This post is the companion to the short film Ten kilometres for one sunset?, which tests a fifty-year-old astronomical claim about the Dorset Cursus against modern LiDAR. The method, numbers and caveats follow.
The Dorset Cursus is a pair of parallel banks with outer ditches running for just over 6¼ miles, nearly 10 km, across Cranborne Chase, from Thickthorn Down in the south-west to Martin Down in the north-east (RCHME 1975; Dorset HER MDO5608). It was built in two parts. The first, the Gussage cursus, ran from Thickthorn to a terminal on Bottlebush Down; the second, the Pentridge cursus, continued it on a slightly different alignment to Martin Down. Measured along the published points, the two are 5.65 km and 4.33 km long. Radiocarbon dates span the second half of the fourth millennium BC (Bowden et al. 1983; Barrett, Bradley & Green 1991). For most of its length it survives only as a cropmark or soilmark; the ends at Thickthorn and Martin Down remain as earthworks.
Long barrows lie beside it, inside it and in its banks. On Gussage Hill one lies across the interior, at right angles to the banks: a mound 155 ft long, 65 ft across and up to 10 ft high (RCHME 1975; Dorset HER MDO5615).
The Dorset Cursus on EA 1 m LiDAR, from the south-east. Height ×1.5.
The claim
Penny and Wood (1973) proposed that an observer at the Bottlebush terminal of the Gussage cursus would see the midwinter sun set behind the Gussage Hill long barrow. Barrett, Bradley and Green (1991) accepted this as the most plausible of Penny and Wood's alignments. It has since become the standard astronomical reading of the monument.
The test
The Environment Agency's LIDAR Composite DTM at 1 m (2022) covers the whole cursus. The observer is placed 30 m inside the Bottlebush terminal (RCHME SU 01581566) on the line of the Gussage cursus, at E401558 N115639, with the eye 1.6 m above the ground. The barrow mound is located directly in the DTM: its summit stands 3.1 m above its surroundings, and the part higher than 0.6 m is 43 m long on a bearing of 143°, matching the RCHME description.
From the observer the barrow is 2.89 km away on a bearing of 231.2°; the mound spans 230.8° to 231.7°. Sight lines are traced across the DTM out to 8 km with the curvature of the earth and terrestrial refraction (coefficient 0.13) included.
Section along the sight line from the Bottlebush terminal. The barrow is the skyline; the ground beyond stays below the line. Height exaggerated as marked.
The barrow is the skyline. On bearings from 224° to 238° the horizon is the Gussage Hill ridge, 2.83 to 2.95 km away, and the mound lifts it from about 0.24° to 0.29° above level. Nothing further away shows above it.
The sun. In 3300 BC the obliquity of the ecliptic was 24.05° (Laskar 1986), so the midwinter sun's declination was −24.05°. At latitude 50.93° N, with standard refraction at the horizon's altitude, the last gleam of the sun, its upper limb touching the skyline, goes down on a bearing of 230.60°. Moving the date to 3500 BC or 3000 BC shifts this by less than 0.05°.
The skyline from the middle of the Bottlebush terminal at true angular scale, with the midwinter sun of 3300 BC at three moments of its setting. The last gleam goes down just south of the mound.
From the middle of the terminal it just misses. The last gleam goes down 0.20° south of the mound's south-west end, about 10 m at the distance of the barrow. The sun's disc is 0.53° across, so the miss is less than half its width.
From the north-west half it works. Moving across the cursus changes the barrow's bearing by about 0.02° for every metre. From anywhere between 10 m and 55 m north-west of the axis, which is the north-west half of the terminal, the last of the midwinter sun goes down behind the mound.
The Bottlebush terminal on LiDAR local relief. Gold: the last gleam goes down behind the barrow. Blue: it goes down just beside it. Banks as traced on the flyover.
What could move it
Refraction. Near the horizon in winter it varies from night to night; a change of 30% moves the last gleam by about 0.27°, more than the miss from the middle of the terminal.
The terminal. The Gussage terminal bank on Bottlebush is ploughed. Its position comes from an eight-figure RCHME grid reference, and the axis there from the line between published points, not from surviving banks. Either can shift the zone by several metres.
