The publication of the British Geological Survey memoir on Rum and the Small Isles in 1997 was followed by a period of intense petrological and mineralogical research, leading to some 40 papers, books and other publications. The research progress since then is reviewed here and integrated with the information previously available to provide an overview of the current status of understanding of the centre. New data on the acidic and mixed acid/basic magmas of the early Rum caldera demonstrate that frequent mafic replenishments were the main driver for magmatic activity at Rum right from its initial stages. The caldera is bound by the Main Ring Fault, a structure which probably also exercised an influence on the emplacement of the subsequent basic and ultrabasic intrusions. The later emplacement of gabbros and ultrabasic rocks caused only limited thermal metamorphism of the surrounding Torridonian sandstones, contrasting markedly with the crustal isotope signatures of the early intracaldera ignimbrite magmas and the intense alteration of uplifted masses of Lewisian gneiss within the ring fault. Rare picritic dykes provide an indication of the possible parent magma for the mafic and ultrabasic rocks, but these, as with most other magmatic rocks on Rum, have undergone varying degrees of crustal contamination, involving both Lewisian granulite and amphibolite-type crust but, notably, no Moine metasedimentary compositions as is the case at the nearby Ardnamurchan centre. Detailed textural studies on the gabbroic and ultrabasic rocks allow a distinction between intrusive peridotites and peridotite that forms part of the classic layered cumulate units of Rum and, furthermore, this work and that on the chromite seams and veins in these rocks shows that movement of trapped magma and magma derived from later intrusions, may produce textures regarded previously as of primary cumulate origin. Sulfides in the chromitite seams and ultrabasic rocks, in turn, show possible influences from assimilated Mesozoic sediments. Igneous activity on Rum was short-lived, possibly only between 0.5 and 1 m.y. in duration and commenced at ∼60.5 Ma. The Rum Central Complex was extinct by the time the main activity at the nearby Skye Central Complex commenced (∼59 Ma). From recent apatite fission-track studies it seems probable that Rum, in common with other Palaeogene centres, underwent a brief, but significantly later heating event (∼45 Ma).

Arsenic-rich magmatic sulphide mineralization is hosted by a diorite intrusion at Talnotry, southwest Scotland. A relatively abundant and diverse platinum-group mineral assemblage is present and is dominated by sperrylite, irarsite and electrum with subordinate merenskyite, michenerite and froodite. Early euhedral gersdorffite is enriched with respect to Rh, Ir and Pt and in some cases contains exsolved blebs of irarsite or euhedral grains of sperrylite. Sperrylite is also enclosed within silicates and sulphides indicating that it crystallized directly from an As-rich sulphide liquid. Pyrrhotite-chalcopyrite mineral assemblages are consistent with the fractional crystallization of monosulphide solid solution and are overlain by PGE-, Ni- and As-rich mineral assemblages indicative of crystallization from a NiAs liquid. Late-stage, cross-cutting, electrum-bearing chalcopyrite veins are consistent with the crystallization of Cu- and Au-rich intermediate solid solution. The chemistry, mineralogy and lithological relationships of the diorite suggest that it may be an appinite and as such is potentially analogous to the Au-rich lamprophyre dykes present within southwest Scotland.

We report preliminary petrographic and mineral chemical data for a rock hosting an unusual mineral assemblage from Långban, Värmland, Sweden. The rock is a two feldspar-scapolite-spessartine-romeite skarn. The bulk composition and high degree of enrichment in Pb, Sb and As suggest that the rock was formed by reaction between a pre-existing Mn skarn containing the chalcophiles and a potassic granite, with loss of silica, alkalis and CO2. The alkali feldspar is a Pb-rich hyalophane, averaging Or63Ab19Cs15Pb03, the plagioclase feldspar a Pb-rich labradorite, An48Ab48Or02Pb02, and the scapolite a 'mizzonite' (Ca/(Na+Ca) = 0.66—0.70). These minerals show their highest Pb contents recorded in nature to date: up to a maximum of 5.7 wt.% PbO in the hyalophane, 2.1% PbO in the plagioclase, and 5.3% PbO in the scapolite. Laser ablation ICP-MS of a scapolite grain detected substantial Be up to 1.7 wt.% BeO (0.6 Be per 12 tetrahedral cations), as well as Pb up to 7.05 wt.% PbO. The Be is incorporated into scapolite via the coupled exchange [Be(OH)][Al(CO3,SO4)]—1. This is the first documentation of scapolite as the major repository for Be in a rock.

