Journal of Petrology

This Article Abstract FREE Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Search for citing articles in: ISI Web of Science (23) Request Permissions Google Scholar Articles by JANNEY, P. E. Articles by VILJOEN, K. S. Search for Related Content GeoRef GeoRef Citation Social Bookmarking          What's this?

Journal of Petrology | Volume 43 | Number 12 | Pages 2339-2370 | 2002
© Oxford University Press 2002

A Chemical and Multi-Isotope Study of the Western Cape Olivine Melilitite Province, South Africa: Implications for the Sources of Kimberlites and the Origin of the HIMU Signature in Africa

P. E. JANNEY^1,2,*, A. P. LE ROEX¹, R. W. CARLSON² and K. S. VILJOEN³

¹DEPARTMENT OF GEOLOGICAL SCIENCES, UNIVERSITY OF CAPE TOWN, RONDEBOSCH 7700, SOUTH AFRICA
²DEPARTMENT OF TERRESTRIAL MAGNETISM, CARNEGIE INSTITUTION OF WASHINGTON, 5241 BROAD BRANCH ROAD, NW, WASHINGTON, DC 20015, USA
³BERNARD PRICE INSTITUTE OF GEOPHYSICAL RESEARCH, UNIVERSITY OF THE WITWATERSRAND, JOHANNESBURG, 2050, SOUTH AFRICA

Received October 30, 2001; Revised typescript accepted July 4, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION TECTONIC SETTING AND AGE... PETROGRAPHY ANALYTICAL METHODS RESULTS DISCUSSION AND CONCLUSIONS REFERENCES

 
We present major and trace element and Sr–Nd–Pb–Hf–Os isotopic data for the 76–58 Ma Western Cape melilitite province, an age-progressive magmatic lineation in which primitive olivine melilitite intrusives and alkali basalt lavas have been emplaced on the southwestern margin of South Africa. The magmas range from alkali basalts with strong HIMU isotopic and trace element affinities on the continental shelf to melilitites with kimberlite-like incompatible element compositions and EM 1 isotopic affinities on thick Proterozoic lithosphere (i.e. ⁸⁷Sr/⁸⁶Sr_i = 0·7029–0·7043, _Nd(t) = +5 to +2, ²⁰⁶Pb/²⁰⁴Pb_i = 20·5–18·1). The samples are characterized by radiogenic ¹⁸⁷Os/¹⁸⁸Os_i (0·15–0·21) and unradiogenic Hf isotope ratios, causing some of them to fall outside the compositional range of oceanic basalts. This suggests that they tap a mafic HIMU source component that is less dilute than or compositionally different from that tapped by HIMU ocean island basalts. Os–Pb and Nd–Hf isotopic co-variations suggest that the sources of the Western Cape melilitites were generated by a two-stage mixing process in which (1) mafic material (e.g. ancient oceanic crust) underwent solid-state mixing with a small proportion of asthenospheric peridotite and then (2) high-degree melts of this hybrid material mixed with low-degree melts of metasomatized continental lithospheric mantle. A similar process may help to explain the isotopic variations of Group 1 kimberlites, but the available data do not allow us to rule out a major role for convecting mantle components in kimberlite genesis. Using the Western Cape melilitites as an analog, we propose a model in which the strong HIMU isotopic signature present in most Late Cretaceous to Recent alkaline magmatic provinces across Africa is supplied not by conventional mantle plumes, but rather by pods of recycled oceanic crust brought up from the deep mantle within a long-lived, and laterally broad mantle upwelling feature located beneath the African plate.

KEY WORDS: melilitite; kimberlite; isotopes, HIMU; lithospheric mantle

	INTRODUCTION

TOP ABSTRACT INTRODUCTION TECTONIC SETTING AND AGE... PETROGRAPHY ANALYTICAL METHODS RESULTS DISCUSSION AND CONCLUSIONS REFERENCES

 
Determining the origins and sources of mafic and ultramafic alkaline continental magmas, including kimberlites, remains an important but poorly resolved problem that has major implications for the compositional structure of the deep continental lithospheric mantle (CLM) and the nature of interactions between the continental lithosphere and the underlying convecting mantle. The pioneering work of Smith (1983) and others (e.g. Kramers et al., 1981; McCulloch et al., 1983; Fraser et al., 1985; Nelson et al., 1986) revealed that kimberlites, lamproites and other mafic–ultramafic continental alkaline magmas can be divided into two principal groups on the basis of their Sr, Nd and Pb isotopic compositions. One group requires sources with long-term incompatible element enrichments relative to bulk silicate earth (BSE) as well as low ²³⁸U/²⁰⁴Pb (µ) ratios, consistent with derivation from ancient, highly metasomatized CLM. The other requires sources with long-term incompatible element depletions relative to BSE, but with relatively high µ values, similar to the sources of many ocean island basalts (OIB). This discovery has resulted in a simple and widely used isotopic classification system in which magmas such as southern African Group 2 kimberlites and lamproites are attributed to ‘lithospheric’ sources, whereas magmas such as Group 1 kimberlites, alnöites and melilitites are attributed to ‘sublithospheric’ sources (i.e. the asthenosphere or mantle plumes; e.g. Smith, 1983; Hegner et al., 1995).

Although rocks such as Group 2 kimberlites and lamproites, which have no isotopic equivalents in the ocean basins (e.g. Fraser et al., 1985), require sources dominated by ancient metasomatized CLM, the intermediate isotopic compositions of Group 1 kimberlites and other mafic–ultramafic alkaline igneous rocks with OIB-like isotopic compositions may be derived from mantle sources located either within or below the continental lithosphere. One way to test whether the CLM constitutes a source of these magmas without a priori assumptions of source reservoir compositions is to examine a cogenetic alkaline magmatic province that penetrates two or more adjacent lithospheric terranes of differing age and/or thickness. If radiogenic isotope and source-sensitive trace element ratios display coherent variations that correspond to these lithospheric variations, this would demonstrate that the CLM provides at least one source component for kimberlitic magmatism, and would allow the isotopic characteristics of the lithospheric and the other, possibly sublithospheric, component(s) to be defined.

The Western Cape olivine melilitite province of southwestern South Africa provides an excellent study area for investigating the sources of isotopically OIB-like continental alkaline magmas. Like kimberlites, olivine melilitites are generated by small degrees of melting of carbonated, hydrous, garnet peridotite, albeit at somewhat lower pressures (30–45 kbar as opposed to >50 kbar for kimberlites; e.g. Brey, 1978) and therefore from a slightly different source mineralogy (e.g. with dolomite rather than magnesite as the stable carbonate phase; e.g. Eggler, 1989). However, unlike kimberlites, which occur only on Archean cratons and the thickest portions of Proterozoic mobile belts, olivine melilitites occur in a variety of tectonic settings and on lithosphere of widely varying thickness (i.e. from mature oceanic islands to continental rifts and mobile belts; e.g. Brey, 1978; Clague & Frey, 1982; Wilson et al., 1995).

The Western Cape melilitite province is one of three age-progressive, Late Cretaceous–Early Tertiary alkaline magmatic lineaments that intersect the southwestern continental margin of southern Africa (e.g. Duncan et al., 1978; Moore, 1979; Spriggs, 1988; Fig. 1) and it straddles the boundary between the mid-Proterozoic Namaqua–Natal mobile belt and the latest Proterozoic Cape Fold Belt. The northern end of the province lies on thick mobile belt lithosphere only 150–200 km from numerous true kimberlite occurrences to the NE (Smith et al., 1994) and the southern end of the province includes submarine alkali basalt lavas erupted on the continental shelf, the thin edge of the continental lithosphere. The large variation in the thickness and age of the terranes traversed by the Western Cape province magmas makes it ideal for constraining contributions to these rocks from lithospheric and sublithospheric mantle sources.

View larger version (29K): [in this window] [in a new window]   Fig. 1. Location maps. (a) Regional map of southwestern Africa showing alkaline magmatic lineations (all of Late Cretaceous–Early Tertiary age) and spatially–temporally associated kimberlite provinces (outlined; Duncan et al., 1978; Moore, 1979; Spriggs, 1988; Smith et al., 1994). (b) Sketch map of the five localities constituting the Western Cape melilitite province. K–Ar ages are from Dingle & Gentle (1972) and Duncan et al. (1978).

 

Our primary objective is to place constraints on the origin and mixing systematics of end-member components in the source regions of the Western Cape magmas, and to determine whether the sources and petrogeneses of the Western Cape melilitites are analogous to those of southern African kimberlites. Toward this goal we employ conventional trace element and Sr–Nd–Pb isotopic ratios, as well as Hf and Os isotopic data that provide crucial information on the mineralogy of mantle source regions and the nature of possible source mixing. Our secondary objective is to evaluate possible origins of the HIMU isotopic signature (e.g. Zindler & Hart, 1986) that is nearly ubiquitous in Cretaceous–Recent mafic alkaline magmatic provinces throughout Africa, including the Western Cape melilitites and southern African kimberlites. In particular, we wish to determine if the conventional hypothesis, that this signature is derived from a coincidental concentration of numerous HIMU-type mantle plumes (e.g. Burke, 1996), is the best explanation of the available data.

	TECTONIC SETTING AND AGE CONSTRAINTS

TOP ABSTRACT INTRODUCTION TECTONIC SETTING AND AGE... PETROGRAPHY ANALYTICAL METHODS RESULTS DISCUSSION AND CONCLUSIONS REFERENCES

 
The Western Cape melilitite province spans 360 km from the southern Karoo region to the Alphard Bank on the continental shelf SE of Cape Agulhas (Fig. 1). The province consists of five main localities of which only the four northern, subaerial ones contain true olivine melilitite. The northernmost localities, Sutherland and Saltpetre Kop, are situated on the relatively thick, Early to mid-Proterozoic Namaqua–Natal mobile belt terrane (Doucouré et al., 1996). The Robertson and Spiegel River melilitites are located on thinner Neo-Proterozoic (Pan-African age) Cape Fold Belt lithosphere (Söhnge & Hälbich, 1983). Furthest south, the submarine volcanic vents of the Alphard Bank are located on the continental shelf 50 km SE of Cape Agulhas, presumably on the seaward extension of Cape Fold Belt lithosphere.

Morphological and volcanological descriptions of the Western Cape olivine melilitites and associated igneous rock occurrences have been reported in some detail (Rogers & DuToit, 1904; Taljaard, 1936; Dingle & Gentle, 1972; Viljoen, 1988; Verwoerd, 1990) and will only be briefly summarized here. Most of the melilitites and associated igneous rocks of the province were emplaced as pipe-like bodies, some occurring as solitary intrusions (e.g. Spiegel River) and some having multiple satellite intrusions within an area of a few square kilometers (e.g. Robertson; Verwoerd, 1990). The main pipes at Robertson and Spiegel River each have diameters of 200 m. The Sutherland and Saltpetre Kop localities are more complex than the simple diatremes at Spiegel River and Robertson. The Sutherland melilitites appear to have been intruded in three rapid, successive episodes, starting with the emplacement of an 800 m diameter ring dike, followed by two episodes of sill intrusion inside the dike perimeter (Viljoen, 1988). The melilitites at Saltpetre Kop occur as pipes and dikes on the periphery (2–3 km radial distance) of a prominent volcanic ring structure consisting of potassic trachyte, carbonatite and associated pyroclastics (DeWet, 1975; Verwoerd, 1990). The Alphard Bank volcanic rocks occur as numerous small submarine vents over an area of 100 km², in a down-faulted trough. From the results of limited dredging, the Alphard Bank lavas appear to be constituted mainly of trachyte and phonolite with smaller quantities of alkali basalt (Dingle & Gentle, 1972).

Duncan et al. (1978) obtained K–Ar ages of 75·8–62·6 Ma for whole rocks and phlogopite separates from the four Western Cape melilitite localities (Fig. 1). These ages, in combination with a K–Ar age of 58·0 ± 2·4 Ma for an Alphard Bank trachyte sanidine separate (Dingle & Gentle, 1972), yield a nearly linear age progression of 19 km/Myr. This indicates a steady southward migration of magmatism in the Western Cape that is similar in magnitude to, but about 25° oblique from, the absolute motion of the African plate during the Late Cretaceous and Early Tertiary (Hartnady & le Roex, 1985). Duncan et al. (1978) interpreted this age progression to be the result of the northeasterly passage of southern Africa over the Bouvet hotspot. However, the more recent plate motion reconstructions of Hartnady & le Roex (1985) and O’Connor & le Roex (1992) have placed the trace of the Bouvet hotspot roughly 1000 km to the east of the Western Cape melilitite province, and no other currently recognized South Atlantic hotspots appear to have been present in the region during the Late Cretaceous or Early Tertiary.

	PETROGRAPHY

TOP ABSTRACT INTRODUCTION TECTONIC SETTING AND AGE... PETROGRAPHY ANALYTICAL METHODS RESULTS DISCUSSION AND CONCLUSIONS REFERENCES

 
The petrography and mineralogy of Western Cape melilitites and associated rocks have been described individually by locality (e.g. Taljaard, 1936; Dingle & Gentle, 1972; McIver & Ferguson, 1979; Boctor & Yoder, 1986; Viljoen, 1988), but we will briefly summarize the petrography of the province as a whole. Olivine is the dominant (and most often only) phenocryst phase in magmas from all localities and also is a major groundmass constituent. Olivine phenocryst populations in the Western Cape melilitites, like those of Namaqualand–Bushmanland (Moore & Erlank, 1979), reveal a complex history of minor differentiation punctuated by recharge and mixing with primary melts (Boctor & Yoder, 1986; Viljoen, 1988). The phenocrysts fall into two main categories: complex, anhedral ‘megacryst’ grains (1–15 mm) that tend to show strain are more common in the northern Sutherland and Saltpetre Kop localities, and euhedral ‘hopper’ olivines (<2 mm) are dominant in the southern localities of Robertson, Spiegel River and the Alphard Bank. The megacrystic olivines do not appear to be derived from disaggregated peridotite, as they have compositions that are less forsteritic (Fo_75–87) than typical mantle olivine, nor is there any other evidence of mantle-derived xenoliths or xenocrysts in the Western Cape melilitites. The fact that the olivine megacrysts are often mantled by more forsteritic rims, and that the ‘hopper’ phenocrysts and groundmass olivines tend to be more forsteritic than the megacrysts (typically Fo_85–90; McIver & Ferguson, 1979; Boctor & Yoder, 1986; Viljoen, 1988), suggests that the olivine megacrysts may not be strictly cognate.

The other important mineral constituents of the Western Cape melilitites are melilite, nepheline, clinopyroxene, phlogopite, spinel (both Fe–Ti- and Mg–Al–Cr-rich varieties) and perovskite, in decreasing order of abundance. All but melilite occur exclusively as groundmass phases. Although melilite, nepheline, perovskite and spinel are major groundmass phases in all of the subaerial localities, clinopyroxene is abundant (up to 10 vol. %) only in the Robertson and Spiegel River melilitites, as well as in the Alphard Bank basalts. Conversely, phlogopite is present in minor amounts (1 vol. %) in the Robertson and Spiegel River melilitites, but is abundant at Sutherland and Saltpetre Kop (2–10 vol. %). Accessory phases such as monticellite (often occurring as rims on olivine), garnet, zeolite and calcite are present only in the northern localities. The increased modal proportions of phlogopite, zeolite and calcite, and the greater overall mineralogical similarity of the northern melilitites to kimberlites, suggests that the Sutherland and Saltpetre Kop melilitites were derived by melting of a more volatile-rich source at greater depth than the Robertson and Spiegel River melilitites.