The barrow. The DTM records the mound as it is now, ploughed and eroded. A taller mound would widen the zone, not narrow it.
Trees. Neolithic woodland on Gussage Hill, if any, is unknown and not modelled.
Conclusion
The 1973 claim survives a test its authors could not run. The Gussage Hill long barrow is the skyline from the Bottlebush terminal, and the midwinter sun of the late fourth millennium BC sets behind it when seen from the north-west half of the terminal, and just beside it from the south-east half. Whether the builders intended this cannot be settled from the ground.
How the line is drawn. No complete survey plan of the cursus is published. The banks on the flyover run between the published positions of the terminals and of points along the line (Historic England 1002785; RCHME long barrow 14, SU 011152, SU 01581566, SU 025169, SU 04051920), moved onto the ridges where the LiDAR still shows them, and left on the straight line between points across ploughed gaps. The code for the test is in the film's folder on GitHub (align.py, sens.py, diagrams.py).
References
Atkinson, R. J. C. 1955. The Dorset Cursus. Antiquity 29, 4–9.
Barrett, J., Bradley, R. & Green, M. 1991. Landscape, Monuments and Society: the prehistory of Cranborne Chase. Cambridge University Press.
Bowden, M., Bradley, R., Gaffney, V. & Mepham, L. 1983. The date of the Dorset Cursus. Proceedings of the Prehistoric Society 49, 376–379. doi:10.1017/S0079497X00008069
This post is the companion to the short film Eighty metres up. In under three minutes the film sets out why the erratic boulders of North Devon matter to the claim that ice carried the Stonehenge bluestones to Salisbury Plain, and what they actually show. The full argument, with sources, follows.
The claim
Glacial transport of the bluestones needs Irish Sea ice that was not merely present in the Bristol Channel but standing high on the land, high enough to override the coast and carry on towards Wiltshire. The case for that has leaned on erratics. One abstract puts it this way: "of the scores of known glacial erratics on the shores of the Bristol Channel, many are found at altitudes in excess of 100m" (John 2024, ResearchGate abstract to his Quaternary Newsletter 162 article; the sentence does not appear in the article itself, as set out in The myth of Bristol Channel high-level glacial erratics).
South of the Channel, North Devon is where such evidence would have to be found. Three bodies of evidence there bear on the question: the shoreline erratics, the Fremington clay, and a handful of claimed high-level sites, of which the Ramson Cliff boulder is the last and the most cited.
The shoreline band
The Saunton–Croyde erratics, Madgett & Inglis (1987), plotted from their grid references
Far-travelled boulders have been reported at the foot of the Saunton cliffs since 1837. Madgett & Inglis (1987) catalogued 37 in the Saunton–Croyde area (listed with grid references in the North Devon erratics master catalogue). Thirty-six of them lie on the foreshore, the raised shore platform or the raised beach, all below about 30 m OD. Every erratic reported up to 1969 lay at the base of the cliffs. The largest is the granulite gneiss near Baggy House at Croyde known as the '50-tonner' (No. 2, 420 × 220 × 200+ cm, first recorded by Hughes in 1887), resting on the planed shore platform; its weight "has been estimated to be 50 tonnes" (Bennett et al. 2024, p. 91).
The '50-tonner' granulite gneiss erratic at Croyde. Archive photograph.
These boulders are genuinely far-travelled and they matter. Their distribution is the evidence. Stephens (1966) pointed out that the largest are confined to a narrow coastal band within reach of storm waves: a selective distribution that is hard to explain by an ice sheet burying the ground, and easy to explain by boulders grounded from floating ice at a former shoreline. Scourse et al. (2024) have since supplied a dated mechanism. Early in the last cold stage (Marine Isotope Stages 4 and 3), ice-loaded crust held relative sea level high while calving ice margins still existed offshore, so floating ice could raft material onto the southern shore platforms. Bennett et al. (2024, p. 91) set out the same alternative, that the platform erratics "were delivered by icebergs calved from glacier ice farther north", and conclude that "both modes of transport (ice-rafting and glacial transport)" may have been involved in the South-West. The shoreline band is the signature of ice that reached the coast, not of ice that climbed it. The argument is set out more fully in Two boulders fewer and Thinking three dimensional.
