Search

This paper presents major, trace element and Sr-Nd isotope data on two large, east-west trending dolerite dykes in the southeastern part of the Karoo Basin, the South Gap (155 km long, ≤275 m wide) and the North Gap (150 km long, ≤190 m wide) dykes. The Gap dykes represent a rare case in the Karoo Large Igneous Province (LIP) where low-Ti (Gap1

Datasets emanating from research undertaken at the Department of Geology at Rhodes University.

The Lesotho remnant contains the type succession for Karoo low-Ti basalts of central southern Africa. The 40Ar/39Ar dating indicates that the sequence was emplaced within a very short period at about 180 Ma and consists of a monotonous pile of compound basalt lava flows which lacks significant palaeosols and persistent sedimentary intercalations. We have used geochemistry to establish a stratigraphic subdivision of the lava pile. Thin units of basalt flows, the Moshesh's Ford, Golden Gate, Sani, Roma, Letele, and Wonderkop units, with diverse geochemical character and restricted geographical distribution, are present at the base of the succession. Location of this section is described in Marsh et al. (1997) AGU Geophysical Monograph, 100, 247-272. Title of data set: Oxbow Section The Mahlasela Pass road has been reconstructed and lies to the N of the stream whereas the old pass, sampled for this collection, lay to the south. Rahacek (1995) PhD thesis. Washington State University, designated samples OXB54-60 as belonging to the Pitseng Unit.

The Lesotho remnant contains the type succession for Karoo low-Ti basalts of central southern Africa. The 40Ar/39Ar dating indicates that the sequence was emplaced within a very short period at about 180 Ma and consists of a monotonous pile of compound basalt lava flows which lacks significant palaeosols and persistent sedimentary intercalations. We have used geochemistry to establish a stratigraphic subdivision of the lava pile. Thin units of basalt flows, the Moshesh's Ford, Golden Gate, Sani, Roma, Letele, and Wonderkop units, with diverse geochemical character and restricted geographical distribution, are present at the base of the succession. These are overlain by extensive units of compositionally more uniform basalt, the Mafika Lisiu, Maloti, Senqu and Mothae units, which build the bulk of the sequence.Location of this section is described in Marsh et al. (1997) AGU Geophysical Monograph, 100, 247-272. Title of data set: Roma Section

The Lesotho remnant contains the type succession for Karoo low-Ti basalts of central southern Africa. The 40Ar/39Ar dating indicates that the sequence was emplaced within a very short period at about 180 Ma and consists of a monotonous pile of compound basalt lava flows which lacks significant palaeosols and persistent sedimentary intercalations. We have used geochemistry to establish a stratigraphic subdivision of the lava pile. Thin units of basalt flows, the Moshesh's Ford, Golden Gate, Sani, Roma, Letele, and Wonderkop units, with diverse geochemical character and restricted geographical distribution, are present at the base of the succession. These are overlain by extensive units of compositionally more uniform basalt, the Mafika Lisiu, Maloti, Senqu and Mothae units, which build the bulk of the sequence.Location of this section is described in Marsh et al. (1997) AGU Geophysical Monograph, 100, 247-272. Title of data set: Springbok Flats Location of Borehole RL1 (SF samples) – S24.9367 deg; E 28.3750 deg Location of Borehole RTL1 – S 24.4400 deg; E 29.1767 deg Location of Borehole WD4 – S 24.6483 deg; E 28.7450 deg Location of Borehole LB1 – S 24.8817 deg; E 28.5833 deg Borehole TF2 – base of volcanic sequence – 768m Borehole TF1 – base of volcanic sequence – 357m All Sr-, Nd- and Pb-isotope values are MEASURED values.

The Lesotho remnant contains the type succession for Karoo low-Ti basalts of central southern Africa. The 40Ar/39Ar dating indicates that the sequence was emplaced within a very short period at about 180 Ma and consists of a monotonous pile of compound basalt lava flows which lacks significant palaeosols and persistent sedimentary intercalations. We have used geochemistry to establish a stratigraphic subdivision of the lava pile. Thin units of basalt flows, the Moshesh's Ford, Golden Gate, Sani, Roma, Letele, and Wonderkop units, with diverse geochemical character and restricted geographical distribution, are present at the base of the succession. These are overlain by extensive units of compositionally more uniform basalt, the Mafika Lisiu, Maloti, Senqu and Mothae units, which build the bulk of the sequence.Location of this section is described in Marsh et al. (1997) AGU Geophysical Monograph, 100, 247-272. Title of data set: Mafika Lisiu Pass section.

The Lesotho remnant contains the type succession for Karoo low-Ti basalts of central southern Africa. The 40Ar/39Ar dating indicates that the sequence was emplaced within a very short period at about 180 Ma and consists of a monotonous pile of compound basalt lava flows which lacks significant palaeosols and persistent sedimentary intercalations. We have used geochemistry to establish a stratigraphic subdivision of the lava pile. Thin units of basalt flows, the Moshesh's Ford, Golden Gate, Sani, Roma, Letele, and Wonderkop units, with diverse geochemical character and restricted geographical distribution, are present at the base of the succession. These are overlain by extensive units of compositionally more uniform basalt, the Mafika Lisiu, Maloti, Senqu and Mothae units, which build the bulk of the sequence. Location of this section is described in Marsh et al. (1997) AGU Geophysical Monograph, 100, 247-272. Title of data set: Semongkong Section This section is entirely located in the Lesotho Formation and does not go down to the Clarens Formation contact with the base of the volcanic sequence.

The Lesotho remnant contains the type succession for Karoo low-Ti basalts of central southern Africa. The 40Ar/39Ar dating indicates that the sequence was emplaced within a very short period at about 180 Ma and consists of a monotonous pile of compound basalt lava flows which lacks significant palaeosols and persistent sedimentary intercalations. We have used geochemistry to establish a stratigraphic subdivision of the lava pile. Thin units of basalt flows, the Moshesh's Ford, Golden Gate, Sani, Roma, Letele, and Wonderkop units, with diverse geochemical character and restricted geographical distribution, are present at the base of the succession. These are overlain by extensive units of compositionally more uniform basalt, the Mafika Lisiu, Maloti, Senqu and Mothae units, which build the bulk of the sequence.Location of this section is described in Marsh et al. (1997) AGU Geophysical Monograph, 100, 247-272. Title of data set: Ongeluksnek Pass

The Lesotho remnant contains the type succession for Karoo low-Ti basalts of central southern Africa. The 40Ar/39Ar dating indicates that the sequence was emplaced within a very short period at about 180 Ma and consists of a monotonous pile of compound basalt lava flows which lacks significant palaeosols and persistent sedimentary intercalations. We have used geochemistry to establish a stratigraphic subdivision of the lava pile. Thin units of basalt flows, the Moshesh's Ford, Golden Gate, Sani, Roma, Letele, and Wonderkop units, with diverse geochemical character and restricted geographical distribution, are present at the base of the succession. These are overlain by extensive units of compositionally more uniform basalt, the Mafika Lisiu, Maloti, Senqu and Mothae units, which build the bulk of the sequence.

