ELSEVIER Earth and Planetary Science Letters 164 (1998) 353–369

A multielement geochronologic study of the Great Dyke, Zimbabwe:

significance of the robust and reset ages
S.B. Mukasa a,Ł , A.H. Wilson b , R.W. Carlson c
a Departmentof Geological Sciences, University of Michigan, Ann Arbor, MI 48109-1063, USA
b
Department of Geology and Applied Geology, University of Natal, Durban 4041, South Africa
c Department of Terrestrial Magnetism, Carnegie Institution of Washington, 5241 Broad Branch Rd., Washington DC 20015, USA

Received 26 January 1998; revised version received 6 August 1998; accepted 8 September 1998

Abstract

New Sm–Nd, U–Pb, and Pb–Pb age determinations indicate that the Great Dyke of Zimbabwe, an elongate intrusion
of mafic and ultramafic rocks some 550 km long and between 3 and 10 km wide, is over 100 Ma older than previously
believed based on Rb–Sr ages. The intrusion was emplaced as a series of subchambers with similar stratigraphy,
comprising a lower ultramafic sequence with cyclic layering of dunite or harzburgite grading upwards into bronzitite, the
top sections of which include Pt-enriched sulfide zones, and an upper mafic sequence of pyroxenites capped by olivine
gabbro and gabbronorite. The Sm–Nd method has yielded a combined mineral=whole-rock isochron of 2586 š 16 Ma
and "Nd .t/ of C1.1 for samples from the Darwendale, Sebakwe, and Wedza Subchambers as well as the satellite East
Dyke. This isochron age is in excellent agreement with the U–Pb age for three concordant rutile fractions extracted
from a feldspathic pyroxenite of the Selukwe Subchamber with an error-weighted mean at 2587 š 8 Ma. Two zircon
fractions from the same feldspathic pyroxenite sample as the rutile are discordant, and although not well constrained,
suggest Pb loss from the zircons at ca. 830 Ma. This may be related to the onset of the widespread and diachronous
Pan-African tectonothermal event in southern Africa. Whole–rock samples and clinopyroxene and plagioclase separates
from a Darwendale Subchamber drill core yielded a 207 Pb=204 Pb vs. 206 Pb=204 Pb isochron age of 2596 š 14 Ma, which
is in agreement with the Sm–Nd isochron and the rutile U–Pb crystallization age. This new age information shows that
emplacement of the Great Dyke and its satellite dikes closely followed the amalgamation of the Kaapvaal and Zimbabwe
Cratons, and was contemporaneous with emplacement of the youngest of the trondhjemite–tonalite–granodiorite granitoid
suite in the Zimbabwe Craton. Assuming that amalgamation of the Kaapvaal and Zimbabwe Cratons was largely by
NNW-directed convergence, it follows that the source of the Great Dyke was asthenospheric mantle hydrated and enriched
in incompatible elements by subduction processes. Isochrons of 206 Pb=204 Pb vs. 238 U=204 Pb and 207 Pb=204 Pb vs. 235 U=204 Pb
yield ages with large errors, but well constrained initial Pb ratios (206 Pb=204 Pb D 14:15 š 0:30 and 207 Pb=204 Pb D
15:04 š 0:06). Assuming a two-stage model for common lead evolution, this result yields a ¼ value of 9.5. Along with
the calculated initial Sr and Nd isotopic compositions, these data are consistent with derivation of the Great Dyke magmas
by large volume melting of a mantle that has been hydrated and enriched by subduction. While a small amount of crustal
contamination of magma derived from depleted mantle could produce the composition of the Great Dyke, the uniformity
of initial ratios between subchambers supports the notion of enrichment in incompatible elements being an intrinsic
characteristic of the mantle source.  1998 Elsevier Science B.V. All rights reserved.

Keywords: Great Dyke; layered intrusions; isotopes; absolute age; mantle; Archean

Ł Corresponding author. Fax: C1 313 763 4690; E-mail: [email protected]

0012-821X/98/$ – see front matter  1998 Elsevier Science B.V. All rights reserved.
PII: S 0 0 1 2 - 8 2 1 X ( 9 8 ) 0 0 2 2 8 - 3
354 S.B. Mukasa et al. / Earth and Planetary Science Letters 164 (1998) 353–369

1. Introduction

The Great Dyke is a major intrusion of mafic

and ultramafic rocks that cuts across the dominantly
Archean rocks of the Zimbabwe Craton and is re-
garded as a major magmatic event that marks the
Archean–Proterozoic boundary in that province [1].
Based on Nd model ages, the Archean craton of Zim-
babwe appears to have undergone two episodes of
crustal generation at 3.5 and 2.9 Ga [2] by the devel-
opment of dominantly tonalite–trondhjemite intru-
sions subsequently metamorphosed to orthogneisses.
These granitoid domains were the basement to the
major greenstone belt sequences and comprise up to
60% of preserved Archean crust in Zimbabwe [3].
The craton is bounded by the Zambezi metamorphic
belt in the north, the Mozambique belt to the east
and the Limpopo belt to the south. Proterozoic and
Phanerozoic basin deposits cover the Archean rocks
in the northern part of the craton (Fig. 1).
The earliest widespread formation of greenstones
in Zimbabwe, together with associated granitoid
suites, was the Lower Greenstone sequences of the
Belingwean Group emplaced at 2.9–2.8 Ga. These
comprised mafic volcanic sequences in the lower
Fig. 1. Geological setting of the Great Dyke in relation to
parts passing upwards into a dominantly bimodal the Zimbabwe Archean craton, the older and younger granitoid
succession with alternating mafic and felsic volcanic bodies, and the 2.6 Ga Chilimanzi granitoid suite
rocks. The upper part of the Belingwean Group is a
komatiite and mafic rock volcanic sequence that also
incorporates sedimentary rocks [4]. An unconformity tercept is 2:66 š 17 Ga [4]. Intrusive into the Lower
separates the Upper Belingwean from the base of the Greenstone successions are a number of granitoid
Upper Greenstone succession. The latter is called bodies, some of which may be correlated with felsic
the Lower Bulawayan and consists of intermediate volcanism [4].
to felsic volcanic rocks and volcaniclastic sediments The younger granites post-date the Upper Green-
yielding single zircon U–Pb ages of 2.83–2.79 Ga stones and the Shamvaian Group and consist of
[4]. Where not in contact with Belingwean rocks it two groups [7]: (1) the earlier tonalite–granodiorite
is intruded by granitoids. The extensive Upper Bu- Sesombi suite was dated at approximately 2.7 Ga
lawayan Group includes clastic sediments, argillites [5,8] and more recently has been dated at 2673 š 5
and limestones together with basalts and komatiitic Ma [9]; (2) the later group comprises tabular mon-
basalts. The Upper Bulawayan Group has yielded zogranites of large areal extent and generally simi-
Rb–Sr whole rock ages of 2:66 š 7 to 2:48 š 14 Ga lar composition. This craton-wide multiphase grani-
[5,6]. Single zircon U–Pb dating [4] shows the Upper toid Chilimanzi suite marks the last major pre-Great
Bulawayan to have been emplaced at 2.70–2.64 Ga. Dyke granitic event [4]. The post-tectonic Murahwa
The locally occurring Shamvaian Group comprises granite (north), also associated with the Chilimanzi
sedimentary associations intercalated with felsic vol- granitoid suite, has been dated at 2601 š 14 [10] uti-
canic rocks and is the youngest of the greenstone lizing U–Pb, which agrees well with the Rb–Sr ages
successions but single zircon U–Pb determinations of 2574 š 14 Ma [11] and 2583 š 52 Ma [12] for the
yielded an array for which the concordia upper in- southern area of this large body near Masvingo. The
 S.B. Mukasa et al. / Earth and Planetary Science Letters 164 (1998) 353–369 355

compositionally and tectonically equivalent Glen-

dale tonalite in the north has been dated using U–
Pb at 2618 š 6 Ma. These post-tectonic granitoids
are characterized by relatively high Sr initial ratios
(0:7040 š 10 to 0:706 š 14) and were generated
through remobilization of basement granitoid [10].
The Great Dyke is the first magmatic event in the
stabilized Zimbabwe Craton, and therefore, new age
data on this major intrusion have important impli-
cations for the stabilization of the crust as well as
the intrusion’s temporal relation to the prevailing
tectonism and extensive late granitoid magmatism.
Hence the main purpose of this study is to determine
the time of crystallization for the Great Dyke using
the Sm–Nd and U–Pb methods and to compare the
results to our new and previously published Rb–Sr
ages.

