Glencoe caldera volcano, Scotland. Classical areas of British geology

By B P Kokelaar I D Moore Contributors: T Bradwell, D Stephenson

Bibliographical reference: Kokelaar, B P, And Moore, I D. 2006. Classical areas of British geology: Glencoe caldera volcano, Scotland. (Keyworth, Nottingham: British Geological Survey.)

British Geological Survey

Authors B P Kokelaar Department of Earth and Ocean Sciences, University of Liverpool, Liverpool L69 3BX, UK

I D Moore Formerly University of Liverpool; present address Chevron-Texaco, Seafield House, Hill of Rubislaw, Aberdeen AB15 6XL, UK

Contributors T Bradwell D Stephenson British Geological Survey, Murchison House, Edinburgh

The grid, where it is used on the figures, is the National Grid taken from Ordnance Survey mapping. Maps and diagrams in this book use topography based on Ordnance Survey maps. © Crown Copyright reserved Ordnance Survey licence number 100017897/2006.

(Front cover) Glen Coe viewed towards the south-west from The Study (Photographer: B P Kokelaar) (P611763).

(Rear cover)

Acknowledgements

The geological reappraisal of Glencoe volcano was substantially funded by an Isle of Man Department of Education Postgraduate Studentship (awarded to I D M) and a NERC/BGS contract (GA/96E/13), both held at the Earth Sciences Department of the University of Liverpool. D Stephenson (BGS) wrote the section on the Dalradian metamorphic ‘basement’, and T Bradwell (BGS) Chapter nine — Shaping the landscape. M F Howells and S C Loughlin provided helpful reviews.

Scientific editing on behalf of BGS was by D Stephenson, with review and compilation by E A Pickett; the series editor is A A Jackson. Final versions of figures were produced by BGS Cartography, Edinburgh; pagesetting by A Hill, BGS Keyworth.

Photographs, with the exception of (Plate 1) and (Plate 27), were taken by the authors, and have been deposited for reference in the BGS National Archive of Geological Photographs.

The 3D image used in (Figure 4)a was generated by Harlan P Foote, Image Processing and Analysis Group, Battelle-Pacific Northwest Laboratories, Richland, Washington (state); the geological overprint is from USGS Map I-571, 1970.

(Frontispiece) Viewed due east from the summit of Bidean nam Bian (1150 m), the successive ridges of Beinn Fhada, Buachaille Etive Beag and Buachaille Etive Mòr provide serial sections through the volcanic succession of the Glencoe caldera volcano. The prominent summit in the distance (middle) is of Stob Dearg (1022 m), which presents remarkable exposures of three successive volcanic cones formed during powerful explosive interactions of magma with water. Beyond Stob Dearg lies the desolate Rannoch Moor, mostly underlain by a granitic pluton that was unroofed just before the volcanism at Glen Coe (Photographer: B P Kokelaar) (P611764).

Foreword

The inspiring mountain scenery of Glen Coe, with its famous history of clan rivalry and murder, creates an almost palpable atmosphere of roughness tinged with romance. For many scientists, the awe is enhanced by the importance of the place in the history of geology and, specifically, of volcanology. It was here, close to 100 years ago, that Geological Survey geologists E B Bailey, C T Clough, G W Grabham, H Kynaston and H B Maufe exploited the superb three-dimensional exposure in the rugged terrain to produce the first detailed analysis of how a large, long-lived volcano is successively plumbed by magma. Their interpretation, of ‘cauldron subsidence’, still influences volcanologists who have to infer what they cannot see beneath the surface at modern volcanoes.

In the 1950s and 60s, the recognition of ignimbrites (ash-flow tuffs) and their association with caldera-forming eruptions caused a renewal of interest in the Siluro-Devonian, Glencoe volcano. However, no refinement of existing maps was accomplished, possibly because the prospect of the steep mountain ridges and the midge-infested hollows proved too daunting. By the 1980s and early 90s, when it was clear that understanding of the behaviour of modern, commonly hazardous, caldera volcanoes was hindered by uncertainties at depth, reappraisal of the outstanding natural volcano laboratory at Glen Coe was long overdue.

This book is the result of a complete resurvey of the volcanic rocks and their relationships by Dr Ian Moore, as a PhD student, and Dr Peter Kokelaar, at the University of Liverpool. It is accompanied by a map, at 1:25 000 scale, with several cross sections and a new refined stratigraphy. The early ideas have been modified substantially and the volcano now perhaps constitutes a new archetype, as ‘the world’s best exposed, tectonically controlled, multi-subsidence, piecemeal caldera volcano’! Parts of this complicated interpretation have been published in the relevant journals, but this book presents the complete reappraisal. It has been written with an emphasis on volcanic processes and environments, and on the regional significance of this ancient volcano, to be of interest not only to volcano aficionados but also to the keen amateur and student.

The book and map represent the first major revision of this classical geology since the early pioneering work of those Survey geologists. It is a testament to the techniques of field-based geology, although, inevitably on such a dramatic stage, involving a significant element of mountain guile.

David Falvey, PhD. Director. British Geological Survey. Kingsley Dunham Centre, Keyworth, Nottingham

Chapter 1 Introduction

The deep valley of Glen Coe and the surrounding mountains together form one of the most spectacular natural landscapes in Britain. Glen Coe is in the south-western Grampian Highlands of Scotland (Plate 1) and descends westwards from the Rannoch Moor plateau at about 300 m above sea level to the sea at Loch Leven (Figure 1). The River Coe flows in the glen, over several waterfalls and through Loch Achtriochtan, entering Loch Leven at Glencoe village. The Three Sisters (Aonach Dubh, Geàrr Aonach and Beinn Fhada), great rocky buttresses separated by steep-sided hanging valleys, dominate the south side of the glen, extending north-eastwards from Bidean nam Bian (1150 m), which is the highest mountain in the Glencoe area. The northern wall of the glen rises steeply to the sharp, serrated edges of the Aonach Eagach and Am Bodach. At the eastern end of Glen Coe, the imposing bulk of Stob Dearg stands guard at the junction of Glen Coe and Glen Etive, while to the east of Glen Etive lie the remote peaks of Meall a’ Bhùiridh, Creise and Clach Leathad. Farther to the east is Rannoch Moor, a wild expanse of peat bogs and lochans.

The Glencoe area is famous for its long and sometimes turbulent history, particularly the treacherous massacre there in 1692. Glen Coe was the home of the legendary Fingal, or Fionn MacCumhail, one of the great heroes of Celtic mythology; a cave high above Loch Achtriochtan is named after Fingal’s son, the poet Ossian. Today, the glen is popular with visitors from all over the world who come to enjoy the dramatic mountain scenery. Many come to walk and climb; several Munros (Scottish mountains with summits over 3000 feet above sea level) rise above the glen and the numerous buttresses and rock faces afford spectacular routes for climbers. In winter the slopes of Meall a’ Bhùiridh above the White Corries centre are used for skiing. A large part of the Glencoe area, including the Aonach Eagach, Bidean nam Bian, Buachaille Etive Beag, Stob na Bròige and Stob Dearg, is in the care of The National Trust for Scotland.

In this book, information is provided to enable the interested geologist to locate and examine key features in the field; a map and table of key field sites is provided (Appendix). Throughout the text, localities are given in square brackets as references to the UK Ordnance Survey National Grid; all refer to the 100 km square NN, which is omitted for brevity.

For those intending to visit the Glencoe area, it is stressed that:

much of the terrain is very rugged and remote
many localities and mapped lines are accessible only by rock climbing
the weather can become extremely inclement with surprising rapidity
some descent routes are not straightforward and can be arduous.

Most of the described locations are relatively accessible to the fit and experienced field geologist, weather permitting, and it is intended that they should be visited (without the use of a geological hammer). However, no site is perfectly safe and field safety is not indicated by the following descriptions that imply accessibility.

Outline of the geology

The Glencoe area exposes the roots of an ancient volcano that has been deeply dissected by the action of natural forces over millions of years, culminating in successive glaciations that ended only 11 000 years ago. It has been recognised as a classical area of British geology for almost a century. Meticulous fieldwork here by the Geological Survey in the early 1900s led to the first detailed analysis of the relationships between a volcano and its underlying intrusions, and hence between surface and subsurface magmatic processes. This early analysis was all the more remarkable for having been accomplished in terrain that is amongst the most rugged and physically demanding for fieldwork in Britain. The early interpretations of the Glencoe volcano profoundly influenced subsequent studies throughout the world, particularly of modern volcanoes where processes at depth could only be inferred indirectly. Now, the Glencoe Caldera-volcano Complex, as it is more formally named here, is renowned internationally both for its early influence in volcanology and for important new insights derived from a modern reappraisal. Detailed remapping of the entire volcanic succession at Glen Coe, utilising far richer volcanological understanding than was available nearly a century ago, has led to considerable modification of the original interpretations. This book, together with a new 1:25 000 scale geological map (British Geological Survey, 2005), summarises the new discoveries; important technical terms, volcanic features and processes are explained in the next section.

The geology of the Glencoe area is dominated by an intra-caldera succession of silicic pyroclastic rocks, intermediate and silicic lavas and a wide variety of intrusions; unconformities with associated alluvial deposits are common in the sequence (Figure 2). This succession overlies a deeply eroded ‘basement’ of Neoproterozoic to early Palaeozoic metasedimentary rocks, as well as pre-caldera andesite sills with intervening sedimentary strata. Seven thick ignimbrites constitute most of the intra-caldera succession. These represent major caldera-forming explosive eruptions that were associated with volcanotectonic subsidence and piecemeal caldera collapse. The subsidence was incremental and involved the movement of numerous cross-cutting faults, many of which also formed the plumbing system that tapped the underlying magma chambers. Tectonic faulting continued during the periods between eruptions, causing changes in drainage and sedimentation that are recorded in the intercalated sedimentary rocks. A ring-fault system and associated intrusions formed after the early incremental caldera development, and these were succeeded by the emplacement of voluminous silicic magmas to form the Clach Leathad Pluton (new name; see p.98). This large intrusion formed during foundering of a considerable chunk of the caldera succession, which is thus largely obliterated in the south-east; it merges locally with the ring-fault intrusions and is truncated by the large Etive Pluton to the south (Figure 2); (Figure 3).

The Glencoe Caldera-volcano Complex forms part of an extensive magmatic province that is marked by numerous volcano remnants and plutons e.g. (Figure 3) and extends from Shetland in the north-east to Donegal (Ireland) in the south-west. The magmatic activity occurred during strike-slip faulting, uplift and erosion following the continental plate collision known as the Scandian Event of the Caledonian Orogeny, which is thought to have occurred 435 to 425 million years ago (Soper et al., 1992; Dewey and Strachan, 2003). At the time of the magmatism, the newly amalgamated continent was near southern tropical latitudes (Cocks and Torsvik, 2002) and climatic conditions in the region were semi-arid; rapidly eroding metamorphic and igneous massifs were flanked by alluvial plains characterised by flash floods and ephemeral lakes.

Isotopic dating suggests that the Glencoe volcano was active some 421 ± 4 million years ago (Thirlwall, 1988), in late Silurian times according to current time scales (e.g. Gradstein and Ogg, 1996; Tucker et al., 1998). However, plant remains and sporomorphs in pre-caldera strata indicate an earliest Devonian age (Kidston and Lang, 1924; Wellman, 1994) of around 415 million years (Lochkovian Stage). The apparent age contradiction, at least in part, reflects uncertainties attributable to the dating methods employed; it appears that the caldera-related volcanism may have lasted for no more than two or three million years (see p.104). Intrusions that cut the volcano complex, and thus postdate its activity, have yielded ages that suggest emplacement some 412 to 401 million years ago (Clayburn et al., 1983; Thirlwall, 1988), in early Devonian times (Pragian–Emsian ages).

What a caldera is, and why Glencoe caldera volcano is important

Catastrophic eruption of tens to thousands of cubic kilometres of magma from a chamber beneath a volcano typically leads to collapse of the chamber roof and consequent subsidence of the overlying volcanic pile. The subsidence forms a surface depression called a caldera (Figure 4). Calderas can be many tens of kilometres across and hundreds of metres deep: many become partially filled by lakes or form embayments of the sea. Well-known modern caldera volcanoes include Mount Mazama (Crater Lake) in the Cascades Range of the USA, Santorini in the Greek islands, and Toba in Sumatra (see Francis and Oppenheimer, 2004). Calderas form at long-lived volcanoes and they should not be confused with volcanic craters, which are relatively small depressions that form around vents by erosion during an explosive eruption or by build-up of ejecta.

Most caldera-forming eruptions are explosive (Figure 4)b; they involve sustained jets and gas-rich flows of vitric ash particles, pumice fragments and crystals, collectively called pyroclasts, usually with some included rock fragments called lithic clasts. Much of the erupted mixture is denser than atmosphere so that it falls to the ground, as in a fountain, and moves away from the vent as a pyroclastic density current (also known as a pyroclastic flow; see Branney and Kokelaar, 2002). The magma that erupts to form most medium-sized or large calderas is normally of intermediate or silicic composition, commonly dacite, rhyodacite or rhyolite, and its initial temperature is usually in the range 700° to 900°C. Since the eruptive vents tend to lie within or around the periphery of calderas, pyroclastic material usually partially infills the topographical depression (Figure 4)b. Parts of both the eruptive fountain and the related pyroclastic density current (or currents) mix with and heat surrounding air and thus become buoyant, forming vigorously ascending plumes of pumice fragments and ash. Such plumes can be vast and can spread in the stratosphere so as to influence both regional and global climate, such as occurred with the caldera-forming eruptions of Mount Pinatubo (1991), Krakatau (1883), Tambora (1815) and Toba (about 74 000 years ago).

The deposits of the ground-hugging pyroclastic density currents that occur in caldera-forming eruptions are known as ignimbrites (synonymous with ash-flow tuffs). These normally are poorly sorted mixtures of ash, crystals, pumice fragments and lithic clasts, and tend to be thick within topographical depressions. Caldera-related ignimbrites can have volumes of up to thousands of cubic kilometres and can extend over tens of thousands of square kilometres. It is common for the fragmented magmatic material in ignimbrites initially to retain temperatures above 550° to 600°C, in which case the glassy constituents (vitric ash and pumice fragments) deform in a ductile manner and become fused together. Such ignimbrites are described as welded, and they typically show distinctive flattened pumice fragments, known as fiamme, and streakiness of the finer matrix to form eutaxitic texture (Plate 2a)(Plate 2b). Fallout from the buoyant plumes typically produces extensive blanket-like deposits of pumice or ash. Seven ignimbrites are now recognised in the Glencoe Caldera-volcano Complex; each is preserved within the caldera that formed during its eruption and accumulation, so that multiple subsidence is recorded. The total volume of material erupted in each case is unknown, because of incomplete preservation within the caldera and because an unknown proportion is likely to have escaped beyond the confines of the caldera during the eruption. Best estimates of the minimum volumes of magma erupted during formation of the most substantial preserved ignimbrites are in the order of ten cubic kilometres, which is a small volume by caldera volcano standards.

Because calderas are topographical depressions that are liable to become at least partially flooded with water, eruptions that occur within them commonly involve the interaction of magma and water. Such hydrovolcanic activity can be highly explosive, particularly where the interaction occurs beneath the ground surface, as in an aquifer, in which case it is referred to as phreatomagmatic explosivity. Phreatomagmatic explosions typically build low-profile tuff cones composed of erupted ash and fragments of the aquifer. Tuff cones normally include distinctive layers in which parallel- and cross-stratification record deposition from powerful ground-hugging ash clouds (another type of pyroclastic density current, commonly referred to as ‘ash hurricane’ or ‘pyroclastic surge’), while bomb-sags register the impact of blocks hurled ballistically from the vent. Phreatomagmatic explosivity is now known to have punctuated the evolution of the Glencoe caldera volcano at least five times and, although not recognised until recently; extensive exposures of one pyroclastic cone with its underlying aquifer constitute a rare and exceptionally instructive section of a silicic hydrovolcano.

Calderas normally form within earlier-erupted, genetically related volcanic rocks. In many cases they are defined by steep and nearly circular, scalloped topographical margins with subannular arrangements of post-caldera intrusions, vents and lavas e.g. (Figure 4)a. Geologists have interpreted this ring-form as reflecting subsidence of a coherent, piston-like block of volcanic and crustal rocks into the magma chamber along a relatively simple bounding ring-fault (e.g. Smith and Bailey, 1968). The piston-subsidence model, however, did not originate from a modern volcano, but primarily from the early Geological Survey work on the ancient Glencoe volcano and associated intrusions (Clough et al., 1909), which was soon supplemented with perspectives from the nearby Ben Nevis volcano (Maufe, 1910). The subsided block and the associated intrusions that represent the contemporary magmatic plumbing system at these sites were originally referred to as ‘cauldrons’, emphasising the perceived fundamental role of ring-faults (Figure 5)a. It was as the archetypal piston-subsidence cauldron, with its excellent exposure and striking mountain setting, that Glen Coe and its environs became recognised as a classical area of British geology.

Although control of caldera subsidence by relatively simple ring-faults has often been inferred at volcanoes where there is little or no erosional dissection (e.g. the Valles caldera; (Figure 4)a, more complex and diverse structural processes are now known to be involved in many cases. These processes include downsag, which is flexural subsidence without development of a main controlling fault (Walker, 1984), although extensional structures around the edges of the depression are common. The latter include fractures that may be hundreds of metres deep and are referred to as crevasses, and in many cases also narrow graben or half-graben between tilted strata (Figure 4)b; Branney and Kokelaar, 1994; Branney, 1995; Moore and Kokelaar, 1998). Also, subsidence has been shown to develop incrementally, involving a succession of events and haphazard piecemeal movements of numerous caldera-floor fault-blocks (Figure 4)b; Branney and Kokelaar, 1994; Moore and Kokelaar, 1998). Controls on the structural evolution of multiple-subsidence calderas are complicated and difficult to determine precisely, because early-formed structures tend to be reactivated or cross-cut during later movements, or they are obliterated by later intrusions. Studies at the surface or at shallow levels of caldera volcanoes can overlook complexity that occurs at deeper levels, in the floors and early infilling deposits, mainly because early-formed features tend to be buried by later deposits. Thus, studies of ancient volcanoes that have been deeply dissected by erosion to expose their internal structure and stratigraphy are essential to complement any understanding derived from analysis of younger counterparts. The extremely rugged topography of the Glencoe Caldera-volcano Complex has provided excellent opportunities to examine the internal volcanic structure at various levels through to the underlying basement (see 1:25 000 scale geological map; British Geological Survey, 2005), with numerous valleys and intervening ridges virtually providing serial sections up to a kilometre high, and with extensive glacially scoured outcrops. It is because of this exceptional exposure that the Glencoe volcano is a world-class natural volcanological laboratory.

The studies at Glencoe reported by Moore (1996) and by Moore and Kokelaar (1997, 1998) were undertaken with the knowledge that caldera volcanoes commonly form on major faults or fault zones (e.g. Smith and Luedke, 1984; Self et al., 1986; Kokelaar, 1988, 1992; Kokelaar et al., 1994; Milner et al., 2002; Manville and Wilson, 2003), and that crustal structure and active tectonism influence caldera form and evolution. When the Glencoe volcano developed, the regional tectonic regime was dominated by strike-slip and normal-slip movements on major crustal discontinuities such as the nearby Ericht–Laidon Fault, Great Glen Fault and Highland Boundary Fault (Figure 3); (Plate 1); Watson, 1984; Hutton, 1987; Treagus, 1991). The new studies found that the evolution of the volcano was far more complicated than previously thought, and that it was strongly influenced by the location and activity of underlying faults. Now the Glencoe volcano is considered as a different archetype; having once been the original piston-subsidence cauldron, it is now perhaps the prime example of a tectonically controlled, piecemeal, multiple-subsidence caldera volcano. The term ‘cauldron’ has been dropped, because of the recognition that much of the caldera subsidence was unrelated to any ring-fault. The new interpretations in this book largely derive from studies of the volcanic succession and structure within the ring-fault, but the intruded ring-fault system is also reappraised and interpreted within the new perspective. While there is now a far fuller understanding of the history of the volcano, much remains unknown; there is both need and scope for further research, particularly concerning the origins of the magmas and the relationships of the volcano to the nearby, broadly coeval plutons.

Chapter 2 Previous studies

Cauldron subsidence

Published in 1909, the paper entitled The cauldron-subsidence of Glen Coe, and the associated igneous phenomena, by C T Clough, H B Maufe and E B Bailey, provided the first detailed description of a deeply dissected caldera volcano; it included four geological cross-sections. The striking occurrence of an Old Red Sandstone (Siluro-Devonian) volcanic and sedimentary succession, over 1 km thick, juxtaposed with Neoproterozoic to early Proterozoic (Dalradian Supergroup) metasedimentary rocks along a steep ‘boundary-fault’ was interpreted as recording syn- to post-eruptive piston-like subsidence of a virtually cylindrical block of crust (Figure 5)a. The authors presented considerable evidence that subsidence was accompanied by the complementary rise of magma, which formed semi-continuous ‘fault-intrusions’ around the downthrown block. It was demonstrated that the subsidence occurred in at least two stages. Along northern outcrops, where there are two subparallel strands of the boundary-fault, subsidence was found to have taken place first on the outer ‘Early Fault’ and then on the inner ‘Main Fault’. In the east, the earlier fault was found to be the inner of the two strands there. The authors described in detail their findings of ‘flinty crush-rock’, which they took to represent frictionally melted rock (pseudotachylite; Shand, 1916) along the boundary-faults. At Stob Mhic Mhartuin [NN 208 575], excellent and relatively accessible exposures of the Early Fault and Main Fault, flinty crush-rock, and fault-intrusions, led to this relatively small and accessible peak (707 m) becoming the type locality of the boundary-fault and thereafter probably the most visited and vigorously debated locality of the entire volcano complex.

The ‘cauldron-fill’ type succession (see (Table 2), p. 24) of Clough et al. (1909) was described from the precipitous crags between Aonach Dubh [NN 148 559], at the western entrance to the Pass of Glencoe, and the lofty Bidean nam Bian [NN 143 542], 2 km to the south. The succession was thought to consist mainly of lavas, and the complicated internal structure that was detected across the cauldron led the authors to conclude that eruptions had occurred at more than one centre. From their interpretations of the Glencoe cauldron, together with their analysis of the cross-cutting Clach Leathad Pluton (referred to by them as the ‘Cruachan Granite’; see p.98) and the Etive Pluton, Clough et al. (1909) formulated models of various styles of piston-like cauldron subsidence (Figure 5)a. These early, detailed and perceptive studies of a dissected volcano were also recorded in a memoir of the Geological Survey, Scotland, entitled The Geology of Ben Nevis and Glen Coe, and the surrounding country (Bailey and Maufe, 1916). This included the previously published four cross-sections of the cauldron (Bailey and Maufe, 1916, fig. 15, p.105) and was an explanation of the geology depicted on the one-inch to one-mile geological map, Sheet 53 (Glen Coe), the first edition of which was published in 1921. A second edition of the memoir was published in 1960 (Bailey, 1960), with some minor advances on interpretation of the Glencoe volcano. This edition included in a footnote a rather important correction concerning one of the cross-sections of the cauldron (Bailey, 1960, fig. 21, p.134), noting that dips on the boundary faults shown dipping inwards in fact dip outwards (for the significance of this see pp.14; 29). It is worthy of note, and admiration, that H B Maufe mapped the greater part of the volcanic rocks of the Glencoe cauldron as well as Ben Nevis, and also that E B Bailey’s huge contribution to unravelling the complicated geology of this region spanned some sixty years. This is not to belittle the important and wide-ranging contributions in this area by other Geological Survey geologists such as C T Clough, G W Grabham and H Kynaston.

Explosive volcanism

Roberts (1963, 1966a) recognised two major rhyodacitic welded ignimbrites within the volcanic succession; as was usual before the first recognition of welded ignimbrite (Marshall, 1935), the Geological Survey geologists had originally considered the rocks to be rhyolite lavas (see (Table 2)).

This first identification of major explosive activity at the Glencoe volcano heralded renewed interest both in the volcanic succession and in the boundary fault and its associated intrusions. Roberts (1963, 1966a) considered that the two ignimbrites recorded major cycles of volcanism that correlated with caldera subsidence and emplacement of fault-intrusions at depth. Like Clough et al. (1909), he attributed the two subparallel strands of the boundary fault to separate subsidence episodes, although he used the more prevalent term ‘ring-fault’ rather than ‘boundary-fault’. In a review, Roberts (1974) incorporated his own work with that of Taubeneck (1967) and Hardie (1968) and concluded that early explosive volcanism involved asymmetrical, trapdoor-like subsidence of the caldera floor, with magma withdrawal and consequent fault-block subsidence greatest at the north-eastern margin, towards Stob Mhic Mhartuin [NN 208 575] and Stob Dearg [NN 22 54] (Figure 5)b. He also interpreted certain steep contacts between metamorphic rocks and breccias inside the ring-fault in this north-eastern part of the volcano complex as part of a topographical wall of an early caldera.

Bailey (1960) suggested that some of the breccias adjacent to metamorphic basement just inside the ring-fault system in the north-east, between Stob Mhic Mhartuin [NN 208 575] and Meall a’ Bhùiridh [NN 25 50], might represent volcanic vents; thus he revised earlier Geological Survey conclusions that these were outliers of basal sedimentary rocks, with some representing infills of ‘landslip-cracks and earthquake-rents’ (Clough et al., 1909). However, Bailey offered no firm evidence for the revised interpretation. Hardie (1963, 1968) described the various breccias composed predominantly of fragments of metamorphic basement and postulated that they represent parts of a linear north-west-trending fissure-vent system; he interpreted them as explosion breccias. Taubeneck (1967) considered that breccias towards the north-west along the linear system discussed by Hardie (1968) were sedimentary, and he speculated that to the south-east some dyke-like tuffaceous breccias with invasive rhyolite represent vents. Moore and Kokelaar’s (1998) work, discussed more fully below, indicates that all of the earlier authors were partly correct. The breccias towards the north-west, on the slopes beneath Stob Mhic Mhartuin [NN 20 56] and beneath Stob Dearg [NN 228 547], are indeed mainly parts of the volcanic succession, including sedimentary deposits. The breccias farther south-east are in places invasive into the faulted caldera floor of metasedimentary rocks, and elsewhere occur in deep crevasses between metamorphic basement and volcanic rocks. These south-eastern occurrences probably do represent vents. Much of the earlier uncertainty was a consequence of the rather poor quality of exposure in the low but relatively accessible ground where those who followed the Geological Survey geologists chose to devote their attentions.

Ring-fault rocks and geometry

The ring-fault and its associated intrusions became the subject of close scrutiny by several geologists as processes of explosive volcanism in general were becoming topical. The view that the veins of flinty crush-rock had been produced by frictional melting of rocks along the ring-fault during large-scale subsidence, as originally interpreted by Clough et al. (1909), fell from favour as models involving streaming of particulate debris in a flow of escaping magmatic gas (fluidisation) were preferred. Reynolds (1956) suggested that the very fine-grained material represented intrusive tuff, whereas Hardie (1963) considered it was much-modified explosion breccia and Roberts (1966b) argued that it represented fluidised microbreccia, a view with which Taubeneck (1967) tended to agree. These problematic rocks are described later in this book along with fuller accounts of the previous interpretations, following which a new unifying model is proposed. Suffice it to say here that problems concerning the flinty crush-rock are not fully resolved, although an origin involving frictional melting is back in favour.

Taubeneck (1967) and (Roberts 1974) took the dips on the ring-fault mainly to be near-vertical or towards the inside of the volcano complex. Like Reynolds (1956), they interpreted inward-dipping strata within the Glencoe volcanic succession as evidence for compressional shortening of the subsided block owing to downthrow between inward-dipping faults. In addition, they suggested that the ring-fault, which they described as being shaped like an ‘upward-opening cone’ (see Figure 5)b, might have formed initially by upwards-directed magmatic pressure. This fracturing mechanism is like that which leads to the formation of cone-sheets at central volcanoes, but no direct evidence for magmatically induced uplift was presented. Later in this book (p.83) it is demonstrated both that there is no simple cross-section of the ring-fault system that has bounding faults that diverge (‘open’) upwards, and that the inward dips of the volcanic strata resulted from (extensional) downsag rather than compressional shortening. There are long sections of the ring-fault system with outward dipping fault planes, especially along the northern strands, including the type locality at Stob Mhic Mhartuin [NN 208 575]. It is possible that the erroneous view of upwards-divergent bounding faults arose from a mistake in the original paper by Clough et al. (1909). In their Plate XXXIII, three of their four cross-sections of the ‘cauldron-subsidence’ show the bounding faults to be parallel or slightly upwards-convergent, but their main north-west to south-east section shows upwards-divergent faults. The four cross-sections were reproduced showing the same fault geometry in both subsequent editions of the Geological Survey memoir (Bailey and Maufe, 1916, p.105; Bailey, 1960, p.134). Only an easily overlooked footnote in the second edition admitted that faults in the one section showing upwards divergence were wrongly drawn; faults in the south-east shown with inward dips in fact dip outwards, so that this section too should show parallel caldera-bounding faults.

In the far west of the volcano complex, Bussell (1979) mapped various breccias and intrusive rocks on the slopes of An t-Sròn [NN 13 55] and considered that they record the former presence of a volcanic centre at this locality. He found evidence for early intrusions that formed ‘felsite’ (rhyolite) and then granite, with later emplacement of mainly ‘dioritic’ rocks to form the Main Fault intrusion here. He argued that brecciation, involving explosive release of volatiles and followed by metasomatism, occurred in advance of the ascent of both the silicic and the intermediate magmas. Garnham (1988) reappraised the various fault-intrusions and studied their petrography and geochemistry. Importantly, she concluded that most of the ring-fault intrusions are chemically and mineralogically distinct from the volcanic rocks in the down-faulted caldera-fill succession and thus that the intrusions were unlikely to be their feeders. However, like Bussell (1979), she considered that the An t-Sròn composite intrusion (comprising mainly tonalite, with granite and diorite) might represent the root of a volcano, and that the marked contact metamorphism around the intrusion, which has sillimanite overprinted onto the regional garnetiferous mica-schists (Bussell, 1979), was due to protracted throughput of magma. Major movement on the ring-fault here must have postdated this protracted heating, as the inner (downthrown) rocks are not so metamorphosed; the chilled margin of the intrusion is slightly tectonised along the fault plane.

Sediment provenance

A recurrent topic of discussion in early studies was whether the granitic boulders that occur in several conglomerates immediately beneath and also within the volcanic succession were derived from the Rannoch Moor Pluton see (Figure 3). This pluton is cut by, and therefore predates, fault-intrusions of the eastern part of the Glencoe complex, such as on the slopes of Creag Dhubh [NN 25 52]. It consists predominantly of an outer, foliated K-feldspar-phyric biotite-granite, up to 1 km wide at outcrop, and an inner, foliated hornblende-granodiorite (Bailey and Maufe, 1916; Hinxman et al., 1923; Bailey, 1960); quartz-diorite, monzodiorite, monzogranite and syenogranite occur locally and the term ‘granitic’ is used in the broad sense. Petrographical analyses by Taubeneck (1967) confirmed that the granitic boulders were probably derived from the outer parts of the Rannoch Moor Pluton, which is what the Geological Survey geologists had suspected specifically for occurrences in Cam Ghleann [NN 249 520] (e.g. Bailey, 1960, pp.131, 147). This provenance is significant, because the pluton is one of the suite of numerous large, composite granitic intrusions that are genetically related to, and in some instances cut, the central volcanoes, like those of Glen Coe, Ben Nevis and Starav–Cruachan (see (Figure 3) and Read, 1961). The evidence from the boulders is that the Rannoch Moor Pluton was emplaced, uplifted and unroofed, before caldera volcanism at Glen Coe. As the pluton was probably emplaced at depths no less than some 2 to 3 km (compare with Droop and Treloar, 1981; Key et al., 1993), this unroofing suggests vigorous crustal uplift during the magmatic cycle.

Geochemistry and the origin of the magmatism

Although there has been considerable interest in the genesis of the volcanic and plutonic rocks of the region (Figure 3), particularly concerning the extensive lavas of Lorn (e.g. Groome and Hall, 1974; Thirlwall, 1979, 1981, 1982, 1986, 1988; Fitton et al., 1982; Trewin and Thirlwall, 2002) and the nearby plutons (Harmon and Halliday, 1980; Clayburn et al., 1983; Halliday, 1984; Harmon et al., 1984; Frost and O’Nions, 1985; Halliday et al., 1985; Batchelor, 1987; Holden et al., 1987; Tarney and Jones, 1994), there has until now been little systematic geochemical examination of the Glencoe Caldera-volcano Complex. Nevertheless, it is clear that the caldera volcano formed an integral part of the regional magmatic system (see p.103), sharing the same origin and overall tectonic setting. The geochemical studies of Siluro-Devonian volcanic rocks in northern Britain by Thirlwall (references above) included whole-rock analyses of 23 samples from the Glencoe volcanic succession, and Garnham (1988) presented analyses of 56 samples from the Glencoe fault-intrusions, with a few from the volcanic rocks. These results, together with more recently obtained data (J C Neilson, B P Kokelaar, J G Fitton and M F Thirlwall, unpublished results, 2005) and some regional findings, are briefly synthesised below.

The rocks of the Glencoe Volcanic Formation range from 52 to 78 weight per cent SiO₂ and fall within the range basaltic trachyandesite to rhyolite on a total alkalis versus silica classification (weight per cent Na₂O + K₂O versus SiO₂; see Trewin and Thirlwall, 2002, p.217). They have generally high levels of Ba, Sr and light rare-earth elements and are fairly typical of the high-K calc-alkaline suites that are common in the volcanoes of modern continental margins beneath which there is subduction of oceanic lithosphere. Garnham (1988) found a similar high-K calc-alkaline signature in the fault-intrusions, which range between about 52 and 72 weight per cent SiO₂, from gabbro through diorite and monzonite to granite. However, she thought that the rhyolitic rocks now preserved within the volcanic succession and intruding its basement have no counterpart in the fault-intrusions, except possibly in a granophyric granite that forms part of the An t-Sròn composite intrusion. All of the studies have found many of the rocks to be mildly or strongly altered, apparently mainly by hydrothermal activity.

Thirlwall (1981) identified geochemical features that unite most of the Siluro-Devonian volcanic rocks of northern Britain in a single province, and he interpreted spatial variations in concentrations of elements such as Sr, Ba, K, P and the light rare-earth elements as reflecting magmatism directly related to subduction of the Iapetus oceanic lithosphere beneath the Laurentian continent. He highlighted the existence of rocks that, from their unusually high Mg, Ni and Cr content, seem to have been little-evolved from primary magma compositions: that is, representing partial melts of mantle and having undergone little fractional crystallisation or contamination during relatively rapid ascent. From studies of trace elements and of isotopes of Nd, Sr and Pb in these relatively primitive rocks, Thirlwall (1982, 1986) subsequently invoked the mixing of the mantle melts with subducted oceanic sedimentary material akin to the Lower Palaeozoic strata exposed in the Southern Uplands. He recognised that the mantle source beneath the south-west Highlands was mildly enriched in incompatible elements and that it differed from the depleted mantle that was the source for the majority of the coeval lavas in the Midland Valley of Scotland. To account for these features he invoked layering in the lithospheric mantle, with the south-west Highland magmas originating at relatively deep levels, possibly at depths in excess of 200 km in the lowest lithosphere, or in the asthenosphere, close to an active or very recently defunct subduction zone that supplied the sedimentary material. Contamination of the primitive magmas by continental basement material was considered to be minimal, whereas this appears not to be true regarding the more evolved magmas.

From the geochemistry of the intermediate to silicic plutonic rocks in the south-west Highlands, Harmon and Halliday (1980), Clayburn et al. (1983), Harmon et al. (1984) and Halliday et al. (1985) showed that the derivation of the intrusive magmas did not significantly involve asthenospheric mantle influenced by active subduction, but that the magmas were derived from sources in enriched lithospheric mantle and various levels of the continental crust, commonly with hybridisation of different melts during ascent.

Although it had been recognised for a long time that the regional magmatism was ‘late’ in relation to closure of the ocean adjacent to the Laurentian continent, recent refinements in reconstructions and dating of the collisions of Laurentia with Baltica and Avalonia (e.g. Armstrong and Owen, 2001; Dewey and Strachan, 2003) have tended to suggest that oceanic closure was complete by about 425 million years ago, at just about the same time as the onset of the major magmatic activity that was to continue for some 25 million years. This timing, if it is correct, renders the fundamental cause of the magmatism somewhat problematical. Where there is continuous subduction of oceanic lithosphere, fluids from the down-going slab persistently induce partial melting of the overlying mantle, in the lithosphere or in the convecting asthenosphere, or in both. However, it is not clear for how long the down-going slab may sustain magmatism after oceanic closure and the consequent deformation due to plate collision; ultimately, when subduction of relatively buoyant continental crust becomes impossible, no new slab will be supplied. In the case of the former Laurentian margin, the recently proposed plate-tectonic reconstructions imply that magmatism was sustained with considerable intensity for a long time after obliteration of the ocean. It is not straightforward to explain the cause of such a long-lived and intense thermal anomaly. Conceivably subduction continued for some time after the oceanic basin ceased to exist, but for how long?

