Content and licensingview original scan buy a printed copy

St. Kilda: an illustrated account of the geology

By R. R. Harding, R. J. Merriman and P. H. A. Nancarrow. With special chapters by M. Brooks, G. P. Durant and G. E. Morgan

Bibliographic reference: Harding, R.R. and Nancarrow, P.H.A. 1984. St. Kilda: an illustrated account of the geology. BGS Report Vol. 16, No. 7. Keyworth: British Geological Survey

British Geological Survey, Natural Environment Research Council

BGS Report Vol. 16, No. 7

Her Majesty's Stationery Office 1984. © Crown copyright 1984. ISBN 0 11 884360 5

Chapter 1 St. Kilda: an illustrated account of the geology

Summary

The islands of St Kilda consist of a range of intrusive igneous rocks which were formed in a Tertiary volcano about 60 million years (Ma) ago. The oldest rock is the Western Gabbro (E^w), a banded and layered intrusion with an overall dip of 45°E in the north and 45°N in the south. At its eastern margin parts of the Gabbro have been intensely sheared and recrystallised prior to intrusion of the younger rocks of the Mullach Sgar Complex. The Cambir Dolerite intrudes the Western Gabbro on the western cliffs of the Cambir as thin veins which have a fine-grained metamorphic granoblastic texture. Blocks of gabbro, identical to the Western Gabbro, have been mixed with other gabbros and dolerites and form a widespread breccia (EK) which comprises most of Boreray, Soay and Glacan Mor. The gabbros may once have formed a large layered intrusion which underwent metamorphism, disintegration and finally intrusion by tholeiites and microbreccias. South of the tunnel in Glen Bay the igneous breccias (EK) is cut by the Glen Bay Gabbro (E^G). The gabbro has been chilled to a glassy basalt against the breccias, and it itself divided into two parts by intrusion of the Glen Bay Granite (G). This represents the first evidence of granitic activity on St Kilda and immediately preceded formation of the Mullach Sgar Complex. The Complex consists of four major intrusive phases, each having a mafic and a felsic component, some resembling ring dykes and others extensively fragmented. The last major intrusion on St Kilda was the Conachair Granite (I), a leucocratic, very drusy rock intruded 55 Ma ago with an initial ⁸⁷Sr/⁸⁶Sr ratio of 0.7041. Late dykes and sheets of a range of dolerites and felsites cut the major intrusions and represent the last phase of igneous activity. Neither lavas nor Tertiary sediments have been found on St Kilda although minerals from contact and regional metamorphic environments occur in the stream sediments and probably represent the residues from rocks that originally enclosed the volcanic centre. Early faults trend NW–SE on Hirta and a later series of NE–SW tensional faults are partly responsible for such features as the Dun Passage and the Cambir neck. Palaeomagnetic data indicate that the major intrusions crystallised in a reversed-polarity magnetic field between 50 and 60 Ma ago, and that St Kilda has drifted north about 19° since that time.

The dominant topographical features of St Kilda are the result of Quaternary glaciation and a small glacier probably occupied Village Bay during the Devensian glacial maximum. Today St Kilda is part of a drowned landscape.

Introduction

The islands of St Kilda lie 65 km WNW of North Uist at the edge of the Outer Hebridean shelf. This part of the earth's crust is composed largely of Precambrian gneisses, with Mesozoic sediments in the Flannan trough and a thin impersistent cover of Recent sediments. At the edge of the continental shelf Mesozoic and Recent sediments thicken westwards to more than 2 km at the north end of the Rockall Trough. Although the structure of the submarine rocks is complex near St Kilda, seismic data indicate that basic intrusions are close by, and St Kilda lies close to a significant change in direction of the continental margin. From the latitude of St Kilda the margin runs due south for 350 km to the latitude of northern Ireland, and to the north' it trends NE for 900 km towards Norway. Very old shear directions trending NE–SW and NW–SE have been mapped in the Precambrian of the Outer Hebrides and detected in the submarine shelf by seismic work, and the intersection in the St Kilda area of a major shear trending NW from South Uist with tensional faults trending N–S and NE–SW indicate that the location of the volcanic centre was influenced by ancient planes of weakness in the crust.

The first comprehensive account of the geology of St Kilda was written by A. M. Cockburn in 1935 and he gave summaries of the earlier accounts of Macculloch (1819), Ross (1884) and Sir Archibald Geikie (1897) before providing detailed descriptions of the rocks he had mapped during the summers of 1927 and 1928. Since then aspects of the glacial features of the islands have been discussed by Wager (1953) and Sutherland and others (1982); aspects of the petrology and geochemistry have been published by Harding (1966; 1967; 1982) and Meighan (1979); the regional setting of St Kilda was described by Jones (1981); and summaries of the general geology have appeared in numerous books and articles whose main theme has been some other aspect of St Kilda's natural history. A list of references and suggestions for further reading is given at the end of the report.

The importance of St Kilda from cultural, regional and scientific points of view is now recognised and one of the fundamental requirements for making decisions about future developments affecting the islands is a source of reliable scientific data. To fulfill the geological aspect of this requirement, a team was formed in 1978 to produce a geological map and detailed report on the geology of the island group. Field mapping was carried out by the authors during the summers of 1978 and 1979 and sample collecting for laboratory analysis was shared with those contributing special chapters. The map contains its own summary of the geology and four itineraries to enable certain aspects of the geology to be seen in half or full day periods. The Report has been designed along modular lines and the list of contents shows that in general each major topic occupies a double page.

Chapter 2 Western Gabbro: E^W Field description

Keywords: banding, cumulate, alteration

The western cliffs of Hirta and Dun consist largely of dark gabbroic rocks (the Western Gabbro), only superficially weathered but massively jointed and faulted into a spectacular landscape. They are the oldest rocks exposed in the St Kilda island group and are cut by offshoots of major intrusions on the Cambir (by the Cambir Dolerite), in Gleann Mor (by the Glen Bay Gabbro and Mullach Sgar Complex), and on Ruaival (by the Mullach Sgar Complex). In addition there are numerous minor intrusions of dolerite and felsite in the form of inclined sheets and dykes. The western limit of the Western Gabbro lies under the sea and its form is not known but the eastern limit lies north-south in Gleann Mor and northwest-south-east on Ruaival and Dun. The present thickness of the gabbro is more than 360 m but the easterly dip of the banding on the Cambir and the northeasterly dip on Dun indicate that it may be a remnant of a much bigger saucer-shaped intrusion whose centre lay between Hirta and Boreray.

The Western Gabbro is made up of coarse-grained rocks consisting essentially of plagioclase feldspar, calcium-rich pyroxene and olivine, with minor amounts (< 5%) of orthopyroxene, magnetite, ilmenite, amphibole, chlorite or spinels. The definition of this rock as a gabbro (as with the naming of other igneous rocks on St Kilda) follows the scheme proposed by the LUGS Subcommission and described by Streckeisen (1976). Different mineral proportions and different mineral textures, which are visible particularly on weathered slabs, have enabled division of the gabbros into 3 types. Type 1 occurs on the Cambir, lower slopes of Mullach Bi and south east Dun, and consists of grains of plagioclase, pyroxene and olivine of sub-rounded or sub-angular shape up to 1 cm across in relatively consistent proportions, although some feldspar-rich or pyroxene-rich bands are present (Figure 3B). Type 2 gabbro lies above Type 1 on Dun and south of Mullach Bi and constitutes the higher parts of Ruaival. it is characterised by black pyroxene grains up to 3 cm across with inclusions of small white plagioclase crystals, the contrast creating an overall speckled appearance on the weathered surface (Figure 3C). Rounded olivine grains are of similar size to the feldspar in this rock and weather honey brown. The third group of gabbros (Type 3) contains many textural and structural varieties (Figure 3A), (Figure 3D) and it is this variability that distinguishes it from Types 1 and 2. The best exposures are on the upper slopes of Mullach Bi where fine granulitic and coarse pegmatitic rocks are interbanded with granular and poikilitic gabbros similar to Types 1 and 2.

The attitude of the boundary between the gabbros of Types 1 and 2 is indistinct and gradational but broadly coplanar with the sporadic banding and layering which dips at about 45° to the east on the Cambir and to the north-east on Dun. The transition between Types 1 and 2 and Type 3 is also gradational but in the latter the character and attitude of the banding is more disturbed. In Type 3 gabbros, quite different textural types are in close proximity, and many structures visible in the cliffs between Mullach Bi and Claigeann Mor resemble those in sedimentary rocks. Size grading, mineral grading, detached blocks, intraformational slumping and faulting and local breccias are all features visible in this part of the Western Gabbro.

The disposition of the types of gabbro and the smaller scale banding and textural features which they display indicate that the Western Gabbro was formed mainly by crystal accumulation. The consistent textures of Types 1 and 2 suggest that tranquil conditions prevailed during their formation but the variability in Type 3 gabbros indicates rapid (if small) changes in magma composition or in pressure and temperature conditions, with associated differential movement of liquids, mushes and even blocks of gabbro. The layering and textural variation in the Western Gabbro is mainly the result of concentrations of primary or cumulus minerals. This is reflected in the chemical composition of the main rock types which do not correspond to known basaltic lavas. It seems likely that the concentration of variable amounts of olivine, pyroxene or plagioclase was effected firstly by mechanical crystal-sorting during magma flow and intrusion, and secondly by nucleation-diffusion during solidification of the basaltic magma. Both processes probably operated to different degrees in formation of the three gabbro Types: the structures and textures in Type 3 gabbros indicate that crystal-sorting was a dominant factor in their formation, in Type 2 the bulk composition and mineral textures suggest that crystal-sorting and diffusion were equally significant processes, while in Type 1 there is less evidence for either process.

In many parts of the Western Gabbro, gabbro pegmatites and narrow fine-grained green veins stand out on the weathered surfaces. Both features were formed after solidification of the gabbro, the former filling tension cracks and the latter occupying both tension and shear fractures. The shear veins are more abundant near the eastern margin of the gabbro and in places relative rotation of the banded rocks has occurred. At the top of Carn Mor, faulting is probably responsible for the close juxtaposition of shallow and steep banding in the Gabbro. Faults and tension joints with a south-west-north-east trend are preferential sites for erosion and are a major cause of the irregular shape of the west coast.

Chapter 3 Western Gabbro: petrology

Keywords: mineralogy, texture, microprobe analyses

Feldspar, clinopyroxene and olivine form a number of interlocking granular and poikilitic textures in the Western Gabbro and some are illustrated in the photomicrographs (30 µm thick sections of rock photographed through a polarising microscope). The optical properties and composition of many minerals may be determined from such sections, and textural details of their relationships within the rock give some idea of how the rock was formed.

Plagioclase, a mixture of anorthite (An, CaAl₂Si₂O₈) and albite (Ab, NaA1Si₃O₈), is the most abundant mineral, generally forming 50–75% of the rock although in places it may be scarce ( < 30%) or, as in (Figure 3B), concentrated in thin anorthosite bands (100%). In the Western Gabbro most grains have calcium-rich cores of composition between An₈₈Ab₁₂ and An₇₅Ab₂₅, and one or more outer zones that vary between An₈₀Ab₂₀ and An₅₅Ab₄₅. Sharply defined zone boundaries in some crystals indicate rapid changes in conditions of crystallisation and suggest that these grains have been moved into different parts of magma chamber, perhaps by flow in a silicate melt or crystal mush, whereas gradational zoning shown by a large proportion of the plagioclase indicates much slower changes in the chemical and physical environment during crystallisation. Rocks containing unzoned crystals are probably a result of static conditions where the process of ionic diffusion supplied material from the main body of magma to the growing crystals. Each plagioclase grain within the area of a thin section (2 cm²) does not necessarily show the same zonal or compositional pattern, some grains may be unzoned, some with 2 or 3 zones and a few with 5 or 6, and this suggests that these crystals have accumulated from different environments. In many gabbros the first-formed plagioclase crystals are calcic and, as the intrusion cools, more sodic compositions are added. This process may be recorded in the feldspars as a 'normal' zoning pattern where progressively more sodic zones crystallise towards the grain margin. Normal and reverse zoning patterns are common in the Western Gabbro feldspars but an unusual feature of many grains is that, regardless of their inner zonal variation, they have relatively calcic rims.

In Type 2 gabbros (from Mullach Bi to Ruaival and Dun) olivine and plagioclase form a granular mosaic in which crystal shape has been determined by the mutual interference of grains during cooling and solidification of the rock. One can envisage a slowly cooling basic magma from which primary or cumulus crystals of olivine and plagioclase accumulated to the extent of about 60% of the mass. At this stage mechanical movement probably ceased and on further cooling of the liquid between the crystals (intercumulus liquid) more plagioclase and olivine formed on the primary grains, and large poikilitic clinopyroxenes grew into the remaining spaces. Nearly all the pyroxene in Type 2 gabbro was formed in this way, and its average size of 20 mm and its skeletal form is in marked contrast to the granular habit of pyroxene in some Type 3 gabbros (Figure 4B), and the modified granular habit characteristic of Type 1 gabbro. In Type 3 gabbros near Mullach Bi and Claigeann Mor, a third kind of pyroxene is part cumulus and part intercumulus in origin (Figure 4C), the cores having commenced crystallisation early and containing feldspars which are much smaller than those outside the pyroxene and the rims enclosing grains at a much later stage in their growth. These pyroxenes resist weathering to a greater extent than olivine and plagioclase and stand out on the gabbro surfaces creating a characteristic knobbly appearance (Figure 3D).

Despite these different habits, the compositional range of the clinopyroxene is small and, in terms of its three major components, the molecular percentages are: MgSiO₃, 42–45; CaSiO₃, 43–45; and FeSiO₃, 11–15. These values were obtained by analysing grains in polished thin sections with an electron microprobe. On the pyroxene diagram (Figure 5) they lie in the augite field, close to augites from gabbros in Glacan Mor, Boreray and Soay, but quite separate from clinopyroxenes in the Cambir Dolerite and the Glen Bay Gabbro. Orthopyroxene (hypersthene) and olivine compositions are shown on the same diagram and where these occur in the same slide as the augite this is indicated by a tie-line. Olivine composition in the Western Gabbro expressed as molecular proportions of forsterite (Mg₂SiO₄) and fayalite (Fe₂SiO₄) generally varies between Fo₇₈Fa₂₂ and Fo₇₀Fa₃₀ but individual crystals are unzoned. Orthopyroxene also shows a small range in composition and is typically found intergrown with opaque minerals (Figure 4A).

Magnetite (Fe₃O₄) and ilmenite (FeTiO₃) occur in small amounts as discrete irregular grains or intergrown with orthopyroxene (Figure 4A). In thin section they are opaque but examination in reflected light reveals a considerable amount of intergrowth. Microprobe analysis of the grains further indicates wide variation in magnetite composition with up to 25% TiO₂, 19% Cr₂O₃, 1% NiO or 1% V₂O₃ present in different grains, and up to 10% spinel (MgAl₂O₄) in the magnetite intergrown with orthopyroxene. Minor amounts of relatively pure magnetite occur as needles in clinopyroxene and as granules on the edges of altered olivine grains. Green spinel is a minor and sporadic constituent of Type 3 gabbros and occurs next to olivine or surrounded by amphibole and chlorite in zones of alteration. Its composition is (Mg₀.₆Fe₀.₄)Al₂O₄ with possibly a trace of chromium. Minute quantities of sulphide minerals (mainly chalcopyrite with a little pyrite) occur either on the boundaries of larger grains or included in plagioclase.

The wide range in abundance of the minerals in different parts of the Western Gabbro is associated with a comparable range in whole-rock chemical composition, and it is difficult to determine the average composition of the gabbro. For this reason selected analyses which illustrate the range are given in the table and in (Figure 5) the range of the olivine and pyroxene composition is set in the context of values obtained from analyses of these minerals in other St Kilda rocks.

Chapter 4 Cambir Dolerite D

Keywords: mineral analyses, annealed textures, cooling history

At the north west end of the Cambir the Western Gabbro is intruded by sheets and veins of the fine-grained Cambir Dolerite. It is exposed on the Cambir cliffs facing the island of Soay, and also near sea level below Mullach Bi where south-westerly-dipping sheets of similar rock cut the Western Gabbro. On weathered surfaces small variations in grain size give the Dolerite a streaky appearance and altered poikilitic olivine crystals form vague bands of elongate black blotches. Contacts with the Western Gabbro are sharp and transgressive to the banding and lamination but variable in attitude and orientation; overall the Dolerite probably dips steeply south-east. Outcrops of Dolerite near the grass-mantled gully below and west of the Cambir summit are variable in grain size and in places distinction from the Western Gabbro is difficult. Near the top of the gully at about 500 ft (160 m) there is an isolated intrusion of medium-grained gabbro similar to some on Soay. Like the Western Gabbro, the Cambir Dolerite is intruded by coarse gabbroic and pegmatitic veins and affected by the same kinds of faulting and shearing which gave rise to the network of veins.

More than 90% of the Dolerite is made up of plagioclase, augite and opaque minerals, and the rest consists of olivine, orthopyroxene and a small quantity of alteration products. Most plagioclase grains are small and unzoned, ranging in composition from An₅₀ to An₆₈ in different rocks; some larger crystals are more calcic with complex zoning patterns and may be xenocrysts derived from the Western Gabbro. Clinopyroxene grains, making up about 40% of the rock, have a relatively constant composition of En₃₈ Wo₄₂ Fs₂₀, and are generally less magnesian than the augite(s) of the Western Gabbro. the opaque grains are mostly magnetite with variable contents of either ilmenite lamellae or ulvospinel (Fe₂TiO₄) and spinel (MgFeA1₄O₈) granules, and this is reflected in the TiO₂ content which in some grains reaches 25 % . Irregularly shaped ilmenite grains occur in association with magnetite but are not common. Orthopyroxene grains, of smaller size and shape to the granular clinopyroxenes, are found in a few rocks, but this mineral occurs more commonly as poikilitic platy crystals enclosing plagioclase. Olivine also occurs in this form but not generally in the same rock and this mutually exclusive development of olivine in some rocks and orthopyroxene in others may be related to local temperature and oxidation conditions. Commonly, magnetite mantles the poikilitic olivine grains and there is some alteration to chlorite in the interior; fresh grains however range in composition from Fo₆₃Fa₃₇ to Fo₅₃Fa₄₇ (Figure 5). The sheets of dolerite below Mullach Bi have a similar mineralogy but they are coarser, with granular olivines and an overall intergranular texture. Parts of the Dolerite on the Cambir also are intergranular but the main mass is granulitic in texture, resembling a metamorphic rather than an igneous rock. This texture is characteristic of basic rocks that have been heated and maintained at a high enough temperature for sub-solidus recrystallisation (annealing) of the minerals to have occurred. Thus the sharp contacts and the transgressive nature of the veins of Cambir Dolerite in the Western Gabbro indicate that it was a liquid or a crystal mush when intruded, while the textural evidence suggests that much of the Dolerite has suffered a complex cooling history.