2. General geology of the Great Dyke

The Great Dyke (Fig. 2), aligned approximately

NNE, is 550 km in length and 3 to 11 km wide
[13]. Parallel to the intrusion and almost over the
same length are a number of gabbro or quartz gab-
bro satellite dikes. The most prominent of these Fig. 2. Map showing the locations of the samples, and the
are the Umvimeela and East Dykes situated on the subdivision of the Great Dyke into chambers and subchambers.
west and east sides of the Great Dyke, respectively,
and between 3 and 25 km distance from it. These
satellite dikes are closely associated with a major The Great Dyke is longitudinally subdivided into
fracture pattern postulated to be the result of the a series of narrow contiguous layered chambers and
craton-wide tectonic control which gave rise to the subchambers (Fig. 2). Three magma chambers have
form and alignment of the major Great Dyke in- been recognized with further subdivision totaling
trusion [14]. Closely associated with the East Dyke five subchambers on the basis of structure, style
are fine-grained mafic rocks [15] regarded as the of layering and continuity of layers [18–20]. The
best estimates of the Great Dyke initial magma. The Mvuradonha Chamber is at the extreme north of
Southern Satellite Dyke complex is a series of nar- the Great Dyke. South of it is the North Chamber
row intrusions almost in line with the Great Dyke, which is subdivided into the Darwendale and Se-
and in contrast to the East and Umvimeela Dykes bakwe Subchambers. The South Chamber comprises
comprises mainly olivine bearing feldspathic pyrox- the northern Selukwe Subchamber and the southern
enites. The Southern Satellite Dyke complex shows Wedza Subchamber. Geophysical evidence indicates
evidence of the tectonothermal Limpopo Metamor- that the North Chamber has a much greater vol-
phic event at 1950 Ma, a compressional phase of ume than the South Chamber [14] and also has the
deformation best developed in the Limpopo Belt im- greatest stratigraphic thickness of 3500 m.
mediately to the south [16]. The northern end of The transverse structure of the Great Dyke is
the Great Dyke has been rotated and deformed as it shown by gravity studies to have inward dipping
impinged upon the ca. 550 Ma Pan-African Zambezi walls with a flared upper layered structure and a
Metamorphic Belt [17]. lower steep-sided deep section that may include a
356 S.B. Mukasa et al. / Earth and Planetary Science Letters 164 (1998) 353–369

feeder system [20,21]. The lateral extensions of the is a magnesian tholeiite containing 15.6% MgO and
upper part of the intrusion, now completely eroded, 52.8% SiO2 [23].
may have extended beyond the present wall rock
boundaries.
The synclinal layered succession of the Great 3. Previous isotopic studies on the Great Dyke
Dyke is divided into a lower Ultramafic Sequence
and an upper Mafic Sequence. The Ultramafic Se- All previous age determinations of the Great Dyke
quence is characterized by the development of cyclic were carried out using Rb–Sr (all Rb–Sr ages dis-
units in which the lower Dunite Succession is dom- cussed here were recalculated for an 87 Rb–decay
inated by olivine cumulates bounded by chromitite constant of 1:42 ð 10 11 yr 1 ). The first isotopic
layers. Cyclic units in the upper Pyroxenite Suc- investigation of the Great Dyke was by Faure et
cession have a dunite or harzburgite layer which al. [24] which suggested an equivalent age to the
grades through an olivine pyroxenite into the over- Bushveld Complex of 1950 š 150 Ma obtained by
lying orthopyroxenite unit. The uppermost cyclic Nicolaysen et al. [25]. Allsopp [26] on the basis of
unit representing the top of the Ultramafic Sequence Rb–Sr and Ar–Ar age determination on two biotites
is a complexly layered association of chromitites, set a lower limit of 2476 š 30 Ma for the layered
harzburgites and pyroxenites and is of economic sequence. Davies et al. [27] reported the first Rb–Sr
importance in that it hosts several sulfide zones en- isochron age on the Great Dyke as 2477 š 90 Ma
riched in the platinum group elements. The cyclic with an initial 87 Sr=86 Sr ratio of 0:7024 š 8 based
units are explained by repeated injection of magma on eight whole–rock determinations. Samples of the
into the chamber with the overall composition be- Southern Satellite Dyke Complex yielded an age of
coming gradually more evolved [14]. 2545 š 120 Ma with an initial 87 Sr=86 Sr ratio of
The Mafic Sequence is marked by the first appear- 0:7049 š 12 [28]. This important age determination
ance of cumulus plagioclase and is subdivided into indicated that the Southern Satellite Dyke complex
three major units called the Lower, Middle and Up- was coeval and cogenetic with the Great Dyke, but
per mafic Successions [22]. These rocks are mainly the high initial ratio is indicative of crustal contami-
norites and gabbronorites with olivine bearing gab- nation by evolved crustal material.
bros marking the base of the Lower and Middle The most recent age determination of the Great
Mafic Successions. The Upper Mafic Succession is Dyke was by Hamilton [29] whose Rb–Sr study
marked by the presence of inverted pigeonite. of five-whole rock and selected mineral separates
The Great Dyke does not have a chill margin and yielded an age more precise than earlier studies
estimates of the composition of the initial magma at 2455 š 16 Ma with an initial 87 Sr=86 Sr ratio of
are based on the chill zone of an offshoot of the 0:70261 š 4. (The original age of 2514 Ma has been
cogenetic East Dyke and indirect evidence from geo- recalculated using the currently accepted 87 Rb decay
chemical projections of the layered succession. In constant). The exact locations and lithologies of
common with many other layered intrusions of this the samples, four from the Darwendale Subchamber
type and age, the Great Dyke magma was charac- and one from the Wedza Subchamber, and the rock
teristically high in SiO2 and MgO. This is evident types are not described in that paper. However, on
from the highly magnesian olivine and pyroxene the basis of matching Rb and Sr concentrations
compositions (Mg# 92), the relative abundance of with other studies on the Great Dyke, two samples
orthopyroxenites in the Ultramafic Sequence and the from the Darwendale Subchamber are indicated to
modeled crystallization path [14,15]. The apparent be gabbronorites and two pyroxenites. The sample
disparity of mafic to ultramafic rocks may also be from the Wedza Subchamber is indicated to be a
indicative of the generally ultramafic nature of the pyroxenite. Four mineral separates of plagioclase and
Great Dyke magma but this also has to be seen in clinopyroxene from ultramafic rocks in the Wedza
the context of repeated emplacement of magma dur- Subchamber are concordant with the whole rock
ing the early stages of development of the magma data. The mineral separates were from the Wedza
chamber [14]. The best estimate of the initial magma Subchamber but there is also no indication as to
 S.B. Mukasa et al. / Earth and Planetary Science Letters 164 (1998) 353–369 357

whether these were from surface samples or drill

core. The consistency of the data is remarkable given
the wide geographic spread of the samples and also
that they represent both the Mafic and Ultramafic
Sequence. Until our study, only a single sample of
the Great Dyke had ever been analyzed with the
Sm–Nd method, and this yielded a depleted mantle
model age of 3.0 Ga [30].
The present study applies Sm–Nd, U–Pb and
Pb–Pb isotopic dating to the Great Dyke to assess
the robustness of the Rb–Sr age of Hamilton [29]
which is the most widely accepted to date, and also
to investigate the homogeneity of the intrusion us-
ing whole rock and mineral samples from different
stratigraphic and geographic localities. Descriptions
of the samples used for this isotopic study are pro-
vided in the Appendix (EPSL Online Background
Dataset 1 ) and the sample localities are marked on
the map in Fig. 2 and the stratigraphic column in
Fig. 3.

