Journal of Petrology Pages 1117-1146 © 1998 Oxford University Press

Isotope and Trace Element Geochemistry of Cretaceous Damaraland Lamprophyres and Carbonatites, Northwestern Namibia: Evidence for Plume-Lithosphere Interactions
Introduction
Geological Setting
   Regional geology
   The Damaraland complexes
Analytical Techniques
Results
   Okenyenya lamprophyres
   Okorusu and Ondurakorume lamprophyres and nephelinites
   Damaraland carbonatites
   Carbonate-bearing ocelli within the Okenyenya and Okorusu dykes
Discussion
   Petrogenesis of Damaraland lamprophyres and nephelinites
   Petrogenesis of Damaraland carbonatites
   Source of Damaraland lamprophyre and carbonatite magmatism
Acknowledgements
References

Footnote Table

Isotope and Trace Element Geochemistry of Cretaceous Damaraland Lamprophyres and Carbonatites, Northwestern Namibia: Evidence for Plume-Lithosphere Interactions

ANTON P. LE ROEX* AND RUTH LANYON[dagger]

DEPARTMENT OF GEOLOGICAL SCIENCES, UNIVERSITY OF CAPE TOWN, RONDEBOSCH 7700, SOUTH AFRICA

RECEIVED JULY 7, 1997; REVISED TYPESCRIPT ACCEPTED DECEMBER 12, 1997

Major and trace element and radiogenic and stable isotope data are reported for Cretaceous lamprophyres and carbonatites associated with alkaline igneous complexes in Damaraland, northwestern Namibia. Lamprophyres, emplaced as late-stage dykes, stocks and diatremes, include both alkaline and ultramafic varieties. Many of the latter contain abundant carbonate-rich ocelli. The carbonatites are primarily sövites and Fe-rich beforsites emplaced as plugs and dykes. Both lamprophyres and carbonatites are characterized by strong, but variable, enrichment in incompatible trace elements (e.g. La 45-7232 ppm; Nb 89-1961 ppm; Sr 700-43 000 ppm), and variable but coherent initial 87Sr/86Sr (0·70351-0·70468), 143Nd/144Nd (0·51244-0·51266) and Pb isotope ratios (e.g. 206Pb/204Pb = 17·77-19·99). Individual lamprophyre types show distinct, but restricted, ranges in isotopes and incompatible element ratios (Zr/Nb = 0·9-2·0; Y/Nb = 0·15-0·34; La/Ybn = 20-58), indicating derivation from a trace element enriched, heterogeneous mantle source. Lamprophyres and carbonatites found within the same complexes have similar isotope ratios, indicating derivation from a common source. Isotope and trace element variations, coupled with the temporal and spatial association with the palaeoposition of the Tristan plume, are thought to indicate that the Damaraland lamprophyre and carbonatite magmatism formed as a consequence of melting of metasomatic vein material introduced into an isotopically depleted subcontinental lithospheric mantle by alkaline melts or fluids derived from the upwelling Tristan mantle plume at the time of continental break-up.

Keywords: geochemistry; lamprophyre carbonatite ; Namibia ; Tristan plume

INTRODUCTION

The close association of alkaline silicate rocks (ijolites, nephelinites and phonolites) and carbonatites within certain intra-plate intrusive complexes is a well-recognized and geographically widespread phenomenon (Le Bas, 1987; Barker, 1989). However, the genetic relationship between them and the processes involved in their generation are still not fully understood. Considerable debate exists as to whether carbonatite magmas are derived via the direct partial melting of mantle peridotite (e.g. Gittins, 1988; Green & Wallace, 1988; Wallace & Green, 1988; Yaxley et al., 1991), or whether they represent mantle-derived magmas which have undergone modification at lower pressures by a process of silicate-carbonate liquid immiscibility from, or extensive fractionation of, a carbonated silicate parental magma (e.g. Middlemost, 1974; Koster van Groos, 1975; Le Bas, 1987, , 1989; Kjarsgaard & Hamilton, 1989; Lee & Wyllie, 1994).

A linear series of anorogenic igneous complexes extend through Damaraland in northwestern Namibia (Fig. 1) and provide an example of the contiguous alkaline silicate and carbonatite lithologies. The composition of these complexes shows a general northeasterly spatial trend from granitic associations close to the Atlantic coast, through differentiated basic, to peralkaline and carbonatitic compositions located ~250-350 km inland (Martin et al., 1960). In addition to the silicate rock-carbonatite associations within three of the Damaraland complexes, carbonate-rich globular structures (ocelli) have also been described within the late-stage nephelinitic and/or lamprophyric intrusions associated with the Okenyenya, Okorusu and Ondurakorume complexes (Prins, 1981; Lanyon & le Roex, 1995). The Okenyenya igneous complex, the most northeasterly located of the differentiated basic complexes, marks the position northeast of which carbonatites occur as discrete intrusive phases. In addition, Okenyenya also contains the most extensive and chemically varied set of lamprophyre intrusions.


Figure 1. Simplified geological map of northwestern Namibia, showing the main outcrop regions of the Etendeka Group volcanic rocks, Cretaceous Damaraland igneous complexes, and basement ages (from Milner & le Roex, 1996).


This study is aimed at geochemically characterizing the Damaraland lamprophyres, nephelinites and associated carbonatites, so as to place constraints on their petrogenesis and location of their source regions, and evaluating the possible role of the Tristan mantle plume in initiating this Cretaceous alkaline magmatism. An investigation into the relationship between the carbonatites and the carbonate ocelli-bearing lamprophyres and nephelinites associated with the Okenyenya, Okorusu and Ondurakorume complexes was also undertakenusing trace elements as well as stable and radiogenicisotopes.

GEOLOGICAL SETTING

Regional geology

The Mesozoic Damaraland complexes of northwestern Namibia intrude a predominantly Late Proterozoicbasement that consists of Pan-African (930-470 Ma) Damara Sequence metamorphic rocks and syntectonic bodies of Salem granite (Martin et al., 1960; Miller, 1983). Although they define a broad northeasterly-trending linear feature which extends from Cape Cross on the Atlantic coast to Okorusu ~350 km inland (Fig. 1), no temporal significance can be attached to their linear arrangement (Milner et al., 1995). However, their emplacement between ~137 and 123 Ma (Milner et al., 1995, and references therein), coincident with the eruption of the voluminous Paraná-Etendeka flood basalts of Brazil and northwest Namibia and the opening of the South Atlantic Ocean (Siedner & Mitchell, 1976; Renne et al., 1992; Turner et al., 1994), suggests that they are a product of thermal activity associated with the break-up of western Gondwana (Milner et al., 1995; Milner & le Roex, 1996). Initial upwelling of the Tristan mantle plume head beneath the previously contiguous regions of South America and southern Africa resulted in both extensive flood volcanism (e.g. Duncan, 1984; White & McKenzie, 1989; Thompson & Gibson, 1991; Hawkesworth et al., 1992) and the emplacement of a series of high-level subvolcanic intrusions, the Damaraland igneous complexes, along an ancient crustal suture (Milner et al., 1995).

The Damaraland complexes

Okenyenya igneous complex

The Okenyenya igneous complex, located midway between Cape Cross and Okorusu, comprises a series of intrusive phases emplaced between ~129 and 123 Ma (Milner et al., 1993; Watkins et al., 1994) and is thought to represent the sub-volcanic remnants of a central volcano complex (Simpson, 1954; Watkins & le Roex, 1991, , 1994). Tholeiitic rocks dominate the southern two-thirds of the complex and comprise a suite of olivine gabbro to quartz monzodiorite rock types (~129 Ma) which are both rimmed and cross-cut by later picritic gabbro, syenite and quartz syenite dykes (~127 Ma). Tholeiitic magmatism at Okenyenya alternated with the emplacement of plugs of alkaline gabbro (~128 Ma) within the central and northwestern regions, and a series of concentric andesine essexite, oligoclase essexite (~126 Ma), and nepheline syenite (~123 Ma) intrusions in the northeast. The final highly alkaline phase of magmatic activity at Okenyenya, of particular interest to this study, consists of lamprophyric rock types which form a series of plugs, dykes and diatremes and intrude both the tholeiitic and alkaline gabbros in the southern half of the complex. Their geochemistry has been described in detail by le Roex et al., (1996).

Carbonatite complexes

The Damaraland carbonatite complexes are all located further inland and to the east of Okenyenya. The closest, and smallest, is the Osongombo complex, which outcrops ~70 km east-northeast of Okenyenya (Fig. 1). It is dominated by volcanic breccia with a central plug of Fe-rich beforsite and scattered outcrops of iron ore, and lacks associated silicate rock types (Martin et al., 1960; Verwoerd, 1966; Prins, 1981). The Kalkfeld igneous complex lies ~15 km to the northeast of Osongombo (Fig. 1) and consists of a central carbonatite plug of equigranular grey sövite with micaceous- and apatite-rich portions and ankeritic veins (Prins, 1981), and is surrounded by fenitized intrusions of nepheline syenite and syenite.

About 15 km northeast of the Kalkfeld igneous complex, the Ondurakorume complex is made up of a central carbonatite plug which penetrates volcanic breccia and which is surrounded by scattered and fenitized outcrops of syenite and nepheline syenite, and cross-cut by olivine dolerite dykes and oxidized iron ore dyke-like bodies (Verwoerd, 1966; Prins, 1981). The Ondurakorume carbonatite consists of micaceous sövite, sövite, apatite-rich beforsite and amphibole- or rare earth-beforsite (Prins, 1981).

