
DCr for pyroxene/melt -e.g., Duke 1976; Dostal et al. Textural and chemical evidence suggests that the diopside
1983; Francalanci 1989) than the value of 4 obtained at phenocryst and Ti-augite pyroxenes crystallized directly
30 kbar by Hart and Dunn -1993). The CrMgDi com- out of the Skien lavas, although their nonequilibrated
positions appear to be too depleted in trace elements to isotopic signatures suggest that there has been some
be cognate, and Fig. 8 con®rms that they fall outside the magma mixing and/or crustal contamination since the
®eld of known compositions from volcanic lavas crystallization of the diopside compositions. This leaves
-Koloskov and Zharinov 1993), suggesting that a the moderately depleted MgDi compositions which show
xenocrystic origin may be more appropriate for these the greatest textural variations ± glomeroporphyritic
compositions.
clusters of twinned subhedral crystals, strained anhedral
crystals, and resorbed and partially melted cores. We
favour a cognate origin related to the same magmatic
event that gave rise to the Skien lavas, but perhaps related
to early cumulates formed near the MOHO or within the
Geothermobarometry
An inherent problem in attemptingPT estimates of py- lower crust -Olsen et al. 1987) and which were subse-
roxene formation in the Skien lavas is the lack of other quently incorporated into the lavas which reached the
phenocryst phases -e.g., olivine), which thus restricts the surface. Work in progress on MgDi- and CrMgDi-hosted
number and type of geothermometers and barometers melt inclusions should help to resolve this issue in the
available for use. As shown above, the Putirka et al. near future -Kirstein et al., in preparation), and enable us
-1996) geothermobarometer gives ꢀ12 kbar and to derive an overall model consistent with all the petro-
ꢀ1,250 °C for the MgDi pyroxenes from sample SKX logical, thermobarometric, and geophysical data avail-
14, whereas the Soesoo -1997) geothermobarometer able for this part of the Oslo Rift.
gives <3 kbar and 1,115±1,160 °C for the Ti-augite
compositions. The CrMgDi and MgDi compositions fall
outside the calibrated ®eld of Soesoo -1997; see Fig. 3f). Isotopic disequilibrium ± real or artefact?
It should be noted that clinopyroxene crystallized ex-
perimentally as a result of orthopyroxene dissolution Table 4 lists Sr-Nd isotope data for individual samples.
within a basanitic magma under dry disequilibrium At ®rst glance, this provides evidence of intrasample
conditions at 1 GPa -Shaw 1999) would plot almost isotopic disequilibrium -see above). However, it is also
exactly on top of the MgDi and CrMgDi compositions probable that some of the variations are secondary in
in Fig. 3f -data not shown), supporting the Putirka et al. origin. Trace-element whole-rock analyses from the
-1996) PT estimates for the MgDi crystallization con- basalts throughout the Skien sequence show that Rb and
ditions, and con®rmingthat the egothermobarometer
of Soesoo -1997) is not correctly calibrated for these haviour since crystallization -Dunworth, unpublished
compositions. data), due to postmagmatic -hydrothermal) alteration. It
U in particular have been subjected to open-system be-
The MgDi and Ti-augite PT estimate place the ear- is for this reason that both the whole-rock sample
liest, cognate pyroxene crystallization at the MOHO, powders and pyroxenes were heavily leached in hot HCl
and the last stage of crystallization at upper crustal prior to analysis, in order to remove the majority of the
levels. This appears to be geologically reasonable, and low-temperature alteration and to identify the original
matches geophysical evidence for MOHO-depth and isotopic signatures of the rocks. The presence of crustal
deep-crustal cumulates -Neumann 1994, and references -secondary) Sr is evident when the Sr isotope ratios of
therein).
leached and nonleached powders are compared, a dis-
Similar MgDi pyroxene cores are also found in the B1 crepancy that cannot be due only to the radioactive
Vestfold lavas, northeast of Skien -Fig. 1; Dunworth, decay of primary magmatic Rb over the last 300 Ma.
unpublished data). The more fractionated rhomb-por- The leachingprocess, however, tends to remove much of
phyry latite lavas which lie above the Vestfold B1 basalts the original Rb from the whole-rock powders, hence the
were estimated to have formed as the result of frac- leached Rb/Sr concentrations measured by isotope di-
tionation processes at 6.5±10 kbar -Neumann 1994). lution are arti®cially low, yieldingback-corrected initial
This suggests that the earlier, more ma®c lavas, whose 87Sr/86Sr ratios which are too high -Table 4).
