Geology of Rum and the adjacent islands Memoir for 1:50 000 Geological Sheet 60 (Scotland)

By C H Emeleus

This memoir, and the 1:50 000 scale geological map that it describes, are the products of a mapping contract between the Natural Environment Research Council and the University of Durham. The interpretations presented are those of the authors.

Bibliographical reference: Emeleus, C H. 1997. Geology of Rum and the adjacent islands. Memoir of the British Geological Survey, Sheet 60 (Scotland).

British Geological Survey

Geology of Rum and the adjacent islands. Memoir for 1:50 000 Geological Sheet 60 (Scotland)

C H Emeleus

This memoir, and the 1:50 000 scale geological map that it describes, are the products of a mapping contract between the Natural Environment Research Council and the University of Durham. The interpretations presented are those of the authors.

Bibliographical reference: Emeleus, C H. 1997. Geology of Rum and the adjacent islands. Memoir of the British Geological Survey, Sheet 60 (Scotland).

Contributors

© NERC copyright 1997 First published 1997. ISBN 0 11 884517 9. Printed in the UK for The Stationery Office J27713 C6 12/97

The grid used on the figures is the National Grid taken from the Ordnance Survey map. Replace (Figure 2) is based on material from Ordnance Survey 1:50 000 scale map number 39 with the permission of The Controller of Her Majesty's Stationery Office. © Crown copyright. Ordnance Survey Licence No. GD272191/1997.

Author

C H Emeleus, MSc, DPhil, DSc University of Durham

Contributors

Other publications of the Survey dealing with this district and adjoining districts

Books

Maps

Preface

This memoir and the 1:50 000 scale geological map are the products of a mapping contract awarded to the University of Durham by the Natural Environment Research Council. The funding provided by the NERC allowed for ongoing fieldwork to be continued and the completion of further laboratory studies. The contract stems from an NERC policy of encouraging academics to transfer into the public domain data and information that they hold on various parts of the country. The intention was that the detailed field maps and observations in particular be lodged in the National Geosciences Record Centre at the BGS, and that a new 1:50 000 scale map and accompanying memoir be published by the BGS.

The original memoir was published in 1908, an excellent piece of work, full of acute observations and comments, many of which are still valid today. Since that date, and particularly in the period following the Second World War, the geology of Rum and adjacent islands has been the subject of active research by many scientists and this memoir presents a new synthesis of the geology using data from those sources. The memoir benefits greatly from the guidance and detailed local knowledge of Dr C H Emeleus, who has many years of research experience on Rum and the adjacent islands.

Let me make a personal observation here for I first met Henry Emeleus in 1957 when he was a Tutor at University College Durham and I was a first-year student. At that stage I was proposing to major in chemistry but fortunately for me (and perhaps for chemistry!) I was exposed to the inspired teaching of Professor Kingsley Dunham and his staff, including Henry Emeleus who gave a superb series of lectures on igneous rocks in general and layered intrusions in particular. This volume bears further testimony to the talent and dedication of Henry Emeleus and I am confident that it will excite the interest and imagination of readers for many years to come, just as he was able to help to trigger my interest in geology some 37 years ago.

Complementing Dr Emeleus' account of the igneous rocks are contributions by scientists expert in other parts of the geology: Dr Nicholson (Torridonian), Professor Steel (Triassic), Professor Hudson (Jurassic and Cretaceous), Dr Peacock (Quaternary), Professor Bott (Geophysics), Dr Harmon, Dr Boyce and Dr Greenwood (Isotope geology) and Dr Gallagher (Economic geology). Some years ago Professor A C Dunham and Dr W J Wadsworth prepared accounts of aspects of the igneous geology, in preparation for a new edition of this memoir; this material has been of considerable use in the preparation of the accounts of the early, acid magmatism and the ultrabasic rocks. The result is that the new memoir is an extremely valuable and informative account of one of the classic areas within the British Tertiary Volcanic Province.

The geology of Rum and the adjacent islands present a record of the development of a Palaeocene central volcano, and displays the roots of the volcano and its relations with pre-Palaeocene rocks. The peridotites and gabbros of the Rum Central Complex, in particular, have attracted the attention of many researchers over the years and the complex has become a testing ground for theories of emplacement of magmas and the origins of layering in igneous rocks. The earlier sedimentary rocks of the Precambrian Torridon Group and the Mesozoic Era represent the basement on which the volcanic rocks erupted, while small exposures of Lewisian gneiss indicate uplift events within the root zone of the volcano. The exposures of Mesozoic rocks are important in providing visible sections through parts of the Sea of the Hebrides and Inner Hebrides sedimentary basins, which have been foci for oil exploration. During the Quaternary the islands were affected by both Scottish mainland ice and a later local Rum glaciation, resulting in a patchy and thin cover of various glacial and fluvioglacial deposits.

Rum and the adjacent islands constitute an area of outstanding beauty and include some of the finest coastal and mountain scenery in the British Isles. The islands display a wealth of important geological detail, now mainly preserved in areas designated to be Sites of Special Scientific Interest. It is hoped that this memoir will lead to a fuller appreciation of the geological history of an area that provides such excellent displays of the results of igneous processes, and which forms an important part of our natural heritage.

Peter J Cook, CBE, DSc, CGeol, FGS Director Kingsley Dunham Centre British Geological Survey Keyworth Nottingham NG12 5GG

Acknowledgements

The 1: 10 000 scale compilation and partial resurvey of Sheet 60 was carried out by Dr C H Emeleus. Parts of the Torridonian rocks of Rum were resurveyed by Dr P G Nicholson. Dr J D Peacock carried out a reconnaissance survey of the Pleistocene and Recent deposits of Rum, Canna and northern Eigg. Dr M J Gallagher surveyed the olivine and chromite deposits on Rum and organised the offshore survey. The work was performed under the supervision of Dr D I J Mallick, Regional Geologist, Highlands and Islands Group and Dr D Stephenson of this group, who edited the memoir.

The greater part of the present memoir was written by Dr Emeleus. Dr Nicholson described the stratigraphy and sedimentology of the Torridonian rocks (Chapter 3). In Chapter 4, Professor R Steel gives an account of the stratigraphy and sedimentology of the Triassic rocks and the account of the biostratigraphy of the Middle and Upper Jurassic rocks and the Cretaceous beds was written by Professor J D Hudson. The results of oxygen isotope studies on Rum were described by Dr R S Harmon, Dr R Greenwood and Dr A Boyce (in Chapter 7), these also incorporate results from Dr R W Forester, Dr I E G Meighan and Dr Z Palacz. The account of the studies on gravity is by Professor M H P Bott (Chapter 11). The Pleistocene and Recent deposits (Chapter 12) are described by Dr J D Peacock and the Economic Deposits (Chapter 13) by the late Dr M J Gallagher.

The work on Rum has been actively encouraged and facilitated by Scottish Natural Heritage (formerly the Nature Conservancy Council) through the Wardens and Chief Wardens, P Wormell, L Johnston, R Sutton and C T Eatough. During the resurvey, accommodation and other facilities were kindly made available on Canna by the National Trust for Scotland, on Muck by L McEwen and E McEwen and on Eigg by K Schellenberg (the former owner), and C and M Carr. The islets of Oighsgeir were not visited during the resurvey but rock samples were obtained through the good offices of the Northern Lights Commissioners' Principal Keeper, D R McGaw, and R Sutton when Chief Warden on Rum.

A large amount of unpublished information was provided by the British Geological Survey (NERC) through Dr Mallick and Dr Stephenson who are also warmly thanked for much editorial and other advice and Dr P J Henney (BGS, Nottingham) is thanked for his constructive comments about the geochemistry of the Palaeocene igneous rocks. Many persons have generously made available their unpublished work, much of it contained in doctorate and other theses. Particular thanks are due to E A Allwright, J E Andrews, S Braley, A R Butcher, R Coppin, A C Dunham, M Errington, J W Faithfull, R M Forster, R Greenwood, J P Harris, A C Kerr, J E McClurg, R Renner, N J Smith, J A Volker, W J Wadsworth and M I Wakefield.

The work has benefitted from many constructive discussions with B R Bell, G P Black, G M Brown, M J Cheadle, C H Donaldson, A C Dunham, S A Gibson, R Greenwood, R H Hunter, I G Meighan, R S J Sparks, R N Thompson, B G J Upton and W J Wadsworth. Additional chemical analyses were made by R Hardy.

C H Emeleus Department of Geological Sciences University of Durham March, 1996

Notes

Figures in square brackets are National Grid references within 100 km squares NG and NM.

Key to the specimen numbers mentioned in the text:

Example of sample number Collection
S 1250 British Geological Survey. Samples from Scotland
DU 13747 Durham University, Accession Collection
SR 157 C H Emeleus. Research collection1, (Small Isles)
HE 7456 E A Allwright. Research collection1, (Eigg)
HM 7556 E A Allwright. Research collection1, (Muck)
HMD 33 E A Allwright. Research collection1, (Muck)
HC 7543 E A Allwright. Research collection1, (Canna/Sanday)
RG 81 R Greenwood Research collection2 (Rum)
A 421 A C Dunham. Research collection3 (Rum)
C 53900 A Harker collection4

Held at: 1. Durham University; 2. St Andrews University; 3. Leicester University; 4. Cambridge University.

Specimen numbers relating to other collections are explained in the text.

Samples and thin-sections of many of the specimens cited in the text are deposited with BGS Edinburgh.

Geology of Rum and the adjacent island—summary

The Small Isles of Inverness-shire, comprising Rum, Eigg, Muck and Canna, are largely formed of Palaeogene volcanic and subvolcanic rocks which rest on or intrude a basement of Lewisian Complex gneisses overlain by Torridonian and Mesozoic sedimentary rocks. Igneous intrusions form the highest peaks on Rum; basaltic lavas make the flat-topped hills on Canna, Muck, and Eigg; and the serrated ridge of the Sgurr of Eigg is formed by a pitch-stone. The Mesozoic sedimentary rocks form low, fertile ground on Eigg, and also occur on Muck and Rum. Torridonian sandstones make a tract of rough, hilly ground on northern and eastern Rum. Lewisian gneisses are restricted to central Rum.

The late Precambrian Torridon Group strata, laid down as braided river deposits, are predominantly sandstones, sometimes pebbly, with subordinate shales and silty sandstones. Mesozoic strata comprise Triassic sandstones and breccias of fluviatile origin, with cornstones; they crop out on Rum and mark the eastern, feather-edge of the Sea of the Hebrides Basin. Sandstones, shales and limestones of shallow-marine and brackish-water origin on Eigg and Muck are predominantly of Middle Jurassic age; they form part of the Inner Hebrides Basin deposits. Upper Cretaceous sandstone and limestone occur on Eigg.

The earliest manifestations of the Palaeogene volcanic activity were the extrusions of Palaeocene basaltic and hawaiite lavas of the Eigg Lava Formation on Eigg, Muck and south-east Rum. These were followed by the development of the Rum Central Complex, which began with an early phase of caldera-controlled rhyodacitic ash flows and intrusions, together with collapse and explosion breccias, and later granitic intrusions. Uplift and subsidence on ring faults are responsible for the preservation within the igneous complex of Lewisian gneisses, basal members of the Torridon Group, sparse Lower Jurassic strata and some Palaeocene lavas. Numerous basaltic dykes form a regional NW- to NNW-trending swarm across Muck, Eigg and much of Rum and a radial swarm of basaltic dykes, and basaltic cone-sheets, focus on Rum. The final phase of the central complex involved intrusion of pulses of basaltic magma, some notably magnesium-rich, to form the thick core of layered ultrabasic and gabbroic rocks. These rocks are the focus of a pronounced gravity high and a strong magnetic anomaly. During the Palaeocene, vigorous subaerial erosion of Rum exposed the roots of the igneous complex and formed thick river deposits. This period of erosion coincided with the eruption of further basaltic and more-evolved lava flows of the Canna Lava Formation on northwest Rum and Canna, which are now found interbedded with the fluviatile deposits. Igneous activity closed in the Eocene with the eruption of subaerial ash and lava flows of the Sgurr of Eigg Pitchstone Formation on Eigg and Oigh-sgeir.

A major fault beneath the Sound of Rum separates Rum from Eigg. This is considered to be part of the Camasunary–Skerryvore fault system. On Rum, the Long Loch Fault has probably been long-lived and is considered to have controlled emplacement of the ultrabasic rocks. Glaciation during the Quaternary Period produced corries, moraines and U-shaped valleys on Rum, and limited till deposits on all the islands. Elevated marine benches testify to changes in sea level. Landslips and rock falls are extensive in northern Eigg. Marine placer deposits rich in chromite and olivine, derived from the ultrabasic rocks, occur offshore around Rum. (Table 1)

Chapter 1 Introduction

Introduction

This memoir describes the rocks on Sheet 60 and small areas of Sheets 61 and 70 of the 1:50 000 geological map of Scotland. The region covered lies in the Inner Hebrides and includes the islands of Rum (109 km2), Eigg (31 km2), Muck (6.5 km2), Canna (11.7 km2) and Sanday (2.6 km2), and the islets of Humla and Oigh-sgeir, all of which comprise the Small Isles of Inverness-shire (Figure 1). The area overlaps the boundary between the Tiree and Little Minch sheets of the 1:250 000 geological map of Great Britain.

The area is one of outstanding natural beauty and includes some of the finest coastal and mountainous scenery in Great Britain. Rum is a National Nature Reserve owned and managed by Scottish Natural Heritage (the successor to the Nature Conservancy Council). Canna and Sanday are owned by the National Trust for Scotland and worked as a farming community. Eigg is owned and managed by the Isle of Eigg Heritage Trust, a consortium of Eigg residents, the Highland Council and the Scottish Wildlife Trust. It has a small crofting community in the north and there is recent afforestation in the south. Muck is privately owned and is managed as a farm. The geological importance of these islands is indicated by the presence of designated Sites of Special Scientific Importance (SSSIs) on each one (Emeleus and Gyopari, 1992). Access to the islands is by ferry from Mallaig and, in season, by launch from Arisaig.

The marked scenic contrasts between the islands reflect the varied geology. The rugged hills of central Rum dominate the district. The resistant gabbroic and ultrabasic rocks of the Palaeocene central complex forming these mountains culminate in the spectacularly terraced slopes of Askival (812 m) and Hallival (722 m) (see Cover and Frontispiece), where easily weathered feldspathic peridotite layers are sandwiched between resistant allivalite (bytownite-troctolite). In western Rum, smooth-weathering granitic hills are fringed by steep cliffs which rise from a spectacular rock platform at 20–30 m above sea level to over 250 m height along the southwest coast (Plate 10). Precambrian sandstones of the Torridon Group occupy all of northern Rum; this monotonous succession of WNW-dipping beds forms a series of dip and scarp slopes across the island.

The flat-lying Palaeocene lavas and sedimentary rocks of Canna, Sanday and Muck provide a complete contrast to Rum. Northern Canna is margined by steep basalt cliffs, rising to over 200 m in height. Terracing, or trap-featuring is pronounced on the southern slopes, leading down to the softer ground eroded from conglomerates and other sedimentary rocks around Canna Harbour. The south-east of Sanday is noted for cliffs and sea stacks of interbedded lavas and conglomerates (Harker, 1908, fig. 12). Muck is generally low-lying but it too has fine basalt and gabbro sea cliffs on its south coast and sheltered rocky bays and shell sand beaches on the north side. On southern Eigg a thick succession of gently dipping lavas rises north-eastwards to cap the hills in the northern part of the island where spectacular landslips mantle and largely conceal the underlying Mesozoic rocks (Plate 1). The terraced basaltic lavas of southern Eigg are capped by the steep-sided rocky ridge of the Sgurr of Eigg, an Eocene pitchstone flow originally filling a valley system eroded in the lavas. The Sgurr rises steeply to almost 400 m above OD, forming a distinctive landmark visible from many places in the Inner Hebrides and the adjacent mainland (Plate 26).

Regional setting and summary of geology

Rum lies on a NNE-trending ridge of Precambrian rocks that extends from central Skye to Coll, Tiree and the Skerryvore Light. The mountainous area of southern Rum is the site of one of the deeply eroded Palaeocene central igneous complexes which include Skye (Harker, 1904), Mull (Bailey et al., 1924), St Kilda (Harding et al., 1984), Ardnamurchan (Richey and Thomas, 1930), the Blackstones Bank (McQuillin et al., 1975), Arran (Tyrrell, 1928) and others farther afield in Lundy, northeast Ireland and the North Atlantic (Replace (Figure 2); Upton, 1988). To the east, a Mesozoic basin underlying Eigg and the surrounding seas, the Inner Hebrides Basin, is down-faulted against the ridge by faults between Camasunary, Skye and Skerryvore. To the west the feather edge of the Mesozoic Sea of the Hebrides Basin touches the northwest coast of Rum (Figure 1); McQuillin and Binns, 1973; Binns et al., 1974, fig. 2; Fyfe et al., 1993)). Both basins are overlain by Palaeocene lavas. The flows on Eigg and Muck appear to continue to the lava fields of Ardnamurchan and Mull whilst the Canna and Sanday lavas lie on a basalt ridge that extends from south-west Skye to a point about 15 km south-west of Oigh-sgeir (Figure 1). On Rum, lavas in a fault zone in the south-east of the island are correlated with lavas on Eigg but the lavas in the the north-west of Rum are part of the later succession of flows on Canna and Sanday.

The geological succession in the Rum district is summarised in (Table 1). The oldest rocks are Archaean, consisting of feldspathic gneisses and amphibolites of the Lewisian Complex. These occur within the Main Ring Fault of the Palaeocene Rum Central Complex where they are overlain by sedimentary breccias and fine-grained clastic rocks at the base of the Neoproterozoic Torridon Group. The principal development of the Torridon Group is on northern Rum where at least 2500 m of sandstones, feldspathic sandstones, pebbly sandstones and siltstones of subaerial floodplain origin form a succession of WNW-dipping beds. In north-west Rum the Torridon Group rocks are overlain unconformably by sedimentary breccias, cornstones, sandstones and shales of Triassic age, at the featheredge of the thick succession of Mesozoic rocks of the Sea of the Hebrides Basin. A thin sequence of fossiliferous shale, limestone, sandstone and rare ironstone caught up in the Main Ring Fault of eastern Rum has been equated with part of the Broadford Beds of Skye, of Early Jurassic age. A thick development of Jurassic rocks occurs on Eigg where Bathonian and Bajocian sandstones, shales and limestones of the shallow water to deltaic Great Estuarine Group crop out in the north of the island. Part of this succession is also found at Camas Mór on Muck. On Eigg these beds adjoin Oxfordian limestones and shales at the Bay of Laig, where they are overlain by thin Upper Cretaceous sandstones and limestones at Laig Gorge.

Palaeocene intrusive and effusive igneous products form the majority of the rocks exposed on the Small Isles. The earliest activity was on Muck and Eigg where thin tuffs and rare volcanic breccias are succeeded by subaerial lavas of predominantly basaltic composition forming the Eigg Lava Formation. Initiation of the Rum Central Complex followed the lavas. The intrusive activity was strongly influenced by the Main Ring Fault, a system of arcuate faults that was established at an early stage in the growth of the central intrusive complex. Prior to central collapse, there had been pronounced central uplift of between 1 and 2 km. Lewisian gneisses and the basal members of the Torridon Group were raised within the ring-fault. Doming and deformation of the Torridonian rocks took place outside the fault system; the usual low WNW dips change to very steep dips outwards from the central complex. Low-angled faults inclined to the east or ENE at Welshman's Rock and Mullach Ard may have formed at this stage as part of the sedimentary cover in effect slid off the side of a developing dome.

Prior to, or possibly during, uplift intrusion of gabbros and peridotites took place and imparted a strong thermal overprint to the Lewisian gneisses. Fragments of these early basic and ultrabasic blocks are contained in the Am Màm Breccias, formed during a phase of subsidence on the Main Ring Fault. Subsidence was accompanied by caldera formation, eruption of rhyodacitic ash flows, the formation of breccias by explosive activity and collapse of the caldera walls, and granite intrusion. Tuffisites were injected. Jurassic rocks and Eigg Lava Formation rocks subsided on the ring-faults. There followed renewed upift on the ring-faults when Torridonian rocks were reverse-faulted over the Jurassic and Eigg Lava Formation rocks. These events comprise Stage 1 (Table 1).

On Eigg, and especially on Muck, a dense swarm of basaltic dykes intrudes the lavas and earlier rocks. This swarm may also be recognised on the south-east side of Rum where it intrudes early silicic members of the Rum Central Complex but it largely predates the Layered Suite. The peak of basaltic dyke intrusion, including the radial dykes about the Rum Central Complex, marked the change from the dominantly silicic magmatism of Stage 1 to the second stage of the Rum Central Complex, which involved the input of large amounts of mafic magmas into the subjacent crust. The first indication of highly centralised activity in Stage 2 was the intrusion of numerous basaltic cone-sheets which focus on upper Glen Harris. The sheets cut many of the acid rocks and probably postdate all but the latest movements on the Main Ring Fault system. They were followed by intrusion of gabbros and the formation of the layered ultrabasic rocks of eastern and south-western Rum. Emplacement of these rocks, the Layered Suite, was guided in the east by the pre-existing zone of weakness provided by the Main Ring Fault but to the west the fracture occurred within the earlier granitic rocks. Three groups of mafic rock are recognised: the ultrabasic and gabbroic rocks of the Eastern and the Western Layered Intrusions, and a later, transgressive, cross-cutting central group consisting of feldspathic peridotite, ultrabasic breccias, bytownite-troctolite and gabbro (Figure 1). This late group, termed the Central Intrusion, cuts the earlier mafic rocks and the Main Ring Fault. It may be continued into north-west Rum in the line of small peridotite and gabbro plugs that intrude Torridon Group rocks. The coincidence between the Central Intrusion and the Long Loch Fault, which probably has a history of pre-Palaeocene and later movement, and whose last movements postdate the Rum Central Complex, has led to the suggestion that the mafic rocks of Rum were primarily fed from sources along this zone of weakness. The composition of the magmas responsible for the ultrabasic rocks of Rum has been the subject of some debate; the current view is that both basaltic and highly magnesian, picritic magmas were involved. The existence of the latter is necessary to explain the large, positive Bouguer anomaly over the southern part of the island.

Igneous activity did not cease with the emplacement of the Rum mafic rocks. Basaltic and more evolved lavas with interbedded fluviatile conglomerates of Stage 3 form several of the hilltops of north-west Rum. These deposits fill valleys which were eroded in the Torridon Group rocks and the Palaeocene Western Granite intrusion. The inter-lava conglomerates contain a wide variety of clasts including porphyritic rhyodacite granophyre, gabbro and troctolite from erosion of the Rum Central Complex which was unroofed during the Palaeocene. Similar conglomerates occur interbedded with flows of the Canna Lava Formation on Sanday and Canna. There is thus a long and complicated history of activity with clear evidence that the Rum Central Complex was emplaced between the early Eigg Lava Formation, and the later Canna Lava Formation, and that a period of deep, subaerial erosion separated the last two events. The limited radiometric dating of the Palaeocene igneous rocks, and the more detailed palaeomagnetic studies, suggest that the time span involved was at most 5 Ma (c.63–59 Ma) and it is likely to have been less. It should also be noted that lavas on Skye, to the south-west of Glen Brittle, contain clasts of granophyre identical with the Rum granophyre. Since the Skye lavas are intruded by the gabbros of the Cuillin Central Complex, this may indicate that the Rum Central Complex is distinctly older than that of Skye. A few north-trending basalt dykes cut the Canna Lava Formation on Rum and Canna. They have normal magnetic polarity, in contrast to the lavas and intrusions, which are invariably reversedly magnetised. The dykes thus belong to a distinctly later igneous event. They may correlate with dykes in south-west Skye. The youngest igneous rocks are the Eocene pitchstone flows of the Sgurr of Eigg and Oigh-sgeir. On Eigg, large boulders of Torridonian sandstone found in the underlying conglomerates indicate that high ground farmed by Torridonian rocks persisted from the time of formation of the Canna Lava Formation into the Eocene.

Evidence for Quaternary glaciation is found everywhere on the Small Isles. Probably all the land area, with the exception of the highest peaks of Rum, was overlain by ice at the maximum extent of the Scottish mainland glaciation. Local, corrie glaciers were subsequently established in the Rum mountains during the Loch Lomond Stadial. Till is widespread in the valleys of Rum and in parts of the other islands; morainic mounds are common in the north-facing valleys of Rum and raised beach deposits of Late-Glacial and Post-Glacial age occur widely. Extensive blockfields occur on the granitic mountains of south-west Rum. Some of the most spectacular remnants from the glacial period are the marine notches which are developed around the south-west and south-east coasts of Rum (Frontispiece), (Plate 10) and on parts of the north coast of Canna. On Eigg, the numerous landslips around the north and east of the island continue to be active locally to the present. Deposits of peat are widespread and are locally thick in valleys on all the islands.

History of geological investigations

Geological investigations of the Small Isles began in the late 18th and early 19th centuries with observations by, amongst others, Jameson (1800) and MacCulloch (1819; 1824). A later, notable discovery was made by Hugh Miller (1858), who found plesiosaur remains on Eigg. Detailed accounts of the sedimentary rocks, structure and igneous geology were given by Judd (1874; 1878; 1885a; 1885b; 1889) and Geikie (1867; 1871; 1888; 1896; 1897). It was recognised that Rum was the site of a central volcano, although Judd's and Geikie's interpretations differed. The early investigations are fully documented in the first edition of this memoir (Harker, 1908).

It is to Harker that we owe the first comprehensive maps and account of the area. With help from Barrow in northern Eigg, Harker made the Geological Survey's 1:10 560 and original One-Inch (1:63 360) map. The maps and memoir contain a wealth of factual information about the geology of Sheet 60, including detailed descriptions of the Jurassic rocks, the Tertiary lavas and the plutonic igneous rocks of Rum. Amongst the spectacularly layered ultrabasic rocks Harker defined two new rock types, allivalite and harrisite. Some of Harker's interpretations have subsequently been revised; in particular the numerous sills mapped in the lava areas are now recognised to be hard centres to flows (Tomkeieff, 1942; Anderson and Dunham, 1966); gneissose rocks within the Rum Central Complex are not interpreted as deformed magmatic hybrid rocks but as Lewisian gneisses; and the role of lateral thrusting is largely discounted as an explanation of arcuate boundaries between rock types and as the cause of the numerous breccias within the Rum Central Complex. Much of the reinterpretation of Harker's work was made by Bailey (1945, 1956) who, during a short but perceptive visit to Rum with J E Richey made a number of crucial observations. Bailey recognised that coarse clastic rocks in northwest Rum were Triassic rocks and not overthrust Cambrian and Torridonian beds. He also corroborated Tilley's view (1944) that there were Lewisian gneisses on Rum, and demonstrated that these and the basal Torridonian beds were confined within a major arcuate structure which was not a thrust but a ring-fault. Significantly, Bailey calculated that uplift of 1–2 km had occurred within the ring-fault. Many of the breccias were identified as explosion breccias accompanying the emplacement of porphyritic felsite, an association already well known from other central complexes in the British Tertiary Volcanic Province (Tyrrell, 1928; Richey and Thomas, 1930; 1932). Following investigation of the layered gabbros of the Skaergaard Intrusion, east Greenland (Wager and Deer, 1939), Wager and Brown (1951) recognised that there were structural and petrological similarities with the layered ultrabasic rocks of Hallival and Askival on Rum, an interpretation confirmed and refined by Brown (1956), Wadsworth (1961) and Wager and Brown (1968). These investigations, together with studies on the acid rocks of Rum (Black, 1954; Hughes, 1960a; Dunham, 1965a; 1968), marked the start of a large increase in research on many aspects of the Rum Central Complex (e.g. papers in a special issue of the Geological Magazine, 1985). This work has concentrated especially on the superbly exposed and little-altered layered peridotites and gabbros, which have become a testing ground for petrogenetic theories, including those on the origins of layering in igneous rocks (e.g. Wager and Brown, 1968; Sparks et al., 1993), the fluid dynamics of magma chambers (Huppert and Sparks, 1980; Sparks and Huppert, 1984), and the roles of picritic and peridotitic liquids in magma chambers (e.g. Gibb, 1976; Young et al., 1988).

Much work has also been carried out on the Mesozoic sedimentary rocks, especially by J D Hudson and his co-workers (e.g. Hudson, 1960; Harris and Hudson, 1980; Hudson and Andrews, 1987), and on the Pleistocene and Recent geology (e.g. McCann, 1969; McCann and Richards, 1969; Peacock, 1969; Ryder and McCann, 1971).

This memoir presents an updated account of the geology of Rum and the Small Isles based on previous published and unpublished work and selective resurvey of many parts of the islands in the period from 1969 to 1992. In addition to these published and other sources, the advances in our knowledge of the geology of Sheet 60 since the Second World War are also due in no small measure to studies by university undergraduate students engaged in mapping and dissertation projects.

Descriptions of the petrography, mineralogy, geochemistry and other features of rock samples appear throughout this memoir. These are taken from several collections and are referred to by distinctive numbers in the text. The key to these numbers is given in Notes p.xii.

Chapter 2 Lewisian Complex

Introduction

Gneiss outcrops are confined to Rum, where they are almost entirely restricted to areas within the Main Ring Fault (Figure 3). Their close resemblance to mainland banded Lewisian gneisses has long been recognised (Geikie, 1897) but Harker regarded them as hybrids, resulting from the partial intermingling of Tertiary basic and acid magmas (Harker, 1908, pp.105–114). They were recognised to be Lewisian by Bailey (1945, pp.168–170), and Tilley (1944) showed that many of the gneisses had undergone intense thermal metamorphism. The restriction of gneisses, and the basal member of the Torridon Group (Chapter 3), to areas within the Main Ring Fault was a key element in the recognition by Bailey (1945) of major central uplift in the Rum Central Complex.

The close connection between faulting and occurrences of gneiss is evident on the northern and eastern sides of the central intrusive complex, from the Long Loch to Dibidil. The only major exception is the large outcrop of gneiss in the Sandy Corrie area ((Figure 3), locality 1).

Details

Eastern Fiachanis<span data-type="footnote">(Numbers 1–12 refer to the areas indicated in (Figure 3.)</span>

The largest area of gneiss is in Sandy Corrie [NM 370 940], eastern Fiachanis. The principal outcrops are in the cliffs on the northern slopes of Leac a'Chaisteil [NM 368 938], north-east of Ainshval [NM 374 944], in whaleback ridges in the lower parts of the corrie and in and near the headwaters of the streams draining the corrie. Elsewhere, the gneiss is largely obscured by thick moraine. Gneiss is overlain by the Fiachanis Gritty Sandstone Member of the Torridon Group about 500 m southwest of Ainshval [NM 3742 9402]. On the northern cliffs of Leac a'Chaisteil [NM 3722 9388], the gneissic banding is vertical or steeply inclined to the north and strikes between east-west and SE-NW. The western and northern margins of the gneiss are generally separated from gabbros and ultrabasic rocks by a zone of intrusion breccia (cf. Harker, 1908, pl. 2). The gneiss becomes increasingly fragmented to the south, passing into coarse breccias of Palaeocene age (Chapter 5).

Ard Nev (numbered locality 2, 3 in (Figure 3))

Thermally metamorphosed, banded feldspathic gneiss forms an ESE-clipping cap on the summit of Ard Nev [NM 346 986]. About 700 m to the SSE, gneiss crops out in a 300 m-long strip on the north-west side of a prominent gully [NM 3480 9790]. This gneiss (2) is more highly metamorphosed than that on Ard Nev (3) since it is situated between the Western Granite and the Ard Mheall feldspathic peridotites; it is margined by, and involved with spectacular net-veining and intrusion breccia. This area of gneiss and the small cap on Ard Nev summit are considered to be relicts of the roof to the Western Granite, which is elsewhere either fault bounded or intruded by mafic rocks (Dunham and Emeleus, 1967, fig. 3; cf. Harker, 1908, fig. 30).

Priomh-lochs and the Long Loch area (numbered locality 4, 5, 6, 7 in (Figure 3))

Glaciated pavements of E-W-striking, vertical to steep, north-dipping banded feldspathic gneiss and thin amphibolites occur east of Priomh-loch Mór (7) [NM 3712 9871] and between Priomh-loch Mór and Loch Duncan to the north-east (Plate 2). The gneisses are either overlain unconformably by the basal breccias of the Torridon Group or are in faulted contact with these rocks, as at 150 m east of Priomh-loch Mot. [NM 3709 9875]. Other scattered gneiss outcrops form an elongate zone between 100 and 200 m in width west of the Priomh-lochs (6) and a strip of gneiss margins the north-eastern side of the Long Loch (5). The isolated area of thermally metamorphosed gneiss about 180 m southeast of Priomh-loch Beag [NM 3710 9840] is probably continuous with gneiss east of Priomh-loch Mór (7).

Sheared and crushed gneiss is exposed on the west bank of the Kilmory River [NM 3635 9950], [NM 3635 9924], close to the Long Loch Fault. Gneiss also crops out 150 m south-west and 100 m west of the upper bridge over the Kilmory River (4) [NM 3636 9943]. Both areas are surrounded by peridotite and the rocks are intensely altered to pyroxene hornfels; the gneiss from the northerly area is highly magnetic (sample SR449).

Am Màm, Meall Breac and Coire Dubh (numbered locality 8, 9 in (Figure 3); see also (Figure 20))

There are small gneiss outcrops (8) on the northern slopes of Meall Breac [NM 3875 9864] and east of Am Màm [NM 3857 9873] on the south side of the Main Ring Fault. Gneiss contributes fragments several metres in diameter to the Am Màm Breccias on the north of Am Màm. On the northern slopes of Meall Breac [NM 3874 9862], the Am Màm Breccias contain abundant gneiss fragments. Fragments of baked gneiss occur in an narrow, east–west-elongated zone of brecciated and bleached feldspathic sandstone on the north side of Loch Bealach Mhic Néill [NM 3792 9920]. This zone is probably of similar origin to the Fissure Breccias of northern Rum (Chapter 6).

Remnants of crushed, thermally metamorphosed feldspathic gneiss occur in WNW–ESE crush zones east of Allt Slugain a'Choilich (9) [NM 3945 9834]; [NM 3952 9828]. The gneiss is surrounded by steeply dipping, splintery Torridonian siltstones and was presumably thermally altered at depth, and subsequently tectonically emplaced during movement on the Main Ring Fault (Chapter 10).

Beinn nan Stac (numbered locality 10 in (Figure 3); see also (Figure 11))

About 1 km SE of Beinn nan Stac, banded feldspathic gneiss with thin amphibolite layers forms a prominent whaleback outcrop [NM 4035 9345]. A 40–70 m-wide strip of gneiss extends north for about 600 m; this is bounded on the east by the Outer Main Ring Fault, on the west by the Central Main Ring Fault, and by intrusive porphyritic rhyolite to the south. The gneissose banding is steep to vertical, it strikes between NNW and WNW, across the direction of elongation of the outcrop. The gneisses are generally thermally metamorphosed, with signs of partial melting in the felsic varieties. This thermal metamorphic overprint is not found in the adjoining rocks nor in the porphyritic rhyolite; thus, the gneiss was subjected to intense thermal alteration prior to its emplacement at the present level during movements on the Main Ring Fault system.

Dibidil (numbered locality 11, 12 in (Figure 3); see also (Figure 21))

Thin, faultbounded slivers of gneiss crop out on the coast (11) between 600 and 400 m east of the Dibidil bothy [NM 3929 9274] and crushed gneiss is present in a small NE-facing scarp on Cnoc na Cuilean [NM 398 929], close to the Main Ring Fault. Gneiss crops out intermittently in the Dibidil River, about 100–250 m from the coast. In this section it is injected by anastomosing, irregular intrusions of tuffisite. Outcrops of gneiss also occur for about 1.5 km to the west of the Dibidil bothy, on and near the inner, northern side of the Main Ring Fault. These areas, on the southern and south-western slopes of Sgurr nan Gillean, are faultbounded or else margined by the extensive tuffisites and breccia bodies to which they contribute.

Petrography

The gneisses of Rum are divided into three interbanded types of regionally metamorphosed rocks, and a fourth type of thermally-altered rocks (Tilley, 1944). The types are:

  1. Basic hornblende-plagioclase-gneiss of high colour index (amphibolite), with or without quartz, locally with small amounts of garnet or clinopyroxene.
  2. Hornblende-biotite-gneiss, with or without plagioclase, with quartz-bearing leucocratic types.
  3. Granodioritic biotite-gneiss, with microcline perthite, sodic plagioclase and quartz. Plagioclase may be the predominant feldspar.
  4. Assemblages derived from the other three by thermal metamorphism. The principal changes are the replacement of hornblende by pyroxene plus Fe-Tioxides, a change in composition of the biotite, clouding of the feldspars, and ultimately the melting of the felsic parts of the rocks.

Type 2 rocks represent a spectrum between Type 1 and Type 3. The Type 1, 2 and 3 mineral assemblages listed above are characteristic of amphibolite-facies metamorphism. Type 4 gneiss forms the bulk of the samples collected from Rum. They are considered further in Chapter 7.

Hornblende-gneisses and their derivative types are commonest in the southern areas (Sandy Corrie (1), Dibidil (11, 12)) whereas biotite-gneisses and their derivatives predominate in the northern areas.

Gneiss clasts are common in the fluviatile conglomerates interbedded with the Palaeocene Canna Lava Formation flows of north-west Rum (Chapter 8). In contrast to the gneisses within the Main Ring Fault, the gneisses in the conglomerates are either free from any obvious thermal overprint or are only slightly affected. No examples of altered gneisses of either pyroxene hornfels or sanidinite facies have been collected from the conglomerates. The clasts are principally of Type 1 gneiss.

Chemical analyses of a range of Rum gneisses have been reported by Dunham (1968, (Table 7)); these and other representative analyses of the gneisses are given in Appendix 5a. The analysed rocks show varying degrees of thermal alteration. They range in composition from basic (originally amphibolitic) gneiss (mainly Type 1) through tonalitic (mainly Type 2) to granodioritic (Type 3) varieties. All, with the exception of a pegmatite rich in potash-feldspar, show the characteristic dominance of Na2O over K2O.

The gneisses of Rum are of such limited outcrop and are so commonly thermally metamorphosed that no attempt has been made to identify their position in the Lewisian chronology. Geographically, they are situated between little-altered Archaean gneisses recognised on Tiree and Barra, and the southern region of gneisses affected by the Mesoproterozoic Laxfordian metamorphism (Park, 1991, fig. 2.3). The high-grade thermal alteration imposed by the Palaeocene intrusions has led to partial melting of the more felsic varieties and it is probable that partial melts of Lewisian rocks have contributed to the acid rocks on Rum and have also contaminated the mafic magmas (Chapter 9).

Chapter 3 Torridon Group

Introduction

A belt of Middle to Upper Proterozoic sedimentary rocks stretches along the northwest coast of Scotland for about 200 km from Cape Wrath to Rum. These predominantly fluvial clastic deposits (red beds) are collectively referred to as the 'Torridonian'.

Stratigraphically, the Torridonian is subdivided into the Stoer, Sleat and Torridon groups. The Stoer and Torridon groups are separated by a major angular unconformity (Lawson, 1965; Stewart, 1969) and were deposited approximately 1050 Ma and approximately 850 Ma ago respectively (see Stewart, 1988a). The undated Sleat Group, which conformably underlies the Torridon Group, is confined to Skye and the adjacent mainland (Stewart, 1969; 1982; 1988b).

All the Torridonian rocks in Sheet 60 are on Rum and belong to the Torridon Group (Stewart, 1966; 1975; 1988b; Nicholson, 1992a; 1992b). This group is approximately 4 to 5 km thick on the mainland, where it is subdivided into four constituent formations (Stewart, 1969; 1988b). Directly overlying the Lewisian Gneiss basement is the Diabaig Formation, consisting of grey shales and siltstones with subordinate basal breccias and sandstones. These deposits are overstepped by the predominantly coarse-grained sandstones (and local pebble conglomerates) of the Applecross Formation, which in turn gradually pass upwards into more homogeneous medium-grained sandstones of the Aultbea Formation. The overlying Cailleach Head Formation (outcrops of which have not been identified at present on Rum) is composed of grey shale-to-sandstone cyclothems (Stewart, 1988b).

The earliest, detailed published study of the Rum Torridonian strata was by Harker (1903; 1908), who provided detailed maps and descriptions of the sedimentary rocks over the entire island, and subdivided them into two stratigraphical units, namely a 'lower group of dark shales' (427 m in thickness) and an 'upper group of feldspathic sandstones' (2743 m thick). Bailey (1945) revised Harker's lithostratigraphy by identifying a third subdivision, the 'Basal Grit', underlying the lower shale unit. Moreover, Bailey proposed that this 'basal grit' and the 'lower shales' collectively correlated with the former Diabaig 'group' on Skye, and that the overlying 'feldspathic sandstones' were equivalent to the former Apple-cross 'group'. Black and Welsh (1961) subdivided the Rum Torridonian strata into five units based on lithological characteristics (cf. (Table 2) and discussions below). They correlated their four lower subdivisions ('Basal Grit' through to 'Loch nan Eala Arkose' inclusive) with the Diabaig 'group' on Skye, and the remaining 'Guirdil Arkose' with the Applecross 'group' in the same area. However, this correlation was rejected by Stewart (1966) who instead equated the 'Bàgh na h-Uamha Shale' on Rum with the Diabaig Formation on the mainland, and the audha na Roinne Grit' and higher divisions with the Applecross Formation.

Based on detailed field mapping (Nicholson,1992b), a thoroughly revised stratigraphy for the Torridon Group sedimentary rocks on Rum is proposed (Table 2). There are three key reasons why this revised terminology is required:

  1. The previous stratigraphical subdivisions proposed by Black and Welsh (1961; cf. (Table 2)) display a number of major inconsistencies in the field, and have few consistently mappable sedimentological characteristics by which the subdivisions can be defined. These subdivisions, which have neither formation nor member status, should be abandoned.
  2. An accurate assessment of vertical (stratigraphical) trends in sediment lithology, grain size, pebble content and overall depositional style within the Torridon Group can only be made through lengthy, regional traverses covering several hundreds of metres of stratigraphical thickness. Subtle variations in sedimentary rock texture and structure between discontinuous exposures (particularly inland as opposed to foreshore outcrops) over shorter distances often give misleading stratigraphical impressions. This is due to the Applecross and Aultbea formations' relative homogeneity of petrography and facies (a 4 to 5 km thickness of coarse- to medium-grained cross-stratified sandstone with a variable pebble content), together with the obvious difficulties associated with lack of chronostratigraphical control in Proterozoic fluvial successions, and the additional structural complications on Rum resulting from Tertiary igneous activity.
  3. The formation names 'Diabaig', 'Applecross' and 'Aultbea' are applied directly to Rum in this study. Usage of these terms is justified in that they are already well established, and equally importantly, they provide a framework through which comparisons with the mainland Torridonian succession can be made.

Diabaig Formation

Fiachanis Gritty Sandstone Member

The Fiachanis Gritty Sandstone Member consists primarily of coarse-grained sandstone with subordinate granular sandstone and basal breccia. Exposures are restricted to localities within the Main Ring Fault and consequently the rocks have been faulted, brecciated and thermally metamorphosed, commonly to the point of being partially melted. Explosion breccias, megabreccias and other volcaniclastic breccias (which occur prominently within the Rum Central Complex; Chapter 5) commonly obscure original depositional features and stratigraphical relationships. In the type locality in upper Fiachanis [NM 3740 9396] to [NM 3750 9396], a 35–70 m-thick stratigraphical succession through the member is exposed, including contacts with both the underlying Lewisian Gneiss and the Laimhrig Shale Member above. The unconformable contact with the basement is also visible near the Priomhlochs [NM 3710 9875] and at Dibidil [NM 3953 9277].

Breccias and coarse (granular) sandstone are found at the base of the member, either directly overlying, or in the immediate vicinity of, exposures of Lewisian Gneiss. The breccias are matrix-supported, and contain angular clasts of gneiss typically 10–15 mm in size but reaching a maximum of 30–40 mm. The granular sandstones contain an abundance of white angular quartz grains up to 10 mm in size. Bedding is commonly crude and difficult to recognise within these basal deposits; where discernable, beds are massive, of highly variable thickness (<10 cm to more rarely >100 cm thick), and exhibit discontinuous bedding planes with either curved and non-parallel or broadly planar and parallel geometries (the descriptive terminology of Collinson and Thompson, 1989, figs. 2.3, 2.6 and 6.14, is used for bedding and lamination). Trough cross-stratification is often visible within the granular sandstones, with sets reaching a maximum of 15–20 cm thick. Palaeocurrent directions are highly variable. Breccias and granular sandstones are not always present at the base of the member, and coarse-grained sandstone may instead be observed directly overlying the basement gneiss, as near the Priomh-lochs [NM 3694 9903].

Moving away from the basement unconformity in either a vertical or lateral direction, the maximum clast size and matrix grain size of the basal breccias and the granular sandstones decrease, and these deposits grade over a short distance into the coarse-grained sandstones which constitute the majority of the Fiachanis Member. The coarse-grained sandstones are also characterised by an abundance of white angular quartz grains, 1–2 mm in maximum size, which protrude noticeably on the buff-grey weathered surfaces. Thin (<10 cm thick) parallel-bedding to parallel-lamination is typical; ripple cross-lamination (1–2 cm thick), small-scale cross-bedding (3–5 cm thick) and rarer 5–10 cm-thick sets of trough cross-strata are also visible. Gradually towards the top of the member, the sandstones become more medium grained, the 'gritty' (coarse- to very coarse-grained) beds become scarce, and parallel- and cross-lamination predominate.

The boundary between the Fiachanis Gritty Sandstone Member (TCDF on the 1:50 000 map) and the overlying Laimhrig Shale Member (TCDL) is gradational, but it is also highly variable in Ethological character. The upwards transition may occur over several tens of metres of strata, where a thick succession of medium-grained sandstone with sparse gritty bands (TCDF) is interbedded with fine-grained sandstones and subordinate shales (TCDL), as observed near the Priomh-lochs [NM 3695 9905]. Structural complications within the central complex commonly make it difficult to assign such successions to either member with certainty (exposures east of Cnapan Breaca c.[NM 3980 9750] are a typical example). Alternatively, coarse-grained, gritty sandstones (TCDF) may grade upwards over several metres into beds of shale and fine sandstone (TCDL), as in Fiachanis [NM 3730 9378].

Laimhrig Shale Member

This member consists of laminated dark grey 'shales' (composed of laminated mudstone, siltstone and very fine-to fine-grained sandstone) with intercalcated beds of light grey fine-grained sandstone (Plate 3a). These deposits are best exposed along the south-east coast of Rum from Bàgh na h-Uamha [NM 421 972] to Dibidil [NM 4010 9295], where they are gently dipping and essentially unmetamorphosed. In the type locality [NM 4200 9700] to [NM4186 9604], the member is at least 275 m thick (its base lies below mean low water). Several exposures of the Laimhrig Shale Member also occur immediately adjacent to the Main Ring Fault and within the central complex; here, the sedimentary rocks have been subjected to varying degrees of induration, faulting, brecciation and partial melting.

The typical dark grey Laimhrig Member 'shales' consist of two interbedded lithofacies: very thin to thin (<1–3 mm) alternating parallel laminae of mudstone and siltstone (commonly arranged as graded, fining-upwards silt–mud couplets) constitute the finer lithofacies; wave-rippled, parallel-laminated and/or massive (structureless) very fine- to fine-grained sandstone constitutes the coarser. These two lithofacies are thinly interbedded (beds of either being predominantly 1–5 cm thick), and display parallel and planar to wavy bedding planes. Intercalcated with these 'shales' is a third lithofacies consisting of light grey, fine-grained sandstone beds; these typically range in thickness from 5–20 cm in exposures within and immediately adjacent to the central complex, for example on the Dibidil path about 450 m south of Allt na h-Uamha [NM 4110 9640], to between 20 cm and 1 m in stratigraphically higher exposures within the type locality. Together with this increase in bed thickness, the fine sandstone beds also become progressively more abundant towards the top of the member. The beds have a sharp erosional base with rare small load casts (<5 cm) and either a sharp or gradational fining-upwards top (the latter grading upwards into silty sandstone). Internally, they are either massive, parallel-laminated and/or ripple-drift cross-laminated (ripples 5–10 mm in size). Palaeocurrents measured from current ripples in the light grey fine-grained sandstones indicate a south to south-westerly transport direction.

Depositional environment of the Diabaig Formation

Sedimentation of the Diabaig Formation was largely controlled by the irregular topography of the underlying Lewisian Gneiss basement. Breccias and gritty to coarse-grained sandstones of the Fiachanis Member were deposited by channelised flows and sheetfloods on alluvial fans draping the lower slopes of hills of gneiss. The lithology and texture (angular clasts and grains) of these deposits indicate that the sediment was derived locally from the gneiss, and that transport distances were short. Laterally down-palaeoslope, these fans fed into lakes which formed in valleys and topographic lows on the basement surface (Stewart, 1982; 1988b; Stewart and Parker, 1979). The laminated silt–mud couplets of the Laimhrig Member are thought to represent varves or rhythmites, possibly seasonal, which accumulated in the deeper water (below wave base) of one such lake. Thinly interbedded very fine- to fine-grained sandstones (which constitute the coarser lithofacies of the shales) were likely deposited by storm events (wave ripple lamination) or by density currents during periods of increased alluvial discharge into the lake from the surrounding fans (Rodd, 1983). The variable nature of the stratigraphical boundary between the Fiachanis Member and the Laimhrig Member is a result of varying amounts of interdigitation between their respective depositional environments. The light grey fine-grained sandstones interbedded with the shales in the Laimhrig Member, which increase in both thickness and abundance towards the top of the member, are interpreted as turbidite deposits triggered by discharge into the lacustrine basin from rivers encroaching from the west at the start of Apple-cross Formation time (Rodd, 1983; Stewart, 1988b).

Applecross Formation

Allt Mór na h-Uamha Member

This member consists of cyclically interbedded 'siltstones' and fine- to medium-grained cross-bedded sandstones (Plate 3b). A complete section through the member is exposed in the type locality in the Allt Mór na h-Uamha [NM 4205 9722] to [NM4108 9751], where its strati-graphical thickness is 415 m. Previous workers have assigned part of this member to the Diabaig 'group' (Bailey, 1945, plate VIII; Black and Welsh, 1961; Stewart, 1966). However, owing to their physical similarity to the overlying fluvial arkoses, these deposits are all assigned to the Applecross Formation in this study (also cf. Rodd, 1983, who defined an analogous member at the base of the Applecross Formation at the Diabaig Formation type locality). Exposures of this member are absent around Rubha Port na Caranean [NM 427 985] due to the easterly down throw on the Mullach Ard Fault.

The Allt Mór na h-Uamha Member can be subdivided broadly into two lithofacies ((Figure 4)a. The finer-grained lithofacies consists of laminated and thinly lenticular to wavy bedded 'siltstones' (actually composed of siltstones and very fine- to fine-grained sandstones with subordinate mudstones), which are collectively interbedded with light grey fine-grained sandstone beds typically 10–50 cm thick; current ripples and parallel-lamination are common (wave ripples are subordinate); the fine-grained sandstone beds commonly display low-angle (<5–10°) accretion surfaces internally, and small-scale load casts at their base. The coarser-grained lithofacies consists of tabular, very thickly bedded fine to medium-grained sandstones with an orange-pink to red-brown weathered surface; cross-bedding is common (both planar-tabular and trough types), soft sediment convolutions are abundant and horizontal lamination displaying parting lineation on plan surfaces is present but is rare; the tops of many tabular sandstone beds are current-rippled. These coarser-grained beds are typically 2–6 m thick and occur as either single-or multi-storey sandbodies, the latter reaching a maximum of 45 m thickness at Bàgh na h-Uamha [NM 4202 9722]. Palaeocurrents within both the coarser- and finer-grained lithofacies are generally towards the south to south-east.

Interbedding of the coarser- and finer-grained lithofacies of the Allt Mór na h-Uamha Member produced both coarsening-upward and fining-upward cycles, each several metres in thickness ((Figure 4)b). The cross-bedded sandstone beds of the coarser-grained lithofacies increase in frequency and average thickness upwards through the succession. Towards the top of the member, these sandstones become coarse-grained and more feldspathic, until ultimately the arkoses of the overlying Scresort Sandstone Member prevail. Stratigraphically, the base of the Allt Mór na h-Uamha Member is defined by the lowest thickly bedded, cross-stratified (or convoluted) fine- to medium-grained sandstone bed greater than 2 m in thickness. The top is defined at the base of the lowest thickly bedded, cross-stratified (or convoluted) coarse grained sandstone (arkose) containing common 'exotic' pebbles characteristic of the Applecross Formation (Williams, 1969; Stewart, 1969; 1988b).

Depositional environment of the Allt Mór na h-Uamha Member

The Allt Mór na h-Uamha Member is interpreted as fluviodeltaic in origin, deposited as the prograding Applecross Formation rivers reached the Diabaig Formation lacustrine basin in the Rum area. Delta progradation and abandonment produced coarsening- and fining-upward cycles respectively. Sedimentation of the coarser-grained lithofacies occurred in distributary channels (cross-bedded sandstones) and on crevasse splays (horizontal lamination with parting lineation) collectively on subaerial delta lobes. The finer-grained lithofacies was likely deposited by turbidites (current ripples and parallel lamination) and surface waves (wave ripples) in either a prodelta or a delta slope setting (the latter is indicated by low-angle accretion surfaces in the fine-grained sandstones), or alternatively, it was deposited on the delta lobes themselves during periods of lacustrine incursion associated with delta abandonment and subsidence. Fluviodeltaic cycles are a characteristic feature of the base of the Applecross Formation in other southern Torridon Group localities (e.g. Scalpay, Raasay and Diabaig; Nicholson, 1992b).

Scresort Sandstone Member

The majority of the Torridonian on Rum belongs to the Scresort Sandstone Member, which crops out over most of the northern half of the island. These deposits are particularly well exposed in their type locality along the northern shore of Loch Scresort [NG 4230 0014] to [NM 4085 9982]. The Scresort Sandstone Member includes exposures formerly assigned to the Rudha na Roinne Grit, the Loch nan Eala Arkose and the lower part of the Guirdhil Arkose (Table 2); its total stratigraphical thickness is estimated at 2000 ± 500 m (allowing for uncertainty in fault displacements).

Medium- to coarse-grained sandstones dominate this member ((Figure 5), Plate 3c); minor deposits of mudstone, siltstone and fine sandstone are also present. The sandstones are thickly to very thickly bedded, with individual beds typically 1–3 m thick; multistorey sand bodies (consisting of two or more genetically related beds) reach a maximum thickness of 6 m at Camas Pliasgaig [NG 3924 0346]. Petrographically, the sandstones contain an abundance of relatively fresh feldspar grains (primarly micro-dine, together with other alkali feldspars) and are thus classified as arkoses. These feldspars, combined with an oxidised ferruginous clay matrix, give the sandstones their orange brown to reddish brown colour. The coarse-grained sandstones contain well-rounded 'exotic' pebbles which are diagnostic of the Applecross Formation (Williams, 1969; Stewart, 1969; 1988b). These pebbles are typically 10–30 mm in size (reaching a maximum of 64 mm at Allt Rubha na Moine [NG 3828 0413]), and consist of a variety of lithologies including vein quartz, chert, various fine-grained volcanic rocks, quartzite and greywacke (cf. Stewart, 1966 for further details). Some individual beds of medium-grained sandstone are very micaceous (e.g. at [NM 4189 9994]) and/or may contain prominent dark laminae of detrital heavy minerals (e.g. at [NG 3924 0346]).

Cross-stratification is abundant in the Scresort Sandstone Member (cf. (Figure 5)). Trough cross-sets are most common, with set thicknesses typically between 5 and 30 cm and widths between 30 and 300 cm; these cross-sets usually occur in cosets that are 50–300 cm thick. Compound (downcurrent-dipping) cross-sets are typically 5–50 cm thick, arranged in cosets 1–3 m in thickness. Isolated sets of planar-tabular cross-strata occur less commonly, and are generally 75–150 cm thick. Although individual sets or cosets of cross-strata may display highly variable palaeocurrent directions (Figure 5), overall they yield a mean palaeoflow direction towards the east to south-east for the entire Scresort Sandstone Member (Nicholson, 1992a; 1992b; 1993). Other (minor) sedimentary structures within these sandstones include primary current lineation (on plan surfaces of horizontal stratification) and small-scale current ripples (<2 cm in height).

Soft-sediment deformation structures ('water escape' or 'quicksand' structures of secondary origin) are ubiquitous in both the Scresort and Sgorr Mhór sandstone members. Several types of structure are visible, including vertical intrusions or 'cusps' of sediment reaching several metres in height, a variety of folds ranging from complex convolutions to gentle synclines that are commonly linked with sharp anticlines (cusps), and overturned cross-strata (recumbent foresets). Most structures show that the predominant direction of sediment movement was vertical and that deformation occurred by the upward explusion of water (Selley, 1969; Nicholson, 1992b; 1993). The majority of structures also display only one stage of deformation and are commonly confined to the upper part of individual sand bodies (Nicholson, 1992b; 1993). Nicholson (1992b; 1993) has suggested that these structures were produced by a two-stage process consisting of liquefaction triggered by vigorous fluvial currents, followed by fluidisation caused by the upward expulsion of porewater.

Depositional environment of the Applecross Formation

Deposition of the Applecross Formation is likely to have occurred on a major fluvial braidplain (Nicholson, 1992a; 1992b; 1993) within rivers similar in scale and appearance to the present-day South Saskatchewan River (cf. Cant, 1978; Cant and Walker, 1978) and Platte River (cf. Smith, 1971; Crowley, 1983). The Scresort Sandstone Member represents the deposits of channel-floor dunes (trough cross-stratified cosets), transverse bars (solitary sets of planar cross-strata), larger compound bars (two or more compound cross-stratified sets or cosets) and bar-top bed-forms (small-scale cross-strata overlying transverse or compound bars) (Nicholson, 1992b). The occurrence of horizontally stratified, current-lineated medium-grained sandstone suggests that overbank deposition on the floodplain was quite significant. The deposits on Rum suggest that the Applecross Formation rivers were typically 2–3 m deep, but may have reached a maximum depth of over 6 m (Nicholson, 1992b). These rivers may have commonly been 500–1000 m wide (Nicholson, 1992b; 1993), given that Applecross Formation sandbodies typically display sheet geometries (cf. Friend, 1983, fig. 6.) and were likely between 200 and 500 km+ in length (Nicholson, 1992b; 1993). Overall, it is thought that the deposits of the Scresort Sandstone Member on Rum represent only a fragment of what was once an extensive fluvial basin, stretching from Cape Wrath in the north, past Tiree in the south, to beyond the Outer Hebrides in the west. This basin was filled with sandy, braided rivers flowing to the east to south-east (Nicholson, 1992b; 1993, fig. 9).

Aultbea Formation

Sgorr Mhór Sandstone Member

Outcrops of the Sgorr Mhór Sandstone Member are restricted to the western and southern regions of Rum, owing largely to their stratigraphical position at the top of the westwardly dipping Torridon Group succession on the island. These deposits are assigned to the Aultbea Formation (Nicholson, 1992b) on the basis of their resemblance to equivalent deposits on the mainland. In its type locality on the west of Bloodstone Hill ( [NG 3149 0119] to [NM 3010 9962]), this member is separated from the underlying Scresort Sandstone Member by the subvertical Bloodstone Hill Fault (Chapter 10). A continuous section between the two members occurs farther north between Glen Shellesder (e.g. [NG 3361 0175]) and the nearby coastline [NG 3300 0250]. At the southern end of the island, Sgorr Mhór sandstones occur in a structurally downthrown position relative to those of the type locality, due to displacement on the Main Ring Fault and the Long Loch Fault. The minimum thickness of the Sgorr Mh6r Sandstone Member (measured from [NM 3698 9106] to [NM 3659 9172]) is 175 m.

Although the Sgorr Mhór Member sandstones represent an uninterrupted continuation of the underlying Scresort Sandstone Member, they display distinct differences in their depositional character ((Figure 6), (Plate 3d). The sandstones are predominantly fine to medium grained (coarse-grained sandstones are subordinate, as are siltstones and mudstones) and, more significantly, they do not contain any of the 'exotic' pebbles found within the Applecross Formation. Petrographically, the Sgorr Mhór Member sandstones vary from arkoses to lithic arkoses; in addition to an abundance of feldspar, these deposits commonly contain a substantial amount of detrital muscovite and detrital heavy minerals of silt to fine sand grade. Typically, the heavy minerals are concentrated to form dark laminae ((Figure 6)) or dark beds up to 1 m thick, the latter of which, at Rubha nam Meirleach [NM 3690 9115], were previously mistaken as shale beds belonging to the Diabaig Formation (see Bailey, 1945, plate VIII, and Black and Welsh, 1961, fig. 1). Current-lineated horizontal stratification and very low-angle cross-stratification (<5°), together with soft-sediment deformation structures (described above) are the most common structures within the member (Plate 3d). Trough, compound and planar cross-strata are also present, but are not as prevalent as in the Scresort Member sandstones. In summary, the transition between the Scresort and Sgorr MUT members, which occurs over several tens of metres of stratigraphical thickness, is characterised by a gradual decline in the sediment grain size, an increase in the relative proportion of horizontal stratification, and a significant decrease in the 'exotic' pebble content; the stratigraphical boundary between the two members is marked by the disappearance of these 'exotic' pebbles characteristic of the Applecross Formation.

Depositional environment of the Aultbea Formation

The Aultbea Formation on Rum was deposited in a braidplain setting, but by a fluvial system with at least three key differences relative to that postulated for the Applecross Formation. Firstly, the prominence of horizontal stratification and low-angle cross-stratification within the Sgorr Mhór Sandstone Member suggests that the Applecross Formation rivers either became shallower during Aultbea Formation time, or alternatively, became less numerous and that more sedimentation occurred in overbank areas during flooding. Secondly, the mean palaeoflow direction of rivers during deposition of the Sgorr Mhór Member was towards the south to south-east (Nicholson, 1992b), as opposed to towards the east to south-east during deposition of the Scresort Sandstone Member. Thirdly, the sediment load of the Aultbea Formation rivers was finer grained and probably contained a higher percentage of suspended material. Overall, outcrops of the Sgorr Mhór Sandstone Member on Rum appear to represent the most distal deposits (in terms of depositional environment) and the most mature deposits (in terms of sediment texture and composition) of the Torridon Group fluvial system (Nicholson, 1992b).

Chapter 4 Mesozoic rocks

Introduction

Small areas of Mesozoic sedimentary rocks crop out in western Scotland from Arran to Sutherland and thick successions are present in several offshore basins (Johnstone and Mykura, 1989, fig. 37). Within Sheet 60, sedimentary rocks of Triassic, Jurassic and Late Cretaceous age form small outcrops on several islands (Figure 7) and they are also present offshore (Binns et al., 1974, fig. 2; Fyfe et al., 1993).

Jurassic rocks were described from Eigg and Muck, and a small outcrop of Cretaceous beds was discovered on Eigg during the original survey (Harker, 1908). Subsequently, sandstones, conglomerates and nodular limestones in north-west Rum, which had been interpreted as Torridonian overthrust on Cambrian rocks, were shown to be fossiliferous Triassic beds resting on eroded Torridonian sandstones (Bailey, 1945). Hudson (1960) recognised the Cretaceous age of an outcrop on Eigg, formerly regarded as Jurassic. Small areas of marmorised limestone and calc-silicate rocks in south-east Rum, which are preserved as downfaulted slivers within the Main Ring Fault system (Chapters 5 and 10; Emeleus et al., 1985), were originally correlated with Archaean limestones on Tiree (Hughes, 1960b). However, these were later found to be fossiliferous and of Jurassic age (Dunham and Emeleus, 1967); they are correlated with the Lower Jurassic Broadford Beds of Skye (Smith, 1985).

The Rum Triassic and Jurassic rocks form part of the eastern, feather edge of the Mesozoic succession in the Sea of the Hebrides Basin. They are separated from the younger Jurassic rocks of Eigg and Muck which form part of the fill of the Inner Hebrides Basin ((Figure 1); Binns et al., 1974). General reviews of the Mesozoic rocks were given by Hudson (1983), Steel (1977) and Hallam (1991). The history of subsidence and sedimentation during the Jurassic was discussed by Morton et al. (1987) and Morton (1989).

Triassic

Investigations of the rocks of possible Permo-Triassic age in the Hebridean province (Steel, 1971; 1974a; Steel et al., 1975; Steel and Wilson, 1975) suggest that the earliest Mesozoic sedimentation took place in a number of NE-trending continental basins (Figure 8). Palaeocurrents from the Rum Triassic sequence indicate sediment transport and palaeoslope mainly towards the north and north-west, emphasising that the succession belongs to the Sea of the Hebrides Basin and is, therefore, more closely related to the Triassic rocks of Raasay and the Minch than those farther south in Mull and Ardnamurchan (Steel, 1977). In most of the Hebridean basins, faulting is likely to have been important in controlling the basin margins and in allowing the development of wedges of conglomerate which thin towards the basin axis. In this context the NE-trending monoclinal structure of the north-west coast of Rum, with its evidence of contemporaneous tectonic instability, is of considerable interest. It is possible that the monocline divides mainly alluvial floodplain from mainly alluvial fan sequences on Rum and was a part of the margin control of the Sea of the Hebrides Triassic basin.

Rocks of Triassic age, the Monadh Dubh Sandstone Formation, occur in two outliers that total some 1.5 km2 in area on Monadh Dubh [NG 340 030], north of Glen Shellesder, north-west Rum (Figure 7), (Figure 9). They were regarded by Harker (1908) as overthrust Torridonian and Lower Palaeozoic strata but were recognised as Triassic by Bailey (1945). Fossils of possible Late Triassic age, suggesting a correlation with the Mercia Mudstone Group or Penarth Group of more southern areas, have been recovered from the topmost part of the succession.

The Monadh Dubh Formation consists largely of clastic and continental carbonate sedimentary rocks. The clastic rocks are dominantly coarse grained in the lower part of the succession but become medium to fine grained in the upper half (Figure 10). It comprises four members the distribution of which are shown in (Figure 9); (Figure 10) is a composite section constructed from exposures at localities 2 and 10 (Figure 9).

The Shellesder and A' Mharagach members, which may be in part laterally equivalent, overlie a thick basal cornstone which penetrates downwards into Torridonian sandstones (Plate 4). Amongst the fossils found in the uppermost strata, the Allt Dubh Member, poorly preserved plant remains are abundant and occur with ostracods at locality 10.

Details

North-Western Outlier

Shellesder Member

The Shellesder Member is best exposed at the southern edges of this outlier, on a prominent east–west ridge on the northern side of Glen Shellesder, and has a maximum exposed thickness of 16 m near localities 4 (the type locality, [NG 3335 0255]) and 5 (Figure 9).

A basal cornstone (Figure 10) permeates the joints and bedding planes of the underlying Torridonian sandstone for up to 3 m (Plate 4). The overlying conglomerates are flat bedded and relatively coarse grained. Measurements of maximum particle size and bed thickness of all the conglomerates in this area show that the sequence becomes finer grained between localities 4–5 and locality 2, i.e. from east to west. The conglomerates are usually poorly sorted, but thin lenses of well-sorted, fine-grained conglomerate occur at the top of many of the main conglomerate units. Pebble imbrication is seen in several beds, but cross-stratification is rare. Some conglomerates display an unusual fabric in which blade and rod-shaped clasts, completely supported in a sandstone matrix, stand vertically with respect to the subhorizontal bedding planes. The pebbles are usually of Torridonian sandstone although corn-stone pebbles are important locally.

The sandstones are arkosic and are commonly thinner and less persistent laterally than conglomerates with which they are interbedded. In places they show low-angle cross-stratification which consistently indicates transport to the north-west.

Cornstone, or concretionary carbonate, occurs on the basal unconformity and at other levels within the conglomerate/ sandstone sequence (Figure 10). These upper cornstones show clearly that the carbonate has replaced quartz and feldspar grains of the host sediment, and is therefore authigenic (Steel, 1974b).

The member in the southern area appears to thin westwards, in the same direction as the fining of the conglomerates, and the direction of sediment transport as indicated by cross-stratification and pebble imbrication.

In the northern part of the outlier (Figure 9), occurrences of the Shellesder Member are sparse and are commonly confined to gullies on the Torridonian surface. In the stream south of A' Mharagach, 500 m inland, some 10 m of coarse conglomerate with angular pebbles fills an erosional hollow in the Torridonian (locality 9 [NG 3421 0329]). In this area there is evidence for fault or fold movement in the Torridonian basement during or immediately after deposition of the Triassic beds. Torridonian sandstone is seen near sea level at localities 6 and 7 but crops out at 100 m above sea level at locality 8. This suggests the presence of a monoclinal structure which also occurs some 2 km to the south-west (locality 2). Conglomerates of the Shellesder Member are not present at the coast, but where seen at localities 8 and 9 show synsedimentary deformation in the sandier horizons. This may be an additional indication of unstable conditions during deposition of the conglomerate.

In their poor sorting, local clast content and lack of cross-stratification, the rocks of the Shellesder Member have the character of debris-flow deposits (Steel, 1974a). Gravel deposits with vertical fabric elements, like those described above, have been observed in Recent debris flows (Bull, 1964, p.23; Lindsay, 1968). In addition, the conglomerates of the formation appear to become finer grained away from the source in a manner well known for debris-flow and stream-flood deposits on recent and ancient alluvial fans (Bluck, 1964; 1967). These features, together with the north-westward thinning of the beds, suggest that the conglomerates represent a wedge-shaped alluvial fan sequence which thickens and then onlaps towards a Torridonian sandstone source area to the south-east.

A' Mharagach Member

In the western part of the outlier some 40 m of red beds, in which sandstones or conglomeratic sandstones alternate with fine-grained sandstones and corn-stones, overlie the Shellesder Member, and constitute the A' Mharagach Member. At the type locality (locality 2, (Figure 9); [NG 3300 0265]), the sequence comprises some 15 m of upward-fining cyclothems of the type widely recorded in ancient alluvial deposits (e.g. Allen, 1970). The sandstones and conglomeratic sandstones which constitute the coarse-grained parts of the cyclothems are arkosic and commonly show parallel lamination and large-scale cross-stratification. The finer-grained sandstones are deeper red in colour and commonly exhibit ripple lamination. The cliff section exposing the Shellesder–A' Mharagach succession at locality 2 is of considerable interest. Viewed from the beach the NW-dipping beds of the Shellesder Member are seen to descend from the hillside at a height of 100 m above OD to sea level within a distance of less than 200 m. Together with the underlying Torridonian strata, the Shellesder Member forms a NE-trending, NW-facing monoclinal structure, with some possible minor faulting associated with the steep limb. The A' Mharagach Member does not appear to conform to the folding but shows some evidence of being banked against the steep limb of the monocline. A possible explanation is that faulting or folding movements took place in the Torridonian basement after the deposition of the basal facies but prior to the deposition of the A' Mharagach Member.

In the northern part of the outlier, the best exposures (localities 6 and 7) are in stream sections. Here the A' Mharagach Member, resting directly on Torridonian sandstones, comprises a succession of fining-upward cyclothems, each capped by corn-stone. As in the southern part of the outlier, palaeocurrent vectors indicate transport towards the north and north-east. Although detailed correlation between sections in the north and south of the outlier is not possible, it is noteworthy that the unusually thick cornstone present at the top of the member at locality 2 is also present at locality 6. At both localities this corn-stone is overlain directly by the quartzite-bearing basal conglomerate of the Sgaorishal Member.

As already noted, the A' Mharagach Member is composed of fining-upward cyclothems similar to those recorded from many ancient alluvial sequences. Such cyclothems are generally accepted as representing episodes of river-floodplain construction; the coarse-grained components accumulated as bars in river channels and the fine-grained sediment as overbank deposits on the river floodplain. The cornstone at the top of the individual cyclothems is not so commonly recorded in ancient alluvial sequences but probably represents prolonged periods of subaerial exposure and pedogenesis between successive flood-plain construction events (Steel, 1974b). The contrast in lithofacies and in sediment transport direction between the A' Mharagach and Shellesder members suggests that the former may have accumulated as floodplain sediments in a valley system sloping to the north-east, and that the latter represents alluvial fans which built out in a westerly and north-westerly direction onto these floodplains from highlands to the east.

Sgaorishal Member

The base of this member, exposed atlocality 7 of (Figure 9), is marked by a cross-stratified conglomerate which contains abundant pebbles of cornstone and quartzite (cf. Bailey, 1945, p.172) in addition to clasts of Tonidonian sandstone. It is generally better sorted than the conglomerates of the Shellesder Member but is unusually coarse grained with some clasts reaching a maximum diameter of150 mm (large cobbles). The coarseness and the large number of cornstone pebbles suggest vigorous erosion by strong currents of the underlying cyclothem of the topmost A' Mharagach Member. The presence of quartzite pebbles, possibly derived from the basal Cambrian of the west Highlands, indicates an important new provenance.

The deposits overlying the basal conglomerate consist of thinly bedded, cross-stratified and flat-laminated, white and green sandstones, with subordinate green siltstone and well-sorted, quartzite-bearing conglomerate. Ripple marks and climbing ripple lamination are relatively common; cornstone is notably absent. These beds are accessible in low cliffs immediately north-east of locality 2.

In the northern part of the outlier, at locality 7, coarse-grained conglomerates at the base of the member infill channels up to 20 m wide and 5 m deep which cut down into the A' Mharagach Member. As at locality 2, the basal conglomerate of the Sgaorishal Member is overlain by thinly bedded, cross-stratified sandstones and siltstones.

The conditions of deposition of the Sgaorishal Member are less clear than those of the A' Mharagach Member, but the conglomerates are probably fluvial on account of their cross-stratification, sorting and channel form characteristics. The absence of red coloration and of cornstone in the associated sandstones may anticipate the upwards change from continental conditions into the semimarine and marine conditions of the Allt Dubh Member. A change in sediment provenance from that of the underlying beds is indicated by the quartz-rich sandstones and quartzite-bearing conglomerates. Cross-stratification indicates transport towards the east and north-east.

Allt Dubh Member

The highest beds in the succession are fossiliferous and some 18 m thick. They are best seen at the type locality in the second stream south of A' Mharagach (locality 10, (Figure 9); [NG 3403 0300] ). The lower part of the member comprises grey, well-bedded sandstones, with abundant plant and wood fragments (up to 18 cm in length) and a few small shells. The upper part, 3–4 m in thickness, comprises calcareous sandstones with thinly bedded calcareous siltstones and mudstones and contains abundant fish teeth, fish scales, ostracods and small, indeterminate shells. Bailey (1945, pp.172–173) recorded Euestheria minuta from these beds, indicating a possible Late Triassic age.

The Allt Dubh Member crops out near to the base of the Triassic sequence in the east and south-east. Taking into account the dip of these and adjacent beds it appears that they are underlain by a relatively thin succession, mainly of the Shellesder Member. The fossiliferous rocks may, therefore overstep the Sgaorishal and A' Mhagarach members towards the south-east, or the Shellesder Member may pass laterally northwestwards into the A' Mhagarach Member. The change in lithology and the appearance of numerous fossils signify a change in environment towards estuarine and shallow-water conditions, associated with the oncoming Jurassic transgression.

South-Eastern Outlier

A NE–SW-elongated, downfaulted area of Triassic rocks extends for 1 km to the north-west of Loch Sgaorishal [NG 347 022]. The area is bounded to the north-west by a fault and the beds rest unconformably on Torridonian sandstones on their south-eastern edge (Figure 9). The rocks are coarse-grained conglomerates with carbonate impregnations, cornstones and coarse-grained sandstones. Cornstone on the south-east side is marked by small depressions and sink holes in the grass and peat-covered ground north-west of Loch Sgaorishal [NG 3450 0260]. The rocks are correlated with the Shellesder Member of the main outlier to the north-west.

Jurassic

Lower Jurassic of Rum

Small outcrops of sandstones, sandy limestones, limestones and shales occur on the south-east side of Beinn nan Stac between the north-eastern slopes of Dibidil [NM 398 933] and the stream Allt nam Bà [NM 407 944]. The beds crop out along the line of the Main Ring Fault and were preserved by downfaulting in this Palaeocene fault system. All the beds have been more or less indurated and thermally metamorphosed by adjacent Paleocene gabbros and ultrabasic rocks. These effects are most notable near Allt nam Bà where high-temperature talc-silicate mineral assemblages have been formed (Chapter 7). The rocks were first mentioned by Geikie (1897, p.351) who refered to altered grey and white marble or limestone on the northern slopes of Glen Dibidil. However, there is no description of this or any other occurrence in the original memoir (Harker, 1908). Hughes (1960b) described limestone and talc-silicate rocks from Allt nam Bà and tentatively correlated them with Precambrian (Lewisian) limestones on Coll and Tiree. Subsequently poorly preserved fossils with Jurassic affinities were reported from outcrops south of Allt nam Bà (Dunham and Emeleus, 1967; Emeleus and Forster 1979). Smith (1985; 1987) has provided the most detailed descriptions of the lithology, fauna and structural setting of these beds, which he correlated with the Lower Jurassic Broad-ford Beds of Skye (Hallam, 1959) on the basis of the fauna (Smith, 1987, tables 3.2 and 3.3) and the close similarities between the lithologies on Rum and those in the Broadford area of Skye. The closest similarities were considered to be between the Skye Ob Lusa and Ob Breakish coral beds (Hallam, 1959) and the Rum limestones, and the petrography of the sandstones in both areas (Smith, 1987, pp.58–60).

The beds occur between the Inner and Central members of the Main Ring Fault south of Allt nam Bà (Chapter 10) where they form a steep, westerly dipping succession some 35 m thick consisting of, from west to east, shales, sandy shales, limestones and sandstones (Figure 11). They are fault-bounded to the west and on the east they appear to be unconformably overlain by altered basalt flows assigned to the Paleocene Eigg Lava Formation (Chapter 8). Since such dips as are seen in the sedimentary rocks are steeply inclined to the west, the implication is that the Jurassic beds and the overlying lavas have been overturned within the zone of faulting.

The shale was assigned originally to the Torridonian Bàgh na h-Uamha Shale Member (present Laimhrig Shale Member of the Diabaig Formation), prior to the discovery of a belerianite guard and other pyritised fossil remains. The shale crops out over a small area at [NM 4051 9415] 200 m south of Allt nam Bà.

At its northern end, the limestone is strongly recrystallised to a coarse, grey marble, well exposed in a low scarp [NM 4059 9424] 130 m south of Allt nam Bà. On the southern slopes of the valley, at about 200 m elevation, near-vertical beds of limestone and sandy limestone contain numerous relicts of deformed, poorly preserved bivalves and less common casts of belemnites. Although exposure of the limestone is poor, its course south of Allt nam Bà is indicated by lines of sink holes in the peaty ground [NM 4054 9413]. Sink holes also occur along the east side of the Inner Main Ring Fault on south-east Beinn nan Stac where caves occur beneath Torridonian sandstone in the hanging wall of this fault [NM 4037 9386]. Outcrops of calc-silicate rocks within the marginal gabbro in Allt nam Bà [NM 4063 9438] and in ultrabasic rocks 200 m to the NNW, are large xenoliths of the Jurassic rocks (Figure 11).

