Lithos 118 (2010) 156–168

Contents lists available at ScienceDirect

Lithos
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / l i t h o s

Trace element chemistry of mineral inclusions in eclogitic diamonds from the

Premier (Cullinan) and Finsch kimberlites, South Africa: Implications for the
evolution of their mantle source
Fanus Viljoen a,b,⁎, René Dobbe c, Jeff Harris d, Braam Smit e
a
Department of Geology, University of Johannesburg, P.O. Box 524, Auckland Park, 2006, South Africa
b
De Beers Geoscience Centre, PO Box 82232, Southdale, 2135, South Africa
c
FEI Electron Optics B.V., PO Box 80066, 5600 KA Eindhoven, The Netherlands
d
Department of Geographical and Earth Sciences, University of Glasgow, Glasgow G12 8QQ, United Kingdom
e
Anglo Research, 8 Schonland Street, Theta, Johannesburg, South Africa

a r t i c l e i n f o a b s t r a c t

Article history: Although diamonds of eclogitic paragenesis are commonly encountered in the productions of many
Received 18 November 2009 Southern Africa kimberlites, the nature and evolution of the protolith to eclogitic diamonds are still poorly
Accepted 19 April 2010 understood. There is some evidence that these protoliths (and possibly also the diamonds) may be related
Available online 28 April 2010
to subduction of oceanic crust, although this is not a universally accepted view. In order to further
investigate the protolith/diamond relationship, garnets and (in some cases) clinopyroxene inclusions in
Keywords:
Diamond
23 diamonds from Premier mine and 16 diamonds from Finsch were analysed for their trace element
Kimberlite composition.
Eclogite From both mines a strong correlation between the garnet Ca content and the chondrite-normalised rare
Lithosphere earth element (REE) pattern is evident. Garnets with comparatively low Ca content are characterised by
Trace elements REE patterns which show a steady increase in abundance from light rare earths (LREE) to heavy rare
earths (HREE). With increasing Ca content in garnet, the abundance of LREE (La, Ce, Pr, and Nd) as well as
the middle rare earths (MREE; Sm, Eu, Gd, and Tb) progressively increases, ultimately giving the trace
element pattern a distinct ‘humped’ appearance.
Bulk-rock trace element abundance patterns have been reconstructed from measured trace element
contents in garnet as well as calculated trace element concentrations in clinopyroxene, based on known
clinopyroxene–garnet partition coefficients (Harte and Kirkley, 1997). At both Premier and Finsch, the
low-Ca group samples (2.6 to 5.0 wt.% CaO in garnet) are LREE depleted, and have relatively flat calculated
bulk-rock trace element abundance patterns at approximately 10 times chondrite concentrations, but
with marked positive Sr and negative Zr anomalies. The intermediate-Ca group samples (5.2 to ∼9 wt.%
CaO in garnet) are LREE depleted, show Sr and Zr anomalies, have somewhat higher concentrations of Zr
and MREE, and have HREE contents that overlap with the low-Ca group (Fig. 6). High-Ca group samples
(∼ 9 to 14.8 wt.% CaO in garnet) are LREE depleted, show Sr and Zr anomalies, are MREE-enriched, and
have HREE contents that are slightly less than the low- and intermediate-Ca group samples.
Based on both the calculated bulk eclogite trace element abundances and their patterns, as well as
previously published radiogenic isotope data, our preferred model of protolith evolution for the eclogitic
diamonds from Premier and Finsch is one in which both the major and trace element chemistry of the
inclusions are ultimately inherited from low-pressure oceanic protoliths, consisting of varying mixtures of
oceanic basalt + cumulate gabbro for diamonds from both Premier and Finsch. Of particular importance in
the current data are the presence of marked negative Zr anomalies, marked positive Sr anomalies, and a
general absence of Eu anomalies in all compositional groupings. The Zr anomaly can arise in reconstructed
bulk eclogite trace element abundance patterns if rutile is not included in the calculations, but the Sr
anomalies (coupled with an absence of Eu anomalies) can only be explained through the mixing of oceanic
gabbro and mid-ocean ridge basalt. The averaged eclogite bulk trace element compositions for Premier
and Finsch are also markedly similar to that of clinopyroxene in a typical cumulate gabbro, and a role for
cumulate clinopyroxene in protolith evolution may therefore also be inferred.

⁎ Corresponding author. Department of Geology, University of Johannesburg, P.O. Box 524, Auckland Park, 2006, South Africa. Tel.: + 27 11 5594710; fax: + 27 11 5594702.
E-mail address: fanusv@uj.ac.za (F. Viljoen).

0024-4937/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.lithos.2010.04.011
 F. Viljoen et al. / Lithos 118 (2010) 156–168
It is likely that prior to and during diamond crystallisation, the major and particularly the trace element
compositions of the high-pressure eclogite source rock to these diamonds may have been slightly
modified by metasomatic fluids and melts. However large-scale fluid- or melt-related metasomatic
processes are not indicated.
1. Introduction
Due to the chemically inert nature of diamonds, minerals
occurring as inclusions record conditions of protolith evolution and
diamond crystallisation up to the time of encapsulation (Gurney,
1991), being thereafter isolated from any later metasomatic modifi-
cation of the high-pressure diamond host rock by transiting melts,
fluids, and the carrier kimberlite (Ireland et al., 1994). This situation
markedly contrasts with xenoliths (including diamondiferous ones),
of peridotite and eclogite found in kimberlite, which may have been
extensively and repeatedly modified in the mantle until sampled by
the host kimberlite. Hence inclusions represent pristine micro-
samples of the deep lithosphere.
The two dominant parageneses of mineral inclusions are termed
peridotitic and eclogitic (Stachel and Harris, 2008). Peridotitic
inclusion assemblages typically consist of olivine, chromite, orthopyr-
oxene and purple (chrome pyrope) garnet whereas eclogitic assem-
blages predominantly comprise orange pyrope–almandine garnet and
pale green omphacitic clinopyroxene. A websteritic paragenesis and
an ultrahigh-pressure sub-lithospheric paragenesis are also recog-
nised, but these diamonds are rare at most localities (Stachel and
Harris, 2008).
Diamonds of eclogitic paragenesis vary widely in abundance in the
productions from kimberlite mines in Southern Africa. For example, in
diamonds of about 2 mm in maximum dimension they range from
b5% at Finsch (Phillips et al., 2004) to N60% in the case of Orapa in
Botswana (Gurney et al., 1984). The abundance at Premier in this size
range is 28% and for both Premier and Finsch the proportion of
eclogitic diamonds steadily increases with size so that with ∼4 mm
diamonds at Premier abundance is 72% whilst at Finsch for ∼3 mm
diamonds the abundance is 15.9% (Stachel and Harris, 2008).
Comparison of the mineral chemistries of diamond-bearing
eclogites with that of eclogitic minerals occurring as inclusions in
diamonds, clearly indicates a genetic relationship (Gurney, 1991).
The petrogenesis of cratonic eclogite xenoliths remains controver-
sial with a crustal origin (the subduction hypothesis) as well as a
mantle origin (ponding of asthenosphere-derived melts) being
proposed by various authors (see Jacob, 2004 and references
therein). Although not a universally accepted view (Griffin and
O'Reilly, 2007), the majority of eclogites from the subcratonic
lithosphere are probably best explained as samples of subducted
oceanic crust (Jacob, 2004). This view is also held by those who
have studied eclogitic inclusions in diamond (Stachel and Harris,
2008; references therein). Within this framework, Ireland et al.
(1994) suggested that some eclogitic inclusions in diamond may
derive from an eclogite protolith consisting of Archean basalt
depleted by the extraction of tonalitic–trondhjemitic melts at high
pressure during subduction, a hypothesis considered likely in the
case of eclogitic inclusion-bearing diamonds from Mwadui (Stachel
et al., 1999) and Venetia (Aulbach et al., 2002).
The proposed models for the nature and timing of diamond
crystallisation in the eclogite host rock, range from the syngenetic
crystallisation of inclusions and the diamond host (Navon, 1999) to
models advocating diamond formation from subducted organic
matter through the direct conversion of graphite to diamond (Tappert
et al., 2005a,b), or through a metasomatic origin of diamond in which
protolith formation and diamond crystallisation may be completely
separate events (Cartigny et al., 1998a,b; Stachel and Harris, 2008;
references therein).
157
© 2010 Elsevier B.V. All rights reserved.
The trace element geochemistry of eclogitic garnet and clinopy-
roxene inclusions in ‘normal’ lithospheric diamonds (as opposed to
eclogitic deep sublithospheric diamonds) have been studied at many
localities world-wide. For information from a) Udachnaya, Mir and
Aikhal in Siberia see Ireland et al. (1994); Taylor and Anand (2004); b)
Jagersfontein, Monastery and Venetia in South Africa see Moore et al.
(1991); Aulbach et al. (2002); Tappert et al. (2005a,b); c) Mwadui in
Tanzania see Stachel et al. (1999); d) KanKan in Guinea see Stachel et
al. (2000); e) DO27, Snap Lake, A154 South, Buffalo Hills and Jericho in
Canada see Davies et al. (2004a,b); Promprated et al. (2004); Donnelly
et al. (2007); De Stefano et al. (2009); f) Guaniamo in Venezuela see
Kaminsky et al. (2000); g) and placer deposits in Namibia and Brazil
see Stachel et al. (2004) and Tappert et al. (2006), respectively. In
most cases these studies comprise up to five eclogitic diamonds per
locality, but at Guaniamo, Jericho, Jagersfontein, and Namibia more
substantial sample suites were examined.
Comprehensive geochemical studies of diamond-bearing eclogite
xenoliths in kimberlites are not numerous, but have been completed
for Mir and Udachnaya in Siberia (Jacob et al., 1994; Snyder et al.,
1997), Kaalvallei and Newlands in South Africa (Menzies et al., 2003;
Viljoen et al., 2005), and Jericho in Canada (Smart et al., 2009).
The purpose of the present paper, therefore, is to report on an
investigation into the nature and genesis of the protolith to eclogitic
diamonds, through the study of compositional trends amongst a large
and chemically diverse suite of eclogitic minerals occurring as
inclusions in diamonds from the Premier and Finsch kimberlites in
South Africa. For this study, 22 garnets and 4 clinopyroxenes in 23
eclogitic diamonds from the Premier mine and 15 garnets and 4
clinopyroxene inclusions in 16 eclogitic diamonds from the Finsch
diamond mine, were analysed for their major and trace element
composition (Tables 1–3).
2. Samples and methods
The Premier kimberlite (Field et al., 2008) is of the Group I variety
(Smith, 1983) and has a preferred emplacement age of 1180 Ma ± 30
Ma (Richardson et al., 1993). From Sm–Nd studies on the garnets and
clinopyroxenes of the lherzolitic component of the peridotitic
inclusion suite, a genesis isochron age of 1930 ± 60 Ma was obtained,
but with the harzburgitic component Sm/Nd and 143Nd/144Nd ratios
were not sufficiently different from each other to yield either a genesis
isochron age, or a meaningful model age (Richardson et al., 1993).
Similar inclusion studies on garnets and clinopyroxenes of the
eclogitic paragenesis yielded a genesis isochron age of 1150 ±
60 Ma. Considering the errors of the analytical methods (see above),
the eclogite isochron age allows the crystallisation of eclogitic
diamonds at Premier to be contemporaneous with the age of pipe
emplacement (Richardson, 1986). However the level of nitrogen
aggregation within the Type I (nitrogen-bearing) eclogitic diamond
population clearly indicates a residence time within the upper mantle
(Richardson, 1986) making these diamond xenocrysts within this
kimberlite, as is the case world-wide.
The Finsch kimberlite (Field et al., 2008) is of the Group II variety
(Smith, 1983), with an estimated Rb–Sr pipe emplacement age of
118 ± 3 Ma (Smith et al., 1985). As previously noted, the inclusion
suite of small diamonds at Finsch is dominated by inclusions of the
peridotitic (harzburgitic) paragenesis (Harris and Gurney, 1979)
from which an Archean Sm–Nd model age of 3.3 ± 0.1 Ga was
obtained (Richardson et al., 1984). Similar studies on eclogitic
158 F. Viljoen et al. / Lithos 118 (2010) 156–168