4. Analytical procedures

Samples were chosen as those having the least

alteration. Rocks of the Great Dyke show strong
Fig. 3. Samples used in the present study shown on a relative
alteration in general and care must be exercised in stratigraphic basis. The section shown for reference (with appro-
their selection. In particular the gabbroic rocks show priate thicknesses of rocks) is from the Darwendale Subchamber,
some alteration of the feldspars even in least altered but all subchambers have the same sequence of rock types.
samples. Pyroxenite may show pervasive and intense
alteration even in drill core from depths up to 400
m. This is generally the result of the high magnesian plished with HCl-based ion chromatography while
nature of the rocks which renders them susceptible that for rutile used a two-stage HBr and HCl cation-
to alteration and also the influence of faulting. exchange column chemistry procedure described by
Rutile and zircon fractions were obtained using Parrish et al. [33].
standard separation techniques the details to which Samples for the Sm–Nd, Rb–Sr and Pb–Pb
are described by Krogh [31], Parrish [32], and Par- isochrons were crushed in steel and milled in carbon
rish et al. [33]. Zircon grains were abraded for 48 steel. For the analyses at the University of Michi-
hours to minimize the effects due to surface-corre- gan, standard dissolution and column procedures
lated Pb loss, and then washed (like the rutile) in described by Mukasa et al. [34,35] were followed.
15% HNO3 and ultra-pure water. They were then Each sample was then dried, treated with a drop of
weighed into Teflon microcapsules and spiked with 14 N HNO3 , re-dried and then loaded on appropriate
205
Pb–235 U tracer. Both minerals were dissolved in filaments (single rhenium for U, Pb, Rb, and Sr,
an HF–HNO3 acid solution at 240ºC for 72 hours. and triple tantalum–rhenium–tantalum for Nd and
Separation of Pb and U from zircon was accom- Sm). Lead was loaded with a silica gel–phosphoric
acid solution, Sr with tantalum tetrachloride, and Rb,
1http:www.elsevier.nl=locate=epsl, mirror site: Sm and Nd with a 10% nitric acid solution. The
http:==www.elsevier.com=locate=epsl samples were run on VG Sector thermal ionization
 358
S.B. Mukasa et al. / Earth and Planetary Science Letters 164 (1998) 353–369
Table 1
Rb–Sr, Sm–Nd and Pb–Pb isotopic and concentration data for mineral separates and whole rock samples from core GDX17 from the Darwendale Subchamber of the Great Dyke

Sample and material Rock type Rb Sr 87 Rb=86 Sr 87 Sr=86 Sr Sm Nd 147 Sm=144 Nd 143 Nd=144 Nd "Nd .t/ U Pb 206 Pb=204 Pb 207 Pb=204 Pb208 Pb=204 Pb
analyzed (ppm) (ppm) (ppm) (ppm) (2.6 Ga) (ppm) (ppm)
325.85–326.15 Cpx Gabbronorite 0.390 10.76 0.10426 0.70606 š 1 1.83 5.61 0.1971 0.512621 š 11 0.5 0.121 0.255 36.090 18.835 57.862
325.85–326.15 Plag Gabbronorite 3.850 343.2 0.03227 0.70337 š 1 0.10 0.739 0.0795 0.510458 š 11 3.5 0.014 1.678 14.355 15.076 33.916
325.85–326.15 WR Gabbronorite 5.311 229.2 0.06668 0.70449 š 1 0.54 1.95 0.1658 0.512071 š 13 0.8 0.151 1.225 17.621 15.637 36.375
619.85–620.15 Cpx Gabbronorite 0.272 11.41 0.06850 0.70461 š 1 0.73 1.83 0.2418 0.513341 š 13 1.4 0.072 0.148 31.389 18.102 45.207
619.85–620.15 Plag Gabbronorite 2.374 304.8 0.02241 0.70379 š 1 0.05 0.310 0.1004 0.510963 š 27 0.6 0.013 0.657 14.815 15.194 34.070
619.85–620.15 WR Gabbronorite 3.026 235.5 0.03698 0.70472 š 2 0.16 0.534 0.1842 0.512321 š 18 2.1 0.035 0.675 17.360 15.788 35.176
627.90–628.10 Cpx Websterite 0.476 12.34 0.11095 0.70654 š 1 1.25 4.15 0.1827 0.512382 š 13 0.4 0.091 0.418 20.638 16.195 39.272
627.90–628.10 WR Websterite 5.490 35.58 0.44465 0.71825 š 1 0.85 2.86 0.1801 0.512371 š 15 C0.3 0.213 1.696 17.640 15.653 36.619
649.90–650.10 Cpx Websterite 0.367 13.45 0.07847 0.70528 š 1 0.93 2.86 0.1962 0.512614 š 13 0.3 0.089 0.467 19.038 15.883 37.528
649.90–650.10 WR Websterite 1.719 21.10 0.23456 – 0.63 1.93 0.1967 0.512616 š 10 0.5 0.095 1.469 15.831 15.337 35.108
651.91–652.11 WR Orthopyroxenite 3.993 40.89 0.28126 0.71272 š 1 0.28 1.07 0.1606 0.511972 š 13 1.0 0.185 2.103 16.759 15.532 35.711

Analytical errors for 87 Sr=86 Sr and 143 Nd=144 Nd are based on 2¦m of the within-run statistics. Errors for the Pb isotopic ratios are 0.1%, based on replicate analyses of NIST standard
SRM-981. Abbreviations for the material analyzed are as follows: Cpx, clinopyroxene; Plag, plagioclase; and WR, whole rock. The εNd values were calculated using the 143 Nd=144 Nd value
for chondritic uniform reservoir (CHUR) at t D 2:6 Ga.
 S.B. Mukasa et al. / Earth and Planetary Science Letters 164 (1998) 353–369 359

mass spectrometers. All elements were run statically,

except for Nd and Sr, which were run multidynam-
ically. Lead isotopic compositions are corrected for
fractionation using a factor of 0:12 š 0:02% per
atomic mass unit, based on replicate analyses of
the US National Institute of Standards and Technol-
ogy (NIST) Standard NBS-981. Nd and Sr ratios
were normalized to 146 Nd=144 Nd D 0.721900 and
86
Sr=88 Sr D 0.119400, respectively. Measurements
for the NIST Standard SRM-987 give 87 Sr=86 Sr D
0:710245 š 10, and for the La Jolla Nd Standard val-
ues of 143 Nd=144 Nd D 0:511842 š 10. Total blanks
averaged 0.04 ng for Pb, 0.02 ng for Nd, 0.02 ng for
Sm, 0.07 ng for Rb and 0.1 ng for Sr.
For Sm–Nd analyses at the Department of Ter-
restrial Magnetism (DTM), Carnegie Institution of
Fig. 4. Rb–Sr mineral=whole-rock isochron for drill core sample
Washington, 200 mg of powdered whole rock sample GDX17 from the Darwendale Subchamber. Note that most of the
or mineral separate was decomposed in a HF–HNO3 scatter occurs in plagioclase separates and gabbronorite whole-
mixture in a sealed Teflon vessel after the addition of rock samples from shallow levels in the intrusion, probably reset
isotopically enriched tracers 149 Sm and 150 Nd. Proce- by hydrothermal processes.
dures for chemical separation and mass spectrometry
followed those described by Wilson and Carlson
[36]. Nd, measured as NdOC on the DTM 38-cm ment with the age of 2455 š 16 Ma age determined
radius single collector instrument, is fractionation by Hamilton [29]. These ages, however, are consid-
corrected to 146 NdOC =144 NdOC D 0.722250 which erably lower than the Sm–Nd, U–Pb and Pb–Pb ages
is equivalent to 146 Nd=144 Nd D 0.7219 when Nd is discussed below and are therefore judged to be reset.
run as a metal (NdC ). Both sets of Nd data are re- Also, the initial Sr ratios for this new errorchron
ported relative to a value of 143 Nd=144 Nd D 0.511860 and the Hamilton [29] isochron (0:7026 š 4 and
for the La Jolla Nd standard. During the course of 0:70261 š 4, respectively) are indistinguishable.
the study, measured procedural blanks were Sm D 6
pg and Nd D 50 pg. 5.2. Sm–Nd

