3.1. Palaeosurfaces and unconformities

A complex unconformity surface separates the TF and units of the underlying PBCF from elements of the earlier lava field of west-central Skye (Figs 3, 4). The stratigraphically highest units below the PBCF (Fig. 3) preserve complex interflow relationships, including local valley-fills, pinch-swell features indicative of flow channelling around topographical obstacles and the marginal overtopping of a localised palaeo-topography, flow terminations and wider-scale onlap and downlap relationships. Thin interflow volcaniclastic units and lateritic palaeosols are commonplace. Locally, the base of the TF directly overlies the Gleann Oraid Formation, but at most locations the PCBF is sandwiched between them and in part represents the partial fill of a pre-existing fault-controlled topographic low, possibly a shallow graben (cf. Holm & Cloud Reference Holm and Cloud1990). This complex sequence of volcanic and sedimentary rocks is now preserved within a narrow, fault-bounded, north-south oriented valley, the Talisker Graben (Figs 2, 4).

3.1.3. North face of Preshal Beg

Along the entire northern face of Preshal Beg the irregular base of the TF overlies a thick development of the PBCF (Figs 2, 4, 7a). Overall, the unconformity surface is inclined gently to the NW, although locally the base of the lava appears centroclinal. The base of the PBCF is nowhere exposed, but may be closely approximated in the field as a distinctive grassy slope similar to that noted along the south-eastern side of Preshal More; craggy exposures of trachyte lava confirm the presence of the Sleadale Member directly below and extending across the entire width of the graben (ca. 1 km at this point). The Sleadale Member here comprises a locally developed basal trachytic tuff and a single lava (or flow field) of pale-weathering, highly vesiculated, porphyritic trachyte with a maximum thickness of c. 55 m. To the north, in Sleadale, stratigraphically lower lavas within the Arnaval Member are predominantly hawaiites and thin basalts. Regional mapping of the lava field (Williamson Reference Williamson1979; Williamson & Bell Reference Williamson and Bell1994; BGS 2000) suggests that this development of the Arnaval Member may interdigitate with part of the upper (Skridan) Member of the Glen Caladale Formation (Fig. 4).

Figure 7 (a) Main exposure of facies PBCF-1 on the north side of Preshal Beg at c.[NG 3313 2792], comprising stacked lobate masses of boulder conglomerate, overlain by columnar-jointed Talisker Formation (TF) lava, with approximate locations of (b) (c). (b) Detail illustrating poorly-sorted character of PBCF-1 in exposure at c.[NG 3313 2792]; arrows indicate orientation of stratification, pole c.1 m long. (c) Detail illustrating stratification of PBCF-1 in exposure at c.[NG 3313 2792], arrows indicate orientation of stratification, pole c.1 m long. (d) Typical exposure of facies PBCF-2, comprising moderately well-sorted, bedded volcaniclastic sandstones, pebbly sandstones and pebble conglomerates, on the SE side of the small satellite hill SE of Preshal Beg at c.[NG 3320 2763], arrows indicate orientation of stratification, hammer shaft c.60 cm long; (e) Exposure of facies PBCF-3, comprising medium- to coarse-grained poorly-sorted breccias (B), invaded by sheets of lava (L), brecciated basalt (hyaloclastite, H) with abundant hydrothermal veins and cavity fill, and associated basalt pillows (P), on the north face of Preshal More at c.[NG 3346 3003], ruler is 30 cm long.

4.1. Boulder conglomerate (facies PBCF-1)

This facies is the most common and reaches its maximum development of between 20 m and 40 m on the north face of Preshal Beg (Figs 5, 7a) (Williamson & Bell Reference Williamson and Bell1994, fig. 25); minor exposures, slope morphology and a distinctive vegetational change indicate its presence on the western- and southern-facing slopes of the hill. It is also present in lenticular exposures beneath the TF along parts of the northern flank of Preshal More. PBCF-1 comprises a thick but internally variable sequence of massive, immature, poorly- to very poorly-sorted, mostly clast supported, pebble- to boulder-grade conglomerates with a variable matrix of dark lithic-volcanic sand, silt and clay, and some rare interbedded thin, commonly pebbly, sandstones. It is almost wholly volcaniclastic, with little siliciclastic content (rare quartz, alkali feldspar).

These conglomerates form an overlapping series of stacked, lenticular bedforms, individually up to 10 m thick but generally less than 4 m thick. Clasts within the conglomerates range from less than 1 cm across to tabular blocks over 2×0.45 m, are sub-rounded to angular (exfoliation gives a false impression of roundness and sphericity) and appear to be derived exclusively from locally available volcanic rocks comprising various basalts, hawaiites and mugearites and their scoriaceous and amygdaloidal variants. There are also rare clasts of basaltic tuff, reddened pebble conglomerate and indurated laterite. Clasts of trachyte are absent from PBCF-1 facies exposures at Preshal Beg, despite the Sleadale Member being the immediately underlying unit (Fig. 2); however, lower PBCF beds and the contact between these and the trachyte are not seen and so their total absence from the Formation must remain speculative. A few small angular clasts of trachyte were however located in the PBCF at Preshal More. Also absent from the clast assemblage in this facies and, indeed, throughout the rest of the Formation, are clasts of epiclastic origin i.e. the pre-Palaeogene rocks and coarse-grained intrusive suites (cf. the Minginish Conglomerate Formation in the older part of the lava field to the south: Williamson & Bell Reference Williamson and Bell1994). Sedimentary structures are not common and the facies is rather more typified by its chaotic, unpredictable variations both in vertical and lateral sections. A weak sub-horizontal to moderately-inclined fabric may be present in some places and, whereas there are hints of clast imbrication locally, these are ambiguous and contradictory. Locally, some of the larger conglomerate lenticular bedforms appear to be crudely, but irregularly, massively- to thickly-bedded. Individual smaller bedforms may be planar or lenticular. Most appear wholly unsorted, but a few display either a crude reverse- or, more commonly, a normally-graded profile. Thin, irregular, wavy, highly-inclined (>60°) rootless zones of relatively finer-grained conglomerate (with a sand grade matrix) appear to locally cut through one or more of these lenticular bedforms. Some zones preserve a crude vertical, clast- and grain-size foliation with an irregular central sand-rich portion containing clay-filled vesicle-like structures. Little is seen of facies PBCF-1 on the satellite outlier immediately east of the main hill of Preshal Beg and detailed mapping here suggests that it may be much reduced in thickness on the northern side of this minor hill and possibly entirely absent from the southern side, where only facies PBCF-2 appears sandwiched between the TF and the massive hawaiites of the Arnaval Formation (Figs 2, 4).