Floating ice grounding on the shore platform at a high relative sea level. Diagram, not to scale, after Scourse et al. (2024).
The Fremington clay
Boreholes along the Higher Gorse–Roundswell line. Red: lake clay; gold: uncertain; grey: head on the ridge. From BGS borehole logs.
The red potter's clay of Fremington, near Barnstaple, was described as boulder clay by Maw in 1864 and for a century was read as till from an Irish Sea ice sheet. Bennett et al. (2024, p. 87) still call these deposits "the only widely (though not universally) accepted evidence that glacier ice ever reached the peninsula". The borehole record points elsewhere. The Brannam campaign logs (BGS SS53SW62–79) and later holes confirm fine lake clay only at low levels: its base and the underlying gravel lie at about 15–28 m OD, the highest confirmed clay tops at 31.7–32.8 m OD, so no higher than about 33 m, and there is none south of the Hele–Bickington ridge. On the ridge itself, at 38–40 m OD, the clayey material is head, not lake clay. That fits the lake surface of about 30 m OD that Edmonds inferred, as reported by Bennett et al. (2024, p. 88). The clay records a modest lake ponded in the Taw–Torridge estuary behind an ice and outwash dam at the coast, not a till sheet laid by grounded ice overriding Devon to 80–90 m. The boreholes are mapped in the Fremington Clay gazetteer; the analysis is in New borehole analysis sharply constrains the Fremington Clay and Caution in attributing the Fremington Clay Series to Irish Sea glaciation.
Local Lundy granite moved about on the island, not far-travelled erratics (Rolfe et al. 2014). Carr reads them as residual boulders from two-stage weathering rather than glacially transported.
Shebbear (Devil's Stone)
150 m
A sarsen, a local silcrete on or near its parent duricrust: not a travelled stone (The Shebbear erratic sarsen).
The Ramson Cliff erratic beside the coast path, 2025. Photo: T. Daw
The Ramson Cliff erratic is a 378 kg block of altered epidiorite (105 × 55 × 38 cm; Madgett & Inglis 1987, No. 8) now lying beside the coast path at about 80 m OD on Baggy Point. It had been treated as decisive. Bennett et al. (2024, p. 91) wrote that "an isolated block of epidiorite was found at about 80 m OD on Baggy Point promontory [SS 4356 4070] by Madgett and Madgett (1974) which can only have been emplaced by an ice sheet." They qualified it at once: "Whilst this implies ice-sheet transport, it is important to consider the role of tectonic uplift of the region early in the Pleistocene." It has also been cited in support of glacial transport of the bluestones (John 2024).
The question is how the boulder came to be where it was found. Daw, Ixer & Madgett (2026) went back to the original thin section, the archive and the correspondence:
It was first recorded in 1969, standing upright in the middle of a pasture field on the crest of Baggy Point. It does not appear on earlier maps or aerial photographs.
In the early 1970s the field was ploughed; the boulder was knocked over and then dragged to the field edge, where it lies now.
It is angular and rough-surfaced, with none of the abrasion of the shoreline boulders.
It is the only claimed erratic on the south shore of the Bristol Channel above about 30 m OD, with no supporting deposits, striae or erratics at intermediate heights.
The petrography adds to the emplacement problem. A second thin section, cut from the original hand specimen and described by Rob Ixer in September 2026, confirms the published description. The rock is a highly altered ophitic microgabbro with relict clinopyroxene, no olivine and no quartz; the secondary minerals are brown-green hornblende and later colourless-to-green actinolite, with minor chlorite, epidote and clinozoisite. That assemblage matches the outer aureole of the Dartmoor granite and is consistent with Cornubian contact metamorphism. It does not match the Welsh Ordovician metadolerites on the Irish Sea ice route. A Scottish source cannot be excluded on the amphibole evidence alone.
Irish Sea ice came from the north. Dartmoor lies inland, to the south. A Cornubian source does not prove that people moved the stone, but it makes a lone high-level glacial drop even less likely. An angular block of probably local greenstone, standing upright, unworn, undocumented before 1969 and alone at its height is better read as a manuport, perhaps a standing stone, than as proof of an ice sheet 80 m above the sea.