Table 1
Electron microprobe analyses of eclogitic garnet and clinopyroxene in diamonds from Premier.

Sample AP85 AP86 AP90 AP91 AP92 AP93 AP94 AP96 AP97 AP98 AP100 AP101A AP101B

Mineral GNT GNT GNT GNT GNT GNT GNT GNT GNT GNT GNT GNT CPX

wt.%
SiO2 39.13 40.13 40.42 38.48 40.24 41.84 40.89 39.73 39.52 40.47 40.61 41.31 55.03
TiO2 0.67 0.36 0.37 0.82 0.73 0.37 0.41 0.90 0.38 0.41 0.36 0.46 0.53
Al2O3 22.20 22.17 21.93 20.19 22.53 22.29 22.47 22.01 22.82 22.27 22.37 22.69 8.29
Cr2O3 ND 0.17 0.15 0.03 0.05 0.51 0.16 0.07 0.17 0.13 0.09 0.15 0.19
FeO 14.91 16.61 16.34 22.41 16.01 14.13 16.60 18.33 15.45 16.30 16.71 15.47 7.47
MnO 0.28 0.43 0.41 0.47 0.28 0.28 0.45 0.33 0.34 0.34 0.41 0.31 0.14
NiO ND ND ND ND ND ND ND ND ND ND ND ND 0.05
MgO 7.74 15.17 15.37 8.70 11.31 17.43 15.17 10.02 17.44 15.48 14.83 16.71 13.62
CaO 14.81 4.08 4.15 7.73 9.38 3.38 4.12 8.56 2.96 3.55 4.39 3.06 8.85
Na2O 0.26 0.11 0.09 0.24 0.20 0.09 0.12 0.20 0.17 0.14 0.11 0.15 4.81
K2O ND ND ND ND ND ND ND ND ND ND ND ND 0.03
Total 100.03 99.24 99.23 99.09 100.74 100.34 100.39 100.16 99.27 99.12 99.89 100.35 99.03

GNT = garnet; CPX = clinopyroxene.

ND = not detected.
AP102A AP102B AP104 AP106A AP106B AP108A AP108B AP109 AP112 AP113 AP114 AP117 AP125 AP127

GNT CPX GNT GNT CPX GNT CPX GNT GNT GNT GNT GNT GNT GNT

wt.%
inclusions show a significantly younger diamond population of
1580 ± 50 Ma (Richardson et al., 1990). Sm–Nd model ages for
eclogitic garnets in large single diamonds ranged from 1443 ± 166
to 1657 ± 77 Ma (Smith et al., 1991).
The eclogitic diamonds from Premier are a new collection assembled
for the present study. The Finsch inclusions derive from the study of
Appleyard et al. (2004). Based on the chemical composition of the
inclusions from both sources (see Fig. 1) the eclogitic diamond suites
are believed to be fully representative of the two mines. The inclusions
were broken out through crushing the host, then mounted in epoxy
resin contained in brass stubs, polished and analysed for major elements
on a Cameca SX-50 electron microprobe at the De Beers GeoScience
Centre in Johannesburg. Methods and detection limits are as described
in Appleyard et al. (2004) and Viljoen et al. (2009) for samples from
Premier and Finsch respectively. Each analysis reported represents the
average of three spot analyses per inclusion.
The trace element analyses were conducted on the Cameca ims-4f
ion microprobe facility at the University of Edinburgh. Prior to
analysis all samples were carefully cleaned before coating with gold.
Analytical procedures and precision are as described in Harris et al.
(2004). In all cases rare earth elements (REE), large-ion lithophile
elements (LILE; Sr and Ba) and high field-strength elements (HFSE;
Nb, Ti, Zr, Hf, and Y) were analysed. Analyses were calibrated using
synthetic glass standard NBS-610 (which is doped with 500 ppm of
the trace elements concerned). In order to monitor analytical quality,
garnet standards Dutsen Dushowa (Irving and Frey, 1978) and GHR1
(Campbell et al., 1996; Viljoen et al., 2009) were used. Again reported
data are the average of two or three analyses per inclusion. Accuracy
and precision are typically better than 20% and in most cases better
than 10% (see Table 2). However in cases where analyses were
conducted for concentrations at or close to the detection limit (0.01–
0.005 ppm), accuracy and precision are typically lower and may be as

Table 2
Trace element compositions (ppm) of eclogitic garnet and clinopyroxene inclusions in diamonds from Premier, and garnet standards Dutsen Dushowa as well as GHR1.

Sample AP85 AP86 AP90 AP91 AP92 AP93 AP94 AP96 AP97 AP98 AP100 AP101A AP101B AP102A AP102B AP106A

Mineral GNT GNT GNT GNT GNT GNT GNT GNT GNT GNT GNT GNT CPX GNT CPX GNT

ppm
Ti 3440 1730 1810 4090 3830 1810 1900 4440 1780 1900 1640 2120 2500 3910 3070 3920
Sr 12.5 0.45 0.44 2.32 2.06 0.72 0.46 1.91 0.51 0.87 0.31 0.35 50.1 0.69 93.2 1.02
Y 20.9 8.15 8.86 53.9 33.4 7.70 8.06 32.4 18.9 39.2 7.40 21.0 4.25 36.8 6.48 31.8
Zr 43.1 8.92 9.48 26.7 27.9 15.3 9.61 24.6 13.6 20.5 8.23 20.0 21.2 20.7 17.7 10.4
Nb 0.70 0.09 0.11 0.38 0.30 0.28 0.08 0.06 0.08 0.12 0.08 0.08 0.23 0.08 1.21 0.49
Ba 0.01 ND ND ND ND ND ND ND 0.02 0.01 ND 0.02 0.07 ND 2.27 ND
La 0.34 0.02 0.01 0.11 0.06 0.02 0.01 0.03 0.01 ND ND ND 0.30 0.01 0.64 ND
Ce 2.55 0.14 0.09 0.85 0.47 0.16 0.08 0.30 0.14 0.07 0.06 0.05 1.15 0.13 2.26 0.12
Pr 0.85 0.04 0.03 0.25 0.17 0.05 0.03 0.12 0.04 0.03 0.02 0.02 0.22 0.05 0.45 0.10
Nd 8.15 0.41 0.37 3.07 1.90 0.48 0.33 1.26 0.42 0.38 0.25 0.27 1.57 0.81 3.01 1.82
Sm 4.05 0.28 0.26 2.23 1.53 0.30 0.20 1.22 0.35 0.51 0.22 0.24 1.11 0.97 1.68 1.76
Eu 2.78 0.14 0.16 1.02 0.82 0.18 0.14 0.78 0.22 0.34 0.11 0.14 0.26 0.41 0.44 0.77
Gd 5.65 0.58 0.57 4.84 3.53 0.67 0.49 2.91 0.74 1.80 0.45 0.89 0.62 2.07 1.54 3.45
Tb 1.01 0.14 0.16 1.16 0.84 0.16 0.14 0.78 0.25 0.59 0.12 0.25 0.14 0.61 0.27 0.77
Dy 5.88 1.35 1.38 9.26 6.51 1.16 1.24 6.16 2.71 6.00 1.19 2.88 1.03 5.41 1.64 5.74
Ho 0.97 0.35 0.40 2.14 1.48 0.34 0.35 1.56 0.79 1.63 0.33 0.89 0.21 1.43 0.30 1.35
Er 2.33 1.36 1.57 6.35 4.66 1.35 1.20 4.35 3.31 5.96 1.30 3.55 0.56 5.40 0.69 4.15
Tm 0.34 0.29 0.28 1.07 0.71 0.20 0.24 0.72 0.64 1.05 0.26 0.67 0.08 0.89 0.10 0.71
Yb 1.88 2.04 2.25 6.46 4.56 1.41 2.02 4.60 5.11 7.32 2.08 5.24 0.48 6.53 0.78 4.43
Lu 0.24 0.34 0.38 1.08 0.65 0.27 0.37 0.78 0.97 1.21 0.35 0.91 0.08 1.15 0.08 0.64
Hf 1.13 0.31 0.32 1.09 1.00 0.40 0.31 0.76 0.44 0.70 0.23 0.63 1.19 0.81 0.72 0.83