Using whole rock samples of gabbro, olivine gab-

5. Results of the isotope systematics bro, and pyroxenite from the Darwendale, Sebakwe
and Wedza Subchambers of the layered intrusion,
5.1. Rb–Sr an olivine basalt chill zone of the satellite East
Dyke, and layered intrusion mineral separates of
Rb–Sr data for whole–rock samples of orthopy- clinopyroxene, orthopyroxene, and plagioclase (Ta-
roxenite, gabbronorite and websterite and mineral ble 2), a Sm–Nd isochron of 2586 š 16 Ma and
separates of clinopyroxene and plagioclase from the an initial 143 Nd=144 Nd value of 0:50934 š 4 ("Nd .t/
GDX17 drill core of the Darwendale Subchamber D C1:1 š 0:8) have been generated for the Great
(Table 1) cannot be fitted to a straight line if it Dyke (Fig. 5a). Exclusion of the East Dyke sample
is assumed that all scatter is due to analytical er- changes the age by only 0.1% to 2589š16 Ma which
ror. If some of the scatter is attributed to normally is inconsequential. The isochron age of 2586 š 16
distributed errors in the initial 87 Sr=86 Sr, then an er- Ma is therefore interpreted to represent crystalliza-
rorchron of 2467 š 85 Ma and an initial Sr ratio of tion inasmuch as the data points are highly co-linear
0:7026 š 4 are produced (Fig. 4). Although highly (MSWD D 4.5), and their Sm=Nd values cover a
imprecise, this age determination is broadly in agree- wide range. Whole rock Nd in samples from the lay-
360 S.B. Mukasa et al. / Earth and Planetary Science Letters 164 (1998) 353–369

Table 2
Sm–Nd isotopic and concentration data for mineral separates and whole rock samples for the Great Dyke and its satellite East Dyke
plotted on the isochron in Fig. 5a

Sample and material Rock type Sm Nd 147 Sm=144 Nd 143 Nd=144 Nd "Nd .t/
analyzed (ppm) (ppm) (2.6 Ga)
Sel Brnz Opx Orthopyroxenite 0.314 0.870 0.2190 0:513074 š 10 C1.1
East Dyke WR Olivine basalt 2.134 9.627 0.1340 0:511633 š 19 C1.3
Wedza WR Gabbro 0.852 3.963 0.1300 0:511549 š 22 C1.0
HC38 407 WR Websterite 1.237 4.360 0.1715 0:512252 š 14 C0.9
HC38 402 Opx Gabbro 0.519 2.149 0.1459 0:511835 š 18 C1.3
P1=48 crse Opx Coarse websterite 0.0485 0.1230 0.2385 0:513402 š 21 C1.0
P1=48 crse Cpx Coarse websterite 1.002 3.421 0.1770 0:512360 š 12 C1.1
P1=48 fine Opx Fine websterite 0.0202 0.0389 0.3137 0:514724 š 21 C1.8
P1=48 fine Cpx Fine websterite 0.6019 1.616 0.2252 0:513195 š 18 C1.4
ND15 WR Olivine gabbro 0.1571 0.4417 0.2150 0:512983 š 29 C0.7
ND15 Cpx Olivine gabbro 0.5640 1.343 0.2538 0:513629 š 20 C0.4
ND15 Opx1 Olivine gabbro 0.0347 0.0625 0.3351 0:51506 š 20 C1.1
ND15 Opx2 Olivine gabbro 0.0386 0.0706 0.3302 0:51517 š 51 C5.0
ND15 Plag1 Olivine gabbro 0.0365 0.2189 0.1008 0:511068 š 22 C1.3
ND15 Plag 2 Olivine gabbro 0.0399 0.2390 0.1008 0:511107 š 32 C2.1

Analytical errors for 143 Nd=144 Nd are based 2¦m of the within-run statistics. Abbreviations for the material analyzed are as follows:
Cpx, clinopyroxene; Opx, orthopyroxene; Plag, plagioclase; and WR, whole rock. The " Nd values were calculated using the 143 Nd=144 Nd
value for chondritic uniform reservoir (CHUR) at t D 2:6 Ga.

ered intrusion ranges from 0.44 to 4.4 ppm while the 5.3. Common Pb and U–Pb isochrons
chill zone of the satellite East Dyke has a concen-
tration of 9.63 ppm. Whole–rock Sm concentrations The whole–rock samples and mineral separates
vary sympathetically in the layered intrusion samples from drill core GDX17 were also analyzed for their
between 0.16 and 1.2 ppm, and reach 2.13 ppm in Pb isotopic compositions and U and Pb concen-
the East Dyke. Clinopyroxene and plagioclase sep- trations (Table 1). The 207 Pb=204 Pb vs. 206 Pb=204 Pb
arates have reasonably high concentrations of both isochron age obtained of 2596 š 14 Ma (Fig. 6a)
Nd and Sm (0.22–3.4 ppm and 0.04–1.0 ppm, re- is in remarkably good agreement with the two Sm–
spectively), allowing precise isotopic measurements. Nd isochrons and the rutile U–Pb data discussed
However, orthopyroxene from gabbro sample ND15 below. Isochrons of 206 Pb=204 Pb vs. 238 U=204 Pb and
207
has generally low concentrations of Nd (<0.07 ppm) Pb=204 Pb vs. 235 U=204 Pb (Fig. 6b,c) yield ages
and Sm (<0.04 ppm) resulting in poor precision for (ca. 2.2–2.6 Ga) with large errors, but well con-
the measured isotopic ratios and has therefore been strained initial Pb ratios (206 Pb=204 Pb D 14:15 š 0:30
excluded from the isochron. and 207 Pb=204 Pb D 15:04 š 0:06) that are indis-
A second Sm–Nd age determination using whole tinguishable from those of the feldspar cogenetic
rock samples of bronzitite, gabbronorite, and web- with rutile and zircon in sample MR89 (206 Pb=204 Pb
sterite and their mineral separates of clinopyrox- D 14:193 š 14, 207 Pb=204 Pb D 15:056 š 23, and
208
ene and plagioclase all from drill core GDX17 in Pb=204 Pb D 33:889 š 34). These values require
the Darwendale Subchamber (Table 1 and Fig. 5b) the source materials for the Great Dyke magmas to
yielded an isochron age of 2572 š 43 Ma and an have evolved with a higher ¼ value than that for
initial 143 Nd=144 Nd value of 0:50926 š 14 ("Nd .t/ depleted mantle or bulk earth.
D 0:46 š 2:7). This result is in agreement with
the Sm–Nd isochron presented above, but lacks 5.4. Rutile and zircon U–Pb
high precision because of the smaller spread in
147
Sm=144 Nd and the relatively high MSWD of Rutile and zircon U–Pb data for a feldspathic
10.7. pyroxenite at 186.79–187.53 m depth of drill core
 S.B. Mukasa et al. / Earth and Planetary Science Letters 164 (1998) 353–369 361

tile and zircon are magmatic rather than xenocrystic.

For example, these minerals are invariably associated
with pockets of late stage crystallization assemblages
(biotite, K-feldspar, quartz and apatite) and are not
observed on pyroxene boundaries. Relevant to the
origin of zircon grains, Zr contents of whole rock
samples are 20–50 ppm compared with 3–4 ppm
in the cumulus minerals, correlating well with ele-
ments such as K, P, Rb and Y, and providing a good
measure for the amount of trapped liquid. Such a re-
lationship would not be observed if the zircons were
xenocrystic. Supporting this interpretation is the fact
that the concordia upper intercept age obtained from
the rutile and zircon fractions is identical to the
Sm–Nd isochron age generated from the principal
rock constituents. Three fractions of rutile based on
crystal size (45–75, 75–100, and 100–140 µm) were
recovered and these are concordant yielding an er-
ror-weighted mean for the U–Pb ages of 2587 š 8
Ma. This age is virtually identical to the age cal-
culated for the better of the two Sm–Nd isochrons
presented here (Fig. 5a), and is interpreted to repre-
sent extremely rapid cooling. Three zircon fractions
from the same feldspathic pyroxenite sample are dis-
cordant, and although not well constrained, yield a
lower concordia intercept that suggests Pb loss from
the zircons at 816 š 24 Ma.