Although the chemical analyses of the dolerites on the Cambir (col. 1) and Mullach Bi (col. 2) are similar, the latter is slightly more magnesian and less silicic; there is some variation in the trace elements and it is possible that the example of dolerite from Mullach Bi is related more to the suite of gabbros and dolerites which form parts of Soay, Glacan Mor and Boreray. Certainly the range of sulphides found in this dolerite was greater than in the main Cambir dolerite and these include chalcopyrite, nickeliferous pyrite and a member of the linnaeite–violarite family (Ni_1.8Co_0.8Fe₀.₆)S₄.

Chapter 5 Breccia of gabbros and dolerites EK: Field description

Keywords: complex rock mixture, age relations, microbreccias, explosive vulcanicity

Viewed from Village Bay (Figure 8A), the basic rocks of Mullach Geal and Mullach Mor form sporadic outcrops and tors along the skyline above the quarry. They are the southermost part of an igneous Breccia (EK) that consists of basalt, dolerites and gabbros ranging in texture from fine-grained granular rocks to coarse and pegmatitic varieties, some with black pyroxenes up to 10 cm long which are prominent on weathered surfaces. Coarse and fine rocks are intimately mixed such that contacts can rarely be traced for more than a few metres. Gabbro-dolerite contacts may be sharp or transitional although both rock types show similar patterns of shearing and granulation. Finer-grained, flinty basalts cut both the gabbro and granular dolerite, often in tortuous fashion, and may be distinguished from the latter by their chilled contacts, abrupt grain-size variation and less intense shearing.

On Hirta the most extensive exposures of the Breccia are found on the cliffs of Glacan Mor and around Gob na h-Airde with similar rocks occurring low on the Cambir (Figure 8B). Boreray, Soay and the stacs also consist largely of igneous Breccia (Figure 9A). The dominant rock type in the Breccia is gabbro. Various kinds are present although much is texturally similar to the Western Gabbro (E^W). These gabbros have been brecciated, veined and intruded by basaltic fluids which display a range of textures. Some have been chilled against cold gabbro blocks and show black glassy margins, others have come into contact with hot gabbros and have more crystalline edges, and yet others are fragmental rocks which consist of basalt and dolerite fragments of various sizes and compositions resting in a fine-grained matrix of the same material. In some rocks brecciation and veining by basalt has occurred repeatedly so that changes to wall-rock temperatures have resulted in different degrees of recrystallisation of the veins, and this has led to a variety of textures from doleritic to fine-grained granoblastic being developed over distances of a few metres. The granoblastic rocks are commonly hard and flinty, dark grey, black or dark green depending on the degree and nature of subsequent hydrothermal alteration. The most abundant basaltic rock on Glacan Mor is extremely fine-grained and splintery, almost black, and closely jointed. On some weathered surfaces the closely spaced jointing of the ,basalt veins contrasts with the less jointed gabbro blocks and reveals the inconstant and unpredictable nature of the contacts. The general fracture pattern suggests processes of formation which have involved the transport and mixing of large solid blocks of basic rock by mobile basaltic material. These structures are exposed in Glacan Mor and in places on the eastern rock shelves of Glen Bay, especially south of the tunnel where the straight, steeply dipping chilled margin of the Glen Bay Gabbro cuts across disoriented blocks in the Breccia. A type of gabbro common as blocks on Glacan Mor, Soay and Boreray is exposed for about 200 ft above the quarry in Abhainn Mhor. Here, it is an ophitic gabbro very variable in grain size and texture, largely consisting of feldspar and pyroxene in a coarse ophitic texture with minor but significant quantities of fine-grained and pegmatitic varieties. Amygdales, filled with green amphibole, chlorite and clay minerals are present in a few places and some form elongate groups parallel to mineral lamination displayed by the feldspar. The lamination and the zones of amygdales dip at about 20°N into Mullach Mor, roughly parallel to the inferred contact between the Breccia and the Mullach Sgar Complex.

On Boreray and Soay some large blocks of banded, layered and sheared gabbro within the breccia are indistinguishable from the Western Gabbro (E^W) of Hirta and Dun. One block, texturally identical with the E^W Type 2 gabbro on Ruaival, extends for about 100 m along the top of the main south-facing grass slope on Boreray. To the west and east of this block are dolerites and gabbros of varying shades of dark green and dark grey, generally fine-grained but with sporadic coarser patches where white feldspar and dark pyroxene or amphibole form conspicuous ophitic textures. In places along the ridge north of the summit of Boreray (Figure 9A), fragmental textures similar to those on Glacan Mor are visible on weathered surfaces and there is little doubt that the assemblages of rock types, their structures and the textures are the same as those on Hirta. Breccias on Soay show the same features, with dolerites of different kinds intrusive into a range of gabbros. A large raft of gabbro similar to the Type 1 variety of E^W forms a prominent cliff at 700 ft facing the Cambir, and other blocks of similar type occur on the northern and western cliffs of Soay. Many blocks are sheared and show metamorphic features similar to the gabbros of Mullach Mor and Mullach Geal.

Evidence of the age of the Breccia relative to other phases of intrusive activity on St Kilda is indicated by field relationships on Hirta. On Mullach Geal dolerite and microgranite sheets cutting the Breccia are offshoots of the Mullach Sgar Complex (MSC) well exposed in the quarry beneath. The contact between the Breccia and MSC appears to dip gently north. On the cliffs east of Glacan Mor igneous breccias are cut by dolerites and granites of the MSC. In the overhanging cliffs above Na Cleitean both MSC and EK rocks are cut by veins of Conachair Granite (I). Thus the breccias appear to predate the Mullach Sgar Complex. Field evidence from the east side of Glen Bay suggests that breccias also predate the Glen Bay Gabbro (E^G) which is chilled against it. However, within the Glen Bay Gabbro, a coarse texture with sub-vertical layering is developed at a distance of 100–120 m west of the contact, and the bulk of the Gabbro has clearly crystallised under plutonic conditions. It thus seems that chilling is more likely the result of collapse of a large block of cold country rock (EK gabbros and dolerites) into the magma rather than of upward intrusions into relatively cold country rocks in sub-plutonic conditions. The possible size of such a collapse block is difficult to estimate, but it seems to have been brecciated and veined by basalt and in places hydrothermally altered prior to making contact with the Glen Bay Gabbro. No doubt further brecciation and basalt veining (by E^G) accompanied collapse. Thus the field evidence suggests that on Hirta some igneous brecciation was contemporaneous with the crystallisation of the Glen Bay Gabbro. The breccias have formed from layered gabbroic country (E^W) rocks at fairly high sub-solidus temperatures which were close to or within magma conduits. Collapse following rapid evacuation of magma, perhaps accompanying surface eruption, is a likely feature of sub-volcanic magmatism. So also is the variety of dolerites and basalts intrusive in single or multiple veins, together with explosive microbreccias and hydrothermal alteration. These breccias, which are the most widespread rock type in the island group, are the most direct evidence of the explosive surface volcanism which was centred on St Kilda.

Chapter 6 Breccia of gabbros and dolerites: petrology

Keywords: textural range, annealing, hydrothermal minerals

The gabbros and dolerites of the Breccia display a wide range of textures and grain-size, from coarse to fine and occasionally glassy types, including fresh rocks and hydrothermally altered varieties. Coarse gabbros are rarely fresh, most showing evidence of some alteration with amphibole, chlorite and serpentine developed in place of olivine, veins of albite visible in the plagioclase crystals (Figure 10A), and significant amounts of epidote in a few places. Fresh olivine (ranging from Fo₆₅ to Fo₆₀) however remains in some gabbros as the cores of rounded grains with marginal alteration to fibrous chlorite (pycnochlorite) and amphibole (cummingtonite) with a scattering of magnetite grains. Feldspar compositions generally range from An₇₂Ab₂₈ to An₄₀Ab₆₀ and most grains are zoned over part of this range, but some show more extreme zoning to rims of An₂₆Ab₇₄ in areas of residual crystallisation. Late veins of albite (An₅Ab₉₅) occur throughout the breccia but alkali feldspar is found only in interstitial patches with quartz in a few gabbros, a scarcity reflected in the low content of K₂O in the chemical analysis (column 3). Orthopyroxene (En₆₇Wo₂Fs₃₁), like olivine, is uncommon and susceptible to alteration, but clinopyroxene is abundant, is relatively fresh and has a limited range of composition from En₄₂Wo₄₄Fs₁₄ to En₄₀Wo₄₀Fs₂₀ (Figure 5). Ilmenite and magnetite are common and show considerable variation in their contents of MnO, Cr₂O₃ and V₂O₃. Tiny amounts of iron and copper sulphides occur in the groundmass. Gabbros with this mineralogy make up perhaps one fifth of the Breccia and are of similar abundance to gabbros thought to be fragments of the Western Gabbro (E^W). The latter in general are so recognised on the basis of mineralogy and structure but some blocks, notably on Mullach Mor, Mullach Geal and on Soay, have suffered more intense metamorphism than the rocks on the west coast. They consist of large crystals (relics of the original gabbro) which have been broken and partly recrystallised to form a matrix of tiny granules of olivine, pyroxene, feldspar and opaques (Figure 10B). The composition of the relict olivine and the newly-formed granules are very similar (Fo₆₅ to Fo₆₂), but there is considerable variation in the compositions of the other minerals. In some areas, granules of Ann lie adjacent to a large crystal of that composition whereas, in another part of the slide, granules of An₆₅ exist next to zoned larger grains; the rocks are markedly inhomogeneous both texturally and compositionally. These petrographical features are consistent with highly localised brittle failure, granulation and varying degrees of recrystallisation in a hot, but essentially solid, gabbro.

The commonest basaltic dolerite on Glacan Mor is dark grey with a sharp, subconchoidal fracture. It consists of sparse plagioclase phenocrysts in a groundmass of granular to subhedral pyroxenes, small laths of plagioclase (An₅₀Ab₅₀), and opaque minerals which range in size from 2 to 50 µm (Figure 10C). Some patches of rock are richer in amphibole, chlorite and magnetite than others and may represent volatile-rich regions of the basalt. At different times the basalt has been fractured and intruded by a variety of different veins, some consisting of opaque minerals, some of amphibole, or chlorite, or epidote, or clay minerals and yet others of mixtures of these minerals. They indicate a considerable range and persistence of hydrothermal activity. Dolerites throughout the Breccia display a range of textures (Figure 10C), (Figure 11A) and (Figure 11B) which suggest that in different places the dolerites have undergone varying degrees of sub-solidus recrystallisation. In some the texture is recognisably igneous and represents a magma that has chilled against cool host rock's; in others the lath-like form of the feldspar is preserved but the shape of the interstitial pyroxenes and opaque minerals is distinctly granular and indicates that the basaltic magma was intruded into a substantially hotter environment (Figure 11A). In (Figure 11B) the texture is entirely granoblastic with no obvious trace of an igneous texture and it is likely that after initial solidification this particular metabasite remained at a high sub-solidus temperature long enough to undergo complete recrystallisation. Similar ranges in texture occur in the coarser rocks also and some gabbros on Boreray and Soay have ophitic textures modified by recrystallisation to smaller and more rounded grain sizes.

Hydrothermal alteration affects all rocks in the Breccia to greater or lesser degree, and some dolerites have been extensively altered to fine-grained chlorite-albite- epidote- sphene- carbonate assemblages. In addition to these minerals prehnite, pyrite and zeolites occur in amygdales, found in a few dolerites, and small amounts of pink epidote occur as a thin coating on some joints. The pink colouration is caused by up to 0.8% MnO. The breccias and microbreccias consist of rounded and irregular fragments of dolerite, gabbro, basaltic glass, feldspar and pyroxene crystals in a finely comminuted or glassy base that has undergone partial recrystallisation. Again local conditions have probably determined whether this is minimal (cool conditions), extensive (hot conditions with devitrification of glass and formation of subhedral crystals), or whether enough water and carbon dioxide were present to form hydrous minerals and carbonates. The rarity or absence of quartz and alkali feldspar, either in the original igneous textures or as veins, is further evidence that the Breccias predate the granitic rocks abundantly exposed nearby.

The chemical analyses of three common rock types in the Breccia show that the fine-grained flinty dolerite of Glacan Mor (column 1) is very similar (except in content of water and fluorine) to a dolerite from Mullach Geal (column 2) that has undergone partial recrystallisation (Figure 11A), and that both are similar to a gabbro from Boreray, despite evidence of hydrothermal alteration (Figure 10A). Both dolerite analyses indicate that the rocks are tholeiites with particularly low contents of potassium and titanium, suggesting oceanic affinities.

Chapter 7 Glen Bay Gabbro E^G

Keywords: chilled margin, vertical banding, shearing, mineral analysis

On the rock shelves at the west side of Glen Bay there is abundant exposure of the Glen Bay Gabbro and of its contact with the Glen Bay Granite. Here the contact trends north-south and is steeply dipping although exposures of granite at the top of the rock shelves and others partly covered by glacial deposits indicate that the boundary is irregular and not predictable over any distance. The western contact of the Glen Bay Gabbro with the Western Gabbro is largely covered by hill slope deposits although at its northern end it is vertical, and in some prominent crags to the south, coarse and unchilled gabbro veins cut sheared Western Gabbro. The contact of the Gabbro with the Glen Bay Granite is extensively sheared by Granite veins in the gabbro and the finer grain size of the Granite at its margin clearly indicate that it is the younger intrusion. The Granite has split the gabbro into western and eastern outcrops. In the western part of the Gabbro concentrations of feldspar or of pyroxene and magnetite form impersistent bands or flattened lenses with a steep easterly dip. This type of banding was not seen on the eastern side but, in contrast, 100 m north of the outfall of Abhainn a Ghlinne Mhor there is a zone of vertical rhythmic banding about 3 m wide which consists of feldspathic bands alternating with ferromagnesian bands, each about 5 mm thick (Figure 12). these are sinuous and impersistent but in general strike north-south, and are parallel to the inferred contact of the gabbro with the breccia of Glacan Mor. At its eastern contact, south of the tunnel at Gob na h-Airde, the Gabbro is a splintery black glassy basalt with sparse phenocrysts of plagioclase and olivine and some small rounded inclusions of dolerite. Between this point and the coarse gabbro to the south, there is a gradation in grain size from basaltic dolerite through dolerite to gabbro. The attitude of the contact and the fine vertical banding indicate a vertical zone of chilling extending over a width of 100–120 m. The Gabbro is cut by small pegmatitic veins and pods of inconsistent attitude and by north-east-trending dykes of dolerite and microgranite (the Glen Bay Dykes) similar to Mullach Sgar Complex rocks. These indicate that the Gabbro predates the Complex. An isolated block of magnetite-gabbro similar to some lenses in the western part of the Gabbro (Figure 13C), occurs on Mullach Sgar and constitutes further evidence that the Complex developed after the Gabbro.

The chilled margin of the Gabbro is glassy for a width of 10 mm, and contains xenocrysts of plagioclase and olivine up to 3 mm long and inclusions of foreign gabbro and dolerite. At distances of 5 to 45 m from the margin, xenocrysts up to 4 mm long lie in a groundmass of plagioclase, olivine, calcium-rich and calcium-poor pyroxenes, opaques, quartz and apatite in a size range of 0.05–0.35 mm (Figure 13A). In the banded gabbro feldspars in the light bands commonly reach 7 mm in length but the pyroxenes and opaques in the dark bands average only 1 mm. The pyroxenes in the Gabbro between the chilled margin and the banding are generally subhedral with variable clustering or clotting (Figure 13B), but west of the banded rock inclusion of plagioclase crystals is a feature of the calcium-rich pyroxenes and a subophitic to ophitic texture in the rock is more pronounced. Plagioclase crystals in the groundmass have cores of An₅₅Ab₄₅ and exhibit a simple normal zoning pattern to An₃₀Ab₇₀. Some larger crystals have corroded cores and complex zoning patterns, and they resemble grains from some Western Gabbros so closely that these seem a likely source. A few augite grains with compositions similar to those in the Western Gabbro are also present but unlike the feldspars they appear to have been more susceptible to alteration in the gabbroic magma so that only skeletal 'blebby' relics remain. The augite of the Glen Bay Gabbro differs from the xenocrysts in showing slight pleochroism and variably developed exsolution of Ca-poor pyroxene lamellae along both 100 and 001 directions in the crystal. Chemical analyses indicate that the Gabbro contains 2.6% TiO₂ and most of this is present as ilmenite. It is invariably associated with magnetite either in irregularly shaped masses or in acicular grains. Some titanium also occurs in biotite which commonly mantles grains of magnetite and ilmenite (Figure 13B), and in titanite (sphene), another alteration product associated with shear planes. Sparse, large olivine grains are zoned from Fo₆₀Fa₄₀ at their cores to Fo₃₀Fa₇₀ at the rims, and the rim compositions correspond with those of the smaller groundmass olivines which vary between Fo₃₀Fa₇₀ and Fo₂₅Fa₇₅. Olivine constitutes between 1 and 2% of the rock, a low but consistent presence also shown by quartz and alkali feldspar which occur in sporadic patches throughout the rock and represent crystallisation of the last residues of liquid. Apatite varies in abundance and has recrystallised over a range of temperatures in a number of different crystal habits. Near the gabbro margin it occurs as long needles; these are joined more than 30 m from the margin by crystals of stumpy prismatic habit, and west of the banded gabbro irregularly-shaped interstitial grains also appear.