Discontinuous exposures of lithofacies PBCF-1 occur along the north-facing slopes of Preshal More and are best developed towards the east (at Grid Reference [NG 33425 30062]; Figs 2, 4). The sequence, locally up to 6 m thick, comprises two distinct beds with horizontal bounding surfaces. The lower bed, 0.5–1.5 m thick, is of poorly-sorted breccia and conglomerate, with abundant rounded pebble and cobble grade clasts. It is overlain by 3.5–4 m of upward-fining boulder conglomerate, with clasts up to c. 0.5×0.25 m. Intimately associated with these conglomerates, mostly as small sedimentary drapes, are thin (2–3 cm thick) beds of lithofacies PBCF-2. Another characteristic of the facies is the presence, mostly between bedforms, of numerous cuspate slip-surfaces and shear-planes, many at a high angle (>45°) and with slickensides; these may be accompanied by thin developments of breccia attributed to PBCF-3.

4.1.1. Interpretation

This lithofacies has all the hallmarks of coarse-grained, perhaps concentrated, debris flows. These may have been generated from either a series of proximal alluvial fans or lahars.

Whilst there are some rare and highly localised and minor interbedded facies as described above, which perhaps indicate pulsed or intermittent deposition, these coarse and chaotic conglomerates were most likely emplaced extremely rapidly as remobilised, unconsolidated volcanic material, in-part probably fluvially derived (?flash-flooding), within a confined and certainly restricted depression, perhaps fault-bounded river valley or gorge on the lava field (cf. Holm & Cloud Reference Holm and Cloud1990). Whilst the majority of the clasts are sub- to reasonably well-rounded, this need not indicate a fluvial origin as basaltic rocks are prone to exfoliation and this would be accentuated especially under the type of climatic conditions that are thought to have prevailed during the Palaeocene. Though we cannot entirely rule out an element of fluvial deposition, any primary fluvial input was probably small. The thoroughly chaotic nature, lack of any major sorting and the apparent lack of clasts of the immediately preceding Sleadale trachyte lava, coupled with the rare presence of highly inclined diffuse zones of finer-grained sediment that may be regarded as fluid escape structures, suggest that the debris flows abruptly deposited their sediment load either as they met obstructions on the floor of the valley, or as gradients decreased and the channel flattened out. The dominance of basaltic clasts over more evolved rock-types, and especially the near absence of trachytic clasts both within the facies and in the sub-formational regolith, despite this being the immediately subjacent lava lithology, suggest that the bulk of the facies was actually sourced from parts of the lava field, most likely outside, but possibly also marginally flanking, the immediate Talisker area. The nature and architectures of both the PBCF and the TF imply an overall source area for the former, mainly to the south to SE of the present outcrops.

The presence of the slip surfaces is taken as evidence for post-depositional adjustment and limited mass movement, although some may have been induced by loading during emplacement of subsequent deposits or the overlying TF lava. In addition to alluvial fan debris flow deposits, we suggest that some of these physical and sedimentological features also point to possible similarities with the channel facies and run-out of massive, cohesive, clay-poor volcanic debris avalanche processes (lahars) and channel-fill deposition from the debris flow phases associated with them (e.g. Smith Reference Smith1986; Arguden & Rodolfo Reference Arguden and Rodolfo1990; Glicken Reference Glicken, Smith and Fisher1991; Rodolfo & Arguden Reference Rodolfo, Arguden, Fisher and Smith1991; Smith & Lowe Reference Smith, Lowe, Fisher and Smith1991; Capra & Macias Reference Capra and Macias2000; Vallance Reference Vallance and Sigurdsson2000). Cohesive debris flows commonly ‘freeze’ where palaeoslopes drop to below 0.5–1.0° (Capra & Macias Reference Capra and Macias2000). These mass flow deposits at Talisker, although requiring only low-angled palaeoslopes in order to travel many kilometres from source, could have originated as slope failures on the flanks of a volcanic edifice, possibly an early Cuillin Volcano (see below). However, Reubi et al. (Reference Reubi, Ross and White2005), in documenting a series of larger-scale debris avalanche deposits that formed during uplift associated with large igneous province magmatism, also suggest that avalanches can be triggered by dyke intrusion without a major volcanic edifice being produced. For PBCF-1, the initiating mechanism might therefore have been surface volcanic, shallow intrusive or seismic activity, but may also conceivably have been due to, and accentuated by, periodically exceptional rainfall producing flash flooding from upland areas of deep weathering and wasting (see also Brown et al. Reference Brown, Halohan and Bell2009).

The architectures and topographical dispositions of the PBCF-1 deposits are complex. Whereas it is clear that some units, especially those seen associated with the small outlier east of Preshal Beg, are older than the main development of PBCF-2 facies (see below), there is a possibility that others are, in fact, some of the youngest sedimentary events. This could be the case for those PBCF-1 units observed at the western end of Preshal Beg (Fig. 4). Here, not only are they the lowest topographically, but they are overlain by a thin development of a pillow-breccia facies at the base of the TF (Fig. 8a). These PBCF-1 conglomerates may have been deposited within a fresh erosional channel incised within pre-existing deposits immediately prior to emplacement of the TF lava.