ND = not detected (b 0.005 ppm); DD(IMP) = average of 12 ion microprobe analyses of Dutsen Dushowa garnet standard; DD(SDEV) = standard deviation on DD(IMP);%SDEV =
standard deviation on DD(IMP) expressed as a percentage; GHR1(IMP) = average of 4 ion microprobe analyses of Monastery garnet standard GHR1; GHR1(ICP) = average of 46
LA-ICP-MS analyses of Monastery garnet standard GHR1; %DIFF = % difference.
 F. Viljoen et al. / Lithos 118 (2010) 156–168 159

Table 1 (continued)
AP102A AP102B AP104 AP106A AP106B AP108A AP108B AP109 AP112 AP113 AP114 AP117 AP125 AP127

GNT CPX GNT GNT CPX GNT CPX GNT GNT GNT GNT GNT GNT GNT

40.02 54.06 41.23 39.33 52.31 41.45 55.39 40.57 41.42 39.01 39.85 40.40 40.26 42.28
0.83 0.59 0.42 0.80 0.23 0.56 0.57 0.26 0.49 0.43 0.26 0.50 0.29 0.33
21.43 5.43 22.54 20.74 5.37 23.08 7.27 22.64 22.66 21.25 21.55 22.19 22.64 23.06
0.44 0.23 0.12 0.06 ND 0.20 0.17 0.04 0.14 0.12 0.13 0.04 0.03 0.26
18.29 9.64 14.72 14.37 7.36 14.00 6.03 15.52 13.95 22.53 16.84 16.62 15.52 11.65
0.33 0.12 0.33 0.35 0.10 0.23 0.07 0.28 0.33 1.12 0.29 0.37 0.30 0.28
ND 0.06 ND ND ND ND 0.06 ND ND ND ND ND ND ND
14.87 13.59 17.74 8.70 12.33 16.75 13.69 12.41 18.20 12.12 14.86 14.43 11.24 19.10
4.08 12.12 3.35 14.53 19.57 4.18 12.42 8.76 3.19 2.60 4.68 5.03 9.78 2.96
0.15 3.29 0.15 0.11 2.23 0.15 3.88 0.14 0.15 0.16 0.14 0.17 0.14 0.10
ND 0.08 ND ND ND ND 0.13 ND ND ND ND ND ND ND
100.46 99.22 100.61 98.99 99.53 100.61 99.67 100.63 100.54 99.34 98.60 99.75 100.22 100.04

poor as 50% (e.g. ≤0.02 ppm of Sr, Nb, Ba, La and Pr in garnet, and
≤0.02 ppm of Ba, Nb, Tm and Lu in clinopyroxene; Harte and Kirkley,
1997).
3. Major element chemistry
Major element data for Premier eclogitic inclusions are given in
Table 1, whilst similar compositional data for inclusions from Finsch are
located in Appleyard et al. (2004). At Premier the eclogitic garnets have
variable FeO (6–23 wt.%), MnO (0.28–1.12 wt.%), MgO (7–19 wt.%), and
CaO (2–15 wt.%), but these compositions are broadly similar to those in
the world-wide database for eclogitic garnet inclusions (see Fig. 1 and
Stachel and Harris, 2008). Eclogitic garnet inclusions at Finsch are
enriched in FeO (15 –27 wt.%) and MnO (0.7–1.6 wt.%), relative to
Premier, but are depleted in CaO (8–16 wt.%) and MgO (3–12 wt.%),
when compared to the world-wide database (Appleyard et al., 2004;
Fig. 1). At both localities, Na2O contents in garnet are typical of garnet
inclusions in diamond world-wide.
The chemical compositions of eclogitic inclusion clinopyroxenes
from both mines broadly mirror the Fe–Mg–Ca chemistry of the
garnets with lower FeO in clinopyroxenes from Premier when
compared to Finsch and to the world-wide database (Appleyard et
al., 2004).
4. Trace element chemistry
Trace element data for eclogitic inclusions in diamonds from
Premier and Finsch are given in Tables 2 and 3. Although the
concentrations of Ti, Sr, Y, Zr, Nb, Ba and Hf in eclogitic garnet
inclusions in diamonds at Premier (Table 2) are highly variable, for
example, Sr in the range of 0.29 to 12.48 ppm, Y in the range of 7.40 to
53.9 ppm, Zr in the range of 8.23 to 43.1 ppm, and Ba ranging up to

