Experiments have been performed in the multicomponent (natural) bulk system to constrain the conditions of generation and differentiation of a K-rich group II kimberlite (now also referred to as orangeite). The group II composition examined was saturated in olivine, orthopyroxene, and garnet at near liquidus conditions in the pressure range 4 to 10 GPa. In the range 2 to 3 GPa, the liquidus phase was olivine only. The potassic nature of the melts in the bulk compositions studied was ensured by the absence of any K-bearing phase in the residual assemblage at P > 4 GPa. Phlogopite is destabilized toward higher pressures by a carbonation reaction of the type phlogopite + CO2 = enstatite + garnet + K2CO3 (liquid) + H2O leading to alkalic, carbonatitic liquids coexisting with a garnet–peridotite (harzburgite or lherzolite) residue over a wide pressure–temperature space at pressures in excess of 4 GPa. Evidently, CO2-bearing systems do not favor the stability of phlogopite and/or K-richterite amphibole at pressures in excess of 4 to 5 GPa, and it is suggested that the carbonate-bearing and potassic character of any mantle melt originating from this depth is most likely the product of a two-stage process: either a carbonate-bearing protolith is invaded by a potassic melt or fluid (probably supercritical), or a potassic protolith (after metasomatism) has been invaded by a carbonatite melt.

Diamond exploration focuses on geochemical analysis of indicator minerals that are more abundant than diamond itself. Among such indicators, low-Cr (Cr2O3 < 1 wt%) garnets from mantle eclogites are problematic since they overlap compositionally with many lower-crust-derived garnets also transported by kimberlite. Misclassification of these garnets may create “false positive” mantle signatures and possible misdirection of exploration efforts. Statistical solutions using major elements in low-Cr garnet (Hardman et al. in J Geochem Explor 186:24–35, 2018) provide improved error rates for the discrimination of low-Cr crustal and mantle garnets recovered from kimberlite. In this study we analysed a large suite of garnets (n = 571) from both crustal and mantle settings, already characterised for major elements, for a wide range of trace elements by laser ablation inductively-coupled plasma mass spectrometry and use these new data along with literature data (n = 169) to evaluate the effectiveness of adding trace elements to garnet-based diamond exploration programs. A new garnet classification scheme, initially using a major-element based filter, uses garnet Sr contents and Eu anomalies to help identify low-Cr garnets that are misclassified using major element methods. Combined with existing methods, our new trace element classifiers offer improvement in classification error rates for low-Cr, crustal and mantle garnets to as low as 4.7% for calibration data.

Aims In recent years there has been an increasing recognition of the importance of non-technical skills training in medical education. Strong communication and interpersonal skills are crucial to good clinical practice. These are of particular relavence when communicating with team members, parents and patients in challenging situations. We identified a lack of postgraduate training opportunities for paediatric trainees to develop these skills. Our aim was to devise an interdisciplinary training opportunity which would enable senior paediatric trainees to develop their communciation skills by undertaking simualted scenarios based on challenging clinical situations. Methods Level 3 trainees in our deanery were invited to attend the ACTup course. We recruited a diverse faculty consisting of paediatricians, paediatric nurses, social workers and psychologists. In a unique collaboration, we worked closely with drama students from our local universty. The drama students acted in the role of parents for the simulations. The scenarios focused on challenging clinical encounters which may be seen in paediatric practice. These included safeguarding cases, sudden unexpected death of an infant and discussions around withdrawal of care. Feedback was collected through questionnaires and focus groups conducted by the psychology team. Results Feedback on the course was excellent. All participants found input from a diverse faculty particularly beneficial. Themes identified on analysis of the focus groups included the importance of the drama students in enhancing the fidelity of the simulations, the necessity of further non-technical skills training in paediatric training and the benefits of debrief following challenging clinical encounters. Conclusions This course provided an opportunity for senior paediatric trainees to develop their communication skills in challenging situations. Initial feedback was excellent. We strive to establish this course as a regular training opportunity for paediatric trainees in our deanery. The course has also been adopted as an essential component of the undergraduate drama module entitled ‘Drama, Health and Social Care’ now offered in our local university.

