Wednesday, 9 September 2026

The Wiltshire Long Barrow Gazetteer

The Wiltshire Long Barrow Gazetteer - https://timdaw37.github.io/wiltshire-long-barrows/

A hundred and twenty-nine Neolithic long barrows on the Wiltshire chalk — HER certain and possible, plus one modern mound at All Cannings that is mine. Gold markers you can trust; grey ones you should treat as possible. Where the earthwork still shows on the LiDAR chip, a rim arrow marks the long axis of the mound: the line of the bank, not the way a façade “faces”. Eighty-six arrows. Forty-three sites with no arrow, because nothing usable is there to measure.

People have always wanted a sentence for the set. They face the rising sun. They run along the ridges. They stand on the skyline, landscape controllers, meant to be seen from below.

We measured all three.

They do not face the sun as a shared design. The eighty-six axes spread. There is a gentle east–west smear and no spike at midsummer or midwinter. Give a random set of arrows the same two solstice targets and chance is as close as the real mounds. They sit nearer the contour than luck would put them. They do not glue themselves to the ridge. Half the sample is still more than thirty degrees off the local ground. West Kennet rides the nose of its spur. East Kennet, a mile away and the same tradition, cuts across it.

They do not stand on the skyline always either. Adam’s Grave does, and West Kennet, and a handful of others. South Street and Lanhill, good surviving mounds, fail the test: often visible, almost never the crest against the sky.

Field, McOmish, Ruggles and Roberts already said so in other words. The map and the two notes are that argument with arrows on it.

Eighty-six Wiltshire long-barrow axes, each drawn both ways. The dashed lines are sunrise on a flat horizon at this latitude: midsummer, equinox, midwinter. They are drawn so you can see them. They are not a finding.

The longer notes:

Do Wiltshire’s long barrows face the sun? — bearings, ridges, the 86.
Do Wiltshire’s long barrows stand on the skyline? — length-shows approaches, Kitchen and West Kennet photographs, the maps.

Do Wiltshire’s long barrows stand on the skyline?

The short answer is no as a shared design — same family as the solar result. Some do (Adam’s Grave class). Many famous ones do not. Being visible on a ridge is not the rule.

Every arrow is on the map:

The Wiltshire Long Barrow Gazetteer

What follows is a first cut at a different question from the ridge-axis work: not whether the mound runs along high ground, but whether it was meant to stand on the skyline from approaches where you can see its length.

Standing on other people’s work

None of this starts from a blank field.

David Field and David McOmish, walking the chalk, kept saying the same thing in different words: long barrows sit with care in local ground; there is no single shared recipe. Clive Ruggles insisted prehistoric “display” claims have to survive a proper test, not a pretty photograph. Dave Roberts and colleagues, around Stonehenge, put local topography first and published the table. David Wheatley’s HER recreation and the Environment Agency county DTM are what let a county-scale check exist at all.

We add a systematic skyline-from-below sample on top of that work, not instead of it. If this note is useful, it is because those people already did the hard looking.

NOTE - Click photos to embiggen, and use The Wiltshire Long Barrow Gazetteer

Two sentences that get stuck together

People say long barrows “sit on the ridge” and mean two different things.

One is siting: the mound is on high ground. West Kennet is. Many others are. That is true and not the test here. The companion note already showed that long axes sit nearer the contour than chance, without gluing themselves to it. East Kennet, a mile from West Kennet and the same Cotswold–Severn tradition, cuts across the ridge. Same downs, two recipes.

The other is design as a skyline marker — a landscape controller — from approaches where the length of the mound shows: broadside views, not end-on. Does the crest sit against sky, silhouetted, from those wedges? Or is it only visible against rising ground beyond?

That is the question I wanted answered. “Along the ridge” was the wrong sentence.

Two photographs show the phenomenon the maps are trying to score. They are ground truth for those viewpoints. They are not a county rule.

Figure 1. Kitchen Barrow from a lower southern approach, 9 September 2026. The mound is the bump on the distant ridge, silhouetted against sky. That is a skyline-from-below catch.

Figure 2. West Kennet from the north, 22 October 2021. The long mound sits on the ridge skyline; people on the crest give the scale.

Kitchen’s southern silhouette is what an Adam’s-class neighbour looks like in a photograph: a distant ridge bump, length showing, sky behind it. West Kennet from the north can do the same. The rest of this note asks whether those catches are typical of the defined approaches, or rare sweet spots.

What was measured

Take each barrow’s trusted long axis (the undirected bearing already on the gazetteer).

From that axis, open two perpendicular wedges — the bearings where the length is visible, about ±32° either side of broadside. Inside those wedges only, drop observers on a dense grid at road-to-ridge scale (roughly 0.3–2.5 km). For each cell, ask whether a line of sight to the crest is clear on the county Environment Agency DTM, and whether ground beyond the crest rises above that line. If it does not, the mound is the local horizon peak: skyline from below.

