Emeleus, C.H. & Gyopari, M.C. 1992. British Tertiary Volcanic Province, Geological Conservation Review Series No. 4. JNCC, Peterborough, ISBN 0 412 47980 X. The original source material for these web pages has been made available by the JNCC under the Open Government Licence 3.0. Full details in the JNCC Open Data Policy
Chapter 1 Introduction to the British Tertiary Volcanic Province
The general setting
The north-western British Isles was the site of intense igneous activity during the Palaeocene and early Eocene (c. 63–52 Ma) which accompanied continental separation and lithospheric attenuation during the early stages of the opening of the North Atlantic. In Great Britain, volcanism was most vigorous in the Inner Hebrides and adjoining north-west Scotland but also extended to southern Scotland, north-east England and the Outer Hebrides. Intrusions of similar age in North Wales and the English Midlands are probably outlying representatives of contemporaneous activity in north-east Ireland; additional activity in the Bristol Channel centred on Lundy. The region encompassing this activity is known as the British Tertiary Volcanic Province (BTVP) or the British Tertiary Igneous Province (BTIP)
The igneous activity took many forms and involved a wide variety of magmas and rock types. The remains of large accumulations of dominantly basaltic lava flows cover extensive areas in Skye and Mull. Laterally extensive swarms of basaltic dykes are most intense near Skye, Rum, Mull, and Arran but extend to the Outer Hebrides, southern Scotland, Cleveland in north Yorkshire and parts of North Wales and central England. Central intrusive complexes consisting of granite, gabbro, peridotite and other rock types occur on a line from Skye to the Bristol Channel and at several places in the northeast Atlantic
The varied igneous rocks, together with associated sediments and metamorphic rocks have responded in different ways to the profound erosion of the last fifty million years. Gabbro and peridotite have given rise to the rugged mountain scenery of St Kilda, the Skye Cuillins and Rum. The considerable, but generally less-rugged, mountains of northern Arran and the Skye Red Hills are composed of granite, while the piles of flat-lying lavas form tabular, 'trap-featured' hills in northern Skye, rising sometimes to form high mountains, as at Ben More, Mull.
Geologists, mineralogists, mountaineers and others have been attracted to the Province for over two centuries by the spectacular mountain and coastal scenery, and by the abundance of fresh rock exposures from sea-level to over 1000 metres altitude. Pennant, Necker de Saussure, Ami Boue and MacCulloch were among early visitors; notable scientists who worked on the Province in later years included Judd, Geikie, Harker, Bowen, Holmes and Wager; and the area is one where at the present time research into fundamental problems of igneous geology is actively pursued. The intensity of scientific investigation has made the BTVP one of the most historically important and deeply studied igneous provinces in the world. Furthermore, it is acknowledged as one of the best areas in the British Isles in which to demonstrate igneous rocks in the field, and consequently it is visited annually by numerous groups from universities, schools and scientific societies from Britain and abroad.
GCR sites have been selected to cover the features of the BTVP within Great Britain. Additional sites have been identified in Northern Ireland, but these are outside the terms of reference of this published review. The sites vary considerably in importance, size and scope. Some are whole mountain groups (for example, the Skye Cuillin Hills); others may be merely stream sections or small quarries but all have been selected as scientifically significant examples of their kind, where the features described can be observed and appreciated at present and which should be preserved for future study and research.
The Tertiary igneous activity occurred in geographically well-defined areas which usually include a central volcano and surrounding lavas. These areas have been made the basis of individual chapters.
The igneous sequence
Volcanism in the BTVP extended over a period of about twelve million years, largely within the Palaeocene Epoch, apart from the later intrusions within the Eastern Red Hills Centre of Skye (Dickin, 1981), the Sgurr of Eigg pitchstone (Dickin and Jones, 1983), the Lundy granite complex (Hampton and Taylor, 1983) and the youngest intrusion in the Mourne Mountains Western Centre, Ireland (Meighan et al., 1988). The Palaeocene volcanism was preceded by Cretaceous igneous activity in the eastern Atlantic (Harrison, 1982) and it is also possible that the submarine central complex at the Blackstones, south-west of Mull, is Cretaceous in age (Durant et al., 1982). Within the BTVP, the life span of individual central complexes was short, of the order of two or three million years (or even less) and the thick lava accumulations built up over even shorter periods
Volcanic activity in a given area often started with the formation of small amounts of basaltic ash and other volcaniclastic accumulations. These were quickly followed by voluminous subaerial eruptions of basaltic lavas which covered the peneplaned surfaces of older sedimentary and metamorphic rocks ranging in age from the Precambrian to the Cretaceous. Occasionally, the lavas covered landscapes of considerable relief, filling valleys, burying hills and sometimes flowing into shallow lakes, where pillow lavas and hyaloclastites formed. Sedimentary horizons are not common in the lavas, but fluviatile conglomerates, sandstones and fine-grained plant-bearing horizons do occur and provide valuable stratigraphic and palaeogeographical information. The lavas were often subjected to intense weathering between flow extrusion, with the formation of bright red lateritic deposits. With increasing thickness of lavas, heated waters circulating through the flows altered the basalts and deposited distinctive suites of zeolite minerals, for which Skye and Mull are particularly noted.