In diamond exploration, the accurate distinction between garnets from the crust or mantle, or from those having a cognate origin with kimberlite (low-Cr megacrysts), is important for the assessment of indicator mineral samples; misclassifications potentially result in costly misdirection of exploration efforts. Existing literature databases and graphical classification schemes for garnets suffer from a paucity of craton-derived, lower-crustal garnets that - as shown here - are among the most difficult to distinguish from garnets of mantle origin. To improve this situation, a large database of new and literature garnet major element analyses has been compiled. Using this dataset, it is shown that the conventionally used Mg# (Mg/(Mg + Fe)) vs. Ca# (Ca/(Mg + Ca)) plot (Schulze, 2003) for discrimination of crust and mantle garnets results in significant overlap (39.2% crustal failure rate using our dataset). We propose a new graphical classification scheme that uses the parameters In(Ti/Si) and In(Mg/Fe) to discriminate low-Cr garnets of crust origin from those of a mantle eclogite-pyroxenite origin with an error rate of 10.1 +/- 2.1% at the 95% confidence level (assessed via K-fold cross-validation with ten random test datasets), significantly lower than existing methods. Multivariate statistical solutions, which incorporate a wide selection of geochemical variables, represent additional possibilities for discrimination. Using our new database, logistic regression (LR) and linear discriminant analysis (LDA) approaches are evaluated and new crust-mantle garnet discrimination schemes derived. The resulting solutions, using a wide variety of cations in garnet, provide lower misclassification rates than existing literature schemes. Both LR and LDA are successful discrimination techniques with error rates for the discrimination of crust from mantle eclogite-pyroxenite of 7.5 +/- 1.9% and 8.2 +/- 2.3%, respectively. LR, however, involves fewer stipulations about the distribution of training data (i.e., it is more "robust") and can return an estimate for probability of classification certainty for single garnets. New data from diamond exploration programs can be readily classified using these new graphical and statistical methods. As the discrimination of low-Cr megacrysts from mantle eclogite-pyroxenite is not the focus of this study, we recommend the method of Schulze (2003) or Grutter et al. (2004) for low-Cr megacryst discrimination to identify megacrysts in the "mantle" suite. Runstreams for our LDA and LR approaches using the freeware "R" are provided for quick implementation.

This paper reports on the identification of peperite, a breccia comprising angular to lobate variably vesiculated basalt clasts in a mudstone matrix, at the upper margin of a basalt sheet of the Sable River Basalt Formation, northern Lebombo. The peperite formed as a result of emplacement of the basalt sheet as a shallow sill that burrowed into and migrated laterally within an unconsolidated water-saturated sediment pile. Although breccias at the upper margin of basalt units are widely recognized from the Karoo of southern Africa, they have generally been interpreted as blocky 'autobreccia' or 'flow-top breccia' on lavas. Consequently, many mixed basalt-sediment breccias may have been viewed as sedimentary infill of these 'flow-top breccias'. To the best of our knowledge, this paper represents the first documentation of an intrusive or invasive mode of formation. The new shallow sill or invasive flow interpretation has significant implications on the perceived sedimentary architecture and basalt emplacement model for the Karoo, and perhaps other flood basalt provinces. It is suggested that the widespread distribution and consistent thickness of individual 'flows' in the Sable River Basalt Formation can sometimes be attributed to emplacement as shallow sills that have propagated through a regionally extensive sediment pile.

This paper presents some of the microanalytical problems and challenges in minerals research and the geosciences in general. It deals with the evolving possibilities for advanced analysis in the geological end mineralogical field, with emphasis on ion-beam based techniques. The geosciences have a large array of analytical techniques that they rely on, but there are still a number of areas where problems persist. The different analytical techniques have also been more complementary than competitive up to now, but this is set to change as new techniques evolve and are optimized. The strengths and weaknesses of ion-beam techniques are highlighted, with emphasis on minerals characterization.

A ΔE–E telescope was used in coincidence to analyse diamond samples by recoil of hydrogen (ERDA) using 4.1 MeV He+. Comparison of the results from silicates with known H-abundances suggested sensitivities of <50 atomic ppm. A slice of type Ib diamond was prepared and implanted with known amounts of hydrogen (1000, 150, 40 and 10 atomic ppm). The analysis of this manufactured standard by ERDA showed that the 1000 and 150 ppm implants were resolvable. The presence of intrinsic hydrogen in the diamond prevented resolution of the lower H implants. A number of diamonds from southern Africa were analysed for their H-content using ERDA and Fourier transform infrared (FTIR) techniques. There was no correlation between the results of both techniques. It was concluded that much of the hydrogen in diamond (typically 100–4000 atomic ppm in the specimens analysed) was not infrared active (i.e. not bonded to C or N in the diamond lattice).