Okorusu, the most inland of the Damaraland complexes, is located ~100 km north-northeast of the Ondurakorume complex. Okorusu is characterized by an extensive fluorspar deposit and comprises a series of confocal alkaline intrusions, including hortonolite monzonite, syenite, foyaite and urtite, which are intruded by dykes and plugs of foyaite, tinguaite, bostonite, melanephelinite, nephelinite (Van Zijl, 1962; Prins, 1981) and rare lamprophyre. Homogeneous sövitic carbonatite occurs only as minor plugs (Verwoerd, 1966) associated with pyroxene fenite (Van Zijl, 1962).

ANALYTICAL TECHNIQUES

Whole-rock major and trace element abundances were determined by X-ray fluorescence (XRF) spectrometry following the procedures routinely applied in the Department of Geological Sciences at the University of Cape Town, with errors and detection limits similar to those quoted by le Roex, (1985). Whole-rock CO2 abundances were calculated following duplicate determinations of CaCO3 using the Karbonat-bombe method of Birch, (1981). Whereas the precision for samples with >5% CaCO3 is 2% relative, precision decreases to ~4% relative for samples with <5% CaCO3.

Whole-rock REE abundances were determined using high-pressure ion chromatography (HPIC) following the procedures outlined by le Roex & Watkins, (1990); errors are typically <5% relative. Stable isotope (C and O) analyses of whole-rock and ocellar carbonates were obtained following the procedures outlined by Martinez et al., (1996). Carbonatite and whole-rock lamprophyre analyses were performed on the same powder splits as used for all other analytical procedures, whereas chips ( >= 180 µm) of lamprophyre ocellar and groundmass phases were hand separated before analysis. An internal standard calibrated against NBS-19 was used to correct raw data to the SMOW and PDB scales assuming [delta]18O and [delta]13C values for NBS-19 to be 28·64%° and 1·95%°, respectively. Typical reproducibility for analyses of the in-house standard (Namaqualand marble) are 0·05%° ([delta]13C) and 0·1%° ([delta]18O).

Sr, Nd and Pb were separated using conventional ion-exchange techniques and all radiogenic isotope analyses were performed on a VG Sector multi-collector mass spectrometer operated in either static (Pb) or dynamic (Sr and Nd) multi-collector mode. To correct for mass fractionation effects, measured 87Sr/86Sr and 143Nd/144Nd values were normalized to 86Sr/88Sr = 0·1194 and 146Nd/144Nd = 0·7219, respectively. Lead isotopes were corrected for fractionation using the values of Catanzaro et al., (1968) for international standard NBS 981: 206Pb/204Pb = 16·937, 207Pb/204Pb = 15·491, and 208Pb/204Pb = 36·721; average fractionation factors were 0·14% per a.m.u. Mean and 2[sgr]pop errors for repeated analyses of standards performed during the course of this study are as follows: NBS 987 87Sr/86Sr = 0·710222 ± 24 (n = 9); La Jolla 143Nd/144Nd = 0·511820 ± 12 (n = 8); and NBS 981 206Pb/204Pb = 16·902 ± 6, 207Pb/204Pb = 15·453 ± 9, 208Pb/204Pb = 36·575 ± 30 (n = 11), where the errors relate to the least significant digit/s. Where possible, the trace element and REE concentrations used to calculate the initial isotopic ratios were obtained by XRF and HPIC, respectively; trace element and REE concentrations in the remaining samples were obtained by isotope dilution using mixed 87Rb-84Sr and 149Sm-150Nd spikes.

RESULTS

Okenyenya lamprophyres

Both alkaline and ultramafic lamprophyres have been identified at Okenyenya (Lanyon & le Roex, 1995, and references therein). Detailed petrographic descriptions, accompanied by representative electron microprobe mineral analyses, have been presented by Lanyon & le Roex, (1995), and a petrographic summary is provided in Table 1.


Table 1. Mineralogy of lamprophyre and nephelinite intrusions associated with the Okenyenya and Okorusu igneous complexes

The alkaline lamprophyres have been further classified as camptonites in that they possess groundmass plagioclase and feldspathic segregations (Rock, 1991; Lanyon & le Roex, 1995). They occur as thin dykes cross-cutting gabbros in the southeast of the complex and as a stock-like body near the centre of the complex, comprising both camptonite and differentiated tinguaite. Four distinctive petrographic varieties of ultramafic lamprophyre dykes occur within the southern half of the complex, and as two brecciated and xenolith-rich diatremes which outcrop in the centre of the complex. The classification of the southeastern diatreme as an alnöite is based primarily on the presence of groundmass melilite and perovskite (Baumgartner, 1994). Most of the ultramafic lamprophyre dykes, dominated by phenocrystic phlogopite and abundant groundmass carbonate, are classified as damkjernites (Lanyon & le Roex, 1995). The remainder, as well as the more centrally located diatreme, have been subdivided on the basis of their petrographic textures into either `seriate-textured' or `ocelli-rich' varieties (Lanyon & le Roex, 1995). Although Lanyon & le Roex, (1995) originally considered the seriate-textured, ultramafic lamprophyres to lack distinctive enough modal mineralogy to allow classification beyond ultramafic, it is possible that, with their lack of groundmass feldspar and melilite, they might represent ouachitites, a term that will be used here.

Major elements

Distinction between the alkaline and ultramafic lamprophyres was originally made primarily on the basis of mineralogy (Lanyon & le Roex, 1995), but differences are also evident in terms of their whole-rock major and trace element chemistry. In particular, the ultramafic lamprophyres are readily distinguished from the alkaline varieties by their lower SiO2 and Al2O3 abundances, their higher TiO2 and CaO contents, and their higher CaO/Al2O3 ratios (Table 2; Fig. 2). The alkaline lamprophyres (camptonites) show a wide range of mg-number (68-39; assumed Fe2O3/FeO = 0·20), with the most primitive having mg-number close to expected values for primary magmas. The most evolved in terms of mg-number is classified as tinguaite. The considerable internal variation shown by the camptonite plug (Fig. 2) is qualitatively consistent with fractionation of observed phenocryst phases, namely, olivine, clinopyroxene, amphibole and Fe-Ti-oxide.


Table 2. Whole-rock major, trace and REE chemistry of the lamprophyre and nephelinite intrusions associated with the Okenyenya, Okorusu and Ondurakorume igneous complexes


Figure 2. Selected major element variations of the Okenyenya alkaline and ultramafic lamprophyre intrusions: [wdbullf], ouachitite dykes;N, Čouachititediatreme; [utrif], ocelli-rich ultramafic dyke; [utri], alnöite diatreme; [circle], damkjernite dykes; [squ], camptonite dykes; *, differentiated camptonite plug.


With the exception of a single evolved sample (OKJ-94-3: mg-number ~50), the ultramafic lamprophyres have relatively primitive compositions (mg-number = 60-74), and the different sub-types can be readily distinguished from one another on the basis of major elements (Fig. 2). The significantly higher LOI values (6·42-15·91 wt %) and CO2 contents (5·35-14·22 wt %) of the damkjernites, as compared with the other ultramafic lamprophyres (LOI = 2·80-6·45 wt %; CO2 <1·00 to 4·69 wt %), is reflected in the greater abundance of groundmass carbonate.

Trace elements

Compatible trace element abundances (e.g. Ni, Cr, Sc) show a regular decrease with decreasing mg-number; the more primitive ultramafic and alkaline lamprophyres have >200 ppm Ni, and >500 ppm Cr (Fig. 3; Table 2). In terms of incompatible trace element abundance variations, the individual lamprophyre types have generally distinct and coherent characteristics (Fig. 3). All of the Okenyenya lamprophyres have REE patterns characterized by strong chondrite-normalized light rare earth element (LREE) enrichment (Fig. 4). The camptonites are less enriched in the LREE (La/Ybn = 20-30) than the ultramafic lamprophyres (La/Ybn = 27-58). The camptonites also have a more concave-upwards chondrite-normalized REE pattern than the other varieties (Fig. 4).


Figure 3. Selected trace element variations of Okenyenya alkaline and ultramafic lamprophyre intrusions. Symbols as for Fig. 2.



Figure 4. Chondrite-normalized rare earth element variations of Okenyenya and Ondurakorume lamprophyre intrusions and Okorusu lamprophyre and nephelinite dykes. Chondrite normalizing values from Sun & McDonough, (1989).


The different lamprophyre types present at Okenyenya are readily distinguished from one another in terms of their incompatible element ratios, which show remarkably restricted ranges for each type. The damkjernites are distinguished by their low Zr/Nb and Y/Nb ratios and their high Ce/Yb, K/Nb, K/Th, Rb/Nb and K/U ratios compared with the other lamprophyres (Table 3; Fig. 5). The ouachitite and ocelli-rich ultramafic varieties have the lowest K/U and K/Nb ratios and the highest Zr/Nb ratios, whereas the alnöites have the highest La/Nb and Y/Nb at intermediate Zr/Nb ratio (Table 3; Fig. 5). The alkaline lamprophyres have lower Ce/Yb, La/Nb and Ti/Zr ratios than the ultramafic varieties.


Table 3. Selected incompatible element ratios for primitive lamprophyres (mg-number >0·60) from the Okenyenya igneous complex


Figure 5. Covariations of selected incompatible trace element ratios in the Okenyenya lamprophyre intrusions. Approximate 2[sgr] errors are shown. Symbols as for Fig. 2.