The apparent persistence of the the magmatism after continental collision was considered by Zhou (1985), who cited possible modern analogues in Turkey, Iran and Tibet, where post-collision calc-alkaline magmatism is voluminous. He did not, however, explain the fundamental cause of the (putative) long duration of post-collision magmatism. Trewin and Thirlwall (2002, pp.247–249) reviewed evidence critical to understanding the magmatism, and re-emphasised that the volcanic and plutonic rocks in the Midland Valley and Highlands of Scotland are older than and differ isotopically from the superficially similar igneous rocks farther south. It is easier to reconcile the northern suite with continuing subduction of Iapetus lithosphere, whilst accepting that the younger rocks to the south do require an alternative origin. However, Atherton and Ghani (2002) have suggested that the Baltica–Laurentia continental collision, which caused the Scandian orogenic event (Soper et al., 1992), led to break-off and sinking of the subducted slab of oceanic lithosphere into the asthenosphere. They postulated that the substantial thermal event recorded by the regional magmatism in the Highlands could have resulted from buoyant ascent of hot (but relatively ‘dry’) asthenosphere through the gap created by the slab break-off, with consequent heating and partial melting of the enriched (relatively ‘wet’) subcontinental lithospheric mantle, as appears to have occurred in the Alps (see Davies and von Blanckenburg, 1995; von Blanckenburg and Davies, 1995). Ongoing research, including high-precision age determinations, should shed further light on this problematical magmatic episode.

New models

Moore and Kokelaar (1997, 1998) studied the volcanic and sedimentary rocks that comprise the Glencoe volcanic succession and showed that caldera subsidence and infilling were too complex to be reconciled with any simple model of coherent-block movement on a ring-fault. Early subsidence was shown to have been incremental and piecemeal, with the locations and styles of downthrow of caldera-floor rocks profoundly affected by tectonically active, intersecting basement faults. The basement faults facilitated both the movement of magmas towards the surface and the localisation of the tectonic and volcanotectonic subsidence in the vicinity. The localisation of the subsidence in turn caused a major fluvial system to drain across the site during much of the volcanic history.

Consequently, rivers repeatedly incised the intra-caldera rocks, and fluvial and lacustrine deposits became extensively intercalated in the volcanic succession, along with alluvial-fan deposits that formed near active fault-scarps. Two instances of uplift have been detected from the more recent studies; they involve minor doming of the caldera floor due to shallow intrusion of andesitic sills. There is no evidence of cone-sheets or for major doming by resurgent magma (as occurred at the Valles caldera; (Figure 4)a.

The ring-fault system and associated intrusions formed only after the early incremental caldera developments and it seems that even then coherent-block subsidence on a ring-form discontinuity may not have occurred. While there is no doubt that downthrow on the ring-fault system, amounting to some 700 m in places, was an important factor in the preservation of the volcanic succession, it is also clear that early subsidence occurred entirely within the yet-to-form bounding ring-structure and that late caldera subsidence overlapped south-westwards a considerable distance beyond it (see p.81). Thus the role of the ring-fault system is not as originally conceived and, had there been less erosion so that south-western fault strands remained buried, a rather different volcanic structure would have been evident. Both because of its early influence in volcanological interpretations and because this superbly displayed and well-described system yields much in a modern analysis, it is thoroughly reappraised in this book.

Chapter 3 Overview of the succession and structure

‘Glencoe Caldera-volcano Complex’ is the new term applied here to include all of the volcanic and sedimentary rocks, coeval intrusions and volcanotectonic structures that relate to development of a volcano centred in the vicinity of the ‘cauldron’ originally defined by Clough et al. (1909). The caldera-volcano complex and its associated metamorphic substrate and host rocks are the subject of a new geological map at 1:25 000 scale (British Geological Survey, 2005), which includes several cross-sections. An overview of the entire succession and structure is provided here as an introduction and framework to set the scene for the following detailed descriptions and explanations of the successive volcanic stratigraphical units. The succession and structure are now interpreted in terms of caldera development that was tectonically influenced, incremental and piecemeal (Moore and Kokelaar, 1997, 1998), and in which successive depocentres enlarged and migrated with time. Many units of the volcanic succession are newly recognised or have revised names (see below). Erosional unconformities are common between successive units and in some instances there are minor intervening fluvial sedimentary strata; these are mostly palaeocanyon-confined conglomerates or thin sandstones that are too limited in outcrop to portray at 1:25 000 scale and have not been named. The volcanic and sedimentary rocks are collectively named the Glencoe Volcanic Formation. The volcanic formation unconformably overlies a ‘basement’ of intensely deformed metasedimentary rocks that are part of the Neoproterozoic to early Palaeozoic Dalradian Supergroup, which dominates the Grampian Terrane, between the Great Glen and Highland Boundary faults (Plate 1).

Dalradian metamorphic ‘basement’

The Dalradian metasedimentary rock sequence (Table 1), which is described in detail in the second edition of the Geological Survey memoir (Bailey, 1960) and by Hickman (1975), represents deltaic, intertidal and shallow marine sedimentation on a slowly subsiding rifted continental shelf that was to become the passive margin of the Laurentian Supercontinent (Anderton, 1985; Wright, 1988). The alternating quartzites and pelitic to semipelitic schists of the Lochaber Subgroup form a nearshore sequence reflecting periodic changes in sea level. Incursions of sand from deltaic areas to the south resulted in tidal sand bodies (the quartzites), which become finer grained and taper out north-eastwards towards more distal areas, now represented by the more uniformly semipelitic Loch Treig Schist and Quartzite Formation of the Glen Spean area (Key et al., 1997). Evidence for lagoonal basins with carbonate deposition is first seen in the upper part of the Leven Schist Formation, where thin metacarbonate and calcsilicate beds are found locally in the otherwise dominantly semipelitic rocks. Such metacarbonate and calcsilicate facies become dominant in the Ballachulish Limestone Formation, and in the vicinity of the Glencoe fault-intrusions these rocks are altered to calcsilicate hornfels. The overlying Ballachulish Slate Formation is generally graphitic and pelitic, representing anoxic basin conditions, but it is only present in the area of the 1:25 000 geological map as thin tectonic slices of black schist.

The Dalradian strata were deformed and metamorphosed at depths of between 15 and 30 km during the early to mid Ordovician (about 470 million years ago; Soper et al., 1999), largely in what is referred to as the Grampian Event of the Caledonian Orogeny. Several phases of folding can be recognised and, although there is now general agreement on the overall geometry of the major folds, there are various interpretations of their relative ages and how they relate to each other.

Interpretations of the structure of the area to the north and north-east of the Glencoe Caldera-volcano Complex have been published by Bailey (1934, 1960), Treagus (1974) and Hickman (1978) (see Stephenson and Gould, 1995, fig. 22). All of these authors agreed that the dominant structure is a large, recumbent, north-west-facing, isoclinal D₁ fold termed the Kinlochleven Anticline (Figure 6). The axial plane trace of this anticline is cut by the Glencoe fault-intrusions to the north of Meall Dearg [NN around 165 595]. To the west of here, strata from the Eilde Schist Member to the Leven Schist Formation are on the upper limb of the fold; they are the right way up and dip generally to the west. To the east of Meall Dearg, the strata are on the lower limb and are generally inverted. This limb has been affected by further close to tight, steeply inclined folds (D₁ or D₂ depending on author), and hence the strata vary considerably in their orientation (The regional fold phase D₂ was originally termed D₃ locally by Treagus, 1974). A complex antiform, containing both D₁ and D₂ folds according to Treagus (1974) and termed the Mamore Antiform (D₂) by Hickman (1978), is responsible for a core of Binnein Quartzite Member to the north of Sròn Gharbh [NN 180 595]. The wide outcrop of Eilde Flags Formation around the Devil’s Staircase [NN 215 575] lies in the core of the D₂ Blackwater Synform and farther east, close to the margin of the Rannoch Moor Pluton around [NN 235 570], outcrops of the Eilde Quartzite and Eilde Schist members define the core of the Blackwater Antiform (_D2; Treagus, 1974) or Treig Syncline (D₁; Hickman, 1978).

The Dalradian rocks to the west and south-west of the volcano complex were assigned by Bailey (1960) to the D₁ Ballachulish Syncline, which structurally overlies the Kinlochleven Anticline. The two recumbent regional folds are separated by the Ballachulish Slide, one of the major tectonic dislocations of the Grampian Highlands. The slide replaces much of the common limb and over most of this area it juxtaposes a widespread inverted sequence of Leven Schist Formation and Ballachulish Limestone Formation of the upper limb of the Ballachulish Syncline upon right-way-up rocks of the lower limb, generally at the level of the Glen Coe Quartzite Member and basal Leven Schist Formation (Bailey, 1960, figs 7G, 7H). The slide can be traced along the north-east side of lower Glen Coe, where it dips generally to the south-west, between Loch Leven and the Clachaig Hotel [NN 128 567]. From there the strike swings to the north-east close to the bounding faults of the volcano complex, and south-east-dipping strata of the Ballachulish Limestone Formation, much invaded by fault-intrusions, can be traced up the flanks of the Aonach Eagach to a point just to the east of Sgorr nam Fiannaidh [NN 143 583]. The slide is also present in several outcrops on the flanks of Glen Etive around Dalness, which have become known as the ‘Etive Windows’ (see 1:25 000 scale geological map section 3). In Gleann Chàrnan [NN 135 505] and Gleann Fhaolain [NN 155 515], near-identical structural inliers expose a right-way-up sequence of the Glen Coe Quartzite Member and Leven Schist Formation below the slide. On the south-east side of Glen Etive, outcrops of the Glen Coe Quartzite Member overlying the Leven Schist Formation on Stobh Dubh [NN 166 488] and faulted against the Leven Schist Formation on Beinn Ceitlein [NN 178 492] have less certain relationships. However, a capping of the Glen Coe Quartzite Member resting on the Leven Schist Formation on Beinn Maol Chaluim [NN 135 526] is demonstrably the right way up from current-bedding evidence and hence must overlie another slide at a higher structural level than the Ballachulish Slide. It should be noted that, in the classic section on the north side of Loch Leven (Figure 6)b, Hickman (1978) did not recognise the existence of a recumbent D₁ Ballachulish Syncline, only an upright D₂ Stob Ban Synform. To the west and south-west of Glen Coe, the Ballachulish Syncline, as described by Bailey (1960), is difficult to trace on existing maps and there have been no recent investigations. However, such a structure could be a continuation of the D₁ Beinn Sgulaird Recumbent Syncline identified to the south-west in Glean Creran by Litherland (1982).

Within the caldera-volcano complex, Dalradian rocks underlie the volcanic succession (see 1:25 000 scale geological map, section 1). In the west, around Loch Achtriochtan [NN 143 567], they are assigned to the Leven Schist Formation and, according to Bailey (1960, fig. 7E), they are all within the gently dipping inverted upper limb of the Ballachulish Syncline, above the Ballachulish Slide. In places adjacent to the bounding fault system, the strata dip steeply so that the slide and the immediately overlying Ballachulish Limestone Formation are exposed, for example at Meall Dearg [NN 165 589] and Coire Mhorair [NN 188 584] in the north and at Coire an Easain [NN 256 492] in the south-east. A perceived lower grade of regional metamorphism in the Leven Schist Formation inside the caldera-volcano complex was used as supporting evidence for the considerable subsidence of the central block (Elles and Tilley, 1930; Bailey, 1960). However, the evidence rested mainly on the absence of garnet, which is widely present in the Leven Schist Formation outside the caldera-volcano complex, and more precise determinations of metamorphic grade are lacking. In the east of the complex, between Stob Mhic Mhartuin [NN 208 574] and Coire an Easain, the Eilde Flag Formation and Eilde Quartzite Member crop out beneath the volcanic succession and are assumed to be part of the right-way-up sequence below the Ballachulish Slide.

The Dalradian rocks were modified further during the later phases of the Caledonian Orogeny. Following the Scandian Event there was substantial uplift and considerable strike-slip and dip-slip movements occurred on major crustal discontinuities (e.g. Treagus, 1991; Dewey and Strachan, 2003), such as the Ericht–Laidon Fault close to Glen Coe, the Great Glen Fault and the Highland Boundary Fault (Figure 3); (Plate 1). Contemporaneous magmatism included emplacement of the Rannoch Moor Pluton (Figure 3) and various minor appinitic intrusions, well before caldera-related explosive activity began at the Glencoe volcano. Erosion, presumably related to active uplift, produced a marked topography and was sufficiently rapid, at least in this vicinity, for the Rannoch Moor Pluton to be unroofed by the time of caldera volcanism (see pp.15; 40; 104).

Glencoe Volcanic Formation

A revised stratigraphy of the volcanic and related sedimentary rocks of the Glencoe Caldera-volcano Complex is summarised in (Figure 2)b and in (Table 2). The thick sequence of basaltic to andesitic sheets that was previously interpreted as a pile of lavas (Group 1 of Clough et al., 1909; Roberts, 1974; see (Table 2)) is reinterpreted as a stack of sills with intervening sedimentary strata. This reinterpretation is based on the recognition of peperites along the top contacts of the sheets; these are coarse- to fine-grained breccias with admixed vesicular sedimentary rocks, and they show that the sheets were intruded into unlithified wet deposits in which steam was generated (see Kokelaar, 1982, 1986). The sedimentary component of the sequence records fluvial deposition in a small basin taken to be a graben or half-graben. The igneous sheets and intercalated sedimentary strata are now referred to as the Basal Andesite Sill-complex. These rocks, as well as a former cover of lavas, were extensively eroded by rivers before the onset of the silicic magmatism that led to caldera formation. Preservation of the sills involved considerable fault movement, and it seems that the interval that separated the earlier magmatism from the initiation of caldera volcanism may have lasted for hundreds of thousands of years, possibly even longer.

The succession that reflects caldera development includes seven major silicic ignimbrites (rhyodacitic and rhyolitic compositions), each of which records a substantial explosive eruption with concomitant caldera subsidence. These ignimbrites are assigned to three distinctive members, each named after the vicinity of its thickest occurrence: Etive, Three Sisters and Dalness (Table 2). The upper surface of each ignimbrite shows evidence of fluvial erosion and incision. Alluvial-fan deposits, silicic tuffs, and various breccias are interstratified with the ignimbrites and are named after local features close to their type sections. In places, there are minor fluvial sedimentary rocks that have not been named. The various erosion surfaces and deposits between the ignimbrites indicate long intervals between the caldera-forming explosive eruptions, probably measured in several thousands to many tens of thousands of years. Thick andesitic units, comprising both lavas and high-level sills, lie between the silicic members and also register protracted intervals of time. It is tentatively estimated that the entire preserved caldera-related succession, which is not a complete record of the magmatism at this volcano, records between one and two million years of volcanism (see p.104).

The Etive Rhyolite Member consists predominantly of three flow-laminated rhyolites: the Lower, Middle and Upper Etive rhyolites (Table 2). These previously constituted the lower part of Group 2 of Clough et al. (1909) and much of the Lower Group 2 of Roberts (1974). Originally the rocks were thought to represent lava flows, but Moore and Kokelaar (1998) recognised that they formed from three discrete explosive eruptions, each associated with caldera subsidence. The lava-like quality of the three rather enigmatic units belies their explosive origins and reflects rapid near-vent deposition and coalescence of hot fragmented magma followed by fluid-lava-like laminar flow. The Etive rhyolites were erupted from vents that formed at the intersection of tectonic faults. Each rhyolite is underlain by fluvial or alluvial-fan sedimentary rocks, and silicic tuffs (Table 2).

The Lower Streaky Andesites occur as lavas and as sheets that intrude the Etive rhyolites. The characteristic millimetre- to centimetre-scale streaky texture is the result of incomplete mixing (here referred to as mingling) of coexisting rhyolitic and andesitic magmas. Earlier workers referred to these rocks as the Lower Group 2 Andesites (Table 2).

The succeeding Three Sisters Ignimbrite Member consists predominantly of two strongly welded (eutaxitic) ignimbrites: the Lower and Upper Three Sisters ignimbrites. These record two major explosive eruptions that differed in style from those that produced the lava-like Etive rhyolites. The ignimbrites are underlain and overlain by various sedimentary rocks and thick breccias (Table 2), and the assemblage mostly equates with the upper part of Group 2 of Clough et al. (1909). Roberts (1966a, 1966b, 1974), who first recognised the rocks as ignimbrite and referred to them as Upper Group 2, considered there to be only one eruption recorded at this level in the succession. In the original type section, at Coire nam Beitheach [NN 142 548], only the upper of the two welded ignimbrites is present and the lower unit initially was not distinguished.

The succeeding Upper Streaky Andesites record a further phase of activity in which andesitic magma mingled with rhyolitic magma. These andesites occur in a plug that forms part of a vent infill and as numerous irregular sill-like sheets that ramify widely within the Three Sisters Ignimbrite Member. An extrusive counterpart is also distinguished, where the upper contact is an unconformity.

The Glas Choire Sandstone Member comprises various conglomerates, sandstones and siltstones that record a period of fluvial incision succeeded by alluvial-fan sedimentation and development of a temporary caldera lake partly filled with lacustrine turbidites. These deposits, which include reworked parts of underlying thick breccias that are assigned to the Three Sisters Ignimbrite Member, formed part of the Group 3 Agglomerates and shales of Clough et al. (1909).

The succeeding Bidean nam Bian Andesite Member comprises thick andesite and dacite lavas that form the striking upper cliffs of Bidean nam Bian [NN 144 543] and Stob Coire nan Lochan [NN 148 548]. Clough et al. (1909) referred to these lavas as the Group 4 Hornblende-andesites (Table 2); their substantial thickness and strikingly monotonous character suggest that the erupted magma was topographically ponded and formed one or more deep lava lakes.

The Dalness Ignimbrite Member comprises two welded (eutaxitic) ignimbrites, named the Lower and Upper Dalness ignimbrites, each underlain by a silicic tuff (Table 2). These ignimbrites are thickest near Dalness, which indicates a south-westwards shift of the main locus of intracaldera deposition. The Lower Dalness Ignimbrite corresponds to the Group 5 Rhyolite of Clough et al. (1909) and to the Group 5 Ignimbrite of Roberts (1974). Sedimentary rocks and silicic tuffs overlying the Lower Dalness Ignimbrite were previously called Group 6 Shales and grits, while what is now the Upper Dalness Ignimbrite would have been within the former Group 7 Andesites and rhyolites (Clough et al., 1909).

Structural framework of the caldera volcano

Rather than occupying the entire area delimited by the ring-fault, all of the five extensively preserved ignimbrites of the Etive Rhyolite and Three Sisters members show clear evidence that they were thickly ponded in volcanotectonic fault-bounded basins lying within (not formed by) the ring-fault. Similarly, their eruptive vents were sited within the ring-fault system (see pp.54; 64; 69). Moore and Kokelaar (1997, 1998) demonstrated that the Dalradian metamorphic basement and the lower strata of the Glencoe Caldera-volcano Complex are fragmented by numerous faults and fault zones. Two sets of faults acted repeatedly to define a main north-west-trending graben, the Glencoe Graben, which is approximately 4 km wide and cross-cut by several faults (Figure 7). The pattern of faults within the ring-fault is probably more complicated than is portrayed in the diagram. Many of the faults along the graben are probably splays from deeper discontinuities and thus are linked, although at times they acted independently or reversed their sense of downthrow, or both. In places, the faults or fault-zones acted as extensional hinges for flexural subsidence (downsag), with associated development of crevasse-like, downward-tapering fissures that in some instances were hundreds of metres deep. Determination of the structural framework that is described here depends on both accessible contacts and complete stratigraphical sections, and it is likely that there are additional structures that have not been detected in particularly rugged terrain and beneath drift. All of the structures that comprise the framework depicted in (Figure 7) are validated on more than one line of evidence.

The Glencoe Graben is bounded by the Northeastern and Southwestern graben faults (Figure 7), which are zones, up to 1 km wide, across which there was repeated, often large-scale, normal-sense movement with various surface manifestations. In many instances the bounding structures comprised one or more distinct faults and fault scarps, but, either locally or at different times, they were defined by more gradual thinning of units, particularly of ignimbrite and also of sills. This gradual thinning reflects topographical barriers with moderate slopes that formed either by flexural draping over a deeper fault or by splaying of numerous faults from a deep fault, each with a minor increment of displacement. Minor downthrow in the opposite direction to the main sense of downthrow (antithetic displacement) occurred in some places where several fault strands were involved. Locally the faults are marked by crevasses that opened up to the surface, and in some cases the faults were exploited by ascending magma and acted as conduits to vents. In several instances, wedge-shaped deposits of coarse avalanche debris provide evidence of a steep and active fault scarp in the immediate vicinity.

Faults and fault scarps that define part of the Northeastern Graben Fault are well exposed in the steep crags on the flank of Am Bodach [NN 167 576] and can be traced above the cottage at Allt-na-Ruigh [NN 179 570] to the vicinity of the rocky prominence known as The Study [NN 183 564]. Farther south-east, the fault-zone is unexposed for almost 3 km where there are extensive superficial deposits. It is again exposed on the north-west flank of Stob Dearg at Coire na Tulaich [NN 218 547] and also on the south-east face near the prominent deep gully known as The Chasm [NN 225 539]. The fault-zone here is almost 1 km wide. South-east of Glen Etive, in lower parts of Cam Ghleann, fault strands and extensive zones of brecciation are exposed in the quartzitic basement and locally the fault traces are marked by tuff and breccia-filled dykes, some with invasive rhyolite [NN 247 522]. At a shallower level, a large volcanotectonic crevasse forms the major bounding structure for over 2 km south-eastwards onto the lower eastern slopes of Meall a’ Bhùiridh [NN 2562 5078], where it is cut by strands of the ring-fault and the associated intrusion.

Faults and fault scarps of the Southwestern Graben Fault are well exposed in the steep north-facing cliffs of Stob Coire nam Beith, close to the prominent Summit Gully [NN 138 548]; a small horst occurs between constituent fault strands towards the north-east [NN 1400 5495] (Figure 14), p.51. The fault-zone is intermittently exposed in the north-west-trending ridge of Bidean nam Bian, such as at the foot of the towering Church Door Buttress at the head of Coire nam Beitheach [NN 1415 5442], but farther south-east it is largely buried by the Bidean nam Bian Andesites and is mainly marked by pronounced thinning of ignimbrites and sills. Farther south-east, the Clach Leathad intrusion obscures the structure.

A linear zone of maximum caldera subsidence between the graben-bounding structures, referred to as the Glencoe Graben axis, lies approximately 1 km from the Northeastern Graben Fault; fluvial incision occurred repeatedly along this axis or close to the fault.

The Glencoe Graben is cut at right angles (orthogonally) by several faults that were active during volcanism. From north-west to south-east they are: Ossian Fault, Queen’s Cairn Fault, Devil’s Staircase Fault, Glen Etive Fault, and White Corries Fault (Figure 7). The middle three lie along the major north-east-trending valleys of the Lairig Eilde, the Lairig Gartain and Glen Etive, respectively, and these define the fault-bounded caldera-floor blocks represented in the massifs of the Three Sisters, Buachaille Etive Beag, Buachaille Etive Mòr, and Stob a’ Ghlais Choire, each of which has a significantly different volcanic and sedimentary succession (see later detailed descriptions and also Bailey (1960, pp.132–149)). The Ossian Fault, in the north-west of the volcanic complex, is clearly exposed in cross-section on Aonach Dubh (see (Plate 4); (Plate 14a), p.55)), near the steep cleft known as Ossian’s Cave [NN 155 563], which is where a post-caldera dyke has weathered out. Here the fault has two obvious strands roughly 200 m apart. The north-western strand was exploited by erosion that formed a palaeocanyon in the unconformity surface on top of the sill-complex (Plate 4). This feature, which is partially infilled with coarse fluvial conglomerate, establishes the early existence of the Ossian Fault. The later effects of the Ossian Fault on the strata immediately overlying the conglomerate are clear to see, but its trace towards the south-west is mainly buried beneath higher units. The fault is also obvious on the north side of Glen Coe, where it forms the deep gully adjacent to the cliffs of Am Bodach [NN 166 575] (Plate 14b). South-east of Glen Etive, the poorly exposed White Corries Fault is tentatively positioned south-east of Cam Ghleann [NN 24 51]; its location is inferred from aspects of the stratigraphy on Meall a’ Bhùiridh [NN 25 50] and in Coire an Easain [NN 25 49] (see pp.49; 64).

Many of the rectilinear caldera faults are truncated by the ring-fault system and its associated intrusions (Figure 7), and outside the ring-fault their possible extensions are largely obscured by later intrusions. However, the Devil’s Staircase Fault, locally exploited by dykes, is traceable north-eastwards beyond the ring-fault [NN 226 570] and forms part of the regional Etive–Laggan Fault. Moore and Kokelaar (1997, 1998) considered that this and the adjacent Queen’s Cairn and Glen Etive faults are splays from a deep north-east-trending crustal discontinuity. Similarly, the bounding north-eastern and south-western faults of the Glencoe Graben were inferred to splay from a deep north-west-trending crustal discontinuity, called the Glencoe Lineament (p.102). This feature evidently extended outside the ring-fault, as it persistently guided major fluvial drainage across the entire caldera complex. Minor dykes of andesite, dacite and rhyolite, which cut various levels of the volcanic succession but predate the Clach Leathad Pluton, mimic elements of the structural framework; they tend to lie parallel to the Glencoe Graben in the north-west, but towards the centre of the volcano complex, between the Queen’s Cairn Fault and Devil’s Staircase Fault, they strike roughly north–south (Figure 7), as if influenced by the cross-cutting structures. The most prominent multiple dyke of the area, part of the Etive swarm in Gleann Fhaolain [NN 15 51] (see Bailey and Maufe, 1916, fig. 28), evidently exploited a south-westward continuation of the Queen’s Cairn Fault.

The apparently simple, smoothly curving continuity of the ring-fault, as originally depicted by the Geological Survey (e.g. Clough et al., 1909, p.627; Bailey, 1960, p.132), is partly conjectural, because of incomplete exposure and partial obliteration by intrusions, and in places is an artefact of the ways in which the variously inclined and sharply intersecting fault planes trace across the steep topography. Constituent faults of the ring-fault system along the north, north-east and south-west (two strands) of the volcano complex are quite planar and generally parallel to the earlier Glencoe Graben except for several distinct right-angled steps in their traces (fault jogs). The fault traces at the north-west and south-east ends of the system connect to the longer sides via sharp angular bends (pp.85; 86). In a horizontal plane, as if the topography were flat, the overall shape of the fault system would appear rectilinear to polygonal, consistent with the considerable orthogonal faulting of the basement that was already present by the time of activity on the ring-fault system. Faults that constitute the ring-fault system cut the youngest part of the Glencoe Volcanic Formation, and it seems that many of them only developed large-scale downthrow after accumulation of the early ignimbrites that form most of the preserved volcanic succession (Etive Rhyolite Member and Three Sisters Ignimbrite Member). Even then, as in the earlier volcanotectonic history, it is probable that subsidence on the ring-fault system was not coherent, with certain sections moving more than others or at different times.

The ‘fault-intrusion’ complex of Clough et al. (1909) comprises numerous discontinuous intrusions that crop out extensively along and outside the ring-fault system (Figure 7); there are only a few (coeval) small intrusions inside the system. The early Geological Survey geologists made much of the fact that the inner contacts of the ring-intrusions are mainly smoothly planar, mostly with distinct chilled margins (some were later shown to be a separate rhyolite). From these features they argued that the intrusions had invaded the ring-fault and that where the fault was not exposed its original presence could be inferred from the continuity of an intrusion.

Numerous north-east-trending dykes, referred to as the Etive Dyke Swarm (Clough et al., 1909; Bailey and Maufe, 1916; p.100), cut both the caldera volcano rocks and the Clach Leathad Pluton; many, but by no means all (p.99), are related to late silicic magmatism towards the south-west, centred in the vicinity of the Etive Pluton (Anderson, 1937; (Figure 3)). The dykes have not been remapped or reappraised since the early work, and only some of them are represented on the 1:25 000 scale geological map (British Geological Survey, 2005), but their emplacement bears on the overall structure of the Glencoe volcano complex. Measurements of the cumulative width of Etive dykes cutting the Clach Leathad Pluton in the River Etive north-east of Alltchaorunn [NN 20 51], not far from the centre of the volcano complex, yielded a total of 306 m for 31 dykes in an outcrop length of 1036 m (Clough et al., 1909); this is almost one third of the outcrop in a north-west to south-east direction, amounting to 42 per cent extension. Although it was acknowledged (C T Clough in Bailey, 1960, pp.199–200) that this was probably too high a value to apply generally, later workers (e.g. Taubeneck, 1967) have used the original result to infer that the overall shape of the volcano complex, as defined by the ring-fault system, would have been subcircular before the Etive dykes were intruded. However, examination of the major outcrops across the whole volcano complex reveals that the dykes do not widely constitute as much as one third of the rock in a north-west to south-east transect. In some of the main ridges, as in Beinn Fhada [NN 17 55] and Buachaille Etive Beag [NN 19 55], the extension may amount to some 20 per cent overall, but it is considerably less elsewhere. At three other transects of the Etive swarm, with a total of 120 dykes, Anderson (1937, table 1) determined extension amounting to 9.3, 9.4 and 18.2 per cent. It is reasonable to suppose that the section from which the original measurement was derived is more highly populated by Etive dykes than elsewhere within the main massifs, because it is located in the vicinity of a pre-existing crustal weakness, recognised now as the Glen Etive Fault (a splay of the basement discontinuity represented by the Etive–Laggan Fault; p.28). The Etive dykes may well have exploited this weakness, as well as the other north-east-trending faults: for example, the Devil’s Staircase and Queen’s Cairn faults. Thus the overall extension is probably no more than about 10 to 15 per cent and the original outline shape of the Glencoe Caldera-volcano Complex would have been significantly elongate in a north-west to south-east direction before emplacement of the Etive dykes, and not nearly circular, as some have suggested.

While the preservation of the thick volcanic succession by faulting down within metamorphic basement remains perhaps the most striking feature of the Glencoe volcano, the original idea of coherent subsidence on a simple structure is no longer tenable. Much remains to be understood regarding the development of the bounding fault system, particularly in discriminating tectonic from volcanotectonic influences in fault development and in magmatic plumbing. These topics are more fully considered in later sections (pp.83; 92). From a historical perspective, it is interesting to contemplate how interpretations of caldera volcanoes might have differed in the past if this archetypal ‘cauldron’ had at the outset been described as involving tectonically fragmented crust and rectilinear to polygonal intersecting bounding faults. Would so many less-dissected volcanoes subsequently have been depicted as having simple continuous ring-faults around a coherent crustal block at depth? Similarly, would their inferred cross-sections have so frequently shown upwards-diverging bounding faults if the only cross-section of the Glencoe volcano showing this had been correctly drawn in the first place, without such divergence?

Chapter 4 Volcanic history: Glencoe Volcanic Formation

In the following pages, the Glencoe Volcanic Formation is described in detail according to a modern understanding of volcanic processes. This understanding derives substantially from the study of relatively young volcanoes, knowledge of which was limited at the time of the original work in the early 1900s. The aim here is to establish the geological evolution in terms of volcanic, sedimentary and tectonic processes. There is emphasis on those features that characterise the Glencoe volcano as probably the world’s most instructively exposed, tectonically controlled, piecemeal, multi-subsidence caldera volcano. For detailed petrographical descriptions of the rocks the reader should refer to the original Geological Survey publications, especially Bailey (1960); many of the rocks have been horribly altered by hydrothermal activity.

Basal Andesite Sill-complex (Precaldera fluvial sedimentation and andesitic magmatism, followed by protracted erosion)

The Basal Andesite Sill-complex (Plate 3); (Plate 4) is a unit, up to about 500 m thick, composed of a stack of basalt, basaltic andesite and andesite sheets with intercalated sedimentary layers. The stack rests unconformably on deeply eroded metamorphic rocks and is mainly preserved north-west of the Queen’s Cairn Fault and along the south-west flank of the volcano remnant, as far as the vicinity of Dalness (Figure 8). The sheets in the western outcrops, previously interpreted as lavas (Group 1 of Clough et al., 1909; Roberts, 1974; (Table 2)), show evidence of having been intruded as shallow-level sills in unlithified wet sediments that occupied a small sedimentary basin. It is possible that some of the sheets near Dalness are lavas, but they are not sufficiently well exposed to establish this. The preserved extent of the sheets is strongly controlled by faults and the original extent was certainly greater before faulting and erosional planation; the top of the sill stack is marked by a major unconformity (see p.104). To the north and north-west, such as at Stob Coire Leith [NN 152 583], An t-Sròn [NN 1355 5526] and Stob Coire nam Beith [NN 1378 5461], the sill stack is truncated by the ring-fault; to the south, near Dalness, it is also cut by the Clach Leathad intrusion (Figure 8). It is absent farther east.

A succession of steep scarps with intervening ledges, well displayed on the north and north-west faces of Aonach Dubh above Loch Achtriochtan [NN 14 56], shows that the sill stack comprises approximately 17 sheets, a few of which terminate or bifurcate along the outcrop (Plate 3); (Plate 4). The separation into distinct sheets is largely due to (ledge-forming) sedimentary intercalations and associated breccia layers, which occur throughout the stack.

Layers and pockets of conglomerates with bedded sandstones and siltstones are preserved beneath as well as within the sill stack, and these display an asymmetrical distribution on either side of the valley in lower parts of the Pass of Glencoe. On the northern side, moderately sorted to well-sorted conglomerates partially infill both small V-shaped gullies and larger steep-sided palaeocanyons, up to 20 m deep and 300 m wide, incised into the metamorphic basement. The conglomerates comprise rounded pebbles, cobbles and boulders (some over a metre in diameter), mainly of basement quartzite and semipelite, with some of andesite, ‘biotite granite’, ‘quartz diorite’ and ‘basic plutonic rock resembling kentallenite’ (Bailey and Maufe, 1916). These features are well exposed on the south-facing slopes beneath the Aonach Eagach (Plate 5), in a prominent outcrop that extends south-eastwards along the base of steep cliffs from Coire Liath [NN 151 580] towards the farm buildings at Achtriochtan [NN 157 574]. Plant remains have been recovered from micaceous siltstones at the base of the sequence. Higher in the sill stack here, and on the opposite side of the valley, such as below Aonach Dubh e.g. [NN 1412 5608], impersistent thin-bedded, medium- to coarse-grained brown sandstones, commonly with normal grading and lobate basal surfaces, are interstratified with red fine-grained sandstones and siltstones.

The deep canyons on the north side of the present valley are interpreted as having been cut by a major north-west-trending river, from which the conglomerates were deposited. To the south, beneath Aonach Dubh, bedded sandstones and siltstones were deposited mainly from suspension during fluvial floods, probably in overbank areas. The river is thought to have formed part of an extensive drainage system and, although its flow direction cannot be proved from this occurrence, later rivers in the same vicinity (at stratigraphically higher levels) evidently did flow from south-east to north-west.

The sill stack is dominated by sheets of dark, blue-grey basaltic andesite and andesite, which are 20 to 30 m thick. The andesites are generally fine grained, lack phenocrysts (aphyric texture) and are nonvesicular, but varieties with sparse phenocrysts, mainly of pyroxene, and some with vesicles, also occur. The upper and lower margins of the sheets are marked by various red-to-purple breccias and peperite, which is a mixture of quenched magmatic particles and sediment. These features reflect brittle fragmentation of magma due to stresses imposed by flow and cooling-contraction, or explosive activity caused by interaction of magma with water in wet sediment, or both (see Kokelaar, 1986). The red coloration is due to oxidation of iron. Along the upper contacts of most sheets, peperites and vesicular sedimentary rocks, with evidence of sediment fluidisation by steam, are taken as diagnostic of intrusion into wet sediment (Kokelaar, 1982). Hence the sheets are sills and not extrusive lavas as was previously thought.