	ANALYTICAL METHODS

TOP ABSTRACT INTRODUCTION TECTONIC SETTING AND AGE... PETROGRAPHY ANALYTICAL METHODS RESULTS DISCUSSION AND CONCLUSIONS REFERENCES

 
Samples were trimmed of weathered surfaces using a hydraulic rock-splitter and crushed in steel or alumina jaw-crushers. Fresh, homogeneous rock chips free of veins and rare segregation vesicles were picked, washed with distilled water in an ultrasonic bath and powdered in steel (for major and trace elements and Sr–Nd–Pb isotope ratios) and alumina (for Hf and Re–Os isotope ratios). Major and trace element analyses were performed by X-ray fluorescence spectrometry (XRF) in the Department of Geological Sciences, University of Cape Town (UCT), with errors and detection limits equal to or lower than those of le Roex (1985). Although some samples had previously been analyzed for trace elements by XRF, all samples were reanalyzed for this study to obtain an internally consistent dataset with improved analytical precision. The elements Rb, Sr, Y, Zr, Nb and Th were measured using doubled counting times, yielding reproducibilities of better than 5% (1). Trace elements in a subset of samples were analyzed by inductively coupled plasma mass spectrometry (ICPMS) using a Perkin–Elmer Sciex Elan 6000 system at UCT, following the sample preparation and analysis methods of Janney & Castillo (1996), or by high-pressure ion chromatography following the method of le Roex & Watkins (1990).

Most Sr and Nd and all Pb isotope measurements were performed by thermal ionization mass spectrometry (TIMS) using the VG Sector 54 facility at UCT and standard techniques. Additional Sr and Nd isotopic ratios were determined by TIMS on VG 354 instruments at the Bernard Price Institute (BPI) of the University of the Witwatersrand and the Department of Terrestrial Magnetism (DTM), Carnegie Institution of Washington. To remove the effects of possible Sr isotopic contamination by groundwater (or seawater, in the case of the Alphard Bank basalts), powders used for Sr isotopic analysis (except those analyzed at BPI) were leached three times with 2·5N HCl for 20 min in an ultrasonic bath, interspersed with ultrapure water rinses. One Nd and all Hf isotope ratios were measured by multicollector inductively coupled plasma mass spectrometry (MC-ICPMS) using a VG Plasma 54 instrument at DTM. Hf separations were performed following high-temperature bomb dissolution using a cation–anion–cation exchange chemistry sequence adapted from the procedures of Patchett & Tatsumoto (1980a) and Blichert-Toft et al. (1997). Os isotope ratios and Re and Os concentrations were determined by negative ion-TIMS using a single-collector, custom-built 15 inch mass spectrometer at DTM following the method of Creaser et al. (1991). Os and Re separations were accomplished by Carius tube digestion, followed by solvent extraction and micro-distillation of Os and anion-exchange chemistry for Re separation (Shirey & Walker, 1995).

	RESULTS

TOP ABSTRACT INTRODUCTION TECTONIC SETTING AND AGE... PETROGRAPHY ANALYTICAL METHODS RESULTS DISCUSSION AND CONCLUSIONS REFERENCES

 
Major elements
Chemical analyses for basalts and melilitites of the Western Cape province are presented in Table 1. Apart from the Alphard Bank alkali basalts, which are moderately differentiated (Mg-number = 51–55, where Mg-number = 100Mg/(Mg + Fe²⁺) and Fe²⁺ = 0·85Fe^tot) all of the Western Cape magmas represent near-primary melt compositions (Mg-number = 70–80). The exceptionally high Mg-numbers (76–79) in Sutherland and Saltpetre Kop samples are probably due to accumulation of olivine during ascent (9–15% addition of accumulated equilibrium olivine, assuming a primary magmatic Mg-number of 73). Conversely, the melilitites with the lowest Mg-numbers (70), found at Robertson, may have experienced up to 8% olivine fractionation. In terms of silica undersaturation, MgO and Al₂O₃ contents, the Western Cape melilitites are intermediate between the compositions of ocean island and continental rift-related melilitites (e.g. from the Canary Islands and the Rhine Graben) and southern African Group 1 kimberlites (e.g. Smith et al., 1985; Davies et al., 2001; see Fig. 2).

View this table: [in this window] [in a new window]   Table 1: Major and trace element contents of the Alphard Bank alkali basalts and Western Cape olivine melilitites

 

View larger version (24K): [in this window] [in a new window]   Fig. 2. (a) MgO–SiO2 and (b) MgO–Al2O3 diagrams showing the intermediate nature of the Western Cape olivine melilitites (not including the Alphard Bank lavas) between ocean island melilitites (Clague & Frey, 1982; Hoernle & Schminke, 1993), rift-related melilitites from the Rhine Graben (Wilson et al., 1995) and southern African Group 1 kimberlites (Smith et al., 1985; Spriggs, 1988).

 

On the major element variation diagrams in Fig. 3, data from each melilitite locality tend to plot as distinct groups, with some elements displaying coherent variations with latitude. In general, SiO₂, TiO₂ and Fe₂O₃ contents increase from north to south, and Mg-number, CaO and P₂O₅ contents decrease. Fractionation corrected data for the Alphard Bank lavas (calculated by addition of 30–35% equilibrium olivine until Mg-number = 73) extend some of these trends (CaO, P₂O₅) but break others (Fe₂O₃, TiO₂). The alkali contents of the melilitites show little systematic variation, although Na₂O contents exceed K₂O in almost all samples. The Sutherland melilitites in particular display wide ranges in Na₂O and K₂O contents, which appear to be largely the result of mobilization of alkalis during deuteric or surficial alteration (e.g. Na₂O values from this locality show a strong negative correlation with LOI).

View larger version (24K): [in this window] [in a new window]   Fig. 3. Major element variation diagrams for the Western Cape olivine melilitites and alkali basalts. Fractionation-corrected Alphard Bank sample compositions were calculated by adding equilibrium olivine in 2% increments until a composition with an Mg-number of 73 (i.e. in equilibrium with mantle olivine of Fo91) was reached.

 

Trace elements
The alkali basalts and melilitites of the Western Cape province are strongly enriched in moderately and highly incompatible elements, spanning a range equivalent to that between HIMU-type ocean island basalts (OIB) and Group 1 kimberlites (e.g. chondrite-normalized La/Sm_N = 3·4–8·2; Table 1). They also display significant depletions in the heavy rare earth elements (HREE; Tb/Yb_N = 2·7–5) relative to the middle REE (MREE), indicating that all were generated at least partly in the presence of residual garnet. Compatible element abundances are high in the four melilitite localities (Ni 400–600 ppm; Cr 650–850 ppm), even after accounting for possible accumulation of olivine and chrome spinel (Ni >370 ppm; Cr >600 ppm).

The Alphard Bank basalts display incompatible element patterns with strong positive Nb anomalies and negative K anomalies, remarkably similar to lavas from St. Helena and other ocean islands with HIMU isotopic affinities (Fig. 4). Northward from the Alphard Bank, the patterns change progressively in two ways. First, there is an increasing overall fractionation between the most and least incompatible elements, demonstrated by the steepening of REE patterns, which can be explained mainly by decreasing degrees of partial melting and to a lesser extent by an increasing proportion of residual garnet in the source (Fig. 5), suggesting an increase in the depth of melting. Second, marked depletions in Zr, Hf and Ti and enrichments in Ba and Th develop relative to other elements of similar incompatibility (such as Sm and Nb, respectively). These two chemical trends are most developed in the Saltpetre Kop melilitite, which has the highest La/Sm, Nb/Zr, Ba/Nb and Th/Nb ratios of the five localities.

View larger version (33K): [in this window] [in a new window]   Fig. 4. Primitive mantle-normalized (Sun & McDonough, 1989) incompatible element diagrams for melilitites and alkali basalts from (a) the Alphard Bank, (b) Robertson and Spiegel River, (c) Sutherland, and (d) Saltpetre Kop. Range of compositions for Gibeon province (S. Namibia) Group 1 kimberlites is from Spriggs (1988), and those for St. Helena lavas are from Chaffey et al. (1989) and Thirlwall (1997). The overall steepening of incompatible element patterns from south to north, and the progressive development of depletions in Zr, Hf and Ti and enrichments in Ba and Th, should be noted.

 

View larger version (28K): [in this window] [in a new window]   Fig. 5. (Tb/Yb)N vs (Sm/Tb)N (chondrite-normalized) ratios for the Western Cape melilitites, Group 1 kimberlites from the Kimberley area (A. P. le Roex & P. Davis, unpublished data, 2001), Rhine Graben melilitites (Wilson et al., 1995) and ocean island (O.I.) melilitites (Clague & Frey, 1982; Hoernle & Schminke, 1993). Shown for reference is a grid of calculated melt compositions generated by varying degrees of melting with varying proportions of residual garnet from a mildly enriched source. Melt compositions are for batch, non-modal melting with modal proportions of 0·5/0·3/0·1/0·1 and melting proportions of 0·25/0·25/0·25/0·25 for olivine, orthopyroxene, clinopyroxene and garnet + spinel, respectively. Partition coefficients for olivine, orthopyroxene and spinel are from McKenzie & O’Nions (1991). Partition coefficients for clinopyroxene and garnet were taken or interpolated from Salters & Longhi (1999), using values for Sm, Tb and Yb of 0·253, 0·339 and 0·445 (in clinopyroxene), and 0·337, 1·10 and 6·07 (in garnet), respectively.

 

The plots in Fig. 6 show that although variable extents of melting in the presence of exotic residual phases may have played a role in producing some of the observed inter-elemental fractionations within and between rocks from the five localities (e.g. the likely role of residual phlogopite in producing the large range in Nb/Rb and Nb/K ratios in the Sutherland melilitites; e.g. Rogers et al., 1992), they cannot explain the extreme enrichments in Ba or the relative depletions in Zr and Hf observed between Alphard–Robertson–Spiegel River and Sutherland–Saltpetre Kop. In particular, the pronounced negative Zr, Hf and Ti anomalies seen in some Saltpetre Kop and Sutherland melilitites cannot be the result of melting in the presence of a residual Ti-bearing phase, such as ilmenite or rutile, because these minerals have a stronger affinity for Nb than for Zr or Hf (Zack & Brumm, 1998; Foley et al., 2000). Thus, melting of a rutile- or ilmenite-bearing assemblage would produce melts with low Nb concentrations and low Nb/Zr ratios (relative to melts derived from rutile- or ilmenite-free assemblages), whereas the Sutherland and Saltpetre Kop melilitites have the highest Nb and Nb/Zr values in our dataset (Fig. 6).

View larger version (29K): [in this window] [in a new window]   Fig. 6. (a) Nb vs Nb/Zr and (b) Nb/Rb vs Ba/Rb ratio plots for the Western Cape melilitites and alkali basalts. It should be noted that the melting trajectories for peridotite assemblages containing the exotic phases ilmenite, rutile, phlogopite and amphibole cannot explain the composition of the samples (primarily from Sutherland and Saltpetre Kop) with the highest Nb/Zr and Ba/Rb ratios. However, these samples are well explained by sources that have experienced addition of small amounts of carbonatite [carbonatite composition from le Roex & Lanyon (1998)]. Melt compositions were calculated assuming batch, non-modal melting with modal proportions of 0·5/0·3/0·12/0·06/0·02 for olivine, orthopyroxene, clinopyroxene, garnet and exotic phase, respectively, and melting proportions of 0·2/0·2/0·3/0·2/0·1 (for ilmenite and rutile-bearing assemblages) and 0·15/0·15/0·3/0·2/0·2 (for phlogopite and amphibole-bearing assemblages). Partition coefficients for olivine, orthopyroxene, clinopyroxene and garnet are from McKenzie & O’Nions (1991), those for phlogopite and amphibole are from Adam et al. (1993), those for ilmenite are from Zack & Brumm (1998) and those for rutile are from Foley et al. (2000).

 

Rather, the data appear to require significant heterogeneity in incompatible element concentrations and ratios of the mantle source regions of the different melilitites. Carbonatite metasomatism, in particular, appears to be a likely cause of the extreme enrichments in Ba and the relative depletions in Zr, Hf and Ti seen in the Sutherland and Saltpetre Kop melilitites, as these are common traits of both carbonatites (e.g. le Roex & Lanyon, 1998) and peridotites that have interacted with carbonatitic fluids (e.g. Yaxley et al., 1991). The involvement of carbonatite metasomatism in the sources of the northern Western Cape melilitites is supported by the observation (Fig. 6) that addition of minor amounts of carbonatite (<3%) to their mantle sources can explain the high Nb/Zr and Ba/Rb (and Ba/Nb) ratios of some Sutherland and most Saltpetre Kop melilitite samples.

Sr, Nd and Pb isotopes
Initial Sr, Nd and Pb isotope data (Table 2) for the Western Cape alkaline magmas confirm that there is significant source heterogeneity between the five localities. The principal variation is between the Alphard Bank lavas, which have the highest initial ²⁰⁶Pb/²⁰⁴Pb and ¹⁴³Nd/¹⁴⁴Nd ratios [e.g. ²⁰⁶Pb/²⁰⁴Pb 20·5; _Nd(t) +4·6], and the lowest ⁸⁷Sr/⁸⁶Sr (0·7029), nearly identical to lavas from St. Helena, and the Saltpetre Kop melilitites, which have moderate ²⁰⁶Pb/²⁰⁴Pb (18·6), with high 7/4 and 8/4 (the vertical distance from the Northern Hemisphere Reference Line in ²⁰⁶Pb/²⁰⁴Pb–²⁰⁷Pb/²⁰⁴Pb and ²⁰⁶Pb/²⁰⁴Pb–²⁰⁸Pb/²⁰⁴Pb space; Hart, 1984), the lowest ¹⁴³Nd/¹⁴⁴Nd (_Nd +1·9) and highest ⁸⁷Sr/⁸⁶Sr (0·7042). Samples from Robertson and Sutherland fall between these two extremes; the Spiegel River isotope data will be discussed below.

View this table: [in this window] [in a new window]   Table 2: Sr–Nd–Pb isotope data for Western Cape melilitites and alkali basalts

 

On the Sr–Nd isotope ratio diagram in Fig. 7, data for the five localities fall on or very near a mixing curve between a HIMU (high time-integrated U/Pb) end-member, represented by the 2 Ga recycled oceanic crust composition of Chauvel et al. (1992), and a metasomatized CLM end-member, represented by the average bulk composition of metasomatized garnet peridotites from the Bultfontein kimberlite at Kimberley, South Africa (Hawkesworth et al., 1990). The isotopic composition of the Western Cape magmas can be reproduced by solid-state mixing of 20% (Saltpetre Kop) to 90% (Alphard Bank) of the HIMU end-member with CLM; mixtures of melts of these two end-members would fall on similar curvilinear mixing trajectories, but would require a range of 50–95% HIMU–recycled crust component. The strong correlation between source trace element characteristics and isotopic composition in the Western Cape magmas is demonstrated by the good correlations of initial Sr isotope ratios with La/Sm_N, Th/Nb, Nb/Zr and Ba/Nb ratios (r² = 0·97, 0·93, 0·84 and 0·8, respectively).