Eight garnet-peridotites (5 harzburgites and 3 lherzolites) and two garnet-bearing ol-orthopyroxenites from Bultfontein kimberlite dumps (Kimberley pipes, South Africa) were studied. They present protogranular textures with well-equilibrated primary mineralogical assemblage, consisting of large olivine (3-5 mm), orthopyroxene (3-5 mm), garnet (2-6 mm), phlogopite (up to 1 mm), and smaller clinopyroxene (2-3 mm), and spinel (up to 0.5 mm) crystals. Three types of secondary textures, superimposed on the primary paragenesis, were distinguished on the basis of geometric relationships and relative proportions of primary and secondary phases. Type A is characterised by the development of reaction coronas around garnets, which beside a dark keliphitic rim, consist of small subeuhedral crystals of phlogopite (up to 0.05 mm), spinel (100-400 mm) and clinopyroxene (0.05- 0.1 mm). Type B is characterised by reaction areas around orthopyroxene. The secondary assemblage is constituted by small grains of olivine (200-300 mm), clinopyroxene (200- 400 mm) and phlogopite (up to 0.05 mm) which usually include subeuhedral spinels. Type C is characterised by veins and patches filled with secondary olivine (200-300 mm), clinopyroxene (200-300 mm), phlogopite (400-600 mm) and spinel (50-100 mm). Small amounts of carbonates (80-100 mm) and apatite (90- 100 mm) set in serpentinitic matrix may also be found. Primary olivine, orthopyroxene and garnet show consistently lower Mg# values (Mg/(Mg+Fe)*100) in lherzolites (ol, 90.7-91.3; opx, 91.5-92.3; gt, 79.9-81.1) with respect to harzburgites (ol, 91.5-92.8; opx, 92.3-93.7; gt, 80.6-83.9). On the other hand, in contrast with what usually observed in spinel-bearing metasomatized peridotites, secondary phases record systematically lower mg values than the primary phases. Chondrite-normalized REE patterns for garnets are typically “humped” with higher values in MREE (Gd and Eu) and (La/Yb)N ratios varying between 0.010-0.017. They are remarkably homogeneous in composition from core to rim and display a strong negative Ti anomaly. Clinopyroxenes show the widest range of major element compositions, clearly related to the various textural positions. Primary clinopyroxenes are very homogeneous in composition with moderate Al2O3 (2.38-2.87 wt%), and high Na2O (up to 3 wt%) contents. With respect to the primary clinopyroxene, secondary clinopyroxenes in type B textures (associated with orthopyroxene) are depleted in Al2O3 and Na2O and enriched in TiO2 contents, while secondary clinopyroxenes in type A textures (related to garnet) are enriched in Al2O3 (6.07-9.07 wt%) and FeO and depleted in SiO2 and Na2O. As a result, clinopyroxenes related to orthopyroxene are easily distinguished from clinopyroxenes related to garnet on the basis of their quite different SiO2 and Al2O3 contents. As far as trace element are concerned, in chondrite-normalized diagram, primary clinopyroxenes show a remarkable Ti and Zr negative anomalies (Ti/((Eu+Gd)/2)), Ti*=0.01-0.06; Zr/((Nd+Sm)/2), Zr*=0.04-0.05) and enriched (La/Yb)N ratios (42.67-45.62), which would speak in favour of equilibrium condition with garnet. On the other hand, both secondary clinopyroxenes display similar patterns, which are characterised by i) a decrease in the La/Yb ratios ((La/Yb)N=5.08-27.4), caused by a HREE enrichment, ii) a less pronounced Ti negative anomaly, iii) the appearance of a slight Sr negative anomaly, and iv) the disappearance of Zr negative anomaly. Particularly Ti content results higher than every primary Ti-bearing phases (including phlogopite). In accordance with what already observed for the other primary phases, phlogopite in lherzolites tends to have lower Mg# values with respect to that in harzburgites, although with a wide overlap probably due to the difficulties in calculating the Fe3+/Fe2+ ratios in hydrous minerals. Secondary phlogopite is enriched in TiO2 and FeO and depleted in MgO. In chondrite-normalized patterns primary phlogopite are characterised by higher amount of Ba, Sr and Nb and lower Ti contents with respect to the secondary crystals. So far, no clear geochemical markers have been found for discriminating between phlogopite in type A, B or C textures. Spinels display a wide range of both Cr# (Cr/(Cr+Al)*100) and Mg# values irrespective from textural positions (Cr#=32.5-87.4; Mg#=41.3-64.3). Only in pyroxenites they are characterised by a substantially lower Mg# values (9.16-16.37). Carbonates are mainly constituted by calcite, but rare crystals of dolomite were also found within a serpentinitic veins. Apatite present a fairly homogeneous composition, with a highly fractionated chondrite-normalized REE patterns ((La/Yb)N ~120) and a Sr content up to 17,500 ppm. Amphibole is very rare. It has been observed only in type B reaction zones (in orthopyroxenite). It has pargasitic composition. Several chemical balances, using both major and trace elements, were put forward in order to constrain the process which caused the secondary paragenesis observed in type A, B and C textures. Trace element contents in secondary clinopyroxenes point toward a link with garnet destabilization, as testified by the HREE enrichment. The contribution of orthopyroxene in producing secondary clinopyroxene is also highly supported by the high SiO2 content (type B textures). Primary phlogopite does not participate to any reaction as evidenced by the petrographic characteristics, thus the remarkable Ti enrichment observed both in secondary clinopyroxene and phlogopite remain unbalanced. Also apatite crystallization seems affect secondary clinopyroxene formation as suggested by its lower LREE and Sr contents. In conclusion these preliminary data may indicate an enrichment event which affected the garnet peridotite. Corona around garnet will not (at least only) result by a simple decrease in pressure conditions, which may have caused the crossing of the garnet/spinel stability field. Garnet (and orthopyroxene) may have reacted with a metasomatic fluid/s (enriched in Ti) giving rise to secondary clinopyroxene, phlogopite, apatite and, possibly, carbonate minerals.