Statistically it is also an outlier. Of 49 recorded erratic heights in the area, about 86% lie between 5 and 25 m OD; the 80 m value is flagged by the interquartile-range test, the z-score (z = 5.92), Grubbs' test and Dixon's Q test.
What is left
With Ramson Cliff and Shebbear set aside, there is no demonstrated far-travelled boulder above the coastal zone of North Devon, and the claimed high-level sites further east do not survive inspection either. Every proven far-travelled boulder south of the Channel sits at or near the shore. Irish Sea ice reached the coast; nothing shows that it climbed inland, let alone that it went on to Salisbury Plain.
Caveats. This is the better-supported position, not an agreed one. Bennett et al. (2024, p. 91) had previously read the Ramson Cliff boulder as glacial, though with the uplift caveat noted above; Daw, Ixer & Madgett (2026) was written in response. Croot et al. (1996; reviewed by Bennett et al. 2024, p. 88) concluded that the Fremington Clay Series, though probably glaciolacustrine, was overridden by glacier ice. Bennett et al. themselves allow that the Fremington till and gravels may have been deposited near the ice limit "directly or indirectly (i.e., as ice-rafted debris)" (p. 87). Low-level grounded ice reaching about the present coastline remains possible for North Devon, and local glaciation in Devon is not challenged. What it removes is the evidence that was said to require thick ice high on the land.
Notes on the film. The coastline on the maps is drawn from ONS administrative boundaries (Open Government Licence). The red dots are Madgett & Inglis (1987) Nos 1–7 and 9–37, plotted from the catalogue grid references. Claimed sites are plotted from published grid references where they exist (Court Hill ST 473 723, Nightingale Valley ST 450 752, the Devil's Stone SS 4388 0925, Ramson Cliff SS 4356 4070); Lundy and the Ilfracombe plateau are approximate. The Fremington section is redrawn from the borehole data in the Fremington Clay gazetteer. The Croyde photograph is an archive image; the thin section is Madgett & Madgett's. Ramson Cliff and bluestone photographs: T. Daw. Music: original.
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.
Campbell, S., Hunt, C.O., Scourse, J.D., Keen, D.H. & Stephens, N. (1998). Quaternary of South-West England. Geological Conservation Review Series 14. Chapman & Hall / JNCC. Chapter 7 (PDF)
Croot, D.G., Gilbert, A., Griffiths, J. & van der Meer, J.J.M. (1996). The character, age and depositional environments of the Fremington Clay Series, North Devon. Quaternary Newsletter 80, 1–15.
Daw, T., Ixer, R. & Madgett, P. (2026). A review of the Ramson Cliff erratic: evidence of high-level ice flow? Quaternary Newsletter 167, 13–19. doi:10.64926/qn.20517
John, B. (2024). An igneous erratic at Limeslade, Gower, and the glaciation of the Bristol Channel. Quaternary Newsletter 162; ResearchGate abstract: publication 381775577.
Madgett, P. & Madgett, R. (1974). High level erratic on Baggy Point. Quaternary Newsletter 14, 1–2.
Madgett, P.A. & Inglis, A.E. (1987). A re-appraisal of the erratic suite of the Saunton and Croyde areas, North Devon. Transactions of the Devonshire Association 119, 135–144. PDF
Maw, G. (1864). On a supposed deposit of boulder-clay in North Devon. Quarterly Journal of the Geological Society of London 20, 445–451.
Rolfe, C.J., Hughes, P.D. & Brown, A.G. (2014). Timing of the maximum extent of Late Pleistocene glaciation in NW Europe: evidence from Lundy. Journal of the Lundy Field Society 4, 7–18. PDF
Scourse, J.D. et al. (2024). The timing and magnitude of the British–Irish Ice Sheet between Marine Isotope Stages 5d and 2: implications for glacio-isostatic adjustment, high relative sea levels and 'giant erratic' emplacement. Journal of Quaternary Science. doi:10.1002/jqs.3611
Stephens, N. (1966). Some Pleistocene deposits in North Devon. Biuletyn Peryglacjalny 15, 103–114.