AP106B AP108A AP108B AP109 AP112 AP113 AP114 AP117 AP125 AP127 DD DD % GHR1 GHR1 %

CPX GNT CPX GNT GNT GNT GNT GNT GNT GNT IMP SDEV SDEV IMP ICP DIFF

1110 2480 2740 1190 2200 1970 1160 2300 1390 1520 1630 30.29 2% 4890 6030 19%
35.6 0.71 125 1.71 0.34 0.61 2.43 0.52 1.52 0.29 0.17 0.02 12% 0.59 0.686 14%
1.54 24.8 4.70 12.4 10.9 18.2 22.9 13.0 16.2 9.12 44.8 1.01 2% 28.8 32.1 10%
2.61 25.6 21.2 12.1 17.7 30.8 8.96 17.5 16.4 20.5 39.2 1.48 4% 92.2 86.5 7%
0.10 0.10 0.23 0.13 0.13 0.15 0.07 0.17 0.17 0.16 0.06 0.01 17% 0.18 0.179 1%
21.2 0.08 0.58 ND ND ND ND 0.03 ND ND ND 0.00 – 0.03 – –
0.02 0.02 1.16 0.03 ND 0.03 0.11 0.01 0.04 ND 0.01 0.00 0% 0.04 0.043 7%
0.36 0.20 3.75 0.30 0.07 0.20 0.33 0.15 0.36 0.08 0.14 0.01 7% 0.40 0.473 15%
0.15 0.07 0.63 0.10 0.03 0.08 0.09 0.05 0.13 0.03 0.05 0.01 20% 0.14 0.159 12%
1.33 0.65 3.39 1.00 0.45 1.04 0.81 0.50 1.44 0.28 0.74 0.09 12% 1.53 1.61 5%
0.46 0.54 1.18 0.69 0.28 0.60 0.56 0.45 0.75 0.35 0.84 0.07 8% 1.26 1.37 8%
0.17 0.38 0.41 0.54 0.19 0.31 0.45 0.22 0.50 0.15 0.48 0.04 8% 0.65 0.746 13%
0.49 1.88 1.35 1.20 0.61 1.26 1.58 0.97 1.56 0.67 2.51 0.16 6% 2.87 2.84 1%
0.08 0.52 0.26 0.30 0.17 0.35 0.41 0.23 0.43 0.18 0.73 0.05 7% 0.65 0.686 5%
0.35 4.36 1.44 2.19 1.54 3.08 3.44 2.25 3.01 1.53 7.00 0.39 6% 5.18 5.16 0%
0.07 1.02 0.22 0.54 0.45 0.82 0.89 0.55 0.68 0.38 1.91 0.12 6% 1.24 1.27 2%
0.14 3.08 0.61 1.64 1.92 2.91 3.05 1.82 2.12 1.41 7.11 0.21 3% 4.06 3.72 9%
0.01 0.52 0.07 0.26 0.39 0.51 0.55 0.31 0.32 0.25 1.26 0.06 5% 0.61 0.574 6%
0.32 3.38 0.45 1.78 3.57 3.88 3.89 2.10 2.04 1.83 9.62 0.39 4% 4.20 4.11 2%
ND 0.53 0.04 0.28 0.69 0.66 0.61 0.34 0.27 0.30 1.66 0.11 7% 0.67 0.623 8%
0.16 0.43 0.92 0.30 0.37 0.66 0.48 0.54 0.43 0.46 0.98 0.13 13% 2.58 2.15 20%
160 F. Viljoen et al. / Lithos 118 (2010) 156–168
0.08 ppm, similar values are also recorded at Finsch (Table 3), with Sr
ranging from 0.25 to 9.49 ppm, Y from 7.61 to 31.5 ppm, Zr from 2.91
to 21.5 ppm, and Ba ranging up to 0.09 ppm. All these compositions,
however, are broadly similar to the world-wide database for eclogitic
garnets in diamond given by Stachel et al. (2004).
Particularly high Sr abundances are recorded in a few of the garnets
analysed, notably Premier AP85 with 12.48 ppm, and Finsch F12A and
F77A with 4.55 ppm Sr and 9.49 ppm Sr, respectively. Such relatively
elevated values are not considered as artifacts related to analytical error
or contamination, because similar values have not only been noted
previously at Premier (up to 3 ppm Sr; Richardson, 1986) and Finsch (up
to 7.09 ppm Sr—Richardson et al., 1990 and up to 28.48 ppm—Smith et
al., 1991), but also found (by different analytical methods) at Argyle
(88 ppm—Griffin et al., 1988), at Mir in Siberia (9 ppm—Sobolev et al.,
1998a) and from Guaniamo alluvials in Venezuela (26.5 ppm—
Kaminsky et al., 2000).
Concentrations of Ti, Sr, Y, Zr, Nb, Ba and Hf in the 8 eclogitic
clinopyroxene inclusions from both localities are also variable (see
Tables 2 and 3), Sr values, for example, lying between 35.6 and 190 ppm,
Y between 1.54 and 6.48 ppm, Zr between 2.61 and 21.2 ppm and Ba
between 0.02 and 21.2 ppm, yet these concentration ranges are broadly
similar to the world-wide database for eclogitic inclusion clinopyrox-
enes (Stachel et al., 2004).
The rare earth element abundance patterns for both eclogitic
garnet and clinopyroxene inclusions from both mines are shown in
Fig. 2, with garnet data divided according to the level of Ca present
(see below). As can be seen, the garnet light rare earth elements at
Premier have concentrations which increase with increasing Ca
content for the low and intermediate diagrams, but there is
insufficient data where the high-Ca garnets are concerned. At Finsch,
a similar abundance trend is noted, which in this case, includes the
high-Ca garnets (see Fig. 2). For the heavy REE (HREE) the garnet
patterns between the two mines show similar characteristics. When
compared to the garnets, 7 of the 8 eclogitic clinopyroxenes show very
flat patterns with slight enrichment in light REE (LREE) but depletion
in HREE. The data from both garnet and clinopyroxene REE patterns
are similar to those described for eclogitic inclusions in diamonds
from localities world-wide (Stachel et al., 2004).
5. Discussion
5.1. Compositional groupings and comparison to other localities
From the Ca–Mg–Fe major element compositions, three distinct
compositional groupings (low-Ca, intermediate-Ca, and high-Ca) can
be defined for the garnets analysed from Premier and Finsch (Fig. 1).
The similarity of these groupings for both localities suggests a
comparable composition for the protolith(s)? at the time of diamond
crystallisation, although with an increased level of Fe and Mn (and a
lower Ca) in the case of Finsch. Also there is a strong correlation
between these compositional groupings and the chondrite-normal-
ised rare earth element patterns of the inclusion garnets (Fig. 2).
Garnets with low calcium (2.6 to 5.0 wt.% CaO for Premier and 3.1 to
4.5 wt.% for Finsch) are characterised by a strong increase from the
LREE to the middle REE (MREE) and a continuous moderate increase
towards the HREE. Garnets with intermediate calcium (7.7 to 9.8 CaO
for Premier and 5.2 to 7.5 wt.% CaO for Finsch) are also characterised
by a strong increase from the LREE to the MREE, but have broadly flat
patterns towards the HREE. Included in this grouping are two garnets
from Finsch with calcium values outside the above range; F71B with
3.9 wt.% CaO, and F96A with 4.5 wt.% CaO. These two specimens are
assigned to this group due to similar rare earth element patterns
when compared to the rest of the garnets with intermediate Ca.
Garnets with high calcium (14.5 to 14.8 wt.% CaO for Premier and 9 to
11.5 wt.% for Finsch) are characterised by unusual, often ‘humped’
REE patterns in which there is typically a strong increase from the
LREE to the MREE, and a broadly flat to decreasing REE pattern
towards the HREE.
Chondrite-normalised HFSE and LILE patterns appear not to have a
strong correlation with garnet calcium composition (Fig. 3), but there
is a markedly lower concentration of Nb evident in garnet and
clinopyroxene from Finsch. This relatively low Nb is likely a reflection
of possible rutile in the source protolith, which will preferentially
retain elements such as Zr, Nb, and Ta (Schmidt et al., 2009). Rutile is a
common accessory mineral in eclogites (Jacob, 2004) and has also
been recognised as an eclogitic type inclusion in diamond (Meyer,
1987).
The relationship of calcium to trace element characteristics in
garnet inclusions has previously been described for eclogitic dia-
monds from world-wide sources. For example, at the Jagersfontein
mine in South Africa, Tappert et al. (2005a,b) showed garnets with Ca
contents in the range of 2.87–8.6 wt.% CaO (equivalent to the low-Ca
and intermediate-Ca groups defined here for Premier and Finsch),
being characterised by REE patterns showing a strong increase from
the LREE towards the MREE and a continuous moderate increase
towards the HREE (∼20–80 times chondritic Lu). The two most
calcium-rich eclogitic garnets from Jagersfontein (13–13.5 wt.% CaO;
equivalent to the high-Ca group defined here) exhibited a different
pattern characterised by flat, ∼10 times chondritic MREE (Sm–Ho)
and HREE, and both show a small positive Eu anomaly.
At Venetia (Aulbach et al., 2002), eclogitic inclusion garnets with
Ca contents in the range of 7–12 wt.% CaO have a steep positive slope
in LREE and flat MREE and HREE (Aulbach et al., 2002), and are
therefore broadly similar to garnets with intermediate Ca from
Premier and Finsch, in terms of chondrite-normalised REE abundance
patterns (although with somewhat higher CaO in garnet). One garnet
from Venetia with 16.8 wt.% CaO differs markedly from the others in
that it has superchondritic LREE (i.e. a hump-shaped LREE) and a
trough in the MREE together with a positive Eu anomaly (Aulbach et
al., 2002). This REE pattern is comparable to some of the high-Ca
group garnets at Premier and Finsch.
Three eclogitic garnets from KanKan (Stachel et al., 2000) mirror
the REE patterns seen at Premier and Finsch, with the low-Ca garnet
(5.4 wt.% CaO) characterised by the lowest LREE, a strong increase
from the light REE towards the MREE, and a continuous moderate
increase towards the HREE. Higher-Ca garnets (10.3 and 14.1 wt.%
CaO) have higher LREE, a strong increase from the LREE to the MREE,
and a broadly flat REE pattern towards the HREE.
Finally, seven eclogitic garnets from Guaniamo with Ca contents in
the range of 6.6 to 12.4 wt.% CaO (Kaminsky et al., 2000) are
chemically equivalent to intermediate-Ca and high-Ca garnets from
Premier and Finsch. These all have similar REE patterns and are
characterised by a strong increase from the LREE towards the MREE,
and a broadly flat MREE and HREE pattern. They are, however,
unusually enriched in LREE (e.g. Ce = 3–12× chondrite) when
compared to garnets with similar Ca contents from Premier and
Finsch.
When viewed collectively, similar correlations of major and trace
elements, as well as REE abundance patterns, are also seen for low
numbers of eclogitic garnets occurring as inclusions in diamonds from
Boa Vista (Tappert et al., 2006), Monastery (Moore et al., 1991),
Mwadui (Stachel et al., 1999), A154 South (Donnelly et al., 2007),
Snap Lake (Promprated et al., 2004), Mir (Taylor et al., 1996; Sobolev
et al., 1998b), Udachnaya (Taylor et al., 1996) and Buffalo Hills (Davies
et al., 2004b).
In contrast to the above, eclogitic garnets occurring as inclusions in
diamonds from Namibian placers (Stachel et al., 2004), with calcium
contents ranging from 2.7 to 18.5 wt.% CaO, have no clear correlation
of REE patterns with respect to calcium contents. Also eclogitic garnet
inclusions in diamonds from DO27 in Canada (Davies et al., 2004a),
with calcium contents ranging from 8.4 to 18.3 wt.% CaO, do not show
any clear correlation. Similarly eclogitic inclusion garnets from the
 F. Viljoen et al. / Lithos 118 (2010) 156–168 161

Table 3
Trace element compositions (ppm) of eclogitic garnet and clinopyroxene inclusions in diamonds from Finsch.

Sample F5A F9A F12A F23A F23B F33B F37B F46A F53A F59C F67B F71B F77A F80B F90A F90C F92C F92E F96A

Mineral GNT GNT GNT GNT GNT GNT GNT GNT GNT GNT CPX GNT GNT CPX GNT CPX GNT CPX GNT

ppm
Ti 1650 1480 2290 1680 1800 1480 1580 572 1880 3120 1330 1930 1380 2020 2000 2120 1700 2420 1690
Sr 1.91 0.90 4.55 1.02 3.53 1.24 3.33 0.45 3.75 0.55 59.3 1.18 9.49 60.7 1.29 190 0.25 76.8 1.65
Y 26.4 18.9 9.82 27.3 31.5 23.6 22.2 7.61 24.3 27.4 2.28 25.6 15.9 2.40 29.1 3.32 20.9 2.65 23.2
Zr 9.86 6.49 21.5 11.3 13.7 7.89 11.8 2.91 17.3 17.0 9.08 16.1 14.5 16.1 10.9 14.0 18.2 21.2 10.5
Nb 0.03 0.05 0.02 0.04 0.04 0.02 0.03 0.05 0.03 0.02 0.04 0.03 0.02 0.02 0.03 0.08 0.03 0.11 0.05
Ba ND 0.09 0.03 ND 0.03 ND ND 0.04 ND ND 0.02 ND 0.03 0.20 ND 0.75 0.02 0.10 ND
La 0.04 0.02 0.14 ND 0.08 0.05 0.08 0.02 0.07 0.10 0.18 0.02 0.24 0.27 0.03 1.06 ND 0.15 0.06
Ce 0.24 0.16 0.65 0.11 0.47 0.31 0.64 0.11 0.45 0.35 0.81 0.19 3.00 0.97 0.26 3.59 0.02 0.91 0.26
Pr 0.08 0.05 0.15 0.04 0.14 0.10 0.20 0.05 0.13 0.03 0.17 0.06 1.20 0.22 0.07 0.59 0.01 0.24 0.09
Nd 1.21 0.52 1.31 0.70 1.84 0.78 2.29 0.43 1.66 0.23 1.09 0.96 13.4 1.67 0.75 3.12 0.28 2.02 0.82
Sm 1.32 0.35 0.69 1.00 2.44 0.60 1.95 0.27 1.67 0.84 0.41 1.34 7.26 0.83 0.68 0.86 0.36 0.70 0.87
Eu 0.78 0.25 0.70 0.65 1.37 0.41 1.12 0.14 0.91 0.78 0.18 0.72 2.72 0.31 0.43 0.29 0.19 0.22 0.47
Gd 3.08 1.08 1.43 2.55 5.05 1.82 3.58 0.54 3.21 2.98 0.54 3.00 7.19 0.91 2.19 1.06 0.99 0.77 2.06
Tb 0.66 0.28 0.28 0.60 0.93 0.46 0.65 0.13 0.68 0.62 0.09 0.59 0.84 0.15 0.54 0.16 0.29 0.13 0.46
Dy 4.49 2.86 1.84 4.61 6.22 3.71 4.32 1.05 4.65 4.79 0.57 4.29 3.95 0.70 4.31 0.92 3.02 0.78 3.63
Ho 1.05 0.73 0.40 1.00 1.18 0.90 0.83 0.31 0.90 1.11 0.10 0.94 0.66 0.11 1.05 0.14 0.83 0.12 0.87
Er 3.04 2.43 1.25 3.14 3.22 2.93 2.43 1.31 2.73 3.29 0.23 3.06 1.70 0.16 3.38 0.31 3.32 0.38 2.57
Tm 0.49 0.40 0.21 0.49 0.44 0.47 0.37 0.25 0.40 0.47 0.03 0.46 0.26 0.03 0.53 0.05 0.63 0.04 0.45
Yb 3.21 2.74 1.33 3.20 2.80 3.30 2.55 2.09 2.77 2.93 0.23 3.12 1.54 0.29 3.61 0.31 4.73 0.31 3.08
Lu 0.49 0.42 0.22 0.49 0.39 0.52 0.40 0.38 0.42 0.45 0.03 0.49 0.23 0.02 0.63 0.03 0.80 0.03 0.50
Hf 0.44 0.26 0.58 0.49 0.52 0.42 0.31 0.11 0.50 0.55 0.54 0.65 0.30 0.73 0.59 0.94 0.47 1.29 0.47

ND = not detected (b 0.005 ppm).

