Monday, 24 August 2026

Rock on a Rope, how to level the lintels

 


In an earlier post on setting out the Stonehenge sarsens I suggested a simple way of establishing a horizontal: https://www.sarsen.org/2014/08/how-to-construct-level-sarsen-circle.html , place two posts either side of a small pool of water (an “Aurochs skin full of water”) and sight across their tops. Adjust the posts so their tops both have the same length of post above the water surface, the line of sight is level and can be transferred around the circle. 

A complementary approach starts from a different but equally basic physical fact: a free-swinging pendulum released from rest reaches essentially the same height on the opposite side of its arc. A tall central timber mast, a long fibre rope, and a heavy stone or timber bob are all materials the Late Neolithic builders already used. Release the bob from one marked position and it will rise to the matching height on the far side; adjust a second post until the bob just reaches it. For more accuracy, you release the pendulum from post A and it marks on post B the end of the first swing. When it comes back to post A the second time, it's going to be just a little bit lower because of air resistance. Halfway between the first and the second mark on post A will be the same as a mark on post B. Subtle and easy to get an accurate reading. Repetition is quick, free and easy so a reliable and repeatable measure can be recorded.  Rotate the plane of swing or shift the mast a short distance and a small set of levelled reference posts can be established around the intended circle. Once three or four reliable points exist, ordinary sighting or taut-string transfer finishes the job.

The two methods form a small family of “gravity levels”:

  • Water-pool / skin method – immediate visual reference, excellent for a local datum, but needs a still surface and a reliable container.
  • Pendulum method – needs no liquid, generates widely spaced equal-height points from one central mast, and is cheap to repeat when wind deflects the swing.

Neither is attested archaeologically. Both are simply possible with the technology of the time and address the practical problem of creating a consistent horizontal on sloping chalk so that the sarsen uprights could finish with level tops for the continuous lintel ring. And avoid the need for a water filled pig intestine or portable troughs as are often suggested as alternatives.


Pendulum method being tested


Friday, 21 August 2026

The Sarsen Transport Routes - A Least Cost Model

A terrain model you can argue with. Open this to understand this post: https://sarsenroute.netlify.app/ 


Fifty of the fifty-two sarsens at Stonehenge came from West Woods, near Marlborough. That was settled in 2020 by matching their chemistry to the source. It leaves the interesting part open: how thirty-tonne stones travelled 25 km south, across the Vale of Pewsey, to the monument.

Several routes have been proposed over the years, mine among them. None has field evidence behind it. So rather than argue for another line on a map, I set out to ask a narrower question:

If the only things that mattered were the shape of the ground, what it is made of, and the water on it — where would the cheapest road run?

The answer is not a line. It is a corridor: 37 km², about 5% of the landscape searched. Within that band the terrain barely cares which way you go.

Doing it blind

I am one of the people with a published route, which is a problem. So the method was written down and committed before any data arrived, and the published proposals — including my own — were sealed in a file that stayed closed until the result was frozen. No proposed route, mine included, is scored anywhere in the paper.


What the ground says

Four things came out that I did not expect.

The vale crossings are on sand, not clay. National geological mapping shows the Vale of Pewsey floor as one great smear of Gault clay and greensand. Zoom in to 1:50,000 and the five places where roads actually cross the vale turn out to sit on greensand benches — firm ground — with the real clay lying to the west, between Stert and Urchfont. Across all 360 model runs, not one metre of route touches Gault. The clay is real. Nothing goes near it.

There are only five crossings in eleven kilometres of vale — and adding every footpath and bridleway to the metalled roads adds no new ones. The same five places. When footpaths don't create new crossings, the crossings are being fixed by the ground.

Better data made a question harder, not easier. Two ways down off the Marlborough Downs, 2.8 km apart, are what the model has to choose between. At 50 m resolution they differ by 1.7% in cost. At 1 m, measuring the gentlest line the ground allows, they differ by 1.2 percentage points of gradient. More resolution made them more alike. Huish is not steep ground.

What separates them isn't the slope. It's use. One descent — Walkers Hill — carries a worn way running continuously from 219 m down to 140 m at a steady 6%, with the dished profile of a holloway. Andrews and Dury drew a road down it in 1773 and none down the other. The Huish face has a narrow farm track and nothing else.

Andrews and Dury's map of 1773

The bit I find most interesting

With no thumb on the scale, the model preferred the unused descent.