Maps are contour fields of that clearance, not a single corridor percentage. Green (positive clearance) means the crest sits above the far horizon. Red means you can often still see the mound, but against higher chalk beyond. Percentages are a blunt summary of the green.

Figure 3. The local four on the same frame. Blue dashed wedges are the length-shows approaches; yellow tick is the long axis; red dots are skyline hits. Green is silhouette; red is crest against rising far ground. Adam’s Grave is almost all green. Kitchen has a coherent southern catchment. West and East Kennet are mostly red, with small green pockets near the mound.

Ground truth before the computer: Kitchen’s southern silhouette (Figure 1); West Kennet from the north (Figure 2); Adam’s Grave from the Lockeridge approach; East Kennet from the nearer road. Pins and photos check that the model can catch real views when they exist.

Crest height is the 1 m LiDAR chip plus a modest reconstruction stick-up of 2 m. Eye height is standing height (1.7 m). Mild earth curvature and refraction are on. Contour-band nulls — random points on the same height band — are secondary. Named barrows should beat them if “designed skyline marker” is the claim.

This is not a ridge-axis test, not a Tilley-style walk, and not the Bulford solstitial horizon-altitude protocol. Do not mix the three.

The local four

Adam’s Grave, Kitchen, West Kennet, East Kennet. Same stretch of down. Same method.

Adam’s Grave

Figure 4. Adam’s Grave. About 88% of the broadside sample puts the mound on the skyline. The cyan X is the Lockeridge pin (~593 m north of the crest): clearance ≈ +5.4°. That is what an Adam’s-class silhouette looks like in these scores.

From Pewsey Vale to the south, and from the Lockeridge approach to the north, the scarp-edge mound is almost always the horizon peak when it is visible at all. Contour-band nulls average ~14%. A fake crest shoved 200 m downslope, in the earlier corridor run, collapsed to 0%. The Lockeridge pin sits squarely in the green.

Kitchen

Figure 5. Kitchen Barrow. About 32% skyline in the same frame, beating local nulls (~8%) by roughly four times. The southern wedge is the catchment that matches Figure 1.

Kitchen lands lower than Adam’s Grave but still real. The map shows a coherent southern / vale catchment and a thinner northern one. That is the photograph: a distant ridge bump from below, length showing. Not as dominant as Adam’s Grave. The named site outperforms random same-height ground.

West Kennet

Figure 6. West Kennet. About 60% of the broadside cells can see the mound; only 19% put it on the skyline. Median clearance is slightly negative. The green pockets sit close in, north and south of the crest. Most of both wedges are red: the ray through the mound continues onto higher chalk beyond.

West Kennet is the awkward one relative to reputation, and relative to Figure 2.

Corridor-style sampling — vale cells 1–5 km out, not restricted to broadside — made it look almost absent as a true skyline peak (~1%). Restricting to length-shows wedges at nearer range improves it to ~19%, and beats the local contour null (~6%). So there are pockets of silhouette, including from the north. Figure 2 is real. The model finds them.

Typical cells in the defined approaches still see the mound against higher chalk beyond more often than against sky. Geometry says why: from much of the northern Kennet valley the ray through West Kennet continues onto the East Kennet ridge and related ground. Raising the target (a taller reconstructed mound) barely moved the corridor skyline percentage. WK can silhouette from the north. As a fraction of the defined approaches it is not Adam’s Grave.

East Kennet

Figure 7. East Kennet. About 19% skyline, similar to its neighbour. The cyan X is the nearer road pin (~678 m north-east of the crest): clearance ≈ +1.2°, a genuine high-contour catch. The south-west wedge is almost entirely red.

East Kennet’s road pin sits in a coherent positive-clearance pocket on the north-east side. The earlier sample was too far; at ~0.7 km the mound does silhouette broadside. That does not make the rest of the approach an Adam’s-class catchment. Most of the south-west wedge sees the crest against rising far ground.

So: two strong local markers, two famous neighbours that do not carry the same score. Same downs, different recipes — again.

Site	Axis	n	% visible	% skyline	Median clr	vs null
Adam’s Grave	136.0°	224	87.5	87.5	+2.66°	6.1×
Kitchen	43.5°	230	39.1	31.7	+0.59°	3.9×
West Kennet	84.5°	226	59.7	19.0	−0.29°	3.1×
East Kennet	139.0°	226	44.7	18.6	−0.13°	2.4×

Ten more, same test

I flagged prominent and good-survival sites, same perp-wedge method: Winterbourne Stoke Crossroads (WS1), Whitebarrow, Giants Grave, Cold Kitchen Hill, Tinhead Hill, South Street, Horton Down, Lanhill, King Barrow, Amesbury 42. (Cold Kitchen Hill is not Kitchen.)

