Saturday, 4 April 2026

Caution in Attributing the Fremington Clay Series to Irish Sea Glaciation

A Case for Predominantly Fluvial and Periglacial Origins in North Devon

Tim Daw

Cannings Cross, Wiltshire, UK

tim.daw@gmail.com

Note: This paper is a revised and expanded version of an earlier synthesis published in 2025. The original intention was to defer any substantial update until full petrographic results for Taylor's No. 7 hyalopilitic andesite — rediscovered in 2025 at Combrew Farm — were available, since that analysis bears directly on the question of distant versus local provenance for the most potentially significant exotic clast in the Fremington inventory. However, the degree of critical interest the original paper has attracted, particularly regarding the ice-dam height argument and its implications for the Ramson Cliff erratic, suggested that an interim revision incorporating the composite dam model and the findings of Daw et al. (2026) was warranted rather than leaving those points unaddressed. The petrographic results for No. 7, when available, will be reported separately and may prompt further revision.

Fremington Area Superficial Geology - light blue, till which contains the Fremington Clays , the Bickington-Hele ridge lies between them and the estuary - from https://geologyviewer.bgs.ac.uk/

Abstract

The Fremington Clay Series of north Devon has been central to debates on the extent of Middle Pleistocene glaciation in south-west England, often interpreted as evidence of Irish Sea ice incursion during the Wolstonian Stage (MIS 6). Undisputed glacial-marginal and ice-rafted processes demonstrably reached the present north Devon coastline, with associated deposits and erratics at low elevations (~10–30 m OD) and glaciofluvial outwash influence extending to ~55 m OD on the pre-existing Bickington–Hele bedrock ridge. However, stratigraphic, sedimentological, petrological, geomorphological, and chronological evidence—now reinforced by Daw et al. (2026) on the Ramson Cliff erratic (~80 m OD, for which glacial emplacement is considered highly improbable) and the 2025 rediscovery and re-examination of key clasts—warrants caution against designating the series as direct till from a high-standing ice sheet.

The deposits primarily reflect fluvial and paraglacial sedimentation in an ice-marginal Taw–Torridge setting, with overwhelmingly local (Dartmoor/Cornubian) provenance for embedded erratics. Sparse “exotic” signals—small clasts with possible Irish Sea affinities (including Taylor’s No. 7 hyalopilitic andesite) and abraded Irish Sea-type foraminifera—are best explained as dropstones or reworked rafted debris delivered into a proglacial lake at ~30 m OD. A refined composite ice-sediment dam model, drawing on modern proglacial analogues from New Zealand, Svalbard, and Patagonia, demonstrates that lake impoundment at this level can be achieved by a modest marginal ice lobe augmented by high sediment flux from periglacial catchments and confined by pre-existing topography, without requiring extensive onshore overriding. The ephemeral nature of such estuarine barriers explains the absence of a prominent terminal moraine while preserving perched glaciofluvial witnesses such as the Hele gravels. This framework confines significant glacial influence to the coastal fringe and aligns with offshore Bristol Channel evidence.

Keywords: Fremington Clay, low-level glaciation, fluvial-periglacial deposition, erratic provenance, ice-rafting, composite ice-sediment dam, proglacial lake dynamics, north Devon Quaternary.

1. Introduction

The Quaternary glacial history of south-west England remains contentious, particularly regarding the southerly limits of Irish Sea ice. Offshore evidence from the Celtic Sea and Western Approaches—including ice-rafted detritus records and glacigenic sediment distributions documented by Scourse et al. (2009)[1] and synthesised by Gibbard & Clark (2011)[2]—indicates a long-lived ice margin near 51°N during multiple cold stages, yet onshore corroboration is sparse and disputed. In north Devon, glacial-marginal or ice-rafted processes clearly reached the present coastline, depositing erratics and glacigenic materials on platforms and in low-elevation sequences around Saunton, Croyde, and the Taw estuary (~10–30 m OD). The Fremington Clay Series (up to ~30 m thick) lies at these low levels, while capping sands and gravels on the Bickington–Hele ridge reach ~55 m OD. These features confirm ice or floating ice impinging on the coast.

However, the extent of inland overriding remains disputed. This revised synthesis argues for continued caution in attributing the Fremington Clay Series to direct Irish Sea till from a high-standing ice sheet. It integrates new data: the re-evaluation of the Ramson Cliff erratic at ~80 m OD on Baggy Point (Daw et al., 2026[3]—for which glacial emplacement is now considered highly improbable), the 2025 rediscovery of Taylor’s No. 7 hyalopilitic andesite (pending full petrography), and a fully developed model of composite ice-sediment damming that explains the reconstructed ~30 m lake level without requiring extensive onshore ice. The series is interpreted as predominantly fluvial and paraglacial, with local sources dominant and any exotic signals compatible with limited rafting into a proglacial lake.

2. Stratigraphy and Sedimentology

The Fremington Clay Series overlies bedrock or basal gravels at low elevations (~10–26 m OD in exposures). Croot et al. (1996) carried out the most detailed modern investigation of the sequence, including a purpose-dug excavation from surface to bedrock at the former Brannam's Pottery Clay Pits and Higher Gorse Claypits (SS529317). They delineated five units (Table 1). Horizontal to pseudo-laminated bedding, fining-upward trends, and weak clast fabrics are inconsistent with subglacial lodgement till. They are, however, entirely compatible with low-energy fluvial or glaciolacustrine settling in a valley-confined proglacial lake at ~30 m OD — which is the depositional environment that the bulk of the sedimentological evidence supports, and which the present paper accepts as the most parsimonious interpretation for the upper clay units.

Table 1. Fremington Clay Series stratigraphy (adapted from Croot et al. 1996).

Unit	Description	Thickness	Key Features	Interpretation
E (Head)	Gravelly sand/clay; angular local clasts	1–1.5 m	Cryoturbated; gradational base	Periglacial solifluction (Devensian+)
D	Clast-rich weathered red clayey silt	0.5–1.0 m	Over-consolidated; CaCO₃ 10–20%	Weathered glaciolacustrine/fluvial
C	Irregular sand/silt lenses; reworked fossils	2–2.5 m	Sharp contacts; no strong grading	Ice-proximal fluvial sands
B	Dark brown clay; stoneless base to clast-rich top	8–9 m	Pseudo-laminae; weak fabrics; >1,500 clasts analysed	Low-energy lacustrine; episodic flood inputs
A (Basal)	Clast-supported subangular gravels	1.5–2.0 m	Weak imbrication; local clasts	High-energy fluvial/proglacial outwash

Croot et al. examined more than 1,500 clasts in the 16–256 mm size range, extracted from grab samples at various levels within Unit B. All but a single exception could be accounted for in the bedrock geology within a 10 km radius of the site. That exception — a cobble-sized clast at approximately 5 m depth displaying typical flat-iron subglacial morphology with exceptionally well-striated faces — represents the only unequivocally glacially-transported in-situ clast recovered from the clay pits. Croot et al. were explicit about the weight this single find carried in their interpretation: without it, they stated, they "would have been forced to consider a much wider range of possible origins for the Fremington Clay Series." The glacial character of the deposit thus rests, in the primary investigators' own assessment, on a single clast rather than on the bulk sedimentology.

The critical question is not whether a proglacial lake existed — the fining-upward sequence, dropstones, and laminated silts make that case persuasively — but what kind of ice margin was needed to dam it, and what that implies for the lateral extent of glaciation along the open coast.

3. Provenance of Erratics: Altitude and Source Constraints

A fundamental observation is the consistent low elevation of in-situ erratics within the Fremington Clay: ~10–26 m OD in clay pits and cuttings, as recorded by Taylor (1956)[4] and Arber (1964).[5] These elevation data, it should be noted, predate modern differential GPS levelling, and the original Ordnance Datum benchmarks used by Taylor and Arber carry uncertainties of ±1–2 m. Updated fieldwork in 2025 confirmed the approximate range by reference to current OS mapping. Coastal erratics on Saunton–Croyde platforms and in head deposits occur at or near sea level to slightly elevated positions consistent with raised beaches or periglacial reworking. No verified erratics in the main clay body exceed ~30 m OD.

The Hele gravels capping the Bickington–Hele bedrock ridge to ~55 m OD contain erratic clasts deposited by meltwater streams linked to the same marginal system. These include igneous types (dolerites, possible andesites) alongside dominant local Devonian–Carboniferous and Dartmoor-derived material. No detailed petrographic inventory exists specifically for the ridge gravels, but the assemblage overlaps with the sparse, mixed-provenance clasts in the lower clay and reflects fluvial transport and abrasion rather than direct ice rafting or high-level till.