Figure 8 (a) Basal facies (TF-1) of the Preshal Beg lava outcrop, comprising a thin interval of pillow lava, overlying facies PBCF-1 conglomerates and breccias, giving way, upward, to the regular columnar facies (TF-2), on the west side of Preshal Beg at c.[NG 3261 2797]. (b) Sill-like protrusion of basalt (TF-2) into underlying facies PBCF-1 conglomerates on the west side of Preshal Beg at c.[NG 3260 2799].

4.2. Bedded sandstone and conglomerate (facies PBCF-2)

Facies PBCF-2 is best developed and reaches its maximum directly observed thickness of c. 15 m around the small satellite outlier of lava east of the main mass of Preshal Beg (Figs 2, 4, 7d). The true maximum thickness, based upon detailed mapping and a consideration of height differences along the south of the main hill, may be considerably more. Facies PBCF-2 appears to be topographically the highest, and consequently most likely overlies, or is at least younger than, the main development of lithofacies PBCF-1. Elsewhere on Preshal Beg, at lower topographical levels, it is absent and the TF lava directly overlies facies PBCF-1. Along the north face of Preshal More there are a few thin, discontinuous beds of probable facies PBCF-2.

Lithologically, as seen on Preshal Beg, the facies is wholly volcaniclastic and comprises a thin- to medium-bedded sequence of moderately well-sorted sandstones, pebbly sandstones and pebble conglomerates (Figs 5, 7b). Clasts range from less than 1 cm, to over 20 cm, but most fall within the range 2–10 cm. Clasts and matrix alike are locally derived lava types, many of them reddened due to subaerial exposure prior to burial, and laterites; most are sub-rounded to rounded. Unlike the drab, dark grey coloration typifying lithofacies PBCF-1, these rocks are usually pale-weathering in hues of pink and pinkish-grey. Stratification is clear but imperfect; most beds are flat-lying and many pass laterally into either finer- or coarser-grained units. Some of the beds, however, exhibit a small degree of normal (fining-upwards) grading and are crudely cross-stratified in broad trough-shaped bedforms. Less commonly, other thin beds, especially those towards the top of the deposit, exhibit minor coarsening-upward motifs and broad, ripple-drift, cross laminations. There is an overall post-depositional reddening motif to PBCF-2, most prevalent towards the upper parts of the deposits.

There are minor developments of other lithofacies intimately associated with PBCF-2. Fine-grained breccia, sandstone and siltstone locally form thin (a few centimetres) lenticular beds of no great lateral extent within the main body of PBCF-2. The finer-grained lithologies are coarsely- to finely-laminated, and, in places, very irregular and discontinuous ‘drifts’ of granule to sand grade material are present. Colours vary from pinkish-grey and brown through to dark grey and some thin beds may contain finely comminuted carbonaceous fragments. Also present are minor beds of reddened siltstone and mudstone and associated reddened surfaces. Some beds are considerably indurated, with a localised development of secondary carbonate and iron-oxide cements. Others have irregular, possibly erosional, undulating upper surfaces with, in places, thin drapes of intensely reddish brown silty mudstone and clayey siltstone. These drapes are generally only a few millimetres thick and laterally impersistent. Similar undulose surfaces and some accompanying reddened mudstones are locally, but rarely, present between some beds within facies PBCF-1, also on Preshal Beg. It is these relatively clay-rich beds that commonly display evidence of post-depositional instability and slippage.

4.3. Medium- and coarse-grained angular breccia (facies PBCF-3)

Medium- and coarse-grained breccias comprising basalt and hawaiite clasts are best displayed along the NE flank of Preshal More (Figs 4, 7e). Here, the main development has a distinct wedge-shaped geometry, thickest c. 25 m east of the Talisker Fault (East) (Figs 2, 4), and thinning out to the west; there are no exposures on the southeast flank of the hill, the contact zone being obscured by thick talus, soil and vegetation. Its thickness tends to be somewhat exaggerated as it is intruded/invaded by irregular bodies of both basal massive and partially brecciated facies of the overlying TF lava. It appears to be entirely unbedded, clast-supported and unsorted, with clasts mainly, but not exclusively, of large pebble and cobble grade. In comparison with similar clast lithologies in facies PBCF-1, the clasts are angular to sub-angular and commonly surprisingly fresh, with few signs of either weathering or having been transported far from their source. Locally within this unit, and increasingly more common towards the base, especially at its western termination, there are minor, thin lenticular bodies of conglomerate similar to facies PBCF-1. Some crudely stratified breccias that may simply be locally reworked facies PCBF-3 breccias, occur as lenticular bodies along the base of the TF to the NW. At the NW nose of Preshal Beg (Fig. 4), another thin, but highly localised unit of a similar but finer-grained breccia facies overlies facies PBCF-1 conglomerates. It, too, has been intruded by apophyses of the overlying lava.

Along the northern flank of Preshal More immediately below the base of the TF lava and always intimately associated with facies PBCF-2 and PBCF-3, there are intermittent lenticular bodies, up to 0.75 m thick, of massive medium-grained breccia. These are mainly composed of poorly-sorted sub-angular to sub-rounded clasts of partially decomposed massive aphyric and amygdaloidal basalt, hawaiite and laterite, although there are also uncommon angular clasts of chilled vesicular rock showing only minor zeolitisation. This mixed clast assemblage sets them apart from the monolithic basal flow facies breccias of the TF. Some units are locally mineralised (principally by calcite-chabazite assemblages (Williamson Reference Williamson1979) and some minor sulfide mineralisation) and invaded by chilled tongues of the basal breccia facies of the overlying lava.

5.1. Architecture and internal structure

Some of the most compelling field-based evidence pointing to an intracanyon-style emplacement mechanism for the TF lava stems from our detailed study of flow morphology. The thickness, internal structure and overall architecture of the lava contrasts markedly with the more regular, relatively thin and extensive sheet-forms of the lavas forming the bulk of the underlying Gleann Oraid and Glen Caladale formations flooring the Talisker Graben and the surrounding terrain (Fig. 4).