Sandstone found sporadically along the east side of the limestone was originally thought to be Torridonian but it is unlike the Rum Torridonian sequence. It consists of subrounded quartz grains of bimodal size distribution (0.2–0.5 mm) with a dusty coating and supported by a silica cement. In contrast with the Torridonian sandstones, feldspar is uncommon A poorly preserved cross-section of a coral (? Thecosmilia sp.) was recovered from the calcareous sandstone.

A similar succession, apparently unfossiliferous, crops out between about 160 and 240 m on the north-east side of Dibidil, about 800 m north-west of Dibidil Bothy [NM 3929 9274]. The outcrops are terminated to the north and south-east by steep, reversed faults which uplift Torridonian strata, and end to the south against altered basalt lavas and upfaulted Torridonian sandstones. Outcrops of thermally altered ironstone within the Jurassic sandstone and shale (M Errington, personal communication, 1990) may correlate with a prominent ironstone in the upper Broadford Beds near Ob Breakish, Skye (Hallam, 1959, fig. 2). Relicts of a pale, calcareous sandstone also occur in the faulted foreshore exposures at Dibidil [NM 3936 9270] (M Errington, personal communication, 1990).

Middle and Upper Jurassic of Eigg and Muck (Table 1)

Sedimentary rocks underlie the Paleocene basalt lavas and form the low ground of the northern part of Eigg and the shore platform of Camas Mór on Muck (Figure 7). They are mainly of Middle Jurassic age, overlain locally by thin Upper Cretaceous strata. The oldest, exposed in northeast Eigg, belong to the upper (Bajocian) part of the Bearreraig Sandstone Formation. The most widely distributed and best developed part of the succession comprises the Lealt Shale, Valtos Sandstone, Duntulm and Kilmaluag formations of the Bathonian Great Estuarine Group. Callovian to Oxfordian strata of the Staffin Shale Forma-don have a small outcrop at Laig on Eigg.

Bearreraig Sandstone Formation

The oldest rocks found in situ on Eigg belong to the upper part of the Bearreraig Sandstone Formation which, in the type area of north-east Skye, is of latest Toarcian to Bajocian age (Morton, 1976). The outcrops form low scars, accessible only at low tide, exposed intermittently along the north-east coast north of Rhubha nan Tri Chlach for some 900 m [NM 498 895]–[NM 496 903], dipping gently inland and forming a strike section. The rocks are pale calcareous sandstones with abundant fragmentary echinoderms, bivalves, bryozoa and belemnites; conspicuous cross-bedding foresets in the lower beds dip northwards. The higher parts of the succession are darker, with shale partings. The thickness is approximately 6 to 7 m; the immediately overlying beds are not seen. Some scars of dolerite, probably sills, cross the shore near the northern end of the exposures. Shales exposed at the top of the beach [NM 496 901] belong to the Great Estuarine Group (see below) and the gap between the outcrops is less than 10 m.

Although no diagnostic fossils have been found, the facies is clearly marine and closely resembles that of the uppermost sandstones of the Bearreraig Sandstone Formation, of Bajocian age, as developed in Skye and Raasay immediately beneath the Garantiana Clay Member (Morton, 1965), and the argillaceous sandstone at the exposed top on Eigg suggests a passage into a clay or shale facies that is continued in the overlying Great Estuarine Group.

Great Estuarine Group

By far the most extensive outcrops of Mesozoic rocks in the Small Isles are those of the Great Estuarine Group on the Isle of Eigg. The upper beds are also well exposed at Camas Mór, Muck. Eigg was the original type section of the Great Estuarine 'Series' (Judd, 1878, pp.722–723). Previously, MacCulloch (1819) had correctly correlated the strata with those of Trotternish, Skye, and Hugh Miller had visited Eigg, making many acute observations and discovering the first Scottish plesiosaur. Harris and Hudson (1980) replaced the term Series by Group in conformity with modern usage, and defined formations within it with type sections mainly in Skye (Table 1).

The base of the Group is not exposed on Eigg. The lowest exposures giving a continuous section are those of the type section of the Kildonnan Member, Lealt Shale Formation, about 2 km south of the Bearreraig Sandstone outcrop (Figure 13) and Appendix 1. (Modified from Hudson, 1966 fig.1A.)" data-name="images/P936583.jpg">(Figure 12). The small outcrop north of Rhubha nan Tri Chlach [NM 496 901], referred to above, shows 1–1.5 m of black shale, slightly baked by irregular basalt sills. The shale is laminated and contains cycloid fish scales, closely resembling the facies of the Cullaidh Shale Formation at its type locality at Elgol, Strathaird, Skye (Harris and Hudson, 1980). In Skye the Cullaidh Shale Formation is separated from the Kildonnan Member by the Elgol Sandstone Formation, but no sandstone is exposed on Eigg. Discontinuous shaly mudstone exposures on the north-east coast, north of the Bearreraig Sandstone outcrop, are silty or contain thin beds of fine sand, and the Kildonnan Member type section itself is silty, especially in its lower part. It seems most practical to assign all these outcrops to the Kildonnan Member, while recognising that the lateral equivalents of the Cullaidh Shale and Elgol Sandstone Formations of Skye may be included.

The exposed top of the Great Estuarine Group on Eigg and Muck is a disconformity with Cretaceous or Paleocene deposits, and the Callovian–Oxfordian rocks of Laig Bay are probably faulted against it. The Skudiburgh Formation of Skye is not known in the Small Isles and the Kilmaluag Formation is incomplete. Eigg provides the best development known of the Kildonnan Member, a fairly continuous but metamorphosed section of the Lonfearn Member (Lealt Shale Formation) and excellent exposures of the main part of the Valtos Sandstone Formation. The upper Valtos and lower Duntulm formations crop out at Camas Mór on Muck; the Duntulm Formation–Kilmaluag Formation boundary is well exposed in Laig Gorge, Eigg, and a higher part of the Kilmaluag Formation at Camas Mór. So, between them the two islands provide a nearly continuous section through the Great Estuarine Group.

Lealt Shale Formation, Kildonnan Member

The Kildonnan Member is well exposed and highly fossiliferous at its type section, 2.5 km north of Kildonnan, eastern Eigg [NM 496 872], (Hudson, 1966; Harris and Hudson, 1980). Less extensive exposures occur in the small bay west of Talm on the northern coast [NM 4725 9075] where, unlike the type section, the Kildonnan Member–Lonfearn Member junction is continuously exposed. The base of the Member represents the base of the Great Estuarine Group on Eigg (see above). Its top is a stromatolitic algal limestone which occurs throughout the Inner Hebrides (Harris and Hudson, 1980).

The predominant lithology is a grey, silty shale or mudstone, with a thin bed of coarse calcareous sandstone and several of shelly limestone, often with fish debris; two of these are bone-beds and one of them, Hugh Miller's Reptile Bed, yields plesiosaur bones. These strata were discovered by Miller in 1844, but for more than a century thereafter little was written about them. They were included in Barrow's Lower Shales of the 1908 Memoir. Revision was started by Hudson (1962) and the succession and fauna discussed in subsequent works (Hudson, 1963b; 1966; 1968; 1970; Tan and Hudson, 1974; Hudson and Harris, 1979; Harris and Hudson, 1980; Hudson, 1980; Chen and Hudson, 1991; Riding et al., 1991). The palaeontology is summarised in Hudson et al. (1995).

Details: Kildonnan

A sketch-map of the exposures at the type section of the Kildonnan Member is given in (Figure 13) and Appendix 1. (Modified from Hudson, 1966 fig.1A.)" data-name="images/P936583.jpg">(Figure 12), and the succession is summarised in (Figure 13). A detailed measured section appears as Appendix 1. The bed numbering scheme of Hudson (1966) is retained with minor modification. Exposures between tidemarks and on the storm beach have provided an almost continuous section through 27 m of strata, but parts on the storm beach are sometimes covered. The marker beds shown on (Figure 13) and Appendix 1. (Modified from Hudson, 1966 fig.1A.)" data-name="images/P936583.jpg">(Figure 12) and (Figure 13) can be used to correlate between the outcrops. It is now thought that Section 4 of Hudson (1966) is a scatter of loose blocks of the harder limestone beds, perhaps the remains of a higher landslip block now largely eroded away. Exposures elsewhere in the type section area, and on the north coast of Eigg, are consistent with the order of succession of the limestones in Hudson (1966) but the thicknesses estimated between them are probably too low. The gap shown below the Algal Bed (9b in (Figure 13)) is an estimate based on extrapolation from the north shore exposures.

The variable westerly dips observed are probably connected with the landslips that affect this part of the coast. Away from the thin dykes and sills the rocks are unmetamorphosed, and the mollusc shells, especially Praemytilus, are preserved in glistening white aragonite (Hudson, 1968). The sequence of faunas and of palynomorphs allows increasingly accurate interpretation of the conditions of deposition of the beds, particularly in respect of salinity variations (Hudson, 1963a; 1966; Tan and Hudson, 1974; Riding et al., 1991; Wakefield, 1994). In the following text, fossil names are generally give at generic level; see Hudson (1963b; 1968; 1980) for details of the molluscs, Chen and Hudson (1991) for the conchostracans, Wakefield (1991) for the ostracods and Riding et al. (1991) for the palynomorphs.

The base of the exposed succession can be seen near the north end of the exposures, beneath the Reptile Bed which serves as a marker. Fossils are less abundant and less well preserved than in the higher beds; shelly layers occur towards the base of the section, and plant and fish fragments are common throughout. Small, mytilid bivalves, probably juveniles of Praemytilus strathairdensis, are the commonest molluscs, accompanied by small gastropods: Valvata and less common neritids. Unionoid bivalves also occur, as do the conchostracans, Euestheria and Neopolygrapta. This represents a low-salinity, probably nearly freshwater fauna, as also indicated by palynomorphs (Riding et al., 1991).

The Reptile Bed, Bed 2, has a limited in-situ exposure but is readily recognised in loose blocks because the fine-grained siderite within it weathers red. Blocks with conspicuous disartiulated black plesiosaur bones are not as common as they were, but fish teeth and scales are consistently present. The rock is mainly a sideritic gastropod biosparite; siderite mudstone lenses within it contain well-preserved small neritids (the Globalaria of the 1908 memoir), Valvata and 'Cylindrobullina' (probably an ellobiid). This gastropod association recurs at several horizons within the Kildonnan Member and forms part of a low-salinity assemblage (Hudson et al., 1995). Unio occurs less abundantly.

The lower part of Bed 3 contains the same gastropod fauna as Bed 2, together with Viviparus and small Praemytilus, forming shell layers and thin shell beds in silty mudstone. In the higher parts of the bed Praemytilus becomes increasingly dominant, but the shells are all small and probably juveniles. In the topmost 5 cm, large articulated specimens of Unio occur. This probably signals an influx of fresh water prior to the deposition of coarser sediment in Bed 4.

Bed 4 is the most useful marker bed in the lower part of the section. Its lower part consists of irregularly lenticular beds of coarse, calcite-cemented sandstone, separated by shale lenses. Some of the sandstone beds have rippled tops and mudcrack fills at their bases. The sandstones contain variable but often high proportions of worn phosphatic debris, the 'vertebrate sand' of Reif (1971). Some of the shales contain small gastropods, as in the underlying beds. The upper part of Bed 4 is a sandy molluscan biosparite composed of mainly small Praemytilus.

Bed 4 is the only part of the Kildonnan Member to include sandstones that are both coarse grained and well sorted. Hudson (1966) suggested that an influx of coarse sand into the Eigg area might correlate with the coarse top of the Elgol Sandstone Formation delta in Skye; see also Harris (1989). The incorporation of phosphatic material suggests reworking on the lagoon floor. The shale laminae represent periods of slack water between influxes of sand.

Bed 5 is the thickest development of silty mudstone in the section, and suggested the name Mytilus Shales used for the Kildonnan Member by Hudson (1966). Thin siltstones and Praemytilus biosparites, some with fibrous calcite layers, provide arbitrary divisions within the bed. Its lower part contains a similar fauna to Bed 3. In the higher parts, especially Bed 5e (see (Figure 13) and Appendix 1. (Modified from Hudson, 1966 fig.1A.)" data-name="images/P936583.jpg">(Figure 12)), large Praemytilus cover entire bedding planes, often showing current lineation (Hudson, 1968). Some beds contain Valvata. The ostracods fauna includes Darwinula, Limnocythere and progoncytherids (Wakefield, 1994). Oxygen isotope ratios in Praemytilus (Tan and Hudson, 1974), ostracod distribution (Wakefield, 1994) and palynomorph abundances, particularly of Botryococcus (Riding et al., 1991), all suggest low and fluctuating salinity; of all the beds in the Great Estuarine Group these most deserve the term estuarine.

The upper part of Bed 5e and Bed 5f contain ovoid septarian concretions, about 1 m across and 20 cm thick, with well-formed rhombic calcite crystals in the otherwise empty septarian cracks. In Bed 5f, unusually, fossils occur at all angles with respect to bedding. They are a mixed assemblage of Unio, Viviparus (up to 4 cm) and Praemytilus, perhaps transported with an influx of coarse sediment. Above this is the Fish Bed (Bed 5g), exposed resting directly on a concretion at the back of the beach below the Shelter Stone; it also forms a marker bed farther south (Figure 13) and Appendix 1. (Modified from Hudson, 1966 fig.1A.)" data-name="images/P936583.jpg">(Figure 12). It is a thin Shelly limestone, crowded with hybodont shark teeth and fish scales. In Bed 5h, Praemytilus shells are preserved as clusters, rather than spread out on bedding planes, and heterodont bivalves occur.

The base of Bed 6 marks the onset of a sequence of biofacies in which heterodont bivalves, probably best referred to as Tancredia, replace Praemytilus in dominance, and a sharper differentiation between marine–brackish and freshwater–oligohaline biotas than in the lower beds becomes evident. The more marine beds (Bed 6b, middle of Bed 6d, 6e, base Bed 6f) contain abundant Tancredia, sometimes associated with Praemytilus and Cuspidaria. Low salinity facies (upper Bed 6d, most of Beds 6f, 7b, 8a, 8c, 8e) are fine-grained shales with Neomiodon, conchostracans and numerous ostracods, mainly Darwinula and Limnocythere. A 5 cm-thick shell-bed of small gastropods, mainly neritids and Cylindrobullina' , occurs at the base of Bed 6d. Three marker beds occur in this part of the section. Bed 6e, the Bivalve–Septarian Bed, helps to correlate exposures between the south end of the section and the region of the Shelter Stone. The Unio Bed (Bed 7a), contains large Unio with conjoined valves, Viviparus, wood and bone fragments. It is the only limestone in the upper part of the section with an unequivocal freshwater fauna. Bed 8b is a hard, blue-grey, unfossiliferous calcilutite resembling the limestones associated with Bed 9. It has a thin layer of fibrous calcite at its top, beneath which Placunopsis occurs, the first appearance of a marker genus for marine episodes higher in the succession. The shales above and below, however, contain low-salinity faunas. Bed 8b is the highest consistently exposed marker bed at the type locality.

The stromatolitic algal limestone, Bed 9b, is no longer exposed in situ at the type section (see above). At exposures elsewhere (north Eigg and Skye), a calcilutite with abundant Placunopsis and Cuspidaria occurs at its base. Hudson (1966) recorded a similar fauna, also including Modiolus and Lopha, from what was regarded as a separate limestone bed, Bed 8d. In the present account such limestones are designated as Bed 9a; it is uncertain whether they were originally contiguous with Bed 9b (Figure 13). These occurrences record the most convincing high-salinity faunas in the succession. Pseudomorphs after gypsum occur in the stromatolite (Hudson, 1970). It is possible that the lagoons, as well as their exposed margins, became hypersaline at this time; no species of ostracod crosses the Kildonnan Member–Lonfearn Member boundary (Wakefield, 1994).

Details: North Eigg

There are several small exposures of the Kildonnan Member on the north and north-east shores of Eigg, between a small bay west of Talm [NM 4725 9070] and the coast east of Fhaing Ruadh [NM 4890 9080]. At the former locality the top of the Kildonnan Member is seen in continuous section beneath the Lonfearn Member; a measured section is given in Appendix 2. Correlations with the type section are suggested below.

The stromatolitic algal bed is well developed with Placunopsis limestone at its base (Bed 9). Beneath this are greenish mud-stones with a sparse low-salinity fauna, not usually well exposed (?Beds 7–8). After a short non-exposed interval, silty mud-stones with Tancredia and conspicuous mudcracked horizons are seen (Bed 6), underlain by mudstones with Praemytilus (Bed 5h). This part of the section occurs between tidemarks, dipping gently west, and farther east exposures are discontinuous.

Near the top of the storm beach in the same vicinity [NM 4730 9075] to [NM 4735 9075] are several excellent exposures of the stromatolite equivalent to Bed 9, including that figured by Harris and Hudson (1980), dipping south at variable steep angles. These form part of the eroded toes of the landslips that fringe this part of the coastline. Beneath them, shales and thin limestones resembling Beds 5h to 8 of the type section are exposed from time to time. Between this locality and Eilean Thuilm the beach lacks exposures but the low cliff cut into landslip has yielded exposures across the Kildonnan Member–Lonfearn Member boundary.

The shore opposite Eilean Thuilm is the discovery site of Hugh Miller's Reptile Bed and thus of marine reptiles in the Scottish Jurassic (Miller, 1858; Hudson, 1966). Many eroded blocks of the distinctively red-weathering sideritic limestone occur in the storm beach and the raised beach, and plesiosaur bones are common (although less so than they once were). The outcrop has not been located. There is no doubt about Miller's correlation with Bed 2 of the type section. East of Fhaing Ruadh [NM 4890 9080] ostracod-rich siltstones crop out on the foreshore and are evidently not slipped; they probably belong low in the Kildonnan Member.

Lealt Shale Formation, Lonfearn Member

The Lonfearn Member (formerly the Estheria Shales, Harris and Hudson, 1980) lies above the stromatolitic algal limestone that marks the top of the Kildonnan Member and is overlain gradationally by the Valtos Sandstone Formation (Harris and Hudson, 1980). It consists mainly of shales, generally darker, fine grained and more fissile than those of the Kildonnan Member. The shales are olive-brown when fresh, but almost black when they are slightly baked by numerous thin basaltic sills. They have a predominantly low-salinity fauna of Neomiodon, ostracods and conchostracans, except near the base where marine–brackish forms occur. The shales are interbedded with thin limestones, some composed of variably broken and abraded Neomiodon shells, presumably winnowed from the muds on the lagoon floor; others are oolitic, testifying to carbonate precipitation. Micrite envelopes and 'algal borings' are well developed (Hudson and Andrews, 1987).

There are only two long, fairly continuous sections in the Lonfearn Member on Eigg, and both are intruded by numerous thin basaltic sills that partly disrupt the sequence. The most important is on the shore between the top of the Kildonnan Member west of Talm [NM 4725 9070] to the transitional contact with the Valtos Sandstone Formation near the north-west tip of the island [NM 4690 9050], the North Shore section. The other is in a large landslipped mass in the Allt na h'Airde Mheadhonaich above the east coast, below the old shieling [NM 4970 8885], the Shieling Burn section; it exposes the top but not the base of the Member. There are several other small exposures on the landslip zone fringing the north and east coasts; their relationships, however, are obscure. One, above the south end of the Kildonnan Member type section, includes an unusually thick (2.5 m) mollusc-shell biosparite limestone, and is intruded by a columnar sill (Figure 13) and Appendix 1. (Modified from Hudson, 1966 fig.1A.)" data-name="images/P936583.jpg">(Figure 12).

Details: North Shore section

The lower part of the section is well exposed and, away from the two thin sills, is not metamorphosed; it is detailed in Appendix 2. Beds 1 to 7 have a marine–brackish bivalve fauna including Cuspidaria, Placunopsis and Isognomon, and a high-salinity ostracod fauna, also found at this horizon in Skye (Wakefield, 1994). Bed 9 is a prominent fine-grained limestone with mudcracked horizons, and the freshwater gastropod Viviparus in its upper part. Above Bed 12, a prominent sill closes the small bay in which these outcrops occur, and to the west the section is disrupted by numerous sills. The beds are dark shales with thin shelly limestones, dominated by Neomiodon. At the top of the Member, the shales are again unmetamorphosed, yielding ostracods and conchostracans. Hudson and Wakefield measured a complete, but approximate, section in 1990 (Wakefield, 1991). It includes 27 sills totalling 17.5 m, with 16.5 m of observed sedimentary rocks, and 17 m of estimated gaps, presumably mainly sedimentary rocks, giving approximately 50 m of total section.

Details: Shieling Burn section

Here the strata dip at 26°W, and form part of a large landslip block. The lower part of this section (Appendix 3) compares closely to that given by Barrow (Harker, 1908, pp.20–21). The Lonfearn Member consists of shales and thin limestones, mostly metamorphosed but some fresh enough to yield well preserved ostracod faunas. Some of the limestones are conspicuously oolitic, others are composed of broken Neomiodon shells. Wakefield (1991) gives a section including 12 sills (some multiple) totalling 11 m, 13.75 m of observed sedimentary rocks, and 9 m of estimated gaps; about 34 m of total section. At the base of the Valtos Sandstone Formation, below the typical sandstones, is a bed of unconsolidated shell-hash including aragonitic Unio shells.

Valtos Sandstone Formation

The most conspicuous Mesozoic outcrops in the Small Isles are the cliffs of the Valtos Sandstone Formation in north-west Eigg. From the Bay of Laig to Blàr Mór, sandstone cliffs up to 40 m high form the seaward limit to the undulating lowland of Cleadale (Plate 1). The pale sandstone give rise to pure white beach sands, the famous 'singing sands' of Camas Sgiotaig. The sandstones are criss-crossed by dykes which have weathered into trenches, leaving walls of irregularly baked sandstone standing proud on either side (Frontispiece). Calcareous concretions stud the cliffs and the shore (Plate 5). The formation outcrop continues, though partly concealed, around the north-east and east coasts, descending to sea level at Kildonnan. Here, small outcrops of pale, cream to white indurated sandstone occur at Pol nam Partain [NM 489 849], as screens between the basic intrusions (the Kildonnan Sheets, Chapter 6). Soft, pale cream to white sandstone is also exposed nearby in the stream by the old mill [NM 4878 8519]. To the east and to the north-east the sedimentary rocks are faulted out by the south-east extension of the Laig Gorge Fault which brings basalt flows down against them.

The base of the Valtos Sandstone Formation can be seen near the north-west tip of Eigg. Its top cannot be clearly seen on that island, although its position can be defined in Cleadale. The top is well exposed at Camas Mór Muck, where the upper part of the formation is the lowest stratigraphical level exposed on the island. The succession in the Small Isles has been divided into six informal divisions, A–F (Figure 14) and Figure 15), each generally consisting of a facies sequence from five of the seven facies recognised by Harris (1992). Division A is composite, containing two facies sequences, Ai and Aii (Figure 14). Only the upper part of E and division F crop out on Muck. The facies may be summarised as follows:

  1. Neomiodon Mudstone–Siltstone Facies, representing offshore, low-salinity lagoonal mudstones at the base of each facies sequence.
  2. Coarsening-upwards Sandstone Facies, representing delta-front and lower–middle shoreface sands in the lower part of each sand body.
  3. Coarse-Pebbly Sandstone Facies, representing distributary channel and sand-flat environments (locally split into subfacies: 3a, pebbly facies, and 3b, thin-bedded, trough cross-stratified coarse-grained sandstone).
  4. Wave-Formed Sandstone Facies, representing wave-reworked, fluvially supplied, mid–upper shoreface, foreshore and backshore sands.
  5. Neomiodon-debris Limestone Facies, representing transgressive deposits capping abandoned lagoonal delta and shoreline sequences; Neomiodon limestones also occur within facies 1, generally as thinner, less sandy beds.

Each of the divisions, with the exception of F, contains a sand body, generally coarsening upwards, and usually capped by a coarse-grained, shelly, cemented unit representing reworking, and overlain by the fine-grained base of the next division. All the divisions are predominantly sandy, except for division F which is shale and limestone (Figure 15). The formation was interpreted by Harris and Hudson (1979) and by Harris (1992) as representing a series of delta lobes built into the lagoons in which the Lealt Shale Formation had been deposited.

Details: North-west Eigg

The succession of these facies can be studied in detail over 1.8 km of excellent coastal exposures between the north side of the Bay of Laig [NM 472 885] and the north-west tip of the island (Figure 14). A summary of the facies distribution is given in (Figure 15), and sedimentary logs in Harris (1984, 1992). Some comments on particular exposures follow.

At the north-west tip of the island the gradational contact with the Lonfearn Member of the Lealt Shale Formation is exposed. Thin beds of the sandstones appear intercalated with dark shales, which are intruded by thin and somewhat irregular basic sills. The proportion of sandstone beds increases, and that of sills decreases, upwards. The sandstone beds are strongly bioturbated. The upper part of Division A contains prominent concretions which are well exposed on the shore platform south-east of Bogha na Brice-nis [NM 464 904]. Sandstones in Division B north of Camas Sgiotaig contain large masses of rafted logs.

The massive sandstone of Division C forms most of the lower cliff exposures in the complex headland between Laig Bay and Camas Sgiotaig. It is pebbly, the pebbles indicating derivation mainly from the metamorphic rocks of the Scottish mainland It is also intersected by several deeply eroded dykes (Frontispiece). Alongside these dykes the sandstone is cemented by masses of centimetre-scale concretions, which die out a few decimetres from the dyke; presumably carbonate was mobilised during dyke intrusion (Hudson and Andrews, 1987). Near the path down to Camas Sgiotaig [NM 471 899] the sandstone shows prominent south-dipping foresets. Where it overlies thin beds of greenish argillaceous sandstone at the base of the cliff, differential erosion causes prominent overhangs and there is a natural arch.

On the north side of Laig Bay, north of the chapel path [NM 472 885], Division E is well exposed, giving a particularly good view of the calcerous concretions that gave the formation its former name of Concretionary Sandstone (Plate 5). In the lower part of the sandstone body relatively isolated concretions occur but are often coalesced into botryoidal masses. The top of the sandstone body is more continuously, but still not completely, cemented. Wilkinson (1992) gives a detailed account of the geometry and geochemistry of the concretions. They are believed to have formed during the late Jurassic burial of the sand bodies, when calcite precipitated from meteoric-derived pore waters at a depth of 200 to 500 m and temperatures of 31 to 34°C. Using the equations of Wilkinson and Dampier (1990), the small concretions that were investigated in detail, of average radius 11.8 cm, would have taken 0.36 to 0.84 million years to have formed; metre-scale concretions would have taken several million years. In the cliff above the shore, a grassy interval corresponds to shales of Division F; a Neomiodon biosparite occurs, associated with fibrous calcite ('beef) which was studied by Marshall (1982).

Details: North-east Eigg

The exposures in this region add little sedimentologically to what has already been described. In the stream section of Fhaing Ruadh, at the foot of the basalt cliffs [NM 4884 9048], probable Cretaceous sandstone rests directly on the Valtos Sandstone Formation, from which it was probably largely derived. In this part of the island the sandstone occurs in situ beneath the Tertiary volcanic rocks as part of the main inland cliff. At the 'Shieling Burn' (see above), the sandstone is involved with the Lealt Shale Formation in a major landslip (Appendix 3).

Details: Camas Mór, Muck

The upper part of the Valtos Sandstone Formation is exposed in Camas Mór [NM 408 792], west of a fault that is occupied by a prominent dyke and has a downthrow to the east (Figure 17), ((Figure 18)) and Appendix 4." data-name="images/P936587.jpg">(Figure 16). The lowest beds exposed are sandstones with concretions, correlated with Division E of Eigg; about 6 m are visible at low tide. Above are alternating shales and Neomiodon limestones; one limestone is over 1 m thick and has complex loadcasts at its base. Thick veins of fibrous calcite occur within the shales. These beds crop out as prominent intertidal crags immediately west of the 'fault dyke' (Figure 17), ((Figure 18)) and Appendix 4." data-name="images/P936587.jpg">(Figure 16) and are referred to the lower part of Division F; their total thickness is 3.5 m. Boulders of the storm beach usually separate these outcrops from more regularly bedded limestone–shale alternations that form broad intertidal scars on the western side of Camas Mór, dipping gently north-west and passing gradationally up into the Duntulm Formation. These comprise the upper part of Division F, with a thickness of approximately 7 m. A limestone with conspicuous load casts, which penetrate more regularly and less deeply into the shale beneath than that referred to above, is a convenient marker bed. A section starting at this bed and continuing into the Duntulm Formation is given in (Figure 17) and Appendix 4. The formation boundary is taken at the change from the freshwater–brackish Neomiodon-Viviparus- conchostracan faunas of the Valtos Formation to the marine–brackish faunas with Placunopsis of the Duntulm Formation (Harris and Hudson, 1980; Andrews and Walton, 1990). This occurs 1.3 m below the incoming of the massive oyster beds typical of the Duntulm Formation.

Duntulm Formation

With its abundance of oysters (Praeexogyra hebridica (Forbes): see Hudson and Palmer, 1976), the Duntulm Formation is instantly recognisable, and can be correlated easily in a general way from the type locality in North Skye to Muck. Within the Small Isles the lower part of the formation is particularly well exposed in Camas Mór, Muck, and the top in the Laig Gorge, Eigg. These exposures have been correlated in detail with those of Strathaird, Skye by Andrews and Walton (1990); see (Figure 17). There are several small exposures in Cleadale, Eigg, where the oyster limestones make a small scarp above Division F of the Valtos Sandstone Formation near the Chapel path, where there is an old lime kiln [NM 4746 8854]. There are no coastal exposures on Eigg.

The Duntulm Formation represents the closest approach to a marine facies within the Great Estuarine Group, but lacks ammonites, belemnites, corals, brachiopods apart from Kallirhynchia, and most stenohaline bivalves. The dominance of one species, in this instance Praeexogyra hebridica, is characteristic of brackish-water deposits (Hudson, 1963a; 1980; Andrews and Walton, 1990). In the Small Isles the lithologies are dominated by limestones and shales, with subordinate 'algal' limestone; the sandy facies that occur in North Skye are absent, and so are the freshwater intercalations found there. There is little evidence of tidal currents, and the depositional environment is envisaged as one of microtidal, shallow lagoons, with low oyster banks, and fringed by 'algal' (cyanobacterial) marshes.

Of the lithofacies recognised in this formation by Andrews and Walton (1990), three occur in the Small Isles. They correlate well with the biofacies, and may be summarised as follows:

Facies 1 consists of oyster limestones, variably argillaceous, and closely interbedded shales or silty shales. Praeexogyra is a rock-former in many of the limestones but is less dominant as the clastic content of the beds increases. Although the oysters are commonly attached to one another, and modern analogues suggest that they probably formed mounds or even small biostromes, the original relief is not preserved; however the good preservation of the shells implies minimal transport. Some beds yield the brachiopod Kallirhynchia as well as oysters; this probably represents a higher salinity facies. The shales and siltstones have a more varied bivalve fauna, and probably represent inter-mound areas. There is no sharp distinction with Facies 2.

Facies 2 is also characterised by limestone–shale alternations, but the limestones are finer grained, usually argillaceous and lack abundant oysters; some may be concretionary. The shale:limestone ratio is higher than in Facies 1. This facies is particularly characteristic of the Valtos–Duntulm and Duntulm–Kilmaluag formational boundaries, at Camas Mór and Laig Gorge respectively. At these transitions the faunas are rather restricted, but similar lithologies within the main part of the formation yield more diverse faunas in which the bivalves Placunopsis, Cuspidaria, and Corbula are particularly characteristic. Some of the limestones contain abundant miliolid foraminifera. In general the facies probably represents quieter water conditions than Facies 1, perhaps protected by oyster banks.

Facies 3 consists of 'algal' (cyanobacterial) limestones. In the field they are hard, blue-grey, knobbly weathering limestones composed of nodules 2 to 10 mm across that are generally unfossiliferous. The beds are typically 30–50 cm thick. Petrographic examination reveals much complexity, and several subfacies can be distinguished (Hudson, 1970; Andrews, 1986). The most characteristic subfacies bears numerous thalli of the cyanobacterium Cayeuxia nodosa Anderson with intergrown organic-walled tubes, closely comparable to the Scytonema–Schizothrix association described from the Bahamas by Monty (1967). The facies is interpreted as representing schizohaline, supra- to intertidal marshes that probably fringed the oyster lagoons. The fact that some individual beds of this facies can be traced over long distances suggests either small-scale diachronism or that at times the cyanobacterial marshes extended across the lagoonal area.

Andrews and Walton (1990) gave sedimentary logs of the Camas Mór and Laig Gorge exposures and proposed a correlation with the continuous sections in Strathaird, Skye (Figure 17). Faunal lists for the facies within the formation are included in Andrews and Walton (1990).

Details: Camas Mór, Muck (see also (Plate 12)

The gently NW-dipping beds in Camas Mór form scars extending from between tide-marks into the coarse storm beach which obscures the higher beds (Figure 17), ((Figure 18)) and Appendix 4." data-name="images/P936587.jpg">(Figure 16). The section is separated from the Kilmaluag Formation exposed 25 m to the south-west by dolerite intrusions. The rocks are well preserved but the shales are slightly hardened by thermal metamorphism connected with the dyke swarm that crosses the area. The transitional boundary with the Valtos Sandstone Formation is well exposed (see above). The basal 0.8 m of the Duntulm Formation yields Placunopsis and Cuspidaria, in part pyritised, but not Praeexogyra. Once it appears, Praeexogyra rapidly comes to dominate the fauna, forming the typical oyster beds of the formation, which are exceptionally well exposed at this locality. Kallirhynchia occurs in at least two beds, tending to form clusters. Bed 14 ((Figure 17), Appendix 4) is the lowest nodular algal limestone in the succession. Between beds 24 and 33 is a distinctive sequence of alternating shales and fine-grained limestones, 1 m thick, lacking oysters but yielding Corbula, Placunopsis, ostracods and miliolid foraminifera. It is capped by a discontinuous nodular algal bed. Above this, the section is mainly oyster-rich limestone and shale, with one conspicuous algal bed.

Details: Laig Gorge, Eigg

A good section of the upper part of the Duntulm Formation occurs on the left bank of the Abhainn a' Chaim Loin in the lower part of Laig Gorge [NM 473 875], where the stream cuts through the Paleocene lavas into Cretaceous and Jurassic sedimentary rocks before flowing out on to the plain south-east of the Bay of Laig. A log of the section is included in (Figure 17). The sedimentary rocks were only slightly metamorphosed by the quartz porphyry which crops out on the north bank of the stream. Near the base of the exposed section is a prominent nodular algal bed interbedded with typical oyster-rich shale–limestone alternations. This bed is correlated by Andrews and Walton (1990) with the highest algal bed of Strathaird and probably lies above the section exposed in Camas Mot., Muck. The upper 5 m of the section is mainly shale, with fine-grained, partly nodular limestones; oysters are not common. The top of the Duntulm Formation is taken at the last thin oyster-bearing limestone, from which thin sections indicate the shells are heavily abraded and bored. The ostracod fauna typical of the Kilmaluag Formation enters a short distance above, the contact between the formations is perfectly gradational.

Kilmaluag Formation

The Kilmaluag Formation is typified by alternations of fine-grained limestones or dolomites and calcareous shales; bed contacts often show conspicuous mudcracks. Calcareous macrofossils are generally absent, apart from Viviparus in a few beds, but ostracods of fresh to brackish affinity (Wakefield, 1994), conchostracans (Chen and Hudson, 1991) and palynomorphs (Riding et al., 1991) are abundant. The formation marks a return to low-salinity environments after the brackish–marine episode of the Duntulm Formation.

Exposures of the Kilmaluag Formation in the Small Isles are limited to a small area around Laig Gorge in Eigg and Camas Men-, Muck. The section in Laig Gorge exposes the lower part of the Formation, disconformably overlain by Cretaceous rocks. In Camas Mór the base is not exposed; correlation with the complete succession in Strathaird, Skye (Figure 18), suggests that the beds are high in the formation (Andrews, 1985). They are overlain by the Paleocene Camas Mór Breccia (Chapter 8, (Figure 41).

Details: Laig Gorge, Eigg

Six metres of shale–limestone alternations (Figure 18) overlie the Duntulm Formation with a transitional boundary. There is an abundant ostracod and conchostracan (Antronestheria) fauna, showing salinity oscillations across the formation boundary (Wakefield, 1994). The top limestone is dolomitic. A shale occurs beneath the disconformable base of the Cretaceous Laig Gorge Sandstone. The outcrop, above the south bank of the stream, is sometimes obscured by vegetation but can be easily exposed. A similar section is exposed in a stream section about 200 m north-east of Laig Gorge.

Details: Camas Mór, Muck

The principal exposures are on the west side of the bay, about 50 m south-west of the most southerly Duntulm Formation out and separated from them by irregular dolerite intrusions. The lowest Kilmaluag Formation exposures are seen only at extreme low water, topographically below and approximately along strike of the highest Duntulm Formation exposures. A fault intervenes but its position and orientation are not well constrained (Figure 17), ((Figure 18)) and Appendix 4." data-name="images/P936587.jpg">(Figure 16). There are two main exposures of the Kilmaluag Formation, separated by a poorly exposed interval, together representing about 14 m of strata. The lowest beds are fine-grained dolomites separated by shales, which show deep mudcracks. Beds 13–16 of Andrews (1985) yield well-preserved conchostracans of the genus Pseudograpta, a fauna believed to be the youngest in the Great Estuarine Group and also known in China (Chen and Hudson, 1991). The higher group of beds includes limestones with Viviparus, as well as dolomites. The geochemistry and origin of the dolomites is discussed in detail by Andrews et al. (1987). The excellent preservation of their fine-grained fabrics and depositional oxygen isotopic composition testifies to their mild thermal history, despite the proximity of the Paleocene lavas.

On the east side of Camas Mór, the Great Gabbro Dyke [NM 411 792] (Chapter 6) produced intense thermal metamorphism along its margins, and is itself contaminated by interaction with the sedimentary rocks. The country rock consists of disrupted and brecciated Kilmaluag Formation carbonate rocks which contribute to the Camas Mór Breccia. Tilley (1947) concluded from the metamorphic mineralogy that the sedimentary carbonate rocks were dolomite, thus producing the first record of dolomite from the Scottish Jurassic rocks. There are sporadic outcrops of the Kilmaluag Formation between the Great Gabbro Dyke and the fault and dyke that limit the outcrop of the Valtos Sandstone Formation (Figure 17), ((Figure 18)) and Appendix 4." data-name="images/P936587.jpg">(Figure 16). Here, the fault must have a throw to the west of at least 28 m (the exposed thickness of the Valtos and Duntulm formations, plus the missing upper part of the Duntulm Formation and probably part of the Kilmaluag Formation).

Staffin Shale Formation

The Staffin Shale Formation occurs as discontinuous faulted outcrops between tide-marks on the boulder-strewn shore, west of Laig Farm and east of Clach Alasdair. The small outcrop on the south shore of Laig Bay, Eigg, was formerly known as 'Oxford Clay' (Barrow, in Harker, 1908). Wright (1964) showed that ammonite zones from the Callovian Quenstedtoceras lamberti Zone to the Oxfordian Cardioceras cordatum Zone were represented in the outcrops. Turner (1966) referred the beds to the Staffin Shale Formation which has its type section in Staffin Bay, Skye. Sykes (1975) ascribed them as forming the type (and only known) section of the Laig Siltstone Member of the Staffin Shale Formation. This account is largely drawn from that paper.

The base of the succession is not seen, and the top is overlain disconformably by the Upper Cretaceous rocks of Clach Alasdair [NM 4537 8833]. The outcrops are affected by landslips, producing steep dips and minor faults, and the succession may only be determined by examination of the ammonites. The dominant lithology is medium-grey siltstone, usually glauconitic, with calcareous clays associated with nodular limestones. There is a slight coarsening-upward trend, and in the Cardioceras cordatum Zone strongly glauconitic sandstones occur; these are the most distinctive features of the lithology compared to outcrops of the Shaffin Shale Formation on Skye.

Bivalve distribution is facies controlled. Calcareous clays yield low-diversity assemblages dominated by nuculaceans. In the slightly glauconitic siltstones Cucullaea and Oxytoma are also present, and in the strongly glauconitic sandstones large Pleuromya, Pholadomya, Modiolus and Pinna occur in life position.

Ammonites are only locally abundant, but they suffice to establish the zonal sequence (Sykes, 1975, table 2). Because of discontinuous exposure the zonal thicknesses quoted are mostly minima, as is the overall thickness of 35 m.

Upper Cretaceous of Eigg

The thin discontinuous Cretaceous deposits of the Inner Hebrides rest disconformably on various Jurassic and earlier strata; sedimentary rocks of latest Jurassic and early Cretaceous age are everywhere absent. General reviews of these rocks are given by Rawson et al. (1978) and Hudson (1983), and a detailed study was made by Braley (1990). The three occurrences of Cretaceous rocks are all on Eigg. The relationships between them are uncertain, since they are isolated from one another, lithologically dissimilar (Figure 19) and poorly fossiliferous. The three localities will be described in turn.

Details

Laig Gorge [NM 4735 8750]: Strathaird Limestone Formation with Laig Gorge Sandstone Member

The distinctive strata disconformably overlying the Kilmaluag Formation in Laig Gorge were described as Jurassic in the 1908 memoir. Hudson (1960) redescribed them as the Laig Gorge Beds, and Adams (in Hudson, 1960) determined their Late Cretaceous age. Braley (1990; and in Lowden et al., 1992) proposed the current nomenclature. The succession is well exposed in the Laig Gorge where the Laig Gorge Sandstone Member forms bluffs above the lower gorge, and limestone occurs in the smaller upper gorge, between the waterfall caused by the quartz porphyry intrusion and the lowest basalt flow.

The Laig Gorge Sandstone Member has a basal conglomerate containing clasts up to 5 cm long. Some are far-travelled quartz and quartzite, and a few are rip-up clasts from the underlying Kilmaluag Formation shales. Most conspicuous are rounded pebbles of phosphorite, some containing oolitic chamosite. Hudson (1960) compared them to similar pebbles associated with disconformities elsewhere. The matrix in the lowest 1.1 m unit of the sandstone, which includes pebbles like those in the basal conglomerate, is predominantly quartzose and coarse grained, with grain size approximately 0.25–2 mm. There is an irregular contact with 1.2 m of somewhat finer-grained poorly sorted sandstone with rare pebble horizons dipping 18° north-east. The next unit has a disconformable base, and fines upwards from coarse and pebbly to somewhat finer-grained sandstone; it is 46 cm thick. At the top of this bed are the irregular 'nodules' described by Hudson (1960) as the main nodule bed. These are interpreted by Braley (1990) as Thalassinoides burrow infills. This burrowed horizon marks the top of the Laig Gorge Sandstone Member.

The lowest limestone in the Strathaird Limestone Formation is somewhat sandy. The bed is 85 cm thick. The sand content decreases upward, and the top is marked by a burrowed horizon. Disconformably above this is 1.6 m of uniform hard micritic limestone, divided into apparent beds on a decimetric scale by concentrations of stylolites; these 'beds' probably do not correspond to depositional events. This limestone is overlain directly by basalt but shows no obvious metamorphism. It lacks flint nodules.

Hudson (1960) recorded a derived specimen of Cardioceras (Scarburgiceras) sp. from the basal conglomerate, showing that Oxfordian strata were undergoing erosion nearby; other macrofossils recorded by Barrow (in Harker, 1908) from the same horizon are not age diagnostic and were probably also derived. Adams (in Hudson, 1960) was unable at that time to date accurately the planktonic foraminifera of the limestone (they can only be studied in thin section). Braley (1990), however, has determined, among other planktonic forms, Hedbergella praehelvetica (Trukillo), a species diagnostic of the Early to Mid Turonian, so giving more precision to the dating. Two species of dinoflagellate were recorded but are long-ranging taxa (Braley, 1990). The limestone also contains abundant calcispheres and a few Inoceramus prisms; it is almost devoid of benthic foraminifera.

Clach Alasdair [NM 4540 8830]: Clack Alasdair Conglomerate Member of the Strathaird Limestone Formation; horizon uncertain

The exposure at Clach Alasdair was recognised as Cretaceous in the 1908 Memoir, partly by a poor microfauna but mainly by lithological comparison. The succession is less than 1 m thick. It consists of complex siliceous flint conglomerates, overlying a sandstone which rests disconformably on the Oxfordian Staffin Shale Formation. The exposure in the shore section is not extensive but Cretaceous strata must continue in the land-slipped zone between the shales and the basalts for some 600 m to the east, as loose blocks are common on the shore. This description is based on a visit by J D Hudson and R G Clements in 1985, and on Braley (1990); see (Figure 19).

The top few centimetres of the Staffin Shale Formation are oxidised, and contain large (3 cm) and small glauconitic burrow fills (Chondrites?) extending from the bed above. This is a glauconitic mudstone possibly comprising Jurassic mud reworked on a Cretaceous sea floor; this was commented upon by Barrow (in Harker, 1908, p.34). The next unit is 15–18 cm thick at the seaward limit of exposure, thickening south to 40 cm. It is a glauconitic sandstone with Skiolithos and Chondrites burrows, and contains rounded white-weathering phosphatic clasts. The upper contact of this bed appears discordant in places and gradational in others, at least in part due to irregular silicification of the matrix. The two overlying beds weather out as one block, due to strong silicification. Their total thickness is about 40 cm, although each part varies considerably because of the irregular contact between them. The lower bed is a coarse-grained quartz-sandstone with an originally micritic matrix, now strongly silicified and containing clasts of silicified chalk, flint, and black phosphate. Silicified chalk clasts are up to 20 cm long, subangular to subrounded and chaotically arranged within the bed. Above a highly irregular contact, the upper bed is somewhat finer grained, and its matrix less obviously silicified. Subangular silicified chalk and flint clasts are distributed throughout the bed. Larger subrounded clasts, up to 15 cm, occur in distinct layers. The bed is capped by a gently undulating surface overlain by basalt.

The conglomerate beds were defined by Braley (1990) as the type section of the Clach Alasdair Conglomerate Member. Because similar conglomerates occur in Skye intercalated with typical Strathaird Limestone Formation, she referred the member to that formation, but in the absence of diagnostic fauna it is not possible to refer it to a specific horizon. The glauconitic sandstone beneath cannot be assigned confidently to Braley's lithostratigraphic scheme. In the author's opinion it is probably not much older than the conglomerate.

Allt Ceann a'Gharaid, North Eigg [NM 4885 9049]

Immediately below the Paleocene volcanic rocks, and resting with slight discordance on the Valtos Sandstone Formation, is a thin sequence of sedimentary rocks discovered by Dr A H F Robertson (Figure 19). The rocks consist largely of sand reworked from the beds below and lack an indigenous fauna, but they contain large irregular flint clasts that are almost certainly Upper Cretaceous.

The exposures are at about 155 m altitude, in and around the steep bed of the stream, beneath its cascade over the basalts. Several metres of typical Valtos Formation sandstone are exposed, intruded near the top by a 50 cm basalt sill. Overlying these sandstones is a coarse breccia, less that 50 cm thick in the stream but about 1 m thick to the south. Its base apparently transgresses the bedding in the underlying sandstone. The breccia contains angular, pale grey to white chert pebbles with pink margins, very similar to (baked?) chalk flints. The cherts are commonly up to 10 cm across, and occur together with rounded 1 cm quartz pebbles in a matrix of medium- to coarse-grained calcareous sandstone. The breccia is overlain by about 1 m of non-pebbly calcareous sandstone with rare glauconite grains, and this by about 3 m of amygdaloidal basaltic lava. Above this is 2 m of fine-grained, brown sandstone with beds of lithic tuff, and then at least 10 m of coarsely brecciated basalt before the lowest columnar basalt flow. This considerable development of sedimentary and volcaniclastic rocks suggests that the area may have been a pre-volcanic valley, and that the cherts are reworked. They cannot have travelled far because of their size, concentration and angularity. It is unknown whether the sandstone matrix of the flints is Cretaceous or Paleocene.

In thin section, the flints show numerous calcispheres, spicules and planktonic foraminifera in a matrix of very fine-grained silica. The sandstone matrix contains a few grains of glauconite, which cannot have been derived from the underlying Great Estuarine Group. The only likely age of the flints is Late Cretaceous.

Correlation of the Upper Cretaceous deposits

The only firm date is Early to Mid Turonian for the Strathaird Limestone Formation at Laig Gorge. The difference in facies between Laig Gorge and Clach Alasdair over so short a distance suggests that the two successions are not coeval. According to Braley's (1990) review of the regional evidence, the Clach Alasdair Conglomerate Member is likely to be the younger, and the local evidence supports this. The occurrence of a derived Oxfordian ammonite in the basal conglomerate at Laig Gorge is consistent with erosion of the west Eigg area at that time. The absence of flints in the Laig Gorge conglomerate argues against the former presence there of deposits like those at Clach Alasdair, because flints are virtually indestructible once formed. A period of erosion after Cretaceous deposition and before the outbreak of volcanism is suggested by the derived flints in the North Eigg breccia, and during the same interval any Clach Alasdair-type deposits were presumably removed from the Laig Gorge area. However, this period of erosion was minor compared with that taking place prior to the Late Cretaceous. The evidence from North Eigg shows that the erosion of Jurassic strata from this part of the island was accomplished by Late Cretaceous times, as previously shown for the Laig Gorge district. This is of interest in a regional context, where the relative importance of pre-Late Cretaceous and pre-Paleocene volcanic erosion is debated (Hudson, 1983).

Chapter 5 Palaeogene 1: Introduction and Rum Central Complex Stage 1

Introduction

The terminology used for the divisions of the Palaeogene igneous rocks is summarised in (Table 1).

Palaeogene igneous activity in the Small Isles part of the Hebridean Province, of the Thulean Superprovince, probably began with the basaltic eruptions of the Eigg Lava Formation in the Danian (c.63 Ma, Mussett et al., 1988) and ended in the Ypresian with the formation of the Sgurr of Eigg Pitchstone (c.52 Ma, Dickin and Jones, 1983). Emplacement of the Rum Central Complex took place over about 1 Ma at 58 Ma (Mussett et al., 1988) and consisted of an early episode when acid magmatism was dominant and was closely linked with ring-faulting (Stage 1 of (Table 1)); this was followed by the Layered Suite (Stages 2). Finally, the complex was unroofed and as this was taking place the Canna Lava Formation accumulated. Basaltic dykes, including the Muck regional swarm and the subsidiary Rum radial swarm, were intruded throughout the Danian and the Thanetian, after which they largely ceased. All the Palaeogene igneous rocks have reversed magnetic polarities except for a few normally polarised basaltic dykes that cut the youngest lavas (Chapter 6). Most of the igneous activity took place during the reversed magnetic interval 26r (Harland et al., 1982) but the Sgurr of Eigg Pitchstone Formation is a distinctly later event (23r, (Figure 60))

In this memoir the Palaeogene igneous rocks are described as follows:

Rum Central Complex: Stage One, acid magmatism

Porphyritic rhyodacites form conspicuous outcrops in the Northern Marginal Zone and the Southern Mountains Zone (Plate 6); RR on ((Figure 20) and (Figure 21)). These rocks are termed porphyritic felsite or felsite in earlier accounts (Harker, 1908; Hughes, 1960a; Dunham, 1968). Together with tuffisites, breccias, granophyres and microgranites, they represent an early acid magmatic stage in the development of the Rum Central Complex, which was closely linked with the formation of the Main Ring Fault system (Stage 1 of Emeleus et al., 1985). There are close compositional similarities between the rhyodacites and the Rum granophyres and microgranites (Figure 24) and (Figure 25), despite differences in mode of occurrence, texture and mineralogy.

Porphyritic rhyodacites (Stage 1c on (Table 1))

Judd (1874) suggested that the rhyodacites are lavas but Geikie (1897) recognised they have intrusive features. Harker (1908) gave the first full account of the field relations, petrography and chemistry of these rocks. He believed that the rhyodacites are younger than the gabbros margining the mafic layered rocks since veins of rhyodacite cut and brecciate the gabbros. However Bailey (1945) considered that the gross field relationships show that the gabbros truncate the rhyodacites. Hughes (1960a) and Dunham (1964) showed that the veining and formation of intrusion breccias was the result of ultrametamorphism of the rhyodacite by the later gabbros. They also demonstrated the close association between the rhyodacites, coarse breccias and intrusive tuffs or tuffisites (Hughes, 1955; 1960a; Dunham, 1968), and Dunham (1964) made a study of the intrusion breccias and hybrid rocks formed at the contact of the rhyodacites with the marginal gabbros. Williams (1985) made the significant observation that the rhyodacite on Cnapan Breaca [NM 394 976] has many of the features of pyroclastic rocks including welded tuffs. He concluded that the rhyodacite sheet there is similar in many respects to an ignimbrite. This interpretation has been corroborated by M Errington (personal communication, 1989) from observations in the Southern Mountains Zone, and Bell and Emeleus (1988) have suggested that the rhyodacitic welded ash flows and associated breccias may be caldera-fill deposits. However, some of the rhyodacite bodies in both the Southern Mountains and Northern Marginal zones are intrusive, and these also have some of the features of pyroclastic rocks (cf. Bell and Emeleus, 1988, pp.369–371, pp.374–376). The interpretation of these rocks has thus come almost full circle back to Judd's original suggestions.

Northern Marginal Zone

Four sizeable bodies of rhyodacite crop out in the Northern Marginal Zone (Figure 20). The largest, on Meall Breac [NM 386 983], is at least 100 m thick near the southern end of the ridge. It is truncated to the south by gabbro, where the contact is marked by intrusion breccia and hybrid rocks (Dunham, 1964). The upper surface of the rhyodacite is not seen but at the south-western and south-eastern limits, the rhyodacite is underlain by and intermixed with fragmental rocks, which appear to dip beneath it at low angles. At the south-west end of Meall Breac, about 2 m of pale, unbedded feldspathic sandstone overlain by 2 m of coarse breccia containing fragments of well-bedded sandstone are exposed in low crags [NM 3837 9803]. The coarse breccia is overlain by a variable thickness of bedded tuffaceous rock, followed by the rhyodacite. Two or three metres above the base of the rhyodacite there is a 10–30 cm-thick layer of breccia with sandstone and siltstone clasts in a comminuted sandstone matrix. The upper contact with rhyodacite is sharply defined. However, the lower surface is irregular and gradational. The junction is cuspate; wispy streaks of rhyodacite occur within the breccia and, conversely, fragments of breccia and sandstone are found within the top few centimetres of the rhyodacite. This thin layer of breccia maintains its thickness and position within the rhyodacite for a distance of about 200 m north of the gabbro, ending as several detached pieces which together form a south-facing, S-shaped overfold. The relationships suggest that the fragmental layer may have been disrupted during rheomorphic flow of the rhyodacite. Similar fragmental rocks also occur near the base of the rhyodacite south-east of Meall Breac [NM 3876 9790]; at both localities they have been interpreted as tuffisite (Dunham, 1968, fig. 1).

Many rhyodacite outcrops display a pronounced, discontinuous low-angle streaky banding caused by numerous subparallel, lens-like areas of dark, semivitreous rock in paler surroundings (Plate 7). In addition to the Meall Breac occurrences, examples are common on Cnapan Breaca and Am Màm, and in Dibidil and upper Fiachanis in the Southern Mountains Zone. The structures are interpreted as fiamme (Williams, 1985; Bell and Emeleus, 1988, fig. 6; cf. Cas and Wright, 1987, fig. 8.38; Fisher and Schmincke, 1984, fig. 8.2.).

The ill-exposed valley west of the Meall Breac rhyodacite has been interpreted as fault-controlled (Dunham, 1962; 1968, pl. 25). Outcrops are scanty, and an alternative explanation is that the valley is largely floored by breccia which was formerly covered by a rhyodacite sheet that once extended unfaulted from Meall Breac to Am Màm.

Rhyodacite is in steep to vertical intrusive contact with breccia at and just south-east of the northern summit of Meall Breac [NM 3875 9846]. Nearby, a steep-sided, dyke-like protrusion cuts sandstone 100 m ESE of the summit. Sharp, vertical contacts occur on the slopes to the WNW, and a broad rhyodacite dyke intrudes sandstone and the Am Main Breccia on a shelf [NM 3873 9861] about 130 m north of the summit. Steep to vertical banded structures, striking parallel to the contacts, are common in rhyolite near and to the north-west of the summit. The individual bands may be continuous for several metres and are interpreted as primary flow banding. They contrast with the streaky discontinuous banding at the south end of this hill, which dips at low angles.

The Meall Breac rhyodacite forms a steep-sided, intrusive body at its north end but to the south it conformably overlies and is interlayered with fragmental deposits, some of which are clearly bedded tuffs. Furthermore, it has eutaxitic textures and structures that are characteristic of welded tuffs.

The sheet of rhyodacite on Am Màm [NM 383 987] has a broadly similar structure to the Meall Breac rhyodacite. At its southern end it overlies pale-coloured sandstone that in turn overlies breccia. To the north, the sheet is in steep, intrusive contact with the Am Màm Breccia and coarse-grained gabbro and dyke-like apophyses of rhyodacite also intrude both the gabbro and the Am Màm Breccia.

The rhyodacite of Cnapan Breaca [NM 393 976] has a simpler structure. It consists of a sheet that dips south at 35°. The basal contact is marked by an overhang on the north face of the hill (Plate 6). No roof is preserved but the sheet is at least 80 m thick. The base of the rhyodacite is in fairly sharp contact with fragmental rocks identified as tuffisite by Dunham (1968, fig. 1 and pl. 25) but which are now interpreted to be pyroclastic breccia grading up into tuffaceous rhyodacite (Williams, 1985). The relationships are similar to those at the south-west end of Meall Breac. A thin layer of breccia, in which sandstone clasts are abundant, occurs several metres above the base of the rhyodacite and a further thin breccia crops out about 30 m above the rhyodacite base towards the Coire Dubh end of the hill [NM 3903 9786]. About 100 m south-west of the lochan [NM 3961 9765] east of Cnapan Breaca, bedded lithic and crystal tuffs are overlain by rhyodacite with conspicuous streaky banding. The contact is irregular and the banding is deflected around rounded, upward-protruding masses of the tuffs. Where this happens the bands are highly attenuated, whereas they may have very low aspect ratios between these irregularities. A layer of tuffaceous rock up to 1 m thick occurs 2–3 m above the base of the rhyodacite. The layer has a sharp basal contact but it grades into the overlying rhyodacite.

At the western end of Cnapan Breaca, below the level of the main rhyodacite sheet, an isolated area of rhyodacite is underlain by tuffaceous sediments and pale grey sandstone. Towards its eastern end [NM 3919 9784] this rhyodacite is cut by a micro-granite sheet. While it is possible that the mass slipped downhill to a level about 20 m below the main sheet, it is thoroughly coherent along 150 m of strike and does not have the accentuated jointing commonly present in landslip blocks. It may be a thin ash flow that filled a shallow valley in the breccias, and thus slightly predates the flow(s) on Cnapan Breaca.

A crescentic body of porphyritic rhyodacite intrudes sandstone and breccias about 400 m south of the Coire Dubh hydro electric dam [NM 3931 9828]. It is in steep contact with country rocks and differs in other respects from the Cnapan Breaca rhyodacite. Angular and subangular inclusions of sandstone up to 30 cm across are common and are accompanied by rounded areas of dense, dark rock up to 20 cm diameter, many of which have lobate, cuspate margins against the rhyodacite (Plate 8c). The dark inclusions are of a basic igneous rock and their shapes suggest a liquid-liquid relationship with the rhyodacite (cf. Blake et al., 1965). Thorough, turbulent mixing must have taken place at a high structural level in the complex to achieve the fairly uniform distribution of the basic inclusions and sandstone fragments.

Isolated rhyodacite intrusions occur along the Main Ring Fault. Rhyodacite crops out within the strip of breccia forming a pale-coloured ridge [NM 367 994] 300 m east of the upper bridge over the Kilmory River; it also contributes fragments to this breccia (Dunham, 1968). A rhyodacite plug less than 10 m in diameter intrudes sandstone immediately adjacent to the Main Ring Fault 550 m ESE of the Coire Dubh hydroelectric dam.

Southern Mountains Zone

By far the largest body of rhyodacite on Rum forms the ridge joining Sgurr nan Gillean [NM 380 930] and Ainshval [NM 378 943] (Figure 21). Small areas of breccia cap the rhyodacite at the summit of Sgurr nan Gillean and 400 m to the north-west. The latter exposure is probably an outlier of the layer of breccia that occurs between rhyodacites on the northern and southwestern slopes of Sgurr nan Gillean. The rhyodacite on the ridge forms a sheet that is at least 200 m thick. For the most part rhyodacite overlies breccias but on the north-west side of Ainshval it rests on gneiss and to the south of Bealach an Fhuarain [NM 379 948] it overlies sandstones of the Torridonian Fiachanis Gritty Sandstone Member. On the south-east side of Sgurr nan Gillean, a thin layer of pale grey feldspathic sandstone crops out for a distance of about 400 m, from 450 to 500 m above OD, between the base of the rhyodacite and underlying breccias. This sandstone is similar to sandstones underlying the rhyodacites in the Northern Marginal Zone. A NW–SE-elongated body of rhyodacite occurs between 550 and 400 m above OD on the south-west side of Sgurr nan Gillean where it is cut by the Papadil Granite (M Errington, personal communication, 1989). A sheet of rhyodacite is sandwiched between breccias in the upper part of Sandy Corrie [NM 376 939]. The rhyodacite has well-developed fiamme, especially close to its lower contact with breccia. Fiamme also occur in the base of the rhyodacite on the Sgurr nan Gillean–Ainshval ridge, and in rhyodacite that crops out from Nameless Corrie [NM 386 935], on the north-east side of Sgurr nan Gillean, to the south-east slopes of Ainshval. The fiamme in rhyodacite at 260 m above OD, on the north side of a stream at [NM 3855 9351] in Nameless Corrie, are locally highly contorted, attenuated and folded, suggesting wholesale slumping and rheomorphic flow shortly after deposition. On the north-western and north-eastern slopes of Ainshval, spectacular intrusion breccias define the contact with gabbro and ultrabasic rocks (Hughes, 1960a, pl. 10, fig. 2).

M Errington (personal communication, 1989) has mapped extensive areas of intrusive rhyodacite south-east of Ainshval, south of Sgurr nan Gillean and, in smaller bodies, cutting faulted rocks on the coast near the foot of the Dibidil River [NM 3945 9273]. The contacts with breccias, country rocks and other extrusive rhyodacites are sharply defined and cross-cutting.

Rhyodacite intrudes gneiss and sandstones in the Main Ring Fault zone 500 m WNW of Stac nam Faoileann [NM 407 932], it also caps breccia and sandstone on Beinn nan Stac [NM 397 940] and forms a roof outlier (with breccia) to the ultrabasic rocks about 400 m to the north-west. At several localities in the Southern Mountains Zone, particularly on the eastern slopes of Ainshval, it is not always clear whether the rhyodacite is extrusive, or intrusive and more or less conformable with the associated breccias.

Petrography and mineralogy

The rhyodacites are tough, grey-weathering rocks with conspicuous phenocrysts of white feldspar (to 3 mm long) and grey quartz (1–2 mm in diameter). The groundmass varies in appearance from black and semi-vitreous to dull, matt grey and 'stony'. The rock may be banded; the bands are most obvious on weathered surfaces where they form resistant parallel ridges generally about 2–3 cm apart, or streaks of dark and light coloured rock. The bands are generally several tens of centimetres in length although they may be continuous over several metres, and impart a distinctive streaky appearance to the rocks (cf. (Plate 7)). Both fiamme and flow-banding are most common near the margins of rhyodacites where they strike parallel with the contacts (Figure 21).

Averages and ranges of modes are summarised in (Table 3). The magnitude of the modal ranges suggests that the rhyodacites are not homogenous bodies. However, no systematic variation in modes has been observed.

Fresh rhyodacite contains discrete phenocrysts of bipyramidal quartz and glomeroporphyritic groups of plagioclase, Ca-poor clinopyroxene (Ca9Mg37Fe54), rare Ca-rich clinopyroxene (Ca37Mg23Fe40) and orthopyroxene (Ca4Mg43Fe53), and Fe-Ti-oxides. The pyroxenes are usually altered to amphibole and chlorite but fresh crystals are invariably iron-rich ((Figure 22); Emeleus et al., 1971). Normally zoned phenocrysts of plagioclase (An37 to rims of An containing up to 3% Or) are common and oscillatory zoning may occur (Plate 8a). The sudden shifts in composition have been attributed to a rapid lowering of pH2O on fracturing of the magma chamber (Dunham, 1968). Earlier identifications of orthoclase and sanidine phenocrysts (Geikie, 1897; Harker, 1908) were not confirmed. Quartz phenocrysts occur in every rhyodacite examined but are never present in large amounts (Table 3). The crystals generally have rounded, embayed outlines which indicate resorbtion after inital crystallisation, possibly in response to decreasing pressure as the magma rose towards the surface (cf. Tuttle and Bowen, 1958, fig. 38). Relicts of well-formed bipyramidal crystals are common. Magnetite with a few broad lamellae of ilmenite, and separate ilmenite grains, forms about 1 per cent of the rhyolites. The opaque oxides also exhibit a wide range of ulvospinel-magnetite solid solutions (A C Dunham, personal communication, 1980).

Phenocryst populations in the rhyodacites of Am Marn, Meall Breac, Cnapan Breaca, Coire Dubh and Sgurr nan Gillean are all similar, but that on Beinn nan Stac is different (Table 3); here two generations of zoned plagioclase (An45-19 and An29-18), Ca-rich and Ca-poor clinopyroxenes and orthopyroxene have been recorded (Hughes, 1960a; Emeleus et al., 1971). Although most rhyodacite phenocrysts are euhedral or resorbed, thin sections commonly contain crystals with at least one ragged edge and some of these are clearly fragments of original, euhedral grains.

The fresh rhyodacite matrices vary from holocrystalline assemblages of quartz, alkali feldspar, amphibole and Fe-Ti-oxides to semivitreous, cryptocrystalline rock which is banded on a fine (millimetre) scale (Plate 8b). Rarely, perlitic cracks are found in the marginal rocks, suggesting original glass (Dunham, 1965a). The rhyodacites commonly have indications of a thermal metamorphic overprint. This is generally evidenced by aggregates of opaque oxide grains which pseudomorph original mafic minerals but in extreme instances plagioclase phenocrysts develop sieve-like, fingerprint texture. Although thermal alteration may obliterate the details of the groundmass textures, unaffected rocks in the margins of the rhyodacite bodies commonly have fine-scale, discontinuous streaky structures that resemble shards or, more commonly, collapsed shards that have been compressed and possibly deformed by rheomorphic flow (Plate 8b). Rhyodacites have other distinctive features, especially in their margins. Numerous small (c.1 mm) xenoliths of sandstone, basalt and other igneous rocks are present, some of the rhyodacite crystals are crystal fragments and the distinctive fine-scale banding may consist of close-packed, subparallel pieces of devitrified siliceous glass (e.g. SR 526).

Conclusions

The Am Màm and Meall Breac rhyodacites consist of steep feeder dykes which pass south into extrusive ash flows. The rhyodacite forming Cnapan Breaca is similarly made of ash flows and probably the same situation pertains in the Southern Mountains with the additional complication that there were several distinct episodes of rhyodacite intrusion and, presumably, extrusion.

Breccias in the Northern Marginal Zone and the Southern Mountains Zone

Coarse fragmental rocks occur within the Main Ring Fault in both the Northern Marginal and Southern Mountains zones (Figure 20) and (Figure 21). They are closely associated with intrusive and extrusive rhyodacites and tuffisites. Two distinct types of breccia are recognised: those with a matrix composed typically of comminuted sedimentary rocks are termed the Corrie Dubh Breccias; and those with an igneous matrix are called the Am Màm Breccias. Both contain numerous angular to subangular blocks of Tonidonian sandstones, Lewisian gneiss and, in the Am Màm Breccias, coarse gabbro and rare feldspathic peridotite.

The breccias of Rum were believed by Judd (1874, p.253) to be 'felstone lavas intermingled with agglomerate'. Geikie (1897, p.353) was not certain 'whether the breccias be regarded as the result of earlier rock crush ings, or as due to explosions during the Tertiary period'. Harker (1903, 1908) attributed the breccias to crushing associated with a major thrust plane. This thrust plane was shown to be a ring-fault by Bailey (1945) who emphasised the close association between intrusive rhyodacite and breccia. Hughes (1955, 1960a) and Dunham (1968) envisaged that the breccias were formed when volatiles expelled from crystallising acid magma shattered the country rocks. The breccias were termed 'explosion breccias' in conformity with the use of this term by Richey and Thomas (1932, p.811).

It has been recognised increasingly that coarse breccias in volcanic terrains may also arise because of failures of oversteepened slopes, for example caldera walls and the sides of volcanic cones or domes. The resultant breccias may cover large areas and contain fragments ranging in size from hundreds of metres (megablocks) to centimetres. These coarse deposits may be interbedded with lava flows, ash deposits and pyroclastic flows, but they need not result directly from explosive volcanism. They are essentially epiclastic deposits formed by mass-flow processes (Cas and Wright, 1987).

Coire Dubh breccias and associated tuffaceous rocks

Typical Coire Dubh Breccia consists of angular to sub-angular fragments of Torridonian sandstone or siltstone in a matrix of comminuted sedimentary rocks (Plate 9). Fragments of Lewisian gneiss may be present, and contribute to the matrix; small clasts of basalt or dolerite occur sparingly. The clasts vary from a few centimetres diameter to many tens of metres in the megabreccias. The deposits are generally chaotic (Plate 9) although rough bedding occurs sparingly and the finer-grained tuffaceous rocks are commonly bedded. The breccias may be matrix- or clast-supported.

Northern Marginal Zone

The most extensive area of breccia is in Coire Dubh, to the east of Meall Breac and north of Cnapan Breaca ((Plate 6); (Figure 20)). The contact with the Torridonian Fiachanis Gritty Sandstone Member is exposed on slabs and low crags close to Allt Slugan a'Choilich [NM 3918 9805]. It is inclined to the south at angles varying from about 45° to near vertical. Close to the contact, the clast-supported fragments are 2–6 cm across, coarsening to 5–15 cm after about 1.5 m. Rhyodacite intrudes the breccia on the northern end of Meall Breac [NM 3880 9830] and occurs as a plug partly in breccia south-east of Allt Slugan a'Choilich. On the north side of Cnapan Breaca at [NM 397 977] the breccias are separated from the rhyodacite sheet by several metres of pale grey or cream-coloured, coarse-grained sandstone and bedded tuffs. Tuffs are interbedded with breccia south-east of the lochan east of Cnapan Breaca, at [NM 3964 9757] (Figure 20).

Although the breccia exposures in Coire Dubh, and elsewhere, generally lack obvious structure, poorly developed bedding at several localities dips from SSW to WSW at moderate to steep angles (Figure 20). Three hundred metres SSW of the hydroelectric dam on Ant Slugan a'Choilich, a layer of close packed blocks up to 40 cm across overlies normal breccia. One hundred metres to the south-east a north–south ridge at [NM 3920 9802] exposes roughly bedded breccia with a steep (50–60°) southerly dip. Normally the breccias in Coire Dubh consist entirely of sedimentary material but a slightly finer-grained breccia layer in the southern half of this exposure contains small (<5 mm) black, scoriaceous fragments (SR 437). Crags about 500 m N070° from the dam expose irregular blocks of sandstone, 20–30 m across, in steep contact with the enclosing breccia. These are megablocks, similar to those found in Dibidil (see below). East of Cnapan Breaca, bedded crystal and lithic tuffs crop out for about 150 m south-east of a lochan [NM 3960 9765]. The tuffs dip WSW and grade up into essentially unbedded, coarse breccias. The breccias are in turn overlain by coarse- to medium-grained, well-bedded tuffs which in turn are interbedded with rhyodacite at [NM 3955 9750] (Figure 23). Here, and on the north side of Cnapan Breaca, the tuffs were originally mapped as tuffisites (Dunham, 1968, pl. 25). Similar tuffs underlie rhyodacite at the base of the prominent crag on the north face of Cnapan Breaca where they are cross-bedded (Williams, 1985). Beneath the line of rhyodacite crags (Plate 6), the tuffs are underlain by a variable thickness (0–2 m) of coarse-grained, pale grey or cream-coloured sandstone. The sandstone fines upwards, it is very weakly bedded and the lowest exposures, of gritty sandstone, are brecciated

Breccia also occurs west of Meall Breac and south of Am Màm. At the south-west corner of Meall Breac [NM 3836 9804], about 6 m of breccia underlies bedded tuffs and pale sandstone, and rests on Torridonian sandstone. On the south side of Am Màm [NM 3822 9844], the Coire Dubh Breccias crop out near Am Màm Breccias but the contact has not been located. Pale sandstone underlies the rhyodacite at this locality.

Southern Mountains Zone

The earlier maps of Ainshval and Sgurr nan Gillean depict breccia, 'felsite' and Torridonian sedimentary rocks interlayered on the hillsides, particularly on the north-eastern slopes of Sgurr nan Gillean (Hughes, 1960a, pl. XIV; BGS One-Inch Sheet 60 Rhum, 2nd (provisional) edition, 1971; Emeleus, 1980). Mapping by M Errington (personal communication, 1990) has shown that the Torridonian outcrops consist of megablocks and that much of the 'felsite' formed as rhyodacitic ash flows. The megabreccias consist of sandstone blocks up to tens of metres in diameter, generally separated by normal Corrie Dubh Breccias, for example on prominent, rounded crags at [NM 3875 9415] on the south-east side of Nameless Corrie. The exposures appear at first sight to be normally bedded Torridonian sandstone but are in fact an assemblage of large, close-packed, disoriented blocks of sandstone.

The general features of these coarse clastic deposits are clearly exposed on the slopes north of the Sgurr nan Gillean summit area between about [NM 379 932]–[NM 382 932] and in upper Fiachanis [NM 376 938] (Figure 21). At about 550 m above OD on the north side of Sgurr nan Gillean, a wedge of megabreccia between layers of normal breccia thickens eastwards and is traceable for over 1 km east and south around the hillside. The overlying, 50 m thick layer of normal Coire Dubh Breccia contains rare bedding and is separated from the thick rhyodacite sheets that cap Sgurr nan Gillean by a thin layer of pale grey to cream sandstone, similar to that found in Coire Dubh. The thick summit rhyodacites on Sgurr nan Gillean enclose at least one substantial, ESE-dipping breccia layer and a small area of breccia caps the uppermost rhyodacite (Figure 21).

Breccia crops out near the base of the rhyodacite sheet on Beinn nan Stac and a small area of breccia overlies rhyodacite at the summit [NM 3961 9406]. Nearby, breccia and rhyodacite form a prominent small crag at [NM 3922 9422] underlain by feldspathic peridotite. In Sandy Corrie, at about [NM 376 939], a sinuous sheet of breccia dips between 10–15° ESE. The sheet overlies gneiss and basal Torridonian rocks, it is overstepped by rhyodacite 250 m west of the Ainshval summit cairn, at about [NM 3760 9433] and is truncated by gabbro on the south-west slopes of Leac a'Chaisteil at about [NM 368 934].

The breccias and much of the associated rhyodacite in the Southern Mountains Zone form a sequence of interlayered sheets that are generally inclined at low angles towards Dibidil. However, certain rhyodacite bodies are steeply inclined, strongly discordant and clearly intrusive (see above).

Petrography

The breccia matrices consist of finely comminuted, unbedded sedimentary rocks. Fragments of siltstone and sandstone, from the lower members of the Torridonian succession, form the majority of the clasts in the Coire Dubh breccias. There is little conclusive evidence that the fragments had been affected by thermal metamorphism prior to incorporation in the breccias; rarely, very thin, fine-grained dusty films separate quartz and feldspar grains. A thermal overprint is, however, apparent in breccias and tuffs near to the gabbros and ultrabasic rocks at the eastern end of Cnapan Breaca [NM 395 975].

Small fragments of an opaque-charged, very fine-grained scoriaceous rock occur in certain of the Coire Dubh breccias (SR 437), (SR 438). The fragments consist of innumerable minute opaque granules in a felsic matrix, with chlorite pseudomorphs after original feldspar microphenocrysts. Their bulk composition is not known but they appear to have a high iron content. They bear a resemblance to fine-grained, opaque-rich devitrified glass found in nearby tuffisite dykes.

Fragments of Lewisian gneiss contribute to the breccias on the lower slopes of Sgurr nan Gillean. Elsewhere in the Southern Mountains Zone, breccias in upper Sandy Corrie [NM 377 940] contain fragments of baked shale with abundant, minute crystals of biotite with a strong preferred orientation (SR 527). In this area, breccia predominantly of gneiss and sandstone clasts also contains fragments of an alkali feldspar-bearing quartz porphyry (SR 526). This rock is quite unlike the plagioclase-phyric rhyodacites in the Northern and Southern zones, but it is similar to quartz-porphyry fragments in a breccia clast in the conglomerate of the Canna Lava Formation of West Minishal (Chapter 8). The tuffaceous rocks interbedded with or underlying the rhyodacites, and in a few instances interbedded with Coire Dubh breccias, vary from finely comminuted sandstone to crystal tuffs rich in plagioclase identical to the phenocrysts in the rhyodacite (SR 367).

Conclusions

The virtual absence of igneous rock fragments is a striking feature of the Corrie Dubh breccias in both the Northern Marginal and Southern Mountains zones. Rare fragments of gabbro have been reported (Dunham, 1962) and small pieces of basalt have been found in the pale sandstone breccia (SR 473) beneath the Cnapan Breaca rhyodacite sheet. However, not only is there a lack of igneous clasts but igneous material does not contribute materially to the finely comminuted fragments in the breccia matrices. The breccias were thus formed at a stage in the evolution of the Rum Central Complex when neither the Lewisian gneiss nor the Torridonian sedimentary rocks in southern Rum and its surroundings had been cut to any great extent, at the presently exposed structural level, by Paleocene (or earlier) dykes or other intrusions. The breccias therefore predate the main NW–SE dyke swarm, the dykes that radiate from the Rum centre, the inclined basaltic sheets (cone-sheets) and probably also the majority of the peridotite, gabbro and dolerite plugs.

The features described, together with the rare bedded structures in the coarser facies of the breccias, are compatible with an origin as collapse breccias that formed either directly as landslips off steep caldera walls, or from the reworking of landslip deposits. However, the close temporal and spatial association with rhyodacitic intrusions, ash flows and ignimbrites make it highly probable that some of the breccias are explosion breccias as envisaged by Richey and others, and reworking of this material will also have contributed to the epiclastic breccias. The bedded tuffs that underlie or are interbedded with rhyodacites commonly contain crystal debris derived from rhyodacitic magma. There can be little doubt that they formed during the explosive volcanic events that gave rise to the rhyodacitic ash flows. The pale-coloured sandstones that overlie the breccias and occur beneath rhyodacite at several well-separated localities probably formed from fine material washed out of the breccia matrices.