The Lesotho remnant contains the type succession for Karoo low-Ti basalts of central southern Africa. The 40Ar/39Ar dating indicates that the sequence was emplaced within a very short period at about 180 Ma and consists of a monotonous pile of compound basalt lava flows which lacks significant palaeosols and persistent sedimentary intercalations. We have used geochemistry to establish a stratigraphic subdivision of the lava pile. Thin units of basalt flows, the Moshesh's Ford, Golden Gate, Sani, Roma, Letele, and Wonderkop units, with diverse geochemical character and restricted geographical distribution, are present at the base of the succession. These are overlain by extensive units of compositionally more uniform basalt, the Mafika Lisiu, Maloti, Senqu and Mothae units, which build the bulk of the sequence.Location of this section is described in Marsh et al. (1997) AGU Geophysical Monograph, 100, 247-272. Title of data set: Sani Pass

One of the most significant results emerging from the Karoo Volcanics Project of the NGP ls the recognition of a number of geochemically distinct basalt magma types occurring within the lower part of the Karoo volcanic pile in the Northeastern Cape and Southern Lesotho.NE Free State

Geochemical data and mapping from a Karoo flood basalt crater complex reveals new information about the ascent and eruption of magma batches during the earliest phases of flood basalt volcanism. Flood basalt eruptions at Sterkspruit, South Africa began with emplacement of thin lava flows before abruptly switching to explosive phreatomagmatic and magmatic activity that formed a nest of craters, spatter and tuff rings and cones that collectively comprise a crater complex >40 km2 filled by 9–18 km3 of volcaniclastic debris. Rising magma flux rates combined with reduced access of magma to external water led to effusion of thick Karoo flood basalts, burying the crater-complex beneath the >1.5 km-thick Lesotho lava pile.Sterkspruit dataset

Analyses in this dataset are from the following primary sources: Bowen, MP (1979) Some petrological aspects of the Omega Member flow unit, Barkly East District. Unpublished BSc Honours project, Rhodes University, 83pp. Pemberton, M (1979) A study of the massive unit of the Omega Member near Barkly East. Unpublished BSc Honours project, Rhodes University, 44pp Minne, HFD (1975) A study of a sequence of lava flows in the lower part of the Drakensberg Subgroup near Barkly East. Unpublished BSC Honours project, Rhodes University, 38pp Pemberton, J (1978) The geochemistry and petrology of Karoo basalts of the Barkly East area, north-eastern Cape. Unpublished MSc thesis, Rhodes University, 139pp Barrie, PD (1977) The Kraai River Formation: a petrographical and geochemical study of an occurrence in the vicinity of Barkly East. Unpublished BSC Honours project, Rhodes University. Major element analyses from original source. All trace elements redetermined in 2004.

This chapter contains sections titled:IntroductionOverview of the Karoo ProvinceSampling and Analytical TechniquesOverview of Stratigraphic VariationsThe Basis for Geochemical SubdivisionCorrelation Between Palaeomagnetic and Geochemical StratigraphyImplications for Structure and Emplacement of the Lava SequenceThe Age of Karoo VolcanismCorrelations with Other Karoo Volcanic SuccessionsConclusionsAppendixRESEARCH DATA. Location of this section is described in Marsh et al. (1997) AGU Geophysical Monograph, 100, 247-272. Title of data set: Ben McDhui Section. Altitudes measured by anaeroid altimeter.

High precision dating of Large Igneous Provinces (LIP) is not only useful to understand their link to environmental changes and mass extinctions (Courtillot and Renne, 2003), but they also provide insights into the geodynamic setting in which they form (Encarnación et al., 1996). The Drakensberg continental flood basalts of South Africa and Lesotho are part of the Karoo LIP, which is presumably responsible for a phase of global climate change and disturbance of the oceanic ecosystems (the so-called Toarcian oceanic anoxic event T-OAE; Pálfy and Smith (2000)). However, the paucity of zircon or baddeleyite in most continental flood basalts renders is difficult to match the sub-permil age precision and accuracy that is typical for high-precision U/Pb CA-ID-TIMS age determination. Previous attempts to date the Karoo lavas using the 40Ar-39Ar method failed to yield sufficient precision and accuracy for resolving the sequential stacking of the different basalt units. For example, 40Ar-39Ar analyses of carefully selected plagioclase separates yielded dates that are inverted relative to their stratigraphic position, with uncertainties that encompass the entire duration of volcanism in the area (Jourdan et al., 2007; Moulin et al., 2017). Here we test the hypothesis that previous, inconsistent 40Ar-39Ar dates of plagioclase were a consequence of degassing of primary, metasomatic and alteration phases (mainly zeolites with subordinate sericite and carbonate) within single or multiple crystals. The lavas are mainly tholeiitic basalts that display two distinct sizes of plagioclase, which can be dated separately. Petrological characterization of these two size fractions shows that the larger plagioclase crystals (100-400 μm) are more altered and fractured than the smaller grains and are therefore more likely affected by post-crystallization disturbance of the Ar isotopic system. We present preliminary 40Ar-39Ar data from i) untreated plagioclase that hosts visible alteration phases, ii) untreated plagioclase that is devoid of visible alteration phases (2 grain size aliquots), and iii) leached plagioclase that is devoid of visible alteration phases (2 grain size aliquots). The results of this study may enhance the effectiveness of the 40Ar-39Ar dating technique to accurately constrain the crystallisation ages of altered mafic lavas, which form the majority of the exposed Karoo LIP flood basalts. Ar isotope data were collected using a multi-collector Argus VI mass spectrometer, and irradiated in an un-shielded reactor position to optimize the formation of 38Ar from Cl to permit identification of different gas reservoirs in the sample through isochemical dating, based on Ca, K and Cl in-situ concentration (EPMA) and Ar isotopic ratios.

An extension of the Etendeka-Parana Igneous Province into SW Angola occurs as minor basalt lavas, intrusive gabbro sheets, minor mafic dykes and thick sheets and lava flows (with minor pyroclastics) of quartz latite composition This suite crops out along the eastern margin of the Cretaceous Namibe Basin in SW Angola. The quartz latites from one locality have been referred to informally as the Giraul volcanics but the name ‘Giraul’ has previously been used for Cretaceous conglomerates. We propose the name Bero Volcanic Complex for this suite of intrusive and extrusive rocks on the basis that the full compositional range of this diverse suite crops out along the Rio Bero. Major and trace element compositions and Sr-, Nd-, and Pb-isotopic compositions indicate that the basalts and gabbros are equivalent to the high-Ti Khumib/Urubici and Pitanga types from the Etendeka and Parana. The basalts underlie the quartz latites which are cut by mafic dykes some of which are compositionally equivalent to the Paranapanema lavas in the Parana. Five different geochemical types of high-Ti quartz latite are recognised amongst the silicic volcanics, 3 of which have very close geochemical affinities to the Ventura, Sarusas, and Khoraseb types of the northern Etendeka. Their relative stratigraphic position in the Bero volcanic sequence is the same as in the Etendeka sequence and extend significantly the area over which these types were erupted. The two remaining types, Chinguau and High-Nb are not known from either the Etendeka or the Parana provinces.