Winterbourne Stoke 1

Figure 8. Winterbourne Stoke 1, the great surviving mound at Longbarrow Crossroads. About 66% skyline — every visible cell in this sample is also a skyline hit. Median clearance +0.53°. The ratio versus null is only 2.6× because ridge nulls on this ground are high (~25%). Still the standout of the second batch.

WS1 is the pilot-2 headline. Field already flagged the mound as a topographic object first. The skyline maps agree: from the length-shows wedges at road-to-ridge scale, it behaves like an Adam’s-class catchment, not a solar instrument and not a coincidence of a pretty photograph.

The rest of the ten

Figure 9. Ten prominent / good-survival mounds on the same frame as Figure 3. Green catchments at WS1, Giants Grave, Amesbury 42, Cold Kitchen Hill and Tinhead Hill. South Street and Lanhill are dark red throughout.

Giants Grave and Amesbury 42 sit in a second tier near 42% skyline, both with slightly positive median clearance.

Figure 10. Giants Grave, above the Wylye. One wedge is a coherent skyline catchment; the other is almost empty. About 42% skyline overall — Kitchen-class coverage, strongly one-sided.

Cold Kitchen Hill and Tinhead Hill land around 30%, Kitchen-class coverage with the strongest positive clearance cores in the batch (about +1.2° and +1.1°). Cold Kitchen Hill is a separate monument from Kitchen; it happens to score in the same band.

Figure 11. Cold Kitchen Hill (ST83NW101), not Kitchen. About 30% skyline; median clearance +1.21°, the strongest angular core in this batch. The north-east wedge carries the catchment; the south-west wedge does not.

Horton Down and King Barrow beat their local nulls but are weaker in absolute skyline cover (~23% and ~20%).

Whitebarrow is often visible and rarely the true crest-against-sky silhouette (~14%).

The fails

Survival and fame are not the same as skyline design.

Figure 12. South Street. About 55% of the broadside sample can see the mound. Skyline hits: 0.4%. Median clearance −0.53°. The crest sits against rising far ground from almost every length-shows cell.

Figure 13. Lanhill. About 51% visible; 0% skyline. A well-known surviving mound, length showing from the sides, never the local horizon peak in this frame.

South Street sits in the Avebury landscape, a short walk from West Kennet. Lanhill is a Cotswold–Severn chambered long barrow with a good earthwork. Neither behaves as a skyline marker from the approaches where the length shows. That is the point of doing the work this way: the method that lights up Adam’s Grave leaves these two dark.

Site	Axis	% visible	% skyline	Median clr	vs null
Winterbourne Stoke 1	35.0°	66.2	66.2	+0.53°	2.6×
Amesbury 42	5.0°	50.5	42.4	+0.24°	1.9×
Giants Grave	9.0°	46.5	42.0	+0.25°	2.2×
Tinhead Hill	64.3°	31.7	31.7	+1.05°	1.4×
Cold Kitchen Hill	148.5°	29.7	29.7	+1.21°	1.9×
Horton Down	97.7°	48.7	23.4	−0.14°	3.9×
King Barrow	161.5°	69.8	19.8	−0.51°	3.6×
Whitebarrow	91.7°	44.6	14.4	−0.14°	2.0×
South Street	68.9°	55.3	0.4	−0.53°	0.2×
Lanhill	90.5°	51.4	0.0	−0.29°	0.0×

Combined ranking (blunt)

Fourteen mounds. Rank by skyline fraction in the length-shows wedges. Numbers will move if the distance band, wedge width or vegetation model moves. The rank order is what I trust for now.

Class	Sites (approx. % skyline)
Strong	Adam’s Grave (~88%); WS1 (~66%); Giants Grave / Amesbury 42 (~42%); Kitchen (~32%); Cold Kitchen / Tinhead (~30%)
Mid	Horton Down (~23%); King Barrow / West Kennet / East Kennet (~19–20%); Whitebarrow (~14%)
Fail	South Street (~0%); Lanhill (0%)

A few mounds behave like length-on-skyline markers. A larger set of well-known names do not. West and East Kennet sit in the middle once you stop scoring them as corridor percentages and stop cherry-picking the sweet-spot photograph. The photograph remains true. It is not typical.

What that means

Visible on the ridge, like sunrise alignment, does not seem to be a factor — not a shared design rule for Wiltshire’s long barrows.

Local effects exist. Adam’s Grave and WS1 are not accidents in these maps. Kitchen, Giants Grave, Amesbury 42, Cold Kitchen and Tinhead show real catchments. That is careful siting and, in places, effective display.