6. Discussion

Fig. 5. (a) Sm–Nd isochron determined at DTM using whole 6.1. Significance of the Rb–Sr ages
rock samples of gabbro, olivine gabbro, and pyroxenite from
the Darwendale, Sebakwe and Wedza Subchambers of the Great There is fairly good agreement between the Rb–
Dyke, an olivine basalt chill zone of the satellite East Dyke, and Sr isochron of 2467 š 85 Ma presented here (Table 1
Great Dyke mineral separates of clinopyroxene, orthopyroxene, and Fig. 4) and decay constant-adjusted isochrons of
and plagioclase. The isochron age of 2586 š 16 Ma is interpreted
to represent the time of crystallization. (b) Sm–Nd isochron 2455 š 16 Ma and 2477 š 90 Ma by Hamilton [29]
determined at the University of Michigan using whole rock and Davies et al. [27], respectively. Our isochron and
samples of bronzitite, gabbronorite, and websterite and their the one by Davies et al. [27] show similar scatter
mineral separates of clinopyroxene and plagioclase all from and imprecision compared to the isochron produced
drill core GDX17 in the Darwendale Subchamber. The isochron by Hamilton [29]. However, these ages are at least
yields an age of 2572 š 43 Ma, which is less precise largely
because of the poor spread in 147 Sm=144 Nd and the relatively 100 million years younger than ages obtained by the
high MSWD of 10.7. Sm–Nd isochron and rutile U–Pb methods. Sm–Nd
ages can appear artificially old because of minor
crustal contamination that would be less obvious in
MR89 from the Darwendale Subchamber are sum- the Rb–Sr data inasmuch as the spread in Rb=Sr
marized in Table 3 and plotted on a concordia dia- values of the rocks used to make the isochron is
gram in Fig. 7. There is ample evidence that both ru- bigger compared to the spread in Sm=Nd values.
 362
S.B. Mukasa et al. / Earth and Planetary Science Letters 164 (1998) 353–369
Table 3
Rutile and zircon U–Pb data (in columns) and feldspar common-Pb isotopic compositions a for drill core MR89 from the Darwendale Subchamber of the Great Dyke

Size (µm) Mass (mg) U (ppm) Pb (ppm) Measured ratios Atomic ratios Apparent ages (Ma)
206 Pb=204 Pb 207 Pb=206 Pb 208 Pb=206 Pb 206 Pb* =238 U 207 Pb* =235 U 207 Pb* =206 Pb* 206 Pb* =238 U207 Pb* =235 U207 Pb* =206 Pb*
Rutile fractions
100–140 0.93 2.47 1.56 264.5 0.21851 0.16130 0.5037 (59) 11.86 (16) 0.1708 (12) 2630 2594 2566 š 12
75–100 1.68 2.66 1.44 1182 0.18234 0.07540 0.4929 (27) 11.668 (70) 0.17169 (30) 2583 2578 2574 š 3
45–75 0.99 2.86 1.51 2254 0.17763 0.05516 0.4949 (43) 11.74 (10) 0.17204 (44) 2592 2584 2578 š 5

Zircon fractions
75–100 0.08 42.7 74.8 299.7 0.20104 4.0870 0.3650 (35) 7.970 (8) 0.15838 (70) 2006 2228 2438 š 7
45–75 0.05 47.4 62.9 6242 0.16070 3.9421 0.3301 (25) 6.889 (6) 0.15138 (48) 1839 2097 2362 š 4
<45 0.05 63.1 78.0 841.2 0.15344 3.1633 0.2955 (20) 5.919 (5) 0.14529 (33) 1669 1964 2291 š 6
a Isotopic compositions and concentrations were determined using a VG Sector thermal ionization mass spectrometer (in static mode) at University of Michigan. Total

Pb processing blanks during the study averaged 0.04 ng. Common Pb from laboratory procedures (blank Pb) are as follows: 206 Pb=204 Pb D 19:4 š 1:2, 207 Pb=204 Pb
D 15:79 š 0:12, and 208 Pb=204 Pb D 39:0 š 0:8. Within-run precision for Pb measurements was generally š0.06% or lower. Uncertainties in U and Pb concentrations
are š0.39% and 0.53%, respectively, or better. Uncertainty in Pb=U values is š0.82% or better, and was calculated using propagated analytical errors oil the Pb and
U concentrations. All isotopic ratios have been corrected for blank and for fractionation using replicate analyses of NIST standards NBS 981 and U500 (0:11 š 0:05%
amu 1 for Pb and 0:17 š 0:02% amu 1 for U). Common Pb corrections are based oil the following determinations for feldspar cogenetic with the dated rutile and zircon:
206 Pb=204 Pb D 14:193 š 14, 207 Pb=204 Pb D 15.056š23 , and 208 Pb=204 Pb D 33:889 š 34. Ages calculated using the Pbdat reduction program of Ludwig [39] using decay

constants ½238 U D 1:55125 ð 10 10 yr 1 and ½235 U D 9:8485 ð 10 10 yr 1 [40], and present 238 U=235 U D 137.88.
* Denotes radiogenic lead.
S.B. Mukasa et al. / Earth and Planetary Science Letters 164 (1998) 353–369 363

There is not, however, a viable mechanism to make

rutile U–Pb ages identically incorrect.
Therefore, the Rb–Sr ages clearly do not represent
the time of crystallization and should be evaluated
from three other possibilities: (1) They are indicative
of slow cooling in the Great Dyke, though this pos-
sibility is easier to appreciate for a mineral isochron,
but harder to explain on a whole–rock scale. (2)
The highly vulnerable whole–rock Rb–Sr system-
atics were affected by low-grade metamorphism.
Although this could indeed account for the scatter
exhibited by our isochron and the one by Davies et
al. [27], it fails to explain the near-perfect co-lin-
earity of the data points defining the Hamilton [29]
isochron. Also, at least for the better constrained
Hamilton [29] isochron, minerals give nearly the
same age (2464 Ma) as the whole rock samples, and
there is no good reason why both mineral and whole
rock samples would be altered to give the same erro-
neous age. (3) The Great Dyke rocks were subjected
to some fluid-enhanced ionic exchange while still
hot that altered the Rb–Sr systematics, but not the
Sm–Nd in whole–rock and mineral samples or U–
Pb in rutile. Some scatter would be expected in this
case as well, but the possibility of isotopic reequili-
bration that might minimize scatter is enhanced par-
ticularly at deep levels in the stratigraphic section.
The Hamilton [29] Rb–Sr isochron was generated
on core samples from a considerably deeper level
(>1000 m) than the Davies et al. [27] isochron and
the one presented here which might explain its ro-
bustness. Thus, of these three possibilities, the third
seems the most reasonable.

Fig. 6. (a) 207 Pb=204 Pb vs. 206 Pb=204 Pb isochron for whole-rock
samples and mineral separates from drill core GDX17. Its age
of 2596 š 14 Ma is in remarkably good agreement with the
two Sm–Nd isochrons and the rutile U–Pb data illustrated in
Fig. 5a, b and Fig. 7. (b) and (c) Isochrons of 206 Pb=204 Pb vs.
238 U=204 Pb and 207 Pb=204 Pb vs. 235 U=204 Pb, respectively. These

yield ages (ca. 2.2–2.6 Ga) with large errors, but well constrained
initial Pb ratios (206 Pb=204 Pb D 14:15 š 0:30 and 207 Pb=204 Pb
D 15:04 š 0:06), in agreement with values of 206 Pb=204 Pb D
14:193 š 14, 207 Pb=204 Pb D 15:056 š 23, and 208 Pb=204 Pb D
33:889 š 34 for feldspar cogenetic with the dated rutile and
zircon in sample MR89.
364 S.B. Mukasa et al. / Earth and Planetary Science Letters 164 (1998) 353–369

Fig. 7. Concordia diagram showing the rutile and zircon U–Pb data for a feldspathic pyroxenite at 186.79–187.53 m depth of drill core
MR89 from the Darwendale Subchamber. The three fractions of rutile are concordant yielding an error-weighted mean for the U–Pb ages
of 2587 š 8 Ma. Three zircon fractions from the same feldspathic pyroxenite sample are discordant, and although not well constrained,
yield a lower concordia intercept that suggests Pb loss from the zircons at ca. 830 Ma.