This paper reports new paleomagnetic and geochronologic data from a ~1500 m-thick composite section belonging to the Drakensberg group, the thickest remnant of the Karoo lavas in Northern Lesotho. Flow-by-flow analysis of paleomagnetic directions reveals twenty-one magnetic directional groups (DG), corresponding to single eruptive events (SEE), and sixteen individual lava flows. The new age determinations of lava flows range from 180.1 ± 1.4 to 182.8 ± 2.6 Ma. These data, combined with previous results allow us to propose that the main part of the Drakensberg group and the Karoo intrusive complex dated around 181-183 Ma may have been erupted over a period as short as 250 kyr, and may have coincided with the two main phases of extinction in the Early Toarcian. This scenario agrees well with the discontinuous rhythm of environmental and biotic perturbations in the late Pliensbachian-Toarcian interval.

Physical volcanological features are presented for a 710-m-thick section, of the Naude’s Nek Pass, within the lower part of the Lesotho remnant of the Karoo Large Igneous Province. The section consists of inflated pāhoehoe lava with thin, impersistent sedimentary interbeds towards the base. There are seven discreet packages of compound and hummocky pāhoehoe lobes containing flow-lobe tumuli, making up approximately 50% of the section. Approximately 45% of the sequence consists of 14 sheet lobes, between 10 and 52-m-thick. The majority of the sheet lobes are in two packages indicating prolonged periods of lava supply capable of producing thick sheet lobes. The other sheet lobes are as individual lobes or pairs, within compound flows, suggesting brief increases in lava supply rate. We suggest, contrary to current belief, that there is no evidence that compound flows are proximal to source and sheet lobes (simple flows) are distal to source and we propose that the presence of flow-lobe tumuli in compound flows could be an indicator that a flow is distal to source. We use detailed, previously published, studies of the Thakurvadi Formation (Deccan Traps) as an example. We show that the length of a lobe and therefore the sections that are ‘medial or distal to source’ are specific to each individual lobe and are dependent on the lava supply of each eruptive event, and as such flow lobe tumuli can be used as an indicator of relative distance from source.

Large igneous provinces (LIPs) are sites of the most frequently recurring, largest volume basaltic and silicic eruptions in Earth history. These large-volume (N1000 km3 dense rock equivalent) and large-magnitude (NM8) eruptions produce areally extensive (104–105 km2) basaltic lava flow fields and silicic ignimbrites that are the main building blocks of LIPs. Available information on the largest eruptive units are primarily from the Columbia River and Deccan provinces for the dimensions of flood basalt eruptions, and the Parana–Etendeka and Afro-Arabian provinces for the silicic ignimbrite eruptions. In addition, three large-volume (675– 2000 km3) silicic lava flows have also been mapped out in the Proterozoic Gawler Range province (Australia), an interpreted LIP remnant. Magma volumes of N1000 km3 have also been emplaced as high-level basaltic and rhyolitic sills in LIPs. The data sets indicate comparable eruption magnitudes between the basaltic and silicic eruptions, but due to considerable volumes residing as co-ignimbrite ash deposits, the current volume constraints for the silicic ignimbrite eruptions may be considerably underestimated. Magma composition thus appears to be no barrier to the volume of magma emitted during an individual eruption. Despite this general similarity in magnitude, flood basaltic and silicic eruptions are very different in terms of eruption style, duration, intensity, vent configuration, and emplacement style. Flood basaltic eruptions are dominantly effusive and Hawaiian–Strombolian in style, with magma discharge rates of ~106–108 kg s−1 and eruption durations estimated at years to tens of years that emplace dominantly compound pahoehoe lava flow fields. Effusive and fissural eruptions have also emplaced some large-volume silicic lavas, but discharge rates are unknown, and may be up to an order of magnitude greater than those of flood basalt lava eruptions for emplacement to be on realistic time scales (b10 years). Most silicic eruptions, however, are moderately to highly explosive, producing co-current pyroclastic fountains (rarely Plinian) with discharge rates of 109– 1011 kg s−1 that emplace welded to rheomorphic ignimbrites. At present, durations for the large-magnitude silicic eruptions are unconstrained; at discharge rates of 109 kg s−1, equivalent to the peak of the 1991 Mt Pinatubo eruption, the largest silicic eruptions would take many months to evacuate N5000 km3 of magma. The generally simple deposit structure is more suggestive of short-duration (hours to days) and high intensity (~1011 kg s−1) eruptions, perhaps with hiatuses in some cases. These extreme discharge rates would be facilitated by multiple point, fissure and/or ring fracture venting of magma. Eruption frequencies are much elevated for large-magnitude eruptions of both magma types during LIP-forming episodes. However, in basaltdominated provinces (continental and ocean basin flood basalt provinces, oceanic plateaus, volcanic rifted margins), large magnitude (NM8) basaltic eruptions have much shorter recurrence intervals of 103–104 years, whereas similar magnitude silicic eruptions may have recurrence intervals of up to 105 years. The Parana– Etendeka province was the site of at least nine NM8 silicic eruptions over an ~1 Myr period at ~132 Ma; a similar eruption frequency, although with a fewer number of silicic eruptions is also observed for the Afro- Arabian Province. The huge volumes of basaltic and silicic magma erupted in quick succession during LIP events raises several unresolved issues in terms of locus of magma generation and storage (if any) in the crust prior to eruption, and paths and rates of ascent from magma reservoirs to the surface.

Phonolites of Palaeogene age occur at two different localities in central Namibia, occurring as clusters of eroded monogenetic lava domes. Six phonolites at Aris collectively exhibit a narrow compositional range and represent a single magma system of highly evolved phonolite with high concentrations of Na2O, Rb, Th, Zr, Zn, Pb, and REE and very low abundances of MgO, P2O5, Sr and Ba. No mafic rocks are associated with the Aris phonolites and their ultimate petrogenetic origin remains obscure. The more abundant Staalhart phonolites exhibit more variable but scattered compositional variation consistent with fractional crystallization dominated by sanidine (and nepheline) and pyroxene. The behaviour of REE is decoupled from Zr, Nb, and Th in this suite indicating a controlling role for minor phases in their evolution. The least evolved phonolites have initial 87Sr/86Sr ≈ 0.7043, which is identical to a small occurrence of associated plagioclase-bearing nephelinite indicating a possible petrogenetic link. Higher isotopic ratios in more evolved phonolites suggest their evolution was accompanied by some crustal interaction. Comparisons of phonolites world-wide indicate that they all have similar trace element abundance patterns when normalized to associated and possibly parental mafic compositions, regardless of proposed petrogenetic origins (melting, fractional crystallization, AFC, etc.). These patterns of similar pronounced enrichments of Rb, Th, Zr, strong depletions in Ti and P, and YbN > GdN (i.e. depletion of middle REE relative to heavy REE) confirm the suggestion that fractional crystallization is the dominant process by which phonolites evolve from mafic alkaline compositions. Despite similar published values of DEu for sanidine in silicic and phonolitic compositions, highly evolved phonolites develop only modest negative Eu anomalies compared to evolved silicic rocks. This might reflect differences in melt structure, fO2, halogen complexing or fractionating mineral assemblages in phonolite and silicic liquids. Sparse data on Eu partition coefficients in phonolite bulk compositions does not provide insight into the role of these controls of Eu partitioning in phonolite.