Updated erratic inventory (selected examples):

Taylor’s No. 7 (Combrew Farm driveway wall): Hyalopilitic andesite, ~16 inches, well-rounded, glassy, brittle, no augite; closely matches the original Scottish (Dumfries/Argyll) description in Dewey (1910).[6] Rediscovered 2025; full petrography awaited. Even if the pending analysis confirms a genuinely distant Scottish origin, the clast would remain one of very few exotics among >1,500 locally derived stones—entirely consistent with delivery as a single dropstone from floating ice calved from a marginal lobe in Barnstaple Bay, rather than requiring direct till transport by an overriding ice sheet.

Other clasts (spilites, dolerites, quartz porphyry) align with Dartmoor aureole or Cornubian sources (Madgett & Inglis, 1987).[7] Sparse possible Irish Sea-affinity lithologies (<1%) fit occasional dropstones or reworked rafted debris into the proglacial lake. The Ramson Cliff altered epidiorite/greenstone at ~80 m OD is petrographically compatible with Cornubian/Dartmoor aureole sources, was discovered upright with no pre-1969 record, and shows no beach abrasion—rendering glacial emplacement highly improbable (Daw et al., 2026).

4. The “Exotic” Signal: Foraminifera and Far-Travelled Clasts

While the series is overwhelmingly local in provenance, a subtle “exotic” signal is present and requires careful interpretation. A small number of clasts show possible Irish Sea affinities, including certain quartz-dolerites and hypersthene-andesites (notably Taylor’s No. 7) that parallel material in the Saunton/Croyde coastal deposits. These clasts are typically small pebbles or cobbles rather than large boulders. They are best explained as dropstones delivered by floating ice from a modest ice lobe in Barnstaple Bay, or as minor reworked material from the adjacent sea floor, rather than primary till.

The clay also contains a derived microfauna, including damaged and abraded Irish Sea-type foraminifera. Kidson & Wood (1974)[8] reported up to eleven species, including Ammonia beccarii and Nonion labradoricum, from samples in the Fremington exposures. The diagnostic feature is their physical condition: tests are abraded, broken, and frequently infilled with secondary minerals, indicating significant transport and reworking prior to deposition. This is consistent with incorporation via glacial meltwater or ice-rafting into a freshwater or brackish lake, rather than in-situ glaciomarine deposition in which tests would be expected to show better preservation. The foraminifera therefore record proximity to Irish Sea-derived material without requiring that the ice itself advanced over the site.

5. The “Missing” Dam: Post-Glacial Erosion and the Hele Gravels

A common critique of the proglacial lake model is the absence of a prominent terminal moraine or dam remnant at the Taw estuary mouth. This absence is, in fact, precisely what modern analogues predict for composite ice-sediment dams in high-energy settings.

At the Nordenskiöldbreen glacier margin in Svalbard, Strzelecki et al. (2017)[9] documented a small ice-dammed lake that formed in the early 1990s, was progressively infilled by a Gilbert-type fan delta, and was largely obliterated as the ice nucleus melted and the remaining sediment was breached and redistributed—all within approximately two decades. The Svalbard case demonstrates how rapidly an ice-sediment barrier can be created and destroyed when the ice component is removed and fluvial processes resume. In the Taw estuary, a high-energy estuarine environment subject to repeated periglacial reworking and Holocene tidal scour would have been at least as effective at removing any central dam remnant. The estuarine plug occupied the lowest point of the system—exactly where post-glacial rivers would re-establish their channels and tidal currents would focus erosion.

The primary surviving evidence of the dam’s effective height is preserved not at the estuary mouth but on higher ground to the south: the Hele gravels, which cap the pre-existing Bickington–Hele bedrock ridge at ~55 m OD.[10] These perched glaciofluvial deposits, laid down by meltwater streams linked to the same marginal system, were high enough above the valley floor to escape the main phase of post-glacial fluvial incision. They contain erratic clasts (including igneous types such as dolerites and possible andesites) alongside dominant local Devonian–Carboniferous and Dartmoor-derived material. The basal Unit A gravels of the Fremington sequence itself represent the “roots” of the dam: the initial high-energy outwash phase before the basin deepened into a more tranquil lake environment.

6. Composite Ice-Sediment Dams: The Modern Evidence

The standard mental model of an ice-dammed lake is a clean wall of glacier ice blocking a valley, with the lake surface pressing directly against the ice face. In this model, the dam height is simply the ice thickness, and the ice surface must exceed the lake surface to prevent overtopping. This is the model implicitly invoked when critics argue that a ~30 m lake at Fremington requires ~30 m or more of solid ice—and therefore an ice surface capable of reaching the ~80 m Ramson Cliff erratic on Baggy Point.

Modern proglacial geomorphology shows that this clean-ice model is the exception rather than the rule. In most settings where glaciers interact with rivers carrying significant sediment loads, the actual barrier is a composite structure: ice, outwash gravel, glaciofluvial sand, debris flows, and—in marine-terminating cases—subaqueous fans. The effective dam height is the combined elevation of all these components, not the ice thickness alone.

6.1 The New Zealand Last Glacial Maximum: Outwash Fan-Head Damming

The most comprehensive modern analogue comes from the Southern Alps of New Zealand, where Sutherland et al. (2019)[11] reconstructed the ice-contact proglacial lake systems associated with the Last Glacial Maximum across the entire mountain range. Their central finding is that the major LGM lakes—Tekapo, Pukaki, Ohau, Wanaka, Hawea, and Wakatipu—were not dammed primarily by moraine ridges or by monolithic ice barriers, but by outwash fan-heads: massive aggradational gravel bodies built by high-sediment-load braided rivers against and around the ice margin.

The mechanism is straightforward. New Zealand’s Southern Alps glaciers during the LGM were characterised by extremely high rates of subglacial erosion and sediment production, fed by rapidly uplifting Torlesse Terrane greywacke. Rivers draining from the ice margins carried enormous bedloads. Where these rivers encountered the glacier terminus, they could not maintain their gradient, and deposited vast outwash fans that aggraded against the ice front. The fan surfaces built up to and above the level of the ice terminus, creating a sedimentary barrier that contributed directly to lake impoundment.

Sutherland et al. (2022)[12] developed this model in detail at Lake Tekapo, where the glacier terminus was entirely buried by the outwash fan-head, with no terminal moraine visible as a discrete landform. The ‘moraine’ at Tekapo is in fact a subdued ridge coalesced into the fan surface. The dam that created the lake—which persists today, long after the ice vanished—was the sedimentary mass itself. Shulmeister et al. (2019)[13] provide wider context, noting that this pattern is typical across the western South Island, where high sediment supply from the actively eroding Alps overwhelmed the capacity of terminal moraines to form as distinct features.

The key insight for the Fremington debate is that these New Zealand lakes stood at substantial depths behind barriers whose ice component was only a fraction of the total dam height. Once the ice melted, the lakes persisted—Tekapo, Pukaki, and Ohau still exist today—held in by the residual sediment mass. The permanence of the lakes derives from the sediment, not the ice.

6.2 High-Arctic Svalbard: Rapid Infilling and Stabilisation

The Svalbard example, already cited in Section 5 for the ephemeral nature of ice-sediment barriers, also illustrates the positive side of composite damming. At Nordenskiöldbreen, fluvial sediment rapidly aggraded against the ice barrier, building a composite dam whose crest exceeded the ice surface in places. The effective barrier height fluctuated as sediment accumulated, eroded, and redistributed. At no point was the dam a simple wall of ice with a measurable freeboard; it was a dynamic, composite structure in which ice and sediment alternately dominated. The process operates even at very small scales and over very short timescales; in a Pleistocene setting with larger ice volumes, greater sediment supply, and longer duration, the composite damming effect would be amplified.

6.3 Perito Moreno, Patagonia: Ice, Debris, and Dynamic Water Levels

The Perito Moreno Glacier in southern Patagonia provides perhaps the most dramatic modern example of composite damming. The glacier periodically advances across a narrow strait to dam Brazo Rico, a branch of Lago Argentino, raising its water level by up to 20 m above the main lake before catastrophic drainage ensues. The damming mechanism involves not just the ice front but a complex assemblage of subaqueous moraine, calved debris, and sediment redistributed by currents and waves against the ice face.