View larger version (28K): [in this window] [in a new window]   Fig. 7. (a) 87Sr/86Sri–Nd(t), (b) 206Pb/204Pbi–87Sr/86Sri, and (c) 206Pb/204Pbi–Nd(t) isotope ratio diagrams showing the Western Cape samples and fields for southern African Group 1 kimberlites (Kramers et al., 1981; Smith, 1983; Davies et al., 2001), whole-rock Archean metasomatized peridotites from Kimberley, South Africa [Hawkesworth et al., 1990; square fields in (b) and (c) show only absolute ranges of initial Pb, Sr, and Nd isotope ratios], ocean island basalts from St. Helena, Bouvet (B), Tristan and Discovery hotspots (T&D) and Vema Seamount (V) (O’Nions et al., 1977; Sun, 1980; le Roex 1985; Chaffey et al., 1989; le Roex et al., 1990; A. P. le Roex & M. D. Kurz, unpublished data, 1991) and Atlantic MORB [Dosso et al. (1991) and references therein]. Assumed end-member compositions: HIMU [recycled crust of Chauvel et al. (1992)] 94 ppm Sr, 9·3 ppm Nd, 0·23 ppm Pb; CLM (Hawkesworth et al., 1990) 110 ppm Sr, 8·0 ppm Nd, 0·5 ppm Pb. It should be noted that to explain the Spiegel River melilitite samples by this mixing scheme, the CLM end-member must have an extremely high Pb concentration (dashed curve, 8 ppb Pb).

 

For the most part, isotopic variations in the Western Cape alkaline rocks fall on similar mixing curves between the HIMU and CLM end-members in the Pb–Sr, Pb–Nd and Pb–Pb isotope diagrams of Figs 7 and 8. Interestingly, isotopic data for on- and off-craton Group 1 kimberlites from southern Africa plot near this same trend and overlap with the Sutherland and Saltpetre Kop data. An exception to this good fit are the samples from Spiegel River, which have low ²⁰⁶Pb/²⁰⁴Pb (18·1–18·3) for their relatively high _Nd and low ⁸⁷Sr/⁸⁶Sr, causing them to plot off the main mixing array in Pb–Sr and Pb–Nd isotope ratio plots.

View larger version (38K): [in this window] [in a new window]   Fig. 8. (a) 206Pb/204Pbi–207Pb/204Pbi and (b) 206Pb/204Pbi–208Pb/204Pbi isotope ratio diagrams showing the Western Cape samples and fields for southern African Group 1 and Group 2 kimberlites, whole-rock Kimberley peridotites, South Atlantic ocean island basalts and Atlantic MORB (abbreviations, references and data sources are the same as in Fig. 6). The Northern Hemisphere Reference Line (NHRL) is from Hart (1984). Also shown are the Stacey–Kramers curves (marked ‘S&K’; Stacey & Kramers, 1975) for terrestrial Pb evolution with tick marks in 250 Ma increments; the rightmost tick mark indicates 0 Ma.

 

The cause of the anomalous Pb isotopic composition of the Spiegel River melilitite is not clear. The fact that it is so close in its trace element composition and Sr and Nd isotopic ratios to the Robertson melilitite (apart from having higher average Pb content) suggests that the primary Spiegel River melilitite magma interacted with a material having high Pb/Sr and Pb/Nd ratios and a relatively unradiogenic Pb isotopic composition (but with fairly high 7/4) probably located in the lithosphere. However, this material could represent either highly Pb-enriched CLM or continental crust. In the mixing scenario described above, a CLM end-member having a Pb concentration of 8 ppm (many times higher than the 0·5 ppm concentration used to construct the solid mixing curves in Fig. 7, but still within the range of Pb concentrations in South African peridotites; e.g. Hawkesworth et al., 1990) could explain the Pb isotopic composition of the Spiegel River melilitite. Lower Pb concentrations would be required if the Spiegel River CLM end-member had a ²⁰⁶Pb/²⁰⁴Pb value <17·8. Alternatively, it is possible that continental crustal contamination could account for the unradiogenic Pb isotopic composition of the Spiegel River melilitite. Similar and even less radiogenic Pb isotopic compositions have been reported for some melilitites and nephelinites from Namaqualand (Rogers et al., 1992), which approach or overlap those of the regional crustal basement (Reid et al., 1997). However, these rocks are more differentiated than those from Spiegel River (with Mg-numbers of 68–49 vs 72–74, respectively; Moore, 1979), and are therefore more likely to have resided in crustal magma chambers before reaching the surface. It would require addition of a large amount (20–25%) of continental crust to produce the Pb isotopic composition of the Spiegel River melilitite from a primary magma with Pb concentrations and isotope ratios similar to those of the Robertson melilitite (conservatively estimating a crustal Pb concentration of 20 ppm and a ²⁰⁶Pb/²⁰⁴Pb ratio of 17). Because the effects of such a large amount of crustal assimilation should be readily apparent in the major and trace element composition of the Spiegel River melilitite but are not seen (i.e. the Spiegel River melilitite is chemically nearly identical to that at Robertson) we tentatively favor a mantle origin for the unradiogenic Pb isotopic composition of the Spiegel River melilitite.

Hf isotopes
Initial Hf isotope ratios span a small range in the Western Cape melilitites [_Hf(t) = +0·5 to +3·4] and do not correlate with Nd, Sr or Pb isotope ratios (Table 3). However, the Hf isotope data are most clearly understood by evaluating where they plot in ¹⁴³Nd/¹⁴⁴Nd–¹⁷⁶Hf/¹⁷⁷Hf isotope space, because of the very similar behavior of the Sm–Nd and Lu–Hf isotope systems. Nearly all mantle-derived basalts that have been measured fall within 7 _Hf units of a regression line drawn through the global oceanic basalt dataset, the ‘mantle array’ of Vervoort et al. (1999; Fig. 9). Recently, however, extremely unradiogenic Hf isotope ratios that fall 8–10 _Hf units below the mantle array have been measured in alkaline lavas from the continental portion of the Cameroon Line (Ballantine et al., 1997) and southern African kimberlites (Nowell et al., 1998b). Like these, samples from the southern Western Cape localities, Alphard Bank, Robertson and Spiegel River, have isotopic compositions that lie well below the mantle array (by 7·1–8·4 _Hf units), outside the range of oceanic basalts, whereas the Sutherland and Saltpetre Kop melilitites fall only about 5 _Hf units below this line. The vertical deviation of a given sample from the mantle array has been termed _Hf (Johnson & Beard, 1993) and in the Western Cape melilitites _Hf correlates positively with ⁸⁷Sr/⁸⁶Sr_i and negatively with ²⁰⁶Pb/²⁰⁴Pb_i and ¹⁸⁷Os/¹⁸⁸Os_i (Fig. 10).

View this table: [in this window] [in a new window]   Table 3: Hf and Os isotope data for the Western Cape olivine melilitites and alkali basalts

 

View larger version (40K): [in this window] [in a new window]   Fig. 9. (a) Nd(t) vs Hf(t) isotope ratio diagram showing the Western Cape melilitites and fields for Group 1 kimberlites (Nowell et al., 1998b), continental and oceanic sections of the Cameroon Line (Ballantine et al., 1997) and data for MORB and OIB (Patchett & Tatsumoto, 1980b; Patchett, 1983; Stille et al., 1986; Salters & Hart, 1991; Chauvel et al., 1992; Salters, 1996; Nowell et al., 1998a; Salters & White, 1998; Blichert-Toft et al., 1999). Fields labelled ‘St. H’ and ‘Rur.-Tub’ enclose data for the islands of St. Helena and Rurutu–Tubuai, respectively. The ‘Mantle array’ line is a linear regression through all OIB and MORB data (Vervoort et al., 1999; Hf = 1·33Nd + 3·19). Also shown is a grid of recycled oceanic crust compositions reflecting the present-day compositions of ancient oceanic crust generated at 1, 1·5 and 2 Ga from a depleted mantle source (formed at 2 Ga) for a range of estimated N-MORB compositions (Hofmann, 1988; Sun & McDonough, 1989; and the mean circum-African MORB from the southern Mid-Atlantic, Southwest Indian and Central Indian ridges: le Roex et al., 1983; Rehkämper & Hofmann, 1997; Janney et al., 1998; Le Roux et al., 2002). (b) Nd(t) vs Hf(t) isotopic diagram; symbols, fields and grid are as in (a). The curves in the CLM fields are intended to show the composition of mantle metasomatized to varying extents by ancient low-degree silicate melts. +, mixtures of depleted mantle (Nd = 11, Hf = 0) with 1-Gyr-old, 1% melts of garnet and spinel peridotite derived from a depleted mantle source. Melt compositions were calculated using the same melting parameters and partition coefficient sources as in Fig. 6. Vectors labelled 1 and 2 schematically show how (1) mixtures of ancient recycled oceanic crust with a large proportion of convecting mantle (with the composition of FOZO/C) could have produced the vertically aligned fields of HIMU ocean islands, whereas (2) ancient recycled oceanic crust that mixed with little or no FOZO/C-type material could produce the shallower slope of the Western Cape data array by mixing with continental lithospheric mantle having been metasomatized by spinel peridotite melts and/or carbonatitic fluids.

 

View larger version (15K): [in this window] [in a new window]   Fig. 10. Variations of Hf with (a) 206Pb/204Pbi, (b) 87Sr/86Sri and (c) 187Os/188Osi isotope ratios. The good correlations of Hf with other isotopic systems should be noted.

 

The Lu–Hf system differs most significantly from the Sm–Nd system in that the parent element Lu is very highly compatible in garnet and therefore melting of garnet-bearing lithologies will produce melts with very low Lu/Hf ratios relative to the source, and these melts will evolve outside of the OIB array toward unradiogenic ¹⁷⁶Hf/¹⁷⁷Hf ratios over time. Thus, generation of the anomalously unradiogenic Hf signatures in the Robertson, Spiegel River and Alphard Bank rocks could have been most easily accomplished if their sources, or main source end-member (if they represent mixtures) incorporated an ancient basaltic melt component generated from a garnet-bearing mantle source.

Os isotopes
Initial ¹⁸⁷Os/¹⁸⁸Os isotope ratios of the melilitites (see Table 3) span a range from mildly supra-chondritic for Saltpetre Kop (0·15), to radiogenic for Robertson and Spiegel River (0·19 and 0·21, respectively). These values extend well outside the range measured for OIB, which generally have ¹⁸⁷Os/¹⁸⁸Os <0·16 (excluding samples with <40 ppt Os, which are often contaminated with altered crust). The Alphard Bank lavas were not measured for Os isotope ratios because they are moderately fractionated and probably have extremely low Os concentrations.

The Os contents of the Western Cape melilitites are more than an order of magnitude lower than those of southern African Group 1 kimberlites (e.g. Pearson et al., 1995b), and are slightly lower than those of melilitites from the Bushmanland plateau of northwestern South Africa (P. E. Janney et al., unpublished data, 2000; Fig. 11) but they are similar to those of primitive basanites and nephelinites from ocean islands (e.g. Widom et al., 1999). Os isotope compositions of the Western Cape melilitites correlate well with inverse Os concentration, forming a linear trend that passes very near the mean Group 1 kimberlite composition (¹⁸⁷Os/¹⁸⁸Os 0·13; Pearson et al., 1995b). Because Os is compatible and Re is moderately incompatible in mantle peridotite, mantle melts typically have low Os concentrations and high Re/Os ratios, and rapidly evolve radiogenic Os isotopic compositions with time. Thus, the linear array in Fig. 11 is simply explained as the result of mixing in the melilitites’ mantle source regions between an unradiogenic, Os-rich peridotite-derived component and a radiogenic, Os-poor, mafic material. However, it is also possible that crustal contamination or a crustal assimilation–fractional crystallization (AFC) process could produce such an array (e.g. Lassiter & Luhr, 2001; Fig. 11). This possibility is evaluated in the discussion below. Os isotopes in the Western Cape melilitites display a negative correlation with ⁸⁷Sr/⁸⁶Sr, and (except for Spiegel River) a positive but strongly curved correlation with ²⁰⁶Pb/²⁰⁴Pb_i (Fig. 12).

View larger version (25K): [in this window] [in a new window]   Fig. 11. Inverse Os concentration vs 187Os/188Osi for the Western Cape melilitites and southern African Cretaceous Group 1 kimberlites (Pearson et al., 1995b). Also shown are data for melilitites from the Bushmanland plateau and lower-crustal xenoliths from the Namaqua–Natal belt (P. E. Janney et al., unpublished data, 2000), and the compositional range of southern African lithospheric peridotite (Pearson et al., 1995a) and uncontaminated ocean island basalts (i.e. those with Os concentrations >40 ppt). The curved vector illustrates the effect of assimilation–fractional crystallization (AFC, occurring at a mass ratio 1·25:1) in which a parental melilitite undergoes assimilation of mean Namaqua–Natal lower crust (187Os/188Os = 0·28, 130 ppt Os) and fractional crystallization (assuming bulk DOs = 15). Tick marks indicate the percentage of crust assimilated. It should be noted that the melilitite and kimberlite data form a linear array, which could be due either to mixing with an ancient mafic component or to an AFC process.

 

View larger version (27K): [in this window] [in a new window]   Fig. 12. 206Pb/204Pbi–187Os/188Osi isotopic diagram showing possible end-member mixing models to explain the isotopic variation in the Western Cape melilitites, HIMU type OIB and Group 1 kimberlites. Symbols as in Fig. 11. Mixing occurs in two stages: (1) an ancient mafic HIMU end-member (calculated assuming an age of 1·5 Ga and 187Re/188Os = 40) with highly radiogenic Os and Pb undergoes solid mixing with depleted mantle (DM) peridotite, resulting in the strongly concave-upward dashed curve; (2) HIMU–DM mixtures undergo melting and mix with melts of metasomatized CLM to produce the Western Cape melilitites, or mix with melts of depleted mantle to produce HIMU OIB (continuous curves). Tick marks on all curves indicate the proportion of mixture derived from the HIMU or hybrid HIMU–DM end-members. It should be noted that the Western Cape melilitites require much less solid mixing with DM than OIB to produce ‘HIMU’ end-members with appropriate Pb and Os isotopic compositions. Results would essentially be the same if a FOZO/C end-member were used in place of DM. OIB data are from Hauri & Hart (1993), Reisberg et al. (1993) and Widom et al. (1999). Kimberlite data are from Smith (1983) and Pearson et al. (1995b). Assumed end-member compositions: HIMU (solid: 200 ppb Pb, 80 ppt Os,), DM (solid: 20 ppb Pb, 1000 ppt Os; melt: 200 ppb Pb, 200 ppt Os), CLM (melt: 1000 ppb Pb, 250 ppt Os). Pb concentration of CLM melt end-member for leftmost melt-mixing curve (through Spiegel River melilitite) is 5000 ppb. HIMU–DM hybrid melt compositions were approximated by calculating the composition of the solid mixture, multiplying the Pb concentration by two and dividing the Os concentration by four. The r values are ratios of Pb/Os in the HIMU (or HIMU–DM hybrid) end-member to those in the DM or CLM end-member.

 

	DISCUSSION AND CONCLUSIONS

TOP ABSTRACT INTRODUCTION TECTONIC SETTING AND AGE... PETROGRAPHY ANALYTICAL METHODS RESULTS DISCUSSION AND CONCLUSIONS REFERENCES

 
The data presented above indicate that the Western Cape melilitites and associated alkali basalts were derived from a range of sources with coherently varying incompatible element enrichments and isotopic compositions. For the most part, the chemical and isotopic variations of these rocks are well explained as the product of binary mixtures of two main source components. One, which we propose to be a metasomatized CLM component, has Sr–Nd–Pb isotopic characteristics approaching the EM 1 mantle end-member, a relatively unradiogenic Os isotope composition and trace element characteristics indicative of carbonatite metasomatism. This component is expressed most strongly on the oldest, thickest lithosphere. The second component, which we propose to be dominantly composed of ancient mafic material, is characterized by radiogenic Os, exceptionally low _Hf at intermediate _Nd, and Sr and Pb isotope ratios similar to HIMU-type OIB. The fact that this proposed mafic HIMU component is expressed most clearly on the thinnest continental lithosphere suggests that it resides either near the base of the lithosphere (e.g. within the thermal boundary layer; Wilson et al., 1995) or in the underlying convecting mantle.