Twelve Group II and fourteen Group I kimberlite diatremes in central South Africa were examined for upper crustal xenoliths in order to estimate the extent of various lithological units of the Karoo Supergroup in the main Karoo basin at times of kimberlite eruption, the Cretaceous erosional history of the area, and the approximate vertical extent of the kimberlite diatremes prior to erosion. Sandstone and amygdaloidal basaltic lava xenoliths from the Karoo Supergroup were specifically selected as their modal mineralogies and geochemical compositions respectively can be attributed to specific stratigraphic positions within the Karoo Supergroup. Results indicated that, at the time of Group II kimberlite eruption (~120 Ma), basaltic lavas of the Drakensberg Group covered the entire area, but by the time of Group I kimberlite eruption (~85 Ma), they were restricted to the south-eastern half of the study area. At the latter time, an escarpment is proposed to have existed at the basalt outcrop limit, some 180 km west of its current position. Sandstones of the Stormberg Group had a restricted original distribution in the north and east of the study area, whereas sandstones from all other Karoo groups occurred throughout the entire area. In the Kimberley area, approximately 500 m of erosion is estimated to have occurred from 120 to 85 Ma and 850 meters from 85 Ma to the present day at average rates of approximately 15 m/Ma and 10 m/Ma respectively. Both Group I and II kimberlite diatremes had vertical extents of approximately 1350 m at eruption. An inland scarp-retreat model is proposed for the Cretaceous erosion cycle in central South Africa.

The Insizwa sill is an ~ 1 km-thick subhorizontal layered mafic intrusion and part of the Karoo Large Igneous Province in South Africa. This well-exposed intrusion consists of several superimposed petrologically and geochemically distinct units. Magnetic methods were used to study the intrusion in order to constrain the physical processes active in these types of bodies during crystallization. Rock magnetism studies indicate that within different petrologic units bulk susceptibility is controlled by primary magnetite (with minor pyrrhotite) and/or paramagnetic minerals (olivine, pyroxene). New magnetic data based on 659 specimens obtained from 3 vertical borehole cores, each spaced 5 km apart, confirm the prominent vertical zonation in low field magnetic susceptibility ( K lf ), degree of anisotropy ( P j ) and orientation of the anisotropy of magnetic susceptibility (AMS) axes. The magnetic susceptibility correlates very well with petrographic units and the lateral continuity of magnetic units between boreholes is very consistent. Petrologic units with high, but variable, K lf , also show moderate anisotropy and dominantly vertical foliations. We interpret these patterns to result from inverse fabrics from single domain magnetite. The degree of anisotropy is low in petrologic units with low K lf , which also show shallowly dipping magnetic foliations. We interpret that the magnetic properties of these units are dominated by the paramagnetic minerals. These low K lf petrologic units also show no systematic increase in K lf , suggesting that only minor differentiation is occurring in these units. The dataset derived from 2 surface sampling traverses are consistent with borehole core AMS data, showing a pattern of dominantly steep magnetic foliation and variably plunging magnetic lineation with a NW–SE trend.

Geochemical data and mapping from a Karoo flood basalt crater complex reveals new information about the ascent and eruption of magma batches during the earliest phases of flood basalt volcanism. Flood basalt eruptions at Sterkspruit, South Africa began with emplacement of thin lava flows before abruptly switching to explosive phreatomagmatic and magmatic activity that formed a nest of craters, spatter and tuff rings and cones that collectively comprise a crater complex >40 km2 filled by 9–18 km3 of volcaniclastic debris. Rising magma flux rates combined with reduced access of magma to external water led to effusion of thick Karoo flood basalts, burying the crater-complex beneath the >1.5 km-thick Lesotho lava pile. Geochemical data is consistent with flood basalt effusion from local dikes, and some lava flows likely shared or re-occupied vent sites active during explosive eruptions at Sterkspruit. Flood basalt magmas involved in Sterkspruit eruptions were chemically heterogenous. This study documents the rapid (perhaps simultaneous) eruption of three chemically distinct basaltic magmas which cannot be simply related to one another from one vent site within the Sterkspruit crater complex. Stratigraphic and map relationships indicate that eruption of the same three magma types took place from closely spaced vents over a short time during formation of the bulk of the crater-complex. Two magma types recognized there have not been recognized in the Karoo province before. The variable composition of flood basalts at Sterkspruit argues that magma batches in flood basalt fields may be small (0.5–1 km3) and not simply related to one another. This implies in turn that heterogeneities in the magma source region may be close to each other in time and space, and that eruptions of chemically distinct magmas may take place over short intervals of space and time without significant hybridisation in flood basalt fields.

Detailed field and geochemical investigations in the vicinity of the type section of the Tafelberg Formation of the early Cretaceous Etendeka Group, NW Namibia, have revealed the existence of a large eruptive vent in the lower part of the regional volcanic sequence produced by Strombolian and Vulcanian eruptions. The vent is filled with the thick, differentiated, ponded Kudu-Run olivine-rich basaltic flow, which has a distinctive low Zr/Y geochemical signature as well as a Tafelberg-type tabular basalt and the Nil Desperandum latite. Field evidence indicates that the Kudu-Run basalt and the latite were erupted from fissures located within the vent. Associated with the vent is an extensive pyroclastic apron extending from the vent edge and which is interbedded with the regional stratigraphy. Blocks of Precambrain basement lithologies occur within this deposit and indicate that the vent was excavated to a depth of at least 350 m below the palaeosurface at that time. The original Tafelberg Formation type section described by Erlank et al. [Erlank, A.J., Marsh, J.S., Duncan, A.R., Miller, R.McG., Hawkesworth, C.H., Betton, P.J., Rex, D.C. 1984. Geochemistry and petrogenesis of the Etendeka volcanic rocks from SWA/Namibia, 195–247. In: Erlank, A.J. (Ed.), Petrogenesis of Volcanic Rocks of the Karoo province. Special Publication of the Geological Society of South Africa, vol. 13, 395 p.] the Tafelberg Gully section, crosses from the lower part of the regional sequence into the intra-vent sequence and returns to the regional sequence higher up. In doing so it includes some of the localized intra-vent flows and excludes a number of flows which are part of the regional sequence in its lower part, thus rendering it inappropriate as a type section. A revised type section for the Tafelberg Formation is described by combining the upper part of the Tafelberg Gully section with a new section of 14 flows at the base of the regional sequence in the Tafelberg North (TBN) section some 2 km N of the Tafeleberg Gully. Distinctive flows in the TBN section can be mapped southwards where their precise stratigraphic relationship to the northward-thinning Springbok and Goboboseb Quarts Latite members of the Awahab Formation can be demonstrated. These stratigraphic relationships are entirely consistent with palaeomagnetic reversal stratigraphy and demonstrate that the same N–R–N polarity sequence occurs in the type sections of the Awahab and Tafelberg formations. Thus, the Awahab and Tafelberg magma systems were contemporaneous but the Tafelberg system outlived that of the Awahab. The Awahab system was built from eruptive centres located S of the Huab River whereas the Tafelberg vents were located further north.