The lavas were principally fed from fissure eruptions similar to those of present day Iceland. The actual feeders are among the multitude of dykes forming the swarms which extend across the Province; dykes intrude virtually all intrusions and extrusions in the BTVP, so it is likely that lava effusion also occurred throughout the life span of the Province. However, the thick sequences of lavas now preserved in Mull, Skye, and the Small Isles built up between about 63 Ma and 60 Ma, early in the life of the BTVP. Occasionally, lavas must have been erupted as the central complexes developed; there is good evidence from several centres that silicic and intermediate lavas were closely associated with central complexes but there are few substantiated examples of basaltic lavas, with the exception of pillow lavas within the Mull centre. Not all the (predominantly basaltic) magma reached the surface to form flows, some froze in conduits as dykes and plugs and quite large amounts spread laterally through the Mesozoic sediments beneath the lavas to form the prominent dolerite sills of northern Skye and Arran.
The central complexes generally post-date the adjoining lavas, but they were intruded by later members of the dyke swarms. Within the complexes the magmatic sequences were rarely straightforward: in Mull early granite intrusions were followed by numerous basaltic, intermediate and acid cone-sheets, by gabbros and peridotites and by further granites; in Skye, the sequence was apparently simpler, the gabbros and peridotites of the Cuillins were cut by numerous basaltic cone-sheets and subsequently by granites of the Red Hills. Thus, the central complexes record varied intrusive sequences in which basaltic and granitic magmas have been intimately associated. Occasionally, it may be shown that contrasted magmas must have coexisted, forming, for example, the composite basalt/quartz-porphyry sheets and dykes of the Province and the complicated intrusion breccias found in Ardnamurchan, Skye and other central complexes.
Within the central complexes, individual centres of activity are defined by arcuate intrusions — cone-sheets, ring-dykes and stocks — which have a common focus. In most central complexes, and in particular Ardnamurchan, Skye and Mull, the intrusions of one centre may cut those of an earlier one, recording movements in the focus of magmatic activity with time. Ardnamurchan provides an exceptionally clear and often cited example of this phenomenon.
The central complexes are generally spatially well-separated from one another. This has made correlation between centres difficult; furthermore, the present precision of radiometric age determinations is not sufficiently good to obtain relative ages, let alone distinguish between events within a central complex (cf.
By combining the field evidence and the available radiometric age determinations with detailed palaeomagnetic measurements, Mussett et al. (1988) have been able to build up a stratigraphic framework for the BTVP, and at the same time have shown that intense igneous activity in any one lava field, or within a given centre, was usually of geologically short duration (cf.
A review of research
'The gradual development of opinion regarding the nature and history of volcanic rocks is thus in no small measure bound up with the progress of observation and inference in regard to the Tertiary volcanic series.' Sir Archibald Geikie's words, written nearly a hundred years ago about the fresh and extensive exposures of volcanic rocks in western Scotland and Northern Ireland (Geikie, 1897), remain true to the present day. Since the early days of geology in the late eighteenth century, investigations of these rocks have been to the fore in the advancement of geology and mineralogy, and, in the last few decades, of geochemistry and geophysics.