That isn't a bug, it's a bias — and I suspect it's general. A 50 m terrain model averages a scarp face. A smooth, unwalked hillside averages nicely. A hillside with a deep worn track in it averages badly. So the model kept choosing the easier- looking slope precisely because nobody had ever used it.

Feed in the evidence that one descent was used and the other wasn't, and it moves. But the model needed telling. It could not work that out from the ground alone.

What it can't do

A least-cost path is not a route. It knows nothing about ownership, tradition, season, weather or who you were allowed to walk past. It says what the ground permits, not what people did.

And it doesn't reproduce my own route. I think the stones came down Walkers Hill and crossed the vale beside Honey Street, on a low rise you can see on the ground. The rise is real — 1 m laser survey and the terrain model both find it, standing a couple of metres above the ground either side. What isn't shown is that it was made, or that anyone used it. The model doesn't go that way, and I've left that disagreement in the paper rather than tidy it out. A model that agreed with me would be worth less.

Have a go yourself

The model is online as a set of sliders: sarsenroute.netlify.app

Three hundred and sixty routes, pre-solved. Turn up the fear of hills, or of wet feet, or of river crossings, and watch the corridor move. There is no correct setting — that's the point. The route is a function of what you think a hauling party was most afraid of, and the tool makes that dependence visible instead of hiding it behind one confident line.

The full technical paper, with the method, the data and the things that went wrong, is on https://www.researchgate.net/publication/413532460_Modelling_the_sarsen_transport_route_West_Woods_to_Stonehenge_A_blind_least-cost_analysis_over_layered_open_data and https://www.academia.edu/172184320/Modelling_the_sarsen_transport_route_West_Woods_to_Stonehenge_A_blind_least_cost_analysis_over_layered_open_data.

Data: OS Terrain 50, BGS Geology, Environment Agency LIDAR (all Open Government Licence), OpenStreetMap. Analysis and drafting with Claude and Grok; the errors are mine.


Wednesday, 19 August 2026

Avebury 3D Fly Through

A 3D topographical fly through of the Avebury Landscape - https://avebury-3d-flythru.netlify.app/ . Play with it and get back to me with comments. It is schematic and only shows a sample of the monuments, but importantly it shows Silbury in relation to its neighbouring hills.

All built with Open Source Data so is freely shareable.

The various periods which you can switch between in the 3d model.





Click to expand


Tuesday, 18 August 2026

Why We Still Don’t Know Where Britain’s “Missing” Neolithic Ancestry Came From

I’m not a geneticist by trade. I’m an amateur who got stuck on a question that the big Beaker papers leave half-open, and then spent a long time trying to close it with public data only—the same call sets, tables and annotations everyone else can download.


The question sounds small. After the great ancestry turnover in Britain around 2450 BC, English Chalcolithic–Early Bronze Age genomes still carry roughly 7–9% ancestry that looks Middle/Late Neolithic-related. Where did that residue come from?

• Local British farmers absorbed into incoming communities?
• Or continental people who were already mixed before they arrived?

Those are different pasts. One implies survival and incorporation on this island. The other implies the “British-looking” fraction was already baked into migrant groups. Group averages hide the difference. You need to know whether almost everyone carries a thin layer, or whether a few people carry a lot and most carry none.

I thought public IBD, f-statistics, Y/mt labels, and the shape of published ancestry estimates would settle it. They don’t. What I think I have produced—and why I’m bothering to post—is a clearer map of why they don’t, and therefore where the problem has to be solved next. Not a funding pitch. A diagnosis.

What I actually found

1. The residue looks population-wide, not a mosaic of survivors.
On the Booth × Olalde overlap (n = 28), once one extreme individual (I2462) is set aside, the group is statistically homogeneous at about 7.15%. Between-person scatter is small; models where only a minority carry the component and everyone else carries none are excluded. Booth’s gradual rise through time reproduces—and it is a rise in everyone. That is a real result, and as far as I can tell it hasn’t been stated this way with a scripted dispersion bound behind it.

It still doesn’t tell you provenance. An already-mixed import is uniform from day one. A local pulse absorbed over ten to fifteen generations is uniform by the time we sample Early Bronze Age people. Same snapshot. Different histories.

2. The “trace it home” instruments fail on public products.
Long shared segments (IBD), allele-frequency contrasts aimed at British private drift, and uniparental “insular” screens all break under ordinary audits: leave-one-out, site pruning, shared-depth truncation, positive controls. The discriminating information—rare variants, fine Y structure, long-segment continuity at time depth—isn’t reliably in the 1240k / AADR-shaped data we all use. Sometimes the burial still holds it; the assay and the label conventions don’t.