Oxygen isotope compositions of Karoo and Etendeka large igneous province (LIP) picrites and picrite basalts are presented to constrain the effects of crustal contamination versus mantle source variation. Olivine and orthopyroxene phenocrysts from lavas and dykes (Mg# 64–80) from the Tuli and Mwenezi (Nuanetsi) regions of the ca 180 Ma Karoo LIP have δ18O values that range from 6.0 to 6.7 ‰. They appear to have crystallized from magmas having δ18O values about 1–1.5 ‰ higher than expected in an entirely mantle-derived magma. Olivines from picrite and picrite basalt dykes from the ca 135 Ma Etendeka LIP of Namibia and Karoo-age picrite dykes from Dronning Maud Land, Antarctica, do not have such elevated δ18O values. A range of δ18O values from 4.9 to 6.0 ‰, and good correlations between δ18O value and Sr, Nd and Pb isotope ratios for the Etendeka picrites are consistent with previously proposed models of crustal contamination. Explanations for the high δ18O values in Tuli/Mwenezi picrites are limited to (1) alteration, (2) crustal contamination, and (3) derivation from mantle with an abnormally high δ18O. Previously, a variety of models that range from crustal contamination to derivation from the ‘enriched’ mantle lithosphere have been suggested to explain high concentrations of incompatible elements such as K, and average eNd and eSr values of −8 and +16 in Mwenezi (Nuanetsi) picrites. However, the primitive character of the magmas (Mg# 73), combined with the lack of correlation between δ18O values and radiogenic isotopic compositions, MgO content, or Mg# is inconsistent with crustal contamination. Thus, an 18O-enriched mantle source having high incompatible trace element concentration and enriched radiogenic isotope composition is indicated. High δ18O values are accompanied by negative Nb and Ta anomalies, consistent with the involvement of the mantle lithosphere, whereas the high δ18O themselves are consistent with an eclogitic source. Magma δ18O values about 1 ‰ higher than expected for mantle-derived magma are also a feature of the Bushveld mafic and ultramafic magmas, and the possibility exists that a long-lived 18O-enriched mantle source has existed beneath southern Africa. A mixed eclogite peridotite source could have developed by emplacement of oceanic lithosphere into the cratonic keel during Archaean subduction.

Field and Scott Smith [Field, M., Scott Smith, B.H., 1999. Contrasting geology and near-surface emplacement of kimberlite pipes in southern Africa and Canada. Proc. 7th Int. Kimb. Conf. (Eds. Gurney et al.) 1, 214–237.] propose that kimberlite pipes can be grouped into three types or classes. Classical or Class 1 pipes are the only class with characteristic low temperature, diatreme-facies kimberlite in addition to hypabyssal- and crater-facies kimberlite. Class 2 and 3 pipes are characterized only by hypabyssal-and crater-facies kimberlite. In an increasing number of Class 1 pipes a new kimberlite facies, transitional-facies kimberlite, is being found. In most cases this facies forms a zone several metres wide at the interface between the hypabyssal- and diatreme-facies. The transitional-facies exhibits textural and mineralogical features, which are continuously gradational between the hypabyssal and the diatreme types. The textural gradations are from a coherent magmatic texture to one where the rock becomes increasingly magmaclastic and this is accompanied by concomitant mineralogical gradations involving the decline and eventual elimination of primary calcite at the expense of microlitic diopside. Both transitional- and diatreme-facies kimberlites are considered to have formed in situ from intruding hypabyssal kimberlite magma as a consequence of exsolution of initially CO 2 -rich volatiles from the volatile-rich kimberlite magma. The transitional-facies is initiated by volatile exsolution at depths of about 3 km below the original surface. With subsequent cracking through to the surface and resultant rapid decompression, the further catastrophic exsolution of volatiles and their expansion leads to the formation of the diatreme facies. Thus diatreme-facies kimberlite and Class 1 pipes are emplaced by essentially magmatic processes rather than by phreatomagmatism. Distinctly different petrographic features characterize crater-facies kimberlite in each of the three pipe classes. In crater-facies kimberlites of Class 1 pipes, small pelletal magmaclasts and abundant microlitic diopside are characteristic. These features appear to reflect the derivation of the crater-facies material from the underlying diatreme zone. Most Class 2 pipes have shallow craters and the crater-facies rocks are predominantly pyroclastic kimberlites with diagnostic amoeboid lapilli, which are sometimes welded and have vesicles as well as glass. Possible kimberlite lava also occurs at two Class 2 pipes in N Angola. The possible presence of lava as well as the features of the pyroclastic kimberlite is indicative of hot kimberlite magma being able to rise to levels close to the surface to form Class 2 pipes. Most Class 3 kimberlites have very steep craters and crater-facies rocks are predominantly resedimented volcaniclastic kimberlites, in some cases characterized by the presence of abundant angular magmaclasts, which are petrographically very similar to typical hypabyssal-facies kimberlite found in Class 1 pipes. The differences in crater-facies kimberlite of the three classes of pipe reflect different formation and depositional processes as well as differences in kimberlite composition, specifically volatile composition. Kimberlite forming pipe Classes 1 and 3 is thought to be relatively water-rich and is emplaced by processes involving magmatic exsolution of volatiles. The kimberlite magma forming Class 2 pipes is CO 2 -rich, can rise to shallow levels, and can initiate phreatomagmatic emplacement processes.

As a result of their relative concentration towards the respective Atlantic margins, the silicic eruptives of the Parana (Brazil)-Etendeka large igneous province are disproportionately abundant in the Etendeka of Namibia. The NW Etendeka silicic units, dated at similar to132 Ma, occupy the upper stratigraphic levels of the volcanic sequences, restricted to the coastal zone, and comprise three latites and five quartz latites (QL). The large-volume Fria QL is the only low-Ti type. Its trace element and isotopic signatures indicate massive crustal input. The remaining NW Etendeka silicic units are enigmatic high-Ti types, geochemically different from low-Ti types. They exhibit chemical affinities with the temporally overlapping Khumib high-Ti basalt (see Ewart et al. Part 1) and high crystallization temperatures (greater than or equal to980 to 1120degreesC) inferred from augite and pigeonite phenocrysts, both consistent with their evolution from a mafic source. Geochemically, the high-Ti units define three groups, thought genetically related. We test whether these represent independent liquid lines of descent from a common high-Ti mafic parent. Although the recognition of latites reduces the apparent silica gap, difficulty is encountered in fractional crystallization models by the large volumes of two QL units. Numerical modelling does, however, support large-scale open-system fractional crystallization, assimilation of silicic to basaltic materials, and magma mixing, but cannot entirely exclude partial melting processes within the temporally active extensional environment. The fractional crystallization and mixing signatures add to the complexity of these enigmatic and controversial silicic magmas. The existence, however, of temporally and spatially overlapping high-Ti basalts is, in our view, not coincidental and the high-Ti character of the silicic magmas ultimately reflects a mantle signature.