Jericho mine in Canada, with Ca contents in the range of 4.1 to 7.5 wt.%
CaO, but distinctly magnesium-rich (12.8–20.7 wt.% MgO), whilst
characterised by steadily increasing REE patterns from the light REE to
the heavy REE (De Stefano et al., 2009), show no apparent
compositional groupings. In fact the Jericho garnets may be unique
because of their high MgO contents and are totally unlike the eclogitic
diamond paragenesis present at Premier and Finsch.
5.2. Reconstruction of eclogite bulk compositions
Eclogite is essentially a bimineralic rock consisting of garnet and
clinopyroxene. As garnet is a major host for e.g. the MREE and
(particularly) the HREE, whilst clinopyroxene is a major host for e.g. Sr
as well as the LREE, the trace element content and chondrite-
normalised trace element abundance patterns of eclogitic garnet and
clinopyroxene inclusions, when viewed in isolation, are unlikely to
accurately reflect that of the host eclogite (Fig. 4). In order to obtain
this information it is necessary to calculate bulk compositions from
combinations of these two inclusions, but this is not always possible in
the present study as typically only one of the necessary inclusions
(garnet in the present study) are encountered as an inclusion in the
diamonds. An additional complication is that there is a marked
correlation of trace element partition coefficients with clinopyrox-
ene–garnet Ca contents (Harte and Kirkley, 1997; references therein).
However based on known clinopyroxene–garnet partition coefficients
which incorporate the effect of Ca (Harte and Kirkley, 1997), trace
element contents for theoretical, ‘co-existing’ clinopyroxenes can be
calculated from the garnet compositions. From these, the composition
of the eclogite host may be calculated, assuming the proportion of
garnet to clinopyroxene in the bulk rock is set to some figure, which in
this case will be 50%.
The calculated (‘theoretical’) trace element abundances for
clinopyroxenes at Premier and Finsch are compared to actual
(‘measured’) clinopyroxene abundances in Fig. 5. As can be seen
measured and calculated patterns are invariably close, with differ-
ences most likely being due to uncertainties in the determination of
distribution coefficients for these elements as a result of analytical
uncertainty at low concentration levels (Harte and Kirkley, 1997).
Differences between measured and calculated values may also be due
to disequilibrium between garnet and clinopyroxene. For instance,
two separate garnet inclusions in Finsch diamond F23 do show
slightly different concentrations of the light and middle REE, and a
degree of disequilibrium is therefore indicated in this instance
(Table 3 and Fig. 5). However for the most part it would appear that
equilibrium between co-existing phases can be assumed (Fig. 5). Thus
from the above evidence, the calculated inclusion trace element
abundances will reflect the trace element composition of the eclogite
protolith to the diamonds from Premier and Finsch. This is important
to this discussion, as disequilibrium between co-existing phases
occurring as inclusions in diamonds invalidates the calculations, see
for example Taylor and Anand (2004).
At both Premier and Finsch, the low-Ca group samples are LREE
depleted, and have relatively flat calculated bulk-rock trace element
abundance patterns at approximately 10 times chondrite concentra-
tions, but with marked positive Sr and negative Zr anomalies (Fig. 6).
The intermediate-Ca group samples are LREE depleted, show Sr and Zr
anomalies, have somewhat higher concentrations of Zr and MREE, and
have HREE contents that overlap with the low-Ca group (Fig. 6). High-
Ca group samples are LREE depleted, show Sr and Zr anomalies, are
MREE-enriched, and have HREE contents that are slightly less than the
low- and intermediate-Ca group samples (Fig. 6). Although these
calculated bulk-rock trace element abundance patterns are somewhat
different from that defined by the garnets (Figs. 2 and 3), it is clear
that the markedly discrete compositional groupings recognised on the
basis of garnet major and trace element compositions alone, are still
seen. Hence, in spite of the correlation of trace element partition
coefficients with clinopyroxene–garnet Ca contents (Harte and
Kirkley, 1997), the effect is not large enough to hide compositional
differences (and is adequately accounted for through application of
Harte and Kirkley, 1997). It is therefore concluded that three distinct
eclogite protolith compositions are present at Premier and Finsch.
Similar correlations seen for eclogitic diamonds from many other
localities (Section 5.1) suggests that this may be a world-wide
phenomenon.
162 F. Viljoen et al. / Lithos 118 (2010) 156–168
Fig. 1. Ternary plot of the CaO, MgO and FeO contents of eclogitic garnets in diamonds
world-wide (Stachel and Harris, 2008 and references therein) as well as for diamonds
from the Premier (Gurney et al., 1985) and Finsch (Gurney et al., 1979; Appleyard et al.,
2004) kimberlites. Solid symbols refer to garnets analysed for their trace element
content by ion microprobe.
5.3. Petrogenesis of the eclogite protolith
REE patterns with nearly flat HREE (typical of mid-ocean ridge
basalt, MORB) but with positive Eu anomalies are displayed
by ∼49% of kimberlite-derived eclogite xenoliths, and are especially
common amongst coesite-, kyanite- and corundum-bearing eclo-
gites (Jacob, 2004). This is often interpreted as evidence of
plagioclase accumulation, implying that the protoliths to these
eclogite xenoliths underwent magmatic fractionation at shallow
depths prior to subduction and high-pressure metamorphism
(Jacob, 2004).
The calculated average bulk-rock REE contents of the eclogite
source rock to eclogitic diamonds from localities world-wide can be
modelled as that of MORB that has lost about 10% of a partial melt in
the eclogite stability field, and that subsequently experienced some
re-enrichment in the light REE, probably through metasomatism
prior to, or during, diamond crystallisation (Stachel et al., 2004).
Linking the evidence for the two hypotheses to the data from
Premier and Finsch suggests a more complex petrogenetic model. Of
particular importance in the current data are the presence of marked
negative Zr anomalies as well as marked positive Sr anomalies in all
compositional groupings, and a general absence of Eu anomalies
(Figs. 6 and 7). The Zr anomaly can arise in reconstructed bulk eclogite
trace element abundance patterns if rutile, which may contain
substantial concentrations of Zr, is not included in the calculations.
For example, addition of just 1% rutile (with a Zr content of just over
1000 ppm; see Jacob et al., 2003) to the calculated bulk eclogite
compositions for Premier and Finsch would result in flat Zr
abundances. The rutile addition would have no effect on the observed
positive Sr anomalies, which as previously noted, when associated
with Eu anomalies, are commonly interpreted as evidence in support
of an eclogite genesis involving subducted plagioclase-rich cumulates
in oceanic crust (Jacob, 2004; references therein). However, in the
present data, the absence of Eu anomalies in the calculated bulk
eclogite compositions for Premier and Finsch (Figs. 6 and 7) does not
lend support to the latter model.
A possible alternative is that the protoliths may have consisted of
varying mixtures of oceanic gabbro and MORB. Typical oceanic gabbro
is characterised by positive anomalies in Sr and Eu, and negative Zr
(e.g. Benoit et al., 1996; Kelemen et al., 1997; Coogan et al., 2001),
whilst MORB has a smooth trace element abundance pattern (Fig. 6)
with abundances higher than that of oceanic gabbro (Klein, 2004).
Mixing of gabbro and MORB (e.g. a 50:50 ratio) results in a trace
element abundance pattern with no Eu and Zr anomaly, but which
still shows a marked positive Sr anomaly (Fig. 6). The discrepancy
between Ba abundances in reconstructed eclogites and potential
crustal protoliths (Figs. 6 and 7) might be due to mobilization of Ba
during subduction-related dehydration.
The averaged eclogite bulk compositions for Premier and Finsch
are also markedly similar to that of clinopyroxene in a typical
cumulate gabbro (Fig. 7), and a role for cumulate clinopyroxene in
protolith evolution (e.g. a cumulate rock of gabbroic affinity,
comprising of 90%, or more, clinopyroxene) may therefore also be
inferred.
In an earlier study, Aulbach et al. (2002) reconstructed, from the
trace element chemistries of eclogitic inclusions in diamonds from
Venetia, a bulk eclogite source rock similar to that for diamonds from
Premier and Finsch (i.e. trace element abundances intermediate
between that of oceanic gabbro and enriched MORB), but which is
depleted in highly incompatible elements relative to these possible
precursors. Quantitative modelling of this relationship showed that
greater compatibility is achieved if these oceanic precursors were
subjected to a loss of partial melts after subduction into the eclogite
stability field (Aulbach et al., 2002).
The calculated bulk-rock REE patterns for the host eclogite to the
diamonds from Premier and Finsch are similar to those of diamond-
bearing eclogites from the Mir kimberlite in Russia (Beard et al.,
1996). In that case, the authors postulated that low-Ca group
eclogites represent low-temperature, sea-water altered extrusive
basalts, in the upper section of an ophiolite complex, whilst high-Ca
group eclogites represent a similar, but high-temperature alter-
ation, in the mid to lower sections of the ophiolite, involving
feldspathic cumulates. However, the presence of marked Zr and Sr
anomalies in the calculated bulk eclogite compositions at Premier
and Finsch argues against a protolith consisting solely of subducted
basalt (Figs. 6 and 7), whilst the lack of Eu anomalies does not
support a model of protolith evolution involving only feldspathic
cumulates (Fig. 7).
The possibility of a high-pressure igneous or metasomatic origin
for all of the chemical variability of mineral inclusions in eclogitic
diamonds should also be considered (Taylor and Anand, 2004;
Griffin and O'Reilly, 2007). For instance, observed ranges in the
major and trace element mineral chemistry of eclogitic garnets in a
 F. Viljoen et al. / Lithos 118 (2010) 156–168
Fig. 2. Rare earth element abundance patterns for garnet and clinopyroxene in eclogitic diamonds from Premier and Finsch, determined by ion microprobe. All data normalised to the
CI carbonaceous chondrite of McDonough and Sun (1995).
single diamond from Mir are attributed to changing fluid chemistry
in successive metasomatic fronts moving through a subducted
oceanic slab in the mantle (Sobolev et al., 1998b), with the
metasomatic fluids also being the carbon source for the diamond.
Diamonds, and diamond-bearing eclogites in the Jericho kimberlite,
provide evidence for multiple episodes of melt- and metasomatism-
related modification of the high-pressure eclogite source to these
diamonds, ultimately resulting in substantial modification of major
and trace element compositions (De Stefano et al., 2009; Smart
et al., 2009).
Metasomatic re-enrichment in LREE of the eclogite precursor
subsequent to subduction, metamorphism and partial melting
(Stachel et al., 2004) may be accompanied by the introduction of Sr
to form the Sr anomalies observed. However most reconstructed REE
patterns for Premier and Finsch (Fig. 6) appear mildly LREE depleted
and a secondary Sr enrichment in the absence of other metasomatic
indicators (e.g. LREE enrichment) seems implausible. On this basis,
melt-, and metasomatism-related models of eclogite evolution are
therefore not preferred for the source to the eclogitic diamonds from
Premier and Finsch. Discrete eclogite compositions are also seen, and
not a continuum, as would be expected from the evolution of a
magma, or from pervasive metasomatism. The Sr isotopic composi-
tions of eclogitic inclusions in diamonds from Premier and Finsch
(Richardson, 1986; Richardson et al., 1990; Smith et al., 1991) also do
not support a model of metasomatic introduction of Sr as they do not
plot in the field of old, metasomatised lithosphere on the Sr–Nd
isotope diagram (enriched Sr and enriched Nd; Pearson et al., 2003;
see below).
163
Our preferred model of protolith evolution for the eclogitic
diamonds from Premier and Finsch, therefore, is one in which both
the major and trace element chemistry of the inclusions are ultimately
inherited from three discrete low-pressure oceanic protoliths, con-
sisting of varying mixtures of oceanic basalt + gabbro for diamonds
from both Premier and Finsch. It is likely that prior to and during
diamond crystallisation the major and trace element compositions of
the high-pressure eclogite source rock to these diamonds may have
been slightly modified by very minor interaction with metasomatic
fluids and melts, probably associated with the diamond crystallisation
event (Appleyard et al., 2004), but large-scale melt- or metasoma-
tism-related modification is not indicated.
As a result of the similarity in trace element abundance patterns
for the host eclogite to diamonds at Premier and Finsch (Figs. 6 and 7),
no clear insights into the ultimate origin of the Fe- and Mn-
enrichment seen in eclogitic inclusions in diamonds from Finsch
(but not seen at Premier) can be obtained from the trace element data.
5.4. The evidence from isotope studies
The isotopic composition of eclogitic mineral inclusions in
diamond, as well as that of the diamond host, may provide important
additional clues to the petrogenesis of the protoliths to these eclogitic
diamond source rocks.
Based on the data of Richardson (1986), eclogitic inclusions in
diamonds from Premier plot in the depleted quadrant of the Sr–Nd
isotope diagram as they have higher initial Nd isotopic ratios and
lower initial Sr isotopic ratios than those of the Bulk Earth at the age
164 F. Viljoen et al. / Lithos 118 (2010) 156–168
of diamond crystallisation and inclusion encapsulation. These
isotopic compositions are characteristic of LREE depleted upper
mantle such as the source to mid-ocean ridge basalt (Allègre, 2008;
references therein), perhaps indicating a link to subducted oceanic
crust.
In the case of Finsch the inclusion data of Richardson et al. (1990)
as well as Smith et al. (1991) plot predominantly in a field of depleted
Nd and enriched Sr at the age of diamond crystallisation, typical of
altered oceanic crust (Staudigel, 2004, references therein). It is
reasonable to infer that these diamonds may have crystallised in an
eclogite source rock formed through subduction and metamorphism
of altered oceanic crust (Smith et al., 1991). Alteration of oceanic crust
may also result in the precipitation of Fe- and Mn-oxides (Staudigel,
2004). The unusually high FeO and MnO contents of inclusions in the
Finsch diamonds, and the positive correlation between inclusion MnO
contents and Sr isotopic compositions therefore provide further
support for a subducted slab origin for the source rock to these
diamonds (Smith et al., 1991).
In many cases, the carbon isotopic composition of diamonds
containing eclogitic inclusions is either higher (e.g. δ13C = +5‰;
Cartigny, 2005) or lower (e.g. δ13C = − 41‰; De Stefano et al., 2009)
than typical mantle values for carbon (approximately −5‰; Deines,
2002). This variation in δ13C for eclogitic diamonds is present at
Premier and Finsch (Deines et al., 1984), and the δ13C values also
correlate with the major element chemistry of the garnet and
clinopyroxene inclusions (Deines et al., 1989).
Mn-enriched eclogitic inclusions in diamonds from Roberts Victor
are characterised by unusually light δ13C isotopic compositions
Fig. 3. Large-ion lithophile element and high field-strength element abundance patterns for garnet and clinopyroxene in eclogitic diamonds from Premier and Finsch, determined by
ion microprobe. All data normalised to the CI carbonaceous chondrite of McDonough and Sun (1995).
(−15.1‰ to − 16.3‰; Deines et al., 1987) in the diamond host,
which might be derived from Mn-rich oceanic sediment containing
organic carbon. However a correlation between Mn-enriched inclu-
sions and light δ13C in the diamond host is not observed at Finsch,
where the δ13C of the host diamond for such inclusions ranges
between −3‰ and −8‰ (Deines et al., 1984; Deines et al., 1989).
The source of this anomalous carbon in the upper mantle has been
the subject of some debate. Deines (1980), for example, suggested
that the carbon source originated from heterogeneities generated
during the original accretion of the earth. An alternative explanation
(Kirkley et al., 1991; Schulze et al., 2004), is that the carbon derives
from a crustal (biogenic) source. At present there is no consensus
(Deines et al., 1993; Cartigny, 2005) and the subject is further
complicated when the nitrogen isotopes are considered. For instance,
at Jwaneng a recycled sediment contribution to eclogitic diamond
formation is ruled out by the lack of a heavy nitrogen isotope
signature in the diamonds (Cartigny et al., 1998b).
If protolith formation and diamond crystallisation are separate
and unrelated events, these difficulties may be overcome as
diamond formation would be from infiltrating fluids within a pre-
existing protolith (Navon, 1999; Stachel and Harris, 2009). In such a
case, isotopic signatures that exist within eclogitic silicate and
sulphide inclusions, such as those from O, S, Pb and Os, and which
provide evidence for a crustal origin (Eldridge et al., 1991;
Richardson et al., 2001; Farquhar et al., 2002; Schulze et al.,
2003), would result from subducted material within the protolith
recrystallising during diamond formation during or after a metaso-
matic event.
 F. Viljoen et al. / Lithos 118 (2010) 156–168 165