In contrast to the relatively undisturbed nature of the gabbro of east Glen Bay, the gabbro on the western rock shelves is very sheared and granulated. A few lenses however retain the original igneous texture and some exhibit mineral lamination broadly coplanar with the bands which dip at 60° to the east. Apart from the absence of fresh olivine the constituent minerals are the same as those of the eastern part of the Gabbro, although the proportion of calcic amphibole (ferroedenite), biotite, chlorite and albite developed at the expense of the original minerals is much greater. In addition to mineralogical changes, the main textural difference in the shear zones is that large crystals have been broken into smaller domains and this has resulted in a fine-grained mortar texture.

Chemically, the Glen Bay Gabbro is richer in TiO₂, K₂O, P₂O₅ and rare earth elements than are earlier gabbros and dolerites on St Kilda (E^W, D and EK), and it almost certainly has a different source. It marks a separate stage in the evolution of the St Kilda volcanic centre and may represent part of a magma chamber developed at a relatively high level in the earth's crust. One way of explaining the unusually thick chilled margin of the Gabbro is by postulating that a large block or segment of cold gabbro-dolerite breccia (EK) collapsed into such a magma chamber, following perhaps a particularly violent surface eruption.

Chapter 8 Pegmatites

Keywords: residual fluids, mineralogy, zirkelite

Throughout the Western Gabbro a relatively coarse facies of gabbro occupies irregular fractures parallel to or transgressing the banding and layering. In some places on the eastern slopes of the Cambir, on Dun or below Mullach Bi (Figure 14A) the veins are coarsely pegmatitic with crystals of pyroxene reaching 10 inches (250 mm) in length. The smaller and thinner veins have a mineralogy comparable with that of the gabbro except that olivine is scarcer, but the coarser pegmatites are altered to varying degrees and the primary augite, labradorite–andesine and magnetite–ilmenite grains are partly replaced by chlorite (pycnochlorite), epidote, albite and prehnite. The pale-coloured centres of the wider pegmatite veins consist almost entirely of these secondary minerals (Figure 14B) and here also accessory zircon is found. The composition of the augite is comparable both in major elements (En₄₂Wo₄₃ Fs₁₅) and in TiO₂ (0.5%) and A1₂O₃ (1.6%) contents to augites in the Western Gabbro and it is likely that these pegmatites represent the residual fluids after crystallisation of most of the E^W magma. They indicate a residue rich in Na, Ca and Fe and poor in Si and K, for neither quartz nor alkali feldspar were found with the albite–chlorite–epidote–prehnite association.

In contrast the pegmatites which cut the Glen Bay Gabbro on the eastern rock shelves of Glen Bay contain a different and more extensive suite of minerals. Ferroaugite (see (Figure 5)), amphibole (ferroedenite), chlorite (diabantite), magnetite, ilmenite, quartz, oligoclase and orthoclase are the essential minerals with accessory biotite, epidote, allanite, sphene, apatite, zircon and zirkelite (Figure 14C). Some minerals from this suite occur in tiny patches of residual crystallisation in Glen Bay Gabbro and it is likely that the pegmatites represent, on a slightly larger scale, concentration and segregation of many of the incompatible elements (K, P, Ti, Zr, Y, rare earths) in the tholeiitic magna that could not be incorporated in the olivine, pyroxenes and feldspar of the Gabbro. The rare earth elements are largely concentrated in the allanite grains (22%) but they also occur in small quantities (7%) in the rare Zr –Y titanate, zirkelite, and in trace quantities in sphene.

Although the pegmatitic patches and veins in the gabbros of Glacan Mor, Boreray and Soay vary in composition between the two extremes described above, most contain the albite–chlorite–epidote–prehnite assemblage and there are relatively small quantities of quartz and alkali feldspar.

Chapter 9 Metamorphism

Keywords: granulite–zeolite facies range, mineralogy, textures

A prominent feature of the Western Gabbro is the network of veins that stand out on weathered surfaces (Figure 15A). These consist predominantly of amphibole and chlorite and range from widely-spaced thin veins, which appear to be joint infillings, to denser and more disoriented styles of veining, especially near the eastern margin of the Gabbro where igneous banding indicates that there has been relative rotation of discrete blocks. Veining in these highly sheared rocks may be accompanied by more pervasive zones of granulation, for example in the gabbro north of Mullach Bi and around Claigeann Mor. These penetrate all parts of the original, igneous texture and in places extremely tough rocks have resulted, composed of the same anhydrous gabbroic minerals but with a granulitic texture (Figure 10B). Highly localised shearing facilitated this high-temperature, anhydrous recrystallisation which appears to affect only the Western Gabbro. It may predate the metamorphism characterised by amphibole and chlorite veining, but possibly the two were generally synchronous, the veining representing recrystallisation along joints or fractures permitting ingress of water. Good examples of the latter are found on the slopes of the ridge north of Mullach Bi where some rocks have been sheared with development of hydrous minerals and others retain their igneous texture and show only minor alteration at vein margins (Figure 15B). The composition of the amphiboles in the veins ranges from actinolite to magnesiohornblende to edenite, and zoning of individual grains is common. In the sheared rocks the range extends to tremolite and tschermakite, both varieties commonly fringing or entirely replacing pyroxene and olivine grains. Local variation is marked and reflects the particular minerals and fluids in any one area which have contributed to a particular composition not only of amphibole but also of chlorite (diabantite, penninite and clinochlore are found in different areas), serpentine (magnesian and iron-rich varieties), biotite and spinel (with different Mg/Fe ratios) and talc. Green spinel (Mg_0.6Fe_0.4A1₂O₄) is of sporadic occurrence but is associated with feldspathic rocks and generally surrounded by pale green amphibole (Figure 15C).

The high grade alteration described above is also present in large blocks of typical Western Gabbro, within the igneous breccia EK. However the most common secondary mineral assemblage encountered in the EK breccias of Soay, Boreray and Glacan Mor is amphibole-chlorite- epidote-albite, suggesting a low-grade, greenschist facies alteration involving hydration of mafic rocks. Petrographical evidence shows that this alteration postdates the granulite- and amphibolite-facies alteration of the gabbros, but since quartz and feldspar veining is conspicuously absent, it probably predates granitic magmatism in the St Kilda complex. The restricted nature of the alteration suggests it is due to hydrothermal activity, perhaps associated with the cessation of surface volcanism. Relatively high initial Sr⁸⁷/Sr⁸⁶ ratios in the Boreray gabbro (pp. 40–41), indicates that alteration was accompanied by minor enrichment in alkalis and perhaps other lithophile elements, but there is no evidence of widespread mineralisation in the breccias.

A distinctly different hydrothermal alteration affected the Mullach Sgar Complex and, to a limited extent, the Conachair Granite. Alteration is concentrated along mineralised fractures and faults (p. 37), and petrographically is less pervasive than that affecting the EK breccia. Typical mineral assemblages of apophyllite–chabazite–calcite–actinolite–epidote–prehnite–chlorite–quartz–albite occur in the MSC, while orthoclase–albite–quartz assemblages occur on fault planes cutting the Conachair Granite. Assemblages in the MSC indicate a very low-grade alteration, over a temperature range of 100°–260°C, and equivalent to the zeolite facies through to the prehnite–pumpellyite facies of regional metamorphism. These minerals occur in fractures in the MSC associated with NW–SE faults that cut both MSC and Conachair Granite. Thus alteration of the MSC must have occurred after consolidation of the Conachair Granite, and may therefore be the result of fluid release following uplift and early faulting of the Granite. Some zeolite-grade alteration may have accompanied the subsequent intrusion of late cone sheets and dykes.

Chapter 10 Glen Bay Granite G

Keywords: age relations, chemical analysis, rare earth minerals

The Glen Bay Granite forms the rock shelves at the southern end or Head of Glen Bay. It extends for a further 200 m south along A. a Ghlinne Mhoir and may underlie the Quaternary sediments to the west of this stream. Fine-grained (probably chilled) granite is present both at the western margin against the western part of the Glen Bay Gabbro and on its eastern side next to the Gabbro at the outfall of A. a Ghlinne Mhoir, and it seems likely that the Granite was intruded as a steeply-dipping north–south sheet into cold gabbro. Gabbro and Granite are both sheared at the contacts (Figure 16A) and shear zones are variably developed throughout the Granite and western part of the Gabbro. Movement associated with the shearing was possibly responsible for the position of the detached outcrop of Granite on the western rock shelves (see foreground of (Figure 16B)) and for the sinuous nature of the Granite–Gabbro contact in the low rubbly cliffs, south of the shelves visible on the right of (Figure 16B). The Granite is intruded by a dyke of dark purplish grey porphyritic felsite near the outfall of A. a Ghlinne Mhoir and this in turn is cut by prominent dykes with a steep south-easterly dip. The latter consist of dolerites, microdiorites and microgranites identical to some of the rock types found in the Mullach Sgar Complex. Late, rusty-weathering, shallow-dipping sheets of dolerite cut both the Granite and the dykes.

The Granite is medium to fine-grained with variable contents of dark minerals and, in contrast with the granites of Na h-Eagan and Conachair, is almost free of drusy cavities. The first minerals to crystallise were andesine (An₃₅Ab₆₅), two pyroxenes, ferroaugite (Ca₄₄Mg₁₃Fe₄₃) and ferropigeonite (Ca₁₀Mg₄₄Fe₄₆), and a calcic amphibole (ferrohornblende or ferroedenite). Most amphibole is euhedral with abundant opaque inclusions but some forms mantles on pyroxene grains, where it clearly crystallised after the pyroxene, in association with more sodic plagioclase (oligoclase), magnetite and ilmenite. There is some tendency for amphibole, pyroxene and opaque minerals, to cluster with accessory biotite, zircon and apatite; these mafic clots show a sporadic distribution through the rock. Most plagioclase grains are zoned from andesine cores to rims of An₀₈Ab₉₂ and they are commonly mantled with alkali feldspar microperthite (Figure 17A). With a 5-µm wide electron probe beam, patches of pure albite mixed with patches of composition Ab₃₈Or₆₂ were detected but these are not distinguishable under an optical microscope. The alkali feldspar is either intergrown with subhedral grains of quartz, some with prism or rhombohedral faces, or, less commonly, the two minerals are granophyrically intergrown. Isolated euhedral and skeletal crystals of apatite and zircon, and elongate or stumpy grains of the rare-earth mineral chevkinite, are included in alkali feldspar and quartz. Chevkinite contains about 22% Ce₂O₃ and the crystallisation of this rare mineral rather than other Ce-bearing phases such as allanite or monazite suggests that conditions in the later stages of crystallisation of the Glen Bay Granite were unusually rich in Ti and poor in Al and P. Alteration of the original mineralogy is variable throughout the rock and depends partly on the degree of granulation in any one area (Figure 17B). Pyroxene is the least stable mineral and many grains are altered to an iron-rich chlorite (ripidolite). Chlorite is thus a common secondary mineral in the mafic clots referred to above and is accompanied in some places by tiny areas of sphene which adopts the unusual habit of sheafs of radiating fibres. This habit is similar to that of some sphene occurrences in the granites on Na h-Eagan. Chemically the sphene in both granites is not pure CaTiSiO₅, and variable contents of Fe, Mg and Al, with up to 8% A1₂O₃ in grains , in the Glen Bay Granite are present. In the sheared granite, fragmented crystals of amphibole, feldspar and quartz are penetrated by zones of incipient granulation, resulting in recrystallisation of grain margins to a fine-grained mosaic or mortar texture (Figure 17B).

The small amounts of chevkinite and sphene are not sufficient to increase the TiO₂ content of the Glen Bay Granite unduly (Table 17) and chemically it is similar to the less siliceous granites in other Hebridean centres, notably Grigadale (Ardnamurchan), Papadil (Rhum) and Glamaig (Skye). It has higher Mg, Fe and Ca contents than the Conachair Granite and this is reflected in a higher colour index of 7–9. Again the trace element content is not significantly different from those in other Hebridean granites although Ba, Zr and rare earth elements (p. 43) are relatively high.

Chapter 11 Mullach Sgar Complex. Field description g^D, F^M and P^M

Keywords: basalt, granite, hybrid rocks, fragment shapes, textures, map

The Mullach Sgar Complex consists of fragmented intrusions of dolerite, microdiorite and granite, and of blocks of gabbro and dolerite derived from the Western Gabbro, the Breccia of Mullach Mor and the Glen Bay Gabbro. At its south-west margin, granites and dolerites are clearly intrusive into the Western Gabbro, the contact dipping 80°SW away from the centre of the Complex, but the northern margin of the complex is ill-defined in Gleann Mor and beneath Mullach Geal it probably has an irregular gentle northerly dip. Its eastern margin dips approximately 45° west and runs from Mullach Mor across to Glacan Chonachair where veins of the later intrusive Conachair Granite cut dolerites and granites of the Complex.

Excellent exposures on Na h-Eagan and in the Dun Passage have enabled four major phases of intrusion to be recognised (Figure 19). The first of these phases consists of grey-green dolerite, microdiorite and microgranite and occurs near sea level between Ruaival and Na h-Eagan. The component rocks of the first phase have suffered disruption by the subsequent phases, and brecciated and partially or completely hydridised fragments can be found throughout the Complex. The second phase is represented by dolerite, granodiorite and granite which is marginally coarser than the majority of rocks in the Complex and weathers more easily to a brown crumbly rock. It occurs on the rock shelves of the Dun Passage in elongate lenses trending south-west with a steep dip, and is also present in Abhainn a Ghlinne Mhoir. In the Dun Passage the Phase 2 rocks are intruded by fine-grained dark grey dolerites which characteristically form elongate lobate and generally parallel dyke-like masses, 10 cm to 10 m long in a granite matrix. These form Phase 3 and are approximately parallel to the margin of the Complex. Phase 3 dolerites are cut by Phase 4 microdiorite and granite which on Na h-Eagan forms a large composite dyke with chilled lobate margins dipping steeply south-west (Figure 18). Another composite dyke of microdiorite chilled against granite, with similar orientation, occurs on the rock shelves beneath the site of St Brianan's Church. Similar sheets of microgranite containing lobate inclusions of dolerite or microdiorite intrude the Western Gabbro 400 m north of Mullach Bi, on Claigeann Mor and in many places on Ruaival and Dun. Dykes consisting of a similar range of rock types with similar structures also intrude the granite and gabbro in Glen Bay. Xenoliths of gabbro are largely unaffected by the invading granites and dolerites, and at most appear to suffer only marginal alteration, where hydrous minerals have formed at the expense of olivine and pyroxene and some hybridisation by quartzofeldspathic material may be apparent. The major features of the Complex include the intimate association of mafic and felsic rocks (many in sheet structures), the large amount of shattering and net-veining and the rarity of shearing, the occurrence of mafic rocks chilled against felsic rocks, and the lack of felsic inclusions in the dolerites and microdiorites. The chilled margins of the lobate dolerites and the ubiquity of felsic rock between mafic fragments suggest that the sequence of magma intrusion was basaltic followed by granitic followed by basaltic and the complex may have developed in the following way. Firstly subsidence of the gabbro and dolerite country rocks (E^W, EK and E^G) took place and granitic magma welled up and occupied the fractures and space created by the subsidence. Then further subsidence and fracturing enabled alternate intrusion of granite and either basalt or microdiorite in a number of pulses or phases. Enough time must have elapsed for phase 2 rocks to solidify before phase 3 magmas were intruded, and perhaps movements of various competent blocks within the Complex contributed to the unusual lobate shape of many dolerite fragments.

Chapter 12 Mullach Sgar Complex. Petrology

Keywords: magmatic and metamorphic textures, chemical analysis, mineralogy.

The mafic rocks range in composition from dolerite to microdiorite and display a variety of textures. Some are finely granular and microphenocrystic basalts, with or without olivine; others are coarser dolerites, and these grade with increasing quartz into microdiorites. Phenocrysts of plagioclase, An₇₅ in the basic rocks to An₅₀ in the intermediate ones, are common and outnumber the sparser olivine or orthopyroxene phenocrysts. The latter are commonly altered around the margins and contrast with the fresh clinopyroxene phenocrysts which occur sporadically in the dolerites but more commonly in the microdiorites. The most basic rocks typically consist of feldspar phenocrysts in a groundmass of plagioclase, augite and opaques whose textures vary between inter-granular, sheaf, and subophitic. The mafic component of Phase 2 is microdiorite ((Figure 20A) and analysis) and although fine-grained at fragment margins in the Dun Passage, is hypidiomorphic in texture. The first minerals to crystallise were orthopyroxene (Figure 20A) and plagioclase, followed by clinopyroxene, opaques and amphibole. Orthopyroxene is rimmed with chlorite and appears to have been unstable in the conditions that favoured simultaneous formation of clinopyroxene and calcic amphibole (ferrohornblende). Pockets of K-feldspar and quartz occur in the groundmass and there are patchy areas where accessory zircon or apatite are common. Olivine-dolerites, dolerites and quartz-dolerites form the mafic rocks of Phase 3. Two examples of the textures are shown in (Figure 20C) and (Figure 21B) and these and the other textures are typically fine-grained. The small grainsize and content of opaque minerals probably determine how dark the rocks appear in hand specimen. The mafic component of Phase 4 is microdiorite and pale-coloured in comparison with the Phase 3 dolerites. Its composition is given in the Table of analyses and, following Irvine and Baragar (1971), it is tholeiitic andesite. Compared with the mafic component the Phase 2 granodiorites contain less pyroxene and opaque minerals, and show a corresponding increase in K-feldspar and quartz, which is reflected in their CaO, MgO and K₂O contents (see Table). The felsic rocks enclosing the mafic fragments of Phase 3 are indistinguishable from Phase 4 felsic rocks and both are distinctly richer in quartz and K-feldspar than are Phase 2 rocks. The texture of one of the steeply-dipping sheets on Na h-Eagan (Figure 18) is shown in (Figure 20B). Calcic amphibole (ferro hornblende or ferroedenite), opaque minerals and andesine-oligioclase were the first minerals to form, followed by biotite, K-feldspar and quartz. Accessory zircon, apatite and sphene (in small radiating sheafs associated with chlorite) are common and the rare-earth minerals chevkinite and allanite are present in some samples (although not together in the same hand specimen). Plagioclase crystals are progressively zoned from cores of An₃₅Ab₆₅ to margins of An₃₅Ab₆₅ and these are mantled by turbid orthoclase microperthite. Granophyric intergrowth of alkali feldspar and quartz is common in some parts and microgranitic texture predominates in others.

