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

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.

The Breast on Stone 16: An Update

Tim Daw  |  www.sarsen.org  |  2026

 

The Original Observation (2015)

In March 2015 I noted a feature on Stone 16 of the outer sarsen circle at Stonehenge that I could not explain. The north-eastern face of the stone — the side visible from inside the monument, on the line of the midwinter sunset alignment — has been finely pecked to a smooth, straight surface. The dressing is meticulous. Yet at just below head height, a rounded protrusion was deliberately retained. Its outermost part appears to have been broken off at some point, but there is no question that it was left in place while the rest of the face was worked flat around it.

I cautiously suggested that the protrusion resembled a breast, while acknowledging the risk of pareidolia. An update to the original post drew attention to the Kerloas menhir at Plouarzel in Brittany, where two projecting bosses were deliberately left on an otherwise smoothed 9.5-metre standing stone, and were later associated with fertility rites.

In the decade since, new evidence has come to light — particularly from Cornwall — that strengthens the case for reading Stone 16’s protrusion as an intentional breast motif, and allows a more developed speculation about the stone’s role within the monument.

Comparanda: Breast Motifs on Atlantic Megaliths

Boscawen-ûn, Cornwall. The most significant British parallel emerged in the same year as my original post. In July 2015, the archaeologist and 3D specialist Tom Goskar recorded the central leaning stone at Boscawen-ûn stone circle near St Buryan using photogrammetry. His digital surface analysis revealed that the features on the north-east (inner leaning) face, previously interpreted as a pair of axe-head carvings, are more probably a pair of feet carved in low relief, soles outwards, with a row of toes discernible on the right foot. More striking still, roughly 500mm above the feet, Goskar identified a pair of circular features, also in low relief, which he noted “appear very similar to carvings interpreted as breasts on some allées couvertes in Brittany (Tressé, Prajou-Menhir, etc.)” (Goskar 2015).

Goskar’s interpretation is that the Boscawen-ûn central stone may have been reused from a dismantled chambered tomb that incorporated decorated stones in the Breton style — a practice well attested in Brittany itself (Scarre 2011, p.147). He has since identified a second set of possible breast motifs at Carn Leskys near St Just, describing the Boscawen-ûn pair as “the first set of breast motifs that I have found in Cornwall” (Goskar 2024). These are, as far as I am aware, the only identified examples of the Breton-style breast motif in mainland Britain.

The Kerloas menhir, Plouarzel, Finistère. This remains the closest technical parallel to Stone 16. The Kerloas menhir, the tallest standing menhir in Europe at 9.5 metres, was deliberately smoothed but retains two hemispherical protuberances about one metre from the ground, on opposite faces (east and west). The 18th-century antiquary Jacques Cambry described them as “espèces de mamelles” (a kind of breast), while Charles Blin also considered them feminine in character. According to tradition, newly-wed couples would come naked to the stone and rub their bellies against the bosses to ensure fertility. The parallel with Stone 16 lies not in carved relief but in the retention of a protrusion on an otherwise dressed surface — the same deliberate act of leaving a boss intact while working the surrounding stone flat.

The Guernsey statue-menhirs. Two anthropomorphic menhirs on Guernsey — the Castel Church statue-menhir and La Gran’mère du Chimquière at St Martin’s — are relevant as intermediate examples between Brittany and Britain. The Castel statue-menhir, a granite slab approximately 1.65 metres high found under the church chancel in 1878, bears carved breasts and a necklace. These features are shared with statue-menhirs in northern France, Corsica, and the wider western Mediterranean. The La Gran’mère, originally a Neolithic or early Bronze Age standing stone, was later re-carved with a head and shoulders, probably in the Celtic or Roman period. Both demonstrate that the breast motif was current in the Channel Islands — the cultural midpoint between Brittany and southern Britain.

