Quaternary geology of the Outer Hebrides

By J. D. Peacock

Bibliographical reference: Peacock, J. D. 1984. Quaternary geology of the Outer Hebrides. Report British Geological Survey, Vol. 16, No. 2

British Geological Survey Natural Environment Research Council

Quaternary geology of the Outer Hebrides

Author J. D. Peacock, BSc, PhD, FRSE, British Geological Survey, Murchison House, West Mains Road, Edinburgh EH9 3LA

London Her Majesty's Stationery Office 1984. © Crown copyright 1984 ISBN 0 11 884353 2. Printed in England for Her Majesty's Stationery Office by Commercial Colour Press, London E7

(Front cover)

(Rear cover)

Quaternary geology of the Outer Hebrides—summary

The earliest recorded event of probable Pleistocene age in the Outer Hebrides was the formation of a raised platform and cliff of marine erosion of which remnants are found only in north and east Lewis. This was followed by a glacial phase tentatively correlated with either the Wolstonian stade, or a Devensian event pre-dating the Late Devensian, during which ice from the Scottish mainland crossed the extreme north of Lewis and probably also the more southerly islands. Following a period of cold climate with evidence of frost and possible solifluction, the tills formed during this phase were partly reworked by the sea to form a raised gravel beach of which remains occur in north Lewis and Barra. During a subsequent episode thought to represent the Late Devensian maximum, the islands were glaciated chiefly by local ice, with outward flow from an elongated ice dome near the west coast of Barra and the Uists, another over Harris and south-west Lewis and perhaps a third over the low plateau of central northern Lewis. Ice from the Scottish mainland may have been in contact at this time with the local ice near the east coast of Lewis from the Eye Peninsula northwards, with a tongue of mainland ice extending on shore in the extreme north. The local Late Devensian ice failed to cross the extreme north-west of Lewis. The retreat of the local Hebridean ice-sheet was followed by a vigorous valley glaciation which resembled in some of its landforms the Loch Lomond Readvance of the Scottish mainland, but may have predated it.

Till, chiefly lodgement till, is extensive in central Lewis below the peat, and interbedded deposits including possible subaerial and subaqueous flow-tills occur in the north of the island. In South Harris and the Uists there are areas of moundy till formed of little-moved rock fragments. Hummocky moraines including moundy till and push moraines are associated with the former valley glaciers of Harris, and fluted drift aligned in the direction of ice-movement occur both in Harris and south-west Lewis. Glaciofluvial sand and gravel deposits are found in the extreme north of Lewis and also in south-west Lewis, where they are associated in part with a north-eastward directed subglacial drainage system which includes the spectacular Glen Valtos meltwater channel. Periglacial features including boulder lobes, probably formed during the cold Loch Lomond stade, occur on the highest hills and on low ground in north Lewis. There are large areas of peat in central and northern Lewis and smaller deposits in the Uists. A feature of the islands, particularly the Uists, is the widespread cover of blown sand, partly calcareous, on the landward side of the sand and gravel beaches of the west coast.

Introduction

Despite the fact that much detailed work on bedrock geology has been carried out in the Outer Hebrides in recent years, little systematic mapping of the superficial deposits and glacial features has been attempted and no comprehensive report covering the Quaternary history of the islands has been published since the classic papers of James Geikie, summarised in the third edition of his book, 'The Great Ice Age', published in 1894. The following account is largely based on the results of a reconnaissance survey of the islands at the 1:50 000 scale between 1977 and 1979, following interpretation of air photographs at the 1:10 000 and 1:25 000 (approximate) scales. St Kilda, the Monach Islands and islands south of Vatersay were not visited, but some information is available from recent research, from older literature, and from records of the British Geological Survey (BGS).

The Outer Hebrides, a chain of islands which lies about 50 km west of Skye and the north-west coast of Scotland, are some 200 km long in a NNE to SSW direction with a maximum width of 50 km (Figure 1). In the north, the peat-covered plateau of northern Lewis, much of it between 80 and 140 m above OD, contrasts with the bare mountains of southern Lewis and Harris, where the highest peaks are between 700 and 800 m above OD, and where the pattern of valleys and sea-lochs is controlled by faults and glacial excavation. Large areas of the Uists and Benbecula are only a little above sea-level, resembling the low strandflat of western Norway (Holtedahl, 1960, p. 525). The hummocky, rocky moorland is studded with numerous lochs, and the rich green pasture of the sandy machair of the west coast contrasts with the line of higher hills along the east coast and in the northern part of North Uist. The southerly islands between Eriskay and Barra Head are more generally hilly. The whole archipelago is commonly referred to as 'The Long Island'.

Interest in the Quaternary history of the Outer Hebrides has been spasmodic. Between 1870 and 1880, the islands were a battleground between proponents of the marine drift ice hypothesis (glaciation from the north-west by floating ice-floes) and supporters of glaciation by land-based glaciers (with a probable source on the Scottish mainland). The former view was that of D. Milne Home, convenor of the Boulder Committee of the Royal Society of Edinburgh, whose supporters included the eminent Scottish mineralogist F. Heddle. Their hypothesis was based chiefly on the contrast between the supposedly smooth north-west facing hillsides and the roughened appearance of those facing the south-east. Heddle provided more convincing evidence when he pointed out that boulders of hornblende-gneiss near Tarbert, Harris, were derived from an exposure of this rock a few hundred yards to the west (Milne Home, 1880b, p. 609), suggesting west to east ice movement. However, J. Geikie and his Geological Survey colleague R. Etheridge (1873 and 1878) proposed that the islands were crossed in a westerly or north-westerly direction by ice derived from the Scottish mainland, which covered all but the highest of the Harris hills. The rarity or absence of erratics derived from the mainland was explained as being due to the diversion of the basal debris-carrying layers of the ice northwards and southwards along the Minch.

In the twentieth century, Jehu and Craig (between 1923 and 1934) mainly investigated the solid geology, but added valuable detail to the distribution of erratics foreign to the islands. Deposits thought to be of interglacial age and a raised beach were described from North Lewis (Baden-Powell, 1938; Baden-Powell and Elton, 1937). Charlesworth (1956) adduced various retreat stages for local glaciers and Godard (1965) and McCann (1968) reported the presence of a raised marine bench in Lewis. More recent research has resulted in a drastic revision in the role thought to have been played by mainland ice. In Lewis and Harris, von Weymarn (1974; 1979) adduced evidence for the presence of a large local ice cap during the last glaciation, with external ice impinging on the islands only in the extreme north and north-east. With Edwards, he described a radiocarbon-dated interstadial site in North Lewis (1973). In South Uist, Coward (1977) showed that erratics of the quarto-feldspathic gneisses typical of the rocks west of the Outer Hebrides Thrust Zone are found east of its outcrop, and the implied carry by ice moving across the island from west to east has been confirmed by a re-examination of other features of ice-movement in the Uists and Benbecula by the writer and Ross (1978). Flinn (1979) has suggested an ice-shed situated just east of the west coast of the Uists.

In the field of post-glacial geomorphology the relationship between blown sand and marine processes as well as the evidence for coastal submergence were described by Ritchie in a series of publications between 1966 and 1979. Little pollen analytical work has yet been published for the Outer Hebrides (Birks in Shotton, 1977) and the vegetational changes in both late-glacial and post-glacial times are poorly known.

Pre-Quaternary geology

The Outer Hebrides are composed chiefly of Precambrian basement rocks collectively known as Lewisian (Smith and Fettes, 1979; Fettes and others, in press). These mostly comprise monotonous grey gneisses with hornblendic streaks and patches, which are traversed by dykes. and irregular bodies of metadolerite and amphibolite. In south-west Lewis and Harris there are extensive areas of granite veins, and bodies of anorthosite occur in South Harris and east of the Butt of Lewis. Metasedimentary bodies comprising calcareous, graphitic and biotitic gneisses, quartzite, and rare marbles occur locally on a variety of scales, but generally form only a small part of the assemblage. In south Harris these are associated with large bodies of metamorphosed intermediate and basic igneous rocks. Black, glassy pseudotachylite is found at many localities in the islands but is especially common along the Outer Hebrides Thrust Zone which extends along the length of the Long Island generally near the east coast (Figure 1). East of the Outer Hebrides Thrust Zone in South Uist, the Eastern Gneisses include granulite facies gneisses perhaps derived from a large igneous body (Coward, 1972), together with mylonites and intensely crushed rocks. The Lewisian rocks are cut by a variety of basic igneous dykes of Permo-Carboniferous and Tertiary age.

The Lewisian rocks of the islands seem to form a horstlike feature flanked by a series of Mesozoic basins not far offshore. To the east they are bounded by the Minch Fault, the line of which runs only a few kilometres east of the east shore of the Long Island. At Stornoway, the thick conglomerates and sandstones of the Stornoway Formation, which are thought to be of Permo-Triassic age (Steel and Wilson, 1975), lie unconformably on the gneisses. They occur on the downthrow (eastern) side of faults belonging to the same system as the Minch Fault (Watson in discussion of Steel and Wilson, 1975), and thus provide a link with the Mesozoic offshore basin which occupies the Minch. From the geomorphological point of view, the chief interest is the relatively uniform nature of the lithology and the numerous lines of structural weakness which control both the coastline (Watson, 1977) and the major through valleys.

Decomposed bedrock is found at numerous localities in the Outer Hebrides, both in areas where the effects of glaciation are relatively slight, and in heavily glaciated mountain and lowland districts. In many cases such decomposition is related to faulting. Examples in Lewis are the belt of decomposed rock extending from a locality [NB 363 200] a little west of Caversta on the south side of Loch Erisort to Loch Crois Ailein [NB 383 162] 1 km north of Gravir; and, on a large scale, the decomposed rocks in the crush zone extending NNW from Aline Lodge [NB 197 118] to Little Loch Roag and Carishader 1.5 km south-east of Miavaig [NB 085 345]. Farther south, the fault passing through Tarbert into Scalpay is flanked by deeply decomposed bedrock, and in North Uist, a cutting [NG 833 634] in bedrock decomposed to a depth of 6 m at the head of Loch Eport (Godard, 1965; Ritchie, 1968, p. 5) lies close to a fault parallel to the principal Loch Eport Fault.

Outwith faults and crush belts, gneisses within areas affected by the cataclasis and low-grade metamorphism associated with the Outer Hebrides Thrust Zone tend to weather more readily than do those elsewhere. In north Lewis, where solifluction and frost shattering on low ground are more in evidence than farther south in the islands, decomposition of bedrock seems to be more extensive. It is difficult, however, to isolate the effects of weathering from decomposition or accelerated weathering associated with the thrust zone referred to above, with the Minch Fault, and with the unconformity below the Stornoway Formation. For example, the weathering, reddening and fracturing of the rocks forming the coastal cliffs of Tolsta Head and Port of Ness could be associated with any or all of these.

Raised rock platform and cliff of marine erosion

Remnants of a raised rock platform and cliff of probable marine origin (Figure 1) and (Figure 2) occur extensively in north Lewis between Shader [NB 390 547] and Cunndal and at various points in the Eye Peninsula (Godard, 1965, p. 210–212; McCann, 1968, p. 30–33; von Weymarn, 1974, p. 61–75). Typically this takes the form of a shelf a few metres above OD backed by a degraded cliff. The shelf, which reaches its greatest width of about 150 m between South Galson and Dell Sands (von Weymarn, 1974), sometimes slopes seawards to join the modern erosional intertidal platform but in most cases is separated from it by a cliff. Degraded former stacks rise above the general surface, as at Toa Dibadale. Von Weymarn records an average height for the base of the back feature of 9.2 m above OD. This figure is based on four measurements on the north-west coast, and two (7.3 and 7.8 m above OD) in the Eye Peninsula. A section of the former cliff line survives as a 'slope over wall' feature above the modern sea cliff along some 1.5 km of coast adjacent to Sguiman nan Creagan Briste [NB 550 530] (von Weymarn, 1974) and in patches farther south for 2 km.