Contact between enclosed fragments and host rock in the Complex is of four main kinds: sharp and angular; lobate; torn or shredded; and gradational or hybridised. Chilled margins are found in the first three contact types but occur in mafic fragments only. Some typical contact textures are shown in (Figure 20C), (Figure 21A) and (Figure 21B). In (Figure 20C), quenched plagioclase microlites and microphenocrysts lie in a basaltic groundmass of granular pyroxene, plagioclase and an opaque mineral. Plagioclase grain size increases away from the margin indicating that the basaltic magma was chilled against the granitic material. However, in spite of the presence of quenched plagioclase microlites, there is no evidence of glass in the finely granular chilled margin of the basalt. Such textures suggest that initially glassy chilled margins have been annealed by contact with hot granitic material. Indeed the sinuous nature of the margin suggests that both rock types were plastic at the time of contact. An example of a basaltic rock cut by a granitic vein is shown in (Figure 21A). The sharp angular contact and the lack of chilling indicate that the mafic rock was brittle with a well-developed texture prior to intrusion by the microgranite. But though the vein is thin (9 mm) the microgranite is unchilled, suggesting that the dolerite was itself hot when fractured. The first stage of dolerite assimilation by granite is also shown in (Figure 21A) where K-feldspar is developed in the margin, and all gradations to dioritic hybrid rocks are present elsewhere in the complex. The shredded fragments are generally of fine-grained basaltic dolerite bounded partly by sinuous, rounded chilled margins and partly by ragged or irregular edges which are unchilled and show little if any evidence of impregnation by potassic fluids. The example shown in (Figure 21B) is an olivine-tholeiite but chilled and 'shredded' marginal features are shown by a range of mafic rocks through quartz-tholeiites to microdiorites. Two main kinds of alteration are shown by the mafic fragments. One is hybridisation by felsic material referred to above and the other results in development of biotite, hornblende and chlorite at the expense of pyroxene, plagioclase and olivine. These are features of retrogressive hydration of essentially anhydrous mafic rocks. The sustained high temperature conditions that caused formation of granoblastic textures in the Glacan Mor dolerites were evidently not achieved in the Mullach Sgar Complex, and it therefore seems that granitic magma, not basaltic, was the dominant invading liquid in the complex.

The analyses of Phase 2 and Phase 4 rocks given in the Table are broadly representative of a narrow compositional range, but in Phase 3 dolerite compositions range from olivine to quartz-tholeiite and the analysis given (of the latter) is only one example. Soda exceeds potash in the acid rocks of the Complex and this is reflected in some of the minerals found lining drusy cavities. In addition to prehnite, epidote and sphene, calcite, apophyllite, scolecite, chabazite and amethyst are present in small quantities, and represent late hydrothermal activity at temperatures possibly as low as 100°C.

Chapter 13 Conachair Granite

Keywords: leucocratic drusy rock, textures, chevkinite, chemical analysis

Much of the eastern portion of Hirta, about one third of its area, is occupied by the youngest of the major intrusions comprising the St Kilda Tertiary igneous complex. The Conachair Granite(I) is a cream-coloured medium-grained leucogranite, containing scattered druses vapour cavities which formed as the granite magma cooled and consolidated. Glassy or milky quartz crystals and pale creamy-yellow grains of feldspar are the principal mineral constituents, and both minerals can be found with perfect crystal terminations where they project into the druses. Conachair granite is usually the first rock encountered by visitors to Hirta, for blocks of it have been used in the construction of the jetty and natural outcrops flank the slipway to the jetty. It has also been extensively used in the building of cottages, cleits and walls in the Village Bay area. Above the village, the granite forms the prominent outcrops of Glacan Chonachair, Oiseval and, of course, Conachair. The spectacular cliffs below the Gap, Conachair and Oiseval are also mostly granite, as are the rocky islets of Bradastac and Mina Stac and many smaller islets nearby. The slabby or blocky appearance of granite outcrops is due to the pattern of joints. These are vertical and sub-horizontal planar surfaces which develop as the rock contracts after cooling; weathering tends to enhance the joint pattern.

The granite was intruded into dolerites and microgranites of the Mullach Sgar Complex along a western junction which dips at 30°–70° to the WSW. In the Village Bay this junction is obscured by Quaternary deposits, but it can be followed up the slopes of Glacan Chonachair and, after disappearing in boggy ground, is again found in the cliffs above Bradastac. Contact between granite and gabbro' is seen in crags 300 m bearing 293° from Conachair summit. From this point, at a height of about 300 m, the junction can be followed down the cliffs in a north-westerly direction to about the 200 m contour. It is again seen at the water line in cliffs overhanging Na Cleitean, where medium-grained granite is in contact with mixed acid/basic rocks of the Mullach Sgar Complex. To the south and south-east the submarine extension of the granite may terminate against NW–SE faults, but in any case cannot extend beyond the arcuate exposures of the Western Gabbro represented by Levenish and submarine outcrops of E^w or EK (pp. 34–35). North-eastwards, the occurrence of gabbroic rocks forming submarine outcrops 2.25 km SSW of Stac Lee limits the thickness of this sheet-like granitic intrusion to a maximum of 4 km for an overall south-westerly dip of 50°.

Thin section petrography shows that the granite consists largely of quartz, orthoclase and low-albite, forming respectively 25–30%, 40–50% and 25–30% of the rock. All three minerals are commonly closely intergrown, orthoclase and albite forming a microperthitic feldspar which itself may contain granophyric intergrowths of quartz (Figure 23B), (Figure 23C)). Other minerals occur in accessory amounts only and are commonly found in and adjacent to drusy cavities; they include calcic-amphibole, chlorite replacing amphibole, biotite, titaniferous-magnetite, zircon, rutile, fluorite, sphene, anatase, and chevkinite (Figure 23B). The typical granite of Oiseval and Conachair has a microgranitic texture with up to 20% of granophyre, and consists mostly of anhedral perthite grains, up to 5 mm across, enclosing unstrained, subhedral quartz grains up to 3 mm across. Quartz grains are commonly embayed and serrated relics of stubby bipyramids, indicating that they are inverted β -quartz phenocrysts that have suffered magmatic corrosion (Figure 23A). Granophyric quartz usually forms an optically continuous systems of blebs, stems and wedges whose domain is controlled by the enclosing perthite grain. Finely vermicular quartz characterises interstitial granophyre moulded onto perthite grains or forming outgrowths on corroded quartz phenocrysts. Tablets of lamellar-twinned plagioclase are enclosed in some perthite grains, often occurring as ghost euhedra overgrown and replaced by perthite, and occasionally mantled by vermicular quartz. Late albite An₆Ab₉₄ to An₈Ab₉₂ occurs as clear rims on perthite grains bordering druses and also as discrete euhedra within the cavities.

No evidence of chilling is seen at the margins of the granite. Contacts with gabbro to the west are sharp and often show microgranite-filled fissures extending 1–30 cm into the country rock. Within 2–3 mm of contacts, metasomatic alteration of the gabbro has resulted in alkali-feldspar overgrowths on plagioclase and replacement of pyroxene by amphibole and biotite. An intrusive contact with microgranite of the Mullach Sgar Complex has led to coarse granophyric intergrowths in the Conachair granite, which develop an elongate habit, 10–15 mm long, perpendicular to the junction (Figure 23C). Passing inwards, the granophyre becomes patchy and equant at 20 mm from the contact, and at 30 mm or more becomes finer-grained and microgranitic in texture. Such contact textures suggest that the country rocks (MSC and EK) may have been hot at the time of intrusion.

Thin veins (up to 15 cm) of pale grey rhyolite and aplite cut the granite and possibly represent fissure-fillings, associated with the episode of late sheet intrusion (p. 24). The rhyolite veins show corroded phenocrysts of inverted β -quartz and perthitic alkali feldspar in a microcrystalline felsitic groundmass; tiny xenoliths of granophyre may also be present. Some veins show flow alignment of groundmass microlites; others display a slight increase in grain size at the margins of the veins, indicating limited annealing, of the chilled groundmass against hot, fissured microgranite. Aplitic veins consist of a saccharoidal intergrowth of quartz and alkali-feldspar with a grain size of 0.2–0.5 mm. Quartz-feldspar grain boundaries are mostly straight and triple-junctions approximate to angles of 120° (Figure 23D). A few larger (up to 1.2 mm) perthite grains enclose blebs of quartz, reminiscent of the granophyre xenoliths noted above. The equilibrium textures of the aplites suggest solid-state recrystallisation, possibly the result of prolonged annealing of rhyolitic vein-filling in hot microgranite.

Chemically the Conachair granite shows many similarities with other Tertiary acid intrusions (Bell, 1982).Typically it is more siliceous (mean 76.2% (Table 23)) and alkaline (mean Na₂O + K₂O = 8.58%), but less aluminous (mean 12.33%), magnesian (mean 0.20%) and calcic (mean 0.31%) than Le Maitre's (1976) average granite, and indeed more closely approaches his average rhyolite. All the analysed rocks are peraluminous and are sparingly corundum-normative. Low normative anorthite (0.13–4.8) and a colour index ranging from 1.2–5.2, emphasise the alkaline and extreme leucocratic nature of the granite. In terms of the synthetic system Ab–Or–Q it plots close to the thermal minimum at 1 kbar P_H2O.Such a composition could have arisen from either extreme fractionation or fusion of sialic crust. Chemically the latter is unlikely because the initial ⁸⁷Sr/⁸⁶Sr ratio is low (p. 40), and the REE distribution (p. 43) is inconsistent with fusion of Torridonian or Lewisian basement rocks (Thompson, 1983). The geochemical evidence suggests that the Conachair Granite resulted from fractionation of a mafic magma.

Chapter 14 The minor intrusions

Keywords: basalt–rhyolite range, regional structure

A considerable number of inclined sheets and dykes are exposed on St Kilda which cut all the major plutonic intrusions. Differential erosion of these minor intrusions has resulted in some of the more spectacular scenic features of the islands including caves (Figure 24A), arches and numerous gullies and ledges which have become the nesting sites for thousands of sea birds.

The presence of minor intrusions in the St Kilda igneous complex was first noted by Macculloch (1819) who described the ‘syenite’ (Conachair Granite) as being 'traversed near the bay by two long and nearly horizontal basaltic veins at no great distance from each other, the fragment of a third being also seen near the village'. Ross (1884) also noted the 'veins of compact basalt' cutting the granite of Oiseval in addition to the 'several trap dykes, of many feet in thickness, penetrating the granite (of Glen Bay) at right angles to the line of ... a great junction of greenstone with the granite', and also dykes on the north shore of Village Bay. He reported that microscopically the basalt dykes comprise minute prisms of plagioclase feldspar, mainly granular augite and mostly serpentinised olivine. Geikie (1897) figured some of the inclined sheets cutting the Conachair Granite (Figure 24C) and reported that although numerous dykes traverse both the gabbros and granites, they are more abundant in the gabbros. He believed that a large number of these intrusions pre-dated the granite although correctly noting that an actual example of a basic dyke truncated by the Conachair Granite could not be seen. The sheets and dykes were examined and extensively sampled by Cockburn whose collection and notes are housed in the Royal Scottish Museum, Edinburgh. Cockburn (1935) described the age relationships of these bodies and recognised 3 groups: first, a restricted group of thin pre-Glen Bay Granite dykes which cut the Western Gabbro (pp. 2–3 of this report); secondly, a group of dykes and sheets which cut all the major intrusions except the Conachair Granite; and thirdly, a post-Conachair Granite group of sheets and dykes. Cockburn recognised that a considerable variety of rock types were emplaced during these intrusive episodes. The earliest intrusions are olivine-free basalt with some variolitic types whereas the second group varies widely from basalt through andesite to spherulitic and nonspherulitic microgranites, granophyres and felsites. The latest intrusions were predominantly basaltic although some granitic intrusions were emplaced late in the history of the complex.

The overall structure of the sheets is best appreciated by considering the St Kilda islands in turn. On the western coasts of Dun and Hirta the predominant strike of dykes and sheets is NW or NNW, the sheets being inclined NE at angles between 30° and 60°. In Glen Bay and along the northern cliffs on Mullach Mor, Conachair and Oiseval the dykes and sheets strike NE, the sheets dipping SE at about 60° in the west and at much shallower angles in the Conachair Granite. Some sheets depart significantly from the dominant trend as, for example, the 2 m-thick sheet exposed at the top of the cliffs south of Ruaival which strikes north-west and dips south-west. Most sheets have been intruded along dilational fractures with little additional disturbance to the country rock, although there are some shear zones of crumbly-weathering gabbro on the west coast that are adjacent and parallel to the dolerite sheets, and it is probable that some compressional shearing movements preceded the sheet intrusion in these regions.

On Soay the late sheets generally show a south-easterly dip and, indeed, the gently sloping top of the island appears to follow the surface of an eroded sheet. Late sheets on Boreray and the Stacs dip at low angles to the south and south-west (Figure 9A), and an erosion surface developed along one of these intrusions has given rise to the smooth, grassy slopes of southern Boreray. The swing in dip direction from mainly north-east on Dun, east on Hirta and south-east on Soay to mainly southerly on Boreray suggests a convergence of sheet attitude to the east of Hirta, in the manner of the classic cone sheet intrusions of the Inner Hebrides.

Many of the dykes in the complex have a north-easterly trend, a feature clearly shown by the older dykes of Glen Bay (Figure 24B). A similar trend is displayed by basic dykes which cut late sheets on Boreray.

A close association of basaltic and granitic magmas is displayed by several composite intrusions, and one in Glen Bay is exposed in the roof of the arch at Gob na h-Airde (Figure 24A). Others occur on the western side of Glen Bay, in Abhainn Gleshgill (a precipitous gully south-west of Mullach Sgar), above An Torc on Na h-Eagan (Figure 25B), and on Mol Ghiasgar below Conachair. A composite sheet occurs on the eastern cliffs of Soay, and another apparently occurs on Levenish (Cockburn slides L2–9). The dyke pictured in (Figure 25B) was regarded by Cockburn as a multiple rather than a composite intrusion, as fragments of the 50 cm-wide basic margin were enclosed within the partially devitrified porphyritic pitchstone central part. The internal margins of the basic components are lobate which suggests that the basalt was still hot and mobile when it was intruded by the granitic magma.

Chapter 15 Minor intrusions. Petrology

Keywords: textures, composite dykes, metamorphic effects

The dykes and sheets show considerable textural and compositional variation and although most of them can broadly be described as basaltic or doleritic, in detail these consist of glassy, variolitic, intersertal and ophitic ground-mass textures, with or without phenocrysts. Andesite or microdiorite sheets are much less common than dolerites but felsites, microgranites or pitchstones are relatively common, principally in dyke form but also in gently inclined sheets north of Mullach Bi, on Glacan Mor and on Soay.

A selection of textural types is shown in the photomicrographs. In (Figure 26A) and (Figure 26B) laths of plagioclase lie adjacent to irregularly shaped grains of pyroxene in an intergranular or subophitic texture. Sparse rounded grains of olivine have been altered to serpentine and saponite with the dark cracks across the grains occupied by iron oxides. In (Figure 26B) the pyroxene is pale brown with the lilac hue typical of members of this mineral group that contain appreciable contents of TiO₂ (an analysis of this rock is given in (Table 29), col. 1). Purplish augitic pyroxene (Wo₄₅En₃₆Fe₁₉) with 2.75% TiO₂ occurs in the groundmass of a basalt sheet from An Fhaing, Dun (Figure 26C) where it is associated with labradorite laths (An_64–66), grains of magnetite and interstitial chlorite and clay minerals. Feldspar phenocrysts in this rock show reverse zoning with cores of labradorite (An₆₁) and relatively calcic rims (An₆₆). These may either have crystallised from the same basaltic magma, in which case they have failed to equilibrate during changes of temperature of water pressure (that is, they are phenocrysts), or have been picked up by the magma from the country rocks during its ascent (that is, xenocrysts). The rounded shape of the feldspar indicates that the part with composition An₆₁ was unstable in the basaltic magma and was being resorbed, but enough survived to form the nucleus for crystallisation of a rim of granular pyroxene and magnetite, and for the formation of new feldspar (An₆₆) which grew in crystallographic continuity with the core. In a basalt sheet from Glen Hay (denoted by 'R'in (Figure 24B)) clusters of plagioclase (either An₆₄ ?xenocrysts or An₇₃–An₈₂ oscillatory-zoned phenocrysts) and pyroxene form glomeroporphyritic aggregates in an intersertal groundmass (Figure 27A). The clinopyroxene phenocrysts are slightly zoned with a compositional range Wo_42–36En_46–41Fs_19–15 and are richer in MgO and SiO₂ and poorer in TiO₂ (1.7 %) than the groundmass pyroxenes. The latter (Wo₄₄En₃₉Fs₁₇) are richer in A1₂O₃ and TiO₂ (2.2%) and occur with plagioclase (An₆₆), magnetite, ilmenite and interstitial alteration products (analysis in (Table 29), col. 7). A basaltic sheet which cuts the Mullach Sgar Complex on Na h-Eagan ((Table 29), col. 3) consists of extensively serpentinised olivine (Fo_78–81) and plagioclase (An₇₇) phenocrysts resting in a ground-mass of less calcic plagioclase (An_70–58), augite (Wo₄₄En₂₉Fs₂₇; TiO₂ 1.2–2.2%) and opaque minerals.

Possible contamination by foreign material in some sheets has been referred to above in connection with very calcic plagioclase grains, but more positive indications are provided by quartz xenocrysts. One example is shown in (Figure 27B) where the quartz has been rounded and corroded and has formed the nucleus for precipitation of pyroxene. Nearby, a spherical amygdale filled with dark greenish brown chlorite and clay minerals is surrounded by a tangential arrangement of feldspar laths, suggesting that the vesicle expanded during the crystallisation of the groundmass.