The Okenyenya lamprophyres have primitive mantle normalized trace element patterns that are characterized by an increase in abundance with increasing incompatibility (Fig. 6). Superimposed on this general increase are a number of variably pronounced negative anomalies (relative to adjacent elements). The alkaline lamprophyres have slightly negative Ti and Sm anomalies but no significant K anomalies. The damkjernites display negative Zr and P anomalies, strong negative Ti anomalies, and have no significant K anomalies. In contrast, the remaining varieties of ultramafic lamprophyre have strong negative K and Rb anomalies. Whereas the alnöite diatreme samples also possess negative Zr and minor negative Ti anomalies, the ocelli-rich ultramafic lamprophyre lacks a significant negative Zr anomaly and the ouachitites show no significant negative Ti or Zr anomalies.


Figure 6. Primitive mantle normalized incompatible trace element abundance patterns of primitive (mg-number >60) Okenyenya and Ondurakorume lamprophyre intrusions and the Okorusu nephelinite dyke; primitive mantle normalizing values from Sun & McDonough, (1989). Shaded fields correspond to primitive basanites from Tristan da Cunha (le Roex et al., 1990), representing the present-day composition of Tristan plume eruptives.


Radiogenic and stable isotopes

The initial radiogenic isotope compositions (calculated at 124 Ma, Table 4; Fig. 7) of the alkaline lamprophyres are virtually indistinguishable. They have high [epsilon]Nd and low 87Sr/86Sri ratios, and Pb isotope ratios intermediate to the range displayed by the ultramafic lamprophyres (Table 5; Fig. 8). The ultramafic lamprophyres define a relatively narrow range of initial [epsilon]Nd values (+1·6 to +3·1), signifying time-integrated LREE depletion of their source region relative to Bulk Earth, and a moderate range of initial 87Sr/86Sr ratios (0·70385-0·70461). With the exception of the ouachitite diatreme sample (OKJ-94-10), the ultramafic lamprophyres define a narrow range of initial 207Pb/204Pb (Table 5), but with significant 206Pb/204Pb variation (17·769-19·070), and plot well above the Northern Hemisphere Reference Line (NHRL) in 207Pb/204Pb vs 206Pb/204Pb space (Fig. 8). However, they form an approximately linear array along the NHRL on a plot of 208Pb/204Pb vs 206Pb/204Pb (Fig. 8).


Table 4. Measured and initial whole-rock Sr and Nd isotope ratios for the Okenyenya, Okorusu and Ondurakorume lamprophyre and nephelinite intrusions and selected Damaraland carbonatite samples


Table 5. Whole-rock Pb isotope data for the Okenyenya and Okorusu lamprophyre intrusions and selected Damaraland carbonatite samples


Figure 7. Initial Sr and Nd isotope ratios of Damaraland lamprophyre and carbonatite intrusions, as calculated for 124 Ma. Osongombo (+) and Kalkfeld ([utrif]) are carbonatites; Ondurakorume samples include carbonatites ([squf]) and lamprophyre ([squ]); Okorusu samples include carbonatites () and lamprophyre-nephelinite dykes ([diamonds]); Okenyenya lamprophyre symbols as for Fig. 2. Small crosses indicate carbonate ocelli compositions, with associated tie lines indicating host rock composition. BE124, Bulk Earth at 124 Ma. The 2[sgr] errors are less than symbol size.



Figure 8. Initial Pb isotope ratios of Damaraland lamprophyre and carbonatite intrusions, as calculated for 124 Ma. Field for the Tristan plume encompasses data from Tristan da Cunha, Inaccessible Island, Gough Island and DSDP holes 527 and 528 on the Walvis Ridge (Richardson et al., 1982; le Roex, 1985; le Roex et al., 1990; Cliff et al., 1991). µ, 238U/204Pb; [kappa], 232Th/204Pb. Data for contemporaneous Brazilian carbonatites from Toyoda et al., (1994). NHRL, Northern Hemisphere Reference Line (Hart, 1984). The 2[sgr] errors are less than symbol size. Symbols as for Fig. 7.


Small, yet coherent and distinct, differences amongst the initial radiogenic isotope signatures of the different lamprophyre types found at Okenyenya are evident in Figs 7 and 8. Whereas the ocelli-rich ultramafic lamprophyre dyke has a similar Sr, Nd and Pb isotopic signature to the camptonites, the ouachitite dykes have slightly lower [epsilon]Nd and higher 87Sr/86Sr values (Table 4). The ouachitite diatreme (OKJ-94-10) has the highest initial ratios of 87Sr/86Sr (0·70461) and 207Pb/204Pb (15·705) of all the Okenyenya lamprophyres (Fig. 8), and the latter in particular is considered to reflect crustal contamination. The damkjernite dykes have a limited range of initial 87Sr/86Sr and [epsilon]Nd values (Table 4) which plot within the depleted mantle quadrant at 124 Ma (Fig. 7). They also have the least radiogenic initial 206Pb/204Pb values (Table 5; Fig. 8). The alnöite diatreme has similar initial Sr and Nd isotope ratio to the damkjernites, but is more similar to the camptonites and ocelli-rich ultramafic dyke with respect to Pb isotopes (Fig. 8).

Groundmass carbonates within the Okenyenya damkjernite dykes have a narrow range of [delta]13C values (-5·7 to -5·0%°) and a virtually constant [delta]18O signature (~9·2%°), and lie within the field defined by Deines, (1989) for most primary carbonatites (Table 6; Fig. 9). This groundmass carbonate can therefore be considered to have a primary magmatic origin.


Table 6. Carbon and oxygen itotope data for the Damaraland carbonatites and groundmass and ocellar carbonate phases within the Okenyenya and Okorusu lamprophyre and nephelinite intrusions


Figure 9. Plot of [delta]13C vs [delta]18O compositions of Damaraland lamprophyre and carbonatite intrusions. Data for Brazilian carbonatites from Santos & Clayton, (1995); field for primary carbonatites from Deines, (1989). The 2[sgr] errors are less than symbol size.


Okorusu and Ondurakorume lamprophyres and nephelinites

Late-stage ocelli-bearing ultramafic lamprophyre and olivine nephelinite dykes also occur at Okorusu and Ondurakorume. The mineralogy of these dykes is summarized in Table 1 and their whole-rock geochemistry is presented in Table 2. Ondurakorume sample ON-6 was originally described as a carbonatite by Prins, (1981) but its low CO2 (4·80 wt %) content and generally similar major and trace element composition to some of the Okenyenya ultramafic lamprophyres and the Okorusu olivine nephelinite (Table 2) suggests that it is more correctly classified as an olivine nephelinite, a term adopted here. The pisolitic structures in this sample are similar to the carbonate-bearing ocelli found in ultramafic lamprophyres from Okenyenya and Okorusu. The Okorusu and Ondurakorume nephelinite dykes have similarly high mg-number to the primitive ultramafic lamprophyres at Okenyenya (Table 2), but have lower TiO2 and P2O5, and slightly lower LOI and CaO contents. In contrast, the ultramafic lamprophyre (OKU-94-1) from Okorusu has a more evolved composition than the nephelinite dyke (Table 2), and is highly enriched in incompatible trace elements.

The Okorusu and Ondurakorume dyke samples are strongly LREE enriched (La/Ybn = 28-84) and have concave-up chondrite-normalized REE patterns similar to the Okenyenya camptonites (Fig. 4). With the exception of the unusual enrichment in Ba in the Ondurakorume dyke, the incompatible trace element patterns of the Okorusu and Ondurakorume nephelinites are similar when normalized to primitive mantle abundances (Fig. 6), with both samples showing negative Ti, P, K and Rb anomalies.

The Okorusu lamprophyre and nephelinite dykes have similar initial 87Sr/86Sr but lower initial [epsilon]Nd values than the Okenyenya lamprophyres (Table 4; Fig. 7), whereas the Ondurakorume nephelinite has significantly lower initial 87Sr/86Sr but intermediate [epsilon]Nd (Table 4; Fig. 7). Pb isotope values for the Okorusu ultramafic lamprophyre dyke are similar to those obtained for the Okenyenya damkjernites (Table 5; Fig. 8). Groundmass carbonate within the Okorusu ultramafic lamprophyre dyke has slightly lower [delta]13C (-6·8%°) and [delta]18O (7·4%°) values than those within the Okenyenya ultramafic lamprophyres (Table 6; Fig. 9). However, it too lies within the primary carbonatite field of Deines, (1989), and therefore also appears to be mantle-derived with no significant crustal influence.

Damaraland carbonatites

The Damaraland carbonatites analysed in this study were originally studied in detail by Prins, (1981). Specific sampling localities, general petrographic and mineralogical descriptions, and the classifications of the various types of carbonatites have been given by Prins, (1981, and references therein).

The two samples analysed from Osongombo are classified as Fe-rich beforsites; sample OS-4 was collected from the main intrusive body, whereas sample OS-5 was collected from a cross-cutting beforsite vein. Kalkfeld carbonatite samples K-2 and K-22 were classified by Prins, (1981) as micaceous sövite and ankeritic sövite, respectively, although sample K-22 is hydrothermally altered. The four samples from Ondurakorume are thought to represent the four main stages of carbonatite intrusion, and include sövite (ON-8 and ON-10), apatite-rich beforsite (ON-9) and rare earth-beforsite (ON-4). Okorusu carbonatite samples (OKU-6 and OKU-18) were collected by Prins, (1981) from two sövite plugs.