groundmass compositions are trachyandesitic, should Some researchers have preferred to leach whole-rock
also have begun to fractionate at this depth, a fact powders to determine the unaltered, measured isotopic
supported by the MgDi ꢀ12 kbar crystallization esti- values, while usingRb-Sr and Sm-Nd concentration
mate given above.
data from nonleached powders to back-correct the lea-
Overall, we favour a xenocrystic source for the Cr- ched isotopic values to their ``initial'' values. The e€ects
diopside cores, although this source is unlikely to be of this procedure on the Skien basalts is seen in Table 4
lherzolitic in composition. If cognate, their extremely in the ``leached/ICP-MS'' lines, where the leached

717
measured 87Sr/86Sr and 143Nd/144Nd signatures have Magma mixing
been back-corrected to initial values usingICP-MS
concentration data on the same nonleached powders. It Evidence for crustal-level magma mixing in the Skien
can be seen that the resultingSr epsilon values are basalts can be demonstrated usingboth the geochemis-
exceptionally low, presumably due to the fact that some try and petrography of the pyroxenes. There are clear
increases in 87Sr/86Sri from core to rim compositions,
of the Rb present in the nonleached powders is of
secondary origin. Overall, these results imply that bulk-
rock isotope data, obtained from leached or nonleached
material from the Skien lavas, should be treated with
caution.
indicatingmixingof a primitive, isotopically depleted
magma with either a more enriched magma or crustal
contaminants -Table 4, Fig. 7). Figure 5 also provides
detailed evidence of magma mixing and the processes
associated with it. The crystal shown in Fig. 5a contains
a small MgDi core originally overgrown, presumably at
low pressures -Fig. 3f), by concentrically and sector-
zoned Ti-augite. The crystal was subsequently subjected
to a sector-selective dissolution event within the magma
chamber. Data on the chemical compositions of sector-
zoned ``hourglass'' clinopyroxenes -Ferguson 1973;
Downes 1974) imply that the [111] sectors -with central
external apices) would have been more Si-Mg-Ti-rich
than the [ab0] sectors which have been selectively melted
in this sample, thus implyingthat the [111] sectors had
higher solidus temperatures, or were closer to equilibri-
um with the host magma. The [ab0] sectors now contain
a honeycomb network of small diopside cores -con-
Since clinopyroxenes tend to have naturally low Rb/
Sr values, little correction is needed to determine their
initial 87Sr/86Sr ratios -e.g., Rosenbaum et al. 1996), and
thus acid-leachingshould preserve the primary isotopic
signature of the clinopyroxene while removing second-
ary alteration -but not genuine pyroxene overgrowth).
Thus, assumingisotopic equilibrium at the time of
crystallization, each generation of pyroxene should
record the isotopic composition of the magma from
which it formed. If magma mixing or crustal contami-
nation of the source magma occurred continuously
duringpyroxene growth, the compositions we observe
would represent an average of these processes. The
di€erent generations of pyroxene thus preserve
snapshots of magmatic evolution within individual ¯ows
which would be missed if only bulk-rock samples were
analyzed.
taining<0.4 wt% Cr O3) which are overgrown with
2
thin ꢀCr-free Ti-augite rims, similar in composition to
the Ti-augite found throughout the [111] sectors -Fig. 5b,
c). It is believed that this meltingand recrystallization
process was due to an in¯ux of hot magma into the
magma chamber. When the newly replenished magma in
the magma chamber cooled suciently, it precipitated a
thin fresh rim of high-Cr -up to 1.0 wt% Cr2O3) Mg -
poor diopside -Fig. 5), followed by an outer rim of ꢀCr-
free Ti-augite. Given the extreme compatibility of Mg
and Cr in clinopyroxene, it is inconceivable that the late-
stage high-Cr Mg-poor rim -shown white in Fig. 5a, b)
could have been precipitated simply by a sudden rapid
cooling event within the host magma, given that the Cr-
rich rim lies outside -and has thus overgrown) an earlier
phase of Ti-augite growth. The most logical reason for
the observed textures and compositions lies in the
magma-mixing process described above. A similar ar-
gument can be put forward to explain the thin -white)
diopside overgrowth in Fig. 2a that lies between two
generations of Ti-augite precipitation.
