Some of the ideas discussed in this blog are published in my new book called "The Stonehenge Bluestones" -- available by post and through good bookshops everywhere. Bad bookshops might not have it....
To order, click

Friday 22 December 2017

More fun and games on Lundy Island

On this map from Carr, Hiemstra and Owen (2017) the "established" Late Devensian ice limit is shown with a solid line. Established by whom?  It is inaccurate in Pembrokeshire, and the dashed line showing an ice edge close to the coasts of Devon and Cornwall seems to be a much better fit for the field evidence

Here is a new paper:

Landscape evolution of Lundy Island: challenging the proposed MIS 3 glaciation of SW Britain
Simon J. Carr, John F. Hiemstra, Geraint Owen of the Geologists’ Association 128 (2017) 722–741

A B S T R A C TLundy Island, in the Bristol Channel of south-west Britain, holds a pivotal place in understanding theextent and timing of Quaternary glaciations in southern Britain, in particular the timing, extent anddynamics of the Irish Sea Ice Stream during the Devensian glaciation. New geomorphologicalobservations and revised interpretations of geomorphological and cosmogenic exposure data lead tothe conclusion that Lundy was not covered by ice in the last (Devensian) glaciation. Geomorphological features are related to surface lowering by means of granite weathering under mainly periglacial and cool-temperate conditions. Previously reported cosmogenic ages are re-interpreted to reflect a dynamic equilibrium of cosmogenic nuclide production and surface lowering during a prolonged period of subaerial granite weathering. This re-evaluation of the geomorphology of Lundy Island challenges recently proposed interpretations of early glacial cover of Lundy (MIS 4-3) and for cold-based ice cover at the Last Glacial Maximum (MIS 2), and instead supports existing regional ice sheet reconstructions. This study demonstrates that a robust, coherent geomorphological framework is fundamentally important to support the validity of detailed geochronological and stratigraphic investigations

It’s always good to see an old-fashioned geomorphological row going on — conducted, as ever, in the politest possible terms. In this paper, Carr, Hiemstra and Owen go after Chris Rolfe and others who have suggested (rather uniquely, it has to be said) that the most dramatic recent glaciation of Lundy Island was during the Early Devensian, at a time when most workers consider the Bristol Channel basin to have been ice-free. We have considered the Rolfe thesis here:
and you can find other discussions on Lundy by typing "Rolfe Lundy" into the search box........

I have been somewhat sceptical about the "Rolfe theory", largely on the basis that it appears to be unsupported elsewhere. My own feeling was, on reading the Rolfe et al (2012) paper, that the cosmogenic dates / exposure ages on which it depends were all inaccurate for some reason or another. Maybe the cause of the false (?) readings was intermittent rock surface cover or rock surface breakdown by weathering……… 

Anyway, when I got at this PDF I was rather interested in what the new authors had to say. The abstract is above.

I was rather concerned at the outset by the authors’ assumption that there are accepted “regional reconstructions” and “determined ice margins” for other parts of the Celtic Sea / Bristol Channel arena. They also refer to "the established regional model of glaciation of the Southwest British Isles” as if it is soundly based and widely accepted. Well, I don’t think it is, and since the LGM maximum of the Irish Sea Glacier is in the wrong place in West Wales on Figure 1, it is very probably in the wrong places elsewhere as well.

The authors give a summary of the geology and regional context of Lundy, and when they consider the borehole and sedimentary evidence from the floor of the Bristol Channel. Strangely, they are disinclined to believe the data with respect to glacigenic sediments beneath sea-level, and suggest that the deposits referred to as “glacigenic” may not actually have a glacial origin because there are no known scouring or streamlining features in the vicinity. I don’t follow that line of reasoning at all, since some glaciated areas have streamlined features and others do not. Anyway, at an early stage in the paper it’s fairly obvious where the authors are heading.

