Peter Kokelaar, a retired volcanologist and field geologist with extensive experience in the Quaternary history of the Gower Peninsula, has made a thoughtful and detailed contribution to the long-running debate over the transport of Stonehenge’s bluestones. In two online articles, Towards Stonehenge: the Anglian Glaciation of Gower and Stonehenge, he argues that glacial processes, specifically the Anglian-stage Irish Sea Ice Stream (c. 450,000 years ago), played a major role in moving bluestones from their Pembrokeshire sources at least part of the way towards Salisbury Plain. While acknowledging the absence of direct proof, Kokelaar proposes that ice performed “most of the work,” depositing stones “near enough” for later Neolithic collection and incorporation into the monument. His work is richly illustrated with high-quality photographs, maps, and LiDAR imagery, reflecting sustained and meticulous field observation.
Kokelaar’s contributions deserve serious attention, particularly for their geological rigour on Gower, where he documents compelling evidence of Anglian ice overriding the peninsula, transporting far-travelled erratics, and reshaping landscapes in ways that have often been under-appreciated or misattributed to later glaciations. His work strengthens understanding of the southern reach and dynamic behaviour of the Irish Sea Ice Stream. At the same time, when these regional observations are extrapolated to Stonehenge itself, the hypothesis encounters substantial difficulties. By early 2026, the prevailing multidisciplinary consensus, drawing on geological provenancing, archaeological excavation, and Quaternary geomorphology, overwhelmingly favours deliberate Neolithic human transport from specific Preseli outcrops, with no compelling evidence for glacial delivery to the Salisbury Plain region.
Core Arguments and Their Limitations
Kokelaar’s case rests on several interconnected claims,
primarily grounded in his observations on Gower and projected eastward.
Ice Flow Trajectory and Transport Capability
Kokelaar anchors his hypothesis in detailed geological observations from the Gower Peninsula, which he interprets as evidence that Anglian ice carried Pembrokeshire-derived material eastward in a direction broadly aligned towards Stonehenge. Gower provides one of the most southerly and best-exposed records of Irish Sea Ice Stream activity, and Kokelaar documents this record with considerable care.
A central line of evidence is the diverse assemblage of far-travelled erratics preserved on Gower beaches, many reworked from Anglian deposits. Kokelaar records more than twenty igneous lithologies, including gabbros consistent with St David’s Head, silicic volcanic rocks comparable to Ramsey Island, and non-spotted dolerites matching Pembrokeshire sources. Particularly significant are erratics of riebeckite microgranite derived from Ailsa Craig in western Scotland, a well-established tracer of Irish Sea Ice Stream transport. These lithologies collectively confirm long-distance entrainment, southward flow through the Irish Sea basin, and subsequent east–southeast deflection across Gower into the Bristol Channel.
Ice-flow direction is further constrained by mapped erratic dispersal patterns and geomorphological indicators. Kokelaar’s reconstructions (for example, his Figure 12) depict ice advancing from the west-northwest across Gower, a trajectory that in broad compass terms points towards southern Britain. Supporting evidence includes marine shells incorporated within till at sites such as Cockle Pot, interpreted as material entrained from Carmarthen Bay and redistributed by advancing ice. Together, these observations demonstrate that Anglian ice on Gower was thick, coherent, and dynamically capable of transporting lithologies over considerable distances.
Kokelaar also emphasises subglacial and erosional features indicative of warm-based ice overriding a karst landscape. Cave systems such as Ogof New Park exhibit sediment infill, constricted passages, and mixed corrosion features plausibly interpreted as subglacial modification. At the surface, bedrock-incised channels (for example at Melins Lake), rounded coastal cliffs, and large fluvioglacial boulders at Pwll-du point to vigorous erosion and meltwater activity during deglaciation. Fossil material associated with some Pwll-du deposits has been dated to around 425,000 years ago, consistent with an Anglian age. More broadly, Kokelaar argues that Anglian ice overrode a pre-existing marine platform on Gower, substantially modifying landforms often attributed to later (Devensian) glaciation.
These observations convincingly establish the power and reach of Anglian ice on Gower. The difficulty arises when this evidence is extended beyond the Bristol Channel.
Critique:
While regional ice-flow alignment shows that ice moved eastward across Gower,
this does not in itself demonstrate penetration onto Salisbury Plain. Most
glaciological reconstructions constrain the Anglian southern margin to the
Bristol Channel and adjacent lowlands, with higher ground promoting thinning,
stagnation, or deflection rather than sustained eastward advance. Crucially,
the geomorphological and sedimentary signatures so clearly preserved on
Gower, tills, erratic spreads, and subglacial erosion, are absent from Salisbury
Plain and its margins, despite extensive Quaternary mapping and repeated
archaeological exposure.