Fig. 4. Reconstructed bulk eclogite trace element patterns for the host rocks to a selection of eclogitic diamonds from Premier and Finsch, calculated for varying clinopyroxene–garnet
ratios (25% clinopyroxene to 75% clinopyroxene). All data normalised to the CI carbonaceous chondrite of McDonough and Sun (1995). Sample numbers indicated.

Fig. 5. Trace element abundances for co-existing garnet and clinopyroxene in diamonds from Premier and Finsch, illustrating the deviation of measured and calculated trace element
values for clinopyroxene (based on clinopyroxene–garnet trace element distribution and using equations and constants of Harte and Kirkley, 1997). All data normalised to the CI
carbonaceous chondrite of McDonough and Sun (1995).
166 F. Viljoen et al. / Lithos 118 (2010) 156–168

Fig. 6. N-MORB (Klein, 2004), gabbro sample 92OG53 (Benoit et al., 1996), and a 50:50 mixture of N-MORB + gabbro, plotted in comparison to the reconstructed bulk eclogite trace
element abundance patterns for the host rocks to eclogitic diamonds from Premier and Finsch (calculated for a 50:50 clinopyroxene–garnet ratio). All data normalised to the CI
carbonaceous chondrite of McDonough and Sun (1995); element order after Hofmann (1988).

Fig. 7. N-MORB (Klein, 2004) and gabbro sample 92OG53 (Benoit et al., 1996) as well as constituent plagioclase and clinopyroxene, plotted in comparison to the averaged 50:50 gnt:
cpx bulk eclogite trace element pattern for each of the compositional groupings identified at Premier and Finsch in this study. All data normalised to the CI carbonaceous chondrite of
McDonough and Sun (1995); element order after Hofmann (1988).