The romeite also contains substantial Pb, and shows extensive solid solution towards end-members containing Fe3+, Ti and Sb3+. In some analyses, the dominant end-members are and its Pb analogue rather than (Ca,Pb)2Sb2O7. Complex exsolution textures are displayed in the hyalophane, by hancockite-epidote, romeite-bindheimite and hedyphane-johnbaumite. Ca-rich scapolite and hancockite appear to be new minerals for the Långban deposit.

The mineralogy appears consistent with the regional peak conditions of P = 3 kbar, T > 600°C. Several potential thermobarometers for Mn-rich skarns are identified in this rock.

Flamite (Ca,Na,K)2(Si,P)O4 (P63; a = 43.3726(18), c = 6.8270(4) Å; V = 11122.2(9) Å3), a natural analogue of the P,Na,K-doped high-temperature α-Ca2SiO4 modification, is a new mineral from Ca- and Al-rich paralava, an ultrahigh-temperature combustion metamorphic melt rock. The type locality is situated in the southern Hatrurim Basin, the Negev Desert, Israel. Flamite occurs as regular lamellar intergrowths with partially hydrated larnite, together with rock-forming gehlenite, rankinite and Ti-rich andradite, minor ferrian perovskite, magnesioferrite, hematite, and retrograde ettringite and calcium silicate hydrates. The mineral is greyish to yellowish, transparent with a vitreous lustre, non-fluorescent under ultraviolet light and shows no parting or cleavage; Mohs hardness is 5–5½; calculated density is 3.264 g cm–3. The empirical formula of holotype flamite (mean of 21 analyses) is (Ca1.82Na0.09K0.06(Mg,Fe,Sr,Ba)0.02)Σ1.99(Si0.82P0.18)Σ1.00O4. The strongest lines in the powder X-ray diffraction pattern are [d, Å (Iobs)]: 2.713(100), 2.765(44), 2.759(42), 1.762(32), 2.518(29), 2.402(23), 2.897(19), 1.967(18), 2.220(15), 1.813(15). The strongest bands in the Raman spectrum are 170, 260, 520, 538, 850, 863, 885, 952 and 1003 cm–1.

Approaches currently used for remediating acid mine drainage (chiefly active chemical treatment and passive bioremediation systems) have a number of major detractions, including their failure to recover potentially valuable metals from these waters. Bioremediation strategies that utilizereactor-housed microorganisms can circumvent this problem, but have tended not to be widely used due to their relatively high costs. We have devised innovative approaches for remediating mine waters that use acidophilic bacteria to remove metals either as oxidized or reduced phases, usingmodular bioreactors that are designed to operate at minimal cost and to generate products that have commercial value. A composite system is described that combines microbial oxidation of ferrous iron with abiotic precipitation of ferric iron as schwertmannite, a mineral that has commercialvalue as an absorbent of arsenate and other environmental pollutants, and as a pigment. Sulfidogenic bioreactors maintained at acidic pH values are used to selectively precipitate metal sulfides, such as CuS. Tests with synthetic mine drainage containing mixtures of soluble metals confirmedthat these systems can generate relatively pure mineral deposits from complex acid waters. The units are designed to be configured differently, according to the nature of the mine water requiring treatment.