Major advances to c. 1945
The early history of investigation of the Province is admirably recounted in the second volume of Geikie's Ancient Volcanoes of Great Britain (Geikie, 1897), in which the important role of these rocks in the controversies of the Neptunist and Plutonist schools is recorded. Continental visitors, particularly from France, who were familiar with active and recently active volcanoes, unequivocally interpreted the basalts as the result of volcanic eruptions. Notwithstanding, others including Robert Jameson, the great Scottish mineralogist and Professor of Natural History at Edinburgh University, strongly advocated the Neptunists', or Wernerian, viewpoint and ridiculed any connection between these rocks and volcanic activity. The controversy continued well into the nineteenth century, with the gradual ascendency of the Plutonist viewpoint, partly through observations on the dykes of north-east England. Recognition of the Tertiary age of the rocks was due to the Duke of Argyll (1851) who discovered and described plant remains in sediments intercalcated with basalt flows on Mull, although arguments about the ages persisted and older dates were assigned to the rocks of Skye where difficulties were experienced in distinguishing sills in the Mesozoic rocks from true lava flows.
In the latter half of the nineteenth century very significant advances were achieved. The noteworthy investigations by A. Geikie (eg. 1888, 1894, 1897) and J. W. Judd (eg. 1885, 1886, 1889) were mainly concerned with the overall structure and mode of accumulation of the Hebridean lava plateau. Geikie drew on his experiences in the western USA and Iceland to assert that the lavas had built up by successive fissure eruptions, whereas Judd recognized that the central complexes were the eroded remains of major Tertiary volcanoes, and was strongly of the opinion that they were the sources of the lavas.
Partly in an attempt to resolve this controversy, A. Geikie, the then Director of the Geological Survey, arranged for Alfred Harker of Cambridge University to make a detailed survey of the Tertiary igneous rocks of Skye, thereby initiating Survey investigations which were to cover Skye (Harker, 1904), Rum and the other Small Isles of Inverness-shire (Harker, 1908), Mull (Bailey et al., 1924), Arran (Tyrrell, 1928) and Ardnamurchan (Richey and Thomas, 1930). The impact of these memoirs and the accompanying maps was immense. The Skye Memoir detailed swarms of basaltic sheets inclined towards a common focus beneath the gabbros and granites, the coexistence and hybridization of magmas of contrasted compositions, and the formation of a central complex by multiple intrusions. The Mull Memoir confirmed the abundance of centrally inclined sheets (termed cone-sheets) in the central complex, described ring-dykes, including the type example at Loch Bà, and demonstrated the migration of igneous activity within a central complex with time. The relatively large number of chemical analyses made on the Mull igneous rocks allowed the Memoir authors to introduce the concepts of distinct chemical lineages of igneous rocks forming 'Magma Types' and 'Magma Series'; the importance of crystal differentiation in the development of these associations was also stressed. Richey and Thomas's work on Ardnamurchan (1930) built on ideas generated in the Mull Memoir, numerous ring-dykes and several suites of cone-sheets were recognized and the presence of three distinct centres was demonstrated.
Research in the BTVP was relatively limited between the early 1930s and the mid 1940s. The lavas and sills of northern Skye were mapped by the Geological Survey (Anderson and Dunham, 1966). A reassessment of the composite Glen More ring-dyke of Mull by Holmes (1936) and by Fenner (1937) concluded that the upward passage from olivine dolerite to granitic rocks resulted from mixing of compositionally contrasted magmas rather than by crystal fractionation, as originally proposed by Bailey et al. (1924) and subsequently supported by Koomans and Kuenen (1938). Valuable theoretical contributions based largely on data from the BTVP were made by Kennedy (1931a, 1933) and Kennedy and Anderson (1938) on the possible origins of the basaltic magmas, and by Anderson (1936) on the conditions of formation of ring-dykes and cone-sheets.
Research in the post-war years
The immediate post-war period saw the culmination of a controversy on the origins of granite, and other plutonic rocks, rivalling that between the early nineteenth century Plutonists and Neptunists. The question was, did plutonic granites form from the crystallization of magmas, or were pre-existing rocks transformed, or granitized, in situ to form granites? Evidence from the Province was regarded as crucial to the arguments of both sides, granites in the eastern Red Hills, Skye and western Rum were regarded as transformed Torridonian and Jurassic sediments respectively (e.g. Black, 1954, 1955), and it was also suggested that feldspar-phyric gabbros on Skye, were transformed amygdaloidal basalts (Reynolds, 1951; King, 1953). While these claims were strongly contested as they arose, a key discovery which decisively helped to swing the debate on granite origins in favour of the magmatists was made on the Beinn an Dubhaich granite on Skye, where Tuttle and Keith (1954) conclusively demonstrated that the mineralogy of this granite had characteristics of both the problematical plutonic granites and extrusive rhyolites of undoubted magmatic origin. It, and other BTVP granites, provided the essential 'missing link'.