The precise timing of metasomatic events in the Kaapvaal craton lithospheric mantle is poorly constrained and therefore it has proven difficult to relate mantle metasomatism in the source to igneous activity at the surface. A suite of mica–amphibole–rutile–ilmenite–diopside-bearing xenoliths (MARID) represents the product of such metasomatic fluids or melts. In this paper we report results of a SHRIMP and cathodoluminescence study on zircons from four MARID samples from the Kimberley cluster of pipes. Textures combined with experimental evidence suggest that zircons are late crystallizing phases in MARIDs that form close to the solidus where the major precipitating phases are phlogopite and K-richterite. Cathodoluminescence patterns reveal a complex growth history of the zircons involving strong deformation, repeated events of crystallization, resorption and/or modification of the zircon composition, possibly over several millions of years. Many of the U–Pb ages are significantly older than the intrusion age of the group I kimberlites of the Kimberley cluster. Zircon ages in a MARID xenolith from the Wesselton kimberlite vary in the range 113±3 Ma to 142±3 Ma, compared with a pipe age of 84±3 Ma. In two samples from the De Beers, Bultfontein or Dutoitspan pipes, ages range from 85±5 to 119±2 Ma in one and 80±4 to 90±2 Ma in the other. The ages obtained for the latter are indistinguishable from the pipe age of kimberlites belonging to the younger group I kimberlite event (ca. 80–95 Ma). Zircons in this sample, as well as the younger ages in the other, may reflect real ages but most likely have lost radiogenic Pb as a result of deformation and fluid or melt interaction. The presence of pre-group I ages in two of the samples demonstrates unambiguously that the metasomatism that yielded the older crystallization ages cannot be related to the group I kimberlite event. Instead, minimum ages for the MARID zircons coincide with the period of group II kimberlite magmatism in the central to southeastern Kaapvaal craton. This coincidence in ages would be consistent with the contention based on experimental evidence, that MARID-type rocks can be derived from group II precursor magmas by olivine fractionation and exsolution of a carbonatitic component. An alternative possibility, not excluded by the zircon minimum ages, is a genetic relation between MARID metasomatism and Karoo magmatism.

A ΔE-E telescope was used in coincidence to separate the isotopes of hydrogen and to improve the sensitivity of hydrogen detection when recoiled with 3.8 MeV He+. This was achieved by recording coincident events in both detectors (corresponding to the recoil of H, D or T from the target) thereby excluding any contribution from D (or T) beam contamination to the ERDA spectrum for hydrogen and also eliminating electronic noise from the background. The analysis of standard hydrogen-bearing silicates (6–189 wt ppm H) suggested that the concentrations at the lower part of this range were resolvable. The lowest counts were accumulated on two fragments of colourless diamond and were <20% of the counts obtained for the standard bearing 6 wt ppm H. We suggest sensitivities of better than 0.0056 wt% H2O (or 75 at. ppm H in carbon-rich matrices) are achievable with relative precisions <20% in the range 0–400 at. ppm H.