Above the boulder beach at Mol Ghiasgar, a 3 m-wide composite dyke cuts the Conachair Granite and also a late basic sheet cutting the Granite. The dyke consists of a thick (up to 2 m) central portion of spherulitic felsite in an envelope of dolerite. Corroded phenocrysts of sodic plagioclase and inverted β-quartz have acted as nuclei for spherulitic devitrification of the felsite (Figure 27C). As an result the original glassy groundmass now consists of finely intergrown orthoclase, albite and quartz, which together with the phenocrysts gives this rock the modal composition of an alkali rhyolite. The felsite is intruded centrally into an earlier, poorly vesicular feldsparphyric granular dolerite dyke, originally 1–2 m wide, which contains rare epidotised xenoliths of coarser mafic rock. In places a trachytoid-textured basalt occurs between the felsite and the dolerite, and this appears to represent a basic fluid that lubricated the intrusion of the more viscous rhyolitic magma (Figure 27D). Both the basalt and the dolerite show patchy development of epidote, chlorite and calcite resulting from contact alteration by the felsite. In contrast, alteration in the olivine-dolerite sheet cut by the dyke is characterised by replacement of groundmass olivine and interstitial material by smectite. This zeolite grade alteration and the glassy nature of the felsite indicate that both minor intrusions were emplaced at a high level into essentially cold (<150°C) granite.

Chapter 16 Minor intrusions. Geochemistry

Keywords: chemical analysis, magma mixing and differentiation

Chemical analyses of 24 minor intrusive rocks from St Kilda were carried out; 16, representing the range, are given in (Table 29). A chemical classification of the rocks is based on the scheme of Irvine and Baragar (1971) and uses normative values recalculated after the manner of Thompson and others (1972) in which the values for Fe₂O₃ are adjusted to 1.5% or 2% depending on the alkali content of the rock. This scheme shows that a range of basaltic compositions is present, with most being classified as olivine + hypersthene normative tholeiites and only a small number being quartz normative. Only a few sheets are strongly quartz normative basaltic andesite, although one cuts the Mullach Sgar Complex at the northern end of Na h-Eagan [NM 0937 9850], and is comparable with some mafic pillows (Phase 3) within the Mullach Sgar Complex itself. The composition of the inclined sheet is probably a mixture of basaltic magma and granitic or sedimentary material as a partially resorbed quartz xenocryst can be seen in thin section. Cockburn (1935) recorded partially resorbed quartz crystals in another inclined sheet below Mullach Bi (Figure 27B). Some olivine + hypersthene normative basalts fall in the alkaline field of a Ne′Ol′Q′ triangular diagram (Irvine and Baragar, 1971) and the compositions of two late sheets are nepheline normative. The analysis of one marked R in (Figure 24B), is given in (Table 29), col. 7, and the other, which cuts the granite on Oiseval, appears in col. 8.

The range of normative values is shown in (Figure 29), and this distribution of the compositions is consistent with that of a series of rocks which have undergone fractionation at low pressure, suggesting that the St Kilda rocks were derived from a high level magma chamber. These were possibly developed within the continental basement since crustal contamination of the basalts can be demonstrated on a TiO₂–K₂O–P₂O₅ diagram (Figure 28). The St Kildan basalts become increasingly potassic at a fairly constant ratio of Ti:P and follow a similar trend to basalts from Ardnamurchan which only differ in lying closer to the P₂O₅ apex. Thompson (1982a) has demonstrated that such enrichment in potash in the Ardnamurchan cone sheets cannot be due to closed system fractionation and is therefore due to contamination by sialic crust. However two rocks, one from a sheet cutting gabbro at Ruaival, the other from a sheet cutting gabbro at An Fhaing, Dun, are enriched in phosphorus and lie close to the trend for closed system fractionation. High values for other trace elements in these two rocks with corresponding low values for Ni and Cr are a further indication that fractionation has occurred. Geochemically the basic minor intrusives are similar to basalts from other Tertiary centres in that they possess low trace element abundances relative to typical continental tholeiitic rocks, but differ in respect of the proportions of these trace elements. In addition to differences in Ti–K–P proportions noted above, the St Kildan rocks have higher yttrium : zirconium ratios (0.15–0.27) than most basalts in the British Tertiary Volcanic Province, these ratios being exceeded only in sheets cutting the submarine Blackstones igneous complex WSW of Mull (Durant, unpublished data).

Analyses of both basic and acid components of two composite dykes are given in columns 13–16 of (Table 29). In both intrusions the outer part is basaltic and the inner is granitic. Both granitic components differ petrographically but have a similar SiO₂ content of 72%, which, along' with Ca, Mg and Fe values, lies between those for the Glen Bay Granite (69%) and the Conachair Granite (76%).

Chapter 17 Quaternary deposits

Keywords: glacial erratics, organic sand, protalus rampart, till, blockfield

A considerable gap in the geological record occurs between the intrusion of late dykes and sheets and the deposition of Quaternary tills of the Village Bay and Glen Mor areas. Neither lavas nor Tertiary sediments have been found on St Kilda, but since these rocks almost certainly occur to the west and north-west of the islands (Jones, 1981), it seems likely that the gap in the record represents an extended interval of erosion which ultimately exposed the roots of a complex volcanic centre comparable in size with Tertiary centres found in the Inner Hebrides.

Glacial erosion during the Quaternary era produced many of the present day topographical features of St Kilda and also deposited highly characteristic drifts of boulder, gravels and sands. Detailed studies of these deposits by Sutherland, Ballantyne and Walker (1982) suggest that three Quaternary cold periods can be recognised on Hirta: an early local glaciation, which deposited the Ruaival till, followed by another local glaciation responsible for the more extensive Village Bay till, the two glacial events being separated by an interstadial during which pollen-bearing sands accumulated. A final cold period was characterised by extensive frost shattering giving rise to blockfields, screes and protalus ramparts.

The Ruaival till forms the lower part of thick drift sections exposed below St Brianans Church and in the A. Ruaival (Figure 30A). It is a massive deposit, up to 20 m thick, consisting of locally-derived blocks and cobbles in a gravel and sand matrix, and is weakly iron-cemented towards its base. In the banks of the A. Ruaival the till is overlain by 1 m to 3 m of grey stratified slope deposits, the Ruaival head, containing platy joint-blocks from the Mullach Sgar Complex which are aligned parallel to the slope. Sutherland and others (1982) used measurements of the thickness of weathering rinds on dolerite clasts to show that the Ruaival till is significantly older than the overlying Ruaival head, and they suggest that the latter formed in periglacial conditions contemporaneously with the deposition of the Village Bay till. Pockets of organic sand with embedded blocks are found in the base of the Ruaival head in A. Ruaival (Figure 30B). These sands, up to 20 cm thick, are thought to have formed by fluvial reworking of a pollen-bearing soil horizon which developed during a mild climatic interval; radio-carbon dating indicates that they are at least Middle Devensian in age (Sutherland and others, 1982).

The Village Bay till is exposed in the low cliffs above the storm beach, extending from just east of the pier westwards to the A. Mhor. At the pier, up to 4 m of granitic till can be seen resting on shattered bedrock (Figure 30C), whereas in the centre of the bay (below the fuel store), some 8 m of till consist of a mixture of local rock, granite, dolerite and gabbro blocks, in a gravel and sand matrix. Petrological analysis of the matrix shows that plagioclase is the main constituent of the coarse sand fractions, while illite and vermiculite predominate in the finest material. Up to 5 m of till, containing a mixture of local rock types, are exposed in the A. Mhor, near to the end of the village street. The local glacier responsible for these deposits probably occupied much of the bay from An Lag to the eastern slope of Mullach Sgar, but actual drift limits are mostly obscured by later flows of soliflucted debris and hill slope deposits. Springs commonly appear where the leading edges of these later deposits rest on the Village Bay till, for example at just above 100 ft contour west of the A. Mhor, and at a higher level, about 400 ft below Glacan Chonachair. A similar line of springs breaks at the 300 ft contour in Gleann Mor (Figure 31A), suggesting that here also hill slope deposits have flowed onto a more consolidated till. The latter, which may be contemporaneous with the Village Bay till, is well exposed above the rock shelves at the mouth of the bay, where some 2–3 m of unstratified blocks and cobbles in a gravel and sand matrix are seen, all of it apparently locally derived.

On Hirta most of the higher slopes are mantled by a varying thickness of hill slope deposits, material formed during periglacial conditions when frost shattering of exposed crags provided a plentiful supply of rock fragments. The most spectacular periglacial deposits can be seen in the Village Bay area where two protalus ramparts have developed over the Village Bay till. The best preserved rampart is found beneath Glacan Chonachair where a mound 250 m long and up to 8 m high has accumulated at the foot of the crags (Figure 31B). Thick screes can also be found on the eastern slopes of Mullach Sgar where a former protalus rampart has been breached by later flows of scree and hill slope material. In An Lag, mass movement of hill slope deposits is preserved as soliflucted gravel lobes. The lobes, up to 60 m long, form mounds 3–6 m across and 0.5 m high and display a regularity reminiscent of cultivation strips, with which they have been confused. Less regular solifluction sheets are found on the southern slopes of Mullach Geal and a sheet flow from the western slopes of Glacan Chonachair has diverted the course of the A. Mhor at about 350 ft, leaving a dry gully at 300–225 ft.

Prolonged frost shattering has developed highly characteristic block fields on the gabbroic rocks of Hirta. The most extensive is found on Cam Mor where blocks up to 7 m across form a massive scree below Mullach Bi (Figure 31C). On Ruaival and Dun the blockfields are essentially residual and associated with tor-like features on the summit ridges. Phosphatic cementation of the residual blocks on Ruaival points to a long history of seabird activity since, and perhaps during, the periglacial period. Similar blockfields occur on the south eastern side of Soay, on the northern part of Boreray, and one, albeit small, is perched on the top of Stac an Armin.

Apart from a few boulders of gneiss, believed by Cockburn (1935) to have been transported as ships' ballast, no large foreign erratics have been found on St Kilda. However, detailed petrological analyses of coarse sand fractions (0.5 mm - 2.0 mm) from drift deposits reveal the presence of rare clasts of red sandstone, reddened feldspars and rounded quartz grains in samples of the Ruaival and Village Bay tills and the Ruaival head. Rare grains of garnet also occur and other heavy minerals have been identified in stream sediments (pp. 32, 33). Since these foreign grains occur in the oldest till deposit, they must have been introduced prior to the local glaciation. Sutherland and others (1982) suggest that the Scottish ice sheet may have encroached on St Kilda depositing erratics which were subsequently reworked into locally derived deposits. These erratics could have been derived from rocks outcropping close to the St Kildan volcanic centre.

Chapter 18 The mineralogy of stream sediments on Hirta

Keywords: heavy minerals, exotic minerals, mineral origins

Small pockets of sediments have been deposited in steps and hollows along the stream beds that drain Hirta. The contents and nature of these sediments reflect the mineralogy and geochemistry of nearby intrusions and also constitute a partial record of the development of the form and shape of the island. Stream sediments are natural collections of material from all parts of the catchment area and contain concentrations of the minerals which are most resistant to weathering and abrasion. The denser minerals are further concentrated by the winnowing effect of turbulent flowing water. Although relative abundances of minerals in the sediments and the source material are therefore not directly related, comparison may be made on a broad scale between one catchment and another, and it is possible to distinguish between indigenous minerals and those whose original source was outside the present catchment.

Only a small proportion of the sediments can be examined in detail and the sampling procedure used on St Kilda was the same as that developed for use in regional geochemical mapping by the Applied Geochemistry Unit of IGS (Plant and Moore, 1979). Sampling sites were carefully chosen to be representative of a particular catchment, and to avoid where possible, sources of contamination such as dwellings or industrial activity. Even in a location as remote as St Kilda metallic particles from crashed aircraft and fragments from building work on the hilltops can contaminate the sediments. Locations of the St Kilda sites are shown in (Figure 32)

The tools used in sampling include a small pointed shovel (with all traces of paint removed), nylon sieves of 2.0 mm and 150 µm mesh, in wooden frames, and a shallow wooden dish for panning. At the sample sites notes are taken of various factors which may affect the sedimentation, including stream flow conditions and the nature of bedrock and any drift deposits nearby. The top few centimetres of sediment are removed to dispose of recent organic contamination and transient precipitates, and pebbles and boulders examined and sampled if they show any interesting features (such as mineralisation; possible erratics). A sample of sediment is passed through the sieves, the < 150 µm fraction being collected in the wooden dish and retained, while the > 2.0 mm material is discarded. The 150 µm - 2.0 mm fraction is then panned to remove the bulk of the lighter minerals such as quartz and feldspar, and the heavy concentrate retained. Panned concentrates of the size fraction 150 µm- 2.0 mm were examined in detail to determine the nature and provenance of the mineral grains. The samples were subdivided in three ways; firstly into size fractions by sieving, secondly according to their magnetic susceptibility, by hand magnet and electromagnetic separator, and thirdly by density using a heavy liquid (bromoform SG 2.89). Most minerals were readily identified under the binocular microscope, but in addition a selection of identified and unidentified minerals were analysed qualitatively with the electron microprobe and further quantitative analysis or X-ray diffraction analyses were undertaken where necessary for positive identification.

Mineral grains in the panned concentrates are of two types: (i) indigenous minerals which occur in rocks outcropping within the catchment area of the sampling site; (ii) exotic minerals not known in rocks on St Kilda, forming at the most only a few percent of each concentrate.

In the An Lag concentrates, the most abundant indigenous minerals are angular to subrounded clinopyroxene and mottled pink translucent zircon grains with smaller amounts of euhedral magnetite. Very few of the grains are larger than 0.5 mm. The pyroxenes, probably derived mostly from late basic sheets, have a range of composition, with a distinct group of clear green angular diopside grains, many of which contain Cr. The most numerous zircons are commonly broken euhedra of simple prism-and-pyramid form, showing little sign of abrasion. They are chemically very pure, and only a few have a trace of Hf or Th, a characteristic of the zircons found in druses in the Conachair granite. A few zircons have thin discontinuous coatings or minor intergrowths of xenotime. A second group of zircons consists of sharply euhedral, very pale yellow transparent crystals, often with inclusions. A third group consists of well rounded pink, dark red, and brown zircons and these are much rarer. Euhedral anatase was found in a variety of forms ranging from very dark, sharply-pointed bipyramidal crystals to thin square plates showing strong green colouring, similar to crystals occurring in the Conachair Granite. Most of the anatase contains niobium, up to a maximum of about 7 wt% Nb₂O₅. Other indigenous minerals include a variety of amphiboles, epidote, orthopyroxene, chlorite, biotite, ilmenite, chromite, apatite, rutile, sphene and chevkinite. The most abundant exotic minerals in An Lag are garnets and these show a wide range of both composition and form. Although less than 0.25 mm in diameter the clear pink, red, orange-brown and lilac grains are conspicuous in the panned concentrates even though they form a very small proportion of the sample. Most are angular to subrounded fragments, some are well rounded grains, and a few unworn dodecahedra. The composition ranges from almandine, almandine-pyrope, almandine-spessartine through andradite, grossular-andradite, to spessartine. Andalusite, kyanite and sillimanite (polymorphs of Al₂SiO₅), staurolite, and a few small angular fragments of clear blue corundum also occur. Minerals of uncertain origin, but which could be derived from rocks on Hirta, are tourmaline, including euhedral stumpy prisms and rounded grains, goyazite (a single subrounded grain) and fergusonite (three angular or subhedral grains).

Magnetite is the commonest indigenous mineral in the panned concentrates from Abhainn Mhor and generally occurs as angular fragments, with lesser amounts of ilmenite, pyroxene, and amphiboles. Apatite, biotite–phlogopite, chlorite, chromite, epidote, olivine, prehnite, pyrite, rutile, and sphene are of rarer occurrence and all are typical accessory and secondary minerals in nearby basic intrusions. Exotic minerals are mostly garnets, with grossular and almandine-grossular occurring in addition to those types found in An Lag. Staurolite is also present with a few grains of pale blue corundum ranging from clear angular to 'frosted' subrounded shapes.

In Abhainn a Ghlinne Mhoir the suite of indigenous minerals is similar to that found in Abhainn Mhor. However, exotic grains are very scarce and are mostly almandine-pyrope, with some grossular-andradite.

The range of indigenous minerals present in Abhainn Ruaival is similar to that in Abhainn Mhor, although mica, chlorite, prehnite and sphene were not found. A few small (< 0.25 mm) angular, subhedral fragments of clear blue spinel are however present. Angular to subrounded garnets are the most common exotic mineral. They form two distinct groups, one consisting of pale pink almandine-pyrope and almandine-spessartine, including two euhedral almandine crystals, and the other consisting of orange-brown grossular-andradite grains. Rare angular to subrounded grains of staurolite, andalusite, and tourmaline (black, and pleochroic green–brown) are also present.

Most of the heavy minerals in each stream sediment sample were locally derived. Many of these grains show very little evidence of abrasion, and have probably been moved by running water or solifluction very short distances downslope from their source rock. In contrast to these, many exotic mineral grains and also some zircon, rutile, and tourmaline grains show evidence of prolonged abrasion. The garnets in particular, exhibit a wide range of characteristics, both in grain shape and composition. These features suggest that the history of the grains may be quite complex, some having existed as detrital grains for comparatively long periods. The compositional variation of the garnets and the nature of other foreign mineral grains indicates that the exotic heavy mineral suite was derived from two types of metamorphic rock. Corundum, andalusite, andradite, grossular-andradite, and grossular most commonly occur in contact metamorphosed rocks, the first two minerals in altered aluminous sediments, the garnets characteristically in skarn or calc-silicate hornfelses. The remaining exotic minerals — sillimanite, kyanite, staurolite, tourmaline, pyralspite, almandinegrossular and spessartine — were most likely derived from medium to high-grade regional metamorphic rocks.

The work of Sutherland and Ballantyne (1982) suggests that the earliest, pre-Devensian glaciation of St Kilda was responsible for introducing foreign rock fragments which were subsequently reworked into locally derived glacial and periglacial deposits (pp. 30–31). It is not necessary to look far beyond the shelf surrounding St Kilda for a source of the exotic minerals which are found in stream sediments. The regional metamorphic suite of minerals were most likely derived from the Lewisian basement which Jones (1981) suggests is at or near to the submarine surface east of St Kilda.

Jones also suggests that a belt of Mesozoic sediments resting on the Lewisian basement extends to the northeast and south of St Kilda, and westwards is covered by a seaward-thickening sequence of Cainozoic sediments. By analogy with the Inner Hebrides, particularly Skye, the Mesozoic sediments, overlain by Tertiary basic lavas, may well have provided the high level country rock into which the St Kilda complex was intruded. Both pelitic and calcareous Mesozoic lithologies appear to have been thermally metamorphosed, producing a characteristic heavy mineral suite. These minerals must be locally derived, possibly from a roof pendant, or high level xenoliths, which were eroded by early glacial activity and deposited in a local till which was subsequently reworked.