The following discussion aims to characterize the sources of the Western Cape melilitites and determine the nature of the mixing processes that may have played a role in their formation. First, we discuss the possible role of crustal contamination in controlling Os isotopic variations in the melilitites. Second, we employ the unique constraints provided by Os and Hf isotopes to constrain the nature of the source components and mixing processes involved in the genesis of the Western Cape melilitites. Third, we assess whether similar sources and mixing processes can account for the isotopic variations of southern African Group 1 kimberlites. Finally, we explore the possible origins and implications of the HIMU isotopic signature that is strongly expressed in the Western Cape melilitites, but is also nearly ubiquitous in Late Cretaceous to Recent mafic–ultramafic alkaline rocks throughout Africa.

Is Os isotopic heterogeneity in the Western Cape melilitites controlled by crustal contamination?
Despite their near-primary major element compositions, the facts that the Western Cape melilitites do not have high Os contents (mostly <100 ppt) and that they contain olivine megacrysts with relatively low forsterite contents, which must have crystallized from moderately fractionated liquids, could indicate that their parent magmas resided for a time in crustal magma chambers where assimilation may have occurred. Thus, the Os isotopic heterogeneity observed in these rocks, and its correlation with inverse Os concentration, could simply be due to variable contamination with continental crust, combined with Os loss through minor fractional crystallization (e.g. Lassiter & Luhr, 2001), rather than variations in the age, mineralogy or composition of their mantle source regions.

However, there are three lines of evidence that lead us to believe that crustal contamination and assimilation are not responsible for the main observed Os isotopic heterogeneity in the Western Cape melilitites. First, although Os isotope data for lower or upper crust from the Cape Fold and Namaqua–Natal belts are very few, two kimberlite-borne lower-crustal xenoliths from the southwestern Namaqua–Natal belt have been analyzed (a pyroxenite and an amphibolite; P. E. Janney et al., unpublished data, 2000), and these have highly radiogenic Os isotope ratios combined with moderate Os concentrations (¹⁸⁷Os/¹⁸⁸Os = 0·245 and 0·317, and 83 and 192 ppt Os, respectively) similar to mafic lower crust examined elsewhere (e.g. Saal et al., 1998). If these samples are representative, then it would require unrealistically large degrees of assimilation of Namaqua–Natal lower crust (10–25%, based on the range of lower-crust compositions) combined with significant fractional crystallization (12–18%) to produce the full observed variation in Os isotopic composition and concentration (see Fig. 11 for details). The fact that the most radiogenic melilitites, Robertson and Spiegel River, have the lowest Re contents (50–93 ppt) and Re/Os ratios also argues against such a large amount of crustal assimilation.

Second, if crustal contamination were a major contributor to Os isotope heterogeneity in the Western Cape melilitites, then one should expect that this would result in significant Os isotopic heterogeneity within each locality (as a result of variable extents of contamination), correlating with variations in Os concentration. Instead, Os isotope ratios within Saltpetre Kop, Sutherland and Robertson (the only localities with multiple analyses) are nearly constant (varying by <0·003, or three times the nominal 2 error, in ¹⁸⁷Os/¹⁸⁸Os_i) even though Os concentrations vary by 15–40% in each of these localities.

Third, primitive melilitites of the ‘Gamoep’ cluster on the Bushmanland plateau of northwestern South Africa (Moore, 1979; Fig. 1) form an array in Fig. 11 that is essentially collinear with that of the Western Cape melilitites, but these data are displaced toward higher Os concentrations and less radiogenic Os isotope ratios (P. E. Janney et al., unpublished data, 2002). Similar to those from the Western Cape, the Bushmanland melilitites display Sr–Nd–Pb isotopic variations intermediate between EM 1 and HIMU-type compositions (Rogers et al., 1992) that could be indicative of mixing between peridotitic and ancient mafic mantle sources, but given the high Os concentrations of these primitive rocks (105–399 ppt Os, Mg-number = 70–77), it would require extreme amounts of crustal assimilation to produce their Os isotope variations. Thus, the significant Os isotopic variations in the Western Cape and other South African melilitites are not well explained as being primarily the result of crustal contamination. Rather, it appears that they are at least dominantly controlled by Os isotopic heterogeneity in their mantle source regions.

Os and Hf isotopic constraints on mixing processes in the sources of the Western Cape melilitites
The Sr, Nd and Pb isotope compositions of the Western Cape melilitites clearly indicate that one of their compositional end-members is similar to the HIMU mantle component (e.g. Zindler & Hart, 1986) tapped at oceanic islands such as St. Helena or Mangaia (e.g. Chaffey et al., 1989; Woodhead, 1996). From trace element and isotopic studies of ocean island basalts, the HIMU mantle component is most often proposed to originate from oceanic crust that was chemically fractionated during subduction, isolated at a boundary layer in the deep mantle for 1·5–2·5 Gyr, and then returned to the surface in a mantle plume (Hofmann & White, 1982; Chauvel et al., 1992; Woodhead, 1996). The two key aspects of this model, generation of a melt (oceanic crust) and its storage over a long time period, are processes to which the Re–Os and Lu–Hf systems are particularly sensitive. In the Re–Os system, mantle melting effectively fractionates the moderately incompatible parent Re from the highly compatible daughter Os, yielding melts with high Re/Os ratios that rapidly evolve radiogenic Os isotopic compositions. In the Lu–Hf system it is the daughter, Hf, that preferentially migrates into the melt. The retention of Lu in the residue (and the fractionation of the Lu/Hf ratio of the melt) is dependent upon the degree of melting and the amount of residual garnet present. Even if a small amount of garnet remains in the residue or, in the case of mid-ocean ridge or plume melting, if a relatively small proportion of melting occurs in the presence of garnet, Lu/Hf ratios can be lowered enough for the melt to evolve well outside of the main MORB–OIB Nd–Hf isotopic data array within a billion years.

Os and Hf isotopic data for the Western Cape melilitites strongly suggest that, in this instance, the HIMU end-member is dominantly composed of ancient mafic material that was at least partly generated in the presence of garnet. Moreover, the fact that the ¹⁸⁷Os/¹⁸⁸Os_i and _Hf values of the melilitites extend outside the range of HIMU-type ocean island basalts (which have the highest ¹⁸⁷Os/¹⁸⁸Os and lowest _Hf values of all OIB; Hauri & Hart, 1993; Salters & White, 1998) suggests either that the HIMU component was less dilute in the mantle sources of the Western Cape melilitites than in the sources of HIMU-type OIB, or that the composition, origin or age of the HIMU component in this continental setting is different from that in the ocean basins.

One way to address the nature of mixing and extent of dilution of the HIMU component in the sources of the Western Cape melilitite magmas is to examine how isotope systems based on incompatible elements, such as Sr, Nd and Pb, vary with one based on the compatible element Os. On the ²⁰⁶Pb/²⁰⁴Pb–¹⁸⁷Os/¹⁸⁸Os diagram in Fig. 12, the shape of the Western Cape melilitite data array, excluding Spiegel River, is mildly convex upward. If this represents a mixing array, the Pb/Os ratio of the HIMU component must be less than that of the end-member having unradiogenic Pb and Os (presumably metasomatized lithospheric peridotite). This is very different from what would be expected for mixing of solid mafic material with peridotite, where the extremely low Pb/Os ratio of the peridotite would produce a strongly concave-upward mixing array (e.g. Becker, 2000), similar to the HIMU-depleted mantle (DM) mixing curve in Fig. 12.

The most plausible way for such an array to be produced is by mixing taking place dominantly in the melt phase. Small-degree melts of metasomatized CLM would probably have high Pb abundances and low Os concentrations relative to the residual peridotite. Even so, for the array in Fig. 12 to be the product of pure binary mixing, melts of the mafic end-member would have to have extremely low Pb/Os ratios, lower even than typical MORB. This might be possible if the mafic end-member consisted largely of cumulates (such as an oceanic crustal gabbro) and was itself almost entirely melted. If this were the case, Pb/Os ratios of melts of a HIMU end-member might be as low as 1000 (i.e. values typical of the lowest Pb/Os oceanic crustal gabbros; Hart et al., 1999). South African Group 2 kimberlites, which may represent low-degree melts of metasomatized lithosphere, have Pb/Os ratios as high as 4000–10 000 (e.g. Fraser et al., 1985; Pearson et al., 1995b) and so binary mixtures of these two types of melts might satisfy the required contrast in Pb/Os ratios.

However, a mafic component need not have an extraordinarily low Pb/Os ratio to produce the curvature of the Western Cape melilitite array in Fig. 12 if the mixing is not purely binary. Petrological arguments suggest that, before melting, the mafic HIMU end-member experienced some amount of solid-phase mixing with peridotite (from either the convecting mantle or CLM), which would significantly raise the Os content and lower the Pb/Os ratio (and ¹⁸⁷Os/¹⁸⁸Os value) of the mixture (see Fig. 12). Such mixing appears to be necessary because it would be nearly impossible to generate sufficiently SiO₂- and Al₂O₃-poor melilitite magmas with moderate Os contents (50 ppt) from a purely basaltic or pyroxenitic source (e.g. Kogiso & Hirschmann, 2001) but it would not be so for a mixed peridotite–pyroxenite (i.e. roughly wehrlitic) lithology (Lee & Wyllie, 2000). The amount of peridotite incorporated into the HIMU end-member source of the Western Cape melilitites need not be very large, as pyroxenitic sources appear able to produce silica-undersaturated, CaO- and MgO-rich melt compositions approaching those of olivine melilitites (Schiano et al., 2000). We propose that a two-stage mixing model is most likely, in which (1) an ancient mafic–basaltic material such as subducted oceanic crust undergoes solid-state mixing with a small proportion of peridotite in the convecting mantle or lithosphere, followed by (2) melting of this composite material and blending of the resulting melts with low-degree melts of metasomatized CLM.

An important implication of this model is that the Western Cape melilitites and HIMU-type OIB may have both tapped the same type of mafic HIMU material but, if this is the case, the HIMU end-member sampled by ocean island magmatism must have undergone much more solid-phase mixing with peridotite before melting. Fluid-dynamic models of mantle plume ascent (e.g. Hauri et al., 1994) suggest that a large extent of solid-phase mixing occurs between an upwelling plume and surrounding ambient mantle as a result of entrainment. Such extensive mixing may explain the relatively unradiogenic Os isotopic compositions of HIMU OIB relative to the Western Cape melilitites. For the melt mixing models shown in Fig. 12 (see caption for details), the Os isotopic range of HIMU OIB can be explained by solid-phase mixing of a mafic HIMU end-member with 75–80% peridotite, whereas the Os isotopic range of the Western Cape melilitites is explained by solid mixing of a mafic end-member with only about 10–40% peridotite. Although melt mixing is probably also an important process in the generation of OIB (e.g. Class & Goldstein, 1997), in this scenario the HIMU end-member would have already been diluted by a much larger proportion of solid peridotite before the commencement of melting. One possible interpretation of this difference in mixing proportions is that the HIMU end-member beneath the Western Cape was not delivered by a conventional mantle plume (i.e. a narrow jet of hot, low-viscosity material rising from great depth in the mantle).

Contrasting mixing and dilution processes may also help to explain the difference in Nd–Hf isotope compositions between OIB and the melilitites and alkali basalts from the Western Cape. Oceanic basalts with strong HIMU isotopic affinities, such as from the islands of Tubuai and Rurutu in the South Pacific and St. Helena in the South Atlantic, tend to display lower _Hf values for their moderate positive _Nd values than other OIB, falling at most 7 _Hf units below the mantle array (e.g. Salters & White, 1998). The Western Cape localities with the strongest HIMU affinities, Alphard Bank and Robertson, plot just below the St. Helena field, with identical _Nd but slightly lower _Hf.

In the _Hf–_Nd diagram in Fig. 9, a grid of recycled oceanic crust compositions is shown calculated for a range of isolation times and Lu/Hf ratios (encompassing the range of estimated mean compositions for global MORB and MORB from circum-African spreading centers). All compositions of 1–2 Ga crust fall well below the mantle array (with _Hf values as low as -16), and compositions near the center of the grid make plausible low-_Hf end-members for HIMU OIB and the Western Cape melilitites (as well as for alkaline basalts from the continental portion of the Cameroon Line; Ballantine et al., 1997).

The different shapes of the St. Helena, Cook–Austral Islands and Western Cape melilitite fields in Fig. 9 suggest an explanation for the lower _Hf values of the Western Cape magmas relative to HIMU OIB. The fields for the ocean islands are elongated almost vertically in _Nd–_Hf isotope space, suggesting that they are the product of mixing between a low-_Hf HIMU end-member and a component with _Nd of +4 to +6 and _Hf that lies on or near the OIB array. The hypothetical entrained deep mantle components ‘FOZO’ and ‘C’ (Hart et al., 1992; Hanan & Graham, 1996), which are proposed to be common to all plumes, have appropriate Nd isotopic compositions (_Nd +5) for this end-member. In contrast, the Western Cape melilitites form a linear array of essentially constant _Hf values that is nearly orthogonal to the HIMU–FOZO mixing array.

A regression line drawn through the Western Cape data (see Fig. 9b) intersects the mantle array at _Nd -3, a typical value for metasomatized continental peridotite (e.g. Hawkesworth et al., 1990). In accordance with our inferences from Re–Os and other isotope systems, the linear array described by the Western Cape melilitites in Fig. 9b is most easily explained as a mixing trajectory between a low-_Hf HIMU end-member and a metasomatized lithospheric component with a _Hf value near or greater than zero, with little mixing with convecting mantle components (e.g. FOZO or depleted MORB-source mantle). The isotopic composition of this lithospheric end-member could have been produced by ancient metasomatism by small-degree melts derived from spinel peridotite (which tend to have moderate Lu/Hf ratios; e.g. Beard & Johnson, 1993), possibly with some involvement of carbonatitic melts (which typically have high Lu/Hf ratios), but not by small-degree melts of garnet-bearing lithologies. Thus, both Os–Pb and Nd–Hf isotope variations in the Western Cape melilitites indicate much less mixing of the HIMU end-member source of the Western Cape melilitites and alkaline basalts with convecting mantle than in the sources of HIMU-type OIB. Dilution of the HIMU end-member tapped by the Western Cape melilitites appears to have occurred mainly during interaction with the African lithosphere.

Do the Western Cape melilitites and southern African Group 1 kimberlites share similar sources?
Smith (1983) established that southern African Group 1 kimberlites fall within a relatively narrow Sr–Nd–Pb isotopic range compared with the gap between Group 1 and Group 2 kimberlites. However, compared with many intraplate volcanic provinces Group 1 kimberlites display a significant isotopic variation (e.g. Smith, 1983; Davies et al., 2001; see Fig. 7) ranging from mild EM 1 to mild HIMU isotopic compositions. Relatively little attention has been paid to the processes responsible for isotopic variation within the Group 1 kimberlite population, but their identification seems essential for discerning the ultimate source(s) of kimberlite magmatism. A mixing scenario involving a mafic HIMU component and metasomatized CLM, like that which appears to explain the Western Cape melilitites, may also provide an explanation for isotopic heterogeneity within southern African Group 1 kimberlites.

In significant ways, the Western Cape melilitites appear to be analogous to Group 1 kimberlites: the two groups overlap to a large extent in Sr–Nd–Pb isotope ratio space, and the two melilitite localities compositionally most similar to kimberlites, Sutherland and Saltpetre Kop, are located only a few hundred kilometers from true Group 1 kimberlite occurrences (e.g. Smith et al., 1994). However, the Western Cape melilitites display a significantly wider range of trace element and Sr, Nd and Pb isotope ratio variations (especially in ²⁰⁶Pb/²⁰⁴Pb) than southern African Group 1 kimberlites, and because of this they much more clearly illustrate mixing of an EM 1-type with a HIMU-type mantle end-member.