The use of elastic recoil detection analysis (ERDA) can measure bulk H in silicates at trace concentrations. This technique is non-destructive and uses a 4He beam which may be routinely focused to dimensions of 5 × 5 μm. The technique is suitable for the analysis of both materials of appreciable H content (e.g. silicate glasses) and materials with trace H content (e.g. olivines). H contents are determined by primary calibrations (using standards) and from first principles (spectra simulations). The calibration was conducted in the range 0–2 wt% H2O using a set of mineral standards of known H2O concentration with sensitivities of 0.04 wt% H2O achieved. There is a good correlation between H2O contents derived by spectra simulations and concentrations derived by calibration. Concentrations determined by both methods were affected by H-loss due to diffusion and release away from the beam spot. The magnitude of this loss is quantifiable.

Hydrogen (reported as H2O) affects the physical properties of silicate materials profoundly. Silicates (with Mg and Fe) dominate a significant portion of the earth's crust and mantle (0–2900 km depth). The melting point of the silicate assemblage in the upper mantle (olivine dominated) is decreased by up to 400°C at 3 GPa with the addition of 0.14 wt.% H2O. The strength of olivine is also reduced by two orders of magnitude with the incorporation of small amounts of water (e.g., 0.001 wt.%). H occurs in the earth's upper mantle (30–670 km depth) in trace proportions (0.05–0.2 wt.% H2O equivalent). Over the depth range of the upper mantle there is a two orders of magnitude increase with depth in the ability of the dominant mineral (olivine) to accommodate hydrogen. This suggests that silicates derived from deeper (minerals or partial melts of those minerals) may be more hydrous. There is some speculation that Mg-perovskite (stable at 670–2900 km depths) may also accommodate significant hydrogen.

A nuclear microbeam technique called elastic recoil detection analysis (ERDA) or forward recoil spectroscopy which is capable of yielding bulk H in silicates at ppm sensitivities is described. This technique is nondestructive and uses a 4He+ beam which may be routinely focused to dimensions less than 5 × 5 μm. The technique is suitable for the analysis of both materials of appreciable H content (e.g., amphiboles) and materials with trace H content. The technique is calibrated in the range 0–2 wt% H2O using a set of mineral standards of known H2O concentration with sensitivities of 0.04 wt% H2O achieved. There is a good correlation between H2O contents derived by spectra simulations and concentrations derived by empirical calibration, although the former yield data 10–20% lower when compared to known values. The equilibration of olivine with a potassic silicate melt at high pressures (1.5 to 10 GPa) in experiments shows more H is accommodated in the mineral with increasing pressure. The olivine-melt KH2O (expressing H as wt% H2O concentration in mineral/concentration in melt) at 1.5GPa (1400°C) was ca. 0.04 ± 0.015. At 5.8–6 GPa (1740°C), olivine-melt KH2O increased to 0.13 ± 0.03. A single experiment at 10 GPa (ca. 2000°C) yielded a minimum KH2O of 0.12. The amount of H which minerals accommodate is also highly correlated with bulk system composition (which controlled melt composition). In alkali-absent bulk systems, the KH2O for olivine equilibrated with a MgSi melt was 0.30 at 1 GPa (1400°C), an order of magnitude increase over the alkali-bearing system at this pressure. This reflects the reduced facility of a wholly MgSi melt to accommodate H2O relative to an alkali-bearing melt. The increase in KH2O with pressure for olivine-melt, combined with data for KH2O of natural olivine (and orthopyroxene) in basaltic glass at P < 0.3 GPa (<0.005), implies that a deep residual mantle would be more H-rich than the shallow mantle for the same degree of melt extraction.