The raised marine platform is being exhumed from beneath a cover of till at many localities, and is certainly glaciated where it occurs in the Eye Peninsula. Former stacks such as Toa Dibadale, mentioned above, and Dorus Beag [NB 558 338] in the Eye Peninsula appear to have been rounded by the passage of ice. On the north coast at Bragar and Arnol [NB 310 485], von Weymarn (1974) records that the platform, cliff and accompanying stacks have almost been destroyed by ice passing over them from the south. The platform clearly predates the raised beach gravel described next, which itself post-dates one glacial episode in north Lewis. It thus seems to have formed prior to the maximum of the last (Devensian) glaciation and could be very much older. Given time, it could certainly have been cut by very high energy marine processes such as those forming the present marine platform and cliff on exposed coasts, but recent work suggests that very rapid formation under periglacial conditions is more likely (compare with Gray, 1978). The absence of the platform and cliff elsewhere in the Outer Hebrides could result from glacial erosion, a protective cover of glacier ice (see Sissons, 1982), or a tilt below sea-level. Similar limited distributions are known for marine platforms and cliffs of diverse ages elsewhere in western Scotland (Sissons, 1982) and other factors including tectonism may be involved.

Raised beach gravels of Lewis And Barra

A raised gravel beach, first described from Galson by Baden-Powell and Elton (1937) is now known to occur extensively but patchily along the north-west coast of Lewis and the east coast of the Eye Peninsula ((Figure 1) and (Figure 2); von Weymarn, 1974; 1979). It is overlain in some places by solifluction deposits and in others by lodgement till. If, as seems likely, it is all of one age, it can be taken as a useful marker horizon in a discussion of the glacial history of the island. A raised gravel beach overlain by till and possibly of the same age also occurs on the north-west coast of Barra.

Lewis

The beach deposit commonly overlies a water-worn, pitted, bedrock surface, which is merely a slight modification of the Raised Rock Platform. At other localities there is a thin (up to 0.8 m thick) layer of angular rock fragments, probably head, between it and the solid rock, or the beach may be underlain by till. It sometimes includes bouldery gravels, and lag boulders up to a metre across have been noted, for instance at Cladach Lag na Greine [NB 387 558], 1 km north-east of Rubha Bhlanisgaidh, but there is a general impression that the clasts are commonly of gravel and cobble size, and smaller than clasts on the modern storm beaches in the area. The matrix is of sand, and beds of sand are intercalated with the gravels. Though most clasts are of Lewisian gneiss, pebbles of red siliceous sandstone and arkose probably derived from Torridonian strata are generally common and quartzites, lithologically similar to the Cambrian quartzites of the north-west Highlands, also occur. There is little evidence for a separate back feature associated with the gravels, and it is likely, in view of the height of the raised beach (commonly 10 to 14 m above OD according to von Weymarn) that the older cliff of the Raised Rock Platform came under wave attack at the time of deposition. This however shows no evidence of 'freshening-up', and marine erosion at the highest level of the gravels must have been slight.

The most extensive development of the raised beach is at Galson, where it extends continuously for 2 km ENE from a locality [NB 420 586] to the mouth of the Galson River (Figure 2). Here, as at many other places, the beach rests on the raised rock platform with the upper surface between 10 and 14 m above OD (von Weymarn, 1974). The inner edge against the degraded cliff is concealed by slope wash or head, but the beach possibly reaches a maximum width of over 200 m, with traces of storm ridge development on the coast due north of South Galson. At this locality the shingle reaches a thickness of 5 m and is formed of very well rounded pebbles of high sphericity. The pebble size is commonly between 1 and 5 cm, but coarser beds occur southwards. As noted by von Weymarn, the clasts are mainly of Lewisian gneisses, but red siliceous sandstones are common, up to the size of small boulders. With the exception of the section of beach due north of South Galson, many of the exposures show soliflucted till up to 1.5 m thick overlying the gravel; this is overlain locally by what seems to be further beach gravel with a red, clayey matrix (Plate 1, fig. 2). If the latter is genuine beach gravel, the interbedded soliflucted till is contemporaneous with beach formation (compare with von Weymarn, 1974, p. 101), but the presence of the clayey matrix suggests that this too could be a solifluction deposit post dating formation of the raised beach. It is noteworthy that the soliflucted till here contains very few Torridonian erratics. The upper metre or so of the deposits is cryoturbated.

North-eastwards from South Galson (Figure 2) there are many exposures of the raised beach a few metres above high water mark as far north as Cunndal [NB 512 657]. These frequently show the gravel resting on frost-shattered rock (head) and at Toa Galson [NB 452 602], the gravels, which reach a level of 20 m above OD according to von Weymarn, appear to be banked against and overlain by this material. The upper surface of the gravel is commonly cryoturbated and is overlain by cryoturbated solifluction deposits at many localities. At Cunndal [NB 512 656] there is a 6 m section of alternating beds of soliflucted till and raised beach gravel (von Weymarn, 1974, p. 101). Only the lowest gravel, however, may be genuine beach, the other gravel beds having been soliflucted (see above).

Deposits other than frost-shattered rock occur below the raised beach at four localities on the north-west coast of Lewis (Figure 2). At Cladach na Luinge [NB 465 611] the raised beach overlies 0.5 m of grey till (possibly soliflucted), which in turn lies on up to 2 m of bedded sand and gravel, the latter with subangular to well rounded pebbles including sparse Torridonian clasts. Von Weymarn records over 2 m of till below the raised beach at Toa Dibadale, and, at Cunndal, red till similar to the soliflucted till above the beach occurs below it at two close localities [NB 513 655]. At a locality [NB 508 642] a little south of Traigh Sands, calcite-cemented raised beach gravel recorded by von Weymarn (1974, p. 99) is up to 1.5 m thick and is banked against reddish brown sandy till, which also slumps across it (Plate 2, fig. 1). The gravel is underlain by 0.3 m of the same till, which contains wisps of sand and silt and is possibly flow-till. The till passes northwards into interbedded sand, gravel and diamicton typical of the multiple deposits of this area ( see p. 11). This important section apparently confirms the evidence from the other three, that the raised beach postdated the last evidence of glaciation on this stretch of coast. Moreover, this glaciation, on the evidence of the soliflucted till, was followed by periglacial activity before the deposition of the raised beach gravel. These matters are discussed further on p. 21.

South-westward from Breivig (Figure 2) the raised beach can be traced as remnants as far as a locality [NB 372 544] on the coast north-west of Ballantrushal. In this sector, there is clear evidence that ice crossed the coast in a NNW direction, overriding the beach. Where this survives, it is commonly overlain by grey sandy till to a maximum thickness of 4 m [NB 412 580]. In places a thin layer of angular rock fragments intervenes between the beach and till, this being either head or material transported a short distance by the ice. Where the beach has been removed, till, with or without the thin head-like deposit below, rests on the striated surface of the raised rock platform.

A critical section showing the relationships of the raised beach to glacial deposits and features is at Cladach Lag na Greine [NB 387 558] 1 km north-east of Rubha Bhlanisgaigh. Here the beach deposit overlies a waterworn, pitted surface and is overlain by till. Seawards, the shingle thins to zero and the water-worn surface becomes striated where it projects slightly above the level of the beach. This firm evidence that the raised beach was overrun by ice supports the view, based on field examination, that much overlying till exposed on the coast from Breivig to the westward beach limit [NB 372 544] is lodgement till.

Another good section showing the relationship of the till to the raised beach is that of von Weymarn (1974), near Melbost Borve [NB 408 573], where the beach gravel 1 m thick, which overlies about 0.5 m of head, is in turn overlain by 3 metres of hard brown till (Figure 2). A partial fabric analysis by von Weymarn confirms an origin as lodgement till. The bed of beach gravel in this section dips to the south-east and could be either in situ or a raft transported a short distance by the Late Devensian ice (Peacock, 1981).

On the east side of the Eye Peninsula (Figure 1), traces of the raised beach are known at Sheshader Bay and Bayble Bay (Godard, 1965, p. 211; von Weymarn, 1974). At the former, a line of pebbles lies below thin till or head at one locality [NB 556 343], and in the north part of the bay, till up to 5 m thick overlies water-worn rock possibly associated with the raised beach. In Lower Bayble Bay [NB 527 304] beach gravel up to 1 m thick underlies several metres of till. The relationships here are thus similar to those on the north-west coast south-west of Galson. Occurrences of raised beach gravels below till at two further east coast localities [NB 529 654] and [554 615] (von Weymarn, 1974) have not been confirmed, but a similar section showing till on gravel of possible marine origin occurs at Caiashader Shore [NB 559 611] near the latter.

Barra

An isolated section of raised beach gravel occurs at [NF 673 048] near Cliad (Plate 2, fig. 2). Up to 3 m of sandy gravel and cobbles with well rounded clasts are exposed over a distance of 20 metres. The deposit lies on, and is banked up against, water-worn rock and is overlain by an unsorted bouldery deposit (till), up to 6 m thick, of angular blocks of gneiss up to 1 m across in a hard brown sandy and silty matrix. The gravel is composed chiefly of local gneisses, but pebbles of red siliceous sandstone of Torridonian aspect, pale sedimentary quartzite and micaceous psammite of Moine type also occur. The overlying deposit also contains a few rounded pebbles, including clasts of Torridonian aspect. The top of the gravel is about 5 m above high water mark, at approximately the same level as the raised beach in Lewis. Though the bouldery deposit above the gravel could be a product of solifluction, it lies immediately down slope from ice-smoothed rock and is identical to till seen in a few sections elsewhere in Barra and Vatersay. Possibly, therefore, the Barra raised beach, like that in Lewis, predates the maximum of the last (local) glaciation in the Hebrides.

Discussion

To summarise: the characteristics of the beach in Lewis are its low altitude (10–14 m, locally up to 20 m above OD), lack of back feature, and the small size of the clasts compared with those in the modern storm beaches (the section in Barra is too isolated to be included in a general comment). The generally uniform altitude suggests that it was formed at a time of minimal glacio-isostasy, as in an interglacial period, though this apparent uniformity may be a function of the small stretch of coastline along which the beach occurs, its orientation and the small number of localities where the maximum height can be measured. The lack of a back feature adjacent to a coastline, where marine erosion of solid rock is taking place at the present day, suggests that the sea was at beach level for a short time only. The small size of the clasts might be due to their derivation from till, but could also indicate a relatively low-energy depositional environment compared to the present. The very well rounded nature of the shingle would not appear to support the latter interpretation but rounding is a function of time as well as energy, and continual reworking of deposits might occur continually over a long period during a marine transgression. The beach may thus mark the culmination of such a transgression, similar to that which formed the (much later) main post-glacial beach of the Scottish mainland. Two situations can be envisaged. If the beach is interglacial (Ipswichian), its formation probably occupied only a very small part of the time available (about 11 000 years, for instance, Bowen, 1978, p. 103), and the 'main' Ipswichian beach must be sought at or below modern sea-level in Lewis and perhaps in Barra also. If it is periglacial, an interpretation equally feasible, then a low energy environment could possibly have been caused by the presence of pack ice throughout most of the year. The latter interpretation would be considerably strengthened if the gravels interbedded with the soliflucted till above the raised beach could be shown to be in situ beach deposits rather than solifluction sheets.

Interstadial deposits on Tolsta Head

On Tolsta Head [NB 5572 4682] in north Lewis, laminated silt and dark brown organic lake detritus some 0.5 m thick rest on bedrock and are overlain by up to 2.5 m of a deposit interpreted as till (von Weymarn and Edwards, 1973). The pollen spectra from the lake deposits indicate an open landscape with a flora consistent with a cool maritime climate. A radiocarbon date of 27 333 ± 240 BP (SRR-87) was obtained from these sediments. According to von Weymarn and Edwards, the overlying deposit contains numerous erratics of Torridonian sandstone derived from the Minch. Other evidence, however, suggests a westerly origin for some of the erratics (see below). Examination of the deposit and its setting suggests that an origin by solifluction cannot be ruled out, and the site as a whole would clearly repay detailed examination.

Features of ice movement

The Outer Hebrides were probably wholly glaciated during the last (Devensian) glaciation, with the possible exception of the extreme north of Lewis (which is discussed separately below). The glaciation can be divided into an earlier Hebridean ice-sheet phase, when ice extended over the extramontane areas and a later valley glacier phase, when glaciers were confined to the mountain.

Hebridean Ice-Sheet phase

North Harris and South Lewis

On low ground, numerous features of ice movement, such as striations, plucked surfaces, roches mountonnees and crag-and-tail, suggest that ice flowed radially outwards from centres in North Harris and south Lewis (Figure 1). The belt of crag-and-tail features extending in an arc of a circle from Uig Sands [NB 040 330] in the west to Loch Erisort [NB 300 200] in the east is particularly marked and can be traced as far north as Stornoway.

These features are accompanied by large-scale streamlining in rock and till and the implied ice-flow pattern is further confirmed by plucked and striated rock surfaces.

North of a line between Cellar Head [NB 565 560] and Galson, features of ice movement are scarce, only one striated surface having been recorded (see below).