Breton allées couvertes. The breast-on-stone motif is most densely concentrated in the gallery graves of Brittany and the Paris Basin. The allée couverte at Tressé (La Maison des Feins) preserves multiple breast-pairs with collars carved in high relief on sidestones. Similar paired bosses appear at Prajou-Menhir and Kergüntuil. In the gallery graves of the Seine-Oise-Marne culture, breast imagery appears alongside axes and necklaces on tomb orthostats, notably in the Petit Morin valley. The consistent association is with funerary contexts: the breasts appear at the threshold or within the chamber, at the boundary between the living and the dead.

Across all of these examples, one pattern is emphatic: breasts on megaliths almost invariably appear in pairs. Bilateral symmetry is the norm. This has direct implications for Stone 16.

Stone 16 and Stone 15: A Paired-Breast Hypothesis

Stone 16 sits on the south-western arc of the sarsen circle, on the primary solstitial axis — the line connecting the midsummer sunrise to the north-east with the midwinter sunset to the south-west. The breast-like protrusion is on its north-eastern face: the side facing the interior of the monument, and the direction from which the last light of the midwinter sun would have entered the circle before setting between the uprights of the Great Trilithon.

Stone 15, its immediate neighbour to the south, survives only as a small fragment. Its stonehole and former position are marked on all modern plans, but the surviving piece preserves nothing of the original dressed faces. We cannot know how those faces were finished or whether they carried any features. What we can say is that Stones 15 and 16, standing side by side on the south-western arc, would have flanked the view from inside the circle toward the midwinter sunset.

Given that the Atlantic Neolithic breast motif overwhelmingly appears in bilateral pairs, it is reasonable to ask whether Stone 15 once carried a matching protrusion on its own north-eastern face. If so, the two stones together would have presented a pair of breasts to anyone standing inside the circle and looking toward the solstitial sunset — a pair of retained bosses on otherwise finely dressed surfaces, flanking the solar corridor.

Interpretation: Fertility at the Solstitial Threshold

The placement of such a motif on the solstitial alignment would not be anomalous. It would, in fact, be entirely consistent with the broader Neolithic use of breast imagery.

In the Breton allées couvertes, breasts appear at the threshold of the burial chamber — the boundary between the world of the living and the world of the dead. In passage graves, the corridor itself functions as a symbolic birth canal, and the chamber as a womb; breast imagery near the entrance reinforces the association with nurture and regeneration. At the Kerloas menhir, the breast-bosses are associated with fertility rites in which the stone mediates between human aspiration and cosmic process.

At Stonehenge, the midwinter sunset marks the nadir of the solar year: the shortest day, the moment when the sun “dies.” English Heritage’s own analysis emphasises that midwinter, rather than midsummer, was probably the more important focus for the builders, given the monument’s orientation and the evidence for large-scale midwinter feasting at Durrington Walls. The midwinter sunset, sinking between the uprights of the Great Trilithon and falling onto the Altar Stone, was the central dramatic event of the monument.

To place a breast — or a pair of breasts — at the inner threshold of the circle, on the solstitial line, would have been to frame that moment of solar death with a symbol of nurture and renewal. This is precisely the logic we see at Newgrange, where the midwinter sunrise penetrates a womb-like chamber, and at the Breton passage graves, where breast imagery guards the passage between death and regeneration. The builders of Stonehenge were participants in the same Atlantic cultural world that produced the breast carvings at Tressé, Prajou-Menhir, and now Boscawen-ûn. It would be more surprising if that symbolic vocabulary were entirely absent from the monument.

Caveats

Several important caveats must be stated. First, Stone 16’s protrusion has not been subjected to the kind of photogrammetric or RTI analysis that Goskar applied at Boscawen-ûn. A 3D surface survey would help to establish whether the feature was shaped or merely retained, and how it compares dimensionally and morphologically with the Kerloas bosses and the Boscawen-ûn reliefs.

Second, Stone 15 survives only as an uninformative fragment. The paired-breast hypothesis therefore rests on analogy with the wider tradition rather than on direct evidence from the missing stone. It is an inference, not a demonstrated fact.