The bimodal NW Etendeka province is located at the continental end of the Tristan plume trace in coastal Namibia. It comprises a high-Ti (Khumib type) and three low-Ti basalt (Tafelberg, Kuidas and Esmeralda types) suites, with, at stratigraphically higher level, interstratified high-Ti latites (three units) and quartz latites (five units), and one low-Ti quartz latite. Khumib basalts are enriched in high field strength elements and light rare earth elements relative to low-Ti types and exhibit trace element affinities with Tristan da Cunha lavas. The unradiogenic Pb-206/Pb-204 ratios of Khumib basalts are distinctive, most plotting to the left of the 132 Ma Geochron, together with elevated Pb-207/Pb-204 ratios, and Sr-Nd isotopic compositions plotting in the lower Nd-143/Nd-144 part of mantle array (EM1-like). The low-Ti basalts have less coherent trace element patterns and variable, radiogenic initial Sr (similar to0.707-0.717) and Pb isotope compositions, implying crustal contamination. Four samples, however, have less radiogenic Pb and Sr that we suggest approximate their uncontaminated source. All basalt types, but particularly the low-Ti types, contain samples with trace element characteristics (e.g. Nb/Nb-*) suggesting metasediment input, considered source-related. Radiogenic isotope compositions of these samples require long-term isolation of the source in the mantle and depletions (relative to unmodified sediment) in certain elements (e.g. Cs, Pb, U), which are possibly subduction-related. A geodynamic model is proposed in which the emerging Tristan plume entrained subducted material in the Transition Zone region, and further entrained asthenosphere during plume head expansion. Mixing calculations suggest that the main features of the Etendeka basalt types can be explained without sub-continental lithospheric mantle input. Crustal contamination is evident in most low-Ti basalts, but is distinct from the incorporation of a metasedimentary source component at mantle depths.

A reliable 40 Ar/ 39 Ar plateau age of 180 ± 1.2 Ma (1σ) has been obtained for fresh basalt lava interbedded with aeolian Etjo Formation sandstones south-southwest of Grootfontein in northeast Namibia. This indicates that the Early Jurassic Karoo flood basalt sequence extended from Botswana into northeast Namibia at least as far as 18 east and that this may mark the eastern extent of the Early Cretaceous Etendeka Igneous Province. If so it would confirm the marked asymmetry of the Parana-Etendeka Flood Basalt Province relative to the Atlantic Rift as noted by others. This age also supports the correlation of the Etjo Formation of the Waterberg Plateau, Namibia with the Clarens Formation of the Karoo Sequence in South Africa, and the use of the name ‘Twyfelfontein Formation’ for the aeolianites in the Etendeka region and its incorporation into the Cretaceous Etendeka Group.

Concentrations of the platinum-group elements have been determined in several suites of southern African flood-type basalts and mid-ocean ridge basalt (MORB), covering some 3 Ga of geologic evolution and including the Etendeka, Karoo, Soutpansberg, Machadodorp, Hekpoort, Ventersdorp and Dominion magmas. The magmas cover a compositional range from 3.7 to 18.7% MgO, 26–720 ppm Ni, 16–250 ppm Cu, and

Most of the Large Igneous Provinces (LIPs) formed during the last 260 million years are associated with climatic changes, oceanic anoxia, or extinctions in marine and terrestrial environments. Current hypotheses involve (1) degassing of carbon from either oceans or shallow sea-bed reservoirs, (2) degassing from flood basalts, or from (3) sedimentary basins heavily intruded by LIP-related sills. These hypotheses are based on detailed geological and geochemical studies from LIPSs or relevant proxy data sequences. Here we present new data on gas generation and degassing from a LIP, based on the LA1/68 borehole north of the Ladybrand area in the Karoo Basin, South Africa. The borehole was drilled in the middle of a phreatic breccia pipe and penetrated 11 sills before reaching the basement at 1710 m depth. We present new data on the lowermost 15 m thick sill emplaced in shale, and on the breccia comprising the uppermost 154 m of the core. We show that (1) a reduction in organic matter within a contact aureole can be explained by heating and the formation of CH4, (2) a phreatic eruption and breccia formation was initiated from pore fluid boiling around sills emplaced in Beaufort Group sandstones at 420–570 m depth, (3) the phreatic eruption cut through a cover of solidified and partly molten lava flows that subsequently filled the crater, and (4) the pipe has been used as a fluid flow pathway for millions of years, demonstrated by fossil and active oil seeps. We conclude that the sub-volcanic LIP environment hold the key to understand the relationships between large scale volcanism and rapid environmental perturbations.

The Mount Ayliff Complex of the Eastern Cape province of South Africa is a layered intrusion of some 800 km2 surface area and up to 1,200 m thickness. On the basis of compositional similarities and spatial association, it is generally interpreted to form part of the Karoo igneous province. Similarities between the Mount Ayliff Complex and the staging chambers and feeder conduits to flood basalts that host magmatic sulfide ores elsewhere in the world suggest that the Mount Ayliff Complex may have an enhanced potential for Noril’sk-Talnakh–type massive Ni-Cu sulfide ores, an idea that is supported by the well-known sulfide occurrence at Waterfall Gorge. Here, we present major-, trace-, and noble-element data for 30 samples of cumulate rocks from a continuous 1,200 m drill core through the Insizwa lobe of the complex, as well as six samples of a footwall sill, the Taylor’s Koppie dike, that was considered a possible feeder zone to the lobe, and through a massive sulfide lens in the footwall of the lobe at Ndzongiseni. We show that most of the Insizwa cumulate rocks contain small amounts of cumulus sulfides. The sulfides, including those at Waterfall Gorge, can be explained by a model in which the sulfides were segregated from magma having chilled-margin composition. On the basis of the composition of the chilled margins, the Insizwa magma contained no entrained sulfides and was depleted in platinum group elements (PGE), although not in Cu and Ni, before emplacement. These compositional characteristics are also observed in lavas of the central Karoo igneous province and suggest only limited sulfide segregation from these magmas at depth. Our findings thus indicate relatively little potential for economic magmatic Ni-Cu-PGE sulfide deposits in the Insizwa lobe.