In the Eye Peninsula, east of Stornoway, von Weymarn (1974) noted that the red till derived from the Stornoway Formation transgresses eastwards over the Lewisian, suggesting an easterly ice movement. Confirmation is provided by striated, plucked and grooved surfaces at two close localities [NB 559 337] south-east of Sheshader (Figure 1) and a fabric analysis from the same area by von Weymarn. The presence of erratics of Torridonian rocks has long been known in the Eye Peninsula (Geikie, 1878, p. 828, as 'Cambrian grit and sandstone') and cannot be accounted for by the eastward ice-movement suggested by the other evidence. Similarly, though von Weymarn and Edwards (1973) suggested that Scottish mainland ice reached Tolsta Head during the last glaciation, the plucked surfaces elsewhere on the coast in this area indicate only an easterly movement of ice. This agrees with Flinn's (1979) suggestion that some of the Lewisian erratics in the till at the Tolsta Head interstadial site were derived from the west. An earlier westerly, or northerly, movement of ice seems likely at both localities though it is not yet clear whether this was due to fluctuations of the boundary between contemporaneous Late Devensian mainland and local ice, or whether the mainland ice crossed the Eye Peninsula during an earlier glacial stade only. These erratics and the shelly tills and other deposits at Garrabost and Tolsta are referred to later (p. 10).

South of Loch Erisort a former east to north-east movement of ice is indicated, but in the vicinity of Lemreway [NB 380 120], south-easterly directed striae suggest the possible presence of a local ice centre. Roches moutonees in the peninsula south of Loch Sealg [NB 315 090], together with striations near the summit of Gormol [NB 303 069] show an ice-movement to the east or ENE (Figure 4). Jehu and Craig (1934) noted a few boulders of Torridonian rocks on low coastal ground in the Park district.

Between Uig Sands [NB 040 330] and Brenish [NA 993 263], striations and plucked surfaces show that ice-flow was directed to the west or north-west, but larger streamlined forms in several places and striations at one locality [NA 992 250] suggest a possible earlier, northward movement of ice (Figure 1) and (Figure 3).

South Harris

In the area about West Loch Tarbert, ice flowed southwestwards across the island of Scarp (Jehu and Craig, 1934), south-westwards out of the North Harris mountains, and southwards across Taransay into South Harris (Figure 1). A south-eastward flow of ice in the Tarbert area is shown by numerous striated and plucked surfaces, and this south-easterly or SSE direction is maintained throughout South Harris. The evidence suggests that at one time the major North Harris ice centre was situated over the extreme west of the island and that ice issuing from the mountainous area to the east was deflected south-eastwards across South Harris. This western ice-centre could also have led to the earlier southward-directed striae at Hushinish [NA 985 123] and the possible early northward movement of ice in the Brenish area (Figure 1).

North Uist, Benecula and South Uist

In the Uists and Benbecula generally, innumerable plucked surfaces, striations and roches moutonnees show that, over much of the islands, the last general movement of ice on low ground was towards the ESE (Figure 1). The exception is on the western seaboard, where the sense of ice-movement is ambiguous. Flinn (1979) however reported westward-directed striae at six localities on this coast and a north-westerly ice movement has been suggested for the Monach Islands (Catterall, 1975). In South Uist the diversion of the ice around the base of the mountain massif of Hecla and Beinn Mhor noted by Geikie (1878, p. 581) can be confirmed, but the direction of ice-movement was generally eastward rather than westwards as shown by well formed roches mountonnees south of Loch Skipport and on the southern flank of Beinn Mhor. The roches moutonnees at the latter locality, with the correct (south-eastward) sense of ice-movement, were noted by Geikie (1878, p. 853), but ascribed to a phase subsequent to ice-sheet glaciation and prior to local glaciation. Furthermore, Coward (1977) has shown, on the basis of a detailed investigation of the bedrock geology, that erratics of gneisses from the west part of the island are found in the east, but not the reverse. Jehu and Craig (1926, p. 488) suggest that boulders of crushed and sheared gneisses found in the drift and on the surface west of the Outer Hebrides Thrust Zone were transported there by a westerly movement of ice, but as such crushed and sheared rocks are now known in situ at many localities in the islands, this evidence must be discounted.

Eriskay, Barra and Vatersay

Farther south, numerous striated and plucked surfaces show that the easterly to south-easterly direction of ice-flow continued across Eriskay and the central and east parts of Barra and Vatersay (Plate 1, fig. 1). The rocky exposures a little north-west of the summit of Heaval (383 m) [NL 678 994] in Barra exhibit rounding produced by an easterly or south-easterly flow of ice.

On the west of Barra and Vatersay the situation is more complex. At Greian Head [NF 650 050], the moutonnee forms, though weathered, suggest an ice movement from east rather than from west. Further south on the low ground westward at Tangasdale [NF 650 004], many of the plucked surfaces seem to be on the west sides of projecting rocks though there is some smoothing on this side also. An earlier westward flow of ice followed by a weaker easterly flow is suggested; and the latter probably produced the striations. Near the north-west point of Vatersay, though most striae were clearly produced by ice moving south-eastwards, striations on the south-easterly (lee) side of projecting rocks at two localities [NL 623 976] and [NL 617 975] indicate an earlier east–west or west–east movement.

Discussion

The above evidence suggests that the ice-shed, during the maximum of the Hebridean ice-sheet phase, was situated in the area of weakly developed features of ice-movement, very close to the west coast of Barra and Vatersay; and that the features showing a westerly ice-flow date from either a time immediately preceding this, or from an earlier glacial episode. The ice-shed probably extended northwards along the west coast of South Uist and Benbecula into the western part of North Uist (compare with Flinn, 1979). The distribution of striations (Figure 1) suggests that the ice in the Uists originated in a centre separate from that of Harris.

The presence of erratics foreign to the Uists, Barra and the islands to the south was noted by Jehu and Craig (1923a, 1923b and 1926). Boulders and pebbles of Torridonian sandstone and arkose are scarce on the west coast of Benbecula but become less rare southwards, where they are accompanied by quartzites which may be of Cambrian age. They are particularly frequent in Vatersay: for example, on the beach at Bagh Siar [NL 627 947] where there are also pebbles of Moine-type granulite, porphyry, granite, hard chalk and chalk flint. Jehu and Craig also recorded boulders and pebbles of Torridonian type in the glacial drift on low ground in South Uist. During the present investigation these were observed in till at Cliad [NF 673 048] and Greian Head [NF 652 046] in Barra. Boulders of hornblende-porphyry are common along the western seaboard of North Uist and Benbecula and also in the Monach Islands. Other distinctive erratics found more locally are fossiliferous Carboniferous limestone on the west and south coast of South Uist and red sandstone of possible Triassic age at the extreme north-west corner of this island.

In view of the overwhelming evidence that the last ice-sheet to occupy the Hebrides was local in origin, the source of these foreign erratics requires some discussion. Torridonian and Cambrian pebbles are found in the raised beach at Chad in Barra below local till and thus date from some earlier episode. Clearly this beach could have been a source for some of the erratics of these rocks both locally and elsewhere in the islands. If derived partly or wholly from the mainland (see detailed discussion on p. 17), they may furnish evidence that the southern part of the Outer Hebrides was crossed by mainland ice at some stage, though all glacial deposits associated with this event have now disappeared.

Ice-movements in North Lewis

On the north-west coast of Lewis (Figure 1 and (Figure 2) there is abundant evidence from striations and plucked surfaces that ice of the Hebridean ice-sheet phase extended as far north as Breivig [NB 414 582] beyond which there are marked periglacial features and deposits (p. 17). On the east coast, moutonnee forms indicating an eastward movement associated with the Hebridean ice-sheet phase can be traced for about 3 km north of New Tolsta. North of a line from Galson to Cellar Head only one striated rock surface is known (at [NB 533 651], below grey till, Figure 2), but shelly tills and erratics foreign to Lewis occur at many places around the coast. Von Weymarn (1974) records partial fabric analyses from till at Port Skigersta [NB 556 614] and Eoropie (Figure 2) which suggest a westerly ice movement, but in view of possible debris flow deposits in this area (p. 11), the type of material selected for investigation needs further examination. However, the evidence as a whole suggests that the northward limit of the Hebridean ice sheet coincided roughly with the southern limit of abundant Torridonian erratics (Figure 1) and the southern limit of thick till on the east coast (Figure 2). North of this line the glacial features and deposits point to glaciation from the east and could be of more than one generation (p. 19).

Valley glaciation

There is abundant evidence, particularly in south Lewis and north Harris, of a vigorous valley or small ice-cap glaciation in mountainous areas. The evidence in some areas suggests that this was part of the retreat phase from ice-sheet glaciation, but in others there appears to have been a readvance of the ice before its final disappearance. The landforms and deposits associated with the valley glaciation are discussed on pp. 13–15 and the glacial history on p. 21.

Glacial and glaciofluvial deposits and landforms

Introduction

In the outer Hebrides there is a variety of deposits and landforms. Some have been formed directly by active glacier ice and others reflect some degree of control by ice in both an active and a passive capacity. Thus at one end of the scale is lodgement till, deposited in response to the conditions of stress, loading and temperature at the base of moving ice, and at the other, glaciofluvial sediments, the deposition of which by meltwater was more or less controlled by distance from an ice margin. The latter are accompanied in places by debris flow deposits including flow-tills. Some of the hummocky moraine-like mounds are formed of little moved basal debris. Ordered morainic features include fluted drift and transverse ridges.

Lodgement till of the Hebridean Ice Sheet phase

Lewis

Much of Lewis north of a line from Stornoway to Bragar [NB 300 480] is underlain by a peat-covered sheet of till (Figure 1). South of this line the till is discontinuous and occurs chiefly on the distal side of rock knobs. Numerous exposures of till near the roadside between Stornoway and Lower Barvas show the deposit to be up to 4 m thick. Much of it is a soft to hard, grey or brown, sandy till with subangular to subrounded clasts of local rocks up to about 1.5 m across. This is locally underlain by a more gravelly deposit with angular clasts and overlain by a layer up to 1 m thick of 'washed' till with water-sorted sand and gravel and numerous large boulders (up to 2 m across). The bedrock below the till is broken and rarely preserves striae. Similar till is exposed in coastal sections between Rubha Bhlanisgaidh [NB 380 550] and Breivig (p. 6) where it contains a few pebbles of Torridonian rocks probably derived from the underlying raised beach. At Lower Barvas [NB 360 495], the till is exceptionally bouldery, (Jehu and Craig, 1934) and a low ridge of this very bouldery material extends NNW in the direction of ice-movement.

North of Stornoway, between Newmarket and Tong, hard red till, the colour of which reflects of the underlying Stornoway Formation, is exposed in a 5 m-high section in a river bank [NB 436 366]. An excellent section some 100 m long and 6 m high [NB 453 307] near Holm exposes hard brown lodgement till lying on a glaciated pavement of conglomerate. Clasts in the till are imbricated and dip gently westwards. Partial fabric diagrams from these two sites suggests an eastward ice movement (von Weymarn, 1974). On the Eye Peninsula there is a discontinuous cover of brown weathering brownish grey till, thickest in the eastward-facing bays, where sections 4 to 6 m high occur.

North Harris

Lodgement till occurs locally in the area mapped as morainic drift on (Figure 4). A temporary road section at [NB 135 033] exposed between 2 and 6 m of lithologically varied silty to bouldery till overlying striated bedrock.

South Harris

Elsewhere in the islands, lodgement till tends to occur as small, isolated patches. In South Harris, there are exposures of greyish brown till up to 10 m thick with local stones, on both sides of a small bay [NF 995 903] near Northton. At the highest part of the Rodel to Leverburgh road [NG 037 849], thin till with well rounded pebbles rests on rock. Nearby, a roadside pit at about 67 m above OD shows 0.5 to 1.0 m of bedded gravel with well rounded to sub-angular clasts below this scree, which is possibly the source of rounded clasts in the till.

North Uist to Eriskay

In the northern part of North Uist, several metres of sandy brown till have been recorded at the north end of Hoglan Bay [NF 705 725] near Tigharry (Ritchie, 1968) and north-east of Newtonferry [NF 903 792]. Hard to loose brown sandy till up to 2 m thick forms part of Hornish Point at the north end of South Uist [NF 758 473]. Similar till, 3 m thick, is exposed in a coastal section at Kilbride [NF 752 143] on the south coast. In Eriskay, about 3.5 m of brown sandy and gravelly till are visible just south of the jetty [NF 790 120] overlying a striated rock surface. At all these exposures, the till contains subangular clasts of local rocks and no 'foreign' erratics have been seen.