Good exposures of the key features of the sills are accessible around the top contact of a basaltic andesite sheet that lies close to the base of the stack and forms the lower waterfall of the Allt Coire nam Beithach [NN 1408 5594]. East of the stream, along the peperitic upper contact, irregular fingers of andesite extend upwards into homogenised and vesicular sandstones (Figure 9). The overlying bedded sandstones, exposed in the walls surrounding the uppermost plunge-pool of the waterfall and several metres thick, are intensely convoluted and locally overturned; in places they show discordance along low-angle reverse faults that formed before the sandstones were lithified. At the lobate irregular base of the succeeding sheet, peperite intrudes fractures that extend upwards for several metres into the andesite. These features are very similar to those of the andesitic sills in Ayrshire (Midland Valley of Scotland) that intruded similar sediments and are of similar age (see Kokelaar, 1982, pp.22–28), and contact relationships such as these persist throughout the stack of sills in Glen Coe. Near the top of the stack, spectacular examples of coarse andesitic breccia with peperite, up to 8 m thick, are exposed on extensive glacially smoothed surfaces near the uppermost waterfall where the stream drains out of Coire nam Beitheach [NN 139 551]. This breccia (Plate 6a)(Plate 6b) locally contains jigsaw-fitting andesite blocks surrounded by red siltstone and fine sandstone, which indicates in situ autobrecciation, due to break-up of the intrusion margins during emplacement. In the vicinity, squeeze-up structures record upward interstitial flow and moulding of fluid andesite around earlier-formed breccia-blocks.

The top of the sill stack is marked by an erosional unconformity (Plate 4); (Plate 7), the temporal significance of which has not previously been recognised; it records considerable erosional planation and hence indicates a substantial interval of time between the early andesitic magmatism and the later caldera-volcano activity. Nevertheless, the unconformity is cut into by ancient river canyons in virtually the same area as the palaeocanyons located beneath the sill stack, on the northern side of the Pass of Glencoe (Figure 8). The unconformity cuts metamorphic basement elsewhere in the volcano complex where the andesitic sheets are missing. Fluvial sedimentary rocks overlie the unconformity in places; conglomerates, invariably containing rounded fragments of andesite as well as Dalradian metamorphic basement, occur in the palaeocanyons and pass laterally into various bedded sandstones, which are mostly interpreted as overbank flood deposits. The most instructive and accessible section of these fluvial deposits is exposed at the foot of Stob Dearg, beneath the slab and overhang known as the Waterslide [NN 2285 5455]. Here there is a palaeotopographical depression in the quartzitic basement, the uppermost parts of which are widely brecciated and record development of a regolith. The depression, interpreted as a fluvial palaeocanyon, is partially filled by some 8 m of purple and green conglomerates mainly composed of subrounded to rounded fragments of semipelites and quartzite, with andesite and, according to Hardie (1968), a 15 cm subrounded clast of ‘granite’. These conglomerates are laterally variable and are locally intercalated with purple and green sandstones and siltstones, which also form an infiltrated matrix. Steep shear bands locally cut and deform the conglomerates and, as these bands are parallel to the strands of the Northeastern Graben Fault in this vicinity, they are interpreted as recording minor strain due to activity in that system. Overlying the conglomerates are a further 8 m of thin-bedded to laminated, red, purple and green sandstones and siltstones, which are interpreted as fluvial overbank deposits. It was within these siltstones that plant and algal remains were found during the early Geological Survey work (see Bailey and Maufe, 1916, p.98); these were thought to confirm a ‘Lower Old Red Sandstone’ correlation with the lavas of Lorn (but see Kidston and Lang, 1924; Bailey, 1960; Wellman, 1994). Similar coarse conglomerates with abundant andesite clasts occupy a palaeocanyon that exploited a strand of the Ossian Fault where it cut the Basal Andesite Sill-complex, as seen on Aonach Dubh (Plate 4).

Nearby, at Lorn and in the Midland Valley (Plate 1), piles of andesitic lavas many hundreds of metres thick overlie coeval sills that were intruded into unlithified sediments. These sills were emplaced in active-basin settings similar to that at Glen Coe, and at more or less the same time (regarding the timing see Marshall, 1991; Wellman, 1994). By analogy with these nearby areas, it is inferred that andesite lavas originally covered the Basal Andesite Sill-complex at Glen Coe and that they were subsequently removed by erosion. An extensive early cover of lavas is suggested by the repeated fluvial supply of andesitic boulders from beyond the present limits of the volcano complex, especially from the east along with boulders from the unroofed Rannoch Moor Pluton (see pp.15; 40). The sills may represent the deeply eroded remnant of an andesitic volcano centred in the vicinity of Glen Coe, but no associated vent or conduit is recognised.

The evidence for fluvial systems below, between and above the sills, together with the thick stacking of the sills, reflects a strong tectonic control on basin development, sedimentation and andesite magmatism. The repeated incision by rivers in the same place indicates persistent tectonic control in a north-west-trending graben or half-graben. The preservation of the sill stack records a downthrow to the north-west of about 500 m on the Queen’s Cairn Fault, and it is clear from the palaeocanyon that exploited the Ossian Fault (Plate 4) that this fault also was active after the sills were emplaced. Hence it is apparent that both north-west-trending and north-east-trending faults were active at Glen Coe prior to caldera development. This fault activity is thought to reflect regional transtension, a combination of normal extension and strike-slip displacement (Moore and Kokelaar, 1997; see also Dewey and Strachan, 2003). Given the large amount of tectonic downthrow and the apparently considerable erosion of the lavas and sill stack that is recorded in the unconformity on the sills, an interval of hundreds of thousands of years, or more, may separate the andesite magmatism from the following recorded volcanism (see p.104).

Kingshouse Breccias: tectonic subsidence and alluvial-fan sedimentation

Following prolonged erosion of the precaldera andesites and widespread accumulation of conglomerates and other fluvial sediments, several small alluvial fans formed in the vicinity of the Northeastern Graben Fault. The largest fan, which is the main element of the rather heterogeneous sedimentary association referred to as the Kingshouse Breccias (Table 2), is well exposed on the lower north-eastern flank of Stob Dearg [NN 22 54], part of the Buachaille Etive Mòr massif (Figure 10). The overlying silicic Kingshouse Tuffs constitute a remarkable record of the onset of explosive activity centred close to the Northeastern Graben Fault (see pp.40–46). The type localities of the two units are not far from the Waterslide [NN 2285 5455], where the basal fluvial sedimentary rocks are also well exposed (p.34). They are easily reached using one of the climbers’ paths that lead from the main road, either from the car park at Altnafeadh [NN 2208 5630] to ascend gently, or, less pleasantly and including a river crossing, directly south-west from the road [NN 2368 5539].

The Kingshouse Breccias are up to 35 m thick and are well exposed between Central Buttress [NN 2282 5443] and Tulaich Buttress [NN 223 552], some 600 m to the north (Figure 10); (Plate 8a), (Plate 8b). They comprise thick-bedded breccias interstratified with cross-bedded sandstones and, towards the south-east, are topographically restricted in extent by two closely spaced fault scarps. These scarps are exposed on the eastern lower slopes of Central Buttress [NN 228 545], at the side of the rock slab that forms the Waterslide (Figure 10), (Plate 8). The faults show downthrow to the north-west and are orthogonal to the Northeastern Graben Fault. A small wedge of breccia at the foot of the upper scarp is talus buried by the alluvial deposits (Plate 8a), confirming the contemporaneity of the topographical barrier and the alluvial-fan sedimentation. The near-vertical fault scarp contacts exposed here, between the Kingshouse deposits and the psammite and quartzite of the metamorphic basement, were misinterpreted by Hardie (1968) as the walls of a vent infilled with explosion breccias (localities 5 and 6 on his figures 1 and 2).

The Kingshouse Breccias (Plate 8a), (Plate 8b) are characteristically very poorly sorted, with blocks up to 1 m in diameter predominantly of psammite, supported in a fine breccia to pebbly quartzose sandstone matrix. Most blocks are angular, but some rounded boulders also occur and rounded pebbles of vein quartz are common. Other clasts include angular fragments of porphyritic rhyolite, with rounded fragments of andesite and of granitic rock. Discontinuous lenses of poorly sorted thin-bedded sandstones, which mainly consist of well-rounded quartz grains, are intergradational with the breccias. Channels several metres wide and with steep margins are common in the breccias and these are mostly filled with cross-stratified sandstones. These features are impressively exposed on extensive glacially smoothed rock surfaces at and near the Waterslide [NN 2284 5454] (Plate 8b), but are most accessible at the foot of Great Gully Buttress [NN 2268 5486] (Figure 10)a, b.

Although the abundant angular blocks of psammite in the Kingshouse Breccias probably record development of talus and regolith from the metamorphic basement, the high degree of rounding of some boulders, and especially the rounding of quartz pebbles and granules in the sandstones, suggest considerable fluvial transport. Hence the alluvial-fan deposits are interpreted as recording incorporation of talus and regolith into a fluvial drainage system. The drainage was extensive enough to incorporate boulders and pebbles of volcanic and plutonic rocks that were exposed outside the Glencoe Caldera-volcano Complex. The chaotic organisation of the deposits suggests that they were deposited rapidly during flash floods. The granitic clasts, which probably derive from the Rannoch Moor Pluton (p.15), and possibly also the rounded andesite clasts, indicate provenance from the north-east or east, with dispersal generally towards the graben axis.

Another alluvial-fan deposit in the graben-bounding fault zone, at the same stratigraphical level as the Kingshouse Breccias described above, is exposed in the floor of Cam Ghleann in the south-east of the volcano complex [NN 2484 5181]. Here, unstratified conglomeratic breccias, dominated by boulders and subrounded blocks of Rannoch Moor granite (see petrographical analyses by Taubeneck, 1967, p.1301) and porphyritic andesite, grade generally south-westwards over 500 m into poorly sorted bedded breccias, in places with imbrication of coarse clasts indicating transport broadly from the north-east (Plate 9). The latter beds are intercalated with well-sorted sandstones that indicate subaqueous deposition. The andesitic clasts probably represent continued erosion of lavas remaining from an extensive volcanic field related to the Basal Andesite Sill-complex (pp.36; 38). The large size of the boulders, up to 60 cm, indicates steep drainage and substantial highlands nearby. It is conceivable that while there was more general uplift and erosion, for example causing unroofing of the Rannoch Moor Pluton, relative subsidence along the Glencoe Graben maintained steep slopes into it, as well as causing entrapment of the main fluvial drainage. The Rannoch Moor Pluton is exposed nearby, to the east and south-east of the Glencoe volcano complex. The common granitic clasts indicate that the pluton had been unroofed prior to development of the caldera volcano, and that the graben-axial river flowed from south-east to north-west.

Coarse breccias that rest upon more normal fluvial conglomerates and overbank siltstones, with a sharp contact, as at the Waterslide (see above), occur in several places at the same stratigraphical level along the Glencoe Graben. The breccias, which lack bed-thickening and coarsening-upwards trends and are coarse grained from the base, mark nearby onset of rapid fault-scarp growth with associated mass wasting. Such distinct switches, from relatively normal fluvial sedimentation to more catastrophic alluvial aggradation of sediment bearing abundant angular clasts, recurred throughout the subsequent history of the caldera volcano, commonly without any contemporary eruption and hence without any volcanotectonic subsidence (i.e. due to magma withdrawal). These switches of sedimentation without volcanism are regarded as a characteristic reflection of purely tectonic activity; in many cases, as described below, the switches were precursors to eruptions, which are thus considered as having been closely linked to the tectonic activity.

Kingshouse Tuffs: onset of explosive volcanism

Explosive volcanism at the Glencoe Caldera-volcano Complex began in the vicinity of Stob Dearg [NN 22 54] and is recorded by the Kingshouse Tuffs (Figure 11). Ascending rhyolitic magma interacted with groundwater in the Kingshouse Breccias, causing powerful phreatomagmatic explosions and eruption of abundant fine-grained ash. A tuff cone at least 2 km in diameter and possibly some 100 m high formed over the alluvial fan and is now superbly exposed in tangential section (Figure 10). Beyond the cone slopes the eruption formed a distinctive phreatomagmatic tuff layer, generally 1 to 2 m thick, which is preserved widely across the Glencoe volcano complex (Figure 11); Plate 10). On Aonach Dubh, near Ossian’s Cave (Plate 4), and on Dinner-Time Buttress [NN 146 559], canyon-filling ignimbrites up to 10 m thick beneath the tuff layer are interpreted as the result of channelling of pyroclastic currents some 7 km away from the vent. In this vicinity, the phreatomagmatic tuff layer forms the distinctive and more or less continuous overhang between the terraced and vegetated lower slopes formed by the Basal Andesite Sill-complex and the prominent upper buttresses of the Lower Etive Rhyolite (Plate 3); (Plate 4).

The Kingshouse Tuffs are thickest, at up to 70 m, on the lower slopes of North Buttress on Stob Dearg (Figure 10); 12a). They generally fine upwards from tuffaceous lithic breccias at the base (Figure 12)c to stratified and cross-stratified tuffs above (Figure 12)b. The deposits are interpreted as part of the outer flank of a low-profile tuff cone; there is no evidence of the extensive unconformities and slump structures that characterise the relatively steep inner slopes and vent areas of such cones (e.g. see Fisher and Schmincke, 1984). The lower tuffaceous breccias contain abundant blocks of psammite and smaller sandstone clasts derived from the underlying alluvial fan. Higher in the succession, thin-bedded tuffaceous breccias locally display bomb-impact sags that indicate ballistic trajectories from a vent that lay to the west or south-west (Plate 11a). Upper parts of the tuff-cone succession comprise planar-stratified and cross-stratified silicic tuffs with abundant accretionary lapilli formed by the aggregation of moist ash particles in an eruption plume.

Cross-stratified tuffs show long-wavelength (more than 1 m), low-amplitude (less than 40 cm) bedforms and low-angle scour surfaces (Figure 12)b; (Plate 11b). In places these are cut by steep south-west-facing scour surfaces, with original slopes at up to 45°, some of which are draped by steeply dipping laminated tuff that locally shows slump structures. These features are interpreted as scours that faced and were eroded by energetic pyroclastic currents, and were then plastered with wet ash. Excellent examples of the various stratified tuffs are exposed in the lower flanks of Great Gully [NN 2266 5479].

The tuff cone was constructed by both pyroclastic density currents and pyroclastic fallout during eruption from a vent that probably lay only a short distance south-west of Great Gully Buttress (no more than a few hundred metres into the hill). During early stages of the eruption, significant amounts of alluvium and interstitial water were incorporated in low eruption columns and these generated mostly high-concentration pyroclastic currents that rapidly deposited the mainly massive tuffaceous lithic-breccias. The fining-upward sequence reflects gradual clearing of debris from the vent, with progressive development of more efficient magma-water interaction and hence more explosive and higher eruption columns. The more energetic columns incorporated more ambient air and produced more dilute and more mobile currents that in turn deposited planar-stratified and cross-stratified silicic tuffs. The extensive scours within the succession (e.g. (Figure 12b); (Plate 11a), (Plate 11b)) mark periods when the pyroclastic currents were erosive. The regionally widespread accretionary lapilli-bearing tuff layer (Figure 11a; (Plate 10), which is a hybrid deposit formed by pyroclastic fallout and dilute pyroclastic currents, was deposited during construction of the upper parts of the tuff-cone succession.

On the north-east face of Sròn na Creise [NN 242 524], 2 km south-east of the tuff cone and vent, the correlative succession thickens abruptly to 40 m in a small fault-bounded basin (Figure 11); (Plate 12a), where it shows evidence of subaqueous deposition (Plate 12b). The strata here consist predominantly of tuffaceous sandstones with intergradational massive, graded, planar-laminated and cross-laminated divisions. These are turbidites, deposited from aqueous density currents, and they become thinner bedded and finer grained towards the south-east. Some tuffaceous sandstone beds show irregular downward loading into parallel-stratified tuffs. Liquefaction structures, such as convolute lamination and flame structures, are common (Plate 12b) and the strata show evidence of widespread sliding and slumping along low-angle detachment surfaces. This succession records syneruptive lacustrine (lake-bed) deposition from turbidity currents, with intermittent fallout of fine phreatomagmatic ash from suspension. It is inferred that the tuff cone grew across the Northeastern Graben Fault (zone) and the Glencoe Graben axis and dammed the graben-axial river, forming a lake to the south-east (Figure 11). South-east-directed pyroclastic currents apparently crossed the lake shoreline and formed turbidity currents, while sedimentation of ash from suspension formed the plane-parallel stratified tuffs (Plate 12b). Much of the ejecta that by-passed the cone’s south-eastern subaerial flank was trapped and deposited in the lake.

Lower Etive Rhyolite: large-volume rhyolitic eruption and graben-like caldera subsidence

Caldera subsidence was initiated with eruption of the first of the three enigmatic flow-laminated rhyolites that constitute the bulk of the Etive Rhyolite Member (Table 2). Earlier workers did not distinguish the three units, and, quite reasonably for the time, considered that the rhyolites originated as slowly extruded, viscous lavas. Now they are recognised as having formed from three discrete episodes of explosive eruptive fountaining, in which the fragments of magma remained sufficiently hot and fluid during pyroclastic dispersal to coalesce during deposition and only then form a lava-like flow (Moore and Kokelaar, 1998). The rhyolites differ from slowly extruded lavas in showing evidence for the combination of (1) explosive (pyroclastic) origins, (2) progressive deposition from the base upwards, and (3) rates of eruption sufficiently high to cause contemporary caldera subsidence and infilling. Such rocks are known from other volcanoes, both within and outside calderas. Examples occur in the USA (Ekren et al., 1984) and in England (Branney et al., 1992) and here they are referred to as lava-like ignimbrites (Moore and Kokelaar, 1998). The three lava-like ignimbrites at Glen Coe, now called the Lower, Middle and Upper Etive rhyolites, are each underlain by silicic tuffs: the Kingshouse Tuffs (see above), the Raven’s Gully Tuffs and the Crowberry Ridge Tuffs, respectively. These tuffs constitute remnants of cones stacked in vertical succession in the position of Stob Dearg (Figure 10), with the intervening Etive lava-like ignimbrites recording increments of caldera subsidence. The repeated eruptions at this site reflect basement-fault control of the magmatic plumbing. The opening phreatomagmatic (tuff-cone) activity of each eruption was a consequence of eruption through wet alluvial sediments, the localisation of which also reflects the action of underlying faults. All three Etive rhyolites and their associated tuff cones are well exposed on Stob Dearg (Figure 10), but the Lower Etive Rhyolite has its type section in the north-west of the complex, on the north-west face of Aonach Dubh [NN 143 554].

The Lower Etive Rhyolite is exposed from Meall a’ Bhùiridh [NN 257 501] in the far south-east of the Glencoe volcano complex to Stob Coire nam Beith [NN 139 548] in the far north-west. It is up to 110 m thick and has a low aspect ratio (thickness: lateral extent) of about 1:140. It shows ponding relationships against faults and flexures of both the Northeastern and the Southwestern graben faults and displays some substantial thickness changes across cross-graben faults (Figure 13). The rhyolite is generally crystal poor and dominantly flow laminated. Locally, especially towards the south-east, it contains several volume per cent of feldspar crystals and sparse small lithic clasts (Plate 13a), (Plate 13b). North-west of the Ossian Fault, stratified lithic tuffs occur at the base of the sheet. Autobreccias (brecciated rhyolite) are widely present along the top and also within the sheet.

In the north-west, the shape of the Lower Etive Rhyolite in cross-section between its bounding structures to the north-east and south-west suggests that it infilled a half-graben. On the north-west face of Stob Coire nam Beith [NN 139 547], close to Summit Gully, the sheet is ponded against the scarps of two closely spaced normal faults with combined downthrow of more than 80 m (Figure 14). These faults, which mark the Southwestern Graben Fault (Figure 13), have been rotated by at least 35° owing to later caldera-related subsidence, and the Lower Etive Rhyolite does not extend farther to the south-west. From this location, the rhyolite sheet can be traced across Glen Coe towards The Study [NN 180 564], where it wedges out across a steep flexure defined by an approximately 0.5 km-wide zone of steepening dips in the underlying Kingshouse Tuffs and Basal Andesite Sill-complex. The flexure marks the Northeastern Graben Fault (Figure 13) and is most clearly displayed south-east of the Meeting of Three Waters [NN 176 562].

North-west of the Ossian Fault, in the buttresses on the north-west face of Aonach Dubh [NN 14 55], the Lower Etive Rhyolite is not altogether lava-like and shows large-scale subparallel lithological stratification (Plate 3). This stratification is particularly striking when the buttresses are viewed from the north-west, with the summer sun low in the evening sky. The most south-westerly of the buttresses, G Buttress [NN 143 554], is the type section for the Lower Etive Rhyolite (Figure 15). Here, up to 20 m of stratified lithic-tuffs at the base of the sheet show alternations of lithic-rich and lithic-poor layers. The tuffs contain abundant blocks of flow-laminated rhyolite, with others of andesite and of quartzite, in a welded matrix with flattened pumice clasts (fiamme), especially in the lithic-poor layers. These features record pyroclastic deposition; the flow-laminated clasts could represent early-deposited lava-like ignimbrite shed from the Ossian Fault scarp as it grew during the eruption (see below). The stratified lithic tuffs grade upwards via crudely stratified tuffaceous breccias into rhyolite breccia, up to 20 m thick, which in turn passes upwards into coarsely flow-laminated (lava-like) rhyolite and further into finely flow-laminated rhyolite. The latter is up to 50 m thick and is characterised by intense flow-folding. Autobreccias gradationally and irregularly overlie the flow-folded rhyolite.

To the south-east, the lava-like ignimbrite buried a north-west-facing scarp of the Ossian Fault that was approximately 30 m high. The fault scarp is clearly exposed in cross-section on the north-east face of Aonach Dubh, a short distance to the east of Ossian’s Cave [NN 157 563], where it offsets the top of the Basal Andesite Sill-complex (Plate 4). The basal stratified lithic tuffs of the type section are absent south-east of the Ossian Fault and the unit throughout much of the rest of the extensive outcrop is represented by finely flow-laminated rhyolite. In the far south-east of the volcano complex, the rhyolite shows abrupt thinning from 80 m to less than 30 m, which is inferred to register initial topographical restriction of the lava-like ignimbrite by the escarpment of the White Corries Fault (Figure 13).

On the flank of Stob Dearg, near Creag na h-Uamhaidh [NN 2202 5523], a short distance to the south-east of the path into Coire na Tulaich, the finely flow-laminated rhyolite locally includes metamorphic basement lithic fragments and sedimentary inclusions in more or less planar layers. The debris was apparently derived from unconsolidated Kingshouse Breccias and Kingshouse Tuffs. Autobreccia at the base comprises clasts of (nonvesicular) rhyolite in a matrix with relict pumice fragments from pyroclastic debris. Although no vent for the Lower Etive Rhyolite is exposed, the included clasts derived from the Kingshouse Breccias and Kingshouse Tuffs suggest that the vent, or a series of closely spaced vents, cut these deposits, probably within the Buachaille Etive Mòr–Stob Dearg massif in the vicinity of the vent of the Kingshouse Tuffs. This manifestly was the case for the Upper Etive Rhyolite (see p.54).

Unlike slowly extruded rhyolite lavas, which normally form domes flanked by blocky talus or have steep, free-standing brecciated flow margins, the Lower Etive Rhyolite has no such original upper surface. Its upper contact is generally more or less planar, while the sheet as a whole has a low aspect ratio of about 1:140. The restriction of the sheet against topographical barriers that developed in response to the eruption (i.e. by caldera subsidence), without brecciation at the contacts, indicates both fluid behaviour and high rates of eruption of a substantial volume of magma. The absence of vesicles and any larger cavities that were originally gas-filled (lithophysae) indicates that the rhyolite was substantially degassed by the time of its final emplacement. Hence it is inferred that the rhyolite formed from an efficiently degassing fountain of fluid pyroclastic particles, reflecting catastrophic venting. An explosive eruptive origin for the sheet is also suggested by the gradation from stratified eutaxitic tuffs containing diverse lithic clasts at the base of the sheet north-west of the Ossian Fault, and by other relict pumiceous textures, although these alone are not diagnostic. It is envisaged that a lava-like flow formed directly from the coalescence of fluid magma particles in lower parts of the pyroclastic fountain close to the vent. The parallel layers of lithic fragments within the laminated rhyolite probably represent periodic incorporation of clasts into the vent or conduit during the progressive build up of the flow. Much of the flow flooded into and was ponded within the actively subsiding Glencoe Graben. Large-scale stratification of the sheet in the north-west of the volcano complex proves progressive accumulation there. It is possible that, in the early stages of the eruption, erosion and nondeposition occurred close to the vent while accumulation occurred in the north-west of the graben, and that at a late-stage flow-laminated rhyolite was emplaced across most of the graben.

Raven’s Gully Tuffs, Middle Etive Rhyolite, Crowberry Ridge Tuffs and Upper Etive Rhyolite: further cycles of rhyolitic phreatomagmatic and magmatic activity with intervening fluvial incision

On Stob Dearg, at Broad Buttress [NN 2252 5485] and at Tulaich Buttress [NN 2232 5500] and [NN 2222 5507], there are remnants of a steep-sided palaeocanyon that was eroded up to 15 m deep into the Lower Etive Rhyolite and filled with breccias and boulders (Figure 10)b. This feature records re-establishment of the fluvial system on the north-east side of the Glencoe Graben. The silicic Raven’s Gully Tuffs overlie this palaeocanyon and its associated unconformity (Table 2), and, like the Kingshouse Tuffs, their outcrop on the north-east face of Stob Dearg [NN 22 54] represents a section through the outer flanks of a tuff cone (Figure 10). Abundant bomb-impact sags reveal that the vent, as in the earlier explosivity, lay to the west or south-west of the present outcrops. Deposits forming the low flanks of the cone pass laterally into a thin (1–2 m) phreatomagmatic tuff layer. North-west of the path leading into Coire na Tulaich, at Creag a’ Bhancair [NN 2188 5505], the tuff layer is locally interstratified with, and overlain by, lenses of tuffaceous breccia up to 10 m thick, derived from collapse of a fault scarp at or near the Devil’s Staircase Fault. Although there are no exposures of extensive sedimentary deposits formed by the re-established river, fragments of psammite within the Raven’s Gully Tuffs suggest that the eruption involved reworking of underlying alluvium. As with the Kingshouse Tuffs, the Raven’s Gully Tuffs phreatomagmatic eruption probably also resulted from the interaction of ascending magma with groundwater and surface water.

The overlying Middle Etive Rhyolite, known only on Stob Dearg (Figure 10), was topographically confined by the faults bounding the Glencoe Graben and by the cross-graben Glen Etive Fault and Devil’s Staircase Fault. The sheet is predominantly flow laminated and is thickest, at 100 m, near Central Buttress [NN 226 541]. To the south, along the eastern face of Stob Dearg [NN 22 53] (Figure 10)c, it thins progressively and wedges out in the vicinity of a series of closely spaced normal faults, referred to as The Chasm step-fault system (Figure 7). To the north-west, close to Creag na h-Uamhaidh [NN 2185 5502] and the path leading into Coire na Tulaich, the rhyolite thins to 30 m across a normal-fault scarp 850 m south-east of the Devil’s Staircase Fault. Although the termination of the Middle Etive Rhyolite is not exposed, it is considered that the scarp of the Devil’s Staircase Fault (indicated by breccias) formed its north-western limit. The rhyolite records a second major magmatic eruption from a vent in the vicinity of Buachaille Etive Mòr–Stob Dearg. As in the first cycle, there was an initial phreatomagmatic phase and collapse of fault scarps.

Following emplacement of the Middle Etive Rhyolite, the graben-axial river was re-established and another palaeocanyon was eroded close to the Northeastern Graben Fault. This palaeocanyon is well exposed north and north-east of the prominent rocky knoll known as The Study [NN 183 564] and farther north-west on the lower slopes of Am Bodach [NN 172 573], where the steep north-eastern wall of the canyon is overturned south-westwards owing to subsequent caldera downsag. Fluvial pebble-bearing red sandstones and siltstones, up to 3 m thick, occur in the palaeocanyon and also form overbank deposits elsewhere, such as at Creag Doire-bheith on the lower slopes of Beinn Fhada [NN 1782 5618]. However, the sedimentary deposits have been largely obliterated by intrusions of the Lower Streaky Andesites. Part of a deep topographical depression cut into the Lower Etive Rhyolite close to the Northeastern Graben Fault is also exposed on the north-east flank of Sròn na Creise [NN 242 521] (see (Plate 17)), but it remains unclear when the erosion was initiated here relative to the timing of emplacement of the Middle Etive Rhyolite. The feature has no fluvial infill and the geometry of the exposure suggests that the topography was cut across the trend of the main graben fault.

On Stob Dearg [NN 22 54], the silicic Crowberry Ridge Tuffs are part of a third tuff cone stacked above the two earlier cones (Figure 10). The tuffs take their name from Crowberry Ridge [NN 2255 5440], which is near the tuff-cone outcrop and is formed of rhyolite that cuts the cone centre. To the south-east and north-west of Stob Dearg, the tuff-cone succession is represented solely by an extensive accretionary lapilli-bearing phreatomagmatic tuff layer, 1 to 5 m thick, which can be traced for up to 6 km from the vent as far as the Ossian Fault in the north-west and some 4 km into Cam Ghleann [NN 2457 5146] in the south-east; it thickens into the base of the topographical depression exposed on Sròn na Creise (see above and (Plate 17)) and is missing on the south-east flank of this. The Crowberry Ridge Tuffs record a further phreatomagmatic eruption centred near the Northeastern Graben Fault and, although no sedimentary layer is exposed beneath the tuffs, groundwater or standing water must again have been involved with the ascending magma.

The overlying Upper Etive Rhyolite, up to 120 m thick, is an extensive flow-laminated lava-like ignimbrite ponded against graben and cross-graben faults and flexures. North-west of the Ossian Fault, in the north-west face of Aonach Dubh [NN 14 55], the ignimbrite is less than 30 m thick and overlies the autobrecciated top of the Lower Etive Rhyolite. To the south-east, it thickens within a few hundred metres to 120 m, across a steep flexure and fault at the position of the Ossian Fault (Plate 14a). The sense of downthrow is reversed relative to the subsidence that influenced the emplacement of the Lower Etive Rhyolite. South-east of the flexure, thick autobreccias occur at the base of the laminated rhyolite and, within the sheet, a large-scale fold of the lamination (with a wavelength of 500 m) is truncated by a major internal slump detachment. These features are well exposed at Guala Laidir, on the south-east face of Aonach Dubh [NN 1572 5592], and they suggest late-stage slumping away from the nearby flexure that developed over the Ossian Fault. On the northern side of the Pass of Glencoe, the lava-like ignimbrite is ponded, without brecciation, against a steep scarp of the Northeastern Graben Fault (Plate 14b). The fault scarp is well exposed in the south-west-facing crags of Am Bodach [NN 1665 5762], where they tower above the deep gully formed on the Ossian Fault. To the south-west, at the head of Coire Gabhail [NN 156 545], the ignimbrite thins dramatically across a flexure that defines the Southwestern Graben Fault zone. These relationships indicate that during the eruption of the Upper Etive Rhyolite the 4 km-wide volcanotectonic structure at Glen Coe had the form of a trapdoor-like half-graben, with asymmetry opposite to that of the half-graben that formed during the Lower Etive Rhyolite eruption. South-east of Glen Etive, the Upper Etive Rhyolite is about 100 m thick on Sròn na Creise [NN 240 522] and some 40 to 50 m thick in Cam Ghleann [NN 2454 5145], beyond which it is not found.

On Stob Dearg, in the conspicuous upper part of North Buttress [NN 2256 5455], a near-vertical and roughly cylindrical body of flow-laminated rhyolite is exposed in oblique section, cutting downwards for almost 200 m through the thickest sections of the Crowberry Ridge Tuffs, Middle Etive Rhyolite, and Raven’s Gully Tuffs (Figure 10). The flow laminations are steep and apparently continuous with those in the main part of the Upper Etive Rhyolite and the feature is interpreted as part of an infilled vent from which the lava-like ignimbrite was erupted. The eruption resulted in large-scale subsidence along the Glencoe Graben, within which flow was restricted by steep cross-graben flexures and fault scarps, notably the Ossian Fault in the north-west and, presumably, the White Corries Fault in the south-east (Figure 16).

Lower Streaky Andesites: extrusion and shallow intrusion of andesite followed by fluvial incision

Emplacement of the Etive rhyolites was followed by a phase of andesitic magmatism represented by the Lower Streaky Andesites (Table 2), which form sheets up to 100 m thick in northern parts of the volcano complex, on both sides of the lower reaches of the Pass of Glencoe. The pervasive streaky texture of the rocks is the result of mingling (incomplete mixing) of andesitic and rhyolitic magmas (Plate 15a). Andesite is volumetrically dominant, with the rhyolite forming 10 to 50 per cent of the rock; irregular bands of purple-grey andesite alternate with cream-coloured rhyolitic bands and in places there are bleb-like patches of the rhyolite.

Within the Glencoe Graben, mainly on the southern side of the present valley, the andesitic sheets were emplaced as sills along the unconformity surface between the Lower and Upper Etive rhyolites. They have peperitic lower and upper contacts, which, like those around the sheets in the Basal Andesite Sill-complex, record shallow-level interaction of the magma with wet sediments. The largest sill is spectacularly exposed in the steep north-eastern face of Gearr Aonach [NN 1675 5608], where it is approximately 100 m thick between the Lower and Upper Etive rhyolites (Plate 15b). Its rather abrupt south-westward termination is exposed in the south-eastern slopes of Gearr Aonach, towards the prominent rockfall boulder field in Coire Gabhail [NN 1676 5576]. Similar abrupt terminations are evident in the lower part of Coire nan Lochan, in the north-west face of Gearr Aonach [NN 1608 5584] and in the south-east face of Aonach Dubh [NN 1581 5593] (Plate 15b). Towards the north-east the sheet thins at Creag Doire-bheith [NN 1781 5617], on the gentle slopes north of Beinn Fhada. Where the unconformity surface between the Lower and Upper Etive rhyolites includes a deep palaeocanyon, as in the vicinity of The Study [NN 1820 5637], the intrusive andesite appears to have been restricted by the original canyon walls. Much of the steep palaeoscarp that follows the Northeastern Graben Fault, from the Allt Coire Meannarclach [NN 1879 5642]; near the large beehive-like cairn] north-westwards towards the lower slopes of Am Bodach [NN 1695 5750], also acted to confine the intrusive andesite (LSA-sill in (Plate 14b)). This originally near-vertical, south-west-facing scarp of Basal Andesite Sill-complex has been rotated, top south-westwards, by later caldera downsag, so that the older rocks now overlie the younger rocks originally restricted by the scarp (both Upper Etive Rhyolite and intrusive andesite). This distribution of the sill indicates that the andesitic intrusion was restricted between the Ossian Fault and the two faults that bound the Glencoe Graben, with distinct thickening (up to 100 m) towards the middle ground. Its emplacement evidently was at least partly facilitated by volcanotectonic subsidence, as the overlying strata were not pushed up so far as to accommodate the full thickness of the lenticular sheet. Some relatively minor uplift of the caldera floor probably did result from the sill intrusion, and this case constitutes one of the only two instances where uplift due to magma emplacement can be inferred fairly confidently at this volcano (see also p.74).

An extensive unit of andesites, over 3.5 km 2 in area and widely up to 50 m thick, is exposed on the ridges between Meall Dearg [NN 162 583], Am Bodach [NN 169 580] and Sròn Gharbh [NN 175 584] (Plate 14b), and extends towards A’ Chailleach [NN 184 571]. This unit lies outside and to the north-east of the Glencoe Graben, where it was probably extruded as lavas. Its original extent is unknown; it is truncated to the north by the ring-fault system. Given that there was contemporary volcanotectonic subsidence within the Glencoe Graben, to accommodate the intruded andesite, it is probable that the lavas were erupted from one or more vents in the vicinity of the graben.

The Lower Streaky Andesites and other parts of the caldera-fill succession were eroded by rivers, which, like the earlier fluvial systems, formed a main canyon close to the Northeastern Graben Fault. This palaeocanyon can be traced for approximately 1 km between The Study [NN 1826 5647] and above the cottage at Allt-na-Ruigh [NN 1742 5690]. Like its predecessor, this palaeocanyon was subsequently intruded, in this case by magmas forming the Upper Streaky Andesites. Although the intrusion obliterated most of the record of sedimentary deposition within the palaeocanyon (thin pebble-bearing sandstones are preserved in places beneath the andesitic sheet), it did at least preserve the original canyon form. Between The Study and Allt-na-Ruigh [NN 176 567] it is possible to demonstrate that the later river incised more deeply than the one before; the later palaeo-canyon cuts through the floor of its predecessor.