Chapter 19 The seabed and coastal features

Keywords: bottom profiles, dredge samples, drowned landscape

A series of traverses using a recording echo sounder abroad the mv Golden Chance revealed strong contrasts in the submarine topography around St Kilda. North and east of the main island group, a gently undulating platform at depths between 45 and 70 m extends to Boreray, whereas along the south-west coast the sea floor is very irregular, with ridges and gullies sloping steeply to depths of 80 m or more. East of Dun, the crest of this slope forms a hummocky ridge ((Figure 35)B,C) extending to about 5 km ENE of Levenish and forming an arcuate southern margin to the platform east of Village Bay. A similar feature extends NNW from Soay but is less distinct due to the extremely rugged topography of this area. North of Soay there is a transition from the rough ground of the west coast eastwards to a platform, which rises very gradually to an indistinct culmination between Oiseval and Boreray and then slopes down eastwards, reaching a depth of 90 m some 3 km east of Boreray. Away from the coasts the platform appears to be relatively featureless, except for a group of isolated 'peaks' rising to −42 m about 2 km north of Mina Stac, and a ridge running east-west at the same latitude about 2 km south of Stac Lee, near which divers recovered gabbro from a smooth, well-jointed rock surface. Gabbro was also found at 50 m depth about 1.5 km ENE of Levenish near the southern margin of the platform, indicating that the Western Gabbro, or substantial blocks of it, extends eastwards from Dun for several kilometres past Levenish.

The precipitous cliffs around Boreray, Soay and the north side of Hirta commonly continue beneath sea level with no change in slope to depths exceeding 30 m. Underwater overhangs occur beneath Stac Lee; and Scarbhstac, off the south coast of Boreray, lies above a submerged natural arch, with its apex at 30 m below sea level (Ridley, 1980). On the sea bed near the cliffs in many places are abundant large boulders scattered over bare rock surfaces, but the relative scarcity of loose talus indicates that large volumes of rock have been removed from the vicinity. The greatest amount of cliff debris is at the base of the north face of Conachair, but even this lies mostly below sea level. No wave cut platform associated with long term marine erosion at present sea level is found anywhere around the islands, and the stacs are separated from the main islands by deep water (more than 40 m between Stac Lee and Boreray). The coast most exposed to prevailing wave action, from Soay to Dun, is almost devoid of beach deposits and along the south-west coast of Hirta, divers found 'cliffs, overhangs and gullies with rock walls covered with encrusting life'. (Ridley, 1980.) Only small patches of boulder beach west of Stac a Langa and by the Cambir 'neck' occur on the north coast of Hirta. The comparatively sheltered Glen Bay has no significant beach deposits, and soundings revealed a hard smooth floor with a U-shaped east–west profile and a concave slope northwards to about 40 m depth at the mouth of the bay. Along the east side of Glen Bay underwater cliffs down to about −25 m are covered with anemonies. The beach in Village Bay lies along the most sheltered shore of the islands, and at the foot of comparatively gentle slopes.

The greatest area of sediment lies on the floor of Village Bay and extends seawards to about 1 km south-east of the base of the cliffs of Oiseval forming a gentle convex 'mound' with a steep slope eastwards which breaks abruptly to the platform at 50 m depth (Figure 35). A dredged sample was collected from this slope between about 1.25 km SSE of Rubha an Uisge (50 m depth), and about 250 m south of the HWM due south of the summit of Oiseval (30 m depth). This consists of 2245 g (total weight) of which 87% is rock material (with encrustations) and 13% shell debris. It is a loose gravel formed largely of pebbles, mostly sub-angular to rounded, up to 60 mm diameter, with very few rock fragments less than 10 mm across. Most have primary and secondary encrusting bryozoa and worm tubes, which although commonly covering the pebbles, yet form very few aggregates. More than half the rock fraction consists of Conachair Granite, and most of the remainder are dolerites and granites of the Mullach Sgar Complex. About 10% of the total sample consists of rock types not known on St Kilda, and these include biotite-granite with pink and colourless feldspar, amphibolite, granitic and quartzose gneisses, a variety of quartzites, and a silica cemented arkosic sandstone. The latter, which is reminiscent of Torridonian Sandstone, is deep buff-pink and consists of well rounded to sub-angular quartz with pink to red feldspar, and contains a fragment of granitic gneiss with biotite and pink feldspar. Most of the shell material consists of broken, worn fragments heavily encrusted with bryozoa, sponges and worm tubes. A wide variety of species is present including mussels and other bivalves, gastropods, brachiopods, echinoid spines, plates and complete small tests, and siliceous sponge spicules and skeletons.

No samples were recovered from several dredge runs in the area south-west of Boreray and much of the sea bed consists of bare rocks. However, one run was more fruitful and 3233 g were dredged from about 2 km south of Stac Lee. About 35% of this gravel sample consists of smooth, clean, well rounded pebbles and rock granules, only a very few having minor encrustations of bryozoa and worm tubes. The remaining 65% consists mostly of rounded clean shell debris up to about 10 mm across with a few large complete shells. The faunal types are similar to those from Village Bay. The rock fraction of the sample consists of dolerite and gabbro, with some pebbles of basalt and of a hard, grey, very fine-grained rock similar to the early basic sheets. Granules of greenish-grey tuff are also present and may represent extrusive components of the St Kilda igneous suite which have not been preserved above sea level. A single (40 mm) pebble of Conachair Granite was found. Pebbles and granules of rock types not known locally form about 5% of the rock fraction and include biotite-granites, granitic gneiss (some with epidote), amphibolite (some garnetiferous), coarse-grained micaceous quartzite, psammitic schists and epidote-rock. Sedimentary rock fragments include quartzites, siltstones and a variety of sandstones, some red with well rounded quartz grains, and some arkosic with pink feldspar.

A small amount of sand was also obtained from the seabed about 1 km north-east of Mina Stac. It consists of angular quartz sand with some white to buff translucent feldspar, a few pyroxene and magnetite grains and some basic rock fragments.

Although sampling was insufficient to define the exact distribution of sediment, both the divers' observations and the echo-sounding data indicate that much of the sea floor near the islands is free of unconsolidated sediment, and that which does occur is generally present as isolated thin patches. Discrepancies between depths sounded in Village Bay during the present survey and those published on Admiralty charts and between soundings taken a few days apart suggest that some of the sediment present may be moved frequently. (Strong tides are demonstrated in the area by overfalls along the submarine ridge east of Dun). On the other hand, the angularity of the pebbles and their encrustation with delicate bryozoa and worm tubes indicates that the deposit is only rarely disturbed, perhaps during violent storms, which would prevent aggregation of the pebbles and the deposition of fine sediment. Some depth discrepancies may be due to uncharacterised local tide anomalies. The gravel deposit sampled south of Stac Lee is one which has suffered much more abrasion than that in Village Bay, and probably represents material which is frequently disturbed by tidal current and forms patches lying in hollows on the rock platform.

The submarine topography around St Kilda indicates that the present coastline is a very immature one, with many submerged features, well below present sea level, which have been caused by severe erosion. The rugged gully topography to depths of more than 90 m along the west coast may be due to subaerial erosion, and the distinct platform north and east of Hirta, together with the nature of the cliffs and stacs suggests that there was a considerable period when the sea level was 40 to 50 m lower than it is at present. The large bank of heavily encrusted pebbles and shells in Village Bay may therefore be regarded as the remains of a drowned beach, since the outer edge of this deposit lies at about 50 m depth, which is similar to the base of the concave slope below the cliffs of Oiseval.

Studies of sea levels around Scotland (Jardine, 1982) show that although global sea level stood at about -130 m at the Devensian glacial maximum (about 18 000 years ago), relative sea levels around the Scottish mainland were at only about -60 m due to isostatic depression of the land mass near the main centre of glaciation. Peripheral to this basin, isostatic effects are less marked and relative sea levels are more closely related to absolute sea level changes. Some features of the St Kilda coast and sea floor are strikingly similar to parts of the Shetland coast, where submergent cliffs with no wave cut platform or beaches, and concave slopes below sea level from the base of the cliffs to great depths are attributed to a continuing net rise in relative sea level over a long period covering several glacial episodes (Flinn, 1964). Such a process would seem the most likely to account for the present St Kilda coast and submarine features, particularly the great heights of the cliffs and the large area of platform at 40 to 50 m depth.

Chapter 20 Fracturing and faulting

Keywords: fracturing and faulting sequence, topographic expression, mineral coating

This section is concerned with the brittle failure of consolidated rocks in the St Kilda igneous complex. Localised failure, commonly confined to a specific intrusion, is referred to as fracturing or brecciation, whereas more extensive failure usually affecting more than one intrusion, is treated as faulting. Auto-brecciation, the result of preconsolidational magmatic processes, is dealt with in preceding sections.

The earliest evidence of fracturing occurs in the Western Gabbro where under high confining pressures, hot, solid gabbro failed along discrete shears and resulted in high temperature recrystallisation of the mafic mineral assemblage (p. 15). This early event appears to have been the precursor to a more intensive episode of fracturing and disintegration of parts of the Western Gabbro which was accompanied by the intrusion of basaltic magma. The resulting igneous breccia EK almost certainly represents the root zone of a large basaltic volcano, parts of which have suffered collapse and further disintegration during a subsequent episode of basaltic magmatism.

In the area of Village Bay several outcrops of the Mullach Sgar Complex display brittle structures which clearly post-date auto-brecciation associated with the intrusion of the complex. Such structures are well displayed in the quarry below Mullach Geal, where the earliest are steep, eastward-dipping fractures trending 356°, with slickensides indicating movement towards the NNE at shallow plunges. These fractures are best developed at the northern and southern extremities of the quarry face and are also found farther east in the Abhainn Mhor where they appear to control the course of the stream above 140 m and also below 60 m. The more highly fractured central part of the quarry lies at the intersection of a number of differently orientated mineralised surfaces all of which post-date the north-trending fractures. Intermediate dips to the north-east and south-west characterise the earliest of the mineralised fractures, which carry alkali feldspar, calcite, prehnite, epidote, apophyllite and chabazite and a slickensided blue-green coating of chlorite and amphibole. They are cut by fractures carrying a similar mineral suite which trend east–west and dip steeply to the north; slickensides indicate movement on these planes to the north-east at intermediate plunges. The latest mineralised surfaces are less numerous than earlier ones and may be heavily coated with blue-green slickensided material, while carrying less of the whitish minerals noted above. These fractures trend NE–SW and dip steeply south-east; easterly movement at intermediate plunges is indicated by slickensides. A considerable concentration of fractures also occurs in outcrops of the Mullach Sgar Complex, west of the boulder beach in Village Bay. The earliest of these are generally north and NNW-trending surfaces with variable dips north-east and south-west, carrying pale-green to whitish mineral coatings of similar composition to those found in the quarry. A few early east–west trending, northward-dipping slickensided surfaces are also seen and have an orientation similar to fractures occurring in the Abhainn Ruaival; their relationship with the north or NNW-trending fractures is not clear. Steep or vertical, NE-trending slickensided planes appear to represent the latest episode of fracturing. Slickensides on these surfaces plunge north-east at intermediate angles, as indeed do slickensides on the other fracture surfaces.

Several structures are seen in exposures of the Conachair Granite in the cliffs and tidal rock shelves which extend south-eastwards from the jetty in Village Bay. On the rock shelves a conspicuous reddened fault plane trending at 300° and dipping 75° north-east can be followed for about 150 m up to the jetty, cutting an earlier fault trending at 330° and dipping 60° north-east. The younger fault extends into a zone of shattered Mullach Sgar Complex forming the course of the Abhainn Mhor between 60 m and 140 m. Similar 300°-trending faults coated with orthoclase and quartz are exposed in the cliffs surrounding the cave at the south-east end of the shelves, where they in turn are cut by vertical planes trending at 220° which are also mineralised with felsitic material. Post-mineralisation slickensides on the later set of fractures suggests that the most recent movement was mainly vertical with a small north-easterly horizontal component. Vertical slickensided planes with a similar orientation (220°) can be seen beside the slipway to the jetty, where again they cut the earlier faults dipping at 55° north-east and trending 305°.

In Glen Mor several drift covered NW-trending faults can be located from stream exposures of shattered rock; the kink in the Abhainn a Ghlinne Mhoir shown in Plate 37A follows one of these faults. They appear to be terminated by NE–SW-trending faults along which such features as the Dun Passage, Cambir Neck and probably Soay Sound have been formed by erosion. The latest set of faults appears to have developed more or less contemporaneously with the intrusion of the suite of late dykes and cone sheets. For example, at Geo na Lashulaich (Figure 36) several cone sheets converge on a pre-existing NE–SW fault, and while a similar fault cutting Conachair Granite at Geo nan Sgarbh displaces some basic sheets, another sheet appears to have been intruded along the fault plane for several metres.

Field relationships suggest that the following sequence of structural events can be distinguished on St Kilda:

Chapter 21 Palaeomagnetism

Keywords: sampling, statistics, reverse magnetisation, Tertiary pole

Oriented cores for palaeomagnetic investigations were obtained from 19 sites in a variety of lithologies on the main island (Figure 38A). Samples from 17 of these sites responded extremely well to alternating-field (af) demagnetisation and yielded stable well-grouped end points. These stable magnetisations were approximately reversed with respect to the present day field direction, having southerly declinations and moderately steep negative (upward) inclinations. The palaeomagnetic statistics of each of these 17 sites, together with the overall final statistics, are given in (Table 38).

At most sites the total NRM (that is, uncleaned) directions of the individual samples were already moderately or well grouped with southerly declinations and negative inclinations, and af demagnetisation simply improved the grouping and generally slightly steepened the inclinations. At some sites, however, the total NRM directions were either loosely grouped around the earth's present Field direction: they had northerly declinations and steep positive inclinations, or they were more or less randomly distributed, but in both cases af demagnetisation caused the directions to move to stable well grouped end points with southerly negative inclinations very similar to those shown by the other sites. A number of samples were subjected to thermal demagnetisation and this technique produced stable directions indistinguishable from those obtained using af demagnetisation.

The two sites which gave unsatisfactory results were site 1, a rusty-weathering dolerite sheet in the Dun Passage, and site 4, in the Conachair Granite. The total NRM directions at site 1 were well grouped, almost exactly parallel to the earth's present field direction, and af and thermal demagnetisation did not produce any appreciable change in the mean direction, but caused the grouping to deteriorate and also indicated that the stability or remanence was much less than for any of the 17 reversed sites. Although it is of course impossible to be entirely certain that the loose grouping of cleaned directions does not represent a primary normal remanence, it is felt that the balance of evidence suggests that this magnetisation is of very recent origin, and consequently this site was excluded from the overall statistics. At site 4 the cleaned directions were spread out between the earth's present field direction and the reversed direction found in the 17 stable sites, suggesting that this site contains both these components, but with overlapping stability spectra so that they cannot be isolated by demagnetisation. This site was also excluded from the final statistics; whether its northerly component is of recent origin or represents a primary normal remanence is difficult to decide.

Thus all 17 sites which showed good stability have stable magnetisations with southerly declinations and moderately steep negative inclinations. The sites cover a variety of lithological types and include the Western Gabbro, the gabbro on Mullach Geal, the Glen Bay Gabbro, the Glen Bay Granite, and the rocks in the Mullach Sgar Complex. At each site at least 7 separately oriented cores were analysed, and (Table 38) shows that the within-site grouping of cleaned directions was always good, with most sites having a circle of confidence (α 95) of less than 5°, and no site having an α 95 of more than 12°. The 17 cleaned site mean directions are plotted in (Figure 39A), which shows that there was also very close between-site grouping.

The fact that all the stable magnetisations on St Kilda are of reversed polarity suggests that the intrusion of the whole complex occurred over a fairly short time period during one polarity interval. The most recent version of the Cainozoic polarity reversal time scale (Lowrie and Alvarez, 1981) shows an uninterrupted reversal interval of longer than average duration between about 53.7 and 55.8 Ma, so it is evident that a suitable period of reversed polarity occurred at the time which radiometric dating (pp. 40–1) indicates is the most probable age of intrusion. The pole corresponding to the overall mean direction from St Kilda is shown in (Figure 39B), together with a recent version of the Cainozoic and late Mesozoic apparent polar wander (apw) path for northern Eurasia. Although both the apw path and the St Kilda pole are quite precisely defined, the confidence limits associated with both, together with the relatively slow rate of apw during this period, preclude using the position of the pole for accurate dating. It is noted however, that the St Kilda pole is indistinguishable at the 95% probability level from the 55 Ma pole, and so is consistent with this age assignment. The position of the St Kilda palaeomagnetic pole on the far side of the geographic pole implies that Britain has drifted about 19° northwards in the last 55 million years.

Chapter 22 The age of the Conachair Granite

Keywords: Rb–Sr isotopes, initial ⁸⁷Sr/⁸⁶Sr ratios, granite, gabbro

The islands of St Kilda are made up of a series of intrusive igneous rocks whose relative ages can be determined by examining their mutual contacts, but the boundary between the whole complex and the surrounding country rock has so far proved inaccessible beneath more than 50 m of water. So in this case one cannot assign an older age limit to the complex by identifying the country rock, and, similarly, a younger limit cannot be assigned from field mapping methods because (apart from glacial deposits) neither lavas, sediments nor any other country rock remains of the cover that must have overlain the complex. The intrusions have been considered Tertiary solely on the basis of their similarity with those in Mull, Skye and other Tertiary centres.

One way to determine rock age uses the radioactive decay of part of the trace element rubidium (Rb). Rubidium consists of two isotopes with different masses, ⁸⁵Rb and ⁸⁷Rb; the latter breaks down to ⁸⁷Sr by losing an electron. The rate of this breakdown or decay is constant, so if the amount of Rb in the rock and the amount of Sr that has formed as a result of the decay are known, the time taken to form the Sr can be calculated. This radiogenic ⁸⁷Sr joins the isotopes of Sr (88, 87, 86 and 84) initially present in the rocks, the extra amount being most conveniently measured as an increase in the ⁸⁷Sr/⁸⁶Sr ratio. The parts of the rock that contain high Rb will, through time, gain more ⁸⁷Sr than those parts with low Rb and this is best illustrated by measuring and plotting the ⁸⁷Rb/⁸⁶Sr and ⁸⁷Sr/⁸⁶Sr isotope ratios on a graph known as an isochron diagram.