There is no county recipe. The same method that lights up Adam’s Grave leaves South Street and Lanhill dark as silhouettes, and leaves West and East Kennet looking ordinary once you stop at the viewpoint that makes the postcard. Field, McOmish, Ruggles and Roberts already told us to expect local topography without a single bearing or a single display trick. The skyline scores say the same in a different dialect.

The two photographs at the top of this note are doing honest work. Figure 1 is what Kitchen’s southern catchment looks like from a real vale. Figure 2 is what West Kennet’s northern pocket looks like on a clear October afternoon. The maps are there so that neither picture has to carry a theory by itself.

Caveats (keep these next to the maps)

Bare-earth DTM — Environment Agency terrain without trees. Real woodland on approaches or crests would shrink catchments. Treat percentages as an upper bound.
Trees and chalk brightness — a fresh chalk mound would have been a louder signal than a green DTM bump; vegetation would have muffled it. Neither is modelled.
Chronology — Neolithic land cover and original mound height are not in the machine.
Nulls — raw skyline % on chalk scarps can flatter any high point. Contour-band and downslope checks matter when claiming “designed marker.”
Approach definition — results are conditional on the perp wedges and the 0.3–2.5 km band. A different sampling frame changes the percentages; the photo and pin ground truths remain valid for their viewpoints.
20 m county DTM — fine for vale→ridge lines of sight; crest detail comes from the 1 m chips. Coarse cells near the crest can smear local ridge texture.
This is not a ridge-axis test and not a solstitial horizon-altitude protocol. Do not mix the three.

What we are not saying

We are not saying no long barrow was ever seen on a skyline. Some clearly were, and still are. Figures 1 and 2 are two of them.

We are not saying topography did not matter. We are saying skyline display from length-shows approaches mattered locally and variously, when it mattered at all.

Orientation / ridge companion: Do Wiltshire’s long barrows face the sun?

Compilation © Tim Daw / sarsen.org · CC BY-SA 4.0.
Photographs © Tim Daw.
NHLE © Historic England / OGL · EA LiDAR © Environment Agency / OGL.
Roberts et al. 2018, Internet Archaeology 47, CC BY.
HER seed: Wheatley recreation 10.5281/zenodo.11005373.

With thanks

Field, McOmish and Brown; Ruggles; Roberts and the Historic England Stonehenge landscape team; Wheatley; the Wiltshire and Swindon HER; Historic England NHLE; Environment Agency LiDAR. The mistakes are ours.

The Devil’s Arrows and the glacier that probably didn’t

Clarke, Leary and Kirkland have published a detrital zircon–apatite provenance study of the Devil’s Arrows (Boroughbridge, North Yorkshire) in Proceedings of the Royal Society A. Open access: doi:10.1098/rspa.2026.0504.

The stones are Millstone Grit standing on Triassic Sherwood Sandstone; the nearest grit outcrop is about 5 km west, so the megaliths were moved — by people, by ice, or both. The paper fingerprints the three pillars to Brimham Rocks (~18 km west) rather than Plumpton Rocks, and states that regional ice-flow reconstructions exclude glacial transport from that source.

This updates an earlier sarsen.org note that followed secondary sources to Plumpton. That claim is superseded.

Grit on the wrong bedrock

Lithology already said the Arrows are not local. Sherwood Sandstone is red–brown, quartzose, Triassic fluvial–aeolian stuff; the pillars are grey, coarse, lithic Namurian Millstone Grit. Zircon agrees: two-sample KS against published Sherwood spectra gives D = 0.17, p = 2.9×10⁻⁴. The paper concludes the megaliths were not sourced locally.

Heights: south 6.86 m, centre 6.7 m, north 5.5 m. The paper specifies these as Britain’s tallest surviving standing stone row, with the southern stone the second-tallest British megalith after Rudston and the central stone the third. Erection date remains Late Neolithic to Early Bronze Age by convention — the paper states the “precise date remains unknown” for want of recent excavation. The U–Pb ages date the mineral grains, not the erection of the row.

People or ice?

Thorpe and Williams-Thorpe (1991) floated glacial erratics as an alternative to long-distance human haul — for the Arrows, grit moved south from the Northallerton area and dumped near Boroughbridge, then erected. Burl pushed back the same year. Plumpton Rocks (~13 km), with lookalike fluting, became the usual human source in the literature and in the NHLE entry. Brimham (~18 km west) was already on the menu as another grit tor field.

The paper’s contribution is isotopic fingerprinting combined with ice-flow-geometry analysis, drawing on the existing BRITICE-CHRONO and Veness et al. reconstructions, not new excavation.