In the “evidence” part of the paper, the authors seek to systematically dismantle the claims by Rolfe et a that there are streamlined granite surfaces and associated features, whaleback forms, grooves and meltwater channels, erratic scatters and perched boulders, tors, faceted clasts and diamictons, and periglacial materials. Then they reinterpret the cosmogenic dates as corrupt because of the gradual lowering of granite surfaces by granular disintegration and “grus” production.

These are the conclusions:
An evaluation of the geomorphological evidence previously reported by Mitchell (1968) and Rolfe et al. (2012, 2014) in support of Devensian glaciation of Lundy Island cannot be reconciled with any glaciologically-plausible or coherent glacial landsystems. Re-examination of the geomorphology of Lundy Island identifies a suite of related landforms that reflect a combination of subaerial processes of pneumatolysis, granite weathering and slope movement operating under cool-maritime temperate and periglacial conditions, with more recent impact of wildfires and human agency. Critically, this study finds no evidence for past glaciation, contrary to previous work. A revised model of Late Quaternary landscape evolution of Lundy Island dominated by granite weathering periodically enhanced by periglacial conditions offers a simple and realistic explanation of previously-published cosmogenic nuclide dates. This revised model removes the need to explain an additional MIS 3/4 glacial advance not recorded elsewhere in the southwestern sector of the BIIS, and conforms to the established regional glacial stratigraphy of the Irish Sea Ice Stream, as well as previously-reported periglacial development of tors on nearby Dartmoor.
So where does this leave us? Some of the evidence presented here is convincing, and the authors are right, I think, to question some of the assumptions of Rolfe et al about what happened on Lundy in the Early and Late Devensan. But just as Rolfe et al have “forced” the evidence into their hypothesis, and have found traces of glaciation everywhere, Carr et al have gone to the other extreme, and have forced all their evidence into a non-glacial hypothesis. In particular, they seek to argue that the so-called “glacial features” described by Rolfe et al "cannot be reconciled with any glaciologically-plausible or coherent glacial land systems” — I find that unsettling, since landforms and deposits in close juxtaposition do not need to be contemporaneous or even to belong to the same “glaciological plausible” episode. Just because granite surfaces degrade and break up to create scatters of grus and even broken granite cobbles, that does not mean that the surfaces are not glaciated; and just because assorted degradational processes are at work in grooves and bedrock channels, that does not mean that they have not also been affected by glacial meltwater. Just because there are tors nearby, that does not mean that nearby surfaces cannot have been streamlined; and I have seen far too many meltwater channels adjacent to streamlined surfaces to think that the two are mutually exclusive.

The jury is still out. I remain convinced (not having been to Lundy, and having to depend on published evidence) that the island has been affected by overriding ice during the Devensian, and that there are real problems of timing or dating. The reference to an Early Devensian glaciation is anomalous, since one would expect to see evidence of it in Pembrokeshire and elsewhere. And I remain convinced that this new paper by Carr et al does nothing to confirm the correctness of "the established regional model of glaciation of the Southwest British Isles”. I still think that strange lobe of ice pushing out into the Celtic Sea does not make sense.

Monday 18 December 2017

BRITICE is not much help on the Stonehenge front......

Thanks to Peter for drawing our attention to this post from Tim Daw.

Tim seems to think that the glacial transport hypothesis is dead, because nothing shows up on the BRITICE map which I referred to in a recent post. Sorry Tim, but before committing yourself to print you should have read the BRITICE stuff more carefully, since it refers specifically to the Late Devensian glaciation.  It has no relevance for the Stonehenge debate.  Wrong glaciation.  Forget it.

And as for Jim Scourse's article from 1997 -- for goodness' sake, that is 20 years out of date.  Hundreds of relevant articles have been published since that was published.  The debate has moved on, and as I have shown on this blog, it is now accepted that on glaciological grounds it is perfectly possible that ice during the larger past glaciations did indeed reach Salisbury Plain. As I have said on this blog, I am not at all happy with that old article anyway.  Read my past posts.