Other than its unusually great thickness, perhaps the most prominent feature of the TF lava is that both outliers preserve spectacularly well-developed suites of columnar joints, comparable in many respects to the grandeur of other better known and classical examples from the NAIP, e.g. Giant's Causeway, Co. Antrim and Fingal's Cave on the island of Staffa, Mull (Williamson & Bell Reference Williamson and Bell2012; see also below). The original top of the TF lava is not, however, preserved, having been removed through erosion. The thickest development of the formation is on Preshal More, where c. 120 m is preserved. The Preshal Beg outcrop is a little thinner, c. 100 m, but its topmost exposures may preserve features that originated closer to the original flow top than those on Preshal More. Such thick lavas are relatively uncommon within the NAIP lava fields, suggesting something special about their environment and mode of emplacement. Other than where more differentiated or evolved flows have formed as lava domes, thicknesses of this magnitude may be attributed most readily to either ponding effects and/or to volumetrically huge eruptions and high effusion rates.

Gross morphology, in particular the overall joint patterns that have developed in both the TF outcrops, make it possible to subdivide the lava into a number of separate structural ‘facies’ (Fig. 6). From base to top these are: (i) basal facies (TF-1); (ii) regular columnar facies (colonnade) (TF-2); (iii) platy facies (TF-3); (iv) irregular columnar facies (entablature) (which may be multiple) (TF-4); and (v) intrusive/sheeted facies (TF-5).

The architecture of these ‘joint facies’ is such that the TF lava would be classified as a Type III flow in the scheme developed by Long & Wood (Reference Long and Wood1986, Reference Long and Wood1987) for characterising lavas within the CRBG. This flow type is typically 30–80 m thick and has a single colonnade-entablature pair with a sharp mutual contact, and with no evidence for the previous existence of an upper colonnade.

The five ‘facies’ of the TF lava are considered in turn, starting with the lowest (basal facies).

5.2. Basal facies (TF-1) and relationships to the pre-Talisker Formation units

The TF lava has complex relationships with earlier formations. Schematic cross-sections illustrating the nature and inter-relationships of the various facies of both outcrops and the subjacent Preshal Beg Conglomerate Formation (PBCF) and Gleann Oraid Formation are presented in Figures 3 and 6. The landscape and environment at the time of the eruption of the TF flow field is summarised in succeeding sections.

TF-1 forms the lowest 0.5–5 m of both outcrops and exists in two, locally intermingled, forms: an irregular body of breccia, composed wholly of volcanic material and commonly associated with fractured, but otherwise intact, globe-shaped pillow-like masses of basalt; or as massive lava up to 1.5 m thick, with crude, irregular joint patterns and a chilled lobate base (Fig. 8). In both instances, the lava is a fine-grained to glassy (hypohyaline) melanocratic rock, with only rare small and widely distributed amygdales. This facies is particularly well-developed on the west and south-west sides of Preshal Beg, where a thin development (c. 1.75 m) of breccia overlies the PBCF (lithofacies PBCF-1; Fig. 8a). These breccias comprise clasts of basalt up to small pebble size and discrete, semi-intact but shattered, lobate-margined and globe-like masses of basalt in a very fine-grained to glassy and rarely fragmental matrix. The junction with the underlying formation is irregular and generally sharp. There is a sharp planar contact between breccia and the overlying flow-laminated, massive, chilled lava (see also below).

Locally, along the base of the lava, for example on the north side of Preshal More, the basal facies comprises either very fine-grained to glassy lobate (some pillow-like, others more akin to pāhoehoe lobes) coherent masses of basalt with crude radial joint patterns, or a breccia of angular fragments of basalt cemented in a zeolite and carbonate matrix with rare pillow-like breccia masses. This is best developed within the NE quadrant of the Preshal More outcrop, where the basal facies appears to have auto-brecciated (Fig. 7e). This is thought to have occurred during the flow's overriding of, and shallow intrusion into, wet (connate-water rich, or shallow water bodies) poorly- or un-lithified sediments of the subjacent PBCF (lithofacies PBCF-3). Irregularly lobate and locally megapillow-like bodies, or large-scale pāhoehoe-like tongues up to several metres across, are common within parts of the PBCF-3 outcrop wedge. Similar and related intrusive phenomena occur at a number of localities on both outliers, with the flow base(es) intrusive into the PBCF in the form of sill-like apophyses, with obvious chilled and locally brecciated margins (Fig. 8b). Similar phenomena have been described by Carr & Jones (Reference Carr and Jones2001) within the basal facies of Permian lavas from the Sydney Basin (Australia); this supports the contention that the TF lava locally intruded into semi-consolidated breccias and likely contemporaneous pyroclastic deposits. In places, the matrix of these breccias may also comprise fine, laminated, pale-coloured (bleached) ‘basaltic sand’ against which the clasts, many of them ragged in form, have quenched. This is highly suggestive of developing peperitic textures. Here, again, the contact relationships indicate that the various deposits of the PBCF were essentially unlithified and still wet when the TF lava was erupted.

Also present within the more massive variant of the basal lava facies is a well-developed, fine, sub-horizontal flow lamination, as, for example, at SW Preshal Beg. This part of the lava displays prominent flow banding in its lowermost 0.5 m and incorporates rare small xenoliths of amygdaloidal basalt. The banding is inclined, at angles of up to 65°, and the attitude of small folds and cross-lamination structures within this zone suggests a localised basal flow direction towards WNW. On the western nose of Preshal More, where massive columnar-jointed lava directly overlies eroded hawaiites of the Gleann Oraid Formation, a similar banding is present within TF-1. These features reflect laminar flow planes, grain-size and modal variations (Williamson Reference Williamson1979) and density contrasts developed at the base of the flow; some may be small pāhoehoe lobes. There is no accompanying platy jointing developed along the base of either outcrop, as for example exemplified by the basal zone of thick CRBG flows described by Self et al. (Reference Self, Thordarson, Keszthelyi, Mahoney and Coffin1997) and Thordarson & Self (Reference Thordarson and Self1998).