Third, the Boscawen-ûn breast identifications, while produced by rigorous digital methods, remain tentative. Goskar himself frames them as features that “appear very similar to” Breton examples, not as certain identifications. The coarse Lands End granite makes fine detail difficult to resolve.

Fourth, Stone 16’s protrusion is a single feature on a single face, not a carved pair. If the paired-breast reading is to hold, it requires the cooperation of a stone that no longer exists. This is speculative archaeology in the fullest sense.

Conclusion

The protrusion on Stone 16 is not a natural accident left by careless workmanship. The stone’s north-eastern face was dressed to a standard of finish that demonstrates complete control of the material. The boss was retained by choice. Its position — on the solstitial alignment face, at roughly head height, on the inner side of the circle — is consistent with the placement of breast motifs in the broader Atlantic Neolithic tradition: at thresholds, on alignment faces, in contexts associated with death and regeneration.

With the identification of probable breast carvings at Boscawen-ûn in 2015 and at Carn Leskys in 2024, the breast motif is no longer confined to Brittany and the Channel Islands. It is present in Cornwall, within the same broadly Atlantic cultural zone that Stonehenge’s builders inhabited. The case for reading Stone 16’s protrusion as an intentional breast, and for hypothesising a matching feature on the largely lost Stone 15, is stronger than it was a decade ago.

A photogrammetric survey of Stone 16’s north-eastern face would be a worthwhile next step.

Photographs - Click to embiggen

 

 

References

Goskar, T. A. (2015). “Neolithic Breton-Style Rock Art at Boscawen-ûn Stone Circle.” tom.goskar.com, 14 September 2015.

Goskar, T. A. (2024). “A Cornish Rock Art Discovery at Carn Leskys — the Carn of Burnings.” tom.goskar.com, 25 June 2024.

Scarre, C. (2011). Landscapes of Neolithic Brittany. Oxford: Oxford University Press.

Shee Twohig, E. (1981). The Megalithic Art of Western Europe. Oxford: Clarendon Press.

L’Helgouac’h, J. (1995). “La sculpture des mégalithes armoricains.” In Scarre, C. & Healy, F. (eds), Trade and Exchange in Prehistoric Europe. Oxford: Oxbow.

Kinnes, I. (1995). “Statue-menhirs and allied representations in Northern France and the Channel Islands.” In Casini, S., de Marinis, R. C. & Pedrotti, A. (eds), Statue-Stele e massi Incisi nell’Europa dell’Età del Rame. Notizie Archeologiche Bergomensi 3, pp. 131–41.

Kohring, S. (2015). “Stepping Stones: Art and Community on Prehistoric Guernsey, Channel Islands.” Oxford Journal of Archaeology 34(3), pp. 301–16.

The Ox That Moved a Mountain — or Didn't?

 

How a New Study on Cattle Traction Reshapes a Speculative Hypothesis

A cow that walked from Wales to Salisbury Plain five thousand years ago may have helped drag the bluestones to Stonehenge. Or she may simply have been a milk cow who calved at the wrong moment. A landmark 2026 study has just made the second interpretation more likely — but it has also, almost accidentally, handed us the tools to settle the question.

The hypothesis rests on a single tooth. Excavated from Stonehenge's ditch and dated to around 3350–2920 BC, the third molar belonged to a young female Bos taurus whose strontium isotopes trace a journey from the Preseli Hills in Wales to the chalk downlands of Wessex — exactly the route taken by the famous bluestones. Sequential sampling of the enamel revealed a sharp spike in lead (Pb) during the final months of the cow's life. The original researchers attributed this to the metabolic upheaval of calving or lactation, which mobilises lead stored in bone. But an alternative reading is possible: the spike could record fracture healing or intense physical stress caused by harness-related injury during heavy draught work. If so, this animal wasn't just walking the bluestone route — she was working it.

For that idea to hold, Neolithic farmers would have needed to yoke cattle — probably cows rather than specialised oxen — and use them for heavy labour at a time when organised draught technology is otherwise invisible in Britain. This is where the new evidence arrives.