6.2. Differences in behavior between rutile and 6.3. Sm–Nd and U–Pb suggest rapid cooling
zircon and the geologic significance of the zircon Pb
loss event The closure temperature for Sm–Nd in mafic
whole–rock samples and magmatic minerals is
All three rutile fractions of the feldspathic pyrox- ¾800ºC, which is a factor of two higher than the
enite from drill core MR89 that we have dated by estimated closure temperature for U–Pb in rutile
U–Pb are concordant while all of its three zircon [37,38]. Although published closure temperatures for
fractions are discordant owing to some Pb loss. One rutile of 400–430ºC apply only to those in metamor-
obvious difference between the two minerals that phic belts that underwent cooling rates of 2–5ºC=Ma,
might account for their different behaviors is U con- there are no data to indicate that magmatic rutile has
centrations (rutile: 2.46–2.86 ppm; zircon: 42.6–63.1 a substantially different closure temperature. This is
ppm). With its order of magnitude higher concentra- because rutile is typically >99% TiO2 which makes
tion in U, zircon is more susceptible to radiation the compositional dependence of closure tempera-
damage and therefore Pb loss either by leaching or ture quite small. Also, it has been demonstrated (cf.
ionic diffusion. A discordia through the zircon points [41–44]) that for most systems, closure temperature
and an upper concordia intercept fixed by the rutile increases by only 30–50ºC if the cooling rate in-
data yields a lower concordia intercept of 816 š 24 creases by a factor of 10. For a 400ºC difference
Ma, the geologic significance of which remains to be in closure temperature between the Sm–Nd whole
explored fully. This date may be related to the on- rock and U–Pb rutile systems, the uncertainties of
set of the widespread and diachronous Pan-African the Great Dyke ages reported here allow a maximum
tectonothermal event in southern Africa (cf. [4]). of 23 ð 106 yr age difference between the Sm–Nd
isochron age and the U–Pb rutile mineral age. This
 S.B. Mukasa et al. / Earth and Planetary Science Letters 164 (1998) 353–369 365

translates to a minimum cooling rate of 17ºC=106 yr,

though much higher cooling rates are implied by the
overlap in rutile and whole rock ages.

6.4. Petrogenetic implications of the isochrons

Indistinguishable initial 87 Sr=86 Sr values derived

from the Rb–Sr isochrons for samples from widely
separated areas of the Great Dyke show that parental
magmas in the intrusion were homogeneous on a
large scale. Also, besides yielding a well constrained
age, the better of the two Sm–Nd isochrons shows
that: (1) magmas forming the Mafic and Ultramafic
Sequences of the Great Dyke had the same initial
isotopic ratio; (2) there was isotopic homogeneity
between subchambers; (3) rare medium- and fine-
grained layers in a Darwendale Subchamber web-
sterite represent in-situ magma chamber processes
rather than dramatic changes in magma sources; and
(4) satellite dikes, such as the East Dyke, which also
fall on the same isochron share a common origin
Fig. 8. Initial 206 Pb=204 Pb for the Great Dyke plotted against
with the Great Dyke. e½t 1 and compared to values for the Stillwater layered mafic
intrusion. The conformable ore evolution curve (COL), and de-
6.5. Pb and Sr isotopic evolution pleted mantle curve (DM) (same as the bulk earth (BE) curve)
from Kwon et al. [45] are also included for reference. See text
for discussion.
Initial 206 Pb=204 Pb for the Great Dyke plotted
against e½t 1 and compared to values for the Still-
water layered mafic intrusion is shown in Fig. 8. The
conformable ore evolution curve (COL), depleted
mantle curve (DM), and bulk earth curve (BE) from
Kwon et al. [45] are also included for reference. Note
that the BE and DM evolution curves are essentially
unresolvable. The diagram shows that sources for the
Great Dyke magmas either had higher ¼ values than
BE=DM or the magmas acquired their high ¼ char-
acter through crustal contamination. Similarly, the
87
Sr=86 Sr evolution diagram (cf. [46]) (Fig. 9) shows
that like the Stillwater complex [47], the Great Dyke
magma appears to have been influenced by crustal
materials not unlike those with typical Archean evo-
lution trajectories. The slightly positive initial "Nd
of the Sm–Nd whole rock isochron (C1.1) is some-
what below that expected for the depleted mantle in
the latest Archean ("Nd D C2 to C4) which also is
consistent with the addition to the magma of some
crustally derived components. Fig. 9. 87 Sr=86 Sr evolution diagram comparing the Great Dyke
to crustal materials from Archean cratons [46] and the Stillwater
The highly uniform isotopic data and well con-
complex [47]. See text for discussion.
strained isochrons for a wide variety of rock types
366 S.B. Mukasa et al. / Earth and Planetary Science Letters 164 (1998) 353–369

sampled from the entire length of the Great Dyke the Limpopo Belt with the rigid Zimbabwe Craton
indicates that contamination at crustal levels is not [49]. Kamber et al. [48] caution, however, that mod-
likely to have been an important factor. Thorough els of simple continent–continent collision of the
homogenization throughout all the magma subcham- Kaapvaal and Zimbabwe Cratons [50], or of colli-
bers is unlikely if contamination was at the crustal sion between the Kaapvaal Craton and the Central
level. It is more likely that the introduction of mate- Zone of the Limpopo Belt followed by collision
rials rich in incompatible elements occurred directly with the Zimbabwe Craton [51], may be simplistic.
in the source region by subduction. This interpreta- Nevertheless, evidence is now compelling that em-
tion is favored by the new crystallization age for the placement of the Great Dyke and its satellite dikes
Great Dyke that we report here which ties temporally closely followed the amalgamation of the Kaap-
the emplacement of this intrusion to amalgamation vaal and Zimbabwe Cratons, and was contempo-
of the Kaapvaal and Zimbabwe Cratons by NNW- raneous with emplacement of the youngest of the
directed convergence. trondhjemite–tonalite–granodiorite granitoid suite in
the Zimbabwe Craton. Assuming that amalgamation
6.6. The Great Dyke emplacement event in the of the Kaapvaal and Zimbabwe Cratons was largely
broader context by NNW-directed convergence, it follows that the
source of the Great Dyke was asthenospheric mantle
The Great Dyke magmatic event has important hydrated and enriched in incompatible elements by
implications for the late Archean–early Proterozoic subduction processes. Rapid decompression result-
tectonic development of the Zimbabwe Craton. The ing from the fracturing of the Zimbabwe Craton by
emplacement of the Great Dyke is commonly ac- collision promoted the large degree of partial melting
cepted as marking the first major magmatic event necessary to produce large volumes of mafic magma
in which the Zimbabwe Province and the North in a narrow rifted zone. Prior melting and meta-
Marginal Zone of the Limpopo Belt (NMZ) acted somatism of the mantle connected with production
as a rigid cratonic entity [48]. This is also accepted of the Chilimanzi granitoid suite may explain the
as conveniently marking the boundary between the boninite-like petrologic character of the Great Dyke
Proterozoic and Archean periods in Zimbabwe [1]. and its satellite dikes, typified by both high MgO and
There is also much recent evidence to suggest that at SiO2 .
about 2.6 Ga extensive felsic magmatism coincided
with the peak of metamorphism in the NMZ of the
Limpopo Belt. Kamber et al. [48] show that on the 7. Conclusions
basis of U–Pb zircon geochronology, the peak of
high-temperature metamorphism in the NMZ of the Our new Sm–Nd, U–Pb, and Pb–Pb age determi-
Limpopo Belt was 2591 š 4 Ma, slightly preceding nations indicate that the Great Dyke of Zimbabwe
the time of emplacement of the Great Dyke (but and its satellite dikes are over 100 Myr older than
in error indistinguishable), and following extensive previously believed based on Rb–Sr ages. The Rb–
felsic plutonism of the Chilimanzi Suite and related Sr method records a resetting probably related to
rocks between 2601 š 14 and 2618 š 6 Ma [10]. The high-temperature hydrothermal processes.
age of emplacement for the Great Dyke as presented The Sm–Nd method has yielded a combined
in this work suggests that major mafic magmatism mineral=whole–rock isochron of 2586 š 16 Ma and
was also part of the tectonothermal event. The Great "Nd .t/ of C1.1 for samples from the Darwendale,
Dyke mafic intrusion thus appears to be part of Sebakwe, and Wedza Subchambers as well as the
the complex series of events embracing a narrow satellite East Dyke. This isochron age is in excel-
time span and is in itself not related to an isolated lent agreement with the U–Pb age for three con-
emplacement event arising from the completion of cordant rutile fractions extracted from a feldspathic
cratonization. pyroxenite of the Selukwe Subchamber with an er-
The tectonic control of the Great Dyke emplace- ror-weighted mean at 2587 š 8 Ma, and also with
ment event was certainly related to the collision of the mineral=whole-rock 207 Pb=204 Pb vs. 206 Pb=204 Pb
 S.B. Mukasa et al. / Earth and Planetary Science Letters 164 (1998) 353–369 367