Lower Three Sisters Ignimbrite: ignimbrite-forming eruption with graben-like caldera subsidence and downsag

The Three Sisters Ignimbrite Member consists predominantly of two welded (eutaxitic) ignimbrites: the Lower and Upper Three Sisters ignimbrites, up to 150 m and 200 m thick, respectively. These record major rhyodacitic explosive eruptions that were substantially different from the eruptions that produced the Etive rhyolites. They were accompanied by large-scale caldera subsidence and the ignimbrites are intercalated with coarse breccias. The Three Sisters Ignimbrite Member is thickest close to the Pass of Glencoe at Buachaille Etive Beag [NN 19 55] and [NN 20 55] and in the three prominent ridges known as the Three Sisters: Beinn Fhada [NN 17 55], Gearr Aonach [NN 16 55] and Aonach Dubh [NN 15 56]. The type section is at Stob nan Cabar [NN 19 55], at the north-eastern end of the Buachaille Etive Beag massif. The Lower Three Sisters Ignimbrite is underlain by fluvial sedimentary rocks and, in the far south-east of the complex, on Meall a’ Bhùiridh [NN 25 50], it is sharply overlain by thick breccias (White Corries Breccias; Table 2). After emplacement, the Lower Three Sisters Ignimbrite was eroded by a drainage system that in places formed deep palaeocanyons, which, as on Stob nan Cabar [NN 19 55], are partially filled with alluvial-fan deposits referred to as the Queen’s Cairn Conglomerates. On Stob nan Cabar these deposits are overlain by thick breccias and megabreccias, the Lower Queen’s Cairn Breccias. During the last phases of emplacement of the Upper Three Sisters Ignimbrite, and for some time afterwards, breccias were shed locally from fault scarps around the caldera. Thick tuffaceous breccias exposed towards the summit of Stob nan Cabar are the Upper Queen’s Cairn Breccias, and various other breccias and sandstones in the north-west of the volcano complex at and near this stratigraphical level are collectively referred to as the Church Door Buttress Breccias.

The Lower Three Sisters Ignimbrite consists mainly of massive strongly welded (eutaxitic) tuff (Plate 16a), (Plate 16b), but it also includes mesobreccias and, locally, megabreccias (breccias with clasts tens to hundreds of metres across). It filled an elongate volcanotectonic depression that can be traced from Loch an Easain [NN 25 49], in the extreme south-east of the volcano complex, as far as Ossian’s Cave [NN 15 56] in the north-west (Figure 17). The ignimbrite is interpreted as the product of a single sustained eruption and it shows marked lateral changes in thickness and character that suggest it was emplaced over fault scarps, tilted strata, and, in places, deep open fractures (crevasses).

Between Loch an Easain [NN 2518 4960] and the slopes of Meall a’ Bhùiridh above Cam Ghleann [NN 2532 5105], the ignimbrite is approximately 35 m thick and comprises thick but laterally impersistent mesobreccias and megabreccias, intercalated with thin layers of eutaxitic tuff. On the western side of Cam Ghleann, on the slopes of Stob a’ Ghlais Choire [NN 2436 5178], the ignimbrite thickens considerably towards the south-west, and, farther north-westwards, on Buachaille Etive Mòr [NN 22 53] and Buachaille Etive Beag [NN 20 55], it is at least 150 m thick. The south-east face of Stob Dearg [NN 22 54] and [NN 22 53] shows spectacularly the great thickening of the ignimbrite away from the flanks of the Glencoe Graben, across three fault scarps of The Chasm step-fault system (Figure 10)c. Similar thickening relationships are exposed on the eastern slopes of Stob a’ Ghlais Choire in Cam Ghleann [NN 24 51], where the upper parts of fault blocks of the Etive rhyolites in the step-fault system are cut by rotational slumps (Figure 18); (Plate 17). Mesobreccias at the base of the ignimbrite are ponded against palaeoscarps and slumps of The Chasm step-fault system 1 in (Figure 18), and, in places, megablocks of Etive rhyolite up to 40 m across are included. This suggests that fault scarps were initiated in the underlying rhyolites during early stages of the eruption, causing large blocks to be shed. Subsequently, emplacement of eutaxitic tuffs with lithic-breccia layers buried the basal breccias, fault scarps and slumps 2 in (Figure 18), and onlap with fanning dips of the internal layering demonstrates progressive downsag towards the south-west. The ignimbrite is strongly welded across the whole of the downsag area. For example, at Stob nan Cabar on Buachaille Etive Beag [NN 202 556], upper parts of the ignimbrite comprise monotonous, massive, intensely welded eutaxitic tuff over 100 m thick. The pronounced flattening fabrics show extreme pinching and moulding around lithic fragments and crystals, with aspect ratios of up to 1:20 in some fiamme.

The combination of step-faulting and downsag that deformed the caldera floor around Sròn na Creise and Buachaille Etive Mòr gave way to relatively simple downsag farther north-west. For example, where the Lower Three Sisters Ignimbrite is traced from north-east to south-west on the upper slopes of each of the Three Sisters — Beinn Fhada [NN 17 55] towards [NN 16 54], Gearr Aonach [NN 16 55] towards [NN 15 55] and Aonach Dubh [NN 15 56] to [NN 15 55] — it thins gradually and wedges out where a linear hinge separated untilted strata on the flanks of the Glencoe Graben from tilted (downsagged) strata within it (Plate 18). Close to this bounding structure, crevasse-like fissures up to 30 m deep and infilled with eutaxitic tuffs and breccias can be traced in Coire nan Lochan [NN 1547 5573], on the south-east flank of Gearr Aonach [NN 1578 5508] and onto the north-west flank of Beinn Fhada [NN 1627 5484].

The fault block north-west of the Ossian Fault apparently did not subside so much during eruption of the Lower Three Sisters Ignimbrite, if at all. In the steep cliffs at Ossian’s Cave [NN 1537 5618], the ignimbrite, almost 40 m thick, is ponded against a fault scarp (Plate 14a) and farther north-west it is only exposed in a linear crevasse that extends over 1 km along the buttresses of the north-west face of Aonach Dubh. The most accessible exposures of the crevasse fill are near G Buttress [NN 1423 5525], where the crevasse tapers downwards for 40 m into the Etive Rhyolite Member and is partially filled with moderately welded lithic tuffs. The tuffs are themselves incised and overlain by cobble conglomerates, which suggests that a small energetic river exploited the crevasse. This crevasse system north-west of the contemporary subsidence area is considered to represent peripheral extension due to the major caldera downsag towards the south-east.

In the south-east of the Glencoe volcano complex, a crevasse system can be traced for approximately 2.5 km from Meall a’ Bhùiridh [NN 259 497] towards Cam Ghleann [NN 251 516]. It lies along the Northeastern Graben Fault and separates Dalradian metasedimentary rocks to the north-east from strata of the Glencoe Volcanic Formation to the south-west 3 in (Figure 18); the volcanic strata are tilted at 40° to the south-west. On the north-eastern slopes of Meall a’ Bhùiridh, strongly welded eutaxitic tuffs infill a crevasse-like fissure 5 to 20 m wide and at least 50 m deep. Both the inner and outer contacts of the tuffs in the crevasse are sharp and the eutaxitic fabrics have near-vertical or steep north-eastward (outward) dips. Although physical continuity is not preserved, the tuff is indistinguishable from the eutaxitic tuff of the Lower Three Sisters Ignimbrite, with which it is correlated. The crevasse fill is well exposed near the huts at the base of the upper ski tows above White Corries [NN 2558 5078]. Farther north-west, on the upper, north-west facing slopes of Cam Ghleann [NN 2508 5168], the crevasse is 200 m wide and contains megablocks, mesobreccia and tuff. It is approximately 500 m deep and tapers downwards to 20 m wide near the valley bottom [NN 2495 5200], in the subvolcanic basement 3 in (Figure 18). Locally, at deeper levels, mesobreccias with tuff matrix occur in sinuous north-west-trending dykes cutting metasedimentary rocks along strike in the fault zone. Locally, these breccia dykes contain irregularly shaped intrusions of rhyolite and are well exposed on the glacially smoothed lower slopes of Sròn na Creise and Stob a’ Ghlais Choire e.g. [NN 2468 5210] and [NN 2449 5256] (Plate 17).

It seems likely that the tuff-filled and mesobreccia-filled crevasses exposed along the trace of the Northeastern Graben Fault in the south-east of the volcano complex, and the linear breccia dykes occurring at deeper levels, represent part of the vent and conduit system for the Lower Three Sisters Ignimbrite eruption. Repeated collapse of adjacent fault scarps periodically fed blocks into the pyroclastic density current, subsequently to be emplaced as breccias along successive surfaces as the ignimbrite progressively accumulated. This magmatic plumbing is in marked contrast to that of the more centrally located vents that supplied the three Etive rhyolites and initially produced tuff cones due to explosive magma-water interaction there (compare (Figure 16)a and b.

Following the Lower Three Sisters Ignimbrite eruption, catastrophic collapse of fault scarps in the extreme south-east of the volcano complex led to the accumulation of a wedge of breccias, the White Corries Breccias. These breccias sharply overlie the Lower Three Sisters Ignimbrite and contain blocks of Dalradian metasedimentary rocks up to 20 m across. They are well exposed and up to 35 m thick on the north-east flank of Meall a’ Bhùiridh [NN 2557 5064], and they can be traced from the slopes above Loch an Easain [NN 2500 4980] almost as far as the south-eastern slopes of Cam Ghleann [NN 2525 5092].

Queen’s Cairn Conglomerates: localised fluvial incision and alluvial sedimentation

On the crags of Stob nan Cabar at the north-eastern end of Buachaille Etive Beag, several steepsided palaeocanyons are incised up to 20 m deep into the Lower Three Sisters Ignimbrite e.g.[NN 2004 5564]. These are partially filled with poorly sorted boulder and pebble conglomerates, which are composed of rounded Dalradian metasedimentary clasts supported in a matrix of quartzose sandstone and reworked tuff. To the south-west, along the flanks of Buachaille Etive Beag [NN 1982 5495] and [NN 1875 2627], an extensive sedimentary layer, up to 2 m thick, comprises laterally variable conglomerates, sandstones and laminated siltstones. These deposits, collectively named the Queen’s Cairn Conglomerates, together with the underlying erosion surface, reflect incision and sedimentation by a river system that originated beyond the site of the caldera volcano, entered the Glencoe Graben from the north-east, and was susceptible to flash floods. The widespread sedimentary layer is interpreted as recording short-lived braided streams and overbank areas of a small alluvial fan.

Lower Queen’s Cairn Breccias: tectonically induced collapse of scarps along the North-eastern Graben Fault

On the north-east-facing crags of Stob nan Cabar [NN 2006 5556], the Queen’s Cairn Conglomerates are sharply overlain by a 50 m-thick layer of breccias and megabreccias, the Lower Queen’s Cairn Breccias (Figure 17). These are crudely stratified, contain blocks of various metasedimentary rocks, with megablocks up to 30 m across, and they can be traced from Stob nan Cabar for 1 km to the south-west along both the south-east and north-west faces of Buachaille Etive Beag. They terminate sharply at Coire Raineach [NN 1921 5537]. The sparse matrix of the breccias apparently formed by breakage and abrasion of blocks during transport; in places, jigsaw-fitting clasts record the disintegration of blocks during late stages of their transport (Figure 19). Stratigraphically equivalent breccias are exposed on the relatively smooth rock slopes between A’ Chailleach and An t-Innean Mòr [NN 1875 5704], on the north side of the Pass of Glencoe, where they thicken abruptly to 60 m over a short distance towards the south and south-east. These breccias comprise blocks, typically 10 to 20 cm in diameter, almost exclusively derived from the Lower Streaky Andesites and Etive Rhyolite Member, which are exposed in this vicinity along the footwall of the Northeastern Graben Fault. The composition, geometry and position of these breccias, which lie north-west of the trace of the Queen’s Cairn Fault (Figure 17), suggest that they were banked against a south-east facing, degraded scarp of the Queen’s Cairn Fault (Figure 20)a. The Lower Queen’s Cairn Breccias represent catastrophic avalanching that followed alluvial-fan sedimentation and occurred without contemporaneous eruption, which implies that the recorded fault-scarp growth, with collapse, was tectonic in origin.

The Dalness Breccias, which rest unconformably upon the Basal Andesite Sill-complex along the south-western side of the volcano complex, from high on the flanks of Stob Coire Sgreamhach [NN 151 529] to the northern flanks of Beinn Ceitlein [NN 180 506], may include parts that formed during the tectonism that produced the Lower Queen’s Cairn Breccias. Like the latter breccias, the Dalness Breccias are overlain by the Upper Three Sisters Ignimbrite, but their full context is not known. They lie to the south-west of, or bury, the projected trace of the Southwestern Graben Fault (zone) and could relate to movements on strands of the ring-fault, as is registered in the uppermost Church Door Buttress Breccias, which succeed the Upper Three Sisters Ignimbrite (p.72).

Upper Three Sisters Ignimbrite: major ignimbrite- forming eruption with caldera subsidence

The Upper Three Sisters Ignimbrite, which is up to 200 m thick, is the most voluminous preserved pyroclastic unit at Glen Coe. It can be traced for 13 km from Meall a’ Bhùiridh [NN 25 50] in the south-east of the caldera-volcano complex to Stob Coire nam Beith [NN 14 54] in the north-west (Figure 17). It shows marked thickness changes across tilted strata and filled an elongate flexural volcanotectonic depression (downsag) between the major graben faults. Maximum subsidence was in the vicinity of Stob nan Cabar [NN 19 55] and [NN 20 55] on Buachaille Etive Beag. Here, the ignimbrite grades upwards from moderately to strongly welded lithic eutaxitic tuff, and includes several layers of tuffaceous mesobreccia. The strongly welded tuff is overlain by nonwelded bedded tuffs that in turn are overlain locally by laminated tuffs. The deposits characteristically contain a wide variety of lithic fragments, including metasedimentary rocks, Etive rhyolite, peperitic andesite and streaky andesite.

The Upper Three Sisters Ignimbrite is well exposed in the prominent upper crags of each of the Three Sisters — Beinn Fhada, Gearr Aonach and Aonach Dubh — and its three-dimensional geometry is well defined in the intervening Coire Gabhail [NN 16 55] and Coire nan Lochan [NN 15 55]. At the north-eastern end of each ridge, the ignimbrite grades upwards from moderately to strongly welded lithic eutaxitic tuff (e.g. on the north face of Aonach Dubh [NN 153 560]). However, to the south-west the ignimbrite thins rapidly (onlaps) onto underlying strata that were tilted during, and as a consequence of, the ignimbrite-forming eruption. The hinge separating tilted from nontilted strata coincides with the hinge that earlier confined the Lower Three Sisters Ignimbrite; both are well exposed towards the heads of Coire Gabhail (Plate 18) and Coire nan Lochan. The lower, moderately welded part of the Upper Three Sisters Ignimbrite thins south-westwards and wedges out some distance inside the hinge. The overlying strongly welded tuff thins similarly, but overlaps the moderately welded tuff and occurs extensively as a thin veneer over the flanks of the graben as far as the south-western slopes of Bidean nan Bian [NN 1448 5363] in Gleann Fhaolain.

In contrast, north-west of the Ossian Fault, the caldera floor subsided without flexural downsag. On the north-west face of Aonach Dubh [NN 14 55], the ignimbrite, 80 to 100 m thick, grades up from moderately welded to intensely welded tuff, but contains numerous thin intercalated mesobreccia-rich layers. Here, it forms pale outcrops in the upper parts of steep rock buttresses and to the south-west is ponded against the scarp of the Southwestern Graben Fault. These ponding relationships, schematically represented in (Figure 14), are well exposed in Coire nam Beitheach and on the lower slopes of Stob Coire nam Beith [NN 141 547]. The edge of the Upper Three Sisters Ignimbrite, where it pinches out against the steeply tilted Etive rhyolites, can be traced close to the summit of Stob Coire nam Beith [NN 1405 5468]. Away from the structural margin, thick tuffaceous mesobreccias at the base of the ignimbrite locally include megablocks, up to 50 m across, of Etive rhyolite (Figure 14). In places, as beside the Allt Coire nam Beithach [NN 1425 5486], these basal mesobreccias are crudely stratified and underlain by stratified tuff (Figure 21); (Plate 19a), (Plate 19b). Higher in the ignimbrite, mesobreccia lenses extend from Stob Coire nam Beith north-eastwards for up to 300 m across the floor of Coire nam Beitheach and show irregular basal contacts with the ignimbrite, recording loading-related foundering. It is inferred from evidence in overlying breccias (Church Door Buttress Breccias; p.71) that the ignimbrite in the north-western part of the volcano complex was also ponded against the Northeastern Graben Fault.

The Upper Three Sisters Ignimbrite is less well exposed in south-eastern parts of the caldera-volcano complex, but it can be traced more or less continuously around the ridges of Buachaille Etive Mòr [NN 21 54] and Stob a’ Ghlais Choire [NN 23 51]. On the south-eastern slopes of Buachaille Etive Mòr, in Coire Cloiche Finne [NN 21 53], the ignimbrite thickens progressively south-westwards away from the flanks of the Glencoe Graben, as identified in Stob Dearg (Figure 10)c. Here, it comprises moderately welded lithic tuff only, with mesobreccias interstratified at and near the base. In the extreme south-east of the volcano complex, on the slopes of Meall a’ Bhùiridh [NN 2533 5078], the ignimbrite is 40 m thick and includes breccias with blocks up to 2 m across.

It appears that the Upper Three Sisters Ignimbrite was deposited from a single sustained pyroclastic density current. Although no vent is exposed, the distinctive mixed lithic-clast population that occurs throughout the ignimbrite suggests a single source, which would have lain close to where the Queen’s Cairn and Devil’s Staircase faults intersect the Glencoe Graben (Figure 20)b; here, all of the distinctive rock types were available to supply clasts into the pyroclastic current. Layers of meso-breccia within the ignimbrite probably record repeated collapse of near-vent fault scarps during ignimbrite emplacement. The successive ponding of moderately welded tuff and then strongly welded tuff, with overlap onto tilted caldera-floor strata, indicates that during ignimbrite emplacement the pyroclastic density current became less topographically constrained by the downsag that developed.

Seemingly, strongly welded tuffs accumulated only in the vicinity and north-west of the Devil’s Staircase Fault. The laminated tuffs overlying the ignimbrite on Stob nan Cabar record fallout of ash suspended above the pyroclastic density current; they were preserved owing to the closely succeeding deposition of the Upper Queen’s Cairn Breccias.

Upper Queen’s Cairn Breccias fault-scarp degradation along the Northeastern Graben Fault

At Stob nan Cabar on Buachaille Etive Beag, the Upper Three Sisters Ignimbrite is overlain by thick-bedded tuffaceous breccias, named the Upper Queen’s Cairn Breccias (Figure 23)." data-name="images/P988028.jpg">(Figure 22). These have the form of a wedge that is thickest, at 60 m, in the north-east and thins south-westwards for 2.5 km. The relationships are well exposed from the summit of Stob nan Cabar [NN 2000 5546] south-westwards along Buachaille Etive Beag through Coire Raineach [NN 1918 5507] towards Coire Dubh [NN 1836 5455], where the breccias dip gently to the south-west. As with the Lower Queen’s Cairn Breccias, the clasts, which are mostly of psammite and streaky andesite, suggest that the talus was derived mainly from collapse along the footwall of the Northeastern Graben Fault where this extended between the Queen’s Cairn and Devil’s Staircase faults. Unlike the Lower Queen’s Cairn Breccias, however, pyroclastic debris in the matrix of these breccias indicates that the collapse accompanied or was closely followed by the final stages of emplacement of the Upper Three Sisters Ignimbrite; this suggests that the faulting in this instance was probably volcanotectonic in origin.

In the north-west of the Glencoe Caldera-volcano Complex, the Church Door Buttress Breccias (Figure 23)." data-name="images/P988028.jpg">(Figure 22); (Figure 23) comprise various layers of breccia, sandstone and tuff. These are exposed more or less continuously from their type section, around Church Door Buttress on the northern flank of Bidean nam Bian [NN 142 544], to Coire nan Lochan [NN 15 55]. They are also exposed on the south-west slopes of Bidean nam Bian in Gleann Fhaolain [NN 14 53]. The complex stacking geometry of the breccia layers (Figure 23) records successive failures of different fault scarps north-west of the Queen’s Cairn Fault.

On the steep crags beneath the summit of Stob Coire nam Beith [NN 1410 5470], 12 m-thick stratified tuffaceous mesobreccias are intercalated with, and show loading into the top of, the Upper Three Sisters Ignimbrite (Figure 14). These mesobreccias, with clasts mainly of Basal Andesite Sill-complex rocks and Etive rhyolites, record an increment of collapse along the scarp of the Southwestern Graben Fault during the late stages of ignimbrite emplacement. Eutaxitic fabrics that are deflected around lithic blocks show that the loading occurred while the ignimbrite was sufficiently hot (more than 550°–600°C) for the glassy constituents to deform in a ductile manner. The mesobreccias thin gradually north-eastwards across the floor of Coire nam Beitheach (Figure 23) to beneath the scree-covered south-western slopes of Stob Coire nan Lochan [NN 1442 5493]. In Coire nan Lochan [NN 1508 5533] the Church Door Buttress Breccias overlying the ignimbrite are stratified and dominated by clasts of streaky andesite, and here the top of the underlying ignimbrite shows evidence of sedimentary reworking. The streaky andesite clasts are lithologically indistinuishable from the Lower Streaky Andesites, which crop out widely on the north-eastern flank of the Glencoe Graben (e.g. between Meall Dearg [NN 162 583] and Am Bodach [NN 169 580]).

Across the north-west face of Aonach Dubh [NN 14 55], the breccias dominated by streaky andesite thin south-westwards and overlap another breccia that thins in the opposite direction (Figure 23). This lower breccia thins north-eastwards from near G Buttress [NN 1437 5520] to B Buttress [NN 1475 5566] and rests with a sharp, slightly erosional contact on the Upper Three Sisters Ignimbrite. Like the tuffaceous breccias deposited during emplacement of the ignimbrite (see above), it is dominated by clasts from the Basal Andesite Sill-complex and Etive rhyolites. The two overlapping breccia layers mark successive collapses following ignimbrite deposition, first along the Southwestern Graben Fault and later on the Northeastern Graben Fault. From their long run-out distances they are inferred to record mainly catastrophic avalanching, as opposed to gradual accumulation of talus, but their upper and distal parts show some evidence of aqueous reworking and grade laterally into sandstones with scour-and-fill cross-stratification.

An uppermost breccia layer is well exposed at the head of Coire nam Beitheach. It is mainly non-stratified and is composed predominantly of angular blocks, 10 to 30 cm in diameter, which are derived from the Basal Andesite Sill-complex and set in a sparse matrix of comminuted andesite. It forms the 50 m-high Church Door Buttress [NN 1416 5437], where it overlies and effectively buries the Southwestern Graben Fault against which previous units were banked. The breccia layer thins gradually north-eastwards for approximately 1 km across Coire nam Beitheach and onto the south-western slopes of Stob Coire nan Lochan [NN 1440 5463], and it also thins south-eastwards across the southern slopes of Bidean nam Bian into Gleann Fhaolain [NN 1452 5364] (Figure 23). The significance of this breccia layer is that, unlike those beneath it, it marks large-scale failure of one or more fault scarps beyond the former south-west limit of the Glencoe Graben. The scarp or scarps that exposed the Basal Andesite Sill-complex may have been in the vicinity of, or beyond, the ring-fault mapped in this vicinity. The south-westward shift of the limit of subsidence that is recorded appears to anticipate later developments, wherein the former Southwestern Graben Fault became deeply buried within successive volcanotectonic centres of deposition (see p.82).

Upper Streaky Andesites: further phases of andesite intrusion and extrusion, with vent formation and fluvial incision

Emplacement of the Three Sisters ignimbrites and overlying collapse breccias was followed by a phase of andesitic magmatism represented by the Upper Streaky Andesites (Table 2). The andesites are extensively brecciated and contain bleb-like rhyolitic patches, locally forming a pervasive millimetre- to centimetre-scale streaky texture, which records magma mingling. Within the Glencoe Graben, the sheets occur as transgressive and locally bifurcating sills at shallow levels in or beneath the Three Sisters Ignimbrite Member, and variations in thickness of the sills mimic those of the ignimbrites. For example, on the south-east face of Buachaille Etive Mòr, on the lower slopes of Coire Cloiche Finne [NN 21 53], thick andesite sheets intrude lower parts of the Lower Three Sisters Ignimbrite and terminate against fault scarps that are part of The Chasm step-fault system [NN 2254 5326]. Similar relationships are evident on the north-west face of Buachaille Etive Mòr at Creag a’ Bhancair [NN 2154 5504] and in Coire na Tulaich [NN 217 548], where andesitic sills densely intrude the Lower Three Sisters Ignimbrite and abut the same fault system. No correlative andesites are observed north-east of this fault system on Stob Dearg [NN 22 54]. In the north-west of the volcano complex, on Beinn Fhada [NN 17 55] and Gearr Aonach [NN 16 55], the andesitic intrusions occur at higher levels and cut the Upper Three Sisters Ignimbrite. Here, multiple and bifurcating sills thin rapidly and pinch out to the south-west in much the same manner as the enclosing ignimbrites (Plate 18). Where the cumulative thickness of andesitic sills is great, there is likely to have been some uplift of the caldera floor together with volcanotectonic subsidence, as seemed likely to have occurred with emplacement of sills of the Lower Streaky Andesites (see p.57).

Two conduits or infilled vents of the Upper Streaky Andesites have been identified. The more impressive example is well exposed on the north-eastern shoulder of Aonach Dubh [NN 157 561], less than 200 m east of Ossian’s Cave (Plates 14a); (Plate 15b). Here, a near-cylindrical body of andesite, with steeply inclined streaky fabrics, cuts upwards through the Etive Rhyolite Member and into the Lower Three Sisters Ignimbrite. In places, thin veins of the streaky andesite interfinger with the latter ignimbrite and show that the intrusion caused reheating sufficient to allow ductile remobilisation of the tuff. Locally within the cross-cutting streaky andesite there are irregular bodies of agglomerate. These are rich in incorporated lithic fragments (Plate 20) and indicate explosive activity, possibly involving a vent at the contemporary surface. Another, larger cross-cutting intrusion occurs where the Upper Streaky Andesites crop out along the plane of the Queen’s Cairn Fault. On the north side of the Pass of Glencoe, the contact between the streaky andesites and the Basal Andesite Sill-complex lies along the Allt Coire Meannarclach [NN 1905 5703]. From here the contact, against the Upper Etive Rhyolite, can be traced south-westwards through low ground across the road close to the large beehive-shaped cairn [NN 1872 5634], and some way into the Lairig Eilde [NN 1815 5550]. Here, and to the west, numerous ramifying sills of the andesite obscure the detail of the Three Sisters Ignimbrite Member. The intrusions record both mingling of magmas and emplacement associated with further volcanotectonic subsidence. The cross-cutting bodies interpreted as conduits or infilled vents are located close to where cross-graben faults intersect the Glencoe Graben axis, like the vents and associated tuff cones of the Etive rhyolites.

High on the north side of the Pass of Glencoe, streaky andesites, similar to those that are intrusive, are more than 200 m thick and overlie an unconformity that cuts intrusions of the Upper Streaky Andesites. Along the unconformity, which is poorly exposed on the north-west slope of Coire Mhorair [NN 1860 5795], there is a deep palaeocanyon partially filled with conglomerate and sandstone. Clough et al. (1909) recorded sedimentary intercalations within the overlying andesite succession. Although the stratigraphical relationships are somewhat obscure, it seems that fluvial incision and sedimentation occurred after emplacement of some of the Upper Streaky Andesites, and that andesite lavas were subsequently extruded across the sedimentary deposits while shallow-level sills were intruded within them.

Glas Choire Sandstone Member fluvial incision followed by alluvial-fan and lacustrine sedimentation

In the south-eastern part of the caldera-volcano complex, a canyon was cut into the Upper Three Sisters Ignimbrite; there is no evidence there for the extrusion of lavas of the Upper Streaky Andesites, although related intrusions occur at shallow depth and lavas may have been eroded away. At Glas Choire in Cam Ghleann [NN 241 514], the canyon is 300 m wide and up to 40 m deep (Figure 24); it trends north-west, as shown by counterpart exposures on the north-west flank of Stob a’ Ghlais Choire [NN 237 518], and is exposed in a slightly oblique, north–south section. North-west of the Glen Etive Fault, the erosion surface is less deeply incised where it is exposed at Feadan Ban [NN 21 54], near the ridge of Buachaille Etive Mòr, and also on the slopes of Beinn Fhada [NN 17 55]. This change marks the opening of the palaeocanyon north-westwards onto a broad plain.

Deposits assigned to the Glas Choire Sandstone Member (Table 2) fill the palaeocanyon and extend laterally beyond its immediate confines; to the north-west they form an extensive layer on the broad fluvially reworked surface (Figure 23)." data-name="images/P988028.jpg">(Figure 22); (Figure 24). The sedimentary architecture of the palaeocanyon fill is well exposed at Glas Choire [NN 2416 5143], where undulose-stratified sandstones (Plate 21a) enclose poorly sorted conglomerates resting within broad (several tens of metres) asymmetrical scours. The conglomerates contain angular blocks as well as rounded boulders, both up to 1 m in diameter; numerous boulders are granitic and from the Rannoch Moor Pluton and large clasts of metamorphic basement are abundant, all in a matrix of pebbles, granules and rounded sand grains (Plate 21b). The lenticular conglomeratic units represent filled channels, and the deposits in the palaeocanyon contain at least four discrete groups of them, stacked with local overlap (Figure 24); they record the migration of successive stream channels during rapid aggradation and possibly formed during waxing and waning of only a few flash floods. Planar-bedded and laminated fine-grained sandstones and siltstones (Plate 21c), up to 10 m thick, overlie the main scour-and-fill succession and extend laterally beyond it. Although these are widely intruded and disrupted by sills related to the overlying Bidean nam Bian andesites, they clearly record deposition from aqueous suspension and from turbidity currents, and hence indicate standing water. The palaeocanyon and the sedimentary rocks within it record a switch from fluvial erosion to fluvial deposition, and the overlying strata record ensuing formation of a lake.

At Feadan Ban [NN 2108 5400], north-west of the River Etive, extensive parallel-stratified sheet-like pebbly sandstones, up to 25 m thick, are interpreted as alluvial plain deposits, downstream from the fluvial palaeocanyon. Farther north-west, distal correlative sandstones and siltstones are restricted and onlap against the earlier collapse-breccia wedges (Figure 23)." data-name="images/P988028.jpg">(Figure 22); (Figure 23). For example, in the vicinity of Coire Dubh on Buachaille Etive Beag [NN 1820 5446], planar-bedded sandstones and siltstones thin with onlap onto the south-western edge of the Upper Queen’s Cairn Breccias, and, in Coire nan Lochan [NN 1550 5510], similar deposits thin onto the south-eastern edge of the Church Door Buttress Breccias. Farther north-west, on Aonach Dubh [NN 14 55], sandstones and siltstones, up to 5 m thick, accumulated in the topographical depression on the surface of the Church Door Buttress Breccias where the two contrasting breccia layers overlap (Figure 23). Some of these sandstones show normal grading, from medium- or fine-grained sand to silt and, where they rest on laminated siltstones, small load casts are common at their bases. Preferential recent erosion of the siltstones has formed the topographical feature on Aonach Dubh known as The Amphitheatre [NN 1456 5545].

It is inferred that a river draining from the east, with some catchment on the Rannoch Moor Pluton, initially caused substantial erosion in the east and then became dammed by changed topography so as to form a caldera lake. Damming of the river led to sedimentation across the broad eroded plain and back up into the topographical confines of the Glas Choire palaeocanyon. The planar-bedded and laminated strata that are prevalent towards the west and also form the upper part of the sequence in the east record lacustrine sedimentation in which the turbidity currents probably arose from periodic fluvial floods. The depth of the caldera lake is unknown; the existence of standing water and turbidite sedimentation at the site of the former palaeocanyon suggests development of a considerable topographical barrier, or dam, in the north-west. The lake may have been at least many tens of metres deep, possibly much deeper.

Bidean nam Bian Andesite Member: andesite and dacite extrusion with volcanotectonic ponding

Thick sheets of andesite and dacite overlie the Glas Choire sedimentary rocks and are collectively named the Bidean nam Bian Andesite Member (Table 2). They represent effusive eruptions of unusually large volume for magma of intermediate composition in an intracaldera setting. In the west of the volcano complex, a massive to extensively columnar jointed sheet forms the spectacular cliffs and pinnacles of Stob Coire nan Lochan (Plate 22a), (Plate 22b) and the peaks of Bidean nam Bian [NN 14 54], from which the member takes its name. These are the rocks from which Thirlwall (1988) obtained an age of 421 ± 4 million years. In the precipitous north-east face of Stob Coire nan Lochan [NN 148 550] and in the sheer walls around Central Gully and Diamond Buttress on Bidean nam Bian [NN 144 544], columnar jointing extends for more than 200 m, virtually from the top to the bottom, and there is no evidence of any internal flow-unit boundary. The unit is thickest, at 450 m, near the head of Coire Gabhail [NN 15 54], where it overlies the Southwestern Graben Fault, and from here it thins progressively for up to 4 km to the north-east, as seen along the ridges of Beinn Fhada [NN 16 54] and Buachaille Etive Beag [NN 18 54] and [NN 192 548]. Between Feadan Ban [NN 210 539] and the flanks of Stob na Doire [NN 207 533], on Buachaille Etive Mòr, the Bidean nam Bian member thickens dramatically south-westwards as the dip of the underlying pebbly sandstones steepens from a few degrees south-west to almost vertical. The unit thins to the south-east through Stob Coire Sgreamhach [NN 15 53] and Coire nan Easan [NN 160 518], becoming considerably thinner in the south-west slopes of Sròn an Fhorsair [NN 17 51], near Dalness.

South-east of the River Etive, Bidean nam Bian andesite and dacite, locally more than 200 m thick, forms the upper cliffs and summits of Stob a’ Ghlais Choire, Creise [NN 24 51] (Plate 17) and Meall a’ Bhùiridh [NN 25 50]. Here the columnar jointing is not as obviously continuous as elsewhere, but the rocks are monotonously uniform, dark blue and fine grained, and there is no sign of any internal flow-unit boundary. Around the high flanks of Stob a’ Ghlais Choire [NN 2410 5176], one or more sills of the same rock type lie beneath the main unit, within the upper fine-grained sandstones and siltstones of the Glas Choire Sandstone Member (see previous section). These sills show pillowed and fragmental margins indicative of intrusion when the sediments were wet and unlithified. Elsewhere there are minor occurrences of autobreccia and peperite at basal contacts of the unit, as well as small-scale loading-related deformation of the underlying Glas Choire deposits, such as at Coire Dubh [NN 179 541]. The upper contact of the member is preserved only in the south-west of the volcano complex, where phreatomagmatic tuffs and ignimbrites lie unconformably upon it (see p.81). Upper parts of the unit are autobrecciated locally and in places the breccias show crude stratification, which may reflect some sedimentary reworking. These features of the upper contact suggest that the andesites and dacites were extrusively emplaced, as lavas, while the features of the basal contact indicate emplacement onto wet caldera-lake-floor sediments.

The Bidean nam Bian andesites and dacites are remarkably massive and homogeneous throughout their thickness, across the entire caldera volcano complex, with no well-defined internal contacts such as pronounced discontinuities of jointing, erosion or weathering surfaces, or sedimentary intercalations. In the north-west of the volcano complex, the andesites are distinctly porphyritic with well-formed (euhedral) phenocrysts of plagioclase and hornblende, typically 5 mm across, and plane-parallel flow lamination is common. Around the summit of Stob Coire nan Lochan [NN 148 548] this lamination occurs in domains with contrasting orientation, as if laminated material had been broken up in a brittle fashion and had then foundered into fluid. The andesites and dacites in the east are far less porphyritic than in the west, which may be interpreted as recording coexistence of two eruption centres, possibly tapping different parts of the same magma chamber.