What Liu and Albarella Found

In the most systematic osteological investigation of British cattle traction yet published, Phoebe Liu and Umberto Albarella examined metapodials and phalanges from 22 archaeological sites spanning the Neolithic to the post-Medieval period. Fourteen of these were core pre-Iron Age sites; eight were later comparators from the Iron Age, Roman, and Medieval periods that provide positive controls for known traction use. The authors applied two independent methods: a modified Pathological Index (mPI) that detects bone stress, and biometric shape ratios — particularly the Bd/GLpe ratio in first phalanges — that distinguish draught from non-draught animals.

The results are clear. Across Early, Middle, and Late Neolithic assemblages — including sites contemporary with bluestone transport such as Durrington Walls and the Orcadian settlements of Skara Brae and Links of Noltland — pathological changes are minimal and bone shapes match modern non-working cattle. Even where mature animals are well represented, as at Runnymede (60% adult/elderly) and Durrington Walls (58%), mPI values remain low. Only in the Middle to Late Bronze Age (1600–700 BC) do both methods converge on clear evidence of regular draught use, most prominently at Clay Farm and Potterne. The authors conclude that animal traction, at least at an intensity sufficient to leave skeletal traces, arrived in Britain centuries after the bluestones.

Importantly, the paper also challenges one of the key supporting references for Neolithic traction in the British Isles. The 2023 Kilshane study, which proposed draught oxen in Ireland from around 3600 BC, used comparative Pathological Index values drawn from second rather than first phalanges — making the comparison unreliable. This weakens what had been one of the stronger pieces of circumstantial evidence for early cattle traction in the region.

What This Means for the Hypothesis

Does this kill the idea of Neolithic cattle hauling bluestones? Not quite — and thinking in Bayesian terms helps explain why.

We begin with a prior probability that should be described honestly: moderately low, but not negligible. It rests on the cow's Welsh provenance matching the bluestone source, the ritual deposition of her jaw at Stonehenge, the lead spike in her enamel, and continental evidence that cattle were occasionally used for light traction elsewhere in Neolithic Europe. None of this is proof, but it amounts to more than idle speculation.

The likelihood of finding clear osteological evidence, however, was always going to be low if the work was rare, short-lived, and performed by unspecialised animals. Picture the scenario concretely: not a standing workforce of trained oxen ploughing fields season after season, but a small number of cows yoked for days or weeks during an exceptional communal project — hauling stones over a journey of roughly 250 kilometres, with frequent rest stops, across terrain that would have demanded intermittent rather than continuous effort. Neolithic farming in southern Britain was small-scale and intensive; cows were primarily valued for milk, and the energetic cost of maintaining working animals would have been tolerated only for extraordinary purposes. In such a scenario, we would expect little or no detectable skeletal damage at the population level — exactly what Liu and Albarella found. Their method is calibrated against modern draught oxen performing sustained, heavy work; as they themselves note, it is mainly designed to identify draught oxen, and an earlier use of cows for traction cannot be excluded.

The posterior probability therefore shifts downward, but not to zero. The absence of evidence is genuinely informative: it makes routine, organised cattle haulage of multi-tonne stones less plausible than we might once have thought. But it is not decisive evidence of absence. Light, opportunistic use of a few animals over a short period could have occurred without leaving the kind of widespread bone remodelling that survives in the archaeological record five millennia later.

Liu and Albarella's broader archaeological narrative actually reinforces one dimension of the hypothesis by a different route. They argue persuasively that the Late Neolithic in southern Britain was predominantly pastoral, with arable farming in decline — meaning there was no agricultural motivation for routine traction. But that is precisely the point: the bluestone transport hypothesis doesn't require routine traction. It requires an exceptional effort for an exceptional project. The paper is about farming economies and regular draught use; Stonehenge was neither regular nor ordinary.

A Way to Test It

This is where the story becomes more interesting than a simple verdict of "weakened but not refuted." The Liu and Albarella study doesn't just update our beliefs — it equips us with the material and methods to test the hypothesis more rigorously than ever before.