Monday 11 December 2017

Ice in the Bristol Channel -- the latest modelling

This is a comprehensive and important paper which draws together a vast amount of information on the Devensian Eurasian ice sheet complex.There is a huge reference list -- including many articles that can be open-accessed as PDFs.

Henry Patton, Alun Hubbard, Karin Andreassen, Amandine Auriac, Pippa L.Whitehouse, Arjen P. Stroeven, Calvin Shackleton, Monica Winsborrow, Jakob Heymane, Adrian M. Hall. (2017)
"Deglaciation of the Eurasian ice sheet complex"
Quaternary Science Reviews
Volume 169, 1 August 2017, Pages 148-172

Here is the Abstract:


The Eurasian ice sheet complex (EISC) was the third largest ice mass during the Last Glacial Maximum with a span of over 4500 km and responsible for around 20 m of eustatic sea-level lowering. Whilst recent terrestrial and marine empirical insights have improved understanding of the chronology, pattern and rates of retreat of this vast ice sheet, a concerted attempt to model the deglaciation of the EISC honouring these new constraints is conspicuously lacking. Here, we apply a first-order, thermomechanical ice sheet model, validated against a diverse suite of empirical data, to investigate the retreat of the EISC after 23 ka BP, directly extending the work of Patton et al. (2016) who modelled the build-up to its maximum extent. Retreat of the ice sheet complex was highly asynchronous, reflecting contrasting regional sensitivities to climate forcing, oceanic influence, and internal dynamics. Most rapid retreat was experienced across the Barents Sea sector after 17.8 ka BP when this marine-based ice sheet disintegrated at a rate of ∼670 gigatonnes per year (Gt a−1) through enhanced calving and interior dynamic thinning, driven by oceanic/atmospheric warming and exacerbated by eustatic sea-level rise. From 14.9 to 12.9 ka BP the EISC lost on average 750 Gt a−1, peaking at rates >3000 Gt a−1, roughly equally partitioned between surface melt and dynamic losses, and potentially contributing up to 2.5 m to global sea-level rise during Meltwater Pulse 1A. Independent glacio-isostatic modelling constrained by an extensive inventory of relative sea-level change corroborates our ice sheet loading history of the Barents Sea sector. Subglacial conditions were predominately temperate during deglaciation, with over 6000 subglacial lakes predicted along with an extensive subglacial drainage network. Moreover, the maximum EISC and its isostatic footprint had a profound impact on the proglacial hydrological network, forming the Fleuve Manche mega-catchment which had an area of ∼2.5 × 106 km2 and drained the present day Vistula, Elbe, Rhine and Thames rivers through the Seine Estuary. During the Bølling/Allerød oscillation after c. 14.6 ka BP, two major proglacial lakes formed in the Baltic and White seas, buffering meltwater pulses from eastern Fennoscandia through to the Younger Dryas when these massive proglacial freshwater lakes flooded into the North Atlantic Ocean. Deglaciation temporarily abated during the Younger Dryas stadial at 12.9 ka BP, when remnant ice across Svalbard, Franz Josef Land, Novaya Zemlya, Fennoscandia and Scotland experienced a short-lived but dynamic re-advance. The final stage of deglaciation converged on present day ice cover around the Scandes mountains and the Barents Sea by 8.7 ka BP, although the phase-lagged isostatic recovery still continues today.

Most of this paper is concerned with the wastage phase of this vast ice sheet complex -- following on from the modelling done by Patton et al in 2016.  We have discussed that work before, here:

Henry Patton's early modelling was intriguing since it showed a Devensian ice margin far beyond the conventional one in the Celtic Sea arena -- with an ice-filled Bristol Channel, Irish Sea ice impinging on the coasts of Cornwall and Devon, and pressing well inland in Somerset too.  This was the 2016 published model for the Devensian maximum ice margin:

The modelled Devensian ice extent, as determined by Patton et al in 2016.  The red line is the older assumed ice limit.  Note that in the Midlands and in the Bristol Channel area, modelling suggests that ice could have progressed more than 100 km further south than assumed.  Ground truthing holds the key......