isochron age of 2596 š 14 Ma for a Darwendale References

Subchamber drill core.
This new age information shows that emplace- [1] E.G. Nisbet, Definition of ‘Archaean’ — comment on
a proposal on the recommendations of the International
ment of the Great Dyke and its satellite dikes Sub-Commission on Precambrian Stratigraphy, Precam-
closely followed the amalgamation of the Kaap- brian Res. 19 (1982) 111–118.
vaal and Zimbabwe Cratons, and was contempo- [2] B. Luais, C.J. Hawkesworth, The generation of continental
raneous with emplacement of the youngest of the crust: An integrated study of crust-forming processes in the
trondhjemite–tonalite–granodiorite granitoid suite in Archean of Zimbabwe, J. Petrol. 35 (1994) 43–93.
[3] K.C. Condie, Archean greenstone belts, Elsevier, Amster-
the Zimbabwe Craton. Assuming that amalgamation dam, 1981, 434 pp.
of the Kaapvaal and Zimbabwe Cratons was largely [4] J.F. Wilson, R.W. Nesbitt, C.M. Fanning, Zircon
by NNW-directed convergence, it follows that the geochronology of Archean felsic sequences in the Zim-
source of the Great Dyke was asthenospheric mantle babwe craton: A revision of greenstone stratigraphy and
hydrated and enriched in incompatible elements by a model for crustal growth, in: M.P. Coward, A.C. Ries
(Eds.), Early Precambrian Processes, Geol. Soc. London
subduction process. Spec. Publ. 95 (1995) 109–126.
Initial Pb, Sr, and Nd isotopic compositions are [5] C.J. Hawkesworth, S. Moorbath, K. O’Nions, J.F. Wilson,
consistent with derivation of the Great Dyke magmas Age relationships between greenstone belts and ‘granites’
by large volume melting of a mantle that has been in the Rhodesian Archean craton, Earth Planet. Sci. Lett. 25
hydrated and enriched by subduction. While a small (1975) 239–250.
[6] B.-M. Jahn, K.C. Condie, On the age of Rhodesian green-
amount of crustal contamination of magma derived
stones, Contrib. Mineral. Petrol. 57 (1976) 317–330.
from depleted mantle could produce the composition [7] J.F. Wilson, M.J. Bickle, C.J. Hawkesworth, A. Martin,
of the Great Dyke, the uniformity of initial ratios E.G. Nisbet, J.L. Orpen, The granite–greenstone terrains of
between subchambers supports the notion of enrich- the Rhodesian Archaean Craton, Nature 271 (1978) 23–27.
ment in incompatible elements being an intrinsic [8] S. Moorbath, J.F. Wilson, R. Goodwin, M. Humm, Fur-
characteristic of the mantle source. ther Rb–Sr age and isotope data on early and late Ar-
chaean rocks from the Rhodesian Craton, Precambrian Res.
The high MgO and SiO2 character of the Great 5 (1977) 229–239.
Dyke, which resembles boninite magmas in modern [9] J.S. Dougherty-Page, The evolution of the Archean conti-
arc settings, is explicable by derivation of the intru- nental crust of northern Zimbabwe, The Open University,
sion’s magma from a mantle that has undergone prior 1994.
melting and subsequent enrichment by subduction. [10] H.A. Jelsma, M.L. Vinyu, P.J. Valbracht, G.R. Davies,
J.R. Wijbrans, E.A.T. Verdurmen, Constraints on Archean
Two zircon fractions from the same feldspathic crustal evolution of the Zimbabwe craton: A U–Pb zir-
pyroxenite sample as the rutile are discordant, and con, Sm–Nd and Pb–Pb whole–rock isotope study, Contrib.
although not well constrained, suggest Pb loss from Mineral. Petrol. 124 (1996) 55–70.
the zircons at ca. 830 Ma. This may be related [11] M.H. Hickman, Geochronological studies of the Limpopo
to the onset of the widespread and diachronous mobile belt and part of the adjacent Rhodesian Craton,
Ph.D., University of Leeds, 1976.
Pan-African tectonothermal event in southern Africa. [12] B. Kamber, T. Blenkinsop, H. Rollinson, J. Kramers, M.
Berger, Dating of an important tectono-magmatic event in
the Northern Marginal Zone of the Limpopo Belt, Zim-
Acknowledgements babwe: first results, in: T. Blenkinsop, H. Rollinson (Eds.),
North Limpopo Field Workshop, University of Zimbabwe,
Harare, 1992, pp. 39.
We are indebted to Charlie Murahwi of Anglo
[13] B.G. Worst, The differentiation and structure of the Great
American Corporation, Zimbabwe, for facilitating Dyke of Southern Rhodesia, Trans. Geol. Soc. S. Afr. 61
our field work and the sampling of drill cores, and (1958) 283–354.
Sandy Zeff for the analytical work performed at the [14] A.H. Wilson, The Great Dyke of Zimbabwe, in: R.G.
University of Michigan. We also acknowledge the Cawthorn (Ed.), Layered Intrusions, Elsevier, Amsterdam,
1996, pp. 365–402.
constructive reviews provided by K. Condie and S.A.
[15] A.H. Wilson, The geology of the Great Dyke, Zimbabwe:
Morse. This work was supported by a grant from the Crystallization, layering, and cumulate formation in the P1
Office of the Vice Provost for Academic and Multi- Pyroxenite of cyclic unit 1 of the Darwendale Subchamber,
cultural Affairs at the University of Michigan. [CL] J. Petrol. 33 (3) (1992) 611–663.
368 S.B. Mukasa et al. / Earth and Planetary Science Letters 164 (1998) 353–369