Their project includes mandibular teeth from the Neolithic and Bronze Age cattle they analysed, still curated in museums across Britain. Applying the identical sequential enamel micro-sampling technique used on the Stonehenge cow tooth — the method developed by Evans et al. (2025) — would allow a direct comparison of lead profiles across three groups:

Neolithic teeth with no traction signatures would establish the baseline. If Pb spikes appear in these animals, they are most likely linked to reproduction, seasonal nutritional stress, or other non-labour causes. This tells us what "normal" looks like for a Neolithic cow that wasn't hauling stones.

Bronze Age teeth with confirmed traction stress — from sites like Potterne and Clay Farm, where both pathological and biometric evidence converge on draught use — would serve as positive controls. If labour-induced bone mobilisation produces a distinctive Pb signature, it should show up here.

The Stonehenge tooth can then be compared against both groups. If its Pb spike is indistinguishable from the Neolithic baseline, the haulage interpretation loses ground. If it is anomalous — sharper, differently patterned, or more consistent with the Bronze Age draught profile — the hypothesis gains credible, testable support.

The raw data from the Liu and Albarella study are publicly available on Mendeley, and the relevant teeth are already in museum collections. This is not a speculative call for some future technological breakthrough; it is a logistically feasible, relatively low-cost follow-up that could be undertaken with existing methods and existing material. It would also address a broader archaeological question that the paper itself raises: when exactly did Britain adopt animal traction, and what does the transition look like at the level of individual animal biographies rather than population-level bone morphology?

The Sharper Question

The Liu and Albarella paper does two valuable things at once. It injects a necessary note of caution into romantic narratives of Neolithic oxen dragging bluestones across Wales, reminding us that the osteological evidence for organised traction in Britain begins later than the hypothesis requires. Yet it simultaneously provides the comparative framework — the dated, curated, analysed teeth from known traction and non-traction populations — that could transform the debate from speculation into data.

That cow walked from the Preseli Hills to Salisbury Plain around five thousand years ago. Her jawbone was placed in Stonehenge's ditch, and her tooth is still in a museum. The dataset she needs to be compared against now exists. The question is sharper than it has ever been. It just needs the right experiment.

References:

https://www.sarsen.org/2026/02/a-speculative-hypothesis-neolithic.html 

Liu, P., Albarella, U. The origins of animal traction in Britain: implications for technological and social developments in the Bronze Age. Archaeol Anthropol Sci 18, 83 (2026). https://doi.org/10.1007/s12520-026-02455-z

The Stones Are Losing Their Pull - What's Going Wrong

(A guest Op-Ed)

For a family of four arriving at Stonehenge on a summer Saturday, the bill starts at roughly £80 before anyone has set foot near the stones.⁹ They will queue for a shuttle bus, walk a roped-off path at a polite distance from the monument, and be back in the car park inside two hours. Meanwhile, the Natural History Museum — which last year welcomed a record-breaking 7.1 million visitors — costs nothing at all.¹³ Increasingly, families are making exactly that calculation. And the numbers show it.

In 2025, Stonehenge welcomed 1,253,405 visitors — an 8 per cent fall on the previous year and roughly 22 per cent below its pre-pandemic peak of approximately 1.6 million in 2019.^2,19 It slipped to 25th in the Association of Leading Visitor Attractions rankings, behind Windsor Castle and RHS Garden Wisley.¹ This happened in a year when the broader sector managed a modest 2 per cent increase to 165 million total visits across 409 ALVA sites. Even that aggregate figure remained 7 per cent below 2019.¹ Stonehenge, in other words, is not merely failing to recover from the pandemic — it is actively falling behind in a sector that is itself still struggling.