Major and trace element analyses of the Damaraland carbonatites are reported in Table 7. All are greatly enriched in REE and incompatible elements and generally display steep chondrite-normalized patterns of LREE/HREE enrichment with (La/Er)n ratios ranging from 34·4 to 649 (Fig. 10). The primitive mantle normalized trace element patterns reflect the strong incompatible element enrichment and are generally similar (Fig. 11), with variable but marked depletions in Ti, Zr, P, K and Rb. These depletions are similar to, although considerably more enhanced than, those observed in the Damaraland lamprophyres and nephelinites.


Table 7. Whole-rock major, trace element and REE chemistry of carbonatites associated with the Damaraland intrusive complexes


Figure 10. Chondrite-normalized rare earth element variations of Damaraland carbonatites. Chondrite normalizing values from Sun & McDonough, (1989).



Figure 11. Primitive mantle normalized incompatible trace element abundance patterns of Damaraland carbonatites; primitive mantle normalizing values from Sun & McDonough, (1989).


Considered as a whole, the Damaraland carbonatitesdefine a limited range of initial [epsilon]Nd values which cluster around Bulk Earth, overlap with values for the associated lamprophyre and nephelinite dykes at Okorusu and Ondurakorume, but plot well below the Okenyenya lamprophyres over a similar range of initial 87Sr/86Sr (Table 4; Fig. 7). Whereas the Osongombo and Okorusu carbonatites have similar initial Sr isotope signatures, the initial 87Sr/86Sr ratios of the Kalkfeld and Ondurakorume carbonatites are distinctly lower (Table 4). The well below Bulk Earth initial [epsilon]Sr values (-14·8 to -11·4) of the Kalkfeld and Ondurakorume carbonatites imply a time-integrated source depletion in Rb/Sr, without associated depletion in Nd/Sm. In this respect the Kalkfeld and Ondurakorume carbonatites are similar to East African carbonatites in plotting below the mantle array (Fig. 7; Bell & Blenkinsop, 1989), and are unlike comparable aged Brazilian carbonatites, which have higher 87Sr/86Sr at similar 143Nd/144Nd ratios (Roden et al., 1985).

Although only one Pb isotope dataset is available for each of the Damaraland carbonatite complexes (Table 5), it is obvious that samples from the Kalkfeld and Ondurakorume complexes, which are the least radiogenic in terms of initial 87Sr/86Sr, are similar in having the lowest initial 206Pb/204Pb ratios of all the analysed Damaraland samples (Fig. 8). Okorusu carbonatite sample OKU-18 has initial Pb isotope ratios which are similar to the Okenyenya and Okorusu lamprophyres. Osongombo carbonatite sample OS-4 has anomalously high initial Pb isotope values, which are particularly evident in terms of 206Pb/204Pb and 208Pb/204Pb. Brazilian carbonatites of comparable age have significantly lower 206Pb/204Pb ratios ( Toyoda et al., 1994) than those found in the Damaraland carbonatites (Fig. 8).

Whole-rock stable isotope values for the Damaraland carbonatites are variable (Table 6; Fig. 9), with all but two samples plotting within the field encompassed by most primary carbonatites (Deines, 1989). Although carbonatite samples K-22 and OS-5 have [delta]13C values consistent with a mantle origin they have significantly higher [delta]18O signatures. The description of Kalkfeld sample K-22 as hydrothermally altered sövite (Prins, 1981) suggests that its high [delta]18O value (23·1%°) is probably a product of equilibration with low-temperature fluids (e.g. Deines, 1989). The high [delta]18O value (24·8%°) of Osongombo sample OS-5 may result from the sample being a carbonatite dyke. According to Deines, (1989), although [delta]13C values do not vary significantly between different intrusive carbonatite types, carbonatite dykes and veins are less likely to retain their original oxygen isotope signature because they are less massive than carbonatite plutons. Although sample OS-5 has the highest [delta]18O and initial 87Sr/86Sr values of any of the analysed carbonatites, the latter is still representative of a mantle value and within the range of initial Sr isotope ratios shown by the Damaraland lamprophyres. Thus it seems unlikely that crustal contamination has played a significant role in the genesis of the Osongombo carbonatites. It is noteworthy that the Damaraland carbonatites have similar [delta]18O values but slightly higher [delta]13C values than the non-contaminated Brazilian carbonatites of equivalent age, but significantly lower [delta]13C values than the limestone-contaminated, Brazilian Mata Peto carbonatites (Fig. 9; Santos & Clayton, 1995).

Carbonate-bearing ocelli within the Okenyenya and Okorusu dykes

Leucocratic globular structures within all of the Okenyenya lamprophyres (Lanyon & le Roex, 1995), as well as the Okorusu and Ondurakorume nephelinite and ultramafic lamprophyre dykes, consist of spherical ocelli or more irregularly shaped segregations, and their mineralogy varies according to the host rock (Table 1). Of the three mineralogical types of globular structures defined by Rock, (1991), the feldspathic variety occurs within the Okenyenya camptonites, whereas the ultramafic lamprophyre and nephelinite dykes at Okenyenya, Okorusu and Ondurakorume are dominated by carbonate-analcime ocelli.

Ocellar carbonates within the Okenyenya lamprophyres have stable isotope signatures within the prescribed range of Deines, (1989) for primary carbonatites ( Table 6; Fig. 9). They have [delta]13C values (-3·9 to -3·0%°) higher than those of the groundmass carbonates (-5·7 to -5·0%°), and more similar to the generally higher Damaraland whole-rock carbonatite values (-5·0 to -2·4%°); the slight shift is attributed to fractionation during formation of the ocelli. Ocelli from Okorusu have [delta]13C values (-6·6 to -5·7%°) which are similar to, or even lower than, the Okenyenya groundmass values. The groundmass and ocellar carbonates within Okorusu ultramafic lamprophyre sample OKU-94-1 have similarly low [delta]13C values (-6·8 to -6·6%°), although the ocellar carbonates have a significantly lower [delta]18O signature (+5·0%°) compared with groundmass carbonate (+7·4%°).

The initial radiogenic Sr and Nd isotope signatures of ocelli within the lamprophyre and nephelinite intrusions at Okenyenya and Okorusu have also been determined (Table 8). Ocelli separates from two of the Okenyenya ultramafic lamprophyre dykes have slightly higher initial Sr and Nd isotope ratios than their corresponding groundmass components (Fig. 7). However, they lie well within the field defined by the Okenyenya lamprophyres and can be considered to derive from the same source as their host rocks. The lower initial [epsilon]Nd values (-0·7 to +0·3) of ocelli within the Okorusu dykes, as compared with the Okenyenya lamprophyre ocelli (+2·0 to +3·1), are consistent with the lower host-rock initial [epsilon]Nd values and those obtained for the associated carbonatite, suggesting derivation from a similar source.


Table 8. Sr and Nd isotope data for the groundmass and ocellar components of selected Okenyenya and Okorusu lamprophyre and nephelinite intrusions

DISCUSSION

Petrogenesis of Damaraland lamprophyres and nephelinites

Lamprophyres are highly alkaline, H2O- and/or CO2-rich magmas, generated by very small degrees of partial melting of a hydrous mantle (e.g. Rock, 1991). The mantle-like low 87Sr/86Sr and high 143Nd/144Nd ratios obtained for all the Damaraland lamprophyre intrusions, coupled with the presence of hydrous mafic and ultramafic mantle xenoliths (amphibole-bearing lherzolite, wehrlite and clinopyroxenite) within the Okenyenya alnöite diatreme, are consistent with derivation from hydrous mantle. Together with their primitive nature (mg-number >60), these features make the Okenyenya lamprophyres ideal for placing constraints on the petrogenesis of lamprophyric magmas in general. Although reference will be made to the Okorusu and Ondurakorume lamprophyres and nephelinites, the following discussion will emphasize the Okenyenya lamprophyres in view of their greater abundance, wider variety, and close spatial proximity.

Of the lamprophyre types recognized at Okenyenya, the camptonites and ouachitites have representatives close in composition to primary magmas (i.e. mg-number >67). The most primitive damkjernites and alnöites have mg-number ~65 and, although they may have experienced as little as 10% olivine fractionation, their compositions closely reflect those of primary magmas.

The most noticeable features of the Okenyenya lamprophyres are the distinct and restricted range in incompatible trace element ratios of each of the different varieties (Fig. 5), their unique primitive mantle normalized trace element abundance patterns (Fig. 6), and the associated distinct radiogenic isotope signatures (Fig. 7). These geochemical features are inconsistent with derivation of the different Okenyenya lamprophyres from a homogeneous mantle source. The damkjernites must have derived from a mantle source with lower 206Pb/204Pb and 143Nd/144Nd and higher 87Sr/86Sr ratios than the other ultramafic lamprophyre dyke varieties. The high 87Sr/86Sr, 206Pb/204Pb and particularly 207Pb/204Pb ratios of the ouachitite diatreme sample OKJ-94-10 are attributed to shallow-level crustal contamination. The alnöite diatreme has similar 143Nd/144Nd and Pb isotope ratios to the camptonites, ouachitites and ocelli-rich ultramafic lamprophyre dyke, and its slightly elevated 87Sr/86Sr ratio could also reflect minor crustal contamination or late-stage alteration.