The modelling work published in 2017 is based on a vastly increased amount of data from many sources, and the segment of the ice sheet complex incorporating the British Isles (which the authors refer to as the "Celtic Ice Sheet") looks like this:

The extent of the Devensian glaciated area is very similar indeed, adding to our confidence that it fits well with glaciological, climatic, topographical, sea-level and many other parameters.    The garish green colour used here relates to the rate of ice edge retreat.  Solid green indicates rapid retreat, yellow suggests approximate stillstands or slow retreat, and red suggests a more prolonged stillstand. So the map shows a relatively prolonged stillstand of the ice edge along the coast of Cardigan bay and on the northern flank of Preseli -- I can live with that, since it is supported by rather a lot of evidence on the ground.

Now let's put some flowlines onto the map -- ice in ice sheets and ice caps always flows broadly perpendicular to the ice edge, so the patters will have been something like this.  (There will of course have been many internal irregularities caused by uplands, glacier discharge routes and changes in the surface topography and the bed conditions of the ice we have discussed many times regarding the interactions between ice from the Welsh ice cap and the ice of the Irish Sea Glacier or ice stream.)

This is in my view far more realistic than some of the maps published by the BRITICE team over recent years.  Since we are all interested in how the bluestones got to Stonehenge, note that the modelling shows -- yet again-- that Devensian ice COULD have reached Salisbury Plain.  I personally doubt that it did in reality, but since this modelling work will also hold pretty accurately for pre-Devensian glacial episodes, my view is reinforced that during the Anglian Glaciation (around 450,000 years ago) ice might well have carried those dearly beloved bluestones almost all the way from Preseli to Stonehenge.

Thursday 7 December 2017

More on the Carn Goedog dolerite sill

This rather splendid oblique aerial photo has just been published in Pembrokeshire Life magazine -- it was taken in 2009 by Toby Driver of the Royal Commission.  It shows Carn Alw in the foreground and Carn Goedog in the distance. Because of the light snow cover there is tremendous detail in the image -- I have added a few pointers for those who do not know the area. Click to enlarge the image.

The photo shows just how extensive the outcrops of the Carn Goedog sill actually are.  As I have explained in previous posts, the assumption that there is a Neolithic bluestone quarry at Carn Goedog is very dodgy indeed, and is based in turn on the assumption that the spotted dolerites ar Stonehenge have been precisely provenanced to the Carn Goedog tor.  But these spotted dolerites outcrop all the way along the sill to the vicinity of Carn Alw, and I have seen nothing in the geological papers that demonstrates that any of the spotted dolerite monoliths actually come from the tor rather than from outcrops further to the east or indeed towards the west.  The sampling point density is just not great enough for any definitive conclusions to be drawn.  The geologists (Richard Bevins and Rob Ixer) sampled the tor because it is a prominent feature  and therefore "attracts" the sampling process.

Confirmation bias yet again? I'm not sure I would put it as strongly as that, since I suppose many of the geological samples were taken years ago, long before Neolithic quarries were being thought of in this area.  But you must always be careful in this work not to attach "artificial significance" to places simply because they happen to be the places you have sampled........

Monday 4 December 2017

Ice in the Bristol Channel

One of the figures from the article.   I do not believe that the shown "Late Devensian limit" is well supported by the published evidence, and I am inclined to accept that the so-called Early / Middle Devensian ice limit, drawn largely on the basis of the borehole evidence, is in 
fact of Late Devensian age.