Detailed geochemical and field data for the volcanic sequence and intrusions of the Etendeka Igneous Province are used to construct a stratigraphic framework for petrogenetic interpretation of the evolution of the Etendeka-Parana continental flood volcanic event. Geochemical and petrographic characterization of over 1,000 analyzed samples allows 8 mafic and 17 silicic magma types to be recognized. Both silicic and mafic types can be grouped into high-Ti and low-Ti suites on the basis of elevated Ti relative to other elements. The mafic magmas are: Khumib (high-Ti), Tafelberg, Kuidas, Horingbaai, Huab, Tafelkop, Albin, and Esmeralda (all low-Ti). Amongst the silicic types, the Goboboseb, Springbok, Wereldsend, Grootberg, and Beacon low-Ti quartz latites, and the Nil Desperandum high-Ti latite have been described previously. In addition, the Hoas (low-Ti), Nadas, Sechomib, and Hoarusib, (all high-Ti) latites and the Fria (low-Ti), Sarusas, Ventura, Khoraseb, Naude, and Elliott (all high-Ti) quartz latites are described for the first time here. There is a marked provinciality in the distribution of the high- and low-Ti suites, with the former concentrated in the Northern Etendeka region and the latter dominant in the Southern Etendeka. Stratigraphic distribution of magma types allows two new formations to be defined in the Northern Etendeka – the Khumib Formation of basaltic flows and the Skeleton Coast Formation dominated by silicic sheets. The geochemical provinciality hampers precise correlations between Northern and Southern Etendeka. Available evidence suggests that the lower part of the Awahab Formation in the Southern Etendeka is coeval with the lower part of the Khumib Formation and that the silicic units in the upper part of the Tafelberg Formation probably correlate with the Skeleton Coast Formation. The paucity of Khumib dykes in relation to Tafelberg dykes and their field relationships with regard to the volcanic sequence in the Northern Etendeka suggests that the main Khumib eruptive centers lay further north, consistent with southward thinning of the Khumib basalts. In the Southern Etendeka, the Doros complex is the eruptive center of the plume-derived Tafelkop basalts, which probably built a shield volcano within the regional flood lava field. This work indicates that the Etendeka has a greater variety of both silicic and low-Ti mafic magma types than are known from the SE Parana. Detailed comparisons reveal that all the important silicic types in the Parana have geochemical equivalents in the Etendeka. New correlations are Santa Maria=Fria, Ourinhos=Khoraseb, Guarapuava=Sarusas, thus extending the previously recognized correlations between Southern Etendeka quartz latites and the Caixas do Sul and Anita Garibaldi "rhyolites." These correlations emphasize that very large volume silicic systems develop with pronounced lithospheric rifting in continental flood basalt provinces.

The Messum complex of NW Namibia, a part of the Parana– INTRODUCTION Etendeka volcanic province, consists of a dominantly felsic central The Etendeka Igneous province in the Damaraland core, surrounded by older gabbros. The igneous rocks of the core region of northwest Namibia (Fig. 1) contains numerous can be divided, in order of decreasing age, into (1) a sub-alkaline anorogenic igneous complexes of Mesozoic age (Martin suite, (2) an outer quartz syenite suite, and (3) an inner silicaet al., 1960; Piranjo, 1994), one of which is the Messum undersaturated suite dominated by nepheline syenite. Compositional complex. The range in chemical composition, both within differences within the quartz syenite suite can be explained by and between the various complexes, is remarkable and fractional crystallization, but Srand O-isotope data indicate that includes both silica-oversaturated and silica-underthese rocks contain a significant crustal component. The younger saturated mafic and felsic rocks, including carbonatite. central nepheline syenites trend towards the 0·1 GPa phonolite Messum was classified as a ‘mixed complex’ by Harris minimum and appear to have evolved by closed system fractional (1995), as it contains substantial amounts of both silicacrystallization of a mantle-derived magma, whereas earlier nepheline undersaturated and silica-oversaturated rock types. syenites appear to be contaminated with quartz syenite. The Dating of several Damaraland complexes yields ages decreasing effects of crustal contamination with time may reflect that lie between 137 and 124 Ma (Siedner & Miller, progressive armouring of the magmatic plumbing system by less 1968; Milner et al., 1995b; Renne et al., 1996). The contaminated material. Basanite dykes that intrude the plutonic alignment of the complexes along two parallel NE–SW rock of the core show trace element and Ndand Sr-isotope trends suggests that they represent a hotspot trace caused characteristics that are similar to those of basanites from Tristan by the motion of the African plate over a fixed mantle da Cunha. This suggests that the mantle-derived component of the plume (e.g. Rhodes, 1971). Because there is no systematic age progression along the lineament, this view has genMessum core rocks came from the Tristan plume. erally been discounted (Marsh, 1973). Recent ideas that large continental flood basalt (CFB) provinces, such as the Parana–Etendeka province, are manifestations of mantle plume activity (White & Mackenzie, 1989) has refocused attention on the relationship between plumes and linear arrays of complexes in CFB provinces. Milner

The Marinkas Quellen Carbonatite Complex, in southern Namibia, forms part of the ca. 490–550 Ma Kuboos-Bremen Igneous Province, a zone of alkaline igneous rocks that intrude the Mesoproterzoic to Paleoproterozoic Namaqualand Metamorphic Province and Neoproterozoic platform sediments. The carbonatite complex includes Ca-rich, Mg-rich and Fe-rich carbonatites as well as late ferrocarbonatite dykes that show extreme enrichments in Mn, REE and Th. Compared to most carbonatites, those at Marinkas Quellen are generally depleted in most trace elements, particularly Ba and the LREE. These depletions are accompanied by what are possibly the highest ϵNd values (>+5.3) observed for carbonatites from a continental setting, and combined with low initial 87 Sr / 86 Sr values (

A volcanic event of immense scale occurred within a relatively short period in early Jurassic time over large regions of the contiguous Gondwana supercontinent. In southern Africa, associated remnants of thick volcanic successions of lava flows and extensive dike and sill complexes of similar composition have been grouped together as the Karoo Igneous Province. Correlative volcanic and plutonic rocks occur in Antarctica and Australia as the Ferrar Province. Thirty-two new 40Ar-39Ar incremental heating experiments on feldspars and whole rocks from Namibia, South Africa and East Antarctica produce highly resolved ages with a vast majority at 183±1 Ma and a total range of 184 to 179 Ma. These are indistinguishable from recent, high-resolution 40Ar-39Ar and U-Pb age determinations reported from the Antarctic portion of the province. Initial Karoo volcanism (Lesotho-type compositions) occurred across the entire South African craton. The ubiquitous distribution of a plexus of generally nonoriented feeder dikes and sills intruding Precambrian crystalline rocks and Phanerozoic sediments indicates that these magmas penetrated the craton over a broad region. Lithosphere thinning of the continent followed the main pulse of igneous activity, with volcanism focused in the Lebombo-Nuanetsi region, near the eventual split between Africa and Antarctica. Seafloor spreading and dispersion of east and west Gondwana followed some 10–20 m.y. afterward. The volume of the combined Karoo-Ferrar province (∼2.5×106 km3) makes it one of the largest continental flood basalt events. The timing of this event correlates with a moderate mass extinction (Toarcian-Aalenian), affecting largely marine invertebrates. This extinction event was not as severe as those recorded at the Permian-Triassic or Cretaceous-Tertiary boundaries associated with the Siberian and Deccan flood basalts events, respectively. The difference may be due to the high southerly latitude and somewhat lower eruption rates of the Karoo event.