Explosion Breccias

A strip of breccia forms a low, WNW-trending ridge east of the upper bridge over the Kilmory River [NM 3638 9944]. The Main Ring Fault margins the northern edge of the breccia. This breccia is not associated with any major rhyodacite body, although small intrusions of rhyodacite occur within it and there are sparse rhyodacite and gabbro clasts. The breccia ends at the Long Loch Fault and is cut by gabbro plugs (Figure 20). This area of breccia is regarded as true explosion breccia (cf. Tyrrell, 1928), as are small areas in the Southern Mountains Zone (Figure 21). Explosion breccia fills a small vent which cuts the Western Granite in the Glen Duian River at [NM 3376 9781]. Subangular fragments of granophyre up to 8 cm diameter are set in a matrix of comminuted granophyre and rare, small fragments of fine-grained basaltic rock.

Tuffisites

In addition to the breccias already described, intrusive fragmental rocks occur in both the Northern Marginal and Southern Mountains zones. These rocks were first recognised by Hughes (1960a) in the Southern Mountains and subsequently further examples were found in Coire Dubh (Dunham, 1968). The major tuffisite outcrops in the Southern Mountains Zone are shown on (Figure 21).

The tuffisites differ from the Coire Dubh type breccias in that they are clearly intrusive and, in addition to fragments of the local country rock, they contain country rock fragments brought up from depth, rare pieces of rhyodacite, blebs of devitrified glass and in some instances there are many crystals of plagioclase. Flow structures are brought out by variation in crystal and fragment contents, and in size. Hughes (1960a) called these rocks intrusive tuffs. However, the term tuffisite is preferred (Dunham, 1968) and is used in the sense described by Reynolds (1951, 1954) and Holmes (1965, 269; see also discussion of Hughes, 1960a).

Details

Two large masses of tuffisite occur in the Southern Mountains Zone. One exposed in the lower part of the Dibidil River [NM 393 931] extends south-westward for about 500 m. The second is about 600 m south of the summit of Sgurr nan Gillean and trends due east to the coast, more or less parallel to the Main Ring Fault ((Figure 21)). Except where truncated by later faults, the tuffisites have steep, intrusive contacts against Torridonian rocks, Lewisian gneisses and Paleocene breccias. Intrusive contacts with gneiss and sedimentary rocks are exposed on the waterworn slabs which extend from the ford on the Dibidil River [NM 3933 9307] for some distance downstream. The tuffisites form a complex meshwork of intrusive bodies enveloping relicts of the country rocks (Hughes, 1960a, pl. XIII, fig. 1).

The tuffisites in the Northern Marginal Zone are mostly steeply inclined, thin sheets or dykes that strike in various directions. Two SE-dipping sheets occur to the north-west of the isolated rhyodacite plug in Coire Dubh, at [NM 3930 9794]. The western sheet is 150 m long and up to 70 cm wide; it cuts across the contact between Coire Dubh breccia and Torridonian sandstones. Its lower contact is sharp and the first 10 cm contain black glassy layers parallel to the contact. Above this, isolated angular sandstone fragments up to 5 cm across occur in a grey, fine-grained, compact matrix. The larger inclusions are concentrated in the centre of the sheet The upper contact is gradational towards the Coire Dubh breccia. The eastern sheet is cut by the rhyodacite plug.

North-east of Cnapan Breaca, approximately 100 m west of a lochan [NM 3960 9765], a dull black tuffisite dyke up to 50 cm wide intrudes Coire Dubh breccia. The dyke terminates in a bulbous end, close to the rhyodacite. It is cut by several fine-grained basic dykes and sheets that cut rhyodacite and breccia. Two NW-trending tuffisite dykes up to 1 m in width also intrude the shale and sandstone about 700 m N130° of the hydroelectric dam on Allt Slugain a'Choilich. They have drusy cavities that contain epidote, calcite and prismatic quartz, a feature not found in other tuffisites in the Northern Marginal Zone.

Several occurrences of fragmental rocks originally identified as tuffisite sheets are now interpreted as tuffs in the Coire Dubh Breccias or interbedded with rhyodacite ash flows (e.g. Dunham, 1968, fig. 1: D and C, and the two areas shown at the south end of Meall Breac).

Petrography

Tuffisites in the Southern Marginal Zone contain fragments of Torridonian sandstones and shales, Lewisian gneiss, dolerite, and less abundant gabbro and granophyre. Rhyodacite fragments are present in tuffisites in the Dibidil River section. In the Northern Marginal Zone, the fragments are similar except that granophyre and rhyodacite have not been found. The Dibidil tuffisite matrix consists of 'rounded grains of quartz, feldspar, ore and chlorite, with an average grain-size of about 0.1 mm with occasional larger crystals of augite, zircon, apatite and ore' (Hughes, 1960a, p.121, pl. XIII, fig. 2). Crystals from gneiss and rhyodacite are present, as well as fragments of porphyritic granophyre and rounded pieces of dolerite. In the Dibidil River section, the tuffisite matrix is cemented by quartz, epidote, grossular, sphene and chlorite of hydrothermal origin. In some samples the rhyodacitic contribution is large. Dunham (1968, pp.330–331) noted numerous blebs of devitrified acid glass in the Northern Marginal Zone tuffisites, generally containing 'plagioclase crystals identical to those in the felsite' (= rhyodacite). Spindle-shaped glass blebs are aligned parallel with the tuffisite sheet margins. Epidote, calcite and low-quartz occur in drusy cavities. In a few instances, perlitic cracks occur in the matrix. Glassy areas are flow banded and have relict vesicular textures. The devitrified glassy areas in the tuffisites resemble the rare scoriaceous inclusions in the Coire Dubh breccias.

Am Màm Breccias

Coarse breccias crop out west and north of Am Màm, in the col at the north end of the valley to the east and on a shelf at about 300 m elevation on the north end of Meal Breac (Figure 20). Small patches of similar rocks occur in Coire Dubh, close to the Main Ring Fault at [NM 391 984].

The breccias contain angular to subangular blocks of coarse-grained gabbro, gneiss, sandstone, basaltic rocks and rare feldspathic peridotite. The matrix is a pale creamy white or grey coloured, smooth-weathering, fine-grained microgranodiorite or micro-quartz-diorite. The matrices have pitted surfaces where small (1–20 cm diameter) basic inclusions have been partly weathered away; unlike the inclusions already listed, these are rounded and may have rather diffuse, lobate outlines.

The Am Màm Breccias are quite different to the breccias found in Coire Dubh and the Southern Mountains Zone. In these areas inclusions in the breccias are rarely of igneous rocks and the matrices are largely of comminuted sandstone; however the two types of breccia are not separated on the earlier published maps (BGS Sheet 60 provisional edition, 1971; Emeleus, 1980), despite their differences having been recognised by Dunham (1968). No contact has been found between the two types of breccia, although they are juxtaposed near Loch Gainmich [NM 3814 9848] (Emeleus, 1994), and no fragment of one has been recognised in the other.

Rhyodacite dykes intrude the Am Màm Breccias north of Mean Breac and on Am Màm. The breccias are also cut by numerous basaltic dykes, some of which are affected by the Main Ring Fault immediately to the north, thus demonstrating that the breccias are older than at least the later movements on this fault system.

Angular blocks of coarse-grained gabbro, ranging in diameter from about 1 m to over 10 m, are common in the Am Màm Breccias. Much larger areas of a similar gabbro have been mapped as intrusive plugs east of Loch Gainmich and in the col at the north end of Three Lochs Valley (e.g. Emeleus, 1980). The gabbro east of Loch Gainmich [NM 382 988] is cut by Am Màm Breccia on its western side and by a rhyodacite dyke to the east. The 100 m-diameter gabbro mass in the Three Lochs Valley [NM 3855 9865] is surrounded by breccia and intruded by several breccia dykes. Both gabbros are extremely coarse grained up to their contacts with breccia; they are now interpreted as megablocks of an early gabbro that became entrained in the breccias. Additional evidence for pre-breccia plutonic mafic intrusions is found at a locality [NM 3876 9859] on the north of Meall Breac where gabbro and rare inclusions of a coarse feldspathic peridotite (e.g. SR 381) occur in sheets of breccia that intrude bedded sandstones.

Petrography

The breccia matrices have a speckled, inhomogenous appearance due to small, ill-defined darkish areas and the presence of acicular mafic minerals. A typical example (SR 401) from the north end of Meall Breac consist of fine-grained crystals (c.0.1 mm diameter) of greenish brown biotite, dusty plagioclase, patches of opaque oxides (which may be after original mafic minerals) and interstitial quartz, alkali feldspar and minor calcite. Apatite occurs sparingly and there are quartz xenocrysts rimmed by amphibole. Scattered plagioclase crystals up to 2 mm across have clear cores and sieve-textured rims similar to those in altered gneisses (Chapter 2), from which they may have been derived. Fresh microgabbro inclusions and fragments of sandstone are present. The small dark, rounded, lobate inclusions e.g. (SR 406) are of dolerite or basalt. Breccia with a coarse-grained matrix (SR 398) from Am Màm contains augite, and plagioclase with pronounced marginal normal zoning.

Conclusions

The Am Màm Breccias are an early event in the Northern Marginal Zone since they predate the rhyodacite intrusions. They are intrusive breccias and are quite different from the breccias of Coire Dubh type. The rocks they most closely resemble, but clearly predate, are the intrusion breccias that margin the Layered Suite. The Am Màm Breccias provide crucial evidence of the presence of plutonic mafic rocks in the Rum Central Complex prior to emplacement of the acid rocks of the Northern Marginal Zone. When this evidence is considered with that provided by the rounded basaltic and doleritic inclusions, which suggest a liquid–liquid' relationship between the acid material of the matrices and basaltic magma, it is likely that the Am Màm Breccias originated as marginal intrusion breccias, formed when mafic intrusions were emplaced into acid (Lewisan gneiss?) surroundings. The breccia did not remain where it was generated, but intruded forcibly to higher structural levels, possibly guided by the Main Ring Fault.

Western Granite and related intrusions

There are three areas of granitic rock in Rum. Granophyre, generally drusy, and fine-grained drusy micro-granite are the commonest rock types. The largest intrusion is the Western Granite which covers about 14 km2 in western Rum (Figure 1). The Long Loch Granite occupies about 0.25 km2 to the north-east of the loch (Figure 20). The third mass is termed the Papadil Granite and covers about 0.25 km2 on the south-west slopes of Sgurr nan Gillean (Figure 21).

The granitic rocks weather to smooth, well rounded hills. Orval [NM 334 991] at 571 m above OD, is the highest point. The south-west margin of the Western Granite is formed by a line of steep cliffs up to 200 m high, with aprons of scree which partly mantle a well-developed rock platform at about 20–30 m altitude (Plate 10). Judd (1874) believed the granites to be older than the lavas on Orval, whereas Geikie (1897) considered the lavas to form the roof of the intrusion. From the evidence of acid veins that freely cut the marginal gabbros and peridotites of the Layered Suite (Plate 20), Geikie also suggested that the granites are younger than the plutonic mafic rocks rocks to the east. Harker (1908, Chapter IX) provided a brief account of the granites and also concluded that they are later than the mafic rocks. He considered that the gneiss on Ard Nev [NM 346 986], in common with other gneisses on Rum, had been formed when 'eucrite' (= bytownite-gabbro) was permeated by acid magma. The Ard Nev occurrence was thought to be a more advanced stage in the invasion of acid magma than the hybridisation and veining found at the eastern contact of the Western Granite. After initially agreeing with Geikie's views on the age of the granites, Bailey (1945) reversed his opinion when Black (1952b) demonstrated that weathered granite underlies the lavas on the north side of Orval [NM 3343 9927]. Black (1954) identified a central microgranitic facies of the Western Granite which grades into an outer granophyre. From the evidence of an apparently transitional contact at Minishal [NG 354 005], he also suggested that the Western Granite originated through the in-situ granitisation of Torridonian arkosic sandstone. This view was challenged by Hughes et al. (1957) and Hughes (1960a), who showed that the supposed transitional zone was a thermally metamorphosed fault breccia. Dunham (1968) stressed the chemical and petrographic similarities between the granites and the porphyritic rhyodacites of the Northern Marginal Zone and the Southern Mountains Zone. Dunham and Thompson (1967) considered that the granitic magmas came from the fusion of Lewisian and Torridonian rocks and disagreed with the suggestion by Kleeman (1965) that the rocks formed from differentiated or contaminated and differentiated basic magmas (see also Kleeman, 1967).

Details

A number of varieties of acid rock were encountered during the remapping of the Western Granite. However, the generally indifferent outcrop made it difficult to confirm the pattern depicted by Black (1954, fig. 1; Emeleus, 1980). The northern exposures from the Main Ring Fault to the ridge between Sròn an t-Saighdeir [NM 318 989] and Orval are of medium- to fine-grained granophyre with feldspar phenocrysts (1–2 mm long) and small drusy cavities. Medium- to coarse-grained granophyre crops out on Ard Nev (even close to the gneiss), in the upper reaches of Glen Duian, and in the river 300–400 m upstream from a group of ruined shielings [NM 3352 9736]. To the west, fine-grained, spherulitic granophyre forms white-weathering outcrops about 600 m NNW of Loch Monica [NM 332 966].

An internal contact occurs at 340 m altitude in the Glen Duian River, at [NM 3402 9843]. In about 400 m of intermittent stream exposures, coarse granophyre in the north gives way abruptly to a fine-grained feldspar-phyric microgranite which coarsens southwards to medium-grained microgranite. The granophyre on the coast, north-westwards from Harris, is medium to coarse grained. It shows a compositional break about 2.8 km north-west of Harris although no textural change is found. There may, therefore, be a central, later plug of microgranite and granophyre but of smaller size than previously envisaged by Black (1954).

The faulted, northern contact between the Western Granite and the Torridonian rocks was recognised by Black (1954). Exposure is generally poor but a steep fault, inclined to the south, is exposed in the east-facing cliff [NG 320 000] about 800 m south-east of Bloodstone Hill. The sinuous course of the contact between granite and Torridonian sandstones from the cliff-tops south-east of Camas na h-Atha [NM 302 997] to Minishal is an indication that the contact is steeply inclined to the south-east or south. A faulted contact is also exposed south-west and south of Minishal; close to the gabbros on the south-east of this hill (at [NG 3548 0024] ), the fault breccias are thermally altered and the cataclastic features are overprinted and rather obscured, giving an apparent transition from granite to sandstone.

On the south-western edge of the granite, sea cliffs expose granite intruded by numerous thin, NW-trending basic dykes and inclined sheets (Plate 10). The straight coastline between Wreck Bay [NM 309 981] and Harris is probably controlled by the south-eastern continuation of a right-lateral fault from Wreck Bay to Camas na h-Atha (Figure 53). Several other NW- to NNW-trending faults offset the northern boundary of the Western Granite; all show a right-lateral sense of horizontal displacement, each between 200 and 400 m (Figure 53).

Granite is exposed in two small outcrops adjoining the Harris Bay bytownite-gabbro ('eucrite') at the north-west and south-east ends of Harris Bay [NM 3360 9545] and [NM 3405 9505] respectively where the complex relationships between granophyre and the later mafic rocks are clearly seen in low cliff exposures. There is excellent evidence for rheomorphism of the acid rock with back-veining of the gabbro (Plate 20). Similar features occur wherever the granophyre/gabbro junction is exposed between Harris and Minishal.

A possible roof to the Western Granite occurs on Ard Nev [NM 346 986]. The summit is formed by a poorly exposed area of baked Lewisian gneiss. Another small area of gneiss occurs 700 m to the south-east of the summit (see Chapter 2) which may also be part of the original roof granophyre (Dunham and Emeleus, 1967, fig. 3, p.399).

The Long Loch Granite is badly exposed. The faulted western margin is visible in an aqueduct [NM 3637 9926] that drains the Long Loch north into the Kinloch River. Crushed granite abuts against gneiss or peridotite and marks the line of the Long Loch Fault. The northern margin is defined by a ridge of explosion breccia but no contact has been seen and no granite fragment has been found in the breccias. Later mafic intrusions form parts of the east and south-east margins and a small composite gabbro/feldspathic peridotite mass intrudes the centre of the granophyre. Elsewhere, gneiss borders the granophye, but the contact is nowhere exposed.

The Papadil Granite is generally poorly exposed. Crushed granite on the southern margin has been affected by the Main Ring Fault. Elsewhere, -the margins are largely obscured by scree although exposures about 600 m south-west of the summit of Sgurr nan Gillean [NM 3805 9305] show that the granite is younger than the porphyritic rhyolite (M Errington, personal communication, 1989). Marginal gabbro and peridotite of the Central Intrusion (Layered Suite) have thermally altered the western edge of the Papadil Granite (Dunham and Emeleus, 1967).

The Papadil Granite, the Long Loch Granite, and a narrow, faulted strip of granite about 500 m north of the road bridge at [NM 3637 9964], are probably all parts of the Western Granite.

Petrography

The granophyres and microgranites commonly weather to light brown or grey colour rock. In places they contain conspicuous drusy cavities. Feldspar phenocrysts (<3 mm) are usually visible, especially in the finer-grained varieties.

The granophyres contain phenocrysts of plagioclase (An27–15), alkali feldspar, iron-rich pyroxenes (ferroaugite, Ca32Mg21Fe47; ferropigeonite, Ca9Mg23Fe68; (Figure 22); Emeleus et al., 1971), rare fayalitic olivine, hornblende, biotite and Fe-Ti-oxides. The groundmass forms the bulk of the granophyres (60 to 70 vol.%). There is considerable textural variation, especially within the Western Granite. Over much of this intrusion, and in the Long Loch Granite, there is a well-developed granophyric texture with quartz and turbid, brown alkali feldspar in micrographic intergrowth. This is of radiating fringe to irregular type around the phenocrysts (Leighton, 1954). In the interstitial areas an almost granitic texture may be developed, although on a fine scale. Small amounts of apatite, amphibole, biotite, chlorite and magnetite also occur in the groundmass. Drusy cavities may be present, especially in the Papadil and the Long Loch granites, and in the northern part of the Western Granite. The cavities contain prismatic quartz, chlorite and calcite.

The modal proportions of quartz and feldspar have been measured in some of the coarser-grained granophyres. This proportion, while highly variable on the square millimetre scale, is remarkably uniform on the thin-section scale, at around 43:57 of quartz to feldspar. Analyses of groundmass feldspar from the Long Loch Granite show a wide range, from K-rich phases through possible anorthoclase to potassic oligoclase.

Geochemistry of the rhyodacites and granitic rocks

Selected major and trace element analyses of the rhyodacites and granitic rocks are given in Appendix 5(b, c). The silicic ash flows and minor intrusions in the Northern Marginal and Southern Mountains zones are rhyodacites with compositions that vary little from area to area despite apparent modal variation (Table 3) The granophyric and microgranitic rocks are generally quartz-monzogranites (cf. LeMaitre, 1989) but some granitic rocks do occur (Appendix 5c). There is close compositional identity between the rhyodacites and the least-evolved granitic rocks (Figure 24), (Figure 25); all have Na2O>K2O together with relatively high Ca, Mg and total Fe contents when compared with other silicic rocks from the. Hebridean Province (Dunham and Thompson, 1967; Thompson, 1982). Normalised incompatible element plots of the Rum rocks emphasise the compositional similarities of the rhyodacites and granitic rocks (Figure 25); compared with the granitic rocks of Skye and Mull (e.g. (SK69), (SK127) and (M177), (Figure 25) the Rum rocks are most comparable to the least evolved Glamaig Granite (SK69) although they have less Ba and P. They are distinctly different from the Beinn a'Ghraig Granophyre of Mull (M177, taken as representative of many granitic rocks in the Province), and none of the Rum rocks resembles the highly evolved Southern Porphyritic Granite of Skye (SK127). Overall, they are closer in composition to the least-evolved granitic rocks which crystallised early in episodes of siliceous magmatism in the Hebridean and Irish Palaeocene central complexes.

A suite of samples of granitic rocks was collected by A C Dunham from the Western Granite in a traverse some 2.8 km in length along the cliff tops north-west of Harris. The suite crosses the boundary between the granophyre and the south-eastern part of the area mapped as microgranite by Black (1954, fig. 1), where a distinct composition change occurs (see Appendix 5c): SiO2 rises towards the south-east, whereas P2O5, TiO2, CaO and Na2O decrease. In addition, the Na2O/K2O ratio changes from greater than unity in the granophyre to less than unity in the microgranite. These data support Black's view that an internal boundary exists in the Western Granite hereabouts. Taken together with the limited evidence of an internal intrusive contact in Glen Duian (see above), there appears to have been a later pulse of more-evolved granitic magma which crystallised as sparsely porphyritic microgranite in the central part of the Western Granite.

Meighan et al. (1982) have found out that there are very close geochemical and petrographic similarities between the Rum granophyres and granophyre clasts that occur in some abundance in the conglomerates interbedded with the Skye Main Lava 'Series' south-west of Glen Brittle (see Chapter 8). This evidence, together with the presence of clasts of Rum granophyre in the conglomerates of Sanday, provide crucial evidence in the relative dating of Palaeocene igneous activity in the Small Isles and Skye.

Chapter 6 Palaeogene 2: Minor intrusions

Introduction

Dykes, plugs, sills and other sheets of a variety of rock types intrude many of the Palaeocene extrusive and intrusive rocks and are locally abundant in the pre-Palaeocene sedimentary and metamorphic rocks. These minor intrusions are most common in and around the Rum Central Complex but also occur widely on Eigg and Muck, which are on a regional NNW- to NW-trending dyke swarm. On Canna and Sanday, a few dykes cut the relatively young lavas of the Canna Lava Formation. A single composite dyke and several felsite sheets intrude the Eocene Sgurr of Eigg Pitchstone Formation on Eigg (Table 1).

Minor mafic intrusions:gabbro, dolerite and peridotite plugs<span data-type="footnote">A comprehensive account of these peridotite plugs appeared after this memoir went to press (Wadsworth, 1994) </span>

Details

Rum

Plugs of gabbro, dolerite and peridotite intrude the Torridonian sandstones and members of the Rum Central Complex (Figure 27) for the localities of plugs within the Northern Marginal Zone, (Figure 21) for the Southern Mountains Zone and (Figure 35) for detail within the Layered Suite. The numbers 1-13 refer to the localities where dyke orientations have been measured (see (Figure 28))." data-name="images/P936601.jpg">(Figure 26). The intrusions range in size from elongate bodies several hundred metres in length, to smaller masses less than 10 m in diameter. The plugs bake and bleach the sandstones which also become splintery and close-jointed, giving rise to patches of conspicuous, pale-coloured scree amongst the Torridonian scarps. At least 40 plugs have been mapped in northern Rum.

The peridotite plugs commonly form rough, brown-coloured outcrops, as at Loch Beauty [NG 359 017] and west of Loch Sgaorishal [NG 347 022]. A large peridotite plug on the north side of Kinloch Glen has, however, been deeply eroded by the Allt Airigh Thalabhairt [NG 379 006]. The peridotite plugs generally lack internal structure apart from blocky jointing, etched preferentially by weathering, and matrix banding (Dunham, 1965b) which occurs in the large plug west of Loch Sgaorishal and is similar to that present in peridotite tongues in the central complex. Angular fragments of dunitic peridotite (DU 9856) occur in a feldspathic peridotite plug at the south end of the lochan [NG 356 009] north of Minishal but inclusions are otherwise rare. The peridotite plugs of northern Rum are generally elongate and appear to radiate from the northern end of the Rum Central Complex (Figure 27) for the localities of plugs within the Northern Marginal Zone, (Figure 21) for the Southern Mountains Zone and (Figure 35) for detail within the Layered Suite. The numbers 1-13 refer to the localities where dyke orientations have been measured (see (Figure 28))." data-name="images/P936601.jpg">(Figure 26). No dyke or other minor intrusion has been found intruding these plugs which are therefore late in the Paleocene intrusive sequence and probably coeval with peridotites, including plugs, in the central complex (Table 1).

The dark-weathering gabbro and dolerite plugs are less conspicuous than the peridotites and are easily overlooked. West of Kinloch, a peaty hillock [NM 3965 9970] within 150 m of the road marks the site of a 30 m diameter dolerite plug. A slightly larger plug forms low crags [NM 397 985] beside a stream south of Kinloch; this is relatively conspicuous by reason of an associated train of erratic boulders. Several substantial plugs on the hillside south of the bridge over Allt Bealach Mhic Néill [NM 3802 9982] are difficult to distinguish from slabs formed by the north-dipping sedimentary rocks. The elusive character of the intrusions makes it likely that other dolerite and gabbro plugs remain undetected. It is very rare to find minor intrusions that cut the gabbro or dolerite plugs; an example is seen in a NNE-trending basalt dyke which cuts a plug of crumbling gabbro exposed in a small quarry [NM 3778 9985] in Kinloch Glen. Further evidence for the relatively young age of the gabbro plugs is seen west of Loch Duncan [NM 372 991] and probably also north of Loch Gainmich [NM 380 980] where gabbros cut the Main Ring Fault. Plug-like masses of coarse-grained gabbro east of Loch Gainmich and east of Am Màm are interpreted as megablocks in the Am Màm Breccias (Chapter 5; Dunham, 1968, pl. 25, 'early gabbros').

Petrography and chemistry of plugs on Rum

The peridotite plugs consist of variable amounts of rounded to euhedral Mg-rich olivine crystals in a matrix of clinopyroxene and calcic plagioclase. The large plug west of Sgaorishal consists of close-packed euhedral to subhedral olivine crystals up to 2 mm in length and small euhedral chromite crystals, in a sparse matrix of anhedral plagioclase and clinopyroxene. On the south-west edge of the plug, there is almost no grain size change at the contact with thermally altered arkosic sandstone. However, at its northern contact, the peridotite adjoining calcareous Triassic sandstone contains less olivine and the feldspathic matrix has many bunched groups of slender plagioclase crystals that resemble quench texture (DU 9854). Rounded patches (c.1 cm diameter) of sulphide minerals, including chalcopyrite (DU 9854), are present. The peridotite plug at the bridge over Allt Bealach Mhic Néill is much finer grained. The close-packed, subhedral olivine crystals are less than 0.5 mm in length (DU 9858) and the marginal rock is slightly finer grained than in the central parts. Peridotite on the north side of Kinloch Glen, at [NG 380 007], has a similar texture to some feldspathic peridotites of the central complex; euhedral olivine is completely enclosed by large poikilitic crystals of calcic plagioclase and diopsidic augite. Elsewhere, as in the plug 450 m north of Rubha na Meirlich [NM 368 911], large olivine crystals (to c.1 cm in length) lie in a doleritic matrix of normally zoned plagioclase laths, anhedral augite, opaques, biotite and secondary chlorite.

The bulk chemical compositions of the peridotite plugs reflect the high modal proportions of magnesian olivine (Fo85): Mg ranges from 25 to >40 wt%, Ni from 1300 to 2500 ppm and Cr 2200 to 4500 ppm, Al2O3 generally <10%, K2O <0.12% and Na2O <1% (Forster, 1980). Spinel as discrete crystals and as inclusions in olivine gives rise to the high Cr in these rocks.

The plugs of basaltic composition vary from coarse-grained gabbro, with plagioclase laths up to 1 cm in length, to normal dolerite. A typical rock consists of normally zoned plagioclase, euhedral to subhedral augite, opaque oxides (which may be skeletal) and interstitial quartz, micropegmatite, biotite, chlorite and epidote. Secondary alteration of both feldspar and pyroxene is common. If present, olivine is generally altered to serpentine and opaque oxides, but fresh olivine-dolerite or gabbro does occur, as in a small plug (SR 422) on the east side of the peridotite at Allt Airigh Thalabhairt. Chemical data are limited; they indicate that the basic plugs are mainly of transitional basaltic compositions (Forster, 1980), with quartz-basalt and alkali basalt present but less common.

Petrography of contact metamorphosed Torridonian sandstones

The thermal alteration around the plugs and other minor intrusions rarely extends for more than 20 m into the country rocks. Alteration is somewhat more pronounced around the gabbro plugs as compared with the peridotites. The effects are essentially similar to those described from rocks adjoining the mafic members of the central complex (Chapter 7), except that evidence for melting is less common. The alteration of sandstone in the fissure breccias is also similar (see below).

The unaltered sedimentary rocks are normally reddish brown in colour. They consist of quartz, microcline, small amounts of plagioclase, micas and heavy minerals including opaque oxides, zircon and epidote. The initial alteration involves a change to a dull grey colour in hand specimen, with the reduction of haematite to magnetite. Closer to the plugs, light-coloured areas, 2–5 mm in diameter, develop in which quartz and alkali feldspar are in spherulitic intergrowth. With more extreme alteration the spherulitic intergrowths coalesce to form polygonal structures up to 1 cm in diameter, for example, next to a small gabbro plug in Allt Bealach Mhic Néill (SR 317 [NM 3802 9967]). Initially, original microcline or orthoclase is converted to sanidine but discrete alkali feldspar disappears on reaction with quartz to give a turbid, felsitic or microspherulitic matrix in which relict grains of quartz are generally associated with acicular quartz paramorphs after tridymite (Plate 21a). Cordierite and orthopyroxene occur sparingly (SR 481). At the eastern contact of the peridotite on the north side of Kinloch Glen, at [NG 3803 0073], a thin zone (c.3 mm thick) of glassy buchite (SR 177) occurs against peridotite. A limited amount of hybridisation occurs at the contacts of a few of the gabbro plugs.

In a few rare instances the alteration was locally sufficiently intense to produce small amounts of partial melting. Thin acid sheets and veins are found in a few of the mafic plugs, for example in the gabbro 150 m south of the Allt Bealach Mhic Néill bridge [NM 3802 9982]. The rheomorphic rocks, which are texturally similar to those found in rocks marginal to the central complex, consist of subrounded, irregular quartz grains in a fine-grained felsic matrix of intergrown quartz and turbid alkali feldspar, which also contains needle-like, inverted tridymite.

The intense thermal metamorphism and limited partial melting indicate that considerable volumes of magma have passed through the plugs. The gabbroic plugs may have fed surface flows, although no lavas of comparable compositions have been identified. The general lack of chilled contacts to the peridotite plugs suggests that they were emplaced as dense suspensions of olivine crystals in picritic or basaltic liquids.

Eigg

Two dolerite plugs intrude the Eigg Lava Formation. One forms a low NNW–SSE hill [NM 4843 8634] 600 m ESE of Eigg post office. The adjoining fine-grained basalt and mugearite flows are unaltered. The plug is cut by thin, flat-lying sheets of fine-grained basalt, possibly of Kildonnan-type (see below). The prominent crag of Gualainn na Sgurra [NM 4676 8465] is formed by an elongate NNW–SSE plug of coarse ophitic olivine-dolerite (SR 559). The plug, which is about 120 m in length, intrudes but does not alter olivine-basalt flows.

Muck

A small plug of analcime-dolerite (at [NM 4155 7899]) intrudes feldspar-phyric basaltic hawaiite 150 m N346° from Fionnairdh. The rock (SR 576) is coarse grained, with ophitic crystals of lilac-brown hourglass-zoned augite, normally zoned plagioclase laths, irregular, angular opaque crystals commonly intergrown with brown, strongly pleochroic biotite, and anhedral areas of analcime. Thin rims of aegirine-augite occur on pyroxene adjoining analcite. There is much secondary alteration with extensive replacement of feldspar by zeolites. Irregular areas of deep red-brown (?)iddingsite partly replace augite.

Sills and conformable sheets

During the original survey, broadly conformable thick sills of dolerite and (less commonly) mugearite were mapped as intruding the Palaeocene lava flows (Harker, 1908, chapter 10; also 1st edition of Sheet 60, 1903). These are now interpreted to be the massive central parts of the lava flows (cf. Anderson and Dunham, 1966).

Details

Rum

Thin basaltic sills intrude the Torridonian rocks. The sills, which are generally less than 1 m thick, are common in the cliff exposures. They are cut by the Mullach Ard and Welshman's Rock faults (Chapter 10), where they contribute basalt fragments to the carbonate-impregnated fault breccias at Bàgh na h-Uamha [NM 4231 9741] and the south end of Welshman's Rock [NM 4172 9464]. No intersections were found with dykes or other intrusions. The rock is generally a fine-grained basalt with scattered calcic plagioclase phenocrysts and rare pseudo-morphs after olivine. At Welshman's Rock there is evidence of multiple intrusion, the dolerite is cut by very fine-grained basalt with tachylitic selvedges.

Basic sills, from 1 to 4 m thick, intrude the Torridonian sedimentary rocks between Dibidil and Papadil, south and east of Loch Meirlich [NM 373 914], and crop out intermittently for about 1.6 km to the ENE. A sill cuts a NW-trending basalt dyke 400 m east of Loch Dubh an Sgoir [NM 381 916], and a small dyke or elongate plug of feldspathic peridotite intrudes a sill 150 m north-west of the loch. Harker (1908, p.164) classified these sills as basic andesites; analysed samples (Forster, 1980) are tholeiitic basaltic andesites.

Several NW-dipping basic sills intrude the Torridonian and Triassic rocks of north-west Rum. The sills extend for about 2 km south and south-west from the vicinity of Kilmory Lodge [NG 3576 0391]. They are 0.5 to 2 m thick and form prominent, SE-facing scarps and long, pale grey, rubble-covered dip slopes. None of the sills was found in contact with other minor intrusions. The rock is microporphyritic with phenocrysts (to 1 mm in length) of fresh, normally zoned plagioclase, smectite pseudomorphs after olivine and rare fresh augite. Turbid, interstitial areas (of orthoclase?) occur in the otherwise fresh matrices. The rocks are tholeiitic basalts or basaltic andesites similar to those described from south-east Rum (see above).

Eigg

Numerous thin basaltic sills intrude rocks of the Staffin Shale Formation east of Clach Alasdair [NM 453 884] and the Leak Shale Formation on the coastal platform north-west of Blàr Mór [NM 472 904]. A few sills intrude the Valtos Sandstone Formation. Thin basaltic sheets cut the quartz-porphyry body on Eilean Thuilm [NM 482 913] and at the east end of the cliffs at Uaimh Mhic' lc Ailein [NM 488 910]. In Laig Gorge [NM 473 875], the junction of the Cretaceous limestone and Paleocene lavas is marked by a thin basalt sill. Scattered thin (c.1 m) inclined basalt sheets intrude the lavas of north-west Eigg near Bidein Boidheach [NM 440 865].

A distinctive group of fine-grained basic sheets intrudes the Paleocene lavas and the underlying Valtos Sandstone Formation at Kildonan [NM 4900 8503]. The sheets, which commonly form low scarps, are generally <1 m and rarely >2 m thick and characteristically have thin (2–3 mm) tachylitic selvedges. Many crop out in the vicinity of Poll nam Partain [NM 487 848] and in the cliffs east and north east of Kildonnan. They form the reefs of Garbh Sgeir [NM 490 840] at the entrance to Eigg Harbour (Harker, 1908, pp.165–168). A pitchstone dyke cuts a sheet at the North Pier [NM 4832 8415] and at Na Gurrabain [NM 4867 8440] thin sheets intrude a NNW-trending basalt dyke. These are termed the Kildonnan-type basic sheets. They are generally aphyric and contain prismatic augite, zoned plagioclase, abundant opaque oxides, apatite and brown, anhedral areas probably of devitrified glass. The tachylitic selvedges have microphenocrysts of plagioclase. Vesicles contain chlorite, clay minerals and calcite. Harker classified the sheets as basic andesites but analyses by Allwright (1980) show that they range from basalt to basaltic hawaiite.

Muck

A 30 m-thick sheet of massive, columnar-jointed olivine-dolerite forms Am Maol [NM 426 805]. The sheet overlies the reddened, fragmental deposits on top of an olivine-basalt flow at Fang a'Ghille Ruaidh [NM 4268 8034], where both are cut by a NNW-trending basalt dyke. The upper contact of the sheet is not preserved. The rock (SR 571) contains numerous phenocrysts of zoned olivine, between 1–2 mm in length, and glomero-porphyritic aggregates of olivine in a coarse, ophitic matrix. The rock has been analysed (Appendix 5g; Ridley, 1973, table 4, Analysis M.4).

Inclined sheets on Rum

A large number of thin (c.1 m thick) inclined sheets of basalt and feldspar-phyric basalt intrude the Torridonian rocks on the northern edge of the Northern Marginal Zone (Figure 27). They are also abundant in the Dibidil area where they intrude all members of the Southern Mountains Zone (Plate 11) , with the possible exception of the poorly exposed Papadil Granite. The inclined sheets are members of cone-sheet systems.

North of the Northern Marginal Zone the inclined sheets crop out in Allt Slugan a'Choilich [NM 395 987], between about 25 and 140 m above OD. About 1.3 km to the west, many inclined sheets crop out as low ridges in the flat ground north of Loch Gainmhich and Loch Bealach Mhic Néill, and continue almost as far west as the Long Loch. The sheets dip in a generally southerly direction at angles between 30 and 70°, with the majority between 30 and 40°. They change strike progressively from NW–SE in the east to near NE–SW close to the Long Loch, and form a cone-sheet system which focuses 1 to 1.5 km depth beneath Glen Harris (Figure 27), 1 to 2 km north-east of the foci of the radial dyke swarm (s). In Coire Dubh, to the south of the Main Ring Fault, a few inclined sheets intrude rhyodacite and Coire Dubh Breccias. Farther west, between the Long Loch and Loch Duncan [NM 372 991], they are present in the Torridonian gritty sandstones, the Lewisian gneiss and the granophyre. The cone-sheet system postdates at least the earlier movements on the Main Ring Fault since it appears to be unaffected by the fault system north-west of Loch Bealach Mhic Néill (Figure 27). There are sparse intersections with other minor basic intrusions; three NNE-trending basalt dykes cut inclined sheets in and near Allt Slugan a'Choilich and inclined sheets are cut by gabbro plugs north of Loch Bealach Mhic Néill and west of Loch Duncan.

In the Southern Mountains Zone, bare slabs and buttresses of rhyodacite in Dibidil are cut by a multiplicity of minor intrusions (Plate 11). These include basaltic dykes and inclined sheets with varied dip and strike directions. Hughes (1955) noted three sets of inclined sheets: an early set dipping at about 25° to the N, a later set at 40° to the NNE and another inclined to the south west. However, there are many exceptions to the sequence and it is difficult to detect any systematic variation in direction with time. Detailed mapping of the dykes and sheets in Nameless Corrie [NM 385 936] (Forster, 1980) demonstrated two principal direction of dip: one is approximately north, the other from WSW to NW. Thus, it is likely that there are two or more sets of cone-sheets in the Southern Mountains Zone. The inclined sheets are cut by the Layered Suite; close to the contact they are intruded by rheomorphic acid veins and contribute angular blocks to the intrusion breccias.

Aphyric and sparsely feldspar-phyric basalts make up the majority of these sheets. They contain small (<0.5 mm) plagioclase and rare augite microphenocrysts in a matrix of fine-grained plagioclase laths, prismatic to subophitic augite, and opaque oxides. Olivine is uncommon. Chlorite and carbonate occur as secondary minerals and both may be found in small amygdales (<0.2 mm diameter). The majority of the sheets are of transitional basaltic compositions, with a few more fractionated examples (Appendix 5j; Forster, 1980, groups 3 and 4).

Dykes

The Small Isles are towards the north-western end of a Palaeocene basaltic dyke swarm that extends to Ardnamurchan and the south-east, where it merges with the Mull dyke swarm (Speight et al., 1982, fig. 33.3). The greatest concentration of dykes occurs on Muck (with up to 8 per cent dilation) and the south-east coast of Rum (up to 4 per cent dilation). The dykes are predominantly NW trending, but there are indications of a radial swarm, or swarms, around the Rum Central Complex (Figure 27) for the localities of plugs within the Northern Marginal Zone, (Figure 21) for the Southern Mountains Zone and (Figure 35) for detail within the Layered Suite. The numbers 1-13 refer to the localities where dyke orientations have been measured (see (Figure 28))." data-name="images/P936601.jpg">(Figure 26); Harker, 1908, pl. 3).

Details

Rum

The mafic dykes on Rum largely predate the Rum Layered Suite and the Canna Lava Formation (Table 1). Dykes are numerous in south-east Rum, in the Torridonian rocks between Kinloch Glen and the Main Ring Fault, and on the south-eastern and north-western coasts. Within the Main Ring Fault, many dykes intrude the earlier felsic rocks of the Rum Central Complex as well as the pre-Tertiary rocks. They are generally between 0.2 and 1.5 m in width although a few are considerably wider. Few may be traced for more than 20 m, partly because of a tendency to weather into hollows and clefts. Detailed work by Forster (1980, e.g. figs. 2.5 and 2.6) shows that dyke and cone-sheet intrusions overlapped and that there are several generations of dykes. The orientations of the dykes were measured by Speight et al. (1982, fig. 33.3), and by Forster (1980, fig. 2.1, summarised as (Figure 28). Many of the dykes on Rum appear to radiate from foci in Glen Harris (Figure 27) for the localities of plugs within the Northern Marginal Zone, (Figure 21) for the Southern Mountains Zone and (Figure 35) for detail within the Layered Suite. The numbers 1-13 refer to the localities where dyke orientations have been measured (see (Figure 28))." data-name="images/P936601.jpg">(Figure 26).

Forster (1980, chapter 3) used petrographical and chemical criteria to divide the Rum dykes and sheets into several groups. Members of the least evolved group of picrite and olivine dolerite intrusions intrude felsic and mafic members of the Rum Central Complex and form part of the radial swarm. The majority of the intrusions, which are of transitional basaltic compositions, cut the early felsic members of the Complex (Stage 1 of (Table 1)) but very few intrude the Layered Suite. These intrusions include many of the NW-trending dykes, the majority of the cone-sheets, and the thin conformable sheets in the Torridonian rocks. A distinctive group of dykes of fresh ophitic dolerite with sparse olivine microphenocrysts occurs mainly in north-west Rum and postdates the Canna Lava Formation. Several NNE- to NNW-trending feldspar-phyric dykes form another group which is largely restricted to the northern margin of the central complex where they cut the felsic rocks of Stage 1 but are invariably cut by the gabbros of the Layered Suite. A group of tholeiitic basaltic andesites includes the thick sills or conformable sheets that intrude Torridonian rocks east of Papadil and south-west of Kilmory, and thin dykes with ‘liquid–liquid' contacts with rhyodacite in Nameless Corrie. Several dykes also assigned to this group intrude peridotites of the Eastern Layered Intrusion. The rare evolved dykes of hawaiite, mugearite and benmoreite mostly postdate the Layered Suite (Stage 2, (Table 1)) but none has been found intruding the Canna Lava Formation.

Specific features of several distinctive dykes merit comment. Thick dykes (10–30 m) of coarse-grained dolerite or microgabbro intrude Torridonian rocks south of Allt nam Bà, Triassic sandstones on Monadh Dubh [NG 338 028], and granophyre east and south-west of Ard Nev. Some 300–500 m south of the ford at Allt nam Bà [NM 4091 9432], several of these dykes appear to terminate abruptly 200–250 m outside the Outer Main Ring Fault and reappear, with apparent sinistral off-sets, at about 350 m above OD on the south-eastern slopes of Beinn nan Stac. On the north-west spur of Am Màm [NM 3816 9893], a feldspar-phyric dyke is offset dextrally on small fractures where the Main Ring Fault system cuts the Am Màm Breccias. In the Central Intrusion about 1 km south-east of Harris Bay, several of the picritic dykes that intrude gabbro and peridotite are thermally altered; olivine crystals are full of small semi-opaque inclusions which give them a dull, clouded appearance, and pyroxene in the matrices is largely replaced by red-brown amphibole. A similar amphibole, generally accompanied by biotite, occurs in dolerite and basalt dykes and marginal gabbros on the eastern edge of the Eastern Layered Intrusion, where the marginal mafic rocks have been contaminated by the adjoining felsic country rocks. A variety of picritic dolerite crops out near An Dornabac and south-west of the southern end of the Long Loch. The rock (RMF 77/024) contains numerous skeletal crystals of dark, clouded olivine, 3–4 mm in length, in a fine-grained matrix of feldspar laths and granular augite. Although these dykes intrude mafic rocks, they appear to have been thermally altered. In upper Glen Harris, on the north side of Abhainn a' Ghlinne [NM 3748 9642], several thin (10–40 cm wide) basic dykes cut allivalitic gabbro. They are fine-grained basic rocks (SR 445) with scattered olivine microphenocrysts, their texture resembles that of the beerbachite inclusions in the Eastern Layered Intrusion (Chapter 7). In Nameless Corrie [NM 385 936], 'liquid–liquid' contacts occur between basaltic andesite dykes and rhyodacite.

Eigg

NW-trending basaltic dykes intrude the Jurassic sedimentary rocks and the Paleocene flows of the Eigg Lava Formation. They are most numerous in southern and central Eigg and around Laig Bay, decreasing in number towards the north end of the island. Basic dykes cut the quartz-porphyry intrusions at Eilean Thuilm [NM 482 913] but few intrude the Grulin Felsite and the Eocene Sgurr of Eigg Pitchstone Formation.

The dykes are generally less than 2 m in width. The thick dyke which Harker (1908) noted east of the Sgurr is an elongate dolerite plug. However, 5 m wide olivine dolerite dykes intrude lavas at the south end of Eilean Chathastail [NM 4844 8285] and near Kildonnan (Ridley, 1973, table 2, E 2, E 3). Numerous dykes crop out in the coastal sections and they are found inland when exposure is good, as in recent roadcuts in the plantations around Glac an Dorchadais [NM 476 862]. The dykes may be deeply eroded. Substantial trenches have been excavated by the sea along dykes on the foreshore at the Bay of Laig, near [NM 470 890] and these are accentuated by thin, rampart-like walls formed by indurated and veined calcareous sandstones of the Valtos Sandstone Formation (Frontispiece).

The majority of the dykes are aphyric or sparsely feldspar-phyric and olivine-phyric basalts. The olivine crystals are up to 4 mm in length and enclose deep brown chrome spinels. They resemble some of the less-evolved dykes on Rum but none is picritic. A feldspar-phyric dyke that intrudes a basalt flow immediately west of Laig Gorge (at [NM 4721 8746]) contains labradorite phenocrysts up to 1 cm across together with large olivine crystals that enclose chrome spinels. Aggregates of feldspar, clinopyroxene and olivine in this rock (EA 7613) resemble igneous cumulates. An irregular dyke of feldspar-phyric dolerite (SR 553) crops out on the shore at Uamh Fhraing (the Massacre Cave) [NM 475 834]. A central zone is crowded with tabular plagioclase crystals, commonly 5 x 1 cm and rarely up to 10 x 1.5 cm. The dyke strikes NNW and is exposed intermittently together with a thin aphyric basalt for 400 m to the NNW. A feldspar-phyric basalt dyke that intrudes the south-west side of the Gulainn na Sgurra plug [NM 4675 8465] is probably a continuation of this dyke. Also in southern Eigg, the Sgurr of Eigg Pitchstone is cut by a composite pitchstone-basalt dyke in the access gully 270 m west of the ancient fort on An Sgurr [NM 4608 8474]. The mafic rock (SR 625) is a fine-grained basalt with numerous olivine pseudomorphs. It is net-veined by a pale grey felsic rock that merges into the pitch-stone dyke. This is the only basic dyke found to intrude the Sgurr Pitchstone. An aphyric basalt, (EA 33), with vesicles filled by analcime and thompsonite, intrudes the Grulin microsyenite (= 'felsite') sheet.

Muck

A dense swarm of basaltic dykes that trends N320–340° intrudes the Jurassic sedimentary rocks and the Eigg Lava Formation (Plate 12). The dykes and subparallel faults impart a distinct NNW–SSE 'grain' to Muck. The dyke swarm was examined in detail by Harker (1908, pp.147–152, pl. 4). Over 130 dykes averaging 1.3 m in width were mapped on the south coast, which represents up to 8 per cent crustal dilation in the densest part of the swarm (Speight et al., 1982, fig. 33.3).

The Muck dykes commonly form distinct walls where they intrude the lavas in coastal sections, although the deep NNW-to NW-trending trenches on Eilean Aird nan Uan [NM 399 800] and the reefs east of Bàgh a'Ghallanaich [NM 408 806] were probably eroded along dykes. Wall-like dykes on the coast at Camas na Cairidh [NM 415 804] have thin, flow-banded tachylitic selvedges. The attitude of the flow-banding in the selvedges indicates that the magma flowed obliquely along the fissures (Harker, 1908, pp.155–161, figs. 42, 47). Individual dykes in this area and at Camas Mór [NM 407 792] have small sinistral offsets of <1–3 m (Harker, 1908, figs. 45, 47), while at Camas Mór a north-trending dyke on the foreshore at [NM 4065 7924] passes up into a sill and, after about 8 m of dextral offset, once more becomes a dyke, trending approximately north-west.

Harker considered that the dykes on the south side of Beinn Airein and in the cliffs of southern Muck showed a tendency to die out upwards against resistant layers of fragmental rocks or massive parts of the lava succession. During the resurvey it was not possible to decide whether this was a real effect or due to the generally poorer outcrop at higher levels and inland.

The dykes range from deep brown, tachylitic glass (SR 606) through basalts, commonly with olivine microphenocrysts, to coarse-grained dolerite and olivine-dolerite which is commonly ophitic. Olivine generally occurs in the matrices and as phenocrysts. The dykes are usually fresh, although partial serpentinisation of olivine is common. Smectite, zeolites and calcite fill vesicles. A few dykes depart from the types mentioned above. A multiple dyke at Rubh' Leam na Laraich [NM 3922 7970] strikes NNW–SSE for over 300 m and consists (from west to east) of: a 1.3 m basalt, a 3 m coarse-grained olivine-dolerite (SR 591), and a 2 m feldspar-phyric olivine-dolerite (SR 592). The 3 m-wide dyke is crowded with olivine as single crystals (to 4 mm diameter) and in granular aggregates, brown augite (c.2–3 mm diameter), short plagioclase laths with pronounced normal zoning, abundant small (c.2 mm diameter) euhedral opaque oxide crystals, and a matrix of zeolites and chlorite crystals. The 2 m-wide dyke is similar but also contains zoned plagioclase phenocrysts. The dykes may be separated by lava screens or adjoin one another. A 4 m-wide dyke about 50 m east of Achad na Creige [NM 399 800] changes from olivine-dolerite to feldspar-phyric olivine-dolerite and back to olivine dolerite as it is traced for 500 m SSE from the coast. Some 300 m SW of Gallanach [NM 4072 8012], an extremely decomposed, coarse-grained feldspar-phyric dolerite dyke contains plagioclase phenocrysts up to 5 cm in length. It is similar to the dyke at Uamh Fhraing on Eigg. The thick gabbroic dyke at Camas Mór is described below.

Camas Mór gabbro dyke, Muck

A thick dyke of coarse-grained olivine-dolerite and gabbro forms the rocky coastline for 1400 m SSE between the east end of Camas Mór [NM 411 792] and Bogha na Fionn-aird [NM 4173 7803]. The dyke, which is up to 100 m in width, intrudes lavas of the Eigg Lava Formation for most of its length. It thins towards the NNW at the east end of Camas Mór, where it intrudes rocks of the Kilmaluag Formation and the Camas Mór Breccia (Chapter 8). Farther north, an en echelon continuation of the gabbro dyke forms Druim na Fhardairc [NM 409 796] and Cnoc na Curran [NM 4075 7990]. It is also probable that a segment underlies Gallanach since the low hill [NM 4072 8007] immediately south of the farmhouse is made of tough, thermally metamorphosed basalt (SR 580).

The dyke is chilled to dolerite where it intrudes the lavas, but the interior is coarse-grained dolerite and gabbro with scattered, irregular pegmatitic areas. South of Sloe na Dubhaich [NM 4155 7845] there is a 20 m-wide zone of small-scale layering close to the eastern contact and similar layering occurs north of the inlet. The layering consists of mafic schlieren and bands, 5–8 cm thick, alternating with feldspathic bands 1–2 cm thick. The boundaries are not sharp and the bands merge over about a centimetre. The banding, which dips at 20–30° E, is developed over a vertical distance of about 3 m. The banded structures are somewhat similar to those described from the Tertiary sills of Trotternish, North Skye (Gibson and Jones, 1991), although no assymetrical, graded banding is present. It is similar, at least superficially, to the matrix banding in some peridotite plugs and tongues on Rum.

The gabbro contains olivine; laths of normally zoned plagioclase, ophitic augite, and minor apatite and late zeolites (SR578). In the feldspathic bands (SR 598), augite is markedly ophitic and olivine also tends to be moulded on euhedral plagioclase.The coarse pegmatitic gabbro (SR 584) contains abundant augite intergrown with clear plagioclase and coarse, skeletal opaque oxides; minor amounts of olivine are present, and analcime and zeolites fill irregular areas.

At Camas Mór, Tilley (1947) identified a broad exogenous contact zone, in which the impure Jurassic limestones are altered to sanidinite-facies assemblages, a thin zone (c.2.5–8 cm wide) where skarn minerals are present and an endogenous zone in which the gabbro has been much modified by reaction to form pyroxene-rich, critically undersaturated assemblages. This is one of several classic localities in the British Tertiary Volcanic Province where complex, high-grade thermal metamorphic mineral assemblages formed when calcareous sedimentary rocks were intruded by gabbro and dolerite (e.g. Scawt Hill, Co. Antrim; Tilley and Harwood, 1931). The Camas Mór dyke intrudes limestone with both dolomitic and silty, aluminous components and the mineral assemblages are therefore more varied than, for example, those at Scawt Hill. The principal minerals in the carbonate-bearing contact zone are calcite, grossularite, wollastonite, monticellite, gehlenite, spurrite, merwinite, larnite, rankinite, cuspidine, tilleyite, periclase, brucite, spinel and perovskite. Minerals in the skarn zone consist mainly of brown titanaugite and wollastonite, with minor amounts of soda-sanidine, analcime and plagioclase. The gabbro at the margin of the dyke is usually olivine-free, rich in clinopyroxene with thin green (Na--rich?) rims, and has minor plagioclase and accessory wollastonite. The marginal pyroxene-rich facies contains streaks, veins and lenses of very modified gabbro which contain abundant wollastonite, plagioclase, melilite, nepheline, analcime and minor soda-sanidine. The segregations are coarse-grained, both nepheline and titanaugite are up to 3 mm. Rare segregadons occur which are rich in soda-sanidine with andesine cores, greenish (Na-rich?) clinopyroxene, fayalitic olivine, magnetite and apatite (Tilley, 1947). Examples of modified gabbro and altered limestone are shown in (Plate 13a) and (Plate 13b) respectively.

In addition to the calcareous hornfelses described by Tilley, other altered rocks were noted during the resurvey. Hornfelsed, feldspar-phyric basaltic hawaiite at the northern contact of the gabbro at [NM 4112 7914] ((SR 587); see also (Figure 41), locality C) is a dark matt-grey rock. It contains plagioclase phenocrysts crowded with minute, semi-opaque inclusions in a matrix that has been recrystallised to a felted mass of plagioclase laths poikilitically enclosed by large irregular opaque oxide, clinopyroxene and olivine crystals. A short distance north of the altered lava, carbonate-free sandstone in the Camas AI& Breccia (Figure 41), locality B) is altered to a tough, dark grey, fine-grained hornfels (SR 601). It consists of relict quartz grains in a matrix of cordierite, opaque oxides, needles of sillimanite or mullite, and clear, anhedral quartz.

Canna and Sanday

Only 11 dykes have been recorded intruding the Palaeocene Canna Lava Formation, which clearly postdates the regional dyke swarm which is well represented on the north-west coast or Rum. The dykes trend between north-west and north-east and they range from fine-grained olivine-basalt to coarse, ophitic olivine-dolerite. Two feldspar-phyric dykes on Sanday, including a 4 m-wide dyke (SR 548) that crosses Sanday from west Suileabhaig [NG 2775 0410] to Buaile na h-Uaimh [NG 2780 0468], contain labradorite microphenocrysts 1–3 mm in length. Some 30 m east of Geodha na Nighinn Duibhe [NG 2460 0547], a 9 m-wide dyke of coarse-grained olivine-dolerite intrudes along a small fault that trends N345o; and 300 m to the west, a 4.5 m dolerite cuts lavas on the shoreface. An olivine-dolerite dyke (SR 547) trending at about N020° crops out on the shore platform [NG 2590 0507] at Losaid an t-Sagairt and in the cliffs 250 m to the NNE.

Compositions of the dykes

Analyses of the Eigg dykes are given by Ridley (1973) and Allwright (1980; see also Appendix 5g), together with limited microprobe analyses of selected minerals. The basalts are typically of olivine-hypersthene-normative transitional type. None is quartz-normative and very few are nepheline-normative. No dyke is more fractionated than hawaiite. Unlike the lavas on Eigg, few of the dykes have rock/chondrite normalised multi-element profiles (Thompson, 1982; (Figure 29)) that suggest appreciable contamination by granulite-facies Archaean crustal rocks or partial melts from that source. It therefore seems likely that the lavas and the majority of the dykes represent different magmatic events.

With few exceptions, the dykes on Muck are transitional olivine-basalts with either normative hypersthene or minor nepheline (Appendix 5g; Ridley, 1973, table 4; Allwright, 1980, appendix 6.2.1). Of over 60 analysed samples, only 5 are quartz-normative (e.g. Ridley, 1973, table 4, M17). Amongst the suite, Allwright recognised a small number of basaltic hawaiites and hawaiites, two mugearites and a few dykes belonging to the Preshal Mór and Fairy Bridge basalt types of Skye (Allwright, 1980, appendices A.6.5 and 6.6; cf. Thompson et al., 1972 and Mattey et al., 1977). In contrast to the analysed dykes from Eigg, the Muck dyke multi-element profiles indicate that both uncontaminated and crustally contaminated dykes are present (Figure 29).

Compared with Eigg and Muck, the Rum dykes cover a much greater compositional range from evolved pitch-stones (see below) to picritic types. McClurg (1982) described a dyke of aphyric picritic with MgO = 13.47 wt% and another with a calculated MgO of 20.5 wt% for its quenched groundmass; both dykes intrude the Rum Layered Suite and are proof that high-MgO liquids were present during the formation of the Layered Suite (see (Table 9)). By far the most common dykes are basalts which include tholeiitic, transitional and mildly alkaline varieties (Forster, 1980). Multi-element profiles for a limited number of these dykes show that both uncontaminated and crustally contaminated varieties occur (Figure 29).

On Canna, the fine-grained olivine-basalt dyke (SR 247) that forms Garbh Asgarnish [NG 2810 0582] is the only dyke from Canna for which there is a chemical analysis. It is a transitional olivine-basalt (Allwright, 1980, appendix A6.3)

Fissure breccias

More than 20 linear zones of bleaching, intense thermal metamorphism and brecciation affect the Torridonian and Triassic rocks in northern Rum (Figure 27) for the localities of plugs within the Northern Marginal Zone, (Figure 21) for the Southern Mountains Zone and (Figure 35) for detail within the Layered Suite. The numbers 1-13 refer to the localities where dyke orientations have been measured (see (Figure 28))." data-name="images/P936601.jpg">(Figure 26). The distinctive breccias are common on Mullach Mór, between [NG 370 018] and [NG 396 010], and on Monadh Dubh [NG 340 030]. One occurs on the Dibidil path south of Kinloch at [NM 4037 9885]. The brown-coloured sandstones are bleached pale grey in zones up to 2 m in width and several hundred metres in length. Where brecciation occurs, close-packed pieces of subangular to rounded, bleached sandstone up to 20 cm across are common, as in the fissure breccias that extend from south to north-west of Loch Lewis [NG 3765 0190]. The breccias vary in strike from about N320° in the west to N020° toward the eastern end of the Mullach Mór ridge. They focus on the Rum Central Complex.

The linear zones of alteration vary from simple bleaching of the country rocks to close-packed, thermally metamorphosed sandstone fragments cut by thin basaltic intrusions, as found in a low cliff in the east bank of the Kilmory River [NG 3636 0187]. Near Mullach Mór summit, irregular tachylitic veins with minute pseudomorphs after olivine cut breccia and Harker (1908) noted breccia intruded by variolitic basalt west of Loch an Tairbh [NG 379 019]. The alteration of the sedimentary rocks is similar to that found near mafic intrusions; rounded quartz grains and aggregates of quartz crystals are surrounded by turbid brown intergrowths of fine-grained quartz and alkali feldspar; slender quartz paramorphs after tridymite cut the intergrowths; and individual quartz grains are commonly rimmed by overgrowths of needle-like quartz aggregates. Chlorite pseudomorphs after (?)orthopyroxene occur, and epidote is present. The sedimentary rocks have evidently undergone high-temperature thermal metamorphism.

The fissure breccias were mapped and described in detail by Harker (1908, pp.61–67, figs. 18, 19) who suggested that 'they may be regarded as incipient or aborted channels for the fissure-eruptions of basaltic lava'. The brecciation was attributed to crushing, and the high-temperature alteration to 'some kind of solfataric agency, operating along vertical bands of rock disintegrated by crushing'. The close association of breccias, thermal alteration and, less commonly, thin, basaltic intrusions, together with their roughly radial arrangement about the central complex, prompt the suggestion that they are connected with the radial dyke swarm and might overlie dykes. However, no substantial dyke, or other minor intrusion, has been found in close association with a fissure breccia.

Minor acid intrusions

On Rum, a north-trending, 1.5 m-wide dyke of fresh black pitchstone intrudes the Western Granite in the cliff 800 m north-west of Wreck Bay, at [NM 3041 9876]. Two pitchstone dykes, 2 and 4 m wide, intruding sandstone in southern Rum, are probably part of a single NNW- to NW-trending intrusion. One is in a gully on the east side of Rubha nam Meirleach [NM 3681 9118] and the other is 500 m to the NNW, on the north side of a prominent cliff (SR 533, [NM 3668 9167]).

A small microgranite sheet cuts Torridonian arkose and a basaltic cone-sheet on the south side of Kinloch Glen (at [NM 3783 9955]) and a 30 cm dyke of microgranite intrudes porphyritic rhyolite and associated tuffs in Coire Dubh at [NM 3920 9785]. The origin of the micro-granites is uncertain. They do not resemble the Western Granite and it is possible that they are rheomorphic acid intrusions originating from intrusion breccias bordering underlying mafic rocks (cf. ((Figure 32)b.

On Eigg, several acid dykes and plugs intrude the Palaeocene lavas and Mesozoic sedimentary rocks. At Sròn Laimhrige [NM 4745 8775], a north–south elongated plug of quartz-porphyry extends south for about 400 m from a prominent north-facing bluff. It intrudes beds of the Kilmaluag Formation and the lowest flows of the Eigg Lava Formation. Some 150 m to the west, a separate intrusion of similar rock forms the steep northern face of Laig Gorge [NM 4726 8754]. An intrusive contact with beds of the Duntulm Formation and the Kilmaluag Formation is exposed in the stream section where the quartz-porphyry body appears to have wedged the sedimentary rocks apart. This intrusion is roofed by the Cretaceous Laig Gorge Sandstone Member and is cut off to the east by the Laig Gorge Fault (Chapter 10). At the northern end of Eigg, on Eilean Thuilm [NM 482 914] and the adjoining coast at Sgorr Sgaileach [NM 485 913], columnar jointed quartz-porphyry forms cliffs up to 40 m in height. The intrusion is an irregular sill which cuts thin shales and limestones of the Lealt Shale Formation. The sill is intruded by a dolerite dyke, and by thin basalt sheets in the north-facing cliffs WNW of the caves at [NM 4866 9114].

At Rubh' an Tangaird [NM 4779 8318] two irregular NNW-trending pitchstone dykes intrude basalt lavas. The western dyke is up to 60 cm in width and the eastern one tapers from about 90 cm to less than 30 cm towards the sea. In each dyke, the central part is dark olive-brown in colour with a resinous lustre. This grades through an intermediate zone with a beautiful bluish tint into black glassy rock at the margins (Harker, 1908, p.177). A dark, resinous porphyritic pitchstone dyke up to 4.5 m in width intrudes lavas and a basic Kildonnan-type sheet at the North Pier (Clanranald Pier) [NM 4832 8416] from where it can be traced northwards for about 120 m. A pale, cream-coloured quartz-porphyry crops out a further 100 m to the north, at Ruigh na Tràghad [NM 4834 8438]. Although directly in line with the dyke at the pier, the rock of this intrusion differs petrographically and the two bodies are probably unconnected. On the west coast, about 150 m south of Rubha na Fhasaidh [NM 4390 8724], a NNW-trending pitchstone dyke cuts basalt lavas.

South of Laig Gorge, there is a small outcrop of dark, resinous pitchstone at the foot of a waterfall [NM 4743 8742] in the Abhainn a' Chaim Loin. It is not clear if the pitch-stone is a dyke or a sheet. It is a distinctly different rock from that which forms the adjoining Sròn Laimhrige intrusion and is not regarded as part of that body. The pitchstone protrusion of the Sròn Laimhrige intrusion recorded crossing the Cleadale road (Harker, 1908, p.141) was not found during the resurvey.

A dark felsic sheet intrudes the lava flows on the south side of the Sgurr of Eigg, extending from north of the ruined houses at Grulin lochdracht [NM 447 851], southeast and eastwards to a point about 150 m south of Collie's Cleft [NM 4622 8465]. This sheet, the Grulin Felsite, has been suggested as a dyke-like feeder for the Sgurr of Eigg Pitchstone (Harker, 1908, p.175, figs. 48 and 49).

Harker noted that the felsite cuts at least one basalt dyke and he did not find any dykes that intruded the sheet. However, Allwright (1980) found one later basalt dyke. Since the sheet occurs in the denser part of the regional dyke swarm on Eigg, the rarity of later dykes confirms Harker's view that the sheet is a late intrusion and it may be of similar age to the Sgurr Pitchstone. It is also possible that other acid minor intrusions on Eigg are of Eocene age for, with the exception of the intrusion at Eilean Thuilm, none is cut by basic dykes.

Petrography and chemical composition

The quartz-porphyry intrusions at Eilean Thuilm, Sròn Laimrhige and north of Laig Gorge are petrographically similar. They contain scattered phenocrysts of rounded, embayed quartz and rare sanidine phenocrysts that are largely replaced by calcite. The matrix is an aggregate of turbid, brown alkali feldspar and clear quartz; spherulitic structure is developed rarely, as is a weak granophyric texture. Mafic minerals are conspicuously absent and none are recognised as phenocrysts, but irregular areas of opaque grains in the matrices may be after original mafic minerals. The quartz-porphyry dyke, about 150 m north of the North Pier, contains sparse rounded quartz crystals and relicts of possible feldspar phenocrysts in a carbonated, felsitic matrix. The Rum micro-granite sheets have equigranular textures with a tendency for the turbid alkali feldspar to be moulded on equigranular quartz. They contain opaque oxides and minute amounts of biotite.

The Rubh' an Tangaird pitchstone dykes have a brown glassy matrix and scattered euhedral sanidine phenocrysts; the matrix is devitrified and felsitic in the central parts of the dykes. The North Pier dyke (SR 505) contains complexly zoned plagioclase, clinopyroxene, orthopyroxene, opaque oxides and apatite as phenocrysts up to 3 mm long and as crystal aggregates. The phenocrysts and crystals in the glomeroporphyritic aggregates have euhedral outlines but contain rounded and subhedral areas of matrix, and have marginal embayments. The matrix is dark, flow banded and partially devitrified to aggregates of minute feldspar, pyroxene and opaque oxide microlites which resemble 'quench' texture. The Abhainn a' Chaim Loin pitch-stone (SR 492) is sparsely porphyritic. All the phenocrysts are rather rounded and frequently embayed to the point of being skeletal; they include zoned plagioclase, anorthoclase with clear grid twinning, Ca-rich clinopyroxene and pigeonite. The matrix is a pale brown glass with scattered areas containing abundant microlites.

The pitchstones in southern Rum contain small quartz and alkali feldspar phenocrysts in a flow-banded matrix in which layers of dark, semivitreous rock alternate with white, felsic layers. The felsic parts are crowded with small (c.0.5–1 mm diameter) spherulitic growths that are commonly centred on the phenocrysts. The dyke near Wreck Bay contains sparse phenocrysts of quartz, plagioclase and sanidine in a brown glassy matrix crowded with stellate crystallites ((SR 286), (S 3294); Harker, 1908, p.179, pl. IX, 1, B).

The Grulin Felsite contains small corroded crystals (1 mm and less in length) of alkali feldspar with turbid rims, fresh augite, minute euhedral opaque oxide grains and rare crystals of slightly purple-coloured apatite. The iron-stained, fine-grained matrix is rich in small opaque grains, turbid alkali feldspar, and clinopyroxene. The petrography and chemical composition (Allwright, 1980, table A7.1) show that this is a microsyenite.

A selection of chemical analyses is given in Appendix 5d (based on Ridley, 1973, table 6). The quartz-porphyry intrusions of Sròn Laimhrige and North Eigg are typical of many granitic intrusions in the Hebridean Province, with high silica (>70%), high total alkalis (>7%), K2O>Na2O, and low MgO, CaO. The Sgorr Sgaileach intrusion (EA 55) is one of the most evolved rocks in the Small Isles, with particularly high light rare-earth elements and strongly depleted Sr, P and Ti (Figure 52). The pitchstone and felsite minor intrusions of Eigg have, with one exception, compositions that differ appreciably from the pitchstones and felsitic sheets of the Sgurr of Eigg; the compositions of the Sgorr Laimhrige and Sròn Sgaileach intrusions are close to the granite minumum for low pressures but differ from the Rubh' an Tancaird pitchstone dykes (Figure 51).

Palaeogene 3: Rum Central Complex Stage 2 Layered Suite

Introduction

The layered peridotitic and gabbroic rocks are the most distinctive igneous rocks on Rum. They form the higher peaks, including Askival and Hallival where their contrasted weathering gives rise to strongly terraced topography (see (Cover) and (Frontispiece)). These mafic rocks crop out over about 30 km2 and are the largest area of unserpentinised ultrabasic rocks in the British Isles. They are the exposed top of a substantial body of dense mafic rock underlying the Rum Central Complex (Chapter 11; McQuillin and Tuson, 1963). The Layered Suite is comprised of the Eastern Layered Intrusion, the Western Layered Intrusion and the Central Intrusion (Figure 30).

The classic investigation by Harker (1908) established Rum as an important area for the study of ultrabasic rocks. He introduced the rock names allivalite and harrisite, after localities on Rum, and discussed the origin of the layering in the ultrabasic rocks. He suggested that the alternating layers formed when magma which had differentiated into compositionally contrasted fractions at depth was intruded in successive olivine-rich and feldspar-rich pulses at progressively lower levels; the layered sequence was built from the top downwards.

No further major investigation of the ultrabasic and gabbroic rocks was undertaken for nearly 50 years, although Phillips (1938) made a petrofabric examination of olivine orientation, and Bailey (1945) and Tomkeieff (1945) reconsidered certain aspects of the petrology. By this time Wager and Deer (1939) had published their classic account of the Skaergaard Intrusion in east Greenland, in which they emphasised the importance of crystal settling in the formation of layered igneous rocks. This concept was first applied to Rum by Wager and Brown (1951) and later developed and refined by Brown (1956), Wadsworth (1961) and Wager and Brown (1968). Subsequently, the paramount role of crystal settling in igneous layering was questioned by McBirney and Noyes (1979), who advocated a mechanism involving in situ crystallisation and double-diffusive convection to explain layering in the Skaergaard Intrusion. Their work, together with other studies which suggested picritic rather than basaltic parental magmas for the ultrabasic rocks (e.g. Gibb, 1976), generated renewed interest in mafic layered rocks, which in turn led to numerous studies on the well-exposed layered rocks of Rum (Dunham and Wadsworth, 1978; McClurg, 1982; Volker, 1983; Butcher, 1984; 1985; Palacz, 1984; 1985; Tait, 1985; Young, 1984; Faithfull, 1985; 1986; Palacz and Tait, 1985; Butcher et al., 1985; Young and Donaldson, 1985; Kitchen, 1985; Dunham and Wilkinson, 1985; Huppert and Sparks, 1980; Sparks and Huppert, 1984; Young et al., 1987; Emeleus, 1987; Renner and Palacz, 1987; Volker and Upton, 1990; Elias, 1989; 1991; Wadsworth, 1992; Emeleus et al., 1996b).

The possibility that layered igneous rocks could undergo late-stage, or 'diagenetic', changes was recognised by Sparks et al. (1985) following on from work by Irvine (1980). These ideas were further developed on Rum by Bedard et al. (1988), and the significance of the development of textural equilibrium during crystallisation was investigated by Hunter (1987). Subsequently, Sparks et al. (1993) combined laboratory experiments and mathematical modelling to demonstrate that much of the centimetre- to metre-scale layering in the ultra-basic rocks of Rum could have formed by sedimentation in a convecting magma chamber, marking a return to mechanisms advocated by Wager and Brown (1968) and earlier workers.

The nomenclature of the plutonic ultrabasic and basic rocks presents problems, especially when they exhibit small-scale layering and when all gradations exist between the principal rock types. In this account an attempt is made to formalise the terms so that they more nearly agree with internationally accepted definitions (LeMaitre, 1989) and also with current stratigraphical nomenclature (Whittaker et al., 1991).

The mafic rocks of central Rum have been generally referred to as ultrabasic, despite the fact that the more feldspathic examples do not fall within the chemical definition of the term. The associated gabbroic intrusions (mostly unlayered) have been classified as basic. As there is considerable mineralogical and chemical overlap between all the mafic rocks, a general name is clearly desirable and the term 'Layered Suite' is proposed. Within the suite, the term 'series' has been applied to each of the three major divisions (see above), as well as to some minor ones where 'group' has also been used. These terms now have strict stratigraphical connotations (Whittaker et al., 1991) and are to be avoided in describing and classifying rocks such as those of the Layered Suite of Rum. The terms now proposed to replace the original names for major and minor divisions of the suite are listed in (Table 4).

The more detailed terminology of the layered ultra-basic and gabbroic rocks of Rum may be treated in various ways. In the field it is often convenient to describe rocks rich in magnesian olivine as peridotites or as feldspathic peridotites where calcic plagioclase is visible. However, it should be emphasised that as used on Rum, peridotite and feldspathic peridotite do not always conform to modern definitions; the rocks are commonly mela-troctolites or mela-olivine-gabbros (LeMaitre, 1989, figs. B6 and B7b). The modern terminology is therefore included in (Table 5) which is based on a modal classification proposed by McClurg (1982, p.20), in which a variety of rock types are distinguished between the extremes of dunite and anorthosite. Her divisions amplify those sug gested by Brown (1956). The Rum rocks are type examples of igneous cumulates and one of the first detailed applications of the terminology to layered rocks was in south-west Rum (Wadsworth (1961; see also Irvine, 1982; Wadsworth, 1985). Where precise designations are required, the terminology for cumulate rocks of Wager, Brown and Wadsworth (1960) is employed.

If it were not for the calcic compositions of the plagioclase and the high-Mg nature of the mafic minerals, the majority of the feldspar-rich rocks termed allivalite or eucrite would be called troctolite or olivine-gabbro respectively (Bedard et al., 1988; Renner and Palacz, 1987; cf. LeMaitre, 1989, fig. B.6). They are more strictly defined as bytownite-troctolite and bytownite-gabbro (Le Maitre, 1989) and these terms will be used in this memoir. Allivalite does continue in use as a recognised local term (LeMaitre, 1989; Claydon and Bell, 1992; Wadsworth, 1992) but it has long been recommended that eucrite be abandoned as a rock name (LeMaitre, 1989).

Harrisite (Harker, 1908) is the term applied to unusual-textured, coarse-grained basic and ultrabasic rocks widely developed at Harris in south-west Rum [NM 337 957]. It is defined as 'a variety of troctolite in which the large black lustrous olivines have a branching habit of growth and are orientated perpendicular to the layering' (LeMaitre, 1989, p.72). Since this distinctive texture (Plate 15) is found in a variety of rocks including feldspathic peridotite, bytownite-troctolite, bytownite-gabbro and troctolite, it will be used in the adjectival sense rather than as a rock name. The rocks with harrisitic textures are crescumulates (Wager and Brown, 1968) and their textures and structures have been compared with the spinifex texture of Archaean ultramafic komatiites (Donaldson, 1974).

The three major divisions of the Layered Suite (Figure 30) each have their own distinctive features. The Eastern Layered Intrusion is characterised by gently dipping, large-scale (>10 m) layers rich in olivine or plagioclase (see Cover); the rocks of the Western Intrusion are generally olivine rich and their layering is manifest in subtler textural differences; layering similar to that found in the Eastern Layered Intrusion occurs in the Central Intrusion but the distinctive feature of this intrusion is the abundant development of linear zones of ultrabasic breccia, which include blocks derived from the Eastern and Western layered intrusions as well as members of the Central Intrusion. The Central Intrusion cuts and separates the Eastern and Western layered intrusions (Figure 30); see also Chapter 9, pp.115–116).

Eastern Layered Intrusion

The Eastern Layered Intrusion is a classic area of igneous layering (Harker, 1908; Brown, 1956; Wager and Brown, 1968; Emeleus and Gyopari, 1992). It consists of a succession of thick layers of bytownite-troctolite and feldspathic peridotite (see Cover and (Figure 31) which forms the major peaks of the Rum Cuillin ridge, with the exception of Ainshval and Sgurr nan Gillean. A zone of marginal gabbros, of variable width, borders parts of the layered rocks. Where the gabbros and ultrabasic rocks are in contact with country rocks, heat from the mafic bodies caused high-grade thermal metamorphism and partial melting of the earlier felsic rocks, with the formation of intrusion breccias (Plate 20). Broadly conformable sheets of gabbro, bytownite-gabbro and bytownite-troctolite intrude the layered rocks, along with gabbroic plugs and veins (Figure 32) and (Figure 35).

The layered rocks are divided into 16 major units (Brown, 1956, fig. 2; Volker and Upton, 1990, fig. 3), which are now considered to have built up successively from lower to higher layers. Each major layer consists predominantly of feldspathic peridotite and peridotite at the lower levels, whereas bytownite-troctolite and bytownite-gabbro form the higher parts. Viewed from a distance, the layering has an illusory simplicity but this is quickly dispelled on climbing any hill in the Eastern Layered Intrusion. Within each of the major layered units the passage upwards from olivine-rich to feldspar-rich rock is generally not a simple gradation over the thickness of a unit; the change can be quite abrupt at one locality but more gradational several hundred metres along strike, as in Unit 10 on the north-east side of Hallival ((Figure 33)a), ((Figure 33)b). Small- and large-scale reversals in modal proportions also occur, as found in Units 2 and 3 near Allt nam Bà ((Figure 33)c); Faithfull, 1985, fig. 5) and each unit contains abundant small-scale layering, usually reflecting variation in the modal mineralogy (Plate 14a), each lithology weathering in a characteristic and distinctive manner (Brown, 1956, figs. 14, 22 and 33).