Experiments were performed from 1.0 to 8.5 GPa in a peralkaline system K2O–Na2O–CaO–MgO–Al2O3–SiO2–H2O (KNCMASH) to investigate the stability and composition of richteritic amphiboles in the MARID (mica–amphibole–rutile–ilmenite–diopside) assemblage amphibole + phlogopite + clinopyroxene. The results were compared with phase relations and the composition of natural MARIDs to assess possible mechanisms of formation for MARID-type rocks. K-richterite is stable in a wide range of bulk K/Na ratios in the MARID assemblage to 8.5 GPa and 1300°C. In this assemblage the amphibole can accommodate significant amounts of K on the M(4) site and shows a systematic increase in the K/Na ratio with increasing pressure. At P >7.0 GPa, K-richterite can coexist with garnet. Phase relations of K-richterite in a natural MARID composition are consistent with those in the simplified system and confirm the potential stability of K-richterite and K-richterite + garnet within the diamond stability field. The assemblage K-richterite + phlogopite + clinopyroxene is incompletely buffered in the KNCMASH system, resulting in a systematic relation between bulk- and mineral compositions observed in the experiments. Such a correlation, however, cannot be observed in natural MARIDs. Therefore, MARID-type rocks do not represent the bulk composition from which they formed and, hence, must be products of an open-system crystallization.

Analysis of silicate minerals equilibrated at high pressure with silicate- (clinopyroxene) and carbonatite-melts (olivine, orthopyroxene, clinopyroxene, garnet, phlogopite) using proton- and electron-microprobe techniques, enabled the measurement of single mineral/melt partition coefficients for elements in minor and trace abundances. For carbonatite melt—olivine (18 kb), DTi (<0.05), DNb (0.034), DZr (0.036), DTa (<0.06), DY (<0.03), DSr (0.02); carbonatite melt-orthopyroxene (46 kb), DTi (0.13) DNb (<0.01), DZr (0.17), DTa (<0.01), DY (0.18), DLu (0.52), DSr (0.053); carbonatite melt-garnet (34 and 46 kb), DTi (1.2−0.99), DNb (0.086−0.014), DZr (1.42−0.83), DTa (0.14−.051), DY (1.91−1.53), DLu (5.3), DSr (0.062−0.008) ; carbonatite melt-clinopyroxene (46 kb), DTi (0.12), DNb (0.22), DZr (0.29), DTa (0.22), DY (0.27), DLu (0.33), DSr (0.26), and carbonatite melt-phlogopite (18 kb), DTi (2.1) DNb (0.09), DZr (0.05), DTa (0.21), DY (<0.04), DSr (0.056), and DRb (4.0). The partition coefficients are consistent with substitutions involving the main stoichiometric cations Mg, Ca, Fe, and K (phlogopite) in minerals. The effect of different D's and the trace element content of metasomatizing carbonatite and silicate melts will dictate the extent to which trace element ratios may be fractionated differently in residual mantle mineralogies by these two styles of metasomatism. The effect of partitioning differences is quantified using a model assuming equivalent abundances of incompatible minor and trace elements in silicate and carbonatite melts. This model shows that Ti/Na, Ti/Y, Ti/Lu, Ti/Nb, Ti/Ta, and Ti/Sr ratios would be lower in a peridotite affected by carbonatite metasomatism and therefore, in silicate melts derived from such a mantle, relative to a source affected by silicate metasomatism. From natural data (peridotite and carbonatite), element abundances in the metasomatizing carbonatite will also lower these ratios in residual minerals. In addition, the natural data show that carbonatite metasomatism would also increase LREE/HREE, LREE/Hf, LREE/Ti, and Sr/Hf ratios in residual mantle minerals.

Carbonate-rich melts in the mantle have been recognized as primary melts and important agents for altering mantle geochemistry [1-5]. Experimental methods used to characterize these melts have produced different estimates of composition [5-7]. To resolve these differences experiments were conducted on two carbonate-dominated bulk compositions (fertile and refractory with respect to Al content and Fe/Mg ratio) to approximate carbonatite melt compositions in equilibrium with oceanic (fertile) and subcontinental lithospheric (refractory) peridotites. The K/Na ratio was also varied to correspond to the more sodic character of oceanic peridotites and the more potassic character of lithospheric peridotites [8,9]. Carbonatite melt with Na as the dominant alkali (NaCM) is in equilibrium with phlogopite Iherzolite at ca. 25 kbar and 1170 degrees C, and carbonatite melt with K as the dominant alkali (KCM) is in equilibrium with phlogopite Iherzolite at ca. 32 kbar and 1120 degrees C. Mineral compositions in equilibrium with KCM are similar to those found in phlogopite peridotites from the Kaapvaal lithosphere. In contrast, the NaCM residual mineral compositions are consistent with those from a model fertile oceanic peridotite (Hawaiian pyrolite). The relative stabilities of garnet and diopside with CO2 define limits (20-50 kbar, 950-1200 degrees C) within which a primary dolomitic carbonatite melt may be in equilibrium with a peridotitic mantle. The K/Na ratio of the mantle component would control this ratio in the resulting carbonatite.