3. Those are two different kinds of failure.
Three instruments fail because information was discarded (capture + harmonisation). The residue-shape instrument fails because information was erased (admixture before these people lived). Collapsing both into “we need more aDNA” is how you waste the next decade.

I also built a Sardinian re-imputation instrument to ask whether public IBD nulls are partly old pipelines rather than ascertainment. It validates within poles; the decisive cross-pole test is specified and not run due to the constraint of my available resources.

Where and how to solve it (not “who should fund it”)

If the goal is provenance of that 7–9%, the map is annoyingly specific:

A. Where the assay threw the information away
Hold the same individuals and change the data product: shotgun (or denser) sequencing → modern imputation → rare-variant / fine-lineage / IBD methods. Especially useful where the same person already exists in both capture call sets and shotgun releases (a clean ascertainment test). Published library-quality metrics for the people I’d prioritise suggest this is practical for most of that set, not all—some look too thin on public annotation alone.

B. Where time erased the information
Don’t keep sequencing well-sampled Early Bronze Age individuals expecting residue structure to reappear. Sample closer to the transition—Chalcolithic / earliest Bronze Age—where an absorbed local pulse might still show heterogeneity. That is a dating-and-excavation problem as much as a sequencing problem.

C. What not to confuse with a solution
More Beaker genomes won’t answer this. Ireland isn’t a free continuity control across the same interval. Collapsing Y strings to “I2” and calling it local doesn’t survive a shared-depth check. And a homogeneous ~7% background is evidence about structure, not a passport stamp for “British Neolithic survival.”

Why post this as an amateur?

Because the literature often states the residue, then slides past the provenance question—or treats every failed test as “low power” instead of “wrong information class.” I’m not claiming a final β. I’m claiming a usable negative architecture: four instruments, two kinds of limit, and a concrete split in where the next honest attempt has to go.

The draft paper: 


Full deposit (pre-registrations, results, code, audit trail):

If I’ve got something wrong in the genetics, I want to hear it. If the diagnosis is roughly right, the next useful work isn’t another average—it’s either denser data on the right people, or older people on the right dates.

───

• “Uniform at 7% is a result. It just isn’t a provenance result.”
• “Some missing answers are missing chips. Some are missing centuries.”
• “I’m an amateur. The deposit is public. Please break it if you can.”

Friday, 14 August 2026

The Plain Truth About Wildfires

 

Chalk downland has three available trajectories, and only one preserves the habitat as designated. It can be grazed to a standard. It can burn. Or it can be under-grazed, in which case succession returns it to coarse grass, scrub and ultimately wood — though the route to wood runs through a long phase of high fire hazard, so the third option tends to collapse into the second. The largest chalk grassland in north-west Europe has been demonstrating what the second produces for the better part of a century.

The habitat is semi-natural, a product of clearance and sustained grazing rather than undisturbed succession. The sheep-and-fold system held the downs open by grazing the hill and folding the flock on valley arable, exporting nutrients from hill to field year on year. That impoverishment is the basis of the botanical interest: low herbs persist because nothing vigorous can establish to shade them out. The habitat therefore has no equilibrium of its own.

Prescriptions have shifted over the last two decades from a uniformly short sward with zero tolerance of scrub toward structural mosaics, on the sound grounds that short turf serves a narrow suite of species and little else. Marsh fritillary and Duke of Burgundy, both notified features of Salisbury Plain, require tall tussocky vegetation and successional scrub margins. Structural diversity, however, means standing dead material — upright brome and tor-grass, unpalatable past early summer, accumulating a dense basal layer annually, and favoured by nitrogen deposition into a system whose interest depends on being nutrient-poor. In spring the taller sward retains moisture; in drought the same material cures into fuel.

The uncontrolled experiment

Salisbury Plain Training Area holds roughly 390 square kilometres of chalk under military control since 1897, with the SSSI, SAC and SPA covering 19,690 hectares and notified for twenty-five features. Grazing is logistically constrained: stock must be enclosed and moved around firing schedules, and research on the Plain has largely had to work on ungrazed swards. In the impact areas the constraint is absolute. Unexploded ordnance excludes the fire service from range danger areas, and the standing position on impact-area fires is to suspend firing, monitor, and let them burn out; water cannot be dropped directly, since low flying risks detonating ordnance by heat. Figures obtained under Freedom of Information put wildfires linked to MoD training sites at 1,178 between 2020 and August 2025.