6. Conclusions 2. The presence of marked negative Zr anomalies as well as marked

The following conclusions are considered noteworthy:
1. Based on garnet Ca–Mg–Fe major element compositions, and
calculated bulk eclogite trace element contents and abundance
patterns, it is clear that there are three chemically distinct eclogite
source rocks to the eclogitic diamonds at the Premier and Finsch
diamond mines.
positive Sr anomalies in the calculated bulk eclogite compositions
for Premier and Finsch is inconsistent with a model of eclogite
formation involving only subducted mid-ocean ridge basalt, whilst
the absence of Eu anomalies in the calculated bulk eclogite
compositions does not lend support to a model of eclogite genesis
involving only subducted plagioclase-rich cumulates in oceanic
crust.
 F. Viljoen et al. / Lithos 118 (2010) 156–168
3. Calculated bulk eclogite trace element abundances are intermedi-
ate between that of typical oceanic gabbro and MORB, and can be
modelled as gabbro–MORB mixtures containing on the order of
50% MORB. The averaged eclogite bulk compositions for Premier
and Finsch are also markedly similar to that of clinopyroxene in a
typical cumulate gabbro, and a role for cumulate clinopyroxene in
protolith evolution may therefore also be inferred.
4. Broadly similar compositional groupings for both localities suggest
that comparable eclogite protolith compositions were present at
the time of diamond crystallisation, although with an increased
level of Fe and Mn in the case of Finsch. This difference in major
element composition is not reflected in the trace element data.
5. Models of eclogite protolith evolution involving fractionating
melts, or large-scale melt- or fluid-related metasomatism, are not
preferred for the source to the eclogitic diamonds from Premier
and Finsch, as discrete eclogite compositions are recognised, and
not a continuum, as would be expected from the evolution of a
magma, or from pervasive metasomatism. Calculated eclogite bulk
compositions also do not show the marked LREE enrichment that
would result from melt-related processes, or from metasomatism.
Acknowledgements
The authors would like to thank the management of De Beers
Consolidated Mines Limited for the donation of study material, for
funding the project, and for permission to publish. The main body of
research was completed whilst FV was still employed in the
Exploration Division of De Beers at the GeoScience Centre in
Johannesburg, South Africa. The members of the HOH Kimberley
diamond team at the time of this study, Ray Ferraris, Verlece
Anderson, Gill Parker, Edna van Blerk, Wanita Moore and Ronel
Hamman, are thanked for many hours of work selecting diamonds
from production for this study. We thank John Craven and Richard
Hinton of the Edinburgh ion probe facility for their excellent support,
and Ben Harte for arranging access to the facility. Constructive advice,
criticisms and critical reviews by Thomas Stachel, Sonja Aulbach, and
two anonymous reviewers assisted greatly in the improvement of the
paper. This research was partially supported by funds from the South
African Department of Science and Technology Research Chair in
Geometallurgy (FV).
References
Allègre, C.J., 2008. The Edinburgh Building, Cambridge CB2 8RU, United Kingdom.
Isotope Geology. Cambridge University Press, p. 512.
Appleyard, C.M., Viljoen, K.S., Dobbe, R., 2004. A study of eclogitic diamonds and their
inclusions from the Finsch kimberlite pipe, South Africa. Lithos 77, 317–332.
Aulbach, S., Stachel, T., Viljoen, K.S., Brey, G.P., Harris, J.W., 2002. Eclogitic and
websteritic diamond sources beneath the Limpopo Belt — is slab-melting the link?
Contributions to Mineralogy and Petrology 143, 56–70.
Beard, B.L., Fraracci, K.N., Taylor, L.A., Snyder, G.A., Clayton, R.A., Mayeda, T.K.,
Sobolev, N.V., 1996. Petrography and geochemistry of eclogites from the Mir
kimberlite, Yakutia, Russia. Contributions to Mineralogy and Petrology 125,
293–310.
Benoit, M., Polvé, M., Ceuleneer, G., 1996. Trace element and isotopic characterisation of
mafic cumulates in a fossil mantle diapir (Oman ophiolite). Chemical Geology 134,
199–214.
Campbell, J.L., Teesdale, W.J., Kjarsgaard, B.A., Cabri, L.J., 1996. Micro-PIXE analysis of
silicate reference standards for trace Ni, Cu, Zn, Ga, Ge, As, Rb, Sr, Y, Zr, Nb, Mo and
Pb, with emphasis on Ni for application of the Ni-in-garnet geothermometer. The
Canadian Mineralogist 34, 37–48.
Cartigny, P., Harris, J.W., Phillips, D., Girard, M., Javoy, M., 1998a. Subduction-related
diamonds? — the evidence for a mantle-derived origin from coupled δ13C–δ15N
determinations. Chemical Geology 147, 147–159.
Cartigny, P., Harris, J.W., Javoy, M., 1998b. Eclogitic diamond formation at Jwaneng: no
room for a recycled component. Science 280, 1421–1424.
Cartigny, P., 2005. Stable isotopes and the origin of diamond. Elements 1, 79–84.
Coogan, L.A., MacLeod, C.J., Dick, H.J.B., Edwards, S.J., Kvassnes, A., Natland, J.H.,
Robinson, P.T., Thompson, G., O'Hara, M.J., 2001. Whole-rock geochemistry of
gabbros from the Southwest Indian Ridge: constraints on geochemical
fractionations between the upper and lower oceanic crust and magma chamber
processes at (very) slow-spreading ridges. Chemical Geology 178, 1–22.
167
Davies, R.M., Griffin, W.L., O'Reilly, S.Y., Doyle, B.J., 2004a. Mineral inclusions and
geochemical characteristics of microdiamonds from the DO27, A154, A21, A418,
DO18, DD17 and Ranch Lake kimberlites at Lac de Gras, Slave Craton, Canada. Lithos
77, 39–55.
Davies, R.M., Griffin, W.L., O'Reilly, S.Y., McCandless, T.E., 2004b. Inclusions in diamonds
from the K14 and K10 kimberlites, Buffalo Hills, Alberta, Canada: diamond growth
in a plume? Lithos 77, 99–111.
De Stefano, A., Kopylova, M.G., Cartigny, P., Afanasiev, V., 2009. Diamonds and eclogites
of the Jericho kimberlite (Northern Canada). Contributions to Mineralogy and
Petrology 158, 295–315.
Deines, P., 1980. The carbon isotopic composition of diamonds: relationship to diamond
shape, color, occurrence and vapor composition. Geochimica et Cosmochimica Acta
44, 943–961.
Deines, P., Gurney, J.J., Harris, J.W., 1984. Associated chemical and carbon isotopic
composition variations in diamonds from Finsch and Premier kimberlite, South
Africa. Geochimica et Cosmochimica Acta 48, 325–342.
Deines, P., Harris, J.W., Gurney, J.J., 1987. Carbon isotopic composition, nitrogen content
and inclusion composition of diamonds from the Roberts Victor kimberlite, South
Africa: evidence for 13C depletion in the mantle. Geochimica et Cosmochimica Acta
51, 1227–1243.
Deines, P., Harris, J.W., Spear, P.M., Gurney, J.J., 1989. Nitrogen and 13C content of Finsch
and Premier diamonds and their implications. Geochimica et Cosmochimica Acta
53, 1367–1378.
Deines, P., Harris, J.W., Gurney, J.J., 1993. Depth-related carbon isotope and nitrogen
concentration variability in the mantle below the Orapa kimberlite, Botswana,
Africa. Geochimica et Cosmochimica Acta 57, 2781–2796.
Deines, P., 2002. The carbon isotope geochemistry of mantle xenoliths. Earth-Science
Reviews 58, 247–278.
Donnelly, C.L., Stachel, T., Creighton, S., Muehlenbachs, K., Whiteford, S., 2007.
Diamonds and their mineral inclusions from the A154 South pipe, Diavik Diamond
Mine, Northwest Territories, Canada. Lithos 98, 160–176.
Eldridge, C.S., Compston, W., Williams, I.S., Harris, J.W., Bristow, J.W., 1991. Isotope
evidence for the involvement of recycled sediments in diamond formation. Nature
353, 649–653.
Farquhar, J., Wing, B.A., McKeegan, K.D., Harris, J.W., Cartigny, P., Thiemens, M.H., 2002.
Mass-independent sulfur of inclusions in diamond and sulfur recycling on early
Earth. Science 298, 2369–2372.
Field, M., Stiefenhofer, J., Robey, J., Kurszlaukis, S., 2008. Kimberlite-hosted diamond
deposits of southern Africa: a review. Ore Geology Reviews 34, 33–75.
Griffin, W.L., Jaques, A.L., Sie, S.H., Ryan, C.G., Cousens, D.R., Suter, G.F., 1988. Conditions
of diamond growth: a proton microprobe study of inclusions in West Australian
diamonds. Contributions to Mineralogy and Petrology 99, 143–158.
Griffin, W.L., O'Reilly, S.Y., 2007. Cratonic lithospheric mantle: is anything subducted?
Episodes 30 (1), 43–53.
Gurney, J.J., Harris, J.W., Rickard, R.S., 1979. Silicate and oxide inclusions in diamonds
from the Finsch kimberlite pipe. In: Boyd, F.R., Meyer, H.O.A. (Eds.), Kimberlites,
Diatremes and Diamonds, their Geology, Petrology and Geochemistry. American
Geophysical Union, Washington, pp. 1–15.
Gurney, J.J., Harris, J.W., Rickard, R.S., 1984. Silicate and oxide inclusions in diamonds
from the Orapa Mine, Botswana. In: Kornprobst, J. (Ed.), Kimberlites II: The Mantle
and Crust–Mantle Relationships. Elsevier, Amsterdam, pp. 3–9.
Gurney, J.J., Harris, J.W., Rickard, R.S., Moore, R.O., 1985. Inclusions in Premier
Mine diamonds. Transactions of the Geological Society of South Africa 88,
301–310.
Gurney, J.J., 1991. The diamondiferous roots of our wandering continents. South African
Journal of Geology 93, 424–437.
Harris, J.W., Gurney, J.J., 1979. Inclusions in diamond. In: Field, J.E. (Ed.), The Properties
of Diamond. Academic Press, London, pp. 555–591.