The only two lamprophyre varieties with virtually identical initial isotope ratios, and thus potentially derived from the same source, are the camptonites and the ocelli-rich ultramafic dyke. This would require the former to be derived by higher degrees of melting to account for its lower absolute abundances of Zr, Ce and P (at equivalent mg-number; Table 2) and lower (La/Yb)n ratio (Fig. 5; Table 3). The lower K/Nb (Fig. 5), K/Th and K/U and slightly lower Rb/Nb ratios (Table 3) of the ocelli-rich ultramafic dyke, as compared with the camptonites, are consistent with derivation of the former by lower degrees of partial melting in the presence of residual phlogopite or amphibole. Furthermore, the considerably higher volatile content (H2O and CO2) of the ocelli-rich dyke compared with the camptonite would also be consistent with derivation by lower degrees of melting of a hydrous/carbonated mantle. The absence of a negative K-anomaly in the camptonites (Fig. 5) requires derivation from an amphibole free source, or by a degree of melting that fully consumes any K-bearing phase (e.g. Späth et al., 1996). The slightly higher Zr/Nb and La/Nb of the ocelli-rich ultramafic lamprophyre would require minor residual ilmenite, which is consistent with the negative Ti anomaly evident in Fig. 6, but not obviously consistent with the lower Ti/Zr ratio of the camptonite.

Although there is ample evidence for the individual lamprophyre varieties found at Okenyenya to have derived from a locally heterogeneous mantle source (where scale of heterogeneity is comparable with melting volume for a particular magma), there are many similarities in their overall geochemistry, which allow some general comment to be made on their petrogenesis. Of particular note is their uniformly strong chondrite-normalized LREE enrichment (La/Ybn = 20-58) and substantial enrichment in incompatible elements, as well as H2O and CO2 (Tables 2 and 3). In terms of their REE abundances, La ranges from 200 to 600 times chondrite, whereas Yb abundances are fairly constant at ~10 times chondrite (Fig. 4). Quantitative modelling of REE abundance variations [using partition coefficients from McKenzie & O'Nions, (1991) and modal mineralogy from Späth et al., (1996)] suggest that if derived by 1-2% melting of amphibole-bearing lherzolite, the mantle source of the ultramafic lamprophyres would need to be enriched by a factor of 7-10 over chondritic values for the LREE and by a factor of 1·4-1·5 for the HREE, if melting occurred in the garnet stability field. Choice of lower F (e.g. 0·1-0·2%) would scale these values accordingly. If melting occurred as a one-stage process in the spinel stability field then a similar enrichment of the LREE is required, but HREE concentrations would be lower than chondritic by a factor of 2·5-3·3, which seems unrealistic. Our preferred model is therefore one in which the magmas are either derived by direct melting within the garnet stability field, or are derived by melting within the spinel stability field, but inherit their residual garnet REE pattern by predominantly melting vein material introduced into the spinel lherzolite as a consequence of metasomatism by low volume alkali melts or fluids derived from depths within the garnet stability field (see below for further discussion).

The alkaline lamprophyres and ultramafic damkjernites are the only lamprophyre varieties that do not show evidence for a significant negative K anomaly (Fig. 6), indicating the absence of a residual K-bearing phase such as phlogopite or amphibole in their source region. The damkjernites, however, have marked negative Zr, Ti and to a lesser extent P anomalies, indicating either relative depletion of these elements in the source, or fractionation against residual accessory phases that host these elements (e.g. zircon, ilmenite, and apatite). The higher volatile and CO2 content of this ultramafic lamprophyre variety (Table 2) suggests, furthermore, that their source was more volatile rich, with a higher CO2/H2O ratio. In contrast, the remaining ultramafic lamprophyre varieties (including those at Ondurakorume and Okorusu) have higher H2O/CO2 ratios and all show strong negative K anomalies (Fig. 6; Table 3), suggesting a residual K-bearing phase in their source regions. The broadly constant Ba/Nb and Rb/Nb ratios with varying Nb content of these magmas suggest that amphibole rather than phlogopite is the most likely residual K-bearing phase, as these ratios are strongly fractionated during low degrees of melting in the presence of residual phlogopite (Späth & le Roex, in preparation). The different varieties also variably show negative Ti, Zr and/or P anomalies, which, like the damkjernites, can be taken to indicate either relative depletion in their source regions (as a consequence of a metasomatic overprint; e.g. Hauri et al., 1993), or the presence of residual accessory phases hosting these elements.

In summary, qualitative considerations combined with semi-quantitative modelling suggest that the Damaraland lamprophyres were formed as a consequence of melting of a metasomatically enriched, hydrous or carbonated mantle with heterogeneously distributed accessory phases such as amphibole, ilmenite, zircon and apatite. The alkaline lamprophyres appear to have derived from a less volatile and incompatible element enriched mantle source by greater degrees of melting than the ultramafic lamprophyres. The different ultramafic lamprophyre varieties owe their origin to minor differences in the degree of source enrichment (including volatile composition), degree of partial melting and residual accessory phases. The damkjernites seem to have derived from a more volatile and CO2-rich source than the other ultramafic lamprophyre varieties.

Petrogenesis of Damaraland carbonatites

The most remarkable features of carbonatite magmas the world over are their low SiO2 contents, exceptional enrichment in highly incompatible elements, and their very steep chondrite-normalized REE patterns (Woolley & Kempe, 1989). The Damaraland carbonatites are no exception. They are characterized by enrichment of highly incompatible elements to ~10 000 times primitive mantle values (e.g. Fig. 11), LREE abundances 4000-40 000 times chondrite values, and La/Ybn ratios of >400 (Fig. 10). Comparison of primitive mantle normalized incompatible trace element patterns for the Damaraland carbonatites and the associated ultramafic lamprophyres and nephelinites shows that the carbonate and silicate magmas have remarkably similar overall patterns. However, the carbonatite patterns are considerably more accentuated, displaying greater absolute enrichment in the highly incompatible elements and greater negative K, (Rb), P, Zr and Ti anomalies (Figs 6 and 11). Interpretation of trace element abundance variations in carbonatites is subject to considerable uncertainty; the degree of fractional crystallization is hard to evaluate, carbonatites are notoriously rich in exotic accessory phases, many of which host the otherwise highly incompatible elements, partition coefficients are not as well constrained as for silicate magmas, and, perhaps most importantly, recognition of what constitutes a `primary' magma is difficult.

Considerable debate exists around the origin of carbonatite magmas, and the alternative petrogenetic models have been highlighted in the book edited by Bell, (1989), and more recently reviewed by Lee & Wyllie, (1994, , 1997) from an experimental perspective. The debates are concerned with: (1) whether the composition of primary carbonatite magma is calcic, dolomitic or sodic, and (2) whether carbonatites are derived from parental silicate magmas by extensive fractional crystallization or immiscibility. An underlying cause of much of this debate is the difficulty in distinguishing unequivocally between a `liquid' and a cumulate-enriched magma, and the extent of volatile and alkali loss (to fenitizing fluids) from the original magma composition. In this regard, the Damaraland carbonatites are no exception.

Phase relationships indicate that carbonatites formed either through extensive fractional crystallization or as an immiscible phase from a carbonated alkaline silicate magma should be rich in alkalis (e.g. Lee & Wyllie, 1994). If the Damaraland carbonatites formed at crustal or mantle depths as immiscible carbonate magmas from an alkaline silicate magma similar to the spatially associated ultramafic lamprophyres (which show evidence for the localized development of an immiscible carbonate phase in the form of ocelli) then, from a consideration of phase relationships (e.g. fig. 7 of Lee & Wyllie, 1994), alkalis should form at least 40% of the cation component of the carbonatite magma. The Damaraland carbonatites are notably bereft of alkalis, and either did not form as a consequence of immiscibility, or such a significant proportion of the original alkalis has been lost that a rigorous evaluation of their petrogenesis is negated.

Primary carbonatite magma derived directly by partial melting of carbonated mantle peridotite will be dolomitic, with a maximum calcite content of 50-80% (Lee & Wyllie, 1994, , 1997), and low ca-number [Ca/(Ca + Mg) = 0·51-0·53; Sweeney, 1994]. As the Damaraland carbonatites have high ca-number (0·71-0·94) and are primarily sövites or differentiated Fe-rich beforsites (Prins, 1981) they are unlikely to represent primary magmas. Furthermore, following the arguments of Lee & Wyllie, (1994, , 1997) that all sövitic carbonatites are cumulates, they are also unlikely to represent liquid compositions.

In summary, although there are many uncertainties regarding the petrogenesis of the Damaraland carbonatites, it is unlikely that they represent primary magmas, based on their low MgO contents, nor are they likely to represent liquid compositions formed through a process of immiscibility as they are too calcic and too poor in alkalis. Based on the limited available evidence, and without a more in-depth study of individual complexes to establish a possible fractionation history, the most likely explanation is that the Damaraland carbonatites represent calcite-rich cumulates. There is therefore little value in speculating on detailed compositional relationships amongst and between the Damaraland carbonatites and ultramafic lamprophyres. However, the similar Sr and Nd radiogenic isotope compositions of the carbonatites, their associated ultramafic lamprophyres and the carbonate ocelli found within the latter (e.g. Fig. 7) provide compelling evidence that all are intimately related and derive by petrogenetic processes as yet undefined from a common trace element enriched mantle source.