Isotopic signatures
Many of the pyroxene trace-element variations discussed
above are re¯ected in the Sr-Nd isotope data. The
``green'' pyroxenes in Fig. 7b represent PhenDi compo-
sitions sensu stricto -see Fig. 3) but, given the limitations
of mineral pickingon the basis of colour, they may also
contain some MgDi component. These phenocryst
compositions are thought to represent the geochemical
conditions prevailingin crustal and/or subcrustal mag-
ma chambers and, on average, they have slightly higher
ꢀNd and lower ꢀSr values than the later generation of
Ti-augite growth. The particular similarity between the
Ti-augite and whole-rock compositions -particularly in
ꢀNd space) documented for many of the samples sug-
gests that the compositions of the Ti-augites represent
the conditions prevailing during late-stage magma
crystallization. The changes in isotopic composition
between the early- and late-stage pyroxenes provide
good evidence for magma mixing and/or crustal
contamination duringmagma evolution. More analyses
-includingthose for Pb isotopic composition) are
currently in progress to test these assumptions. Our new
isotope data for the Skien samples agree well with those
obtained in previous studies -Neumann et al. 1988;
Anthony et al. 1989), and con®rm that the Skien lavas
contain contributions derived from at least one mantle
source with low epsilon Sr and Nd values that was not
involved in the production of the rest of the Oslo Rift
lavas -Dunworth et al. 1999).
Further work, involvingthe collection of a laregr
whole-rock and mineral-separate isotope and trace-ele-
ment dataset currently underway, will aim to further
characterize and begin to quantify the nature of these
mixingand contamination processes.
Magma-chamber processes
We propose the followingmodel of magma evolution to
describe the processes operatingat Skien about 300 Ma
ago. The parental magmas were derived from a sub-
lithospheric source which was enriched in trace elements,
and had low but potentially variable ꢀSr and ꢀNd values.

718
The contribution of a mantle component similar to the in Ti-augite at low pressures. Nb, Sr, the LREE, and the
modern-day HIMU reservoir, as suggested by Anthony LILE are incompatible under these conditions.
et al. -1989), is still probable but its source location
8. Irrefutable evidence of magma mixing and possible
cannot be con®rmed at this time. Partial meltingof the crustal contamination have been documented by the
sublithospheric source ultimately produced the ®rst, and geochemical and textural characteristics of di€erent
most extensive B1-basalt sequence within the Oslo Rift, generations of pyroxene growth.
at Skien -Neumann et al. 1992). Between the initiation of
sublithospheric meltingand the eruption of the magmas
at the surface, the magmas assimilated material from the
mantle lithosphere, and fractionated within crustal or
subcrustal magma chambers. It is assumed that any
mantle material assimilated into the lavas was poor in
olivine, and perhaps of garnet-pyroxenite composition
-see above). Any garnet and orthopyroxene would
dissolve in an alkali basaltic or basanitic melt -Eggler
1974; Brearly and Scarfe 1986; Shaw et al. 1998;
Shaw 1999).
We propose that there were periods of melt in¯ux
from the mantle to the crustal plumbingsystem, perhaps
incorporating-CrMgDi) or crystallizing-MgDi and/or
PhenDi) a fresh set of diopsidic pyroxenes alongthe
way, some of which made their way relatively rapidly
towards the surface, and some of which underwent
dissolution and Ti-augite overgrowth processes within
the magma chamber. Towards the top of the main
Skien sequence, the lavas generally become more frac-
tionated and pyroxene-plagioclase-phyric, indicating
longer periods between re®lling of the magma cham-
ber, and prolonged periods of magma-chamber
fractionation.
The intrasample isotopic disequilibrium ®rst docu-
mented by Anthony et al. -1989) has been con®rmed
and extended by this study. We have quanti®ed and
described the di€erent pyroxene types extant in the
formation, and have correlated type with major- and
trace-element as well as Sr-Nd isotopic data. Work in
progress will extend this disequilibrium dataset across
Skien and the more extensive Vestfold and Jeloya
pyroxene-phyric lavas further north, and enable more
extensive constraints to be placed on the various mantle
sources of these lavas and their di€erentiation histories.