Although the Bidean nam Bian andesites and dacites have many characteristics of lavas, the thick units lack features such as rubbly flow-unit boundaries and intercalated sedimentary detritus that would indicate accumulation from a series of eruptions over a long period of time. Instead, it seems that the andesites and dacites were emplaced in a short interval to form one or more deep lava ponds, and the implied rapid effusion rates and thick accumulation are likely to have been linked to volcanotectonic subsidence. Nevertheless, from the evidence for a pre-existing caldera lake (in uppermost Glas Choire sedimentary strata) it appears that there already was some form of basin at the time of onset of the eruption or eruptions; conceivably the first andesitic and dacitic magmas flowed into a lake. The limited preservation of overlying strata makes it difficult to establish details of the possible volcanotectonic subsidence, but the thickening of the unit to the south-west (up to 450 m) is most simply reconciled with asymmetrical or trapdoor-like subsidence primarily on volcanotectonic faults lying to the south-west of the Glencoe Graben, possibly involving the nearby strand of the ring-fault and the fault in Gleann Chàrnan see (Figure 7). The remarkable thickening of the unit along with steepening of substrate, on Buachaille Etive Mòr, is consistent with hinge-like subsidence towards the south-west. The original extent of the Bidean nam Bian Member across, and possibly outside, the Glencoe Caldera-volcano Complex is unknown, but a conservative minimum erupted magma volume of about 12 km³ is estimated, which is unusually large for an intracaldera outpouring of andesite and dacite. The An t-Sròn composite intrusion [NN 13 54] has been interpreted as the root of an early andesitic central volcano (Bussell, 1979; Garnham, 1988; see pp.92–93), and its position in relation to the volcanotectonic faults makes it a likely source for the Bidean nam Bian member. The intrusion has branches that invaded the volcanotectonic faults, both along and south-west of the ring-fault; these include diorite and tonalite, which are the coarse-grained (intrusive) equivalents of andesite and dacite.

Coir Eilde Tuffs and Lower Dalness Ignimbrite: renewed phreatomagmatism and large-volume rhyolitic explosive eruption leading to caldera subsidence

The Dalness Ignimbrite Member comprises two welded (eutaxitic) tuffs, the Lower and Upper Dalness ignimbrites, which are at least 200 m and 100 m thick respectively. Each ignimbrite is underlain by extensive phreatomagmatic tuffs, the Coir Eilde Tuffs and the Coire nan Easan Tuffs, respectively. The sequence records further explosive eruptions that were associated with large-scale caldera subsidence. Unlike the previous major pyroclastic units, which have their thickest deposits and maximum subsidence in the vicinity of Buachaille Etive Mòr and the Three Sisters, on the Glencoe Graben axis, the Dalness ignimbrites are thickest (and have their type sections) towards Dalness in the south-west of the caldera-volcano complex. The ignimbrites are, however, not extensively preserved and their full original form is unknown.

The Coir Eilde Tuffs occur widely as a variably preserved layer, 1 to 2 m thick, of finely laminated porcellanous tuff with abundant accretionary lapilli, overlying the Bidean nam Bian Andesite

Member in the south-west of the volcano complex. They record phreatomagmatism similar to that which preceded eruption of each of the Etive rhyolites earlier in the volcanic history (Table 2).

The tuffs are best exposed at the type section in Coir Eilde, which is in the vicinity of the Southwestern Graben Fault, at the head of the Lairig Eilde. Here, north of the stream [NN 161 537], the tuff thickens to 25 m between palaeofault scarps that define a 200 m-wide basin in which there is evidence of subaqueous deposition. South of the stream [NN 1632 5347], the tuff thickens to 10 m.

The Coir Eilde Tuffs record a return to phreatomagmatic activity, but the location of the vent is uncertain. The extensive thin tuff layer records fallout of fine ash from dilute ash clouds, and the thickening of the sequence southwards to 10 m possibly represents the outer edge of a tuff cone. The fault-bounded aqueous basin probably acted to trap ash from pyroclastic currents during the eruption. The availability of water and evidence of contemporaneous faulting suggest that the Coir Eilde phreatomagmatism may have been influenced by movement or movements on the Southwestern Graben Fault and possibly also one or more intersecting cross-graben faults. The temporal shift of the magmatic plumbing towards the south or south-west, relative to the vents of the Etive rhyolites, would be consistent with the similar shift in the locus of maximum volcanotectonic subsidence evident from the Bidean nam Bian andesites (see p.78) and Dalness ignimbrites (see below).

The Lower Dalness Ignimbrite comprises strongly welded crystal-rich eutaxitic tuff and breccia, and it is ponded against tilted strata and within crevasse-like fissures. At its type section near Coir Eilde [NN 159 538], the ignimbrite is about 200 m thick and it appears to thicken south-eastwards towards Dalness and Stob na Bròige [NN 18 52], where, however, it is substantially altered owing to emplacement of the underlying Clach Leathad Pluton (see p.99); here most rocks are patchily or predominantly white weathering, recording remobilisation of silica and some recrystallisation. To the north-east along the ridge of Beinn Fhada [NN 16 54] and north-east along the upper slopes of Buachaille Etive Beag [NN 183 540], the ignimbrite thins with onlap onto tilted strata of the underlying Coir Eilde Tuffs. It is not possible to determine any original southern limit of the ignimbrite, because it is cut by the ring-fault system, on Beinn Ceitlein [NN 18 50], and is obliterated by the Clach Leathad intrusion.

North-eastwards from the Bealach Dearg [NN 151 537], along the ridge of Beinn Fhada, blocks derived from the Etive Rhyolite Member dominate mesobreccias at the base of the ignimbrite. Elsewhere at the base, blocks of plagioclase- and hornblende-phyric andesite similar to the Bidean nam Bian andesites are dominant. In Coir Eilde, the base of the ignimbrite shows loading into the thick sub-aqueous Coir Eilde Tuffs, and in places it includes clasts of the tuff. The upper part of the ignimbrite, which is up to 180 m thick, is massive and strongly welded crystal-rich eutaxitic tuff, containing whole and fragmented plagioclase phenocrysts with some phenocrysts of quartz, biotite and hornblende.

Around the northern periphery where the ignimbrite thins against tilted substrata, the deposit infills irregular and steep-sided crevasses, up to 150 m deep, which opened in the Coir Eilde Tuffs and Bidean nam Bian Andesite Member. The crevasses are well displayed along Beinn Fhada; towards the head of Coire Gabhail, near the Bealach Dearg [NN 1542 5385], the main crevasse lies on the north-west side of the ridge, but, farther north-east, it is deflected round to the east, crosses the ridge of Beinn Fhada [NN 1656 5444], and is exposed in the Lairig Eilde [NN 1676 5452]. Originally, the fissures probably formed a more or less continuous, somewhat arcuate system from near or beyond the Southwestern Graben Fault generally north-eastwards towards the Glencoe Graben axis. It is possible that other related fissures occur farther to the south-east, on Buachaille Etive Beag, for example on the north and north-west flanks of Stob Dubh [NN 177 538], but they are not so clearly defined.

Although the Lower Dalness Ignimbrite is incompletely preserved, its distribution certainly marks a volcanotectonic centre of deposition south-west of those defined by the earlier ignimbrites and extending across the former bounding structure, the Southwestern Graben Fault. The crevasse-like fissure system represents extension bounding a major downsag, the centre of which lay to the south, near Dalness, but there is no evidence that the ignimbrite was erupted from the fissures. The abundant blocks of Etive rhyolite in the mesobreccias suggest that the rhyolite was exposed in the vicinity of the vent or vents, or along faults that were active during the eruption and associated subsidence.

Coire nan Easan Tuffs and Upper Dalness Ignimbrite: further phreatomagmatism and explosive rhyolitic eruption

A layer of accretionary lapilli-bearing phreatomagmatic tuffs, the Coire nan Easan Tuffs, overlies the Lower Dalness Ignimbrite. Close to the peak of Stob Coire Sgreamhach [NN 1557 5356], bedded tuffaceous sandstone up to 10 m thick intervenes between the phreatomagmatic tuffs and the ignimbrite, but pinches out to the south-east, where the tuffs are finely laminated and porcellanous. These strata constitute the Group 6 ‘Shales and grits’ of earlier workers (Table 2). Towards the top, the phreatomagmatic tuffs become interstratified with ignimbrite layers, and this sequence is overlain by stratified and then massive tuffs, which constitute the Upper Dalness Ignimbrite. This transition is well exposed on the south-facing slopes between Coire nan Easan and Dalness [NN 167 525]. In general, the Upper Dalness Ignimbrite is strongly welded and distinctly crystal-rich, similar to the Lower Dalness Ignimbrite. It is at least 100 m thick, but its original maximum thickness and extent are unknown. On Stob Dubh [NN 17 53] patchy to pervasive silicification and recrystallisation due to emplacement of the Clach Leathad Pluton inhibits discrimination of the Upper and Lower Dalness ignimbrites. Plagioclase- and hornblende-phyric andesites, texturally similar to the Bidean nam Bian andesites, intrude the Upper Dalness Ignimbrite locally with irregular and transgressive contacts; these too are altered in the vicinity of the pluton.

Chapter 5 Ring-fault system and fault-intrusions

The Glencoe volcano became internationally renowned for the striking occurrence of its thick succession of lavas and volcaniclastic deposits deep within metamorphic basement, with contacts along steep bounding faults that were described as a ‘ring-fault’. It was considered by Clough et al. (1909), Bailey and Maufe (1916), Bailey (1960) and Roberts (1974) that a block of crust had subsided coherently along the ring-fault, and that the ‘fault-intrusions’ along the fault trace represent the complementary ascent of magma from an underlying chamber (Figure 5). However, Moore and Kokelaar (1997, 1998) showed that substantial displacements on the ring-fault system occurred only after formation of the lower parts of the volcanic succession. None of the five major silicic eruptions recorded by the Etive rhyolites and Three Sisters ignimbrites caused large-scale subsidence on the ring-fault, although the last one seems to have activated a section of it in the west, as is registered in the andesitic Church Door Buttress Breccias that overlie the Southwestern Graben Fault (Figure 23)." data-name="images/P988028.jpg">(Figure 22); (Figure 23). Similarly, the fault-intrusions have no known extrusive counterparts, except possibly the Bidean nam Bian Andesite Member, which could have been erupted via conduits now represented by the An t-Sròn composite intrusion (see pp.79; 92). Subsidence amounting to 700 m is known to have occurred along parts of the ring-fault system, but the timing of this and the associated intrusive activity remain poorly understood. In this book the descriptions of the ring-fault and associated fault-intrusions are derived substantially from earlier work (in particular Clough et al., 1909; Bailey and Maufe, 1916; Bailey, 1960; Roberts, 1966b; Taubeneck, 1967; Garnham, 1988), although new observations are added and the whole is placed in the modern context. For full petrographical details the reader should refer to Bailey (1960). The term ‘ring-fault’ is preserved here, for simplicity, but the previous connotations of a fault system continuously linked around a coherent block of crust should not be assumed.

Ring-fault system

The trace of the ring-fault system broadly outlines an incomplete and truncated ellipse, 14 km by 8 km (Figure 25), with major strands to the south of this; it is somewhat polygonal and not a continuous annular structure. Previous workers have emphasised the linkage of exposed fault strands to form a continuous dislocation (‘ring’), which has formed the basis for the original piston-subsidence model (Clough et al., 1909; see (Figure 5)a and its successor, the asymmetrical or ‘trapdoor’ model (Roberts, 1974; see (Figure 5)b, but it is important to recognise now that the ‘ring’ is only a part of the system and that there was a southwards shift of the locus of subsidence such that south-western parts of the ‘ring’ did not delimit but lay in the floor of the later structure. In a modern review, Lipman (2000, p.656) expressed the opinion that the overall structure at Glen Coe ‘seems reasonably interpreted as dominated by subsidence along the bounding ring faults, with smaller scale breakup of the floor largely preceding development of the ring-fault subsidence’, but this is contrary to the observation that the maximum downthrow that can be determined on any strand of the ring-fault is equal to the downthrow that is known to have occurred within the Glencoe Graben. It is now too simplistic to view the caldera-volcano development as having been dominated by ring-fault controlled subsidence, and there is much to be learned from an objective reappraisal. The story that emerges involves migration and possible enlargement of successive caldera depocentres, with the various bounding rectilinear structures strongly influenced by a regional tectonic grain and active tectonism.

Long sections of the ring-fault system are poorly exposed, particularly in the north-east, and other parts are obliterated by intrusions, especially in the south by the Clach Leathad Pluton (Figure 25). Early Geological Survey geologists argued, with good reason, that in poorly exposed ground the existence of the fault could be inferred from the intrusions lying along the interpolated trace of the ring, even where the igneous contacts are not exposed. However, many well-exposed parts of the ring-fault system are distinctly polygonal rather than smoothly rounded, with sharply angular intersections of north-east- and north-west-trending fault planes, and with multiple subparallel branches and common bifurcations. There is abundant evidence that different fault strands were active, and were intruded by magma, at different times, and that some strands were reactivated. Thus in the poorly exposed ground in the north-east, between Stob Mhic Mhartuin and Cam Ghleann (Plate 23), the inferred trace of the ring-fault is probably a simplification. Tantalising glimpses here of the metamorphic basement within the mapped ring-fault show it to be considerably broken and locally to include minor intrusions interpreted as of the Early Fault-intrusion suite (Clough et al., 1909), suggesting that the fault system here is far more complicated than has been depicted.

The ring-fault is particularly well exposed in the accessible crag of Stob Mhic Mhartuin [NN 208 575], where there are two north-west-striking fault strands. The inner one, the ‘Main Fault’, is younger than the outer, ‘Early Fault’ (Clough et al., 1909), and both can be traced from Stob Mhic Mhartuin north-westwards as far as Coire Mhorair [NN 187 585]. These fault strands are planar, dip steeply outwards (towards the north-north-east) and include some right-angular fault-jogs. Farther north-west, Bailey (1934, 1960, p.157) located another outer fault strand, on the slopes of Meall Dearg [NN 165 589], but it is uncertain whether this might have been continuous with the outer fault at Stob Mhic Mhartuin, because it is cut by an intrusion. This strand, however, extends north-westwards for at least 1.5 km beyond the sharp angular change of strike of the inner fault, which forms the northernmost corner of the (closed) ring-fault system. This ‘corner’, in Coire Cam [NN 1583 5872], is the intersection of two planar faults, one striking nearly east–west and dipping north at 50° and the other striking north-east–south-west and dipping nearly vertically. In the vicinity of the classic view of the ring-fault (Plate 24), where it forms the deep gully known as The Chasm of An t-Sròn [NN 136 557] and crosses the An t-Sròn ridge towards Stob Coire nam Beith [NN 1355 5520], in the far west of the volcano complex, the fault has a vertical to steep inward dip (about 86°) through a vertical distance of approximately 1 km. However, as noted by Bailey (1960, p.77), this part of the ring-fault does not curve smoothly to the north-east across the floor of Glen Coe, but takes an angular deflection where the major downthrow was transferred from one steep fracture to another that intersected it. (Plate 24) shows the gully that marks the continuation of one fracture where it diverges from the ring-fault on the flank of An t-Sròn; a similar gully on the north side of the valley marks the continuation of the fracture that projects into the Chasm of An t-Sròn.

Inner and outer strands of the fault system are exposed in the far east of the volcano complex, in Coire an Easain [NN 257 490], but here the inner strand is the earlier one. These strands dip outwards and they both show abrupt, almost right-angled, changes in strike from north-east to north-west (see Bailey and Maufe, 1916, fig. 23; (Figure 25). Viewed from the summit region of Meall a’ Bhùiridh [NN 250 503], it is clear that the inner contact of the early fault-intrusion here is planar for more than 1 km across Coire an Easain [NN 25 49], and that it dips steeply outwards (towards the south-east) and intersects the Northeastern Graben Fault almost at a right angle. On the northern and western flanks of Beinn Ceitlein, south-east of Dalness [NN 180 503] and [NN 175 498], there are several subparallel fault strands, some of which are linked by faults almost at right angles (Clough et al., 1909). The fault that is part of the system in Gleann Fhaolain, between Bidean nam Bian [NN 139 540] and Dalness [NN 165 509], dips inwards (north-eastwards) at no less than 80° and is more or less straight and planar, while approximately 2 km to the south-west an outer, divergent strand, in Gleann Chàrnan [NN 13 51] (Figure 25), is also straight and is practically vertical. This latter structure displaces units in the Dalradian basement, throwing the Leven Schist Formation down to the east-north-east by more than 500 m, against the older Glen Coe Quartzite Member (Clough et al., 1909). This structure was considered by early workers also to be a boundary fault of the ‘cauldron subsidence’, linked with the faults on Beinn Ceitlein (Bailey, 1960, p.132, fig. 19). It has fault-intrusions along it and extends north-north-westwards into the An t-Sròn component of the ring-intrusion [NN 12 53]; it was probably active as a volcanotectonic fault at least by the time of emplacement of the Bidean nam Bian Andesite Member (see p.93).

The dips of the fault planes that constitute the ring-fault system evidently range from 50° outwards to no less than 80° inwards (Figure 25). Along northern and eastern sections the faults dip outwards, and elsewhere they are near vertical or dip steeply inwards (Clough et al., 1909; Bailey, 1960). There is no evidence for an upwards-flaring cone-fracture system as discussed by Taubeneck (1967) and Roberts (1974; see (Figure 5)b (see discussion on pp.14–15). Several previous authors interpreted the steepening of dips in volcanic strata near the ring-fault as evidence of shortening due to subsidence along inward-dipping faults (e.g. Taubeneck, 1967), and hence used this to confirm an upward-flaring ring-fault geometry. However, most of the inward dips of the strata at Glen Coe were caused by incremental (extensional) downsag before development of the ring-fault. The report by Clough et al. (1909) and Taubeneck (1967) of inverted strata in the north-west of the volcano complex, at Stob Coire nam Beith [NN 138 548], was a misinterpretation. The contacts concerned, with Etive rhyolite apparently overlain by the Basal Andesite Sill-complex, record top- inwards rotation, by at least 35°, of originally near-vertical fault scarps of the Southwestern Graben Fault see (Figure 14), rather than inversion with some 120º rotation of an originally horizontal, normal contact. Outward dips of strata immediately adjacent to the fault in the south-west of the volcano complex, on the south-western slopes of Stob Coire Sgreamhach [NN 15 53] and Coire nan Easan [NN 15 52], probably register rotations due to listric faulting down to the north-east, or, alternatively, relatively late downsag towards the south-west.

Displacements on the ring-fault system are poorly constrained. The youngest part of the preserved volcanic succession, the Dalness Ignimbrite Member, is cut by the ring-fault system only in one area, approximately 1.5 km east-south-east of Dalness [NN 180 503], where the downthrow appears to be at least 600 m. From this point clockwise around the fault system towards the north at Coire Cam [NN 159 585], the metasedimentary rocks on the outside are mostly juxtaposed against the Basal Andesite Sill-complex and its metasedimentary substrate on the inside. In the west, a minimum downthrow of about 500 m can be deduced from the preservation of the greatest known thickness of the sill stack. However, 600 m or more (Clough et al., 1909, p.627) is apparent in the offset to the north-east within the fault system (to Coire Mhorair [NN 185 585]) of the south-dipping Ballachulish Limestone Formation, relative to its outcrop outside the fault system, high on the southern slopes of Sgorr nam Fiannaidh [NN 141 578] (see 1:25 000 scale geological map; British Geological Survey, 2005). Early Geological Survey geologists considered that in this western sector, near Loch Achtriochtan [NN 139 563], the ‘basement’ within the fault system (Leven Schist Formation; (Plate 24)) was of lower metamorphic grade than that outside, consistent with an origin for the inner rocks at significantly higher structural levels (e.g. Bailey, 1960, p.71; but see p.22). Locally, at Meall Dearg [NN 161 583] and Sròn Gharbh [NN 177 583], the fault system cuts the Lower Streaky Andesites, while towards the ridge of Sròn a’ Choire Odhair-bhig [NN 199 579] it cuts the Upper Streaky Andesites; here a minimum displacement of 400 to 500 m can be inferred (the Basal Andesite Sill-complex is thinner here). Continuing clockwise, the ring-fault system juxtaposes various Dalradian metasedimentary rocks, and, although the amount of displacement cannot be established, substantial downthrow can be inferred, because the Dalradian strata on either side of the fault system are lithologically and structurally different (e.g. at Stob Mhic Mhartuin [NN 208 575]). South-east of the River Etive, towards Coire Pollach [NN 256 512], metasedimentary rocks inside the volcano complex are faulted down against the Rannoch Moor Pluton, although elements of fault-intrusion intervene (Figure 25); (Plate 23). In this eastern part of the Glencoe volcano complex, the volcanic succession is cut by the ring-fault system only on the eastern slopes of Meall a’ Bhùiridh [NN 2578 5057]. Here the ring-fault merges into the margin of the ignimbrite-filled crevasse (Three Sisters Ignimbrite Member) that marks the Northeastern Graben Fault; it coincides with this feature for some 750 m and then cuts abruptly across it at [NN259 497], striking south-westwards.

The maximum known cumulative subsidence of the metamorphic basement within the fault system, relative to a surface outside it, is the sum of the greatest subsidence within the Glencoe Graben, which is entirely within the ring-fault (about 700 m) plus the later downthrow on the ring-fault system (also about 700 m). Thus the maximum known cumulative subsidence is about 1400 m. It is possible that actual displacement on the ring-fault system exceeds the amount that can be simply determined, at least locally, so that the actual total could have been greater. This amount of subsidence is quite normal for caldera volcanoes, although such a piecemeal and multistage method of achieving it is either uncommon or has been overlooked and rarely recorded elsewhere (but see Branney and Kokelaar, 1994).

The manner in which faulting occurred on the ring-fault system is not clear. Because the basement of the Glencoe caldera volcano was cut by intersecting regional faults that were active before and during volcanism, and because, consequently, there was early fragmentation of the caldera floor, it is unlikely that the later ring-fault system ever formed a simple continuous dislocation such that there was large-scale piston-like or en bloc (coherent) subsidence within it. The upper members of the Glencoe Volcanic Formation appear to record depocentres that straddled the south-western elements of the main ring structure that lie between the general vicinity of Dalness [NN 17 51] and An t-Sròn [NN 13 55], and it is quite possible that the later volcanotectonic subsidence was bounded to the south-west by the outer strand of the ring-fault system, some 2 km to the south-west, in Gleann Chàrnan [NN 13 51]. The later subsidence that is recorded in the emplacement of the Clach Leathad Pluton (p.98), which partly obliterates and partly invades the Glencoe ring-fault system, possibly constitutes another shift or broadening of the influence of volcanotectonic activity; it is possible that some subsidence along north-eastern parts of the ring-fault system was related to early phases of emplacement of this monzogranitic intrusion.

Flinty crush-rock

The faults that constitute the ring-fault system are widely characterised by the presence of an extremely fine-grained black-to-brown rock, which Clough et al. (1909) called ‘flinty crush-rock’. It occurs as a veneer along the fault planes, commonly at the margins of fault-intrusions, and also in veins that cut irregularly into adjacent (Dalradian) metamorphic rocks, various breccias and fault-intrusions. The flinty crush-rock is intimately associated with a dull pink to red porphyritic rhyolite originally referred to as ‘red felsite’ (Roberts, 1966b; Taubeneck, 1967; Garnham, 1988).

The type locality of the flinty crush-rock is at Stob Mhic Mhartuin [NN 208 575]. Here, the (inner) Main Fault is intruded by a sheet of porphyritic monzodiorite, up to about 30 m wide, which separates thick-bedded (Dalradian) quartzite, down thrown to the south-west, from flaggy quartzite and semipelite to the north-east. Along the planar south-western contact of the intrusion, the fault is represented in a zone, up to 1 m wide, which shows a gradation from undisturbed quartzite into brecciated quartzite, through a white fine-grained microbreccia and then a zone of banded micro-breccia with flinty crush-rock matrix, to flinty crush-rock, and then porphyritic rhyolite (Plate 25a), (Plate 25b). The flinty crush-rock is a band, 3 to 5 cm wide, of delicately laminated, black-to-brown, almost cryptocrystalline quartzofeldspathic material with various inclusions (Plate 26a), (Plate 26b). The most distinctive inclusions are abundant small (0.01–1 mm) quartz grains, many of which are obviously rounded. Locally, the flinty crush-rock encloses swarms of quartzite fragments and euhedral to subhedral (igneous) crystals of plagioclase, K-feldspar, chloritic relicts of amphibole and biotite, opaque oxides and zircon. The lamination is deflected around the included grains and is generally parallel to contacts (Plate 26a), (Plate 26b).

The moderately feldspar-phyric rhyolite (‘red felsite’) adjacent to the flinty crush-rock is locally up to 2 m thick and is of rather variable composition. Early workers thought that it was the chilled margin of the (monzodiorite) fault-intrusion (Clough et al., 1909; Bailey, 1960), but Roberts (1966b) and Garnham (1988) demonstrated that it is a separate intrusion, although in places the two are mingled. Thin veins of the rhyolite interfinger with the flinty crush-rock and various mingling relationships occur, ranging from bleb-like inclusions of rhyolite in crush-rock, through fine-scale swirling interlamination, to a thoroughly mixed hybrid rock (Plate 26a), (Plate 26b). Large crystals (phenocrysts) and clumps of crystals (glomerocrysts) in the rhyolite consist mainly of plagioclase and K-feldspar, with the plagioclase commonly occurring as fragments. Some of the feldspars in the flinty crush-rock are surrounded by a thin coating of rhyolite, showing that they are admixed phenocrysts from the rhyolite. However, other K-feldspars in the crush-rock, along with quartz grains, appear to be derived from a comminuted and partially melted metasedimentary protolith (K-feldspar is abundant in the metasedimentary rocks). Lithic fragments composed of granophyric intergrowths of quartz and K-feldspar in the rhyolite (Plate 26c) indicate a granitic derivation, conceivably from part of the Rannoch Moor Pluton extending at depth under this area. A xenolith of Rannoch Moor granite, 8 m in diameter, is enclosed nearby in the Main Fault-intrusion at Stob Mhic Mhartuin [NN 2090 5740] (see below).

Flinty crush-rock with rhyolite also occurs at the irregular (non-planar), outer contacts between the fault-intrusions and metasedimentary rocks (Roberts, 1966b; Taubeneck, 1967). In the vicinity of Stob Mhic Mhartuin, the quartzitic rocks along the outer contact of the Main Fault-intrusion are strongly fractured and, in contrast to the inner contact, tend to have sharp rather than transitional contacts with the flinty crush-rock. Irregular veins, up to 20 cm wide, of flinty crush-rock with cores of rhyolite, penetrate 2 to 3 m into the metasedimentary rocks, while fine stringers of the crush-rock alone extend several metres farther. In places, flinty crush-rock veins up to 5 cm wide penetrate the fault-intrusions; emplacement of one vein in the Main Fault-intrusion appears to have involved pushing aside of phenocrysts, suggesting that the vein penetrated an unconsolidated crystal mush. This vein locally shows that rhyolite melt entered first and it is interpreted as recording back-veining from the composite margin. Flinty crush-rock also invades interstices of sedimentary and volcaniclastic breccias adjacent to both the inner and the outer faults in this vicinity, and Taubeneck (1967) has identified volcanic rock fragments in the crush-rocks, thus confirming the incorporation of diverse extraneous igneous material into them.

Contact relationships similar to those at Stob Mhic Mhartuin occur elsewhere around the ring-fault system, notably near Loch Achtriochtan in the west [NN 139 566] and at Cam Ghleann in the east [NN 250 523], but in other places, for example in Gleann Fhaolain [NN 14 53] and at An t-Sròn [NN 13 55], the exposed downfaulted rocks are volcanic or sedimentary strata rather than metamorphic basement. Close to the faults, these nonmetamorphic strata show numerous dislocations and are brecciated, while along the faults the exposed fault rocks have loose rubbly or gouge-like textures. This is in marked contrast to the compact fault rocks that occur where the inner rocks are metasedimentary basement (e.g. (Plate 25a), (Plate 25b)). In Gleann Fhaolain, the relatively soft rocks along the fault plane have weathered out to form a pronounced hollow in the side of the valley, but in this case there is evidence of reactivation long after the Glencoe magmatism (see Bailey, 1960, p.155). Most previous workers have noted that wherever strands of the ring-fault juxtapose volcanic or sedimentary rocks on the inside against metasedimentary (Dalradian) rocks on the outside, neither flinty crush-rock nor rhyolite (‘red felsite’) is present. However, one exception has been recorded, south-east of Dalness [NN 178 505], where the inner rocks are Dalness ignimbrite.

Clough et al. (1909), Bailey and Maufe (1916) and Bailey (1960) considered that the flinty crush-rock was formed by frictional heating and partial fusion of the country rocks due to rapid movement on the ring-faults. Shand (1916) named this dark, almost vitreous-textured vein-rock ‘pseudotachylyte’ (synonymous with pseudotachylite), from its similarity to glassy basalt (tachylite). In the survey geologists’ interpretations of the Glencoe ‘cauldron subsidence’, the formation of the flinty crush-rock and emplacement of the fault-intrusion were considered as two stages in a single process. However, first Reynolds (1956) and then Roberts (1966b) disputed whether the contact relationships at Stob Mhic Mhartuin could result simply from intense mechanical brecciation and partial fusion along the ring-fault. Their main concern was that the material represented by the flinty crush-rock clearly had been injected to a considerable distance, at least 5 or 6 m, away from the fault plane on which it was supposed to have been generated. They doubted that a near-instantaneously produced friction-melt would be so mobile and that it could be injected so far into relatively cold rocks. From the published descriptions of the type locality, and rocks elsewhere, Reynolds (1956) postulated that the flinty crush-rock was produced from pyroclastic material that had been emplaced by fluidisation as magmatic gases escaped during or ahead of the ascent of the fault-intrusion magma. Roberts (1966b) reinvestigated details of the contact relationships around the (inner) Main Fault-intrusion at the type locality and, to an extent concurring with Reynolds, suggested that the flinty crush-rocks were produced primarily by the comminution of brecciated country rocks entrained in a (streaming) fluidised system, ahead of fragmenting magma. He suggested that the quartz grains were released from the fault plane during frictional sliding and that their distinctive roundness was caused by attrition in a fluidised system under high confining pressures. Roberts (1966b) went so far as to postulate that the main intrusion was also derived from a fluidised system, by coalescence of magma droplets, but its content of country rock (granite) xenoliths up to 8 m in diameter appears to preclude this. These problematic fault rocks are discussed further in following sections.

Fault-intrusions

The fault-intrusions include numerous and diverse bodies that occur extensively along and mostly outside the innermost trace of the ring-fault system (Figure 25). There is no simple ring-dyke. The intrusions are discontinuous and range from 1 m up to 2000 m wide; they generally have relatively planar inner contacts against the subsided rocks while outer margins are highly irregular (nonplanar). Country-rock xenoliths are abundant. The rocks are typically medium or coarse grained and range from gabbros, through diorites, tonalites and monzonites, to granites (52–72 weight per cent SiO₂; Garnham, 1988). Highly porphyritic varieties are common (phenocrysts 40–70 volume per cent), groundmass is microcrystalline to spherulitic, and chilled margins can be prominent. Distinctive, moderately porphyritic rhyolite occurs widely along intrusion contacts and was initially mistaken as representing chilled margins (‘red felsite’; see previous section). The intrusions occur either individually or in hybrid bodies, and, unlike the streaky andesites that occur in the volcanic succession, there are none that show abundant co-magmatic inclusions indicative of substantial in situ or shallow-level magma mingling. The paucity of exposure in many areas, together with marked lithological heterogeneity, confounds both estimation of the relative volumes of the different rock types and determination of their relative ages. The different types show no clear geometrical or temporal arrangement. According to Garnham (1988), the most abundant rock types are porphyritic diorites, tonalites and monzonites, which form most of the intrusions in Cam Ghleann [NN 249 525] and at An t-Sròn [NN 13 55] and [NN 13 54], as well as at Stob Mhic Mhartuin [NN 208 575]. Granitic rocks occur more commonly in the outer and earlier parts of the fault-intrusion system, notably in the west as sheets along bedding and joint planes in the metamorphic basement e.g. [NN 129 569]. Some of the silicic intrusions show clear evidence that they were emplaced earlier than other types, and these commonly record intense hydrothermal alteration. The altered rocks have numerous ramifying veinlets containing quartz or epidote, or both, or networks of fractures along which there is almost pervasive oxidation of iron-bearing minerals (reddening), and, in places, white mica and pyrite. Gabbroic rock types appear to be the least voluminous and least altered.

The presence of chilled margins at intrusive contacts against the ring-fault, and the common lack of fault-related deformation of the intrusive rocks, led the survey geologists (e.g. Clough et al., 1909) to suggest that fault movement and intrusion were simultaneous. Some intrusions show evidence of having been emplaced before or at an early stage of development of the ‘Main’ ring-fault. In the far west of the volcano complex, the An t-Sròn composite intrusion [NN 13 55] and [NN 13 54], which is about 2 km wide, is the largest of this type (Figure 25). It is predominantly composed of varieties of diorite and tonalite, with branching granitic to rhyolitic (‘felsitic’) sheets towards the outer margin (Bussell, 1979; Garnham, 1988). The sharp inner contact against the ring-fault shows shear deformation and brecciation of a chilled margin, and there is no evidence of any substantial contact metamorphism in the Basal Andesite Sill-complex immediately east of the fault, in the downthrown block. In contrast, the outer intrusion margin is irregular and contains abundant xenoliths of Dalradian metasedimentary rocks. Garnham (1988) suggested that marked thermal metamorphism and partial melting of the metasedimentary rocks in the contact zone to the south-west are records of high heat flow due to substantial throughput of magma, and hence that the An t-Sròn intrusion represents the root (plumbing system) of a volcano. It is possible that the Bidean nam Bian andesite and dacite lavas are extrusive counterparts to the diorites and tonalites of this intrusion (see p.79). The lavas appear to have been deeply ponded by an escarpment in the vicinity of the ring-fault or farther south-west, or both, which would tie in with the relationships between the main An t-Sròn intrusion and two intruded ring-fault strands. Intrusive rocks continuous with the An t-Sròn body extend both along the main ring-fault trace trending south-eastwards on the high flanks of Gleann Fhaolain [NN 141 537], and along the outer, somewhat divergent strand of the fault system, approximately 2 km to the south-west and trending into Gleann Chàrnan [NN 132 517] (see Figure 25). In Gleann Fhaolain, from the south-western slopes of Bidean nam Bian towards Dalness, the fault-intrusion is in places as little as 10 m wide, is mainly composed of tonalite, and has relatively planar outer and inner contacts, like a dyke. The intrusion along the outer strand of the fault system in Gleann Chàrnan, with downthrow to the east-north-east of at least 500 m, is similar.

Cross-cutting relationships between strands of the ring-fault and early and late fault-intrusions are evident in the east of the volcano complex at Coire an Easain [NN 25 49], where there is an abrupt change in strike of both of the fault strands and of the later intrusion (see Bailey and Maufe, 1916, fig. 23). This later fault-intrusion element is continuous with the main body of the Clach Leathad Pluton (Figure 25), which was emplaced during foundering of a crustal block (see p.99), and early workers considered it to be the ‘advance guard’ of this large intrusion (e.g. Bailey, 1960). Along the north-east and south-west flanks of Creag Dhubh [NN 25 52] and north-west towards the River Etive [NN 24 54], the contacts of the fault-intrusion or intrusions are poorly exposed but seem to be broadly planar (Plate 23); farther north-west, however, the outer contacts are very irregular and the three-dimensional shape is evidently complicated.

At Stob Mhic Mhartuin [NN 208 575], where there are two parallel fault strands and associated intrusions of porphyritic monzodiorite, the Early Fault-intrusion is variably brecciated along the outer, Early Fault, which is supposed to have been reactivated (Bailey, 1960). Flinty crush-rock here penetrates the intrusion along brittle fractures. The inner, Main Fault-intrusion is not tectonised. Farther north-west, in Coire nan Lab [NN 175 589] and Coire Cam [NN 15 58], there are various cross-cutting intrusions. The outer margins of the intrusions here, and farther north-west on the slopes below Stob Coire Leith [NN 14 58] and Sgorr nam Fiannaidh [NN 13 57], are extremely irregular, and veins intrude the metasedimentary rock far beyond the main bodies.

Both Reynolds (1956) and Roberts (1966b) considered that the accommodation of fault-intrusions along the various strands of the ring-fault was problematic. They supposed that the controlling faults converged downwards, and then argued that, unless there was central uplift, space could not simply be made for intrusive magma. They questioned the survey geologists’ proposal that the fault-intrusions worked their way upwards largely by magmatic stoping (bit-by-bit incorporation and removal of contact rocks in advancing magma) and suggested that gas-coring by fluidisation would have been a more effective mechanism for making space. With the more recent understanding that the caldera subsidence was piecemeal and substantially facilitated by tectonic dilatation (Moore and Kokelaar, 1997, 1998), and that the faults in fact do not converge downwards (see pp.85–86), neither stoping nor gas coring is required as a fundamental mechanism in accommodating the intrusions. However, that is not to say that they played no part. Given the heterogeneous, discontinuous and cross-cutting nature of the ring-fault intrusions, it is unlikely that there was a single climactic eruption involving large-scale subsidence along the entire system. Although large-scale displacements have taken place, the ring-fault system probably developed through several episodes of faulting and intrusion, over a long period of time.