Source of Damaraland lamprophyre and carbonatite magmatism

On the basis of spatial, temporal and geochemical relationships, the magmatism that gave rise to the Cretaceous Damaraland alkaline complexes is believed to be related to the Etendeka volcanism and to mantle melting associated with the upwelling Tristan plume when it was located beneath this region ~130-123 my ago (Milner et al., 1995; Milner & le Roex, 1996). As field and geochronological evidence (Prins, 1981; Milner et al., 1993; Watkins & le Roex, 1994; Watkins et al., 1994) indicates that the lamprophyre intrusions represent the final stage of magmatic activity associated with the Damaraland alkaline complexes, they are therefore also likely to be related to the influence of the Tristan plume (Milner & le Roex, 1996).

The present-day Tristan plume composition is characterized by elevated 87Sr/86Sr (greater than Bulk Earth), low 143Nd/144Nd ([epsilon]Nd <0) and Pb isotope compositions which plot above the NHRL (Figs 8 and 12). Milner & le Roex, (1996) have shown that the Okenyenya alkaline gabbros and Etendeka Tafelkop basalts have similar isotope and trace element compositions to those of the present-day Tristan plume, leading them to the conclusion that the plume composition has remained broadly constant for the past 130 my. It is evident from Fig. 12 that the Okorusu and Osongombo lamprophyres and carbonatite magmas also have Sr and Nd isotope compositions similar to the calculated Tristan plume composition at 124 Ma (based on the age of the Okenyenya lamprophyres and assuming a source Rb/Sr of 0·05 and Sm/Nd of 0·20; values calculated for the source of the Tristan da Cunha basanites), and are thus likewise consistent with derivation by direct melting of Tristan plume material. In contrast, the Okenyenya lamprophyres and the Kalkfeld and Ondurakorume carbonatites and lamprophyre dyke have isotope ratios displaced towards less radiogenic Sr and more radiogenic Nd compositions than the Tristan plume (Fig. 12; Table 4), suggesting involvement of depleted mid-ocean ridge basalt (MORB) mantle (DMM) or a HIMU (high 238U/204Pb) component in their genesis. It is noteworthy, in this regard, that a small subset of Inaccessible Island lavas, similarly displaced towards low 87Sr/86Sr values ( Fig. 12), have also been interpreted as having a depleted asthenosphere or HIMU mantle component involved in their genesis (Cliff et al., 1991), although this is not as evident with respect to Pb isotopes. Also significant is the fact that, unlike the Okenyenya tholeiitic gabbros, the bulk of the Etendeka lavas (Milner & le Roex, 1996) and the Cretaceous potassic and carbonatitic rocks of southern Brazil (Toyoda et al., 1994; Carlson et al., 1996), the Sr-Nd isotope data obtained during this study provide no compelling evidence for a significant contribution from ancient enriched continental lithospheric mantle (i.e. [epsilon]Sr 0) in the genesis of the Damaraland lamprophyres or carbonatites.


Figure 12. Initial Sr and Nd isotope ratios of Damaraland lamprophyre and carbonatite intrusions compared with the inferred composition of the Tristan plume at 124 Ma. Hatched extension to Tristan field is for a subset of Inaccessible Island lavas (data sources as for Fig. 8).


Variations in Pb isotopes are less systematic. The Okorusu lamprophyre and carbonatite magmas and the Okenyenya lamprophyres all have Pb isotope compositions broadly similar to the calculated composition of the Tristan plume at 124 Ma (assuming a plume µ = 15 and [kappa] = 4; Fig. 8). Their Pb isotope compositions are therefore consistent with derivation from the upwelling plume, although all have slightly lower 208Pb/204Pb ratios. As a group the Okenyenya lamprophyres define an isotopic trend across the field of the Tristan plume which extends towards HIMU (Fig. 8); a feature also seen in a plot of 87Sr/86Sr vs 206Pb/204Pb (Fig. 13). The single analysis for Osongombo is similarly displaced to high 207Pb and 206Pb. However, its extremely high 208Pb/204Pb ratio is anomalous (Fig. 8), in part perhaps reflecting problems inherent in age corrections given the exceptionally high Th abundance (relative to Pb) in this sample (Table 5). Alternatively, the magma must have derived from a source with unusually high Th/Pb (and U/Pb) ratio. In contrast, the Ondurakorume and Kalkfeld carbonatites are both displaced to low 206Pb relative to the Tristan plume. These carbonatites also have Sr-Nd isotope compositions that plot well below the mantle array (Fig. 12) and, like the East African carbonatites (Bell & Blenkinsop, 1989), correspond more closely to the LoNd array of Hart et al., (1986) extending between EM-I and HIMU in Sr-Nd space. In this regard, Tilton & Bell, (1994) and Bell & Simonetti, (1996) have pointed out that most young carbonatites appear to be mixtures between HIMU and EM-I components. However, the very low 206Pb/204Pb ratios of the Ondurakorume and Kalkfeld carbonatites negate any involvement of HIMU source material in their formation, and their low 87Sr/86Sr negates significant involvement of an EM-I component (Fig. 13).


Figure 13. Covariation of 87Sr/86Sr, 143Nd/144Nd and 206Pb/204Pb isotope ratios in Damaraland lamprophyres and carbonatites. Shaded field marks composition of Tristan plume at 124 Ma.


The extreme concentrations of incompatible elements (including the LREE) in the Damaraland lamprophyres and carbonatites require a source that is highly enriched in incompatible trace elements, despite evidence for a contribution from a source component with low Sr and high Nd isotope ratios relative to Bulk Earth. Numerous recent studies have emphasized the importance of metasomatism as a precursor to highly alkaline magmatism. Experimental and petrological evidence suggests that such metasomatic fluids could be silicate or carbonatitic in composition (e.g. Wallace & Green, 1988; Yaxley et al., 1991; Hauri et al., 1993; Lee et al., 1996), and would carry the necessary inventory of incompatible trace elements. Invasion of peridotitic mantle by low-viscosity carbonatite or hydrous alkaline melts leads to cryptic or modal metasomatism in which a range of accessory phases such as amphibole, phlogopite, apatite, zircon, ilmenite, titanates, monazite and whitlockite may host many otherwise highly incompatible elements (Haggerty, 1989; Meen et al., 1989; Yaxley et al., 1991; Rudnick et al., 1993). These accessory phases not only result in an absolute increase in certain incompatible trace element abundances, but can also lead to fractionation of certain incompatible trace element ratios. Most notably, partition coefficients for elements such as Ti, Zr, P and K (e.g. Sweeney et al., 1995) indicate that incipient carbonatite melts will be depleted in these elements relative to a silicate melt, and will impose this characteristic on metasomatized peridotite. Characteristic features of carbonatite metasomatism are believed to include a decrease in the Ti/Eu ratio of the host peridotite with increasing La/Yb ratio (Rudnick et al., 1993) and a high Ca/Al ratio as a consequence of increased modal clinopyroxene (Yaxley et al., 1991). The high (>1·1) Ca/Al ratio of Damaraland ultramafic lamprophyres, and their low and decreasing Ti/Eu ratio with increasing (La/Er)n ratio (Fig. 14; Er is used here in place of Yb as many of the carbonatites have Yb values below detection), coupled with the negative Ti, Zr, P and K anomalies superimposed on their otherwise smooth primitive mantle normalized patterns (Fig. 6), is consistent with derivation from a mantle source that has experienced metasomatism by an incipient carbonatitic melt. Of importance is that the strong metasomatic enrichment is evident in the trace elements, but not the isotopes, indicating that this metasomatism is not an ancient feature but must have occurred shortly before the magmas were emplaced at ~125 Ma.


Figure 14. Variation in (La/Er)n with respect to Ti/Eu ratio in Damaraland lamprophyre and carbonatite intrusions. Schematic trend representing the influence of carbonatite metasomatism on source peridotite from Rudnick et al., (1993).


Lee et al., (1996) have shown that major and trace element and radiogenic isotope compositions of mantlederived spinel lherzolite xenoliths from the Cameroon line provide evidence for metasomatic enrichment of previously depleted sub-continental lithospheric mantle. Their isotope compositions indicate that portions of the sub-continental lithosphere beneath this region of West Africa are comparable with sub-oceanic lithosphere, i.e. are isotopically depleted. Petrographic evidence for metasomatism of sub-continental lithosphere beneath northwestern Namibia is similarly found in xenoliths from the Okenyenya complex in the form of abundant amphibole present in peridotitic xenoliths (Baumgartner, 1994). Calculated temperatures and pressures of equilibration of these xenoliths (~950-1050°C; 18-20 kbar) indicate that this metasomatism is likely to have occurred within the sub-continental lithospheric plate, at a depth immediately above the H2O-CO2-peridotite solidus ledge (i.e. ~70 km; Wyllie, 1989).

Although no isotope data are available for the Okenyenya xenoliths, by analogy to the Cameroon line xenoliths, it is proposed that at the time of continental break-up, similar material existed beneath northwestern Namibia. The evidence for a HIMU component in some samples is limited, but if indeed present it could reflect a dispersed component within the depleted sub-continental lithosphere, formed by a previous metasomatic event unrelated to the Tristan plume [e.g. Hart et al., 1986; also, Burke, (1996) has noted the near ubiquitous presence of a HIMU signature in most African alkaline magmatism], or it is present as a minor heterogeneity within the upwelling Tristan plume itself, as perhaps suggested by the Inaccessible Island data (Cliff et al., 1991).