Evidence of topographic relief (palaeo-morphology) on the already sloping Palaeocene valley floors, by up to 20 m, but more commonly in step-wise increments of 5 m or less, is indicated by abrupt changes in the structural height of the bases of the two outcrops. The overall pre-TF basal surface under both outliers is inclined to the NW, but there is evidence from both outliers that the lava was, in places, restricted by vertical walls along the northern flanks of both outcrops. Where such ramp-like irregularities in the flow base are present, there are detached masses of basalt, in the form of pillow- and tube-like structures, which developed within intensely brecciated rock composed of lava and PBCF lithologies.

The invasive nature of low viscosity basaltic flows as they encounter unconsolidated and often water-saturated (‘wet’) sediments, is a major feature described from several flows of the CRBG (Mackin Reference Mackin1961; Byerly & Swanson Reference Byerly and Swanson1978, Reference Byerly and Swanson1987; Wells & Niem Reference Wells and Niem1987; Ross Reference Ross, Reidel and Hooper1989). There, not only are there localised intrusive phenomena similar to, but also an order of magnitude greater than, that seen in association with the TF lava, but many of the associated, but now physically detached sills, are located many kilometres ‘down valley’ from the terminations of the subaerial flow-facies themselves. Invasive facies lavas with peperitic-textured margins have also been reported from the Karoo flood basalt province in South Africa (Rawlings et al. Reference Rawlings, Watkeys and Sweeney1999).

5.3. Regular Columnar facies (TF-2)

This facies is represented by the lower, c. 35–50 m-thick, well-jointed portion of both outcrops. It is essentially, compact, fine-grained, non-amygdaloidal rock, with spectacularly well developed, regular, upright, and most typically hexagonal, columns (Figs 6, 9, 10). It represents between a half and a third of the preserved flow thickness and constitutes the so-called lower colonnade of the threefold division outlined by Tomkieff (Reference Tomkieff1940) and Spry (Reference Spry1962). Its proportion relative to the overlying irregular facies (entablature) may be significantly reduced locally, as on the NE flank of Preshal More.

Figure 9 (a) Preshal More, from A' Chailleach on the north side of Gleann Oraid, looking towards the south. Immediately below the crags of lava that form Preshal More (base indicated by dotted line) is the Preshal Beg Conglomerate Formation (facies PBCF-3) (mainly in shadow, arrowed). The terraces on the hillside below (north of) Preshal More are formed of intercalated flows of hawaiite (H) and mugearite (M) of the Arnaval Member of the Gleann Oraid Formation. The Talisker East Fault (arrowed) juxtaposes the Preshal More outcrop of the TF lava and PBCF-3 with lavas of the Arnaval Member on the eastern flank of Preshal More. Height difference between base of flow and summit is c.120m. (b) Preshal Beg, looking towards the NE. The PBCF is poorly exposed on this side of the hill, other than where indicated (arrowed: facies PBCF-1). Localised pillow-dominated facies (TF-1, not indicated) give way upwards to the well-developed regular columns (colonnade) of TF-2, in turn overlain (with a sharp interface) by the irregular columns (entablature) of TF-4, capped by the intrusive/sheeted facies (TF-5). Boundaries on the SE side of Preshal Beg indicated by dotted lines. Height difference between base of lava and summit is c.100 m.

Figure 10 (a) The sharp boundary (dotted line) between the regular columns (colonnade) of TF-2 and the irregular columns (entablature) of TF-4 on the central part of the south side of Preshal More. A thin layer of platy facies, TF-3, intervenes and is typically <1 m thick. (b) Exposure of platy facies (TF-3) (arrowed) at c. [NG 3302 2778] on the east side of the main outcrop of Preshal Beg. Pole is c.1 m long.

The individual columns are typically of constant width (30 cm to 1 m) throughout their height, although a thin upper tier or subzone of relatively (slightly) narrower columns is locally present on Preshal More. Generally, the columns are upright and close to vertical in aspect, but locally their lowermost portions (between 1 m and 4 m) are significantly curved (outwards) and may even be abruptly upturned, with the basal polygonal surface inwardly inclined at up to 50° at the normal flow base. At scattered localities on both outliers, the long axes of the columns have much lower ‘splayed-outward’ inclinations and may, in a few instances, even approach being horizontal. In cross-section, the columns present a variety of polygons. Hexagons are the most common, although both 5- and 4-sided columns are also noted. No statistical analysis has been carried out. Within the main body of this facies, the rock may have a crude fissile appearance and there is commonly a diffuse grain-size and textural banding present (see also sections below), varying in thickness from the scale of a few millimetres to centimetres. This appears to coincide with the presence of orthogonal joint-sets perpendicular to, and dividing the columns along, their length. These so-called chisel structures (Iddings Reference Iddings1886; James Reference James1920), striae (Ryan & Sammis Reference Ryan and Sammis1978) or plumose structures (DeGraff & Aydin Reference DeGraff and Aydin1987) are typically spaced along the columns several to tens of centimetres apart.

5.4. Platy facies (TF-3)

Along the interface between the two major columnar facies (TF-2 colonnade and TF-4 entablature) on Preshal More, there is commonly a zone of relatively fine-grained, platy or foliated lava c.70–80 cm thick (Fig. 10). There is usually, but not invariably, some continuity of vertical joints from below, and this zone may therefore be considered as an upper, but only locally developed, sub-facies of the regular columnar facies TF-2. The sub-horizontal joints appear to be cut by the later vertical joints and it is possible that there is a sequence of joint formation of entablature-platy-colonnade as described from the Roza Member (Wanapum Formation, CRBG) by Thordarson & Self (Reference Thordarson and Self1998). The zone has a discontinuous, sheet-like, but overall broadly lenticular, geometry, and is characterised by a series of near-horizontal but undulating, narrow-spaced joints that effectively split it into slabs of varying thickness and extent (Fig. 10b). Thicknesses vary from less than 1 cm to over 15 cm and, in profile, vary from a few centimetres to more than a metre in length. These structures are oriented perpendicular to the more typical vertical columnar joints displayed by the rest of the underlying regular columnar facies (TF-2). It is considerably finer-grained than, and typically gradational into, the underlying regular columnar facies but, in a few places, this foliated basalt appears to have recognisable and moderately finer-grained margins, possibly suggesting chilling or an intrusive relationship with the rest of the flow.