Crucially, the lake level in Brazo Rico during damming episodes rises well above what the subaerial ice thickness at the narrow strait would predict, precisely because the underwater and debris-armoured components of the barrier contribute to the effective seal. Leakage occurs not by overtopping but by subglacial drainage when the hydraulic head exceeds the ice overburden pressure—a process described in the broader GLOF (glacial lake outburst flood) literature by Carrivick & Tweed (2013).[14]

6.4 A General Principle

These three examples—from a temperate maritime mountain belt, a High-Arctic archipelago, and a Patagonian ice field—illustrate a general principle well established in the proglacial lake and GLOF literature but not previously applied to the Fremington debate: in any setting where rivers carry significant sediment loads to an ice margin, the resulting dam is composite rather than monolithic. The effective barrier height is the sum of ice thickness, aggraded outwash, moraine material, and subaqueous sediment, not the ice alone.

Carrivick et al. (2022)[15] synthesise this principle in their review of coincident glacier and lake evolution across the Southern Alps, showing that sediment flux is as important as ice dynamics in determining lake existence, extent, and longevity. In their analysis, ice-marginal lakes are fundamentally sedimentary features as much as they are glaciological ones.

7. Application to the Taw Estuary

7.1 Sediment Supply from a Periglacial Hinterland

The Taw and Torridge drain catchments that include the northern margins of Dartmoor and the extensive periglacial plateau surfaces of Exmoor. Evans et al. (2012)[16] describe Dartmoor during Pleistocene cold stages as an independent ice cap with extensive periglacial slopes generating clitter fields, solifluction mantles, and thick head deposits. Edmonds (1972)[17] documents the terrace stratigraphy of the Taw valley, recording multiple episodes of fluvial aggradation consistent with high-sediment-load braided river systems.

The rivers feeding the Fremington lake basin were not clear-water streams draining stable, vegetated catchments. They were periglacial braided rivers with extremely high bedload transport rates, carrying gravel, sand, and silt derived from frost-shattered bedrock, solifluction deposits, and reworked older drift. The analogy with New Zealand’s high-sediment-load glacial rivers is direct: these are exactly the conditions under which outwash fan-heads build against ice margins.

7.2 The Composite Dam Model for Fremington

Applying the composite dam model to the Taw estuary, the scenario runs as follows. An Irish Sea ice lobe advances into Barnstaple Bay, blocking the combined outflow of the Taw and Torridge. The ice margin need not be a towering cliff; a modest, debris-charged lobe grounding in the relatively shallow waters of the inner bay would suffice. As the rivers back up, sediment-laden flow deposits outwash fans against the upstream (southern) face of the ice barrier. Gravel and sand aggrade rapidly, building fan-head surfaces that supplement the ice and raise the effective dam height.

The Fremington Clay itself records this process. The basal gravels—containing both local and far-travelled lithologies—represent the initial outwash phase, when coarse bedload was deposited in a proglacial setting. The overlying laminated silts and sands record a transitional phase as the basin deepened and the dam became more effective, trapping finer sediment. The uppermost stoneless clay represents full lacustrine conditions, with suspension settling of the finest fraction in a quiet, deep-water environment. Dropstones from floating ice derived from the margin punctuate this otherwise tranquil record.

In this model, the 30 m lake surface does not require 30 m of solid ice. It requires a composite barrier—ice plus aggraded sediment plus debris—whose combined crest reached or exceeded 30 m OD at the topographically constricted estuary mouth. The pre-existing Bickington–Hele ridge at ~55 m OD provided lateral confinement on the southern side, preventing the lake from draining southward. The ice component of the dam may have been substantially less than 30 m thick, supplemented by the very outwash and debris that the Fremington sequence itself records.

7.3 The Self-Reinforcing Dam: Sediment as Both Product and Barrier

This interpretation is not ad hoc. The fining-upward sequence of the Fremington Clay Series—basal gravels to laminated silts to stoneless clay—is precisely what the outwash fan-head model predicts. In the New Zealand analogues, Sutherland et al. (2022) describe ice-contact lakes with identical stratigraphic signatures: coarse outwash at the base grading upward into laminated lacustrine fines as the sediment trap becomes more effective and the lake deepens. The Fremington Clay is the sedimentary product of the process that created its own dam.

This self-reinforcing dynamic—in which the lake deposits contribute to the barrier that impounds the lake—is a standard feature of composite ice-sediment dams. The dam grows in effectiveness over time without requiring any increase in ice thickness. A modest ice lobe that initially created a shallow, gravel-floored pond could, through progressive outwash aggradation, give rise to a deep lacustrine basin over decades or centuries. The Hele gravels at ~55 m OD, perched on the bedrock ridge above the zone of later fluvial incision, are the surviving high-water witnesses of this process: they record the meltwater component of the dam system at an elevation consistent with the composite barrier crest, not with the ice surface alone.

8. Implications for the Ramson Cliff Erratic

The composite dam model has a direct and important consequence for the debate over the Ramson Cliff erratic. If a 30 m lake can be explained by a modest ice lobe at the estuary mouth, supplemented by outwash aggradation and laterally confined by pre-existing topography, then the lake level tells us nothing about the height, thickness, or lateral extent of ice on the open coast to the west.

The Ramson Cliff erratic sits at approximately 80 m OD on Baggy Point, a fully exposed coastal headland several kilometres west of the Taw estuary. Its petrography—an altered epidiorite or greenstone of approximately 700 kg—has been re-examined by Daw, Ixer & Madgett (2026), who argue that it aligns with local Cornubian or Dartmoor sources rather than distant Irish Sea material. But even setting aside the provenance question, the erratic at 80 m OD is not explained by a composite dam at 30 m in the valley below.

The argument that ‘the ice must have been higher than the lake’ is true in a trivial sense—the ice surface at the dam must exceed the water surface, or the lake drains. But the dam was at the estuary mouth, in a topographically constricted setting. The ice there could have been 35 m thick (to maintain a few metres of freeboard above a 30 m lake) without extending laterally at that thickness along 10 km of open coastline to reach Baggy Point at 80 m. These are geometrically and physically distinct situations. A lobate ice margin blocking an estuary is not a uniform ice sheet overriding a headland.

The absence of intermediate-elevation erratics between the ~30 m clay body and the ~80 m Ramson Cliff boulder reinforces this point. If ice were thick enough and laterally extensive enough to reach 80 m on Baggy Point, a scatter of erratics at 40 m, 50 m, 60 m, and 70 m along the intervening coastline would be expected. No such scatter has been identified despite over a century of fieldwork. The Ramson Cliff erratic therefore remains isolated and anomalous: well above both the reconstructed lake level and the ~55 m Bickington–Hele ridge, with no supporting suite of high-level deposits, striae, or erratics along the intervening coast. Whatever mechanism emplaced it, the Fremington lake level provides no evidence for that mechanism.

9. Discussion: Separating Dam Height from Ice Extent

The confusion at the heart of the ice-dam height argument is a failure to distinguish between two very different physical situations: the height of a composite barrier at a topographically constricted point, and the regional extent and thickness of an ice mass across open terrain.

In every modern analogue considered here—New Zealand, Svalbard, Patagonia—the dam height at the constriction point exceeds the ice thickness by a substantial margin because of sediment aggradation. The ice does not need to be thick everywhere, or present everywhere, or even the dominant component of the dam. It needs to provide a nucleus around which sediment accumulates. This is particularly true in high-sediment-load systems like periglacial braided rivers, where outwash fan construction is rapid and volumetrically significant.

The Fremington case is a near-ideal candidate for composite damming. The Taw–Torridge system drains a periglacial hinterland with prodigious sediment supply. The estuary mouth is a topographic constriction. The sedimentary record preserves a classic fining-upward proglacial sequence consistent with progressive dam growth. And the absence of a discrete moraine at the dam site is exactly what the New Zealand model predicts: in high-sediment-load settings, moraines are subdued or buried, and the dam is the outwash mass itself.

None of this denies the presence of Irish Sea ice in Barnstaple Bay. A composite dam still requires an ice nucleus. What it denies is the extrapolation from local dam height to regional ice extent. A 30 m lake in the Taw valley is consistent with a modest, topographically controlled ice lobe, not with a thick onshore ice sheet capable of overriding headlands at 80 m.

The same logic applies to the question of inland penetration. If the ice dam sat at the estuary mouth rather than deep within the valley system, the Fremington Clay records the lake that formed behind the barrier, not the advancing front of a glacier. The deposits are confined to a low-level bench along the southern side of the Taw estuary between Barnstaple and Instow (Croot et al. 1996, Fig. 5), with no continuation of till or glacigenic sediment southward up the Taw valley. No erratics have been reported at or south of Barnstaple itself. The Hele gravels at ~55 m OD represent glaciofluvial outwash — material deposited by meltwater — not direct till from an overriding ice mass.