The units on Rum conform with the term 'cyclic units' (Jackson, 1970; Irvine, 1987). Originally, they were described as 'rhythmic units' (Brown, 1956), but this term is now used to describe smaller-scale, repeated layering (Irvine, 1987). The layered units generally dip gently (10–20°) towards Glen Harris (Figure 30). However, layering in the units on the Trallval summit ridge steepens westward from about 12° NNW to 60–80° WNW. The deformation of these layers, adjoining the later Central Intrusion, resulted from a combination of fault drag with slumping of the poorly consolidated cumulates into a graben to the west. Layering in the lowest units dips steeply (c.50° W) into the intrusion north of Allt nam Bà. The total thickness of the layered succession is about 750 m. Individual units are on average about 50 m thick but they range from 15 to 150 m. However, the thickness of units and the proportions of peridotite and bytownite-troctolite within a unit vary along strike, as for example on the northern slopes of Barkeval [NM 375 972] and east of the Hallival–Askiva1 ridge. Furthermore, there is a systematic increase in the proportion of olivine-rich cumulates to the west (Figure 31).

The steep outer margin of the Eastern Layered Intrusion is approximately coincident with the Main Ring Fault between Allt Mór na h-Uamha [NM 406 972] and Allt nam Bà [NM 407 944] although it extends a short distance beyond the projected continuation of the fault system (Figure 30). The mafic rocks are not faulted but initially they may have followed the Main Ring Fault, thermally eroding the low melting-point Torridonian rocks outside the fault zone. The contact is marked by intrusion breccia where the sedimentary rocks have been partly melted and mobilised (see below), and there is chemical and isotopic evidence that the marginal gabbros hereabouts crystallised from picritic basalt magma contaminated by Torridonian rocks (Greenwood, 1987). Outward-dipping roof contacts are present on Beinn nan Stac [NM 396 940] and from Cnapan Breaca [NM 395 975] westwards to the south end of Meall Breac [NM 385 980]; these hills are underlain by mafic layered rocks ((Figure 32)a), ((Figure 32)b). The Eastern Layered Intrusion also underlies much of the Southern Mountains Zone and probably also extends beneath the Northen Marginal Zone west of Meall Breac. When the inclined contact at Beinn nan Stac is projected in a NNW direction it is found that Askival summit may have been only 200 m below the roof of the intrusion ((Figure 32)a).

Western Layered Intrusion

The Western Layered Intrusion, as redefined by Volker (1983), includes the Harris Bay, Transitional and Ard Mheall members of Wadsworth (1961, fig. 2). The layered succession is about 500 m thick and the layering dips at low angles (10–20°) towards Glen Harris (Figure 30), mirroring the structure (but not the lithological succession) in the Eastern Layered Intrusion. The members are not as clear-cut as the units in the Eastern Intrusion since there is an absence of plagioclase-rich cumulates. The layering tends to reflect textural rather than modal mineral variation although the latter is common on a small scale (e.g. Wadsworth, 1961, pl. 3, figs. 6, 7). The principal divisions are based on compositional change from the bytownite-gabbro Harris Bay Member upwards into the feldspathic peridotites and peridotites of the Transitional and Ard Mheall members. Layered structures are common, they are defined by coarse-grained harrisitic layers that alternate with equigranular peridotites and gabbroic rocks, and by more subtle textural variation in the finer-grained rocks (Plate 15)

The contact with the Western Granite is observed to be vertical to steeply outward-dipping in the cove 100 m south-west of the Harris Mausoleum [NM 3362 9565] and is probably eqully steep for about 1.7 km to the north in the rather ill-exposed lower part of Glen Duian. The course of the contact is largely obscured by drift on the north-western slopes of Ard Mheall. Between Ard Mheall and the south and south-eastern slopes of Ard Nev, the peridotites pass beneath a roof formed by the Western Granite and a wedge of gneiss ((Figure 32)c). A zone of complex intrusion breccia generally marks the line of the contact. Two of the clearest examples of this phenomenon in the Hebridean Province are at Harris Bay, where bytownite-gabbro has intruded the Western Granite (see below).

Central Intrusion

The Central Intrusion was initially recognised by McClurg (1982) and its limits defined by Volker (1983). It occupies a swathe of ground which extends south from Minishal to the coast between Papadil and the east side of Harris Bay (Figure 30), and it cuts the Western and Eastern layered intrusions and the Main Ring Fault. It consists of thick layered units of peridotite and bytownite-troctolite, each of which contains small-scale layering, and extensive linear zones of ultrabasic breccia. Intrusion breccia occurs at several localities on the contact with the Torridonian rocks, and with the Western Granite at Monadh Mhiltich [NM 354 995] and Minishal.

The Central Intrusion is structurally complex. Small-scale layering in bytownite-troctolites resembles layering in the Eastern Layered Intrusion as, for example, in outcrops on the ridge about 150 m north-west of the north end of the Long Loch (Plate 14b). A layered succession is present (Table 6) but it is interrupted by numerous linear zones of ultrabasic breccia (Wadsworth, 1961; 1992; Volker and Upton, 1990, fig. 24). Near the Long Loch the breccia zones strike at about N210°; to the south of Loch Fiachanis [NM 357 947] they are more numerous and fan out towards the south (Volker and Upton, 1990, fig. 24). Breccia also occurs along the contact with the Western Layered Intrusion and to a lesser extent against the Eastern Layered Intrusion. The layering in members of the Central Intrusion dips at 20–60° and strikes approximately north–south to define a shallow syncline that extends from west of the Long Loch to Ruinsival (Figure 30); Wadsworth, 1992). The steeper dips are probably of tectonic origin since bytownite-troctolite and bytownite-gabbro generally develop 'soft-sediment' deformational structures on slopes greater than 20°. Volker and Upton (1990) consider that there was a complex interplay between bytownite-troctolite accumulation and breccia formation: during periods of tectonic stability evolved intercumulus liquids,- expelled from large bodies of layered, unconsolidated olivine cumulates, crystallised to form bytownite-troctolite; but when extension was occuring, major fractures in the incompletely consolidated olivine-rich cumulates were filled with the evolved liquids and fragments from consolidated parts of the cumulate pile, giving rise to the linear breccias.

The overall intrusive behaviour of the Central Intrusion (Figure 30) is also seen at outcrop level. Central Intrusion breccias cut well-layered peridotites of the Ard Mheall Member (Western Layered Intrusion) in crags (at [NM 3582 9800] ) 1 km NNE of An Dornabac. From 250 m NNE to 400 m SE of the southern tip of the Long Loch [NM 363 981], the breccias intrude bytownite-troctolite and peridotite in the Eastern Layered Intrusion, as well as an earlier peridotite tongue. On the north-western slopes of Trallval, 300 m west to north-west of Triangular Buttress [NM 3737 9538], a thin marginal olivine-gabbro and breccias of the Central Intrusion intrude the highest units in the Eastern Layered Intrusion. The external contacts of the Central Intrusion are generally vertical or steeply inward-dipping, as found on Trallval and also on the western side of An Dornabac [NM 351 971], where olivine-gabbro of the Central Intrusion overlies peridotites of the Ard Mheall Member of the Western Layered Intrusion.

Minor structures in the layered rocks

Small-scale layered structures characterise the peridotites and especially the bytownite-troctolites. These include a wide range of structures analogous to those present in subaqueous and subaerial sedimentary deposits (Wadsworth, 1973), certain structures similar to those arising from 'soft sediment' deformation (Volker, 1983; Volker and Upton, 1990) and structures resulting from replacement processes operating within the layered successions. The Rum examples are particularly significant both for the range of structures exhibited and the clarity of the outcrops.

The most obvious and widespread structure encountered in the Layered Suite is the regular small-scale layering (Plate 14a), now termed rhythmic layering (Irvine, 1987). This is analogous to bedding or stratification in sedimentary rocks. It consists of laminae (<1 cm thick) or layers (>1 cm thick), with contrasted mineralogy and/or textures. Often these contrasts are not very marked unless accentuated by differential weathering. The layers and laminae generally reflect variations in the modal proportions and textural relationships of olivine, plagioclase and clinopyroxene. Less common, but highly distinctive, are thin laminae (1 to 5 mm thick) rich in euhedral chrome spinel crystals. The chromite-rich laminae are also laterally extensive; they commonly occur at the lower contacts of peridotite layers with bytownite-troctolite or anorthositic rocks (Brown, 1956, pl. 3, figs. 26 and 27; see also (Plate 19c) but they are also found within peridotite and anorthositic bytownite-troctolite (McClurg, 1982, figs. 2.6 and 2.34; Bedard et al., 1988, fig. 10). Small-scale irregularities resembling load casts in sedimentary rocks are fairly common (SR 389).

The layers are generally persistent laterally and layering extends upwards through stratigraphical thicknesses of many metres. At any given locality, the layering is found to be broadly conformable with the enclosing major lithological unit or member. Preferred orientation(igneous lamination) of tabular or prismatic crystals is also common (Plate 19b). Although the lamination is usually conformable with small-scale layering this is not always so (e.g. (SR 389)).

There are many deviations from the regular patterns of layering. Some resemble sedimentary structures found in fluviatile or marine arenaceous deposits. These include lateral variations in the thickness of individual layers, cross-stratification, graded layering and slump structures (Plate 14b), (Plate 14c). The graded layering is largely restricted to bytownite-troctolite, the gradation being generally in the proportions of minerals of differing density rather than size, with dense olivine concentrated towards the base of the layer and feldspar progressively more abundant upwards. The graded and cross-stratified layering indicates the action of currents in the magma, at least on a local scale. The graded layers result from sporadic episodes of exceptionally vigorous magmatic circulation, possibly involving localised turbidity currents. The latter origin may explain distinctive graded layering which occurs in the Long Loch Member of the Central Intrusion, on the ridge [NM 3618 9910] west of the Long Loch, where a layer is crowded with rounded pieces of peridotite which grade upwards over about 1 m thickness from 10 cm to under 1 cm in diameter. The graded layer overlies bytownite-troctolite, its basal contact is scalloped and irregular, and flame-like structures project upwards.

Some of the most spectacular sedimentary structures in the Rum Layered Ultrabasic Suite are related to the instability of the crystal cumulates, which has resulted in downslope sliding and slumping. The deformed rocks are commonly associated with relatively steep layering (>15–20°), which suggests either that the depositional slope was too steep for cumulates to build up or that the tilting occurred before the cumulates consolidated. Other factors such as strong magmatic currents, or seismic events likely to be associated with a subvolcanic magma chamber, could also lead to slumping. These slump structures (Plate 14b), (Plate 14c) include convolute bedding and slump folds (characterised by their 'disorganised' form and complex geometry), slump breccias (involving the break up of previously consolidated material), a general irregularity of the dips, and the impersistence of individual layers (see Wadsworth, 1992). The structures are common thoughout bytownite-troctolites in the Eastern Layered and Central intrusions. Individual slump horizons are generally less than 2 m thick, which gives an indication of the minimum amount of unconsolidated crystal mush beneath the magma at a given time. However, in the Eastern Layered Intrusion up to 30 m of contorted layering occurs in bytownite-troctolite in Unit 15 on Askival summit, and thick slumped zones involving both peridotite and bytownite-troctolite are present within Unit 13 on the south-east side of Hallival (at [NM 3952 9607]). Both localities provide good evidence for the presence of thick, poorly consolidated cumulates Elias (1989) suggested that cumulates near the outer edges of the intrusion were unstable and that large volumes of unconsolidated material were transported by mass flow towards the central parts of the complex. He identified four mass-movement depositional episodes in the bytownite-troctolite and peridotite in the upper part of Unit 12 on Hallival.

Harrisitic peridotite and bytownite-gabbro are some of the most distinctive components of the Rum Layered Suite. The harrisite layers are generally an integral, non-transgressive part of the layered sequence rather than later, pegmatitic sheets (but see Donaldson, 1982). They reach their best development in the Western Intrusion where they form a high proportion of the succession between Harris Bay and the summit of Ard Mheall. They also occur sparingly elsewhere in the Layered Suite. The harrisitic structure is formed by elongate olivine grains with a branching or skeletal (parallel growth) habit generally at high angles to the layering (Plate 15). The texture is emphasised by the interstitial feldspar and, to a lesser extent, by pyroxene. The top surface of a harrisitic layer is frequently irregular as the result of differential growth rates and it is always defined by an abrupt decrease in grain size. Harrisitic layers range from centimetres to over a metre in thickness and individual layers are rather uniform. Some layers are also composed of moderately large olivines in apparent random orientation. The latter are common in the lower, marginal part of the Harris Bay Member at Harris (as at [NM 3351 9562] ). Originally described by Harker as 'pegmatoid', the harrisites were reinterpreted by Wager and Brown (1951) and Wadsworth (1961) as upward growth structures developed at the magma–cumulate interface during periods of non-deposition of cumulus crystals. Donaldson (1974, 1976) has made a detailed investigation of harrisitic olivines and interpreted the textures to be the result of rapid growth under conditions of extreme supersaturation induced by slow cooling.

An equally striking but less common structure, described by Donaldson et al. (1973), consists of radiating, bifurcating rays of plagioclase up to 40 cm in length which enclose innumerable small olivine crystals. The type examples of these 'poikilomacrospherulitic' feldspars occur in the Dornabac Member of the Central Intrusion on a shelf [NM 3569 9772] east of Loch an Dornabac. The crystals grew in situ, possibly from a hydrous feldspathic magma (Donaldson et al., 1973).

Replacement structures

Small-scale replacement of bytownite-troctolite by peridotite is quite common. Structures originating by replacement include 'upward-growing pyroxene structures' (Brown, 1956, pl. 2, fig. 20), now termed 'finger structures' (e.g. Robbins, 1982; Butcher et al., 1985) and late-stage veins (Butcher, 1985). The finger structures are widespread. They consist of thin, elongated bodies of dunitic or feldspathic peridotite, 2–3 cm in diameter, that extend from feldspathic or dunitic peridotite upwards for several centimetres, or tens of centimetres, into more feldspathic peridotite or bytownite-troctolite (Plate 16). Cumulus olivine and plagioclase are cemented by poikilitic, upward-tapering crystals of clinopyroxene. Disturbance of small-scale layering is uncommon in the overlying rocks where both igneous lamination and slump-structures may be cut by fingers. These features, which are not restricted to Rum (e.g. Robbins, 1982; Claydon and Bell, 1992, fig. 14), are probably related to post-depositional, 'diagenetic' adjustments within the crystal sediments. They were interpreted by Butcher et al. (1985) as replacement structures formed as intercumulus magma expelled during compaction of the underlying peridotite migrated up into the overlying, more feldspathic rocks. Morse et al. (1987) suggested that the fingers developed on the upper surfaces of sills of mafic magma intruded into nearly solid bytownite-troctolite.

Less-regular, larger-scale replacement of bytownite-troctolite and anorthositic rocks by peridotite occurs in units high in the Eastern Layered Intrusion (Bedard et al., 1988). On the northern shoulder of Hallival [NM 390 968], the classic anorthositic bytownite-troctolite-chromitite–feldspathic peridotite sequence at the Unit 11/Unit 12 boundary (Brown, 1956, pl. 3, figs. 26, 27) occurs as isolated areas within peridotite that has replaced considerable amounts of the bytownite-troctolite at the top of Unit 11. Here, and elsewhere in the layered sequence, the chromitite layer may persist as a relict in the replacement peridotite (Butcher et al., 1985, fig. 8; Emeleus, 1987, fig. 13).

Near the top of Unit 9 bytownite-troctolite, on a shelf [NM 395 969] 700 m due North of Hallival summit, there is a wavy, irregular contact between bytownite-troctolite and overlying dark, clinopyroxene-rich bytownite-gabbro. The contact is discordant to igneous lamination, anorthosite schlieren and coarse-grained layers, some of which cross the boundary. Pipe-like masses of bytownite-troctolite extend upwards into the pyroxene-rich rock which also contains isolated, rounded areas of bytownite-troctolite (Brown, 1956, figs. 24, 25; Young and Donaldson, 1985, figs. 1, 3; Bedard et al., 1988, fig. 19; Elias, 1992). Young and Donaldson (1985) suggest that the undulations formed spontaneously at the interface between two fluids with a reverse density gradient. Identical structures in Unit 14 on Trallval [NM 3732 9543] are described by Volker and Upton (1990, p.78, fig. 17) who interpret them as flame-structures caused by loading; similar features in the Bushveld Intrusion are also attributed to loading (Lee, 1981, fig. 8A, B). If the structures are caused by loading, it follows that the laminae, igneous lamination and textural layering formed at a later stage in the consolidation of the cumulates, implying a high degree of post-cumulus, diagenetic re-crystallisation (cf. Irvine, 1980 and Hunter, 1987). An alternative explanation of these phenomena is that the pyroxene-rich bytownite-gabbro formed metasomatically from bytownite-troctolite permeated by basaltic melt (Bedard et al., 1988; Bedard and Sparks, 1991; Elias, 1992; but see Volker and Upton, 1992).

The peridotites in the Eastern Layered Intrusion have generally been regarded as an intergral part of the cumulate sequence (e.g. Brown, 1956; Wager and Brown, 1968). However, it is now recognised that there are many instances where peridotite forms intrusions. Peridotite and rare dunite plugs cut the layered rocks (Figure 35); McClurg, 1982; Butcher, 1985, fig. 1), peridotite originally mapped as the basal part of Unit 9 on the northern slopes of Hallival [NM 391 970] intrudes parts of Units 9 and 10 (Bedard et al., 1988, figs. 15, 16), and thin tongues and apophyses of peridotite intrude bytownite-troctolite near the top of Unit 14 on Hallival [NM 3959 9629] (Renner and Palacz, 1987, fig. 3), however disruption of layered peridotite during slumping may explain the relationships seen at this locality. Bedard et al. (1988) have suggested that many of the thick peridotites in the Eastern Layered Intrusion may be intrusive sheets rather than members of the cumulate sequence. However, Volker and Upton (1990; 1991) maintain that the majority of the conformable peridotite layers in the Trallval succession are depositional rather than intrusive, but do identify intrusive peridotites which they suggest were formed when crystal mush in unconsolidated olivine cumulates became squeezed out during periods of compression. Elias (1992) has shown that apparently intrusive features may arise when layering is disrupted during mass flow.

Veining in the Layered Suite

Thin ultrabasic and basic veins commonly cut members of the Layered Suite. The veins, which are usually between 2 and 3 cm thick but range up to 30 cm thick, are generally concordant to the layered structures but may be transgressive (Butcher, 1985, fig. 3). They are commonly pyroxene-rich and of basic composition, many are gabbroic. From a study of veins in Unit 10 in the Eastern Layered Intrusion and in parts of the Western Layered Intrusion, Butcher (1985) concluded that they formed when late-stage liquids were injected into fractures in crystallising cumulates. The cumulus minerals in the host rocks have been considerably modified through reaction with the migrating liquids for up to 1 cm either side of the veins. Butcher (1985, fig. 8b) found continuous changes in Mg and Fe (but not Ni or Mn) in olivine; peridotite adjoining the veins contains olivine with a composition of about Fo77 compared with a normal composition of about Fo84. Butcher terms this very localised alteration 'channelled metasomatism'.

Veins and segregations of a different type have been described by Kitchen (1985) in peridotite of the Central Intrusion. They are exposed in the banks of the Kilmory River for some hundreds of metres south of Salisbury's Dam [NM 3642 9989]. Pale grey to chalky white teschenite forms irregular segregations up to about 30 cm in diameter, thin veins which vary from a few millimetres to 12 cm in width, and rare dykes. The rock consists of coarse olivine (Fo70-74), Ca-rich clinopyroxene and plagioclase (Anal, zoned to Ano rims), with minor amounts of kaersutite, phlogopite, ilmenite, magnetite and apatite. Interstitial areas of analcime form from 10 to 15 vol. % of the rock, they enclose euhedral actinolite, albite, sphene and zircon, skeletal crystals of arfvedsonite, apatite and epidote, and spherulitic chlorite. The occurrence of these veins is an indication that crystallisation of peridotites on Rum can lead to the formation of undersaturated, alkali-rich residua (Appendix 5i; see also Chapter 9).

Ultrabasic breccias

Ultrabasic breccias are a distinctive feature of the Central Intrusion but are less common in the other parts of the Layered Suite. Breccias occur along the contact between the Central Intrusion and the Western Layered Intrusion from the Abhainn Rangail [NM 347 957] northwards but are less common at the contact with the Eastern Layered Intrusion. Linear breccia zones form a high proportion of the Central Intrusion itself; they appear to radiate from several foci in the Ruinsival area [NM 356 940] (Volker and Upton, 1990, fig. 24).

The breccias consist of blocks of peridotite, feldspathic peridotite or bytownite-troctolite (and in some places all three), in a matrix of feldspathic peridotite, bytownite-troctolite or, less commonly, of peridotite. Breccias near the margins of the Central Intrusion contain blcicks derived from the Western or Eastern layered intrusions, elsewhere blocks in the breccias come from members of the Central Intrusion. The blocks are angular, rounded, or subrounded and may appear to be plastically deformed. They are close-packed or scattered through matrices which are structureless or show streaky, discontinuous banding. Commonly, any banding in the matrices is deformed around blocks. The fragments range from a few centimetres to blocks of well-layered rock several metres in diameter (Plate 17). The area south-east of the Long Loch abounds in exposures that suggest flowing masses of feldspathic peridotite with entrained blocks; low crags at [NM 3635 9790] contain disoriented blocks of layered bytownite-troctolite and peridotite in streaky bytownite-troctolite and feldspathic peridotite, whilst nearby, banded feldspathic peridotite and bytownite-troctolite divide over a large layered block (Plate 18).

Wadsworth (1961) regarded the breccias as scree-like deposits formed at fault scarps within the magma chamber. Breccias at the south-eastern end of Harris Bay [NM 342 948] and in the Ard Mheall Member suggest to Donaldson (1975) that the fragmentation resulted from intrusion of feldspathic peridotite magma into consolidated peridotites (see also McClurg, 1982; Volker, 1983; Volker and Upton, 1990, fig. 24). Wadsworth (1992) remapped breccias in the Long Loch–Loch an Dornabac [NM 355 976] area, where the peridotite and bytownite-troctolite fragments have been derived from the adjoining, older Ard Mheall Member. He interpreted these breccias as Intra-formational debris accumulations, perhaps associated with submagmatic faulting in a central graben' and compares them with peridotite breccias on Duke Island, Alaska (Irvine, 1987). Both views are compatible with the formation of the breccias in the developing graben linked with emplacement of the Central Intrusion (McClurg, 1982; Volker, 1983; Volker and Upton, 1990; Wadsworth, 1992); cumulates will have collapsed and avalanched down fault scarps as debris flows, and at the same time crystal- and fragment-laden magma will have been injected amongst the blocks and into fissures in the graben walls and other extensional fractures.

Other peridotites in the Rum Central Complex

Three north–south elongated tongues of peridotite intrude the Northern Marginal Zone between the Long Loch and Loch Bealach Mhic Néill [NM 376 989] (Figure 35). The easterly intrusion forms a prominent ridge of brown-weathering peridotite which intrudes the Main Ring Fault. The westerly tongue is cut by Central Intrusion breccia south-east of the Long Loch. Although no contact was located, all three tongues appear to predate the Eastern Layered Intrusion since layering in adjoining bytownite-troctolite cuts across the strike of the tongues. Steep to vertical banding, striking north–south, is conspicuous in peridotite on the eastern ridge, in contrast to the low dips of the layering in the bytownite-troctolite to the south. The tongues are made of peridotite and less commonly feldspathic peridotite (Dunham, 1965b, table 1). Amounts of olivine are fairly constant and the layered structures reflect variation in the amounts of interstitial plagioclase and pyroxene, this has been termed 'matrix banding' by Dunham (1965b).

Several peridotite plugs intrude rocks of both the Eastern Layered and Central intrusions on and near Barkeval [NM 375 972] (Figure 35). The largest forms the Barkeval summit area and the steep, rough south-western end of the hill. A north–south-elongated dunite plug crops out 500 m south east of the southern end of the Long Loch. It cuts breccias of the Central Intrusion and a gabbro plug which intrudes rocks of the Eastern Layered Intrusion. The gabbro is dull black, due to thermal metamorphism by the dunite.

Petrography and mineralogy

The broad outlines of the petrological variations within the Rum Layered Suite have already been indicated, the principal feature being the cyclic character of the large-scale (unit) layering, within which occurs small-scale, rhythmic layering (cf. Irvine, 1987). Both the petrological and chemical characteristics of each unit are controlled primarily by the cumulus mineralogy (Plate 19a),(Plate 19b),(Plate 19c),(Plate 19d); (Table 7) for modal data; selected chemical analyses are in Appendix 5i).

The peridotites are composed dominantly of cumulus olivine (Plate 19a). Cumulus chromite is ubiquitous and is concentrated locally into thin laminae (Plate 19c). Plagioclase and augite occur as large poikilitic (inter-cumulus) crystals enclosing olivine and chromite. In the more feldspathic peridotites, euhedral plagioclase crystals are an incipient cumulus phase. Most of the variation in the olivine-rich cumulates reflects variation in the shape of olivine grains and in the proportion of intercumulus material (clinopyroxene and plagioclase). Olivine in the normal, non-harrisitic peridotites varies from small equigranular crystals in random orientation to large tabular forms laminated parallel to the layering. Olivines may show strain-induced twinning. The Rum peridotites are commonly extreme adcumulates, with rare marginal zoning of the feldspar. They show a high degree of textural equilibrium (Hunter, 1987), and it is likely that any primary compositional zoning would have been lost during the high-temperature recrystallisation.

The texturally distinctive harrisitic rocks are generally more feldspathic than the normal cumulate layers. The pyroxene content of each is uniformly low, except in the gabbroic Harris Bay Member. Small amounts of a brown amphibole and biotite are common, as are chlorite, colourless amphibole and sericite as alteration products, reflecting the relatively large proportion of liquid held within the embayments and interstices of the upward-growing olivines. There is only slight zoning of the feldspar and pyroxene, and none in the olivine. Donaldson (1974) argued that the common occurrence of skeletal and dendritic olivine crystals reflected rapid growth from slowly cooled magma supersaturated in olivine. He concluded that at least some of the Rum harrisitic rocks formed during the rapid crystallisation of pods and lenses of migrating interstitial melt that were trappedlocally beneath impermeable barriers in the cumulates (Donaldson, 1982). Branching in the layers at Harris (at [NM 3357 9554] perhaps indicates that harrisitic rocks could arise from separate injection of this migrating melt.

The bytownite-troctolites display a much wider modal variation (Table 7), and with increased pyroxene they grade into bytownite-gabbro. With the proposed narrow limit of 5 per cent modal pyroxene in troctolite (LeMaitre, 1989, fig. B6), many of the feldspar-rich layered rocks are more correctly classified as bytownite-gabbro or bytownite-leucogabbro (e.g. in the upper part of Unit 10 (Figure 33); Tait, 1985, figs. 3a, 4a)). Plagioclase is the dominant cumulus mineral, although olivine and augite may be concentrated locally; chromite is scarce or even absent. These rocks are predominantly adcumulates, although mesocumulates are common in Units 1–6 of the Eastern Layered Intrusion. The rocks are laminated and the tabular feldspar are generally parallel to layering (Brothers, 1964). The number of cumulus phases and their relative abundances control the textures. Cumulus olivine and plagioclase are commonly cemented by overgrowths of olivine, plagioclase rims and poikilitic augite crystals. Where augite is also a cumulus phase, the cumulus character of the texture is less obvious (Plate 19b). Extreme plagioclase adcumulates ('anorthosites') are strongly laminated and this texture may be emphasised by scattered poikilitic patches of augite and olivine. The high degree of textural equilibration and the homogeniety of these rocks suggests that they experienced extensive post-cumulus textural reorganisation (Hunter, 1987, fig. 10a), a view reinforced by the occurrence of small, disoriented feldspars enclosed by poikilitic pyroxene upon which lamination in the surrounding feldspars is moulded. Olivine in the bytownite-troctolites commonly contains thin (c.1 pm), parallel-oriented plates of magnetite and brown chrome-spinel which may be dendritic in form.

Uranium-bearing minerals zirconolite and baddeleyite have been found in intercumulus pockets (Williams, 1978; cf. Henderson et al., 1971). Late crystallising sulphides also occur. Dunham and Wilkinson (1985) described and analysed pyrrhotite, pentlandite, cubanite, bornite, digenite, chalcocite, chalcopyrite, troilite, native copper and electrum (with 46.4% Au and 49.9% Ag; Dunham and Wilkinson, 1985, table 5) from small rounded areas within and above the Unit 11/12 chromitite of the Eastern Layered Intrusion. Sulphides also occur lower down in peridotites near Allt nam Bà (Faithful!, 1986) where high background levels of gold and platinum-group elements have been identified, especially in Unit 1 (Chapter 13; Hulbert et al., 1992).

The evidence from oxygen isotopes indicates that minerals throughout the suite have reacted with circulating heated meteroric waters although the degree of equilibration is variable, being most pronounced near the margins of the Layered Suite (see below). Despite this, hydrous alteration is not extensive except near the margins of the complex where circulation of heated meteoric water may have been at its highest (Figure 37). However, very thin veins (1–5 mp) and small patches of sericite, biotite, magnetite, colourless to pale green amphibole, carbonate and chlorite occur in most rocks, and olivine may be altered to brown clay minerals along veinlets and at crystal margins.

Chromite-rich laminae

Thin chromitite and chromite-rich laminae occur in peridotites and between peridotite and bytownite-troctolite at the contacts of some major layered units (Plate 19c); Brown, 1956, figs. 25, 26). The chromite-rich (= chrome-spinel cumulate) seams may extend for over 2 km and are normally much less than 15 mm thick (McClurg, 1982, p.45). The euhedral, deep brown to opaque chromite crystals are generally <0.2 mm in diameter and may enclose rounded areas (<0.05 mm diameter) containing kaersutite, phlogopite, enstatite, apatite and other minerals (Volker, 1983).

The chromite-rich laminae may have originated by sedimentation of chromite crystals and rare geopetal structures support this origin (Volker, 1983). However, such small crystals might not normally sink in mafic magma. They are more likely to have grown in situ, when fresh picritic magma came into contact with cumulates containing trapped, feldspathic liquid (cf. Sharpe and Irvine, 1983). Growth in place is suggested by chromitite rinds on irregular bytownite-troctolite-peridotite surfaces as, for example, where peridotite cuts across the Units 11/12 contact 750 m north-west of Hallival summit. Volker (1983, p.141) also records chromitite coatings on bytownite-troctolite and gabbro enclosed by peridotite.

Mineral compositions, phase layering and cryptic layering

Compositional data are summarised in (Table 8). The minerals in the Rum layered rocks show remarkably limited compositional ranges compared with other layered intrusions (cf. Wager and Brown, 1968, fig. 14). The modal mineral variation in the Layered Suite is accompanied by slight systematic changes in mineral compositions ('cryptic layering'; Wager and Deer, 1939). The overall pattern of change from the base to the top of a unit is from higher- to lower-temperature members of solid-solution series. Olivine is the only cumulus phase which is always present throughout each unit. Consequently, the basic pattern of cryptic layering is best expressed in terms of forsterite/fayalite variation. Other minerals provide supporting evidence.

The cryptic variation in several layered units is summarised in (Figure 33). Iron-enrichment of olivine generally increases from the bottom to the top of a unit, although there are departures from this pattern Similar high- to low-temperature compositional ranges are present in cumulus plagioclase, and less commonly in pyroxene and/or chrome-spinel. Sharp breaks in compositional trends do not always match unit boundaries; the most magnesian olivines may occur a few metres above the base of a unit or the most iron-rich olivines may be below the top. Additionally, olivines may exhibit compositional reversals within a unit, apparently related to the thickness of olivine-rich layers (e.g. Faithfull, 1985, fig. 5; ((Figure 33)c).

Dunham and Wadsworth (1978) consider that the cyclic pattern of phase and cryptic layering is most convincingly explained in terms of cumulate deposition from repeated influxes of fresh magma into a subvolcanic magma chamber, periodically emptied by volcanic activity, much as proposed by Brown (1956; see also Wadsworth, 1961). The cryptic variation in each unit thus records slight progressive fractionation as a batch of magma cooled. The reversals near the top of each unit were attributed to some mixing between the residual magma of one unit with the first incursion of the next pulse of fresh, undifferentiated magma. Alternatively, the pattern of variation could arise when an influx of new, dense picritic magma set up a convective system, displacing the residual, low density evolved interstitial magma in the underlying bytownite-troctolite cumulates (Tait, 1985). Cumulus minerals in the bytownite-troctolite would equilibrate with downwards-percolating primitive liquid to give less-evolved compositions, with complementary effects in the basal part of the later, overlying peridotite. Elsewhere, compositional gradients either side of thin basic veins provide supporting evidence of the ease with which olivine may re-equilibrate (Butcher, 1985).

Other cumulus phases of more limited stratigraphical range may mirror the changes in olivine compositions (Dunham and Wadsworth, 1978; Faithfull, 1985; 1986; Tait, 1985). Cumulus chromite in particular shows significant variation in Mg/ (Mg + Fe). However, when interpreting chromite data, it is important to minimise the effects of possible late-stage reactions between the chromite and surrounding minerals (cf. Henderson, 1975; Henderson and Suddaby, 1971; Henderson and Wood, 1981).

Within Unit 10, there are compositional breaks at a major boundary of peridotite with bytownite-troctolite ((Figure 33)a) and similar breaks occur elsewhere within other units (cf. ((Figure 33)c). The Mg-Fe variation in olivine is gradational but Ni shows a sharp break. This element was possibly not as suceptible to change during olivine re-equilibration, thus the present Ni content may provides a 'memory' of the initially forsterite-rich olivine in the peridotite (see also Irvine, 1980, fig. 8, but note Volker, 1983, p.146; cf. also changes during 'channelled metasomatism' , p.72).

Geochemical characteristics of rocks in the Layered Suite

Unit 10 of the Eastern Layered Intrusion has been treated as the type layered unit by Brown (1956) and the elemental and isotopic compositions of this unit have been examined in considerable detail by Tait (1985) and Tait and Palacz (1985), whose analyses are taken as representative of peridotite, feldspathic peridotite and bytownite-troctolite (Appendix 5i, samples 5, 25 and 29 respectively). In the peridotite, high MgO (37.8%) and Ni (>1900 ppm) reflect the abundance of magnesian olivine, and the high Cr (>6000 ppm) is contained in discrete chromite, in chromite inclusions in the olivine and in clinopyroxene (Tait, 1985). The low amounts of CaO (3.7%), Na2O (0.24%) and A12O3 (4.5%) reflect the small amount of plagioclase ((Figure 33)a); Tait, 1965, figs. 3, 4). The incompatible element concentrations are very low; the rare-earth elements are between one and two times chondrite amounts (Palacz and Tait, 1985, figs. 4, 6) with a slight positive Eu anomaly. The compositional differences between the feldspathic peridotite (Appendix 5i, 25) and peridotite result from increased modal plagioclase rather than major changes in mineral compositions ((Figure 33)b); Tait, 1985, fig. 3; Palacz and Tait, 1885, fig. 4 cf. 22 with 6). The mineral modes and compositions (Tait, 1985, fig. 3)and bulk rock composition (Appendix 5i, 29) of the bytownite-troctolite differ markedly from the peridotites. Notably higher CaO (14.0%), Na2O (1.74%), Al2O3 (16.8%) and Sr (228 ppm) are due to the high proportions of slightly more sodic plagioclase, whereas the lower MgO (13.1%) and Ni (253 ppm) are largely the result of a smaller modal amount of a less forsteritic olivine. There is a progressive decrease (to 752 ppm in 29) in Cr upwards in Unit 10, due to decreasing amounts of chromite. Incompatible elements are generally slightly more abundant in the bytownite-troctolite with the rare-earth elements at about two or three times the chondrite values, there is a well-defined positive Eu anomaly and a suggestion of light rare-earth enrichment. In detail, there are compositional variations within the rock-types of this unit, some of which may have been caused by melt percolation from overlying and underlying units. Dunitic peridotite (Appendix 5i, LP) from near the base of Unit 14 has a more extreme composition (MgO 40.0%, Al2O3 2.0%, CaO 1.23%, Cr 2515 ppm, Ni 1708 ppm) due to very high concentrations of olivine (87.6 vol. % of Fo85; Renner and Palacz, 1987), it resembles an olivine adcumulate from Ard Mheall (Wadsworth, 1961, table 4). Volker (1983) analysed the matrices of ultrabasic breccias in the Central Intrusion and found that their major and trace element contents were essentially similar to the peridotites and bytownite-troctolites; they are close to mixtures of varying proportions of olivine (Fo88) and plagioclase (An84) (Volker, 1983, fig. 6.9). Differences between analyses of a harrisitic cumulate and an olivine adcumulate from the Ard Mheall Member of the Western Layered Intrusion (Wadsworth, 1961, table 4) appear to be caused largely by higher modal amounts of calcic plagioclase in the harrisitic rocks. From the common occurrence of amphibole, phlogopitic mica and apatite in the harrisitic rocks, there might be expected to be significantly more incompatible elements but modern analyses are lacking; however a segregation in harrisitic peridotite (Appendix 5i, HR) has the composition of feldspathic peridotite without any clear indication of incompatible elements enrichment.

The chilled fades of the Eastern Layered Intrusion on Beinn nan Stac (Greenwood et al., 1990; Appendix 5i, SR 209) has the composition of picritic basalt. The slightly higher K2O (0.44%) and Rb (9 ppm) than in the least evolved basalts on Rum (e.g. SR 217, Appendix 5e:i) may be due to contamination by Torridonian feldspathic sandstone (see p.66).

Radiogenic isotope characteristics of the Layered Suite

Studies on radiogenic isotopes in the upper part of the Eastern Layered Intrusion, from Unit 9 to Unit 15, have been made by Palacz (1984, 1985), Palacz and Tait (1985) and Renner and Palacz (1987). The variation in 87Sr/86Sr through the upper part of the Eastern intrusion is summarised in (Figure 34); Palacz (1985) states that age corrections are insignificant since Rb is an excluded element from the cumulates. In addition, Greenwood (1987) determined Sr and Pb isotopes in the marginal contact zone around the Layered Suite (pp.82–83).

In Unit 10, 87Sr/86Sr ratios in the peridotite are low, between 0.7036–0.7043 and 143Nd/144Nd (at 60 Ma) is 0.51281. Sr and Nd isotopes in the feldspathic peridotite show a significant variation up the sequence, 87Sr/86Sr changes from 0.7049 to 0.7053 and 143Nd/144Nd from 0.51271 to 0.51253. There is a marked increase in 87Sr/86Sr in the bytownite-troctolite to around 0.706 and in 143Nd/144Nd to between 0.51249 and 0.51238. Unit 9 is unusual in that the upper half of the unit consists of a gabbroic rock; the change upwards from bytownite-troctolite to gabbro is sudden and is accompanied by a striking change in 87Sr/86Sr from about 0.705 to a fairly uniform 0.704 in the gabbro (Figure 34); see also Chapter 9). The bytownite-troctolites in the Eastern Layered Intrusion are characterised by relatively high radiogenic Sr compared with the olivine-rich rocks and, using all the available isotopic evidence, it is generally thought that they crystallised from magmas contaminated by amphibolite-facies Lewisian crustal material whereas the magmas responsible for the peridotites were relatively uncontaminated (e.g. Palacz, 1985). In Unit 14 on Hallival, however, much of the bytownite-troctolite has low radiogenic Sr (Figure 34) suggesting that uncontaminated basaltic magma entered the magma chamber from time to time (Renner and Palacz, 1987). From an examination of Pb isotopes Palacz (1985) concluded that the layered rocks had undergone little contamination from Torridonian Pb but there had been significant contamination by crustal lead probably derived from amphibolite facies Lewisian gneisses (Palacz, 1985, fig. 5).

Gabbros in the Rum Layered Suite

Gabbro sheets and other intrusions form parts of the Eastern Layered Intrusion, the Central Intrusion and the lowest member of the Western Layered Intrusion (Figure 35). Gabbro, olivine-gabbro, bytownite-(olivine)-gabbro (formerly 'eucrite') and (rarely) quartz-gabbro are all represented.

Details

Numbers in parenthesis in the following descriptions refer to numbered gabbros in (Figure 35).

Eastern Layered Intrusion

Marginal gabbro (1)

Brown (1956, fig. 2) distinguished a zone of structureless gabbro on the northern and eastern margins of the intrusion. The width is variable, from over 100 m to apparently absent west of Allt a' Mhill Bhric [NM 383 980]. The easy weathering of the mafic rocks leads to poor exposure. The mineralogy is rather variable: augite, strongly zoned plagioclase (An77_51) and opaque oxides are present; these may be accompanied by quartz, orthopyroxene or olivine.

The external contacts of the Layered Suite were studied by Greenwood (1987) who found that ultrabasic rocks crop out close to the Torridonian rocks and Western Granite. Furthermore, he noted peridotite amongst the mafic blocks in intrusion breccias, as in upper Dibidil [NM 3831 9553] and north of Allt nam Bà, and found rare examples of chilled picritic rocks in contact with the country rocks (Plate 19d); Greenwood et al., 1990). Greenwood (1987, p.20) contends that the field and geochemical evidence shows that the marginal gabbroic rocks are not a separate intrusion but gabbroic rocks formed as the result of contamination of peridotite and bytownite-troctolite by siliceous country rocks; Pb and Sr isotopic studies confirm that crustal contamination occurred: the Torridon Group feldspathic sandstones (87Sr/86Sr >0.757 at 60 Ma.; Greenwood, 1987, table 6.1) were melted or partially melted at temperatures estimated at 960° ± 40°C, and Greenwood suggests that these anatectic melts mixed with basaltic liquids during boundary flow, eventually crystallising as gabbroic rocks with high radiogenic Sr (87Sr/86Sr from c.0.705 to 0.707). Microgranodioritic rocks (with 87Sr/86Sr from c.0.7107 to 0.7145) also found in the contact zone are thought to have formed by combined assimilation and fractional crystallisation. Oxygen isotope studies have shown that these marginal rocks have undergone large-scale exchange with heated meteoric water (Figure 37); Greenwood, 1987; Greenwood et al., 1992) which may account for their commonly altered condition (e.g. (DU 25665), (DU 25667)).

Conformable sheets (2)

Two sheets of fine-grained, pyroxene-rich bytownite-troctolite and troctolitic gabbro occur on the eastern side of the Hallival-Askival ridge at the levels of Units 4 and 5. The sheets merge north of Beinn nan Stac where their total thickness is about 110 m (Brown, 1956); they probably continue as thick bytownite-troctolite exposed in upper Dibidil. The purple-grey colour of these rocks readily distinguishes them in outcrop from the dull grey of normal bytownite-troctolite and the warm brown colour of peridotite. They have similar chemical compositions (Appendix 5h, 3.18) to the bytownite troctolites in the layered succession.

The rock is fine- to medium-grained with impersistent, wispy banding and quite common feldspar lamination (Faithfull, 1986, but cf. Brown, 1956, p.40). North and south of Allt Mór na h-Uamha [NM 402 970] the lower sheet contains xenoliths of peridotite, bytownite-troctolite, gabbro and beerbachite, a fine-grained basic rock with equigranular texture. Inclusions are also common in the upper sheet, as in east-facing crags at [NM 4010 9675]. Close to this locality the upper part of the sheet is in sharp, intrusive contact with peridotite of Unit 6 where it contains many peridotite xenoliths apparently derived from the overlying unit. The sheets are feldspar-rich olivine gabbros and troctolites, rocks that are texturally indistinguishable from bytownite-troctolite in the layered units. The sheets contain euhedral plagioclase (An62–64) with rare calcic cores (An69; Brown, 1956), pale brown augite is abundant and anhedral olivine (c.Fo79; Faithful, 1985) is common.

Harker (1908) and Brown (1956, fig. 2, p.39) interpreted the sheets as intrusive gabbros. Faithfull (1985; 1986) noted mineralogical and textural similarities between the sheets and bytownite-troctolites in the layered sequence, and found gradational contacts between sheets and their surroundings. He interpreted the upper sheet as bytownite- troctolite of Unit 5 and the lower, xenolith-rich sheet as the upper part of Unit 4 (Faithfull, 1985, p.461). During the resurvey the balance of evidence, as described above, appeared to favour an intrusive origin for these sheets.

Askival Plateau Gabbro (3)

The Askival Plateau Gabbro (Brown, 1956; Faithfull, 1986) consists of numerous sheets and veins of gabbro, of average thickness 8 cm. The gabbro intrudes and disrupts layering in Unit 10 on a prominent shelf [NM 400 952] east of Askival. The gabbro is rich in anhedral crystals of opaque oxides, the feldspars are granular and bent, suggesting intrusion in a partially solid condition (Brown, 1956, p.40). The gabbro may form part of (4).

Atlantic Corrie Gabbro (4)

This gabbro forms a plexus of sheets that intrude and brecciate the Eastern Layered Intrusion in Atlantic Corrie [NM 385 962] and on the northern side of Trallval (eg. at [NM 3822 9564]; Volker and Upton, 1990, p.71). The rock characteristically contains well-laminated plagioclase (Cores of An85, zoned to An61 at the margins and in a few instances to An54), less commonly clinopyroxene may also be laminated; opaque oxides are scarce, subhedral olivine (Fo77 cores, Fo68 margins) always occurs. Biotite is locally abundant. The gabbros have a subdued, undulatory banding on a scale of 5 to 20 cm (Volker, 1983)

Other gabbros

The East Trallval Gabbro is a dyke-like gabbro on the east end of Trallval c.[NM 380 951]. It intrudes the layered ultrabasic rocks and the Atlantic Corrie Gabbro. This rock is pale coloured, fine to medium grained, plagioclase rich (An55-44) and with rare olivine (Fo65); biotite and opaque oxides are common. The West Trallval Gabbro (5) (Volker, 1983, p.108) on the western end of Trallval [NM 369 954] to [NM 373 959] is a thin dyke, of average width 15 m, which extends for almost 1 km in a NNE–SSW direction. The rock is a coarse-grained melagabbro with conspicuous ophitic clinopyroxene up to 15 cm diameter and minor olivine. It is intruded along the fault that separates the Eastern Layered Intrusion from the Central Intrusion.

Several sheets of coarse-grained gabbro intrude the layered rocks between 300 and 1100 m NNW of Hallival summit. Brown (1956, p.40) noted the calcic plagioclase (cores An89–86, rims zoned to An66) and termed these rocks the Barkeval 'eucrite' (bytownite-gabbro) (6). He suggested that they were formed from late interstitial liquid squeezed out of nearby disturbed layered rocks. A gabbro intrudes bytownite-troctolite of Unit 7 and Unit 8 chromitite and peridotite in south-east Coire Dubh [NM 3900 9742], this also contains zoned, calcic plagioclase (cores An80–69, rims An64,69; Bedard et al., 1988, p.213).

A small plug of fine-grained olivine gabbro (11) cuts the contact of the layered intrusion with rhyodacite 1.2 km north of Hallival. The rock is altered, laths of sericitised labradorite are enclosed in ophitic clinopyroxene and anhedral, altered olivine. Skeletal opaque oxide crystals are partly mantled by biotite, brown to green amphibole and chlorite are secondary after mafic minerals.

Western Layered Intrusion

A marginal gabbro similar to that of the Eastern Layered Intrusion occurs at several places between the layered rocks and the earlier Western Granite (Wadsworth, 1961). On the eastern bank of the Glen Duian Burn, immediately upstream from the bridge [NM 3379 9601], the Harris Bay Member is intruded by a suite of slightly transgressive, thin (20 to 40 cm) gabbroic sheets termed the Glen Duian Gabbro (12) (Wadsworth, 1961, p.56). The rock is generally altered. Labradorite, with rare bytownite cores, is present but the original mafic minerals are largely replaced by epidote and chlorite.

Central Intrusion

Glen Harris Gabbro (7)

Coarse-textured olivine-gabbro crops out on the floor of Glen Harris and southwards to Ruinsival and Leac a' Chaisteil c. [NM 367 936]. The gabbro has a 700 m dextral offset on the Long Loch Fault (Figure 30). East of the fault the gabbro forms a steep sided boss, but to the west it forms flat-lying, possibly wedge-like bodies that intrude layered ultrabasic rocks and breccias (Volker, 1983, p.109, fig. 4.2D). It consists of euhedral to subhedral zoned crystals of bytownite (cores An85-84, margins to An65), discrete olivine crystals (c.Fo80) that are somewhat moulded on the plagioclase, poikilitic clinoproxene and magnetite. The rock was termed the Glen Harris Eucrite by Wadsworth (1961). Volker estimates the gabbro west of the Long Loch Fault to be downthrown by 200 to 400 m. Veins and sheets of gabbro are in sharp but unchilled contact with peridotite. Peridotite also occurs as xenoliths, some of which are hornfelsed and others plastically deformed.

Gabbros of An Dornabac (10)

Olivine-gabbro intrudes bytownite-troctolite of the Long Loch Member at the north end of the An Dornabac ridge [NM 353 972]. This dark, coarse-grained rock contains streaky areas of laminated feldspar-rich rock and blocks of basic hornfels. Normally zoned labradorite (An66-50) with rare bytownite cores (Ann) is typically clouded, olivine is rare. On the west side of the hill, a dyke of bytownite-gabbro up to 100 m in width extends for 1 km SSW from Loch an Dornabac [NM 354 974], transgressing layering in the Ard Mheall Member of the Western Intrusion.

Other gabbros

Olivine-bearing and olivine-free gabbro plugs and sheets form the majority of the coastal exposures between Papadil and Rubha Sgorr an t-Snidhe [NM 344 931] (8) and several smaller bodies occur between the coast and the summit of Ruinsival. The gabbros intrude Torridonian sandstones and may cut the ultrabasic rocks. However, at Rubha na Pairce [NM 361 919] gabbro is intruded by peridotite that encloses many rounded, thermally altered xenoliths of gabbro and dolerite (Emeleus and Forster, 1979, p.41). Similar altered gabbro crops out about 600 m north-east of Loch Papadil.

At the northern end of the Central Intrusion, gabbro (9) generally margins the ultrabasic rocks. These gabbros are heterogenous and range from bytownite-olivine-gabbro to apparently contaminated rock which lacks olivine but contains strongly zoned plagioclase, orthopyroxene, biotite and a mesostasis of quartz and alkali-feldspar. Several small, dyke-like gabbro intrusions also occur well within rocks of the Central Intrusion in this northern area.

An intrusive breccia of peridotite blocks in gabbro forms the south-west shoulder of Trallval [NM 366 949].

Chemical compositions of the gabbro intrusions

The gabbroic plugs, sheets and veins that cut the Layered Suite cover a wide range of compositions from highly magnesian, olivine-rich gabbros to varieties with silica about 50% and considerable normative orthopyroxene. The selection of the gabbros analysed by Volker (1983, tables 6.13 and 6.14) given in Appendix 5h illustrates the range he encountered in the Eastern Layered and Central intrusions; the numbers (X9, etc.) refer to these analyses.

The most magnesian gabbro (B9; MgO 27.4%) has high Ni (1403 ppm) and Cr (2305 ppm) together with low Al2O3 (7.54%) and CaO (4,78%) and in several respects is similar to the more feldspathic peridotites (Appendix 5i, 29) except that it has appreciably higher incompatible elements (e.g. K2O 0.51, Ba 165 ppm, Rb 16 ppm, and La, Ce, Nd with 10, 18, 9 ppm respectively). Volker notes that the incompatible elements in the gabbros are generally higher than in the layered rocks and in some instances their high concentrations suggest that they have been contaminated with crustal materials. The gabbros analysed by Volker and by McClurg (1982, table 6.1, 6.2) are generally transitional, very few (e.g. FD1) are nepheline-normative and none is quartz-normative. Sample FD1 has unusually high Ba (5417 ppm) but is not otherwise enriched in the incompatible elements. McClurg (1982) and Volker (1983) identified a sparse suite of transitional, mildly alkaline dykes that cut the Layered Suite, the dykes were thought to have evolved by low-pressure fractionation of olivine, spinel, plagioclase and clinopyroxene; Volker (1983, p.226) considered that the gabbro geochemistry was 'related to that of the dyke suite by cumulus enrichment of plagioclase and, in some cases, by crustal contamination'.

Intrusion breccias and thermal metamorphism

The Rum Layered Suite is emplaced into country rocks which are of granitic or near granitic composition, or contain a significant granitic component. The hot mafic magmas have generated rheomorphic acid melts, giving rise to the intrusion breccias that are present at all these contacts (Plate 20); Harker, 1908, pl. II); commonly, there was wholesale mobilisation of the country rocks, as seen in rocks of the Torridon Group margining gabbro on the coast at [NM 3660 9172] south-east of Loch Papadil. The original chilled margins of major mafic intrusions are rarely preserved, they have been fragmented to form the bulk of the basic material in the breccias. This material is generally in the form of angular blocks that lack chilled edges, less commonly there is chilling on the edges of basic inclusions in acid matrices and these margins may show the bulbous, crennelated forms characteristic of the co-mingling of contrasted magmas (Wiebe, 1991). These contact zones illustrate a paradox common throughout the central complexes of the Hebridean Province: mafic bodies are intruded by acid rocks that are clearly older on other reliable criteria.

The thermal alteration of the main country rock types and other contact features are described in this section.

Details

Contacts between The Layered Suite and Porphyritic Rhyodacites

Fine-grained olivine-gabbro intrudes rhyodacite on Meall Breac, at [NM 3845 9795] (Dunham, 1964). As the contact with gabbro is approached, basalt and feldspar-phyric basalt dykes in the rhyodacite are increasingly cut by acid veins from the rhyodacite. The dykes do not cut the gabbro, but protrude into it for a metre or more, suggesting the removal of some rhyodacite by thermal erosion. At the contact, thin (c.2 cm-wide) rhyolite veins are in sharp contact with angular fragments of mafic rock (mainly fine-grained dolerites) and very rare peridotite blocks (SR 318). The veins are similar to the rhyodacite except that their average groundmass grain size is greater (c.50 microns) and the modal proportion of phenocrysts is less than in the normal rhyodacite. Sieve texture is common in plagioclase phenocrysts in the altered rocks (cf. Reynolds, 1951). Within the gabbro the veins become less and less distinct, they take on a granophyric texture and ultimately lose their identity, merging into a hybrid rock with plagioclase phenocrysts (An30-20) derived from the rhyodacite and feldspars (An45–25) from the gabbro. Gabbro adjoining the indistinct veins shows a gradual increase in micropegmatite towards the hybrids.

Similar intrusion breccia is present on Cnapan Breaca [NM 3915 9765] and in western upper Dibidil where breccia on slabs [NM 3832 9452] north east of Forgotten Corrie contains numerous peridotite fragments (Greenwood 1987, pl. 2b, locality 215).

Contacts between the Layered Suite and the Western Granite

Exceptionally clear contacts between granophyre and harrisitic bytownite-gabbro occur at the west end of Harris Bay (near [NM 335 955] and on the rocky headland [NM 3405 9507] to the south-east. At the west end of the bay, bytownite-gabbro with flat-lying harrisitic layers becomes fine-grained about 2 m from the 1–2 m-wide zone of intrusion breccia which separates the layered rocks from microgranodiorite and quartz microdiorite hybrids (e.g. SR 261). The texturally variable acid hybrids are characterised by acicular amphibole or chlorite, zoned plagioclase (cores An30) and sieve-textured alkali feldspar in a matrix of delicate alkali feldspar–quartz inter-growths. The zone of hybrid rock is up to 15 m wide on the north side of a nearby cove [NM 3348 9564]. Between the Mausoleum and the cove, irregular, flat-lying sheets of acid hybrid intrude the gabbro and a complex zone of felsic box-veining and intrusive granitic pipes affects parts of the marginal gabbro (Figure 36); Greenwood, 1987, fig. 2.4). The contact between the Western Granite and acid hybrids near the Mausoleum (Figure 36) is sharp but neither rock is chilled. The granophyre is a dull blue-grey rock with small (c.3 mm) mossy areas of dark minerals, dull white feldspar and rusty-weathering vugs up to 2 mm in diameter.

One of the best-exposed intrusion breccia zones in the Hebridean Province crops out in low, wave-washed cliffs at the south-east end of Harris Bay [NM 3405 9505] (Plate 20). Dull grey, thermally metamorphosed granophyre is intruded by thin (<1 m-wide) basalt dykes which are progressively veined and finally broken up in the rheomorphic felsic matrix, merging into a zone of intrusion breccia that margins the mafic rocks. Feldspar phenocrysts in the altered granophyre range from normally zoned plagioclase with incipient sieve texture to vague alkali feldspar areas riddled with quartz. Fine-grained granular aggregates of amphibole, biotite and opaques have replaced original pyroxene and amphibole. At this locality there are a few instances where mafic inclusion have complicated chilled, rounded, crenellated margins against the acid matrix (see above).

Contacts between the Layered Suite and Torridon Group rocks

Fine-grained sandstones are in sharp, near-vertical contact with gabbro in a 50 m-high cliff [NM 3660 9174] near Loch Papadil. Near-horizontal bedding in the sandstone steepens to c.30° NW close to the gabbro before passing into intrusion breccia consisting of dolerite and (rarer) gabbro blocks in a fine-grained matrix of mobilised sandstone. Felsic veins in the adjoining gabbro are derived from partially melted sandstone. Similar recrystallised and partly melted sandstone (SR 532) is enclosed in gabbro west of Inbhir Ghil [NM 358 926]. It contains sutured quartz grains, sanidine and heavy mineral bands, the latter now marked by lines of aggregated magnetite crystals rimmed by fine-grained biotite.

At Bealach an Fhuarain [NM 379 980], sandstone and several basalt cone-sheets are intruded by gabbro. The mafic blocks in the intrusion breccia are derived from the break-up of both the cone-sheets and the margin of the gabbro (Hughes, 1960a, pl. X, fig. 2).

Between 250 and 600 m SSE of the ford at Allt na h-Uamha [NM 4092 9679] gabbro and bytownite-troctolite are in contact with shale and siltstone, and intrusion breccia occurs on the west side of a small hill [NM 4096 9645]. The sedimentary rocks are violently disturbed up to 30 m from the contact. Lamination and fine-scale bedding are bent and contorted, basalt dykes are cut by thin felsic veins as the contact is approached, eventually being broken into elongate segments separated by mobilised siltstone.

Contrasted behaviour of the different lithologies is seen 200 m WSW and SW and about 400 m SSE of the hydroelectic dam on Allt Slugain a'Choilich [NM 3931 9828]. Shale layers have undergone brittle brecciation whereas the sandy beds appear to have flowed around and between the shale fragments. This is not a primary sedimentary feature, where the shale would have behaved in an incompetent manner, but an example of the relative ease with which arkosic sandstone may be mobilised during thermal metamorphism. Silty and shaly rocks have been altered to tough siliceous hornfelses, which may be strongly iron stained. The coarser sandstones and arkosic sandstones change to compact, pale grey rocks. commonly with pustular surfaces where small (2–5 mm) spherulitic areas of intergrown quartz-alkali feldspar weather proud. Spherulitic quartz–alkali feldspar intergrowths and quartz paramorphs after tridymite are seen in thin section (cf. (Plate 21)a).

Contacts between the Layered Suite and Lewisian Gneisses

The gneisses commonly have a distinct thermal metamorphic overprint. Feldspar is a dull, chalky white colour and the mafic minerals are dull green or dark grey. Partial melting is common (Plate 22) and the granitic matrices of intrusion breccias have been derived from acid gneisses in Sandy Corrie [NM 3668 9424], on the north-west slopes of Ainshval [NM 3731 9459] and between Ard Mheall and Ard Nev [NM 3490 9797].

Alteration of the gneiss is manifest initially in the formation of iron oxides along the cleavage cracks and crystal margins of hornblende. Granular pyroxenes begins to form, at first on the margins of the yellow amphibole and in the cleavage traces, then by the complete replacement of amphibole by clino- and orthopyroxene and opaque oxides (Plate 21b). Plagioclase becomes turbid and eventually more calcic with increased metamorphic grade. Biotite, generally present in small amounts, survives to the highest grades, the colour changing from dull brown to a foxy red colour with increased thermal alteration. The compositions of co-existing pyroxenes in the high-grade hornfelses show that these rocks have been subjected to sustained, high temperatures (c.1000°C; A C Dunham, personal communication, 1972; Greenwood, 1987).

In the leucocratic gneisses (Type 3, Chapter 2), extreme thermal alteration causes the formation of rims of intergrown quartz and alkali feldspar between the original quartz and alkali feldspar grains. In the more extreme examples the rims form parallel-sided channels up to about 2 mm in width (Plate 21c). These changes probably represent the first stage in the production of rheomorphic acid melts. However, there cannot have been significant removal of melt from these altered gneisses since the original banding persists (cf. (Plate 2) and (Plate 22)).

At the Western Granite–Harris Bay Member (Western Layered Intrusion) contact, east Harris Bay. The elongate dark strips are dykes that have broken up in the remobilised (rheomorphic) acid matrix.

Contacts Between the Layered Suite and the Broadford Beds (Lower Jurassic)

Marginal gabbro of the Eastern Layered Intrusion intrudes and alters sandstones, shales and limestones of the Broadford Beds near Allt nam Bà' [NM 406 944]. The thermal metamorphism is overprinted on deformation imposed during movements on the Main Ring Fault (Chapter 10). Rare calc-silicate xenoliths also occur nearby, in the gabbro and lowest units in the Eastern Layered Intrusion (Figure 11) and (Figure 54); Hughes, 1960b, fig. 1; Smith, 1985, fig. 1).

Pure limestone is recrystallised to coarse, grey marble (DU 13868 [NM 4049 9402]). Sandstones are altered to tough quartzite. A magnetite-rich quartzite (SR 359) is probably the metamorphosed equivalent of ironstone present in Dibidil (Chapter 4). Impure, sandy limestones are altered to varied calc-silicate mineral assemblages. Hughes (1960b) records grossularite, diopside, vesuvianite, leucoxene and tilleyite (Ca3[Si2O7].2CaCO3); spurrite (2Ca2[SiO4].CaCO3) also occurs (SR 203) and harkerite (a complex Ca.Mg carbonateborosilicate; Tilley, 1951) has been tentatively identified (Smith, 1985). A wall-like outcrop within gabbro on the north bank of Allt nam Bà [NM 4060 9436] (Emeleus and Forster, 1979, fig. 5,II.lb; Smith, 1985, fig. 1) consists of pyroxene-rich, contaminated gabbro intermixed with irregular areas of talc-silicate rock, these are cut by thin syenitic veins (SR 204) consisting of perthitic alkali feldspar rimmed by albite, green aegirine-augite, and micaceous areas possibly after nepheline.

Inclusions of basalt of (?)Palaeocene age in The Layered Suite

Fine-grained, granular olivine-plagioclase-clinopyroxene-magnetite-(orthopyroxene) xenoliths are common in the lower bytownite-troctolite layers and sheets of the Eastern Layered Intrusion (Brown, 1956; Faithfull, 1986). They are altered basalts, termed beerbachites (cf. Richey and Thomas, 1930) which were probably derived from basalts of the Eigg Lava Formation in the roof zone of the Rum Central Complex. An analysed sample (Faithfull, 1986, appendix 2, 6.16) is a Mg-rich, aluminous basalt somewhat similar to the less evolved lavas on Eigg. Zoned areas up to 2 cm diameter of calc-silicate minerals in beerbachites have an outer rim of the Al-rich, Na-poor pyroxene fassiaite (Ca[Mg,Fe3+,Al] [(Si,Al)2O6]) and a core of garnet, diopside, less common vermiculite, and rare clinozoisite, low-iron epidote and plagioclase. Faithfull (1986, p.69) suggests that the clots were formed by the alteration of amygdales in basalts.

Oxygen isotope geochemistry of the Layered Suite

Studies of the 18O/16O ratios of rocks from the Palaeogene intrusive complexes on Skye, Mull and Ardnamurchan (Taylor, 1968; Taylor and Forester, 1971; Forester and Taylor, 1976; 1977) have demonstrated that large-scale, subsolidus O-isotope exchange occurred between the epizonal plutons and low-18O hydrothermal meteoric groundwaters which penetrated deeply into the volcanic cover and adjacent country rocks. This extensive water–rock interaction occurred when hydrothermal systems were set up in reponse to the intrusion of hot igneous bodies into permeable, water-saturated strata in the near surface environments. A similar situation has occurred on Rum (Forester and Hannon, 1983; Greenwood et al., 1992).

Considered together for all lithologies sampled on Rum, whole-rock δ18O values exhibit a large range, from — 6.1 to +10.7‰ (Appendix 6). Depletions in 18O are observed for both whole-rocks and mineral separates, with most low- 18O values occurring within the Rum Central Complex and its marginal zone rather than in the country rocks distant from the intrusion. The O-isotope data confirm that extensive subsolidus hydro-thermal interaction with low-18O fluids has affected all the intrusive units of the Central Complex, and the country rocks in the immediate vicinity of the contact zone. Within the Layered Suite, 18O/16O ratios are highly variable, ranging from δ18O values that are characteristic of intraplate basaltic magmas and their closed-system differentiates (Sheppard, 1986), to strongly 18O depleted rocks with negative δ18O values. For the Proterozoic Torridon Group rocks, O-isotope ratios tend to be high away from the igneous complex (i.e. up to 10.7‰), decreasing to low, positive values as the margin of the intrusion is approached ((Figure 37)b). For all lithologies possible the lowest 18O/16O ratios occur within 0.5 km of the contact zone ((Figure 34)a). Low δ18O values, from +4‰ to c.-6‰, also occur in the granophyres (Appendix 6; Greenwood et al., 1992, table 1, fig. 1) but the post-central complex lavas of the Canna Lava Formation have higher values, characteristic of fresh intraplate basalts (>+6‰).

The low18O values observed for gabbros, bytownite-troctolites (allivalites), and peridotites indicate that virtually the entire ultrabasic suite has been subjected to some degree of subsolidus interaction with meteoric waters at high temperature. Only two of the 41 samples analysed have δ18O values that are characteristic of pristine intraplate basalts, a peridotite from Unit 10 at +6.2%° and a gabbro from the top of Unit 8 at +6.1%°. All the other rocks sampled from the suite are depleted in 18O to some extent, with whole-rock δ18O values ranging from -4.8 to +4.7‰ . There is a tendency for the lowest 18O/16O ratios to occur near the margin of the complex, for example δ18O values of -4.8 and -0.4‰ for peridotites at Harris Bay, and near 0‰ in the Cnapan Breaca area. As pointed out by Greenwood et al. (1992), mafic rocks of the contact zone are disrupted by numerous fractures and intrusive breccias, and all rocks of the contact zone have been subjected to intense hydrothermal alteration at relatively low temperatures, such that feldspars are strongly sericitised and primary ferromagnesian minerals have been converted to amphibole or chlorite. Together, these features indicate that the contact zone was a local region of high permeability that focused hydrothermal circulation during cooling of the Central Complex and, therefore, maximised water-rock interaction and O-isotope exchange.

By contrast, fluid-rock interaction within the interior portions of the Layered Suite has not altered the primary mineralogy or texture of the rocks, indicating that the exchange processes occurred at high temperatures above the field of serpentine stability (i.e. >450°C), as advocated for the Skaergaard Intrusion in East Greenland by Taylor and Forester (1979). This is clearly seen in the preservation of the anhydrous mineralogy of the rocks, despite the extensive 18O depletions in feldspar (δ18O -8.5 to +5.9‰ ) and the negative feldspar–pyroxene 18O fractionations that occur throughout the Eastern Layered Intrusion (Figure 38).

The rate of subsolidus O-isotope exchange of plagioclase with high-temperature hydrothermal fluids is much greater than the rate for pyroxene or quartz (Cole and Ohmoto, 1986), consequently, some of the 180 variation observed in the Layered Ultrabasic Suite could be a reflection of differences in modal mineralogy. Additionally, recrystallisation of the bytownite-troctolite units late in the cooling history of the complex could have occurred in the presence of (or been facilitated by) circulating meteoric hydrothermal fluids. Thus, plagioclase-rich members of the layered suite might be expected to have acquired lower δ18O values through fluid–rock exchange than the more ferromagnesian-rich rocks. For example, the 2‰ difference in Unit 8 between the lower peridotite (δ18O = +4‰ ) and the immediately overlying bytownite-troctolite (δ18O = +1.8‰ ) may be a manifestation of the different plagioclase contents of the two rocks (Greenwood et al., 1992).

However, such an explanation can account for only a small proportion of the overall δ18O variation observed in the Layered Suite. Other significant factors, including the extent of fluid–rock exchange or changes in either the O-isotopic composition of the hydrothermal fluids or temperature of reaction, are likely to be responsible for the large range of O-isotope ratios observed for all the lithological types in the suite. The fact that large differences in O-isotope compositions are observed on a small geographical scale, as within individual units of the Eastern Layered Intrusion, implies that the major factor governing the range of O-isotope compositions observed within the Layered Suite was variations in rock/water ratio and not fluid isotopic composition.

Chapter 8 Palaeogene 4: Lavas and associated sedimentary rocks

Introduction

Predominantly basaltic lavas with sparse interbedded sedimentary rocks comprise all the solid outcrop on Canna and Sanday. They form most of Muck and south Eigg and crop out in north-west Rum and on the southeastern side of the Rum Central Complex (Figure 39). Submarine basaltic ridges link Canna and Sanday with south-west Skye, and Muck and Eigg with Ardnamurchan and Mull (Binns et al., 1974, fig. 2). A later formation of acid lavas, the Sgurr of Eigg Pitchstone Formation, is of Eocene age.

The Palaeocene lavas and interbedded sedimentary rocks belong to two formations. The older Eigg Lava Formation includes the flows on Muck, Eigg and in southeast Rum. This is intruded by a NW-trending dyke swarm which in turn is cut by the Rum Layered Suite as are faults which limit the lava outcrops in south-east Rum. The younger Canna Lava Formation consists of the lavas and interbedded fluviatile conglomerates of north-west Rum, Canna and Sanday. On Rum the lavas rest on a surface eroded from the Western Granite and Torridonian rocks. The conglomerates contain clasts of gneiss, sandstone and rocks originating from the Rum Central Complex, including granophyre, rhyodacite, and bytownite-troctolite from the Layered Suite.

Although the lavas of Eigg and Muck appear linked to those of Mull and Ardnamurchan (Fyfe et al., 1993, fig. 40), their relative ages are not known. laser 40Ar/39Ar age determinations place the Muck lavas at 62.5 Ma earlier than those of Mull (Pearson et al., 1996). The Canna Lava Formation may be of similar age to the Skye Main Lava Series in south-west Skye (Williamson, 1979; Meighan et al., 1982; Williamson and Bell, 1994), it is overlain by sedimentary rocks of Oligocene age in the Canna Basin (Smythe and Kenolty, 1975; Fyfe et al., 1993).

The sedimentary rocks and lavas were first described in detail by Harker (1908, chapters V and VI). Numerous sills of dolerite, basalt and mugearite were mapped alternating with the lavas (Harker, 1908, chapter X and fig. 48). It is now recognised the sills were misidentified; for the most part they are the massive, central parts of lava flows (cf. Anderson and Dunham, 1966). Apart from a petrographical, mineralogical and chemical examination of selected flows by Ridley (1971; 1973; 1977) and an account of the lavas in north-west Rum (Emeleus, 1985), the only other detailed study of the lavas of the Small Isles is that by Allwright (1980), who has made her samples, thin sections and unpublished work freely available during the compilation of this memoir.

Eigg Lava Formation

Details

Eigg

From Clach Alasdair [NM 453 884] eastwards around northern Eigg, to the vicinity of Kildonnan [NM 490 851], lavas and rare sedimentary rocks rest unconformably on rocks of the Cretaceous Strathaird Limestone Formation and the Jurassic Great Estuarine Group. The unconformity is largely concealed by landslips and drift (Plate 23). The base of the succession is generally marked by thin tuff deposits which are a few centimetres thick at Dalian Thalasgair [NM 4802 9070] and Bealach Clith [NM 4922 8580] but several metres thick in the Allt Ceann a'Gharaid [NM 4885 9047]. Lavas overlie Cretaceous rocks at Clach Alasdair. A vertical section through the Eigg lavas is given in (Figure 40) where the approximate positions of the divisions recognised by Allwright are indicated. The succession consists mainly of various basalts but there are several prominent flows of mugearite, especially about 50 m above the base of the Formation. The distinctive feldspar-phyric flows of the Cora-bheinn Member include several of basaltic hawaiite composition.

The lavas flows are commonly separated by thin beds of red clay ('bole') and show strong trap featuring. They form a succession of gently south-westward dipping (c.3°) flows, about 400 m thick. Individual flows are commonly between 5 and 10 m in thickness but may be over 20 in thick, as in a massive columnar olivine-basalt that forms Druim an Aoinidh [NM 486 851] and the base of the cliffs north from Kildonnan (Plate 23); at the latter locality, the columnar jointing is locally highly irregular rather than in the usual vertical attitude. Individual flows may be continuously exposed over several kilometres. The most conspicuous example of lateral persistence is provided by the grey-weathering mugearite flows near the base of the lava succession (Plate 23). Two or more flows totalling 30 m thickness but generally about 10–15 m thick crop out, when not obscured by landslips, from the coast 500 m west of Clach Alasdair, around northern Eigg, then south to the old (Clanranald) pier [NM 4832 8415]. This distinctive marker links the lava successions of northern and south-western Eigg, which had proved difficult to correlate in detail (Allwright, 1980). Amygdaloidal flow tops are common and thin flows, as in the cliffs east of Uamh Fhraing [NM 4750 8346], may be amygdaloidal throughout. Arnygdale minerals listed by Harker (1908, pp.58–59) include calcite, quartz, analcime, chabazite, stilbite and mesolite, with less common thomsonite, gyrolite and apophyllite.

Allwright (1980, p.169) described the sedimentary rocks interbedded with the Eigg lavas as friable, fine-grained bright red or orange-red mudstones, some of which are probably of tuffaceous origin. Bedding is rare and she noted that these red beds are less common than on Muck. An impersistent pebbly conglomerate interbedded with reddened, brecciated mugearite occurs on the foreshore at [NM 4825 8417], west of the old pier.

Muck

The Palaeocene lavas and basal sedimentary rocks overlie limestones of the Jurassic Kilmaluag Limestone Formation at Camas Mór [NM 407 793] (Plate 12) where the contact is complicated by faulting and largely obscured by superficial-deposits. The lavas and sparse sedimentary rocks form a gently NNW-dipping (<3°) succession about 140 in in thickness. They are cut by the NNW-trending Muck dyke swarm and by numerous small NW-to NNW-trending normal faults. The sequence is summarised in (Figure 40).

The lowest lavas on Muck are impersistent flows of feldspar-phyric basaltic hawaiite, basalt and mugearite which form the Basal Member (Allwright, 1980). Flow fronts occur at Camas Mór where the basal basaltic hawaiite terminates abruptly to the west, 100 m south-east of Torr nam Fitheach [NM 408 794], and at Port an t-Seilich [NM 419 785] where a mugearite flow ends suddenly to the east. The Basal Member is overlain by four flows of strongly feldspar-phyric basaltic hawaiite with labradorite phenocrysts up to 2 cm in length. These flows constitute the Port an t-Seilich Member and they crop out intermittently all along the south coast of Muck. They are succeeded by the Beinn Airein–An Stac member which consists largely of olivine-basalt except for a feldspar-phyric basaltic flow on Beinn Airein [NM 403 792] and an impersistent feldspar-phyric olivine-basalt on Eilean nan Each [NM 393 894]. The member occupies most of Muck, and gives rise to the strong trap topography of the island.

Bright red and orange-red mudstones are common in the lavas in the cliffs of Beinn Airein and in the shore sections at [NM 404 788], east of Sgorr nan Laogh. Allwright (1980, p.169, fig. 8.3) recorded a possible basaltic bomb in red tuffaceous mudstone at Fang Mór [NM 4085 7927]. Some mudstones grade into reddened flow tops and probably originated by in-situ lateritic weathering, but many form discrete, bedded layers that suggests they are water-lain tuffs (Emeleus et al., 1996a). The Main Red Bed in the cliff 100 m north-east of Sgorr nan Laogh is one such example and another occurs east of Port Mór [NM 422 793] where a 20–30 cm-thick red bed overlies the reddened, brecciated pahoehoe top of a basalt flow exposed at the top of the wave-cut platform (Plate 24) [NM 4260 7901]. The red bed consists of euhedral crystals and angular fragments of fresh sanidine (c.K38Na59Ca3 with rare K90Na10Ca0), sodic clinopyroxene (Mg60(Fe″+Mn) Na to Mg45 (Fe″ +Mn32 Na23) biotite and sphene, and a fine-grained, felsic rock with attached fragments of these crystals, set in a very fine-grained, deep red, ferruginous matrix (SR 651). On Eigg, a somewhat similar reddened tuff (HE 7648) containing sanidine and pale green (?sodic) clinopyroxene fragments occurs in a disused quarry [NM 4818 8525] near the manse. The crystals and lithic clasts suggest that there was a nearby, explosive eruption of peralkaline trachytic magma at an early stage in the Eigg Lava Formation; however no comparable flow is found amongst the lavas on Muck or Eigg. Peralkaline intrusions and trachytic lavas occur elsewhere in the Hebridean Province but the writer is unaware of any record of peralkaline material from Palaeocene tuffs, except for those in the North Sea and northern Jutland (Knox and Morton, 1988). Analyses of the feldspars from Jutland are similar to those from Muck but the alkali pyroxenes are significantly richer in Fe, Na and Ti (Pedersen et al., 1975).

Camas Mor Breccia (Z, (Figure 41)

A coarse sedimentary breccia crops out beneath the boulders on the shoreface at the east end of Camas Mór [NM 410 793]. The exposures occur over a distance of about 200 m, to the west of a prominent cliff of feldspar-phyric basaltic hawaiite [NM 4109 7920]. The breccias are intruded by the northern tip of the Camas Mór gabbro dyke and by several NNW-trending dolerite dykes.

The character of the breccia changes from west to east. The most westerly exposures contain a black, carbonaceous siltstone in the breccia matrix. The clasts consist predominantly of blocks of a grey, marly limestone, limestone with close packed bivalves (from the Duntulm Formation) and a brown argillaceous limestone with bivalves. The clasts are commonly 20 cm diameter but may be up to 30 cm. Eastward, the matrix contains progressively less siltstone and becomes cream, pale grey or orange coloured. In places there is distinct carious weathering. The enclosed blocks are as before but there are many more centimetre-sized fragments which appear slightly sorted and bedded, with a north-easterly dip (10–20°). Small scarps of well-bedded, coarse-grained pebbly rock (SR 583B; [NM 4102 7925]) occur and their bedding is deformed where they underlie large blocks of limestone from the Duntulm and Kilmaluag Formations. Bedded pebbly sandstones near the end of the Camas Mór gabbro dyke pass up into aluminous siltstones, overlain by feldspar-phyric basaltic hawaiite to the east. No igneous material has been found in the breccia and the adjoining lava flows are undisturbed. Both the lavas and the gritty rocks are baked by the gabbro and the sedimentary rocks are folded close to the dyke (Figure 41). Breccia that truncates bedding in the gritty sandstone is in turn cut by the gabbro.