Ideally, the slope of the isochron is proportional to the age of the rock. Each sample of the rock should lie on the isochron but in practice this rarely happens. First because there are small experimental errors in measuring the isotopic ratios and secondly because the isotopic systems may be disturbed and small amounts of radiogenic strontium may be lost from certain samples by weathering and other processes of alteration. Isotopic measurements were made on 33 samples from St Kilda. The results are tabulated in (Table 41). When all the data are plotted on an isochron diagram, the scatter about the best fit line is considerable as indicated by the high value of 31 for the MSWD (mean square weighted deviates, an estimate of the goodness of fit of the points to a single line). When MSWD has a value of 2.5 or less, all the scattering about the line can be attributed to analytical error, but if the number is greater than 2.5, then there is a geological reason for the scattering and the errors on the age of the rock must be increased to account for this. Two samples from St Kilda fall statistically a long way from the line which passes through all the other points: an aplite vein, which is probably younger than the other intrusive bodies; and a marginal sample of the Conachair Granite which may be contaminated. When these samples are removed from the plot, the MSWD becomes 7 (Figure 40) and when the errors on the age are enhanced to allow for the small excess scatter, the age of the St Kilda intrusive complex becomes 55 ± 1 Ma, in good agreement with the conclusions drawn from the Palaeomagnetism results in the previous section.

On this basis one can calculate the apparent initial ⁸⁷Sr/⁸⁶Sr ratios for each point and evaluate the differences in these for each phase of intrusion and thus suggest possible geochemical relationships between the different phases. The Conachair Granite shows the widest range in possible initial strontium ratios which may reflect varying source contamination of the magma or later disturbance of the Rb-Sr systematics ((Table 41)). The calculated possible initial strontium ratios for the Mullach Sgar Diorite, the Mullach Sgar Microgranite and the Glen Bay Granite show remarkable internal consistency. The means of the initial ratios for each phase are significantly different and suggest varying contamination of the source magma for each distinct intrusion. However, the total range in initial ⁸⁷Sr/⁸⁶Sr ratios is very small and may indicate derivation from a common magma reservoir. The sample of aplite, No. 327A, appears to be younger than the main intrusive phases. For an age of 55 Ma, its calculated apparent initial ratio is unacceptably low. With the more realistic value for the initial ⁸⁷Sr/⁸⁶Sr ratio of 0.7039, the age calculated for this phase becomes 50 Ma.

In the context of Tertiary igneous rocks of the North Atlantic region, the initial ⁸⁷Sr/⁸⁶Sr ratios are amongst the lowest yet recorded.

Chapter 23 Geochemistry of the major intrusions

Keywords: chemical ratios, major rock groups, rare earth element patterns

This section attempts to explore some of the relationships of the St Kilda igneous rocks by plotting their chemical compositions on appropriate variation diagrams. In (Figure 42) the iron–magnesium ratio of each rock is plotted against its alkalis-calcium ratio. Each of these ratios is a measure of how well differentiated or how mixed a particular rock may be and although the ratios must be used cautiously they provide some measure of the relationships amongst a variety of rock types in one igneous centre. Analysed samples of the Western Gabbro occupy a field with the lowest iron ratio but is similar to the Cambir Dolerite field in its alkali–lime ratio. Likewise the Cambir Dolerite has iron ratios similar to both basalts in the breccia EK and to some Phase 3 dolerites in the Mullach Sgar Complex, but the different alkali ratios in these three groups of rocks suggest that they have arisen in different ways. Two pegmatites from the Western Gabbro (P in (Figure 42)) are perhaps the closest approximation to compositions of residual liquids resulting from solidification of the Gabbro. The alkali ratio of the upper pegmatite point (Figure 14A) is similar to ratios in the Glen Bay Gabbro but the iron ratios are different. The Glen Bay Gabbro crystallised from a tholeiitic magma with a composition assumed to be that of the chilled margin in east Glen Bay and produced pegmatitic residues richer in potassium, phosphorus and rare earth elements than the E^w pegmatites. Such differences suggest that the Western Gabbro and Glen Bay Gabbros are genetically unrelated.

In the Mullach Sgar Complex some dolerites of Phase 3 are relatively low in iron and alkalis but other Phase 3 dolerites and those of Phase 2 and 4 are higher and approach the ratios found in Phase 2 granodiorite, being distinguished chemically by lower K₂O contents and petrographically by finer grain size. The latest granite phase (4) in the Complex has ratios quite separate from the granodiorite but they are indistinguishable from those of the Glen Bay Granite. This coincidence demonstrates that although the fields in (Figure 42) appear to indicate a correlation between chemical composition and sequence of intrusion, from oldest Western Gabbro to youngest Conachair Granite, in fact at the high alkali end the Glen Bay Granite is the earliest of the granitic intrusions predating those of the MSC, whereas at the low alkali end relatively magnesian dolerites were intruded quite late in the Mullach Sgar Complex.

A similar overall pattern of progressive variation in chemistry with age is also suggested by the rare-earth element (REE) contents plotted in (Figure 43A) as rock to chondritic meteorite ratios. Again however, in detail, such correlation is false since the earliest granite (Glen Bay) has the greatest enrichment of REE's with La being 200 times and Lu 40 times more abundant than the average chondrite. The younger Conachair Granite has only 100 times the chondrite La content and shows a pronounced negative Eu anomaly, distinctive among the REE patterns of St Kilda rocks. Its Eu content is the same as that of the Western Gabbro but here the rest of patterns 8 and 9 are almost mirror images of that of the Granite, with low enrichment (8) or even depletion (9) of the light REE's and 2 and 4 times chondrite content of the heavy REE's. The Cambir Dolerite pattern (7) is similar to that of E^w with relative enrichment of heavy REE's but the two EK rocks (5 and 6) give almost flat patterns with equal enrichment (10–30 times) of light and heavy REE's. The Glen Bay Gabbro is significantly enriched in light REE's and its pattern is very similar to those of the Mullach Sgar Complex rocks. Close links between these intrusions are also indicated by their similar initial ⁸⁷Sr/⁸⁶Sr ratios (about 0.7041, (Figure 40)) and by the presence of rare-earth minerals. Three minerals with significant REE contents occur in pegmatites related to the Glen Bay Gabbro and also in the three main granites. Chevkinite has the highest content of REE's (41%), and is a common accessory mineral in the granites of Glen Bay (Figure 43B), Mullach Sgar and Conachair. It is very enriched in the light REE's and resembles allanite in this respect, although the latter is poorer in total RE contents (22%). Allanite occurs in the Mullach Sgar granites and in pegmatites of the Glen Bay Gabbro where it is associated with zirkelite (Figure 14C), a Zr–Y titanate relatively low in total RE content (7%) and displaying a significantly different REE pattern (Nd> Ce) from that of allanite. The differing mode of crystallisation of these three minerals appears to be reflected in the patterns 1, 2, 3 and 4 of (Figure 43A), forming a group of light REE-enriched rocks. Zirkelite crystallised in residual liquids derived from solidification of basic magma (pattern 4), one of which was perhaps already partially depleted in light REE's by crystal fractionation processes. Chevkinite and allanite crystallised from more granitic magma enriched in light REE's (patterns 1, 2 and 3). The similarity between patterns 1, 3 and 4 suggests that these rocks may be related through fractionation to a common parental magma. However the distinctly different pattern 2 of the Conachair Granite is not easily linked to the same fractionation event. It shows a strongly negative Eu anomaly which is only mirrored by the positive Eu anomaly displayed by the cumulates of the Western Gabbro, and it seems more likely that a separate fractionation process links these two intrusions.

The geochemical data corroborate field and petrographical evidence suggesting that the major intrusions on St Kilda fall into two major groups. The first, a group of mafic rocks, comprises the Western Gabbro, Cambir Dolerite and the rocks of Glacan Mor, Boreray and Soay. In (Figure 42) this group is characterised by low alkali–lime ratios and in (Figure 43A) by a lack of light REE enrichment. The second group range from mafic to granitic compositions and not only have higher alkali and iron ratios but also contain more K, P and REE's, relatively enriched in the light elements. Several members of this second group appear to be related through fractionation to a parental tholeiitic magma of the Glen Bay type. Relationships among the first group are less clear: if the Conachair Granite is related by fractionation to the Western Gabbro, this might explain the relative depletion of the latter in REE's and low alkali and iron ratios. Further sampling of the breccia EK may reveal rocks intermediate between these two groups, and indeed may indicate components from even more sources.

Geological history

Acknowledgements

We wish to acknowledge the generous help afforded by the many people in different organisations in carrying out the St Kilda project. In particular Dr J. Morton Boyd, R. N. Campbell, C. Brown and Dr M. E. Ball of the Nature Conservancy positively encouraged the project as did Mr D. MacLehose of the National Trust for Scotland. The local knowledge of Wally Wright, the Warden on St Kilda representing both organisations, made possible more extensive mapping and collecting than we had hoped for, and the rock-climbing skills of Stewart Murray enabled access to many of the islands. The assistance and cooperation of the St Kilda Detachment, Royal Artillery, Guided Weapon Range under (at various times) Captains M. S. Forsyth, A. Cameron, and D. J. A. Cooke are much appreciated, and the willingness and expertise of our boat captain Andy Miller Mundy and his crew in undertaking the hazardous survey work near the cliffs is now legendary. We are very grateful for submarine samples collected by G. Ridley and his team (1979) and by Dr P. Kokelaar (1983), and their assistance has added significantly to ideas both about extent of intrusions and about Quaternary geology.

Specimens and thin sections were prepared in the Petrology Unit by C. W. Wheatley and R. D. Fakes, and Mr R. K. Harrison Dr M. T. Styles and Mr B. R. Young have assisted materially with discussion and X-ray data. We are grateful to R. T. Smith (Metalliferous Minerals and Applied Geochemistry Unit of BGS) for training and advice in geochemical sampling, and aspects of the Quaternary geology owe much to ideas discussed with both Dr J. D. Peacock (Highlands Unit of BGS) and Dr D. G. Sutherland (Edinburgh University). All the photographs were processed in the Photographic Department and special thanks are due to J. M. Pulsford, H. J. Evans, C. J. Jeffery and R. E. Collins for the help they gave at all stages in the project. Miss L. Wahl drew the map and the diagrams and made many valuable suggestions for their improvement, and the final production of the map and report owes much to G. F. Inzani, J. B. A. Evans, Gill Cutress and Dr T. J. Dhonau.

References

BELL, J. D. 1982. Acid intrusions. Pp. 427–440 in SUTHERLAND, D. S. (Ed.). Igneous rocks of the British Isles. (New York: Wiley.)

COCKBURN,A. M. 1935. The geology of St Kilda. Trans. R. Soc. Edinburgh, Vol. 58, Pt. 2, 511–547.

FLINN, D. 1964. Coastal and submarine features around the Shetland Islands. Proc. Geol. Assoc., Vol. 75, 3, 321–339.

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

HARDING, R. R. 1966. The Mullach Sgar Complex, St Kilda, Outer Hebrides. Scott. J. Geol., Vol. 2, 165–178.

HARDING, R. R. 1967. The major ultrabasic and basic intrusions of St Kilda, Outer Hebrides. Trans. R. Soc. Edinburgh, Vol. 66, 419–444.

HARDING, R. R. MERRIMAN, R. J. and NANCARROW, P.H. A. 1982. A note on the occurrence of chevkinite, allanite and zirkelite on St Kilda, Scotland. Mineral. Mag., Vol. 46, 445–448.

HASKIN, L. A., HASIUN, M. A., FREY, F. A. and WILDEMAN, T. R. 1968. Relative and absolute terrestrial abundances of the rare earths. Pp.889–911 in AHRENS, L. H. (Ed.). Origin and Distribution of the Elements. (Oxford: Oxford University Press)

IRVINE, T. N. and BARAGAR, W. R. A. 1971. A guide to the chemical classification of the common volcanic rocks. Can. J. Earth. Sci., Vol. 8, 523–548.

IRVING, E. 1977. Drift of the major continental blocks since the Devonian. Nature, London, Vol. 270, 304–309.

JARDINE, W. G. 1982. Sea-level changes in Scotland during the last 18 000 years. Proc. Geol. Assoc., Vol. 93, 1, 25–41.

JONES, E. J. W. 1981. Seismic refraction shooting on the continental margin west of the Outer Hebrides, northwest Scotland. J. geophys. Res., Vol. 86, B12, 11553–11574.

LE MAITRE, R. W. 1976. The chemical variability of some common igneous rocks. J. Petrol., Vol. 17, 589–637.

LOWRIE, W. and ALVAREZ, W. 1981. One hundred million years of geomagnetic polarity history. Geology, Vol. 9, 392–397.

MACCULLOCH, J. 1819. A description of the western islands of Scotland including the Isle of Man. Vol. ii. (London: Constable).

MEIGHAN, I. G. 1979. The acid igneous rocks of the British Tertiary Province. Bull. Geol. Surv. G.B., No. 70, 10–22.

PLANT, J. and MOORE, P. J. 1979. Regional geochemical mapping and interpretation in Britain. Phil. Trans. R. Soc. London, Vol. B288, 95–112.

RIDLEY, G. 1980. Diving in Scotland. Part 2. The British Sub-Aqua club St Kilda Survey Expedition 1979. Underwater World, Vol. 3, No. 1, 12–39.

ROSS, A. 1884. A visit to St Kilda. Trans. Inverness Sci. Soc., Vol. iii, 72–91.

STRECKEISEN, A. 1976. To each plutonic rock its proper name. Earth-sci. Rev., Vol. 12, 1–33.

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

TAYLOR, A. B. 1968. The Norsemen in St Kilda. Saga Book Viking Soc., Vol. 17, 116–144.

THOMPSON, R. N. 1982a. Magmatism of the British Tertiary Volcanic Province. Scott. J. Geol., Vol. 18, 49–107.

THOMPSON, R. N. 1982b. Geochemistry and magma genesis. Pp 461–477 in SUTHERLAND, D. S. (Ed.). Igneous Rocks of the British Isles. (New York: Wiley.)

THOMPSON, R. N. 1983. Thermal aspects of the origin of Hebridean Tertiary acid magmas, II. Experimental melting behaviour of granites at 1 kbar P_H2O . Mineral. Mag., Vol. 47, 111–122.

THOMPSON, R. N., ESSON, J. and DUNHAM, A. C. 1972. Major element chemical variation in the Eocene lavas of the Isle of Skye, Scotland. J. Petrol., Vol. 13, 2, 219–253.

WAGER, L. R. 1953. The extent of glaciation in the island of St Kilda. Geol. Mag., Vol. 90, 177–181.

WAGER, L. R. and BAILEY, E. B. 1953. Basic magma chilled against acid magma. Nature, London, Phys. Sci., Vol. 172, 68.

Figures and tables

Figures

(Figure 3A) Banding in Type 3 gabbro at a height of 700 ft (210 m), due south of Claigeann an Tigh Faire. Weathered surfaces accentuate the abrupt changes in mineral content and small-scale shearing in the gabbro.

(Figure 3B) Type 1 gabbro with a small sheared plagioclase-rich lens close to the ruler. Gob Chathaill (south of Mullach Bi) at a height of 80 ft (25 m).

(Figure 3C) Banding in Type 2 gabbro on the southern face of Ruaival at a height of 250 ft (76 m). The black pyroxene and its inclusions of white feldspar give the rock a characteristic speckled appearance.

(Figure 3D) In this type 3 gabbro, pyroxenes with tiny feldspar inclusions are harder than surrounding minerals and form a knobbly weathered surface. Dark green veins of amphibole and chlorite which cut across the banding are a common feature. Locality about 800 m north of Mullach Bi.

(Figure 4A) Rounded grains of olivine (blue and orange) are partly enclosed by augite (pale brown) next to magnetite and ilmenite (black) and to zoned plagioclase (grey). A fine vermicular intergrowth of orthopyroxene with magnetite and ilmenite occurs at one end of the central opaque mass. (S65855), a Type I gabbro similar to that in (Figure 3B); cross polarised light; field 4 mm wide.

(Figure 4B) Mineral lamination in Type 3 gabbro from Claigeann an Tigh Faire. (S65853), cross polarised light; field of view represents 4 mm (width) of rock.

(Figure 4C) Poikilitic augite (yellow) 2 mm across contains rounded inclusions of plagioclase (grey) and olivine (red and orange) which are smaller than the same mineral species outside the pyroxene. (S64305) from the top of Mullach Bi; cross-polarised light.

(Figure 5) Pyroxene and olivine compositions of St Kilda igneous rocks. Compositions of minerals found in the same thin section are joined with tie lines.

(Figure 6A) Sheet of fine-grained Cambir Dolerite dipping gently to the left cuts across Western Gabbro with white feldspar-rich bands dipping steeply to the right. Locality: Gob Chathaill at a height of 80 ft (25 m).

(Figure 6B) Western cliffs of the Cambir. Cambir Dolerite occurs beneath and south of the summit at heights of 300–500 ft (90–150 m).

(Figure 7A) Clinopyroxene (brown), plagioclase (grey) and magnetite (black) grains averaging 0.2 mm across form a granoblastic texture. Orthopyroxene with magnetite inclusions occurs near the centre of the picture. S 67640; western cliffs of Cambir; plane polarised light.

(Figure 7B) Detailed view of poikilitic orthopyroxene and granoblastic feldspar and clinopyroxene. S 67640; field 1 mm wide, cross polarised light.

(Figure 8A) View of Village Glen with Mullach Geal forming the skyline. The highest outcrops of Breccia can be considered to form a capping to the dolerite and microgranite intrusions exposed in the quarry.

(Figure 8B) The cliffs of Glacan Mor, Gob na h-Airde, the tip of the Cambir and Soay consist of a Breccia of gabbros and dolerites. Large rafts of gabbro very similar to the Western Gabbro occur high on Soay but lower down, gabbros with different textures and dolerite dykes are more abundant. Close and irregular jointing is a feature of the dolerites in the breccias and is particularly evident in the cliffs of Glacan Mor.