The contrast between the irregular outer contacts and the planar inner contacts of the ring-fault intrusions (Figure 25) has not been understood. The outer contacts are extremely irregular and commonly include veins that penetrate the country rock extensively. Whereas the inner planar contacts are taken to represent an original fault-plane surface, the outer contacts appear to reflect one or more fault surfaces that have been substantially modified by magmatic stoping. Obviously, because of the great subsidence (order of hundreds of metres), the inner fault surfaces were formed at much shallower levels than the outer irregular contacts with which they are now juxtaposed (across the intrusions). One explanation of the contrasting features of the intrusion contacts could involve fault-plane dilatation and differences in the amount, pressure and temperature of groundwater or magmatic water, or both, present at different levels above the magma storage sites at the time of fault movement (Figure 26). Dilatation would be a natural consequence of subsidence on caldera-bounding faults that dip outwards, but can occur on any fault plane that is uneven, especially where there is slip that juxtaposes walls with differing curvatures. Rapid dilatation of a fault that transects a hydrothermal system will cause explosive transformation of liquid water to steam and vigorous expansion of vapour, with very steep local pressure gradients in fluid pathways around the cavity. The explosivity and fluid flow would effectively blast the hot-water-bearing country rock towards the dilating fault. On the other hand, the rock mass that subsided from shallower levels would not be so fragmented if it was relatively cool or essentially dry, and thus would present a planar fracture surface to the intrusion along its inner margin. This may explain the relationships around the Main Fault-intrusion at Stob Mhic Mhartuin, and the many similar relationships evident elsewhere. At Stob Mhic Mhartuin, the rocks around the Early Fault-intrusion clearly were affected by hydrothermal alteration in breccia zones that predated downthrow on the Main Fault. It is conceivable that dilatation on the Main Fault, following initial slip, caused the hydrothermal system to explode and disintegrate the (outer) fault wallrocks so that rising melts could penetrate and stope them quite easily e.g. (Figure 26).

How such an explosive system might have been linked with activity at the surface is unclear. Explosive fragmentation of the country rock may be recorded by abundant lithic blocks in pyroclastic deposits that formed from an eruption via such a dilated conduit, but the preserved volcanic succession mainly predates the ring-fault and intrusion system. However, the considerable lithicclast contents of the Three Sisters and Dalness ignimbrites (see previous sections) could at least in part reflect such a mechanism.

Substantial magmatically induced (resurgent) uplift is a common, but not ubiquitous, feature of modern caldera volcanoes (e.g. the central Redondo Peak dome and circle of rhyolitic lava vents of the Valles caldera; see (Figure 4)a. Cone-sheets or conical fractures that may indicate magma-chamber pressures sufficiently high to disrupt the overlying crust are unknown in the development of the Glencoe Caldera-volcano Complex. The preserved volcanic and sedimentary succession shows evidence of only limited magmatically induced (resurgent) uplift, in the form of some doming over the shallow-emplaced streaky andesite sills, and in these cases the magmatism was clearly associated with volcanotectonic subsidence. Perhaps the transtensional tectonic setting and graben-like developments of the volcano (Moore and Kokelaar, 1997, 1998) were not conducive to development of strong upward-directed magmatic pressures with attendant substantial uplift and intrusion of cone-sheets. On the contrary, it appears that the eruptions were facilitated by tectonism rather than having been caused by long-term build-up of magmatic pressure with eventual forceful releases of trapped magmas.

A unifying model

The arguments presented by Reynolds (1956) and Roberts (1966b) against the flinty crush-rock being a product of frictional melting (pseudotachylite; p.90) are not well founded. Firstly, there are no data on the mobility of friction-melts that form where fault movements amount to hundreds of metres over periods of hours or days, possibly involving substantial dilatation of fractures and even rocks that were significantly preheated. In thrust-fault systems, movement increments amount to only a few metres and pseudotachylite migrates only for distances of the order of tens of centimetres to occupy minor dilatational jogs (e.g. Camacho et al., 1995). In a volcanotectonic setting, substantial penetration of flinty crush-rock veins away from fault planes does not necessarily preclude an origin by frictional melting. Secondly, rounded grains are normal in melts formed in high-shear-rate experiments (and hence in pseudotachylites), where frictional melting follows inevitable initial comminution (cataclasis) (Spray, 1995). Hence, the small rounded quartz grains in the flinty crush-rock at Glen Coe are not proof of attrition in a particulate dispersion. Estimated rates of displacement on caldera-related (volcanotectonic) faults, in the range 0.1 to 10 cm per second, are sufficient to cause frictional melting to form pseudotachylite (Spray, 1995, 1997), and, consequently, it would be surprising if there was no pseudotachylite along such a well-exposed caldera-fault system. However, it would also be surprising if there were no intrusive (densely welded or coalesced) tuffs containing particles of cataclasite there too. Any melt that contains dissolved volatiles and ascends into an environment of relatively low pressure, such as into a rapidly dilating fracture, will transform into a foam or spray by explosive volatile exsolution. The resultant disrupted melt is likely to be entrained and swept by expanding gas into any opening spaces where, at high temperatures, it is likely to coalesce to a continuous fluid again, as the gas escapes by continued ascent or by diffusion (see Wolff, 1986). Glassy or cryptocrystalline material at the margins of the fault intrusions would represent the first veining by advancing melt, whether of friction-melting or magmatic origin.

The model proposed here is that the flinty crush-rock (pure end-member) did originate by frictional melting, early during fault movement, and that both it and the rhyolite (‘red felsite’) were emplaced largely as fragmented fluids, like a spray, during dilatation of the fault and consequent catastrophic depressurisation of the accessible melts (see (Figure 26)c. That the flinty crush-rock represents an original melt with rheological properties like the magmatic rhyolite is indicated by the ubiquitously similar patterns of intimate mingling of the two components; they show virtually identical millimetre- and submillimetre-scale fine streaks, swirls and lobate interpenetrations, rather like two paints stirred together (Plate 26a), (Plate 26b). Furthermore an origin by frictional melting can readily explain the characteristic rounding of contained small quartz grains (see above), whereas rounding of small grains by gas streaming (as suggested by Roberts, 1966b) while larger grains remain angular seems unlikely. It is inferred that the rhyolite represents intrusive tuff that invaded the fault planes immediately after the first friction-melt was emplaced and shortly before emplacement of the main body of magma into the site to form the fault-intrusion. Such processes would account for the evident considerable mobility of the material that formed the crush-rock and rhyolite, and the sequence of events would readily produce the characteristic spatial succession of lithological units recognised at the type locality (Plate 25). The thorough cross-mixing of the solid constituents of the crush-rock and rhyolite (Plate 26a), (Plate 26b) is readily explained by the mixing of coexisting fragmented melts, while the considerable breakage of the feldspar phenocrysts is like that widely known to be a consequence of explosive magma fragmentation and particulate transport (e.g. Best and Christiansen, 1997; compare Allen and McPhie, 2003).

There is little doubt that the Main Fault at Stob Mhic Mhartuin cuts the Rannoch Moor Pluton at depth, since the Main Fault-intrusion encloses xenoliths, up to 8 m in diameter, of its (outer) distinctively foliated, quartz-rich biotite-granite facies. Without invoking some unlikely intrusive form for the pluton, the estimated minimum depth to the granite beneath Stob Mhic Mhartuin is some 300 to 400 m. It is most probable that the fragments of graphic quartz-feldspar intergrowths in the rhyolite at Stob Mhic Mhartuin were, like the xenoliths, also derived from the Rannoch Moor Pluton. This indicates a considerable mobility of the decompressed magma, which may have been blasted up the fault for several hundreds of metres, following and partly mixing with fragmental friction melt. Thus the mobility of friction melts formed by faulting may be far greater than ever previously conceived. However, cases of friction melt migration involving its fragmentation and transport for distances of tens to hundreds of metres may occur only in volcanotectonic settings.

Chapter 6 Later related magmatism

Clach Leathad and Etive plutons

At a late stage, during or after emplacement of the fault-intrusions, the Glencoe Caldera-volcano Complex was extensively invaded by silicic magma that formed a granitic pluton with a dome-shaped upper (roof ) contact. This intrusion is widely exposed in the south-east and centre of the volcano complex (Figure 25) and it almost certainly underlies much of the rest of it. No base to this pluton is known and most geologists have considered it as part of the large Etive Pluton; it was interpreted as a ‘northern lobe’ of the outer, Cruachan Intrusion part of the Etive Pluton. In this chapter the relationships of the plutonic intrusions are reconsidered and, for reasons explained, the so-called northern lobe within the Glencoe volcano complex is given a new identity, as the Clach Leathad Pluton. Clach Leathad is a mountain summit, at 1099 m, in the north-east of the outcrop [NN 240 493]; the Gaelic name means ‘stone slope’ or ‘stone hillside’, presumably referring to the vast ribbed exposures of the characteristic monzogranite that occur north-east of the summit and extend some 300 m down into Coire an Easain. On a clear day, both the local form of the Clach Leathad Pluton and its scale in relation to the Etive Pluton, with the distinctive (and eponymous) peaks of Ben Cruachan 25 km away to the south-west, are strikingly and beautifully evident from the summit.

The main outcrop of the Etive Pluton extends 22 km from south-west to north-east and 17 km north-west to south-east, an area of some 300 km2. It forms the largest outcrop of the numerous, broadly coeval and genetically related intrusions that densely populate the Grampian Terrane (Figure 3). It consists of an outer part referred to as the Cruachan Intrusion, predominantly of grey quartz-diorite and quartz-monzodiorite that show a tendency to be more silicic in composition with height (Anderson, 1937), within which lies the large Starav Intrusion, composed of porphyritic and nonporphyritic varieties of monzogranite. In the extreme south there are diorites and remnants of another volcano (see Anderson, 1937; Geological Survey Sheet 45, Oban). Topographical outliers of syenogranite, which rests with flat-lying contact on the Cruachan Intrusion, appear to be remnants of one or more near-horizontal or tilted sheets, the intrusive affinities of which are unclear (Anderson, 1937; Bailey, 1960; Batchelor, 1987; Jacques, 1995).

It has long been known that the pinkish monzogranite of what was previously called the northern lobe of the Cruachan Intrusion (now the Clach Leathad Pluton), is petrographically distinct from all of the other plutonic rocks to the south (Kynaston and Hill, 1908; Clough et al., 1909; Bailey and Maufe, 1916; Anderson, 1937; Bailey 1960; Batchelor, 1987; Jacques, 1995). Barritt (1983) has shown that this northern intrusion differs geochemically from the rest, and has its own centred and concentric compositional variations, as illustrated by mapped abundances of the trace elements thorium (Th) and uranium (U). A steep, north-west-trending sheet of leucocratic monzogranite, which is generally 100 to 200 m wide and referred to as the Meall Odhar Intrusion, extends from Meall Odhar [NN 188 463] to Stob Dubh [NN 167 484] and intervenes between the northern intrusion (Clach Leathad Pluton) and the main body of the Cruachan Intrusion to the south (see British Geological Survey, 2005). This sheet appears to be a multiple intrusion that was emplaced along a fault, or fault-zone, and towards its north-east margin it contains centimetre-thick seams of microbreccia and mylonite parallel to its contacts (see Bailey, 1960); this feature along the junction of the two plutonic masses disappears under superficial deposits towards the south-east, where its projection lies along the deep valley occupied by the Allt Dochart [NN 19 45].

The Meall Odhar Intrusion was emplaced after some dykes of the Etive Dyke Swarm (see below), which are cut or altered by the intrusion (Bailey, 1960), and dykes are far more common within the Clach Leathad Pluton than in the rocks just to the south. One dyke that cuts the Glencoe volcano complex and its ring-fault yielded a radiometric Rb-Sr age of 411.7 ± 5.1 million years (Thirlwall, 1988; sample site given as 183 562), which, although perhaps not very reliable, is significantly older than the Rb-Sr dates available from the Etive Pluton: 401 ± 6 million years for the Cruachan Intrusion and 396 ± 12 million years for the central Starav Intrusion (Clayburn et al., 1983). Taken together, these relationships indicate that the Clach Leathad Pluton significantly predated the main Etive intrusions and that it cannot be interpreted simply as an offshoot from the latter. Furthermore, contrasting features of the two bodies suggest that they have been displaced vertically relative to each other; the northern intrusion and its host volcano complex appear to have been displaced downwards, or the Etive Pluton upwards, or both, so that former different levels are now juxtaposed. It is not at all clear what structures might have acted to allow this, but the phenomenon is, in effect, mirrored at the south-western end of the Etive Pluton, where the Lorn lavas are thrown down to the south-west for considerably more than 1 km on the Pass of Brander Fault (Figure 3). This problematic relationship between the plutons is explained below, and is an additional reason for separating the two.

There is no question that the outcrop of the Clach Leathad Pluton is in the roof zone of the intrusion, where magma welled up to replace foundered crustal blocks of Dalradian metasedimentary rocks and Glencoe igneous rocks. The monzogranite cuts the uppermost preserved strata of the Glencoe Volcanic Formation, as well as the ring-fault system, and, although the contacts of the intrusion define a steep dome shape (Figure 25), the host volcanic succession shows no sign of upheaval. These relationships are clearly seen in the flanks of the Lairig Gartain, around [NN 195 538] and across the ridge that links Stob a’ Ghlais Choire [NN 239 516] and Clach Leathad [NN 240 493]. They show that intrusion here involved detachment and subsidence of a large part of the volcanic pile and its basement, as was originally inferred by Clough et al. (1909) and illustrated by them as a type of ‘cauldron subsidence’ (Figure 5)a. The intrusion is rich in drusy cavities, which are common features in uppermost parts of granitic plutons and record accumulation of volatile constituents during rystallisation, and in its overlying roof rocks there has been intense hydrothermal alteration. Patchy to pervasive silicification and recrystallisation have substantially obliterated original rock textures and compositional variations; rather uniform white-weathered surfaces occur widely instead of etched pyroclastic fabrics, while some secondary nodular developments give the false appearance of coarse fragmental deposits. In the east, amongst crags above the River Bà [NN 265 490], and along the river itself [NN 258 478], an offshoot of the pluton has an irregular or sheeted contact with tonalite within the Glencoe fault-intrusion system (see p.93). Good exposures show that the contact lacks a chilled margin (Jacques, 1995), which indicates that the pluton was intruded shortly after the latest fault-intrusion here. There is no evidence, however, that the pluton emplacement was accompanied by any magmatic eruption at the surface.

It is unlikely that much more than 1 km thickness of volcanic succession ever existed above the preserved succession, so it seems that the upper levels of the present exposure, at the top of the Clach Leathad Pluton, represent depths no more than 1 or 2 km beneath the original surface that existed at the time of intrusion. On the other hand, at the same level of exposure now, the Etive Pluton shows no evidence for proximity to any original roof contact, and has a 2 km-wide thermal metamorphic aureole, which, from hornfels near the southern contact, records formation beneath a cover between 3 and 6 km thick (Droop and Treloar, 1981). Such a thick cover cannot have simply extended northwards over the Glencoe volcano complex and Clach Leathad Pluton. Hence it is inferred that the Etive Pluton must have been displaced upwards, or the Glencoe volcano complex downwards, or both. A few of the Etive dykes are continuous between the two plutons, which suggests that any differential vertical motion must have predated these. Relative movement along the line of the Meall Odhar Intrusion, which contains cataclastic seams of microbreccia and mylonite, is indicated, but no tectonic dislocation continuing farther west has been recognised.

In the light of this reappraisal, the newly named Clach Leathad Pluton is viewed as representing a late phase of magmatism in the Glencoe area, and its now less clear association with the Etive Pluton is a major rationale for giving this northern monzogranite a separate identity. It is unclear what may have constituted all of the 3 to 6 km-thick cover of the Etive Pluton at the time of its emplacement, but the upper part is likely to have included a thick sequence of lavas, probably continuous with the Lorn pile now preserved south-west of the Pass of Brander Fault (Figure 3).

Etive Dyke Swarm

The Glencoe Caldera-volcano Complex, the Clach Leathad Pluton, and the Cruachan Intrusion of the Etive Pluton are cut by numerous north-east-trending dykes, referred to collectively as the Etive Dyke Swarm (Clough et al., 1909; Bailey and Maufe, 1916). Overall, this swarm extends north-east–south-west for some 100 km and is up to 20 km wide; it appears to be centred upon the Etive Pluton, where it cuts all but the central Starav Intrusion, but cross-cutting relationships (described above) prove that part of the swarm existed before the Etive plutonic activity. The dykes have not been remapped in the recent study, and many are not represented on the 1:25 000 scale geological map (British Geological Survey, 2005). They are composed of porphyritic microdiorite, micromonzodiorite, microgranodiorite or microgranite, commonly with phenocrysts of plagioclase, quartz or biotite in the more silicic types and hornblende in the less silicic varieties. Typically, the dykes have chilled margins and average 3 to 4 m in width; in some cases they can be traced continuously for more than 10 km. Locally, contrasting rock types occur together in multiple dykes, as for example in Gleann Fhaolain [NN 154 515] (see Bailey and Maufe, 1916, fig. 28). Clearly, emplacement of the Etive Dyke Swarm was associated with north-west to south-east crustal extension, but previously published estimates of the overall extension are too great, and the notion that their removal would restore the caldera volcano to a circular outline is erroneous (see p.29). While in places the dykes make up some 30 per cent or more of the outcrop over several hundreds of metres, a maximum of about 10 to 15 per cent of north-west to south-east extension overall seems more consistent with what can be gleaned from the well-exposed areas. Surveys of the dykes in selected areas in the vicinity of the Glencoe volcano (Morris and Hutton, 1993) have found mismatches of opposed contacts, oblique dyke offshoots and asymmetrical terminations, which all indicate emplacement involving a component of sinistral strike-slip strain (i.e. sinistral transtension), consistent with the more regional picture for the south-west Highlands up to this time. However, the early surveyors provided evidence for pure orthogonal opening of dykes, especially in matching opposed contacts (Clough et al., 1909, pp.642–643; Bailey and Maufe, 1916), and they also noted that the dykes have the same orientation as the numerous joints that break the surrounding rocks: mostly near vertical, but inclined where the joints are inclined. Emplacement of the Etive dykes seemingly started after emplacement of the Clach Leathad Pluton, and was well underway before emplacement of the Cruachan Intrusion, terminating before final emplacement of the Starav Intrusion. Evidently the dyke swarm formed over a long period, and it seems probable that there were various episodes of extension and transtension during this time, or the strain types may have been localised (partitioned). An interesting feature is that the Etive Dyke Swarm seems not to record any stress-field deflection in the immediate vicinity of the Glencoe Caldera-volcano Complex; there is no obvious influence of any north-west-trending discontinuity at the site of the former Glencoe Graben, even though this feature apparently was exploited by dykes at a later time, during the Early Permian (see British Geological Survey, 2005).

Chapter 7 Controls on the location and evolution of the Glencoe Caldera-volcano Complex

The volcano developed at the intersection of substantial crustal discontinuities, and movements occurred on these immediately before and during caldera-forming volcanism. Caldera subsidence involved numerous crustal blocks that subsided incrementally and in various ways before the ring-fault system formed. During caldera-volcano development the sites of both magmatic plumbing and maximum volcanotectonic subsidence shifted south or south-westwards; this can be deduced from the thickness variation of successive eruptive units, preserved vents and buried fault scarps. The succession also demonstrates marked changes in eruption style related to faulting, including punctuation by distinctive phreatomagmatic explosive eruptions, and distinct intervals of erosion and sedimentation between eruptions. The Glencoe Caldera-volcano Complex shows ample evidence of strong tectonic control of piecemeal subsidence during its evolution (Moore and Kokelaar, 1997, 1998).

The orientation and some of the subsidence of the Glencoe Graben were tectonically controlled. Following its establishment, development of the graben was incremental, with faulting commonly unrelated to magma withdrawal. Fluvial drainage was consistently reinstated along the graben at least nine times (Table 2). Furthermore, the structure must have extended and been active outside the volcano complex towards the south-east, in order to maintain the capture of the river system. There is evidence of only minor uplift within the volcano complex, which was due to shallow emplacement of sills (see pp.57; 74); had there been major uplift, either due to tectonism or by large-scale magmatic resurgence, the fluvial system would have been deflected from its rather narrow course across the volcano. The development and persistence of cross-graben structures, along which rivers originating to the north-east of the Glencoe volcano complex were topographically confined, indicate one or more (closely spaced) tectonic faults at right angles to the Glencoe Graben.

Substantial tectonic faulting during the span of volcanic activity at the Glencoe Caldera-volcano Complex is indicated by:

the repeated switches from fluvial incision to fluvial or lacustrine sedimentation
the abrupt switches from fairly normal fluvial-channel and overbank sedimentation to relatively catastrophic and widespread aggradation of coarse-grained alluvial-fan breccias
the collapse of fault scarps to form very coarse debris-avalanche breccias, including megabreccias.

All of these phenomena occurred without contemporary eruption. Such sedimentation is similar to that known from the most actively subsiding strike-slip sedimentary basins, where time-averaged subsidence rates may approach 2 to 3 km per million years (e.g. see Nilsen and Sylvester, 1995). Given that the Glencoe Graben trended at a right angle to the major regional faults on which substantial (sinistral) strike slip is inferred to have occurred during the magmatism (e.g. the Great Glen Fault; (Plate 1)), subsidence along the graben may have been induced, at least in part, by a pull-apart mechanism. Although the volcanic deposits were accommodated mainly by volcanotectonic faulting and downsag, episodes of marked localised subsidence recurrently preceded explosive volcanism. Many volcanotectonic faults originated as tectonic faults and were reactivated so that displacements were accentuated during eruptions. Several sets of closely spaced volcanotectonic faults probably represent splays of tectonic faults (e.g. The Chasm step-fault system). Because many explosive eruptions were preceded by tectonic subsidence in near-vent areas, it seems likely that magma ascent from depth was, in some cases at least, triggered or facilitated by regional tectonism. Tectonism also influenced eruption cycles indirectly so as to lead to formation of contrasting intra-caldera deposits. Each of the Etive rhyolite eruptions, for example, was initiated with phreatomagmatic explosions before switching to magmatic activity, because each ascending magma batch was channelled through water-saturated sediments that accumulated within a tectonically controlled centre of deposition. Thus the location and evolution of the Glencoe Caldera-volcano Complex evidently were strongly influenced by the regional tectonic regime and crustal structure.

The deep north-west-trending fault or fault-zone that controlled the development of the Glencoe Graben is inferred to have been linked at a high angle to the Great Glen Fault (Moore and Kokelaar, 1997, 1998). In the vicinity of Glen Coe there are several north-east-trending faults or shear zones (Figure 3), including the Etive–Laggan Fault, which cuts the volcano complex, and the Ericht–Laidon Fault to the south-east (Hinxman et al., 1923; Treagus, 1991; Jacques and Reavy, 1994). These lie parallel to the Great Glen Fault and were involved in both normal dip-slip and strike-slip displacements during activity on the major fault. Two north-west-trending lineaments have also been recognised, and are believed to mark deep pre-Caledonian crustal structures that influenced Dalradian sedimentation and magmatism as well as subsequent development of tectonometamorphic domains (Fettes et al., 1986): the Cruachan Lineament in the vicinity of the Pass of Brander Fault (Hall, 1985; Graham, 1986) and, parallel to the Strath Ossian Pluton, the Strath Ossian Lineament (Forrest and Key, 1989; Figure 3). The developments of the Glencoe Graben and its fluvial system are strong evidence of a further north-west-trending deep crustal discontinuity, possibly another pre-Caledonian fault or shear zone. This fundamental crustal discontinuity beneath the caldera-volcano complex is named the Glencoe Lineament.

The orthogonal system of faults (and fault splays) at Glen Coe, and the piecemeal and spatially variable subsidence, are considered to reflect a structurally complex response to crustal extension or transtension primarily focused above the intersection of the Glencoe Lineament and the north-east-trending Etive–Laggan Fault (Figure 3). The Devil’s Staircase Fault, which bisects the caldera volcano, is part of the main surface trace of the Etive–Laggan Fault, and the Queen’s Cairn and Glen Etive faults probably formed as splays from depth on this major discontinuity. Subsidence was greatest on the Glencoe Graben axis between the Queen’s Cairn Fault and the Glen Etive Fault (Figure 7). The repeated plumbing of magmas to locations along the Glencoe Graben indicates that magma ascent was focused or accommodated in concert with the extension or transtension at Glen Coe. Interestingly, the few Permian dykes that cut the caldera-volcano complex (see British Geological Survey, 2005) lie parallel to the Glencoe Graben; the most continuous one lies along the graben axis and extends north-west for 2 km beyond the ring-fault.

The base levels of the successive major rivers in the Glencoe Graben were probably controlled by structural developments outside the Glencoe Caldera-volcano Complex, and, as the drainage was to the north-west, these may have involved movements and base-level changes on the Great Glen Fault. Contemporary sedimentary rocks along the Great Glen towards the south-west, for example on Kerrera (Figure 3), include coarse conglomerates with abundant andesite and fewer silicic fragments. These show derivation from the north-east and must have been derived by fluvial input from volcano complexes on lateral drainages, such as towards Ben Nevis and towards Glen Coe.

Throughout the south-west Highlands there are close spatial and temporal relationships between ‘late-Caledonian’ granitic plutons, strike-slip and dip-slip faults, and major tectonic lineaments (Watson, 1984; Hutton and Reavy, 1992; Jacques and Reavy, 1994; Jacques, 1995). Several of the plutons (e.g. Etive and Ben Nevis) truncate and thus postdate volcanic formations, while others preserve no vestige of a central volcano. It appears that the plutons tended to exploit crustal discontinuities that had in places previously focused magmatic plumbing to volcanic centres. It is probable that other ‘late-Caledonian’ plutons succeeded central volcanoes that consequently became obliterated by intrusion or were removed by erosion. Thus caldera-volcano complexes like that at Glen Coe were probably more numerous than is presently apparent in the south-west Highlands; the same applies in the continuation of the magmatic province to the south-west into Ireland (Donegal) and north-eastwards towards the Shetland Isles.

Chapter 8: timing of eruptions

It is difficult to determine the duration of the individual eruptions and the time interval between them. The recognition that the Basal Andesite Sill-complex is only a small remnant of an extensive volcanic field, most of which has been eroded away, implies that a considerable time elapsed after the andesitic magmatism and before the eruption of the Etive rhyolites; there is, however, some ambiguity about how long this gap may have been. Unroofing of the Rannoch Moor Pluton was well advanced before eruption of the Etive rhyolites (clasts of the granite are abundant in the Kingshouse Breccias), but the timing of its unroofing relative to the andesitic magmatism is unclear. Clasts in fluvial conglomerate that rests on the Dalradian metamorphic ‘basement’ beneath the andesite sheets include granite and basic plutonic rocks resembling kentallenite, indicating that plutonism and unroofing were advanced in this vicinity before the andesites were emplaced, but it is unclear whether or not this recorded history predated emplacement of the Rannoch Moor Pluton. If the latter was emplaced after the andesitic magmatism, the time gap represented by the unconformity on the top of the sill stack would have to be a couple of million years or more (the minimum time between pluton intrusion and unroofing; see Harayama, 1992). On the other hand, if the basal conglomerate does relate to the Rannoch plutonism and the andesites were emplaced long after the pluton, the duration marked by the unconformity on the sill stack could be considerably shorter. Nevertheless, the 500 m of tectonic downthrow that preserved the Basal Andesite Sill-complex west of the Queen’s Cairn Fault must have occurred in the interval recorded by the unconformity; for example, if the interval was 500 000 years, the tectonic downthrow rate would have been 1 km per million years or more, which seems reasonable given the strong tectonic control evident in the subsequent piecemeal caldera development. On balance, it is simplest to infer that the basal conglomerate beneath the sills does record early unroofing of rocks of the Rannoch intrusion.

Each of the ignimbrites reflects sustained deposition from a single, albeit perhaps long-lived (hours to days), explosive eruption. The intervals between ignimbrite-forming eruptions were long enough to permit fluvial erosion, involving incision of canyons, with subsequent development of lakes and alluvial fans. The alluvial fans, however, may have formed quickly, since the contemporaneous tectonism was evidently sufficient to cause rapid switches from erosion to normal fluvial or lacustrine sedimentation, as well as switches to relatively catastrophic alluvial-fan aggradation. In at least one instance the interval was long enough to permit emplacement and then removal of a substantial thickness of andesitic lavas (the Lower Streaky Andesites; such an interval may also be recorded by the Upper Streaky Andesites, but the evidence is less clear).

It is inferred that the multiple nature of the collapse and caldera filling at Glen Coe directly relates to the strong tectonic influence. Eight major eruptive events are known to have occurred and probably more were involved; had the crust not been so disrupted there might have been fewer and larger eruptions. By analogy with the Taupo Volcanic Zone, New Zealand, which shows rapid tectonic extension (7–20 mm per year) and consequent high frequency of caldera-forming eruptions (Houghton et al., 1995; Wilson et al., 1995), the intervals between such eruptions at Glen Coe were probably in the order of several thousands to one or two hundreds of thousands of years (see also Smith and Luedke, 1984). It is tentatively estimated that the time between the emplacement of the Lower Etive Rhyolite and the last subsidence and intrusion along the peripheral fault system of the caldera volcano was no more than two or three million years; a large part of the uncertainty lies in not knowing what may have originally overlain the uppermost preserved volcanic strata (Dalness Ignimbrite Member) and what duration of activity is recorded in the diverse ring-fault intrusions. Also, there is considerable uncertainty regarding the time marked by the unconformities in the succession; conceivably, the intervals marked by the unconformities that cut the two streaky andesite units may have been underestimated. Magmatism at the site continued with emplacement of the Clach Leathad Pluton, before a southwards shift of the focus of magmatism to the Etive Pluton, but the time scale for this later activity is at present poorly known.

The incremental, tectonically influenced development of the caldera volcano, which records many pulses of magmatic activity and alternations between emplacement of andesite and rhyolite or rhyodacite, raises the questions of whether such pulsing was fundamentally similar to that involved in the incremental growth of the closely adjacent plutons and whether there ever was a large, longlived magma chamber beneath the Glencoe Caldera-volcano Complex. There are several points bearing on this. Plutonic intrusion temporally bracketed the development of the caldera volcano, with only limited spatial separation (no more than 10–20 km) of the successive foci of magma ascent, so it may be reasonable to suspect that processes and rates of magma generation at depth were similar throughout. Both the Rannoch Moor and Etive plutons show pulsed intrusion of magmas with composition apparently in the same range as in the volcanic succession, namely intermediate (diorite/andesite) to silicic (granite/rhyolite). It appears that subterranean volumes of rhyolite magma were centrally invaded by andesite magma beneath the Glencoe Graben, with consequent very shallow intrusion and eruption to form both of the (mingled) streaky andesite units. Later, large outpourings of andesite and dacite occurred without magma mingling (the volume of the Bidean nam Bian andesites is estimated at more than 12 km³), possibly because of magma ascent to one side of any reservoir of rhyolite, via the An t-Sròn intrusion. The ignimbrites in the succession show no evidence for any substantial compositional zonation of the magma chamber, although it might have occurred but not been tapped. Thus, the weight of evidence presently available suggests that there was no relatively shallow large or long-lived magma body that evolved substantially by magmatic differentiation, but that the volcano tended to have relatively small reservoirs at shallow depth and was supplied relatively frequently by intermediate and silicic magmas that both originated from depth, like the plutons. This may be a characteristic attribute of caldera volcanoes that form on crustal discontinuities during substantial accommodating extension or transtension.

Glen Coe was the site where geologists first addressed the possible relationships between a volcano and underlying coeval intrusions, almost 100 years ago, and despite the numerous subsequent studies much still remains to be understood, not least the fundamental cause and duration of the magmatism. It seems certain that the Glencoe caldera volcano and its adjacent plutons will remain an outstanding natural laboratory for the study of magmatic and volcanic processes for many years to come, and equally likely that unexpected new insights will be derived in the future from these spectacularly exposed rocks.

Chapter 9 Shaping the landscape

Pleistocene: advance and retreat of the glaciers

The scale of Glen Coe, with its towering mountains, make it a place of wonder, but, despite the great age of the caldera volcano, it was only during the last 2 million years, in the Quaternary Period, that the glen took on its spectacular appearance (Plate 27).

Western Scotland was heavily glaciated during the Pleistocene Epoch and, although the exact number of glaciations and their relative severity is not precisely known, it is likely that they started around 2.4 million years ago (Gordon and Sutherland, 1993). Little evidence exists on land for those episodes that predated the last major glaciation, which occurred between about 40 000 and 11 500 years ago.

Over the last 100 000 years, the Scottish climate has been dominated by harsh, arctic or subarctic conditions. This largely cold period was punctuated by brief temperate spells (interstadials) of 10 000 to 15 000 years’ duration; we are currently in such an interstadial, known as the Holocene Epoch (Figure 27). Between about 25 000 and 15 000 years ago, during the last glacial stage of the Pleistocene (late Devensian Age), intense cold prevailed and glaciers grew to form ice caps in the western Highlands, the Cairngorms and the Southern Uplands. These ice caps coalesced to form a large ice-sheet centred over western Scotland. Evidence from glacial erratics and striae suggest that during this glaciation the general direction of ice movement in the Glencoe area was from east to west from a source in the Rannoch Moor area, close to the present-day watershed. The deep valley of Glen Coe was a major route for ice; glacial striae occur at an altitude of 900 m above sea level on Am Bodach [NN 169 579] and Stob Coire Leith [NN 152 588], and glacial erratics occur at this altitude on Stob Dearg [NN 216 542] and Meall Dearg [NN 162 583] (Bailey, 1960), suggesting that even the highest peaks were covered by the ice-sheet. However, much of the evidence for this, and previous glaciations, has since been removed.

Soon after 18 000 years ago the ice-sheet in Britain was in retreat. Deglaciation did not occur smoothly or synchronously across Scotland, but was punctuated by several re-advances of varying magnitude (Robinson and Ballantyne, 1979; Dawson, 1982; Merritt et al., 1995). The climatic implications of these re-advances and the configuration of the ice-sheet in Britain between about 17 000 and 15 000 years ago are currently the subjects of much debate (McCabe et al., 1998; Adams et al., 1999; Bowen et al., 2002).

Evidence for a short-lived interstadial between 14 5000 and 13 000 years ago is widespread in Scotland, but it is not known whether the Late Devensian glaciers disappeared completely at this time. About 13 000 years ago, ice began to accumulate once again in the south-west Highlands. At the height of this glacial re-advance, which is known as the Loch Lomond Stadial, a considerable ice cap was centred just a few kilometres east of the head of Glen Coe and outlet glaciers extended to sea level in many of the major west Highland fjords. Ice in the Glencoe area was topographically constrained; on mountain spurs such as the Three Sisters an upper zone of advanced weathering can be distinguished from a lower zone of ice-scoured rock. This pattern of weathering and erosion in combination with other geomorphological evidence such as the presence or absence of blockfields, solifluction sheets and fossil screes has been used to delimit the maximum altitude of glaciation during the Loch Lomond Stadial (Thorp, 1981, 1986). The Glencoe glacier probably reached a height of 500 to 600 m on Beinn Fhada, 400 to 450 m on Meall Mòr and 300 m at Ballachulish (Figure 28). Much of the Aonach Eagach and the Bidean nam Bian massif would have remained as ice-free nunataks. This brief glacial period lasted less than 1500 years and ended abruptly 11 500 years ago.

The legacy of glaciation is everywhere in Glen Coe. Large erratic boulders lie scattered on the valley floor and perched on the valley sides. Boulders of granitic rock from the Rannoch Moor Pluton occur on Buachaille Etive Mòr and on the Aonach Eagach at over 800 m above sea level, about 300 m higher than the remaining outcrop (Bailey, 1960). Extensive glacial deposits flank the valley sides of Glen Coe, Glen Etive and their tributary valleys. Where gullies cut into these, poorly sorted, matrix-supported, sandy clay-rich diamicton is exposed. This deposit, generally referred to as till, commonly varies greatly in sedimentology and surface expression, recording deposition in various glacial settings. Where compact, clay-rich diamicton forms smooth sheets, a subglacial origin is inferred, in accord with studies of modern, warm-based glaciers (Boulton, 1970, 1986). Good exposures of this type of deposit occur near Loch Achtriochtan [NN 142 570] and on the flanks of An t-Sròn [NN 123 559]. Other, less well-consolidated, sand-rich deposits record an ice-marginal or supraglacial origin. Commonly, these have a hummocky, apparently chaotic morphology, and were probably laid down during glacier retreat. Good examples abound on Rannoch Moor, particularly around Lochan Mathair Eite [NN 285 541]. Well-defined, individual moraines are not common in the Glen Coe–Glen Etive area, where hummocky glacial deposits are more typical. The best-defined end moraines are found at the mouth of the Lairig Gartain, on the north-west flank of Buachaille Etive Mòr [NN 2212 7554], near Alltchaorunn in Glen Etive [NN 194 512], and in Fionn Ghleann [NN 127 539].