On Preshal More, there is a displacement of this facies, in places by more than 10 m, by a number of post-emplacement normal faults (Fig. 2). The facies is only localised, and poorly developed on most of Preshal Beg (Fig. 10b); locally, in its absence, the contact zone has been exploited by thin intrusive sheets of fine-grained, massive basalt.

5.5. Irregular columnar facies (TF-4)

This facies constitutes much, but not all (see following intrusive/sheeted facies), of the upper c. 80 m of the preserved outcrop on both hills (Figs 9, 10, 11); the true flow tops have been removed by erosion, but what remains constitutes a so-called ‘entablature’ and is between a half and two-thirds of total present-day lava thickness. As seen on the north face of Preshal More (Fig. 9a) and on parts of the SW face of Preshal Beg (Fig. 9b), the lower part of this facies, variable in thickness, is characterised by generally regular, upright but slightly wavy and smaller (generally <30 cm across) columns than those of the underlying regular columnar facies (TF-3) and is, as such, also a form of colonnade, as defined by Tomkeieff (1940) and Spry (Reference Spry1962), but would still constitute an entablature as defined by Long & Wood (Reference Long and Wood1986, Reference Long and Wood1987). Elsewhere, this sub-zone is not so evident and, indeed, is usually absent, and the TF lava does not, therefore, contain consistent multi-tiered joint facies.

Figure 11 (a) Section on south side of Preshal Beg showing distribution of facies TF-1–TF-5. (b) Irregular columns (entablature) of TF-4 on the SW side of Preshal Beg (view looking towards NE), giving way to the intrusive/sheeted facies, TF-5, which caps the hill and is interpreted as having formed by auto-intrusive processes (examples arrowed, see also Fig. 12) close to the original top of the flow. Outline rectangle indicates location of (c). See main text for detailed discussion. Rock face is c.80 m high. (c) Detail of (b), with the boundaries of two sheets defined by dotted lines.

Sub-horizontal joints and indistinct, curvilinear, variously-inclined surfaces that appear to bifurcate downwards from the top of the outcrops most likely represent original early-formed master joints. These major fractures in the body of the lava separate ill-defined ‘zones’ or ‘associations’ of minor irregular facies columns in the entablatures and the uppermost parts of the colonnades and are most prominent high on Preshal More. Upwards, on the NW flank of Preshal More, the dominant surfaces are more steeply inclined to the south, so that these zones are progressively tilted and the inclination of the wavy columns decreases. At the eastern margin of the outcrop, in contact with the palaeo-talus breccias (PBCF-3) of the PBCF at the margin of the conjectured palaeo-valley wall formed of lavas of the Glen Oraid Formation (Fig. 4), similar surfaces are observed and merge with those of the underlying regular columnar facies (TF-2). In this case, however, closer to the flow base, the reverse situation applies, with the columns tending towards a vertical orientation, upwards, in stepped increments, each separated by an irregular planar surface; surfaces dip at c. 45° into the lava at the base (i.e. the closest approach to the valley wall) but, upwards, progressively approach a sub-horizontal attitude. In both cases, the columnar joint sets defined by these major joint systems show no evidence, such as chilled or vesicular margins, which might suggest they were separately emplaced flow units. All of these low-angled ‘master joints’ are simply post-solidification relaxation joints.

The higher level parts of TF-4 are, however, typified by slightly more vesicular material and by columns and prismatic joints that are both irregular and commonly arranged in complex, interfering fan- and chevron-shapes, many with convex bases, and a few appear to be separate ovoid bodies with radial joint sets within more irregular masses, as on the western face of Preshal Beg (Figs 9b, 11). This portion of the lava would be designated as an entablature (Tomkieff Reference Tomkieff1940; Spry Reference Spry1962). The relative thicknesses of the more regular- and irregular-jointed portions of this facies in general are quite variable, but typically in the ratio 1:2.

5.6. Intrusive/sheeted facies (TF-5)

The uppermost exposures of the Preshal Beg outcrop are of particular interest and significance and comprise a series of well-developed low amplitude ‘synformal’ and ‘antiformal’ structures that are especially obvious when the hill is viewed from a distance (Figs 9b, 11, 12). When examined in cross-section, these structures are clearly seen to be the product of auto-intrusion, presumably near-surface, inflating the uppermost part of the lava. They were emplaced as a series of thin (<5 m) irregular sheets, tubes and pāhoehoe-style lava lobes, resulting in the development of complex cooling patterns (see below). In the near-vertical sections on the SW side of Preshal Beg, their irregular nature is evident (Fig. 11). Within the summit exposures (Fig. 12), most sheets exhibit single-contact, chilled margins commonly with glassy selvages, indicative of rapid quenching and of having been successively and repeatedly intruded via the same conduit, producing a form of sheeted complex. The antiformal sheets are mostly only gently arched, but there are some steeper examples, with some rare examples including axial splits or fissures. All of the intrusions possess irregular polygonal joint patterns due to volume reduction during their emplacement and crystallisation (Fig. 12). The polygon shapes are highly irregular and hexagonal forms, generally considered to result from uninterrupted cooling, are not common (cf. the regular columnar facies, TF-2). The polygonal fracture patterns on most exposures mainly comprise 4- and 5-sided forms, many with curved or sinuous margins. On a finer-scale, fracture edges are irregular, rough and hackly, unlike those of TF-2 and most of TF-4. Additionally, many surfaces show evidence of secondary fracturing caused by later glacial action. In the cliff sections, but below these sheet-like bodies, there are numerous pod-like bodies, also with chilled margins (Fig. 12), which commonly display crudely developed colonnade–entablature -like joint couplets; they appear to be intrusive into the more irregularly-jointed entablature. Similar shallow intrusive phenomena are common in the upper parts some of the thicker lavas of the Staffa Lava Formation of the Palaeogene Mull Lava Field (Williamson & Bell Reference Williamson and Bell2012).