[16] D.L. Jones, I.D.M. Robertson, P.L. McFadden, A paleo- solution in U–Pb geochronology, Chem. Geol. 66 (1987)
magnetic study of Precambrian dyke swarms associated 99–102.
with the Great Dyke of Rhodesia, Trans. Geol. Soc. S. Afr. [33] R.R. Parrish, J.C. Roddick, W.D. Loveridge, R.W. Sullivan,
77 (1975) 339–345. Uranium–lead analytical techniques at the geochronology
[17] J.W. Wiles, Some aspects of the metamorphism of the laboratory, Geol. Soc. Can. Pap. 87-2 (1987) 3–7.
basement complex in the Sipolilo District, Trans. Geol. [34] S.B. Mukasa, R. McCabe, J.B. Gill, Pb-isotopic compo-
Soc. S. Afr. 71 (1968) 79–88. sitions of volcanic rocks in the West and East Philippine
[18] F. Podmore, A.H. Wilson, A reappraisal of the structure, island arcs: presence of the Dupal isotopic anomaly, Earth
geology and emplacement of the Great Dyke, Zimbabwe, Planet. Sci. Lett. 84 (1987) 153–164.
in: H.C. Halls, W.F. Fahrig (Eds.), Mafic Dyke Swarms. [35] S.B. Mukasa, J.W. Shervais, H.G. Wilshire, J.E. Nielson,
Spec. Paper Geol. Assoc. Can. 34 (1987) 433–444. Intrinsic Nd, Pb, and Sr isotopic heterogeneities exhibited
[19] M. Prendergast, The chromite ore field of the Great Dyke, by the Lherz alpine peridotite massif, French Pyrenees, J.
Zimbabwe, in: C.W. Stowe (Ed.), Evolution of Chromite Petrol. Spec. Lherzolite Issue (1991) 117–134.
Ore Fields, Van Nostrand Reinhold, New York, 1987, pp. [36] A.H. Wilson, R.W. Carlson, A Sm–Nd and Pb isotope
89–108. study of Archean greenstone belts in the southern Kaapvaal
[20] A.H. Wilson, M. Prendergast, The Great Dyke of Zim- Craton, South Africa, Earth Planet. Sci. Lett. 96 (1989)
babwe I: Tectonic setting, stratigraphy, petrology, structure, 89–105.
emplacement and crystallization, in: M.D. Prendergast, M.J. [37] K. Mezger, C.M. Rawnsley, S.R. Bohlen, G.N. Hanson, U–
Jones (Eds.), Magmatic Sulphides — the Zimbabwe Vol- Pb garnet, sphene, monazite, and rutile ages: implications
ume, Institute of Mining and Metallurgy, London, 1989, pp. for the duration of high-grade metamorphism and cooling
1–20. histories, Adirondack Mts., New York, J. Geol. 99 (1991)
[21] F. Podmore, The shape of the Great Dyke as revealed by 415–428.
gravity surveying, Spec. Publ. Geol. Soc. S. Afr. 1 (1970) [38] K. Mezger, G.N. Hanson, S.R. Bohlen, U–Pb ages of meta-
morphic rutiles: Application to the cooling history of high
610–620.
grade terranes, Earth Planet. Sci. Lett. 96 (1989) 106–118.
[22] A.H. Wilson, J.F. Wilson, The Great Dyke, in: D.R. Hunter
[39] K.R. Ludwig, A computer program to convert raw Pb–U–
(Ed.), Precambrian of the Southern Hemisphere, Elsevier,
Th isotope ratios to blank-corrected isotope ratios and con-
Amsterdam, 1981, pp. 572–578.
centrations, with associated errors and error-correlations,
[23] A.H. Wilson, The geology of the Great Dyke, Zimbabwe,
U.S. Geological Survey, Denver, 1982.
J. Petrol. 23 (1982) 240–292.
[40] E.H. Steiger, E. Jäger, Subcommission on geochronology:
[24] G. Faure, P.M. Hurley, H.W. Fairbairn, W.H. Pinson, Age
Convention on the use of decay constants in geo- and
of the Great Dyke of Southern Rhodesia, Nature 200 (1963)
cosmochronology, Earth Planet. Sci. Lett. 36 (1977) 359–
769.
362.
[25] L.O. Nicolaysen, J.W.L. De Villiers, A.J. Burger, F.W.E.
[41] D.J. Cherniak, W.A. Lanford, F.J. Ryerson, Lead diffusion
Strelow, New measurements relating to the absolute age of in apatite and zircon using ion-implantation and Rutherford
the Bushveld Igneous Complex, Trans. Geol. Soc. S. Afr. backscattering techniques, Geochim. Cosmochim. Acta 55
61 (1958) 137–166. (1991) 1663–1673.
[26] H.L. Allsopp, Rb–Sr age measurements on the Great Dyke, [42] E.J. Krogstad, R.J. Walker, High closure temperatures of
Southern Rhodesia, J. Geophys. Res. 70 (1965) 977–984. the U–Pb system in large apatites from the Tin Moun-
[27] R.D. Davies, H.L. Allsopp, A. Erlank, J.W.I. Manton, Sr tain pegmatite, Black Hills, South Dakota, USA, Geochim.
isotopic studies on various layered mafic intrusions in Cosmochim. Acta 58 (1994) 3845–3853.
Southern Africa, Spec. Publ. Geol. Soc. S. Afr. 1 (1970) [43] E.S. Bogomolov, A comparison of lead and uranium dif-
576–593. fusion in crystalline zircons, Geokhimiya 10 (1995) 1423–
[28] I.D.M. Robertson, O. van Breemen, The southern satellites 1432.
of the Great Dyke, Rhodesia, Spec. Publ. Geol. Soc. S. Afr. [44] D.J. Scott, M.R.S. Onge, Constraints on Pb closure tem-
1 (1970) 621–644. perature in titanite based on rocks from Ungava Orogen,
[29] J. Hamilton, Strontium isotope and trace element studies on Canada — Implications for U–Pb geochronology and P–T–
the Great Dyke and Bushveld mafic phase and their relation t path determinations, Geology 23 (1995) 1123–1126.
to early Proterozoic magma genesis in southern Africa, J. [45] S.-T. Kwon, G.R. Tilton, M.H. Grünenfelder, Lead iso-
Petrol. 18 (1977) 24–52. topic relationships in carbonatites and alkali complexes:
[30] D.J. DePaolo, G.J. Wasserburg, Nd isotopic variations and an overview, in: K. Bell (Ed.), Carbonatites: Genesis and
petrogenetic models, Geophys. Res. Lett. 3 (1976) 249– Evolution, Unwin Hyman, London, 1989, pp. 360–368.
252. [46] A. Zindler, S.R. Hart, C. Brooks, The Shabogamo Intrusive
[31] T.E. Krogh, Improved accuracy of U–Pb zircon ages by the Suite, Labrador: Sr and Nd isotopic evidence for contami-
creation of more concordant systems using an air abrasion nated mafic magmas in the Proterozoic, Earth Planet. Sci.
technique, Geochim. Cosmochim. Acta 46 (1982) 637–649. Lett. 54 (1981) 217–235.
[32] R.R. Parrish, An improved micro-capsule for zircon dis- [47] D.J. DePaolo, G.J. Wasserburg, Sm–Nd age of the Stilwater
 S.B. Mukasa et al. / Earth and Planetary Science Letters 164 (1998) 353–369 369

complex and the mantle evolution curve for neodymium, Afr. Earth Sci. 10 (1990) 483–501.
Geochim. Cosmochim. Acta 43 (1979) 999–1008. [50] C. Roering, D.D. van Reenen, C.A. Smit, J.M. Barton, J.H.
[48] B.S. Kamber, G.G. Biino, J.R. Wijbrans, G.R. Davies, I.M. De Beer, M.J. De Wit, E.H. Stettler, J.F. van Schalkwyk, G.
Villa, Archean granulites of the Limpopo Belt, Zimbabwe: Stevens, S. Pretorious, Tectonic model for the evolution of
One slow exhumation or two rapid events?, Tectonics 15 the Limpopo Belt, Precambrian Res. 55 (1992) 539–552.
(1996) 1414–1430. [51] P.J. Treloar, T.G. Blenkinsop, Archean deformation patterns
[49] J.F. Wilson, The craton and its cracks: some of the be- in Zimbabwe — true indicators of Tibetan style crustal
haviour of the Zimbabwe blocks from the Late Archaean to extrusion or not?, in: P.P. Coward, A.C. Ries (Eds.), Early
the Mesozoic in response to horizontal movements, and the Precambrian Processes, Geol. Soc. London Spec. Publ. 95
significance of some of its mafic dyke fracture patterns, J. (1995) 87–108.