The rock composition and electron microprobe analyses of the five constituent minerals are presented for a sapphirine-cordierite-bronzite-phlogopite-spinel rock from Namaqualand, South Africa. These data are considered inrelation to experimental data and the mineralogy of certain associated rock types, and it is suggested that the sapphirine-bearing rock represents a metamorphic residuum (restite) after extraction of anatectic granitic liquid from an original argillaceous sediment.

In the early part of 1931 public interest in South Australia was stimulated by the fall of the Karoonda meteorite on November 25, 1930, and its subsequent discovery by an Adelaide University party led by Prof. Kerr Grant. In consequence of this Prof. Grant was informed separately by Mr. B. Bowman of Tempe Downs and Mr. J. M. Mitchell of Oodnadatta that fragments of meteoric iron were to be found surrounding several crater-like depressions near Henbury Cattle Station in Central Australia. The number of craters was variously described as three and five.

The stabilities of possible Y(R3+ + R2+ + Li+) clusters around the W anion (O1 site) of the tourmaline structure were checked using the bond-valence approach. Arrangements involving R3+ = Al3+ or Fe3+ and R2+ = (Fe, Mn, Mg)2+ were all found to be stable. Structural data show a strong linear correlation between the mean formal valence (MFV) of the Y cations and the long-range average bond valence sum (BVS) at the O1 site, as estimated from bond-valence parameters. This correlation is observed for all chemical compositions of tourmaline, except for fluor-buergerite where the O3 site is dominated by oxygen anions. Results show that the long-range site populations of the Y and O1 sites are related to each other by valence constraints described by the empirical and theoretical equations: BVS(O1) = [0.99 MFV(Y) − 1.20] and MFV(O1) = [1.00 MFV(Y) − 1.00], respectively. The systematic deviation of the empirical equation from the ideal one is ascribed to the occurrence of bond strain involving the O1 site. An important implication of the correlation between MFV(Y) and BVS(O1) is that the (OH) content at the O1 site may be estimated by the equation W(OH) = 2 − [1.01 BVS(O1)] − 0.21 − F.

The Las Aguilas Ni-Cu-PGE deposit is associated with a sequence of basic-ultrabasic rocks made up of dunite, harzurgite, norite and amphibolite. These igneous (partially metamorphosed) rocks, and their host granulites, gneisses and migmatites of probable Precambrian age, are highly folded. The sulphide ore, consisting of pyrrhotite, pentlandite and chalcopyrite, occurs in the cores of both antiform and synform structures, within dunite, harzburgite and mainly along shear zones in bronzitite, replacing small mylonitic subgrains. The platinum-group mineral assemblage is dominated by Pd bismuthotellurides (Pt-free merenskyite, palladian bismuthian melonite and michenerite), with minor sperrylite, and PGE-sulpharsenides. The latter often occur as single, zoned crystals frequently showing cores of irarsite; outside these are concentric zones of cobaltian hollingworthite, rhodian nickelian cobaltite and Fe-rich nickelian cobaltite.

Mineralogical, textural and chemical evidence indicate that the sperrylite and platinum-group element sulpharsenides were formed during a primary magmatic event associated with the fractionation of a basaltic melt, which was contaminated by the assimilation of metamorphic crustal rocks. PGE sulpharsenides crystallized from As-bearing, residual magmatic liquids that collected PGE and segregated after the crystallization of the monosulfide solid solution. During high-grade metamorphism, sulpharsenides were remobilized as solid crystals in the liquated sulfides suffering partial dissolution and fracturing. On the other hand, there is no evidence of a primary concentration of Pd-bismuthotelluride minerals, and their present spatial distribution is only the consequence of their formation under high- to medium-grade metamorphism, down to temperatures of below 500°C. Pd bismuthotellurides crystallize even in fractures of sulpharsenides, attached to the boundaries of highly dissolved sulpharsenide crystals, and intergrown with molybdenite.