Croot et al. found that the Fremington Clay is mildly over-consolidated, with pre-consolidation pressures of 250–350 kPa, and they considered this consistent with loading by an overriding ice mass. However, they were careful to note that the evidence was insufficient to conclude definitively that the overriding force was glacial ice, observing that fast-ice or a floating ice shelf could produce similar effects on coastal estuarine deposits. The composite dam model offers a further possibility: the over-consolidation may reflect the weight of aggraded outwash and debris on the upstream face of the barrier rather than the passage of a glacier across the lake floor. In either case, the evidence points to ice influence at the coastal margin, not to deep inland penetration of a terrestrial ice sheet up the Taw–Torridge system.

This distinction matters. A modest ice lobe grounding in the outer estuary and supplemented by outwash aggradation is a fundamentally different glacial configuration from an ice sheet advancing tens of kilometres inland up a major river valley. The Fremington evidence is consistent with the former and does not require the latter.

The regional context reinforces this interpretation. Even on the most generous reconstruction of Irish Sea ice extent in the Bristol Channel, onshore glacial evidence on the English side is confined to low elevations at the coastal fringe. The GCR review of the Quaternary of South-West England records a broad consensus that Somerset itself was not glaciated (Kidson, 1977; Hunt et al., 1984), while accepting glaciation of the Avon coastlands in the Kenn lowlands and Vale of Gordano at low elevations. The glacial limit in Sedgemoor appears to lie between Greylake, where a basal diamicton may represent the maximum extent of glacial deposits, and Langport Railway Cutting 6 km further south, which is erratic-free (Campbell et al., 1998). Even the enigmatic deposits on Bleadon Hill at 82 m OD on the Mendip flank lack demonstrably glacially transported erratic material. The Fremington deposits thus sit within a consistent regional pattern: Irish Sea ice impinged on low-lying coastal areas and estuaries along the English shore of the Bristol Channel but did not climb significantly above sea level, penetrate far inland, or override higher ground. The notion that this same ice mass could have reached 80 m OD on Baggy Point or advanced deep up the Taw–Torridge valley system finds no support in any regional parallel.

10. Conclusion

The Fremington Clay Series records a modest ice-marginal setting in which fluvial and paraglacial processes dominated. The sparse Irish Sea lithologies and abraded foraminifera indicate proximity of a limited ice lobe in Barnstaple Bay but do not require voluminous direct till or extensive onshore overriding. The reconstructed ~30 m proglacial lake is explained by a composite ice-sediment dam—a modest ice nucleus augmented by rapid outwash fan-head aggradation from periglacial rivers and laterally confined by the pre-existing Bickington–Hele ridge—rather than a monolithic wall of ice.

The ephemeral nature of such estuarine barriers, demonstrated by modern analogues from Svalbard, explains the absence of a prominent terminal moraine at the estuary mouth. The surviving Hele gravels at ~55 m OD preserve the perched meltwater component of the dam system. The Ramson Cliff erratic at ~80 m OD on Baggy Point—petrographically local, lacking any supporting suite of high-level deposits or erratics at intermediate elevations—remains an isolated outlier whose glacial emplacement is highly improbable.

Elevation data remain central to this interpretation: low-level erratics and deposits (~10–30 m OD) support coastal glacial influence; glaciofluvial outwash reaches ~55 m OD on higher ground; the 80 m outlier does not fit. Dam height in the valley and the lateral extent of onshore ice are separate questions with separate answers. Ongoing geochemical analysis of archival clasts—including Taylor’s No. 7—and potential further study of the Hele gravels will continue to refine provenance and test these interpretations against the primary evidence.

The story advances through careful scrutiny of the rocks, the topography, and the modern analogues—not through assumptions about uniform ice dams.

References

Arber, M.A. (1964). Erratic boulders within the Fremington Clay. Geological Magazine 101(3): 282–283.

Bennett, J.A., Cullingford, R.A., Gibbard, P.L., Hughes, P.D. & Murton, J.B. (2024). The Quaternary Geology of Devon. Proceedings of the Ussher Society 15: 84–130.

Carrivick, J.L. & Tweed, F.S. (2013). Proglacial lakes: character, behaviour and geological importance. Quaternary Science Reviews 78: 34–52.

Carrivick, J.L., Sutherland, J.L., Huss, M., Purdie, H., Stringer, C.D., Grimes, M., James, W.H.M. & Lorrey, A.M. (2022). Coincident evolution of glaciers and ice-marginal proglacial lakes across the Southern Alps, New Zealand: Past, present and future. Global and Planetary Change 211: 103792.

Croot, D.G., Gilbert, A., Griffiths, J. & van der Meer, J.J. (1996). The character, age and depositional environments of the Fremington Clay Series, North Devon. Quaternary Newsletter 80: 1–15.

Daw, T., Ixer, R. & Madgett, P. (2026). A review of the Ramson Cliff erratic: evidence of high-level ice flow? Quaternary Newsletter 167: 13–19. https://doi.org/10.64926/qn.20517

Dewey, H. (1910). Notes on some igneous rocks from North Devon. Proceedings of the Geologists’ Association 21(4): 429–434.

Edmonds, E.A. (1972). Terrace stratigraphy in the Taw valley. Exeter Museums Archaeological Field Unit.

Evans, D.J.A., Harrison, S., Vieli, A. & Anderson, E. (2012). The glaciation of Dartmoor: the southernmost independent Pleistocene ice cap in the British Isles. Quaternary Science Reviews 45: 31–53.

Geological Conservation Review volume: Quaternary of South-West England. Joint Nature Conservation Committee.

Gibbard, P.L. & Clark, C.D. (2011). Pleistocene glaciation limits in Great Britain. In: Ehlers, J., Gibbard, P.L. & Hughes, P.D. (eds), Quaternary Glaciations – Extent and Chronology: A Closer Look. Elsevier, 75–93.

Kidson, C. & Wood, R. (1974). The Pleistocene stratigraphy of Barnstaple Bay. Proceedings of the Geologists’ Association 85: 223–237.

Madgett, P.A. & Inglis, E.A. (1987). A re-appraisal of the erratic suite of the Saunton and Croyde areas. Transactions of the Devonshire Association 119: 135–144.

Scourse, J.D., Haapaniemi, A.I., Colmenero-Hidalgo, E., Peck, V.L., Hall, I.R., Austin, W.E.N., Knutz, P.C. & Zahn, R. (2009). Growth, dynamics and deglaciation of the last British–Irish ice sheet: the deep-sea ice-rafted detritus record. Quaternary Science Reviews 28: 3066–3084.

Shulmeister, J., Thackray, G.D., Rittenour, T.M., Fink, D. & Evans, D.J.A. (2019). The Last Glacial Maximum (LGM) in western South Island, New Zealand: implications for the global LGM and MIS 2. Quaternary Science Reviews 213: 44–66.

Stephens, N. (1966). Some Pleistocene deposits in North Devon. Biuletyn Peryglacjalny 15: 103–114.

Strzelecki, M.C., Long, A.J. & Lloyd, J.M. (2017). Rise and fall of a small ice-dammed lake — Role of deglaciation processes and morphology. Geomorphology 295: 228–243.

Sutherland, J.L., Carrivick, J.L., Shulmeister, J., Quincey, D.J. & James, W.H.M. (2019). Ice-contact proglacial lakes associated with the Last Glacial Maximum across the Southern Alps, New Zealand. Quaternary Science Reviews 213: 67–92.

Sutherland, J.L., Carrivick, J.L., Shulmeister, J., Quincey, D.J. & James, W.H.M. (2022). A model of ice-marginal sediment-landform development at Lake Tekapo, Southern Alps, New Zealand. Geografiska Annaler: Series A 104(3): 182–209.

Taylor, C.W. (1956). Erratics of the Saunton and Fremington areas. Report and Transactions of the Devonshire Association 88: 52–64.

[1]Scourse, J.D., Haapaniemi, A.I., Colmenero-Hidalgo, E., Peck, V.L., Hall, I.R., Austin, W.E.N., Knutz, P.C. & Zahn, R. (2009). Growth, dynamics and deglaciation of the last British–Irish ice sheet: the deep-sea ice-rafted detritus record. Quaternary Science Reviews 28: 3066–3084.

[2]Gibbard, P.L. & Clark, C.D. (2011). Pleistocene glaciation limits in Great Britain. In: Ehlers, J., Gibbard, P.L. & Hughes, P.D. (eds), Quaternary Glaciations – Extent and Chronology: A Closer Look. Elsevier, 75–93.