Acknowledgements The authors would like to thank Jon Blundy
and Cli€ Shaw for thought-provoking reviews, Toril Enger for
patient assistance in the chemistry lab, Tom Andersen and Bjùrn
Sundvoll for tamingthe mass spectrometers, Ha kon Austrheim and
Muriel Erambert for troubleshootingon the microprobe, Linda
Kirstein for cheerful assistance and comments on earlier drafts of
this manuscript, and John Craven for his technical wizardry with
the ion microprobe. Partial support for the ion microprobe time
came via NERC grant IMP/123/1097 to JMR. This work has been
carried out as part of an EU-funded TMR network ± Permo-
Carboniferous Riftingin Europe -project reference ERB-
FMRXCT960093).
References
Concluding remarks
Anthony EY, Segalstad TV, Neumann E-R -1989) An unusual
mantle source region for nephelinites from the Oslo Rift,
Norway. Geochim Cosmochim Acta 53:1067±1076
Blundy JD, Robinson JC, Wood BJ -1998) Heavy REE are com-
patible in clinopyroxene on the spinel lherzolite solidus. Earth
Planet Sci Lett 160:493±504
Brearly M, Scarfe CM -1986) Dissolution rates of upper mantle
minerals in an alkali basalt melt at high pressure: an experi-
mental study and implications for ultrama®c xenolith survival.
J Petrol 27:1157±1182
1. The Skien lavas contain multiple generations of
clinopyroxene, ranging in composition from Mg-rich
Cr-diopsides to Ti-augite.
2. The early generations of pyroxene have been de-
rived from a variety of sources, and are not in geo-
chemical or isotopic equilibrium with each other or with
their host melts.
Carswell DA, Clarke DB, Mitchell RH -1979) The petrology and
geochemistry of ultrama®c nodules from Pipe 200, Northern
Lesotho. In: Boyd FR, Meyer HOA -eds) The mantle sample:
inclusions in kimberlites and other volcanics. Proc 2nd Int
Kimberlite Conference, Vol 2. AGU, Washington, DC, pp 127±
144
Conrad WK, Kay RW -1983) Ultrama®c and ma®c inclusions from
Adak Island: crystallization history, and implications for the
nature of primary magmas and crustal evolution in the Aleutian
Arc. J Petrol 25:88±125
Dostal J, Dupuy C, Carron JP, Le Guen de Kerneizon M, Maury
RC -1983) Partition coecients of trace elements: application to
volcanic rocks of St. Vincent, West Indies. Geochim Cosmo-
chim Acta 47:525±533
Downes MJ -1974) Sector and oscillatory zoningin calcic augites
from M. Etna, Sicily. Contrib Mineral Petrol 47:187±196
Duda A, Schmincke H-U -1985) Polybaric di€erentiation of alkali
basaltic magmas: evidence from green-core clinopyroxenes
-Eifel, FRG). Contrib Mineral Petrol 91:340±353
3. Low-jadeite, Cr-diopside cores found in ®ve sam-
ples, which exhibit a depleted trace-element signature,
are believed to represent lithospheric mantle xenocrysts.
4. MgDi compositions are believed to represent very
early cognate growth, or cumulate material preserved at
deep-crustal levels which has been incorporated into
later lavas.
5. The MgDi and PhenDi pyroxenes preserve an
isotopic signature consistent with derivation from a
mantle reservoir with an isotopic signature of approxi-
mately ꢀSr±13 and ꢀNd+3.
6. Textural characteristics, isotopic and trace-element
signatures show that the Ti-augite compositions were the
last to crystallize, and that they crystallized in equilib-
rium with their host melts.