Fire as a management agent

Natural England's Integrated Site Assessment of the SSSI (2014–15) records that wildfires in the Central Impact Area help produce species-rich swards in places by removing thatch build-up, and that Centre for Ecology & Hydrology work there has shown pronounced cyclicity in devil's-bit scabious — the marsh fritillary's foodplant. The feedback is straightforward: undergrazing accumulates dead material, dead material carries fire, fire removes it, herbs recover. The revegetating shell craters produce short-sward grassland supporting rare plants including early gentian, and are thought to act as fire refugia for marsh fritillary. The ordnance provides the firebreaks.

Fire and grazing are not equivalent instruments. Grass fires remove biomass without discrimination — work on grassland invertebrates on the Plain characterised burning and mowing as indiscriminate in what they take, against the selective defoliation a grazing animal provides. Fire takes tussocks and anthills along with the thatch, and anthills do not resprout, nor do the twelve nationally rare and scarce bryophytes for which the Plain is notified, nor the lichens. Woody growth, by contrast, is only top-killed: hawthorn and blackthorn survive at the root and return.

What the regime selects for

The same assessment describes the fires as an unreliable and potentially hazardous management tool that possibly promotes tor-grass. It records tor-grass as having increased substantially on the Central Impact Area since the mid-1990s, with no control available other than grazing and vehicle compaction along tracks; a CEH aerial-photograph study commissioned alongside it found the extent of Brachypodium rupestre significantly expanded there over the preceding decade. Natural England classes tor-grass as a negative indicator species.

Tor-grass is rhizomatous, unpalatable, litter-forming and a vigorous resprouter — a grass optimised for precisely this regime. The trajectory in the impact area is therefore not toward scrub but toward a fire-maintained monoculture: structurally open, botanically impoverished, and self-sustaining, since each burn returns the ground to the species best placed to reoccupy it.

This is the grass–fire feedback operating as expected. Fire is a weapon of grasses, implicated in the expansion of open grassy biomes through the late Miocene, but a weapon for grasses carrying the requisite traits — and on the chalk the grass holding it is the one under suppression. The prairie and steppe analogy fails for a further reason: those systems are grassland because the climate will not carry closed forest. The Wessex chalk sits within the woodland envelope, which is why the downs require management at all, and lightning in oceanic Britain largely arrives with rain. The chalk flora carries no fire-adapted traits of consequence — no serotiny, no fire-cued germination. Against the 2002 condition assessment, the proportion of grassland parcels passing all mandatory attributes rose from 43% to 62%, an improvement Natural England attributes principally to scrub management rather than to any natural process.

Prescription, not abandonment

The term for the third trajectory matters. Withdrawal of grazing on the chalk is not usually abandonment in the economic sense; it is frequently prescribed. On the Plain, grazing runs through Farm Management Plans with some forty-five tenant farmers, which require long grass margins, specify maximum rather than minimum stocking levels, and in special-restriction areas limit grazing to a mapped proportion of the area per year. The basic stocking rate on the chalk grassland is 0.3 livestock units per hectare per year. Natural England's own assessment notes that the plans emphasise avoiding over-grazing but do not adequately emphasise avoiding under-grazing or litter removal — and it identified under-grazing as the principal threat to the chalk grassland, flagged on 83 of the 282 parcels surveyed.

The standing fuel load is therefore not simply what happens when farming retreats. It is in substantial part a consequence of prescription, and graziers reducing numbers are often doing what they have been asked to do.

The transitional hazard

Reduced grazing is no safer a route than fire, for structural reasons. Succession from open sward to closed canopy passes through a phase carrying both fuel types at once: a continuous fine fuel bed of rank grass and litter, and above it establishing hawthorn, blackthorn and dogwood holding fine dead material. Fine fuel governs ignition and rate of spread; woody fuel governs intensity and residence time. The transition holds the maximum of both, and the hazard falls away only at canopy closure, when shading suppresses the sward and fine fuel continuity breaks. On chalk that window opens once scrub establishment is underway, which is the phase under-grazed downland and much recent rewilding now occupies.

The general finding is well supported: withdrawal of grazing drives woody encroachment, encroachment increases flammable biomass, and increased fuel load raises both wildfire probability and intensity, with herbivore grazing identified as the corresponding mitigation.

The choice

Grazing to a standard maintains the notified interest, at cost and with continual attention. Under-grazing leads to woodland, but only across two decades of elevated hazard during which the transition is liable to be interrupted — and interruption returns the ground to the fire-adapted grass. Fire holds it open indefinitely and without expense, and converts it to something structurally similar and botanically far poorer.