The Mendip Hills as a Critical Constraint
A major obstacle to Kokelaar’s mapped ice trajectory is the Mendip Hills, which occupy a key topographic position between the Bristol Channel and Salisbury Plain and are explicitly crossed by ice in his reconstructions. In mainstream Quaternary geology, however, the Mendips lie outside the direct influence of Pleistocene ice sheets, including the Anglian glaciation.
The Mendip plateau and slopes are mantled by extensive periglacial deposits, commonly termed head, rubby drift, or plateau drift, comprising angular rubble derived almost entirely from local Carboniferous limestone and adjacent bedrock. These deposits are classic products of freeze–thaw weathering and solifluction and lack the sorting, matrix support, or lithological diversity characteristic of glacial till. No confirmed far-travelled erratics from Wales, Ireland, or Scotland—and notably no Preseli-type spotted dolerite—have been recorded on the Mendip plateau or higher ground.
There is good evidence that Irish Sea ice entered the Bristol Channel during the Anglian, depositing exotic erratics on low-lying coastal sites and islands such as Flat Holm and Steep Holm, and leaving patchy tills in the Somerset lowlands. These occurrences demonstrate ice proximity, but they also underscore a consistent pattern: far-travelled material is confined to low elevations, while the hills themselves remain unaffected. Mendip landforms—gorges, dolines, dry valleys, and cave systems—are overwhelmingly attributed to karst and periglacial processes rather than ice overriding.
The Mendips rise to approximately 325 metres above sea level. For ice to have crossed the plateau, local ice thickness would need to have exceeded this elevation by a substantial margin. While continental ice sheets can certainly override relief of this scale, conditions at the southern margin of the Irish Sea Ice Stream were likely far more marginal. Kokelaar himself estimates ice thicknesses of only around 250–300 metres over low-lying Gower. Under such conditions, ice would be expected to thin rapidly against rising terrain and deflect around uplands rather than override them. The complete absence of glacial deposits or erratics on the Mendips strongly suggests that they acted as a partial barrier, diverting ice into the Somerset lowlands and along the Bristol Channel rather than permitting direct eastward flow towards Salisbury Plain.
Contrasting Gower and the Mendips
The contrast between Gower and the Mendips highlights a key methodological issue. Gower lay directly in the path of the Irish Sea Ice Stream, at low elevation and with lithologies conducive to both erosion and preservation of glacial deposits. The Mendips formed an elevated limestone massif at or beyond the effective ice margin, dominated by karstic drainage and periglacial reworking. Gower shows what Anglian ice does to landscapes it clearly overrides; the Mendips show what landscapes look like when it does not. Robust evidence from the former cannot be linearly projected across the latter without contradicting well-established geomorphological constraints.
Bluestone Morphologies and Assemblage
Kokelaar argues that the Stonehenge bluestones comprise a
mixture of angular pillars and rounded, abraded forms suggestive of glacial
transport. He points to stones such as the Newall and Boles Barrow boulders as
potential erratics and interprets widespread debitage as evidence of a natural
glacial supply gradually exhausted by builders.
Critique:
Polyhedral and pillar-like forms are readily explained by natural jointing in Preseli dolerite sills, which facilitated Neolithic quarrying with minimal shaping. Apparent surface rounding or smoothing does not, in itself, demonstrate glacial transport. In the Preseli source area, many dolerite and rhyolite blocks occur naturally as detached or partially weathered pillars and boulders with softened edges, produced by long-term chemical weathering, joint-controlled block release, and limited fluvial reworking prior to the Anglian glaciation. As a result, stones may acquire rounded or smoothed profiles in situ, before any human or glacial movement. Similar morphologies are observed at quarry sites such as Carn Goedog and Craig Rhos-y-felin, where blocks detached from outcrops already display a range of angular to moderately rounded forms. Consequently, stone shape alone cannot be taken as diagnostic of glacial transport, particularly in the absence of unequivocal subglacial features such as faceting, striations, or percussion marks.
Detailed analyses of the Newall boulder, a foliated rhyolite excavated in 1924, demonstrate that it lacks subglacial faceting or striations and matches material quarried at Craig Rhos-y-felin. Its surface alterations are best explained by post-depositional weathering rather than ice modification. Similarly, the Boles Barrow boulder has a highly disputed provenance and context; Mike Pitts has described the archaeology of Boles Barrow as "a mess," noting significant uncertainties over whether the specimen in Salisbury Museum originated from a primary Neolithic deposit in the barrow, rendering it far from an "unequivocal erratic". Debitage patterns at Stonehenge reflect phased on-site dressing, reuse, and reworking over centuries, not dispersed glacial moraine.
Source Specificity and the Nunatak Hypothesis
The dominance of spotted dolerite from specific eastern Preseli outcrops poses a further problem for glacial models. Kokelaar suggests that these outcrops protruded as nunataks, enabling selective supraglacial transport.