The breccias are of uncertain age. They postdate the Kilmaluag Formation but appear to underlie the basal Palaeocene lavas. They crop out close to reddish brown basaltic tuffs (ZB, (Figure 41) but the contact between the breccias was not exposed. The Camas Mór Breccia could have formed during phreatic explosions induced by the gabbro as it intruded wet, incompletely lithified Jurassic rocks (cf. England et al., 1993). This interpretation is perhaps supported by the relicts of a strati-graphical succession in the breccia near the eastern end of the bay. However, had this occurred it is difficult to see how the lavas remained intact and, furthermore, complicated contact relationships would be expected between the gabbro dyke and the sedimentary rocks (cf. Kokelaar, 1982). On balance the evidence indicates that the breccias predate the basal lavas on Muck.

Rum

A narrow strip of metamorphosed basalt occurs in the Main Ring Fault and extends for 1 km at about 250 m above OD on the south-eastern slopes of Beinn nan Stac (Figure 54); Smith, 1985, fig. 1). Within this faulted sliver, the basalts overlie the Lower Jurassic Broadford Beds with a landscape unconformity (Smith, 1985). Altered basalt also crops out on the northeastern side of Dibidil at [NM 399 932].

The basalts are tough, splintery dark grey rocks. Scattered plagioclase microphenocrysts may be present and the rocks are commonly amygdaloidal. It is difficult to distinguish individual flows and no inter-lava red beds have been identified. The lavas have been slightly metamorphosed by the nearby gabbros.

Thin sections of the basalts (SR 198, 202) show a variety of rocks. Most common are feldspar-phyric basalts with labradorite phenocrysts in a matrix of feldspar laths, opaque granules and alteration products after original pyroxene. Ophitic olivine-basalts occur in which olivine has been replaced by green chloritic material but labradorite and augite are generally recognisable. Epidote, albitic plagioclase and rare quartz fill the amygdales. The finer-grained basalts are generally crowded with opaque granules which largely obscure the other minerals. All of the basalts show signs of fracturing and small-scale dislocations commonly off-set feldspar phenocrysts. The deformed rocks resemble mylonites locally (SR 204) and all show signs of crushing.

Canna Lava Formation

The formation differs from the Eigg Lava Formation in several important respects: the lavas include a considerable proportion of evolved types; limited amounts of tholeiitic basalt or evolved tholeiitic types occur in addition to mildly alkaline to transitional flows; and there are many beds of coarse fluviatile conglomerate which, on Rum, reflect deposition on a hilly landscape undergoing vigorous erosion.

Canna and Sanday

On Canna and Sanday, the succession of lavas and fluviatile deposits is over 200 m thick (Figure 40). It is difficult to establish an accurate thickness since faults divide the succession into a number of blocks (Chapter 10, (Figure 56) and individual flows, groups of flows and the conglomerates show abrupt variations in thickness and continuity.

The oldest flows occur in south-east Sanday and the youngest on the high ground of east and west Canna (Figure 56). The flows dip at low angles (<3°). A generalised succession is given in (Figure 40). The lower part of the succession, on Sanday, consists of aphyric basalt, basalt with sparse plagioclase phenocrysts and feldsparphyric basalts with small plagioclase crystals (<5 mm in length). The flows are generally less than 7 m thick and interbedded coarse fluviatile conglomerates are common. The highest flow on Sanday, at Creag Ard [NG 263 045] and either side of An Doirlinn [NG 266 049], contains plagioclase, olivine and possibly augite in glomeroporphyritic aggregates.

Fine-grained red beds (boles) are very rare but brecciated, reddened as flow tops are quite common. An excellent example of breccia is seen on the wave-cut platform between Losaid Mhór [NG 253 053] and Geodha na Nighinn Duibhe [NG 246 055]. The lavas accumulated largely subaerially with pillow lavas and hyaloclastites forming when flows entered small lakes.

Aphyric and sparsely feldspar-phyric basalts interbedded with the thick conglomerates and sedimentary breccias form the Tighard and Cnoc Bhrostan members on east Canna. At Compass Hill the sedimentary rocks total 50 m in thickness but about 1.5 km west, on the northern coast, they have virtually disappeared and they also die out towards the west end of Canna Harbour. Pillow breccias and hyaloclastites crop out at Cuil a Bhainne [NG 2625 0528]. Columnar, aphyric basalt flows up to 20 m thick form high cliffs north of the road but thin rapidly to the east at Ealaist [NG 258 054]. A thin mugearite flow occurs amongst the breccias 300 m ESE of An t-Each [NG 2775 0652].

The Carn a'Ghaill and Beul Lama Sgorr members form the high ground on both west and east Canna. They consist generally of thick (8–15 m) flows of basaltic hawaiite and hawaiite, both are feldspar-phyric with plagioclase crystals up to 3 cm long. A 10 m-thick flow of aphyric hawaiite (SR 252) with a distinctive platy fracture provides a marker horizon from Beul Lama Sgorr [NG 268 062] west to Buidhe Sgorr [NG 247 065]. This flow also forms the small outcrop on the flat topped hill 400 m south-west of Cnoc Rùgail [NG 230 056] in western Canna.

A columnar flow of tholeiitic basaltic andesite (SR 254) overlies coarse fluviatile conglomerate on Eilean a'Bhàird [NG 270 050] (Figure 42) and also occurs on the north shore of Canna Harbour [NG 2690 0517]. This flow and the underlying sedimentary rocks form the Eilean a'Bhàird Member (Figure 40). The flow is compositionally distinct from the other lavas on Canna and Sanday and closely resembles basaltic andesites of the Upper Fionchra Member on Rum. It is probably later than the main Sanday–Canna succession since the underlying conglomerate contains boulders of feldspar-phyric basaltic hawaiite similar to flows in the Beul Lama Sgorr Member.

Sedimentary rocks interbedded with the lava flows

Canna and Sanday have arguably the best development of sedimentary rocks interbedded with lavas in the Hebridean Province. They were described by Geikie (1896, 1897) and the succession on Compass Hill was examined in detail by Harker (1908, chapter V, see also fig. 8 reproduced here, with modifications, as (Figure 43). Other classic occurrences include the stacks of Dim Mór [NG 287 037] and Dun Beag on Sanday (Harker 1908, fig. 12).

The lowest bed at Compass Hill has been described as agglomerate. It consists of about 30 m thickness of breccia with blocks of basalt up to 2 m diameter and less common, smaller cobbles and pebbles of red sandstone similar to rocks in the Torridon Group on Rum. The smaller clasts are generally well rounded and the deposit is here regarded as a sedimentary breccia, probably a debris flow, rather than true agglomerate. It passes up into about 14 m of well-bedded fluviatile conglomerate. On Dun Beag [NG 2882 0375] the shape of the conglomerate outcrop suggests that it is part of the wall of a steep-sided gorge that was filled by a basalt flow which spilled over the top of the gorge. Coarse-grained tuffaceous sandstone on the east side of the promontory at [NG 2674 0404], 250 m due east of Tallabric, is intruded by a 2 m-long tongue of the overlying basalt flow. The irregular, lobate edges of the tongue suggest that the sandstone was wet and poorly consolidated when covered by the flow. Scoriaceous basalt below this sandstone is interbedded with thin lenses and pockets of coarse gritty sandstone, siltstone and conglomerate. South of Tallabric [NG 265 041] the same beds contain poorly preserved carbonised plants. Plant remains are also found in sandstone at Creag a' Chairn [NG 2423 0563] (Harker, 1908, fig. 13), behind Canna Post Office [NG 2720 0530] (Harker, 1908, fig. 11) and at the top of a thick bed of conglomerate, gritty sandstone and siltstone at the east end of Camas Thairbearnais [NG 2405 0655] (Plate 25).

The clasts in the fluviatile conglomerates are principally boulders, cobbles and pebbles of rocks similar to those of the local lava succession. In addition clasts of feldspathic sandstone resembling Torridon Group rocks on Rum are common, particularly in the Compass Hill conglomerates and breccias. Harker (1908, p.40) also found clasts of schists and gneiss in the highest conglomerates on Compass Hill. Granophyre pebbles (SR 253) have been recovered from conglomerates on Sanday and porphyritic rhyodacite (SR 248) from the Compass Hill rocks (Emeleus, 1973). These are identical to rocks present in the Rum Central Complex. Similar clasts of granophyre have been found in interlava conglomerates near Glen Brittle and elsewhere on Skye (Williamson and Bell, 1994). No fragments of Mesozoic rocks have been identified in the conglomerates of Canna or Sanday despite the Little Minch Basin extending beneath the islands (Binns et al., 1974, fig. 2) and the presence of boulders of Mesozoic rocks in the shoreface deposits on the south side of Canna Harbour.

Rum

In north-west Rum the Canna Lava Formation consists of four distinct members (Figure 40). The lavas and conglomerates occur on Fionchra [NG 339 004], Bloodstone Hill [NG 315 006], the eastern end of Orval [NM 334 991] and the small 314 m hill [NG 349 003] known as West Minishal (Figure 44).

The conglomerates and flows have irregular, discontinuous boundaries which were interpreted as faults (Sheet 60, 1st edition, 1903), although Harker (1908, p.50) mentions the possibility that the conglomerates and flows were accumulated on an irregular surface. Subsequent examination has shown that the Canna Lava Formation on Rum formed during repeated episodes of vigorous erosion, accumulation of fluviatile conglomerate and lava effusion (Emeleus, 1985). Most of the boundaries of the formation are unfaulted.

Lower Fionchra Member

The base of the member consists of up to 50 m of coarse fluviatile conglomerates. Gritty sandstones in the lowest exposures at Maternity Hollow [NG 3473 0063] contain thin, carbonised logs up to 1 m in length. Individual flows are about 7 to 10 m in thickness but this is difficult to estimate as exposure is poor. The massive, columnar flow centres form the edges of peat-covered shelves. The lower lavas are olivine-basalts with microphenocrysts of plagioclase. Higher flows, in the steep valley about 200 m north-east and east of Orval, are basaltic hawaiites. These flows lie on an eroded surface of weathered granophyre (Black, 1952b).

The lavas and conglomerates of the Lower Fionchra Member fill former valleys in an early Tertiary landsurface eroded in Torridon Group sandstones and members of the Rum Central Complex. East of Fionchra the conglomerates were deposited in the bottom of a V-shaped valley which was further filled by a succession of lava flows. The valley is cut off to the east by the NNW-trending fault at Maternity Hollow. A former valley wall is seen in knolls about 600 m north-east of Fionchra summit (Figure 44).

Upper Fionchra Member

This forms Bloodstone Hill and the ridge of Fionchra (Figure 44). At the base of the member, a succession (0 to 30 m thick) of conglomerates and tuffaceous sandstones with thin plant-bearing silty horizons crops out in stream sections on the north side of Fionchra. The plant remains include delicate leaf and stem impressions (see also Tomkeieff and Blackburn, 1942). These deposits thin rapidly southwards in Coire na Loigh [NG 331 010] and only a small outcrop is present in upper Guirdil, about 250 m west of Fionchra summit. A low conglomerate hill 300 m ENE of the summit provides a cross-section of valley-fill material. Up to 60 m of hyaloclastite breccia with scattered pillow structures overlies the sedimentary rocks. At Coire na Loigh, the breccias pass downward into several metres of strongly columnar-jointed lava which rests with a thin (1–4 mm) tachylitic selvedge on conglomerate. The structure resembles those in hyaloclastite sheet flows described from Iceland (Bergh and Sigvaldason, 1991). The breccia is succeeded by up to 80 m of tholeiitic basaltic andesite flows. A feldspar-phyric basaltic andesite occurs low in the succession on the south side of Fionchra and similar rock forms the basal flows on Bloodstone Hill. On Bloodstone Hill, conglomerate is limited to a pocket at the base of the flows, 400 m south-west of the summit, elsewhere the lavas rest directly on weathered Torridon Group sandstone, as for example at the north end of the hill at [NG 3157 0083]. Small banded agates occur in geodes and green, cryptocrystalline silica flecked with red specks (bloodstone) fills irregular thin fissures on Bloodstone Hill and at Coire na Loigh. Both occur abundantly as pebbles on the beach at Guirdil [NG 318 014].

Guirdil Member

Two flows of icelandite, each overlying conglomerate, fill a former, eastward-draining valley eroded in lavas and Torridonian sandstone on Bloodstone Hill (Figure 44). A 20 m-thick flow of icelandite and 1–2 m of conglomerate form a prominent cliff on the south side of Fionchra. This is part of a flow filling a steep-sided valley. Part of the same flow forms the cliff 100 m north of Bealach a' Bhraigh Bhig [NG 340 000] where the icelandite rests on a rounded, bouldery surface of granophyre.

Orval Member

The eastern face of Orval consists of at least three thick flows of basaltic hawaiite and feldspar-phyric basaltic hawaiite. The flows are difficult to separate since no intervening boles are present. They lie directly on granophyre except in the steep valley north-east of Orval summit where they abut against a former valley wall eroded from the uppermost lavas of the Lower Fionchra Member. Harker (1908) considered that the lavas were thermally altered by the granophyre; however, this has not been confirmed and the field evidence is unequivocal that all the lavas in north-west Rum postdate the Western Granite (Black, 1952b; Emeleus, 1985).

Clasts in the rum fluviatile conglomerates

Three groups of clasts are recognised in the Rum fluviatile conglomerates: Precambrian country rocks; rocks that do not match any types now found on Rum; and rocks from the Rum Central Complex and the lavas of north-west Rum (Black, 1952a; Emeleus, 1985, table 2).

Clasts of Torridon Group arkosic sandstones and banded Lewisian gneisses are common. None of the sandstone clasts show signs of thermal metamorphism and only slight alteration is found in a few of the gneiss fragments; for the most part they are fresh hornblende-gneisses. Quartz-dolerite and vesicular tholeiitic basalt fragments are common; the basalts indicate the former presence of unevolved tholeiitic flows on or near Rum which are not preserved. In addition to tholeiitic basalt, conglomerates high in the sequence contain olivine-basalts and feldspar-phyric basaltic andesites eroded from earlier members of the formation. Of the rocks derived from the Central Complex, clasts of drusy granophyre and porphyritic rhyodacite are the most common and may be large; boulders of rhyodacite up to 1.5 m diameter are present in conglomerate beneath the basaltic andesite flows on Bloodstone Hill, near [NG 317 006].

Bytownite-troctolite fragments are also found but gabbro is rather rare and peridotite is absent. The lack of peridotite is probably a reflection of the instability of the rock type in a high-energy fluviatile environment. Significantly, peridotite is also virtually absent from the boulders in the Pleistocene raised beach deposits. The occurrence of a wide range of igneous rock types that match rocks now exposed on Rum clearly demonstrates that the central complex underwent deep erosion during the Palaeocene.

Petrography of the Palaeocene Lavas

Eigg Lava Formation

The commonest lavas on Eigg and Muck are olivine-basalts in which olivine phenocrysts vary from under 1 mm to as much as 3 mm in length. Ophitic or subophitic, purple-brown augite crystals enclose laths of labradorite. Titanomagnetite is accompanied by small amounts of acicular ilmenite. The highest part of the lava succession on Eigg, the Cora-bheinn Member (Figure 40), includes several feldspar-phyric flows around the base of the Sgurr. Similar rocks occur low in the Muck succession. In these rocks the abundant labradorite laths, commonly 1 cm long, are accompanied by olivine and augite microphenocrysts. Orthopyroxene mantled by augite occurs in a feldspar-phyric basaltic hawaiite flow (SR 561) 150 m north-west of Guailainn na Sgurra [NM 4675 8470]. Two impersistent flows of strongly feldspar-phyric basalt occur low in the Eigg lava succession 300 and 500 m south-west of Laig farm [NM 4672 8768]. The mugearites are fine-grained, aphyric rocks with laths of oligoclase-andesine in a matrix of alkali feldspar, brown biotite, abundant granules of opaque oxides, clinopyroxene and apatite.

Canna Lava Formation

Canna and Sanday

The lowest lavas on Sanday are generally microporphyritic olivine-basalts. They contain labradorite laths (up to 1 mm in length) and olivine in a matrix of ophitic augite, plagioclase and anhedral opaque oxides. The flows of feldspar-phyric basalt are similar except that they contain euhedral to subrounded labradorite phenocrysts up to 1 cm in length. The glomeroporphyritic basalts in north-west Sanday have coarse aggregates of plagioclase (labradorite-bytownite) laths and olivine with iron-rich rims. Intergrown augite appears to be part of the matrix. Rounded 'cognate' inclusions of coarsely olivine-phyric basalt occur in flows on the north shore of Sanday and 150 m NNW of Dim Beag.

The flows in the Tighard and Cnoc Bhrostan members consist of ophitic olivine-basalts. Scattered olivine microphenocrysts are present but the numerous small plagioclase laths that are characteristic of the lavas on Sanday are rare. Extremely fine-grained olivine-basalts form a number of isolated knolls on the raised beach north of The Square [NG 2702 0520]. A coarse-grained olivine-basalt close to the top of the Cnoc Bhrostan Member on Cnoc na Carraig [NG 2763 0628] may be traced for over 2 km to the west. A fine-grained, columnar basalt flow 160 m north of Tighard house [NG 2730 0556] contains elongate ocelli, up to 2 cm in length. These consist of inter-growths of acicular, forked, deep purple titanaugite about 2 mm in length, andesine-labradorite and a meshwork of slender opaque oxide crystals.

The lavas in both the Cam a' Ghaill and Beul Lama Sgorr members are more evolved than flows in the lower members. The platy-jointed hawaiite that forms a marker flow and overlying feldspar-phyric hawaiites in the Beul Lama Sgorr Member contain abundant apatite needles, small amounts of biotite fringing opaque oxides, interstitial alkali feldspar and phenocrysts of olivine with thin, iron-rich rims.

Rum

Details of the petrography and mineralogy of the Canna Lava Formation on Rum have been described, tabulated and figured by Ridley (1971; 1973, 1977) and Emeleus (1985, pp.426–428, tables 4a and 4b, figs. 5, 6 and 7).

On Rum, the lower flows of the Lower Fionchra Member are olivine-basalts. They contain zoned microphenocrysts of bytownite, olivine, corroded chrome spinel, rare green aluminous spinel (SR 217) and rare augite (SR 156). The olivine phenocrysts enclose minute chrome spinel crystals. The matrices consist of olivine, granular augite, flow-aligned labradorite laths and titanomagnetite. The hawaiite flows higher in the member are texturally similar but the augite is subophitic, anorthoclase borders plagioclase laths and interstitial areas of deep purple-brown glass contain apatite needles and skeletal aggregates of augite and opaque oxides.

In the Upper Fionchra Member, the tholeiitic basaltic andesite flows are generally microporphyritic. Phenocrysts of labradorite, augite and rare hypersthene lie in a matrix of small plagioclase laths, granular augite, titanomagnetite and interstitial glass which is commonly altered. The fresh brown glass, which contains quench crystals of opaque oxides, has the composition of a soda-granite (Appendix 5e:i). Feldspar-phyric flows contain large zoned labradorite crystals and olivine phenocrysts. Clear brown glass (SR 239) from the hyaloclastites contains tiny olivine crystals in addition to plagioclase and augite.

In the Guirdil Member, the microporphyritic icelandite flows contain phenocrysts of andesine–labradorite (and rare bytownite — Ridley, 1973), augite, hypersthene, apatite, ilmenite and titanomagnetite with exsolved ilmenite. The matrix is extremely fine grained. Agate and chalcedonic silica fill geodes.

In the overlying Orval Member, the olivine-basalts consist of granular, interlocking olivine, plagioclase, augite and opaque oxides. In the basaltic hawaiites and hawaiites the pyroxene is subophitic or ophitic and the plagioclase is zoned to rims of potassic oligoclase or alkali feldspar. Olivine phenocrysts are strongly zoned to iron-rich rims and the matrix olivine is also iron-rich (Fo47). Biotite fringes opaque oxides, apatite needles are common and rare yellow-brown amphibole is present.

Chemical composition of the Palaeocene lavas

Harker (1908) published four analyses of lavas and additional analytical data were given by Ridley (1971; 1973), Allwright (1980) and Emeleus (1985). These data, largely for major elements, have been supplemented by new analyses on selected suites of lavas from all the islands. Isotopic data for the lavas are limited to two flows on Eigg (Carter et al., 1978, table 1, E.7 and E.10; all other Small Isles basalts quoted in this account come from dykes — cf. Ridley, 1973, tables 2 and 4). A selection of analyses is given in Appendices 5e:i and ii; 5f.

Using conventional plots of normative minerals and of total alkalies against silica (Figure 45) and (Figure 46), there is no obvious discrimination between flows of the Eigg Lava Formation and the Canna Lava Formation, except that on Canna and Sanday more of the flows are nepheline-normative. With the exception of flows in the Guirdil and Upper Fionchra members, the majority of the analysed lavas in these formations are similar to the plateau basalts and hawaiites of Skye and Mull (Thompson et al., 1972; Kerr, 1993; 1995).

The Eigg Lava Formation (Appendix 51) consists of mildly alkaline to transitional olivine-basalts, rare basaltic hawaiites and hawaiites, and mugearites. The lava sequences show no systematic evolutionary trends; mugearites occur low in the successions on Eigg and on Muck where they are accompanied by feldspar-phyric basaltic hawaiites. On Eigg, a suggestion of more-evolved magmas up sequence is provided by a mugearite and several flows of feldspar-phyric basaltic hawaiite and hawaiite high in the succession.

The Canna Lava Formation is the more variable. On Sanday and Canna (Appendix 5e:ii) there is a clear progression from olivine-basalts to more evolved basaltic hawaiite and hawaiite up the succession. The youngest flow, the basaltic andesite (SR 254) of the Eilean a'Bhàird Member (Figure 40), is compositionally comparable with flows in the Upper Fionchra Member on Rum. The earliest record of flows in north-west Rum (Appendix 5e:i) comes from the unevolved tholeiitic basalts found as clasts in conglomerates at the base of the Lower Fionchra Member. The oldest in-situ flows are mildly alkaline to transitional olivine-basalts which are followed up sequence within the Lower Fionchra Member by the hawaiites and basaltic hawaiites. These are succeeded by basaltic andesite and icelandite flows of the Upper Fionchra and Guirdil members. The youngest flows are the basaltic hawaiites and hawaiites of the Orval Member. The several compositionally distinct members identified on Rum are separated from each other by intervals of pronounced erosion. With the exception of the Lower Fionchra Member, the flows within each member show little compositional variation. There is, however, an overall progression to more evolved members with time, but no progression from alkalic to tholeiitic types, or vice versa; the alkaline to transitional types bracket two periods when evolved, tholeiitic flows were erupted, and the earliest alkali basalts were probably preceeded by tholeiitic flows. Since the evidence from Rum indicates that the compositionally varied groups of flows were erupted during the rapid erosion of the central complex and its surroundings, the varied magmas represented by these flows are thought to have been available within a short period of time.

From studies on the Palaeocene lavas of Skye and Mull, Thompson et al. (1982) have demonstrated that rock/ chondrite normalised multi-element profiles ('spider-grams') could be used to fingerprint basaltic lavas. Using lavas for which there were good isotopic data, which had been used to determine the extent to which the magmas had been contaminated by partial melts derived from Lewisian granulite- or amphibolite-facies crustal sources, they showed that uncontaminated basalts, basalt contaminated with predominantly granulite-facies Archaean crust and basalts contaminated by predominantly amphibolite-facies Archaean crust each gave distinct and characteristic profiles (Thompson et al., 1982, fig. 2, A, B and C respectively). The rock/chondrite normalised profiles for trace element and other data for the Small Isles lavas are presented in (Figure 47).

The profiles for the Small Isles lavas show a considerable variety, reflecting the compositional range of the rocks and, more subtly, processes that affected the magmas prior to eruption. A small number of the least evolved lavas from Muck (HM.75140) and the Lower Fionchra Member on Rum (SR 157, 217) have profiles resembling uncontaminated basalts from Skye and Mull, with smooth, concave-downward-facing curves from c.Sr to Ba, possibly reflecting derivation from a source that had been already depleted in these elements during the generation of small-volume alkali basic melts in the Carboniferous and early Permian (Thompson, 1982). The low phosphorus anomalies in most of the Skye and Mull lavas are not evident in the Small Isles uncontaminated lavas. The majority of the lavas have the distinctive profiles of rocks that have been contaminated by granulite-facies Archaean crust (cf. Thompson et al., 1982, fig. 2B). The significant troughs at Nb, and to a lesser extent at Rb reflect the lack of these elements in the contaminant, however, in a number of the Small Isles lavas the strong negative spikes at Rb may be due to leaching of this element during alteration of intersertal glass (e.g. SR 165, in the Upper Fionchra Member and also SR 156 of the Lower Fionchra Member). The behaviour of Sr is variable, especially in the less mildly alkaline to transitional members. The variability does not appear to be wholly related to plagioclase as is shown, for example, by the Eigg Lava Formation where strongly feldspar-phyric basaltic hawaiites (SR 560, 561) show relative depletion whereas an aphyric basalt (HE 7644) has a pronounced peak. The fairly common presence of zeolites will also affects the Sr levels (cf. Thompson et al., 1982; Kerr, 1993). However, in the most evolved hawaiites, the mugearites, and all the tholeiitic basaltic andesite and icelandite flows on Rum, Sr is severely depleted, indicating the extraction of a Sr-rich phase, presumably calcic plagioclase, during fractional crystallisation.

The well-defined sequence of lavas on Rum shows a progressive change with time. The variably crustally contaminated and rarer uncontaminated flows of the Lower Fionchra Member are succeeded by the successively more evolved Upper Fionchra and Guirdil members (Figure 40) and (Figure 47). The profiles in the Upper Fionchra and Guirdil members are similar and they show pronounced enrichment in the most incompatible elements (but see above for Rb) including the light rare-earth elements. In contrast, there are marked depletions in Nb, Sr in both members and in Ti in the Guirdil Formation. Subsequently, there was a reversion to slightly alkaline to transitional basaltic magmatism with the eruption of the Orval Member in which the slightly evolved basalts and basaltic hawaiites have fairly similar profiles, all with indications of contamination by granulite-facies crustal material. Following the model proposed by Morrison et al. (1985, fig. 4), the sequence suggests that the first olivine basalts followed independent paths from their mantle source to the surface with most, but not all, ponding near the Moho where they were variably contaminated by granulite-facies crustal material, depending on the length of residence. By the time the Upper Fionchra flows erupted, a fractionating magmatic system had become established in a high-level magma chamber which produced lavas which were enriched in incompatible elements, particularly those derived from the original crustal contaminants. This high level magmatism (Emeleus, 1985, fig. 10) culminated in the sparse, more evolved flows of the Guirdil Member. Subsequently, with a general decline in activity, only a few batches of cooler, mildly evolved magmas reached the surface from deeper levels; these had been weakly contaminated by granulite-facies Archaean crustal material, and form the Orval Member.

The compositions of the lavas on Canna (Figure 47) overlap the more evolved flows at high stratigraphical levels in the Lower Fionchra Member. They have the characteristics of crustally contaminated flows and their patterns show a much greater degree of coherence than on Rum. It is suggested that the lower flows on Canna are equivalent to the upper part of the Lower Fionchra Member. On Canna the manner in which the more evolved flows increase up the sequence shows that a fractionating magmatic system was becoming well established in their source area. The later basaltic andesite flow of Eilean a'Bhàird is considered equivalent to the Upper Fionchra Member.

Sgurr of Eigg Pitchstone Formation

The formation includes the pitchstone forming the Sgurr of Eigg and adjoining hills, the underlying fluviatile conglomerates and the pitchstone skerries of Oigh Sgeir [NM 157 962] (Figure 39). The Sgurr Pitchstone is of Eocene age (52.1 ± 1.0 Ma, Dickin and Jones, 1983) and is one of the youngest igneous bodies in the Hebridean Province.

Details

Eigg

The pitchstone on Eigg forms a prominent, steep-sided ridge from Bidean Boidheach [NM 441 866] south-east to the Sgurr [NM 463 846] (Figure 48), (Plate 26). It has been the subject of controversy. Geikie (1871, 1897) considered that it filled an ancient river valley eroded in the underlying basalt lavas and floored by fluviatile conglomerate, whereas Harker (1906, 1908) believed that the pitchstone intruded the lavas and interbedded sediments, a view he reiterated in discussion of a contribution by Bailey (1914) which supported Geikie's interpretation. There the matter rested until the pitchstone was remapped by Allwright (1980); her results support the interpretation of Geikie and Bailey and her mapping of the base of the pitchstone demonstrates that the body fills a system of palaeovalleys (Figure 48) and (Figure 49). The valleys appear to have drained westwards since their floors are at higher levels in the east than in the west, even after allowing for the possibility that the general low (c.5°) south-west dip of the lavas postdates eruption of the pitchstones.

The pitchstone has a black, vitreous to dark matt-grey appearance on fresh surfaces but weathers to a pale, grey rock. Phenocrysts of feldspar (up to 3 mm in length) are common and there are less numerous, smaller pyroxenes. Flow banding is commonly well developed, especially towards the base of the Sgurr Pitchstone. This is most obvious on the weathered surfaces and has been mapped (Figure 48). Columnar jointing is ubiquitous, the columns in the lower parts of the pitchstone are massive and perpendicular to valley sides and about 1 m in diameter. At higher levels, however, they commonly form fan-like structures, with bunches of flat-lying to inclined columns (Plate 27) which may merge upwards. The width, height and orientation of the columns appears to change across flat-lying discontinuities which lie within flow units rather than marking flow contacts. Flat-lying, pale-coloured felsitic sheets of devitrified pitchstone occur at several localities (Figure 48). On the southern face of the Sgurr up to four sheets are present between The Nose [NM 4636 8470] and the cliffs 600 to 800 m north-west of Grulin Cottage [NM 4561 8422]. The sheets are generally well defined, especially at their upper edges which are usually marked by slightly overhanging pitch-stone. They intrude the pitchstone and postdate the columnar jointing since they transgress the discontinuities where changes occur in the style of jointing (Plate 27). At several localities on the south face of the Sgurr the sheets appear to be intruded along discontinuitiues in the jointing or along possible flow boundaries. In detail the contacts between the sheets and the pitchstone are gradational over a few centimetres, as at the base of the cliff at [NM 4510 8490]. A single composite dyke of pitchstone with impersistent, fine-grained basaltic margins cuts the Sgurr pitchstone on the west side of the gully [NM 4585 8475] west of the Sgurr (Figure 48).

The base of the Sgurr pitchstone is irregular and undulating. Strongly transgressive contacts with underlying lavas occur at the Nose and approximately 300 m south-east of Loch Beinn Tighe (at [NM 4520 8616]). At Bidein Boidheach the pitchstone overlies coarse fluviatile conglomerate which fills a former valley to a depth of about 50 m (Figure 49). Conglomerate also crops out 10 to 15 m below the base of the pitch-stone at Collie's Cleft [NM 4630 8460] and it underlies pitchstone at the Recess, west of Botterill's Crack [NM 4606 8462]. Fragments of wood found at this locality include the remains of a tree trunk (the Eigg pine; Harker, 1908). In addition to well-rounded boulders and cobbles of basalt, the clasts include large (>1 m) blocks of red Torridonian sandstone although rocks belonging to the Torridon Group are not found in situ on Eigg.

The base of the pitchstone is brecciated at several localities (Figure 48). At Botterill's Crack, disoriented, angular, blocks of flow-banded pitchstone are cemented by yellowish, hydrated pitchstone debris, and at the Recess breccia has been injected for a distance of about 3 m into the overlying pitchstone. Similar breccia occurs at [NM 4473 8551], about 400 m west of Lochan Nighean Dughaill. The altered condition of the pitch-stone fragments in the breccias, together with the presence of small (<3 cm) pieces of basalt and sandstone, suggests that fragmentation occurred during phreatic explosions as the flow passed over water-saturated river deposits. Brecciation may also have occurred when pitchstone flowed over valley sides with irregular, stepped topography, as at the Nose (Allwright, 1980, fig. 6.4).

Oigh-sgeir

Oigh-sgeir [NM 155 960] is a group skerries some 9 km south-west of Canna (Figure 39). The islets are made of columnar pitch-stone identical to that which forms the Sgurr of Eigg. They were described by Sir Archibald Geikie (1897; also in Harker, 1908, p.176) who noted devitrification of the dark, resinous, porphyritic pitchstone.

Oigh-sgeir has not been resurveyed but additional samples were obtained. The rock (SR 303) is a rough-weathering, black to dark grey, semivitreous pitchstone crowded with feldspar crystals 3–7 mm in length. There is no flow alignment of the phenocrysts nor is there any flow banding. The rock is free of inclusions.

Petrography

The pitchstone from the Sgurr of Eigg Formation contains between 20 and 35% of phenocrysts of plagioclase (16–9%), anorthoclase (10.5–6.5%), Fe-rich clinopyroxene (2.1–1.1%) and orthopyroxene (3.0–0.8%), and opaque (Fe-Ti) oxides (2.6–0.85); quartz is rare or absent (modal data, in volume %, from Allwright, 1980, table 7.2). They are normally separate, euhedral crystals, commonly with sieve-like textures (Plate 28a); aggregates and intergrowths of crystals also occur and anorthoclase is found mantling plagioclase and in separate crystals. Mineral compositional data are given in ((Figure 50)a),(Figure b). Neither augite nor orthopyroxene contains exsolution lamellae but augite occurs mantling the weakly pleochroic orthopyroxene. Both titanomagnetite and ilmenite occur (Carmichael, 1967; Ridley, 1973).

The matrix of the porphyritic pitchstone varies from vitreous to thoroughly devitrified; perlitic cracking is common. The glass is generally featureless except towards the base of the pitchstone where microscopic-scale flow banding and rare fiamme-like structures occur. At Bidein Boidheach, a scarp at [NM 4414 8666] exposes the base of the columnar pitchstone resting on conglomerate. The lowest 5–10 cm lacks the usual abundant large feldspar phenocrysts but contains numerous small crystals, crystal fragments and millimetre-sized pieces of basalt in a matrix of vitreous, flat-lying, close-packed wispy fiamme (SR 490a–c; (Plate 28b); see also Bailey, 1914, pl. XXII.2 for a similar rock from the Nose).

The felsite sheets are characterised by fine-grained matrices consisting of opaque granules in a meshwork of minute feldspar laths. The matrix commonly has a small-scale orbicular texture, opaque oxides are abundant inside the grey-coloured rounded areas (c.0.5 mm diameter) and an unidentified mafic mineral (epidote ?) gives the intervening areas a yellow-green colour. The matrix commonly forms over 80 vol.% of the rock. The phenocrysts are principally sodic sanidines, which are distinctly more potassic that feldspar in the pitchstone ((Figure 50)a), and rare ilmenite, quartz, augite, and hypersthene (cf. Allwright, 1980, table 7.2)

Rocks from Oigh-sgeir (S35189), SR 303) contain numerous phenocrysts of anorthoclase, plagioclase ((Figure 50)a), augite and orthopyroxene ((Figure 50)b) , and opaque oxides. Rare apatite crystals were also noted. The phenocrysts are set in pale brown glass crowded with plagioclase and pyroxene crystallites and minute opaque oxide granules. The details of the petrography, mineralogy (including mineral chemistry) are essentially similar to those of the Sgurr of Eigg pitchstone.

Chemical composition

Representative analyses of the pitchstone, separated glassy groundmass, and felsitic sheets are listed in Appendix 5d (see also Carmichael, 1963; Ridley, 1973, table 6; Allwright, 1980, table A7.1). With total alkalis invariably >8%, SiO2 >60% (generally >65%), and normative quartz <18%,the pitchstone is subacid and should strictly be classified as a trachyte (LeMaitre, 1989, fig. B.13).

There are differences between the compositions of the felsitic sheets and the enclosing pitchstone. The sheets are the more fractionated rocks (Figure 51) and the differences are immediately apparent when the data are presented on rock/chondrite normalised multi-element profiles (Figure 52). Ba, Sr, P and Y are relatively depleted and there is enrichment in Rb and Zr. The Ba, Sr, P and Ti depletion reflect the extraction of these elements in anorthoclase, plagioclase, apatite and opaque oxides all of which occur as phenocryst phases in the pitchstone, whereas sodic sanidine, small opaque oxides and rare quartz and pyroxene are the principal phenocrysts in the sheets. There is also considerable depletion in Ca, Fe and Mg in the felsite sheets reflecting the sparcity of pyroxene phenocrysts. The major element analysis of glass separated from fresh pitchstone (Carmichael, 1963) closely resembles the sheets. When compared with other felsic rocks from Eigg (except for the Grulin Felsite), the pitchstones and associated felsites are furthest removed from the temperature minimum for low PH2O in the normative quartz–alkali feldspar system (Figure 51).

Sub-pitchstone conglomerates of Eigg

The principal conglomerate occurrences are at Bidein Boidheach [NM 4410 8668] and on the southern side of the Sgurr, at the Recess [NM 4612 8462] and at [NM 4628 8459] on the slopes to the ESE (Figure 48). Conglomerate is probably also present about 100 m west of Loch Caol na Corabheinne [NM 4536 8547], where there are rounded cobbles of basalt and pebbles of basalt and red sandstone in the brecciated base of the pitchstone (Allwright, 1980, p.138).

The accessible upper part of the Bidean Boidheach outcrop consists of rounded cobbles and pebbles of basalt, amygdaloidal basalt, and red sandstone. Granite and other igneous rock fragments have also been recorded (Harker, 1908, p.52). The deposits are unbedded and the clasts are in a dark matrix of small pieces of basalt and sandstone and grains derived from these rocks. The clast size increases downwards and boulders up to 0.5 m diameter are visible in the cliff face. At the Recess, subangular blocks of red Torridonian sandstone are common, a few examples exceed 2 m in diameter. These are accompanied by cobbles and pebbles of basalt and rare pebbles of porphyritic rhyolite and quartzphyric acid glass. This locality is the site of the Eigg pine (Pinites eiggensis), described over 160 years ago (Witham, 1831; see also Harker, 1908, p.53) and identified as of Eocene age and contemporaneous with the deposits. Harker (1908) also described fragments of brown wood found with locally abundant pieces of white sandstone but more recent work failed to disclose further examples. These plant remains are distinct from the black, largely silicified Eocene wood of the Eigg pine and Harker considered the white sandstone and the plant remains to be of Jurassic age. A sample of the conglomerate matrix from the Recess contains schists, red arkosic sandstone, red-brown sandstone showing graded bedding, plagioclase-phyric basalts and dark, opaque fragments possibly of carbonised wood (Allwright, 1980).

At the locality on the slopes below Collie's Cleft [NM 4630 8460], the deposits are bedded. The coarse and fine conglomerates contain boulders, cobbles and pebbles of basalt, olivine-basalt, hawaiite, mugearite and oligoclase-phyric glass, most of which match members of the Eigg Lava Formation. Other pebbles include red and green arkosic sandstones, similar to Torridon Group rocks on Rum, and a felsic rock containing phenocrysts of plagioclase, anorthoclase, clinopyroxene, orthopyroxene and opaque oxide in a matrix of fine-grained, felted feldspar laths and quartz. The rock closely resembles the Sgurr pitchstone, as does one of the sectioned pebbles from the Recess.

Conclusions

The Sgurr of Eigg Pitchstone Formation on Eigg commenced with coarse fluviatile conglomerates which accumulated in some of the valleys of a westerly draining system eroded in flows of the Eigg Lava Formation. The valleys were then filled by the pitchstone, which started with an ash flow and continued as one or more lava flows. The basal pitchstone was brecciated where it collapsed down steep valley sides and where phreatic activity occurred as the incandescent flow passed over waterlogged sediments. The sediments were also disturbed, mixing with the breccias. The Sgurr pitchstone has usually been interpreted as a single flow unit. Exposures on the southern face of An Sgurr, north of Grulin, suggest that there is more than one flow; there are several distinct, sharply defined, flat-lying discontinuities in the pitchstone forming the 50 to 100 m cliff. The sudden changes in the attitude and style of jointing in the pitchstone might also suggest that there are several distinct flow units but these are not reliable criteria since other single lava flow-units exhibit a similar range of jointing, as seen for example in the lowest olivine-basalt flow of the Laig Member (Eigg Lava Formation) about 1 km NNE of Kildonnan (see also, for example: BGS photo of Staffa, D 2218; Wilson and Manning, 1978, pls. 1 and 13; Holmes, 1993, fig. 13.31). The flat-lying felsitic sheets are intrusive. They are chemically and mineralogically more evolved than the pitchstone and are considered to be differentiates of the pitchstone magma. Their occurrence and that of the composite basalt-pitchstone dyke, suggests that the source of the pitchstone was probably close.

The similarity between the pitchstones of Oigh-sgeir and the Sgurr of Eigg has long been noted (Harker, 1908) and it has been speculated that they were once connected; in Geikie's view, as part of a subaerial lava flow. The structural, petrographic and compositional similarities are certainly impressive. If the pitchstone originated as a silicic ash flow, there would be little difficulty in accepting that a valley-guided flow of this type could readily have traversed the intervening 30 km in a few hours at most. However, convincing evidence for an origin as an ash flow, or flows is confined to the lowest few centimetres at Bidein Boidheach and possibly also at the Nose. The rest of this body, and that forming Oigh Sgeir, contains numerous delicate but unfragmented phenocrysts and crystal aggregates, and lacks fiamme or lithic clasts. In the absence of evidence to the contrary, the pitchstone is interpreted as having originated as a lava flow, or flows, closely asociated with a preliminary, minor ash flow (cf. Henry and Wolff, 1992, especially table 3).

The conglomerate clasts are principally basalts of local origin but include material derived from the weathering of Torridon Group arkosic sandstones and also porphyritic rhyolites or pitchstones. Harker (1908, pp.52, 53) noted that white Jurassic sandstone was locally abundant but these clasts had disappeared by the time of the resurvey. The presence of pitchstone pebbles is an indication that there were flows and/or dykes similar to, but earlier than the Sgurr Pitchstone. The occurrence of Torridonian fragments, including blocks of considerable size, suggests that the high land formed by these rocks during or after emplacement of the Palaeocene Rum Central Complex persisted into the Eocene. However, the westward-directed drainage pattern of the former valleys beneath the Sgurr Pitchstone poses problems for the direct derivation of the clasts from Rum and there may have been a source to the east (cf. Chapter 12, p.133).

Chapter 9 Palaeogene 5: Magma genesis and tectonic setting

Introduction

The Hebridean part of the British Tertiary Volcanic Province is a region where crustal extension and fracturing has occurred since at least the late Palaeozoic. This resulted in the formation of basins containing thick successions of (mainly) Mesozoic sedimentary rocks separated by ridges of Precambrian rocks (Binns et al., 1974; Fyfe et al., 1993). Major Caledonian structures cross the area, including the Moine and Outer Isles Thrusts, and NE-striking dislocations of which the most notable example is the Great Glen Fault. During the Palaeocene, the presence of locally thinned crust over the North Atlantic plume (White and McKenzie, 1989) resulted in dry decompression melting of the asthenosphere and the generation of voluminous basaltic magma (Thompson and Gibson, 1991). The continuing stresses created favourable conditions for the magma to be injected into tensional fissures as dense, NW- to NNW-trending dyke swarms and to pond in the Mesozoic sedimentary rocks as extensive sill systems (England, 1992). The basaltic magma is thought to have reached the land surface from fissure eruptions, building up lava fields which were preferentially located over the sedimentary basins (Walker, 1979). This suggests that subsidence and crustal thinning continued until at least the end of the Palaeocene.

Central intrusive complexes were also established, generally but not exclusively occurring where the intrabasin ridges of Precambrian rocks are now crossed by NW-trending dyke swarms. Evidence of early subaerial volcanism, comprising basaltic and rhyolitic lavas, and pyroclastic rocks, is commonly preserved as relicts amongst the numerous intrusive members of the complexes, adding support to the view that they are the eroded roots of central volcanoes. The siting of the central volcanoes was probably determined by the intersection of the ridges and pre-existing, long-lived faults and other structural breaks (Richey, 1937) and by early representatives of the regional dyke swarms, which were envisaged to have been fed from linear basalt magma chambers at between 20 and 50 km depth (Speight et al., 1982). However, from an examination of the Early Tertiary stress regime in the Hebrides, England (1988) concluded that the regional stress field had no control on the siting of the central complexes. There is, however clear evidence that the centres did influence the stress field in their more immediate surroundings since secondary swarms focus on centres, dilation in both regional and secondary swarms increases towards the centres and dykes of acid and intermediate composition occur within and near the central complexes but are rare elsewhere. The highly localised, intense positive Bouguer gravity anomalies prove unequivocally that basaltic magmatism was dominant in the central volcanoes (Chapter 11). There is, however, also abundant evidence that silicic magmas were important, alternating and co-mingling with the basaltic magmas, and large granitic intrusions are ubiquitous but do not extend to great depths. The central complexes represent sites where igneous activity was sustained for periods ranging from 1 Ma to over 5 Ma (Mussett et al., 1988). During this time large volumes of hot, basaltic magma existed at high levels within the lithosphere where they formed plutons and augmented the regional dyke swarms. They also melted the less refractory components in the adjoining crust and detailed isotopic and elemental geochemical studies have shown that it is these rheomorphic melts, together with differentiates from the large bodies of cooling and fractionating basaltic magma, that are responsible for the acidic intrusive and extrusive rocks of the central complexes and their immediate surroundings (Dickin, 1981; Walsh et al., 1979). The elemental and isotopic signatures of the granitic rocks in the central complexes suggest a component derived from amphibolite-facies crustal rocks, in contrast to the basic lavas whose isotopic signatures generally indicate that they have been contaminated by anatectic melts derived from high-grade granulite-facies Archaean crustal rocks (Carter et al., 1978). Contamination of the magmas that formed the basalt flows is thought to have occurred independently of the central complexes (Morrison et al., 1985).

The Rum Central Complex is situated on the Precambrian ridge between the Sea of the Hebrides Basin, which is partly overlain by the flows of the Canna Lava Formation, and the Eigg Basin where the Eigg Lava Formation rests unconformably on a Middle Jurassic to Upper Cretaceous succession. There is some evidence that the Rum Central Complex is sited where a major, long-lived north–south fracture crosses the Precambrian ridge, providing a possible structural focus for initiation of the centre. The Long Loch Fault bisects the centre from north to south (Figure 53) and although the proven movement on it is post-central complex there is circumstantial evidence that the fault had a history of pre-central complex movement (Chapter 10). The central complex has developed where a regional dyke swarm crosses the Precambrian ridge; however, evidence cited in earlier chapters indicates that Stage 1 of the central complex predates both this regional dyke swarm and the local radial swarm. Most of the dykes were intruded at the start of Stage 2. It is perhaps surprising that no pre-central complex dykes have been recognised since inclusions in the Am Màm Breccias show that plutonic mafic intrusions had formed at a very early stage (Chapter 5), and the pre-central complex basaltic flows of the Eigg Lava Formation on Eigg, Muck and eastern Rum were probably fed from early basaltic fissure eruptions.

The earliest magmatism in the area of Sheet 60 was the eruption of subaerial basaltic lavas in and at the margin of the Eigg Basin, as early as 62.5 Ma (Pearson et al., 1996). At about 59 Ma a central complex dominated by basaltic magmatism became established at Rum. In the complex, early plutonic mafic magmatism was followed by, or more probably accompanied by, intrusion and extrusion of silicic magmas of Stage 1. At present there are insufficient elemental or isotopic geochemical data to enable us to determine the relative contributions to the acid magmatism from lithospheric melting and fractionation of mafic magmas. This activity was followed by intrusion of the majority of the NW-trending (regional) and radial (secondary swarm) dykes together with numerous basaltic sheets. The final major igneous event was emplacement of the gabbroic and ultrabasic rocks of the Rum Layered Suite. The complex and its surroundings were then deeply dissected by erosion, which was accompanied by renewed eruption of basic magmas, fed from sources outwith Rum. This later igneous activity, which included intrusion of a few dykes, was probably linked with a shift of activity towards Skye; the Canna Lava Formation appears to be a continuation of the lava field in south-west Skye (Binns et al., 1974, fig. 2), which in turn is intruded by the Skye Central Complex.

The evidence at present available from Rum and its surroundings does not solve the problem of the siting of the central complexes but it does perhaps suggest that the Palaeocene igneous activity in the area may be part of a discrete tectono-magmatic package. After an early fissure-fed basaltic lava field had formed in and near the Eigg Basin, igneous activity became centralised on Rum, at a ridge of relatively thick crust, beneath and within which large volumes of basaltic and picritic–basaltic magmas were emplaced and to some extent ponded. These in turn generated acid magmas by fractionation and by melting the surrounding crustal rocks. After consolidation of the acid magmas, denser basaltic magmas once more rose to high structural levels, at first as a multitude of dykes, sheets and cone-sheets and, later forming large, high-level intrusions in which picritic basalt and basalt magma fractionated, forming the Rum Layered Suite. The field evidence from Rum is strongly suggestive that the associated regional dyke swarm was emplaced as part of this package and did not preceed the initiation of the Rum Central Complex. Profound erosion then ensued, accompanied by further subaerial lava flows. However, there is no evidence that these lavas originated on or near Rum; it is suggested that at this stage Rum and its surroundings were in effect nearly dead magmatically and that this later activity is an early manifestation of a new package of linked tectonic and igneous activity that was focussed on Skye.

There was however, a brief recurrence of igneous activity during the Eocene. Mineralogically and compositionally identical pitchstone lava flows formed on Eigg and Oigh-sgeir, underlain by an ash flow on Eigg. A small composite pitchstone–basalt dyke cuts the pitchstone flow on Eigg where there is also a pitchstone dyke that is petrographically and chemically similar to the Sgurr of

Eigg Pitchstone. These may have acted as feeders for the Eigg flows. The basaltic heat source which generated this acid magma by partial melting of granulite-facies Lewisian crust and by magmatic fractionation (Dickin and Jones, 1983) has not been identified.

Mafic magmas of The Rum Central Complex

The origins of the rocks in the Rum Layered Suite are problematical: were the contrasted layers formed by separate and distinct magmas; have a variety of magmas been involved in the genesis of the suite; or was there a common parental magma for all members of the suite and, if so, was this of basaltic or ultrabasic composition?

Harker (1908) suggested that the peridotites and troctolites represented separate injections of contrasted olivine-rich and feldspathic magmas which had differentiated from a common ultrabasic parent. However, using the compositions of orthocumulates in Unit 3 of the Eastern Layered Intrusion, Brown (1956) concluded that there was a single parent magma, of aluminous tholeiitic basalt composition, for the Eastern Layered Intrusion ((Table 9)(a)). He found that in sharp contrast to the layered Skaergaard Intrusion (Wager and Deer, 1939), the Eastern Layered Intrusion of Rum lacked major cryptic layering. To explain this, Brown suggested that the layered rocks were built up by the accumulation of high-temperature minerals crystallised from successive pulses of basaltic magma, with the lower-temperature residua being extruded during contemporary surface volcanism. Rum therefore functioned as an open magmatic system, unlike the closed system envisaged for the Skaergaard Intrusion (Wager and Brown, 1968).

Brown's (1956) hypothesis of periodic magmatic replenishments for the Eastern Layered Intrusion is generally accepted. However, in the absence of any contemporaneous lavas of appropriate composition, subsequent investigators have sought evidence for a parent magma (or magmas) closer to the bulk composition of the Layered Suite, thus removing the need for voluminous, low-temperature residual magmas. From observations on Skye and Rum, which included the examination of minor ultrabasic intrusions as well as the larger masses, Gibb (1976) concluded that the parental magma for the ultra-basic rocks was not basaltic but was of 'eucritic' (bytownite-gabbro) composition with suspended olivine crystals. Various observations led Donaldson (1974, 1982) to suggest that hydrous feldspathic peridotite magmas had been present as the complex evolved. McClurg (1982) discovered picritic dykes that intrude the layered rocks of Barkeval, including an aphyric dyke with 13.47 wt% MgO and olivine-phyric dykes with highly magnesian ground-mass compositions ((Table 9)(b) and (d) respectively). Using cumulate olivine compositions, McClurg (1982) and Volker (1983) calculated that the parent liquids of some peridotites must have had about 20 wt% MgO (cf. (Table 9)(d)). From a study of late-stage teschenitic veins in peridotites of the Central Intrusion near Salisbury's Dam [NM 3643 9992], Kitchen (1985) suggested that the parental picritic magma was mildly alkaline rather than tholeiitic.

Attention has centred on the nature of the new influxes of magma and their interaction with resident magma. Huppert and Sparks (1980) proposed that each unit represents the products of a new influx of hot, dense picritic magma (MgO c.20 per cent) into cooler, light resident magma with MgO c.10 per cent (cf. Brown, 1956; see (Table 9)(a)). With each influx, the new magma spread out beneath the resident magma, both magmas convected vigorously, and separately, as heat was transferred upwards and abundant olivine crystallised in the lower body. The olivine was held in suspension until convection ceased when en-masse settling allowed little opportunity for sorting and layering to develop as the peridotite solidified. The residual basaltic liquid then mixed with the overlying resident magma and, as cooling continued, this hybrid magma crystallised olivine-plagioclase-pyroxene cumulates (i.e. bytownite-troctolites and bytownite-gabbros) until further picritic liquid entered the magma chamber.

Studies of Sr, Nd and Pb isotopes in the upper seven units in the Eastern Layered Intrusion (Palacz, 1985) and of Sr and Nb isotopes in Unit 10 (Palacz and Tait, 1985) indicated that the troctolite has been contaminated by crust rich in radiogenic Sr, whereas the peridotites are relatively uncontaminated. The contrasted rock types may, therefore, have arisen from separate magmas; primitive picritic magma crystallising as peridotite and contaminated tholeiitic feldspathic basalt magma forming troctolite. These conclusions are in accord with the model proposed by Young et al. (1988). They note that whereas boundaries between layered units are generally abrupt, those between peridotite and troctolite within units may be abrupt or gradational. Also, in the centrally dipping sequences of peridotite and troctolite, the latter thins down-dip while peridotite thickens (Figure 30),(Figure 31), it is more evolved than peridotite, and shows evidence of contamination by crustal material. They propose that peridotite and troctolite were crystallising simultaneously from picritic magma overlain by basaltic magma. As crystallisation progressed the troctolite cumulates formed from the basalt magma and built out across the underlying peridotites (Young et al., 1988, fig. 5). However, when picritic replenishment occurred, the interface between the magmas rose, allowing peridotite to form above troctolite and hence give rise to a new unit. Areas closer to the source of the picritic magma were most favourable for peridotite formation, while those farthest away were sites where troctolite accumulated (Young et al., 1988, fig. 7). On this model the litho-logical transitions are slightly diachronous and the picritic magmas were fed from a source or sources on or near the line of the Long Loch Fault.

Rocks of basaltic composition are important components of the Layered Suite (Figure 32). Compositions range from quartz-hypersthene-gabbro, much modified by the assimilation of siliceous country rocks, to bytownite- (olivine )-gabbro ('eucrite'). They include rocks of mildly alkaline to transitional basaltic composition, similar to members of the Small Isles dyke swarm and other minor intrusions on Rum (Allwright, 1980; Forster, 1980; McClurg, 1982; Volker, 1983). One of the few preserved examples of a chilled contact at the margin of the Layered Suite, at the base of the cliff [NM 4003 9404] on south-eastern Beinn nan Stac, is a picritic dolerite (Plate 19d); (Table 9)(c)). This is thought to represent an olivine-tholeiite liquid, with about 19% (volume) suspended olivine crystals, from which troctolites crystallised (Greenwood et al., 1990).

The present overall consensus is that dense, high temperature picritic liquids were parental to many of the olivine-rich ultrabasic rocks of Rum. (e.g. McClurg, 1982; Volker, 1983; Faithfull, 1985; Tait, 1985; Kitchen, 1985; Bedard et al., 1988; Young et al., 1988; Volker and Upton, 1990). Basaltic magmas were also available throughout the growth of the complex; they formed the gabbros and the troctolites, the latter being augmented by, or even largely formed from low-temperature residua from peridotite crystallisation. The long-resident basaltic magmas experienced significant contamination by amphibolitefacies crustal rocks (e.g. Palacz and Tait, 1985), and some contamination also occurred at high levels in the central complex (Greenwood, 1987). However, there is evidence, for example from Unit 14, which indicates that from time to time the chamber was replenished by relatively uncontaminated basaltic magmas as well as by picritic ones (Renner and Palacz, 1987).

While this memoir was in press, continuing investigations (1995) of the Rum Layered Suite have shown that elements of the Eastern and Western layered intrusions may be correlated with parts of the Central Intrusion, and that it is probable that the Layered Suite was fed through a system of conduits situated approximately on the line of the Long Loch Fault (cf. McClurg, 1982). Emplacement of the Layered Suite is thus envisaged as a continuous process, with the Central Intrusion as the youngest exposed remnant. Whereas the mafic rocks have been modelled as a steep-sided body continuing to considerable depth beneath Rum (Chapter 11), an alternative interpretation of the gravity and field data is that the form of the Layered Suite is mushroom-like, with ultrabasic and gabbroic cumulates forming a thin, disk-like body some 2 km+ in thickness, fed through dyke-like conduits, the most recent of which are represented by the linear zones of ultrabasic breccia (M J Cheadle and R H Hunter, personal communications, 1994–95). The head of the body built up through the accretion of successive batches, or pulses, of Mg-rich (MgO = 13–20 wt%) magma which formed thin, tabular sill-like bodies. Space for each pulse of magma was made by a combination of roof doming and floor subsidence. The Layered Suite built from the base upwards, each magmatic pulse formed olivine-rich cumulates and generated a pool of residual, more evolved liquid. As in previous models (e.g. Brown, 1956), the volume of evolved magma was significant; while some may have been erupted during surface volcanism, much will have formed 'allivalitic' cumulates, and some may have formed gabbroic intrusions (Emeleus et al., 1996b).

Paleocene lavas

The mildly alkaline to transitional lavas of the Small Isles are similar to members of the Skye Main Lava Series, with which the Canna Lava Formation may be coeval. However, the analytical data for the Small Isles lavas are not yet sufficiently detailed to allow the rigorous examination of the pre-eruption histories of the magmas as has been done for Skye, Mull and elsewhere in the Thulean Super-Province. The elemental analyses indicate that many of the lavas have been contaminated by granulite-facies Lewisian gneisses although a few flows in both the Eigg Lava Formation and the Lower Fionchra Member of the Canna Lava Formation appear to be uncontaminated (Chapter 8).

The flows of the Eigg Lava Formation are similar to the early members of the Canna Lava Formation but lack the more evolved, tholeiitic flows that form much of the latter on Rum (Ridley, 1973; Allwright, 1980; Emeleus, 1985). The Eigg Lava Formation contains few major intralava sedimentary rocks and almost none of the coarse fluviatile conglomerates that are common in the Canna Lava Formation. This feature, together with the monotonous sequence of olivine-basalt lavas through much of the succession, suggests that the formation built up rapidly from closely spaced eruptions. By contrast, the numerous erosional intervals recorded in the Canna Lava Formation, together with the sequence of progressively more-evolved flows on Canna and the alternation of evolved transitional and tholeiitic members on Rum, point to intermittently erupting volcanoes, some of which tapped sources undergoing active fractionation. The close similarities between the multi-element profiles for flows in the Upper Fionchra and Guirdil members (Figure 29) and their bulk compositions (Appendices 5e:i, e;ii) indicate that they were fed from a strongly fractionating magma chamber established at a fairly high crustal level. By contrast the multi-element profiles of the flows of the Eigg Lava Formation and the earlier members of the Canna Lava Formation are much more variable and this together with their bulk compositions suggests that they may largely have followed independent paths from deep sources, in the manner suggested by Morrison et al. (1985, fig. 4).

Dykes and other basaltic minor intrusions

Allwright (1980) found that the dykes and other minor intrusions of Eigg, Muck and Canna are dominantly of olivine-tholeiite composition but a substantial proportion of mildly alkaline (nepheline-normative) intrusions occurs, with the highest proportion on Muck (41 per cent, Allwright, 1980, table 10.1). On Muck rare quartz-normative tholeiitic dykes occur (10 per cent), including icelandite; none of the analysed dykes on Eigg is quartz-normative. On Rum, Forster (1980) found that the dykes span a wide range of compositions and he identified several distinctive groups (Chapter 6). The majority of members of these groups are transitional between the mildly alkaline Skye Main Lava Series (Thompson et al., 1972) and the quartz-tholeiite Ardnamurchan cone-sheets (Holland and Brown, 1972). Both alkalic and tholeiitic groups include some more evolved compositions. Although there is considerable temporal overlap between the groups, Forster (1980) detected a progressive change from tholeiitic to mildly alkalic compositions with time. From the limited analyses available, the dykes on Muck have geochemical signatures that indicate the majority were contaminated by contributions from granulite-facies Archaean crustal rocks whereas signs of contamination are much less common in the dykes from Eigg and Rum.

Silicic magmas

Rum Central Complex

There is strong compositional overlap between the porphyritic rhyodacites and the granites and granophyres of Rum (Figure 24) and (Figure 25), which have been regarded respectively as effusive/high-level intrusive and deeper-seated intrusive products of the same magma (Dunham 1968; Dunham and Emeleus, 1967). As is found elsewhere in the British Tertiary Volcanic Province, the silicic rocks form relatively small bodies overlying dense mafic rock (Chapter 11). From inclusions in the Am Màm Breccias (Chapter 5), it is known that gabbro and peridotite bodies preceded rhyodacite in the Northern Marginal Zone which in turn predates the granites. The magmas responsible for these early mafic and ultramafic rocks would have produced some melting of the country rocks which would be mainly Lewisian gneisses at the level of the magma chamber (Dunham, 1970). There are, however, significant compositional differences between the gneisses and the silicic Palaeocene rocks (Figure 24); Meighan, 1979), and the rhyodacites and granites cannot have formed simply by the wholesale melting of leucocratic gneiss. In both Mull and Skye detailed elemental and isotopic geochemical studies have shown that the Palaeocene granitic rocks have components derived both from the fractionation of basaltic magmas and from partial melting of crustal rocks (Walsh et al., 1979; Thorpe, 1978; Dickin, 1981, 1988). No comparable geochemical studies have yet been published from Rum, but unpublished strontium isotope data for the Western Granite suggest that these rocks include a substantial contribution from a crustal source enriched in radiogenic strontium (87Sr/86Sr at 58.4 Ma ranges from 0.71265 to 0.71397; I G Meighan, personal communication, October 1992). The close geochemical similarity between the granites and granophyres, and the rhyodacites indicates that they originated from the same source.

Sgurr of Eigg Pitchstone

Dickin and Jones (1983) made a detailed isotopic study of the pitchstone and obtained an age of 52 ± 1.0 Ma, the youngest age obtained from an igneous body in the Hebridean area (Chapter 11). Evidence from Pb, Nd and Sr isotopic analyses suggests that the pitchstone is best explained as originating from the melting of lower-crustal granulite-facies Lewisian gneiss, contaminated by Torridonian rocks. The pitchstone dyke at the North Pier (SR 505) [NM 4832 8418] has a similar petrography and comparable radiogenic isotope ratios to the Sgurr Pitch-stone and it was considered to be a possible feeder (Dickin and Jones, 1983). If the interpretations of the isotopic data are correct, they imply that a heat source (presumably basic or ultrabasic magma) capable of producing appreciable melting of granulite-facies gneisses was present in the district during the Eocene, and that both granulite-facies Lewisian and Torridonian rocks occur beneath or near to Eigg. The Grulin Felsite and Sgurr Pitchstone have significantly different Pb-isotope compositions and distinct Sr- and Nd-isotope ratios; hence the Grulin Felsite is unlikely to have been a feeder for the pitchstone (cf. Harker, 1908). The felsite is thought to have had a notable contributions from granulite-facies Lewisian gneiss but a lesser amount from Torridonian sandstone sources than the pitchstone. Dickin and Jones consider that the Rubh' an Tangaird pitchstone dykes on Eigg are mantle-derived basaltic differentiates which were slightly contaminated by Torridonian rocks.

Chapter 10 Structure

Introduction

The structural history of the Small Isles is difficult to trace earlier than the Mesozoic. Despite lying directly in line with the Caledonian thrust zone that extends from Skye to Sutherland (Johnstone and Mykura, 1989), no thrust relicts of that age are identified on any of the islands. The supposed thrust breccias in north-west Rum (Harker, 1908) were shown by Bailey (1945) to be Triassic sedimentary breccias and, as will be shown, the Welshman's Rock Fault (Emeleus, 1981) is now considered to be of Tertiary age. Any rocks thrust over the area of the Small Isles during the Caledonian Orogeny were removed by erosion before the Mesozoic sediments were deposited directly onto the Torridonian rocks of the Caledonian foreland.

Marine investigations around the Small Isles show that the Precambrian ridge, of which Rum is a part, is separated from the Mesozoic Inner Hebrides Basin to the east, within which Eigg and Muck lie, at least in part by a fault (Figure 1); Binns et al., 1974, fig. 2). This fault is considered to be part of the Camasunary Fault of southern Skye (cf. Richey, 1961, fig. 11). In the west of the area, the feather edge of the Mesozoic Sea of the Hebrides Basin onlaps north-west Rum where Triassic rocks overlie beds of the Torridon Group. The islands of Canna and Sanday form part of a basalt ridge that extends from south-west Skye to Oigh-sgeir, within the Sea of the Hebrides Basin. It is apparent that the thickest Palaeocene lava successions coincide with the axes of earlier sedimentary basins, as noted elsewhere in the Hebridean Province and Northern Ireland (Walker, 1979).

Pre-Palaeocene movements

Rum

Pre-Palaeocene folding and probably faulting affected the Torridonian and Triassic rocks of Rum. The westward-dipping Torridonian succession is overlain by flat-lying lavas of the Canna Lava Formation and the full Torridonian succession is cut by the Rum Central Complex. The Long Loch Fault (Figure 53) affects a narrow zone, up to 40 m in width, within the central complex. However, the continuation of this fault for 3 km to the north, in Kilmory Glen, is marked by a flat valley up to 200 m in width. Sandstones on the west side of the valley are crushed and strong north–south joints occur in sandstone outcrops in the badly exposed valley floor. This suggests that a much wider zone was affected by pre-central complex (and possibly pre-Tertiary) faulting. However, a lack of distinctive horizons in the Torridonian succession means that the precise extent of any pre-central complex movement is not known.

A major fault separates the Torridonian Sgor Mhór and Loch Scresort sandstone formations on Bloodstone Hill [NG 315 007]. The fault strikes NNE, is overlain by Canna Lava Formation flows on the hilltop and is cut off by the Main Ring Fault on Sgorr Mhór [NM 309 999]. Another fault of possible pre-Tertiary age brings Triassic rocks against Torridonian beds north-west of Loch Sgaorishal. The fault does not affect a peridotite plug at [NG 343 023]. To the north-west, a prominent trench that trends about 065° crosses the Triassic and Torridonian up to 2 km WSW of Kilmory Lodge [NG 3576 0392] but the boundary shows no displacement across this feature. It is one of several similar features in the sedimentary rocks either side of Kilmory Glen, notably between Kilmory and Baigh Rubha Mhoil Ruaidh [NG 390 036]. A small WNW-facing monoclinal fold affects the Triassic and Torridonian rocks on the north side of Glen Shellesder [NG 330 026], this appears to have developed as the Triassic beds were being deposited (Chapter 4).

Eigg and Muck

Palaeocene lavas of the Eigg Lava Formation overstep successively higher Jurassic formations from the Valtos Sandstone Formation on the north-east coast of Eigg to the Kilmaluag Formation near Laig Gorge. Farther west, lavas overlie Staffin Shales, which may be in (pre-lava) faulted contact with older Jurassic rocks to the east. There is also evidence that the Laig Gorge Fault (Figure 55) moved after deposition of the Cretaceous rocks at Laig Gorge but before the lavas were erupted. On Muck, a fault separating the Kilmaluag and Duntulm formations near the west side of Camas Mór [NM 405 791] shows more displacement of the Mesozoic rocks than of the overlying Paleocene lavas.

Eigg and Muck therefore provide evidence that the Mesozoic rocks were tilted to the west and cut by NNW-trending faults prior to the Palaeocene. The tilting may have been pre-Upper Cretaceous since glauconitic sandstones of presumed Upper Cretaceous age overlie the Staffin Shale, Kilmaluag and Valtos Sandstone formations at various points in northern Eigg (Chapter 4).

Palaeocene and later movements

Rum

Major faulting and associated folding occurred during Stage 1 and Stage 2 of the development of the Rum Central Complex (Table 1). The temporal relationships between faulting, folding and the igneous intrusions are summarised in (Table 10).

Welshman's Rock and Mullach Ard Faults

These faults are inclined at low angles (35–40°) to the ENE and east between Loch Scresort [NM 415 995] and the south end of Welshman's Rock [NM 417 946]. Sandstones of the Scresort Member of the Torridonian are underlain by beds of the Laimhrige Shale Member at the north end of Welshman's Rock. At the south end, shattered dolerite and basalt sheets appear at sea level, beneath the overhanging cliff at [NM 4168 9463] and crushed basic intrusive rocks also occur beneath the Mullach Ard Fault, in the cave at the north end of Bàgh na h-Uamha [NM 4229 9741]. The intrusions are petrographically similar to Tertiary dykes and sheets, and negate the suggestion that the faulting was part of the Caledonian thrust belt (Emeleus, 1981). Both faults are of Palaeocene age.

The Torridonian beds west of both faults, and in the ground around Laimhrige [NM 418 968], show changes in dip progressively from 15° WNW, to horizontal, to low easterly values close to the faults. A small anticlinal structure occurs on the south side of Loch Scresort, just west of the Mullach Ard Fault (Figure 53). This structure, and the changes in dip elsewhere, are attributable to fault drag as the overlying beds moved eastward down the low-angled faults, probably in response to inflation and doming over the Rum Central Complex. A minimum movement of 500 m down to the east on the Mullach Ard fault plane is required to explain the present disposition of the Torridonian rocks. At Welshman's Rock, the fault block has rotated through almost 90° in an anti-clockwise direction, which may account for the minor folding and faulting present at the north end of the block (Emeleus, 1981, fig. 2).

Main Ring Fault

The Main Ring Fault is a near-vertical or steep, inward-dipping fault or system of faults that extends from Camas na h-Atha [NM 303 996] in western Rum around the north and north-east margins of the Rum Central Complex as far as Allt Mór na h-Uamha [NM 403 976], and from Allt nam Bà [NM 407 943] south and south-west for about 5 km to near Papadil (Figure 53). The fault defines the outer margins of the Western Granite, the Northern Marginal Zone and the Southern Mountains Zone. It is cut by peridotite tongues and late gabbro plugs near Loch Duncan [NM 372 992], and by members of the Central Intrusion near Minishal and Papadil, and by the Eastern Layered Intrusion in the east. The Main Ring Fault is offset dextrally for up to 800 m by the Long Loch Fault. Dextral offsets also occur on NW- to NNW-striking faults between Minishal and Camas na h-Atha in western Rum, where the Main Ring Fault ends against a major NW-striking dextral fault (Figure 53).

The Main Ring Fault forms a zone of bleached and crushed granophyre and sandstone in the cliffs at [NG 320 000] on the south-east side of Bloodstone Hill. It is also exposed to the south and south-east of Minishal where fault-brecciated rocks are thermally overprinted by the adjacent gabbros (Hughes et al., 1957). The Main Ring Fault divides into inner and outer branches north of the Coire Dubh dam. The inner member is exposed in Allt Slugan a'Choilich, a few metres above the hydroelectric dam [NM 393 983], where steeply inclined, bleached and thermally altered sandstone is cut by tuffisite. East and west of the stream, the position of the inner fault is defined by sudden changes in rock type, vegetation and slope. The outer member crosses the stream at about 140 m altitude. The rocks of the Laimrigh Shale Member between the branches of the Main Ring Fault strike parallel to the faults and dip steeply to the NNE. This area of dark sedimentary rocks stands out exceptionally clearly on air photographs. Within the sedimentary rocks, short, elongate patches (<1 m wide) of partially melted gneiss occur east of the stream [NM 3945 9833].

The most complex development of the Main Ring Fault system is found at Allt nam Bà and on the south-east slopes of Beinn nan Stac [NM 403 937]. In this area (Figure 54), successively younger outer, central and inner components are recognised, inclined to the west at various angles (Smith, 1985, fig. 1). The steep outer fault brings baked Lewisian gneiss up against sandstone and shales of the Torridonian Allt Mór na h-Uamha Member. The steep central fault downthrows Jurassic beds and altered amygdaloidal basalts (equated with the Palaeocene Eigg Lava Formation) against the gneiss to the east. The inner fault is reversed. Torridonian strata are brought over the Jurassic beds and the Palaeocene lavas on a fault plane that is inclined at about 45° W (Smith, 1985, fig. 3). Further west-dipping reversed faults cut the Torridonian beds on the higher slopes of Beinn nan Stac. The Jurassic beds and the Palaeocene lavas on the slopes north-east of Dibidil at [NM 398 934] form a tectonic inlier, defined by two reversed faults (Figure 54). At Dibidil, the near-vertical outer fault crops out in cliffs at either end of the bay. About 1.5 km WSW of the Dibidil River, steeply dipping, distorted shale and siltstone occupy a lenticular area, about 750 m in length, between the inner and outer faults.

The Torridonian beds just outside the Main Ring Fault generally deviate from the regional 15–20° WNW dip. From Camas na h-Atha to a point east of Coire Dubh, the beds are upturned and dip steeply away from the fault system. This is strikingly demonstrated in the massive dip-slope slabs on Sgorr Mhór [NM 306 999], where the beds dip at >50° NW, and on the south side of Kinloch Glen where dip-defined slabs slope steeply to the north. The progressive change in strike and dip can be observed in a traverse up Allt Slugan a'Choilich, from Kinloch Castle to the hydroelectric dam. By contrast, the Torridonian beds between Allt Mór na h-Uamha and Allt nam Bà., in eastern Rum, are downturned and show a progressive increase in dip down towards the central complex. Farther south, from Dibidil to Papadil, the beds outside the Main Ring Fault are not disturbed, apart from a shallow ENE-trending syncline about 700 m due east of Papadil Lodge [NM 3657 9222] (Figure 53).

Silty and shaly rocks are commonly folded and faulted where they occur within the Main Ring Fault. An anticline occurs in Coire Dubh (Dunham and Emeleus, 1967, fig. 2) and on the south-east side of Beinn nan Stac, shales are thrown into steep folds (Harker, 1908, fig. 5) that strike nearly parallel to the faults. The folds within the Main Ring Fault zone indicate compression, and those outside the Main Ring Fault, north of the Western Granite and the Northern Marginal Zone, indicate doming around the central complex. In the east there is a suggestion of a rim-anticline margining the central complex, and in the south a shallow rim syncline (Figure 53).

The structural significance of the Main Ring Fault was first recognised by Bailey (1945, 1956) who realised that the Lewisian gneiss and basal Torridonian beds of Rum occur entirely within the fault and, furthermore, that they had been brought up against younger members of the Torridonian succession. Bailey (1945) suggested that there had been central uplift of the order of 1–2 km. Subsequently, Emeleus et al. (1985) and Smith (1985) demonstrated that central subsidence of a similar amount had taken place, juxtaposing Jurassic beds and Palaeocene lavas against Precambrian gneisses and sandstones. Furthermore, it appears likely that this subsidence was accompanied by caldera formation and the intrusion and effusion of silicic magma. At Dibidil, Errington (personal communication, 1989) has found extensive megabreccias and bedded breccias derived from Torridonian rocks; these too are thought to have formed when the caldera walls collapsed. In the Beinn nan Stac area, further central uplift caused Torridonian rocks to be pushed outover the Jurassic beds and Palaeocene lavas, but the full significance of the faulting and folding in this ground and in the Dibidil area is not yet clear.

The presence of rocks of the Sgor Mhór Sandstone Member of the Torridon Group in southern Rum implies dextral movement, possibly accompanied by an easterly downthrow on the Long Loch Fault, and/or considerable southerly downthrow on the Main Ring Fault west of Dibidil. Movements on the Long Loch Fault could predate the Central Complex.

Long Loch Fault

The Long Loch Fault extends from Kilmory [NG 360 040] south to Inbhir Ghil [NM 358 926] (Figure 53). North of the central complex the broad valley of the Kilmory River has been eroded along the fault line. Within the central complex the course of the fault is narrower, but clearly defined, as in the 20–40 m-wide trench south of the Long Loch. Components of the Central Complex are displaced dextrally by 700–800 m on the fault. Although it forms a major structural feature on Rum, the fault has not been identified definitely during marine surveys to the north and south of the island.

As mentioned above, the fault may have had a history of pre-Palaeocene movement. The general parallelism between the components and structures in the Central Intrusion and the Long Loch Fault is quite pronounced. It has been suggested that the fault is on a major, long-lived zone of weakness which controlled the feeder channels during intrusion of the magmas responsible for the Rum Layered Ultrabasic Suite (Chapter 7; McClurg, 1982; Volker, 1983; Emeleus et al., 1996b). However, the final movement on the Long Loch Fault clearly postdated the intrusion of the central complex.

Structure of the Torridonian strata and Lewisian rocks within the Rum Central Complex

The overall structure of the gneisses and sedimentary rocks within the Main Ring Fault is largely obscured by the Palaeocene intrusions. There is, however, a pattern to their distribution; the principal areas of gneiss occur west of a line from the Priomh-lochs [NM 369 986] to Fiachanis [NM 370 940] (Figure 3), whereas east of this line, gneiss outcrops are always intimately associated with uplift on the Main Ring Fault. Unconformable relationships between gneiss and sedimentary rocks are found near the Priom-lochs and in Fiachanis, where the unconformity dips ENE and ESE respectively. Rocks of the basal Torridonian Fiachanis Gritty Sandstone Member form most of the outcrops, except towards the east where the Laimhrighe Shale Member crops out on Beinn nan Stac, east of Coire Dubh, and in upper Fiachanis. The general distribution of the outcrops suggests that a large block of gneiss, unconformably overlain by Torridonian strata, was tilted to the east as it was elevated to its present level, possibly at the same time as doming occurred in Stage 1 of the Rum Central Complex.

Other structures

Other displacements have occurred within the central complex. The Cnapan Breaca effusive rhyodacites and tuffs have been tilted as a block to the south, while in upper Dibidil the steep dips of rhyolite with highly attenuated fiamme suggest that deformation and faulting may have accompanied rhyolite extrusion (Chapter 5). However, the most convincing examples of tectonic disturbances contemporaneous with magmatic activity come from the numerous breccias with distorted, fragmented and smeared out layered ultrabasic rocks in the Central Intrusion (Chapter 7, (Plate 17), some of which formed as fault scarps collapsed within the magma chamber.