The layered rocks
N. L. Bowen's classic work The Evolution of the Igneous Rocks (1928), included arguments strongly favouring crystal fractionation as a major mechanism in determining rock compositions. He drew heavily on examples from the BTVP, in particular citing the Mull feldsparphyric Porphyritic Central Magma-type rocks and the peridotite dykes of Skye as products of the process. The efficacy of this mechanism was brought into sharp focus by Wager and Deer's (1939) investigations of the East Greenland Skaergaard Intrusion. The strikingly layered rocks of Skaergaard were considered to have formed by the gravitational accumulation of crops of crystals from fractionating basaltic magma and its derivatives. The literature of the BTVP was already full of examples of banded peridotites and gabbros (for example, Geikie and Teall, 1894; Harker, 1904, 1908; Bailey et al., 1924; Richey and Thomas, 1930) and it was immediately recognized that there were at least superficial similarities to the Skaergaard layering, especially in the Skye Cuillins and on Rum. Brown's (1956) study of the latter provided convincing evidence of crystal sorting and settling forming the layered rocks, but in contrast to Skaergaard, this was not accompanied by the systematic compositional ('cryptic') variations in rock and mineral chemistry. The explanation offered was that the Rum layered rocks built up from a succession of pulses (15) of basaltic magma, whereas there was only a single input of magma at Skaergaard. Thus Rum became, and remains, a classic example of an 'open' magma chamber, fed by successive batches of magma from each of which early, high-temperature crystals separated and accumulated to form the layered rocks of Hallival and Askival.
Minerals in layered igneous rocks frequently exhibit distinctive textural relationships thought to be due to the accumulation of some (cumulus) crystals and the precipitation of others from the liquid (intercumulus) trapped between sedimented crystals. The fresh, unaltered rocks of Skye and Rum provide type examples of the textures and structures found in these 'magmatic sediments' and the two centres are classic examples of layered igneous rocks and 'igneous cumulates' (Wager et al., 1960; Wager and Brown, 1968). During the past decade, ideas on the origins of igneous rocks have undergone radical reappraisal (cf. McBirney and Noyes, 1979) and it has increasingly been recognized that in situ crystallization, double-diffusive convection and movement of trapped, high-temperature liquids through crystal mushes may contribute to the formation of layered igneous rocks or their modification, and that crystal fractionation aided by convection-driven magmatic currents may not always be applicable. The exceptionally well-exposed layered rocks of Rum have provided an excellent natural laboratory where the new hypotheses have been tested and developed (see for example, papers in the Geological Magazine, 122, 1985).
The contribution from geochemical and isotopic studies
The 1950s and 1960s saw the extensive application of the analytical techniques of optical spectroscopy, spectrophotometry and X-ray fluorescence to geological and mineralogical problems. Geochemistry blossomed as a subject, as it became possible to analyse rocks rapidly, and to detect a whole range of hitherto geologically inaccessible elements in concentrations down to a few parts per million. Many of the pioneering applications of these techniques took place in the BTVP: trace element distributions in the Skye rocks were examined (Nockolds and Allen, 1954); the geochemistry of whole suites from the Province analysed and compared with other classic areas (Wager, 1956; Tilley and Muir, 1962, 1967); and detailed successions of lavas and intrusions were analysed (for example, the lavas of Skye, Thompson et al., 1972), the Ardnamurchan cone-sheets (Holland and Brown, 1972) and the dykes of Skye (Mattey et al., 1977). Several distinctive groups of basaltic and associated compositions were recognized in the Hebrides of which the Skye Main Lava Series and the Mull Plateau Group were the most abundant, including olivine basalts, hawaiites and mugearites which varied from transitional to mildly alkaline compositions (Fig. 1.2). Silica-rich, alkali-poor, tholeiitic rocks form late flows on Skye and are found as dykes on Skye and occur elsewhere in the Province; these were termed the Preshal More group
The geological application of mass spectrometry and the development of isotope geochemistry since the 1960s have provided crucial new information about the ages and origins of igneous rocks. The BTVP has figured prominently in the application of these techniques. The Skye granites, gabbros and lavas (Moorbath and Bell, 1965; Moorbath and Welke, 1969) yielded Palaeocene ages, together with strontium and lead isotopic data which showed that, whereas the basaltic rocks had distinct mantle signatures, the granites on the whole did not and were probably derived by partial melting from crustal sources. Lead derived from Lewisian sources was present and it was proposed that the granites came from partial melting of the gneisses. Subsequent oxygen isotope investigations (Taylor and Forester, 1971; Forester and Taylor, 1976, 1977) showed that the granites had been pervasively invaded by heated groundwaters as had the rocks around the central complexes. This raised the possibility of significant post-consolidation changes in the compositions of the rocks and consequently cast doubt on the validity of the earlier Sr- and Pb-isotope studies, with the added corollary that detailed elemental and isotopic studies of rocks from the BTVP (especially in and near central complexes) might be of very limited petrogenetic application. This crucial problem was specifically addressed by Pankhurst et al. (1978) who showed, by means of a careful study of the Loch Uisg Mull granite, that, provided that the samples were carefully selected, post-consolidation changes were in fact minimal.