Figure 12 Intrusive/sheeted facies, TF-5, at the western end of the summit of Preshal Beg at c.[NG 3275 2795], consisting of thin auto-intrusive curved sheets of coarse basalt/dolerite with prismatic/polygonal joint sets and poorly developed chilled margins. Location of one intrusion margin is arrowed. Pole is c. 1 m long.

10.1. Phase 1

Eruption and emplacement of the hawaiite- and mugearite-dominated Arnaval Member of the Gleann Oraid Formation (Figs 3, 4). Waning intensity and frequency of volcanic activity led to lengthy time intervals between the later eruptions and promoted the development of palaeosols, channelling of flows and some localised unconformities. Burial of these lavas to a substantial depth (possibly by as much as 1.5 km locally). The extent and thickness of the lava pile allowed the onset of depth-related (hydrothermal) metamorphism, secondary mineralisation and the establishment of zeolite zonation patterns.

10.2. Phase 2

A hiatus in volcanic activity then followed, during which time there was extensive faulting of the lava field. Following intense weathering, erosion was rapid and significant, exposing lower units of the Arnaval Member. Volcanism, including hydrothermal activity, and some updoming may have continued in the region to the SE (where the present-day Cuillin Hills are located) and close to the axis of the Skye Dyke Swarm. The last pre-TF eruptions in the Talisker area comprised the emplacement of the Sleadale Member (Gleann Oraid Formation) trachyte (crystal-rich lapilli tuff and lava domes) within the proto-Talisker Graben (at 59.81 Ma; Bell & Williamson Reference Bell, Williamson and Trewin2002).

10.3. Phase 3

Effusive volcanic activity in the Talisker area then effectively ceased, although hydrothermal and solfataric activity may have continued for some time. Thereafter, there was a slow but inexorable uplift in the vicinity of the developing Cuillin Volcano / Intrusive Centre. There was also perhaps still some additional or continued concomitant basinal subsidence to the north, in the Bracadale region – the structural ‘depocentre’ of the lava field (England Reference England1994). This resulted in a significant N- to NW-directed palaeo-slope. During this protracted, relatively (volcanically) stable period – the canyon-forming phase – the volcanic uplands may have been semi-arid but prone to torrential flash-flooding as rivers struggled to reach new base levels. Many environments suitable for the development of deep weathering profiles and palaeosols, deep-incision erosion and the establishment of a complex (possibly Cuillin-radial) palaeo-drainage system were created. Escarpments, valleys and drainage systems would also have been strongly influenced by fault patterns.

10.4. Phase 4

Erosion of the existing lava field and any unconsolidated pyroclastic rocks in the region of the Cuillin Volcano continued and the PBCF was emplaced. Initially there was gradual infilling of palaeo-channels and valleys by various sedimentary lithofacies representing talus, alluvial fan and minor stream environments. Although perhaps initially exacerbating slope erosion and deepening canyons, a series of debris flows, as they waned, deposited the heterogeneous conglomerates of facies PBCF-1 resulting in the partial blockage of local valley-drainage systems. Sediment transport may have been initiated by catastrophic failures on the deeply-weathered slopes of the growing Cuillin Volcano to the SE (cf. Brown et al. Reference Brown, Halohan and Bell2009). Increased hydrothermal and seismic activity ahead of the eruption of the TF flow field and heavy seasonal rainfall and flash-flooding are likely contributors. The composition of all the PBCF deposits indicates a purely local lava field origin; there appears not to have been a source of epiclastic material similar to that characterising the earlier Minginish Conglomerate Formation (Williamson & Bell Reference Williamson and Bell1994).

10.5. Phase 5

Local differential uplift, faulting, erosion and sedimentation continued in the Talisker area, with some weathering and minor incision of both the volcanic hinterland and the existing, largely unconsolidated debris-flow deposits of the PBCF by ephemeral streams. The finer-grained and more structured lithofacies of PBCF-2 were deposited under much less energetic stream-flow conditions. The deposits represent the end phases of debris flow deposition, followed by the establishment of shallow lacustrine and minor deltaic environments. The area may still have been seasonally semi-arid to wet-temperate and subject to flash-flooding events. There was probably significant vegetation cover, dominated by a Taxodiacaea forest (Jolley Reference Jolley and Widdowson1997).

10.6. Phase 6

Further central uplift in the Cuillin area was followed by a resumption of surface volcanic activity and some faulting as the Cuillin (shield) Volcano became established. A major flank fissure eruption on the volcano's slope apron, most likely along the site of the present-day dyke swarm axis, produced the TF flow field. Uplift in the Cuillin area by up to 1 km is thought to have been caused by the emplacement at depth of gabbroic and peridotitic ring intrusions, and cone-sheets, and there would also have been uplift along the line of any developing fissure. The geochemistry of cone-sheets emplaced into the Cuillin Intrusive Centre (and much of the regional dyke swarm) matches that of the TF lava, implying these to be genetically linked.

The TF flow field was initially erupted in a series of pulses as a voluminous sheet-flow from a vent-fissure system (situated up- palaeo-slope), most likely to the south or east, of the present-day outcrops. There is no record of any pyroclastic activity having immediately preceded the eruption of the lavas, but it is likely that there was extensive fire-fountaining along the rift (i.e. the axis of the dyke swarm), comparable perhaps to the so-called ‘Skáftar Fires’ that accompanied the Laki-Grímsvötn eruption(s) of 1783 in southern Iceland. Also, by this time, a compound volcanic edifice (Cuillin (shield) Volcano) had presumably developed to the south above the embryonic intrusive complex, with satellite tephra cones perhaps situated on the apron flanks along the axis of the regional dyke swarm. Much of this material may have been poorly- or totally-unconsolidated.