A proportion of the heat added to a body of rock during prograde metamorphism will be absorbed in the chemical work of metamorphic recrystallization. When and where heat is so absorbed will affect the exact thermal histories of the rocks, and hence the metamorphic textures. This paper reports the results of modelling of the inter-relations between reaction progress and thermal histories in a rock column. The results suggest that volumes of rock undergoing reaction at any moment act as heat sinks and absorb heat from the surrounding rock, that reaction generally takes place close to the temperature at which nucleation took place, and that steady heating of a rock pile can give rise to a reaction history in which spurts of reaction are separated by ‘quiet’, non-reactive intervals.

Static lattice energy calculations have been combined with cluster expansion formalism to predict pairwise ordering interactions in the pyrope–grossular solid solution. The ordering interactions, the Js, have been then used to calculate the activity-composition relations over a wide temperature range with the help of the Cluster Variation Method. It is shown that short-range ordering in the system is driven by size mismatch. The prediction of the right signs and magnitudes of the ordering interaction energies requires separation of the mixing enthalpy into the configuration-dependent (chemical) and the configuration-independent (elastic) components. The study predicts the existence of a miscibility gap below 500°C.

During investigations of the system Sc2O3-Al2O3-TiO2-SiO2, a new, unusual microporous compound, Cs3ScSi8O19, was synthesized as colourless plates from a CsF-MoO3 flux. The crystal structure was determined from single-crystal X-ray diffraction data (Mo-Kα radiation, CCD area detector). The compound is orthorhombic, space group Pnma, with a = 11.286(2), b = 7.033(1), c = 26.714(5) Å, and Z = 4 (R1(F) = 2.6% and wR2all(F2) = 7.3%, using 3066 ‘observed’ reflections with Fo > 4σ(Fo)). The crystal structure of Cs3ScSi8O19 represents a new microporous framework structure type (‘MCV-1’), and the compound is exceptional in being the first representative of a mixed octahedral-tetrahedral framework structure, in which the [TO4]:[MO6] ratio is >6:1. The structure is based on isolated, nearly regular ScO6 octahedra [dav(Sc—O) = 2.112 Å] sharing corners with SiO4 tetrahedra to form an open framework with four-, six- and eight-membered rings; the latter are formed by SiO4 tetrahedra only. Two fully occupied Cs positions are located in large framework voids close to the six-membered rings, whereas four partly occupied and disordered Cs positions are close to very large framework voids bordered by the puckered eight-membered rings. The cavities are linked into channels parallel to [100] and [010]. The structure is compared with that of Cs2TiSi6O15 and related microporous scandium-, REE-, titano- and zirconosilicate minerals and compounds. Cs3ScSi8O19 or derivatives may be important in the context of immobilization of radioactive 137Cs waste, cationic conductivity or catalysis.

The Late Pleistocene kamafugite–carbonatite association at San Venanzo-Pian di Celle forms part of the Umbria-Latium Ultra-alkaline District (ULUD) of central Italy and, together with Toro-Ankole, SW Uganda and Mata de Corda, Brazil, represents one of three similar occurrences so far reported worldwide.

Excellent field exposure and stratigraphic control prompted a study of the kamafugite–carbonatite suite and related phase interactions to understand the nature of the distinct mineral assemblages of the pyroclasts, compared to that of the lavas, the former containing essential potassium feldspar and aluminous diopside crystals, absent in the latter.

The pyroclastic rocks represent a small amount of magma characterized by ubiquitous mantle xenocrysts and emplaced by early high-velocity eruptions. All the investigated specimens show a high Mg/(Mg+Fe2+) ratio (0.84–0.93) and high compatible elements (Ni+Cr>1000 ppm). Lavas (venanzite, i.e. leucite melilitite) and a sill (uncompahgrite, i.e. melilitolite) represent final events in the volcanic sequence. They yielded a (Na+K)/Al ratio of c. 1.1 and are larnite-bearing in the CIPW norm. Glass from the lapilli is peralkaline, i.e. (Na+K)/Al>2, and close to the lava in composition. Glass from melilitolite yielded CIPW Or and Hy and is strongly peralkaline, i.e. (Na+K)/Al = 5–6. The lapilli typically exhibit concentrically zoned structures which compound subliquidus venanzite phases, e.g. melilite, leucite, and kalsilite, with mantle xenolithic/xenocrystic debris and carbonatite phases. These lapilli represent a distinct variant of the venanzite liquid, mechanically fractionated and quenched by the diatremic process.