Such a source mixing relationship may also be responsible for the smaller isotopic variations within Group 1 kimberlites, but the isotopic data for these rocks do not form such an easily interpretable pattern. Two factors in particular may have contributed to this. First, the relationship between isotopic composition and lithospheric thickness displayed by the Western Cape melilitites would predict that true Group 1 kimberlites, which mainly intrude thick, cratonic lithosphere, would have a large source contribution from a metasomatized CLM component. This would result in the kimberlite isotopic compositions being concentrated closer to the EM 1-like CLM end-member, and therefore they should not display the large ranges in ²⁰⁶Pb/²⁰⁴Pb that best characterize EM 1–HIMU mixing. Second, Group 1 kimberlites are present across a broad area of southern Africa, penetrating lithosphere of widely varying ages and metasomatic histories (e.g. Walker et al., 1989; Pearson et al., 1995a). Therefore, the CLM end-member would probably be much more heterogeneous for Group 1 kimberlites as a whole than for the Western Cape melilitites, which only penetrate a small area of two Proterozoic terranes. Greater Sr–Nd isotopic scatter would be predicted, and is observed, at the low-²⁰⁶Pb/²⁰⁴Pb, ‘lithospheric’ end of the kimberlite data array (Fig. 7). Even with these complications, it is clear that most of the Sr–Nd–Pb isotopic variation in southern African Group 1 kimberlites could be explained by mixing between a heterogeneous metasomatized CLM end-member and a compositionally more uniform component with a moderate to strong HIMU isotopic affinity. However, the relatively large degree of isotopic scatter within the kimberlite field, combined with the fact that some Group 1 kimberlite megacrysts have initial Sr, Nd and Pb isotopic compositions approaching those of MORB (e.g. Hops, 1992; Smith et al., in preparation) does not allow us to exclude the possibility that additional components, such as depleted mantle or FOZO/C, may contribute significantly to kimberlite sources.

Os and Hf isotope data offer some additional insights into kimberlite sources, but significant ambiguities remain. Os isotope ratios of southern African Group 1 kimberlites extend to supra-chondritic values that are more radiogenic than those found in southern African peridotite xenoliths (i.e. ¹⁸⁷Os/¹⁸⁸Os_i up to 0·14; Pearson et al., 1995b), which could indicate a contribution from an ancient mafic source material. However, Os isotope compositions of the kimberlites are much less radiogenic than those of the Western Cape melilitites, and they fall far below the melilitite data array in Pb–Os isotope space (Fig. 12). This may or may not be significant, as the bulk kimberlite samples analyzed by Pearson et al. (1995b) contain numerous fine xenocrysts from disaggregated peridotite, which could not be entirely removed by careful picking. Because of its very high Os concentrations, even small amounts of continental peridotite contamination could easily shift the Group 1 kimberlite field toward significantly lower ¹⁸⁷Os/¹⁸⁸Os ratios, with little or no effect on Pb isotope ratios. Thus, the ‘magmatic’ ¹⁸⁷Os/¹⁸⁸Os values of Group 1 kimberlites (i.e. unaffected by contamination) remain poorly known and the available Os isotope data do not place strong constraints on the possible involvement of an ancient mafic source material in the genesis of kimberlites.

In their pioneering study, Nowell et al. (1998b) discovered that both Group 1 and Group 2 kimberlites from southern Africa define arrays in Nd–Hf space that are much steeper than the OIB array, with many samples having Nd–Hf isotope compositions that lie far below the OIB field (indeed, some samples fall further below the mantle array than any other mantle-derived magmas yet measured). However, the shape of the field for Group 1 kimberlite bulk-rock samples suggests mixing between an OIB-like component with moderate positive _Nd that lies near the mantle array and an extremely low-_Hf end-member that also has a relatively unradiogenic Nd isotope composition (_Nd 0). This is distinct from the low-_Hf component tapped by the Western Cape melilitites (which has moderately positive _Nd) and requires a time-integrated Sm/Nd similar to or less than that of Bulk Earth. Such low Nd isotope ratios appear to rule out an origin from recycled oceanic crust, but are more consistent with a CLM source metasomatized by small-degree silicate melts derived from a garnet-bearing mantle lithology (e.g. Beard & Johnson, 1993; Nowell et al., 1998b). Complicating this interpretation is the fact that recent Hf–Nd isotopic data for Cr-poor garnet and ilmenite megacrysts from Group 1 kimberlites (Nowell et al., 1999), which probably crystallized from a primitive kimberlite magma before significant interaction or contamination with continental mantle or crust, have low _Hf values similar to the bulk Group 1 kimberlites, but moderate positive _Nd values that could be consistent with a recycled oceanic crustal source.

Thus, the available data for Group 1 kimberlites indicate that these complex rocks share several compositional attributes of the Western Cape melilitites and could have significant source contributions from metasomatized CLM and an ancient mafic melt component, such as subducted oceanic crust. However, the significantly less radiogenic and more restricted range of ¹⁸⁷Os/¹⁸⁸Os ratios, combined with the different position and orientation of the kimberlite data array in Nd–Hf isotope space are problematic for explaining the Group 1 kimberlite data as being primarily derived from these two components. These problems may largely be due to shallow mantle and/or crustal contamination, but a convincing model of the source components contributing to Group 1 kimberlite magmatism must await a more complete, combined Sr–Nd–Pb–Hf–Os isotope dataset collected using samples carefully selected and prepared to minimize shallow mantle and crustal contamination.

Constraints on the origin of the HIMU source component of the Western Cape melilitites and other African alkaline rocks
Strong HIMU isotopic signatures (defined here as ²⁰⁶Pb/²⁰⁴Pb > 20, ⁸⁷Sr/⁸⁶Sr < 0·7032, _Nd between +2·5 and +6) are widespread across Africa in Late Cretaceous to Recent (i.e. 100 Ma) mafic alkaline volcanic rocks, occurring in locations as diverse as the Western Cape of South Africa, the East African Rift, the Cameroon Line, the Hoggar province of Algeria, the Meidob Hills of Sudan and along the Mediterranean margin of Africa (Allègre et al., 1981; Halliday et al., 1988; Simonetti & Bell, 1994, 1995; Esperança & Crisci, 1995; Franz et al., 1999; see Fig. 13). Similarly strong signatures (e.g. ²⁰⁶Pb/²⁰⁴Pb > 20·4) have also been detected in sub-calcic clinopyroxene megacrysts and high-temperature sheared peridotites from Cretaceous southern African Group 1 kimberlites (Walker et al., 1989; Smith et al., in preparation), which appear to represent early crystallization products of, and lithospheric peridotite pervasively infiltrated by, primitive kimberlite liquids, respectively (Jones, 1987; Hops, 1992).

View larger version (38K): [in this window] [in a new window]   Fig. 13. 206Pb/204Pb vs Nd(t), showing isotopic data for products of Late Cretaceous–Recent alkaline magmatism (e.g. alkali basalts, basanites, nephelinites) in Africa that extend to extreme HIMU-type compositions. The vectors associated with the HIMU field show the effects of varying age, U/Pb fractionation (either during crustal generation or subduction) and partial melting involved in the formation of an ancient oceanic crustal HIMU protolith. It should be noted that, of the sample suites shown, only lavas from the Hoggar (Allègre et al., 1981) and sub-calcic clinopyroxene megacrysts from Group 1 kimberlites (C. B. Smith et al., in preparation) show evidence for significant mixing between HIMU and the convecting (depleted) upper mantle. For reference, fields for St. Helena (SH), Tubuai (T) and Mangaia (M) are shown (Chauvel et al., 1992; Woodhead, 1996). Data sources are Simonetti & Bell (1994, 1995), Paslick et al. (1995), Franz et al. (1999) and this study.

 

Volcanic rocks with equally strong HIMU affinities are rare on other continents. Only the volcanoes of the West Antarctic Rift System (e.g. Rocholl et al., 1995; Hart et al., 1997) have produced lavas with ²⁰⁶Pb/²⁰⁴Pb ratios comparable with those from the African localities listed above, although volcanic rocks with weaker HIMU affinities are present in west–central Europe and northeastern North America (e.g. Eby et al., 1985; Hegner et al., 1995). Basalts from several oceanic island or seamount groups adjacent to Africa also possess HIMU signatures (e.g. the Canary and Comores islands and Vema Seamount; e.g. Hoernle et al., 1991; Späth et al., 1996) but, with the exception of St. Helena, these compositions are not as extreme (particularly with respect to ²⁰⁶Pb/²⁰⁴Pb) as those in the most isotopically extreme alkaline volcanic provinces on the African continent.

Most workers studying individual African magmatic provinces have concluded that the HIMU sources of these magmas were supplied directly by mantle plumes (e.g. Vidal et al., 1991; Class et al., 1994; Lee et al., 1994; Franz et al., 1999; Bell & Tilton, 2001). If correct, this hypothesis would require an extraordinarily high concentration of mantle plumes beneath the African plate, nearly all having HIMU isotopic affinities. However, apart from the major igneous provinces in the Afar region and the Kenya Rift, there is little evidence in the temporal and spatial expression of African alkaline magmatic provinces to indicate that they are the product of long-lived (40 Myr), conventional mantle plume activity similar to that giving rise to the Tristan or Hawaiian hotspots (e.g. Ashwal & Burke, 1989; compare Morgan, 1981). An illustrative example is the non-age-progressive late Tertiary–Recent, 1200-km-long Cameroon Line volcanic province, which is superimposed upon the projected 90–120 Ma portion of the St. Helena seamount chain trace (Fitton & Dunlop, 1985; O’Connor & le Roex, 1992). Finally, the differences in Nd–Hf isotopic compositions between African alkaline rocks, such as the Western Cape melilitites, Cameroon Line lavas and southern African Group 1 kimberlites, and ocean island basalts suggest a significant difference in the source composition and/or the nature of source mixing for these continental magmas relative to that for OIB.

We propose that the presence of the HIMU signature in African alkaline magmas is too widespread and compositionally extreme to be the result of a coincidental concentration of numerous conventional mantle plumes, all having a HIMU isotopic affinity. Rather, it seems more likely that it is the result of a regional feature in the African lithosphere or underlying convecting mantle. Possible alternatives to a conventional mantle plume origin for the widespread HIMU signature include: (1) ancient oceanic crust or mantle-derived melts incorporated into the African CLM and allowed to age in situ; (2) a widely distributed ancient mafic component in the asthenospheric mantle sampled by very low-degree partial melting in the sub-African region; (3) ‘plume-type’ HIMU source material advected into the shallow mantle by a laterally broad (>1000 km radius) zone of concentrated mantle upwelling (i.e. a ‘superplume’, e.g. Gurnis et al., 1999). Although a full investigation of these alternatives is beyond the scope of this paper, we will discuss their principal merits and flaws in the light of the available geochemical and geological evidence.

In the case of the Western Cape melilitite province, a lithospheric origin for the HIMU end-member seems unlikely. Although numerous eclogite xenoliths with chemical and isotopic compositions consistent with a hydrothermally altered oceanic crust protolith have been found in African kimberlites (e.g. MacGregor & Manton, 1986) these are largely restricted to Archean cratonic terranes. The Western Cape samples showing the strongest HIMU affinity are located on the Pan-African age Cape Fold Belt lithospheric terrane, which was initially stabilized, or at least subjected to intense deformation and partial melting, at 600–800 Ma (Sönghe & Halbich, 1983). Such an event would probably largely erase any major pre-existing mineralogical heterogeneities and would probably limit the maximum age of a mafic lithospheric component to 0·8 Ga, which is insufficient time to develop the ²⁰⁶Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb and _Hf characteristics of the Alphard Bank lavas from subducted oceanic crust, or small-degree melt veins derived from the depleted upper mantle (e.g. Wilson et al., 1995). More generally, the facts that no kimberlite-derived eclogite xenoliths analyzed thus far have HIMU-like Pb isotopic compositions (i.e. all measured ²⁰⁶Pb/²⁰⁴Pb ratios are <19·2; e.g. Jacob & Jagoutz, 1995; Viljoen et al., 1996) and that South African high-temperature peridotites with HIMU Sr–Nd–Pb compositions also have sub-chondritic Os isotope ratios (Walker et al., 1989), indicative of recent, rather than ancient melt addition, are both inconsistent with widespread in situ evolution of a HIMU source reservoir within the African lithosphere.

Another possible source of the widespread HIMU signature in African volcanic rocks is mafic material present in the convecting upper mantle, which might be efficiently extracted by extremely low degrees of partial melting beneath continental lithosphere (e.g. Fitton & Dunlop, 1985; Hegner et al., 1995). Chemical and isotopic evidence from mid-ocean ridge and near-ridge seamount lavas suggests the widespread presence of mafic (pyroxenitic) material in the asthenosphere (e.g. Hirschmann & Stolper, 1996; Lundstrom et al., 2000), much or most of this material apparently having been derived from recycled oceanic crust of indeterminate age (e.g. Niu & Batiza, 1997; Eiler et al., 2000). However, two lines of evidence argue that the HIMU mantle signature in African volcanic rocks (and HIMU OIB) is not derived from a mafic asthenospheric component. First, perhaps the best estimate of the composition of such a component is based on highly incompatible element-enriched lavas from near-ridge seamounts on the flanks of the extensively studied northern East Pacific Rise (e.g. Niu & Batiza, 1997), located far from major hotspots. The most extreme of these lavas have Sr and Nd isotope ratios that are similar to those of HIMU OIB (⁸⁷Sr/⁸⁶Sr 0·7030; _Nd +5; Zindler et al., 1984), but have ²⁰⁶Pb/²⁰⁴Pb ratios that are considerably less radiogenic (19·4; Graham et al., 1988). These lavas form linear arrays in ²⁰⁶Pb/²⁰⁴Pb–⁸⁷Sr/⁸⁶Sr and ²⁰⁶Pb/²⁰⁴Pb–_Nd space that do not point directly toward the pure HIMU end-member composition (e.g. St. Helena, Mangaia). Rather, they appear to be pointing toward a composition intermediate between HIMU and EM 2, possibly indicating mixing and homogenization of recycled crust with another material, such as pelagic sediment (e.g. Eiler et al., 2000). Similar results have been observed for enriched MORB and near-axis seamount basalts in other ocean basins (e.g. Dosso et al., 1991).

Second, continental magmas associated with rifts and other extensional environments, which are believed to constitute low-degree melts of continental lithosphere and asthenosphere (Perry et al., 1987), do not typically possess HIMU signatures. Although it is true that young extension-related lavas with moderate to very strong HIMU isotopic compositions are present in West Antarctica and west–central Europe, they are virtually absent in other extensional provinces outside of Africa (e.g. Ewart et al., 1988; Beard & Johnson, 1997; Flower et al., 1998; Yarmolyuk et al., 1998). In the case of west–central Europe, there is seismic evidence for a large-scale mantle upwelling feature underlying the region in which HIMU-type lavas have been erupted that may be the source of this geochemical signature (Hoernle et al., 1995). Plume involvement in the West Antarctic rift system been proposed on the basis of the spatial distribution of lithospheric extension and volcanism (Behrendt et al., 1992), as well as geochemistry (Rocholl et al., 1995). Thus, apart from areas where there is some indication of asthenospheric contamination by plume-type material, there is little evidence to support the existence of a widespread HIMU source in the asthenosphere.