Barra

Brown, sandy and silty till up to 2 m thick is exposed at a number of localities in Barra, chiefly on the coast and in road cuttings. Greater thicknesses are seen only rarely, for instance on the coast north-east of Cleat (Cliad), where up to 8 m of brown silty and clayey till are plastered against the steep slopes [NF 674 049]. The raised beach at Cliad [NF 673 048] is overlain by up to 6 m of bouldery till (p. 6). In the south-west of the island, about 4 m of hard brown gravelly till is exposed in the sea-cliff [NL 646 976]; immediately inland the till forms low ridges strewn with boulders up to 1 m across.

Vatersay

There are numerous exposures of till in Vatersay, particularly in sea cliffs. Much of it is a hard, grey to brown, silty or sandy deposit, 2 to 5 m thick, sometimes resting on material consisting almost entirely of angular rock debris. The latter is prominent on the south and east side of Heishival Mor and Heishival Beag.

Shelly tills, diamictons and multiple deposits

General

Shelly tills and diamictons with Torridonian erratics, associated in places with beds of silt, sand and gravel, occur at Garrabost in the Eye Peninsula, at Tolsta further north, and in a coastal belt extending from Port Skigersta to Dell Sands [NB 485 624] in northern Lewis (Figure 1) and (Figure 2). Between Dell Sands and Eoropie, Baden-Powell (1938), modifying Geikie (1878), distinguished two boulder clays separated by an interglacial marine bed. At Port of Ness, beds termed boulder clay by Geikie were referred to a Glacial Marine Bed by Baden-Powell. The field observations at both sites were supported by an analysis of the largely fragmentary molluscan fauna (Etheridge, 1876; Baden-Powell, 1938, pp. 401–403 for detail). Von Weymarn (1974) pointed out the similarity in erratic content of till belonging supposedly to different glaciations. He was unable to substantiate the simple tripartite successions of Geikie and Baden-Powell, and found that the sequence of different types of sediment could occur in many combinations. This latter view, which implies that the multiple successions could be attributed, at least locally, to one glacial phase, is supported here.

Garrabost

Multiple deposits including sand, silt, clay and till occur in a moraine ridge which extends NNW through the village and is truncated by the sea at its northern end. Geikie (1878, pp. 826–828) records that shelly clay up to 16 ft (4.9 m) thick was worked at a height of 175 ft (53 m) above sea-level, in old quarries at a locality [NB 507 333]. According to IGS records, this clay is overlain by interbedded gravel, sand and till between 3 and 6 m thick. The deposits are extensively landslipped at their seaward end, but probably reach a thickness of at least 30 m and possibly cover an area of 1 square kilometre. A section of the east side of one scar [NB 505 337] shows 3 m of red till on several metres of clean sand with a few shell fragments. This apparently rests on dark grey clay. Another section on the west side of the slip shows contorted red till passing downward into brecciated dark clay with some fine-grained gravel. These beds, all of which are shelly, have evidently been involved in the landslip. Further east along the cliffs, interbedded fine-grained well sorted sands, angular gravels and till-like sediments are between 10 and 20 m thick, but detailed examination is hindered by slumping. Possibly the moraine ridge was formed at the contact of local ice and Scottish mainland ice, though this is unlikely since the last ice movement across the area seems to have been by local ice (p. 7).

Tolsta

Black clay with erratics including Jurassic limestone was recorded by Dougal (1928) perhaps at a locality [NB 545 467] indicated by von Weymarn (1974). Here grey till in a slump scar overlies bedded silt and sand. The till yields shells, including well preserved though sometimes fragmentary specimens of Turritella, Tridonta elliptica and Nuculana pernula (field identification). At another locality [NB 545 470], dark shelly till with Torridonian erratics is overlain by up to a metre of hillwash.

Port Skigersta to Cladach Cuiashader

Multiple deposits with Torridonian erratics were noted by von Weymarn (1974), but no details were given. At the mouth of the river at Port Skigersta (Figure 2), about 4 m of interbedded sandy silt, dark grey to light brown stony clay, poorly sorted sand and angular gravel are seen, lying on bedrock on the west bank. At a nearby locality [NB 550 616] similar deposits up to 5 m thick include till-like beds with shell fragments, and overlie angular bouldery gravel with a silty matrix. Sections in the river as far south as [NB 549 615] show dark grey to brown shelly till up to 6 m thick, apparently underlying the above.

Between Port Skigersta and Meall Geal (Figure 2) a number of sections show thin laminated clays and stony silts. At the mouth of a small gully [NB 5525 6180] these are apparently overlain by up to 2 m of a head-like deposit with angular slabs of gneiss up to 1.5 m long. At another locality [NB 552 620] about 4 m of gravel with very large angular boulders rest directly on solid rock, and at Caiashader Shore [NB 559 611] red till, some 2 m thick, overlies a similar thickness of gravel with well rounded to angular clasts, the whole separated from solid rock by a head-like deposit also with angular clasts. The gravel at the last locality resembles the raised beach gravel of north Lewis, both in the roundness of some of its particles and in its height above OD. Sections showing brown or reddish brown till can be traced southwards in the cliff face to a locality halfway between Meall Geal and Cladach Cuiashader [NB 554 586], beyond which there is little or no till.

Port of Ness

In the cliffs south of this village, Geikie (1878) recorded upper and lower boulder clays separated by an 'interglacial' bed, whereas Baden-Powell (1938, p. 399–401) refers to a massive almost unbedded purple-coloured shelly silt with irregular stone beds passing northwards into a deposit of boulders and gravel. Baden-Powell regarded the silt as being in some sense a marine deposit, chiefly on the evidence of the marine shells, and termed it the 'Glacial Marine Bed'.

Examination of this section suggests that it is more complex than envisaged by both authors. At the north end, where a small stream enters the bay, there are about 8 m of crudely bedded bouldery gravel varying from well sorted to very poorly sorted. The clasts in the better sorted beds, which include open-work gravel, show more rounding and a higher sphericity, and at one locality the imbrication of plate-shaped clasts indicates transport to the north-west. On the east side of the stream, the bouldery gravel is at the surface, but elsewhere it appears to be overlain by brown diamicton (possibly till) which can be traced southwards across the remainder of the succession. The gravel passes southwards into a succession containing several tills or debris flow deposits of a reddish brown and dark grey colour interbedded with gravel and sand, the deposits reaching a thickness of 10 m in places and resting on somewhat weathered bedrock. At the south end of the bay the tills and gravels thin and disappear, and at the north end the bouldery gravel rests against a bedrock slope. Pending a detailed survey, these deposits are tentatively interpreted as the remains of a complex fan, the source being ice lying to seaward. Part of the Port Skigersta deposits, described earlier, is interpreted similarly.

Butt of Lewis area

A few sections (Figure 2) show multiple deposits and shelly till, and some problematical exposures suggest that raised beach gravel may both overlie and underlie till. At [NB 533 651], there is the following section:

Rock surface with striae bearing N 030° true

The gravel in this section is a few metres above high water mark, at much the same height as the raised beach at exposures some 2 km to the west; but the similarity of the till above and below it suggests that it is either an intercalation of glaciofluvial origin or an erratic of raised beach (compare with the exposures at Melbost Borve, p. 6). Von Weymarn (1974), however, recorded nearby a similar section at Cladach an Eilean [NB 529 654], in which the gravel below a till had rounded clasts suggesting beach gravel. This section was not found during the present survey.

At Port Sto [NB 524 660] a few metres of dark brown shelly till with interbedded sand and gravel are exposed, these evidently being the remnants of a much larger deposit. Similar till but without shell fragments underlies about 4 m of raised beach gravel at Cunndal [NB 513 655].

Traigh Sands to Dell Sands

Multiple deposits, possibly remants of outwash fans and deltas, are exposed at intervals along this 3.5 km long stretch of coast (Figure 2). They are very complex, including beds comparable with sub-aerial and sub-aqueous flow-tills. Because of slipping, which has locally produced intense folding and brecciation, and slumping of the upper deposits down the cliff faces, it is difficult to measure complete sections at any single locality and no detailed examination could be made. The following notes indicate the range of deposits present, but are not comprehensive.

At Traigh Sands [NB 513 645] a section 4 m high in reddish brown till includes sandy wisps, sand lenses and pockets of gravel. The beds are thrown into folds, probably penecontemporaneous, with locally high dips. The sandy wisps and lenses are similar to those associated with sub-aerial flow-tills (Boulton, 1976, p. 74). A few metres north of the indurated raised beach locality (p. 6) [NB 508 642] the deposits are more varied, and a 6 m-face shows 0.5 m of cross-bedded sand underlying interbedded, poorly sorted cobbly gravel and brown diamicton (till or debris flow deposit).

The deposits at Swainbost Sands [NB 504 637], generally about 4 m thick, are chiefly interbedded dark grey to light brown clays, silts and sands with subsidiary beds of gravel up to 0.5 m thick. At the base there is about a metre of dark grey to brown till. All these beds contain shell fragments, which are particularly common in the till, and the gravel includes clasts of Torridonian sandstone.

Southwards the deposits thicken to about 25 m between localities [NB 501 632] and [NB 496 632] at Traigh Chumil. This is the section illustrated by Baden-Powell (1938, p. 398), who described an upper and lower till separated by about 3 m of shelly sand, thought to be an interglacial marine bed. Examination of undisturbed sections suggests that, while there is an overall capping of brown, gravelly or cobbly diamicton (possibly till) between 2 and 5 m thick, the underlying strata in the upper half of the cliff are complex, with interbedded water-sorted sand and gravel, diamicton, and sandy silt. The lower part of the section includes horizontally bedded locally pebbly silty sands and finely laminated dark silts with numerous beds of dark diamicton upwards of a centimetre thick. The latter are possibly sub-aquatic flow-tills (compare with Evenson and others, 1977). It is noteworthy that the brown till at the top of the section here shows only slight evidence of solifluction, despite the considerable evidence for periglacial activity at other localities on the north-west coast of Lewis (p. 17).

At Cross Sands [NB 493 627] the following section was measured (thicknesses approximate):

The westward limit of the multiple deposits on this part of the coast is at Dell Sands, where at [NB 486 625] there are about 12 m of interbedded gravel, sand, silt and till-like deposits, some shelly.

On the road between Stornoway and Barvas, Robertson (1880, pp. 248–250) recorded a marine micro-fauna probably derived from a gravelly deposit at 300–400 ft (91–122 m) above sea-level. This could be interpreted either as a raised marine deposit or (more likely) as shelly till.

Moundy till

There are many small areas of moundy till in the more southerly islands of the Outer Hebrides, the most extensive being in North Uist between Lochmaddy [NF 915 697] and Carinish [NF 820 605]. Similar material also occurs at isolated localities in south-west Lewis and within areas designated as morainic drift (Figure 3) and (Figure 4). In places the mounds, which are usually 2–5 m high, are aligned in the direction of ice-movement, as at Diraclett [NG 150 990] 1 km south of Tarbert, and Loch Carran [NG 086 960] 1.5 km east of Seilebost in South Harris. At Baleshare in North Uist, ridges in the Samala area [NF 800 620] are apparently aligned at an angle both to the foliation in the country rock and to the direction of ice-movement further east, whereas south of Howmore [NF 760 362] in South Uist, till ridges are parallel to the NNW to SSE foliation and at an angle to the easterly direction of ice-movement. Both these localities are near the ice-shed. At other localities, such as those described below in North Uist, there is little correlation with other linear features and the mounds themselves are more randomly oriented.

Sections suggest that the mounds consist largely of angular gneiss fragments set in a sparse sandy matrix (Plate 3, fig. 1) and (Plate 3, fig. 2). This material grades into true till full of angular gravel. In South Harris, cross-sections of mounds up to 3 m high at a locality [NG 088 960] show bands of debris derived from differing rock lithologies (gneiss and dolerite) passing downward into angular material and bedrock. Two sections are partly capped by greatly shattered granitic gneiss, suggesting that this material has been pushed across the top. In North Uist at a road metal quarry [NF 950 723], striated bedrock at the core of a drumlin-like feature is broken and slightly displaced, probably by ice action (Plate 1, fig. 2). At Greaneclett [NF 927 720] 3 km NNE of Lochmaddy, roadside exposures in mounds 2 to 3 m high show little-transported angular fragments which are in places apparently in continuity with the underlying disturbed bedrock. Similar sections are to be seen at nearby localities, such as [NF 930 716] and [NF 936 716] and at the east end of the causeway connecting Baleshare to North Uist. Elsewhere, as in South Uist [NF 757 305], the angular rock debris overlies striated bedrock, from which it is separated by a very thin silty deposit possibly marking a shear plane. In most of the above sections, the angular fragments range up to boulder size but are rarely more than a metre across. Blocks larger than this are commonly shattered. Exceptionally, where the granitic gneiss is massive as in the area about Loch Laxadale [NG 108 962] in South Harris, mounds are found consisting essentially of collections of large blocks up to 2 m across.