The moraine ridge in Fionn Ghleann is associated with a conspicuous series of sand and gravel terraces, which were probably deposited as ice-contact fans (or deltas) by running water issuing from a melting glacier. How much of the material mapped as alluvium on the floor of Glen Coe and the nearby valleys is of glaciofluvial origin, rather than recent fluvial origin, is uncertain. The coarse, bouldery, modern river deposits in Glen Coe are difficult to distinguish from other material that may have been left behind by glacial meltwater torrents. It is probable that much of the sediment immediately down-valley from Loch Achtriochtan [NN 135 566] was deposited at the end of the Loch Lomond Stadial by meltwater flowing from wasting glaciers.

Holocene: after the glaciers

Radiocarbon dates from organic sediments deposited in enclosed basins have been used to determine the timing of deglaciation. On Rannoch Moor, which was one of the centres of ice accumulation, radiocarbon dates of between 9800 and 10 600 years before present demonstrate that the area was ice-free shortly before then (Lowe and Walker, 1992).

Landscape modification has continued since the retreat of the glaciers. Large amounts of debris and surface sediment were left in potentially unstable settings in Glen Coe when the ice melted. Over the last 11 000 years, various slope processes have created some of the finest examples in Britain of debris cones and mountain alluvial fans (Brazier, 1992; McEwen, 1997). Many of these are currently active and are thought to have undergone renewed activity in the last 500 years (Innes, 1983). The unusually large number of active fans in Glen Coe is largely due to the geology of the surrounding mountains. The abundant microdioritic to microgranitic dykes (Etive Dyke Swarm, see p.100) have weathered preferentially to form spectacular, steep gullies that channel runoff and accentuate erosion. The debris cone at the base of the towering cliffs of Am Bodach is the largest and most striking talus deposit in Glen Coe. It is fed from the north-west flank of Am Bodach and the extensive debris fields beneath The Chancellor (see (Plate 14b)). When swollen, the River Coe erodes the toe of this landform and causes the channels on its surface to incise to a new base level. The last large flow of debris on this cone occurred in AD 1875, but lesser events occur on a timescale of ten years or less. All of the steep debris cones in the valley are commonly reactivated during heavy rain (Innes, 1983; Brazier, 1992); new lobes of debris form as the main channels migrate across the cones.

The River Coe has changed its course several times during the past 130 years (McEwen, 1997), and has probably done so for the past 10 000 years, owing to fluctuations in sediment load and natural instability. Progradation of a delta into Loch Achtriochtan is gradually reducing the size of this lake; as Bailey (1960) commented, the loch is ‘in process of extinction’.

Occasional catastrophic slope processes have resulted in significant landscape change. A large rockfall deposit, which formed during postglacial times in Coire Gabhail (Lost Valley [NN 166 556]), is a spectacular example (Plate 18) and is believed to be the largest catastrophic rockfall feature in Britain (Ballantyne, 1991). Debris from the flank of Gearr Aonach fell into the valley, probably owing to two massive rock-slope failures. The boulders form a talus ‘dam’ over 20 m high; hundreds of thousands of tonnes of debris choke the narrow valley, making it difficult to pass through. Alluvial sediment accumulated behind the dam to form a small alluvial plain that affords good grazing; in historic times cattle were hidden here during times of conflict. The cause and timing of the rock fall are not precisely known.

References

Most of the references listed below are held in the Library of the British Geological Survey, Murchison House, Edinburgh, and at Keyworth, Nottingham. Copies of the references can be purchased subject to current copyright legislation. BGS Library catalogue can be searched online at: http://geolib.bgs.ac.uk

ADAMS, J, MASLIN, M, and THOMAS, E. 1999. Sudden climatic changes during the Quaternary. Progress in Physical Geography, Vol. 23, 1–36.

ALLEN, S R, and MCPHIE, J. 2003. Phenocryst fragments in rhyolitic lavas and lava domes.

Journal of Volcanology and Geothermal Research, Vol. 126, 263–283.

ANDERSON, J G C. 1937. The Etive Granite Complex. Quarterly Journal of the Geological Society of London, Vol. 93, 487–533.

ANDERTON, R. 1985. Sedimentation and tectonics in the Scottish Dalradian. Scottish Journal of Geology, Vol. 21, 407–436.

ARMSTRONG, H A, and OWEN, A W. 2001. Terrane evolution of the paratectonic Caledonides of northern Britain. Journal of the Geological Society of London, Vol. 158, 475–486.

ATHERTON, M P, and GHANI, A A. 2002. Slab breakoff: a model for Caledonian, Late Granite syn-collisional magmatism in the orthotectonic (metamorphic) zone of Scotland and Donegal, Ireland. Lithos, Vol. 62, 65–85.

BAILEY, E B. 1934. West Highland tectonics: Loch Leven to Glen Roy. Quarterly Journal of the Geological Society of London, Vol. 90, 462–523.

BAILEY, E B. 1960. The geology of Ben Nevis and Glen Coe and the surrounding country. Second edition. Memoir of the Geological Survey of Great Britain, Sheet 53 (Scotland). (London: HMSO.)

BAILEY, E B, and MAUFE, H B. 1916. The geology of Ben Nevis and Glen Coe and the surrounding country. Memoir of the Geological Survey of Great Britain, Sheet 53 (Scotland). (London: HMSO.)

BALLANTYNE, C K. 1991. Holocene geomorphic activity in the Scottish Highlands. Scottish Geographical Magazine, Vol. 107, 84–98.

BARRITT, S D. 1983. The controls of radioelement distribution in the Etive and Cairngorm granites: implications for heat production. Unpublished PhD thesis, Open University.

BATCHELOR, R A. 1987. Geochemical and petrological characteristics of the Etive granitoid complex, Argyll. Scottish Journal of Geology, Vol. 23, 227–249.

BEST, M G, and CHRISTIANSEN, E H. 1997. Origin of broken phenocrysts in ash-flow tuffs.

Bulletin of the Geological Society of America, Vol. 109, 63–73.

BOULTON, G S. 1970. On the deposition of subglacial and melt-out tills at the margins of certain Svalbard glaciers. Journal of Glaciology, Vol. 9, 213–229.

BOULTON, G S. 1986. Push-moraines and glacier-contact fans in marine and terrestrial environments. Sedimentology, Vol. 33, 677–698.

BOWEN, D Q, PHILLIPS, F M, MCCABE, A M, KNUTZ, P C, and SYKES, G A. 2002. New data for the last glacial maximum in Great Britain and Ireland. Quaternary Science Reviews, Vol. 21, 89–101.

BRANNEY, M J. 1995. Downsag and extension at calderas: new perspectives on collapse geometries from ice-melt, mining, and volcanic subsidence. Bulletin of Volcanology, Vol. 57, 303–318.

BRANNEY, M J, and KOKELAAR, B P. 1994. Volcanotectonic faulting, soft-state deformation and rheomorphism of tuffs during development of a piecemeal caldera, English Lake District. Bulletin of the Geological Society of America, Vol. 106, 507–530.

BRANNEY, M J, and KOKELAAR, B P. 2002. Pyroclastic density currents and the sedimentation of ignimbrites. Geological Society of London Memoir, No. 27.

BRANNEY, M J, KOKELAAR, B P, and MCCONNELL, B J. 1992. The Bad Step Tuff: a lava-like rheomorphic ignimbrite in a calc-alkaline piecemeal caldera, English Lake District. Bulletin of Volcanology, Vol. 54, 187–199.

BRAZIER, V. 1992. Holocene debris cones in Glencoe and Glen Etive. 67–73 in South-west Scottish Highlands: field guide. WALKER, M J C, GRAY, J M, and LOWE, J J (editors). (Cambridge: Quaternary Research Association.)

BRITISH GEOLOGICAL SURVEY. 2005. Glen Coe. Bedrock. 1:25 000 Geology Series. (Keyworth, Nottingham: British Geological Survey.)

BUSSELL, M A. 1979. Structures at the south-western margin of the Glencoe fault intrusion on Ant-Sron. Scottish Journal of Geology, Vol. 15, 279–286.

CAMACHO, A A, VERNON, R H, and FITZGERALD, J D. 1995. Large volumes of anhydrous pseudotachylyte in the Woodroffe Thrust, eastern Musgrave Ranges, Australia. Journal of Structural Geology, Vol. 17, 371–383.

CLAYBURN, J A P, HARMON, R S, PANKHURST, R J, and BROWN, J F. 1983. Sr, O and Pb isotope evidence for origin and evolution of Etive Igneous Complex, Scotland. Nature, Vol. 303, 492–497.

CLOUGH, C T, MAUFE, H B, and BAILEY, E B. 1909. The cauldron subsidence of Glen Coe and the associated igneous phenomena. Quarterly Journal of the Geological Society of London, Vol. 65, 611–678.

COCKS, L R M, and TORSVIK, T H. 2002. Earth geography from 500 to 400 million years ago: a faunal and palaeomagnetic review. Journal of the Geological Society of London, Vol. 159, 631–644.

DAVIES, J H, and VON BLANCKENBURG, F. 1995. Slab breakoff: a model of lithosphere detachment and its test in the magmatism and deformation of collisional orogens. Earth and Planetary Science Letters, Vol. 129, 85–102.

DAWSON, A G. 1982. Late glacial sea-level changes and ice limits in Islay, Jura and Scarba, Inner Hebrides. Scottish Journal of Geology, Vol. 18, 253–165.

DEWEY, J F, and STRACHAN, R A. 2003. Changing Silurian–Devonian relative plate motion in the Caledonides: sinistral transpression to sinistral transtension. Journal of the Geological Society of London, Vol. 160, 219–229.

DROOP, G T R, and TRELOAR, P J. 1981. Pressures of metamorphism in the thermal aureole of the Etive Granite Complex. Scottish Journal of Geology, Vol. 17, 85–102.

EKREN, E B, MCINTYRE, D H, and BENNETT, E H. 1984. High-temperature large-volume, lavalike ash-flow tuffs without calderas in southwestern Idaho. United States Geological Survey Professional Paper, No. 1272, 1–76.

ELLES, G L, and TILLEY, C E. 1930. Metamorphism in relation to structure in the Scottish Highlands. Transactions of the Royal Society of Edinburgh, Vol. 56, 621–646.

FETTES, D J, GRAHAM, C M, HARTE B, and PLANT, J A. 1986. Lineaments and basement domains: an alternative view of Dalradian evolution. Journal of the Geological Society of London, Vol. 143, 453–464.

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

FITTON J G, THIRLWALL M F, and HUGHES D J. 1982. Volcanism in the Caledonian orogenic belt of Britain. Vol. 611–636 in Andesites. THORPE R S (editor). (Chichester: Wiley.)

FORREST, M D, and KEY, R M. 1989. The Strath Ossian Lineament: geological and geochemical evidence. Terra Abstracts, Vol. 1, 12.

FRANCIS, P, and OPPENHEIMER, C. 2004. Volcanoes. (Oxford: Oxford University Press.) FROST, C D, and O’NIONS, R K. 1985. Caledonian magma genesis and crustal recycling.

Journal of Petrology, Vol. 26, 515–544.

GARNHAM, J A. 1988. Ring-faulting and associated intrusions, Glencoe, Scotland. Unpublished PhD thesis, University of London.

GORDON, J E, and SUTHERLAND, D G (editors). 1993. Quaternary of Scotland. (London: Chapman and Hall.)

GRADSTEIN, F M, and OGG, J G. 1996. A Phanerozoic timescale. Episodes, Vol. 19, 3–5. GRAHAM, C M. 1986. The role of the Cruachan Lineament during Dalradian evolution. Scottish Journal of Geology, Vol. 22, 257–270.

GROOME, D R, and HALL, A. 1974. The geochemistry of the Devonian lavas of the northern Lorne Plateau, Scotland. Mineralogical Magazine, Vol. 39, 621–640.

HALL, A. 1985. Geophysical constraints on crustal structure in the Dalradian region of Scotland. Journal of the Geological Society of London, Vol. 142, 149–155.

HALLIDAY, A N. 1984. Coupled Sm-Nd and U-Pb systematics in late Caledonian granites and the basement under northern Britain. Nature, Vol. 307, 229–233.

HALLIDAY, A N, STEPHENS, W E, HUNTER, R H, MENZIES, M A, DICKIN, A P, and HAMILTON, P J. 1985. Isotopic and chemical constraints on the building of the deep Scottish lithosphere. Scottish Journal of Geology, Vol. 21, 465–491.

HARAYAMA, S. 1992. Youngest exposed granitoid pluton on Earth: cooling and rapid uplift of the Pliocene–Quaternary Takidani granodiorite in the Japan Alps, central Japan. Geology, Vol. 20, 657–660.

HARDIE, W G. 1963. Explosion-breccias near Stob Mhic Mhartuin, Glen Coe, Argyll, and their bearing on the origin of the nearby flinty crush-rock. Transactions of the Edinburgh Geological Society, Vol. 19, 426–438.

HARDIE, W G. 1968. Volcanic breccia and the Lower Old Red Sandstone unconformity, Glencoe, Argyll. Scottish Journal of Geology, Vol. 4, 291–299.

HARMON, R S, and HALLIDAY, A N. 1980. Oxygen and strontium isotope relationships in the British late Caledonian granites. Nature, Vol. 283, 21–25.

HARMON, R S, HALLIDAY, A N, CLAYBURN, J A P, and STEPHENS, W E. 1984. Chemical and isotopic systematics of the Caledonian intrusions of Scotland and Northern England: a guide to magma source region and magma-crust interaction. Philosophical Transactions of the Royal Society of London, Series A, Vol. 310, 709–742.

HICKMAN, A H. 1975. The stratigraphy of late Precambrian metasediments between Glen Roy and Lismore. Scottish Journal of Geology, Vol. 11, 117–142.

HICKMAN, A H. 1978. Recumbent folds between Glen Roy and Lismore. Scottish Journal of Geology, Vol. 14, 191–212.

HINXMAN, L W, CARRUTHERS, R G, and MACGREGOR, M. 1923. The geology of Corrour and the Moor of Rannoch. Memoir of the Geological Survey of Great Britain, Sheet 54 (Scotland). (London: HMSO.)

HOLDEN, P, HALLIDAY, A N, and STEPHENS, W E. 1987. Neodymium and strontium isotope content of microdiorite enclaves points to mantle input to granitoid production. Nature, Vol. 330, 53–56.

HOUGHTON, B F, WILSON, C J N, MCWILLIAMS, M O, LANPHERE, M A, WEAVER, S D, BRIGGS, R M, and PRINGLE, M S. 1995. Chronology and dynamics of a large silicic magmatic system: Central Taupo Volcanic Zone, New Zealand. Geology, Vol. 23, 13–16.

HUTTON, D H W. 1987. Strike-slip terranes and a model for the evolution of the British and Irish Caledonides. Geological Magazine, Vol. 124, 405–425.

HUTTON, D H W, and REAVY, R J. 1992. Strike-slip tectonics and granite petrogenesis. Tectonics, Vol. 11, 960–967.

INNES, J L. 1983. Lichenometric dating of debris flow deposits in the Scottish Highlands. Earth Surface Processes and Landforms, Vol. 8, 579–588.

JACQUES, J M. 1995. Caledonian magmatism and major tectonic structures in the south-west Highlands of Scotland: implications for ascent, siting and emplacement. Unpublished PhD thesis, Durham University.

JACQUES, J M, and REAVY, R J. 1994. Caledonian plutonism and major lineaments in the south-west Scottish Highlands. Journal of the Geological Society of London, Vol. 151, 955–969.

KEY, R M, CLARK, G C, MAY, F, PHILLIPS, E R, CHACKSFIELD, B C, and PEACOCK, J D. 1997. Geology of the Glen Roy district. Memoir of the British Geological Survey, Sheet 63W (Scotland). (London: The Stationery Office.)

KEY, R M, PHILLIPS, E R, and CHACKSFIELD, B C. 1993. Emplacement and thermal metamorphism associated with the post-orogenic Strath Ossian Pluton, Grampian Highlands, Scotland. Geological Magazine, Vol. 130, 379–390.

KIDSTON, R, and LANG, W H. 1924. Notes on fossil plants from the Old Red Sandstone of Scotland. III. On two species of Pachytheca (P. media and P. fasciculata) based on the characters of algal filaments. Transactions of the Royal Society of Edinburgh, Vol. 53, 604–614.

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.

KOKELAAR, B P. 1986. Magma-water interactions in subaqueous and emergent basaltic volcanism. Bulletin of Volcanology, Vol. 48, 275–289.

KOKELAAR, B P. 1988. Tectonic controls of Ordovician arc and marginal basin volcanism in Wales. Journal of the Geological Society of London, Vol. 145, 759–775.

KOKELAAR, B P. 1992. Ordovician marine volcanic and sedimentary record of rifting and volcanotectonism: Snowdon, Wales, United Kingdom. Bulletin of the Geological Society of America, Vol. 104, 1433–1455.

KOKELAAR, B P, BRANNEY, M J, MOORE, I D, and HOWELLS, M F. 1994. Processes and controls of caldera collapse and related ignimbrite emplacement in Snowdonia (Wales), the Lake District (England), and Glen Coe (Scotland), United Kingdom: Field Guide, International Association of Volcanology and Chemistry of the Earth's Interior Commission on Explosive Volcanism Field Workshop, May 18–29, 1994, University of Liverpool, UK, 130pp [unpublished report].

KYNASTON, H, and HILL, J B. 1908. Geology of the country near Oban and Dalmally. Memoir of the Geological Survey of Great Britain, Sheet 45 (Scotland). (London: HMSO.)

LIPMAN, P W. 2000. Calderas. 643–662 in Encyclopedia of volcanoes. SIGURDSSON, H (editor) (San Diego, California; Academic Press.)

LITHERLAND, M. 1982. The structure of the Loch Creran Dalradian and a new model for the south-west Highlands. Scottish Journal of Geology, Vol. 18, 205–225.

LOWE, J J, and WALKER, M J C. 1997. Reconstructing Quaternary environments. (Harlow: Longman.)

MANVILLE, V, and WILSON, C J N. 2003. Interactions between volcanism, rifting and subsidence: implications of intracaldera palaeoshorelines at Taupo volcano, New Zealand. Journal of the Geological Society of London, Vol. 160, 3–6.

MARSHALL, J E A. 1991. Palynology of the Stonehaven Group, Scotland: evidence for a Mid Silurian age and its geological implications. Geological Magazine, Vol. 128, 283–286.

MARSHALL, P. 1935. Acid rocks of the Taupo-Rotorua volcanic district. Transactions of the Royal Society of New Zealand, Vol. 64, 323–366.

MAUFE, H B. 1910. The geological structure of Ben Nevis. Geological Survey of Great Britain Summary of Progress (for 1909), 80–89.

MCCABE, M, KNIGHT, J, and MCCARRON, S. 1998. Evidence for Heinrich Event 1 in the British Isles. Journal of Quaternary Science, Vol. 13, 549–568.

MCEWEN, L J. 1997. Glen Coe: river and slope forms. 72–76 in Fluvial geomorphology of Great Britain. GREGORY, K J (editor). Geological Conservation Review Series, No. 13 (London: Chapman and Hall).

MERRITT, J W, AUTON, C A, and FIRTH, C R. 1995. Ice-proximal glaciomarine sedimentation and sea-level change in the Inverness area, Scotland: a review of the deglaciation of a major ice stream of the British Late Devensian Ice Sheet. Quaternary Science Reviews, Vol. 14, 289–329.

MILNER, D M, COLE, J W, and WOOD, C P. 2002. Asymmetric, multiple-block collapse at Rotorua Caldera, Taupo Volcanic Zone, New Zealand. Bulletin of Volcanology, Vol. 64, 134–149.

MOORE, I D. 1996. The early history of the Glencoe cauldron, Unpublished PhD thesis, University of Liverpool.

MOORE, I D, and KOKELAAR, B P. 1997. Tectonic influences in piecemeal caldera collapse at Glencoe Volcano, Scotland. Journal of the Geological Society of London, Vol. 154, 765–768.

MOORE, I D, and KOKELAAR, B P. 1998. Tectonically controlled piecemeal caldera collapse: a case study of Glencoe volcano, Scotland. Bulletin of the Geological Society of America, Vol. 110, 1448–1466.

MORRIS, G A, and HUTTON, D H W. 1993. Evidence for sinistral shear associated with the emplacement of the early Devonian Etive dyke swarm. Scottish Journal of Geology, Vol. 29, 69–72.

NILSEN, T H, and SYLVESTER, A G. 1995. Strike-slip basins. 425–457 in Tectonics of sedimentary basins. BUSBY C J, and INGERSOLL, R V. (editors). (Oxford: Blackwell Science.)

READ, H H. 1961. Aspects of Caledonian magmatism in Britain. Liverpool and Manchester Geological Journal, Vol. 2, 653–683.

REYNOLDS, D L. 1956. Calderas and ring-complexes. Verhandelingen van het Koninklijk Nederlands Geologisch Mijnbouwkundig Genootschap, Geologische serie, Vol. 16, 355–379.

ROBERTS, J L. 1963. Source of the Glencoe ignimbrites. Nature, Vol. 199, 901.

ROBERTS, J L. 1966a. Ignimbrite eruptions in the volcanic history of the Glencoe cauldron subsidence. Geological Journal, Vol. 5, 173–184.

ROBERTS, J L. 1966b. The emplacement of the Main Glencoe fault-intrusion at Stob Mhic Mhartuin. Geological Magazine, Vol. 103, 299–316.

ROBERTS, J L. 1974. The evolution of the Glencoe Cauldron. Scottish Journal of Geology, Vol. 10, 269–282.

ROBINSON, M, and BALLANTYNE, C K. 1979. Evidence for a glacial readvance pre-dating the Loch Lomond Advance in Wester Ross. Scottish Journal of Geology, Vol. 15, 271–277.

SELF, S, GOFF, F, GARDNER, J N, WRIGHT, J V, and KITE, W M. 1986. Explosive rhyolitic volcanism in the Jemez Mountains: vent locations, caldera development and relation to regional structure. Journal of Geophysical Research, Vol. 91, No. B2, 1779–1798.

SHAND, J. 1916. The pseudotachylyte of Parijs (Orange Free State), and its relation to ‘trap-shotten gneiss’ and ‘flinty crush-rock’. Quarterly Journal of the Geological Society of London, Vol. 72, 198–221.

SMITH, R L, and BAILEY, R A. 1968. Resurgent cauldrons. 153–210 in Studies in Volcanology, COATS, R R, HAY, R L, and ANDERSON, C A (editors). Geological Society of America Memoir,

SMITH, R L, and LUEDKE. 1984. Potentially active volcanic lineaments and loci in western conterminous United States. 47–66 in Explosive volcanism: inception, evolution and hazards (Washington, D C: National Academic Press.)

SOPER, N J, STRACHAN, R A, HOLDSWORTH, R E, GAYER, R A, and GREILING, R O. 1992. Sinistral transpression and the Silurian closure of Iapetus. Journal of the Geological Society of London, Vol. 149, 871–880.

SPRAY, J G. 1995. Pseudotachylyte controversy: Fact or friction? Geology, Vol. 23, 1119–1122. SPRAY, J G. 1997. Superfaults. Geology, Vol. 25, 579–582.

STEPHENSON, D, and GOULD, D. 1995. British regional geology: the Grampian Highlands. Fourth edition. (London: HMSO for the British Geological Survey.)

TARNEY, J, and JONES, C E. 1994. Trace element geochemistry of orogenic igneous rocks and crustal growth models. Journal of the Geological Society of London, Vol. 151, 855–868.

TAUBENECK, W. 1967. Notes on the Glen Coe cauldron subsidence, Argyllshire, Scotland. Geological Society of America Bulletin, Vol. 78, 1295–1316.

THIRLWALL M F. 1979. Petrochemistry of the British Old Red Sandstone volcanic province. Unpublished PhD thesis, University of Edinburgh.

THIRLWALL, M F. 1981. Implications for Caledonian plate tectonic models of chemical data from volcanic rocks of the British Old Red Sandstone. Journal of the Geological Society of London, Vol. 138, 123–138.

THIRLWALL, M F. 1982. Systematic variation in chemistry and Nd-Sr isotopes across a Caledonian calc-alkaline volcanic arc: implications for source materials. Earth and Planetary Science Letters, Vol. 58, 27–50.

THIRLWALL M F. 1986. Lead isotope evidence for the nature of the mantle beneath Caledonian Scotland. Earth and Planetary Science Letters, Vol. 80, 55–70.

THIRLWALL, M F. 1988. Geochronology of Late Caledonian magmatism in northern Britain. Journal of the Geological Society of London, Vol. 145, 951–967.

THORP, P W. 1981. A trimline method for defining the upper limit of Loch Lomond Advance glaciers: examples from the Loch Leven and Glen Coe areas. Scottish Journal of Geology, Vol. 17, 49–64.

THORP, P W. 1986. A mountain icefield of Loch Lomond Stadial age, western Grampians, Scotland. Boreas, Vol. 15, 83–97.

TREAGUS, J E. 1974. A structural cross-section of the Moine and Dalradian rocks of the Kinlochleven area, Scotland. Journal of the Geological Society of London, Vol. 130, 525–544.

TREAGUS, J E. 1991. Fault displacements in the Dalradian of the Central Highlands. Scottish Journal of Geology, Vol. 27, 135–145.

TREWIN, N H, and THIRLWALL, M F. 2002. Old Red Sandstone. 213–249 in The geology of Scotland. Fourth edition. TREWIN, N H (editor). (London: The Geological Society.)

TUCKER, R D, BRADLEY, D C, VER STRAETEN, C A, HARRIS, A G, EBERT, J R, and MCCUTCHEON, S R. 1998. New U-Pb zircon ages and the duration and division of Devonian time. Earth and Planetary Science Letters, Vol. 158, 175–186.

VON BLANCKENBURG, F, and DAVIES, J H. 1995. Slab breakoff: A model for syncollisional magmatism and tectonics in the Alps. Tectonics, Vol. 14, 120–131.

WALKER, G P L. 1984. Downsag calderas, ring-faults and caldera sizes. Journal of Geophysical Research, Vol. 89, No. B10, 8407–8416.

WATSON, J V. 1984. The ending of the Caledonian orogeny in Scotland. Journal of the Geological Society of London, Vol. 141, 193–214.

WELLMAN, C H. 1994. Palynology of the ‘Lower Old Red Sandstone’ at Glen Coe, Scotland. Geological Magazine, Vol. 131, 563–566.

WILSON, C J N, HOUGHTON, B F, MCWILLIAMS, M O, LANPHERE, M A, WEAVER, S D, and BRIGGS, R M. 1995. Volcanic and structural evolution of Taupo Volcanic Zone, New Zealand: a review. Journal of Volcanology and Geothermal Research, Vol. 68, 1–28.

WOLFF, J A. 1986. Welded-tuff dykes, conduit closure, and lava dome growth at the end of explosive eruptions. Journal of Volcanology and Geothermal Research, Vol. 28, 379–384.

WRIGHT, A E. 1988. The Appin Group. 177–199 in Later Proterozoic stratigraphy of the Northern Atlantic regions, WINCHESTER, J A (editor). (Glasgow and London: Blackie.)

ZHOU J. 1985. The timing of calc-alkaline magmatism in parts of the Alpine-Himalayan collision zone and its relevance to the interpretation of Caledonian magmatism. Journal of the Geological Society of London, Vol. 142, 309–317.

Appendix Key field localities

The following text and map are provided to help the interested geologist locate and examine key features in the field. They are best used in conjunction with the 1:25 000 scale geological map (British Geological Survey, 2005). UK National Grid references all refer to the 100 km square NN. The brief notes regarding access are for general guidance only.

It is stressed that this appendix is not intended as a field guide. All terrain in the Glencoe area is potentially hazardous and all field visits should be planned accordingly. Much of the terrain is very steep and remote. Serious deterioration in weather conditions can occur with surprising rapidity and streams and rivers can quickly become impassable torrents. Some descent routes are not straightforward and can be long and arduous. No listed locality is perfectly safe and visits to those described in the notes as steep, exposed, or involving a scramble should only be undertaken by persons who are experienced and confident in mountain environments, with due regard for prevailing and predicted weather conditions. Some of the listed localities are unsuitable for large parties and potential leaders are advised to check access in advance, bearing in mind the abilities of likely participants.

Many localities are geological Sites of Special Scientific Interest (SSSIs). The use of hammers and sampling without permission from Scottish Natural Heritage is prohibited by law.

Location map of key field localities (Figure 29).

Locality	Grid reference	Key features	Relevant page, figure or plate	Notes
1	[NN 153 573]	Palaeocanyons in tilted unconformity surface. Conglomerates with boulders over 1 m in diameter include diverse plutonic rocks. Plant remains recovered from siltstone nearby	31–32; 104 (Plate 5)	Exposures along foot of cliffs
2	[NN 138 563]	Leven Schist Formation constitutes local basement beneath volcanic succession	22; 87	Numerous exposures
3	[NN 1399 5598]	Unconformity. Reddened Leven Schist Formation cut by irregular erosion surface and overlain by schist-breccia, sandstone and andesite of Basal Andesite Sill-complex	30	Small exposure beside stream (east bank)
4	[NN 1408 5594]–[NN 1402 5591]	Peperitic top of basaltic andesite sill. Overlying sandstones show soft-state convolution and small-scale faulting. Basal part of succeeding sill shows lobate form and peperite in fractures	33 (Figure 9)	Steep access down to exposure at upper plunge-pool
5	[NN 1398 5516]	Coarse peperite and autobreccia forming top of andesite sill	33 (Plate 6a)(Plate 6b)	Extensive exposure
6	[NN 1420 5538]	Peperitic top of uppermost sill unconformably overlain by sandstone and then accretionary-lapilli-bearing Kingshouse Tuffs	34; 38; 104 (Plate 7); (Plate 10)	Access traverse is in exposed position
7	[NN 139 548]	Kingshouse Tuffs and Lower Etive Rhyolite steepen and are restricted at overturned scarp of Southwestern Graben Fault	27; 49; 87 (Figure 14a)(Figure 14b)	View, partly in steep cliff face
8	[NN 142 548]	Lava-like ignimbrite of Lower Etive Rhyolite overlain by Upper Three Sisters Ignimbrite, which is eutaxitic and includes mesobreccia	69 (Figure 14); (Figure 21) (Plate 19a)(Plate 19b)	Extensive exposures
9	[NN 1428 5455]–[NN 1434 5420]	Andesitic Church Door Buttress Breccias bury trace of Southwestern Graben Fault. Overlying Bidean nam Bian andesite has columnar joints more than 150 m tall	73–74; 78; 83 (Figure 14); (Figure 23)	Steep cliff exposures above scree-lined gully
10	[NN 1475 5497]	Columnar jointing of Bidean nam Bian andesite; jointing indicates single cooling unit about 200 m thick. Flow-banded domains with diverse orientations locally define solidified lava blocks that foundered into fluid magma	78–79 (Plate 22a)(Plate 22b)	Extensive exposures amidst large blocks; steep cliffs to north-east
11	[NN 1546 5632]	Palaeocanyon about 15 m deep incised in Basal Andesite Sill-complex along the line of Ossian Fault. Palaeocanyon filled with several metres of coarse heterolithic conglomerate dominated by andesite clasts and overlain by ignimbrite (Kingshouse Tuffs)	27; 104 (Plate 4); (Plate 14)	Steep access
12	[NN 157 561]	Upper Streaky Andesite vent with steeply inclined streaking and local bodies of heterolithic agglomerate and breccia. Adjacent Lower Three Sisters Ignimbrite contains veins of the andesite and evidence of ductile deformation due to reheating	74; 105 (Plate 14); (Plate 20)	Access from below is difficult. Exposed position
13	[NN 167 556]	Postglacial catastrophic rock-fall deposits	109 (Plate 18)	Access via footpath
14	[NN 1838 5602]–[NN 1808 5688]	Dykes of Etive Swarm intrude Upper Etive Rhyolite (lava-like ignimbrite) and are cut by post-Caledonian monchiquite dyke. Palaeocanyon is floored by pebbly sandstone and intruded by Upper Streaky Andesite. Upper Etive Rhyolite is restricted against Basal Andesite Sill-complex at overturned scarp of the Northeastern Graben Fault	27; 53; 57; 100	Exposures along Allt Lairig Eilde and in the the vicinity of ‘The Study’
15	[NN 2006 5556]	Lower Queen’s Cairn Breccias, including megablocks, rest on Queen’s Cairn Conglomerates that locally occupy steep-sided palaeocanyons cut into Lower Three Sisters Ignimbrite	64; 66 (Figure 17); (Figure 19); (Figure 20)	Steep access
16	[NN 1875 5704]	Lower Queen’s Cairn Breccias rest on Basal Andesite Sill-complex at degraded scarp of Queen’s Cairn Fault. Overlying Upper Three Sisters Ignimbrite buries several stratigraphical units that were exposed in the footwall of the Northeastern Graben Fault-zone	66 (Figure 17); (Figure 20)	Access steep in places
17	[NN 2082 5742]	Monzodiorite fault-intrusion; inner contact is a planar fault surface marked by microbreccias, ultracataclasite and pseudotachylite. Outer contact is generally irregular with fault-rocks preserved locally and with pseudo-tachylite lining veins and blocky joint surfaces. Numerous xenoliths in intrusion, including an 8 m-block of ‘granite’. Fault-intrusion along earlier fault at the summit shows brecciation and hydrothermal alteration, with pseudotachylite in veins and along the inner contact	84; 88–97 (Figure 26) (Plate 25)(Plate 25); (Plate 26a) (Plate 26b) (Plate 26c)	Access from top of Devil’s Staircase (path); direct ascent from the main road is farther and more arduous than it looks
18	[NN 2285 5455]	Sequence of fluvial conglomerate, sandstone and siltstone rests on irregular unconformity surface; conglomerate shows fracture zones parallel to nearby Northeastern Graben Fault-zone. Plant remains recovered from siltstone. Overlying Kingshouse Breccias are substantially restricted at the scarp of a north-east-trending fault; small fault scarp farther south has scarp-foot talus buried by the breccias	34; 38; 40 (Figure 10) (Plate 8a)(Plate 8b)	Beneath and on the south side of the slabs known as the Waterslide
19	[NN 2266 5479]	Kingshouse Breccias resting on metamorphic ‘basement’ and succeeded by Kingshouse Tuffs tuff-cone succession, lithic-rich towards base and becoming better bedded, finer grained and with accretionary lapilli-rich layers towards the top	38; 40–43; 46 (Figure 12) (Plate 11a)(Plate 11b)	Slabby outcrops steepen upwards to exposed positions
20	[NN 211 540]	Extensive Glas Choire sandstones gradually steepen in dip towards the south and south-west, becoming vertical, in response to downsag that accompanied emplacement of the overlying Bidean nam Bian andesites and dacites	76; 79
21	[NN 1644 5431]–[NN 1549 5365]	Thick eutaxitic Lower Dalness Ignimbrite is overlain by tuffaceous sandstone and accretionary lapilli-bearing uppermost part of the Coire nan Easan Tuffs. Intrusive andesite sheet forms a topographic outlier and is cut by numerous dykes of the Etive Swarm	81–82	Access is steep in places
22	[NN 1606 5324]	Fine-grained Coire nan Easan Tuffs interstratified with thin ignimbrites and succeeded by eutaxitic Upper Dalness Ignimbrite	82
23	[NN 1352 5100]	Monzodiorite fault-intrusion with steep contact against volcanotectonic fault, cut by several dykes of Etive Swarm	86 (Figure 25)	Exposures along river
24	[NN 154 516]	Multiple dyke including porphyritic varieties of microdiorite and microgranite (see Bailey, 1960, p.198)	100	Exposure in river bed
25	[NN 204 512]–[NN 198 507]	Etive Dyke Swarm constitutes approximately one third of outcrop; dykes include porphyritic varieties of microdiorite, quartz-microdiorite and microgranite	29; 100	Exposures along River Etive and in stream above Alltchaorunn
26	[NN 239 528]–[NN 240 516]	Rhyolite dykes in flaggy quartzite ‘basement’. Lacustrine facies Kingshouse Tuffs overlain by Etive rhyolites (lava-like ignimbrites) and eutaxitic Three Sisters ignimbrites. Lowermost parts of Bidean nam Bian Andesite Member show pillow forms with intervening siltstone and peperite, recording shallow intrusion into wet caldera-lake-floor sediments	46; 79 (Figure 11) (Plate 12a)(Plate 12b); (Plate 16a)(Plate 16b); (Plate 17)	Scramble up ridge; steep in places
27	[NN 247 521]	Rhyolite and rhyolitic tuff with breccias form irregular dykes that mark a zone of discontinuous fractures in the metamorphic ‘basement’ along the North-eastern Graben Fault-zone	64 (Figure 17) (Plate 17)
28	[NN 244 516]–[NN 241 514]	Lower Three Sisters Ignimbrite thickens dramatically across Chasm step-faults where Upper Etive Rhyolite megablocks show downsag-related detachment on underlying Crowberry Ridge Tuffs. Glas Choire conglomerates in palaeo-canyon include boulders of metamorphic ‘basement’ and granite	60–61; 75–78 (Figure 18); (Figure 24) (Plate 17); (Plate 21a)(Plate 21b)(Plate 21c)
29	[NN 2506 5168]	Crevasse formed in Northeastern Graben Fault-zone contains megabreccia of metamorphic ‘basement’ with mesobreccia-bearing eutaxitic Lower Three Sisters Ignimbrite in upper levels; eutaxitic fabric locally steep and parallel to crevasse wall. Lower Etive Rhyolite originally restricted at fault-scarp shows downsag towards south-east	62; 64 (Figure 17); (Figure 18); 48
30	[NN 2386 5001]–[NN 2388 5076]	Clach Leathad hornblende-biotite monzogranite; contacts of pluton with volcanic succession define steep wall and shallow-dipping roof	98–100	Magnificent views
31	[NN 178 477]	Meall Odhar monzogranite cuts and alters dykes of Etive Swarm, gives rise to a micromonzogranitic dyke and is cut by late dykes of Etive Swarm	98–100

Figures, plates and tables

Figures

(Figure 1) Topography of the Glencoe area.