Finally, it is possible that the widespread HIMU signature in African alkaline magmas is supplied by the same deep mantle reservoir believed to feed HIMU ocean islands (e.g. Hofmann & White, 1982; Chauvel et al., 1992; Hauri & Hart, 1993), but that this material has been advected into the shallow sub-African mantle via a laterally broad mantle upwelling phenomenon. Studies of the topography, uplift history and seismic tomography of the African plate and the underlying mantle indicate that southern and eastern Africa is underlain by a very large-scale, buoyant, low seismic velocity structure extending from just above the core–mantle boundary to near the base of the African lithosphere (Lithgow-Bertelloni & Silver, 1998; Gurnis et al., 1999). This feature represents one of the two largest low-velocity structures in the deep mantle (Ritsema et al., 1998) and appears to represent both a thermal and an intrinsic density anomaly (Gurnis et al., 1999). A smaller and possibly related low-velocity feature has been identified in the upper mantle beneath a region extending from northwestern Africa across west–central Europe to the northeastern Atlantic (Hoernle et al., 1995). The numerous recently active volcanic provinces of this region span wide ranges of Sr, Nd and Pb isotope ratios but converge at an isotopic composition of ²⁰⁶Pb/²⁰⁴Pb 20, ⁸⁷Sr/⁸⁶Sr 0·703 and _Nd +5. Hoernle et al. (1995) interpreted this HIMU-like composition to represent that of the low-velocity mantle anomaly. If a similar composition dominates the low-velocity material upwelling within a long-lived (90 Ma) African ‘superplume’, it could provide a plausible explanation for the widespread presence of African alkaline volcanic rocks with HIMU isotopic affinities, as well as the behavior of Cenozoic African intraplate volcanism in space and time, which is not well explained by the classical mantle plume model (e.g. Ashwal & Burke, 1989).

On the basis of the relatively good correlation between the distribution of HIMU-type alkaline magmatism and widespread low seismic velocities in the mantle beneath the African plate, we tentatively favor an origin for the HIMU source of Late Cretaceous to recent African magmatism (including generation of the Western Cape melilitites) from an African superplume advecting subducted oceanic crustal material into the shallow sub-African mantle. However, understanding of the physical and chemical structure of this feature remains poor and many questions must be resolved before this hypothesis can be regarded as more than speculative.

One important question relates to whether the African superplume was in existence beneath southern Africa during the period when the Western Cape melilitites were emplaced. Although seismology gives us no clue as to the longevity or past location of this feature, a study of the Agulhas Plateau (Gohl & Uenzelmann-Neben, 2001), located in the Southern Ocean about 600 km SE of the Alphard Bank, indicates that this sizeable edifice (2·5 km high and >3 x 10⁵ km² in area) was constructed by effusive oceanic volcanism between 80 and 100 Ma. This suggests the presence of an unusually warm upper mantle in the same general time and region as the earliest magmatic activity in the Western Cape melilitite province.

A second, more difficult question is why this superplume should appear to be so dominated by the HIMU component. This may be illusory, as plume materials with both HIMU and EM-type isotopic affinities may have been supplied to African alkaline magmas (e.g. Bell & Tilton, 2001), but the EM components have not been recognized as being plume-derived as a result of their isotopic similarity to continental lithospheric mantle and crust. However, the tendency toward increasingly strong HIMU signatures with decreasing age and lithospheric thickness in many African volcanic provinces (e.g. Vidal et al., 1991; Class et al., 1994; Lee et al., 1994; Furman & Graham, 1999) is more broadly supportive of an origin for EM-type compositions dominantly from lithospheric sources. This suggests that the African superplume is indeed HIMU dominated.

Additionally, it is important to determine why at least some of the HIMU source material supplied to African alkaline volcanics, including the Western Cape melilitites and continental Cameroon Line volcanics, appears from Hf and Os isotope constraints to have undergone less mixing with FOZO/C- or depleted-type mantle than that supplied to ocean islands. One possibility is that parcels of oceanic crust entrained within a broad upwelling experience less mixing with surrounding ambient mantle than material entrained in a narrower (and presumably higher-velocity, e.g. Sleep, 1990) conventional mantle plume. However, this does not explain why HIMU ocean islands above other areas of inferred large-scale mantle upwelling (e.g. the eastern Atlantic and south central Pacific; Hoernle et al., 1995; Masters et al., 1996) lack extreme Os and Hf isotopic characteristics. An answer to this question must await further Hf and Os isotopic characterization of continental alkaline volcanic provinces to determine if such ‘undiluted’ HIMU signatures are indeed characteristic of alkaline magmatism in Africa.

We speculate that the column of warm ascending mantle constituting the African superplume may have advected numerous blobs composed dominantly of subducted oceanic crustal materials into the asthenosphere beneath southwestern Africa. The 360-km-long, 18 Myr duration magmatic lineation comprising the Western Cape melilitite province may represent the trace of one of these blobs, moving only slowly in the mantle reference frame, as it shed partial melts that migrated upward into the northwesterly drifting lithosphere (Fig. 14). These melts would have interacted with the CLM, melting and mixing with the easily fusible, volatile-rich metasomatic components. A large proportion of these hybrid melts would most probably have ‘frozen out’ as silicate veins within the lithosphere, but in a few locations, because of local lithospheric weaknesses or intermittent episodes of high melt production, some melts may have erupted at the surface as alkaline volcanics or diatremes forming the five main localities of the Western Cape melilitite province.

View larger version (47K): [in this window] [in a new window]   Fig. 14. Cross-section schematic representation of two time slices in the mantle reference frame showing a possible scenario for the genesis of the Western Cape melilitites. A blob composed largely of ancient subducted oceanic crust is brought into the upper mantle beneath southern South Africa by a deep-seated regional mantle upwelling. This blob undergoes melting and acts for a short time as a ‘mini-hotspot’. Melts from the blob ascend into and extensively interact with the lithosphere, partially melting it. A large proportion of the resulting hybrid melts freeze out within the lithosphere, but in a few locations the melts reach the surface to produce diatremes or volcanic vents. (See text for discussion.)

 

	ACKNOWLEDGEMENTS

 
We are grateful to Fran Pocock and Andreas Späth for assistance with the XRF and ICPMS instruments at UCT. A. Späth is also thanked for his hospitality to the first author during trips to Cape Town to carry out ICPMS work. John Rogers was instrumental in unearthing the Alphard Bank dredge samples from the UCT rock storage facility. Steve Richardson generously shared his time and knowledge with us in the UCT Radiogenic Isotope Facility. Mary Horan developed the initial Hf separation procedure used at DTM, and she and Tim Mock made the isotopic work at DTM a pleasure. David Bell is thanked for many stimulating discussions on the nature of the deep continental lithosphere and the sources of kimberlites and megacrysts. D. G. Pearson, G. M. Nowell and C. B. Smith generously shared their data and ideas on kimberlite petrogenesis and are thanked for comments on early versions of this manuscript. David Graham, John Lassiter and Nick Rogers provided detailed, constructive reviews that improved the clarity and rigor of the manuscript. This work was supported by the South African National Research Foundation (to A.L.R.) and the US National Science Foundation (to R.W.C. and P.E.J.).

	FOOTNOTES

 
^*Corresponding author. Present address: Department of Geology, The Field Museum of Natural History, 1400 Lake Shore Drive, Chicago, IL 60605, USA. Telephone: (312) 665-7099. Fax: (312) 665-7641. E-mail: pjanney{at}fmnh.org

	REFERENCES

TOP ABSTRACT INTRODUCTION TECTONIC SETTING AND AGE... PETROGRAPHY ANALYTICAL METHODS RESULTS DISCUSSION AND CONCLUSIONS REFERENCES

 
Adam, J., Green, T. H. & Sie, S. H. (1993). Proton microprobe determined partitioning of Rb, Sr, Ba, Y, Zr, Nb and Ta between experimentally produced amphiboles and silicate melts with variable F content. Chemical Geology 109, 29–49.[Web of Science]

Allègre, C. J., Dupré, B., Lambret, B. & Richard, P. (1981). The subcontinental versus suboceanic debate, I. Lead–neodymium–strontium isotopes in primary alkali basalts from a shield area: the Ahaggar volcanic suite. Earth and Planetary Science Letters 52, 85–92.[Web of Science]

Ashwal, L. D. & Burke, K. (1989). African lithospheric structure, volcanism and topography. Earth and Planetary Science Letters 96, 8–14.[Web of Science]

Ballantine, C. J., Lee, D.-C. & Halliday, A. N. (1997). Hafnium isotopic studies of the Cameroon Line and new HIMU paradoxes. Chemical Geology 1997, 111–124.

Beard, B. L. & Johnson, C. M. (1993). Hf isotope composition of late Cenozoic basaltic rocks from northwestern Colorado, USA: new constraints on mantle enrichment processes. Earth and Planetary Science Letters 119, 495–509.[Web of Science]

Beard, B. L. & Johnson, C. M. (1997). Hafnium isotope evidence for the origin of Cenozoic basaltic rocks from the southwestern United States. Journal of Geophysical Research 102, 20149–20178.

Becker, H. (2000). Re–Os fractionation in eclogites and blueschists and the implications for recycling of oceanic crust into the mantle. Earth and Planetary Science Letters 177, 287–300.[Web of Science]

Behrendt, J. C., LeMasurier, W. E. & Cooper, A. K. (1992). The West Antarctic Rift System—a propagating rift ‘captured’ by a mantle plume. In: Kaminuma, K. & Yoshida, Y. (eds) Recent Progress in Antarctic Earth Science. Tokyo: Terra, pp. 315–322.

Bell, K. & Tilton, G. R. (2001). Nd, Pb and Sr isotopic compositions of East African carbonatites: evidence for mantle mixing and plume inhomogeneity. Journal of Petrology 42, 1927–1945.[Abstract/Free Full Text]

Blichert-Toft, J. & Albarède, F. (1997). The Lu–Hf isotope geochemistry of chondrites and the evolution of the mantle–crust system. Earth and Planetary Science Letters 148, 243–258.[Web of Science]

Blichert-Toft, J., Chauvel, C. & Albarède, F. (1997). Separation of Hf and Lu for high-precision isotope analysis of rock samples by magnetic sector-multiple collector ICP-MS. Contributions to Mineralogy and Petrology 127, 248–260.[Web of Science]

Blichert-Toft, J., Frey, F. A. & Albarède, F. (1999). Hf isotope evidence for pelagic sediments in the source of Hawaiian basalts. Science 285, 879–882.[Web of Science][Medline]

Boctor, N. Z. & Yoder, H. S. (1986). Petrology of some melilite-bearing rocks from Cape Province, Republic of South Africa: relationship to kimberlites. American Journal of Science 286, 513–539.[Abstract/Free Full Text]

Brey, G. (1978). Origin of olivine melilitites—chemical and experimental constraints. Journal of Volcanology and Geothermal Research 3, 61–88.[Web of Science]

Burke, K. (1996). The African Plate. South African Journal of Geology 99, 341–409.[Web of Science]

Carlson, R. W., Esperança, S. & Svisero, D. P. (1996). Chemical and Os isotopic study of Cretaceous potassic rocks from southern Brazil. Contributions to Mineralogy and Petrology 125, 393–405.[Web of Science]

Chaffey, D. J., Cliff, R. A. & Wilson, B. M. (1989). Characterization of the St. Helena magma source. In: Saunders, A. D. & Norry, M. J. (eds) Magmatism in the Ocean Basins. Geological Society, London, Special Publications 42, 257–276.

Chauvel, C., Hofmann, A. W. & Vidal, P. (1992). HIMU–EM: the French Polynesian connection. Earth and Planetary Science Letters 110, 99–109.[Web of Science]

Clague, D. A. & Frey, F. A. (1982). Petrology and trace element geochemistry of the Honolulu Volcanics, Oahu: implications for the oceanic mantle below Hawaii. Journal of Petrology 23, 447–504.[Abstract/Free Full Text]

Class, C. & Goldstein, S. L. (1997). Plume–lithosphere interactions in the ocean basins: constraints from the source mineralogy. Earth and Planetary Science Letters 150, 245–260.[Web of Science]

Class, C., Altherr, R., Volker, F., Eberz, G. & McCulloch, M. T. (1994). Geochemistry of Pliocene to Quaternary alkali basalts from the Huri Hills, northern Kenya. Chemical Geology 113, 1–22.[Web of Science]

Creaser, R. A., Papanastassiou, D. A. & Wasserburg, G. J. (1991). Negative thermal ion mass spectrometry of osmium, rhenium and iridium. Geochimica et Cosmochimica Acta 55, 397–401.[Web of Science]

Davies, G. R., Spriggs, A. J. & Nixon, P. H. (2001). A non-cognate origin for the Gibeon kimberlite megacryst suite, Namibia: implications for the origin of Namibian kimberlites. Journal of Petrology 42, 159–172.[Abstract/Free Full Text]

DeWet, J. J. (1975). Carbonatites and related rocks at Saltpetre Kop, Sutherland, Cape Province. Annals, University of Stellenbosch (Series A1) 1, 193–232.

Dingle, R. V. & Gentle, R. I. (1972). Early Tertiary volcanic rocks on the Agulhas Bank, South African continental shelf. Geological Magazine 109, 137–146.[Abstract]

Dosso, L., Hanan, B. B., Bougault, H., Schilling, J.-G. & Joron, J.-L. (1991). Sr–Nd–Pb geochemical morphology between 10° and 17° N on the Mid-Atlantic Ridge: a new MORB isotope signature. Earth and Planetary Science Letters 106, 29–43.[Web of Science]

Doucouré, M., De Wit, M. J. & Mushayandebvu, M. F. (1996). Effective elastic thickness of the continental lithosphere in South Africa. Journal of Geophysical Research 101, 11291–11304.

Duncan, R. A., Hargraves, R. B. & Brey, G. P. (1978). Age, paleomagnetism and chemistry of melilite basalts in the Southern Cape, South Africa. Geological Magazine 115, 317–327.[Abstract]

Eby, G. N. (1985). Sr and Pb isotopes, U and Th chemistry of the alkaline Monteregian and White Mountain igneous provinces, eastern North America. Geochimica et Cosmochimica Acta 49, 1143–1153.[Web of Science]

Eggler, D. H. (1989). Kimberlites: how do they form? In: Ross, J., Jacques, A. L., Ferguson, J., Green, D. H., O’Reilly, S. Y., Danchin, R. V. & Janse, A. J. A. (eds) Kimberlites and Related Rocks. Geological Society of Australia Special Publication 14, 489–504.

Eiler, J. M., Schiano, J. M., Kitchen, N. & Stolper, E. M. (2000). Oxygen-isotope evidence for recycled crust in the sources of mid-ocean-ridge basalts. Nature 403, 530–534.

Esperança, S. & Crisci, G. M. (1995). The island of Pantelleria: a case for the development of DMM–HIMU isotopic compositions in a long-lived extensional setting. Earth and Planetary Science Letters 136, 167–182.[Web of Science]

Ewart, A., Chappell, B. W. & Menzies, M. A. (1988). An overview of the geochemical and isotopic characteristics of the eastern Australian Cainozoic volcanic provinces. Journal of Petrology (lithosphere special issue), 225–273.

Fitton, J. G. & Dunlop, H. M. (1985). The Cameroon Line, West Africa, and its bearing on the origin of oceanic and continental alkali basalt. Earth and Planetary Science Letters 72, 23–38.[Web of Science]

Flower, M., Tamaki, K. & Hoang, N. (1998). Mantle extrusion: a model for dispersed volcanism and DUPAL-like asthenosphere for East Asia and the Western Pacific. In : Flower, M. F. J., Chung, S. L., Lo, C. H. & Lee, T. Y. (eds) Mantle Dynamics and Plate Interactions in East Asia. Geodynamics Vol. 27. Washington, DC: American Geophysical Union, pp. 67–88.