7. Calculated partition coecients, based on pyrox-
ene and whole-rock data, show that the middle and
heavy HREE, alongwith Zr, Hf, and Y, are compatible
Duke JM -1976) Distribution of the period four transition elements
amongolivine, calcic clinopyroxene and ma®c silicate liquid:
experimental results. J Petrol 17:499±521

719
Dunworth EA -1991) A geochemical and petrogenic study of oli- Neumann E-R, Wul€-Pedersen E, Simonsen SL, Pearson NJ,
vine melilitites from the Tertiary-Quaternary Volcanic Province
of Southern Germany. MSc Thesis, University of Leeds
Dunworth EA, Bell K -2001) The Turiy Massif, Kola Peninsula,
Martõ J, Mitjavila J -1999) Evidence for fractional crystalliza-
tion of periodically re®lled magma chambers in Tenerife,
Canary Islands. J Petrol 40:1089±1123
Russia: Isotopic and geochemical evidence for multi-source Nimis P, Vannucci R -1995) An ion microprobe study of clinopy-
evolution. J Petrol -in press)
roxenes in websteritic and megacrystic xenoliths from Hyblean
Plateau -SE Sicily, Italy): constraints on HFSE/REE/Sr frac-
tionation at mantle depth. Chem Geol 124:185±197
Dunworth EA, Wilson M -1998) Olivine melilitites of the SW
German Tertiary Volcanic Province: mineralogy and petro-
genesis. J Petrol 39:1805±1836
Nixon P -ed) -1987) Mantle xenoliths. Wiley, New York
Dunworth EA, Neumann E-R, Rosenbaum JM -1999) Mantle Nixon PH, Neal CR -1987) OntongJava Plateau: deep-seated xe-
reservoirs, xenocrysts and magmatic disequilibrium in the Skien
basalts, Oslo Rift. Abstr Vol Eur Union Geosci Conf No 10,
Strasbourg, 28 March±1 April 1999, pp 291
noliths from thick oceanic lithosphere. In: Nixon P -ed) Mantle
xenoliths. Wiley & Sons, New York, pp 335±345
Olsen KH, Baldridge WS, Larsen BT, Neumann ER, Ramberg IB
-1987) A lithospheric transect across the Oslo palaeorift. EOS
68:1480
Eggler DH -1974) E€ects of CO2 on the meltingof peridotite.
Carnegie Inst Wash Yearb 73:215±224
Ferguson AK -1973) On hour-glass zoning in clinopyroxene. Pearce NJG, Perkins WT, Westgate JA, Gorton MP, Jackson SE,
Mineral Mag39:321±325
Neal CR, Chenery SP -1997) A compilation of new and pub-
lished major and trace element data for NIST SRM 610 and
NIST SRM 612 glass reference materials. Geostds Newslett
21:115±144
Francalanci L -1989) Trace element partition coecients for min-
erals in shoshonitic and calc-alkaline rocks from Stromboli
Island -Aeolian Arc). Neues Jahrb Mineral Abh 160:229±247
Francis DM -1976) The origin of amphibole in lherzolite xenoliths Putirka K, Johnson M, Kinzler R, Longhi J, Walker D -1996)
from Nunivak Island, Alaska. J Petrol 17:357±378
Francis DM -1978) The implications of the compositional depen-
dence of texture in spinel lherzolite xenoliths. J Geol 86:473±
485
Thermobarometry of ma®c igneous rocks based on clinopy-
roxene-liquid equilibria, 0±30 kbar. Contrib Mineral Petrol
123:92±108
RambergIB -1976) Gravity interpretation of the Oslo Graben and
associated igneous rocks. Norges Geol Unders 325:1±194
Frey FA, Prinz M -1978) Ultrama®c inclusions from San Carlos,
Arizona: Petrologic and geochemical data bearing on their RambergIB, Larsen BT -1978) Tectonomamgatic evolution.