[3]Daw, T., Ixer, R. & Madgett, P. (2026). A review of the Ramson Cliff erratic: evidence of high-level ice flow? Quaternary Newsletter 167: 13–19. https://doi.org/10.64926/qn.20517

[4]Taylor, C.W. (1956). Erratics of the Saunton and Fremington areas. Report and Transactions of the Devonshire Association 88: 52–64.

[5]Arber, M.A. (1964). Erratic boulders within the Fremington Clay. Geological Magazine 101(3): 282–283.

[6]Dewey, H. (1910). Notes on some igneous rocks from North Devon. Proceedings of the Geologists’ Association 21(4): 429–434.

[7]Madgett, P.A. & Inglis, E.A. (1987). A re-appraisal of the erratic suite of the Saunton and Croyde areas. Transactions of the Devonshire Association 119: 135–144.

[8]Kidson, C. & Wood, R. (1974). The Pleistocene stratigraphy of Barnstaple Bay. Proceedings of the Geologists’ Association 85: 223–237.

[9]Strzelecki, M.C., Long, A.J. & Lloyd, J.M. (2017). Rise and fall of a small ice-dammed lake — Role of deglaciation processes and morphology. Geomorphology 295: 228–243.

[10]Geological Conservation Review volume: Quaternary of South-West England (describes the Bickington–Hele ridge and capping deposits).

[11]Sutherland, J.L., Carrivick, J.L., Shulmeister, J., Quincey, D.J. & James, W.H.M. (2019). Ice-contact proglacial lakes associated with the Last Glacial Maximum across the Southern Alps, New Zealand. Quaternary Science Reviews 213: 67–92.

[12]Sutherland, J.L., Carrivick, J.L., Shulmeister, J., Quincey, D.J. & James, W.H.M. (2022). A model of ice-marginal sediment-landform development at Lake Tekapo, Southern Alps, New Zealand. Geografiska Annaler: Series A 104(3): 182–209.

[13]Shulmeister, J., Thackray, G.D., Rittenour, T.M., Fink, D. & Evans, D.J.A. (2019). The Last Glacial Maximum (LGM) in western South Island, New Zealand: implications for the global LGM and MIS 2. Quaternary Science Reviews 213: 44–66.

[14]Carrivick, J.L. & Tweed, F.S. (2013). Proglacial lakes: character, behaviour and geological importance. Quaternary Science Reviews 78: 34–52.

[15]Carrivick, J.L., Sutherland, J.L., Huss, M., Purdie, H., Stringer, C.D., Grimes, M., James, W.H.M. & Lorrey, A.M. (2022). Coincident evolution of glaciers and ice-marginal proglacial lakes across the Southern Alps, New Zealand: Past, present and future. Global and Planetary Change 211: 103792.

[16]Evans, D.J.A., Harrison, S., Vieli, A. & Anderson, E. (2012). The glaciation of Dartmoor: the southernmost independent Pleistocene ice cap in the British Isles. Quaternary Science Reviews 45: 31–53.

[17]Edmonds, E.A. (1972). Terrace stratigraphy in the Taw valley. Exeter Museums Archaeological Field Unit.

Friday, 3 April 2026

Boots on the Ground vs. Sniping from a Mountain Lair: Testing Glacial Claims in North Devon

Brian John's April 3 post dismisses my Fremington Clay paper as a "pantomime," full of "scraps of geomorphological and glaciological nonsense," and probably written by AI. He devotes roughly half his post to an AI-generated analysis diagnosing my work as AI-generated. The irony writes itself — his AI mischaracterises my paper more egregiously than any fault it claims to detect, attributing to me positions I don't hold.

What the post avoids is substantive engagement with the actual work: the peer-reviewed Ramson Cliff re-examination, the Combrew Farm fieldwork, or the pending geochemistry.

The fieldwork Brian doesn't mention

On 2 November 2025 I visited Combrew Farm, relocated the key Fremington Clay erratic first described by Dewey (1910) and Taylor (1956), Taylor's No. 7, the well-rounded hyalopilitic andesite with possible Scottish affinities, still by the wall where Taylor recorded it. I photographed and measured it and arranged new samples from it, the most diagnostic clast in the suite. Results are pending and will be published.

In February 2026, Rob Ixer, Paul Madgett (one of the original discoverers, who first reported the Ramson Cliff boulder in QN No. 14, 1974), and I published a re-examination of the ~700 kg epidiorite at ~80 m OD on Baggy Point (Quaternary Newsletter 167: 13–19). Petrographic analysis aligns it with Cornubian greenstones, not northern sources. No pre-1969 record of it exists. Its evidential value for high-level ice flow should be substantially downgraded. Brian's post does not engage with any of this.

Fair points and concessions

Brian objects to my paper's characterisation of Bennett et al. (2024). He's partly right: my abstract groups them among "recent critiques," which was imprecise. Bennett et al. provide a major synthesis that supports a glaciolacustrine interpretation with caveats — they are not critiquing glaciation. I should have distinguished their review from my own sceptical position. My body text is more careful, and I stand by the point that their acknowledged ambiguities justify empirical testing of the critical claims, but the abstract wording could have been better.

He also objects to my description of Madgett and Inglis (1987) as characterising these erratics as "sea-ice proxies." That's a fair criticism. M&I 1987 is primarily descriptive — a careful catalogue noting the erratics "may have entered the area by a variety of mechanisms." My phrasing was an interpretive synthesis drawing on multiple sources and overstated what that specific paper concluded.

The ice-dam argument

Brian's one substantive geomorphological point: if a proglacial lake stood at ~30 m OD, the ice dam must have been higher, making an erratic at 80 m unremarkable. But dam height and onshore ice extent are different questions. Sutherland et al. (2019) show that across New Zealand's Southern Alps, LGM proglacial lakes were dammed not by thick ice but by outwash fan-heads, massive aggradational gravel bodies built against the ice margin by high-sediment-load rivers. (See also, Perito Moreno, Patagonia; Nordenskiöldbreen, Svalbard). In the Taw valley, fed by Dartmoor meltwater carrying exactly this kind of coarse sediment load, a modest ice lobe at the estuary mouth supplemented by rapid outwash aggradation would produce a composite dam well above the ice surface. A 30 m lake does not require 30 m of ice, let alone ice extending to 80 m on an exposed headland kilometres to the west. The Ramson Cliff erratic remains isolated, with no supporting suite above ~30 m OD, and its petrography now points to Devon, not Scotland. See https://www.sarsen.org/2026/04/outwash-fan-heads-composite-dams-and.html for more details.

Where we are

The pattern is Popperian: identify the most diagnostic claims in the prevailing model, then test them empirically. The Ramson Cliff erratic has been tested and found wanting. The Fremington clast analyses are underway. Both are boots-on-the-ground work, conducted transparently with public notebooks.

I welcome engagement with the rocks rather than the rhetoric. The story continues through them.

Key References

Daw, T., Ixer, R. & Madgett, P. (2026). QN 167: 13–19. doi:10.64926/qn.20517
Bennett, J.A. et al. (2024). Proc. Ussher Soc. 15: 84–130.
Madgett, P.A. & Inglis, E.A. (1987). Trans. Devon. Ass. 119: 135–144.
Croot, D.G. et al. (1996). QN 80: 1–15.
Sutherland, J.L., Carrivick, J.L., Shulmeister, J., Quincey, D.J. & James, W.H.M. (2019). Ice-contact proglacial lakes associated with the Last Glacial Maximum across the Southern Alps, New Zealand. Quaternary Science Reviews 213: 67–92.

The Neolithic Decline and the Completion of Stonehenge:

Implications of Population Discontinuity in the Paris Basin

Abstract

A 2026 study in Nature Ecology & Evolution documents a clear population discontinuity at the Bury allée sépulcrale in the Paris Basin, linked to the Neolithic decline of c. 3100–2900 BC. Genomic, pathogen, and palaeoecological data reveal a shift from diverse Early Neolithic farmer groups to a more homogeneous population with dominant Middle Neolithic Iberian/southern French ancestry, accompanied by evidence of disease (including early Yersinia pestis) and subsequent forest regrowth. This article evaluates the study's applicability to Stonehenge, whose initial phase (ditch, bank, and Aubrey Holes) dates to c. 3000–2900 BC and whose iconic sarsen trilithons and circle were erected c. 2620–2480 BC. While Britain followed a partially divergent trajectory—retaining Early Neolithic France-derived ancestry longer and experiencing its major turnover only with the arrival of Bell Beaker groups carrying steppe-related ancestry around 2450–2200 BC—the Bury findings illuminate the broader European context of demographic stress, pathogen circulation, and social reorganisation. Stonehenge's continued monumental activity thus emerges as a regionally distinctive phenomenon, potentially reflecting cultural resilience or ritual intensification amid wider instability, rather than a direct parallel to the continental cessation of megalithic tomb construction.