It is suggested that the mounds are formed of material derived entirely from the bedrock below the ice, probably as the result of shear stress at the ice/rock interface acting together with a variable hydraulic pressure on joints and other discontinuities (compare with Broster and others, 1979). The material was transported in some cases a short distance before being deposited by a meltout or lodgement process. This mechanism of till formation was clearly widespread in the Outer Hebrides and could account for the occurrence of some of the head-like deposits at the base of lodgement till discussed earlier. The presence of such mounds in the ice-shed area suggests that this was traversed by moving ice at some earlier stage.

Morainic drift

The term morainic drift has been used for the heterogeneous silty sandy, often loose, bouldery till with local interbedded silt and sand associated particularly with the late valley glaciation of south-west Lewis and north Harris (Figure 3) and (Figure 4). It is commonly characterised by landforms such as hummocks, fluted drift and (at a few places) push moraines.

Hummocky moraines

Some of the hummocks are formed predominantly of angular clasts (see below) but roadside exposures in others suggest that the chief component is a very poorly sorted diamicton composed of a matrix of clayey or sandy silt, charged with subrounded to angular clasts up to boulder size. Most of the rock fragments, of which only a few are striated, are derived from the local gneiss. The diamicton, in which the character of the clasts may vary greatly from bed to bed, may contain pockets and wisps of sand and silt, and be interbedded or interlaminated with a variable proportion of silt and sand which occasionally shows traces of cross-bedding. In places, the diamicton and associated deposits are contorted and dip at high angles, otherwise the bedding is usually gently dipping, either conforming to, or being truncated by, the surface of the mound. An example can be seen at a pit [NB 187 060] 2 km west of Maaruig, where there is the following section:

It is suggested that most of the diamicton in the hummocks is flow-till originating from debris (perhaps preexisting debris: see Hodgson, 1982) incorporated at the base of the ice and deposited near the ice-front.

Morainic hummocks formed mainly of rock fragments with little or no matrix occur in the vicinity of Tarbert. At a locality in the Park district [NB 324 144], a cross-section in a mound shows till with angular clasts and with compositional banding due to variations in the proportions (always small) of sandy matrix. Such features recall those mentioned earlier for hummocky till in the Uists and suggest that some of moraine-like mounds associated with valley glaciers in Lewis and North Harris have been formed by deformation, shearing and lodgement processes.

Fluted drift

Drift fluted by the ice in the direction of movement is common in the valleys of Harris west of Loch Seaforth and is very well developed locally in south-west Lewis (Figure 3) and (Figure 4). Elsewhere it has been seen at the head of Loch Sealg [NB 290 105] and adjacent to the back wall of Coire Roineabhail [NG 047 863] in South Harris. The individual flutes, which are generally formed of till or gravel, are up to 300 m long (though generally much shorter), 6 m high and 30 m across and tend to be best developed adjacent to steep slopes or crags. In places the ridges are broken into chains of mounds. Good examples can be seen in Glen Skeaudale [NB 152 031] and Glen Ulladale [NB 074 134] where they lie on the proximal sides of asymmetrical transverse moraines (Plate 4, fig. 1). Bouldery ridges parallel to striations on the valley side at another locality [NB 147 029] in Glen Skeaudale are probably flutes. In the valley of the Abhainn Cheann Chuisil north of Loch Cheann Chuisil (Figure 3), the flutes locally cross earlier-formed drumlinoid morainic mounds up to 5 m high [NB 032 223].

The flutes discussed above are similar to those described in detail by Hodgson (1982) from the Scottish mainland, some of which he ascribes to reworking of drift already deposited and some to the deposition of new material. Those lying on asymmetrical push moraines would appear to fall into the latter category.

Asymmetrical transverse moraines

Numerous small moraines occurring within National Grid squares [NB 16 14] and [NB 16 15] on the valley side east of Loch Langavat consist of arcuate ridges 1–5 m high, slightly to strongly convex down valley (north), with steep (20–30°) north-facing distal slopes and gentle proximal slopes (Figure 4) and (Plate 5). The ridges are highest at their central, most distal points and the distance along the chord joining the low proximal tips is 10–40 m. In places, for instance at a locality [NB 162 138], the proximal slopes of the ridges display traces of parallel flutes 0.5–1.0 m high, aligned apparently in the northward direction of ice movement. The lobate moraines tend to unite to form ridges running obliquely down slope, perhaps defining former ice front positions. Apart from scattered boulders on the surface, the lithology of the mounds here is concealed by vegetation and peat, but is likely from a general inspection of sections nearby to be till with angular debris of local origin.

At a locality [NB 152 031] in Glen Skeaudale (Figure 4) there is a moraine ridge 4–6 m high and about 100 m long aligned at right angles to the east to west former ice flow in the valley. Adjacent to the stream the ridge shows a steep distal and gentle proximal profile with flutes 0.2–0.5 m high and 15–20 m apart on the proximal face oriented in the direction of ice movement. Farther north, the ridge is more symmetrical. Another moraine ridge, 1–2 m high and concave upstream, with a fluted proximal face, occurs at a locality nearby [NB 153 031]. Though there are no exposures in the immediate vicinity, it is likely that the ridges are formed of sandy and silty till similar to that seen in road cuttings near the valley mouth.

Moraine ridges several metres high with steep distal and very gentle, fluted proximal faces can be seen on the east side of Loch Ulladale, being particularly clear at locality [NB 083 143]. The moraines trend approximately at right angles to the former direction of ice movement, as shown by the flutes. The surface of the ground here is scattered with boulders, but there is no local indication of the drift lithology, though this is likely to be till.

Five asymmetrical ridges [NB 102 105] about 2 m high and 200–300 m long occur on the west side of Glen Meavaig, descending steeply from north to south (Figure 4). Four of these ridges are spaced at intervals of 30 m, and the southernmost is about 150 m from its neighbour. These ridges seem to have been formed, possibly annually, at an active ice front which was receding northwards across the present-day watershed. Two similar ridges 60 m apart with steep distal and gentle proximal slopes and another 170 m farther east can be seen on the north side of the valley of the Abhainn Cheothadail [NB 305 125], 2.5 km west of Eishken. The ridges, which descend eastwards, apparently defined a small part of the ice-front of a valley glacier or ice-sheet lobe which was receding westwards. Similar features may occur on the north side of the valley at the head of Lochg Shealg, according to air photo interpretation.

The characteristic features of many of the moraines described above are their relatively small size (about 5 m high), commonly lobate or partially lobate form, tendency for the lobes to unite to produce transverse ridges, steep distal and gentle proximal slopes, and the local presence of fluted drift on the proximal sides. In shape and size they are in some cases comparable to the terminal moraines of the modern Icelandic outlet glacier Breidamerkurjokull between Fjallsarlon and Breidarlon (Welch and Howarth, 1968, fig. 1). The latter, which have prominent flutes on their gentle proximal surfaces, are the western extension of the large partly overriden push moraines described by Okko (1955, p. 64) and may have been formed in a like manner. However, asymmetrical moraines described by Haldorsen (1982, pp. 31–32) from Astadalen, south-east Norway, which are of similar size to the Hebridean examples, have been attributed to subglacial conditions immediately preceding stagnation of the ice. These however include water-sorted sediments and are not associated with flutes.

To summarise: the morphology of the Hebridean asymmetrical moraines is consistent with subglacial formation by active ice, and they seem to have more in common with push moraines than features formed in other ways.

Other morainic ridges

Arcuate moraines some 8 m high which occur on the south shore of Loch a'Ghlinne [NB 023 126] in Glen Cravadale mark the termination of a small glacier which issued from An Coire to the south-west. The ridges cut across the deposits laid down by the former valley glacier in Glen Cravadale itself and thus clearly post-date it.

In Glen Meavaig (Figure 4), a moraine ridge 30 m high on the east side of the valley [NB 101 087] extends obliquely down valley from north-east to south-west for about 600 m. This has been interpreted as a composite push moraine, being generally steeper on the distal side. The origin of transverse morainic ridges in Glen Scaladale between the road bridge [NB 183 098] and Loch Vistem [NB 153 090] (Figure 4) is less clear: they seem unusually straight for moraines left by a normal valley glacier. The large composite ridge at the bridge is composed predominantly of fine- to coarse-grained sandy gravel interbedded with till, as is a second ridge about 300 m upstream.

Ridges and chains of mounds at the heads of Loch Ashavat [NB 070 123] and on the watershed between the Langadale River and the Vigadale River, which are marked as morainic ridges on (Figure 4), are composed of loose, sandy drift with numerous large boulders. They are thought to mark the separation of relatively active and inactive ice during downwasting and the emergence of rock thresholds as discussed by Wyllie (1956). They may have been partly deposited by meltwater and are transitional to the meltwater deposits described in the next section.

Meltwater deposits and landforms

Deposits and landforms associated with glacial meltwater, other than those interbedded with the multiple deposits referred to above, are virtually restricted to Lewis and north Harris.

North Lewis

At the north end of the island (Figure 2) about 2 square kilometres of undulating ground south of Habost are underlain by gravels which are exposed at a working pit [NB 524 618]. There are up to 6 m of poorly sorted bouldery cobbly sandy gravel with angular, subangular and subrounded clasts. The bedding is generally obscure, but locally parallel to the slope. A bed 0.5 m thick of brown clayey till was observed within the gravel on one face. Boulders and cobbles of purple and red sandstone and of red arkose are present. The ridge extending southwest from this locality to Cladach Cuiashader [NB 542 572] is formed partly of gravel. As a whole, the ridge and the moundy gravels in the north suggest either a fan deposit with a subglacial feeder channel or a moraine, both of which could have been formed at or near the north-west margin of the Late Devensian ice in North Lewis (see p. 21).

Kinlochresort area

Hummocky gravels and a small esker cover about half a square kilometre [NB 200 210] about Loch Langavat (Figure 4). Another small esker trends south to north on the slopes between Gleann Sandig and Gleann a' Gharaidh [NB 141 173], and a small area of terraced sand and gravel largely concealed by peat is near the mouth of the former glen. A peat-covered terrace underlain by several metres of sand and gravel occurs a little over a kilometre south of Kinlochresort on the low ground immediately north of Kerraschlett Beag [NB 110 151]. This, and hummocky sand and gravel immediately to the west, may have been laid down in ponded meltwater at the junction of the two northward draining glens here.

Garnish and Glen Valtos

On the south side of Camus Uig there are a number of deposits, collectively about 1.5 square kilometres in area. Some appear to be related to the meltwater channel of Glen Valtos between Camas Uig and Loch Roag (Figure 3) and (Plate 4, fig. 2). In the south, a mass of kettled, partly terraced gravels north of Flodraskarve Mor passes northwards into morainic mounds. The deposits are locally some 20 m thick and become increasingly bouldery south-westwards towards their source south of the rock-cut meltwater channel occupied by the gorge of the Abhainn Caslavat [NB 037 294].

Between Ardroil [NB 045 320] and Uig Lodge [NB 056 333] there is a complex ridge of bedded sand and gravel extending for about a kilometre in a south-west to northeast direction. Sections are few, but the material seems to consist predominantly of bedded subangular to angular gravel in a sandy matrix, with large blocks in the upper part [NB 052 326]. It has been suggested that this is a terminal moraine of a glacier in the Loch Suainaval valley (von Weymarn, 1974), a moraine related to ice in Camas Uig (Godard, 1965, p. 223) or an esker (Ritchie and Mather, 1970, p. 68). Mapping suggests it is one of a series of small kames, eskers and morainic mounds possibly associated with subglacial drainage towards the Glen Valtos meltwater channel referred to next.

The Glen Valtos channel (Plate 4, fig. 2) extends a little over 2 km from its head at Sidhean Tunga [NB 062 343] eastwards to Loch Roag, and is presently occupied by a small stream. The walls are steep and cliffed in places, up to 60 m high, and locally mantled in thick scree which has been partly quarried away. It is excavated entirely in rock, some of which is decomposed. The eastern half of the channel is straight, probably following a fault, but the western part is sinuous and terminates upstream in three blind ends. Deposits possibly associated with the channel are limited to a mound of morainic gravel at Miavaig [NB 086 345]. Till may occur on the unexposed valley floor and has been seen locally on the valley side (D. Sutherland, personal communication, 1981).