Precision, accuracy and detection limits in elemental geochemistry have also been significantly improved in the past two decades, partly due to the introduction of induced neutron activation analysis (INAA) and inductively coupled plasma mass spectrometry (ICPMS). These, and other new and improved techniques, have been applied to the Province, notably to the problem of granite genesis. Both on Mull (Walsh et al., 1979) and on Skye (Dickin, 1981) there is now a consensus that the granites contain variable, but always significant contributions from both mantle and crustal sources (for example, Dickin et al., 1984). There is considerable compositional and isotopic variation within the basaltic rocks and their derivatives which has, for Mull, been attributed by Beckinsale et al. (1978) to variations in mantle sources. Essentially the same data are amenable to alternative interpretations: while agreeing with the ultimate mantle source, Morrison et al. (1985) and Thompson et al. (1986) suggest that a major factor contributing to the compositional variability is the extent to which a particular batch of magma has interacted with the lower and/or upper crust during its progress towards the surface. Using data obtained from flows low in the Skye and south-west Mull lava successions, they argued that each separate batch of magma may have had a distinctive and complicated history of intermittent ascent, ponding at different levels in the crust where rock types caused density traps, and reacting with the adjoining rocks to a greater or lesser extent. The Skye and Mull lavas have thus been used as geochemical probes to investigate the magmatic plumbing of part of the feeder system for the BTVP lavas.
A tholeiite dyke from the north of England was among the first intrusions to be dated radiometrically (Dubey and Holmes, 1929) and numerous age determinations have now been made on rocks from the BTVP
Deep structure of the central complexes: the geological input
Studies of the gravitational field over the BTVP have yielded far-reaching and important results. In addition to regional surveys (by the British Geological Survey), much detailed local information has come from a series of investigations since the early 1950s (for example, Tuson, 1959). The central complexes are almost invariably the sites of positive Bouguer gravity anomalies which include some of the most pronounced in the British Isles (for example, the submarine Blackstones Bank complex; McQuillin et al., 1975). The interpretations of the Skye, Mull and Ardnamurchan (Bott and Tuson, 1973), Arran and Rum (McQuillin and Tuson, 1963) anomalies show that each centre is underlain by dense rock (3.0–3.3 g cm−1) of the composition of very olivine-rich gabbro or feldspathic peridotite. The masses of mafic rocks are broadly cylindrical or else steep-sided truncated cones, they extend for depths of 12 km or more through the crust, and they may be over 20 km in diameter (cf. Bott and Tuson, 1973, figs 1 and 2). They are thus major features of the Hebridean crust and they underlie not only the mafic parts of centres but also the granites. The implications of the interpretations of the anomalies are far-reaching: mafic magmas must have dominated the BTVP igneous activity for, although areally extensive, the granites are only superficial bodies, perhaps only 2 km thick (Bott and Tantrigoda, 1987). The mafic rocks could have given rise to the granites by magmatic fractionation but equally they were heat sources capable of effecting melting in the surrounding crustal rocks, thus generating the granitic magmas, or a component thereof. The gabbroic and ultrabasic rocks of Rum are probably the upper tip of one of these mafic bodies.
The precise conditions governing the siting of the central complexes are not fully understood, but their location at or near the intersections of the basic dyke swarms with major faults has long attracted attention, and a causative connection has been suggested (Richey, 1937; Vann, 1978; Upton, 1988). If the major faults, such as the Great Glen and Highland Boundary Faults, have influenced siting of the central complexes, then that influence has extended to the base of the Palaeocene crust.
