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

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 in the Fremington debate.

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.

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)[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

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).[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

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 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 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),[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

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.

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.

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. (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.

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.

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.

---

[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.