Other faults

Rum

There is a mismatch of Torridonian strata on the north and south sides of Loch Scresort: the Laimhrig Shale Member does not crop out on the north side; the lithology of the Scresort Sandstone Member on the south of the loch, between Kinloch and the fault, is different from that on the north side; and furthermore, there is no continuation of the Mullach Ard Fault north of the loch. This suggests that there may be an east–west fault concealed by the loch. The fault could continue west to the Long Loch Fault and easily escape detection in the peaty ground near the Kinloch River. Although it is speculative, a fault may also follow Glen Shellesder. This could be a continuation of a fault in Kinloch Glen, from which it would be offset dextrally by about 600 m, approximately the known post central complex movement on the Long Loch Fault. A dextral displacement of about 300 m on the Loch Scresort Fault is necessary at the entrance to the loch to account for the non-appearance of the fine-grained rocks of the Torridonian Allt Mór na h-Uamha Member and the Mullach Ard Fault exposed on the southern shore. The possible positions of these faults are indicated in (Figure 53).

Numerous small faults radiate from the central complex (Figure 53). These displace the Main Ring Fault at Dibidil (Figure 54), affect bytownite-troctolites and peridotites on the north slopes of Barkeval, and displace members of the Central Intrusion on the coast southwest of Runsival. The radial faults may be continued in the Torridonian rocks north of Kinloch Glen where several features suggest faulting, but the absence of distinctive sedimentary horizons makes proof difficult. On Fionchra [NG 335 005], the young Canna Lava Formation flows are systematically downthrown to the north-west by two NE-trending faults, while on West Minishal [NG 349 003], a curving, north–south to NNW-trending fault downthrows the lavas and conglomerates by at least 90 m to the west.

Eigg and Muck

Two subparallel NW-trending faults cross Eigg from Laig Bay [NM 470 880] to Kildonnan [NM 490 850]. The western, Laig Gorge Fault, was reactivated after eruption of the Palaocene lavas. It has a variable north-east downthrow, from 40 m at Kildonnan, to about 20 m at Laig Gorge (Figure 55). The eastern fault downthrows between 10–15 m to the west. Further subparallel faults probably cut the Eigg Lava Formation between the south coast and Laig Bay, but the lack of distinctive horizons in the lavas makes identification difficult. The lavas of Eigg have a fairly uniform dip of 3–5° WSW and are unfolded. The lavas of Muck have a gentle (<3°) NNW dip and are also largely unfolded. Several minor NNW-trending faults cut the Muck lavas; these have throws of a few metres equally to the south-west or north-east (Figure 55). The fault on the south-west side of Torr nam Fitheach [NM 408 794] throws down about 30 m to the east, bringing the basal Palaeocene tuffaceous sediments against the Duntulm Formation.

Canna and Sanday

The Palaeocene Canna Lava Formation on both these islands has been faulted and gently folded. All dips are low (<3°). In the extreme west of Canna, easterly dips change in about 1 km to a more north-easterly direction and the generally north-west dip on Sanday changes to a NNW direction in eastern Canna (Figure 56). The low ground separating west and east Canna, around Tarbert [NG 238 054], coincides with a change of dip from NNE to NNW but there is no faulting across this ground although faults are present to the west and east (Figure 56).

Several NNE- to NNW-trending faults in western Canna downthrow to the west and have an aggregate vertical displacement of over 100 m. The NNE to NE faults in eastern Canna downthrow in various directions and divide the lavas into several blocks. East of Compass Hill [NG 280 060], a NNW-trending fault downthrows about 6 m to the east. At this point, the fault produces a marked feature, as does its continuation across the eastern end of Sanday (Figure 56).

Faulting between Rum and Eigg

Early investigations of the geology of the Sea of the Hebrides were interpreted to show the Camasunary Fault, on southern Skye, continuing south to join up with the fault between Palaocene basalts and Precambrian rocks east of Coll, and extending to Skerryvore (Binns et al., 1974, fig. 2). However, the seismic data are poor and clear geophysical evidence is lacking that there is a fault boundary to the Palaeocene basalt lavas between Coll and Rum, or that the boundary of the Mesozoic rocks east of Rum is a fault; hence the faults at Skerryvore and Coll are not shown linked with the Camasunary Fault on Skye on (Figure 1) (see also BGS 1:250 000 sheets Tiree and Little Minch).

The striking contrast in the geology either side of the 6 km-wide Sound of Rum requires explanation (cf. Frontispiece). If the generalised strike (NW–SE) and dip (c.5° SW) of the Eigg Lava Formation is continued northwest to Rum, the base of the lavas would be at sea level at Stac na Faoilean [NM 407 932], rising north to the contact of the Layered Suite at about 200 m altitude near Allt Mór na h-Uamha. Lavas would crop out to the south and Mesozoic strata to the north. The latter would include beds up to and above the Kilmaluag Formation. The absence of these lavas and sedimentary rocks on Rum outside the Main Ring Fault could result from faulting in the Sound of Rum, doming over the Rum Central Complex, or a combination of these two. Doming of the country rocks outside the Main Ring Fault has occurred in eastern Rum but it is not as pronounced as on the north side of the central complex; for example, there is only slight deflection of the contact between the Torridonian Allt Mór na h-Uamha and Laimrhig Shale members. Furthermore, the strata nowhere dip away from the Main Ring Fault, the general WNW dips are less that the regional dip (commonly 5–15° as against 20° or more), except close to the margin of the Eastern Intrusion where dips steepen to as much as 50° W, suggesting the presence of a shallow rim anticline around this part of the igneous complex. Doming would therefore appear inadequate as the sole explanation of the present relationships either side of the Sound of Rum. An important clue to these relationships is provided by the rocks within the Main Ring Fault system where flows of the Eigg Lava Formation rest unconformably on the Lower Jurassic Broad-ford Beds (Chapter 8; Smith, 1985). Compared with Eigg, a considerable thickness of Jurassic strata is absent and this is closely comparable to the relationships either side of the Camasunary Fault in southern Skye. The most likely explanation of the relationship between Rum and Eigg is that during the Upper Jurassic (post-Staffin Shale Formation), faulting occurred on a southern extension of the Camasunary Fault which acted as a western boundary fault to the Mesozoic basin beneath Eigg. The Middle and Upper Jurassic rocks were eroded off the upthrown western area prior to eruption of the Palaeocene lavas. Renewed movement on the Camasunary Fault (in the same sense) also affected the lavas, as did the (?later) doming over the central complex. From the present contrasting geology either side of the Sound of Rum, it appears inescapable that there is a major fault or faults, which downthow to the east, on the line of the Sound of Rum. These faults are logically the southern continuation of the Camasunary Fault. The relationships also appear to require the lavas to have been faulted, which is precisely what is found east of Coll.

Chapter 11 Geophysical investigations and age determinations

Gravity anomalies over Rum and the adjacent islands

Rum is notable for having one of the highest values of the Bouguer anomaly in the British Isles. Higher values are found only on Barra and over the offshore Tertiary central complexes of Blackstones and St Kilda.

Onshore gravity observations on Rum, Canna, Sanday and Eigg were first made by McQuillin and Tuson (1963) at 66 stations, 40 of which were on Rum. Coppin (1982) added a further 31 stations in the more mountainous parts of Rum. Surface ship gravity lines in the adjacent sea made by the British Geological Survey are included in the gravity map of the Sea of the Hebrides (Binns et al., 1974, fig. 5). The results from these sources are presented in (Figure 58)." data-name="images/P936658.jpg">(Figure 57).

Rum is dominated by an almost circular positive anomaly over the central complex. The background value in the adjacent sea areas and islands is between 10 and 20 mGal, with the lower values being associated with Mesozoic sedimentary rocks overlain by Palaeocene lavas on Canna, Sanday, Eigg and Muck and on the Canna Ridge (Binns et al., 1974, fig. 2). The maximum observed anomaly of 77 mGal indicates that the amplitude of the local positive over Rum is at least 57 mGal. The half-width radius of the anomaly is about 4.7 km and there is no obvious direction of elongation or major local inhomogeneity as is present in the Skye and Mull central complexes (Bott and Tuson, 1973; Bott and Tantrigoda, 1987). The Bouguer anomaly gradient is exceptionally steep, locally reaching 13 mGal/km across the northern edge of the central complex.

McQuillin and Tuson (1963) modelled the Rum anomaly using vertical cylinders and truncated cones, inferring a density of 3050 to 3100 kg/m3 and a 15 km depth extent. Some 2.5D models in a NE–SW direction across the central complex were constructed by Coppin (1982). A new set of interpretations of the deep structure along an approximately north–south line have been modelled for this account (Figure 58).

Rock densities and other assumptions made in modelling the anomaly

Densities of the three main groups of rocks are required for the interpretation of the north–south profile A–A′ (Figure 58)." data-name="images/P936658.jpg">(Figure 57). Along this profile, the rocks of the central complex are mainly ultrabasic. The adjacent country rocks are Torridonian sandstones with the Applecross Formation mainly cropping out at the surface. Evidence from within the central complex shows that the Torridonian sandstones overlie a basement of Lewisian gneisses (Chapters 2 and 3; Bailey, 1945). Torridonian strata form the solid rock of the sea bed along the marine sections of the profile A–A′ (Binns et al., 1974, fig. 2).

The density profile of a typical ultrabasic unit in the Eastern Layered Intrusion of the central complex is given by Wager and Brown (1968, table 18) from which an average density of 3157 kg/m3 has been calculated. This corresponds to feldspathic peridotite which is the most common ultrabasic rock at outcrop on Rum (Chapter 7). Tuson (1959) determined a density of 2650 kg/m3 for the Applecross Formation in Skye, which is used here. The average density of measured Lewisian rocks from Skye is 2790 kg/m3 (Tuson, 1959), closely similar to the value for Lewisian pyroxene-granulites in Sutherland (Bott et al., 1972). A value of 2800 kg/m3 has been adopted for the sub-Torridonian basement rocks.

Interpretation

The close correspondence between the Rum Central Complex and the associated positive anomaly indicates that it is the dense igneous rocks of the complex which give rise to the anomaly. If the anomaly is treated as a vertical cylinder of 4.7 km radius (= the half-width) that extends to infinite depth, then a simple calculation shows that a minimum density contrast of 290 kg/m3 is required. Even when allowance is made for up to 2 km thickness of Torridonian strata, the density must be at least 260 kg/m3 greater than that of the Lewisian basement. This demonstrates conclusively that an exceptionally high density is required to account for the anomaly, in agreement with McQuillin and Tuson (1963).

The deep structure has been modelled along the line A–A' (Figure 58)." data-name="images/P936658.jpg">(Figure 57). The Bouguer anomalies along this line have been reduced using the Bouguer density values given above. Corrections were also made for the effect of steep gradients on anomalies at higher altitude stations.

Neither the Long Loch Fault (Chapter 10) nor the small ultrabasic tongues that extend north from the central complex appear to affect the nearly circular Bouguer anomaly and neither should prejudice the interpretations. The substantial Western Granite (Chapter 5) does not appear to affect the gravity anomaly. The data over this area are sparse but the granite is clearly grossly subordinate to the dense rocks which form the bulk of the complex.

Two models are presented. In one ((Figure 58)a), Rum 1) a uniform density of 3150 kg/m3 (corresponding to feldspathic peridotite) is assumed and gives a body extending to 8 km depth. A composite model ((Figure 58)b), Rum 2), which assumes peridotite (3200 kg/m3) at shallow depth underlain by gabbro (3000 kg/m3), indicates that the dense peridotite extends to at least 2.5 km , when the overall depth to the base of the body is 16 km.

Discussion and conclusions

The depth of the dense igneous body beneath the Rum central complex will lie between 8 and 16 km, depending on whether it is dominantly peridotitic or gabbroic in composition. All modelling of the anomaly produces steep, outward-dipping contacts, with steeper dips for the smaller assumed density contrasts ((Figure 58)a) and ((Figure 58)b).

Similar steep, outward-dipping contacts are inferred for many of the other Tertiary central complexes in the Hebridean Province for which gravity data are available (e.g. Skye — Bott and Tuson, 1973; Mull — Bott and Tantrigoda, 1987; St Kilda — Bott et al., 1979).

Although ultrabasic rocks predominate at surface outcrop within the Layered Suite, gabbro is common and basaltic magmas were available throughout the evolution of the central complex (Chapter 7). The anomaly could therefore result from a variety of rock types; realistically between the extremes of a uniform body of feldspathic peridotite or gabbro and a composite mass with a dense peridotite cap and a deep gabbroic root. The deep structure of Rum is more comparable with the Blackstones and St Kilda central complexes (McQuillin et al., 1975; Bott et al., 1979) than with Skye and Mull where a density contrast of 200 kg/m3, corresponding to average gabbroic compositions, satisfies the anomalies (Bott and Tuson, 1973; Bott and Tantrigoda, 1987).

The nearly circular gravity anomaly and the inverted flower-pot shape of the dense mass forming the bulk of the Rum Central Complex are strongly supportive of ring fracturing and cauldron subsidence as the mechanism of emplacement; furthermore, arcuate faulting has been important during the development of the complex (Chapters 5 and 10). This raises the question as to how such a large body of dense rocks can be emplaced into the upper crust in this way. The explanation appears to be that the magma has a significantly lower density than the solidified rock; a picritic basaltic magma at 1400°C would be at least 50 kg/m3 lower in density than the Lewisian upper crust of the region (Bott and Tantrigoda, 1987). Ring fracturing and other types of stoping are thus mechanically feasible even for the dense complexes of Rum, Blackstones and St Kilda.

Wager and Brown (1968) suggested that the layered ultrabasic rocks of Rum formed a solid block which was forced upwards at least 1 km into its present location along the ring fault. Contrary evidence presented elsewhere (Chapter 7; Emeleus, 1987; Robinson and McClelland, 1987; Greenwood et al., 1990) indicates that these rocks crystallised at about their present level in the central complex. This latter conclusion is supported on mechanical grounds, in that even a picritic basaltic magma extending over a maximum depth range of 16 km would not give rise to high-enough excess pressure to cause isostatic uplift of more than about 300 m at most.

The fairly low Bouguer anomaly on and near Eigg (Figure 58)." data-name="images/P936658.jpg">(Figure 57) is consistent with a significant thickness of Mesozoic sedimentary rocks, but the increasing gravity gradient, from east to west, is in the opposite sense to that expected from a wedge of sedimentary rocks thickening towards the Camasunary Fault (Binns et al., 1974, fig. 2). This indicates overprinting by the Rum Central Complex. The fairly uniform values of gravity north and south of Rum conform with a NNE-trending ridge of Precambrian rocks, and the NNW decrease in gravity across Canna and Sanday is consistent with an increasing thickness of underlying Mesozoic sedimentary rocks towards the Little Minch Basin (Binns et al., 1974, fig. 2).

Aeromagnetic observations, palaeomagnetisation and radiometric age determinations

Aeromagnetic observations

The aeromagnetic map of the Small Isles (Figure 59) shows anomalies over Canna/Sanday, Rum and Eigg/ Muck. These are described and discussed by Binns et al. (1974, pp.4–7, fig. 4). The anomaly at Canna is part of a belt of variable magnetisation that coincides with the Canna Ridge and is attributable to Palaeocene lavas of the Canna Lava Formation. There is a large anomaly of broadly arcuate outline over southern Rum and this also extends some way south of the island. It is caused by the Rum Central Complex and is broadly similar to anomalies over other Tertiary central complexes (cf. Binns et al., 1974, fig. 4). The strong arcuate anomaly at Harris (Figure 59) is probably caused by relatively magnetic gabbroic rocks in the Harris Bay Member of the Western Layered Intrusion (cf. Coppin, 1982, p.35). The Palaeocene basalts of the Eigg Lava Formation and the regional NNW-trending Muck dyke swarm contribute to the anomalies over south Eigg, Muck and the intervening sea. The NNW to NW trend in the anomalies on and near Muck are due mainly to the dyke swarms, which may also account for the south-eastern linear prolongation of the Rum anomaly. A small NNW-trending anomaly, not shown on (Figure 59), also occurs over the minor dyke swarm at the Bay of Laig, Eigg.

Palaeomagnetisation and radiometric age determinations

Palaeomagnetic measurements have been made on the Tertiary igneous rocks of Canna and Rum (Dagley and Mussett, 1981) and Eigg and Muck (Dagley and Musset 1986). These are summarised and discussed together with results from the British Tertiary Igneous Province as a whole by Mussett et al. (1988). The results for the Small Isles, together with radiometric age determinations, are reproduced as (Figure 60).

There are few radiometric age determinations for the igneous rocks. Using the 46Ar-39Ar step heating method, the Eigg Lava Formation gives ages about 63 Ma (Dagley and Mussett, 1986) <reference>Sanidine crystals from two tuff layers on Muck, at Port an t-Seilich [NM 4183 7842] and Port Mór [NM 4263 7899] (Emeleus et al., 1996a) have yielded step-wise Ar release plateau ages of 62.8 + 0.6 (2 sigma) and 62.4 + 0.6 Ma using the laser 40Ar/39Ar technique (Pearson et al., 1996). These determinations, from beds near the base of the Eigg Lava Formation, provide a definitive age for the start of igneous activity in the Small Isles and, from stratigraphical considerations, they provide precise maximum contraints on the posssible age of Skye magmatism, and possibly most of the British Tertiary Volcanic Province.</reference> and an age of 60.1 Ma has been obtained from the Canna Lava Formation on Rum (Mussett, 1984).

Hitherto unpublished K-Ar whole-rock age determination and stable isotope data on the rhyodacite and granophyres are listed in (Table 11) (A C Dunham, personal communication, 1980). The apparent age for the rhyodacites range from 60.0 to 58.2 Ma, with a weighted mean of 59.2 Ma. This spread is probably accounted for by the inherent errors in the method (±2 Ma two sigma). There is no obvious explanation for the high age of sample A.429. A plot of 40K/39Ar against 40Ar/38.Ar for the five rhyodacite samples analysed by BGS (Table 11) defines an isochron of 60 Ma, with an initial 40Ar/39Ar ratio of 281 ± 25. This compares with the present day ratio of 295 ± 5, and suggests that the rocks have remained a closed system with respect to argon retention. New K-Ar age determinations on the granophyres give a range from about 55.3 to 58.2 Ma, and an age of 58.4 Ma was obtained from the Western Granite using the 40Ar-39Ar step heating method (Mussett, 1984). While the lower apparent age of the granophyres is consistent with the field evidence (Chapter 5), the granophyres generally show amphibole after pyroxene, the development of chlorite as an alteration product, and drusy cavities. Thus, the granophyre ages may all be too low. Analytical data on rubidium and strontium in the rhyodacite suggest a range of initial 87Sr/86Sr ratios of 0.714 to 0.716. These values in turn suggest that there may have been substantial incorporation of crustal material in the acid magma from which these rocks crystallised. In particular, these initial ratios are similar to the initial ratios in partially melted Lewisian gneiss (0.719; A C Dunham, personal communication, 1980).

All the igneous rocks have reversed magnetisation (R) with the exception of a few post-Canna Lava Formation dykes in north-west Rum and Canna, and four dykes on Muck which are normally magnetised (N). The magnetic sequence is R-N-R-N and extends from Chron 27R to Chron 26 in the youngest, post-Canna Lava Formation dykes. The reversed magnetised Sgurr of Eigg Pitch-stone (Chron 23R, Dagley and Mussett, 1986; 52.1 Ma, Dickin and Jones, 1983) is the youngest igneous event in the Small Isles and amongst the youngest in the Hebridean Province. Apart from the Sgurr pitchstone, the igneous activity was confined to about 5 Ma, and the Rum Central Complex probably was active for less than 2 Ma (Figure 60).

Robinson and McClelland (1987) determined the palaeomagnetism of the Applecross Formation (Torridon Group) sandstones from a traverse between Camas Pliasgaig c. [NG 400 028] and Loch Gainmich [NM 380 987] (A-B on (Figure 59). At a distance from the central complex, the primary remanence in the sandstones is carried by both magnetite and haematite. Robinson and McClelland (1987, p.473) suspect that 'only magnetite carries a truly primary (detrital?) remanence; the haematite probably formed diagenetically (largely by oxidation of detrital magnetite) at some time shortly after deposition'. Near to the central complex the original magnetisation of the sandstone was reset by heat from the Palaeocene intrusions; this occurred after deformation of the country rocks during emplacement of acid magmas during Stage 1 (Table I; Chapter 5). Temperatures greater than 700°C were calculated to have occurred up to 1 km outside the Main Ring Fault, caused by the ultra-basic rocks at about 1300°C. They suggest that the shape and size of the thermal aureole indicate that there can have been only limited uplift of the mafic intrusives, and that since the size of the aureole is greater than expected it is likely that there are plugs at depth that provided an additional heat source. This accords with the numerous plugs mapped north of the central complex (Figure 27) for the localities of plugs within the Northern Marginal Zone, (Figure 21) for the Southern Mountains Zone and (Figure 35) for detail within the Layered Suite. The numbers 1-13 refer to the localities where dyke orientations have been measured (see (Figure 28))." data-name="images/P936601.jpg">(Figure 26).

Chapter 12 Pleistocene and Recent

Introduction

The first detailed description of the effects of the Pleistocene glaciations in the Small Isles was by Harker and Barrow in the first edition of this Memoir (Harker, 1908). Their main conclusion, that the islands were crossed at some period by ice from the Scottish mainland, remains unchallenged, and their evidence for a later, local glaciation on Rum has been generally sustained and amplified (Charlesworth, 1956). Recent geomorphological work has been mainly concerned with the recognition of a 'pre-glacial' marine platform and other platforms on Rum, the examination of periglacial phenomena, and studies of the deep decomposition of some of the ultrabasic rocks (Ball, 1964a; Godard, 1965; Peacock, 1976; McCann and Richards, 1969; Ryder and McCann, 1971; Ballantyne and Wain-Hobson, 1980; Ballantyne, 1984).

The Small Isles lie between 10 and 15 km west of the mainland of the western Highlands, which constituted the principal local centre of ice accumulation during the last (Late-Devensian) glaciation (26 000–13 000 BP). That this mainland-derived ice was responsible also for the widespread glaciation of Rum, Eigg, Muck and Canna is supported by the presence of erratics of mainland rocks on these islands. At the time of its maximum development the late-Devensian ice sheet generally moved west or north-westwards across the islands. On its retreat it left deposits of till and morainic drift and, very locally, bedded sand and gravel were laid down near the retreating ice margin. As the ice retreated, raised gravel beaches formed on the isostatically depressed islands up to 30 m above OD. Lower beaches were formed as relative sea level fell and isostatic readjustment occurred. Periglacial features such as stone polygons were formed during the cold climate conditions which accompanied the retreat of the main late-Devensian ice sheet on Rum and probably also during the local reappearance of corrie and valley glaciers during the Loch Lomond Stadial (11 000–10 000 BP). No deposits dating to the relatively warm Windermere (Late-glacial) Interstadial (13 000–11 000 BP) have yet been identified on the islands. As the ice wasted away, extensive rock falls and landslips took place on Eigg on the flanks of the lava plateau and similar mass movements have continued locally to the present day. A second rise of sea level relative to the land occurred in Post-glacial times resulting in a well-marked raised beach a few metres above OD.

Pre-late Devensian coastal features

Rock benches, thought to be partly or entirely of marine origin, occur at various levels up to 38 m above sea-level around the coasts of Rum and Canna, but have not been identified with certainty on Muck. On Eigg, the low ground to the west of the great lava escarpment (Plate 1) may have originated as a marine beach, but if so it has been greatly modified by glacial erosion combined with landslipping (p.136). The best-developed and most widespread bench on Rum is glaciated, and certainly predates the maximum of the last glaciation. As such, it is analogous to the 'pre-glacial raised beach' at 30–46 m OD around the coasts of Mull and Ardnamurchan (Wright, 1937), though not necessarily of the same age. Similar features have been reported from the coasts of Skye and the neighbouring mainland (McCann, 1968).

Rum

A raised cliff-line, fronted by a gently shelving platform is widespread on the south-west and east coasts of the island. It is best developed where it is incised into the granophyre of the Western Granite on the south-west coast (Plate 10); McCann and Richards, 1969) but it is also a clearly marked feature of the coast in the east between Bàgh na h-Uamha and Dibidil (see Frontispiece) and in the neighbourhood of Loch Scresort. South of Harris the shelving platform is 150 m wide, but on the east coast the width rarely reaches more than 60 m. The old cliff and bench can be traced around the south corner of the island as a discontinuous ledge, commonly talus-covered as on the south-west coast (Plate 10). North of Creag na h-Iolaire [NG 409 025] on the north-east coast, the raised cliff line is truncated by the present coast. Apart from one locality about 2 km north-east of Guirdil and a more doubtful bench west of Kilmory (Figure 61) the platform seems to be missing from much of the north coast.

The bench and cliff are certainly older than the last widespread glaciation of the island but no maximum age can yet be given. South-west of Wreck Bay [NM 309 981] and at Harris the platform is overlain by Late-glacial marine gravels and by drift including till (McCann and Richards, 1969; Peacock, 1969), and adjacent to Sgeir a'Mhaim-ard [NM 413 938] on the east coast the inner limit of the platform is smoothed and striated by ice-action. At Harris the platform is cut into ultrabasic rocks that are considerably decomposed below the till cover.

At any one locality the height of the notch at the back of the beach shows some variation. This is particularly pronounced where the platform intersects gently dipping Torridonian sandstones, for instance at Laimhrig [NM 419 968] where the variation in height is as much as 4 m. Rock benches originating in other ways occur locally above or below the marine platform and may be confused with it. Examples of these may be seen north of Loch Scresort where there are dip and strike features in the Torridonian. At Bàgh na h-Uamha [NM 421 973] there are several benches between high-tide level and 100 m OD, controlled by features of the bedrock geology such as gently dipping dolerite sills, low-angle joints, and hard beds of sandstone.

Information concerning the heights of marine benches and raised shorelines is at present scanty (Figure 61) but nevertheless suggests that there are significant differences in level around the island of Rum. The back notch of the platform is at its lowest (approximately 12 m above mean sea level) in the north-west where cut in the Triassic rocks north of Guirdil. Along the south-west coast it is generally about 19–23 m above mean sea-level, rising to 30 m along the south-east coast and a little below 38 m at Loch Scresort. These variations in height around the island have been explained either on the basis that the platforms and cliffs are composite features formed by marine erosion at different sea levels (McCann, 1968; McCann and Richards, 1969) or that they are the consequence of tectonic warping (Peacock, 1969). A third possibility, suggested by the grouping of heights around 23 m (south-east coast), 30 m (south coast to Bàgh na h-Uamha) and 38 m (Loch Scresort) is that the platform is faulted by the Long Loch Fault (Chapter 10) and by another fault (or faults) at Bàgh na h-Uamha (Figure 61).

Canna

A rock platform is well developed on the south side of the island, from Rudha Dubh [NG 265 051] westwards to Rubha Sgorr nam Ban-naomha [NG 230 042], and on the north coast from Sloe a'Ghallubaich [NG 255 066] westwards to Garrisdale Point. In detail its form is controlled by the more resistant massive central parts of the gently dipping lava flows, so that its height varies by several metres within a given area. The highest level for the back notch of the platform is between 15 and 18 m above local OD. The platform is overlain in places by raised gravel beaches of Late-glacial age, and the back feature is often concealed by talus.

Glaciation

Rum

Main Late Devensian glaciation

From the distribution of striae (Figure 61) it is evident that the ice impinging on the east coast was split into two streams, one being diverted southwards parallel to the coast, and the other extending west and north-west across the northern half of the island. The picture is somewhat complicated by ice movement associated with the late glaciation referred to below which produced the easterly directed striae on Hallival and Sgurr nan Gillean. The westerly movement of the northern stream is attested by numerous striated surfaces on the more massive beds of the Torridonian. Such surfaces also show crescentic gouges, friction cracks and chatter marks. The basic and ultrabasic rocks south of the Main Ring Fault are strikingly ice-moulded and roches moutonnees abound. At the head of Kilmory Glen the basal layers of ice were diverted abruptly north and south, and a general southwesterly movement of ice in Glen Harris is shown by numerous striae and plucked surfaces. The ice crossed Glen Duian almost at right angles, giving a quarried appearance to the southern slopes of Ard Mheall [NM 340 970]. No striations have so far been recorded from the Western Granite. On the eastern side of the island the ice probably divided at a locality [NM 415 962] just south of Fearann Laimhrige.

Erratics of mica-schist and garnetiferous gneiss foreign to the island are fairly numerous along the east coast, occuring at altitudes up to 200 m, and on Barkeval where they are found as high as 550 m above OD. Farther west such boulders occur in the glens from Glen Shellesder northwards and at about 530 m above OD on Ard Nev. Boulders of the local Torridonian sandstone occur on the basalt plateau of Orval [NM 336 995] at about 500 m above OD and in Glen Harris [NM 358 963], and peridotite has been transported to the head of Glen Guirdil. These facts suggest that, at the time of the late-Devensian maximum, the high ground in the west of the island was entirely covered by ice and that perhaps only the Hallival–Ruinsival–Sgurr nan Gillean triangle in the south supported enough local ice to divert the westwards flow of ice from the Scottish mainland.

Glacial deposits (till, morainic drift and glaciofluvial gravels) attributable to the late-Devensian ice sheet are widespread on Rum, but are restricted in area. A stiff silty till with somewhat featureless topography occurs in the Kinloch and Kilmory glens. In the latter, at a locality [NG 365 010], there is brown till, 1.5 m thick, with boulders and cobbles mainly of Torridonian rocks and rarely Moine psammite. Similar till overlies deeply weathered rock at a nearby quarry [NG 363 008]. Brown till overlain by peat covers a considerable area of the Monadh Mhiltich [NM 350 996]. In the Glen Guirdil burn there are exposures of up to 3 m of clayey till with small angular clasts and at Guirdil itself the cliff backing the Postglacial beach shows over 3 m of till below a similar thickness of gravel. South of Camas na h-Atha [NM 303 988], hard silty till with numerous boulders of granophyre is seen in the burn. East of Harris upwards of 4 m of raised beach gravel can be seen in the side of the Glen Duian River gorge, resting on hard till which in turn rests on decomposed ultrabasic rocks. The till sheet mantling the lower slopes of Glen Duian is probably derived from the north-east since the clasts in the lower part of the glen are mainly of basic and ultrabasic rocks. At the footbridge across the Abhain Raingail [NM 345 955] nearly 4 m of till rests on decomposed ultrabasic rock.

In the middle section of Glen Harris there is an area mapped as morainic drift, the deposit varying from poorly sorted to well-sorted bouldery sand. Part of the material about the mouth of the Abhainn Sgathaig [NM 362 963] is disposed in ridges with a marked north–south orientation and a section at one locality [NM 361 959] shows a ridge made of 5 m of bedded sands and gravels with boulders overlying hard till. The ridges are orientated at a high angle to the local direction of ice movement as indicated by abundant striae and roches moutonnees, and may have been deposited as crevasse fillings. A gravel ridge with kettle holes, possibly an esker, is present at about 180 m above OD west of Loch Fiachanis, and east of the loch gravelly moundy drift is crossed by the small terminal moraine of a later corrie glacier (see below). Glaciofluvial deposits are restricted to Guirdil, where the surfaces of terraces fall from about 60 m above OD to the more or less contemporary Late-glacial beach at about 30 m above OD.

Late Glaciation

The mountains of the southern half of Rum show evidence of a local valley and corrie glaciation which succeeded the main Late Devensian glaciation and probably occurred during the Loch Lomond Stadial. Peacock (1976) provided evidence for seven such glaciers; Ballantyne and Wain-Hobson (1980) suggested an additional five, bringing the total to 12. The evidence of relative age is particularly well seen at the entrance [NM 410 967] to Coire nan Grunnd where the Kinloch–Dibidil path crosses a block moraine derived from a glacier originating on the east side of Hallival and Askival. The relationship of this moraine to the older glacial features can be seen above the path [NM 409 965] where north-westerly directed striae are crossed by later NNE-trending striae (Figure 61). In Glen Dibidil the local ice has left a moundy deposit of loose poorly sorted gravelly drift with local stones, and till can be traced below 23 m OD, i.e. below the height of the Late-glacial storm beach at Harris.

In central Rum the evidence of local glaciation is particularly striking in Glen Harris, where the bouldery unsorted morainic drift in the upper part of the valley is partly arranged in linear ridges up to. 1.5 m high and 4 m across which are aligned between east–west and ESE–WNW, in the direction of ice movement (cf. Peacock, 1967). The former westward extent of the Loch Lomond Stadial glacier is marked by the downstream limit of very bouldery gravel at [NM 372 965]. A series of fresh arcuate moraine ridges occurring on the north-east flank of Ruinsival about 600 m east of Loch Fiachanis may also have been deposited by a small corrie glacier at this time.

In western Rum former corrie glaciers have left fresh-looking moraines on the north side of the Sròn an tSaighdeir–Orval ridge (Figure 61). In the more westerly corrie [NM 317 994] there are traces of flutes and there is a fine arcuate terminal moraine 1–2 m high composed of granophyre blocks. The head walls of both corries are covered in talus. Traces of local glaciation were noted by Charlesworth (1956) on the east flank of Orval. In Kinloch Glen the former presence of a small local glacier confined to the south side of the valley (Figure 61) is recorded by a patch of moundy morainic drift at about [NM 379 998] and by south- to north-trending striae exposed at the roadside nearby.

Eigg

Main Late Devensian Glaciation

Erratics of mainland rocks, mainly various gneisses, are common on the east side of the island, and together with blocks of Torridonian sandstone, have been carried on to the lower parts of the Sgurr of Eigg ridge. The sandstone was possibly derived from the sea floor east of the island, although no Torridonian sandstone has yet been identified in offshore bores between Morar and Eigg. Alternatively, a local source could have been provided by erosion of the conglomerates found intermittently beneath the Sgurr of Eigg Pitchstone (Chapter 8). No mainland erratics have been recorded so far on the northern lava plateau, but several blocks of gneiss have been reported from near Cleadale on the west side of the island. A boulder of typical Moine pelitic gneiss is included in a field wall near the church [NM 473 385]. There is thus little doubt that the whole of the island was covered by ice during the maximum of the last glaciation.

Striated rock surfaces in the southern two-thirds of the island show that the general direction of ice-movement was westerly, at least during the later stages. In the low ground of the central valley the scarp featuring in the lavas has been subdued and rounded off by glacial erosion. Some divergence of the basal layers of ice is indicated around the Sgurr, and the movement through the central valley was towards the north-west.

The glacial deposits on Eigg are in part closely allied to the landslips (Figure 62). Till is sparsely distributed or is absent altogether over most of the island, but at Camas Sgiotaig [NM 472 898] hard grey till with basalt clasts overlies the Valtos Sandstone Formation and is in turn overlain by brown clayey till with boulders and by raised beach gravels. The brown till with boulders, which also covers the hill to the south, is shown as such on the second edition of the 1" geological map, but since it is essentially a less-consolidated deposit than the Camas Sgiotaig till and has a distinctive surface form, it is shown as morainic drift on (Figure 62) and on the 1:50 000 geological map. The ground north-west of Cleadale is also formed of this morainic drift with lava blocks, arranged in low ridges (Figure 62), the trend of which varies from ESE in the south to ENE in the north. It is thought that the moranic debris was plucked from the western edge of the lava plateau by glacial action and laid down as ridges aligned in the direction of ice movement.

At the north-west corner of Eigg, on Blàr Mór [NM 472 904], two major ridges of basaltic debris extend from north-east to south-west. The more easterly (Guala Mhór) is a mass some 60 m high from which there extends to the south-west a sharp-topped ridge of loose debris. The more massive part of the Guala Mhór debris ridge seems to be formed from lava derived more or less bodily by slip or the action of glacier ice from a former northward continuation of the lava plateau. The morainic tail may have been produced by ice flowing round the extreme north of Eigg. The westerly ridge is probably a lateral moraine formed by the same ice mass as it slowly wasted, the basaltic debris in this case being material which fell onto the glacier margin and was then redeposited a short distance to the south-west.

Late Glaciation

Inland, near Cleadale and extending in an arc to Laig, there is a clearly defined area of fresh-looking mounds of basaltic debris (Plate 1). At Cleadale, and again 300 m ESE of Laig there are kettle holes; that at Laig still retaining a lochan. Such debris is too extensive to have been derived from a normal rockfall, and the presence of kettle holes suggests that it was laid down in intimate association with glacier ice. It is probable that the material is morainic debris derived from the lava plateau together with a substantial contribution from rock falls carried across ice preserved on low ground at the base of the plateau. The sharpness of the ridges, which do not seem to have been affected by solifluction, and the fact that the debris near Laig extends to the level of the highest Post-glacial beach, suggests that the glacier ice formed during the Loch Lomond Stadial (Peacock, 1975).

Muck

The following account is partly derived from the first edition of the Memoir. Striated rock surfaces, indicating a north-westerly movement of ice, have been observed at two or three places, and erratics of foreign rocks are especially common in the northern half of the island. Such rocks are chiefly Moine gneisses and pelites with subsidiary pegmatite, granite and Torridonian sandstone (see p.133 for comment about sandstone erratics on Eigg; a source from the Sgurr conglomerates is unlikely in this instance). There are also boulders of granophyre resembling those of Rum and Skye, these being abundant at some places in the north of the island. A white sandstone occurring as blocks on the raised beach on the north side of the island is similar to a sandstone at the Bay of Laig, Eigg.

Trenches recently (1992) excavated for power cables provided several sections through drift. About 200 m north-west of Carn Mhic Asgaill [NM 417 798] over 1 m of peat is underlain by over 1 m of grey boulder clay containing clasts of gneiss, red Torridonian sandstone, quartzite (?Cambro-Ordovician), siltstone with carbonaceous partings, quartz porphyry and spherulitic pitch-stone. Close to the new school at Port Mór [NM 4227 7955], a sloping excavation showed more than 1.5 m of boulder clay resting on basalt lava. At about 10 m above OD the boulder clay was overlain by raised marine beach deposits consisting of cobbles and boulders (to 30 cm diameter) mainly of basalt but accompanied by white (?Valtos) sandstone, quartzite, banded psammite, leucocratic gneiss, amphibolite and granite. At Godfaig House [NM 4212 8032] 0.5 m of sticky orange-brown till is overlain by 30 cm of gravel which provides the well-drained cover at Blàr Mór [NM 421 803]. 300 m south-west, at [NM 4193 8001], sticky grey till contains clasts of basalt, quartz porphyry and quartzite (?Cambro-Ordovician).

Canna and Sanday

Though ice-moulded surfaces are widespread, striae have been observed only on the hillsides above Tarbert, where a movement of ice towards the WSW is indicated. There are, however, numerous erratic boulders of Torridonian sandstone, particularly on the slopes west of Tarbert Bay, these being accompanied by Moine psammite and Lewisian acid gneiss. A block of dark blue-grey limestone similar in appearance to the Cambro-Ordovician of Skye, is built into a sheepfold wall at [NG 264 053], 600 m east of the farm at Rubha na h-Cor. A boulder of cornstone-cemented breccia similar to that of the Triassic rocks of Monadh Dubh, Rum, occurs on the north-east coast of Sanday where loose blocks similar to Jurassic Kilmaluag Formation rocks have also been observed. Dark grey, brown-weathering till is spread thinly but fairly extensively about Tarbert, but only small patches occur elsewhere on the island 200 m ENE of Tarbert, at Na h-Athannan [NG 2402 0549], a low cliff beside the stream exposes 2.5 m of stiff grey boulder clay overlain by coarsely bedded bouldery raised beach deposits. Examination of a sample of till from below the Late-glacial raised beach at Tarbert Bay showed that the sand-size particles are predominantly quartz with very subsidiary basalt. The larger clasts, in decreasing order of abundance, are Torridonian sandstone, other sandstone, basalt, shelly limestone and sandstone, psammite, mica-schist, acid gneiss, vein quartz, and feldspar-porphyry.

Combining the evidence from striae and erratics it is likely that ice flowing in a west or north-west direction impinged on the east coast of Canna and Sanday and was diverted into a more south-westerly path as it crossed the islands, possibly because of the pressure of ice extending south-westwards from Skye.

Oigh-sgeir

Though the reef is devoid of striated surfaces and till there are numerous erratics, chiefly of Torridonian sandstone and conglomerate with subsidiary gneiss, grey granite, pink coarse-grained and other granites, mica-schist, quartzite, felsite, lamprophyre, and white sandstone. Offshore work by BGS suggests that the Torridonian sandstone is probably derived from the sea floor south or south-west of Rum (Binns et al., 1974).

Features of mass movement and associated deposits

Periglacial features, talus

The mountainous tracts of Rum exhibit periglacial features, a few of which are still forming under favourable conditions at the present day. Most, however, are relicts of the very cold climate at the end of the main Late-Devensian glaciation and during the Loch Lomond Stadial; (Ryder and McCann, 1971; Ballantyne, 1984).

Fossil periglacial features are best seen on the granophyre of Sròn an t-Saighdeir and Orval. The summit plateau of this area is mantled by a blockfield (Figure 61) which is sorted into large, partly vegetated stone polygons. On the flanks of the mountains the polygons become elongated down slope and are replaced by stone stripes, stone lobes and talus slopes, the material of which gradually merges into till at the extreme west of the island. McCann and Richards (1969) found that the debris covering the immense cliffs of Sgor Reidh show evidence of truncation by the Late-glacial sea but that the bench deposits of this period were themselves covered by over 6 m of relatively stable slope debris. Mountains formed by the ultrabasic rocks show less evidence of major periglacial structures, but small-scale patterned ground and solifluction terracettes have been noted. Extensive debris flows are reported by Ryder and McCann (1971, p.300) in Coire nan Grunnd and Fiachanis corrie.

The talus slopes associated with the Western Granite of Rum have already been noted. Screes are widespread also on the ultrabasic rocks where the slope deposits are being added to by the rapid disintegration of the bedrock. The lava hills north of the Western Granite are extensively mantled in partly stabilised scree as are those of Canna and Eigg. On Eigg the scree overlies some of the landslip deposits referred to next.

Landslips and associated deposits

On Rum, small landslips which probably postdate the formation of much of the talus have been observed on the south-west coast between Harris and A'Bhrideanach (McCann and Richards, 1969). Others occur on the steep seaward slope of Bloodstone Hill and east of Allt Bealach Mhic Néill, about 200 m SSE of the bridge. There is a small rockfall on the coast of Canna at Compass Hill, and on the west coast of Muck, at Camas Mór, landslips derived from the lavas of Beinn Airein are truncated by the Post-glacial beach.

The spectacular landslips on Eigg were noted in the first edition of this Memoir and were later studied by Godard (1965). The deposits shown as landslip on the published geological maps (1971 Provisional One-Inch and 1994 1:50 000) are probably of diverse origin although there are few lithological differences among them, and there is probably an almost continuous transition from truly glaciogenic material to debris formed by rotational slip (Figure 62). At Cleadale (Plate 1) and also in the Talm area [NM 483 910] there are the remains of slips and rockfalls now extensively covered by scree. Such slips presumably took place because the stablity of the plateau had been disturbed by ice action during the main Late Devensian glaciation.

Rotational slip and rock fall of a fresher appearance can be seen at several localities marginal to the lava plateau on Eigg. West of Laig the Mesozoic strata and basic sills exposed in the foreshore have been extensively buckled, and the overlying lavas have been subjected to slip, which is rotational in part. The slip has occurred partly on WNW-trending faults. Movement took place before the formation of the Post-glacial beach, where it is banked against the slip, but where the beach is absent due to marine erosion, more-recent slip has occurred. On the north and south sides of Blàr Mór (Figure 62) rock falls have taken place in the cliffs of the Valtos Sandstone Formation without disturbance of the underlying strata. The large rotational slip involving Valtos Sandstone and lavas in the Talm area is probably still active since the toe is being actively eroded by the sea. Part of the quartz-porphyry mass at Sgorr Sgaileach [NM 489 909] appears to have foundered on a rotational slip. Ancient, scree-covered slips and more-modern slips occur along the eastern flank of the plateau as far south as Monadh a'Bhraighe [NM 495 870]. Rockfalls continue to occur occur at the present time along this coast and in the Cleadale area a house was damaged in 1950 by rock fall.

North of Laig, three problematic exposures of breccia and lavas forming Cnoc Chroleaman [NM 469 879] and the adjoining Sidhean na Cailleach are partly covered by Post-glacial beach gravels. These may be relicts of ancient slips (Godard, 1965). On the south side of the Sgurr of Eigg at Grulin cottage [NM 456 843], there is a considerable rockfall some 600 m long composed of large blocks of pitchstone (Plate 27).

Raised beaches

As on the neighbouring mainland coast the raised beaches fall into a Late-glacial series, which formed towards the end of the last glaciation, during a period of great glacio-isostatic depression of the land and a Post-glacial series, formed in more recent times when glacioisostatic uplift of the land was overtaken by the rapid eustatic rise of the sea. Between the two periods of relatively high sea level, sea level was lower than at present. Since the last, Post-glacial transgression, slow but continuous rise of the land has resulted in a fall in sea level to its present level.

Late-glacial beaches

Late-glacial beaches occur at all levels from near 30 m above OD to just above the level of the Post-glacial beach at about 6 m above OD. On Canna an extensive gravel beach occurs above Tarbert Bay with storm ridges reaching 21 m above local OD, and at Corogon Mot. [NG 280 056] near Canna House gravels occur at about 30 m above OD. Thin beach gravels also occur on the pre-Late Devensian glacial marine platform on the north side of the island. On Eigg, storm beach gravels at about 25 m above OD occur at Camas Sgiotaig, at two levels differing by 3 m. Traces of a back feature occur at the higher level. A small patch of Late-glacial beach gravel rises to a little above the 30 m level at Camas Mot. in Muck.

The most extensive developments of Late-glacial beaches occur on Rum (Figure 61). At Harris a fine storm beach comprising several shingle ridges rises to a little above 30 m (Plate 29). Most of the blocks and pebbles are of granophyre, probably derived largely from talus which formerly covered the 'pre-glacial' raised marine platform to the north-west. The high storm beach is locally fronted by a narrow gravel terrace at 15–17 m above high water mark. At Schooner Point [NM 305 981], a convex-outwards shingle ridge rising to 25.3 m above high water mark rests on the pre-Late Devensian marine platform, and beach gravels on this coast are locally overlain by 6 m of stable slope deposits (McCann and Richards, 1969).

On the south side of Guirdil Bay there is about 15 m of well-rounded gravel exposed below a terrace surface at about 30 m above mean sea level. This terrace is perhaps slightly above the level of the poorly sorted glaciofluvial gravels seen above Guirdil itself, suggesting that ice was still present higher up the valley at much the same time as the beach gravels were being laid down. Patches of Late-glacial beach gravel occur in terrace remnants at about 18 m above mean sea level between Guirdil and the mouth of Glen Shellesder, and at the latter locality a small area of gravel overlying brown till reaches nearly 30 m above mid-tide level.

All other Late-glacial beaches seen on Rum are at lower levels. Near Kilmory Lodge [NG 3577 0391] heights of 16.4 and 16 m above mean sea level have been measured on two small spreads of gravel, and a pit just east of the lodge shows over 2 m of well-rounded shingle. Raised beach gravels rising to 15 m above high tide level occur at Camas Pliasgaig [NG 398 029], and gravels about 12 m above OD were formerly seen at the site of Kinloch Castle. On the south side of Loch Scresort there is a small area of gravel rising to about 13.4 m above OD at the old settlement [NM 415 990].

Post-glacial beaches

Small areas of Post-glacial beach gravels which rise to about 6 m above OD are present on most of the islands. On Eigg a small section of such a beach occurs at Camas Sgiotaig against a cliff cut in Valtos Sandstone, and another can be seen on the north coast south of Eilean Thuilm where several storm ridges rise eastwards against the porphyry mass of Sgorr Sgaileach. The latter beach is cut into old slip deposits and has been partly obliterated by the Talm landslip. At the Bay of Laig, storm beach ridges rising to 12 m above OD are partly capped by blown sand and front the bay for over a kilometre. Behind the ridges is a small alluvial tract, the site of a former lagoon. The interpretation of these gravels as 'pre-glacial' as suggested by Barrow in the first edition of the Memoir is certainly incorrect.

On Rum, Post-glacial beach gravels occur below the Late-glacial storm beaches at Kilmory, Guirdil and Harris, and gravel patches by the roadside south of Kinloch Castle are probably relicts of this Post-glacial beach. Similar deposits occur between Camas Mór [NM 407 793] and Bàgh a'Ghallanaich [NM 407 801] on Muck.

Modern beaches around the islands are composed of boulders, shingle and sand, these materials being derived from the erosion of drift deposits, the disintegration of local sandstones and basalt and, on the south-west coast of Rum, from the breakdown of granophyre and ultrabasic rocks. The quartz sands of the Bay of Laig and Camas Sgiotaig are derived from the Valtos Sandstone Formation, and the sands at Kilmory and Samhan Insir [NG 378 045] on Rum are the result of the erosion of the Torridonian sandstones. When dry, the 'singing' sand at Camas Sgiotaig, on Eigg, emits a shrill sound when crushed underfoot, a feature noticed by Hugh Miller (1858). Shell sand occurs on the north coast of Muck, the northwest coast of Sanday, and at Poll nam Partain [NM 486 848] on Eigg.

Peat, alluvium and blown sand

Peat of workable extent and thickness is of restricted occurrence in the Small Isles, but thin blanket peat and small stretches of basin peat are widespread, particularly on the lava plateaux and Torridonian sandstones of Rum, Eigg and Muck. On Rum the most extensive tracts of peat are in the west, on Monadh Mhiltich [NM 355 955] and in Kilmory Glen (Figure 61). Peat occupies the sites of former lochans at Cleadale in Eigg, and also occurs on Blàr Dubh south of Laig.

Alluvium is generally confined to small areas bordering streams, such as in Kilmory Glen in Rum. Alluvial fan deposits are well seen on the northern slopes of Bloodstone Hill.

Blown sand occurs as coastal dunes at the Bay of Laig and at Kilmory, and a small tract of machair occurs on the north coast of Muck where the shell sand has blown inland. The disintegration of the ultrabasic rocks of Rum gives rise inland to small areas of blown sand of a brown colour. A good example is seen on the east shore of Loch Fiachanis, but it is too small to show on the map.

Deeply decomposed rock

Deep decomposition of the plutonic basic and ultrabasic rocks of Rum has been variously ascribed to weathering processes ranging in age from Tertiary to the present day (Ball 1964a; Godard, 1965; Peacock, 1969). As far as is known much of the weathering is physical rather than chemical.

In Kinloch Glen a roadside quarry [NM 378 988] shows nearly 5 m of decomposed gabbroic igneous rock which is traversed by thin dykes of fresh basalt. A section described by Ball (1964a) is exposed in a disused quarry [NG 363 008] in Kilmory Glen, within a few metres of the Long Loch Fault. The decomposition (in gabbro) reaches to a maximum depth of about 7.6 m. At the east end of the quarry section the decomposed rock is undercut so that it is overlain and partly underlain by fresh clayey till which truncates a basalt dyke above and below. Veins of saponite (Ball, 1964b) are terminated at the knife-sharp interface with the till, which contains fragments of the vein mineral. These facts suggest that the rock was either frozen when eroded by the ice or had not then been decomposed. At Salisbury's Dam [NM 364 999] an area of decomposed ultrabasic rock, exposed in and near the stream, extends for about 300 m to the north and the south of the ruins of the dam It is crossed by faults of the Long Loch Fault system which are accompanied by breccia and by veins of lizardite (Ball, 1964a) and calcite. The maximum depth of decomposition seen is about 15 m. Specimens examined from localities 3 and 12 m below rockhead show the peridotite, composed essentially of fresh olivine and pyroxene with minor calcic plagioclase and biotite, to be partly or completely disaggregated but not chemically altered. At both localities in Kilmory Glen, Ball (1964a) records a reddened zone thought to have been formed under warm semi-arid conditions.

Above Harris, the Glen Duian River has cut a gorge through thin till into about 21 m of entirely or partly decomposed rock. The decomposition has preferentially affected cumulates of the Harris Bay Member previously altered by processes connected with the intrusion of the Glen Duian Gabbro sheets (Wadsworth, 1961), and it is therefore uncertain to what extent it falls into the same category as the decomposition seen in Kinloch Glen. At Harris, a section [NM 377 957] in the left bank of the river exposes about 5 m of storm beach gravel on 3 m of hard till which in turn lies on 3 m of partly decomposed ultrabasic rock forming the pre-main Late Devensian marine platform. At the footbridge over the Abhainn Rangail [NM 345 955] the till is underlain by decomposed ultrabasic rock with corestones, between which there are veins of undisturbed laminated silt and gravel, presumably material infiltrated along former frost cracks. Some 100 m downstream there is a section through what seems to be a cliff in decomposed rock, flanked by till and overlain by Late-glacial gravels.

On the east side of Rum decomposition of the ultra-basic rocks was recorded by Harker (1908) at the head of Coire Dubh and in the Allt Mór na h-Uamha [NM 397 973]. In the Allt na h-Uamha [NM 408 967] gabbro and ultrabasic rock are decayed to a depth of over 6 m below the Late-glacial moraine of Coire nan Grunnd. Similar deep decomposition occurs in Dibidil below the loose gravelly drift of the Loch Lomond Stadial glacier.

From the detail given above it is clear that deep decomposition of basic and ultrabasic rocks occurs both in strongly glaciated areas such as in Glen Dibidil where the 'pre-glacial' cliff and bench have been removed by the ice, and in areas of the island where some protection from glacial erosion occurred. In addition considerable physical decomposition has certainly taken place since the retreat of the ice in the area north of Loch an Dornabac [NM 355 974] where roches moutonnées are sufficiently disaggregated to be quarried as sand, and on the mountains about Hallival and Askival, where present-day weathering of the rock faces is giving rise to accumulations of forsterite-bearing sand. In Coire Dubh basaltic dykes stand like walls above the surrounding ultrabasic rock, suggesting that up to 2 m of the latter have been removed, mainly by Post-glacial erosion. Whether or not relicts of Tertiary soil horizons have survived glaciation, the processes responsible for the disaggregation of ultra-basic rock have evidently operated during the Pleistocene and are still continuing at the present day, leading to the formation of offshore deposits in which olivine and chromite are concentrated (Chapter 13).

Chapter 13 Economic geology

Introduction

In the first edition of the Small Isles Memoir the presence of peat and beach sand on Eigg and of chromite on Rum are referred to, but Harker (1908, p.195) stated that 'the constitution of the islands affords little prospect of any mining or quarrying industry'. The rocky coastlines obviously have potential for extraction of bulk aggregate. However, as Rum has protected status it is unlikely that it can be regarded as a potential resource of aggregate for the foreseeable future. Locally, there are small workings for road metal on Rum and Canna, and for rough stone on Muck, but aggregate for the metalled roads on Eigg and Muck is imported from the mainland and from Skye. Of most significance are the recent discoveries of concentrations of chromite and olivine in marine sands off southern Rum (Gallagher et al., 1989) and of precious and base metals in the Eastern Layered Intrusion (Dunham and Wilkinson, 1985; Hulbert et al., 1992).

Chromite and other minerals on Rum

Chromite is widespread in rocks of the Rum Layered Suite and there are a few small sulphide occurrences in the Eastern Layered Intrusion (Chapter 7, pp.77–80). Also, Dunham and Wilkinson (1985) described electrum (Au 46.44%; Ag = 49.9%) in sulphide droplets in the chromitite horizon at the Unit 12/Unit 11 junction 800 m north-west of Hallival summit. Sulphides also occur in the margin of a peridotite plug west of Loch Sgaorishal. Hulbert et al. (1992) emphasise the enrichment of chromium, platinum group elements and gold, as well as nickel and copper, in peridotite containing pyrrhotine-chalcopyrite-pentlandite droplets first noticed by Faithfull (1985) at Allt nam Bà. The sulphide-bearing peridotite occurs at the boundary with allivalite in Unit 1 of the Eastern Layered intrusion (Figure 63). The base of the unit is unexposed. Drilling will be required to test the prediction (Hulbert et al., 1992) that the highest metal concentrations are most likely to be found in the basal part of the Eastern Layered Intrusion.

In brown 'soils' of disaggregated peridotite and in morainic deposits derived from rocks of the Rum Central Complex, olivine and chromite are resistate minerals and fresh, cleanly liberated grains are easily recovered. In Atlantic Corrie, in upper Glen Harris, morainic deposits 15 m or more in thickness occupy an area of at least 0.5 km2 (Figure 61) and are the principal source of heavy minerals in streams draining Glen Harris and leading to the well-formed bay, 0.5 km across, at Harris. Up to 1% Cr and 27% MgO are present in the — 150 μm fraction of stream alluvium (Institute of Geological Sciences, 1983; British Geological Survey, 1990) and as much as 5% Cr accompanied by 100 ppm Pt + Pd have been reported from heavy mineral concentrates (J S Coats and A G Gunn, personal communication, 1987). Streams draining south and east from the Eastern Layered and Central intrusions are also geochemically anomalous, notably the Dibidil River which originates from moraine-strewn conies south-west of Askival then descends to a modest bay at Dibidil.

Marine deposits of chromite and olivine off Rum

As Rum is a National Nature Reserve, exploitation of chromite or other ore reserves on land could prove environmentally unacceptable. The marine dispersal of resistate minerals derived by glacial and fluvial erosion of the rocks of the Rum Central Complex has been investigated as a possible alternative for mineral exploitation. Black, heavy mineral sand occurs on the rocky beach in the bay at Harris, although well-defined high-water-mark bands such as are known from the Northumberland coast (Gallagher, 1974) are absent. The inshore area along 28 km of the southern coastline of Rum were surveyed in 1987, between A' Bhrideahach in the west and Loch Scresort in the east (Figure 1) and surficial marine sediment collected at 91 sites in water depths of 50 m or less within 3 km of the shore. After dissolution of shell calcite averaging 20% by weight, analysis of small shipboard samples yielded mean values of 0.12% Cr2O3 and 3.5% Mg (Table 12).

Relatively high chromium and magnesium values were recognised from two areas representing submarine deltas off Harris and Dibidil; new analyses from these areas are presented in (Figure 64) and (Figure 65). In the bay off Harris, 3 km2 of sand sampled from the sea-bed surface in a mean water depth of 23 m contained on average 0.24% Cr and 7% Mg after removal of some 18% of shell calcite. The grade of the heavy mineral sands is therefore about 1% of chromite and 25% of forsteritic olivine from the mineral analyses given in (Table 13). It is concluded that tens of thousands of tonnes of chromite and more than 1 million tonnes of olivine and of calcite are present in the topmost 1 m of the Harris delta (utilising 2.2 g.cm-1 as the wet density). The peak anomaly of 4% chromite and 50% olivine lies 1.8 km offshore.

A smaller area of 1 km2 of heavy mineral sand is present up to 1 km off the mouth of the Dibidil River in a mean water depth of 19 m (Figure 64). After dissolution of approximately 30% contained calcite a small number of surface samples from the submarine delta (Gallagher, 1989) average 0.19% Cr, equivalent to about 0.8% chromite, and 4.5% Mg (15–20% of forsteritic olivine). The peak anomaly of 1.2% chromite occurs only 350 m offshore of the mouth of the Dibidil river. The figures indicate 10 000 tonnes of chromite in the top metre of the delta. Other isolated sample sites off south-west and south-east Rum are somewhat anomalous in chromium and magnesium, as might be expected from tidal dispersion, but further sampling will be needed to properly assess their significance.

A full evaluation of the true lateral extent and thickness of the deposits will require geophysical surveys and drilling of the deltaic sands. Heavy minerals will most likely be concentrated towards the base of the sand accumulations.

Mineralogy

The deltaic sediments are well-sorted medium- to fine-grained sands (125–500 pm) highly suitable for beneficiation by density or magnetic methods. The total amount of heavy minerals present increases with decreasing content of shell calcite, a feature which is likely to occur with depth in the deltas, and of quartz and feldspar. In five representative samples the total heavy mineral content ranged from 36 to 94% with olivine dominant, followed by clinopyroxene and orthopyroxene. Chromite-type spinels, magnetite and ilmenite make up 10–15% of the overall heavy-mineral fraction. Traces of platinum, palladium and gold were detected in a few of the samples (Gallagher, 1989).

From analyses of the main heavy minerals (Table 13), it can be seen that chromite averages 32% Cr2O3 (10.7–44.4%) and the olivine is forsterite-rich (FO87). Detailed investigation of 31 samples from and around the Harris and Dibidil deltas shows that the chromite and olivine, calculated on a calcite-free basis, form 0.22–3.98% and 5–41% respectively (Basham et al., 1989).

Benificiation trials yielded concentrates of 86% chromite and 78% olivine with recoveries of 60% and 50% respectively. The size range and composition of the olivine in the marine sands compared well with the desired range for refractory use (Griffith, 1984). The iron content of the chromite is higher on average than that of South African chromites (25% FeO); nevertheless, in ferro-chrome production the tendency is towards acceptance of increasingly higher iron content (Power, 1985).

References

Most of the references listed below are held in the libraries of the British Geological Survey at Murchison House, Edinburgh and Keyworth, Nottingham. Copies of the references can be purchased subject to the current copyright legislation.

ALLEN, J R L. 1970. Studies in fluviatile sedimentation: a comparison of fining-upwards cyclothems, with special reference to coarse member composition and interpretation. Journal of Sedimentary Petrology, Vol. 40, 298–323.

ALLWRIGHT, E A. 1980. The structure and petrology of the volcanic rocks of Eigg, Muck and Canna, NW Scotland. Unpublished MSc thesis, University of Durham (2 vols.).

ANDERSON, F W, and DUNHAM, K C. 1966. Geology of northern Skye. Memoirs of the Geological Survey of Great Britain, Sheet 80 and parts of 81, 90 and 91 (Scotland).

ANDREWS, J E. 1984. Aspects of sedimentary facies and diagenesis in limestone-shale formations of the (Middle Jurassic) Great Estuarine Group, Inner Hebrides. Unpublished PhD thesis, University of Leicester.

ANDREWS, J E. 1985. The sedimentary facies of a late Bathonian regressive episode: the Kilmaluag and Skudiburgh formations of the Great Estuarine Group, Inner Hebrides, Scotland. Journal of the Geological Society of London, Vol. 142, 1119–1137

ANDREWS, J E. 1986. Microfacies and geochemistry of Middle Jurassic algal limestones from Scotland. Sedimentology, Vol. 33, 499–520.

ANDREWS, J E, HAMILTON, P J, and FALLICK, A E. 1987. The geochemistry of early diagenetic dolostones from a low-salinity Jurassic lagoon. Journal of the Geological Society of London, Vol. 144, 687–698.

ANDREWS, J E, and WALTON, W. 1990. Depositional environments within Middle Jurassic oyster-dominated lagoons: an integrated litho-, bio-, and palynofacies study of the Duntulm Formation (Great Estuarine Group), Inner Hebrides. Transactions of the Royal Society of Edinburgh, Earth Sciences, Vol. 81, 1–22.

BAILEY, E B. 1914. The Sgurr of Eigg. Geological Magazine, Vol. 51, 296–305.

BAILEY, E B. 1945. Tertiary igneous tectonics of Rhum (Inner Hebrides). Quarterly Journal of the Geological Society of London, Vol. 100, 165–191.

BAILEY, E B. 1956. Hebridean notes: Rhum and Skye. Liverpool and Manchester Geological Journal, Vol. 1, 420–426.

BAILEY, E B, CLOUGH, C T, WRIGHT, W B, RICHEY, J E, and WILSON, G V. 1924. Tertiary and Post-Tertiary geology of Mull, Loch Aline, and Oban. Memoirs of the Geological Survey: Scotland, parts of sheet 43, 44, 51 and 52.

BALL, D F. 1964a. Deep weathering profile on the Isle of Rhum, Inverness-shire. Scottish Geographical Magazine, Vol. 80, 22–27.

BALL, D F. 1964b. Saponite and lizardite veins on the Island of Rhum. Clay Minerals Bulletin, Vol. 5, 434–442.

BALLANTYNE, C K. 1984. The Late Devensian periglaciation of upland Scotland. Quaternary Science Reviews, Vol. 3, 311–343.

BALLANTYNE, C K, and WAIN-HOBSON, T. 1980. The Loch Lomond Advance on the island of Rhum. Scottish Journal of Geology, Vol. 16, 1–10.

BASHAM, I R, BEDDOE-STEPHENS, B, and MACDONALD, A. 1989. Mineralogical assessment of submarine heavy mineral sands, southern Rhum. British Geological Survey Technical Report, WG/89/26. 12pp.

BEDARD, J H, and SPARKS, R S J. 1991. Comments on 'The structure and petrogenesis of the Trallval and Ruinsival areas of the Rhum ultrabasic pluton' by J A Volker and B G J Upton. Transactions of the Royal Society of Edinburgh: Earth Sciences, Vol. 82, 389–390.

BEDARD, J H, SPARKS, R S J, RENNER, R, CHEADLE, M J, and HALLWORTH, M A. 1988. Peridotite sills and metasomatic gabbros in the Eastern Layered Series of the Rhum complex. Journal of the Geological Society of London, Vol. 145, 207–224.

BELL, B R, and EMELEUS, C H. 1988. A review of silicic pyroclastic rocks of the British Tertiary Volcanic Province. 365–379 in Early Tertiary volcanism and the opening of the NE Atlantic. MORTON, A C, and PARSON, L M (editors). Special Publication of the Geological Society of London, No. 39.

BERGGREN, W A, KENT, D V, FLYNN, J J, and VAN COUVERING, J A. 1985. Cenozoic geochronology. Bulletin of the Geological Society of America, Vol. 96, 1407–1418.

BERGH, S G, and SIGYALDASON, G E. 1991. Pleistocene mass-flow deposits of basaltic hyaloclastite on a shallow submarine shelf, South Iceland. Bulletin of Volcanology, Vol. 53, 597–611.

BINNS, P E, McQUILLAN, R, and KENOLTY, N. 1974. The geology of the Sea of the Hebrides. Report of the Institute of Geological Sciences, No. 73/14. 43pp.

BLACK, G P. 1952a. The Tertiary volcanic succession on the Isle of Rhum, Inverness-shire. Transactions of the Edinburgh Geological Society, Vol. 15, 39–51.

BLACK, G P. 1952b. The age relationship of the granophyre and basalt of Orval, Isle of Rhum. Geological Magazine, Vol. 89, 106–112.

BLACK, G P. 1954. The acid rocks of western Rhum. Geological Magazine, Vol. 91, 257–272.

BLACK, G P, and WELSH, W. 1961. The Torridonian succession of the Isle of Rhum. Geological Magazine, Vol. 98, 265–276.

BLAKE, D H, ELWELL, R W D, GIBSON, I L, SKELHORN, R R, and WALKER, G P L. 1965. Some relationships resulting from the intimate association of acid and basic magmas. Quarterly Journal of the Geological Society of London, Vol. 121, 31–49.

BLUCH, B J. 1964. Sedimentation of an alluvial fan in southern Nevada. Journal of Sedimentary Petrology, Vol. 34, 395–400.

BLUCH, B J. 1967. Deposition of some Upper Old Red Sandstone conglomerates in the Clyde area: a study in the significance of bedding. Scottish Journal of Geology, Vol. 3, 139–167.

BURR, M H P, ARMOUR, A R, HIMSWORTH, E M, MURPHY, T, and WYLIE, G. 1979. An explosion seismology investigation of the continental margin west of the Hebrides, Scotland, at 58°N. Tectonophysics, Vol. 59, 217–231.

BOTT, M H P, HOLLAND, G, STORRY, P G, and WATTS, A W. 1972. Geophysical evidence concerning the structure of the Lewisian of Sutherland, N.W. Scotland. Journal of the Geological Society of London, Vol. 128, 599–612.

BOTT, M H P, and TANTRIGODA, D A. 1987. Interpretation of the gravity and magnetic anomalies over the Mull Tertiary intrusive complex, NW Scotland. Journal of the Geological Society of London, Vol. 144, 17–28.

BOTT, M H P, and TUSON, J. 1973. Deep structure beneath the Tertiary volcanic regions of Skye, Mull and Ardnamurchan, north-west Scotland. Nature (Physical Sciences), London, Vol. 242, 114–116.

BRALEY, S. 1990. The sedimentology, palaeoeeology and stratigraphy of Cretaceous rocks in NW Scotland. Unpublished PhD thesis, Polytechnic South West, Plymouth.

BRITISH GEOLOGICAL SURVEY. 1990. Geochemical atlas of Great Britain, Argyll. (Keyworth, Nottingham: British Geological Survey.)

BROTHERS, R N. 1964. Petrofabric analyses of Rhum and Skaergaard layered rocks. Journal of Petrology, Vol. 5, 255–274.

BROWN, G M. 1956. The layered ultrabasic rocks of Rhum, Inner Hebrides. Philosophical Transactions of the Royal Society of London, Vol. 240B, 1–53.

BROWN, G M. 1963. Melting relations of Tertiary granitic rocks in Skye and Rhum. Mineralogical Magazine, Vol. 33, 533–562.

BULL, W B. 1964. Alluvial fans and near-surface substance in Western Fresno Co. California. Professional Papers of the United States Geological Survey, Vol. 437A.

BUTCHER, A R. 1984. A study of postcumulus structures in the Rhum layered intrusion, Inner Hebrides. Unpublished PhD thesis, University of Manchester.

BUTCHER, A R. 1985. Channelled metasomatism in Rhum layered cumulates - evidence from late-stage veins. Geological Magazine, Vol. 122, 503–518.

BUTCHER, A R, YOUNG, I M, and FAITHFULL, J W. 1985. Finger structures in the Rhum Complex. Geological Magazine, Vol. 122, 491–502.

CANT, D J. 1978. Bedforms and bar types in the South Saskatchewan River. Journal of Sedimentary Petrology, Vol. 48, 1321–1330.

CANT, D J, and WALKER, R G. 1978. Fluvial processes and facies sequences in the sandy braided South Saskachewan River, Canada. Sedimentology, Vol. 25, 625–648.

CARMICHAEL, I S E. 1960a. The feldspar phenocrysts of some Tertiary acid glasses. Mineralogical Magazine, Vol. 32, 587–608

CARMICHAEL, I S E. 1960b. The pyroxenes and olivines from some Tertiary acid glasses. Journal of Petrology, Vol. 1, 309–336.

CARMICHAEL, I S E. 1963. The crystallization of feldspar in volcanic acid liquids. Quarterly Journal of the Geological Society of London, Vol. 119, 95–131.

CARMICHAEL, I S E. 1967. The iron-titanium oxides of salic volcanic rocks, and their associated ferromagnesian silicates. Contributions to Mineralogy and Petrology, Vol. 14, 36–64.

CARTER, S R, EVENSEN, N M, HAMILTON, P J, and O'NIONS, R K. 1978. Neodymium and strontium isotopic evidence for crustal contamination of continental volcanics. Science, Vol. 202, 743–747.

CAS, R A F, and WRIGHT, J V. 1987. Volcanic successions: modern and ancient. (London: Allen and Unwin.)

CHARLESWORTH, J K 1956. The Late-glacial history of the Highlands and Islands of Scotland. Transactions of the Royal Society of Edinburgh, Vol. 62, 769–928.

CHEN, P-J, and HUDSON, J D. 1991. The conchostrachan fauna of the Great Estuarine Group, Middle Jurassic, Scotland. Palaeontology, Vol. 34, 515–545.

CLAYDON, R V, and BELL, B R. 1992. The structure and petrology of ultrabasic rocks in the southern part of the Cuillin Igneous Complex, Isle of Skye. Transactions of the Royal Society of Edinburgh: Earth Sciences, Vol. 83, 635–653.

COLE, D R, and OHOMOTO, H. 1986. Kinetics of isotope exchange at elevated temperatures and pressures. 41–90 in Stable isotopes in high temperature geological processes. VALLEY, J W, and others (editors). Mineralogical Society of America, Reviews in Mineralogy, Vol. 16.

COLLINSON, J D, and THOMPSON, D B. 1989. Sedimentary structures (2nd edition). (London: Unwin, Hyman.)

COPPIN, R. 1982. A geophysical investigation of the Tertiary igneous complex of Rhum, Inner Hebrides. Unpublished MSc dissertation, University of Durham.

COX, K G. 1980. A model for flood basalt volcanism. Journal of Petrology, Vol. 21, 629–650.

CRAIG, G Y. 1991. Geology of Scotland (3rd edition). (London: The Geological Society.)

CROWLEY, K D. 1983. Large-scale bed configurations (macroforms), Platte River Basin, Colorado and Nebraska: primary structures and formative processes. Bulletin of the Geological Society of America, Vol. 94, 117–133.

DAGLEY, P, and MUSSETT, A E. 1981. Palaeomagnetism of the British Tertiary igneous province: Rhum and Canna. Geophysical Journal of the Royal Astronomical Society, Vol. 65, 475–491.

DAGLEY, P, and MUSSETT, A E. 1986. Palaeomagnetism and radiometric dating of the British Tertiary igneous province: Muck and Eigg. Geophysical Journal of the Royal Astronomical Society, Vol. 85, 221–242.

DICKIN, A P. 1981. Isotope geochemistry of Tertiary igneous rocks from the Isle of Skye, N.W. Scotland. Journal of Petrology, Vol. 22, 155–189.

DICKIN, A P. 1988. The North Atlantic Tertiary Province. 111–149 in Continental flood basalts. Macdonald, J D (editor). (Dordrecht: Kluwer Academic Publishers.)

DICKIN, A P, and JONES, N W. 1983. Isotopic evidence for the age and origin of pitchstones and felsites, Isle of Eigg, NW Scotland. Journal of the Geological Society of London, Vol. 140, 691–700.

DONALDSON, C H. 1974. Olivine crystal types in harrisitic rocks of the Rhum pluton and in Archean spinifex rocks. Bulletin of the Geological Society of America, Vol. 85, 1721–1726.

DONALDSON, C H. 1975. Ultrabasic breccias in layered intrusions - the Rhum complex. Journal of Geology, Vol. 83, 33–45.

DONALDSON, C H. 1976. An experimental investigation of olivine morphology. Contributions to Mineralogy and Petrology, Vol. 57, 187–213.

DONALDSON, C H. 1982. Origin of some of the Rhum harrisite by segregation of intercumulus liquid. Mineralogical Magazine, Vol. 45, 201–209.

DONALDSON, C H, DREATR, H I, and JOHNSTON, R. 1973. Crystallization of poikilo-macro-spherulitic feldspar in a Rhum peridotite. Nature (Physical Sciences), London, Vol. 243, 69–70.

DUNHAM, A C. 1962. The petrology and structure of the northern edge of the Tertiary igneous complex of Rhum. Unpublished DPhil. thesis, University of Oxford.

DUNHAM, A C. 1964. A petrographic and geochemical study of back-veining and hybridization at a gabbro-felsite contact in Coire Dubh, Rhum, Inverness-shire. Mineralogical Magazine, Vol. 33, 887–902.

DUNHAM, A C. 1965a. The nature and origin of groundmass textures in felsites and granophyres from Rhum, Inverness-shire. Geological Magazine, Vol. 102, 8–23.

DUNHAM, A C. 1965b. A new type of banding in ultrabasic rocks from central Rhum, Inverness-shire, Scotland. American Mineralogist, Vol. 50, 1410–1420.

DUNHAM, A C. 1968. The felsites, granophyre, explosion breccias and tuffisites of the north-eastern margin of the Tertiary igneous complex of Rhum, Inverness-shire. Quarterly Journal of the Geological Society of London, Vol. 123, 327–352.

DUNHAM, A C. 1970. The emplacement of the Tertiary igneous complex of Rhum. Geological Journal, Special Issue No. 2, 23–32.

DUNHAM, A C, and EMELEUS, C H. 1967. The Tertiary geology of Rhum, Inner Hebrides. Proceedings of the Geologists' Association, Vol. 78, 391–418.

DUNHAM, A C, and THOMPSON, R N. 1967. The origin of granitic magma: Skye and Rhum. Journal of the Geological Society of Australia, Vol. 14, 339–344.

DUNHAM, A C, and WADSWORTH, WI 1978. Cryptic variation in the Rhum layered intrusion. Mineralogical Magazine, Vol. 42, 347–356.

DUNHAM, A C, and WILKINSON, F C F. 1985. Sulphide droplets and the U11/12 chromite band: a mineralogical study. Geological Magazine, Vol. 122, 539–548.

ELIAS, R. 1989. The origin of cyclic layering in the Eastern Layered Series of the Rhum Intrusion. Newsletter of the Geological Society of London, Vol. 18, 38.

ELIAS, R T. 1991. Cumulus processes and melt-migration in layered intrusions and the use of image analysis to quantify microscopic textures in cumulates. Unpublished PhD thesis, University of Liverpool.

EMELEUS, C H. 1973. Granophyre pebbles in Tertiary conglomerate on the Isle of Canna, Inverness-shire. Scottish Journal of Geology, Vol. 9, 157–159.

EMELEUS, C H. 1980. 1:20 000 solid geology map of Rhum. (Newbury: Nature Conservancy Council.)

EMELEUS, C H. 1981. A thrust fault at Welshman's Rock, Isle of Rhum. Scottish Journal of Geology, Vol. 17, 1–6.

EMELEUS, C H. 1985. The Tertiary lavas and sediments of northwest Rhum, Inner Hebrides. Geological Magazine, Vol. 122, 419–437.

EMELEUS, C H. 1987. The Rhum layered complex, Inner Hebrides, Scotland. 263–286 in Origins of igneous layering. PARSONS, I (editor). (Dordrecht: D Reidel Publishing Company.)

EMELEUS, C H. 1991. Tertiary igneous activity. 455–502 in Geology of Scotland. CRAIG, G Y (editor). (London: The Geological Society.)

EMELEUS, C H. 1994. 1:20 000 solid geology map of Rum (2nd edition). (Edinburgh: Scottish Natural Heritage.)

EMELEUS, C H, ALLWRIGHT, A E, KERR, A C, and WILLIAMSON, I T. 1996a. Red tuffs in the Palaeocene lava successions of the Inner Hebrides. Scottish Journal of Geology, Vol. 32, 83–89.

EMELEUS, C H, CHEADLE, M I HUNTER, R H, UPTON, B G J, and WADSWORTH, WI 1996b. The Rum Layered Suite. 403–440 in Layered igneous rocks. CAWTHORN, R G (editor). (Amsterdam: Elsevier.)

EMELEUS, C H, DUNHAM, A C, and THOMPSON, R N. 1971. Iron-rich pigeonites from acid rocks in the Tertiary Igneous Province of Scotland. American Mineralogist, Vol. 56, 940–951.

EMELEUS, C H, and FORSTER, R M. 1979. Field guide to the Tertiary igneous rocks of Rhum, Inner Hebrides. (Newbury: Geology and Physiography Section, Nature Conservancy Council.)

EMELEUS, C H, and GYOPARI, M C. 1992. British Tertiary Volcanic Province. Geological Conservation Review, Vol. 4, Joint Nature Conservation Committee. (London: Chapman and Hall.)