I have been looking again at the 2017 article by Gibbard, Hughes and Rolfe, in which it is argued that Lundy Island was glaciated during the Middle Devensian -- at a time frequently assumed to have been characterised by a mild or interstadial climate. There is a big dispute going on about the dating of rock surface exposure on the island following a glacial transgression -- more of that anon. But what interests me more at the moment is the very strong evidence for thick glacial deposits in the middle section of the Bristol Channel. These deposits (for the most part assumed to be till) are up to 38m thick in logged boreholes -- they may of course be even thicker in other locations. They are not interpreted as glaciomarine sediments dropped from icebergs or from the underside of a flowing glacier; and so this points to heavy glacial incursion from the west. When did this occur? Because the sediments on top of the till are relatively thin (normally just a few metres) a Devensian age is most likely.

Here is an extract from the paper, dealing with the borehole evidence for the area to the east of Lundy.

New insights into the Quaternary evolution of the Bristol Channel, UKPHILIP L. GIBBARD, PHILIP D. HUGHES and CHRISTOPHER J. ROLFE.JOURNAL OF QUATERNARY SCIENCE (2017) ISSN 0267-8179.DOI: 10.1002/jqs.29513 April 2017


Swansea Bay and Gower

Boreholes from Swansea Bay indicate tills associated with the southernmost extension of the Welsh Ice Cap which reached ca. 15 km south of Swansea (Bowen, 1970, 1973, 1974; Culver, 1976; Culver and Bull, 1979; BGS, 1986). Sedimentological and microfaunal analyses indicate that these glacial deposits were originally land-based and were progressively submerged by a marine transgression between 10 and 2.5 14C ka BP (Culver and Banner, 1978). A series of rock basins in Swansea Bay were occupied by glacial lakes before marine transgression and are likely to have been present from glacier retreat at ca. 16 14C a BP until marine transgression at ca. 10 14Cka BP (Culver and Bull, 1979). As noted earlier, a clear end moraine is also visible at the sea-bed surface in high-resolution bathymetric maps with a network of meltwater channels visible outside of this moraine (Fig. 3). The eastern part of the moraine arc is covered in thick sand accumulations at Scarweather Sands. Glacial deposits have also been identified beyond the limits of the Swansea Bay moraines by Blackley (1978, his fig. 11), south and south-east of the Scarweather Sands area based on continuous seismic profile records.

The Gower peninsula is well known for its Pleistocene sediment record, with glacial deposits and raised beach deposits recorded and reviewed in numerous publications (e.g. Bowen, 1974; Bowen and Sykes, 1988; Bowen et al., 1985). The Gower is important for glacial research because it marks the southern limit of the Devensian Welsh Ice Cap. A boulder associated with this limit has yielded a 36Cl exposure age of 23.2`2 ka (Bowen et al., 2002). On the south and west of Gower the Paviland Moraine, a marginal complex of meltwater gravels and sands, rests on a red clay till (Bowen, 1974, 1999), rich in Millstone Grit clasts derived from southern Welsh Carboniferous rocks. Raised-beach deposits date to the last interglacial and earlier. Uranium-series ages from a cave stalagmite interbedded with marine beds at Minchin Hole indicate an MIS 5 age (Stringer et al., 1986). Bowen et al. (1985) argued that at least three highstands of sea level are recorded, the older two of which are correlated to the interglacials of MIS 5e (Ipswichian, Eemian Stage) and MIS 7 (Late Wolstonian, Late Saalian Substage).

Atlantic Array and Lundy

Till is present throughout the Atlantic Array survey area (Channel Energy Limited, 2010). This till (unit 3) is overlain by sand and gravel deposits (units 1 and 2). The till is present throughout the Atlantic Array survey area and according to the Channel Energy Limited (2010) report by Gardline Geo- survey reached no more than 4 m thickness. However, a later borehole report (Channel Energy Limited, 2011) indicates a much thicker glaciogenic sequence (see below).