Fire is not a third route to preserving chalk grassland but a third outcome, and the one that arrives by default. That it arrives partly by prescription is the part worth examining.

Sunday, 9 August 2026

How big a pit was 10067?

 



When the A344 was lifted in 2013–14, Wessex Archaeology examined what was left beneath the road bed. Seven and a half metres north of the Heel Stone they found a feature, 10067, surviving 0.06 m deep, three finger widths. It took some skill to notice it at all, and there was not a lot to record. Powell et al. interpreted it as a shallow tree-throw hollow.

The interesting question is not what survives but what was cut. Nobody digs a hole six centimetres deep. Whatever 10067 was, what was found is the very bottom of it, and how big a pit it was depends entirely on how far below the contemporary ground surface that bottom sits.


Hawley's photograph, 10067 is the dark mark just this side of the motor car


10067 is marked with a dark mark this side of the Heelstone

The road makes that hard to establish. Until it was tarmacked in 1935, a chalk road was maintained by scarifying, watering and rolling. That takes out the ruts and gives a remarkably smooth surface, at the cost of removing chalk each time, so roads of this kind sank slowly. The early photographs show a dark patch where 10067 lies — a soft spot in the road surface, which is what you would expect from a deep feature with looser fill compacting under traffic. The looser top surface was removed before the road was tarmacked taking it down to a firm surface. On top of all that, the corridor was machine-stripped for the watching brief. The surface Powell recorded from is not the surface 10067 was dug from.


The 2013 excavation - 10067 is marked with the metal stakes, note the section revealed at the bank.

Levels

Two Ordnance Survey benchmarks are cut into the Heel Stone. The lower one is a broad arrow with the levelled bar above it, about 0.1 m above the turf, published at 100.70 m OD. Pitts tied his 1979–80 levels to it, so his sections can be read directly against the stone.

That gives a run of comparable figures for the entrance zone:

  • Pit 97 — about 1 m deep, base 1.60 m below the mark, so about 99.1 m OD. Pitts's fig. 7 is contoured in centimetres below the benchmark, which is why this one is solid.
  • Heel Stone ditch — cut about 0.8 m into the chalk, base just touching below 99 m (Pitts fig. 8).
  • Avenue ditches — 0.75–0.80 m deep, base near 99 m, and consistent between the Vatchers, Pitts and Powell.

Three features, two campaigns, all bottoming out around the same horizon.

Pitt's Fig. 7 - Stonehole 97

Pitt's Fig. 8  Heelstone Ditch - note 99m OD mark

Pitt's Fig. 13 - Avenue Ditch - note 99m OD mark

Click to enlarge.


Where 10067 sits

Not, unfortunately, from the published levels. Powell's GPS heights do not reconcile with the earlier work. He records the natural at the edge of the Heel Stone ditch, inside the road-line, at 102.13 m — that is 1.43 m above a benchmark cut just above ground level at the stone a few metres away. It is not merely an offset, it is the wrong way round: the road ran below the ground at the stone, not above it. Powell notes the inconsistency and leaves it unresolved. So do I.

Which leaves the photographs. The verge in my photograph of the stripped corridor shows the old road in section with the turf line above it, and the perimeter fence for scale — standard chain-link, probably1.8 m. Reading the drop against that, the surface Powell was working from lies something like 1.2 m below the turf beside the Heel Stone, so around 99.4–99.5 m OD. The base of 10067, a further 0.06 m down, comes out somewhere in the region of 99.3–99.4 m.

That is a reconstruction from photographs, not a survey, and it is worth a couple of tens of centimetres either way. But it is enough for the point. The base of 10067 sits close to the level at which Pit 97, the Heel Stone ditch and the Avenue ditches all stop. And it was cut from a surface a metre or so above where it was found, which makes it a substantial pit rather than a scrape.

So what was it?

I am not going to claim it is definitely a stone-hole. The evidence will not carry that, and even if we knew the actual OD of the surface it was found in, and the archive might be able to reveal that, we would still only be able to improve its comparison to the other pits.

But a pit of that size, seven and a half metres north of the Heel Stone, bottoming at much the same depth as 97, is not obviously a tree throw either. It sits in a zone already full of holes nobody can explain: 97, B, C, the A holes, the Heel Stone's own ditch. Stones put up and taken away? The Heel Stone shifted about like a pawn from one hole to the next? Whatever was going on here, 10067 and the other holes deserve deeper scrutiny.