Following Meen et al., (1989), the isotopic variations in the Damaraland lamprophyres and carbonatites are attributed to melting of metasomatized (depleted) lithosphere produced by invasion of carbonated alkalic melt, or incipient carbonatite (Wallace & Green, 1988), derived from the Tristan plume immediately before the opening of the South Atlantic at ~130 Ma. Water- and CO2-rich magmas from depths in excess of ~70 km (i.e. within the garnet stability field) reacted with depleted peridotite as they crossed the solidus ledge to produce a variety of mineral assemblages at temperatures <1100°C and pressures of between 22 and 17 kbar (Wyllie, 1989). The metasomatized mantle produced was a carbonated amphibole-bearing lherzolite which acted as a potential lamprophyre (and carbonatite?) source rock. The products of this high-temperature metasomatism are envisaged as veins enriched in clinopyroxene (wehrlites are common in the xenolith assemblages at Okenyenya), and this metasomatized mantle had Rb/Sr, U/Pb and Th/Pb ratios controlled by partitioning between mantle phases such as clinopyroxene, accessory phases such as apatite, calcite and whitlockite, and melt or fluid (Meen et al., 1989). Later melting of this metasomatized mantle (with dominant contribution from the metasomatic vein material, and variable contributions from more refractory, depleted, host sub-continental lithosphere) yielded CO2-rich ultramafic lamprophyres, carbonated nephelinites and perhaps primary carbonatite magmas (e.g. Sweeney, 1994) with strong enrichment in incompatible elements, but with relatively low 143Nd/144Nd, moderate 87Sr/86Sr and variable Pb isotope ratios.

The strong plume signature in these late-stage alkaline melts contrasts with a similar situation in Hawaii, where post-erosional, highly alkaline lavas show evidence for dominant melting of lithospheric mantle (e.g. Chen & Frey, 1983). The reasons for this difference are not clear, but our preferred model for magma generation and plume-lithosphere interaction beneath northwesternNamibia is illustrated in Fig. 15, and can be summarized as follows:

Stage 1. The Tristan plume head arrives and spreads out beneath the sub-continental lithosphere. Decompression melting within the plume gives rise to magmas that rise through the cold continental lithosphere, and if suitable pathways exist they erupt as basaltic magmas uncontaminated by sub-continental lithospheric mantle; i.e. carry the plume signature (equivalent to the mildly alkaline Etendeka Tafelkop basalts and Okenyenya alkaline gabbros; Milner & le Roex, 1996).

Stage 2. Conductive heating of the sub-continental lithosphere by the plume, coupled with rise of melts into the sub-continental lithosphere, leads to a raised geothermal gradient within the lithosphere. Volatiles or low-volume carbonate-rich alkaline melts escape from rising plume and infiltrate or metasomatize the base of the lithosphere. Magmas derived from the plume are contaminated en route to the surface by sub-continental lithospheric mantle and erupted as flood basalts; contamination is aided by the raised geothermal gradient and development of low-volume melts derived from the sub-continental lithospheric mantle which mix with plume magmas, and/ormagmas are derived by direct melting of (hydrousor metasomatized) sub-continental lithosphere (e.g. Gallagher & Hawkesworth, 1992; equivalent to the majority of Etendeka LTZ- and HTZ-basalt types).

Stage 3. With continued drift of the African plate to the northeast heat flow wanes and leads to a falling geothermal gradient. Melting of the sub-continental lithosphere is restricted to easily fusible components (metasomatic veins-originally derived from the plume), with perhaps a limited contribution from the host peridotite, leading to the formation of lamprophyric and carbonatitic magmas that carry the Tristan plume signature.


Figure 15. Schematic diagram (not to scale) showing evolution of plume-lithosphere interaction beneath northwestern Namibia. (1) Eruption or emplacement of early (mildly alkaline) magmas uncontaminated by cold sub-continental lithosphere (o Etendeka Tafelkop basalts and Okenyenya alkaline gabbros; Milner & le Roex, 1996). (2) Conductive heating of lithosphere and metasomatism by the Tristan plume. Large volume melts contaminated by, or generated within, sub-continental lithosphere as a result of raised geothermal gradient (o Etendeka LTZ- and HTZ-basalts and Okenyenya tholeiitic gabbros; Milner & le Roex, 1996). (3) Because of waning heat flow, melting is largely restricted to metasomatic veins (originally derived from the plume), giving rise to lamprophyre and carbonatite magmas. (See text for further discussion.)


ACKNOWLEDGEMENTS

Logistic support provided by the Geological Survey of Namibia and financial support provided by the Foundation for Research Development and the University of Cape Town are gratefully acknowledged. Petrie Prins is thanked for allowing us access to his Damaraland carbonatite collection housed in the Geology Department at the University of Stellenbosch. Steve Richardson, Andreas Späth, and particularly Richard Armstrong, provided invaluable help with radiogenic isotope analyses; Chris Harris and Kevin Faure provided equivalent assistance with stable isotope analyses; Simon and Debbie Milner are thanked for their hospitality in Windhoek. Informal reviews by Chris Harris and Phil Janney, valuable insights from Conny Class, and formal reviews by Fred Frey, Godfrey Fitton and Keith Bell greatly improved earlier drafts of the manuscript, and for their conscientious efforts we are greatly indebted.

REFERENCES

Barker, D. S. (1989). Field relations of carbonatites. In: Bell, K. (ed.) Carbonatites: Genesis and Evolution. London: Unwin Hyman, pp. 38-69.

Baumgartner, M. C. (1994). The xenoliths of the Okenyenya volcanic breccia. M.Sc. Thesis, University of Cape Town, 144 pp.

Bell, K. (1989). Carbonatites: Genesis and Evolution. London: Unwin Hyman, 618 pp.

Bell, K. & Blenkinsop, J. (1989). Neodymium and strontium isotope geochemistry of carbonatites. In: Bell, K. (ed.) Carbonatites: Genesis and Evolution. London: Unwin Hyman, pp. 278-300.

Bell, K. & Simonetti, A. (1996). Carbonatitic magmatism and plume activity: implications from Nd, Pb and Sr isotope systematics of Oldoinyo Lengai. Journal of Petrology 37, 1321-1339.

Birch, G. F. (1981). The Karbonat-bombe: a precise, rapid and cheap instrument for determining calcium carbonate in sediments and rocks. Transactions of the Geological Society of South Africa 84, 199-203.

Burke, K. (1996). The African plate. South African Journal of Geology 99, 341-409.

Carlson, R. W., Esperanca, S. & Svisero, D. P. (1996). Chemical and Os isotopic study of Cretaceous potassic rocks of southern Brazil. Contributions to Mineralogy and Petrology 125, 393-405.

Catanzaro, E. J., Murphy, T. J., Shields, W. R. & Garner, E. L. (1968). Absolute isotopic abundance ratios of common equal-atom and radiogenic lead isotopic standards. Journal of Research of the National Bureau of Standards A, Physics and Chemistry 72A, 261-267.

Chen, C.-Y. & Frey, F. A. (1983). Origin of Hawaiian tholeiites and alkalic basalts. Nature 302, 785-789.

Cliff, R. A., Baker, P. E. & Mateer, N. J. (1991). Geochemistry of Inaccessible Island volcanics. Chemical Geology 92, 251-260.

Deines, P. (1989). Stable isotope variations in carbonatites. In: Bell, K. (ed.) Carbonatites: Genesis and Evolution. London: Unwin Hyman, pp. 301-359.

Duncan, R. A. (1984). Age progressive volcanism in the New England seamounts and the opening of the central Atlantic Ocean. Journal of Geophysical Research 89, 9980-9990.

Gallagher, K. & Hawkesworth, C. (1992). Dehydration melting and the generation of continental flood basalts. Nature 358, 57-59.

Gittins, J. (1988). The origin of carbonatites. Nature 335, 295-296.

Green, D. H. & Wallace, M. E. (1988). Mantle metasomatism by ephemeral carbonatite melts. Nature 336, 459-462.

Haggerty, S. E. (1989). Mantle metasomes and the kinship between carbonatites and kimberlites. In: Bell, K. (ed.) Carbonatites: Genesis and Evolution. London: Unwin Hyman, pp. 546-560.

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

Hart, S. R., Gerlach, D. C. & White, W. M. (1986). A possible new Sr-Nd-Pb isotopic array and consequences of mantle mixing. Geochimica et Cosmochimica Acta 50, 1551-1557.

Hauri, E. H., Shimizu, N., Dieu, J. J. & Hart, S. R. (1993). Evidence for hotspot-related carbonatite metasomatism in the oceanic upper mantle. Nature 365, 221-227.

Hawkesworth, C. J., Gallagher, K., Kelley, S., Mantovani, M., Peate, D. W., Regelous, M. & Rogers, N. W. (1992). Paraná magmatism and the opening of the South Atlantic. In: Storey, B. C., Alabaster, T. & Pankhurst, R. J. (eds) Magmatism and the Causes of Continental Break-up. Special Publication, Geological Society, London 68, 221-240.

Kjarsgaard, B. A. & Hamilton, D. L. (1989). Carbonatite origin and diversity. Nature 338, 547-548.

Koster van Groos, A. F. (1975). The distribution of strontium between coexisting silicate and carbonate liquids at elevated pressures and temperatures. Geochimica et Cosmochimica Acta 39, 27-34.