At Garnish (Figure 3), a terrace at or a little above 50 m above OD is underlain by sand and gravel covering an area of 0.5 square kilometres. A gravel pit [NB 034 315] exposes several metres of sand with subsidiary gravel, these being disposed chiefly in large foresets dipping east. A metre of two of sandy gravel at the top of the section is flat-bedded. The surface of the terrace is kettled in places and diversified by small channels, which together with the large-scale foresets indicate transport from the west. On the landward side of this area of sand and gravel, Loch Scaslavat seems to be dammed partly by drift deposits and partly by bedrock, as exposures of solid rock occur locally on the surface of the terrace. As suggested by Jehu and Craig (1934) and von Weymarn (1974), the material seems to have been deposited as a delta in a lake held up by ice to the west, though this lake did not necessarily extend as far as Glen Valtos and may have been very restricted in extent. Small esker-like ridges of sand and gravel on the south and west sides of Loch Scaslavat [NB 023 315] and  [NB 016 315] would suggest subglacial transport through the valley leading east from Mangersta. A patch of sand and gravel west of Crowlista [NB 033 340] at 30 m above OD may be related to the Garnish terrace (von Weymarn, 1974), but is more likely to have been deposited in a separate lake.

The widespread deposits and features discussed above indicate that the hydraulic gradient during deglaciation was such that water flowed north-eastwards from the valley occupied by Loch Raonasgail and the other meltwater channels shown on Figure 3 and from west to east from what is now open sea west of Mangersta to Loch Roag. These directions are almost at right angles to the last movement of ice in the area.

Periglacial deposits

With the exception of the extreme north of Lewis (considered below), periglacial features and deposits are almost absent from much of the low ground in the Outer Hebrides. The uppermost parts of sections in till or broken rock in the Barvas, Stornoway and eastern Park districts are, however, locally cryoturbated. Frost-shattering of rock exposures becomes marked at about the 300 m-contour in Lewis and Harris and above this level it becomes increasingly difficult to find glacial striae and unambiguous evidence for directions of ice-movement based on roches moutonnees and plucked rock faces. Such features are best preserved on amphibolite exposures for instance, striations at about 400 m above OD near the summit of Gormol [NB 302 071], in the southern part of the Park district of Lewis (Figure 4). Above about 400 m the stony debris is commonly arranged as boulder lobes on slopes between 10 and 25°. Good examples of such lobes occur on Gillaval Glas [NB 149 022] and Beinn Dubh [NB 090 007] in Harris (Figure 4) where the risers are up to 3 m high, and on the south side of Mealisval [NB 023 270] in south-west Lewis (Figure 3). Similar lobes on the Scottish mainland are thought to have formed during the very cold period of the Loch Lomond Readvance (Sissons, 1976) and it seems reasonable to suggest that those of the Outer Hebrides formed during the same period. They may, therefore, form a useful time marker. Thus, in the Clisham area (Figure 4), boulder lobes occur within the hummocky moraines on the steep slope to the west of Loch nan Eang [NB 145 082], suggesting that this high corrie was free of glacier ice before the time of the Loch Lomond Readvance and may not have been subsequently glaciated. In South Uist, small boulder lobes occur on Hecla and Ben Scalavat [NF 838 348] above 400 m above OD, but in Barra the small area of rock exposure above 300 m about Heaval [NL 677 994] shows little evidence of frost breakage. Solifluction terracettes with more or less parallel risers and treads are more sparsely distributed; examples occur a little below the 400 m-contour on Roineabhal [NG 047 857] in South Harris, and in Oreval in Harris [NB 086 102] where they are associated with boulder lobes (Figure 4).

In North Lewis (Figure 2) periglacial features are widespread on the coast between Breivig [NB 414 582] and the Butt of Lewis, but are much less evident inland and on the east coast. North-east of Breivig, in the area thought to be outside the limit of the Late Devensian glaciation, rock surfaces, though retaining traces of glacial rounding, are considerably frost-shattering in detail, and on the coast adjacent to Sgarbh Sgeir [NB 449 600] boulder-strewn solifluction lobes about a metre high descend gentle slopes on to the raised beach (Figure 2). At South Galson, the raised beach gravels are overlain by up to 1.5 m of interbedded soliflucted till and gravel, the latter clearly derived from the raised beach, but with a red clayey matrix. The upper metre shows well marked cryoturbation. Several sections at Cunndal between localities [NB 512 656] and [NB 510 654] show up to 2.5 m of head on raised beach gravels, the head locally including more than one band of raised beach gravel (von Weymarn, 1974) which is possibly also a solifluction deposit (p. 4). Here, too, the head locally overlies cryoturbated raised beach deposits. The multiple deposits exposed in the cliffs between Dell Sands and Eoropie rarely show marked effects of periglaciation, but at the former locality there is 0.5–1 m of till-like diamiction with angular clasts, possibly head, at the top of the sections.

Some aspects of the glacial history

The problem of foreign erratics on the Long Islands

Details of the distribution of erratics foreign to the Outer Hebrides have been given in earlier pages and by Jehu and Craig (1928–1934). They can be summarised as follows:

1. Pebbles of Torridonian sandstone and of quartzite resembling the Cambrian quartzite of the mainland are common in coastal areas of North Lewis (Figure 1), virtually absent between here and Benbecula, but common again on the west coast of Barra and Vatersay, particularly near the former ice shed. Pebbles of Moine psammite occur on Barra and Vatersay.
2. Less common or very local erratics are Jurassic limestone (Tolsta Head in north-east Lewis), Carboniferous limestone (west and south coast of South Uist), red sandstone of possible Triassic age (north-west corner of South Uist), hard chalk and flint (Barra and Vatersay) and hornblende-porphyry (Monach Islands and the western seaboard of North Uist and Benbecula).

The Torridonian and Cambrian erratics in North Lewis could have been readily derived from the mainland or from the Torridonian rocks cropping out on the floor of the Minch (Chesher and others, 1983, fig. 3). Farther south, the belt of strata including Torridonian, which extends southwards from Rhum to about the latitude of Barra Head (Binns and others, 1974, and IGS records), is a possible source for the more southerly Torridonian erratics on the Long Island. It could also be a source for the Cambrian pebbles, given the reconnaissance nature of the sea floor mapping. The pebbles of Moine psammite, however, are more likely to have been derived from the mainland.

The erratics itemised in (2) above pose a number of problems. The chalk is similiar to the scanty exposures of this rock type in the Inner Hebrides an adjoining mainland, but no outcrops are known on the sea floor between the Inner and Outer Hebrides. No source is known for the erratics of Carboniferous limestones.

On the other hand, Mesozoic rocks (other than chalk) are widespread on the sea floor east of the Outer Hebrides and their paucity as erratics on the islands, together with the apparent absence of erratics of Tertiary igneous rocks from the Inner Hebrides, requires explanation. The latter include distinctive rock types such as felsite, vesicular basalt and layered ultrabasic rocks which should be recognisable, particularly as clean pebbles in the storm beaches. Many less distinctive Tertiary granites and basic rocks, however, resemble similar rocks in the Lewisian and could be overlooked.

In seeking an explanation for these apparent anomalies it can be pointed out firstly that the last general glaciation in the Outer Hebrides was by local ice (see below) and that this ice removed most pre-existing superficial deposits. Thus in most areas no erratics would be preserved and those that did remain, as in the ice-shed areas, would be polycyclic, having survived two periods of glaciation as well as a possible marine episode. Secondly, no detailed systematic examination of the till and storm beaches has yet been carried out and examples of the missing rock types could be found. These questions will clearly repay further study, though it would seem that the apparent absence or paucity of many erratics is real.

Events during the Devensian maximum

There is little doubt that the last ice to cover most of the Outer Hebrides was local in origin, and, as periglacial features are only slightly developed over most of the low ground (with the exception of the extreme north of Lewis), this glacial episode can reasonably be attributed to the Late-Devensian. This ice, with outward flow from an elongated ice dome near the west coast of Barra and the Uists, another over Harris and south-west Lewis and perhaps a third over the low plateau of central northern Lewis, terminated on land only in the extreme north. It post-dated a raised beach in Barra and Lewis and its advance was probably initiated during the period of cold climate which led to marked periglacial activity on low ground in the extreme north of the latter island. Foreign erratics in the north-west of Lewis and again in the more southerly isles probably point to the presence of ice from the Scottish mainland (see above) which crossed parts of the Outer Hebrides prior to the glaciation by local ice. It also apparently reached north-east Lewis again during the Late Devensian, when it was in contact with local ice (p. 21).

The question arises as to whether Scottish mainland ice reached and crossed the Outer Hebrides during the Late Devensian glaciation (excepting the major Harris–South Lewis centre), before the expansion of the local ice. Three situations can be envisaged. First, the Hebridean ice domes developed as a result of climatic changes following the initial westward passage of the Scottish ice-sheet, leading to a build-up of ice in the islands and a reversal of ice-flow on the east side of the Long Island. The second possibility, advocated by Sissons (1980), is that the reversal of flow was caused by the break-up of Scottish mainland ice in the Minch. The first agrees with the evidence in north Lewis, where an advance of local ice can be demonstrated, but is difficult to reconcile with the presence near the ice-shed of the raised beach in Barra. This, if formed during the same period as the raised beach in north Lewis, would require almost total deglaciation of the islands before the advance of the local ice. The second possibility is equally difficult to reconcile with the raised beach evidence. Moreover, the local ice in Barra and the Uists was at least 400 m thick during the reversed flow (Peacock, 1980); the thickness during the period of mainland glaciation would have been considerably greater (1200 to 1400 m according to Boulton and others (1977, fig. 17.2). This is difficult to reconcile with the presence of an enclave in north Lewis which was not crossed by local ice of the last glacial phase, and was subjected to considerable frost shattering and solifluction. The first two possibilities are therefore rejected, despite the persuasive climatic argument for the second (Sissons, 1980).

The third situation requires that the Devensian maximum was a multiple event with at least two peaks separated by deglaciation of areas peripheral to the main ice-sheet. Such appears to have occurred in Ireland (Mitchell, 1972; Synge in Davies and Stephens, 1978), where the ice-sheet of the last glaciation, built up over low ground, is a possible analogue of that in the Outer Hebrides. Perhaps, following glaciation by mainland ice (the first peak), a raised 'late-glacial' beach was formed and was later largely overrun by local ice with mainland ice crossing on to North Lewis (the second peak). This interpretation can be challenged since any such 'late-glacial' beach would be tilted, as are late-glacial beaches on the Scottish mainland. Such objection is not thought to be serious (see p. 6). A potentially more serious objection, however, is that the interstadial lake deposits at Tolsta Head (von Weymarn and Edwards, 1973), formed prior to the main (late) Devensian glaciation, are overlain by a deposit thought to have been laid down either by local ice (p. 7) or by contemporary mainland ice; it is highly unlikely to have survived two glaciations. Moreover, notwithstanding the opinion of von Weymarn and Edwards (1973), the 'till' overlying the interstadial sediments could be a solifluction deposit and this part of Tolsta Head may have escaped the second peak altogether. Whether or not this is the case, the first peak seems likely to have predated the interstadial and should be placed earlier than 28 000 BP; that is, either in the Middle to Early Devensian or in the penultimate (Wolstonian) glaciation.

Another question is the position of the Late Devensian limit in North Lewis, a problem associated with the status of the multiple deposits and their relationships to the raised beach. That ice, probably from the mainland, crossed on to the extreme north of Lewis during the Late Devensian is attested by the absence of major evidence of periglacial activity on the north-east coast as contrasted to the north-west coast (Figure 1); and by the possibility (yet to be confirmed) that such ice post-dated the raised beach. Preliminary results of amino-acid racemisation studies support this view: they suggest that marine shells collected from the glacial deposits in north-east Lewis are of Devensian age (D. Q. Bowen and J. T. Andrews, personal communication, 1982). This enables a minimum southerly position to be sketched for the ice margin (Figure 1), which is in accord with the evidence adduced on p. 6 suggesting that the raised beach near Traigh Sands post-dates the multiple deposits on the north-west coast and that there is an ice limit at Breivig (Figure 1) and (Figure 2). Possibly there are two sets of multiple deposits, one deposited near the Devensian maximum (Port of Ness and Port Skigersta) and the other (Eoropie to Dell Sands) predating the formation of the raised beach. Such an interpretation accords with fossil finds: those from the Port of Ness deposits appear to include only cold water species, whereas those from the beds between Eoropie and Dell Sands include warm as well as cold water forms, together with the supposedly extinct species Turritella pliorescens (Baden-Powell, 1938; 1963).