1. Introduction

Megalithic traditions flourished across northwestern Europe for over a millennium before widespread decline around 3100–3000 BC, often interpreted as reflecting demographic contraction rather than purely cultural change. In Britain, however, large-scale monument building persisted, with Stonehenge's most impressive phase—the erection of the sarsen stones—occurring several centuries later. The recent high-resolution analysis of 132 ancient genomes from the Bury site (Seersholm et al. 2026) provides detailed evidence for one instance of this continental discontinuity and invites comparison with insular developments, even though the authors do not address Britain or Stonehenge directly beyond ancestry modelling notes.

2. Key Findings from Seersholm et al. (2026)

The Bury allée sépulcrale, a Seine-Oise-Marne gallery grave, contains two main burial phases separated by a multi-century hiatus. Phase 1 (c. 3200–3000 BC) comprises genetically diverse individuals with mixed Early Neolithic France and some hunter-gatherer ancestry, organised in large multi-generational pedigrees with evidence of female exogamy. Pathogen screening identified active infections, including Yersinia pestis in four individuals (mostly Phase 1), alongside excess juvenile mortality.

Phase 2 (beginning c. 2900 BC and extending to c. 2470 BC) shows a marked genetic shift: individuals are more homogeneous, with mean ancestry attributions of ~83.8% to Middle Neolithic Iberian/southern French sources. Y-chromosome profiles also change (predominantly I2a1a1 in Phase 2), and social structure appears more patrilineal with smaller kinship groups. Pollen evidence indicates forest regeneration in the Paris Basin during the intervening period, consistent with reduced agricultural activity. The authors link this discontinuity to the end of megalithic tomb construction across much of continental northwestern Europe, citing contributing factors of disease, environmental stress, and migration. Steppe-related ancestry appears in the region only after c. 2500 BC. The study notes that British and Irish Neolithic genomes model primarily as Early Neolithic France ancestry, distinct from the later Iberian pulse documented at Bury.

3. Stonehenge Chronology and British Context

Stonehenge developed over centuries. Its first stage (ditch, bank, and Aubrey Holes) dates to c. 3000–2900 BC, broadly overlapping the onset of the continental Neolithic decline. Bluestone arrangements occurred in multiple episodes, while the sarsen trilithons and outer circle—the monument's defining features—were erected c. 2620–2480 BC. Further modifications continued into the Early Bronze Age.

Britain experienced its own signals of Neolithic decline around 3100–2900 BC (reduced activity in some regions), yet megalithic and henge traditions persisted longer in areas such as Wiltshire. Ancient DNA indicates that the British Neolithic population, derived largely from Early European Farmer ancestry with strong links to continental sources including Early Neolithic France, remained relatively stable until the rapid introduction and spread of Bell Beaker material culture and substantial steppe-related ancestry around 2450–2200 BC. This resulted in ~90% or greater replacement of local Neolithic ancestry within a few centuries. Individuals from the Stonehenge landscape, including the Amesbury Archer (buried c. 2300 BC with continental isotopic signatures), illustrate mobility during this transitional period.

4. Applicability and Regional Variation

The Bury discontinuity precedes Stonehenge's sarsen phase by 400–500 years and involves an intra-Neolithic farmer ancestry shift (northward expansion of Iberian-related groups), whereas Britain's major transformation involved steppe ancestry via Bell Beaker networks. Direct equivalence is therefore limited. The Paris Basin retained farmer-related ancestry longer before steppe admixture, while Britain shows a more complete and later replacement.

Nevertheless, the study offers valuable comparative context. It demonstrates that the cessation of continental megalith building coincided with genuine population discontinuity, pathogen circulation (including pre-Late Neolithic/Bronze Age Y. pestis), and agricultural contraction. Britain appears to have been partially buffered—possibly by insularity, lower population density, or different exposure dynamics—allowing continued capacity for large-scale construction. No pre-Beaker Y. pestis has yet been confirmed in British Neolithic samples (earliest known British cases date to c. 2000 BC in Beaker-associated contexts), but the continental evidence strengthens the case for targeted pathogen screening of British collective burials, such as long barrows.

Stonehenge's persistence may reflect multiple, non-exclusive factors: relative demographic resilience, ritual responses to stress that reinforced social cohesion, or regional intensification of monument building. Kinship data from British sites such as Hazleton North (Fowler et al. 2022), showing patrilineal multi-generational groups, suggest social structures capable of sustaining inter-generational projects, though severe disruption could still impair labour organisation. The sarsen phase thus represents a late expression of Neolithic monumental traditions in a European landscape already transformed in many regions.

5. Discussion and Conclusions

Seersholm et al. (2026) establish that late fourth-millennium population changes in the Paris Basin involved replacement, disease, and environmental shifts linked to the end of megalithic tomb traditions on the continent. Applied to Stonehenge, these findings highlight the monument's anomalous continuation amid broader instability, without implying identical processes. The sarsen builders operated in a context shaped by the same pan-European pressures—pathogens, demographic flux, and mobility—but followed a trajectory that preserved Neolithic farmer ancestry longer before the Beaker-related turnover.

This reading positions Stonehenge not as an untroubled apogee but as a remarkable regional phenomenon: a final major flourish of a tradition waning elsewhere, potentially erected under conditions of underlying stress yet demonstrating organisational continuity. Future integrated aDNA and pathogen studies from British Neolithic assemblages, combined with refined continental datasets, will clarify the extent of shared versus divergent dynamics across northwestern Europe.

References

Darvill, T. et al. (2012) Stonehenge remodelled. Antiquity 86, 1021–1040.

Fowler, C. et al. (2022) A high-resolution picture of kinship practices in an Early Neolithic tomb. Nature 601, 584–587.

Olalde, I. et al. (2018) The Beaker phenomenon and the genomic transformation of northwest Europe. Nature 555, 190–196.

Parker Pearson, M. et al. (various works on Stonehenge chronology, including 2012 models).

Seersholm, F.V. et al. (2026) Population discontinuity in the Paris Basin linked to evidence of the Neolithic decline. Nature Ecology & Evolution. https://doi.org/10.1038/s41559-026-03027-z.

Seersholm, F.V. et al. (2024) Repeated plague infections across six generations of Neolithic farmers. Nature 632, 114–121.

Shennan, S. et al. (2013) Regional population collapse followed initial agriculture booms in mid-Holocene Europe. Nature Communications 4, 2486.

Outwash Fan-Heads, Composite Dams, and the Fremington Lake Level

Why 30 m of Water Doesn’t Require 30 m of Ice

Tim Daw — sarsen.org — April 2026

(My thanks to Brian John for nudging me to put this draft article up - I emphasis it is a draft and not fully checked, treat the reference section with caution, but the geology and implications are robust, and publication is needed to counter misinformation)

Abstract

If the Fremington Clay was deposited in a proglacial lake at approximately 30 m OD in the Taw–Torridge valley, does that necessarily require an ice dam of equivalent or greater height — and therefore thick onshore ice capable of depositing erratics at 80 m on Baggy Point? This paper argues no. Drawing on well-documented modern proglacial analogues from New Zealand, Svalbard, and Patagonia, it shows that lake damming in high-sediment-load settings is routinely achieved by composite ice-sediment barriers whose effective height far exceeds the thickness of the ice itself. Applied to the Taw estuary, this model explains the Fremington lake level without requiring the extensive onshore ice cover that a monolithic ice-dam interpretation implies. The Ramson Cliff erratic at ~80 m OD remains isolated and unexplained by this mechanism.

1. Introduction: The Ice Dam Height Problem

The Fremington Clay Series, exposed along the south shore of the Taw–Torridge estuary in North Devon, has been the subject of sustained debate since Maw (1864) first described its thick sequence of stoneless clay, laminated silts, and basal gravels. Following the influential work of Stephens (1966)[1] and the detailed sedimentological study of Croot et al. (1996),[2] the consensus interpretation has been that the upper, stoneless clay unit — sometimes called the potter’s clay — was deposited in a proglacial lake dammed by an Irish Sea ice lobe that impinged on Barnstaple Bay. Bennett et al. (2024)[3] broadly accept this model, placing the lake surface at approximately 30 m OD.

A criticism raised against any attempt to downplay the extent of Irish Sea ice in North Devon is that a 30 m lake requires a dam of at least 30 m, and an ice dam must be topped by an ice surface higher still. If the ice was that thick at the estuary mouth, runs the argument, then it could easily have extended laterally to emplace the isolated erratic found at ~80 m OD on Ramson Cliff, Baggy Point — making that erratic unremarkable rather than anomalous.