Lanyon, R. & le Roex, A. P. (1995). Petrology of the alkaline and ultramafic lamprophyres associated with the Okenyenya igneous complex, northwestern Namibia. South African Journal of Geology 98, 140-156.

Le Bas, M. J. (1987). Nephelinites and carbonatites. In: Fitton, J. G. & Upton, B. G. J. (eds) Alkaline Igneous Rocks. Special Publication, Geological Society, London, 30, 53-83.

Le Bas, M. J. (1989). Diversification of carbonatite. In: Bell, K. (ed.) Carbonatites: Genesis and Evolution. London: Unwin Hyman, pp. 428-447.

Lee, D.-C., Halliday, A. N., Davies, G. R., Essene, E. J., Fitton, G. & Temdjim, R. (1996). Melt enrichment of shallow depleted mantle: a detailed petrological, trace element and isotopic study of mantle-derived xenoliths and megacrysts from the Cameroon line. Journal of Petrology 37, 415-441.

Lee, W. & Wyllie, P. J. (1994). Experimental data bearing on liquid immiscibility, crystal fractionation and the origin of calciocarbonatites and natrocarbonatites. International Geology Review 36, 797-819.

Lee, W. & Wyllie, P. J. (1997). Liquid immiscibility between nephelinite and carbonatite from 1·0 to 2·5 GPa compared with mantle melt compositions. Contributions to Mineralogy and Petrology 127, 1-16.

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.

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

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.

le Roex, A. P., Watkins, R. T. & Reid, A. M. (1996). Geochemical evolution of the Okenyenya sub-volcanic ring complex, northwestern Namibia. Mineralogical Magazine 133, 645-670.

Martin, H., Mathias, M. & Simpson, E. S. W. (1960). The Damaraland sub-volcanic ring complexes in South West Africa. Report of the International Geological Congress XXI Session 13, Copenhagen: Det Berlingske Bogtrykkeri, pp. 156-174.

Martinez, I. A., Harris, C., le Roex, A. P. & Milner, S. C. (1996). Oxygen isotope evidence for extensive crustal contamination in the Okenyenya igneous complex, Namibia. Geochimica et Cosmochimica Acta 60, 4497-4508.

McKenzie, D. & O'Nions, R. K. (1991). Partial melt distributions from inversion of rare earth element concentrations. Journal of Petrology 32, 1021-1091.

Meen, J. K., Ayers, J. C. & Fregeau, E. J. (1989). A model of mantle metasomatism by carbonated alkaline melts: trace-element and isotopic compositions of mantle source regions of carbonatite and other continental igneous rocks. In: Bell, K. (ed.) Carbonatites: Genesis and Evolution. London: Unwin Hyman, pp. 464-499.

Middlemost, E. A. K. (1974). Petrogenetic model for the origin of carbonatites. Lithos 7, 275-278.

Miller, R. McG. (1983). The pan-African Damaran orogen of South West Africa/Namibia. In: Miller, R. McG. (ed.) Geological Society of South Africa, Special Publication 11, 431-515.

Milner, S. C. & le Roex, A. P. (1996). Isotope characteristics of the Okenyenya igneous complex, northwestern Namibia: constraints on the composition of the early Tristan plume and the EM1 mantle component. Earth and Planetary Science Letters 141, 277-291.

Milner, S. C., le Roex, A. P. & Watkins, R. T. (1993). Rb-Sr age determinations of rocks from the Okenyenya igneous complex, northwestern Namibia. Geological Magazine 130, 335-343.

Milner, S. C., le Roex, A. P. & O'Connor, J. M. (1995). Age of Mesozoic igneous rocks in northwestern Namibia, and their relationship to continental breakup. Journal of the Geological Society, London 151, 97-104.

Prins, P. (1981). The geochemical evolution of the alkaline and carbonatite complexes of the Damaraland igneous province, South West Africa. Annall Universiteit van Stellenbosch Serie A1 (Geologie) 3, 145-278.

Renne, P. R., Ernesto, M., Pacca, I. G., Coe, R. S., Glen, J. M., Prévot, M. & Perrin, M. (1992). The age of Paraná flood volcanism, rifting of Gondwana, and the Jurassic-Cretaceous boundary. Science 258, 975-979.

Richardson, S. R., Erlank, A. J., Duncan, A. R. & Reid, D. L. (1982). Correlated Sr, Nd and Pb isotope variations in Walvis Ridge basalts and implications for the evolution of their mantle source. Earth and Planetary Science Letters 59, 327-342.

Rock, N. M. S. (1991). Lamprophyres. Glasgow: Blackie.

Roden, M. F., Rama Murthy, V. & Gaspar, J. C. (1985). Sr and Nd composition of the Jacupiranga carbonatite. Journal of Geology 93, 212-220.

Rudnick, R. L., McDonough, W. F. & Chappell, B. W. (1993). Carbonatite metasomatism in northern Tanzanian mantle: petrographic and geochemical characteristics. Earth and Planetary Science Letters 114, 463-475.

Santos, R. V. & Clayton, R. N. (1995). Variations of oxygen and carbon isotopes in carbonatites: a study of Brazilian alkaline complexes. Geochimica et Cosmochimica Acta 59, 1339-1352.

Siedner, G. & Mitchell, J. G. (1976). Episodic Mesozoic volcanism in Namibia and Brazil: a K-Ar isochron study bearing on the opening of the South Atlantic. Earth and Planetary Science Letters 30, 292-302.

Simpson, E. S. W. (1954). The Okonjeje igneous complex, South West Africa. Transactions of the Geological Society of South Africa 57, 125-172.

Späth, A., le Roex, A. P. & Duncan, R. A. (1996). The geochemistry of lavas from the Comores Archipelago, western Indian Ocean: petrogenesis and source region characteristics. Journal of Petrology 37, 961-991.

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. Special Publication, Geological Society, London 42, 313-345.

Sweeney, R. J. (1994). Carbonatite melt compositions in the Earth's mantle. Earth and Planetary Science Letters 128, 259-270.

Sweeney, R. J., Prozesky, V. & Przybylowicz, W. (1995). Selected trace and minor element partitioning between peridotite minerals and carbonatite melts at 18-46 kb pressure. Geochimica et Cosmochimica Acta 59, 3671-3683.

Thompson, R. N. & Gibson, S. A. (1991). Subcontinental mantle plumes, hotspots and pre-existing thinspots. Journal of the Geological Society, London 148, 973-977.

Tilton, G. R. & Bell, K. (1994). Sr-Nd-Pb isotope relationships in Late Archean carbonatites and alkaline complexes: applications to the geochemical evolution of the Archean mantle. Geochimica et Cosmochimica Acta 58, 3145-3154.

Toyoda, K., Horiuchi, H. & Tokonami, M. (1994). Dupal anomaly of Brazilian carbonatites: geochemical correlations with hotspots in the South Atlantic and implications for the mantle source. Earth and Planetary Science Letters 126, 315-331.

Turner, S., Regelous, M., Kelley, S., Hawkesworth, C. & Mantovani, M. (1994). Magmatism and continental break-up in the South Atlantic: high precision 40Ar-39Ar geochronology. Earth and Planetary Science Letters 121, 333-348.

Van Zijl, P. J. (1962). The geology, structure and petrology of the alkaline intrusions of Kalkfeld and Okorusu and the invaded Damara rocks. Annall Universiteit van Stellenbosch, Serie 4A 37, 237-346.

Verwoerd, W. J. (1966). South African carbonatites and their probable mode of origin. Annall Universiteit van Stellenbosch, Serie A2 41, 115-233.

Wallace, M. E. & Green, D. H. (1988). An experimental determination of primary carbonatite magma composition. Nature 335, 343-346.

Watkins, R. T. & le Roex, A. P. (1991). Petrology and structure of syenite intrusions of the Okenyenya igneous complex. Communications of the Geological Survey of Namibia 7, 55-70.

Watkins, R. T. & le Roex, A. P. (1994). A revised geology of the Okenyenya igneous complex: structure and history of emplacement. Communications of the Geological Survey of Namibia 9, 13-21.

Watkins, R. T., McDougall, I. & le Roex, A. P. (1994). K-Ar ages of the Brandberg and Okenyenya igneous complexes, northwestern Namibia. Geologische Rundschau 83, 348-356.

White, R. & McKenzie, D. (1989). Magmatism at rift zones: the generation of volcanic continental margins and flood basalts. Journal of Geophysical Research 94, 7685-7729.

Woolley, A. R. & Kempe, D. R. C. (1989). Carbonatites: nomenclature, average chemical compositions, and element distribution. In: Bell, K. (ed.) Carbonatites: Genesis and Evolution. London: Unwin Hyman, pp. 1-14.

Wyllie, P. J. (1989). Origin of carbonatites: evidence from phase equilibria studies. In: Bell, K. (ed.) Carbonatites: Genesis and Evolution. London: Unwin Hyman, pp. 500-545.

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.


Top

*Corresponding author. Telephone: + 27-21-650-2921. Fax: + 27-21-650-3783. e-mail: alr@geology.uct.ac.za
[dagger]Present address: Port Pirie Environmental Health Centre, 117 Gertrude Street, Port Pirie, S.A., Australia 5540.


This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals World Wide Web service.
Comments and feedback: www-admin@oup.co.uk
Last modification: June 1998
Copyright© Oxford University Press, 1998.