However, other evidence suggests that the ice limit could have been north of this minimum position (Figure 1). On p. 11 it was noted that the overlying diamicton in the multiple deposits of the north-west coast is apparently little affected by periglacial processes. This suggests that it too could be of Late Devensian age. If so, the Late Devensian ice between Eoropie and Dell Sands would have extended to or, in a few places, slightly beyond the coastline (Figure 1). Alternatively, the mainland ice could have extended a little west of the Butt of Lewis, damming a lake against the coast into which the delta and fan deposits between Traigh Sands and Dell Sands were laid down. The last two solutions have the advantage of placing all the multiple deposits in one age bracket, but are apparently contradicted by the raised beach evidence mentioned in the preceding paragraph. No final decision on the possible Late Devensian ice limits can yet be made, though the minimum position seems to be more firmly based and is preferred here. However, despite the great differences in interpretation of the age of the deposits, the proposed minimum and maximum ice limits differ by no more than a few kilometres.

Events following the Devensian maximum

The retreat of the ice from its maximum position was followed by a vigorous valley glaciation. In Lewis, the evidence for this is confined to three small areas in the south-west, where it coincides approximately with the areas of drumlinoid or 'fluted' drift (Figure 3). In the through valley from Loch Raonasgail to the mouth of the Abhainn Cheann Chusil, there is hummocky moraine and fluted drift which post-date the outwash and moraines associated with the Uig gravels (p. 15). Though the limit of the moraines is sharply defined on the east side of the valley at its southern end (where the glacier evidently extended below present sea-level), there is no marked terminal feature limiting the north-flowing glacier in the vicinity of Loch Raonasgail. In Glen Dibadale immediately to the east, the former maximum extent of a southward-flowing valley glacier is given by a well defined limit to irregular, bouldery moraines formed chiefly of granite blocks, and by a limit of frequent boulders on the west side of the valley. The third area, at the head of Loch Suainaval, which is defined by fluted drift and a drift limit near the southern end of the Loch, evidently carried a broad but short glacier originating below the ridge to the west. The landforms associated with the former valley glaciers bear a strong resemblance to those of the Loch Lomond Readvance of the Scottish mainland (Sissons, 1976, p. 91), but, as discussed below, there is a suggestion that they could be older.

An extensive area of North Harris (Figure 4), from Glen Cravadale in the west to the head of Loch Sealg in the east, was subjected to a valley glaciation characterised by fluted drift and by other landforms which have been discussed above (p. 13). At some localities the ice-flow direction during valley glaciation was at an angle to that of the ice-sheet phase and in Glen Skeaudale it was almost exactly opposite (Figure 2). However, in other places the local valley glaciation cannot easily be distinguished from the preceding ice-sheet phase as the orientation of striae suggests that valley glaciers extended well beyond the limits of drift mounds (Figure 2). Though no landforms clearly indicating a readvance have been mapped (with one exception, discussed next), the valley glaciation of north Harris is correlated with that of south-west Lewis on the basis of similarity of features and deposits in almost contiguous areas of similar relief and present day climate.

In South Harris there is a small area of fluted drift at the back of Coire Roineabhail [NG 047 863] and there are traces of a moraine ridge and an upper limit of boulders extending westwards to the back wall of the corrie from the stream junction [NG 051 865]. North-east directed striae occur by the stream a short distance east of this locality. Further south, in South Uist, there are traces of valley glaciation in Coire Rudale, north of Hecla, where there is a small area of hummocky moraine about Loch a' Choire [NE 817 357]. A crescentic terminal moraine 3 to 4 m high partly encloses the loch on its east side.

In the absence of direct evidence such as dateable organic deposits, the age of the valley glaciation can be resolved only on the basis of indirect and less reliable evidence such as landform assemblages and their relationship to periglacial features. In many of the valleys examined in Harris, the morainic landforms and till slopes pass upwards at about 300 m above OD without a marked break into areas of frost-shattered bedrock with boulder lobes (Figure 4). This is seen for instance in Glen Scaladale where, in the upper valley, moundy moraines and fluted drift are overlain by stone lobes, but the relationship is more difficult to demonstrate in Lewis where the hills are lower. In the Scottish Highlands, stone lobes are thought to be associated with the cold climate of the Loch Lomond Readvance (Sissons, 1976) and their relationship to the morainic landforms in the Outer Hebrides therefore suggests that the valley glaciation here predated the Loch Lomond Readvance. If this is so, the well marked An Coire terminal moraines in Glen Cravadale (p. 15) which resemble those of the Loch Lomond Readvance (Peacock and Ross, 1978), would seem also be predate this event. Peacock (1981) has suggested that the valley glaciation dates to roughly to 13 500 BP, when there may have been a temporary increase in snowfall consequent on the northward and eastward migration of the polar front across Scotland.

Landslip and scree

In comparison with the Scottish mainland and with the Inner Hebrides, landslips are relatively uncommon in the Long Island. On Tolsta Head there is an area of slip and rockfall in the cliffs at one locality [NB 558 468], and a large area of land adjacent to Giordale Sands [NB 543 465] is slightly slipped. Inland in North Harris (Figure 4), the crag forming the north-west face of Sron Ulladale [NB 078 137] has broken away as a rockfall which passes into a debris flow, the toe of which reaches the south end of Loch Ulladale. Screes are widespread in the mountain areas, particularly in Harris. In places, such as below the steep western face of Sron Scourst [NB 106 098] (Figure 4), the base of the scree is covered by large dislodged blocks 3 to 4 m across and the scree and blocks transgress on to the hummocky moraines below. Such large blocks may have been frost-rived during the cold conditions of the Loch Lomond Stade.

Beaches and blown sand

In many places the coastline of the islands is one of submergence, a feature which is particularly noticeable in north-west Lewis where a number of lochs have been impounded by storm beaches. Elsewhere, there are many records of peat and archaeological sites which are now between tidemarks (Ritchie, 1979). Radiocarbon ages of 5700 BP, have been obtained on organic material below low water mark in the Uists (Ritchie, 1966) and 8800 BP, from peat at about 3 m below OD near Stornoway (von Weymarn, 1974). Without supporting evidence, such ages merely show that the sites were above sea level at the times mentioned, but it is significant that the date of 5700 BP refers to a period when sea level on the mainland was falling from its post-glacial maximum, but was not far below this level (Sissons, 1976). It is likely that the relative sea level curve will prove to be similar in shape to that of north-west England, where a rapid sea level rise took place prior to 7500 BP and was followed by several smaller transgressions and regressions superimposed on an overall gradual rise up to the present day (Tooley, 1978).

Storm beaches, with cobbles and boulders, are widespread, particularly on the west coast of the islands. For the most part, the coasts are of Lewisian rocks, but other rock types, including some foreign to the islands, can be found (p. 17). Most of the beaches are modern; remnants of slightly higher, older beaches at Tong on Lewis (Ritchie and Mather, 1970) may have been deposited at a time when coastal processes were different and are not necessarily related to a period of higher sea level (von Weymarn, 1974).

Many parts of the hinterland backing sandy beaches, particularly in the more southerly islands, are characterised by dunes and by extensive flat and gently sloping stretches of blown sand only a little above sea level termed machair. Such sand is usually siliceous, but the calcium carbonate content varies from zero to 80% according to the percentage of comminuted marine shells (Ritchie, 1971). With this variable but significant lime content, the machair forms the most important resource of agricultural land in the Outer Hebrides.

In Lewis and North Harris, small areas of sand dunes and machair are present on the north-west and west coasts southwards from Eoropie [NB 516 650] to Hushinish [NA 990 120], the most extensive (about 1.5 square kilometre) being west of Barvas [NB 36 50]. On the east coast, dunes behind a sandy beach occupy a little under 1 square kilometre of New Tolsta [NB 54 48] and between here and Stornoway salt marshes occur at several localities on the landwards side of dunes or beach. Dunes with machair are present around the great sandy beaches of Traigh Luskentyre, and Northton, both on the westward-facing coast of South Harris.

In the Uists and Benbecula, together with the islands between South Harris and North Uist, blown sand, as dunes and machair, occupies about 10% of the land surface along the low-lying west coast and almost the whole surface of the Monach Islands west of North Uist. Farther south, in Barra, machair development is slight, though there is an area of blown sand occupying more than 1square kilometre at the north end of the island, and smaller patches extend along the west coast. Both here and in Pabbay to the south, blown sand extends inland locally to heights of over 100 m (Jehu and Craig, 1923) and is locally accompanied by layers of impure limestone on the latter island.

Ritchie (1971; 1979) has described a morphological sequence where the beach—machair system is best developed, as in the Uists. The beach, with or without a shingle ridge, is followed inland by a narrow zone of dunes, a belt of flat machair and a further sand ridge or low, seaward-facing escarpment behind which the machair slopes away gradually to a line of seepage lochs and marshland. The development of this sequence is attributed to deflation within a pre-existing dune system, the sand being blown landward and deposited either subaerially or as deltas in the seepage lochs to form an inner slope or plain. The beaches and dune systems predating the machair are believed to have been initiated at a time of rapid post-glacial marine transgression, during which glacial deposits came under wave attack and greater quantities of sand and gravel became available for redistribution. More recently, the volume of beach and dune sand has probably remained either relatively constant or is slowly declining and is thus not to be thought of as a renewable resources. Similarly, the commercial exploitation of gravel from the storm beaches in areas where the supply is not known to be supplemented by long-shore drift (as it is, for instance at Tong (Ritchie and Mather, 1970)) may expose the coastline to marine erosion.

Peat and alluvium

Away from the machair, peat is widespread in the Outer Hebrides, and many of the pockets too small to be mapped even on a large scale have been exploited for fuel by the crofters. In the absence of large rivers, alluvium is chiefly confined to small haughs adjacent to streams. Such haughs frequently merge with peat.

The most extensive deposit of peat is on the Lewis plateau (Macaulay Institute, 1981), where stream sections show it to overlie both till and bedrock. Thicknesses of over 2 m are common, for instance between Skigersta [NB 547 618] and Cladach Cuiashader [545 582] where 3 to 4 m of peat overlie till. On the west side of the plateau the average depth of peat was reported as 2.3 m thinning to about 1 in on steeper slopes (Department of Agriculture and Fisheries for Scotland, 1965). The margins of the peatlands have been improved for agriculture in many coastal areas and adjacent to some main roads.

Southwards, the vast tract of peat in northern Lewis merges into peat-rock complexes, and the morainic drift of south-west Lewis and Harris is commonly overlain by a patchy cover of peat. However, the bare mountains of south-west Lewis are in places devoid of all superficial deposits and in South Harris the only extensive peatland is between Leverburgh and Loch Langavat.

In the Uists and Benbecula, peat is widespread in North Uist and several areas of peat farther south exceed 1 square kilometre in extent. For the most part, peat in these islands is distributed as pockets between rock knobs and drift mounds and there is thin dissected blanket peat on hill slopes. Only very small areas of peat occur on Barra and islands to the south. Though much of the peat is composed of the remains of heath, mosses and sedges, tree stumps occur in places below and within the peat. Pine and birch imbedded in peat are said to occur along the coast of North Uist (Jehu and Craig, 1926, p. 474; Ritchie, 1968) and birch stumps at Holm, near Stornoway (von Weymarn, 1974). During the present survey, stumps of pine were noted below peat near Ardvourlie [NB 183 099] in Harris, and at Lemreway [NB 375 125] and Leurbost [NB 356 274] in Lewis. The basal peat in west Lewis is said to contain woody fragments, chiefly birch, and in this area diatomaceous earth has been reported below peat (Department of Agriculture and Fisheries for Scotland, 1965).

Peat growth may have started early on the islands (8800 BP, near Stornoway; see above), but it probably became general only at a much later date. By analogy with western Scotland (Birks, 1977) this was probably after 4000 BP.

References

BADEN-POWELL, D. F. W. 1938. On the glacial and interglacial marine beds of Northern Lewis. Geol. Mag., Vol. 75, 395–409.

BADEN-POWELL, D. F. W. 1963. Isostatic recovery in Scotland. Nature, London, Vol. 199, 546–547.

BADEN-POWELL, D. F. W. and ELTON, C. 1937. On the relations between a raised beach and an Iron Age midden on the Isle of Lewis, Outer Hebrides. Proc. Soc. Antiq. Scotl., Vol. 71, 347–365.