(Figure 2) Simplified map, generalised succession and cross-section showing the geology of the Glencoe area. The cross-section is drawn as if viewed looking towards the south, which is the view seen southwards from the main road (A82T) travelling west from the vicinity of the Kingshouse Hotel [NN 26 54] to the lower end of Glen Coe [NN 12 56]. See p.5 for key.

(Figure 3) Distribution of Siluro-Devonian volcanic and plutonic rocks showing faults that were active during the magmatic activity.

(Figure 4) (right) Main features of calderas, caldera-forming eruptions and the associated phenomena. a. Valles caldera, Jemez Mountains, New Mexico, USA (image generated by H P Foote; geology from USGS map I-571, 1970). The main topographical depression in the summit of the volcano is the caldera. This formed via two large-scale explosive eruptions between 1 and 1.5 million years ago. The entire volcano records some 13 million years of activity. The caldera wall shows degradation by collapse, with a typically scalloped form and with wedges of collapse breccia forming part of the caldera fill. Ignimbrites emplaced during the caldera-forming eruptions form fans on the outer flanks of the volcano and a large part of the fill in the caldera. Post-caldera resurgence of magma into the volcano has caused the intracaldera ignimbrites to be forced upwards, forming a central resurgent dome, with an extensional graben across its apex, and a discontinuous ring of vents with lava flows. b. Sketch illustrating eruption within a multi-subsidence, piecemeal caldera. Hypothetical volcano illustrating a large-scale eruption with associated progressive deposition of ignimbrite from the base of the pyroclastic current and collapse of developing volcanotectonic fault scarps. Downsag with related extensional opening of crevasses is depicted for the ongoing eruption. The diagram illustrates how the complexity of an early stage of caldera collapse, represented by the lower of the two intracaldera ignimbrites, can by obscured by burial. The Glencoe Caldera-volcano Complex records seven caldera-forming eruptions with deposition of intracaldera ignimbrites; most involved both downsag and piecemeal volcanotectonic faulting.

(Figure 5) Models of cauldron subsidence. a. The original models of cauldron subsidence derived from studies at Glen Coe (modified after Clough et al., 1909). b. Model of asymmetrical subsidence of a coherent caldera-floor block (after Roberts, 1974). Note the depiction of pronounced inward dip (downward convergence) of the bounding faults. This geometry is implausible for straightforward central-block subsidence and does not occur in reality.

(Figure 6) Structural cross-sections through Dalradian rocks along the classic Loch Leven section, just to the north of the Glencoe Caldera–volcano Complex. a. After Treagus (1974). b.After Hickman (1978). BSL Ballachulish Slide; BS Ballachulish Syncline; BWA Blackwater Antiform; BWS Blackwater Synform; KA Kinlochleven Anticline; MA Mamore Antiform; SBS Stob Ban Synform; TS Treig Syncline

(Figure 7) Major structural features of the Glencoe Caldera-volcano Complex, highlighting intrusions and faults active during volcanism (Etive Dyke Swarm not shown).

(Figure 8) Outcrop of the Basal Andesite Sill-complex relative to the structural framework.

(Figure 9) Vertical section of the peperitic upper contact of one of the basal andesite sills, below Aonach Dubh [1408 5594]. The vesicles were formed in the sandstone by gen- eration of steam in the original wet sediment, and, together with the in-situ swarms of andesite grains formed by magma-water interaction within the sedimentary host, are evidence that the magma was intrusive.

(Figure 10) (right) Stob Dearg [22 54] (original drawings by K Lancaster) (P611773). a and b) Viewed towards the south-west, showing lower elements of the intracaldera stratigraphy and the cross-cutting rhyolite interpreted as infilling the vent from which the Upper Etive Rhyolite was erupted. c. Viewed towards the north-west, showing the rhyolite-filled vent of the Upper Etive Rhyolite and thickening of the Lower Three Sisters Ignimbrite across the Chasm step-fault system. Views b and c both show the restriction of the Kingshouse Breccias at palaeofault scarps that trend north- eastwards, at right angles to the Chasm step-fault system and Glencoe Graben. In view b, three broad V-shaped outcrops of breccia and conglomerate (labelled p) are remnants of a single infilled palaeocanyon that was cut into the Lower Etive Rhyolite parallel to the Glencoe Graben. The outcrops are remnants of the south- west side of the palaeocanyon, which must have trended north-westwards, parallel to the cliff faces (despite their appearance, they are not complete cross-sections of three separate features). 1 Central Buttress; 2 North Buttress; 3 Great Gully Buttress; 4 Broad Buttress; 5 Tulaich Buttress; 6 Waterslide; CRT Crowberry Ridge Tuffs; DA Dalradian metamorphic basement; KHB Kingshouse Breccias; KHT Kingshouse Tuffs; LER Lower Etive Rhyolite; LTS Lower Three Sisters Ignimbrite; MER Middle Etive Rhyolite; RGT Raven’s Gully Tuffs; UER Upper Etive Rhyolite; USA Upper Streaky Andesites; UTS Upper Three Sisters Ignimbrite

(Figure 11) Kingshouse Tuffs (KHT). a Distribution. b Schematic diagram showing a north-west section parallel to the axis of the Glencoe Graben, and the setting of the Kingshouse Tuffs eruption and related deposits. This site was located at the intersection of the Glencoe Graben and a cross-graben between the Glen Etive Fault and the Devil’s Staircase Fault; thus it was tectonically controlled, as were the alluvial fans of the Kingshouse Breccias (KHB). BAS Basal Andesite Sill-complex; DA Dalradian metamorphic basement; DSF Devil’s Staircase Fault; GEF Glen Etive Fault; NEGF Northeastern Graben Fault; OF Ossian Fault; QCF Queen’s Cairn Fault; WCF White Corries Fault

(Figure 12) (right) Kingshouse Tuffs (KHT) at North Buttress on Stob Dearg [NN 2270 5476]. a. Schematic log of the sequence that is part of the outer flank of a tuff cone. It shows an upward change from predominantly massive breccia with abundant lithic fragments to finer grained tuffs characterised by planar and low-angle cross-stratification with layers rich in accretionary lapilli. This section records phreatomagmatic activity transitional from early vent-clearing explosions to later development of a tall eruption column that produced energetic pyroclastic density currents and substantial ash fall. b Cross-stratified silicic tuffs showing low-amplitude bedforms and low-angle scour surfaces recording the passage of highly unsteady, energetic pyroclastic density currents (P611782). c Massive to weakly bedded tuffaceous lithic breccias and lithic-rich tuff formed from initial, vent-clearing, wet explosions; the lithic fragments are quartzite fragments from the underlying Kingshouse Breccias (P611783).

(Figure 13) Lower Etive Rhyolite distribution and key features.

(Figure 14) Schematic cross-section illustrating ignimbrites and breccias restricted at and near the Southwestern Graben Fault (zone) in Coire nam Beitheach [NN 139 547], north-west of the Queen’s Cairn Fault. Original near-vertical fault scarps that ponded the Lower Etive Rhyolite (LER) have been rotated by downsag towards the Glencoe Graben. Blocks and megablocks of Lower Etive Rhyolite have been incorporated at several horizons within the ponded Upper Three Sisters Ignimbrite. Overlying Church Door Buttress Breccias include mesobreccias that were shed from the Southwestern Graben Fault and show evidence of loading into hot ignimbrite; andesite-dominated breccias higher in the section were shed from scarps cutting the Basal Andesite Sill-complex farther to the south-west.

(Figure 15) Vertical section through the Kingshouse Tuffs and Lower Etive Rhyolite, north-west of the Ossian Fault, on G-Buttress, Aonach Dubh [NN 143 554].

(Figure 16) (left) Glencoe Graben: schematic diagrams showing a north-west to south-east section parallel to the axis of the graben and the contrasting eruptions of the Upper Etive Rhyolite and Lower Three Sisters Ignimbrite. a. The three Etive rhyolite eruptions formed mainly lava-like ignimbrites from pyroclastic fountains at the sites of tuff-cones that were active during the initial phreatomagmatic explosivity. The vents were located centrally, at the intersection of the Glencoe Graben and the cross-graben bounded by the Devil’s Staircase and Glen Etive faults. They probably formed in or close to the Chasm Fault system, on the north-east side of the main graben. b. The Lower Three Sisters Ignimbrite appears to have been erupted from the Northeastern Graben Fault (zone) at a location towards the south-east of the volcano complex; its emplacement involved both volcanotectonic faulting and downsag. Buoyant-convective ash plumes that would have risen above both the vent and the pyroclastic currents are omitted for clarity. BAS Basal Andesite Sill-complex; CRT Crowberry Ridge Tuffs; DA Dalradian metamorphic basement; DSF Devil’s Staircase Fault; GEF Glen Etive Fault; KHB Kingshouse Breccias; KHT Kingshouse Tuffs; LER Lower Etive Rhyolite; LTS Lower Three Sisters Ignimbrite; NEGF Northeastern Graben Fault; OF Ossian Fault; QCF Queen’s Cairn Fault; RGT Raven’s Gully Tuffs; UER Upper Etive Rhyolite; WCF White Corries Fault

(Figure 17) Distribution of the Three Sisters ignimbrites, associated sedimentary rocks and contemporary volcanotectonic features. LQB Lower Queen’s Cairn Breccias; LTS Lower Three Sisters Ignimbrite; QCC Queen’s Cairn Conglomerates; UTS Upper Three Sisters Ignimbrite

(Figure 18) Schematic cross-section of the Lower Three Sisters Ignimbrite and its contact relationships transverse to the Glencoe Graben, south-east of the Glen Etive Fault. 1 Mesobreccias and megabreccias shed from fault scarps of the Etive rhyolites during early phases of caldera subsidence in the course of the Lower Three Sisters eruption. 2 Eutaxitic tuff with mesobreccia layers showing onlap and fanning dips that respectively record progressive aggradation of the ignimbrite and progressive downsag during the course of the eruption. 3 Extensional crevasse formed along the Northeastern Graben Fault to accommodate the downsag towards the south-west.

(Figure 19) Texture of coarse-grained lithofacies of the Lower Queen’s Cairn Breccias showing near jigsaw-fitting of clasts and sparse matrix of similar material, which together indicate fragmentation at a late stage of transport. Such textures in sedimentary deposits are characteristically formed in debris avalanches.

(Figure 20) (right) Schematic sections north-west to south-east parallel to the axis of the Glencoe Graben. a. The tectonic development that led to emplacement of the Lower Queen’s Cairn Breccias (LQB). The breccias are banked against a retrogressively degraded scarp of the Queen’s Cairn Fault and spread to the south and south-east across the alluvial surface formed of Queen’s Cairn conglomerates, sandstones, and siltstones (QCC).b. The ensuing eruption that formed the Upper Three Sisters Ignimbrite (UTS) and led to volcano-tectonic subsidence involving both faulting and downsag. Buoyant-convective ash plumes that would have risen above both the vent and the pyroclastic currents are omitted for clarity.BAS Basal Andesite Sill-complex; CRT Crowberry Ridge Tuffs; DA Dalradian metamorphic basement; DSF Devil’s Staircase Fault; GEF Glen Etive Fault; KHB Kingshouse Breccias; KHT Kingshouse Tuffs; LER Lower Etive Rhyolite; LTS Lower Three Sisters Ignimbrite; NEGF Northeastern Graben Fault; OF Ossian Fault; QCF Queen’s Cairn Fault; RGT Raven’s Gully Tuffs; UER Upper Etive Rhyolite; UTS Upper Three Sisters Ignimbrite; WCF White Corries Fault

(Figure 21) Schematic log of the Upper Three Sisters Ignimbrite in Coire nam Beitheach [NN 1414 5473] showing stratified mesobreccias towards the base and, at around 65 m, the characteristic increase from moderate to dense welding picked out by intensification of the eutaxitic fabric (flattening of fiamme). The stratigraphical levels of the ignimbrite shown in (Plate 19a), (Plate 19b) are also located.

(Figure 23)." data-name="images/P988028.jpg">(Figure 22) Locations of outcrops of the Upper Queen’s Cairn Breccias (UQB), the Church Door Buttress Breccias (CDB), and the diverse sedimentary rocks that infill the Glas Choire palaeocanyon and overlie a fluvially eroded surface to the north-west (Glas Choire Sandstone Member; GCS). The inset box shows the location of the detail provided in (Figure 23).

(Figure 23) Locations and key features of individual elements of the Church Door Buttress Breccias and of the distal Glas Choire alluvial deposits, in the north-west of the caldera-volcano complex (see (Figure 14) and (Figure 23)." data-name="images/P988028.jpg">(Figure 22)).

(Figure 24) Alluvial architecture of the fill within the Glas Choire palaeocanyon.a. Schematic section of the Glas Choire palaeocanyon fill, showing four groups of stacked and infilled channels, with overbank siltstone that lies on the shoulder of the palaeocanyon. The uppermost siltstones are lacustrine in origin. b. Detail of the alluvial architecture: vertical scale is approximately 2 the horizontal scale.

(Figure 25) Ring-fault system, associated fault-intrusions and the passively emplaced Clach Leathad Pluton.

(Figure 26) (left) Simplified conceptual model proposed to explain occurrences at Glen Coe of fault-intrusions with planar inner (caldera-side) contacts and extremely irregular outer contacts (see text). a. This shows the setting of a hypothetical hydrothermal system at considerable depths (at least several hundreds of metres) where it will be intersected by a dilating caldera-fault plane. The figure shows a potential releasing bend in the incipient fault plane, although in reality dilatation may be more general where the fault dips outwards or where it is more irregularly curved and juxtaposes parts with different curvatures. The depicted isolation of the hydrothermal system is a diagrammatic simplification. b. The hydrothermal system is shown here as a localised network of fractures containing superheated water under high confining pressure. In reality such a hydrothermal reservoir would probably have greater vertical extent and be connected to both meteoric and magmatic water sources at depth, and to fumaroles at the surface. X0 is an arbitrary reference point in the block that subsides. c. The proposed immediate effect of rapid subsidence on the caldera fault (X⁰ to X¹): rapid dilatation causes the hydrothermal system to explode, via transformation of superheated liquid water to vapour and vigorous expansion of vapour. Such processes would most probably be followed rapidly by ascent of fragmented melts from depth, these too having been disrupted by volatile exsolution and expansion due to pressure relief. d. The final form of the opposed contacts of the fault-intrusion, as seen at outcrop at Glen Coe: the overall shape and relative dimensions of the intrusion are hypothetical. The vertical distance moved by the reference point (X⁰ to X²) would be of the order of several hundreds of metres and the duration of that subsidence a matter of only hours or a few days.

(Figure 27) Relative timing of events affecting the Glencoe area over the last 40 000 years (or since the last Glacial Maximum). The graph on the left shows the oxygen isotope record from the Greenland (GISP) ice-core, a proxy for Northern Hemisphere temperature.

(Figure 28) A reconstruction of the Glencoe glacier approximately 12 000 years ago, during the Loch Lomond Stadial. View towards south-west.

Plates

(Plate 1) Satellite view showing the location of the Glencoe area in Scotland. BGS enhanced image © NERC, 2005. Grid lines in white show latitude and longitude; National Grid is indicated along the margin of the image.

(Plate 2a) Welded ignimbrite. a. At outcrop: the black lenticular fiamme represent pumice fragments that collapsed within the hot- state compacted ash matrix. Locally the flattening fabric is wrapped around a rock fragment (just below centre) that was rigid (P611765).

(Plate 2b) Welded ignimbrite. b. Under the microscope: a eutaxitic texture can be seen. Brown to colourless glass shards are strongly flattened and wrapped in a ductile fashion around a rigid crystal of quartz (centre). Field of view is 4 mm wide: plane-polarised light (P611766).

(Plate 3) Basal Andesite Sill-complex with overlying Kingshouse Tuffs and stratified Lower Etive Rhyolite, viewed looking east-south-east towards Aonach Dubh [NN 15 56]; the valley to the left is Glen Coe and the summit to the right is Stob Coire nan Lochan (P611767). BAS Basal Andesite Sill-complex; DA Dalradian metamorphic basement; KHT Kingshouse Tuffs; LER Lower Etive Rhyolite

(Plate 4) Basal Andesite Sill-complex, Kingshouse Tuffs and Lower Etive Rhyolite cut by strands of the Ossian Fault in the north face of Aonach Dubh [NN 15 56]. A palaeocanyon filled with conglomerate overlain by ignimbrite is located on the trace of the right-hand fault strand and is part of the extensive unconformity surface that cuts the sill-complex. The Lower Etive Rhyolite thickens considerably to the right (north-west) across both fault strands; the right-hand strand shows reactivation in the opposite sense (down to the south-east) (P611768). BAS Basal Andesite Sill-complex; KHB Kingshouse Breccias; KHT Kingshouse Tuffs; LER Lower Etive Rhyolite; UER Upper Etive Rhyolite

(Plate 5) Basal Andesite Sill-complex (BAS) and unconformity on Dalradian metamorphic basement (DA) including palaeocanyons on north side of Glen Coe [NN 15 57]. The sills and unconformity dip south- wards and are cut by a strand of the ring-fault system across the crest of the Aonach Eagach (serrated ridge at middle left) and beyond it on the north side. This dip away from the fault towards the inside of the caldera-volcano complex, and similar geometry elsewhere, was originally interpreted as due to subsidence between inward-dipping faults, but is now interpreted as early caldera downsag (see p.87) (P611769).

(Plate 6a) Basal Andesite Sill-complex in Coire nam Beitheach [NN 139 551]. a. Peperitic and autobrecciated top of a sill near the top of the sill stack (P611770).

(Plate 6b) Basal Andesite Sill-complex in Coire nam Beitheach [NN 139 551]. b. Detail showing jigsaw fit of andesite fragments, due to in situ brecciation, within a purple, homogeneous fine-grained sandstone host. Drink can is 6 cm in diameter (P611771).

(Plate 7) Peperitic top of the uppermost sill of the Basal Andesite Sill-complex (BAS), unconformably overlain by thinly bedded sandstone (KHB) that locally forms the base of the Etive Rhyolite Member. This exposure, high on the very steep buttressed slopes north-east of Coire nam Beitheach [142 554], estab- lishes the intrusive nature of the entire succession of andesitic sheets in this vicinity. The absence of coeval andesitic lavas here, together with several other lines of evidence (see text), is taken to indicate that the unconformity could represent a time break of the order of hundreds of thousands of years, or more (P611772).

(Plate 8a) Psammite breccias. a Psammite talus-breccia at the foot of a small fault scarp (out of view to left), overlain by poorly sorted psammite breccias of the Kingshouse Breccias, near the Waterslide on Central Buttress, Stob Dearg [NN 2282 5440] (P611774).

(Plate 8b) Psammite breccias. b Massive to weakly stratified psammite breccias with cross- stratified quartzose sandstone infilling a scour in the Kingshouse Breccias, at the Waterslide [NN 2284 5454] (P611775).

(Plate 9) Poorly sorted breccia and conglomerate (Kingshouse Breccias) dominated by clasts of andesite (purple) and of Rannoch Moor granite (pink), with interca- lated sandstones. These deposits, which show imbrication of the large clasts, form part of a small wedge mainly of unstratified breccia that extends south- westwards from the Northeastern Graben Fault in Cam Ghleann [NN 2484 5181] (P611776).

(Plate 10) Parallel-bedded distal Kingshouse Tuffs, 7 km from the vent, showing layers rich in small accretionary lapilli (centre); F Buttress on Aonach Dubh [NN 1432 5550] (P611777).

(Plate 11a) Indicators of the position of the vent of the Kingshouse Tuffs (KHT). a. Asymmetrical ballistic-bomb impact sag in bedded and cross-stratified silicic tuffs, indicating that the vent was located to the west or south-west (away from the viewer). The impacted beds lie on a surface scoured by the passage of one or more pyroclastic density currents; Great Gully Buttress, Stob Dearg [NN 226 548] (P611778).

(Plate 11b) Indicators of the position of the vent of the Kingshouse Tuffs (KHT). b. Cross-stratified silicic tuffs showing five scour (truncation) surfaces (indicated by arrows); the strata record deposition and migration of low-amplitude ash dunes down-current from the south-west (right to left), alternating with erosion. The uppermost scour surface is locally steep and here was plastered by damp ash, also indicating that the current direction was from the south-west. The dip of the beds, which is towards the south-west, resulted from caldera downsag; Broad Buttress, Stob Dearg [NN 225 549] (P611779).

(Plate 12a) Aqueous litho- facies of the Kingshouse Tuffs. a. View towards the south-west showing the flanks of Sròn na Creise (buttress on right hand side) and Stob a’ Ghlais Choire (middle summit) with an extensive outcrop of the Kingshouse Tuffs, including the thick lacustrine succession formed in a small fault-bounded basin (P611780).

(Plate 12b) Aqueous litho- facies of the Kingshouse Tuffs. b. Flame structures and convolute laminations in silicic tuffs deposited from aqueous suspension (parallel-stratified tuffs) and by turbidity currents (mainly massive division showing near-vertical water-escape structures near its middle); Sròn na Creise [NN 241 525] (P611781).

(Plate 13a) Lower Etive Rhyolite with sparse feldspar crystals; Stob Dearg [NN 220 552]. Although texturally similar to rhyolite lava at outcrop, the overall field relationships indicate an explosive, pyroclastic origin, so that the rock is referred to as lava-like ignimbrite. a Fine parallel flow-lamination (P611784).

(Plate 13b) Lower Etive Rhyolite with sparse feldspar crystals; Stob Dearg [NN 220 552]. Although texturally similar to rhyolite lava at outcrop, the overall field relationships indicate an explosive, pyroclastic origin, so that the rock is referred to as lava-like ignimbrite. a Fine parallel flow-lamination (P611784). b Convolute flow-lamination and a rare lithic fragment (arrow) (P611785).

(Plate 14a) North face of Aonach Dubh [NN 15 56] and northern side of Glen Coe [NN 16 57]. a. The Upper Etive Rhyolite (UER) becomes thicker across one strand of the Ossian Fault (OF) due to syn-eruptive down-to-the-south-east (to the left) movement; this is opposite to the offset that formed during eruption of the Lower Etive Rhyolite (LER). Ponding of the Lower Three Sisters Ignimbrite (LTS) and thickness change of the Upper Three Sisters Ignimbrite (UTS) are also evident across this fault strand (P611786). BAS Basal Andesite Sill-complex; KHT Kingshouse Tuffs; USA Upper Streaky Andesites

(Plate 14b) North face of Aonach Dubh [NN 15 56] and northern side of Glen Coe [NN 16 57]. b. Traces of the Northeastern Graben Fault (NEGF) and its footwall scarp along the north side of the Pass of Glencoe [NN 16 57]. The scarp, composed of Basal Andesite Sill-complex (BAS), formed a volcanotectonic topographical barrier during emplacement of the Upper Etive Rhyolite (UER), which is ponded against it, as well as forming a subterranean barrier during intrusion of the Lower Streaky Andesites sill (LSA-sill) within the Glencoe Graben. Lower Streaky Andesites lavas, which form much of the ridge crest from Am Bodach to the Aonach Eagach, were extruded onto the footwall block outside of the graben. The talus cone in the lower left of the view is the largest and most active in the vicinity (P611787). CRT Crowberry Ridge Tuffs; OF Ossian Fault

(Plate 15a) Lower Streaky Andesites on Gearr Aonach and Aonach Dubh. A. Typical exposure of the Lower Streaky Andesites showing purple andesite streaked with rhyolite. Both rock types contain up to 5 per cent of plagioclase phenocrysts; Gearr Aonach [NN 167 561] (P611788).

(Plate 15b) Lower Streaky Andesites on Gearr Aonach and Aonach Dubh. b. Two of the Three Sisters, Gearr Aonach (left) and Aonach Dubh (right) showing dramatic thickness variation of a Lower Streaky Andesites sill: viewed towards the west. The sill is about 100 m thick on Gearr Aonach and thins to only a few metres over a distance of less than 750 m away from the viewer, adjacent to the Upper Streaky Andesites vent on Aonach Dubh. The sill is the same intrusion as seen in Plate 14b (P611789). BAS Basal Andesite Sill-complex; CRT Crowberry Ridge Tuffs; KHT Kingshouse Tuffs; LER Lower Etive Rhyolite; LSA Lower Streaky Andesites; LTS Lower Three Sisters Ignimbrite; UER Upper Etive Rhyolite; USA Upper Streaky Andesites; UTS Upper Three Sisters Ignimbrite

(Plate 16a) Lower Three Sisters Ignimbrite showing well-developed eutaxitic texture (welding), with moderately abundant fiamme and small lithic clasts. The contrast between the non-silicified matrix (a) and the silicified matrix (P611790)

(Plate 16b) Lower Three Sisters Ignimbrite showing well-developed eutaxitic texture (welding), with moderately abundant fiamme and small lithic clasts. The contrast between the non-silicified matrix (b) records variable hydrothermal alteration; Sròn na Creise [NN 239 521] (P611791).

(Plate 17) East flanks of Sròn na Creise and Stob a’ Ghlais Choire showing slumping of the Upper Etive Rhyolite (UER) towards the Chasm step-fault system and related thickening of the Lower Three Sisters Ignimbrite (LTS) towards the axis of the Glencoe Graben. Also shown are breccia dykes in the Dalradian metamorphic basement (DA) along the trace of Northeastern Graben Fault, the faulted margin of the lacustrine facies of the Kingshouse Tuffs (KHT; extreme right), a topographical depression eroded deeply into Lower Etive Rhyolite (LER) and partially draped by Crowberry Ridge Tuffs (CRT), the Glas Choire Sandstone Member (GCS) in the vicinity of its type locality, and the overlying Bidean nam Bian Andesite Member (BBA) (P611792).

(Plate 18) Gearr Aonach viewed from the east across Coire Gabhail [NN 16 55] showing ponding of the Lower Three Sisters Ignimbrite (LTS) within a downsag north-west of the Queen’s Cairn Fault. The ignimbrite is up to 50 m thick along the downsag axis (which trends north-west along the Glencoe Graben) and it thins progressively towards the south-western graben hinge, as does the overlying sill of the Upper Streaky Andesites (USA). The Upper Three Sisters Ignimbrite (UTS) shows similar thinning relationships, but overlaps the hinge line. The large composite debris cone in the foreground was formed by catastrophic rockfall following deglaciation (see p.109) (P611793). LSA Lower Streaky Andesites; UER Upper Etive Rhyolite

(Plate 19a) Upper Three Sisters Ignimbrite in Coire nam Beitheach [NN 142 548]. a. Stratification and inverse grading of lithic tuff at the base of the ignimbrite; the substrate is Lower Etive Rhyolite (right-hand side). Scale intervals are 5 cm (P611794).

(Plate 19b) Upper Three Sisters Ignimbrite in Coire nam Beitheach [NN 142 548]. b. Mesobreccia layer including large matrix-supported blocks derived from the Basal Andesite Sill- complex (located with arrows) and Lower Etive Rhyolite. (See also Figure 14) and (Figure 21) (P611795).

(Plate 20) Upper Streaky Andesites agglomerate forming part of the infill of the vent exposed in the north-eastern shoulder of Aonach Dubh (see (Plate 14a) and (Plate 15b). The various angular lithic fragments are intensely altered (P611796).

(Plate 21a) Glas Choire palaeocanyon, infills of sandstone and conglomerate, with overbank siltstones. a Undulose-stratified pebbly sandstone within the palaeocanyon [NN 241 514] (P611797).

(Plate 21b) Glas Choire palaeocanyon, infills of sandstone and conglomerate, with overbank siltstones. b Cobbles and boulders of metamorphic basement and Rannoch Moor granite in poorly sorted conglomerate that infills channels within the palaeocanyon [NN 241 514] (P611798).

(Plate 21c) Glas Choire palaeocanyon, infills of sandstone and conglomerate, with overbank siltstones. c Planar-bedded and laminated siltstone and fine-grained sandstone showing normal grading and loading-related soft-sediment deformation, interpreted as overbank deposits; Stob a’ Ghlais Choire [NN 2418 5168] (P611799).

(Plate 22a) and b Bidean nam Bian Andesite Member at Stob Coire nan Lochan [NN 148 549]; this columnar jointing persists through more than 200 m thickness with no apparent discontinuity, so that the sheet appears to be a single cooling unit (P611800) and (P611801).

(Plate 22b) a and b Bidean nam Bian Andesite Member at Stob Coire nan Lochan [NN 148 549]; this columnar jointing persists through more than 200 m thickness with no apparent discontinuity, so that the sheet appears to be a single cooling unit (P611800) and (P611801).

(Plate 23) View (towards the north-west) illustrating the trace of the ring-fault and fault-intrusions between Cam Ghleann (foreground) and Stob Mhic Mhartuin. The western outcrop of the Rannoch Moor Pluton is also seen, and part of the Northeastern Graben Fault-zone is represented by the breccia dykes that cut the Dalradian metamorphic basement. Pleistocene till with patches of peat and alluvium cover much of the lower ground, which lies about 250 m above sea level (P611802).

(Plate 24) The ring-fault at An t-Sròn and on Stob Coire nam Beith (background) (viewed towards the south) is traceable in 1 km of vertical relief and shows at least 500 m of vertical displacement of the volcanic succession and Basal Andesite Sill-complex. The deep gully formed along the fault is known as The Chasm of An t-Sròn; the other prominent gully to the right marks a related fracture that lies along a planar projection (dotted line) of the ring-fault plane that traces through the lower ground. (A complementary fracture that is a planar northwards continuation of ring-fault plane in An t-Sròn occurs on the valley side behind and to the right of the viewer) (P611803).

(Plate 25a) Main Fault at Stob Mhic Mhartuin [NN 2082 5742]. a The fault-plane exposed here dips outwards (away from the volcano complex) at about 63°; view is towards the south-east (P611804).

(Plate 25b) Main Fault at Stob Mhic Mhartuin [NN 2082 5742]. b Detail of the Main Fault zone (see a). The banded breccias are some 25 cm thick and show a general increase in microbrecciation and streaking with flinty crush-rock towards the main band of crush-rock and the porphyritic rhyolite. Despite the seemingly straightforward succession of zones, detailed study shows that there has been substantial mixing of components (P611805).

(Plate 26a) Photomicrographs of polished thin sections of flinty crush-rock and porphyritic rhyolite in the ring-fault zone at Stob Mhic Mhartuin [NN 2082 5742]. Fragments of quartz appear white and feldspar phenocrysts (f ) are indicated. a. Textures recording intimate fluid-state interlamination and cross-mixing of solids between melts that formed flinty crush-rock (dark and prevalent on right-hand side) and porphyritic rhyolite (pale and prevalent on left-hand side). Fragments of quartz, appearing white and with smaller grains showing rounding, occur mainly in the crush-rock component but are also embedded in the rhyolite (far left). Feldspar phenocrysts of the rhyolite are heavily altered and broken; an original cluster appears to have been attenuated into the flinty crush-rock by laminar flow (middle). Field of view is 3 mm wide: plane-polarised light (P612385).

(Plate 26b) Photomicrographs of polished thin sections of flinty crush-rock and porphyritic rhyolite in the ring-fault zone at Stob Mhic Mhartuin [NN 2082 5742]. Fragments of quartz appear white and feldspar phenocrysts (f ) are indicated. b. Textures recording various degrees of mingling of original melts. The flinty crush-rock component predominates on the left-hand side (dark with numerous quartz fragments appearing white) and contains isolated feldspar crystals of uncertain origin. Rhyolite forms the pale streak in the middle, which is flanked by intimately mingled (finely interlaminated) rhyolite and crush-rock. Fragmented quartzite forms the bright band on the right-hand side. Field of view is 3 mm wide: plane-polarised light (P612386).

(Plate 26c) Photomicrographs of polished thin sections of flinty crush-rock and porphyritic rhyolite in the ring-fault zone at Stob Mhic Mhartuin [NN 2082 5742]. Fragments of quartz appear white and feldspar phenocrysts (f ) are indicated. c Two lithic fragments (arrowed) of granophyric quartz–K-feldspar intergrowths contained in groundmass of (porphyritic) rhyolite; closely similar granophyric textures occur in a nearby xenolith of granite enclosed in the fault-intrusion. Field of view is 3 mm wide: cross-polarised light (P612387).

(Plate 27) Glen Coe, looking westwards. Note the U-shaped cross profile of this classic glacial trough (P000731).

(Rear cover)

(Front cover) Glen Coe viewed towards the south-west from The Study (Photographer: B P Kokelaar) (P611763).

(Frontispiece)Viewed due east from the summit of Bidean nam Bian (1150 m), the successive ridges of Beinn Fhada, Buachaille Etive Beag and Buachaille Etive Mòr provide serial sections through the volcanic succession of the Glencoe caldera volcano. The prominent summit in the distance (middle) is of Stob Dearg (1022 m), which presents remarkable exposures of three successive volcanic cones formed during powerful explosive interactions of magma with water. Beyond Stob Dearg lies the desolate Rannoch Moor, mostly underlain by a granitic pluton that was unroofed just before the volcanism at Glen Coe (Photographer: B P Kokelaar) (P611764).

Tables

(Table 1) Stratigraphy of the Neoproterozoic Dalradian Supergroup in the area of Glen Coe (after Bailey, 1960; Treagus, 1974; Hickman, 1975). Thickness data, from Hickman, are maxima for the area.

(Table 2) Lithostratigraphical and lithodemic nomenclature used in the Glencoe Caldera-volcano Complex.

Tables

Table 1 Stratigraphy of the Neoproterozoic Dalradian Supergroup in the area of Glen Coe (after Bailey, 1960; Treagus, 1974; Hickman, 1975).

Thickness data, from Hickman, are maxima for the area.

Group	Subgroup	Formation	Member	Lithology
Appin		Ballachulish Slate 400 m		Pelite, black; here mostly schistose in tectonic slices.
	Ballachulish	Ballachulish Limestone 250 m		Dolomitic metalimestone, dark grey, impure schistose calcsilicate rock; intercalations of black pelite.
	Lochaber	Leven Schist 1500–3000 m		Upper part pelite and semipelite, grey-green, phyllitic to schistose; some calcareous strata near top. Lower part psammite and semipelite, dark, schistose, slightly graphitic; some quartzite; lower junction gradational.
		Loch Treig Schist and Quartzite	Glen Coe Quartzite 400 m	Very feldspathic, well-bedded, cross-bedded units up to 4 m thick, strongly slumped. Intercalations of grey-green schist near top. Lower junction characterised by K-feldspar pebble beds and mud flakes.
			Binnein Schist 400 m	Semipelite and pelite,schistose, with rare metacarbonate, calcsilicate and graphitic beds; quartzite ribs at both junctions.
			Binnein Quartzite 300 m	Well-bedded, less feldspar but more muscovite than other quartzites; characteristically white weathering.
			Eilde Schist 400 m	Psammite and semipelite, schistose, distinct graded bedding, graphitic seams throughout. Quartzite ribs at both junctions.
			Eilde Quartzite 600 m	Feldspathic, well-bedded. Similar to Glencoe Quartzite, but thinner cross-bedded units and feldspar pebbles less conspicuous. Intercalated schists at top.
Grampian	Glen Spean	Eilde Flag 1350 m+		Feldspatic quartzite and psammite, micaceous, wellbedded,flaggy; minor schistose semipelite. Upper junction marked by 5 m of K-feldspar pebble beds and slump folding.