(Figure 9A) Gabbros of various kinds intruded by dolerites form the western cliffs of Boreray. Prominent joints with a moderate southerly dip are commonly the sites of late, easily-weathered dolerite sheets.

(Figure 9B) Boulder of pegmatitic gabbro intruded by dolerite near a landing place on Soay. Stac Biorach and Soay Stac rise from Soay Sound which at low tide is only 8 m deep.

(Figure 10A) Gabbro from Boreray. Olivine (2 mm across) has been replaced by serpentine and chlorite whereas augite (showing blue interference colours) remains fresh. Twinned plagioclase is extensively net-veined by albite. (S67657A) from near sea level on west coast of Boreray; cross-polarised light.

(Figure 10B) Metamorphosed gabbro. The original olivine, pyroxene, plagioclase and opaque minerals of the gabbro have been subjected to shear stress at high temperature and as a result have bent, cracked and recrystallised into granules of the same mineral species. In detail the granule compositions reflect the minerals they were derived from and the variations in plagioclase granule composition reflect zoning in the original igneous grains. (S64302) from Mullach Geal at 900 ft (275 m); cross-polarised light; area shown represents 1 mm across.

(Figure 10C) Basaltic dolerite with flinty fracture. Small plagioclase laths up to 50 µm long lie in a fine granular groundmass of pyroxene and opaque minerals. Grain size is variable and fractures have been sealed by the development of chlorite and amphibole. (S67653) from Glacan Mor; plane-polarised light.

(Figure 11A) Dolerite vein in gabbro. The larger plagioclase laths up to 150 µm long retain their igneous fabric but their margins are modified and they lie in a recrystallised ground-mass of equant grains of pyroxene, plagioclase and opaque minerals. 564876 from Mullach Geal; cross-polarised light.

(Figure 11B) Xenolithic dolerite recrystallised to granoblastic texture contains an altered olivine inclusion with a reaction rim 100 µm wide. The reaction zone comprises crystals of orthopyroxene and plagioclase roughly perpendicular to the margin. A later period of hydrothermal alteration has caused development of biotite (pale brown) along a narrow zone that crosses the dolerite and runs along the reaction zone margin. (S68262) from near sea level, north end of Soay Stac; plane-polarised light.

(Figure 12) Vertical banding in the eastern part of the Glen Bay Gabbro. Plagioclase bands up to 20 mm across alternate with darker bands in which pyroxenes and opaque minerals are concentrated.

(Figure 13A) Chilled margin of the Glen Bay Gabbro. Xenocrysts of plagioclase and olivine lie in a groundmass of plagioclase, pyroxene and opaque minerals. Both xenocrysts are rounded and corroded; the plagioclase has provided a nucleus for the growth of more sodic plagioclase (pale rim), and pyroxene has nucleated on the olivine. (S64879A), 150m south of tunnel at Gob na h-Airde; cross polarised light; area shown represents 4 mm across.

(Figure 13B) Gabbro consisting largely of plagioclase (2 mm long), rounded olivine enclosed in a mass of smaller pyroxenes (pale brown) and opaque minerals. Brown biotite mantles some of the opaque grains and prismatic apatite lies adjacent to the subhedral plagioclase. 565208, 300 m south of tunnel at Gob na h-Airde; plane-polarised light.

(Figure 13C) Gabbro rich in magnetite, ilmenite (opaque) and augite pyroxene (blue and yellow interference colours). Apatite is abundant as inclusions (grey) in the opaque minerals and its hexagonal (dark) outlines are visible next to pyroxene. Sinuous lines of crushed material penetrate the rock. Although this comes from an isolated block on Mullach Sgar, it is the same as some facies of the gabbro on the western side of Glen Bay. 568268, cross-polarised light.

(Figure 14A) Black pyroxene crystals, up to 6 in (150 mm) long, and opaque minerals surround a leucocratic core of albite, prehnite and epidote. Pegmatite veins at a height of 400 ft (122 m) below Claigeann an Tigh Faire

(Figure 14B) Rhomb-shaped cross section of epidote next to Fibrous chlorite (at extinction) intergrown with prehnite. Pegmatite from south-east end of the Cambir, (S64900), field 1 mm across, cross polarised light

(Figure 14C) Pyroxene partly altered to opaques and iron-stained chlorite, long needles of apatite (colourless) and shorter acicular grains of zirkelite (brown) lie in a groundmass of turbid alkali feldspar intergrown with quartz. Pegmatite in Glen Bay Gabbro; (S65207), field 4 mm across, plane polarised light.

(Figure 15A) A network of veins consisting of amphibole and chlorite stand out on a weathered surface of the Western Gabbro about 800 m north of Mullach Bi.

(Figure 15B) Coarse gabbro consisting of twinned plagioclase (grey) and rounded olivine (yellow and blue) is cut by a vein containing fibrous chlorite and both fibrous and platy amphibole. The partial alteration of olivine to talc and amphibole is well defined. (S65863), 1 km north of Mullach Bi; field 4 mm wide, cross polarised light

(Figure 15C) Colourless plagioclase penetrated by pale green fibrous calcic amphibole (tremolite) with small granular patches of dark green spinel up to 0.5 mm across. Altered Western Gabbro south of Cambir neck; (S67646), plane polarised light.

(Figure 16A) Contact of Glen Bay Gabbro and fine-grained Glen Bay Granite on the western rock shelves of Glen Bay. Both rocks are crossed by shear planes, many of which are sub-parallel to the contact.

(Figure 16B) The light-coloured rock in foreground and middle distance is Glen Bay Granite. Dark rock in lower foreground is sheared gabbro and there is an irregular contact between the two rock types in the rubbly cliffs to the right.

(Figure 17A) Photomicrograph of Glen Bay Granite. Crystals of plagioclase up to 2 mm long rest in a groundmass of smaller crystals of plagioclase mantled with orthoclase, pyroxene, amphibole, quartz and opaque minerals. (S67632) from near eastern end of Granite, cross-polarised light.

(Figure 17B ) Photomicrograph of crushed Glen Bay Granite. Grains up to 1 mm across, penetrated by zones of incipient granulation, lie in a mosaic of plagioclase, orthoclase, quartz, amphibole and opaque minerals. (S64819) near contact with eastern part of Glen Bay Gabbro.

(Figure 18) Lobate fragment of microdiorite, 80 m long, enclosed in steeply-dipping sheet of microgranite (Phase 4) on Na h-Eagan.

(Figure 19) Detailed map of Na h-Eagan and Ruaival.

(Figure 20A) Orthopyroxene grains 0.5 mm across, pale grey and surrounded by yellow green alteration minerals, brown to yellow-green amphibole, smaller grey grains of clinopyroxene and specks of opaque minerals lie in a groundmass of variably turbid feldspar and a little quartz. Mafic microdiorite component of Phase 2 from Dun Passage, (S64883), plane-polarised light.

(Figure 20B) Zoned oligoclase platy crystals and elongate amphiboles (yellow and red interference colours) up to 1.5 mm across, are associated with subhedral quartz, opaque, sphene and interstitial turbid alkali feldspar which forms a characteristic fringe on many oligoclase grains. Microgranite component of Phase 4 from Na h-Eagan, (S65206), cross-polarised light.

(Figure 20C) Basaltic dolerite chilled against microgranite. The margin of the dolerite is sinuous and very finely granular with a concentration of opaque granules and a few thin hollow (quenched) crystals of plagioclase (grey) less than 0.2 mm long. Beach boulder, (S71661A), cross-polarised light.

(Figure 21A) The contact of an angular dolerite block cut by microgranite runs diagonally across the picture. The dolerite is unchilled, and the yellow staining (with sodium cobaltinitrate) illustrates the extent to which K-feldspar, derived from the microgranite, has developed in the dolerite margin. (S64806) from quarry, plane-polarised light.

(Figure 21B) Margin of a basalt fragment that appears to have been shredded or torn apart. Elongate, hollow (quenched) plagioclase laths are abundant, and an olivine phenocryst partially altered to opaques and chlorite lies near the centre of the picture in a grey green groundmass of pyroxene, amphibole and chlorite. The quenched basalt has developed a granular groundmass in contact with hot microgranite. (S64809) from quarry, plane-polarised light.

(Figure 23A) Bipyramids of inverted β -quartz, some showing corroded margins, are enclosed by large grains of perthitic alkali feldspar. (S64823), Conachair Granite from top of Oiseval; cross-polarised light; field 4 mm wide

(Figure 23B) A needle-like grain of chevkinite 0.5 mm long is enclosed in a granophyric intergrowth of clear quartz and perthitic alkali feldspar (stained yellow by sodium cobaltinitrite). (S64810), plane-polarised light.

(Figure 23C) Coarse granophyric intergrowths of clear quartz and perthitic alkali feldspar pass into fine granophyric inter-growths of vermicular quartz and perthite. The fine granophyre is moulded onto coarse granophyre developed within perthitic feldspars near the margin of the Granite north east of Conachair. (S64813), cross-polarised light; field 4 mm wide.

(Figure 23D) Granoblastic intergrowth of clear quartz and perthitic alkali feldspar resulting from annealing of rhyolitic veins which intruded hot Conachair Granite. (S67644), aplite vein in Glacan Chonachair; cross-polarised light; field 4 mm wide.

(Figure 24A) The arch at Gob na h-Airde, Glen Bay, which has formed by the preferential erosion of the point of intersection of a steeply inclined composite dyke (in the roof of the arch) and a shallow-dipping basaltic sheet.

(Figure 24B) The dykes on Leacan an Eitheir, Glen Bay. E°, Glen Bay Gabbro; G, Glen Bay Granite; P, early banded porphyritic felsite; D, dolerite (short line pattern); M, microdiorite (crosses); F, microgranite and felsite (circles); R, late rusty-weathering basalt or dolerite sheets. Felsite P is near-vertical, R has a shallow irregular dip SE, and the remaining dykes dip steeply SE.

(Figure 24C) Inclined sheets cutting granite in the cliffs below Conachair (from Geikie, 1897)

(Figure 25A) Multiple inclined sheets dipping NE at 50°, cutting the Western Gabbro at An Fhaing, Dun. The upper and lower sheets are each 1 m thick.

(Figure 25B) Composite dyke 2 m-wide cutting dolerites and microgranites of the Mullach Sgar Complex on Na h-Eagan. The dyke has a basaltic outer component with a centre of porphyritic pitchstone; the lobate nature of the contact between the two is well displayed.

(Figure 25C) Dolerite sheets on Oiseval display a shallow dip eastwards.

(Figure 26A) Serpentinised olivine, laths of zoned labradorite (2 mm long), intergranular clinopyroxene and irregular grains of magnetite and ilmenite from an inclined sheet below Mullach Bi. Cockburn Collection K402. Cross polarised light.

(Figure 26B) Partially serpentinised olivine phenocrysts (FO₇₅) with purple-brown titanaugite, magnetite and labradorite (An₆₆). Inclined sheet SE of Bioda Mor, Dun. The largest olivine is 0.35 mm across. HMTS 20532; plane polarised light.

(Figure 26C) Large twinned plagioclase grain with a core composition An₆₁, a zone of pyroxene granules, and a rim of An₆₇ lies in a groundmass of intergranular plagioclase (An₆₆), clinopyroxene and opaque minerals. Porphyritic basalt from An Fhaing, Dun; HMTS 20535; cross polarised light.

(Figure 27B) Quartz xenocryst, 0.75 mm across, rimmed with pyroxene, lies next to an amygdale filled with chlorite and clay minerals in a groundmass plagioclase laths with granular pyroxene and opaques. Cockburn Collection K411; cross polarised light.

(Figure 27C) Altered feldspar and corroded β-quartz grains form some of the nuclei for spherules in the centre of a composite dyke at Mol Ghiasgar. (S72419), field 4 mm wide, cross polarised light.

(Figure 27A) Glomeroporphyritic cluster of pyroxene (Wo₄₀En₄₂Fs₁₈) and plagioclase (An₈₀ in basalt sheet from Glen Bay. Compositional zoning is visible in the largest pyroxene which is 1 mm across. HMTS 20503; cross polarised light.

(Figure 27D) Trachytoid basalt lying between spherulitic felsite (Figure 27C) and marginal dolerite of composite dyke. Patches of albite-epidote-calcite are present in the dolerite. (S72420), field 1 mm wide, cross polarised light.

(Figure 28)TiO_2–K₂O–P₂O₅ diagram showing range in K₂O values at relatively constant Ti:P ratios. Broken line denotes closed system fractionation at Skaergaard (Thompson, 1982a). Solid circles are basalts and basaltic andesites; open circles are microgranite and porphyritic pitchstone.

(Figure 29) Normative nepheline-olivine-diopside-hypersthene-quartz diagram with values for St Kilda rocks (circles), non-porphyritic central magmas in other Tertiary centres (triangles), Blackstones rocks (stars), and the indicated fields for Skye Main Lava Series (SMLS), Mull Plateau Group (MPG) and Preshal Mhor Basalts (PMB). Fractionation trends at low and high pressure are shown (based on Thompson, 1982a).

(Figure 30A) Cliffs of Quaternary drift on bedrock below St. Brianan's Church. Up to 20 m of massive Ruaival till are overlain by 2–3 m of stratified slope deposit (Ruaival head).

(Figure 30B) The base of the Ruaival head in A. Ruaival, showing alignment of platey clasts and containing pockets of organic sand with embedded blocks.

(Figure 30C) The Village Bay till, just west of the pier: an unstratified, locally derived deposit of blocks and cobbles in a gravel and sand matrix.

(Figure 31A) Water collecting on a terrace formed by a periglacial sheet flow, west side of Gleann Mor. On the opposite side of the glen, on the same contour, a spring breaks by the ruined building.

(Figure 31B) Protalus rampart below Glacan Chonachair. Scree derived from periglacial weathering of the crags above has accumulated in front of a former residual ice patch.

(Figure 31C) Cam Mor blockfield. Periglacial frost-shattering of the Western Gabbro on Mullach Bi has resulted in scree deposits of massive gabbro blocks, up to 7 m across.

(Figure 32) Sampling localities. A An Lag [NM 1037 9953](270'); B Abhainn Mhor [NM 0963 9970](470'); C Abhainn Mhor [NM 0975 9946] (220'); D Abhainn a Ghlinne Mhoir [NM 0878 9962] (385'); E Abhainn a Ghlinne Mhoir [NM 0860 0027] (100'); F Abhainn Ruaival [NM 0974 9836] (130').

(Figure 33A) A Grains recovered from An Lag and Abhainn Mhor include (left to right): top line - 2 andradites, a euhedral grossular and 2 almandine garnets; middle row - alumino-silicate (probably andalusite), sapphire, 2 pale green diopsides, and 2 euhedral clear zircons; bottom row - 3 anatase crystals, 2 euhedral magnetites and 3 dark zircons. The grossular is 0.5 mm across.

(Figure 33B) Pale blue fragment of sapphire 0.3 mm long showing intersection of rhombohedral with basal parting, and pale green diopside.

(Figure 34) Seabed contours round St Kilda and positions of the profiles.

(Figure 35) A. Profile UV from St Brianan's east–south–east across and outside Village Bay. B. Profile WX from Dun to Levenish (2.3 km).C. Profile YZ in a SW direction over the submarine ridge between Dun and Levenish.

(Figure 36) Major faults affecting the St Kilda igneous complex.

(Figure 37A) Kink in the Abhainn a Ghlinne Mhoir where the stream follows a NW-trending fault

(Figure 37B) Fractured rocks of the Mullach Sgar Complex in the central part of the quarry below Mullach Geal.

(Figure 38) Simplified geological map of St Kilda showing the palaeomagnetic sampling sites

(Figure 39A) The cleaned site-mean directions of magnetisation. Equal area projection; open circles indicate upper hemisphere (negative inclinations).

(Figure 39B) The St Kilda palaeomagnetic pole with its oval of 95% confidence and the apparent polar wander path for the last 90 million years (Irving, 1977). The arcs of 95% confidence are part drawn around the individual poles; polar equal area projection.

(Figure 40) Diagram showing Rb and Sr ratios of St Kilda intrusions.

(Figure 42) Iron-magnesium and alkali-calcium ratios of the major St Kilda intrusions.

(Figure 43A) Chondrite normalised (Haskin and others, 1968, average of 9 chondrites) REE abundances versus atomic number for St Kilda rocks. 1 Glen Bay Granite, RR239. 2 Conachair Granite, RR322. 3 Dotted lines enclose an area representing dolerites, diorites and microdiorites of Phases 2, 3 and 4 in Mullach Sgar Complex, RR181C, 184, 185, 298A, 301A. 4 Glen Bay Gabbro, H7287. 5 Basaltic dolerite, Glacan Mor, RR311B. 6 Banded gabbro, Boreray, RR369B. 7 Cambir Dolerite, RR265. 8 Western Gabbro Type 2, H7640. 9 Western Gabbro Type 1, RR341B. Analyst: Dr S. J. Parry.

(Figure 43B) Acicular red brown chevkinite 0.5 mm long in quartz and turbid alkali feldspar lies adjacent to amphibole and opaque grains. Glen Bay Granite, (S67633), plane polarised light.

Tables

(Table 29) Analyses of inclined sheets and dykes from Hirta, and Dun, St Kilda.

(Table 38) Palaeomagnetic results from St Kilda.

(Table 41) Rb–Sr data for intrusive rocks on St Kilda.

Maps (unnumbered in the original)

Map 1 Western Gabbro.

Map 2 Cambir Dolerite.

Map 3 Breccia of Gabbros and Dolerites.

Map 4 Glen Bay Gabbro.

Map 5 Glen Bay Granite.

Map 6 Mullach Sgar Complex.

Map 7 Conachair Granite.

Map 8 St. Kilda 1:25,000 Solid Edition Special sheet. 1984.

Tables

(Table 29) Analyses of inclined sheets and dykes from Hirta, and Dun, St Kilda

Post magmatic alteration of the minor intrusions is indicated by high Fe₂O₃:FeO ratios. This is consistent with the presence of abundant hydrous minerals, zeolites, calcite and epidote in some rocks and suggest that there has been considerable local hydrothermal alteration.

(Table 38) Palaeomagnetic results from St Kilda

(Table 41) Rb-Sr data for intrusive rocks on St Kilda