Three boreholes (BH 10, 12, 13), out of 15 proposed in a Senergy report for RWE npower renewables (2010), encoun- tered glacial sediments in the Atlantic Array survey area. They include clay and/or sands interpreted as glacial (Pleistocene) with a thickness of 2–8 m, underlying 2–18 m accumulations of sand. Data from four of the 15 proposed borehole locations (BH 6, 11, 12, 15) are presented in a later report by the Danish Geotechnical Institute (GEO) for Channel Energy Limited (2011) and in three of these bore- holes glacial deposits are reported (BH 6A, 12, 15A). The locations of the GEO boreholes are given in Fig. 1. Summa- ries of the borehole logs are illustrated in Fig. 5. The thickest glacial deposits in the Bristol Channel are present between Lundy and the mainland (BH 15A) where 37.9m of glacial deposits are logged (Fig. 5) (Channel Energy Limited, 2011). In this borehole the top 1.9m is composed of sand. This is underlain by three units of silt/clay deposits with occasional subangular and subrounded cobbles. The silt/clay units (1.9– 12.7, 17.2–23.0 and 24.5–39.0m depths) are separated by sand/silt and gravel units. The middle silt/clay units (17.2– 23.0m) and also the sand and gravels separating this unit from the upper silt/clay unit contain ‘rare seams of fine to medium gravel and lignite’. The upper silt/clay unit is predominantly grey although grades upwards into reddish and dark brown gravelly/sandy/silty clays. The lowermost silt/ clay unit is reddish grey and olive-grey in colour.

The borehole log from BH 15A records the entire sediment sequence below 1.9 m as from a glacial environment (Gl) and of glacial age (Gc) (Fig. 5) (Channel Energy Limited, 2011). Some units are logged exclusively as glacial (1.9–5 m), while below this the sediments are recorded as both glacial and marine (Ma) or freshwater (Fw), sometimes all three simultaneously. However, the lithostratigraphy of the borehole log is interpreted in this paper as representing three till units separated by sand/gravel units. The middle till unit and overlying sand and gravels contain organic materials (lignite). Neither the upper nor lower tills contain any references to lignite. While nogeochronology is available, it is possible that the lower till unit is Middle Pleistocene in age and the middle and upper tills are of Late Pleistocene (Devensian Stage) age. The middle till may be Early or Middle Devensian in age, with lignite seams being derived by reworking of interglacial deposits, possibly of freshwater origin. However, further research is required to determine the age and nature of these deposits. Nevertheless, it does appear that multiple till units are recorded in BH 15A.

In BH 12 the clay unit, described as exclusively glacial, is much less thick than in BH 15A and is present for nearly 4 m between 14.6 and 18.5m and is characterized by many colour variations (Channel Energy Limited, 2011). The clay unit is overlain by 14.6 m of sand with shell fragments. BH 12 is associated with a prominent asymmetrical ridge. This has smaller ridges superimposed onto it (ripples) but the main core of the ridge itself could be a moraine associated with ice coming from the west. The prominent ridges in the area of BH 12 (and to the east) look different from other large ripples and large sand waves of the area. The steep western slope of the ridge is consistent with a steeper ice-contact moraine slope.

BH 6A records just over 3 m of clays/silts between 3.5 and 7.6 m. They are described as mudstone and siltstone but are considered glacial in environment and age (with codes Gl and Gc), although codes indicating marine (Ma) and Jura[- ssic] (Ju) are also used (Channel Energy Limited, 2011). These fine-grained deposits are overlain by 3.5 m of sand and gravel mixed with cobbles and boulders. In the borehole log notes these deposits are considered marine/recent or glacial envi- ronment/age and the interpretation is ‘very uncertain’.

The upper contacts of glacigenic silt and clay deposits in all three cores are at similar elevations of 49.1, 48.9 and 48.6 for boreholes 6, 12 and 15A. Sand and gravel with shell fragments overlies this lower unit (Fig. 5) (Channel Energy Limited, 2011).