The diamondiferous Letlhakane kimberlites are intruded into the Proterozoic Magondi Belt of Botswana. Given the general correlation of diamondiferous kimberlites with Archaean cratons, the apparent tectonic setting of these kimberlites is somewhat anomalous. Xenoliths in kimberlite diatremes provide a window into the underlying crust and upper mantle and, with the aid of detailed petrological and geochemical study, can help unravel problems of tectonic setting. To provide relevant data on the deep mantle under eastern Botswana we have studied peridotite xenoliths from the Letlhakane kimberlites. The mantle-derived xenolith suite at Letlhakane includes peridotites, pyroxenites, eclogites, megacrysts, MARID and glimmerite xenoliths. Peridotite xenoliths are represented by garnet-bearing harzburgites and lherzolites as well as spinel-bearing lherzolite xenoliths. Most peridotites are coarse, but some are intensely deformed. Both garnet harzburgites and garnet lherzolites are in many cases variably metasomatised and show the introduction of metasomatic phlogopite, clinopyroxene and ilmenite. The petrography and mineral chemistry of these xenoliths are comparable to that of peridotite xenoliths from the Kaapvaal craton. Calculated temperature-depth relations show a well-developed correlation between the textures of xenoliths and P-T conditions, with the highest temperatures and pressures calculated for the deformed xenoliths. This is comparable to xenoliths from the Kaapvaal craton. However, the P-T gap evident between low-T coarse peridotites and high-T deformed peridotites from the Kaapvaal craton is not seen in the Letlhakane xenoliths. The P-T data indicate the presence of lithospheric mantle beneath Letlhakane, which is at least 150 km thick and which had a 40mW/m2 continental geotherm at the time of pipe emplacement. The peridotite xenoliths were in internal Nd isotopic equilibrium at the time of pipe emplacement but a lherzolite xenolith with a relatively low calculated temperature of equilibration shows evidence for remnant isotopic disequilibrium. Both harzburgite and lherzolite xenoliths bear trace element and isotopic signatures of variously enriched mantle (low Sm/Nd, high Rb/Sr), stabilised in subcontinental lithosphere since the Archaean. It is therefore apparent that the Letlhakane kimberlites are underlain by old, cold and very thick lithosphere, probably related to the Zimbabwe craton. The eastern extremity of the Proterozoic Magondi Belt into which the kimberlites intrude is interpreted as a superficial feature not rooted in the mantle.

The Lesotho remnant contains the type succession for Karoo low-Ti basalts of central southern Africa. The 40 Ar/ 39 Ar dating indicates that the sequence was emplaced within a very short period at about 180 Ma and consists of a monotonous pile of compound basalt lava flows which lacks significant palaeosols and persistent sedimentary intercalations. We have used geochemistry to establish a stratigraphic subdivision of the lava pile. Thin units of basalt flows, the Moshesh's Ford, Golden Gate, Sani, Roma, Letele, and Wonderkop units, with diverse geochemical character and restricted geographical distribution, are present at the base of the succession. These are overlain by extensive units of compositionally more uniform basalt, the Mafika Lisiu, Maloti, Senqu and Mothae units, which build the bulk of the sequence. A single palaeomagnetic polarity reversal occurs within the lower third of the basalt succession and is consistently located within the Mafika Lisiu unit. This and the persistent and relatively uniform thickness of the stratigraphic units suggest that the pile was constructed in a uniform manner by eruption of basalt onto a generally planar surface from a widespread plexus of dykes. The stratigraphic sequence in Lesotho closely resembles that in the thinner sequence of low-Ti basalts of the Springbok Flats remnant, some 400 km to the north. A thin unit of high-Ti basalt within the upper part of the Springbok Flats sequence can be correlated with the thick high-Ti basalt suite along the rift-related Lebombo structure on the eastern margin of the Karoo province. This is the first established correlation between these two important outcrops of Karoo volcanic rocks and demonstrates that the low-Ti basalts of Lesotho and the cratonic interior are the approximate time equivalents of the lower part of the Lebombo sequence. This conclusion has important implications for models for the origin of the Karoo flood basalt province.

Analyses of samples from a weathering profile on Karoo dolerite allow elements to be divided into three groups depending on their behaviour. Si, K, Na, Mg, Ca, Sr, Ba and V are mobilized and removed from weathered products. Fe, Al. Ti, Zr, Hf, Zn, Cu, Sc, Co and Ni are immobile. REE, Y, and to a lesser extent Cr, are mobile and redistributed within the profile without a net loss of these elements from the profile. Large positive Ce anomalies are developed in oxidized weathered products by preferential leaching of the other REE's. Negative Ce anomalies and REE enrichment is a feature of less altered dolerite.

Phonolites of Tertiary age occur as eroded tholoids, lava flows, ignimbrites, and coulees in the Klinghardt Mountains of southern Namibia. Sixty samples have been analyzed for major and trace elements and fourteen of these for 87Sr86Sr. The phonolites lie close to the low-pressure cotectics in Q-Ne-Ks, in keeping with their petrography which indicates that most samples have phenocrysts of both nepheline and sanidine. Na has been variably lost from the rocks during crystallization and devitrification/alteration of hypocrystalline specimens. The concentrations of other elements except perhaps LREE remain unaffected. The composition variation within the phonolites is consistent with fractional crystallization of nepheline, sanidine, and minor clinopyroxene to produce a number of near-parallel lineages from similar, but not identical, parent compositions. Fractional crystallization results in increasing SiO2, Na2O, Zr, Nb, Ce, Rb, Th, Y, Zn, Pb, and REE, but decreasing K2O, TiO2, CaO, MgO, P2O5, Ba, and Sr with differentiation. Enrichment factors for Zr and Nb indicate that the most evolved phonolites represent < 20% residual liquids, a result which is consistent with major element modelling. In the evolution of the most differentiated phonolites fractional crystallization was accompanied or followed by volatile transfer causing depletion of Th, Y, LREE, Zn, and Pb relative to Zr and Nb. Initial 87Sr86Sr for most of the phonolites lie in the range 0.7050–0.7057, rising to 0.7080 in the low-Sr samples. This is suggestive of some interaction with crustal rocks, but this interaction has no detectable influence on the fractionation controlled compositional evolution of the suite. The phonolites are associated on a regional scale with nephelinite and melilitite rocks but there is no favourable evidence for an evolutionary link between these and phonolite. Alternatively, the primitive phonolites may have originated through anatexis of metasomatized, nepheline-normative crustal rocks.

Ratios between elements Mg, Fe, Co, Cr, Ni, V, and Sc are consistently different in mafic rocks of the upper critical zone, and those above the Bastard unit. Within the 300 m section above the Merensky Reef, 87Sr/86Sr ratios increase from c.0.7063 to c.0.7087, irrespective of rock type. Decoupling of Mg/(Mg + Fe2+) ratios and the Ca contents of plagioclase, and wide variations in the proportions of anorthosite within the Bastard, Merensky, and Merensky Footwall units, are inconsistent with anorthosite formation by simple fractional crystallization of magma batches of limited volume. Conversely, significant differences in Sr-isotope ratios show that these anorthosites could not have shared a common parental liquid. These data are used to develop a model whereby (a) the 300 m column above the critical zone represents the mixing of liquids of isotopically and geochemically discrete upper critical and main zone lineages, (b) mafic layers of the Bastard, Merensky, and Merensky Footwall units crystallized from discrete injections of primitive, mafic liquid while (c) the leucocratic upper parts of these units crystallized during progressive hybridization of liquid residua, which remained after significant separation of mafic phases, with a supernatant column representing the liquid residua of earlier cycles, and (d) the buoyancy of plagioclase, and enlargement of the primary phase volume of plagioclase consequent upon an increase in An/Ab ratio of hybrid liquids, were significant factors in the generation of anorthositic layers.