This is not a blip. It is the visible symptom of a deeper malaise at English Heritage, the charity responsible for more than 400 historic sites across England. The organisation’s own 2024/25 Annual Report reveals that pay-to-enter sites collectively attracted only 5.7 million visitors against internal targets, contributing to financial strain so severe that it triggered a “reshaping programme.”⁷ That programme proposes at least 189 job losses — roughly 7 per cent of its 2,535-strong workforce — with cuts expected across curatorial, site management, marketing, and visitor operations roles.^3,17 A further 21 sites face winter closures, and another 22 are being reclassified to open only on select days.⁴ One site, English Heritage has noted, welcomed just 11 visitors over an entire November weekend.⁵

Stonehenge was supposed to be the ballast against this kind of storm. As English Heritage’s highest-revenue generator — a site that operates without government subsidy and funds conservation across the portfolio through ticket sales, membership, and secondary spend — it is the engine that keeps the rest of the machine running. When the engine falters, everything downstream suffers. And English Heritage’s own board knows it is faltering. The Trust Board minutes from March 2025 recorded high visitor satisfaction across 20 surveyed sites, with improvements at many — “excluding Stonehenge.”⁶ The Annual Report quantified the gap: the overall visitor experience score dipped slightly to 8.79 out of 10, but strip out Stonehenge and it held steady at 8.98, with 70 per cent of visitors rating their experience “excellent.” Stonehenge’s increased weighting in the survey data coincided with a measurable drop in those top-tier ratings, prompting the board to commission action plans.⁷

The internal data merely confirms what visitors have been saying publicly for years. One Tripadvisor reviewer in September 2025 reported paying £34.50 for an adult ticket plus £3 for parking, only to find poorly maintained pathways between the visitor centre and the stones, and a general sense of commercial extraction from what should be a transcendent encounter with 5,000 years of human history.⁸ The recurring complaints are consistent: high prices for a managed experience that keeps people at arm’s length from the monument, overcrowding at peak times, a visitor centre that feels like a warehouse with a gift shop attached, and a lingering suspicion that the site is optimised for throughput rather than wonder. Travel bloggers have coined it a “megalithic disappointment” — a one-off bucket-list tick rather than something worth returning to.^8,10

The cruel contrast is with institutions that have invested boldly and reaped the rewards. The Natural History Museum’s trajectory is instructive. Its visitor numbers are now 31 per cent above pre-pandemic levels — driven by reimagined garden spaces that attracted over 5 million visitors in their first year, a new permanent gallery on climate solutions that has already drawn more than 2 million people, and immersive experiences such as Our Story with David Attenborough.^13,14 ALVA’s director, Bernard Donoghue, attributed the success in part to those transformed outdoor spaces, calling the museum an “astonishingly fun, joyful day out.”¹¹ The lesson is not that Stonehenge should become a museum — it is that capital investment, continuously refreshed programming, and a commitment to opening new reasons to visit year after year can transform an institution’s fortunes even in a cost-of-living crisis. The Natural History Museum plans to open a new or revitalised permanent gallery every year until 2031.¹³ Stonehenge’s core offer, by contrast, has remained essentially static since the visitor centre opened in 2013: view the stones from afar, browse the exhibition, ride the shuttle, go home.

This matters because the competitive landscape has fundamentally shifted. Donoghue’s diagnosis of the wider sector is blunt: visitors today are “more tactical than ever in deciding how they spend their leisure pounds and their leisure hours” and their expectations of a great visit “are higher now than they have ever been.”¹¹ The day trip to an attraction now competes within the family budget with free alternatives and streaming subscriptions. In that environment, a paid heritage site charging premium prices must deliver an experience that justifies the cost — not just once, but persuasively enough that visitors recommend it and consider returning. Stonehenge is increasingly failing that test. Academic research stretching back almost two decades has identified access restrictions, perceived over-commercialisation, and the gap between visitor expectations and reality as persistent weaknesses.¹² The specific complaints have barely changed; only the ticket prices have — upward.