Foley, S. F., Barth, M. G. & Jenner, G. A. (2000). Rutile/melt partition coefficients for trace elements and an assessment of the influence of rutile on the trace element characteristics of subduction zone magmas. Geochimica et Cosmochimica Acta 64, 933–938.[Web of Science]

Franz, G., Steiner, G., Volker, F., Pudlo, D. & Hammerschmidt, K. (1999). Plume-related alkaline magmatism in central Africa—the Meidob Hills (W Sudan). Chemical Geology 157, 27–47.[Web of Science]

Fraser, K. J., Hawkesworth, C. J., Erlank, A. J., Mitchell, R. H. & Scott-Smith, B. H. (1985). Sr, Nd and Pb isotope and minor element geochemistry of lamproites and kimberlites. Earth and Planetary Science Letters 76, 57–70.[Web of Science]

Furman, T. & Graham, D. (1999). Erosion of lithospheric mantle beneath the East African Rift system: geochemical evidence from the Kivu volcanic province. Lithos 48, 237–262.[Web of Science]

Gohl, K. & Uenzelmann-Neben, G. (2001). The crustal role of the Agulhas Plateau, southwest Indian Ocean: evidence from seismic profiling. Geophysical Journal International 144, 632–646.[Web of Science]

Graham, D. W., Zindler, A., Kurz, M. D., Jenkins, W. J., Batiza, R. & Staudigel, H. (1988). He, Pb, Sr and Nd isotope constraints on magma genesis and mantle heterogeneity beneath young Pacific seamounts. Contributions to Mineralogy and Petrology 99, 446–463.[Web of Science]

Gurnis, M., Mitrovica, J. X., Ritsema, J. & van Heijst, H.-J. (1999). Constraining mantle density structure using geological evidence of uplift rates: the case of the African superplume. Geochemistry, Geophysics, Geosystems 1, paper 1999GC000035.

Halliday, A. N., Dickin, A. P., Fallick, A. E. & Fitton, J. G. (1988). Mantle dynamics: a Nd, Sr, Pb and O isotopic study of the Cameroon Line volcanic chain. Journal of Petrology 29, 181–211.[Abstract/Free Full Text]

Hanan, B. B. & Graham, D. W. (1996). Lead and helium isotope evidence from oceanic basalts for a common deep source of mantle plumes. Science 272, 991–995.[Web of Science][Medline]

Hart, S. R. (1984). A large-scale anomaly in the Southern Hemisphere mantle. Nature 309, 753–757.

Hart, S. R., Hauri, E. H., Oschmann, L. A. & Whitehead, J. A. (1992). Mantle plumes and entrainment: isotopic evidence. Science 256, 517–520.[Abstract/Free Full Text]

Hart, S. R., Blusztajn, J., LeMasurier, W. E. & Rex, D. C. (1997). Hobbs Coast Cenozoic volcanism: implications for the West Antarctic Rift system. Chemical Geology 139, 223–248.[Web of Science]

Hart, S. R., Blusztajn, J., Dick, H. J. B., Meyer, P. S. & Muehlenbachs, K. (1999). The fingerprint of seawater circulation in a 500 meter section of ocean crust gabbros. Geochimica et Cosmochimica Acta 63, 4059–4080.[Web of Science]

Hartnady, C. J. H. & le Roex, A. P. (1985). Southern Ocean hotspot tracks and the Cenozoic absolute motion of the African, Antarctic, and South American plates. Earth and Planetary Science Letters 75, 245–257.[Web of Science]

Hauri, E. H. & Hart, S. R. (1993). Re–Os isotope systematics of HIMU and EMII oceanic island basalts from the South Pacific Ocean. Earth and Planetary Science Letters 114, 353–371.[Web of Science]

Hauri, E. H., Whitehead, J. A. & Hart, S. R. (1994). Fluid dynamic and geochemical aspects of entrainment in mantle plumes. Journal of Geophysical Research 99, 24275–24300.

Hawkesworth, C. J., Erlank, A. J., Kempton, P. D. & Waters, F. G. (1990). Mantle metasomatism: isotope and trace element trends in xenoliths from Kimberley, South Africa. Chemical Geology 85, 19–34.[Web of Science]

Hegner, E., Walter, H. J. & Satir, M. (1995). Pb–Sr–Nd isotopic compositions and trace element geochemistry of megacrysts and melilitites from the Tertiary Urach volcanic field: source compositions of small-volume melts under SW Germany. Contributions to Mineralogy and Petrology 122, 322–335.[Web of Science]

Hirschmann, M. M. & Stolper, E. M. (1996). A possible role for garnet pyroxenite in the origin of the ‘garnet signature’ in MORB. Contributions to Mineralogy and Petrology 124, 185–208.[Web of Science]

Hoernle, K. & Schmincke, H.-U. (1993). The petrology of the tholeiites through melilite nephelinites on Gran Canaria, Canary Islands: crystal fractionation, accumulation and depths of melting. Journal of Petrology 34, 573–597.[Abstract/Free Full Text]

Hoernle, K. A., Tilton, G. R. & Schminke, H. U. (1991). Sr–Nd–Pb isotope evolution of Gran Canaria: evidence for shallow enriched mantle beneath the Canary Islands. Earth and Planetary Science Letters 106, 44–63.[Web of Science]

Hoernle, K., Zhang, Y.-S. & Graham, D. (1995). Seismic and geochemical evidence for large-scale mantle upwelling beneath the eastern Atlantic and western and central Europe. Nature 374, 34–39.

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.[Web of Science]

Hofmann, A. W. & White, W. M. (1982). Mantle plumes from ancient oceanic crust. Earth and Planetary Science Letters 57, 421–436.[Web of Science]

Hops, J. J. (1992). A petrologic and geochemical study of peridotites and megacrysts from the Jagersfontein kimberlite, South Africa. Ph.D. thesis, University of Cape Town.

Jacob, D. & Jagoutz, E. (1995). A diamond–graphite bearing eclogitic xenolith from Roberts Victor (South Africa): indications for petrogenesis from Pb, Nd and Sr isotopes. In: Meyer, H. O. A. & Leonardos, O. H. (eds) Kimberlites, Related Rocks and Mantle Xenoliths. CPRM Special Publication 1/95. Brasilia: Companhia de Pesquisa de Recursos Minerais, pp. 304–317.

Janney, P. E. & Castillo, P. R. (1996). Basalts from the Central Pacific Basin: evidence for the origin of Cretaceous igneous complexes in the Jurassic western Pacific. Journal of Geophysical Research 101, 2875–2893.

Janney, P. E., le Roex, A. P. & Sclater, J. G. (1998). Major element and REE constraints on melting beneath the central Southwest Indian Ridge and the role of the Marion hotspot. EOS Transactions, American Geophysical Union (Fall Supplement) 79, F878.

Johnson, C. M. & Beard, B. L. (1993). Evidence from hafnium isotopes for ancient sub-oceanic mantle beneath the Rio-Grande Rift. Nature 362, 441–444.

Jones, R. A. (1987). Strontium and neodymium isotopic and rare earth element evidence for the genesis of megacrysts in kimberlites of southern Africa. In: Nixon, P. H. (ed.) Mantle Xenoliths. Chichester: Wiley, pp. 711–724.

Kogiso, T. & Hirschmann, M. M. (2001). Experimental study of clinopyroxenite partial melting and the origin of ultra-calcic melt inclusions. Contributions to Mineralogy and Petrology 142, 347–360.[Web of Science]

Kramers, J. D., Smith, C. B., Lock, N. P., Harmon, R. S. & Boyd, F. R. (1981). Can kimberlites be generated from an ordinary mantle? Nature 291, 53–56.

Lassiter, J. C. & Luhr, J. F. (2001). Osmium abundance and isotope variations in mafic Mexican volcanic rocks: evidence for crustal contamination and constraints on the geochemical behavior of osmium during partial melting and fractional crystallization. Geochemistry, Geophysics, Geosystems 2, paper GC000116.

le Roex, A. P. (1985). Geochemistry, mineralogy and magmatic evolution of the basaltic and trachytic lavas from Gough Island, South Atlantic. Journal of Petrology 26, 149–186.[Abstract/Free Full Text]

le Roex, A. P. & Lanyon, R. (1998). Isotope and trace element geochemistry of Cretaceous Damaraland lamprophyres and carbonatites, northwestern Namibia: evidence for plume–lithosphere interactions. Journal of Petrology 39, 1117–1146.[Abstract/Free Full Text]

le Roex, A. P. & Watkins, R. T. (1990). Analysis of rare-earth elements in geological samples by gradient ion tomography, an alternative to ICP and INAA. Chemical Geology 88, 151–162.[Web of Science]

le Roex, A. P., Dick, H. J. B., Erlank, A. J., Reid, A. M., Frey, F. A. & Hart, S. R. (1983). Geochemistry, mineralogy and petrogenesis of lavas erupted along the Southwest Indian Ridge between the Bouvet triple junction and 11 degrees east. Journal of Petrology 24, 267–318.[Abstract/Free Full Text]

le Roex, A. P., Cliff, R. A., Adair, B. J. I. (1990). Tristan da Cunha, South Atlantic: geochemistry and petrogenesis of a basanite-phonolite lava series. Journal of Petrology 31, 779–812.[Free Full Text]

Le Roux, P. J., le Roex, A. P. & Schilling, J.-G. (2002). MORB melting processes beneath the southern Mid-Atlantic Ridge (40–55°S): a role for pyroxenite and residual garnet. Journal of Petrology (in press).

Lee, D.-C., Halliday, A. N., Fitton, J. G. & Poli, G. (1994). Isotopic variations with distance and time in the volcanic islands of the Cameroon Line: evidence for a mantle plume origin. Earth and Planetary Science Letters 123, 119–138.[Web of Science]

Lee, W. J. & Wyllie, P. J. (2000). The system CaO–MgO–SiO₂–CO₂ at 1 GPa, metasomatic wehrlites, and primary carbonatite magmas. Contributions to Mineralogy and Petrology 138, 214–228.[Web of Science]

Lithgow-Bertelloni, C. & Silver, P. (1998). Dynamic topography, plate driving forces and the African superswell. Nature 395, 269–272.

Lundstrom, C. C., Gill, J. & Williams, Q. (2000). A geochemically consistent hypothesis for MORB generation. Chemical Geology 162, 105–126.[Web of Science]

MacGregor, I. D. & Manton, W. I. (1986). Roberts Victor eclogites: ancient oceanic crust. Journal of Geophysical Research 91, 14063–14079.

Masters, T. G., Johnson, S., Laske, G. & Bolton, H. (1996). A shear-velocity model of the mantle. Philosophical Transactions of the Royal Society of London, Series A 354, 1385–1411.

McCulloch, M. T., Jacques, A. L., Nelson, D. R. & Lewis, J. D. (1983). Nd and Sr isotopes in kimberlites and lamproites from western Australia: an enriched mantle origin. Nature 302, 400–403.

McIver, J. R. & Ferguson, J. (1979). Kimberlitic, melilititic, trachytic and carbonatite eruptives at Saltpetre Kop, Sutherland, South Africa. In: Meyer, H. O. A. & Boyd, F. R. (eds) Kimberlites, Diatremes and Diamonds: their Geology, Petrology, and Geochemistry. Washington, DC: American Geophysical Union, pp. 111–128.

McKenzie, D. & O’Nions, R. K. (1991). Partial melt distributions from inversion of rare-earth element concentrations. Journal of Petrology 32, 1021–1091.[Abstract/Free Full Text]

Moore, A. E. (1979).The geochemistry of the olivine melilitites and related rocks of Namaqualand–Bushmanland, South Africa, Ph.D. thesis, University of Cape Town.

Moore, A. E. & Erlank, A. J. (1979). Unusual olivine zoning—evidence for complex physio-chemical changes during the evolution of olivine melilitite and kimberlite magmas. Contributions to Mineralogy and Petrology 70, 391–405.[Web of Science]

Morgan, W.J. (1981). Hotspot tracks and the opening of the Atlantic and Indian Oceans In: C. Emiliani (ed.) The Sea, Vol. 7. New York: Wiley-Interscience, pp. 443–487.

Nelson, D. R., McCulloch, M. T. & Sun, S. S. (1986). The origins of ultrapotassic rocks as inferred from Sr, Nd and Pb isotopes. Geochimica et Cosmochimica Acta 50, 231–245.[Web of Science]

Niu, Y. & Batiza, R. (1997). Trace element evidence from seamounts for recycled oceanic crust in the eastern Pacific mantle. Earth and Planetary Science Letters 148, 471–483.[Web of Science]

Nowell, G. M., Kempton, P. D., Noble, S. R., Fitton, J. G., Saunders, A. D., Mahoney, J. J. & Taylor, R. N. (1998a). High precision Hf isotope measurements of MORB and OIB by thermal ionisation mass spectrometry: insights into the depleted mantle. Chemical Geology 149, 211–233.[Web of Science]

Nowell, G. M., Pearson, D. G., Kempton, P. D., Noble, S. R. & Smith, C. B. (1998b). Origins of kimberlites: a Hf isotope perspective. In: Gurney, J. J., Gurney, J. L. & Richardson, S. H. (eds) Proceedings of the Seventh International Kimberlite Conference, The Peter Nixon Volume. Cape Town: Red Roof Design, pp. 616–624.

Nowell, G. M., Pearson, D. G. & Kempton, P. D. (1999). Hafnium isotopic systematics of kimberlites, lamproites and megacrysts: implications for mantle reservoirs and the composition of bulk silicate earth. Abstracts of the Ninth Annual V. M. Goldschmidt Conference. Houston, Tx: Lunar and Planetary Institute, pp. 211–212.

O’Connor, J. M. & le Roex, A. P. (1992). South Atlantic hot spot–plume systems: 1. Distribution of volcanism in time and space. Earth and Planetary Science Letters 113, 343–364.[Web of Science]

O’Nions, R. K., Hamilton, P. J. & Evensen, N. M. (1977). Variation in ¹⁴³Nd/¹⁴⁴Nd and ⁸⁷Sr/⁸⁶Sr ratios in oceanic basalts. Earth and Planetary Science Letters 24, 13–22.

Paslick, C., Halliday, A. N., James, D. & Dawson, J. B. (1995). Enrichment of the continental lithosphere by OIB melts: isotopic evidence from the volcanic province of Northern Tanzania. Earth and Planetary Science Letters 130, 109–126.[Web of Science]

Patchett, P. J. (1983). Hafnium isotope results from mid-ocean ridges and Kerguelen. Lithos 16, 47–51.[Web of Science]

Patchett, P. J. & Tatsumoto, M. (1980a). A routine high-precision method for Lu–Hf isotope geochemistry and chronology. Contributions to Mineralogy and Petrology 75, 263–267.[Web of Science]

Patchett, P. J. & Tatsumoto, M. (1980b). Hafnium isotope variations in oceanic basalts. Geophysical Research Letters 7, 1077–1080.[Web of Science]

Pearson, D. G., Carlson, R. W., Shirey, S. B., Boyd, F. R. & Nixon, P. H. (1995a). Stabilisation of Archean lithospheric mantle: a Re–Os isotope study of peridotite xenoliths from the Kaapvaal craton. Earth and Planetary Science Letters 134, 341–357.[Web of Science]

Pearson, D. G., Rogers, N. W., Irving, A. J., Smith, C. B. & Hawkesworth, C. J. (1995b). Source regions of kimberlites and lamproites: constraints from Re–Os isotopes. In: Extended Abstracts of the Sixth International Kimberlite Conference, Novosibirsk, pp. 430–431.

Perry, F. V., Baldridge, W. S. & DePaolo, D. J. (1987). Role of asthenosphere and lithosphere in the genesis of Late Cenozoic basaltic rocks from the Rio Grande Rift and adjacent regions of the southwestern United States. Journal of Geophysical Research 92, 9193–9213.