What they actually sampled

Scheduled Monument Consent, granted 23 October 2024. Sampling used a peel method: 20 cm of adhesive tape pressed for five minutes onto dry, inconspicuous, lichen-free faces of the stones; grains released by dissolving the glue in turpentine, then separated by magnetic and heavy-liquid methods and mounted for LA-ICP-MS at Curtin’s GeoHistory Facility. Comparative Brimham and Plumpton samples were ~1 kg rock, cut and disaggregated by pulsed electric discharge (SelFrag) and sieved to <425 µm — a different preparation route from the peel method used on the standing stones, though a valid basis for comparing age spectra. Peel sampling recovers only near-surface grains; it is not a thin-section of the core, a clast count, or a sample from a demonstrated quarry scar.

Zircon is the workhorse. Stone 1 yielded 67 concordant ages, Stone 2 64, Stone 3 105 (n = 236 pooled). Pairwise KS and Kuiper tests between the three stones all give p > 0.05 (their Table 1): the spectra are statistically indistinguishable at 95% confidence, so the paper pools them into one composite dataset. High p means the test fails to reject “same distribution,” not that the spectra are identical — but the three pillars share one fingerprint family, not three unrelated erratics from three different outcrops.

Against Brimham (n = 75): KS D = 0.08, p = 0.77; Kuiper V = 0.121, p = 0.73. Against Plumpton (n = 76): KS D = 0.29, p = 6.7×10⁻⁵. Brimham and the Arrows share the same polymodal structure — dominant Caledonian (~460–400 Ma), Mesoproterozoic and Palaeoproterozoic components, minor Archean contributions. Plumpton, though the same Namurian Upper Plumpton Grit interval as Brimham, carries a different mix, dominated by more Silurian zircon. The paper attributes this to lateral heterogeneity already documented within the Pennine Basin’s fluvial channels.

MDS residual. In multidimensional scaling of Pennine Basin datasets (stress = 0.11), Brimham plots closest to the Arrows; Plumpton sits further off, pulled by its Silurian-dominated zircon. The least-dissimilar dataset with an overlapping uncertainty ellipse is a Langsettian sandstone at Binchester Crags, ~70 km north — the same broader northerly Pennine River signature. The paper is explicit that this places Brimham as the best match among the outcrops sampled, not as a unique source: distance, morphology and the identified local candidates still do the narrowing.

Apatite is secondary support. Most Arrow grains are reworked phosphatic bioapatite (low U, high Th/U, common-Pb-rich), not primary igneous provenance clocks; the paper reads their Pb-isotope signature as later diagenetic overprinting, not a source-rock age. A smaller detrital igneous subset does match Brimham’s apatite populations (~480 Ma, ~1300 Ma, ~2600 Ma) better than Plumpton’s.

Battle Cross and Peg Bridge masonry were also sampled. Battle Cross zircon carries the same dominant Caledonian and minor Neoarchean components as the Arrows; Peg Bridge zircon falls within the Arrows’ Neo- to Mesoproterozoic population. No grain from either source is inconsistent with the Millstone Grit fingerprint — support for the antiquarian record that both were built from a lost fourth Arrow.

Once it’s Brimham, ice has a geometry problem

The old erratic hypothesis had grit moving south from Northallerton and being dumped near Boroughbridge. Brimham sits west, on elevated Pennine ground outside that trajectory, so a Brimham source changes the transport question from southbound to eastbound.

Late Devensian ice-flow reconstructions for the Vale of York have ice advancing northwest to southeast, carrying material down-valley south — “rather than eastward from the Pennine uplands” (Catt 2007; Clark et al. 2022; Gibbard & Clark 2011; Veness et al. 2025). Brimham lies outside that principal trajectory; the paper calls the entrainment and eastward transport of >25-tonne blocks from there “improbable.”

The paper doesn’t treat the regional ice map as a fixed transport route: erratic journeys can be multistage as ice divides shift, and Veness et al.’s models are spatially coarse — they “cannot resolve or demonstrate a specific transport pathway” over the ~18 km from Brimham to Boroughbridge. What the paper argues is narrower: given a Brimham source, an eastward glacial haul is a poor fit to the reconstructed ice geometry, and all three stones sharing one outcrop-like spectrum is the wrong pattern for a grab-bag of glacial erratics.

Claim	Strength
Local Sherwood source	Ruled out
Plumpton as best tested grit match	Ruled out among sampled candidates
Brimham as best tested match	Strong
Unique Brimham among all Pennine channel sands	Not proven (Binchester ellipse; unsampled twins)
Principal Late Devensian flow carrying Brimham grit east to Boroughbridge	Contradicted by regional reconstructions
Some exotic multistage glacial path for these exact blocks	Not fully excluded by coarse models; called improbable
Human haul from Brimham	Least-bad remaining explanation once provenance + ice geometry are stacked

Not a proof that ice never moved grit in Yorkshire — the Vale has documented erratic trains of other rocks. A case that these three pillars, matching this one western tor field, do not look like ordinary Vale-of-York erratics.