petrogenesis. Earth Planet Sci Lett 38:129±176 Norges Geol Unders 337:55±73
Garcia MO, Foss DJP, West HB, Mahoney JJ -1995) Geochemical Ro HE, Stuevold LM, Faleide JI, Myhre AM -1990) Skagerrak
and isotopic evolution of Loihi volcano, Hawaii. J Petrol Graben ± the o€shore continuation of the Oslo Graben. Tec-
36:1647±1674 tonophysics 178:1±10
Grin WL -1973) Lherzolite nodules from the Fen alkaline com- Rocholl A, LudwigT, Altherr R, Meyer HP, Brey G, Velz S, Seck
plex, Norway. Contrib Mineral Petrol 38:135±146
Hart SR, Dunn T -1993) Experimental cpx/melt partitioningof
24 trace elements. Contrib Mineral Petrol 113:1±8
HA, Bulatov V -1996) Experimental partitioningof trace ele-
ments between clinopyroxene, garnet and basanitic melts
studied by ion microprobe. J Conf Abs 1:555
Heeremans M, Larsen BT, Stel H -1996) Paleostress reconstruction Roeder PL, Emslie RF -1970) Olivine-liquid equilibrium. Contrib
from kinematic indicators in the Oslo Graben, southern Nor- Mineral Petrol 29:275±289
way: new constraints on the mode of rifting. Tectonophysics Rosenbaum JM, Zindler A, Rubenstone JL -1996) Mantle ¯uids:
266:55±79
Evidence from ¯uid inclusions. Geochim Cosmochim Acta
60:3229±3252
surements of National-Institute-of-Standards-and-Technology Schou-Jensen E, Neumann E-R -1988) Volcanic rocks on Jeloya,
Hinton RW, Harte B, Witteickschen G -1995) Ion probe mea-
standard reference material SRM-610 glass, trace-elements. central Oslofjord: The ma®c lavas. Norges Geol Tidsskr
Analyst 120:1315±1319 68:289±308
Kamenetsky VS, Sobolev AV, Joron J-L, Semet P -1995) Petrology Segalstad TV -1979) Petrology of the Skien basaltic rocks, south-
and geochemistry of Cretaceous ultrama®c volcanics from
Eastern Kamchatka. J Petrol 36:637±662
Koloskov AV, Zharinov SE -1993) Multivariate statistical analysis
of clinopyroxene compositions from ma®c and ultrama®c
xenoliths in volcanic rocks. J Petrol 34:173±185
western Oslo Region, Norway. Lithos 12:221±239
Shaw CSJ -1999) Dissolution of orthopyroxene in basanitic magma
between 0.4 and 2 GPs: further implications for the origin of
Si-rich alkaline glass inclusions in mantle xenoliths. Contrib
Mineral Petrol 135:114±132
Kresten P, Persson L -1975) Discrete diopside in alnoite from Alno Shaw CSJ, Edgar AD -1997) Post-entrainment mineral-melt reac-
Island. Lithos 8:187±192
tions in spinel peridotite xenoliths from inver, Donegal, Ireland.
Geol Mag134:771±779
Mearns E -1986) Sm-Nd ages for Norwegian garnet peridotites.
Lithos 19:269±278
McDonough WF, Sun S-S -1995) The composition of the Earth.
Chem Geol 120:223±253
Mysen BO, Kushiro I -1977) Compositional variations of coexis-
tingphases with degree of meltingof peridotite in the upper
mantle. Am Mineral 62:843±865
Shaw CSJ, Thibault Y, Edgar AD, Lloyd FE -1998) Mechanisms
of orthopyroxene dissolution in silica-undersaturated melts at
1 atmosphere and implications for the origin of silica-rich glass
in mantle xenoliths. Contrib Mineral Petrol 132:354±370
Skulski T, Minarik W, Watson EB -1994) High-pressure experi-
mental trace-element partitioningbetween clinopyroxene and
basaltic melts. Chem Geol 117:127±147
Neumann E-R -1986) Mass estimates of cumulates and residues
J
after anatexis in the Oslo Graben.
91-B11):11629±11640
Geophys Res Soesoo A -1997) A multivariate statistical analysis of clinopyroxene
composition: empirical coordinates for the crystallisation
Neumann E-R -1994) The Oslo Rift: P-T relations and lithospheric
structure. Tectonophysics 240:159±172
Neumann E-R, Tilton GR, Tuen E -1988) Sr, Nd and Pb isotope
PT-estimations. Geol Foren Stockholms Forhand 119:55±60
Sundvoll B, Larsen BT -1994) Architecture and early evolution of
the Oslo Rift. Tectonophysics 240:173±189
geochemistry of the Oslo rift igneous province, southeast Sundvoll B, Neumann E-R, Larsen BT, Tuen E -1990) Age
Norway. Geochim Cosmochim Acta 52:1997±2007
Neumann E-R, Sundvoll B, Overli PE -1990) A mildly depleted up-
relations amongOslo Rift mamg atic rocks: implications for
tectonic and magmatic modelling. Tectonophysics 178:67±87
per mantle beneath southeast Norway: evidence from basalts in Swanson SE, Kay SM, Brearly M, Scarfe CM -1987) Arc and
the Permo-Carboniferous Oslo Rift. Tectonophysics 178:89±107
Neumann E-R, Olsen KH, Baldridge WS, Sundvoll B -1992) The
Oslo Rift: a review. Tectonophysics 208:1±18
back-arc xenoliths in Kurile-Kamchatka and western Alaska.
In: Nixon PH -ed) Mantle xenoliths. Wiley, New York,
pp 302±318