Rehkämper, M. & Hofmann, A. W. (1997). Recycled ocean crust and sediment in Indian Ocean MORB. Earth and Planetary Science Letters 147, 93–106.[Web of Science]

Reid, D. L., Smith, C. B., Watkeys, M. K., Welke, H. J. & Betton, P. J. (1997). Whole-rock radiometric age patterns in the Aggeneys–Gamsberg ore district, central Bushmanland, South Africa. South African Journal of Geology 100, 11–22.[Abstract]

Reisberg, L., Zindler, A., Marcantonio, F., White, W., Wyman, D. & Weaver, B. (1993). Os isotope systematics in ocean island basalts. Earth and Planetary Science Letters 120, 149–167.[Web of Science]

Ritsema, J., Ni, S., Helmberger, D. V. & Crotwell, H. P. (1998). Evidence for strong shear velocity reductions and velocity gradients in the lower mantle beneath Africa. Geophysical Research Letters 25, 4245–4248.[Web of Science]

Rocholl, A., Stein, M., Molzahn, M., Hart, S. R. & Wörner, G. (1995). Geochemical evolution of rift magmas by progressive tapping of a stratified mantle source beneath the Ross Sea Rift, northern Victoria Land, Antarctica. Earth and Planetary Science Letters 131, 207–224.[Web of Science]

Rogers, A. W. & DuToit, A. L. (1904). The Sutherland volcanic pipes and their relationship to other vents in South Africa. Transactions of the South African Philosophical Society 39, 281–316.

Rogers, N. W., Hawkesworth, C. J. & Palacz, Z. A. (1992). Phlogopite in the generation of olivine-melilitites from Namaqualand, South Africa and implications for element fractionation processes in the upper mantle. Lithos 28, 347–365.[Web of Science]

Saal, A. E., Rudnick, R. L., Ravizza, G. E. & Hart, S. R. (1998). Re–Os isotope evidence for the composition, formation and age of the lower continental crust. Nature 393, 58–61.

Salters, V. J. M. (1996). The generation of mid-ocean ridge basalts from the Hf and Nd isotope perspective. Earth and Planetary Science Letters 141, 109–123.[Web of Science]

Salters, V. J. M. & Hart, S. R. (1991). The mantle sources of ocean ridges, islands and arcs: the Hf-isotope connection. Earth and Planetary Science Letters 104, 364–380.[Web of Science]

Salters, V. J. M. & Longhi, J. (1999). Trace element partitioning during the initial stages of melting beneath mid-ocean ridges. Earth and Planetary Science Letters 166, 15–30.[Web of Science]

Salters, V. J. M. & White, W. M. (1998). Hf isotope constraints on mantle evolution. Chemical Geology 145, 447–460.[Web of Science]

Schiano, P., Eiler, J. M., Hutcheon, I. D. & Stolper, E. M. (2000). Primitive, CaO-rich silica-undersaturated melts in island arcs: evidence for the involvement of clinopyroxene-rich lithologies in the petrogenesis of arc magmas. Geochemistry, Geophysics, Geosystems 1, paper 1999GC000032.

Shirey, S. B. & Walker, R. J. (1995). Carius tube digestions for low-blank rhenium–osmium analysis. Analytical Chemistry 67, 2136–2141.

Simonetti, A. & Bell, K. (1994). Nd, Pb and Sr isotopic data from the Napak carbonatite–nephelinite centre, eastern Uganda: an example of open-system crystal fractionation. Contributions to Mineralogy and Petrology 115, 356–366.[Web of Science]

Simonetti, A. & Bell, K. (1995). Nd, Pb and Sr isotopic data from the Mount Elgon volcano, eastern Uganda–western Kenya: implications for the origin and evolution of nephelinite magmas. Lithos 36, 141–153.[Web of Science]

Sleep, N. H. (1990). Hotspots and mantle plumes: some phenomenology. Journal of Geophysical Research 95, 6715–6736.

Smith, C. B. (1983). Pb, Sr and Nd isotopic evidence for sources of southern African Cretaceous kimberlites. Nature 304, 51–54.

Smith, C. B., Gurney, J. J., Skinner, E. M. W., Clement, C. R. & Ebrahim, N. (1985). Geochemical character of southern African kimberlites: a new approach based on isotopic constraints. Transactions of the Geological Society of South Africa 88, 267–280.

Smith, C. B., Clark, T. A., Barton, E. S. & Bristow, J. W. (1994). Emplacement ages of kimberlite occurrences in the Prieska region, southwest Kaapvaal Craton, South Africa. Chemical Geology 113, 149–169.[Web of Science]

Söhnge, A. P. G. & Hälbich, I. W. (eds) (1983). Geodynamics of the Cape Fold Belt. Geological Society of South Africa Special Publication 12, 184 pp.

Spath, A., le Roex, A. P. & Duncan, R. A., (1996). The geochemistry of lavas from the Comores Archipelago, western Indian Ocean: petrogenesis and mantle source region characteristics. Journal of Petrology 37, 961–991.[Abstract/Free Full Text]

Spriggs, A. J. (1988). An isotopic and geochemical study of kimberlites and associated alkaline rocks from southern Namibia. Ph.D. thesis, University of Leeds.

Stacey, J. S. & Kramers, J. D. (1975). Approximation of terrestrial lead isotope evolution by a two-stage model. Earth and Planetary Science Letters 26, 207–201.[Web of Science]

Stille, P., Unruh, D. M. & Tatsumoto, M. (1986). Pb, Sr, Nd and Hf isotopic constraints on the origin of Hawaiian basalts and evidence for a unique mantle source. Geochimica et Cosmochimica Acta 50, 2303–2319.[Web of Science]

Sun, S.-S. (1980). Lead isotopic study of young volcanic rocks from mid-ocean ridges, ocean islands and island arcs. Philosophical Transactions of the Royal Society of London, Series A 297, 409–445.

Sun, S.-S. & McDonough, W. F. (1989). Chemical and isotopic systematics of oceanic basalts: implications for mantle composition and processes. In: Saunders, A. D. & Norry, M. J. (eds) Magmatism in the Ocean Basins. Geological Society, London, Special Publications 42, 313–345.

Taljaard, M. S. (1936). South African melilite basalts and their relations. Transactions of the Geological Society of South Africa 39, 281–316.

Thirlwall, M. F. (1997). Pb isotopic and elemental evidence for OIB derivation from young HIMU mantle. Chemical Geology, 139, 51–74.[Web of Science]

Todt, W., Cliff, R. A., Hanser, A. & Hofmann, A. W. (1996). Evaluation of a ²⁰²Pb–²⁰⁵Pb double spike for high-precision lead isotope analysis. In: Basu, A. & Hart, S. R. (eds) Earth Processes: Reading the Isotopic Code. Geophysical Monograph, American Geophysical Union 95, 429–437.

Vervoort, J. D., Patchett, P. J., Blichert-Toft, J. & Albarède, F. (1999). Relationships between Lu–Hf and Sm–Nd isotopic systems in the global sedimentary system. Earth and Planetary Science Letters 168, 79–99.[Web of Science]

Verwoerd, W. J. (1990). The Saltpetrekop ring structure, Cape Province, South Africa. Tectonophysics 171, 275–285.[Web of Science]

Vidal, P., Deniel, C., Vellutini, P. J., Piguet, P., Coulon, C., Vincent, J. & Audin, J. (1991). Changes of mantle sources in the course of a rift evolution: the Afar case. Geophysical Research Letters 18, 1913–1916.[Web of Science]

Viljoen, K. S. (1988). Petrology of the Sutherland Commonage Melilitite Intrusives. M.Sc. thesis, University of Cape Town.

Viljoen, K. S., Smith, C. B. & Sharp, Z. D. (1996). Stable and radiogenic isotope study of eclogite xenoliths from the Orapa kimberlite, Botswana. Chemical Geology 131, 235–255.[Web of Science]

Walker, R. J., Carlson, R. W., Shirey, S. B. & Boyd, F. R. (1989). Os, Sr, Nd and Pb isotope systematics of southern African peridotite xenoliths: implications for the chemical evolution of subcontinental mantle. Geochimica et Cosmochimica Acta 53, 1583–1595.[Web of Science]

Widom, E., Hoernle, K. A., Shirey, S. B. & Schminke, H.-U. (1999). Os isotope systematics in the Canary Islands and Madeira: lithospheric contamination and mantle plume signatures. Journal of Petrology 40, 279–296.[Web of Science]

Wilson, M., Rosenbaum, J. M. & Dunworth, E. A. (1995). Melilitites: partial melts of the thermal boundary layer? Contributions to Mineralogy and Petrology 119, 181–196.[Web of Science]

Woodhead, J. D. (1996). Extreme HIMU in an oceanic setting: the geochemistry of Mangaia Island (Polynesia), and temporal evolution of the Cook–Austral hotspot. Journal of Volcanology and Geothermal Research 72, 1–19.[Web of Science]

Yarmolyuk, V. V., Ivanov, V. G. & Kovalenko, V. I. (1998). Sources of intraplate magmatism of western Transbaikalia in the Late Mesozoic–Cenozoic: trace element and isotope data. Petrology 6, 101–123.[Web of Science]

Yaxley, G. M., Crawford, A. J. & Green, D. H. (1991). Evidence for carbonatite metasomatism in spinel peridotite xenoliths from western Victoria, Australia. Earth and Planetary Science Letters 107, 305–317.[Web of Science]

Zack, T. & Brumm, R. (1998). Ilmenite/liquid partition coefficients of 26 trace elements determined through ilmenite/clinopyroxene partitioning in garnet pyroxenites. In: Extended Abstracts of the Seventh International Kimberlite Conference. Cape Town: University of Cape Town, pp. 986–988.

Zindler, A. & Hart, S. R. (1986). Chemical geodynamics. Annual Review of Earth and Planetary Sciences 14, 493–571.[Web of Science]

Zindler, A., Staudigel, H. & Batiza, R. (1984). Isotope and trace element geochemistry of young Pacific seamounts: implications for the scale of upper mantle heterogeneity. Earth and Planetary Science Letters 70, 175–195.[Web of Science]

CiteULike    Connotea    Del.icio.us    What's this?

This article has been cited by other articles:

				L. Beccaluva, G. Bianchini, R. M. Ellam, M. Marzola, K. M. Oun, F. Siena, and F. M. Stuart The role of HIMU metasomatic components in the North African lithospheric mantle: petrological evidence from the Gharyan lherzolite xenoliths, NW Libya Geological Society, London, Special Publications, January 1, 2008; 293(1): 253 - 277. [Abstract] [Full Text] [PDF]

				C. Paton, J. M. Hergt, D. Phillips, J. D. Woodhead, and S. R. Shee New insights into the genesis of Indian kimberlites from the Dharwar Craton via in situ Sr isotope analysis of groundmass perovskite Geology, November 1, 2007; 35(11): 1011 - 1014. [Abstract] [Full Text] [PDF]

				R. B. Trumbull, D. L. Reid, C. de Beer, D. van Acken, and R. L. Romer Magmatism and continental breakup at the west margin of southern Africa: A geochemical comparison of dolerite dikes from northwestern Namibia and the Western Cape South African Journal of Geology, September 1, 2007; 110(2-3): 477 - 502. [Abstract] [Full Text] [PDF]

				K. S. PANTER, J. BLUSZTAJN, S. R. HART, P. R. KYLE, R. ESSER, and W. C. MCINTOSH The Origin of HIMU in the SW Pacific: Evidence from Intraplate Volcanism in Southern New Zealand and Subantarctic Islands J. Petrology, September 1, 2006; 47(9): 1673 - 1704. [Abstract] [Full Text] [PDF]

				G. R. DAVIES, A. J. STOLZ, I. L. MAHOTKIN, G. M. NOWELL, and D. G. PEARSON Trace Element and Sr-Pb-Nd-Hf Isotope Evidence for Ancient, Fluid-Dominated Enrichment of the Source of Aldan Shield Lamproites J. Petrology, June 1, 2006; 47(6): 1119 - 1146. [Abstract] [Full Text] [PDF]

				T. FURMAN, K. M. KALETA, J. G. BRYCE, and B. B. HANAN Tertiary Mafic Lavas of Turkana, Kenya: Constraints on East African Plume Structure and the Occurrence of High-{micro} Volcanism in Africa J. Petrology, June 1, 2006; 47(6): 1221 - 1244. [Abstract] [Full Text] [PDF]

				R. DASGUPTA, M. M. HIRSCHMANN, and K. STALKER Immiscible Transition from Carbonate-rich to Silicate-rich Melts in the 3 GPa Melting Interval of Eclogite + CO2 and Genesis of Silica-undersaturated Ocean Island Lavas J. Petrology, April 1, 2006; 47(4): 647 - 671. [Abstract] [Full Text] [PDF]

				G. Yirgu, C.J. Ebinger, and P.K.H. Maguire The Afar volcanic province within the East African Rift System: introduction Geological Society, London, Special Publications, January 1, 2006; 259(1): 1 - 6. [Abstract] [PDF]

				T. Furman, J. Bryce, T. Rooney, B. Hanan, G. Yirgu, and D. Ayalew Heads and tails: 30 million years of the Afar plume Geological Society, London, Special Publications, January 1, 2006; 259(1): 95 - 119. [Abstract] [PDF]

				J. M.D. Day, D. R. Hilton, D. G. Pearson, C. G. Macpherson, B. A. Kjarsgaard, and P. E. Janney Absence of a high time-integrated 3He/(U+Th) source in the mantle beneath continents Geology, September 1, 2005; 33(9): 733 - 736. [Abstract] [Full Text] [PDF]

				G. M. NOWELL, D. G. PEARSON, D. R. BELL, R. W. CARLSON, C. B. SMITH, P. D. KEMPTON, and S. R. NOBLE Hf Isotope Systematics of Kimberlites and their Megacrysts: New Constraints on their Source Regions J. Petrology, August 1, 2004; 45(8): 1583 - 1612. [Abstract] [Full Text] [PDF]

				D. R. Bell and R. O. Moore Deep chemical structure of the southern African mantle from kimberlite megacrysts South African Journal of Geology, June 1, 2004; 107(1-2): 59 - 80. [Abstract] [Full Text] [PDF]

				N. V. CHALAPATHI RAO, S. A. GIBSON, D. M. PYLE, and A. P. DICKIN Petrogenesis of Proterozoic Lamproites and Kimberlites from the Cuddapah Basin and Dharwar Craton, Southern India J. Petrology, May 1, 2004; 45(5): 907 - 948. [Abstract] [Full Text] [PDF]

				A. P. LE ROEX, D. R. BELL, and P. DAVIS Petrogenesis of Group I Kimberlites from Kimberley, South Africa: Evidence from Bulk-rock Geochemistry J. Petrology, December 1, 2003; 44(12): 2261 - 2286. [Abstract] [Full Text] [PDF]

This Article Abstract FREE Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Search for citing articles in: ISI Web of Science (23) Request Permissions Google Scholar Articles by JANNEY, P. E. Articles by VILJOEN, K. S. Search for Related Content GeoRef GeoRef Citation Social Bookmarking          What's this?

Online ISSN 1460-2415 - Print ISSN 0022-3530

Copyright © 2009 Oxford University Press

Oxford Journals Oxford University Press

• Site Map
• Privacy Policy
• Frequently Asked Questions

Other Oxford University Press sites: Oxford University Press Oxford Journals China Oxford Journals Japan American National Biography Booksellers' Information Service Children's Fiction and Poetry Children's Reference Corporate & Special Sales Dictionaries Dictionary of National Biography Digital Reference English Language Teaching Higher Education Textbooks Humanities International Education Unit Law Medicine Music Online Products Oxford English Dictionary Reference Rights and Permissions Science School Books Social Sciences Very Short Introductions World's Classics