EMELEUS, C H, WADSWORTH, W. J. and SMITH, N J. 1985. The early igneous and tectonic history of the Rhum Tertiary Volcanic Centre. Geological Magazine, Vol. 122, 451–457.

ENGLAND, R W. 1988. The early Tertiary stress regime in NW Britain: evidence from the patterns of volcanic activity. 381–389 in Early Tertiary volcanism and the opening of the NE Atlantic. MORTON, A C, and PARSON, L M (editors). Special Publication of the Geological Society of London, No. 39.

ENGLAND, R W. 1992. The role of Palaeocene magmatism in the tectonic evolution of the Sea of the Hebrides Basin: implications for basin evolution on the NW Seabord. 163–174 in Basins on the Atlantic Seabord: petroleum geology, sedimentology and basin evolution. PARNELL, J (editor). Special Publication of the Geological Society of London, No. 62.

ENGLAND, R W, BUTLER, R W H, and HUTTON, D H W. 1993. The role of Paleocene magmatism in the Tertiary evolution of basins on the N.W. seaboard. 97–105 in Petroleum Geology of North West Europe: Proceedings of the 4th Conference. PARKER, J R (editor). (London: The Geological Society of London.)

ESSON, J DUNHAM, A C, and THOMPSON, R N. 1975. Low alkali, high Ca olivine tholeiite lavas from the Isle of Skye, Scotland. Journal of Petrology, Vol. 16, 488–497.

FAITHFULL, J W. 1985. The Lower Eastern Layered Series of Rhum. Geological Magazine, Vol. 122, 459–468.

FAITHFULL, J W. 1986. Petrology and geochemistry of gabbroic and ultrabasic rocks from eastern Rhum. Unpublished PhD thesis, University of Durham.

FISHER, R V, and SCHMINCKE, H-U. 1984. Pyroclastic rocks. (Berlin: Springer-Verlag.)

FITZPATRICK, E A. 1963. Deeply weathered rock in Scotland, its occurrence, age and contribution to the soils. Journal of Soil Science, Vol. 14, 33–43.

FORESTER, R W, and TAYLOR, H P. 1976. 180-depleted igneous rocks from the Tertiary complex of the Isle of Mull, Scotland. Earth and Planetary Science Letters, Vol. 32, 11–17.

FORESTER, R W, and TAYLOR, H P. 1977. 180/ 160 , D/H and 13C/12C studies on the Tertiary igneous complex of Skye, Scotland. American Journal of Science, Vol. 277, 136–177.

FORESTER, R W, and HARMON, R S. 1983. Stable isotope evidence for deep meteoric-hydrothermal circulation: Isle of Rhum, Inner Hebrides, Scotland. 135–136 in Proceedings of the 4th International Conference on water-rock interactions, Japan 1983 (Abstract).

FORSTER, R M. 1980. A geochemical and petrological study of the Tertiary minor intrusions of Rhum, northwest Scotland. Unpublished PhD thesis, University of Durham.

FRIEND, P F. 1983. Towards the field classification of alluvial architecture or sequence. 345–354 in Modern and ancient fluvial systems. COLLINSON, J D, and LEWIN, J (editors). Special Publication of the International Association of Sedimentologists, No. 6.

FYFE, J A., LONG, D, and EVANS, D. 1993. United Kingdom offshore regional report: the geology of the Malin-Hebrides sea area. (London: HMSO for the British Geological Survey.)

GALLAGHER, M J. 1974. Rutile and zircon in Northumbrian beach sands. Transactions of the Institution of Mining and Metallurgy, (Section B. Applied earth science.), Vol. 83, B97–98.

GALLAGHER, M J. 1989. Marine deposits of olivine and chromite, Inner Hebrides. British Geological Survey Open File Data Package.

GALLAGHER, M J BASHAM, I R, and 10 others. 1989. Marine deposits of chromite and olivine, Inner Hebrides of Scotland. British Geological Survey Technical Report, WF/89/13. British Geological Survey Mineral Reconnaissance Programme Report, No. 106. 20pp.

GEIKIE, A. 1867. On the Tertiary volcanic rocks of the British Islands. Proceedings of the Royal Society of Edinburgh, Vol. 6, 71–75.

GELKIE, A. 1871. On the Tertiary volcanic rocks of the British Islands. First paper. Quarterly Journal of the Geological Society of London, Vol. 27, 279–310.

GEIKIE, A. 1888. The history of volcanic action during the Tertiary period in the British Isles. Transactions of the Royal Society of Edinburgh, Vol. 35, 21–184.

GEIKIE, A. 1896. The Tertiary basalt-plateaux of North-Western Europe. Quarterly Journal of the Geological Society of London, Vol. 52, 331–405.

GEIKIE, A. 1897. The ancient volcanoes of Great Britain. 2 vols. (London: Macmillan.)

GIBS, F G F. 1976. Ultrabasic rocks of Rhum and Skye: nature of the parent magma. Journal of the Geological Society of London, Vol. 132, 209–222.

GIBSON, S A, and JONES, A P. 1991. Igneous stratigraphy and internal structure of the Little Minch Sill Complex, Trotternish Peninsula, northern Skye, Scotland. Geological Magazine, Vol. 128, 51–66.

GODARD, A. 1965. Recherches de geomorphologie en Ecosse du Nord-Ouest. (Strasbourg: Universite de Strasbourg.)

GREENWOOD, R C. 1987. Geology and petrology of the margin of the Rhum Complex, Inner Hebrides, Scotland. Unpublished PhD thesis, University of St Andrews.

GREENWOOD, R C, DONALDSON, C H, and EMELEUS, C H. 1990. The contact zone of the Rhum ultrabasic intrusion: evidence of peridotite formation from magnesian magmas. Journal of the Geological Society of London, Vol. 147, 209–212.

GREENWOOD, R C, FALLICK, A E, and DONALDSON, C H. 1992. Oxygen isotope evidence for major fluid flow along the contact zone on the Rhum ultrabasic intrusion. Geological Magazine, Vol. 129, 243–246.

GRIFFITHS, J. 1984. Olivine - exchanging new uses for old. Industrial Minerals, September 1984, 65–79.

HALLAM, A. 1959. Stratigraphy of the Broadford Beds of Skye, Raasay and Applecross. Proceedings of the Yorkshire Geological Society, Vol. 32, 165–184.

HALLAM, A. 1991. Jurassic, Cretaceous and Tertiary sediments. 439–454 in Geology of Scotland (3rd edition, revised). CRAIG, G Y (editor). (London: The Geological Society of London.)

HARDING, R R, MERRIMAN, R J, and NANCARROW, P H A. 1984. St Kilda: an illustrated account of the geology. Report of the British Geological Survey, Vol. 16, No.7.

HARKER, A. 1903. The overthrust Torridonian rocks of the Isle of Rum, and the associated gneisses. Quarterly Journal of the Geological Society of London, Vol. 59, 189–216.

HARKER, A. 1904. The Tertiary igneous rocks of Skye. Memoirs of the Geological Survey: Scotland, Sheets 70 and 71.

HARKER, A. 1906. The geological structure of the Spirt- of Eigg. Quarterly Journal of the Geological Society of London, Vol. 62, 40–67.

HARKER, A. 1908. The geology of the Small Isles of Inverness-shire. Memoirs of the Geological Survey: Scotland, Sheet 60.

HARLAND, W B, COX, A V, LLEWELLYN, P G, PICTON, C A G, SMITH, A G, and WALTERS, R. 1982. A geological time-scale. (Cambridge: Cambridge University Press.)

HARRIS, J P. 1984. Environments of deposition of Middle Jurassic sandstones in the Great Estuarine Group, NW Scotland. Unpublished PhD thesis, University of Leicester.

HARRIS, J P. 1989. The sedimentology of a Middle Jurassic lagoonal delta system: Elgol Formation (Great Estuarine Group), NW Scotland. 147–166 in Deltas: sites and traps for fossil fuels. WHATLEY, M K G, and PICKERING, K T (editors). Special Publication of the Geological Society of London, No. 41.

HARRIS, J P. 1992. Mid-Jurassic lagoonal delta systems in the Hebridean Basins: thickness and facies distribution patterns of potential reservior sandbodies. 111–144 in Basins on the Atlantic Seabord: petroleum geology, sedimentology and basin evolution. PARNELL, J (editor). Special Publication of the Geological Society of London, No. 62.

HARRIS, J P, and HUDSON, J D. 1980. Lithostratigraphy of the Great Estuarine Group (Middle Jurassic), Inner Hebrides. Scottish Journal of Geology, Vol. 16, 231–250.

HENDERSON, P. 1975. Reaction trends shown by chrome-spinels of the Rhum layered intrusion. Geochimica et Cosmochimica Acta, Vol. 39, 1035–1044.

HENDERSON, P, and &pus, R. 1976. Trace element indicators of the genesis of the Rhum layered intrusion, Inner Hebrides. Scottish Journal of Geology, Vol. 12, 325–333.

HENDERSON, P, MACKINNON, A, and GALE, N H. 1971. The distribution of uranium in some basic igneous cumulates and its petrological significance. Geochimica et Cosmochimica Acta, Vol. 35, 917–925.

HENDERSON, P, and SUDDABY, P. 1971. The nature and origin of the chrome-spinel of the Rhum layered intrusion. Contributions to Mineralogy and Petrology, Vol. 33, 21–31.

HENDERSON, P, and WOOD, R J. 1981. Reaction relationships of chrome-spinels in igneous rocks - further evidence from the layered intrusions of Rhum and Mull, Inner Hebrides, Scotland. Contributions to Mineralogy and Petrology, Vol. 78, 225–229.

HENRY, C D, and WOLFF, J A. 1992. Distinguishing strongly rheomorphic tuffs from extensive silicic lavas. Bulletin of Volcanology, Vol. 54, 174–186.

HOLLAND, G, and BROWN, G M. 1972. Hebridean tholeiitic magmas: a geochemical study of the Ardnamurchan cone sheets. Contributions to Mineralogy and Petrology, Vol. 37, 139–160.

HOLMES, A. 1965. Physical geology (2nd edition). (Edinburgh: Nelson.)

HOLMES, A. 1993. Physical geology (4th edition). Dm, P. McL D (editor). (London: Chapman and Hall.)

HUDSON, J D. 1960. The Laig Gorge Beds, Isle of Eigg. Geological Magazine, Vol. 97, 313–325.

HUDSON, J D. 1962. The stratigraphy of the Great Estuarine Series (Middle Jurassic) of the Inner Hebrides. Transactions of the Geological Society of Edinburgh, Vol. 19, 135–165.

HUDSON, J D. 1963a. The recognition of salinity-controlled mollusc assemblages in the Great Estuarine Series (Middle Jurassic) of the Inner Hebrides. Palaeontology, Vol. 6, 318–326.

HUDSON, J D. 1963b. The ecology and stratigraphical distribution of the invertebrate fauna of the Great Estuarine Series. Palaeontology, Vol. 6, 327–348.

HUDSON, J D. 1966. Hugh Miller's Reptile Bed and the Mytilus Shales, Middle Jurassic, Isle of Eigg, Scotland. Scottish Journal of Geology, Vol. 2, 265–281.

HUDSON, J D. 1968. The microstructure and mineralogy of the shell of a Jurassic mytilid (Bivalvia). Palaeontology, Vol. 11, 168–182.

HUDSON, J D. 1970. Algal limestones with pseudomorphs after gypsum from the Jurassic of north-west Scotland. Lethia, Vol. 3, 11–40.

HUDSON, J D. 1980. Aspects of brackish-water facies and fauna from the Middle Jurassic of north-west Scotland. Proceedings of the Geologists' Association, Vol. 91, 99–105.

HUDSON, J D. 1983. Mesozoic sedimentation and sedimentary rocks in the Inner Hebrides. Proceedings of the Royal Society of Edinburgh, Vol. 83B, 47–63.

HUDSON, J D, and ANDREWS, J E. 1987. The diagenesis of the Great Estuarine Group, Middle Jurassic, Inner Hebrides, Scotland. 259–276 in Diagenesis of sedimentary sequences. MARSHALL, J D (editor). Special Publication of the Geological Society of London, No. 36.

HUDSON, J D, CLEMENTS, R G, RIDING, J B, WAKEFIELD, M I, and WALTON, W. 1995. Jurassic palaeosalinities and brackish water communities - a case study. Palaios, Vol. 10, 392–407.

HUDSON, J D, and HARRIS, J P. 1979. Sedimentology of the Great Estuarine Group (Middle Jurassic) of north west Scotland. 1–13 in La sedimentation du Jurassique W-Europeen. (No editor named). Association Sedimentologistes Francais Speciale Publication, No. 1.

HUDSON, J D, and PALMER, T J. 1976. A euryhaline oyster from the Middle Jurassic and the origin of true oysters. Palaeontology, Vol. 19, 79–93.

HUGHES, C J. 1955. Petrographical and structural studies in south-east Rhum, Inverness-shire. Unpublished DPhil thesis, University of Oxford.

HUGHES, C J. 1960a. The Southern Mountains Igneous Complex, Isle of Rhum. Quarterly Journal of the Geological Society of London, Vol. 116, 111–138.

HUGHES, C J. 1960b. An occurrence of tilleyite-bearing limestone in the Isle of Rhum, Inner Hebrides. Geological Magazine, Vol. 97, 384–388.

HUGHES, C J, WADSWORTH, W J, and EMELEUS, C H. 1957. The contact between Tertiary granophyre and Torridonian arkose on Minishal, Isle of Rhum. Geological Magazine, Vol. 94, 337–339.

HULBERT, L J, DUKE, J M, ECKSTRAND, O R, SCOATES, R F J, LECHEMINANT, G M, WILLIAMSON, B, KYSER, T K, GALLAGHER, M J, GUNN, A G, and GRINENKO, L N. 1992. Metallogenesis and geochemical evolution of Cyclic Unit 1, Lower Eastern Layered Series, Rhum. Abstract in Mineral deposit modelling in relation to crustal reservoirs of the ore-forming elements. FOSTER, R F (editor). (London: Institution of Mining and Metallurgy.)

HUNTER, R H. 1987. Textual equilibrium in layered igneous rocks. 473–503 in Origins of igneous layering. PARSONS, I (editor). (Dordrecht: D Reidel Publishing Company.)

HUPPERT, H E, and SPARKS, R S J. 1980. The fluid dynamics of a basaltic magma chamber replenished by influx of hot, dense ultrabasic magma. Contributions to Mineralogy and Petrology, Vol. 75, 279–289.

INSTITUTE OF GEOLOGICAL SCIENCES. 1983. Regional geochemical atlas: The Hebrides. (London: Institute of Geological Sciences.)

IRVINE, T N. 1980. Magmatic infiltration metasomatism, double-diffusive fractional crystallization, and adcumulus growth in the Muskox and other layered intrusions. 325–383 in Physics of magmatic processes. HARGRAVES, R B (editor). (Princeton: Princeton University Press.)

IRVINE, T N. 1982. Terminology for layered intrusions. Journal of Petrology, Vol. 23, 127–162.

IRVINE, T N. 1987. Appendix 1. Glossary of terms for layered intrusions. 641–647 in Origins of igneous layering. PARSONS, I (editor). (Dordrecht: D Reidel Publishing Company.)

JACKSON, E D. 1970. The cyclic unit in layered intrusions - a comparison of repetitive stratigraphy in the ultramafic parts of the Stillwater, Muskox, Great Dyke and Bushveld complexes. Special Publication of the Geological Society of South Africa, Vol. 1, 391–424.

JAMESON, R. 1800. Mineralogy of the Scottish Isles. (Edinburgh)

JOHNSTONE, G S, and MYKURA,W. 1989. British regional geology: The Northern Highlands (4th edition). (Edinburgh: HMSO for British Geological Survey.)

JUDD, J W. 1874. The Secondary rocks of Scotland. Second paper. On the ancient volcanoes of the Highlands and the relations of their products to the Mesozoic Strata. Quarterly Journal of the Geological Society of London, Vol. 30, 220–301.

JUDD, J W. 1878. The Secondary rocks of Scotland. Third paper. The strata of the western coast and islands. Quarterly Journal of the Geological Society of London, Vol. 34, 660–743.

JUDD, J W. 1885a. On the Tertiary and older peridotites of Scotland. Quarterly Journal of the Geological Society of London, Vol. 41, 354–418.

JUDD, J W. 1885b. On the gabbros, dolerites, and basalts, of Tertiary age in Scotland and Ireland. Quarterly Journal of the Geological Society of London, Vol. 42, 49–95.

JUDD, J W. 1889. The Tertiary volcanoes of the Western Isles of Scotland. Quarterly Journal of the Geological Society of London, Vol. 45, 187–218.

KERR, A C. 1993. The geochemistry and petrogenesis of the Mull and Morven Tertiary lava succession, Argyll, Scotland. Unpublished PhD thesis, University of Durham.

KERR, A C. 1995. The geochemical stratigraphy, field relations and temporal variation of the Mull-Morvern Tertiary lava succession, NW Scotland. Transactions of the Royal Society of Edinburgh: Earth Sciences, Vol. 86, 35–47.

KITCHEN, D E. 1985. The parental magma on Rhum: evidence from alkaline segregations and veins in the peridotite from Salisbury's Dam. Geological Magazine, Vol. 122, 529–537.

KLEEMAN, A W. 1965. The origin of granitic magmas. Journal of the Geological Society of Australia, Vol. 12, 35–52.

KLEEMAN, A W. 1967. The origin of granitic rocks: Skye and Rhum. A reply. Journal of the Geological Society of Australia, Vol. 14, 345–347.

KNOX, R W O'B, and MORTON, A C. 1988. The record of early Tertiary N Atlantic volcanism in sediments of the North Sea Basin. 407–419 in Early Tertiary volcanism and the opening of the NE Atlantic. MORTON, A C, and PARSON, L M (editors). Special Publication of the Geological Society of London, No. 39.

KOKELAAR, B P. 1982. Fluidization of wet sediments during the emplacement and cooling of various igneous bodies. Journal of the Geological Society of London, Vol. 139, 21–33.

LAWSON, D E. 1965. Lithofacies and correlation within the lower 'Torridonian'. Nature, London, Vol. 207, 706–708.

LE MAITRE, R W. 1989. A classification of igneous rocks and glossary of terms. (Oxford: Blackwells Scientific Publications.)

LEE, C A. 1981. Post-deposition structures in the Bushveld Complex mafic sequence. Journal of the Geological Society of London, Vol. 138, 327–341.

LEIGHTON, M W. 1954. Petrogenesis of a gabbro-granophyre complex in northern Wisconsin. Bulletin of the Geological Society of America, Vol. 65, 401–442.

LINDSAY, J F. 1968. The development of clast fabric in mud flows. Journal of Sedimentary Petrology, Vol. 38, 1242–1253.

LOWDEN, B, BRALEY, S, HURST, A, and LEWIS, J. 1992. Sedimentological studies of the Cretaceous Lochaline Sandstone, NW Scotland. 159–162 in Basins on the Atlantic Seabord: petroleum geology, sedimentology and basin evolution. PARNELL, J (editor). Special Publication of the Geological Society of London, No. 62.

MCBIRNEY, A R, and NOYES, R M. 1979. Crystallization and layering of the Skaergaard Intrusion. Journal of Petrology, Vol. 20, 487–554.

MCCANN, S B. 1968. Raised shore platforms of the Western Isles of Scotland. 22–34 in Geography at Aberystwyth. BOWEN, E G, CARTER, H, and TAYLOR, J A (editors). (Cardiff: University of Wales.)

MCCANN, S B. 1969. Coastal features of Rhum. Scottish Journal of Geology, Vol. 5, 302–304.

MCCANN, S B, and RICHARDS, A. 1969. The coastal features of the Island of Rhum in the Inner Hebrides. Scottish Journal of Geology, Vol. 5, 15–25.

McCLURG, J E. 1982. Petrology and evolution of the northern part of the Rhum ultrabasic complex. Unpublished PhD thesis, University of Edinburgh (2 vols.).

MACCULLOCH, J. 1819. A description of the Western Islands of Scotland. (Edinburgh: Constable.)

MACCULLOCH, J. 1824. The Highlands and Western Isles of Scotland in letters to Sir Walter Scott, Bart. (London: Longman Hurst.)

MACDONALD, G A, and KATSURA, T. 1964. Chemical composition of Hawaiian lavas. Journal of Petrology, Vol. 5, 82–133.

MCQUILLIN, R, BACON, M, and BINNS, P E. 1975. The Blackstones Tertiary igneous complex. Scottish Journal of Geology, Vol. 11, 179–192.

McQUILLIN, R, BACON, M, and BINNS, P E. 1975. The Blackstones Tertiary igneous complex. Scottish Journal of Geology, Vol. 11, 179–192.

MCQUILLIN, R, and BINNS, P E. 1973. Geological structure of the Sea of the Hebrides. Nature (Physical Science), London, Vol. 241, 2–4.

MCQUILLIN, R, and TUSON, J. 1963. Gravity measurements over the Rhum Tertiary plutonic complex. Nature, London, Vol. 199, 1276–1277.

MARSHALL, J D. 1982. Isotopic composition of displacive fibrous calcite veins: reversals in pore-water composition trends during burial diagenesis. Journal of Sedimentary Petrology, Vol. 52, 615–630.

MATTEY, D P, GIBSON, I S, MARRINER, G F, and THOMPSON, R N. 1977. The diagnostic geochemistry, relative abundance and spatial distibution of high-calcium, low-alkali olivine tholeiite dykes in the Lower Tertiary regional swarm of the Isle of Skye, N.W. Scotland. Mineralogical Magazine, Vol. 41, 273–285.

MEIGHAN, I G. 1979. The acid igneous rocks of the British Tertiary Province. Bulletin of the Geological Survey of Great Britain, No. 70, 10–22.

MEIGHAN, I G, HUTCHISON, R, WILLIAMSON, I T, and MACINTYRE, R M. 1982. Geological evidence for the different relative ages of the Rhum and Skye Tertiary central complexes. Journal of the Geological Society, Vol. 139, 659 (abstract).

MILLER, H. 1858. The cruise of the 'Betsey'. Edited by W S Symonds from articles in 'The Witness'. (Edinburgh: Constable.)

MORRISON, M A., THOMPSON, R N, and DICKIN, A P. 1985. Geochemical evidence for complex magmatic plumbing during development of a continental volcanic center. Geology, Vol. 13, 581–584.

MONTY, C. 1967. Distribution and structure of recent stromatolitic algal mats, eastern Andros Island, Bahamas. Annales de la Societe Geologique de Belgique, Vol. 90, B55–100.

MORSE, S A, OWEN, B E, and BUTCHER, A R. 1987. Origin of finger structures in the Rhum Complex: phase equilibrium and heat effects. Geological Magazine, Vol. 124, 205–210.

MORTON, N. 1965. The Bearreraig Sandstone Series (Middle Jurassic) of Skye and Raasay. Scottish Journal of Geology, Vol. 1, 189–216.

MORTON, N. 1976. Bajocian (Jurassic) stratigraphy in Skye, Western Scotland. Scottish Journal of Geology, Vol. 12, 23–33.

MORTON, N. 1983. Palaeocurrents and palaeo-environment of part of the Bearreraig Sandstone (Middle Jurassic) of Skye and Raasay, Inner Hebrides. Scottish Journal of Geology, Vol. 19, 87–95.

MORTON, N. 1989. Jurassic sequence stratigraphy in the Hebrides Basin, NW Scotland. Marine and Petroleum Geology, Vol. 6, 243–260.

MORTON, N, SMITH, R M, GOLDEN, N, and JAMES, A V. 1987. Comparative stratigraphic study of Triassic Jurassic sedimentation and basin evolution in the northern North Sea and north-west of the British Isles. 697–709 in Petroleum geology of North West Europe. BRooxs, J, and GLENNIE, K (editors). (London: The Geological Society of London.)

MUSSETT, A E. 1984. Time and duration of Tertiary igneous activity of Rhum and adjacent areas. Scottish Journal of Geology, Vol. 20, 273–279.

MUSSETT, A E, DAGLEY, P, and SKELHORN, R R. 1988. Time and duration of activity in the British Tertiary Igneous Province. 337–348 in Early Tertiary volcanism and the opening of the North Atlantic. MORTON, A C, and PARSON, L M (editors). Special Publication of the Geological Society of London, No. 39.

NICHOLSON, P G. 1992a. Precambrian: Upper Proterozoic Torridon Group. 5–7 in Atlas of palaeogeography and lithofacies. COPE, j C WINGHAM, J K, and RAWSON, P F (editors). Memoir of the Geological Society of London, No. 13.

NICHOLSON, P G. 1992b. Sedimentation of the fluvial 'Torridonian' Applecross Formation, NW Scotland. Unpublished PhD thesis, University of Glasgow

NICHOLSON, P G. 1993. A basin reappraisal of the Proterozoic Torridon Group, north west Scotland. 183–202 in Sedimentation and tectonics. FROSTIC, L E, and STEEL, R G (editors). Special Publication of the International Association of Sedimentologists, No. 20.

PALACZ, Z A. 1984. Isotopic and geochemical evidence for the evolution of a cyclic unit in the Rhum intrusion, north-west Scotland. Nature, London, Vol. 307, 618–620.

PALACZ, Z A. 1985. Sr-Nd-Pb isotopic evidence for crustal contamination in the Rhum intrusion. Earth and Planetary Science Letters, Vol. 74, 35–44.

PALACZ, Z A, and TAIT, S R. 1985. Isotopic and geochemical investigations of cyclic unit 10 from the Eastern Layered Series of the Rhum intrusion, north west Scotland. Geological Magazine, Vol. 122, 485–490.

PARK, R G. 1991. The Lewisian Complex. 25–65 in Geology of Scotland. CRAIG, G Y (editor). (London: The Geological Society.)

PEACOCK, J D. 1967. West Highland moraine features aligned in the direction of ice flow. Scottish Journal of Geology, Vol. 3, 372–373.

PEACOCK, D. 1969. Coastal features of Rhum. Scottish Journal of Geology, Vol. 5, 301–302.

PEACOCK, J D. 1975. Landslip associated with glacier ice. Scottish Journal of Geology, Vol. 11, 363–364.

PEACOCK, J D. 1976. Quaternary features of Rhum, Inner Hebrides. Quaternary Newsletter, Vol. 20, 1–4.

PEARSON, D G, EMELEUS, C H, and KELLEY, S P. 1996. Precise 40Ar/39Ar age for the initiation of Palaeogene volcanism in the Inner Hebrides and its regional significance. Journal of the Geological Society of London, Vol. 153, 815–818.

PEDERSEN, A K, ENGELL, J, and RoNsno, J G. 1975. Early Tertiary volcanism in the Skagerrak: new chemical evidence from ash-layers in the mo-clay of northern Denmark. Lithos, Vol. 8, 255–268.

PHILLIPS, F C. 1938. Mineral orientation in some olivine-rich rocks from Rum and Skye. Geological Magazine, Vol. 75, 130–135.

POWER, T. 1985. Chromite - the non-metallurgical market. Industrial Minerals, April 1985, 17–51.

RAWSON, P F (editor). 1978. A correlation of Cretaceous rocks in the British Isles. Special Report of the Geological Society of London, No. 9.

REIF, W-E. 1971. Zur Genese des Muschelkalk-Keuper-Grenzebonebeds in Sildwestdeutschland. Neues Jahrbuch fur Geologie and Paliiontologie, Abhandlungen, Vol. 139, 369–404.

RENNER, R, and PALACZ, Z A. 1987. Basaltic replenishment of the Rhum magma chamber: evidence from unit 14. Journal of the Geological Society of London, Vol. 144, 961–970.

REYNOLDS, D L. 1951. The geology of Slieve Gullion, Foughill and Carrickcarnan: an actualistic interpretation of a Tertiary gabbro-granophyre complex. Transactions of the Royal Society of Edinburgh, Vol. 62, 85–143.

REYNOLDS, D L. 1954. Fluidization as a geological process, and its bearing on the problem of intrusive granites. American Journal of Science, Vol. 252, 577–614.

RICHEY, J E. 1937. Some features of Tertiary volcanicity in Scotland and Ireland. Bulletin Volcanologique, Series II, Vol. 1, 13–34.

RICHEY, J E. 1961. British regional geology: The Tertiary Volcanic Districts of Scotland (3rd edition, revised by A G MacGregor and F W Andeson; reprinted 1987). (Edinburgh: HMSO for British Geological Survey.)

RICHEY, J E, and THOMAS, H H. 1930. The geology of Ardnamurchan, North-West Mull and Coll. Memoirs of the Geological Survey: Scotland, Sheets 51 and 52.

RICHEY, J.E. and THOMAS, H H. 1932. The Tertiary ring complex of Slieve Gullion. Quarterly Journal of the Geological Society of London, Vol. 88, 776–849.

RIDING, J B, WALTON, W, and SHAW, D. 1991. Toarcian to Bathonian (Jurassic) palynology of the Inner Hebrides, Scotland. Palynology, Vol. 15, 115–179.

RIDLEY, W I. 1971. The petrology of some volcanic rocks from the British Tertiary Province: the islands of Rhum, Eigg, Canna, and Muck. Contributions to Mineralogy and Petrology, Vol. 32, 251–266.

RIDLEY, W I. 1973. The petrology of volcanic rocks from the Small Isles of Inverness-shire. Report of the Institute of Geological Sciences, No. 73/10. 55pp.

RIDLEY, W I. 1977. The crystallisation trends of spinets in Tertiary basalts from Rhum and Muck and their petrogenetic significance. Contributions to Mineralogy and Petrology, Vol. 64, 243–255.

ROBBINS, B. 1982. Finger structures in the Little Kufjord layered intrusion, Finnmark, Northern Norway. Contributions to Mineralogy and Petrology, Vol. 81, 290–295.

ROBINSON, M A, and MCCLELLAND, E A. 1987. Palaeomagnetism of the Torridonian of Rhum, Scotland: evidence for limited uplift of the Central Intrusive Complex. Earth and Planetary Science Letters, Vol. 85, 473–487.

RODD, J A. 1983. The sedimentology and geochemistry of the Type Diabaig Formation in the Upper Proterozoic Torridonian Group of Scotland. Unpublished PhD thesis, University of Reading.

RYDER, R H, and MCCANN, S B. 1971. Periglacial phenomena on the Island of Rhum in the Inner Hebrides. Scottish Journal of Geology, Vol. 7, 293–304.

SELLEY, R C. 1969. Torridonian alluvium and quicksands. Scottish Journal of Geology, Vol. 5, 328–346.

SHEPPARD, S M F. 1986. Igneous rocks: III. Isotopic case studies of magmatism in Africa, Eurasia and oceanic islands. 319–371 in Stable isotopes in high temperature geological processes. VALLEY, J W and others (editors). Mineralogical Society of America: Reviews in Mineralogy, No. 16.

SHARPE, M R, and IRVINE, T N. 1983. Melting relations of two Bushveld chilled margin rocks and implications for the origins of chromitite. Carnegie Institute of Washington Yearbook, No. 82, 295–300.

SISSONS, J B. 1967. The evolution of Scotland's scenery. (Edinburgh: Oliver and Boyd.)

SMITH, N D. 1971. Transverse bars and braiding in the Lower Platte River, Nebraska. Bulletin of the Geological Society of America, Vol. 82, 3407–3420.

SMITH, N J. 1985. The age and structural setting of limestones and basalts on the Main Ring Fault in southeast Rhum. Geological Magazine, Vol. 122, 439–445.

SMITH, N J. 1987. The age and structural setting of limestone and basalt on the Main Ring Fault of sourth-east Rhum, Inner Hebrides, Scotland. Unpublished MSc thesis, University of Durham.

SMYTHE, D K, and KENOLTY, N. 1975. Tertiary sediments in the Sea of the Hebrides. Journal of the Geological Society of London, Vol. 131, 227–233.

SPARKS, R S J, and HUPPERT, H E. 1984. Density changes during fractional crystallization of basaltic magmas: fluid dynamic implications. Contributions to Mineralogy and Petrology, Vol. 85, 300–309.

SPARKS, R S J, HUPPERT, H E, KERR, R C, MCKENZIE, D P, and TAIT, S R. 1985. Postcumulus processes in layered intrusions. Geological Magazine, Vol. 122, 555–568.

SPARKS, R S J, HUPPERT, H E, KOYAGUCHI, T, and HALLWORTH, M A. 1993. Origin of modal and rhythmic igneous layering by sedimentation in a convecting magma chamber. Nature, London, Vol. 361, 246–249.

SPEIGHT, J M, SKELHORN, R R, SLOAN, T, and KNAAP, R J. 1982. The dyke swarms of Scotland. 449–459 in Igneous rocks of the British Isles. SUTHERLAND, D S (editor.) (Chichester: John Wylie & Sons.)

STEEL, R J. 1971. New Red Sandstone movement on the Minch Fault. Nature (Physical Science), London, Vol. 234, 158–159.

STEEL, R J. 1974a. New Red sandstone piedmont and floodplain sedimentation in the Hebridean province, Scotland. Journal of Sedimentary Petrology, Vol. 44, 336–357.

STEEL, R J. 1974b. Cornstone (fossil caliche) - its origin, stratigraphy and sedimentological importance in the New Red Sandstone, western Scotland. Journal of Geology, Vol. 82, 351–369.

STEEL, R J. 1977. Triassic rift basins of NW Scotland, their configuration, infilling and development. In Proceedings of the Northern North Sea Symposium. FINSTAD, K G, and SELLEY, R C (editors). Stavanger: Norwegian Petroleum Society, Paper, No. 7, 18pp.

STEEL, R J, NICHOLSON, R, and KALANDER, L. 1975. Triassic sedimentation and palaeogeography in Central Skye. Scottish Journal of Geology, Vol. 11, 1–13.

STEEL, R J, and WILSON, A. 1975. Sedimentation and tectonism (?Permo-Triassic) on the margin of the North Minch Basin, Lewis. Journal of the Geological Society of London, Vol. 131, 183–202.

STEWART, A D. 1966. On the correlation of the Torridonian between Rhum and Skye. Geological Magazine, Vol. 103, 432–439.

STEWART, A D. 1969. 'Torridonian' rocks of Scotland reviewed. 595–608 in North Atlantic: geology and continental drift. KAY, M (editor). Memoir of the American Association of Petroleum Geologists, No. 12.

STEWART, A D. 1982. Late Proterozoic rifting in NW Scotland: the genesis of the 'Torridonian'. Journal of the Geological Society of London, Vol. 139, 413–420.

STEWART, A D. 1988a. The Stoer Group, Scotland. 97–103 in Later Proterozoic stratigraphy of the North Atlantic regions. WINCHESTER, J A (editor). (Glasgow: Blackie.)

STEWART, A D. 1988b. The Sleat and Torridon groups. 104–112 in Later Proterozoic stratigraphy of the North Atlantic regions. WINCHESTER, J A (editor). (Glasgow: Blackie.)

STEWART, A D, and PARKER, A. 1979. Palaeosalinity and environmental interpretation of red beds from the Late Precambrian ('Torridonian') of Scotland. Sedimentary Geology, Vol. 22, 229–241.

SYKES, R M. 1975. The stratigraphy of the Callovian and Oxfordian stages (Middle-Upper Jurassic) in northern Scotland. Scottish Journal of Geology, Vol. 11, 51–78.

TAIT, S R. 1985. Fluid dynamic and geochemical evolution of cyclic unit 10, Rhum, Eastern Layered Series. Geological Magazine, Vol. 122, 469–484.

TAN, F C, and HUDSON, J D. 1974. Isotopic studies on the palaeoecology and diagenesis of the Great Estuarine Series (Jurassic) of Scotland. Scottish Journal of Geology, Vol. 10, 91–128.

TAYLOR, H P. 1968. The oxygen isotope geochemistry of igneous rocks. Contributions to Mineralogy and Petrology, Vol. 19, 1–71.

TAYLOR, H P, and FORESTER, R W. 1971. Low-180 igneous rocks from the intrusive complexes of Skye, Mull and Ardnamurchan, western Scotland. Journal of Petrology, Vol. 12, 465–497.

TAYLOR, H P, and FORESTER, R W. 1979. An oxygen and hydrogen study of the Skaergaard Intrusion and its country rocks: a description of a 55-MY old fossil hydrothermal system. Journal of Petrology, Vol. 20, 355–420.

THOMPSON, R N. 1969. Tertiary granites and associated rocks of the Marsco area, Isle of Skye. Quarterly Journal of the Geological Society of London, Vol. 124, 349–385.

THOMPSON, R N. 1982. Magmatism of the British Tertiary Volcanic Province. Scottish Journal of Geology, Vol. 18, 49–107.

THOMPSON, R N, DILKIN, A P, GIBSON, I L, and MORRISON, M A. 1982. Elemental fingerprints of isotopic contamination of Hebridean Palaeocene mantle-derived magmas by Archaean sial. Contributions to Mineralogy and Petrology, Vol. 79, 159–168.

THOMPSON, R N, ESSON, J, and DUNHAM, A C. 1972. Major element chemical variation in the Eocene lavas of the Isle of Skye, Scotland. Journal of Petrology, Vol. 13, 219–253.

THOMPSON, R N, and GIBSON, S A. 1991. Subcontinental mantle plumes, hotspots and pre-existing thinspots. Journal of the Geological Society of London, Vol. 148, 973–978.

THORPE, R S. 1978. The parental basaltic magma of granites from the Isle of Skye, NW Scotland. Mineralogical Magazine, Vol. 42, 157–158.

TILLEY, C E. 1944. A note on the gneisses of Rum. Geological Magazine, Vol. 81, 129–131.

TILLEY, C E. 1947. The gabbro-limestone contact zone of Camas Mór, Muck, Inverness-shire. Comptes Rendues de la Sociite giologique Finlande, Vol. 20, 97–105.

TILLEY, C E, and HARWOOD, H F. 1931. The dolerite-chalk contact at Scawt Hill, Co. Antrim. Mineralogical Magazine, Vol. 22, 439–468.

TOMKEIEFF, S I. 1942. The Tertiary lavas of Rum. Geological Magazine, Vol. 79, 1–13.

TOMKEIEFF, S I. 1945. On the petrology of the ultrabasic and basic plutonic rocks of the Isle of Rum. Mineralogical Magazine, Vol. 27, 127–136.

TOMKEIEFF, S I, and BLACKBURN, K B. 1942. On the remains of fossil wood enclosed in a Tertiary lava on the Isle of Rum, Inner Hebrides. Geological Magazine, Vol. 79, 14–17.

TURNER, J A. 1966. The Oxford Clay of Skye, Scalpay and Eigg. Scottish Journal of Geology, Vol. 2, 243–252.

TUSON, J. 1959. A geophysical investigation of the Tertiary volcanic districts of western Scotland. Unpublished PhD thesis, University of Durham.

TUTTLE, O F, and BOWEN, N L. 1958. Origin of granite in the light of experimental studies in the system NaAlSi3O8 - KAlSi3O8 – SiO2-H2O. Memoir of the Geological Society of America, No. 74,

TYRRELL, G W. 1928. The geology of Arran. Memoirs of the Geological Survey: Scotland. Isle of Arran Special Sheet.

UPTON, B G J. 1988. History of Tertiary igneous activity in the N Atlantic borderlands. 429–453 in Early Tertiary volcanism and the opening of the NE Atlantic. MORTON, A C, and PARSON, L M (editors). Special Publication of the Geological Society of London, No. 39.

VOLKER, J A. 1983. The geology of the Trallval area, Rhum, Inner Hebrides. Unpublished PhD thesis, University of Edinburgh.

VOLKER, J A, and UPTON, B G J. 1990. The structure and petrogenesis of the Trallval and Ruinsival areas of the Rhum ultrabasic complex. Transactions of the Royal Society of Edinburgh: Earth Sciences, Vol. 81, 69–88.

VOLKER, J A. and UPTON, B G J. 1991. Reply to comments by J H Bedard and R S J Sparks. Transactions of the Royal Society of Edinburgh: Earth Sciences, Vol. 82, 391.

WADSWORTH, W J. 1961. The layered ultrabasic rocks of south-west Rhum, Inner Hebrides. Philosophical Transactions of the Royal Society of London, Vol. 244B, 21–64

WADSWORTH, W J. 1973. Magmatic sediments. Mineral Sciences Engineering, Vol. 5, 25–35.

WADSWORTH, W J. 1985. Terminology of postcumulus processes and products in the Rhum layered intrusion. Geological Magazine, Vol. 122, 549–554.

WADSWORTH, W J. 1992. Ultrabasic igneous breccias of the Long Loch area, Isle of Rhum. Scottish Journal of Geology, Vol. 28, 103–113.

WADSWORTH, W J. 1994. The peridotite plugs of northern Rum. Scottish Journal of Geology, Vol. 30, 167–174.

WAGER, L R, and DEER, W A. 1939. Geological investigations in East Greenland: Part III - The petrology of the Skaergaard intrusion, Kangerdlugssuaq. Meddelelser om Gronland, Vol. 105, 1–323.

WAGER, L R, and BROWN, G M. 1951. A note on rhythmic layering in the ultrabasic rocks of Rhum. Geological Magazine, Vol. 88, 166–168.

WAGER, L R, and BROWN, G M. 1968. Layered igneous rocks. (Edinburgh: Oliver and Boyd.)

WAGER, L R, BROWN, G M, and WADSWORTH, WI 1960. Types of igneous cumulates. Journal of Petrology, Vol. 1, 73–85.

WAKEFIELD, M I. 1991. Ostracoda (Crustacea) of the Great Estuarine Group (Bathonian , Middle Jurassic), Inner Hebrides, Scotland. Unpublished PhD thesis, University of Leicester.

WAKEFIELD, M I. 1994. Middle Jurassic ostracoda from the Inner Hebrides, Scotland. Monograph of the Palaeontographical Society, London, No. 596, part of Vol. 148, 1–89.

WALKER, G P L. 1979. The environment of Tertiary igneous activity in the British Isles. 5–6 in Mesozoic and Tertiary volcanism in the North Atlantic and neighbouring regions: Proceedings of the Flett Symposium, 3rd November, 1976. Bulletin of the Geological Survey of Great Britain, No. 70.

WALSH, J N, BECKINSALE, R D, SKELHORN, R R, and THORPE, R S. 1979. Geochemistry and petrogenesis of Tertiary granitic rocks from the Island of Mull, northwest Scotland. Contributions to Mineralogy and Petrology, Vol. 71, 99–116.

WHITE, R S, and McKENZIE, D P. 1989. Magmatism at rift zones: the generation of volcanic continental margins and flood basalts. Journal of Geophysical Research, Vol. 94, 7685–7729.

WHITTAKER, A. and 11 others. 1991. A guide to stratigraphical procedure. Journal of the Geological Society, Vol. 148, 813–824.

WIEBE, R A. 1991. Comingling of contrasted magmas and the generation of mafic enclaves in granitic rocks. 393–402 in Enclaves and granite petrology. DIDIER, J, and BARBARIN, B (editors), Developments in Petrology, Vol. 13. (Amsterdam: Elsevier.)

WILLIAMS, C T. 1978. Uranium-enriched minerals in mesostasis areas of the Rhum layered pluton. Contributions to Mineralogy and Petrology, Vol. 66, 29–39.

WILLIAMS, G E. 1969. Petrography and origin of pebbles from 'Torridonian' strata (Late Precambrian), NW Scotland. 609–629 in North Atlantic: geology and continental drift. KAY, M (editor). Memoir of the American Association of Petroleum Geologists, No. 12.

WILLIAMS, P J. 1985. Pyroclastic rocks in the Cnapan Breaca felsite, Rhum. Geological Magazine, Vol. 122, 447–450.

WILLIAMSON, I T. 1979. The petrology and structure of the Tertiary volcanic rocks of West-Central Skye, NW Scotland. Unpublished PhD thesis, University of Durham.

WILLIAMSON, I T, and BELL, B R. 1994. The Palaeocene lava field of west-central Skye, Scotland: stratigraphy, palaeogeography and structure. Transactions of the Royal Society of Edinburgh: Earth Sciences, Vol. 85, 39–75.

WILKINSON, M. 1992. Concretionary cements in Jurassic sandstones, Isle of Eigg, Inner Hebrides. 145–154 in Basins on the Atlantic Seabord: petroleum geology, sedimentology and basin evolution. PARNELL, J (editor). Special Publication of the Geological Society of London, No. 62.

WILKINSON, M, and DAMPIER, M D. 1990. The rate of growth of sandstone-hosted calcareous concretions. Geochimica et Cosmochimica Acta, Vol. 54, 3391–3399.

WILSON, H E, and MANNING, P I. 1978. Geology of the Causeway Coast. Memoir of the Geological Survey of Northern Ireland

WITHAM, H. 1831. Observations on fossil vegetables, accompanied by representations of their internal structure, as seen through the microscope. (Edinburgh: Blackwood.)

WRIGHT, J K. 1964. The so-called Oxfordian shales of Eigg, Inner Hebrides. Geological Magazine, Vol. 101, 185–186.

WRIGHT, W B. 1937. The Quaternary Ice Age (2nd edition). (London: Macmillan.)

YOUNG, I M. 1984. Mixing of supernatant and interstitial fluids in the Rhum layered intrusion. Mineralogical Magazine, Vol. 48, 345–350.

YOUNG, I M, and DONALDSON, C H. 1985. Formation of granular textured layers and laminae within the Rhum crystal pile. Geological Magazine, Vol. 122, 519–528.

YOUNG, I M, GREENWOOD, R C, and DONALDSON, C H. 1988. Formation of the Eastern Layered Series of the Rhum Complex, northwest Scotland. Canadian Mineralogist, Vol. 26, 225–233.

Appendix 1 Lealt Shale Formation, Eigg [NM 459 870]

Kildonnan Member type section

Bed No. Description Thickness m
9b Stromatolitic algal limestone. Basal 5 cm with planar laminations, middle portion brecciated and upper 10 cm with domes 0.40
9a Placunopsis limestone 0.15
8g Mudstone, grey, 2 cm fibrous calcite at top 0.50
8f Mudstone, brown weathering. Fish scales abundant 0.25
8e Shale, grey, Neomiodon, ostracods and conchostracans near base 0.45
8d Fibrous calcite and shelly layers, small concretions 0.07
8c Shale, grey, Neomiodon at top and bottom, also ostracods 0.45
8b** Limestone, hard, grey, calcilutite. At top a 5 cm layer of fibrous calcite with Placunopsis beneath it 0.15
8a Shale, dark grey with well-preserved Neomiodon and ostracods near top and middle 0.33
7c Limestone, shelly at top, calcilutite below, fibrous calcite at base 0.08
7b Shale, grey at base passing into green mudstone upwards. Valvata, Neomiodon?, fish scales, ostracods and carbonaceous fragments near middle 0.90
7a** UNIO BED
Limestone, sandy and argillaceous, ferruginous base 15 cm. Irregular shale band above ferruginous layer, about 7 cm. Cylindrobullina, neritids, Viviparus, Neomiodon, Unio, heterdont bivalves, indet. plant and fish fragments 0.25
6f Shale, shelly layers persistent over several metres. 8 cm biosparite with 1 cm fibrous calcite layer at base, 20 cm below top of bed. Top 50 cm with Cylindrobullina, Valvata, and ostracods. Main part with Neomiodon. Conchostracans (Dendrostracus) 30 cm from base. Near base Cylindrobullina, neritid gastropods, Neomiodon?, Tancredia 1.80
6e** Bivalve septaria bed. Limestone, argillaceous, poorly preserved bivalves including Tancredia cf. gibbosa (Lycett); also Cylindrobullina and Praemytilus. Underlain by bed of small septarian concretions 0.25
6d Shale, fibrous calcite near middle. Upper and middle parts with ostracods, conchostracans and Valvata?, Cuspidaria, and Tancredia near base. Gastropod concentrations in basal 5 cm; abundant neritids with Cylindrobullina and Valvata 0.60
6c Limestone, sandy, mudcracks at base. Poorly reserved bivalves 0.10
6b Shale, Valvata, Praemytilus and Tancredia? 0.25
6a Limestone, argillaceous, shelly. Top surface is a 1 cm calc-siltstone with shale-filled 0.10 X 0.01 m Rhizocorallium 0.10
5h (v) Shale, grey silty with shells 0.06
5h (iv) Siltstone, calcareous, flat bed. 0.01
5h (iii) Shale, grey, silty with layers of Praemytilus and heterodont bivalves 0.15
5h (ii) Siltstone, calcareous, flat-topped bed 0.04
5h (i) Shale or mudstone, grey, fissile, layers and clusters of Praemytilus jumbled together, not in layers. 'Typical 5h' 0.30
FISH BED 0.10
5g** Variably developed shelly, argillaceous limestone with numerous fish teeth and neritid gastropods, variably cemented 0.05
5f Mudstone, sandy with septaria immediately below 5 g, 20 cm thick. Large (3 cm) shells clustered and across the bedding. Viviparus (abundant), neritids, Valvata, Unio and Praemytilus all together 0.40
5e (iv) Silt, calcareous 0.02
(iii) Shale or fissile mudstone with septaria and lineated Praemytitur, also Valvata and neritids 1.20
(ii) (poor exposure in storm beach) 1.30
(i) Shale or mudstone, Praemytilus abundant and large; best collecting horizon. Cylindrobullina and Valvata 2.80
5d Siltstone, calcareous 0.02
5c (iii) Mudstone, fissile, silty with layers of Praemytilus 2.10
(ii) Two fibrous calcite layers separated by shelly limestone and 6 cm of shale 0.13
(i) Mudstone, fissile, silty, lowest observed large Praemytilus with Valvata 1.20
5b Limestone, shelly with well perserved Praemylilus 0.03
5a (v) Mudstone, fissile, silty, layers of Valvala and small Praemytilus. Fibrous calcite layers 1.30
5a(iv) Limestone, shelly, Praemytilus 0.03
5a (iii) Mudstone, fissile, layers of small Praemytilus SILL. 0.50
5a (iii) Mudstone, fissile, baked for 7 cm below sill 0.25
5a (ii)

Siltstone, shelly, 1mm calcite veins, Valvata and

Praemytilus 0.05–0.12
5a (i) Mudstone, fissile, slightly silty, layers of Valvata and small Praemytilus, conchostracans and ostracods. 10 cm gastropod-rich shell bed at base; Cylindrobullina, neritids and Valvata 1.30
4** COMPLEX BED
4b Limestone, sandy, Praemytilus 0.25
4a Sandstone, coarse, calcareous with abundant fish debris in several thin, lenticular, ripple-marked sandstone beds with shale partings; Cylindrobullina, neritids and Praemytilus 0.70
3h Shale, grey. Large Unio, also Cylindrobullina, neritids, Valvala, Praemytilus, ostracods and fish debris. Praemytilus shell hash forms top 1–2 cm of bed. Large (6 cm long) undamaged Unio occur in this layer 0.05
3g Shale, grey, silty, layers of abundant Valvata, small Praemytilus and subordinate Cylindrobullina and neritids 1.80
3f Limestone, lenticular, composed of small Praemytilus with Cylindrobullina, neritids and Valvata 0.01–0.03
3e Shale 0.04
3d Limestone; lower part fine fibrous calcite, variable 1 cm fissures into bed beneath, upper part shelly with dominant Cylindrobullina and Valvata, occasional neritids and small Praemytilus 0.07
3c Shale, abundant Valvata and Praemytilus 8 cm from base also Viviparus 0.30
3b Limestone, argillaceous; fills mudcracks in bed beneath. With Cylindrobullina, neretids, Valvata and Viviparus 0.02
3a (iii) Shale with fish scales
3a (ii) Lens of limestone like 3b 0.05
3a (i) Shale with fish scales
2** REPTILE BED
Limestone, sideritic and sandy weathers red. Abundant small gastropods (Cylindrobullina, neritids, Valvata), fish scales and teeth, plesiosaur bones. Planar base 0.10
1 Shale, silty, hard, not very fissile. Upper part not very fossiliferous; below 2 m more shelly but preservation poor. Cylindrohullina, Valvata, small Praemytilus, Unio, conchostracans, in various layers and ostracods.
Base not seen

Appendix 2 Lealt Shale Formation, north shore of Eigg [NM 4717 9071] to [NM 4730 9073]

Lonfearn Member

Bed No. Description Thickness m
SILL (1.50)
gap (1.40)
SILL (0.50)
12 Shale, dark, baked. 0.14
11 Limestone, shelly, Neomiodon, fish fragments 0.07
10 Shale, dark, poorly exposed 0.05
9 VIVIPARUS LIMESTONE (0.87 m)
9i Limestone, grey micritic, undulating base; Viviparus 0.20
9h Limestone, shelly, grey 0.30
9g Limestone, micritic, grey. Some shell debris. Conchostracans, ostracods 0.09
9f Limestone, shelly, grey 0.08
9e Limestone, micritic, some shell debris. Conchostracans? 0.06
9d Limestone, micritic, mudcracks filled by shelly material from above 0.06
9c Limestone, grey, shelly 0.01
9b Limestone, grey, micritic, mudcracks in top filled by shelly material 0.05
9a Limestone, shelly, grey. Gradational base 0.02
8b Shale, grey, soft, Neomiodon, Unio?, conchostracans, fish fragments, ostracods 0.20
8a Shale with fibrous calcite 0.02
7b Limestone, shelly, fairly fine grained 0.10
7a Limestone, shelly, laminated, argillaceous, with fibrous calcite, breaks into slabs. Isognomon?, Placunopsis 0.20
6 Shale. Very common Cuspidaria and Quenstedtia? forming shell layers. Aragonite preserved 0.60
5 Limestone, shelly, argillaceous 0.20
4 Shale 0.10
TWO SILLS
3 Shale, grey, Quenstedtia? 0.40
2 Limestone, grey, micritic, fibrous calcite at base 0.10
1 Shale, grey, soft, not very fissile, abundant Cuspidaria, fewer Placunopsis, ostracods 0.50

Kildonnan Member

Bed No. Description Thickness m
ALGAL BED
12 (9b) Algal bed. Domed upper surface. Brecciated middle layer. Laminated basal layer 0.40
11 (9a) Placunopsis limestone; forms base of algal bed 0.10
10 (8) Mudstone; green, shell fragments. Indefinite boundary 2.70
9 (8) Mudstone; grey-green with a concretionary? limestone near top with Viviparus? in it. Mudstone with conchostracans (Dendrostracus and Neopolygrapta? , fish scales, Valvata, ostracods (Darwinula abundant) near top 1.25
SILL (0.60)
8 Shale, darker grey, small bivalves. 0.60
gap (approximate measurement) (4.10)
The 'mudcrack' section:
7 (6d) Siltstone, calcareous, fibrous calcite layers 0.03
6 (6d) Shale, grey, mudcracked top 0.05
5 (6c) Limestone, large shells in finer matrix. Tancredia? 0.04
4 (6b) Shale, grey, mudcracks, heterodont bivalves, 'parallel' cracks. Tancredia? 0.13
3d (6a) Limestone, coarse shelly 0.08
3c (6a) Shale, grey, heterodonts 0.05
3b (6a) Conspicuous layer of Tancredia? 0.01
3a (6a) Shale, grey, heterodonts. 0.05
2 (5h) Shale, lithologically similar with clusters of Praemytilus 0.30
1 Limestone, concretionary 0.10

Appendix 3 Valtos Sandstone Formation and Lonfearn Member (Lealt Shale Formation) , Shieling Burn, Eigg [NM 497 888]

Valtos Sandstone Formation

Bed No. Description Thickness m
SILL coarse dolerite, strongly weathered on top (1.10)
5 Sandstone, baked 0.25
SILL, dolerite (0.65)
4 Sandstone, fine grained 0.45
SILL, forms small waterfall in stream (0.73)
3 Shale, dark grey, baked 0.60
2 Clay, soft, grey 0.50
1c Shell hash, soft and uncemented. Original aragonite Unio 0.30
1b Grey, soft clay with some shell debris 0.07
1a Shell hash as above. Base not seen 0.05
SILL (0.70)

Lealt Shale Formation, Lonfearn Member

Bed No. Description Thickness m
27 Shale 0.25
Complex of thin SILLS, baked shale, fibrous calcite, thin limestones (exposed on S bank) 1.00
gap (below this section in N bank) (4.00)
26 Limestone 0.40
SILL, dolerite, shale caught up in base (2.00)
gap (1.50)
SILL (0.50)
gap (0.50)
25 Limestone, ferruginous, 3 cm fibrous calcite at top 0.35
24 Shale 0.30
23 Limestone, massive, ferruginous, shell fragments parallel to bedding, some superficial ooliths 0.60
SILL, dolerite (0.80)
22 Shale, baked 0.30
SILL, dolerite (0.20)
21 Limestone, massive, ferruginous, superficial ooliths. 4 cm fibrous calcite at top 0.52
gap (3.00)
20 Limestone, massive, fine grained allochems (possibly ostracods) in sparite, 3 cm fibrous calcite at base 0.50
19 Shale, baked 0.05
SILL, dolerite, very prominent (limestone overhangs from above) (1.40)
18 Shale, black, baked with ostracods and conchostracans on surface 0.50
SILL, dolerite (0.50)
17 SILL, dolerite, forms lowest prominent overhang on 0.25
16 Shale, dark then no exposure 2.00
SILL, dolerite (0.40)
15 Limestone with ferruginous micritic intraclasts, 1 cm fibrous calcite at top 0.12
14 Limestone, shelly biomicrite, shell fragments at all angles, soft weathering especially at top 0.60
13 Shale, grey, soft with calcite layers. Sharks teeth, Neomiodon?, Darwinulas 0.60
12 Limestone, ferruginous superficial ooliths and intraclasts. 2 cm fibrous calcite at top 0.28
11 Shale with conchostracans, poorly exposed. Darwinulas in basal 10 cm 0.70
10 Limestone, composed of flat pieces of shell. Slightly ferruginous. 2 cm fibrous calcite at base 0.20
9 Shale, black, baked, Neomiodon 0.30
SILL, dolerite
8 Shale 0.13
7 Limestone, shelly, ferruginous with superficial ooliths? and patches of ferruginous micrite. Shale partings and fibrous calcite veins at bottom and top 0.30
6 Shale, dark 0.04
SILL, dolerite, forms lowest prominent overhang on N bank. Section measured above it (1.00)
5 Shale with fibrous calcite veins. (rather slipped section in S bank) 1.30
4 Shale, dark, slightly baked 0.40
3 Limestone with fibrous calcite 0.14
SILL, dolerite, poorly exposed (0.50)
2b Limestone with superficial ooliths 1.00
2a Limestone, shelly with superficial ooliths. 1 cm fibrous calcite top and bottom 0.12
1 Shale, dark with 1–2 cm fibrous calcite veins, Viviparus, conchostracans 1.50
SILL, dolerite with feldspar phenocrysts (2.00)

Appendix 4 Duntulm Formation and the upper part of the Valtos Sandstone Formation, Camas Mór, Muck [NM 405 792]

Duntulm Formation

Bed No. Description Thickness m
(End of exposure, boulder-covered storm beach above)
39 Limestone, shaly, Praeexogyra-rich. Top not seen 0.80
38 Limestone, pelletal laminated, non- fossiliferous 0.06
37 limestone, nodular and algal, poorly laminated in basal 0.07 m, massive above with irregular top 0.30–0.50
36 Shale, dark, discontinous 0.01
35 Limestone, micritic matrix grades into shale, Praeexogyra hebridica abundant, plus one specimen of P. hebridica var. subrugulosa 1.20
34 Limestone, nodular and algal, discontinuous pods up to 0.05
33 Shale, dark, ostracods, fish fragments, ? bone fragments, pyrite 0.17
32 Limestone, laminated micrite with shelly layers. Corbula?, mm thick algal stringers 0.09
31 Shale, dark, laminated 0.01
30 Limestone, micritic with shelly layers, (thin sections reveal abundant miliolid foraminifera in certain laminations) 0.09
29 Shales and thin limestones; shales locally ostracod rich, Procerithium? Placunopsis, discontinuous algal stringers 0.10
28 Limestone, conspicuous pale grey weathering, rare bivalve debris. Undulose top, concretionary 0.10
27 Shale, dark abundant ostracods and small bivalves indet. 0.08
26 Limestone, laminated micrite, pelletal, top surface is 10 mm thick nodular algal crust 0.10–0.20
25 Limestone, micritic, not obviously fossiliferous 0.09
24 Shale, grey, locally ostracod-rich, small gastropods, Placunopsis
23 Limestone, Praeexogyra-rich, joint planes solution enlarged and infilled by collapse of overlying shale 0.13
22 Shale, dark, Praeexogyra and small bivalves indet 0.02
21 Limestone, Praeexogyra abundant; shaly layer 0.15 m from base, also Praeexogyra-rich 0.64
20 Shale, discontinuous up to 0.02
19 I.imestone, Kallirhynchia, Praeexogyra but more micritic matrix than true oyster limestone; top 2 cm ayer of Corbula or small heterodont bivalves 0.17
18 Limestone and shales gradational, Praeexogyra-rich 0.33
17 Limestone, nodular and algal, massive throughout, irregular top surface 0.17–0.25
16 Limestone, Praeexogyra-rich, reptile bone fragments, discontinuous shale stringer 0.20
15 Mudstone, abraded Kallirhynchia, Praeexogyra-rich 0.22
14 Limestone, nodular and algal, irregular domed top. Pelletal micrite with flow structures between algal heads 0.20–0.40
13 Shale, dark, unfossiliferous 0.04
12 Limestone, Praeexogyra-rich, fish teeth; Kallirhynchia locally very abundant 0.34
11i Shale, green-grey not obviously fossiliferous 0.05
11h Limestone, laterally discontinuous, dark-fine grained, Praeexogyra-rich layer at base 0.07
11g Shale, green-grey ostracod rich in basal 4 cm, Praeexogyra-rich in top 3 cm 0.07
11f Limestone, dark with Praeexogyra weathers red-brown 0.05
11e Shale, grey, small bivalves indet. 0.10
11d Limestone, Praeexogyra 0.01
11c Shale, grey, small bivalves indet. 0.10
11b Limestone, dark, fine-grained, unfossiliferous, weathers red-brown 0.02
11a Mudstone, green-grey, Placunopsis, Quenstedtia, ostracods 0.30
10 Limestone, Praeexogyra-rich, shaly layers, Kallirhynchia 0.10
9 Shale, dark with Praeexogyra 0.05
8 Limestone, Praeexogyra (not rock forming) 0.14
7 Shale, grey, Praeexogyra in small concentrations bivalves indet, contorted bedding 0.14
6 Limestone, Placunopsis, Praeexogyra-rich 0.08
5 Shale, dark, contorted bedding, pyritic Placunopsis, thin limestone stringer laterally impersistent 0.10
4 Limestone, dark green, fine-grained, unfossiliferous 0.10
3 Shale, dark, pyritic Cuspidaria and Placuopteris ostracods 0.20
2 Limestone, green grey, micritic, burrow infills 0.25
1 Shale, dark green at base, laminated, Placunopsis, bivalves smal indet. Fish scales 0.15

Valtos Sandstone Formation Division F
Bed No. Description Thickness m
23 Limestone, green grey, micritic, ostracods on weathered bedding surfaces 0.14
22 Shale, green grey, shelly, fish fragments with limestone stringers, 50 mm-thick shelly limestone near base 0.18
21 Shale, green-grey, Neomiodon, ostracods, conchostracans, Antronestheria 0.15
20 Argillaceous dolomite, yellow-weathering, irregular contact with limestone below, laterally discontinuous up to 0.35
19 Limestone, shelly, pale green-grey, mottled 0.22
18 Mudstone, calcareous, green lenticular, infilling cracks on underlying bed 0.10
17 Limestone, yellow-grey-weathering, Neomiodon, basal

10 cm laminated, surface deeply mudcracked, curl-sided polygons and teepee structures. Patches of red weathering dolomite

 0.30
16 Shale, grey, bivalves indet. 0.18
15 Limestone, shelly, Viviparus, gastropods, Neomiodonrich 0.13
14 Shale, dark, abundant bivalves indet. Grades into limestone at top limestone at top 0.17–0.30
13 Limestone, not obviously fossiliferous, fibrous calcite 0.20
12 Shale, dark, bivalves small indet. 0.07
11 Shale, dark, abundant small bivalves, Neomiodon?, hybodont shark tooth, grades into limestone at top with fibrous calcite 0.30
10 Limestone, grey, shelly buff weathered, Viviparus, Neomiodon 0.12–0.15
9 Shale, dark, small bivalves indet, 1 cm thick green micritic limestone 0.25
8 Shale, dark, limestone cap of fibrous calcite 0.10
7 Limestone, grey shelly, buff weathered, Neomiodon 0.06
6 Shale, dark, Neomiodon rich 0.15
5 Limestone, grey, micritic. Top 2 cm very fine-grained, weathers nodular 0.18
4 Shale, green-grey, Neomiodon?, ostracods 0.10
3 Shales with thin limestones (Neomiodon plasters), fibrous calcite 0.50
2 Limestone; micritic, dark, mottled, conspicuous green weathering ? laminated 0.12
1 Limestone, shelly, Neomiodon, 2 cm-thick fibrous calcite at base. Prominent loadcasting (This is the upper loadcasted bed of Harris (1984) Chapter 3, fig.7. Valtos Sandstone Formation continues below as alternating Neomiodon limestones and shales becomining coarse! sandy toward the base of the exposed section. 0.45
  • (Modified from Andrews, 1984)

    Appendix 5 Chemical analyses of selected rocks

    This appendix contains a selection of rock analyses. Major ele­ments are in weight %, trace elements in ppm. A large volume of rock analyses is contained in several of the unpublished PhD theses and only a very limited selection is given here, together with analyses from the published accounts of Rum and the adjoining islands. In all instances, the specimen numbers are those quoted in the original paper or thesis. Many samples of the Rum dykes and minor intrusions, the dykes on the other islands, and lavas of the Canna Lava Formation and the Eigg Lava Formation have been analysed; however, these data require revision much of which had not been done at the time of writing.

    Contents

    a. Lewisian gneisses from Rum

    A48 A102 A186 A727 DU13773 DU13781 DU13857
    SiO2 66.17 69.92 71.87 81.36 64.86 59.90 63.35
    TiO2 0.35 0.22 0.13 0.11 0.26 0.47 0.48
    Al2O3 17.37 16.43 15.66 9.29 18.06 17.22* 17.75
    FeO 1.85 1.12 1.01 0.93* - - -
    Fe2O3 1.34 0.89 0.60 - 3.24* 6.52* 4.28*
    MnO 0.08 0.06 0.05 0.02 0.05 0.12 0.07
    MgO 1.65 0.93 0.90 0.27 2.15 3.66 2.58
    CaO 3.91 3.20 2.51 0.35 5.57 7.39 5.40
    Na2O 5.01 4.03 4.75 1.48 5.21 3.68 5.01
    K2O 1.00 2.36 1.40 6.44 0.60 1.04 1.08
    P2O5 0.30 0.18 0.20 0.05 n.d. n.d. n.d.
    CO2 0.03 0.40 0.18 - - - -
    Total 99.06 99.74 99.26 100.30 (100) (100) (100)
    Ba 495 1277 453 - - - -
    Co 10 2 4 - - - -
    Cr 39 24 9 - - - -
    Cu 8 63 26 - - - -
    Ga 24 19 19 - - - -
    Ni 27 9 10 - - - -
    Rb 18 112 29 - - - -
    Sr 499 429 286 - - - -
    Zr 465 126 62 - - - -
    Sc 7 3 - - - - -
    Li 16 43 30 - - - -
    La - 54 - - - - -

    Key

    A48 Alkali feldspar-plagioclase-quartz-biotite gneiss, northern end of Meall Breac.
    A102 Alkali feldspar-plagioclase-quartz-biotite gneiss with well-developed interstitial granophyric material. c.180 m NW of the northern end of the Long Loch.
    A186 Alkali feldspar-plagioclase-quartz-biotite gneiss, c.180 m cast of the the northern Priomh-loch.
    A727 Microcline-rich pegmatite 300 m north of the northern end of the Priomh-lochs [NM 3685 9910].
    DU13773 (= SR128). Plagioclase-quartz-alkali feldspar gneiss with altered amphibole. South end of the gneiss outlier on Ard Nev summit [NM 3460 9858].
    DU13781 (= SR137). Plagioclase-hornblende-quartz-biotite gneiss. Hornblende replaced by granular areas of clinopyroxene and orthopyroxene, biotite as poikilitic crystals. High-grade hornfels. Gully 700 m SSE of Ard Nev c. [NM 348 979].
    DU13857 Plagioclase-hornblende-quartz-biotite gneiss. High-grade hornfels with hornblende largely replaced by clinopyroxene and orthopyroxene, also some interstitial granophyric intergrowth between quartz and feldspar. Gully SSE of Ard Nev (as DU13881) [NM 3494 9798]

    b. Porphyritic rhyolites from Rum

    SR333 SR486 A421 A737 A738
    SiO2 73.02 72.84 70.20 72.42 72.03
    TiO2 0.67 0.66 0.68 0.54 0.58
    Al2O3 13.01 12.84 12.61 12.73 12.70
    FeO* 4.01 4.13 4.99 3.75 4.02
    MnO 0.07 0.08 0.11 0.04 0.04
    MgO 0.61 0.46 1.14 0.41 0.46
    CaO 1.54 1.72 1.88 1.54 1.63
    Na2O 4.47 4.71 4.28 4.71 4.70
    K2O 3.45 3.18 3.16 3.17 3.10
    P2O5 0.11 0.10 0.11 0.14 0.13
    Total 100.96 100.72 99.16 99.45 99.39
    Ba 993 1073 — 1243 1317
    Co 2 3 — 2 2
    Cr 3 2 — 7 3
    Cu 7 n.d. 17 18
    Ga 28 27 — 23 28
    Nb 18 15 — — —
    Ni 13 12 — 5 4
    Rb 105 95 — 81 73
    Sc 8 14 — 12 13
    Sr 195 219 — 232 249
    V 25 26 — — —
    Y 34 21 — — —
    Zn 91 85 —— — —
    Zr 339 308 — 306 273
    La 61 30 — — —
    Ce 118 60 — — —
    Nd 57 28 — — —

    c. Granites from Rum

    A701 A702 A705 A706 A707 A708 A709 A710
    SiO2 71.38 70.51 72.00 72.00 72.00 71.56 70.68 70.57
    TiO2 0.44 0.47 0.33 0.39 0.38 0.39 0.47 0.47
    Al2O3 13.24 13.34 13.00 13.18 13.28 13.23 13.48 13.86
    FeO 2.05 2.49 1.54 2.14 1.36 1.80 1.97 1.98
    Fe2O3 2.01 2.25 1.68 1.93 2.61 2.47 2.62 2.41
    MnO 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04
    MgO 0.36 0.50 0.15 0.44 0.41 0.48 0.36 0.34
    CaO 1.35 1.35 1.10 1.37 1.25 0.99 1.38 1.38
    Na2O 4.33 4.01 4.73 3.85 4.11 4.15 4.44 4.52
    K2O 4.37 4.52 4.05 4.30 4.27 4.55 4.13 3.98
    P2O5 0.13 0.14 0.07 0.07 0.07 0.08 0.14 0.14
    Total 99.70 99.62 98.69 99.71 99.78 99.74 99.71 99.69

    c. Granites from Rum continued

    A440 A7904 RH1 RH2 RH3 RH46
    SiO2 69.39 70.20 71.1 70.8 71.1 68.5
    TiO2 0.81 0.67 0.45 0.58 0.57 0.85
    Al2O3 13.82 12.00 12.96 12.95 12.77 13.26
    FeO* 4.95 4.96 3.82 4.21 4.14 5.48
    MnO 0.14 0.10 0.09 0.09 0.09 0.12
    MgO 0.81 1.05 0.29 0.41 0.31 0.61
    CaO 2.24 1.90 1.43 1.51 1.52 2.54
    Na2O 4.06 4.12 4.07 4.32 4.45 4.75
    K2O 2.68 3.31 3.84 3.64 3.66 2.96
    P2O5 0.20 0.13 0.08 0.12 0.12 0.20
    Total 99.10 98.44 98.13 98.63 98.73 99.27
    Ba 1111 1033 960 937 969 934
    Co 4 3 — — — —
    Cr 5 5 — — — —
    Cu 10 14 — — — —
    Ga 20 29 — — — —
    Nb 16 9 — 13 15 12
    Ni 3 4 — — — —
    Rb 99 59 119 118 112 90
    Sc 12 10 — — — —
    Sr 243 197 196 166 159 254
    Y 65 38 — 51 42 42
    Zn 97 48 89 76 70 79
    Zr 297 324 500 390 396 242
    La 59 76 — 52 26 44
    Ce — — — 109 50 101
    Nd — — — 48 22 48

    d. Sgurr of Eigg pitchstone and minor acid intrusions from Eigg

    HE 7406 EA1 EA48 SR303 EC20 EA28 HE7456 EA55
    SiO2 65.51 64.49 69.77 64.37 69.4 61.78 72.14 75.28
    TiO2 L27 1.19 0.74 1.11 0.59 1.44 0.51 0.31
    Al2O3 16.15 15.09 16.57 14.85 14.3 15.44 13.54 12.70
    FeO 3.39 4.54* 1.07* 4.18* 1.0 6.33* 4.43 2.98*
    Fe2O3 1.08 — — — 1.3 — 1.11 —
    MnO 0.14 0.13 0.01 0.13 0.11 0.13 0.23 0.08
    MgO 0.95 1.11 0.14 0.68 0.41 1.96 0.45 0.73
    CaO 2.66 2.57 0.66 2.36 0.92 3.21 1.28 0.83
    Na2O 4.48 4.85 4.33 4.65 4.7 4.60 4.26 3.29
    K2O 4.05 3.71 7.23 4.45 5.9 4.67 2.69 4.38
    P2O5 0.33 0.35 0.11 0.35 0.02 0.72 0.03 0.04
    Total 100.01 98.03 100.63 97.13 98.65 100.28 100.67 100.62
    Ba 2396 2841 1193 2665 — 1647 34 1036
    Co — 3 <1 20 — 6 — 1
    Cr 23 3 <1 2 — 2 10 24
    Cu 13 3 2 2 — 5 5 5
    Ga — 28 25 26 — 27 — 22
    Nb 27 31 31 31 — 18 40 23
    Ni n.d. 16 12 10 — 12 n.d. 12
    Rb 107 71 93 80 — 53 112 125
    Sc — 16 14 16 — 13 — 6
    Sr 270 287 66 269 — 40 59
    V — 54 36 53 — 87 — 11
    Y 46 50 53 46 — 49 92 47
    Zn 97 114 110 108 — 100 234 118
    Zr 581 593 1074 586 — 561 537 446
    La — 77 82 76 — 68 — 110
    Ce — 150 164 158 — 142 — 191
    Nd — 73 75 78 — 78 — 78

    e:i Lavas of the Canna Lava Formation from Rum: Lower Fionchra Member and Orval Member.