This reasoning is intuitively appealing but geomorphologically naïve. It conflates two distinct questions: the local height of a barrier needed to impound water in a topographically constricted valley, and the lateral extent and thickness of onshore ice required to transport and deposit a grounded erratic on an exposed coastal headland several kilometres to the west. This paper addresses the first question by examining what modern proglacial analogues actually tell us about the mechanics of ice-margin lake damming, and then considers the implications for the second.

2. The Fremington Proglacial Lake: Stratigraphy and Setting

The Fremington Clay Series comprises a tripartite sequence: basal gravels containing both local Devonian lithologies and far-travelled erratics, overlain by laminated silts and sands, grading upward into the distinctive stoneless blue-grey clay that gives the formation its name. Dropstones within the clay indicate the presence of floating ice, and the overall fining-upward sequence is consistent with a deepening lake environment. The full thickness of the deposit reaches approximately 30 m in the type sections between Fremington and Penhill.

The lake model, as developed by Croot et al. (1996) and refined by subsequent workers, envisages an Irish Sea ice lobe advancing into or across Barnstaple Bay, blocking the combined drainage of the Taw and Torridge rivers. No discrete terminal moraine has been identified as the dam; the barrier is attributed to the ice lobe itself, with minor overriding of lake sediments during a possible readvance phase. The reconstructed lake surface of ~30 m OD is inferred from the upper limit of lacustrine deposits and the topography of the valley.

2.1 The Bickington–Hele Ridge

An important but often overlooked topographic feature is the Bickington–Hele ridge, a bedrock high on the southern side of the valley that rises to approximately 55 m OD.[4] This ridge is not a glacial constructional feature. It is formed by the underlying Devonian and Carboniferous slates, shales, and sandstones, shaped by long-term differential erosion. Several metres of sand and gravel containing erratic stones — the so-called Hele gravels — cap the ridge and are generally interpreted as glaciofluvial outwash deposited by meltwater associated with the same ice phase that influenced the Fremington sequence.

In practical terms, the ridge acted as a lateral confining feature on the southern margin of the lake. It helped constrict the ponded area and contributed to maintaining the water level, but it was not the primary dam. The northern barrier — blocking the estuary — remained the ice lobe and whatever associated sediment accumulated against it. This distinction between lateral confinement by pre-existing topography and primary damming by an ice-margin barrier becomes critical when assessing what kind of ice was actually needed.

3. Composite Ice-Sediment Dams: The Modern Evidence

Modern proglacial geomorphology shows that this clean-ice model is the exception rather than the rule. In most settings where glaciers interact with rivers carrying significant sediment loads, the actual barrier is a composite structure: ice, outwash gravel, glaciofluvial sand, debris flows, and — in marine-terminating cases — subaqueous fans. The effective dam height is the combined elevation of all these components, not the ice thickness alone.

3.1 The New Zealand Last Glacial Maximum: Outwash Fan-Head Damming

The most comprehensive modern analogue comes from the Southern Alps of New Zealand, where Sutherland et al. (2019)[5] reconstructed the ice-contact proglacial lake systems associated with the Last Glacial Maximum across the entire mountain range. Their central finding is that the major LGM lakes — Tekapo, Pukaki, Ohau, Wanaka, Hawea, and Wakatipu — were not dammed primarily by moraine ridges or by monolithic ice barriers, but by outwash fan-heads: massive aggradational gravel bodies built by high-sediment-load braided rivers against and around the ice margin.

Sutherland et al. (2022)[6] developed this model in detail at Lake Tekapo, where the glacier terminus was entirely buried by the outwash fan-head, with no terminal moraine visible as a discrete landform. The ‘moraine’ at Tekapo is in fact a subdued ridge coalesced into the fan surface. The dam that created the lake — which persists today, long after the ice vanished — was the sedimentary mass itself. Shulmeister et al. (2019)[7] provide wider context, noting that this pattern is typical across the western South Island, where high sediment supply from the actively eroding Alps overwhelmed the capacity of terminal moraines to form as distinct features.

The key insight for the Fremington debate is that these New Zealand lakes stood at substantial depths behind barriers whose ice component was only a fraction of the total dam height. Once the ice melted, the lakes persisted — Tekapo, Pukaki, and Ohau still exist today — held in by the residual sediment mass. The ‘permanence’ of the lakes, as Sutherland et al. put it, derives from the sediment, not the ice.

3.2 High-Arctic Svalbard: Rapid Infilling and Stabilisation

A second class of evidence comes from High-Arctic Svalbard, where Strzelecki et al. (2017)[8] documented the life cycle of a small ice-dammed lake at the Nordenskiöldbreen glacier margin between 1990 and 2012. The lake formed when the glacier retreated and dammed a tributary valley, but within approximately two decades it was almost completely infilled by a Gilbert-type fan delta advancing from the sediment-laden tributary stream.

What is relevant here is not the infilling per se, but the process: fluvial sediment rapidly aggraded against the ice barrier, building a composite dam whose crest exceeded the ice surface in places. The effective barrier height fluctuated as sediment accumulated, eroded, and redistributed. At no point was the dam a simple wall of ice with a measurable freeboard; it was a dynamic, composite structure in which ice and sediment alternately dominated.

The Svalbard case demonstrates that this process operates even at very small scales and over very short timescales. In a Pleistocene setting with larger ice volumes, greater sediment supply, and longer duration, the composite damming effect would be amplified.

3.3 Perito Moreno, Patagonia: Ice, Debris, and Dynamic Water Levels

Crucially, the lake level in Brazo Rico during damming episodes rises well above what the subaerial ice thickness at the narrow strait would predict, precisely because the underwater and debris-armoured components of the barrier contribute to the effective seal. Leakage occurs not by overtopping but by subglacial drainage when the hydraulic head exceeds the ice overburden pressure — a process described in the broader GLOF (glacial lake outburst flood) literature by Carrivick & Tweed (2013).[9]

3.4 A General Principle

These three examples — from a temperate maritime mountain belt, a High-Arctic archipelago, and a Patagonian ice field — illustrate a general principle that has been well established in the proglacial lake and GLOF literature but has not been applied to the Fremington debate: in any setting where rivers carry significant sediment loads to an ice margin, the resulting dam is composite rather than monolithic. The effective barrier height is the sum of ice thickness, aggraded outwash, moraine material, and subaqueous sediment, not the ice alone.

Carrivick et al. (2022)[10] synthesise this principle in their review of coincident glacier and lake evolution across the Southern Alps, showing that sediment flux is as important as ice dynamics in determining lake existence, extent, and longevity. In their analysis, ice-marginal lakes are fundamentally sedimentary features as much as they are glaciological ones.

4. Application to the Taw Estuary

With this framework in place, consider the specific conditions of the Taw–Torridge system at the time the Fremington Clay was deposited.

4.1 Sediment Supply from a Periglacial Hinterland

The Taw and Torridge drain catchments that include the northern margins of Dartmoor and the extensive periglacial plateau surfaces of Exmoor. Evans et al. (2012)[11] describe Dartmoor during Pleistocene cold stages as an independent ice cap with extensive periglacial slopes generating clitter fields, solifluction mantles, and thick head deposits. Edmonds (1972)[12] documents the terrace stratigraphy of the Taw valley, recording multiple episodes of fluvial aggradation consistent with high-sediment-load braided river systems.

This is a critical observation. The rivers feeding the Fremington lake basin were not clear-water streams draining stable, vegetated catchments. They were periglacial braided rivers with extremely high bedload transport rates, carrying gravel, sand, and silt derived from frost-shattered bedrock, solifluction deposits, and reworked older drift. The analogy with New Zealand’s high-sediment-load glacial rivers is direct and compelling: these are exactly the conditions under which outwash fan-heads build against ice margins.

4.2 The Composite Dam Model for Fremington

The Fremington Clay itself arguably records this process. The basal gravels — containing both local and far-travelled lithologies — represent the initial outwash phase, when coarse bedload was deposited in a proglacial setting. The overlying laminated silts and sands record a transitional phase as the basin deepened and the dam became more effective, trapping finer sediment. The uppermost stoneless clay represents full lacustrine conditions, with suspension settling of the finest fraction in a quiet, deep-water environment. Dropstones from floating ice derived from the margin punctuate this otherwise tranquil record.