10.7. Phase 7

Due to irregularities in the topography on the volcano's flanks, the TF lava was not restricted to the area adjacent to the fissure, but was effectively diverted and channelled obliquely down the existing palaeo-drainage until its ultimate arrival in the Talisker area (and, most likely, beyond). The flow travelled far beyond its immediate eruptive source area, initially channelling down upland valleys, then through subsurface magma conduits. The flow most likely arrived as two separate lava streams, more or less simultaneously, and initially infilled deeper, steep-sided fluvial channels and gorges, akin to the Laki 1783 eruption (Thordarson & Self Reference Thordarson and Self1993). The presence of horizontal columnar joints on the northern flanks of both outliers suggests that, locally, these channels may have been aligned SE–NW, or that the flow streams encountered fairly major topographic barriers near their base. This direction is radial to the present-day Cuillin Hills and sub-parallel to the regional dyke swarm, and possibly represents an original palaeo-slope.

Self et al. (Reference Self, Keszthelyi and Thordarson1998) have suggested that the magma plumbing system beneath the lava fields of the NAIP did not usually allow large batches of magma to accumulate at any one time, so that most flows are thin in comparison to major flood basalt provinces such as the CRBG and the Deccan. Evidence from the TF lava suggests that during the development of the Cuillin Volcano, this plumbing system was largely by-passed and larger chambers were established. All indications are that the TF flow field, when compared to the vast majority of basaltic lavas in the NAIP, was comparatively voluminous, fluid and rapidly emplaced. The TF flow field was almost certainly erupted from a major fissure system in a single short-lived event. A similar mechanism is envisaged (Thordarson & Self Reference Thordarson and Self1993) for the 1783 Laki-Grímsvötn eruptions, with which the TF has already been compared.

If certain features of these flow streams are interpreted as being due to shallow auto-intrusion and inflation, then only gentle ground slopes (<4° and perhaps <0.5°) need apply (Kent et al. Reference Kent, Thomson, Skelhorn, Kerr, Norry and Walsh1998).

10.8. Phase 8

Eruption continued. The original channels/canyons were rapidly inundated, either by lava streams entering them directly from the direction of the Cuillin Volcano then overtopping, or by cascading flows originating from an area of uplift along the axis of the regional dyke swarm. In all probability, one or more large ‘lava lakes’ and smaller perched lava ponds were formed as the flow became impounded within the Talisker Graben. The flow inundated and buried the earlier drainage systems to a depth in excess of 120 m, displacing local water courses and bodies of standing water. This buried topographical surface was a complex one, with the TF flow disconformably onlapping various members of the Gleann Oraid Formation and the PBCF. That the principal drainage systems were still active and their associated sediments largely unconsolidated at the time of the eruption of the TF flow field is indicated by the presence of locally ‘invasive’ (shallow intrusive) phenomena within PBCF-3 deposits, chilled margins and a pillow-breccia-like facies at the base of the TF lava.

10.9. Phase 9

In-situ cooling and solidification of the two ponded lava streams (Preshal More and Preshal Beg) began, but was interrupted by further pulses of tube-fed magma emplaced beneath the surface carapace. The nature and style of the columnar joint sets and sheeted bodies towards the top of the present day outcrops (more especially on Preshal Beg) clearly demonstrate that shallow auto-intrusive activity (TF-5) occurred at this late stage during flow emplacement.

10.10. Phase 10

Slow cooling and solidification of the TF lava then proceeded. Joint systems developed and propagated perpendicular to the principal external cooling surfaces and internal isotherms. This resulted in near-vertical columnar joints where the base of the lava was in contact with the wet valley floor, whereas at the valley sides the joints are inclined, reflecting the profile of the valley. Locally, where the valley sides were near-vertical, the columnar joints developed a sub-horizontal attitude. There were other important controls on cooling and solidification, especially at higher structural levels within the flow. The complex arrangement of irregularly interfering, fanning, prismatic and columnar joint systems in the irregular columnar facies (TF-4) is thought to owe its formation to the disruption of isotherms by agencies such as the localised ingress of surface (meteoric) waters into the top of the flow. This could be achieved in part through rainfall during early cooling but was most likely due to the re-establishment of rivers and streams diverted and displaced during the emplacement of the flow itself. Flow-top topography is likely to have been somewhat irregular and shallow stream courses and lakes will have formed. Water ingress to the cooling interior of the flow would also have occurred locally where stream courses, diverted along the sides of the flow, became impounded and dammed.

Almost 59 million years of weathering and erosion have removed any vestige of subsequent Palaeogene sediments (or further lavas) that may have been deposited on top of the TF, but it is clearly possible in the scenario we put forward that significant fluvio-lacustrine sediments accumulated behind lava dams similar in fashion to the spectacular structures described from the late Cenozoic lavas of the western Grand Canyon, USA, by Hamblin (Reference Hamblin1994) and some of the intracanyon flows in the CRBG.

10.11. Phase 11

At some time after the emplacement and solidification of the TF lava, and while the area was still under an extensional regime, a significant amount of block-faulting took place. This resulted in features such as the Talisker Graben as defined today by the Talisker East and Talisker West faults (Fig. 2).

10.12. Phase 12

The TF is deemed to be the youngest preserved volcanic unit within the Skye Lava Field. In all likelihood it was subsequently buried by younger sediments, lava flows and pyroclastic deposits as various Cuillin (or somewhat younger Red Hill) volcanoes developed. However, the lack of zeolite mineralisation in either outcrop may be significant and interpreted as evidence for only minimal (<300 m?) cover. Subsequent to the complete cessation of the volcanic activity, further uplift, basin inversion and prolonged erosion have, over some 59 million years, removed any vestige of these late volcanic products from the lava field. This erosion, culminating in Pleistocene glacial events (a few glacial erratics of gabbro similar to that exposed in the Cuillin Hills are found on both summits for example) has also differentially removed a considerable thickness of the associated, relatively easily weathered and less resistant ‘softer’ older lavas, leaving two large outliers of this olivine tholeiite lava forming an inverted relief topography so characteristic of intracanyon flows.