    SR 156 SR 157 SR 217 SR 213 SR 235 DU 13846 DU 13847 SR 244D
    SiO2 46.33 44.21 43.70 47.71 47.33 47.26 47.72 46.78
    TiO2 2.29 1.99 2.00 2.09 2.16 2.25 2.21 1.80
    Al2O3 17.56 16.68 16.39 16.74 16.58 17.28 17.94 16.44
    FeO 6.97 8.26 7.98 8.82 10.06 10.10 10.92 4.89
    Fe2O3 5.23 5.30 6.11 5.55 3.62 4.79 1.60 2.33
    MnO 0.16 0.19 0.23 0.21 0.18 0.21 0.18 0.20
    MgO 4.33 8.21 8.50 4.96 4.91 3.44 3.19 8.07
    CaO 11.59 11.41 11.46 8.00 10.04 9.12 9.15 11.22
    Na2O 3.93 2.68 2.04 3.29 3.34 3.90 4.84 1.61
    K2O 0.48 0.23 0.20 0.67 0.73 1.04 0.82 0.75
    P2O5 0.22 0.34 0.44 0.29 0.29 0.36 0.31 0.21
    Total 99.09 99.50 99.05 98.33 99.24 99.75 98.88 97.21*
    Ba 382 97 83 304 325 420 334 147
    Co 47 70 71 65 54 60 55 —
    Cr 99 342 345 113 110 69 75 280
    Cu 94 132 127 98 112 99 96 131
    Ga 31 20 21 21 26 27 27 —
    Nb 17 7 8 8 7 10 9 5
    Ni 77 303 306 85 81 63 77 132
    Rb 2 5 6 9 10 18 10 47
    Sc 25 27 25 23 33 21 26 —
    Sr 778 317 326 444 359 401 408 266
    V 318 351 329 366 417 358 360 —
    Y 24 24 22 30 29 34 32 26
    Zn 89 92 93 93 95 103 95 103
    Zr 204 137 131 113 133 156 124 140
    La 29 10 9 13 18 20 17 —
    Ce 69 26 28 33 45 43 43 —
    Nd 42 17 131 24 29 28 29 —

    e:i continued. Lavas of the Canna Lava Formation from Rum: Upper Fionchra Member and Guirdil Member

    DU9871 SR165 SR9189 SR237 SR230 DU3852 DU13853 SR163G
    SiO2 51.07 49.30 53.75 51.70 53.87 57.35 57.26 72.0
    TiO2 2.24 2.71 2.04 2.07 1.98 2.01 1.94 1.1
    Al2O3 14.62 14.47 15.61 15.27 16.89 15.17 15.22 14.3
    FeO 8.80 4.73 5.78 9.18 3.02 4.41 4.31 4.5*
    Fe2O3 6.36 12.85 5.74 4.31 7.65 5.28 5.40 —
    MnO 0.20 0.21 0.18 0.18 0.16 0.20 0.12 0.1
    MgO 2.68 3.00 2.83 3.05 2.42 2.28 1.80 0.3
    CaO 7.92 8.15 7.25 8.05 5.27 5.51 5.25 1.0
    Na2O 3.89 2.60 4.01 4.57 4.57 4.02 4.53 5.2
    K2O 1.28 1.20 1.59 0.87 2.95 2.25 2.51 1.5
    P2O5 0.36 0.46 0.75 0.50 0.88 0.79 0.79 —
    Total 99.42 99.75 99.53 99.74 99.60 99.27 99.22 (100)†
    Ba 1015 920 1213 903 1474 1905 1676 —
    Co 57 61 34 56 23 21 21 —
    Cr 7 <1 19 <1 <1 2 4 —
    Cu 54 70 9 56 3 1 6 —
    Ga 24 23 26 22 27 21 26 —
    Nb 14 13 16 13 20 22 21 —
    Ni 14 17 21 14 11 11 12 —
    Rb 34 7 27 44 47 70 49 —
    Sc 29 24 26 27 24 16 20 —
    Sr 533 525 635 514 586 733 638 —
    V 298 330 173 302 145 107 101 —
    Y 39 36 42 39 51 46 46 —
    Zn 128 128 127 122 131 132 134 —
    Zr 202 230 242 211 314 348 342 —
    La 43 40 53 42 76 81 80 —
    Ce 84 85 110 87 142 161 158 —
    Nd 52 60 68 56 82 91 90 —

    e:ii Lavas of the Canna Lava Formation from Canna and Sanday

    SR 254 SR 251 SR 252 HC 7515 HC 7566 HC 7582 HC 7585 HC 7588 HC 7599
    SiO2 51.47 48.85 48.92 48.02 49.50 46.83 46.96 46.62 46.30
    TiO2 1.89 1.42 2.07 2.25 2.02 2.15 2.30 2.21 2.52
    Al2O3 15.21 17.81 15.06 16.26 15.25 15.73 15.44 15.44 16.06
    Fe2O3* 12.54 12.48 15.97 16.50 15.31 15.19 15.05 15.65 15.37
    MnO 0.17 0.20 0.14 0.19 0.20 0.20 0.21 0.22 0.20
    MgO 3.41 5.47 4.68 5.54 4.96 7.71 7.19 7.64 7.11
    CaO 7.24 8.66 7.58 8.51 7.66 9.19 9.39 8.95 9.05
    Na2O 3.78 3.48 4.01 3.78 3.76 3.02 3.25 3.09 3.06
    K2O 1.37 0.90 1.35 0.85 1.28 0.53 0.54 0.50 0.47
    P2O5 0.46 0.27 0.42 0.29 0.40 0.23 0.26 0.25 0.33
    Total 97.54 99.54 100.20 102.19 100.20 100.78 100.59 100.57 100.47
    Ba 830 540 691 316 724 254 286 260 226
    Co — 46 53 58 59 66 60 68 64
    Cr 39 121 29 79 10 94 143 101 31
    Cu 39 51 48 101 65 111 108 108 48
    Ga — — — 26 25 24 25 28 24
    Nb 21 5 4 8 10 8 8 8 7
    Ni n.d. 62 27 83 36 116 107 122 68
    Rb 38 13 20 6 19 8 8 8 6
    Sc — 26 32 26 26 27 32 27 25
    Sr 209 743 543 797 508 369 345 364 425
    V — — — 400 318 373 384 361 360
    Y 59 26 27 30 33 30 31 31 25
    Zn 88 107 153 110 114 110 105 107 101
    Zr 364 107 153 108 147 126 138 125 120
    La — 17 26 14 33 13 14 13 13
    Ce — 45 54 31 70 33 34 33 38
    Nd — 24 31 23 46 23 24 23 29

    f:i Lavas of the Eigg Lava Formation from Eigg

    HE7412 HE7422 HE7468 HE7472 HE7644 SR503 SR560 SR561
    SiO2 52.83 47.79 48.39 52.14 45.76 46.35 51.94 53.17
    TiO2 2.12 1.78 1.31 1.81 2.73 2.78 1.81 1.76
    Al2O3 16.05 15.17 14.23 17.38 15.88 16.61 17.57 17.00
    Fe2O3* 10.43 13.27 11.97 10.88 16.70 16.59 11.03 10.54
    MnO 0.16 0.21 0.18 0.21 0.22 0.19 0.14 0.16
    MgO 3.18 6.64 12.22 3.77 6.73 5.32 2.84 2.96
    CaO 5.24 12.42 10.17 7.52 8.40 8.40 9.79 7.03
    Na2O 5.24 2.49 2.43 3.91 3.23 2.62 3.54 3.64
    K2O 3.33 0.42 0.74 2.11 0.49 0.44 1.93 2.25
    P2O5 1.36 0.19 0.15 0.49 0.24 0.27 0.45 0.52
    Total 99.94 100.38 101.79 100.22 100.38 99.47 101.04 100.01
    Ba 1499 172 379 948 354 175 953 1087
    Co 16 61 57 27 70 65 35 31
    Cr <1 567 814 18 3 25 28 26
    Cu 2 53 67 44 16 22 55 39
    Ga 24 21 21 24 25 — — —
    Nb 26 6 5 11 8 6 6 7
    Ni 9 177 360 35 33 29 24 20
    Rb 54 6 16 20 2 6 21 29
    Sc 12 32 32 20 23 38 18 16
    Sr 745 388 288 494 845 510 503 465
    V 130 321 271 245 311 301 263 229
    Y 43 23 20 30 33 37 28 33
    Zn 115 109 86 89 95 92 81 87
    Zr 343 140 108 233 151 181 243 255
    La 83 13 9 38 15 9 31 35
    Ce 165 36 21 79 40 34 65 79
    Nd 97 23 13 43 30 21 36 42

    f:ii Lavas of the Eigg Lava Formation from Muck

    IIM IIM HM HM IIM HM HM 41M HM
    7531 7597 75127 75139 75140 75144 7620 7622 7624
    SiO2 45.26 51.80 51.24 45.90 45.26 46.79 46.19 46.16 46.11
    TiO2 2.75 2.38 2.35 1.51 1.51 2.37 1.51 1.69 1.60
    Al2O3 15.58 16.09 16.02 13.95 13.59 18.32 14.12 14.31 14.46
    Fe2O3* 16.91 11.59 11.90 13.07 13.57 12.92 12.79 13.22 12.96
    MnO 0.22 0.15 0.25 0.18 0.19 0.15 0.18 0.18 0.18
    MgO 7.62 3.76 4.49 12.75 14.10 5.08 12.15 11.81 11.52
    CaO 7.30 6.10 3.84 9.66 9.59 8.64 9.66 9.80 10.19
    Na2O 3.57 4.74 5.14 2.93 2.19 3.83 2.77 2.64 2.50
    K2O 0.54 2.93 2.96 0.31 0.30 1.20 0.45 0.38 0.40
    P2O5 0.27 0.69 0.69 0.17 0.17 0.34 0.19 0.19 0.21
    Total 100.02 100.23 99.16 100.45 100.47 99.64 100.03 100.38 100.13
    Ba 285 1113 1116 132 89 483 373 137 120
    Co 76 26 29 71 79 48 69 67 48
    Cr 26 7 5 978 1035 10 826 729 748
    Cu 12 9 9 136 89 11 99 84 73
    Ga 27 24 26 17 17 24 17 20 21
    Nb 8 21 20 6 6 10 9 7 6
    Ni 33 13 12 387 451 25 364 300 305
    Rb 7 41 49 7 6 16 9 7 6
    Sc 20 14 17 28 28 15 26 28 28
    Sr 422 599 562 323 310 588 455 356 371
    V 326 192 196 268 262 277 257 310 299
    Y 34 22 39 26 24 29 25 22 20
    Zn 98 95 126 94 89 95 96 95 99
    Zr 181 137 372 108 107 197 116 137 130
    La 17 54 56 11 9 24 10 11 10
    Ce 37 109 110 26 22 53 26 33 109
    Nd 47 70 78 16 15 35 16 22 78

    g:i Minor basic intrusions from Muck and Eigg

    HM D2 HM D3 HM D101 HM D111 HM D119 HE 7404 HE 7414 HE 7418
    SiO2 45.66 46.20 55.03 49.09 47.03 48.63 49.22 46.72
    TiO2 1.09 1.92 1.75 1.51 1.81 1.77 1.45 2.07
    Al2O3 18.00 14.86 16.23 15.63 16.79 15.01 14.85 14.62
    Fe2O3* 10.34 13.48 10.85 12.41 14.90 12.67 13.16 13.68
    MnO 0.16 0.20 0.18 0.18 0.18 0.20 0.21 0.19
    MgO 11.85 7.52 2.87 7.52 5.83 6.99 6.52 7.37
    CaO 11.01 11.91 4.24 10.38 10.27 11.29 11.86 12.24
    Na2O 2.11 3.01 5.82 2.84 3.19 2.74 2.63 2.87
    K2O 0.24 0.27 2.13 0.48 0.39 0.80 0.33 0.27
    P2O5 0.10 0.18 0.89 0.19 0.21 0.17 0.12 0.17
    Total 100.56 99.55 99.99 100.23 100.60 100.27 100.35 100.20
    Ba 280 37 976 322 198 548 163 73
    Co 53 57 20 54 71 51 52 53
    Cr 552 210 1 153 38 197 199 174
    Cu 105 180 <1 100 48 95 110 129
    Ga 19 25 29 23 24 23 20 24
    Nb 4 6 23 6 6 8 6 5
    Ni 386 102 11 112 64 75 77 108
    Rb 4 3 24 4 6 17 8 4
    Sc 18 29 11 27 24 27 44 36
    Sr 333 385 505 389 385 477 300 263
    V 208 381 26 317 274 355 362 460
    Y 15 22 61 23 28 24 28 21
    Zn 64 89 118 94 25 83 107 91
    Zr 67 129 546 100 85 140 71 105
    La 4 7 60 11 7 12 6 6
    Ce 12 20 133 31 20 33 11 14
    Nb 9 14 87 20 14 20 7 10

    g. Basic dykes and sheets from Rum

    DU23765 DU23766 DU23767 DU23768 DU23769 DU23771 DU23772 DU23778 DU23786 DU23788
    SiO2 43.16 57.18 47.76 49.01 49.21 46.73 50.77 57.76 45.91 48.19
    TiO2 0.84 1.57 1.44 1.57 1.62 1.46 1.50 1.55 1.77 2.10
    Al2O3 6.74 13.99 13.24 14.25 14.79 13.71 14.29 16.15 14.73 14.45
    FeO* 13.03 8.78 12.33 12.59 12.68 12.20 10.61 6.10 12.39 12.87
    MnO 0.17 0.12 0.18 0.18 0.18 0.20 0.16 0.11 0.18 0.17
    MgO 31.36 4.67 13.42 8.34 7.74 9.10 8.99 2.79 10.63 9.11
    CaO 5.24 6.14 10.19 10.49 10.35 10.19 9.80 4.21 12.56 6.31
    Na2O 0.61 3.82 1.67 2.49 3.17 2.78 2.62 4.98 1.71 3.66
    K2O 0.09 1.58 0.32 0.22 (1.59 0.48 0.61 4.26 0.10 1.96
    P2O5 0.10 0.23 0.14 0.15 0.16 0.18 0.16 0.99 0.14 0.51
    Total 101.35 98.70 100.69 99.29 100.50 97.03 99.51 98.90 100.13 99.31
    Ba 10 533 48 90 71 406 199 1588 5 689
    Co 118 23 59 54 52 54 45 6 54 48
    Cr 1907 103 490 276 178 201 248 <1 270 110
    Cu 72 42 99 135 147 ID 126 <1 136 52
    Ga 16 27 24 27 27 20 28 24 24 28
    Nb 1 13 5 5 5 5 8 31 4 12
    Ni 1532 66 300 153 102 152 146 14 160 141
    Rh 6 46 14 10 32 4 31 67 6 30
    Sc 13 20 25 27 27 29 28 12 31 16
    Sr 126 302 203 228 265 381 255 457 213 614
    V 185 251 310 330 362 321 316 92 371 307
    Y 9 33 19 22 26 21 26 42 23 28
    Zn 72 107 108 99 92 83 109 109 81 115
    Zr 52 262 118 140 127 95 170 481 133 188
    La <1 33 6 8 8 10 16 82 5 33
    Ce 3 68 17 19 23 24 37 157 17 72
    Nd 3 38 12 12 16 14 21 80 12 48

    h. Gabbros and mafic dykes, mainly in the Rum Layered Suite

    M9 B65 B62/2 B7 3.18 X9 B9 X6 K8 FD1
    SiO2 41.74 46.09 45.32 45.25 50.55 44.39 45.95 49.04 49.65 44.48
    TiO2 0.55 0.17 1.70 1.13 0.37 0.30 0.66 1.42 1.43 2.99
    Al2O3 5.54 14.29 13.95 12.92 15.78 20.18 7.54 15.88 13.40 13.47
    FeO* 9.37 10.46 11.48 11.15 6.77 7.21 9.86 10.00 11.22 13.72
    MnO 0.15 0.17 0.18 0.18 0.13 0.11 0.16 0.17 0.16 0.17
    MgO 33.72 13.47 12.46 15.77 9.75 13.14 27.44 7.48 10.74 7.81
    CaO 4.72 10.85 10.95 10.04 13.42 12.10 4.78 11.40 12.11 14.35
    Na2O 0.64 1.96 2.08 1.82 2.00 1.32 1.25 2.74 2.13 1.91
    K2O 0.06 0.10 0.17 0.18 0.11 0.05 0.51 0.42 0.22 0.14
    P2O5 0.05 0.08 0.14 0.10 0.03 0.02 0.06 0.01 0.05 0.01
    Total 96.54 98.64 98.43 98.54 98.91 98.82 98.21 98.56 101.11 99.05
    Ba 31 44 46 55 95 20 165 116 71 54
    Cr 2380 828 609 1137 315 431 2305 266 332 660
    Cu 62 158 131 107 134 80 57 140 204 436
    Ni 2038 405 357 536 109 376 1403 121 174 324
    Zn 58 65 65 67 27 39 69 68 70 60
    Rb 2 3 2 4 n.d. 1 16 7 6 3
    Sr 79 171 356 255 238 261 113 245 221 261
    Y 9 16 18 15 5 7 22 23 18 19
    Zr 30 56 97 77 17 9 81 129 47 69
    Sc 16 27 27 26 — 14 18 35 50 58
    Nb 1 2 3 3 — 2 4 4 2 3
    La n.d. 8 6 4 — n.d. 10 17 <1 n.d.
    Ce n.d. 3 12 8 — 1 18 18 6 4
    Nd <1 3 9 7 — 3 9 12 5 7

    i. Ultrabasic rocks of the Eastern Layered Intrusion, Rum and segregations in the Central and Western layered intrusions, Rum

    5 25 29 LT LP SR209 R5 R7 HR
    SiO2 40.44 43.65 47.91 43.55 38.95 46.06 57.68 46.84 43.43
    TiO2 0.20 0.16 0.24 0.05 0.10 0.76 1.37 1.22 0.24
    Al2O3 4.53 8.44 16.81 22.39 2.02 13.39 18.83 16.05 11.02
    FeO* 12.85 10.97 5.98 5.07 15.30 10.22 2.93 10.36 10.93
    MnO 0.20 0.19 0.09 0.05 0.23 0.18 0.03 0.14 0.02
    MgO 37.80 28.82 13.13 14.75 40.04 15.38 2.67 12.63 24.17
    CaO 3.71 7.08 14.04 11.66 1.23 9.11 6.68 7.87 8.66
    Na2O 0.24 0.64 1.74 1.30 0.09 1.64 8.78 3.35 1.34
    K2O 0.02 0.03 0.05 n.d. n.d. 0.44 0.81 1.11 0.05
    P2O5 0.01 0.02 0.01 0.02 0.01 0.11 0.42 0.28 0.04
    Total (100) (100) (100) 98.84 97.97 97.29 100.20 99.85 99.90
    Ba 10 11 16 — — 42 1180 605 15
    Co 140 135 56 — — — — — —
    Cr 6410 2506 752 203 2515 755 — —1659
    Cu 63 374 65 112 146 115 — — 87
    Ni 1913 1617 253 663 1708 563 51 295 1405
    Zn 76 59 24 25 86 76 67 — —
    Rb 3 3 1 — — 9 301 131 n.d.
    Sr 66 115 228 252 28 115 285 384 172
    Y
    Zr 22 23 26 5 6 62 — — 32
    Sc 14 19 32 5 4 — — — 23
    Nb n.d. n.d. n.d. — — 1 51 n.d. n.d.
    La n.d. n.d. n.d. — — 1 — 7 —
    Ce n.d. n.d. n.d. — — 121 — 59 —

    Appendix 6 Oxygen isotope data from Rum

    Oxygen isotope data from Rum a. Whole rock

    Sample number Grid reference Lithology Rock unit δ18O whole rock (%O SMOW)
    Palaeocene igneous rocks
    SR182 [NG 3394 0019] basalt dyke post-RLS +4.8
    SR158 [NG 3350 0044] lava post-RCC +7.9
    SRI 56 [NG 3290 0099] lava post-RCC +7.0
    SR234 [NM 3387 9928] lava post-RCC +6.1
    DU13846 [NM 3382 9981] lava post-RCC +7.3
    810522–4 [NM 3950 9626] ET U14, ELI −2.9
    810525–3 [NM 3955 9520] ET U14, E11 −0.1
    810522–3 [NM 3940 9645] U U14, ELI +4.2
    810522–1 [NM 3760 9715] U plug +2.3
    810522–2 [NM 3819 9728] U U14. ELI +3.1
    810525–2 [NM 3943 9592] U U13, ELI +3.4
    810525–1 [NM 3959 9611] U U12, ELI +3.4
    R20 [NM 3918 9713] ET U10, ELI +0.4
    R18 [NM 3920 9713] U U10, ELI +6.2
    RG.151 [NM 3901 9731] ET U9, ELI +3.6
    RG225 [NM 3984 9694] ET U8, ELI +1.8
    RG224 [NM 3984 9695] U U8, ELI +4.0
    R30A [NM 3942 9730] U U6, ELI +6.3
    R2 [NM 3619 9904] U LLM, CI +4.0
    R41 [NM 3402 9669] U AMM, WLI −1.8
    R55 [NM 3406 9519] byEG HBM, WLI −4.8
    R59 [NM 3530 9593] U UM, CI −1.5
    RG76 [NM 3688 9219] U LLM, CI +4.7
    ZP42157 [NM 3901 9727] byEG U9 + 21 m, ELI +2.9
    ZP42155 [NM 3903 9729] ET U9 + 17 m, ELI +0.9
    ZP41596 [NM 3908 9736] byEG U8, ELI +6.1
    R11 [NM 4027 9711] ET U3 or sheet, ELI +3.3
    RG82 [NM 3370 9546] byEG HBM, WLI −2.8
    810523/16 [NG 3552 0015] EG LLM, CI +2.9
    RG152 [NM 3926 9738] EG MG, ELI +3.1
    RG153 [NM 3935 9741] EG MG, ELI +1.1
    RG154 [NM 3923 9753] EG MG, ELI +0.4
    RG223 [NM 4039 9721] EG U3, ELI +1.5
    RG57 [NM 3882 9366] EG U3, ELI +2.2
    RG62 [NM 3576 9298] EG (+ quartz) CI +2.8
    RG96 [NM 3364 9551] EG (+ quartz) HBM, WLI +0.6
    RG222 [NM 4019 9735] EG (hybrid) MG, ELI −0.5
    RG56 [NM 3874 9357] EG (hybrid) MG, ELI +2.2
    RG89 [NM 3357 9552] EG (hybrid) HBM, WLI −2.8
    RG155 [NM 3923 9753] EG (hybrid) MG, ELI +2.8
    RG75 [NM 3686 9219] EG (hybrid) CI +1.2
    RG69 [NM 3661 9176] EG (+ sst xenos.) CI +3.4
    810524/13 [NM 3240 9907] mgr WG +2.6
    850524/12 [NM 3285 9896] mgr WG +1.7
    810524/11 [NM 3289 9837] mgr WG +3.3
    R6 [NM 3640 9911] grp NMZ −4.5
    R31 [NM 3350 9557] grp WG −5.6
    810524/8 [NM 3363 9900] grp WG +0.7
    810524/9 [NM 3305 9715] grp WG +1.1
    810524/10 [NM 3326 9890] grp WG +3.1
    RG40 [NM 3371 9773] grp WG +0.5
    RG49 [NM 3032 9877] grp WG +2.0
    RG97 [NM 3345 9560] grp WG −5.1
    RH1 [NM 3290 9840] grp WG +0.8
    RH2 [NM 3400 9840] grp WG −0.7
    RH3 [NM 3440 9780] grp WG −2.8
    RH46 [NM 3640 9900] grp NMZ −6.1
    810523/11 [NM 390 933] R SMZ +2.4
    810525/9 [NM 3781 9475] R SMZ 0
    810525/10 [NM 37889460] R SMZ +1.0
    810525/11 [NM 3780 9430] R SMZ +2.4
    RG55 [NM 3866 9345] R SMZ +6.8
    RG156 [NM 3933 9763] R NMZ +1.7
    Pre-Palaeocene rocks
    RG61 [NM 3562 9254] sst (melted) TC +3.0
    RG67 [NM 3661 9176] sst (melted) TC +4.4
    RG227 [NM 4011 9772] sst (melted) TC (in NMZ) +2.8
    SR517 [NM 3902 9222] sst TC +8.5
    SR518 [NM 3679 9149] sst TC +4.2
    810524/2 [NG 3630 0107] sst TC +10.7
    810524/4 [NG 3622 0216] sst TC +8.6
    810523/14 [NM 3814 9150] sst TC +9.8
    RG235 [NM 4115 9904] sst TC +7.8
    RG230 [NM 4043 9794] sst TC +5.1
    RG229 [NM 4035 9778] sst TC +2.3
    RG228 [NM 4032 9771] sst TC +4.3
    RG66 [NM 3576 9278] sst TC +1.4
    RG234 [NG 3931 0100] sst (brecciated) TC +8.4
    DU13737 [NM 3940 9280] gneiss SMZ +2.4
    DU13773 [NM 3460 9858] gneiss WG +1.4
    810523/12 [NM 3937 9302] gneiss SMZ +3.3
    R13 [NM 3946 9770] gneiss NMZ −0.1

    Oxygen isotope data from Rum b. Mineral oxygen isotope data

    Rum Layered Suite

    Sample

    number

    lithology and rock unit

    		 δ18Opx δ18Opx Δpx-pl (‰) (w/r)c (w/r)c

    (- - - ‰ SMOW - - -)
    R26 ET, U13 +3.4 0.3 0.18
    R24A U, U12 +4.0 0.15 0.14
    R23 U, U12 +1.1 0.4 0.34
    R21 ET, U10 +0.5 0.47 0.38
    R20 ET, U10 +5.0 −2.9 −7.9 0.76 0.64
    ZP42158 U, U10 +2.8 0.24 0.22
    R16 U, U10 +4.8 +2.3 −2.5 0.29 0.25
    R15 U, U9 −2.7 0.96 0.67
    ZP42157 byEG, U9 (+21 m) +5.6 +2.0 −3.6 0.23 0.21
    ZP42149 byEG, U9 (+10 m) +5.9 +5.9 0 0.03 0.03
    R14 ET, U9 (+9 m) +5.7 +5.7 −0.2 0.04 0.04
    ZP42144 U, U9 (+9 m) +6.4 +1.9 −4.5 0.03 0.28
    R30A ET, U6 +4.1 +0.5 −3.6 0.11 0.1

    Others

    Others
    Sample number lithology and rock unit δ18Opx δ18Opx Δpx-pl (‰)

    (- - - ‰ SMOW - - -)
    RS g-rp, WG −0.2 −1.5 +1.3
    R31 grp, WG −3.7
    R1A sst (partly melted), ex-TC +4.5

    Appendix 7 Geological Survey photographs

    Photographs illustrating the geology of Rum and the adjacent island are deposited for reference in the headquarter library of the British Geological Survey, Keyworth , Nottingham NG12 5GG; in the library at the BGS, Murchison House, West Mains Road, Edinburgh EH9 3LA; and in the BGS Information Office at the Natural History Museum (Earth Galleries) , Exhibition Road, London SW7 2DE. The photographs depict details of the various rocks and sediments exposed and also include general views and scenery. A list of titles can be supplied on request. The photograph belong to the D and GN series. The photographs can be supplied a black and white or colour prints and 2 X 2 colour transparencies, at a fixed tariff.

    Official BGS photograph numbers are prefixed by D. Other photographs are prefixed by GN; except where indicated otherwise, GN Series photograph were taken by the author, C H Emeleus.

    Figures, plates and tables

    Figures

    (Front cover) Cover photograph Layered ultrabasic rocks of Hallival from Askival, Rum. (GS 459) (Photographer: C H Emeleus)

    (Figure 1) Geological sketch map of the Rum district, Sheet 60. Based on BGS 1:250 000 Series sheets Tiree and Little Minch, with some additions from the 3rd (1:50 000) edition of Sheet 60 (Rum).

    (Figure 2) Sketch map showing the positions of the Palaeocene central complexes, lava fields and dyke swarms of Scotland and north Ireland. Offshore central complexes (some of late Cretaceous age) are also indicated. Modified after Bell and Emeleus (1988), fig. 1.

    (Figure 3) Location of gneiss outcrops on Rum. 1: Eastern Fiachanis; 2,3: Ard Nev; 4,5,6,7: Priomh-lochs and Long Loch area; 8,9: Am Màm, Meall Breac and Coire Dubh; 10: Beinn nan Stac; 11,12: Dibidil. CL: Canna Lava Formation; t: Triassic strata; T: Torridon Group rocks. Components of the Central Complex: CI Central Intrusion; ELI Eastern Layered Intrusion; WLI Western Layered Intrusion; SMZ Southern Mountains Zone; NMZ Northern Marginal Zone.

    (Figure 4a) Stratigraphical log illustrating textures and structures within the Allt Mór na h-Uamha Member (Applecross Formation) [NM 4210 9892 to 4206 9893], Rum. All structures other than ripples are drawn to scale.

    (Figure 4b) Stratigraphical log illustrating cycles and progressive upward coarsening within the Allt Mot- na h-Uamha Member (Applecross Formation), south shore of Loch Scresort [NM 4210 9892] to [NM4155 9905], Rum.

    (Figure 5) Type-locality stratigraphical log of the Scresort Sandstone Member (Applecross Formation), north shore of Loch Scresort [NM 4164 9987] to [NM 4150 9982], Rum. All structures are drawn to scale.

    (Figure 6) Stratigraphical log of the Sgorr Mhór Sandstone Member (Aultbea Formation), coast north of Glen Shellesder [NG 3290 0240] to [NG 3298 0247], Rum. All structures other than ripples are drawn to scale.

    (Figure 7) Occurrences of Mesozoic rocks in the Small Isles. 1. Triassic; 2. Lower Jurassic; 3. Middle Jurassic Bearreraig Sandstone Formation; 4. Middle Jurassic Great Estuarine Group (lined); 5. Upper Jurassic Staffin Shale Formation; 6 Upper Cretaceous: A. Clach Alasdair; B. Laig Gorge; C. Allt Ceann a'Gharaid. The positions of figures 9, 11, 12, 14 and 16 are indicated.

    (Figure 8) Palaeogeographical reconstruction of the Sea of the Hebrides and Inner Hebrides basins in the Triassic Period. (Modified after Steel, 1977).

    (Figure 9) Distribution of constituent members of the Monadh Dubh Sandstone Formation (Triassic) in outliers in northwest Rum; based on field maps by R J Steel.

    (Figure 10) Succession in the Monadh Dubh Sandstone Formation (Triassic), north-west Rum

    (Figure 11) Geological sketch map of the Allt nam Bà area, eastern Rum, showing the setting of the Broadford Beds (Lower Jurassic) and Palaeocene lavas in relation to the Main Ring Fault, and the Lewisian gneisses and Torridonian rocks within the fault system. (Based on 1:10 000 geological sheet NM49NW and Smith, 1985, fig.1.)

    (Figure 13) and Appendix 1. (Modified from Hudson, 1966 fig.1A.)" data-name="images/P936583.jpg">(Figure 12) Geological sketch map of the type locality of the Kildonnan Member (Lealt Shale Formation), eastern Eigg [NM 491 872]. Bed numbers 1–8g relate to (Figure 13) and Appendix 1. (Modified from Hudson, 1966 fig.1A.)

    (Figure 13) Schematic log of the type section of the Kildonnan Member (Lealt Shale Formation), near Kildonnan, Eigg, showing bed numbers, lithology and principal fossils. (After Wakefield, 1991.) For detailed log see Appendix 1.

    (Figure 14) Outcrop map of the Valtos Sandstone Formation from the Bay of Laig to the Blàr Mór area, north-west Eigg. (After Harris, 1984.) Detailed graphic logs 1–6 are given in Harris (1992).

    (Figure 15) Lithostratigraphical and facies correlation of the Valtos Sandstone Formation, Bay of Laig, Eigg. (After Harris, 1984, 1992.) See text for explanation of Facies 1–5. Locations of logs 1–6 are shown on (Figure 14).

    (Figure 17), ((Figure 18)) and Appendix 4." data-name="images/P936587.jpg">(Figure 16) Outcrop map of the Mesozoic rocks at Camas Mór, Muck. (Modified from Harris, 1984.) For detailed logs see (Figure 17), ((Figure 18)) and Appendix 4.

    (Figure 17) Correlation of sections of the Duntulm Formation in the Inner Hebrides (after Andrews and Walton, 1990, fig. 10). See text for explanation of Facies 1–3a.

    (Figure 18) Correlation of sections of the Kilmaluag Formation in the Inner Hebrides (after Andrews, 1985, fig. 4).

    (Figure 19) Sections in the Cretaceous rocks of Eigg.

    (Figure 20) Geological map of the Northern Marginal Zone, Rum Central Complex.

    (Figure 21) Geological map of the Southern Mountains Zone, Rum Central Complex.

    (Figure 22) Pyroxenes from the Rum rhyodacites and Western Granite, and from members of the Sgurr of Eigg Pitchstone Formation. Rhyodacite cobble in conglomerate, West Minishal (R9870c) 2, 2a.Rhyodacite, Beinn nan Stac (H3318) 3. Western Granite (DU9904) Al, A2. Sgurr of Eigg pitchstone. B.Oigh-sgeir pitchstone. TG. Trend for analysed pyroxenes in Tertiary acid glasses (after Carmichael, 1960b). SK. Trend for pyroxenes in the Skaergaard Intrusion (from Wager and Brown, 1968).

    (Figure 23) Diagrammatic cross-sections through the Coire Dubh Breccias and rhyodacite ash flows at the eastern end of Cnapan Breaca [NM 3955 9750] , Northern Marginal Zone, Rum. A. Section NNE to SSW, from Torridon Group Fiachanis Gritty Sandstone Member (TCDF) to rhyodacite. Coarse Coire Dubh Breccia occurs at A, this is overlain by fine-grained, bedded breccia and crystal tuff which are exposed at B (see (b) below), followed by a thick layer of coarse breccia to D. At C the coarse breccias show weak bedding and they are interbedded with sparse fine-grained breccia and tuff. Coarse, bedded tuffs and interbedded rhyodacite and tuff occur from D to F (see diagram (c)) where they are overlain by the main body of rhyodacite ash flows. The total thickness of the section from A to F is about 50 m. T = tuffisite dyke. B. Detail of fine-grained bedded layers near the base of succession, at B. C. Generalised section through the interbedded coarse tuffs and rhyodacite ash flows near top of the succession (D to E on (Figure 23)a).

    (Figure 24) Normative quartz (Qz), albite (Ab) and orthoclase (Or) for the Rum rhyodacites (open squares), granites and granophyres (inverted open triangles). Also data for Rum gneisses (filled circles). The line shows the position of the cotectic at 1 kb pH2O; the field SK is that for granites and granophyres from Skye. (Based on Brown, 1963 and Dunham and Thompson, 1967, and unpublished data by A C Dunham.)

    (Figure 25) Chondrite-normalised multi-element diagram for (a) rhyodacites and (b) microgranites from Rum, with comparative analyses of rocks from Mull and Skye. SR333 Porphyritic rhyodacite. c.200 m east of Beinn nan Stac summit [NM 3970 9408]; SR486 Porphyritic rhyodacite. North summit of Meall Breac [NM 3825 9847]; R112 Microgranite. Upper part of Glen Duian [NM 340 984]. RH3 Microgranite. South of Ard Nev [NM 344 978].; M177 Beinn a'Ghraig Granophyre. North-west end of Loch BA, Mull, c. [NM 555 394].; SK69 Glamaig Granite. Allt na Measarroch, Glen Sligachan, Skye [NG 500 272].; SK127 Southern Porphyritic Granite. North-east side of Marsco, Skye [NG 509 257].; Normalisation as in Thompson, 1982. For additional details, see Appendices 5b and 5c.

    (Figure 27) for the localities of plugs within the Northern Marginal Zone, (Figure 21) for the Southern Mountains Zone and (Figure 35) for detail within the Layered Suite. The numbers 1-13 refer to the localities where dyke orientations have been measured (see (Figure 28))." data-name="images/P936601.jpg">(Figure 26) Distribution of plugs, dykes, sills and fissure breccias on Rum. The map shows occurrences outside the Main Ring Fault. See (Figure 20) and (Figure 27) for the localities of plugs within the Northern Marginal Zone, (Figure 21) for the Southern Mountains Zone and (Figure 35) for detail within the Layered Suite. The numbers 1–13 refer to the localities where dyke orientations have been measured (see (Figure 28)).

    (Figure 27) Distribution of inclined sheets (cone-sheets) within and north of the Northern Marginal Zone, Rum. The principal focus of the sheets is between 1 and 1.5 km below sea level beneath Glen Harris, at about [NM 381 971].

    (Figure 28) Rose diagrams showing the distribution of dyke trends on Rum. The numbered localities are shown in (Figure 27) for the localities of plugs within the Northern Marginal Zone, (Figure 21) for the Southern Mountains Zone and (Figure 35) for detail within the Layered Suite. The numbers 1-13 refer to the localities where dyke orientations have been measured (see (Figure 28))." data-name="images/P936601.jpg">(Figure 26). The numbers in brackets refer to the number of dykes measured. Data from Forster (1980).

    (Figure 29) Chondrite-normalised multi-element diagrams for basic dykes and minor intrusions: a. Eigg - HE7404 Aphyric basalt. Shoreface, north end of Laig Bay [NM 4692 8925]; HE7414 Olivine-dolerite. Path south-east of Laig Farm [NM 471 875]; HE7418 Aphyric basalt. Cliff, to south-east of Cleadale [NM 4796 8829]; b. Muck - HMD2 Plagioclase-olivine-phyric basalt. West side of Port Mór [NM 4221 7913]; HMD3 Plagioclase-olivine-phyric basalt. West side of Port Mór [NM 4223 7903]; HMD101 Olivine-phyric hawaiite. South-west coast, west of Camas Mór; HMD111 Plagioclase-olivine-phyric basalt. Camas Mór [NM 4068 7927]; HMD119 Aphyric basalt. Foreshore at Camas Mór [NM 4056 7913]; SK940 Basalt. Arnisort Church, Skye [NG 348 532] — see Thompson et al., 1972; Thompson, 1982; c. Rum - DU23766 Tholeiitic andesite dyke. Allt Slugain a'Choilich [NM 3935 9835]; DU23767 Basalt inclined sheet. Allt Slugain a'Choilich [NM 3945 9859]; DU23768 Basalt inclined sheet. Allt Slugain a'Choilich [NM 3940 9846]; DU23778 Basalt inclined sheet. Coast near Papadil [NM 3657 9176]; DU23788 Basalt dyke Minishal [NM 3548 9968]; The profile of SK940 is included for comparison. Normalisation as in Thompson (1982). For additional details, see Appendices 5g:i (Eigg, Muck) and 5g:ii (Rum).

    (Figure 30) Structure of the Rum Layered Suite.

    (Figure 31) Stratigraphical correlation in the Eastern Layered Intrusion, Rum. Inset map shows location of sections. (After Volker and Upton, 1990, fig. 5).

    (Figure 32) Cross-sections through the margins of the Layered Suite, Rum. A. Askival–Beinn nan Stac–Sound of Rum. b. NNE through Cnapan Breaca c. Ard Nev and slopes to the south-east. For lines of section, see (Figure 30).

    (Figure 33) Cryptic variation in units of the Eastern Layered Intrusion a. Variation in mode, olivine composition (Mg/Fe as % Fo, and Ni) and whole-rock 87Sr/86Sr across the main peridotite-bytownite-troctolite boundary within Unit 10 [Dashed line wqith points] - from a series of closely spaced samples, [Dashed line without point] from the main series of samples (after Tait, 1985, fig.4). b. Mineral compositional variation in Unit 10 and the adjoin­ing units. • = cumulus minerals, o = intercumulus minerals, [Solid line] = range of compositions of intercumulus plagioclase (after Dunham and Wadsworth, 1978, fig. 1). The Mg number, Mg* = atom. % 100 Mg/ (Mg + Fe). C. Variation in olivine and pyroxene compositions in the lower units of the intrusion. sss = prominent slumping (after Faithfull, 1985, fig. 5).

    (Figure 34) Variation in 87Sr/86Sr through the upper part of the Eastern Layered Intrusion, Hallival, Rum. Based on Palacz, 1985, fig. 3; Palacz and Tait, 1985, fig. 3; Renner and Palacz, 1987, fig. 2.

    (Figure 35) Distribution of gabbro sheets and plugs, and peridotite plugs in the Rum Central Complex and the surrounding rocks.

    (Figure 36) Geological map of western Harris Bay, Rum, showing details of the contact between the Western Layered Intrusion and the Western Granite. (Based on Greenwood, 1987, fig. 2.4 and Emeleus and Forster, 1979, fig. 14.)

    (Figure 37)a Sketch map showing locations of specimens analysed for oxygen isotopes, together with 818O for each sample. The line A–B marks the position of the profile in (Figure 37)b. (See also Appendix 6.) b Profile through the Eastern Layered Intrusion, Rum Central Complex from Askival to Hallival and Loch Scresort showing the variation in δ18O. Black dots = sample localities, Red dots = corresponding δ18O values. Abbreviations: T. Torridonian rocks; EG. gabbro; ET. bytownite-troctolite*; U. peridotite*, P. hybrid rocks.(*, subscript numbers refer to units in the Eastern Layered Intrusion. Based partly on Greenwood et al., 1992.)

    (Figure 38) Plot of O-isotopic variations in the Eastern Layered Intrusion of the Rum Layered Suite. The large range in δ18O plagioclase values (c.9‰ ) for a small range in δ18O pyroxene values (c.2‰ ) is indicative of subsolidus hydrothermal exchanges at elevated temperature. The diagonal lines labelled “ΔF-P at 500°C” and “ΔF-P at 1000°C” denote the equilibrium fractionation of oxygen isotopes between plagioclase and pyroxene at magmatic temperatures. The black dot denotes δ18O values for plagioclase and pyroxene expected for ultramafic magma with a bulk 8180 value of c.+ 6.5‰. The shaded triangular area at the upper right-hand corner of the figure shows the space accessible through closed system internal 0-isotope equilibration with decreasing temperatures from 1000°C to 500°C for a range of mineralogical compositions between 100% plagioclase and 100% pyroxene. The curves radiating from the black dot towards the lower lefthand side of the figure labelled “δ18O = 5”, and “δ18O = 14” respectively indicate the trajectories that would be followed by a rock having +6.5‰ plagioclase and +6.5‰ pyroxene as it is progressively affected by fluid-rock O-isotope exchange during subsolidus hydrothermal alteration with a fluid of the designated O-isotope composition. Meteoric waters in the British Tertiary Volcanic Province likely had δ18O values of about — 10 ± 2‰ (Forester and Taylor, 1976; 1977).

    (Figure 39) Distribution of the lava formations on Rum and the adjacent Small Isles.

    (Figure 40) Generalised vertical sections of the Canna Lava Formation on Canna/Sanday and on Rum, and the Eigg Lava Formation on Eigg and Muck. Equivalent units established by Allwright (1980) on Eigg, Muck and Canna/Sanday are indicated where possible.

    (Figure 41) Sketch map of eastern Camas Me:1r, Muck, showing the distibution of the Camas Mór Breccia and adjoining rocks.

    (Figure 42) Basaltic andesite lava flow and conglomerate of the Eilean a'Bhàird Member, Canna Lava Formation at Eilean a'Bhàird, Canna Harbour. (After Allwright, 1980, figs. 2.4.13 and 14.)

    (Figure 43) Canna Lava Formation succession at Compass Hill, Canna (after Harker, 1908, fig. 8).

    (Figure 44) Sketch map showing the distribution of members of the Canna Lava Formation on north-west Rum, and the direction of drainage in the palaeovalleys. (After Emeleus, 1985, fig. 4.)

    (Figure 45) Distribution of the Small Isles Palaeocene lavas on the Normative 01-Di­Hy-Ne diagram. (Selected data from Allwright, 1980 and Emeleus, 1985; includes information from Thompson, 1982, fig. 2.)

    (Figure 46) Distribution of the Small Isles Palaeocene lavas on the alkali: silica diagram. Based on selected data from Allwright, 1980 and Emeleus, 1985. Also shown are trends for the Skye lavas (after Thompson et al., 1972, fig. 6: I = basalt–benmoreite trend, II = basalt–trachyte trend) and the dividing line between Hawaiian alkalic and tholeiitic basalts (after Macdonald and Katsura, 1964, fig. 1). The dashed line, III, separates hypersthene-normative and nepheline-normative lavas from Skye.

    (Figure 47) Chondrite-normalised multi-element diagrams for lavas from Eigg, Muck, Rum and Canna. a. Eigg Lava Formation on Eigg; HE7412 Mugearite. Cliff, c.400 m south-east of Laig Farm [NM 4700 8730].; HE7422 Olivine-plagioclase-phyric basalt. Dunan Thalasgair [NM 4783 9053].; HE7644 Feldspar-phyric basalt. About 150 m south-east of the school [NM 4816 8619].; HE7468 Olivine-basalt. South-east side of Sliabh Beinn Tighe [NM 4541 8712].; HE7472 Plagioclase-olivine-phyric basalt. Loch Beinn Tighe [NM 4519 8619].; SR560 Feldspar-phyric hawaiite. North-west of Gualainn na Sgurra [NM 4664 8479].; SR561 Feldspar-phyric hawaiite or mugearite. Flow underlying SR560 [NM 4665 8478].; b. Eigg Lava Formation on Muck; HM75127 Aphyric mugearite. North side of Camas Mór [NM 4084 7933].; HM75139 Olivine-basalt. West side of Fionn-aird [NM 4158 7856]. HM75140 Olivine-basalt. South-east end of Fionn-aird [NM 4168 7856].; HM75144 Feldspar-phyric hawaiite. West of Port an t-Seilich [NM 4186 7839].; HM7624 Olivine-basalt. Cliff at south end of Beinn Airein [NM 4024 7899].; c. Canna Lava Formation, Lower Fionchra Member on Rum; SR156 Plagioclase-olivine-pyroxene basaltic hawaiite. c.1200 m WNW of Fionchra summit [NG 328 010].; SR157 Olivine-basalt. Stream section in upper Guirdil, at [NG 3347 0013] .; SR213 Hawaiite. North-east of Orval, at [NM 3347 9952].; SR217 Olivine-basalt. Stream section in upper Guirdil, at; [NG 3368 0012].; d. Canna Lava Formation, Upper Fionchra and Guirdil Members on Rum; Upper Fionchra Member; DU9871 Basaltic andesite. Fionchra summit [NG 3394 0037].; SR165 Basaltic andesite. WNW of Fionchra summit [NG 3373 0055].; SR189 Basaltic andesite. North-east side of Bloodstone Hill [NG 3168 0066].; Guirdil Member; SR230 Icelandite. Cliff on the south-west side of Fionchra, at [NG 3363 0038].; DU13852 Icelandite. Upper flow on Bloodstone Hill; [NG 3161 0048].; e. Canna Lava Formation, Orval Member on Rum; SR235 Basaltic hawaiite. c.450 m ENE of Orval summit, at [NM 3383 9926].; DU13846 Hawaiite. North end of Orval [NM 3382 9981]. DU13847 Hawaiite. Cliff south of Guirdil, at [NM 337 997].; f. Canna Lava Formation on Canna; SR251 Feldspar-phyric basaltic hawaiite. Beinn a'Geugh Sgorr, Canna, c.[NG 260 065].; SR252 Aphyric hawaiite. South-east side of Beinn Tighe, Canna, at [NG 2545 0613].; HC7515 Aphyric olivine-basalt. 14 km north-west of Compass Hill, Canna, at c.30 m depth.; HC7599 Plagioclase-olivine-phyric basalt. An t-Oban, Sanday [NG 2844 0426].

    The profile of SK940, a basalt from Arnisort Church, Skye [NG 348 532] (see Thompson et al., 1972; Thompson, 1982) is included for comparison in diagrams a, b and c. Normalisation as in Thompson, 1982. For additional information see Appendices be (Canna Lava Formation) and 5f (Eigg Lava Formation).

    (Figure 48) Map of the Sgurr of Eigg pitchstone showing structures within the pitchstone and the approximate positions of former valleys eroded in the Eigg Lava Formation, which were filled by pitchstone flows and, locally, by fluviatile conglomerate. (Based on Allwright, 1980, figs. 6.1, 6.7 and 6.8, with additional data from the resurvey.)

    (Figure 49) Sketch of the western end of the Sgurr of Eigg pitchstone at Bidein Boidheach. (Based on Allwright, 1980, fig. 6.4b, and a field sketch.)

    (Figure 50) Mineral compositions in pitchstones from the Sgurr of Eigg. Sources of analyses: dots, from Ridley, 1973; triangles, from Carmichael, 1960a and 1960b; squares, minerals in SR303A (from Oigh-sgeir) analysed by C H Emeleus. A. Feldspar phenocrysts. Dashed line is limit of solid solution in natural feldspars (from Tuttle and Bowen, 1958). Shaded field shows range in composition of feldspar phenocrysts in felsic sheets intruding the Sgurr pitchstone (analyses by C H Emeleus).b. Clinopyroxene and orthopyroxene phenocrysts. TG, trend for analysed pyroxenes in Tertiary acid glasses (after Carmichael, 1960b); SK, trend for analysed pyroxenes in the Skaergaard Intrusion (from Wager and Brown, 1968).

    (Figure 51) Felsic rocks from Eigg and Oigh-sgeir plotted on the normative quartz-albite-orthoclase diagram. (Based on Allwright, 1980, fig. 11.1a.)

    (Figure 52) Chondrite-normalised multi-element diagrams for the Sgurr of Eigg Pitchstone Formation and minor acid and intermediate intrusions from Eigg, with comparative data from Skye and Mull. EA1 Sgurr pitchstone. East end of Cora Bheinn [NM 4577 8555].; EA28 Grulin Felsite. Near Grulin, [NM 4598 8456].; EA48 Felsite sheet cutting Sgurr pitchstone. c.700 m NNW of An Corrach, at [NM 4481 8573]. EA55 Sròn Sgaileach felsite. North Eigg, at c. [NM 4832 9133]. SR303B Oigh-sgeir pitchstone, c.[NM 156 963]. M177 Beinn a'Ghraig Granophyre. North-west end of Loch BA, Mull, c.[NM 555 394]. SK69 Glamaig Granite. Allt na Measarroch, Glen Sligachan, Skye [NG 500 272]. SK127 Southern Porphyritic Granite. Stream on NE side of Marsco, Skye [NG 509 257]. SK69, SK127 (Thompson, 1969; 1982) and M177 (Walsh et al., 1979) are included for comparison. Normalisation as in Thompson, 1982. For additional details see Appendix 5d.

    (Figure 53) Structure of Rum.

    (Figure 54) Components of the Main Ring Fault at Allt flan' Ba and on south-eastern Beinn nan Stac, Rum. (Based on parts of 1:10 000 sheets NM 39 SE and NM 49 SW.)

    (Figure 55) Structure of Eigg and Muck.

    (Figure 56) Structure of Canna and Sanday.

    (Figure 58)." data-name="images/P936658.jpg">(Figure 57) Bouguer anomaly map of Rum, Eigg and Canna, and adjacent sea areas based on observations reported by McQuillin and Tuson (1963) and Coppin (1982) on land, and by BGS at sea (Binns et al., 1974). Gravity contours are at 10 mGal intervals. The gravity values are connected to the National Gravity Net 1973 and use a reduction density of 2670 kg/m3 throughout. The generalised regions occupied by the ultrabasic and gabbroic rocks, and by the acid rocks with associated breccias, gneisses, etc. are shown. The line A–A' marks the position of the profiles in (Figure 58).

    (Figure 58)a. Model Rum 1 for the north–south profile A–A' (Figure 58)." data-name="images/P936658.jpg">(Figure 57). The subsurface density distribution uses end corrections to simulate the three dimensional shape. A uniform density contrast of 350 kg/m3 relative to the sub-Torridonian basement has been assumed. Densities are shown in kg/m3. The Torridonian sedimentary rocks are assumed to be two dimensional. The density of the intrusive complex in this model is approximately that of the feldspathic peridotite which forms the Eastern Layered Intrusion (Chapter 7). b. Model Rum 2 for the north–south profile A–A' uses a density contrast to the basement of 400 kg/m3 down to 2.5 km depth and 200 kg/m3 below this depth. This represents a peridotite underlain by gabbro. Other details are as in (Figure 58)a.

    (Figure 59) Simplified aeromagnetic map of the Small Isles. Based on the 1:250 000 aeromagnetic anomaly maps Tiree and Little Minch. The line A–B refers to the traverse along which Robinson and McClelland (1987) measured the palaeomagnetisation of Torridon Group rocks on Rum (see text).

    (Figure 60) Summary of the palaeomagnetisn and age determinations f the Palaeocene and Eocene igneous rocks of the Small Isles. (Modified after Mussett et al., 1988, fig. 2.)

    (Figure 61) Pleistocene and Recent deposits and features on Rum. Note: height of back-feature (roman) and bench (italic) are in metres above local Ordnance Datum (OD), mean sea-level (mg) or high water mark (hwm). Data from McCann and Richards 1969, Ryder in McCann 1969, local survey by Abney Level, and interpolation (Peacock, 1969). Heights are approximate only.

    (Figure 62) Pleistocene and Recent geology of the area around Cleadale, Eigg. Based on 1:10 560 scale field sheets of J D Peacock, with some additions from resurvey.

    (Figure 63) Section through Cyclic Unit 1 in the lower part of the Eastern Layered Intrusion at Allt nam Bà, Rum [NM 407 945] showing the distribution of nickel and of platinum group elements plus gold (see Hulbert et al., 1992).

    (Figure 64) Chromium and magnesium distribution in surficial marine sediments off Harris, Rum.

    (Figure 65) Chromium and magnesium distribution in surficial marine sediments off Dibidil, Rum.

    Plates

    (Frontispiece) View from Laig Bay, Eigg, across the Sound of Rum to Askival and Hallival, eastern Rum. Rocks of the Jurassic Valtos Sandstone Formation form the shore in the foreground. They are cut by an in-weathering Palaeogene dyke with upstanding walls of indurated sandstone. Layered ultrabasic rocks form the high peaks on Rum, the lower slopes are formed by Torridonian sandstones. A prominent marine notch occurs along the coast of Rum about 30 m above sea level. (GN42)

    (Plate 1) Landslips and other Pleistocene and Recent deposits overlying Mesozoic rocks at Cleadale, Eigg. The high cliffs are formed by lavas of the Palaeocene Eigg Lava Formation. (D1701)

    (Plate 2) Outcrops of banded Lewisian granodioritic biotite gneisses with amphibolitic layers. The rocks are thermally metamorphosed and fall into Tilley's (1944) class 4. East of the Priomh-lochs [NM 371 987], Rum. (GN43)

    (Plate 3a) Torridon Group on Rum. Scale: staff 1.5 m. (Photographs: P G Nicholson) Typical 'shales' (interbedded mudstones, siltstones and very fine- to fine-grained sandstones) of the Laimhrig Shale Member (Diabaig Formation) at Bàgh na h-Uamha [NM 4203 9718]. Staff rests beside a bed of low-angle, cross-stratified medium-grained sandstone lying near the base of the Allt Mór na h-Uamha Member (Applecross Formation). (GN44)

    (Plate 3b) Torridon Group on Rum. Cyclically interbedded siltstones and sandstones of the Allt Mór na hUamha Member (Applecross Formation), Cro nan Laogh, south shore of Loch Scresort [NM 4208 9893]. (GN45)

    (Plate 3c) orridon Group on Rum. Horizontally stratified to low-angle cross-stratified medium-grained sandstone, erosively overlain by coarse-grained sandstone displaying trough and compound cross-strata. Scresort Sandstone Member (Applecross Formation), north side of Loch Scresort [NG 4204 0004]. (GN46)

    (Plate 3d) Torridon Group on Rum. Horizontally stratified and cross-stratified, fine- to medium-grained sandstone from the type-locality of the Sgorr Mhór Sandstone Member (Aultbea Formation), coast south of Guirdil Bay, at [NG 3107 0092]. (GN47)

    (Plate 4) Basal Triassic cornstones overlying and permeating Torridonian rocks. Thin basaltic sheets cut the Torridonian rocks near the base of the cliff. Coast of north-west Rum, about 1 km north of Glen Shellesder. (GN48)

    (Plate 5) Carbonate concretions in the Valtos Sandstone Formation, Bay of Laig, Eigg. (D1709)

    (Plate 6) Cnapan Breaca and Hallival from north, Rum. (GN49) Rocks of the Northern Marginal Zone form the low foreground (Corrie Dubh Breccia) and the pale crags and outcrops on Cnapan Breaca (rhyodacite ash flows). The sharp break at the base of the Cnapan Breaca crags marks the position of bedded tuffs and fine-grained sandstone. Members of the Eastern Layered Intrusion form the grassy area in Coire Dubh (marginal, easy-weathering gabbro) and the terraced slopes leading up to Hallival (layered bytownite-troctolite and peridotite).

    (Plate 7) Aligned fiamme in the base of a rhyodacite ash flow at the southwest end of Meall Breac, Rum [NM 3838 9807]. Scale: knife c.12 cm. (GN50)

    (Plate 8a) Photomicrographs of the Rum rhyodacites a. Zoned plagioclase phenocrysts, quartz crystals and lithic fragments. Beinn nan Stac. (SR333, X10, crossed polarisers) (GN51)

    (Plate 8b) Photomicrographs of the Rum rhyodacites b. Fine-scale banding and flattened shards in rhyodacite matrix. North side of Dibidil. (DU13876, X100, plane polarisers) (GN52)

    (Plate 8c) Photomicrographs of the Rum rhyodacites c. Rounded, lobate mafic areas in rhyodacite of plug, north of Cnapan Breaca. (SR402, X20, plane polarisers) (GN53)

    (Plate 9) Matrix-supported sandstone fragments in the Coire Dubh Breccia in Coire Dubh, Rum [NM 390 980]. Scale: hammer c. 40 cm. (GN54)

    (Plate 10) Cliffs of granophyre cut by basaltic dykes and inclined sheets at A' Bhrideanach, the western end of Rum. The prominent wave-cut platform at about 20 m above the shore­line is a pre-Late Devensian coastal feature. (D2559)

    (Figure 11) Minor intrusions (dykes, cone-sheets) cutting rhyodacite and Coire Dubh-type breccias on the eastern face of Ainshval. Southern Mountains Zone, Rum. (GN55)

    (Figure 13) and Appendix 1. (Modified from Hudson, 1966 fig.1A.)" data-name="images/P936583.jpg">(Figure 12) Camas Mór, Muck; view from Beinn Airein, showing dykes of the Palaeocene Muck Swarm cutting sedimentary rocks of the Jurassic Duntulm and Valtos Sandstone formations on the foreshore. The Camas Mór gabbro dyke forms the cliffs to the right beyond the bay, the other cliffs are made of flows of the Eigg Lava Formation which give the strong trap-featuring in the distance. (GN56)

    (Plate 13a) Photomicrographs of rocks at the margin of the gabbro dyke, Camas Mór., Muck a. Titanaugite zoned to sodic clinopyroxene, and apatite in the modified margin of gabbro. (C53900/3, X 115, plane-polarisers) (GN57)

    (Plate 13b) Photomicrographs of rocks at the margin of the gabbro dyke, Camas Mór., b. Calcite-gehlenite(G)-wollastonite(W) rock in altered limestone of the Kilmaluag Formation. (C50093, X 115, cross-polarisers) (GN58) (From the Harker Collection, Cambridge University. Photomicrographs by G Chinner)

    (Plate 14a) General features of the layering in the bytownite-troctolite of the Layered Suite, Rum. Small-scale rhythmic layering in Unit 13, Eastern Layered Intrusion, on the south-east side of Askival. (GN59)

    (Plate 14b) General features of the layering in the bytownite-troctolite of the Layered Suite, Rum. Graded layering, cut by basalt dykes. Slumping and anorthositic clasts occur below hammer. Long Loch Member, Central Intrusion, Whaleback ridge west of Long Loch. (GN60)

    (Plate 14c) General features of the layering in the bytownite-troctolite of the Layered Suite, Rum. Slumping with deformation of load-casts in of Unit 14, Eastern Layered Intrusion, east side of Askival. (GN127)

    (Plate 15) Upward-growing olivine crystals in harrisitic layers in the Transitional Member, Western Layered Intrusion. Roadside about 650 m NNW of Harris Lodge, Rum. (GN61)

    (Plate 16) Replacement structures ('fingers') in feldspathic peridotite. Central Intrusion, 450 m south-east of Minishal, Rum. (GN62)

    (Plate 17a) Ultrabasic breccias in the Central Intrusion between the Long Loch and Loch an Dornabac, Rum. (Photographs: W J Wadsworth) Scattered small bytownite-troctolite fragments in feldspathic peridotite. (GN63) Scale: hammer handle c.30 cm.

    (Plate 17b) Ultrabasic breccias in the Central Intrusion between the Long Loch and Loch an Dornabac, Rum. (Photographs: W J Wadsworth) b. Close-packed blocks of layered bytownite-troctolite in feldspathic peridotite. (GN64) Scale: lens cap c.5 cm.

    (Plate 17c) Ultrabasic breccias in the Central Intrusion between the Long Loch and Loch an Dornabac, Rum. (Photographs: W J Wadsworth) c.. Dunite and bytownite-troctolite blocks in a typical feldspathic peridotite matrix. (GN65) Scale: hammer handle c.30 cm.

    (Plate 18) Large block of layered bytownite-troctolite and peridotite (note finger structures at contact) with slumped and brecciated bytownite-troctolite draped over the top. The sense of movement is from top left to bottom right. In the Central Intrusion, at [NM 3635 9792], about 120 m SSE of the south end of the Long Loch, Rum. (GN66) Scale: hammer handle c.50 cm.

    (Plate 19a) Photomicrographs of rocks from the Eastern Layered Intrusion, Rum. Feldspathic peridotite with close-packed olivine enclosed by clinopyroxene and plagioclase. Olivine adcumulate. Unit 8, north-east Hallival. (DU5698, X 10, crossed polarisers) (GN67)

    (Plate 19b) Photomicrographs of rocks from the Eastern Layered Intrusion, Rum. Bytownite-troctolite: plagioclase-olivine-clinopyroxene cumulate, with weak crystal lamination. Unit 8, north-east Hallival. (DU5697, X 40, crossed polarisers) (GN68)

    (Plate 19c) Photomicrographs of rocks from the Eastern Layered Intrusion, Rum. Chromite-rich seam at the contact between anorthositic bytownite­troctolite (an extreme plagioclase adcumulate) of Unit 11 (below) and feldspathic peridotite of Unit 12 (above). About 800 m north-west of Hallival summit. (DU5696A, X 20, crossed polarisers) (GN69)

    (Plate 19d) Photomicrographs of rocks from the Eastern Layered Intrusion, Rum. Skeletal olivine phenocryst in fine-grained, variolitic matrix. Chilled contact of the Eastern Layered Intrusion on the north-east side of Beinn nan Stac. (SR209B, X 40, plane-polarised light) (GN70)

    (Plate 20) Intrusion breccia at the contact of the Layered Suite, Rum. (GN71)

    (Plate 21a) Photomicrographs of thermally altered rocks at the contacts of the Rum Central Complex and the peridotite and gabbro plugs. Quartz paramorphs after tridymite fringing relict quartz grains. The matrix consists of spherulitic aggregates of quartz and alkali-feldspar. Torridon Group arkose. At the margin of gabbro plug. Allt Bealach Mhic Néill, [NM 3801 9966]. (SR317, X100, crossed polarisers) (GN72)

    (Plate 21b) Photomicrographs of thermally altered rocks at the contacts of the Rum Central Complex and the peridotite and gabbro plugs. Two-pyroxene hornfels, from the alteration of Lewisian hornblende gneiss. From margin of Western Layered Intrusion south-east of Ard Nev, at [NM 3502 9802]. (DU13781, X40, crossed polarisers) (GN73)

    (Plate 21c) Photomicrographs of thermally altered rocks at the contacts of the Rum Central Complex and the peridotite and gabbro plugs. Quartz-alkali feldspar intergrowths in rims separating original quartz and feldspar of felsic Lewisian gneiss. Inclusion in porphyritic rhyodacite ash flow, east side of Meall Breac [NM 387 980]. (SR363, X 20, crossed polarisers) (GN74)

    (Plate 22) Deformed and contorted banding in Lewisian gneiss in the rheomorphic zone adjoining a tongue of ultrabasic rocks, east of the Priomh­lochs, Rum. Dark amphibolitic layers at bottom left and to right of hammer have been veined and boudinaged. (GN75) Scale: hammer 50 cm.

    (Plate 23) Columnar-jointed basalt and mugearite flows in the Eigg Lava Formation, north-east Eigg. (GN76). The light-coloured band is formed by two mugearite flows, the dark flows are olivine-basalts. Small outcrops of the Jurassic Valtos Sandstone Formation occur to the right below the lavas. Landslips and rockfalls cover all the lower ground down to the shoreline.

    (Figure 24) Bedded, reddened, sanidine-bearing trachytic tuff overlying a reddened, rubbly, pahoehoe basaltic flow top, near the base of the Eigg Lava Formation south-east of Port Mór, Muck. Sanidine from this tuff has provided an age of 62.4 ± 0.6 (2o-) Ma (Pearson et al., 1996). (GN77) Scale: hammer 40 cm.

    (Figure 25) Thick inter-lava fluviatile conglomerate with lenses of coarse sandstone overlain by a basalt flow. Canna Lava Formation, Camas Thairbearnais, north shore of Canna. (GN78)

    (Figure 27) for the localities of plugs within the Northern Marginal Zone, (Figure 21) for the Southern Mountains Zone and (Figure 35) for detail within the Layered Suite. The numbers 1-13 refer to the localities where dyke orientations have been measured (see (Figure 28))." data-name="images/P936601.jpg">(Figure 26) Sgurr of Eigg from the south-east. The lower massive upturned part of the pitchstone forms the Sgurr ridge and cross-cuts flows of the Eigg Lava Formation. A felsite sheet is visible in the pitch-stone about half-way up the cliff face. The overhang at the base of the pitchstone is generally marked by pitchstone breccia and the pitchstone overlies patches of conglomerate. (D1698)

    (Figure 27) Columnar jointing and later felsitic sheets in the pitchstone on the south of the Sgurr, Eigg. (GN79) Columns are nearly vertical in the lower part of the body but vary from horizontal to inclined in the upper part. A prominent felsite sheets cuts inclined columns and wedges out eastwards. Pitchstone blocks at the extreme centre right are part of a major rockfall.

    (Plate 28a) Photomicrographs of pitchstones from the Sgurr of Eigg Pitchstone Formation, Eigg. Phenocrysts of pyroxene, opaque oxides and corroded feldspar (anorthoclase and plagioclase) in a glassy, flow-banded matrix. 0.5 m above base of the pitchstone at the Recess [NM 465 8462]. (SR489, X100, plane-polarisers (GN131))

    (Plate 28b) Photomicrographs of pitchstones from the Sgurr of Eigg Pitchstone Formation, Eigg. Phenocrysts of feldspar, pyroxene and opaque oxides, together with fragments of basalt, feldspar phenocrysts and flow-banded pitchstone in a glassy, tuffaceous matrix. Base of pitchstone, Bidein Boideach [NM 4412 8667]. (SR490c; X 100, plane-polarised (GN80))

    (Plate 29) Raised storm beaches at Harris, Rum. (D1726)

    Tables

    (Table 1) Geological sequence in Rum and the adjacent islands

    (Table 2) Revised Torridonian stratigraphy, Rum, showing the approximate equivalence of former subdivisions; thicknesses are relative only.

    (Table 3) Modal analyses of the porphyritic rhyodacites of Rum (by A C Dunham)(vol. %).

    (Table 4) Terminology for the gabbroic and ultrabasic members of the Layered Suite, Rum Central Complex.

    (Table 5a) Lithologies in the Rum Layered Suite (based on McClurg, 1982).

    (Table 5b) Terminology of cumulate rocks (after Wager et al., 1960).

    Table 6 Members of the Central Intrusion, Rum Layered Suite.

    (Table 7) Modal analyses of representative peridotites, bytownite-troctolites and bytownite-gabbros of the Rum Layered Suite

    (Table 8) Selected mineral compositions from rocks of the Rum

    (Table 9) Possible parent magmas for the Rum Layered Suite.

    (Table 10) Sequence of faulting, folding and intrusion in the Rum Central Complex

    (Table 11) K-Ar analyses and age determination of rhyodacites and granophyres from Rum.

    (Table 12) Composition of 91 marine sediment samples, southern Rum, analysed by X-ray fluorescence (after calcite dissolution.

    (Table 13) Average composition of minerals in marine sediments from Harris Bay, Rum.

    Tables

    (Table 1) Geological sequence in Rum and the adjacent islands

    Quaternary (Recent and Pleistocene) Peat Glacial meltwater deposits
    Rockfalls Coarse glacier deposits, Loch Lomond readvance
    Landslips Late-glacial till
    Alluvium Late-glacial raised beach deposits
    Eocene Sgurr of Eigg Pitchstone Formation Lavas, ash flows and fluviatile conglomerates c.100
    Palaeocene Final movement on the Long Loch Fault, Rum: sparse, NNW- to N-trending basaltic dykes
    Canna Lava Formation Canna and Sanday: olivine-basalt, basaltic hawaiite, hawaiite and mugearite lavas, interbedded conglomerates 300+
    North-west Rum: olivine-basalt, basaltic hawaiite, hawaiite, basaltic andesite, icelandite lavas, interbedded conglomerates 300+
    Unconformity Subaerial erosion of the Rum Central Complex
    Palaeocene Rum Central Complex
    Stage 2: Emplacement of the Rum Layered Suite: b. Central Intrusion
    a. Eastern Layered Intrusion; Western Layered Intrusion (probably coeval with early components of the Central Intrusion)
    Peridotite, gabbro and dolerite plugs were probably coeval with Stage 2 intrusions
    Regional NW-trending dyke swarm (probably continuous throughout much of Stage 1, with sporadic dykes to the end of Stage 2), intrusion of basaltic cone-sheets and radial basaltic dykes (probably overlaps (e) below)
    Stage 1: Ring-faulting, caldera formation, intrusion and effusion of silicic magmas (e) Second, final phase of uplift on Main Ring Fault; basal Torridonian reverse-faulted
    (d) Intrusion of the Western Granite and granitic rocks at Papadil and near the Long Loch. (Coeval with (e) above)
    (c) Subsidence on the Main Ring Fault; silicic volcanism; Am Màm Breccias; Coire Dubh Breccias
    (b) Early gabbros and peridotites (occur only as blocks in Am Màm breccias in (c) above)
    (a) First phase of uplift on the Main Ring Fault, accompanied by doming of the country rocks. (Coeval with (b) above, or possibly later)
    Welshman's Rock and Mullach Ard faults (probably coeval with (a) above)
    Eigg Lava Formation Eigg: Eigg Lava Formation Eigg:Olivine-basalt, basaltic hawaiite, hawaiite and mugearite lavas, thin mudstones and fine-grained ashes 400+
    Muck: Olivine-basalt, basaltic hawaiite, hawaiite and mugearite lavas, thin mudstones and ashes, coarse breccia at base of succession at Camas Mór 150+ thickness not known
    Rum: Amygdaloidal olivine-basalts, faulted within the Main Ring Fault system in SE Rum
    Early (Pre-Rum Central Complex) movement on the Long Loch Fault
    Unconformity
    Cretaceous
    Upper Cretaceous Strathaird Limestone Formation (includes Laig Gorge Sandstone Member and Clach Alasdair Conglomerate Member) Limestone, coarse calcareous sandstone 5--6
    Unconformity
    Jurassic
    Upper Jurassic
    Oxfordian Staffin Shale Formation Dark siltstone, thin limestones c. 27
    Unconformity
    Middle Jurassic Great Estuarine Group:
    Bathonian Kilmaluag Formation Dark shales with thin limestone common 5+
    Duntulm Formation Limestones and shales 8+
    Valtos Sandstone Formation Concretionary sandstone, limestone c. 60
    Lealt Shale Formation Fissile grey shales, algal limestone band, shale, sandstone, limestone 45+
    Bajocian (in part) Bearerraig Sandstone Formation Calcareous sandstone 5+
    Unconformity
    Lower Jurassic
    Sinemurian/Hetangian Broadford Beds Sandstone, limestone, shale c. 35
    Unconformity
    Triassic Monadh Dubh Sandstone Formation Sandstone, conglomerate, siltstone, cornstone c. 80
    Unconformity
    Proterozoic Torridon Group:
    Aultbea Formation Fine sandstone, siltstone 150+
    Applecross Formation Sandstone, pebbly sandstone, siltstone c. 2000
    Diabaig Formation Sandstone, siltstone, shale, sedimentary breccia 500+
    Unconformity
    Archaean Lewisian Complex Feldspathic gneiss, amphibolite

    (Table 3) Modal analyses of the porphyritic rhyodacites of Rum (by A C Dunham)(vol. %).

    Area/Number of samples Feldspar Quartz Pyroxene Opaques Groundmass
    Sgurr nan Gillean (6) 17.4 3.4 6.2 2.2 70.5
    7.4–21.3 2.7–4.9 3.3–8.0 1.8–3.2 67.5–82.6
    Beinn nan Stac (1) 18.6 1.7 5 1.6 73.1
    Cnapan Breaca (5) 17.1 3.5 2.3 1.1 76.1
    15.4–18.9 1.7–4.3 0.7–4.1 0.6–1.6 73.1–80.4
    Coire Dubh (4) 17.3 2.1 2.7 1 76.9
    16.3–18.4 1.0–3.6 1.5–4.1 0.3–1.8 75.6–78.4
    Meall Breac (8) 21.3 2.7 3.4 1.5 71.1
    14.7–30.1 2.1–3.6 1.5–6.0 1.1–1.9 61.4–77.9
    Am Màm (2) 27.9 2.9 3.6 1.1 64.5
    Long Loch (2) 21.6 3.4 3.6 1.9 70.4
    Grand average 19.5 2.9 3.8 1.4 72.2

    (Table 4) Terminology for the gabbroic and ultrabasic members of the Layered Suite, Rum Central Complex.

    This memoir Former terms
    Central Intrusion Central Series1,2
    Ruinsival Member, Units 1 and 2 Ruinsival Group1
    Ruinsival Member, Units 1 and 2 Ruinsival Member2
    Ruinsival Member, Units 1 and 2 Upper Ruinsival Series3
    Long Loch Member, Units 1 to 3 Long Loch Group1
    Long Loch Member, Units 1 to 3 Long Loch Member2
    Long Loch Member, Units 1 to 3 Lower Ruinsival Series3
    Dornabac Member Dornabac Member1,2
    Dornabac Member Dornabac Series3
    Eastern Layered Intrusion Units 1–16 Eastern Layered Series Units 1–161-4
    Western Layered Intrusion Western Layered Series
    Ard Mheall Member Ard Mheall Series3
    Transitional Member Transitional Series3
    Harris Bay Member Harris Bay Series3

    (Table 5a) Lithologies in the Rum Layered Suite (based on McClurg, 1982).

    Rum name used here Standard name Mineral characteristics (approximate modes in vol%
    Anorthosite Anorthosite >95% total plagioclase
    Bytownite-troctolite (formerly allivalite) Bytownite-troctolite, gabbro Plagioclase-olivine cumulates with total plagioclase 95%-50%, cumulus clinopyroxene <5% to c.35% (rare)
    Feldspathic peridotite Troctolite and gabbro Melatroctolite Olivine cumulates with 50%-30% post-cumulus plagioclase
    Peridotite Melatroctolite Feldspathic peridotite Olivine cumulates with 30%-5% post-cumulus plagioclase
    Dunite Dunite >95% cumulus olivine, <5% combined post-cumulus clinopyroxene and plagioclase

    (Table 5b) Terminology of cumulate rocks (after Wager et al., 1960).

    Orthocumulate: one or more cumulus minerals plus the crystallisation products of the intercumulus liquid (= pore material)
    Mesocumulate: one or more cumulus minerals plus a small amount of pore material
    Adcumulate: one or more cumulus minerals with less than 5% pore material.

    (Table 6) Members of the Central Intrusion, Rum Layered Suite.

    (Top)
    Ruinsival Member: Two cyclic units, both peridotite-rich, the lower one capped by c.10 m bytownite-troctolite. About 330 m thick. Restricted to Ruinsival summit area.
    Long Loch Member: Three cyclic units, mainly peridotite with thin (<10 m) bytownite-troctolite at top of each. c.600 m thick NW of Ruinsival and 450 m in lower Glen Harris. Forms core of syncline in Central Intrusion west and NW of the Long Loch.
    Dornabac Member: 150 m+ thick but base not exposed. Consists of thick bytownite-troctolite, well exposed at An Dornabac but thinning northwards. Underlain by peridotite.
    (Base)

    (Table 7) Modal analyses of representative peridotites, bytownite-troctolites and bytownite-gabbros of the Rum Layered Suite

    Unit Rock type Olivine Plagioclase Clinopyroxene Spinel Others Reference

    Eastern Layered Intrusion

    14

    		 ET 32.6 58.6 4.9 1.0 (8)
    UPd 66.5 23.7 8.4 1.5
    UPd 87.6 2.3 6.2 3.9

    10

    		 ET 15.5 82.5 - 2.0 (3)
    ET 7.5 61.0 31.5
    UPd 77.5 13.5 8.5 0.5

    8

    		 ET 32.8 55.6 10.7
    UPd 69.8 24.7 2.3 2.9 (3)

    7

    		 ET 14.1 80.9 5.0
    ET 22.0 66.2 11.6 0.1 (3)
    UPd 70.4 23.0 5.3 1.3

    Western Layered Intrusion

    AM

    		 UPd 69.6 26.7 0.9 2.9 (7)
    UPd 91.6 4.5 3.3 0.5
    mEH 88.6 6.6 1.3 0.6 3.0
    cEH 75.2 11.2 1.9 1.4 10.4

    HBM

    		 byEG 45.4 36.8 15.4 1.8 0.6 (bi) (7)
    EH 39.0 34.8 24.7 1.5 (mt)

    Central Intrusion

    RM

    		 ET 24.2 73.8 1.7 0.2 (7)
    UPd 70.4 19.2 8.3 2.0
    UPd 95.8 1.9 1.3 1.0

    DM

    		 ET 12.0 54.5 33.4 0.1 (7)
    UPd 71.0 26.9 0.4 1.6

    (Table 8) Selected mineral compositions from rocks of the Rum

    Unit Rock type Olivine (Fo) Clinopyroxene (Mg*) Plagioclase (An) Reference
    EASTERN LAYERED INTRUSION
    10a range in units 88.1–79.0 90–84 87–77 (1)
    10b bytownite-troctolite 84–76 87–81 85–62 (2)
    feldspathic peridotite 86–84 86–84 88–65
    peridotite 86–84 87–83 88–55
    10c bytownite-troctolite 86 80 85 (3)
    7 bytownite-troctolite 89–64 (4)
    5 bytownite-troctolite 76–70 82–74 (8)
    4 bytownite-troctolite 76–70 78–74
    2 feldspathic peridotite 86–82 86–85
    bytownite-troctolite 78–71 87–84
    Member Rock type Olivine (Fo Clinopyroxene (Mg*) Plagioclase (An) Reference
    WESTERN LAYERED INTRUSION
    peridotite 88.5–79.3 (1)
    Transitional/Ard Mheall harrisite 83–82 85–84 (1)
    Harris Bay bytownite- gabbro 79–78 (1)
    Ard Mheall breccias 82.2 84.7 81.1–61 (5)
    Member Rock type Olivine (Fo Clinopyroxene (Mg*) Plagioclase (An) Reference
    CENTRAL INTRUSION
    Dornabach range in unit 84.9–78.6 84 85.9–82 (1)
    Long Loch peridotite 82.0 83.6 67.7 (6)
    segregations 73.7 80.2 37.9
    Long Loch breccia 83 86 77–55.2 (5)

    (Table 9) Possible parent magmas for the Rum Layered Suite.

    (a) (b) (c) (d)
    SiO2 46.8 46.09 46.05 44.82
    Al2O3 18.8 14.29 13.39 11.16
    Fe2O3 1.4 11.32 11.03 11.59
    FeO 7.7 nd nd nd
    MgO 11.0 13.47 15.38 20.50
    CaO 10.6 10.85 9.11 9.41
    Na2O 2.7 1.96 1.64 1.31
    K2O 0.3 0.10 0.44 0.11
    TiO2 0.8 0.99 0.76 1.07
    P2O5 0.04 0.08 0.11 0.10
    MnO 0.1 0.17 0.18 0.16
    Cr2O3 0.04 nd nd nd

    (Table 10) Sequence of faulting, folding and intrusion in the Rum Central Complex.

    (Table 11) K-Ar analyses and age determination of rhyodacites and granophyres from Rum.

    Sample Vol. radiogenic 40Ar nl.g-1 %K % atmosp. Ar
    Sample Vol. radiogenic 40Ar nl.g-1 %K % atmosp. Ar Apparent age (Ma)
    Rhyodacites
    735 [NM 392975] 6.117 ± 0.192 2.53 ± 0.08 12.35 59.62 ± 2.76
    736 [NM 386 980] 6.613 ± 0.257 2.72 ± 0.06 19.22 59.95 ± 2.64
    737 [NM 387 984] 6.226 ± 0.208 2.64 ± 0.06 41.98 58.17 ± 2.24
    738 [NM 373 987] 5.468 ± 0.200 2.27 ± 0.06 23.78 59.40 ± 2.44
    740 Ainshval 6.326 ± 0.214 2.68 ± 0.06 23.34 59.32 ± 2.34
    421 [NM 392 976] 2.77 15 57.6 ± 1
    429 [NM 385 984] 1.93 22 73.8
    Granophyres
    710 [NM 333 967] 7.254 ± 0.218 3.24 ± 0.09 27.25 55.27 ± 2.27
    R1 5.065 2.235 35.80 57.3 ± 2.0
    R6 ' 7.677 3.337 7.95 58.2 ± 2.0

    (Table 12) Composition of 91 marine sediment samples, southern Rum, analysed by X-ray fluorescence (after calcite dissolution.

    Mean(%) s (%) Max. (%) Min. (%)
    Mg 3.5 2.6 14 0
    Ca 1.7 0.82 5.3 0.47
    Ti 0.33 0.16 0.92 0.12
    V 0.009 0.005 0.29 0.001
    Cr2O3 0.12 0.17 1.4 0
    Fe 4.0 1.8 10.6 0.72

    (Table 13) Average composition of minerals in marine sediments from Harris Bay, Rum.

    Chromite Olivine Clinopyroxene (Ca-rich) Orthopyroxene
    N = 82 141 40 6
    SiO2 0.03 40.20 51.22 54.67
    TiO2 2.69 0.03 0.94 0.38
    A12O3 19.07 0.03 3.15 0.87
    Cr2O3 32.49 0.58 0.05
    Fe2O3 13.40 — — —
    FeO 20.60 12.33 6.07 14.25
    MnO 0.32 0.20 0.15 0.34
    MgO 10.52 47.35 16.02 28.02
    CaO 0.01 0.08 21.11 1.42
    Total 99.13 100.22 99.24 100.00
    Mg* 47.44 87.23 82.46 77.80
    Cr* 54.29 — — —