In this model, the 30 m lake surface does not require 30 m of solid ice. It requires a composite barrier — ice plus aggraded sediment plus debris — whose combined crest reached or exceeded 30 m OD at the topographically constricted estuary mouth. The pre-existing Bickington–Hele ridge at ~55 m OD provided lateral confinement on the southern side, preventing the lake from draining southward. The ice component of the dam may have been substantially less than 30 m thick, supplemented by the very outwash and debris that the Fremington sequence itself records.

4.3 Consistency with the Sedimentary Record

This interpretation is not ad hoc. The fining-upward sequence of the Fremington Clay Series — basal gravels to laminated silts to stoneless clay — is precisely what the outwash fan-head model predicts. In the New Zealand analogues, Sutherland et al. (2022) describe ice-contact lakes with identical stratigraphic signatures: coarse outwash at the base grading upward into laminated lacustrine fines as the sediment trap becomes more effective and the lake deepens. The Fremington Clay is the sedimentary product of the process that created its own dam.

This self-reinforcing dynamic — in which the lake deposits contribute to the barrier that impounds the lake — is a standard feature of composite ice-sediment dams. It means that the dam grows in effectiveness over time without requiring any increase in ice thickness. A modest ice lobe that initially created a shallow, gravel-floored pond could, through progressive outwash aggradation, give rise to a deep lacustrine basin over decades or centuries.

5. Implications for the Ramson Cliff Erratic

The Ramson Cliff erratic sits at approximately 80 m OD on Baggy Point, a fully exposed coastal headland several kilometres west of the Taw estuary. Its petrography — an altered epidiorite or greenstone of approximately 700 kg — has been re-examined by Daw, Ixer & Madgett (2026),[13] who argue that it aligns with local Cornubian or Dartmoor sources rather than distant Irish Sea material. But even setting aside the provenance question, the erratic at 80 m OD is not explained by a composite dam at 30 m in the valley below.

The argument that ‘the ice must have been higher than the lake’ is true in a trivial sense — the ice surface at the dam must exceed the water surface, or the lake drains. But the dam was at the estuary mouth, in a topographically constricted setting. The ice there could have been 35 m thick (to maintain a few metres of freeboard above a 30 m lake) without extending laterally at that thickness along 10 km of open coastline to reach Baggy Point at 80 m. These are geometrically and physically distinct situations. A lobate ice margin blocking an estuary is not a uniform ice sheet overriding a headland.

The Ramson Cliff erratic therefore remains isolated and anomalous: well above both the reconstructed lake level and the ~55 m Bickington–Hele ridge, with no supporting suite of high-level deposits, striae, or erratics along the intervening coast. Whatever mechanism emplaced it, the Fremington lake level provides no evidence for that mechanism.

6. Discussion: Separating Dam Height from Ice Extent

In every modern analogue considered here — New Zealand, Svalbard, Patagonia — the dam height at the constriction point exceeds the ice thickness by a substantial margin because of sediment aggradation. The ice does not need to be thick everywhere, or present everywhere, or even the dominant component of the dam. It needs to provide a nucleus around which sediment accumulates. This is particularly true in high-sediment-load systems like periglacial braided rivers, where outwash fan construction is rapid and volumetrically significant.

7. Conclusion

The Fremington Clay records a proglacial lake at approximately 30 m OD in the Taw–Torridge valley. The barrier that impounded this lake need not have been a monolithic wall of ice. Modern analogues from New Zealand, Svalbard, and Patagonia demonstrate that composite ice-sediment dams are the norm in high-sediment-load proglacial settings, and that effective dam heights routinely exceed the thickness of the ice component. Applied to the Taw estuary, a modest Irish Sea ice lobe supplemented by rapid outwash fan-head aggradation from periglacial rivers provides a sufficient and parsimonious explanation for the lake level.

Dam height in the valley and the lateral extent of onshore ice are separate questions with separate answers. The Fremington lake level does not require extensive glaciation of the North Devon coast, and it does not explain the Ramson Cliff erratic at 80 m OD on Baggy Point. That erratic remains an isolated outlier demanding its own explanation — an explanation that the composite dam model, applied properly, neither provides nor requires.

The story advances through careful scrutiny of the rocks, the topography, and the modern analogues — not through assumptions about uniform ice dams.[14]

References

Bennett, J.A., Cullingford, R.A., Gibbard, P.L., Hughes, P.D. & Murton, J.B. (2024). The Quaternary Geology of Devon. Proceedings of the Ussher Society 15: 84–130.

Carrivick, J.L. & Tweed, F.S. (2013). Proglacial lakes: character, behaviour and geological importance. Quaternary Science Reviews 78: 34–52.

Croot, D.G., Gilbert, A., Griffiths, J. & van der Meer, J.J. (1996). The character, age and depositional environments of the Fremington Clay Series, North Devon. Quaternary Newsletter 80: 1–15.

Daw, T. (2025). Caution in Attributing the Fremington Clay Series to Irish Sea Glaciation: A Case for Predominantly Fluvial and Periglacial Origins in North Devon. ResearchGate / Academia.edu.

Daw, T., Ixer, R. & Madgett, P. (2026). A review of the Ramson Cliff erratic: evidence of high-level ice flow? Quaternary Newsletter 167: 13–19. https://doi.org/10.64926/qn.20517

Edmonds, E.A. (1972). Terrace stratigraphy in the Taw valley. Exeter Museums Archaeological Field Unit.

Geological Conservation Review volume: Quaternary of South-West England.

Stephens, N. (1966). Some Pleistocene deposits in North Devon. Biuletyn Peryglacjalny 15: 103–114.

Strzelecki, M.C., Long, A.J. & Lloyd, J.M. (2017). Rise and fall of a small ice-dammed lake — Role of deglaciation processes and morphology. Geomorphology 295: 228–243.

[1]Stephens, N. (1966). Some Pleistocene deposits in North Devon. Biuletyn Peryglacjalny 15: 103–114.

[2]Croot, D.G., Gilbert, A., Griffiths, J. & van der Meer, J.J. (1996). The character, age and depositional environments of the Fremington Clay Series, North Devon. Quaternary Newsletter 80: 1–15.

[3]Bennett, J.A., Cullingford, R.A., Gibbard, P.L., Hughes, P.D. & Murton, J.B. (2024). The Quaternary Geology of Devon. Proceedings of the Ussher Society 15: 84–130.

[4]Geological Conservation Review volume: Quaternary of South-West England (describes the Bickington–Hele ridge and capping deposits).

[5]Sutherland, J.L., Carrivick, J.L., Shulmeister, J., Quincey, D.J. & James, W.H.M. (2019). Ice-contact proglacial lakes associated with the Last Glacial Maximum across the Southern Alps, New Zealand. Quaternary Science Reviews 213: 67–92.

[6]Sutherland, J.L., Carrivick, J.L., Shulmeister, J., Quincey, D.J. & James, W.H.M. (2022). A model of ice-marginal sediment-landform development at Lake Tekapo, Southern Alps, New Zealand. Geografiska Annaler: Series A 104(3): 182–209.

[7]Shulmeister, J., Thackray, G.D., Rittenour, T.M., Fink, D. & Evans, D.J.A. (2019). The Last Glacial Maximum (LGM) in western South Island, New Zealand. Quaternary Science Reviews 213: 44–66.

[8]Strzelecki, M.C., Long, A.J. & Lloyd, J.M. (2017). Rise and fall of a small ice-dammed lake — Role of deglaciation processes and morphology. Geomorphology 295: 228–243.

[9]Carrivick, J.L. & Tweed, F.S. (2013). Proglacial lakes: character, behaviour and geological importance. Quaternary Science Reviews 78: 34–52.

[10]Carrivick, J.L., Sutherland, J.L., Huss, M., Purdie, H., Stringer, C.D., Grimes, M., James, W.H.M. & Lorrey, A.M. (2022). Coincident evolution of glaciers and ice-marginal proglacial lakes across the Southern Alps, New Zealand. Global and Planetary Change 211: 103792.

[11]Evans, D.J.A., Harrison, S., Vieli, A. & Anderson, E. (2012). The glaciation of Dartmoor: the southernmost independent Pleistocene ice cap in the British Isles. Quaternary Science Reviews 45: 31–53.

[12]Edmonds, E.A. (1972). Terrace stratigraphy in the Taw valley. Exeter Museums Archaeological Field Unit.

[13]Daw, T., Ixer, R. & Madgett, P. (2026). A review of the Ramson Cliff erratic: evidence of high-level ice flow? Quaternary Newsletter 167: 13–19. https://doi.org/10.64926/qn.20517

[14]Daw, T. (2025). Caution in Attributing the Fremington Clay Series to Irish Sea Glaciation: A Case for Predominantly Fluvial and Periglacial Origins in North Devon. ResearchGate / Academia.edu.