English Heritage is not unaware of the problem. Its 2025–2030 Strategic Plan speaks of ambitions under the banner “Care, Share, and Grow,” including better visitor experiences, enhanced education provision, and financial sustainability.⁷ Planning permission has been granted for a new Learning Centre and Neolithic classroom near the existing visitor facilities, due to open in autumn 2026.¹⁵ These are welcome, if overdue, steps. But a learning centre alone will not reverse a decline driven by value perception, logistical frustration, and emotional disappointment. Bolder measures are needed: dynamic pricing that rewards off-peak visits and makes the site accessible to families priced out of peak weekends; enhanced on-site interpretation that deepens engagement beyond the current walk-and-gawk circuit; investment in the kind of immersive, regularly refreshed programming that has driven growth at institutions like the Natural History Museum and the Ashmolean; and a hard look at the logistics that turn an encounter with prehistory into an exercise in queue management.

The broader heritage sector faces a punishing economic environment. ALVA described 2025 as financially the toughest year since the pandemic, compounded by increased employer National Insurance contributions and above-inflation minimum wage increases.¹¹ English Heritage cannot control macroeconomic headwinds. But it can control what happens when someone arrives at its most famous site and asks whether the experience was worth the money. Right now, too many visitors are answering no.

The stones themselves remain extraordinary — 5,000 years of human ambition and mystery compressed into a circle of sarsen and bluestone on Salisbury Plain. The tragedy is not that Stonehenge has lost its power. It is that the organisation charged with sharing that power is letting it seep away, and taking the financial foundations of England’s wider heritage with it.

 

 Data from ALVA - https://www.alva.org.uk/details.cfm?p=423

References

1. ALVA, ‘Visitor Figures 2025’, 20 March 2026. alva.org.uk/details.cfm?p=403&codeid=895.

2. Museums Association, ‘Natural History Museum breaks all-time record for visitor figures’, 20 March 2026.

3. English Heritage, ‘Reshaping English Heritage’, January 2025. english-heritage.org.uk/about/search-news/pr-reshaping-english-heritage.

4. Museum Observer, ‘English Heritage Plans 200 Redundancies and Winter Closures Amid Financial Struggles’, 2025.

5. Museums Association, ‘English Heritage workforce could shrink by 7% in major restructure’, 30 January 2025.

6. English Heritage Trust Board Meeting Minutes, 12 March 2025. english-heritage.org.uk/about/our-people/our-trustees/trustee-meeting-minutes.

7. English Heritage, Annual Report and Accounts 2024/25.

8. Tripadvisor reviews, Stonehenge Visitor Centre, September 2025; Adventure Brits, ‘Is Visiting Stonehenge A Megalithic Disappointment?’, June 2025.

9. English Heritage Stonehenge ticket prices 2025/26. english-heritage.org.uk; visitstonehenge.co.uk/en/tickets.

10. Girl Takes Mundo, ‘How To Visit Stonehenge For Free’, 9 February 2026.

11. Arts Professional, ‘Visitor attraction numbers see “modest” but “consistent” growth, ALVA finds’, 20 March 2026.

12. Mason, P. & Kuo, I-L. (2007) ‘Visitor Attitudes to Stonehenge: International Icon or National Disgrace?’, Journal of Heritage Tourism, 2(3), pp. 168–183.

13. Natural History Museum, ‘Record breaking 7.1m visitors make it UK’s most popular visitor attraction’, 20 March 2026. nhm.ac.uk.

14. Natural History Museum, ‘Over 5 Million Visitors to Gardens in First Year’, July 2025. nhm.ac.uk.

15. Wikipedia, ‘Stonehenge’, citing English Heritage announcement March 2025: planning permission for Learning Centre and Neolithic classroom, due autumn 2026.

16. Civil Society, ‘English Heritage proposes cutting at least 189 staff roles’, January 2025.

17. Prospect Union, ‘Prospect warns against proposed English Heritage redundancies’, 2024.

18. Artlyst, ‘Natural History Museum Breaks UK Records With 7.1 Million Visitors’, March 2026.

19. Statista / ALVA, ‘Number of visits to Stonehenge in England from 2010 to 2023’. Pre-pandemic peak c. 1.6m in 2019.