The paper places the Arrows alongside other British and Irish monuments where stone was sourced from specific, often distant, locations rather than the nearest convenient outcrop: the Altar Stone traced to the Orcadian Basin (~750 km), the Preseli bluestones (~225 km), the sarsens forming Stonehenge’s trilithons (~30 km), the stone circles of Brodgar and Stenness, and the passage tombs of Newgrange and Knowth. At ~18 km, the Arrows sit at the modest end of that scale — which, if anything, makes deliberate human transport the easier case to argue, not the harder one.

What this still isn’t

No Brimham quarry face with tool marks or debitage. Provenance chemistry is not quarry archaeology.

No new erection date.

No resolution between practical and symbolic motive. The paper notes that favourable joints and bedding planes at Brimham would have reduced the effort needed to detach elongate blocks, and sits this alongside the site’s cup-and-ring rock art and its tor-like landscape as parallel, not competing, explanations for why this source was chosen.

Sources

Clarke A, Leary J, Kirkland C. 2026. Deliberate prehistoric sourcing of the Devil’s Arrows, Britain’s tallest stone row. Proc. R. Soc. A 482: 20260504. https://doi.org/10.1098/rspa.2026.0504

Thorpe RS, Williams-Thorpe O. 1991. The myth of long-distance megalith transport. Antiquity 65: 64–73. Burl A. 1991. Megalithic myth or man the mover? Antiquity 65: 297–298.

Ice context as cited by Clarke et al.: Catt 2007; Clark et al. 2022; Gibbard & Clark 2011; Veness et al. 2025.

Clarke AJI, Kirkland CL, Bevins RE, Pearce NJG, Glorie S, Ixer RA. 2024. A Scottish provenance for the Altar Stone of Stonehenge. Nature 632: 570–575.

Prior sarsen.org note (Plumpton; superseded): https://www.sarsen.org/2024/09/the-long-flight-of-devils-arrows.html

Tuesday, 8 September 2026

Do Wiltshire’s long barrows face the sun?

Tim Daw · sarsen.org · September 2026

The short answer is no — not as a shared design, and not as folklore likes. They do not all run along the ridges either. What they do, mostly, is spread.

Every arrow is on the map:

The Wiltshire Long Barrow Gazetteer

What follows is the evidence behind those arrows.

Standing on other people’s work

None of this starts from a blank field.

Clive Ruggles insisted that prehistoric “astronomy” has to survive a proper test, not a pretty sunrise. His reading of the Wiltshire chalk — and his demolition of Aubrey Burl’s lunar-arc claim for Salisbury Plain long barrows — remains the honest baseline: no clear common celestial alignment, and the land matters.

David Field, David McOmish and Graham Brown, walking the Salisbury Plain Training Area, said the same in plainer boots: long barrows sit with care in local ground, without a single shared bearing.

Around Stonehenge, Historic England’s landscape surveys — Bax, Bowden, Soutar, Field, Barber and colleagues — drew the mounds that are actually there, not the scheduling polygons. Dave Roberts and co-authors then put twenty-one of those WHS long barrows into Internet Archaeology 47: plans, dates, and a calm conclusion that local topography is the key to alignment, with a note that the more easterly end of many mounds seems to have mattered. They published the table.

Timothy Darvill, Corcoran, Ashbee, Kinnes, Piggott: the catalogues and the Cotswold–Severn distinction. Environment Agency LiDAR, Historic England’s NHLE, the Wiltshire HER extracts that David Wheatley’s Zenodo recreation opened for reuse. We add a county-wide, chip-checked set of axes on top of that work, not instead of it.

If this note is useful, it is because those people already did the hard looking.

What was measured

A long barrow has a long axis: the line of the mound, not the way a façade or forecourt “faces”. That line has two ends. We treat it as undirected — a bearing between 0° and 180° from north. Midsummer sunrise and midwinter sunset are the same line once you fold the compass that way. That alone should slow anyone down before saying “it faces the solstice”.

Historic England’s scheduling polygons give a first-pass axis, but the outline of a scheduled plot is not the crest of the mound. Environment Agency 1 m LiDAR can see the earthwork when it survives. Neither is gospel. Each of 128 Wiltshire gazetteer chips was reviewed by eye (a 129th row, the ploughed-out Cuckoo Stone long barrow by Woodhenge, was added from the records; nothing usable shows on the chip, so it has no arrow).