The equilibration of mantle mineralogies with silicate or carbonatite melts may fractionate trace elements in the mantle in distinctively different ways. To address this question in this study, we present new experimental data to quantify the manner in which selected trace elements (Rb, Ba, Na, K, Nb, Ta, Sr, Zr, Ti, Y, Lu, Cr) may distribute between garnet, clinopyroxene and phlogopite and silicateand carbonatite-melts. New data are also presented for describing the partitioning of these elements between olivine and orthopyroxene and a carbonatite melt. In terms of major elements, three bulk systems are considered (1) carbonate dominated, (2) silicate (carbonate-absent) and alkali-rich and (3) silicate (carbonate-absent) and alkali-poor. The composition and phase relations of the carbonate bulk system are given in Sweeney (1994). The alkali-rich silicate bulk system has a MARID (principally comprising mica-amphibole-diopside) bulk composition (AJE137, Sweeney et al., 1993) and the alkali-poor silicate bulk composition is the quartz tholeiite of Kingwood and Green (1966). In terms of their major element composition, the liquids produced in these three bulk systems are (1) alkali-rich dolomitic melts, (2) K-rich silicaundersaturated basaltic melts, and (3) K-poor silica-saturated basaltic melts. It is suggested that these melt compositions cover the likely spectrum of low degree melt compositions in the Earth's upper mantle. Melts (1) and (2) may be produced in a mantle peridotite at low melting degrees and (3) by the partial fusion of a subducting slab. It is these low degree melts which may carry substantial budgets of incompatible elements and thus we consider the bulk systems studied to be relevant to the quantification of mantle metasomatism. Experimental technique

Three distinct basaltic rock types are recognized on the basis of detailed new analyses in the eastern zone (central Lebombo) of the Karoo Igneous Province. The high titanium and zirconium group is composed of low-Fe (henceforth called simply high Ti-Zr basalts) and high-Fe (henceforth called 'high-Fe') varieties and distinguished from the low titanium and zirconium group by 2-4 times greater abundances of Ti, P, Zr, Nb, Y, La, Ce, and Nd (high field strength elements; HFSE), as well as higher but more variable abundances of K, Rb, Ba, and Sr. All rock types are interbedded to some degree, although the high Ti-Zr type (low-Fe) predominates at the base of the section and the 'high-Fe' type occurs mostly at the top of the section. Stratigraphic relationships show that the high Ti-Zr basalts to the north are replaced as the dominant basalt type by low Ti-Zr basalts to the south within a distance of 60 km. Eruption of each rock type from replenished, tapped, and fractionated magma chambers can account for much of the major and trace element variation within each suite, with the notable exceptions of the large ion lithophile elements (LILE; K, Rb, and Ba) and Sr. Neither crystal fractionation nor contamination with granitic crust can produce the variations between the basalt types and does not appear to affect significantly the geochemistry within any basalt type. The North Lebombo high Ti-Zr picrites are the most likely parent magmas to the high Ti-Zr low-MgO group. Consideration of previous studies of the picrites suggested that the high Ti-Zr picrite geochemistry requires a mixed source or the mixing of two picrite endmember magmas discriminated on the basis of degree of incompatible element enrichment. At least one of the endmembers must be within, or derived from, sub-continental lithospheric mantle. The low-MgO high Ti-Zr basalt geochemistry is consistent with evolution from a picritic parent with about a 30-44 wt.% mantle lithospheric component. In contrast, the low Ti-Zr basalt group parental composition is outside the compositional range suggested for the high Ti-Zr picrite endmembers, being more depleted in incompatible elements (e.g., Ti, P, Zr, Nb, LREE, and LILE). Again, by analogy with possible picritic parents, this group may segregate from a source more depleted in incompatible elements than a mid-ocean ridge basalt (MORB) source, leaving a harzburgitic residue. The 'high-Fe' basalt group is very evolved (e.g., MgO<5%) but may still be constrained by Nd-isotope and Zr/Y ratios to have a mantle source geochemically similar to that from which the Walvis Ridge basalts were derived. If a mantle plume contributed to basalt geochemistry significantly, then the late-stage 'high-Fe' basalts represent this component most closely.