Harris, J.W., Stachel, T., Leost, I., Brey, G.P., 2004. Peridotitic diamonds from Namibia:
constraints on the composition and evolution of their mantle source. Lithos 77,
209–223.
Harte, B., Kirkley, M.B., 1997. Partitioning of trace elements between clinopyroxene and
garnet: data from mantle eclogites. Chemical Geology 136, 1–24.
Hofmann, A.W., 1988. Chemical differentiation of the Earth: the relationship between
mantle, continental crust, and oceanic crust. Earth and Planetary Science Letters 90,
297–314.
Ireland, T.R., Rudnick, R.L., Spetsius, Z., 1994. Trace elements in diamond inclusions
from eclogites reveal link to Archean granites. Earth and Planetary Science Letters
128, 199–213.
Irving, A.J., Frey, F.A., 1978. Distribution of trace elements between garnet megacrysts
and host volcanic liquids of kimberlitic to rhyolitic composition. Geochimica et
Cosmochimica Acta 42, 771–787.
Jacob, D., Jagoutz, E., Lowry, D., Mattey, D., Kudrjavtseva, G., 1994. Diamondiferous
eclogites from Siberia: remnants of Archean oceanic crust. Geochimica et
Cosmochimica Acta 58, 5191–5207.
Jacob, D.E., Schmickler, B., Schulze, D.J., 2003. Trace element geochemistry of coesite-
bearing eclogites from the Roberts Victor kimberlite, Kaapvaal Craton. Lithos 71,
337–351.
Jacob, D.E., 2004. Nature and origin of eclogite xenoliths from kimberlites. Lithos 77,
295–316.
Kaminsky, F.V., Zakharchenko, O.D., Griffin, W.L., Channer, D.M.D., Khachatryan-
Blinova, G.K., 2000. Diamond from the Guaniamo area, Venezuela. The Canadian
Mineralogist 38, 1347–1370.
Kelemen, P.B., Koga, K., Shimizu, N., 1997. Geochemistry of gabbro sills in the crust–
mantle transition zone of the Oman ophiolite: implications for the origin of the
oceanic lower crust. Earth and Planetary Science Letters 146, 475–488.
168 F. Viljoen et al. / Lithos 118 (2010) 156–168
Kirkley, M.B., Gurney, J.J., Otter, M.L., Hill, S.J., Daniels, L.R., 1991. The application of C
isotope measurements to the identification of the sources of C in diamonds: a
review. Applied Geochemistry 6, 477–494.
Klein, E.M., 2004. Geochemistry of the igneous oceanic crust. In: The Crust (Ed. R.L.
Rudnick), Treatise on Geochemistry (Eds. H.D. Holland & K.K. Turekian), Elsevier-
Pergamon, Oxford, volume 3, pp. 433–464.
McDonough, W.F., Sun, S.-S., 1995. The composition of the earth. Chemical Geology 120,
223–253.
Menzies, A.H., Carlson, R.W., Shirey, S.B., Gurney, J.J., 2003. Re–Os systematics of
diamond-bearing eclogites from the Newlands kimberlite. Lithos 71, 323–336.
Meyer, H.O.A., 1987. Inclusions in diamond. In: Nixon, P.H. (Ed.), Mantle Xenoliths.
Wiley, London, pp. 501–522.
Moore, R.O., Gurney, J.J., Griffin, W.L., Shimizu, N., 1991. Ultra-high pressure garnet
inclusions in Monastery diamonds: trace element abundance patterns and
conditions of origin. European Journal of Mineralogy 3, 213–230.
Navon, O., 1999. Diamond formation in the earth's mantle. In: Gurney, J.J., Gurney, J.L.,
Pascoe, M.D., Richardson, S.H. (Eds.), Proceedings of the 7th International
Kimberlite Conference Vol 2. Red Roof Design, Cape Town, pp. 584–604.
Pearson, D.G., Canil, D., Shirey, S.B., 2003. Mantle samples included in volcanic rocks:
xenoliths and diamonds. In: The Mantle and Core (ed R.W. Carlson) Treatise on
Geochemistry (eds. H.D. Holland and K.K. Turekian), Elsevier-Pergamon, Oxford,
volume 3, pp. 61–102.
Phillips, D., Harris, J.W., Viljoen, K.S., 2004. Mineral chemistry and thermobarometry of
inclusions from De Beers Pool diamonds, Kimberley, South Africa. Lithos 77,
155–179.
Promprated, P., Taylor, L.A., Anand, M., Floss, C., Sobolev, N.V., Pokhilenko, N.P., 2004.
Multiple-mineral inclusions in diamonds from the Snap Lake/King Lake dike, Slave
Craton, Canada: a trace element perspective. Lithos 77, 69–81.
Richardson, S.H., Gurney, J.J., Erlank, A.J., Harris, J.W., 1984. Origin of diamonds in old
enriched mantle. Nature 310, 198–202.
Richardson, S.H., 1986. Latter-day origin of diamonds of eclogitic paragenesis. Nature
322, 623–626.
Richardson, S.H., Erlank, A.J., Harris, J.W., Hart, S.R., 1990. Eclogitic diamonds of
proterozoic age from Cretaceous kimberlites. Nature 346, 54–56.
Richardson, S.H., Harris, J.W., Gurney, J.J., 1993. Three generations of diamonds from old
continental mantle. Nature 366, 256–259.
Richardson, S.H., Shirey, S.B., Harris, J.W., Carlson, R.W., 2001. Archean subduction
recorded by Re–Os isotopes in eclogitic sulphide inclusions in Kimberley diamonds.
Earth and Planetary Science Letters 191, 257–266.
Schmidt, A., Weyer, S., John, T., Brey, G.P., 2009. HFSE systematics of rutile-bearing
eclogites: new insights into subduction zone processes and implications for the
earth's HFSE budget. Geochimica et Cosmochimica Acta 73, 455–468.
Schulze, D.J., Harte, B., Valley, J.W., Brenan, J.M., Channer, D.M.D., 2003. Extreme crustal
oxygen isotopic signatures preserved in coesite in diamond. Nature 423, 68–70.
Schulze, D.J., Harte, B., Valley, J.W., Channer, D.M.D., 2004. Evidence of subduction and
crust–mantle mixing from a single diamond. Lithos 77, 349–358.
Smart, K.A., Heaman, L.M., Chacko, T., Simonetti, A., Kopylova, M., Mah, D., Daniels, D.,
2009. The origin of diamond eclogites from the Jericho kimberlite, Canada. Earth
and Planetary Science Letters 284, 527–537.
Smith, C.B., 1983. Pb, Sr and Nd isotopic evidence for sources of southern African
Cretaceous kimberlites. Nature 304, 51–54.
Smith, C.B., Allsopp, H.L., Kramers, J.D., Hutchinson, G., Roddick, J.C., 1985. Emplacement
ages of Jurassic – Cretaceous South African kimberlites by the Rb–Sr method on
phlogopite and whole rock samples. Transactions of the Geological Society of South
Africa 88, 249–266.
Smith, C.B., Gurney, J.J., Harris, J.W., Otter, M.L., Kirkley, M.B., Jagoutz, E., 1991.
Neodymium and strontium isotope systematics of eclogite and websterite
paragenesis inclusions from single diamonds, Finsch and Kimberley Pool, RSA.
Geochimica et Cosmochimica Acta 55, 2579–2590.
Snyder, G.A., Taylor, L.A., Crozaz, G., Halliday, A.N., Beard, B.L., Sobolev, V.N., Sobolev,
N.V., 1997. The origins of Yakutian eclogite xenoliths. Journal of Petrology 38,
85–113.
Sobolev, N.V., Yefimova, E.S., Channer, D.M. DeR, Anderson, P.F.N., Barron, K.M., 1998a.
Unusual upper mantle beneath Guaniamo, Guyana shield, Venezuela: evidence
from diamond inclusions. Geology 26, 971–974.
Sobolev, N.V., Snyder, G.A., Taylor, L.A., Keller, R.A., Yefimova, E.S., Sobolev, V.N.,
Shimizu, N., 1998b. Extreme chemical diversity in the mantle during eclogitic
diamond formation: evidence from 35 garnet and 5 pyroxene inclusions in a single
diamond. International Geology Review 40, 567–578.
Stachel, T., Harris, J.W., Brey, G.P., 1999. REE patterns of peridotitic and eclogi-
tic inclusions in diamonds from Mwadui (Tanzania), volume 2. Proceedings of
the Second International Kimberlite Conference, Red Roof Design, Cape Town,
pp. 829–835.
Stachel, T., Brey, G.P., Harris, J.W., 2000. KanKan diamonds (Guinea) 1: from the
lithosphere down to the transition zone. Contributions to Mineralogy and
Petrology 140, 1–15.
Stachel, T., Aulbach, S., Brey, G.P., Harris, J.W., Leost, I., Tappert, R., Viljoen, K.S., 2004.
The trace element composition of silicate inclusions in diamonds: a review. Lithos
77, 1–19.
Stachel, T., Harris, J.W., 2008. The origin of cratonic diamonds — constraints from
mineral inclusions. Ore Geology Reviews 34, 5–32.
Stachel, T., Harris, J.W., 2009. Formation of diamond in the earth's mantle. Journal of
Physics: Condensed Matter 21, 364206 10pp.
Staudigel, H., 2004. Hydrothermal alteration processes in the oceanic crust. In: The
crust (Ed. R.L. Rudnick), Treatise on Geochemistry (Eds. H.D. Holland & K.K.
Turekian), volume. 3, pp. 511 – 535.
Tappert, R., Stachel, T., Harris, J.W., Muehlenbachs, K., Ludwig, T., Brey, G.P., 2005a.
Subducting oceanic crust: the source of deep diamonds. Geology 33, 565–568.
Tappert, R., Stachel, T., Harris, J.W., Muehlenbachs, K., Ludwig, T., Brey, G.P., 2005b.
Diamonds from Jagersfontein (South Africa): messengers from the sublithospheric
mantle. Contributions to Mineralogy and Petrology 150, 505–522.
Tappert, R., Stachel, T., Harris, J.W., Muehlenbachs, K., Brey, G.P., 2006. Placer diamonds
from Brazil: indicators of the composition of the earth's mantle and the distance to
their kimberlite source. Economic Geology 101, 453–470.
Taylor, L.A., Snyder, G.A., Crozaz, G., Sobolev, V.N., Yefimova, E.S., Sobolev, N., 1996.
Eclogitic inclusions in diamonds: evidence of complex mantle processes over time.
Earth and Planetary Science Letters 142, 535–551.
Taylor, L.A., Anand, M., 2004. Diamonds: time capsules from the Siberian mantle.
Chemie der Erde 64, 1–74.
Viljoen, K.S., Schulze, D.J., Quadling, A.G., 2005. Contrasting Group I and Group II
eclogite xenolith petrogenesis: petrological, trace element and isotopic evidence
from eclogite, garnet-websterite and alkremite xenoliths in the Kaalvallei
kimberlite, South Africa. Journal of Petrology 46, 2059–2090.
Viljoen, F., Dobbe, R., Smit, B., 2009. Geochemical processes in peridotite xenoliths from
the Premier diamond mine, South Africa: evidence for the depletion and
refertilisation of subcratonic lithosphere. Lithos 112, Supplement 2, 1133–1142.