Eighty-six mounds have a trusted long axis. Forty-three do not: too faint, ploughed out, or not a long-barrow axis on the chip. Famous names sit in that second group — Fussell’s Lodge, Amesbury 14, Winterbourne Stoke 71, the Cuckoo Stone long barrow — which is a result, not a slight. If you cannot see the mound, you should not invent its bearing.

Those 86 arrows are what the map draws.

Figure 1. Eighty-six Wiltshire long-barrow axes, each drawn both ways. The dashed lines are sunrise on a flat horizon at this latitude: midsummer, equinox, midwinter. They are drawn so you can see them. They are not a finding.

What the 86 actually do

They spread.

There is a gentle pile-up near east–west. There is not a spike at midsummer (~50°) or midwinter (~129°). A proper circular test on undirected data does not reject a uniform scatter for the full sample. Restrict the set to the 80 earthen (Wessex) mounds and the scatter is even more ordinary. The seven peer-cited Cotswold–Severn chambered long barrows in the county are too few to carry a solar argument; several simply run roughly east–west — the mound, not the forecourt.

Around Stonehenge itself — the group Roberts and Historic England know best — there is no preferred bearing. That matches what they already said.

A weak east–west hint remains among HER-certain sites only. It is not a solstice, and it is not a tight cluster. It is the smear you would expect if people sometimes built along a down, sometimes across a spur, and often just along the mound they had room to raise.

Figure 2. The same 86 in ten-degree bins. The dotted line is what an even spread would look like. Midsummer sunrise does not win.

Score each barrow against whichever solstice is closer and the match looks better — until you give a random set of axes the same two targets. Chance is then just as close. That is the whole of the solar result.

Ridges, not a rule

So are they simply laid along the ridges?

We fitted a plane to the Environment Agency terrain in a ring around each mound (close enough to be “here”, far enough out that the barrow itself is not the hill). The long axes sit nearer the contour than chance would put them. They prefer the ridge line to the fall line. They do not glue themselves to the ridge. Half the sample is still more than thirty degrees off the local ground.

That is the picture Field, McOmish, Ruggles and Roberts described in words: careful siting, no single recipe.

Figure 3. Each point is one barrow: long axis against a fitted local plane (a blunt screen — see West Kennet below). If every mound followed that plane, the points would sit on the diagonal. They do not.

Winterbourne Stoke 1, the great mound at Longbarrow Crossroads, is the example Field already flagged. Its axis is about 35°, in the same NE–SW class Historic England surveyed. Midsummer sunrise is about 50°. The ridge of the later round-barrow cemetery runs that way too. Field wrote that the solar match may be coincidence given the topography. A wider look at the DTM says the ground immediately around the mound is almost flat. The land does not force a choice, and we should not pretend it does.

“But West Kennet runs along the top of a ridge”

It does. Anyone who has stood on the mound knows the Kennet is under the north flank. We are not denying that.

Two different sentences get stuck together here.

Where it is: on a spur of the downs, a nose of high ground with the river valley wrapping north and east. That is siting.

Which way the mound points: roughly east–west (84.5°), along the local crest of that nose.

East Kennet, a mile to the south-east and the same Cotswold–Severn tradition, is the foil. Historic England’s scheduling puts it below the crest of a north-east facing slope, long axis north-west/south-east. The eye on the mound is about 139°. That is across the ridge of the same downs, not along it. West Kennet rides the nose of the spur. East Kennet cuts across the high ground. Same tradition, two recipes.

Figure 4. West Kennet (black) along the ridge; East Kennet (brown) across it. Environment Agency terrain, about four kilometres across.

So the old description of West Kennet is fair at valley scale. A sceptic who says “it follows the ridge” is looking at the same ground we are. They cannot then say the Kennet long barrows as a class follow the ridge: East Kennet is the neighbour that doesn’t.

The county-wide computer test is blunter: the best-fit tilt of a ring of ground around each mound, so the barrow itself is not counted as the hill. A ridge falls both ways; the steeper or broader flank wins the plane. At West Kennet that flank is the rise westward onto the downs, so the machine reports the axis as “across the local slope.” At East Kennet the ring is the north-east hillside, so the machine reports the axis as along the scarp, not the watershed. Neither report is the named ridge. Believe the contours.

We are confident of this much:

West Kennet is on a ridge, and its long axis follows the nose of that spur.
East Kennet is on the same high ground, below the crest, and its long axis cuts across the ridge.
Neither is a midsummer alignment (84.5° and 139° are not ~50°).
Two famous mounds on one down do not make a rule for the other 84. The set as a whole is only modestly closer to local contours than chance, which is why Field, McOmish, Ruggles and Roberts refused a single recipe.

If the computer and the eye disagree on a celebrity barrow, believe the eye and the contours. The statistics are for the crowd.