Critique:
While partial nunatak exposure is plausible, no evidence supports the transport
of megalith-sized blocks over 200 km supraglacially in sufficient numbers and
lithological purity to account for the Stonehenge assemblage. The complete
absence of spotted dolerite erratics beyond Preseli—despite extensive
surveys—remains a critical weakness for any glacial explanation.
Archaeological Evidence for Human Transport
Kokelaar questions the interpretation of Neolithic quarries
at Carn Goedog and Craig Rhos-y-felin and emphasises the absence of preserved
transport routes.
Critique:
Excavations have revealed extraction platforms, removal scars, stone tools, loading features, and radiocarbon dates around 3400–3000 BC, consistent with Stonehenge’s early phases. Geochemical provenance is precise and non-random. While transport logistics remain debated, human movement of large stones over long distances is well attested in Neolithic Britain and fits the cultural context of monumentality and interregional connection.
Claims Regarding the Altar Stone
Kokelaar extends his glacial perspective to the Altar Stone (Stone 80), accepting its Scottish provenance but challenging the northeastern constraint proposed in recent studies. He argues that detrital zircon U-Pb age spectra, central to the 2024 identification, could equally match sediments from southwest or central Scotland (southwest of the Great Glen Fault), derived from erosion of Grampian Highland rocks. This broader sourcing, he suggests, aligns better with Anglian Irish Sea Ice Stream transport of Scottish material southward. This interpretation, while highlighting potential overlaps in zircon inheritance, is selective: it engages only with zircon data and overlooks the multi-proxy evidence (including apatite and rutile trace-element chemistry, mineral fabrics, and stratigraphic context) that robustly constrains the Altar Stone to Old Red Sandstone of the Orcadian Basin in northeast Scotland.
Conclusion
Peter Kokelaar’s articles represent a serious and informed challenge from an independent geologist deeply familiar with the Quaternary history of southwest Wales. His work on Gower significantly enriches understanding of Anglian glaciation and rightly cautions against dismissing natural processes in favour of purely anthropogenic explanations. However, the extension of this regional evidence to Stonehenge involves substantial extrapolation across terrain, most notably the Mendip Hills, that shows no trace of glacial overriding. The hybrid model, in which ice performs “most of the work” before depositing stones “near enough,” relies on assumptions that remain difficult to test and unsupported by positive evidence.
In contrast, the human-transport model is underpinned by targeted quarry archaeology, precise provenance matching, and a coherent cultural framework. Kokelaar’s revival of the glacial hypothesis remains an intriguing and valuable provocation, but one that does not displace the accumulated multidisciplinary evidence favouring deliberate Neolithic transport of the bluestones to Stonehenge.
References
- Bevins, R., Ixer, R. A., Pearce, N., Scourse, J., & Daw, T. (2023). Lithological description and provenancing of a collection of bluestones from excavations at Stonehenge by William Hawley in 1924 with implications for the human versus ice transport debate of the monument’s bluestone megaliths. Geoarchaeology: An International Journal, 38(6), 771-785. https://doi.org/10.1002/gea.21971
- Richard E. Bevins, Nick J.G. Pearce, Rob A. Ixer, James Scourse, Tim Daw, Mike Parker Pearson, Mike Pitts, David Field, Duncan Pirrie, Ian Saunders, Matthew Power. The enigmatic ‘Newall boulder’ excavated at Stonehenge in 1924: New data and correcting the record. Journal of Archaeological Science: Reports,Volume 66, 2025, 105303, ISSN 2352-409X, https://doi.org/10.1016/j.jasrep.2025.105303 (https://www.sciencedirect.com/science/article/pii/S2352409X25003360)
- Kokelaar, P. (n.d.-a). Towards Stonehenge: the Anglian glaciation of Gower. Retrieved January 8, 2026, from https://kokelaargower.com/towards-stonehenge-the-anglian-glaciation-of-gower/
- Kokelaar, P. (n.d.-b). Stonehenge. Retrieved January 8, 2026, from https://kokelaargower.com/stonehenge/
- Pearson, Mike Parker, Josh Pollard, Colin Richards, Kate Welham, Chris Casswell, Charles French, and others, ‘Megalith Quarries for Stonehenge’s Bluestones’, Antiquity, 93 (2019), 45–62 http://dx.doi.org/10.15184/aqy.2018.111
- Scourse, J.D. (1997). Transport of the Stonehenge bluestones: Testing the glacial hypothesis. In B. Cunliffe & C. Renfrew (Eds.), Science and Stonehenge (Proceedings of the British Academy 92, pp. 271–314). Oxford University Press.
- Thorpe, R.S., Williams-Thorpe, O., Jenkins, D.G., & Watson, J.S. (1991). The geological sources of the Stonehenge bluestones. Oxford Journal of Archaeology, 10(2), 127–148.
No comments:
Post a Comment
Comments welcome on fresh posts - you just need a Google account to do so.