Geophysical surveys reported by Channel Energy Limited (2010) and RWE npower renewables (2010) have revealed the presence of more than 22 000 boulders (>0.3 m height) in the Atlantic Array survey areas. Each of these boulders has been identified and measured for height with data on each available in the report by Channel Energy Limited (2010, appendix D). This includes the cable routes through Barnstaple Bay. Two concentrated clusters of boulders occur in the Atlantic Array survey area, one in the west and a larger cluster in the east. The extent of the eastern boulder cluster can be extrapolated southwards since concentrations of boulders are also revealed in the three cable routes between Lundy and the mainland (Fig. 1). 

The origin of the boulder clusters can be attributed to either marine or glacial origin, or both. In the former case boulder concentrations may be related to transient shoreline positions. The boulders could also be glacially transported. The fact that till is reported in boreholes in areas where no surface boulders are present, whereas no till is reported in areas where boulders are found, suggests that either the boulders are not of glacial origin or that they represent boulder remnants of a pre-existing till layer, where the fine matrix has been winnowed out. Pleistocene tills are often relatively thin in the central Bristol Channel and no more than 2–8 m in the cores reported from the Atlantic Array survey area (although 16.1m thick between Lundy and mainland Devon – see above). Another possibility is that the boulders are ice-rafted, since erratic boulders are commonly found on shorelines along the coast of Devon and elsewhere in the Bristol Channel, including as far east as Flat Holm (see ‘Cardiff and the Vale of Glamorgan’ below).

The borehole records referred to in the text. There is clearly some variety in the glacial units revealed in the borehole logs -- there may be up to three till units, and some sands and gravels and freshwater deposits / marine deposits. How many glacial episodes are represented?

Boreholes from the BGS north and west of Lundy also report glacial deposits. These include reports from Borehole 72/49 (51 ̊0.860 N, 4 ̊54.820 W), Borehole 72/52 (51 ̊23.030 N, 4 ̊54.450 W) and Borehole 73/60 (51 ̊24.930 N, 4 ̊44.800 W). Borehole 72/49 is situated ca. 25km due west of Hartland Point in Devon. This borehole revealed sand and gravel (0–3 m) underlain by poorly sorted unconsolidated sediments (3–6m), including clay with numerous rounded, subangular and angular rock fragments. The latter includes ‘broken cobbles and pebbles of igneous material’. These unconsolidated sediments are considered ‘Recent and Pleistocene’, while underlying bedrock revealed in the borehole is Carboniferous mudstone. Boreholes 72/52 and 73/60 are situated west of the Atlantic Array survey area between Lundy and Pembrokeshire. In Borehole 72/52, deposits are entirely ‘Pleistocene and Recent’ (i.e. Pleistocene and Holocene), the top 0.5 m recorded as sand, underlain by 3 m of brown-grey ‘Boulder Clay’. In Borehole 73/60 the top 0–2.5 m is ‘Recent’ sand and gravel, below which ‘Pleistocene: Till, red-brown, stiff with small rounded pebbles’ is recorded between 2.5 and 9.5 m.

The offshore evidence clearly illustrates that the Outer Bristol Channel was glaciated during the Pleistocene (Fig. 6). As noted above, Lundy also shows clear evidence of glaciation (Mitchell, 1968; Rolfe et al., 2012, 2014; Fig. 1). The island has large areas of ice-scoured granite bedrock with perched boulders (local lithology) and extensive areas of erratic gravels. Rolfe et al. (2012) dated the granite bedrock surfaces using paired 10Be/26Al analyses, the results demonstrating that there was no evidence of long-term complex exposure (i.e. exposure, burial and then re-exposure etc). However, the island had been exposed for ca. 40ka, i.e. since the Middle Devensian.

Note that the text also refers to great concentrations of boulders in places. No less than 22,000 of them have been logged -- most of them NOT in conjunction with the glacial deposits found in the boreholes.
Note how similar this proposed ice edge position is to that proposed on this blog as the approx ice edge of the Devensian Celtic Sea piedmont glacier. Maybe things are coming together.......