BINNS, P. E., MCQUILLIN, R. and KENOLTY, N. 1974. The geology of the Sea of the Hebrides. Rep. Inst. Geol. Sci., No. 73/14. 43 pp.

BIRKS, H. J. B. 1977. The Flandrian forest history of Scotland: a preliminary synthesis. Pp. 119–136 In SHOTTON, F. W. (Ed.). British Quaternary studies: recent advances. (Oxford: Clarendon Press.)

BOULTON, G. S. 1976. A genetic classification of tills and criteria for distinguishing tills of different origin. Geografia, Vol. 12, pp. 65–80.

BOULTON, G. S., JONES, A. S., CLAYTON, K. M. and KENNING, M. J. 1977. A British ice-sheet model and patterns of glacial erosion and deposition in Britain. Pp. 231–246 In SHOTTON, F. W. (Ed.). British Quaternary studies: recent advances. (Oxford: Clarendon Press.)

BOWEN, D. Q. 1978. Quaternary geology. A stratigraphic framework for multidisciplinary work. (Oxford: Pergamon Press.)

BROSTER, G. E., DREIMANIS, A. and WHITE, J. C. 1979. A sequence of glacial deformation, erosion and deposition at the ice-rock interface during the last glaciation; Cranbrook, British Columbia, Canada. J. Glaciol., Vol. 23, 283–294.

CATTERALL, J. W. 1975. Further evidence of glaciation in the extreme western Isles of Scotland: a note. Scot. Geogr. Mag., Vol. 91, 12–124.

CHARLESWORTH, J. K. 1956 (1955). The Late Glacial history of the Highlands and Islands of Scotland. Trans. R. Soc. Edinburgh, Vol. 62, 769–628.

CHESHER, J. A., SMYTHE, D. K. and BISHOP, P. 1983. The geology of the Minches, Inner Sound and Sound of Raasay. Rep. Inst. Geol. Sci., No. 83/6.

COWARD, M. P. 1972. The Eastern Gneisses of South Uist. Scott. J. Geol., Vol. 8, 1–12.

COWARD, M. P. 1977. Anomalous glacial erratics in the southern part of the Outer Hebrides. Scott. J. Geol., Vol. 13, 185–188.

DEPARTMENT OF AGRICULTURE AND FISHERIES FOR SCOTLAND. 1965. Scottish peat surveys, Vol. 2. Western Highlands and Islands. (Edinburgh: HMSO.)

DAVIES, G. L. H. and STEPHENS, N. 1978. The geomorphology of the British Isles: Ireland. (London: Methuen.)

DOUGAL, J. W. 1928. Observations on the geology of Lewis. Trans. Geol. Soc. Edinburgh, Vol. 12, 12–18.

ETHERIDGE, R. (Jun.). 1876. Note on the fossils from the glacial deposits of the north-west coast of the Island of Lewis, Outer Hebrides. Geol. Mag., New Ser., Dec. 2, Vol. 3, 552–556.

EVENSON, E. B., DREIMANIS, A. and NEWSOME, J. W. 1977. Subaquatic flow tills: a new interpretation for the genesis of some laminated deposits. Boreas, Vol. 6, 115–133.

FETTES, D. J., MENDUM, J. R., SMITH, D. I. and WATSON, J. V. (In press). The geology of the Outer Hebrides. Mem. Inst. Geol. Sci.

FLINN, D. 1979 (1978). The glaciation of the Outer Hebrides. Geol. J., Vol. 13, 195–199.

GEIKIE, J. 1873. On the glacial phenomena of the Long Island or Outer Hebrides. J. Geol. Soc. London, Vol. 29, 532–545.

GEIKIE, J. 1878. On the glacial phenomena of the Long Island or Outer Hebrides. J. Geol. Soc. London, Vol. 34, 819–870.

GEIKIE, J. 1894. The Great Ice Age. 1st ed. 1874; 2nd ed. 1877 (London: Dalay, Ibister and Co.). 3rd Edition, 1894 (London: Stanford).

GODARD, A. 1965. Recherches de geomorphologie en Ecosse du Nord-Ouest. (Strasbourg: University of Strasbourg.) 701 pp.

GRAY, J. M. 1978. Low-level shore platforms in the south-west Scottish Highlands: altitude, age and correlation. Trans. Inst. Br. Geogr., New Series, Vol. 3, 151–164.

HALDORSEN, S. 1982. The genesis of tills from Astadalen, southeastern Norway. Nor. Geol. Tidskr., Vol. 62, 17–38.

HOPESON, D. 1982. Hummocky and fluted moraines parts of north-west Scotland. Unpublished PhD thesis, University of Edinburgh.

HOLTEDAHL, D. (Ed.) 1960. Geology of Norway. Nor. Geol. Unders. Nr 208, 1–540.

JEHU, J. T. and CRAIG, R. M. 1923–1934. Geology of the Outer Hebrides. Trans. R. Soc. Edinburgh. 1923a. I The Barra Isles, Vol. 53, 419–441. 1923b. II S. Uist and Eriskay. Vol. 53, 615–641. 1926. III N. Uist and Benbecula. Vol. 54, 467–489. 1927. IV South Harris. Vol. 55, 457–488. 1934. V North Harris and Lewis. Vol. 57, 839–874.

MACAULAY INSTITUTE FOR SOIL RESEARCH. 1981. Peat and terrain categories of Lewis and North Harris. Map at 1:100 000 scale.

MCCANN, S. B. 1968. Raised shore platforms in the western isles of Scotland. In BOWEN, E. G., CARTER, H. and TAYLOR, J. A. (Eds.). Geography at Aberystwyth, 22–34. (Cardiff: University of Wales Press.)

MILNE HOME, D. (Convener) 1880a. Fifth report of the Boulder Committee. Proc. R. Soc. Edinburgh, Vol. 10, 113–199.

MILNE HOME, D. 1880b. Sixth report of the Boulder Committee. Proc. R. Soc. Edinburgh, Vol. 10, 577–635.

MILNE HOME, D. 1884. Tenth and final report of the Boulder Committee. Proc. R. Soc. Edinburgh, Vol. 12, 765–884.

MITCHELL, G. F. 1972. The Pleistocene history of the Irish Sea: second approximation. Sci. Proc. R. Soc. Dublin, Vol. 4, 181–199.

OKKO, V. 1955. Glacial drift in Iceland: its origin and morphology. Bull. Commun. Geol. Finl., No. 170, p. 1–133.

PEACOCK, J. D. 1980. Glaciation of the Outer Hebrides: a reply. Scott. J. Geol., Vol. 16, 87–89.

PEACOCK, J. D. 1981. Report and excursion guide - Lewis and Harris. Quaternary Newsl., No. 35, 45–54.

PEACOCK, J. D. and ROSS, D. L. 1978. Anomalous glacial erratics in the southern part of the Outer Hebrides. Scott. J. Geol., Vol. 14, 262.

RITCHIE, W. 1966. The post-glacial rise in sea-level and coastal changes in the Uists. Trans. Inst. Br. Geogr., Vol. 39, 79–86.

RITCHIE, W. 1968. The coastal geomorphology of North Uist. O'Dell Mem. Monogr., No. 1. (Aberdeen: Geography Dept., University of Aberdeen.) 32 pp.

RITCHIE, W. 1971. The beaches of Barra and the Uists. (Aberdeen: Geography Dept., University of Aberdeen.) 83 pp.

RITCHIE, W. 1979. Machair development and chronology in the Uists and adjacent islands. Proc. R. Soc. Edinburgh, Vol. 77B, 107–102.

RITCHIE, W. and MATHER, A. J. 1970. The beaches of Lewis and Harris. (Aberdeen: Geography Dept., University of Aberdeen.) 113 pp.

ROBERTSON, D. 1880. On a series of foraminifera and ostracoda from a post-Tertiary deposit in Lewis. Trans. Geol. Soc. Glasgow, Vol. 7, 248–250.

SISSONS, J. B. 1976. The geomorphology of the British Isles: Scotland. (London: Methuen.)

SISSONS, J. B. 1980. The glaciation of the Outer Hebrides. Scott. J. Geol., Vol. 18, 81–84.

SISSONS, J. B. 1982. The so-called high 'interglacial' rock shoreline of western Scotland. Trans. Inst. Br. Geogr., New Ser., Vol. 7, 205–216.

SMITH, D. I. and FETTES, D. J. 1979. The geological framework of the Outer Hebrides. Proc. R. Soc. Edinburgh, Vol. 77B, 75–83.

STEEL, R. J. and WILSON, A. C. 1975. Sedimentation and tectonism (?Permo-Triassic) on the margin of the North Minch basin, Lewis. J. Geol. Soc. London, Vol. 131, 183–202.

SUTHERLAND, D. G., BALLENTYNE, C. K. and WALKER, M. J. C. 1982. A note on the Quaternary deposits and landforms of St Kilda. Quaternary Newsl., No. 37, 1–5.

TOOLEY, M. J. 1978. Sea-level changes: north-west England during the Flandrian Stage. (Oxford: Oxford University Press.)

VON WEYMARN, J. 1974. Coastal development in Lewis and Harris, Outer Hebrides, with particular reference to the effects of glaciation. Unpublished PhD thesis, University of Aberdeen.

VON WEYMARN, J. 1979. A new concept of glaciation in Lewis and Harris. Proc. R. Soc. Edinburgh, Vol. 77B, 97–105.

VON WEYMARN, J. and EDWARDS, R. J. 1973. Interstadial site on the Island of Lewis, Scotland. Nature, London, Vol. 246, 473–474.

WATSON, J. 1977. The Outer Hebrides: a geological perspective. Proc. Geol. Assoc., Vol. 88, 1–14.

WELCH, R. and HOWARTH, P. J. 1968. Photogrammetric measurement of glacial landforms. Photogramm. Rec., Vol. 6, 75–96.

WILKINS, D. A. 1984. The Flandrian woods of Lewis (Scotland). J. Ecol., Vol. 72, 251–258.

WYLLIE, P. J. 1956. Ice recession in Dronning Louise Land, North-east Greenland. J. Glaciol., Vol. 2, 704–708.

Figures, plates and tables

Figures

(Figure 1) Quaternary deposits, landforms and ice movement directions

(Figure 2) Superficial deposits of north Lewis

(Figure 3) Glacial and periglacial features and deposits of south-west Lewis

(Figure 4) Glacial and periglacial features and deposits of north Harris

Plates

(Plate 1, fig. 1) North coast of Vatersay looking towards Castle Bay, Barra, with Heaval (383 m) in background. Roches moutonnees show ice-flow from left to right (north-west to south-east). D2933. P008321.

(Plate 1, fig. 2) Coast north of Galson, adjacent to Tempeall nan Cro Naomb, Lewis [NB 433 593]. Raised beach gravel overlain by soliflucted till and gravel. D3243. P008390.

(Plate 2, fig. 1) Coast 1 km south of Eoropie, north Lewis [NB 508 643]. Till overlain by carbonate cemented raised beach gravel with slumped till above. D 3241. P008388.

(Plate 2, fig. 2) Section [NF 673 048] on the coast near Chad (Cleat), Barra. Raised beach gravel overlain by till. Water smoothed rock below the beach at left. D2929. P008316

(Plate 3, fig. 1) Quarry [NF 950 723], Lochpartain, North Uist. Part of rock cored drumlin with fractured rock passing upwards into till with angular blocks. D2886. P008276

(Plate 3, fig. 2) Samala, Baleshare, North Uist. Section [NF 800 610] in till with angular boulders. D 2898. P008286

(Plate 4, fig. 1) Glen Skeaudale, Harris [NB 153 031]. Asymmetrical tranverse moraine with flutes on its proximal (right) side. D3222. P008378

(Plate 4, fig. 2) Glen Valtos, Lewis [NB 064 343]. This magnificent subglacial channel carried meltwater eastwards towards Loch Roag. D3231. P001015

(Plate 5) Asymmetrical transverse moraines on the valley side east of Loch Langavat [NB 1614] and [NB 1615]. North at top of photograph; distance across photograph (west to east) about 850 m. Ice movement from south (bottom) to north (top). Enlarged from part of Ordnance Survey air photograph OS/66/100, No. 118 and reproduced with the permission of the Director, Ordnance Survey. Crown copyright reserved.

(Front cover)

(Rear cover)

Tables

(Table 1) Summary of Pleistocene History, Outer Hebrides

Tables

Table 1 Summary of Pleistocene History, Outer Hebrides