Mantle-normalized HFSE for both lava and lapilli show typical extrusive-carbonatite patterns. Carbonatitic beds intercalated with the pyroclastic suite are distinct and typically consist of carbonates high in Sr, Ba and REE. Primary carbonate yielded C isotope compositions ranging from –5.0 to –6.0 δ13C‰, falling within the range of mantle compositions. Distinct differentiation trends of the venanzite magma and its derivatives were recognized, hinging on the coexistence of the silicate and carbonatite fractions. Potential sanidine crystallization trends are suggested, distinct from the venanzite→melilitolite trend, reported for Oldoinyo Lengai assemblages.

Unusual aspects of the San Venanzo rock association, relative to similar rock types elsewhere, include the combination of a rare mantle source composition with a lithosphere about 80 km thick. A genetic model for the origin of the San Venanzo kamafugite–carbonatite association and related carbonate-silicate interactions is proposed and discussed. This may be relevant to the petrogenesis of similar rocks elsewhere, particularly in the light of the detailed data on the pyroclasts.

The widespread occurrence of pyrope garnet in Archean lithospheric mantle remains one of the 'holy grails' of mantle petrology. Most garnets found in peridotitic mantle equilibrated with incompatible-trace-element enriched melts or fluids and are the products of metasomatism. Less common are macroscopic intergrowths of pyrope garnet formed by exsolution from orthopyroxene. Spectacular examples of these are preserved in both mantle xenoliths and large, isolated crystals (megacrysts) from the Kaapvaal craton of southern Africa, and provide direct evidence that some garnet inthe sub-continental lithospheric mantle formed initially by isochemical rather than metasomatic processes. The orthopyroxene hosts are enstatites and fully equilibrated with their exsolved phases (low-Cr pyrope garnet ± Cr-diopside). Significantly, P-T estimates of the postexsolution orthopyroxenes plot along an unperturbed conductive Kaapvaal craton geotherm and reveal that they were entrained from a large continuous depth interval (85 to 175 km). They therefore represent snapshots of processes operating throughout almost the entire thickness of the sub-cratonic lithosphericmantle.

New rare-earth element (REE) analyses show that the exsolved garnets occupy the full spectrum recorded by garnets in mantle peridotites and also diamond inclusions. A key finding is that a few low-temperature exsolved garnets, derived from depths of ∼90 km, are more depleted in light rare-earth elements (LREEs) than previously observed in any other mantle sample. Importantly, the REE patterns of these strongly LREE-depleted garnets resemble the hypothetical composition proposed for pre-metasomatic garnets that are thought to pre-date major enrichment events in the sub-continental lithospheric mantle, including those associated with diamond formation. The recalculated compositions of pre-exsolution orthopyroxenes have higher Al2O3 and CaO contents than their post-exsolution counterparts and most probably formed as shallow residues of large amounts of adiabatic decompression melting in the spinel-stability field. It is inferred that exsolution of garnet from Kaapvaal orthopyroxenes may have been widespread, and perhaps accompanied cratonization at ∼2.9 to 2.75 Ga. Such a process would considerably increase the density and stability of the continental lithosphere.

The ionic radius of zinc (0·83 Å.) is identical with that of divalent iron (0·83 Å.) and nearly the same as that of magnesium (0·78 Å.). One would therefore expect zinc to be mineralogically closely related to iron and magnesium. It is, however, well known that this is not the case. An examination of a list of the zinc minerals which have so far been described shows that only in a few cases are the corresponding isomorphous ferromagnesian species known. As a rule zinc minerals and iron-magnesium minerals with corresponding chemical formulae are entirely different in their crystal-structure.