Agreement with the surveyors

On the Stonehenge World Heritage Site, Historic England and Roberts et al. had already classed the long axes (NE–SW, W–E, and so on). Where we both measured the same mound, we land in the same quadrant. Winterbourne Stoke 1, Amesbury 42, the Wilsford and Figheldean groups, Netheravon Bake: the eye on the LiDAR chip is seeing what the earthwork survey saw.

That is the point of doing the work this way. The new county set is not a rival catalogue. It takes their standard of looking — mound, not paperwork — out across Wiltshire, including the unscheduled and the almost-ploughed.

Scheduling polygons remain on the map as evidence. They are sometimes right, and sometimes nearly at right angles to the earthwork. The arrows ignore those failures.

How to read the map

On the gazetteer, gold markers are HER-certain, grey are possible. Rim arrows mark the long axis where we could see one. A plain circle means we could not, and we have not borrowed a polygon to fake it.

Filter by earthen versus Cotswold–Severn (seven chambered sites, membership from the published lists only). Zoom in on a desktop and the 1 m LiDAR chip appears. The detail panel still shows the NHLE and auto-LiDAR numbers as witnesses.

If a name you love has no arrow, that is the honest state of the earthwork on the chip.

Same chalk, different seats

Positioning is not only axis versus ridge. A separate overlay of the gazetteer against Wiltshire’s hillforts asks a coarser question: do the two traditions sit on the same kind of ground?

They share the chalk. They barely share sites. Hillforts hug the scarps — higher, and much more locally proud. Long barrows sit back on the plateaux and gentle rises. The empty space between the pins looks like avoidance; geography explains most of it. The traditions are also about three thousand years apart, and several forts sit on or beside a barrow — the opposite of a taboo.

That comparison is a companion note (map + write-up): Do Wiltshire hillforts avoid long barrows?

Which end is the front?

The undirected long axis is still the main result above. A separate, lighter pass asked a different question: where a façade, forecourt, chambered end, or clearly broader/higher end can be named, which way does that end point?

NHLE and published notes give a firm label for only fourteen mounds. An eye pass on the same LiDAR chips used for the axes — looking for taper and labelled ends, not inventing fronts — raised a further twenty-one as broader-end candidates and left ninety-four undecided. So at best about a quarter of the gazetteer has a directed call. That is not the same sample as the eighty-six axes.

Among those thirty-five directed calls, twenty-eight sit in the eastern half of the compass (E, NE, or SE), five in the western half, and two due N or S. That is an eastern-hemisphere lean once a front is readable. It is not “they all face east,” and it is not a solstice spike: many of those fronts are simply the easterly end of an already east–west mound. Two labelled cases are higher or broader toward the west. Cotswold–Severn chambered façades are still not the mound long axis.

Treat this as a hedged aside, not a second solar paper. Readable taper and published labels are a biased subset. Broader-end calls are eye judgment on chips, not excavation. The undirected-axis result above does not change: the eighty-six still spread, without a shared midsummer or midwinter aim.

More research is needed before any directed façade claim can carry the county: fuller HER and excavation labels, a clearer rule for broader-end calls, and an honest test of whether the readable subset is biased toward already east–west mounds.

What we are not saying

We are not saying Neolithic people never watched the sky. We are saying this set of mound long axes does not show a shared solstitial aim.

We are not saying topography did not matter. We are saying it mattered locally and variously, which is what the people who walked these downs already told us.

We are not promoting broader-end eye calls into a county façade catalogue. Where NHLE or excavation already names an end, that stands; the rest wait for better evidence.

And we are not done. Ashbee’s and Kinnes’s printed lists, and a true viewshed against a wooded or bare horizon, are still sitting on the shelf. A first cut at skyline-from-below is a companion note: Do Wiltshire’s long barrows stand on the skyline? The hillforts overlay is a positioning addendum: Do Wiltshire hillforts avoid long barrows? The map will move as the gazetteer does.

Figures, tables and the full statistical note live with the technical companion: wiltshire-long-barrow-orientations.html.

Compilation © Tim Daw / sarsen.org · CC BY-SA 4.0.
NHLE © Historic England / OGL · EA LiDAR © Environment Agency / OGL.
Roberts et al. 2018, Internet Archaeology 47, CC BY.
HER seed: Wheatley recreation 10.5281/zenodo.11005373; Kutty 2024 10.5281/zenodo.10989406.

With thanks

Ruggles; Field, McOmish and Brown; Roberts, Valdez-Tullett, Last, Oswald, Bowden, Field, Barber, Bax and all the Historic England Stonehenge landscape team; Darvill; Ashbee; Kinnes; the Wiltshire and Swindon HER; Historic England NHLE; Environment Agency LiDAR. The mistakes are ours.