Modelling magmatic evolution
A major problem of the ultrabasic rocks of the BTVP, and elsewhere, has been to decide whether they formed from ultrabasic or basaltic magmas. Their frequent close association with rocks of basaltic compositions makes the latter an attractive proposition (Brown, 1956); however, the possibility that high-temperature ultrabasic magmas may have been present at upper crustal levels in the BTVP has become accepted in recent years. The idea was advanced by Dreyer and Johnston (1958) and Wyllie and Dreyer (1963) from examinations of minor ultrabasic intrusions around the Skye Cuillin site and on the nearby Isle of Soay. The close similarities between the minor ultrabasic intrusions of Skye and Rum and the layered ultrabasic rocks of those islands suggested to Gibb (1976) that the ultrabasic rocks had formed from parental magmas consisting of a suspension of olivine crystals in ultrabasic ('eucritic') liquid. Further research by Donaldson (1975) on the harrisites and ultrabasic breccias of south-west Rum led him to postulate the participation of feldspathic ultrabasic liquids (possibly hydrous) in the formation of these rocks. In addition, the subsequent discovery of quenched, aphyric ultrabasic dykes intruding the Rum layered rocks (McClurg, 1982), the recognition that some at least of the feldspathic peridotite layers in the Rum succession were intrusive (Renner and Palacz, 1987; Bédard et al., 1988), and the presence of a quenched ultrabasic margin to this complex at Beinn nan Stac (Greenwood et al., 1990), provided further strong supporting evidence for ultrabasic liquids. It thus appears inescapable from the evidence furnished by the BTVP that conditions in the central complexes and their immediate surroundings favoured the rise of hot, dense picritic liquids to within short distances (c. 1 km) of the Earth's surface. Although not unique, these examples are unusual and it appears that the special conditions involving rapid, strongly focused throughput of hot basaltic magmas, resulted in the formation of preheated pathways along which the ultrabasic liquids rose to high structural levels before crystallizing and congealing.
The availability of new, quantitative data on the physical and chemical properties of magmas and increasingly sophisticated computing techniques has made it possible to model magmatic processes. Noteworthy studies using examples from the BTVP have included examination of the emplacement of the Cleveland Dyke of north-east England and of the layering in the Rum ultrabasic rocks. MacDonald et al. (1988) demonstrated a close chemical similarity between the Cleveland Dyke tholeiite and rocks in the Mull central complex. They postulated that the dyke had been fed laterally from a source beneath Mull, flowing '... in a manner transitional between laminar and turbulent conditions'. It was calculated that the magma took between one and five days to reach North Yorkshire. Brown (1956) described contrasted olivine- and feldspar-rich layering in the Rum ultrabasic rocks, attributing the layering to successive gravitational settling of olivine and feldspar, the residual low temperature magma being extruded during concomitant surface volcanism. Recent attempts to model the layering (see Young et al., 1988 for summary) accept Brown's concept of a repeatedly replenished open magma chamber but envisage pulses of olivine-phyric, Mg-rich basaltic (olivine-phyric picrite) magma with or without contemporaneous basaltic liquid. In one model, picrite crystallizes copious olivine during strong convective circulation, the crystals only settling when movement ceases. The residual, overlying liquid then crystallizes feldspar-rich cumulates by basalt fractionation (Huppert and Sparks, 1980; Sparks et al., 1984; Tait, 1985).
An alternative view (Young et al., 1988) is that simultaneous crystallization of olivine, and olivine plus plagioclase, occurs in a magma chamber zoned upwards from picritic to basaltic compositions. The model assumes a dipping floor to the magma chamber, corresponding more or less with the dipping layering observed on eastern Rum (cf. Brown, 1956; Volker and Upton, 1990). Simultaneous crystallization in picritic and basaltic liquids would result in the formation of peridotite at low levels on the magma chamber floor and of troctolitic (allivalitic) and gabbroic rocks at higher levels. The conspicuous major layering in eastern Rum would therefore result from variations in the rate of magma injection into the chamber, peridotite forming when this was high, and troctolite and gabbro when it waned and olivine precipitation depleted the picritic proportion of the resident magma. The model finds its support in the manner in which the proportion of peridotite in the layered rocks of eastern Rum increases towards the west where it has been suggested that the magma arose along the line of the present Long Loch Fault (McClurg, 1982; Volker and Upton, 1990).