Cell dimensions and densities derived from least-squares refinement of X-ray powder diffractometer patterns are: orthoenstatite, a 18·2249±0·0013, b 8·8149±0·0008, c 5·1746±0·0008 Å, V 831·30±0·14 Å3, D 3·208 g/cc ; low clinoenstatite, a 9·6065±0·0011, b 8·8146±0·0007, c 5·1688±0·0006 Å, β = 108·335± 0·009, V 415·46±0·09 Å3, D 3·210 gm/cc. Because orthoenstatite transforms reversibly to clinoenstatite at high pressure and low temperature, and because a statistical test shows that orthoenstatite is less dense than clinoenstatite at the 90 % confidence level (in accord with the direction of slope of the experimental transition curve), it is no longer possible to regard low clinoenstatite as a metastable form of MgSiO3.

Whilst some atention has in the past been devoted to the question of the properties of natural glasses, very little has been done to indicate the interrelations of the properties determined. In the case of tile natural volcanic glasses, this is unfortullately only too evident from a study of the standard text-books. A determination of such properties as density, refractivity, ttud chemical composition, and the interrelations of these for particular glasses, is rarely to be found in the literature. The chief contribution to this subject is due to Stark, who has examined the refringence of a large number of volcanic glasses. Unfortunately, there is no record of the density of these rocks, and the interrelation of these remained unknown.

In May 1950 the writer received from Mr. Henry Burgess, of Spiggie, Shetland Isles, a specimen of a material which had been found by a local quarryman. The property of the material which had particularly attracted attention was that when it absorbed water it became soft and pliable and looked like wet leather. The specimen, measuring approximately 4x2X1 inches, had a brownish iron-stained outer skin, and showed slight signs of foliation parallel to its length. Internally it consisted of a white to greyish-white compact material, in which were embedded portions of the syenitie country-rock. This white material has been shown to consist dominantly of the mineral palygorskite (attapulgite).

Li-mica (zinnwaldite and/or lepidolite)—topaz—albite granites in the Tregonning—Godolphin pluton and similar rocks in the St. Austell pluton appear to be petrogenetically unrelated to the spatially associated biotite granites. Evidence is provided by lack of development of Li-mica granites at roof zones of biotite granites and markedly different trends and composition fields in bivariate plots such as Li vs. Cs, Rb vs. Sr and Nb vs. Zr. Thus, differentiation of biotite granite magma is unlikely to have generated Li-mica granite magma, as also, on its own, is partial melting of biotite granite or biotiteabsent residual lower crust. However, partial melting of biotite-rich residual rocks involving biotite breakdown could yield a trace alkali- and F-enriched melt, although this would require marked femic mineral, K-feldspar and anorthite fractionation, and Na-enrichment. It is proposed that volatiles derwed from either a mantle source or the crust/mantle interface have aided metasomatism of either residual S-type crust that earlier provided S-type biotite granite magma, or basic (biotite-rich) granitoid, to produce a low-temperature, low-viscosity Li-mica granite melt that rose rapidly in the crust soon after the emplacement of associated biotite granites.

The crystal structure of caratiite, K4Cu4O2 (SO4)4MeCl (a 13.60(2), c 4.98(1) Å, Z = 2) was solved in space group I4 from single crystal data, and refined to R = 0.035 for 654Fo > 6σ(Fo). The atomic arrangement is characterized by Cu2O(SO4)2 chains parallel to [001], with Cu in (4 + 1) co-ordination and O surrounded tetrahedrally by four Cu atoms. These chains are interconnected by K atoms to a framework, the channels of which house the Me and Cl atoms. The nature of the Me atoms is not completely clear; most probably the position is half occupied by Cu, fully occupied by Na, or occupied by a mixture of the two.