Stonehenge · England · 51.178889, -1.826111 · PYR record in the living atlas
The Altar Stone lies flat at the centre of Stonehenge, nearly five metres long, half buried under a fallen upright, and for two centuries everyone filed it as Welsh. In August 2024 a team led by Anthony Clarke and Christopher Kirkland dated the individual mineral grains inside it, and the answer came back from the Orcadian Basin in northeast Scotland: at least 750 kilometres away. The most studied prehistoric monument in Europe had the wrong address on its most central stone.
Five monuments wearing one name
Stonehenge was not built once. It was built, rebuilt and rearranged for roughly 1,500 years, and the phases barely resemble one another. The first Stonehenge, around 3000 BC, had no giant stones at all: a circular earthwork about 100 metres across with a ring of 56 pits, the Aubrey Holes, just inside the bank. Many held cremated human remains, since analysed as at least 63 individuals, roughly equal numbers of men and women with some children among them. Before it was anything else, Stonehenge was one of the largest cremation cemeteries in Neolithic Britain.
Around 2600 to 2400 BC came the transformation: a ring of thirty sarsen uprights capped with a continuous lintel, enclosing five colossal trilithons in a horseshoe. The bluestones were shuffled around the centre for centuries afterwards, and an avenue was cut linking the monument to the River Avon. What looks eternal was a work in progress for fifty generations, none of which left a written word about what they thought they were doing. The monument, with Avebury and the sites around both, was inscribed on the UNESCO World Heritage list in 1986.
The stones have passports
Geochemistry has done what the legends never could and traced the stones home. In 2020 David Nash and colleagues at the University of Brighton matched a core sample against sarsen outcrops across southern England, using X-ray fluorescence and mass spectrometry, and placed 50 of the 52 surviving sarsens at West Woods near Marlborough, roughly 25 kilometres north. Each upright weighs on the order of 25 tonnes; published weight estimates for individual stones vary between studies, and we would not put a superlative on any single one of them. They were moved across the downs without wheels or draught animals.
The bluestones weigh two to four tonnes each and fingerprint to specific outcrops in the Preseli Hills of Pembrokeshire, in west Wales, somewhere between 225 and 240 kilometres away depending on which account you read. Two of those outcrops have names and excavated dates: Carn Goedog and Craig Rhos-y-felin, both dug by Mike Parker Pearson's UCL-led team and published in Antiquity in 2019, with the extraction hollows still visible on the ground. The Altar Stone was always the odd one out, and the way it was tracked down shows how this work proceeds. In 2023 Nick Pearce and colleagues at Aberystwyth compared it against Welsh Old Red Sandstone and found no match at all, a negative result that got a fraction of the coverage. Only then did the search widen north, and the 2024 dating of the zircon, apatite and rutile grains put its parent rock in Caithness and Orkney. The authors argue for sea transport, since dragging it overland through that terrain is close to unthinkable.
Knowing where a stone came from is not the same as knowing how it got here, and one part of that argument is more open than it is usually made to sound. Human haulage is the mainstream reading and the quarry excavations support it. But the alternative, that glacier ice carried at least some of the bluestone material eastward and the builders collected it closer to home, has never been closed off, and it is still generating peer-reviewed papers. The most recent is Bevins and colleagues in 2025 in the Journal of Archaeological Science: Reports, on the so-called Newall boulder excavated at Stonehenge in 1924, which argues against a glacial origin for that particular piece. One paper is a contribution to an argument, not the end of it. We would rather record the transport mechanism as an open question than describe as discredited a position that specialists are still publishing on.
Look closely at the sarsens and you find carpentry. The lintels sit on mortise and tenon joints and lock to each other with tongue and groove: woodworking techniques executed in 25-tonne blocks of one of Britain's hardest sedimentary rocks, shaped with nothing harder than stone hammers.
The Welsh circle that kept shrinking
This site has a genuine alternative theory attached to it, and it did not come from the fringe. It came from Mike Parker Pearson, one of the most respected excavators working on the monument. In February 2021 his team argued in Antiquity that a stone circle 110 metres across at Waun Mawn in the Preseli Hills had been dismantled and carried to Salisbury Plain, with a pentagonal socket at one site matching a pentagonal bluestone at the other. A BBC documentary took it to a very large audience, and it is the closest anyone has come to putting evidence behind Geoffrey of Monmouth's twelfth-century story of stones brought from the west.
It has not held up, and the retreat began with the team itself, which concluded that the circle was never completed and that no more than about thirteen stones could have been taken from it, against Stonehenge's eighty bluestones. In 2022 Timothy Darvill published a detailed rebuttal in the same journal. Long stretches of the supposed circuit contain no stones and no sockets; the claimed midsummer alignment misses by around two degrees rather than being framed by the stones; only three of thirty-one radiocarbon dates fall in the right window; and, most damaging, no bluestone appears in the primary or secondary fills of any excavated Aubrey Hole, while micromorphology found no sign that stones ever stood in them. In 2024 Brian John concluded in The Holocene that the lost circle never existed, calling the reading interpretative inflation. Parker Pearson made a bold, testable claim, published it properly, and it failed the test. That is the system working, and it is worth more than a claim that never risked anything.
The axis that actually works
Set against all the contested astronomy, one alignment at Stonehenge needs no defending. The monument's main axis is arranged so that the midsummer sun rises over the Heel Stone in one direction, and the midwinter sun sets between the uprights of the tallest trilithon in the other. It is real, repeatable and verified every year by thousands of witnesses, and UNESCO treats it as part of what makes the site outstanding. It is also, on Clive Ruggles and Amanda Chadburn's assessment of the literature through 2024, the only specific astronomical alignment here that commands broad consensus. Everything else on the list is argued about.
The interesting part is which of the two mattered more. Excavation at nearby Durrington Walls, where the builders appear to have lived and feasted, found pig bones from animals slaughtered in midwinter, which points to the winter solstice gathering as the larger event. The version most visitors carry away may have the season backwards. The gatherings drew people from a long way off: strontium isotope analysis indicates that some of the individuals cremated and buried at Stonehenge did not grow up in Wiltshire.
The calendar argument, weighed
In 2022 Timothy Darvill proposed in Antiquity that the sarsen structures encode a perpetual solar calendar of 365.25 days. In his scheme the thirty uprights of the sarsen circle each stand for one day of a repeating thirty-day month, with two noticeably smaller stones dividing the ring into three ten-day weeks; twelve such months make 360 days; the five trilithons supply the five extra days; and the four station stones track the quarter day that accumulates into a leap year. He noted the close parallel with the Egyptian civil calendar and suggested the idea may have travelled.
The rebuttal came the following year in the same journal, from Giulio Magli and Juan Antonio Belmonte, and it is worth stating fairly because it is the stronger paper. Their objections are three: the scheme is numerology, since with thirty and five and four to hand you can build almost any calendar you like; the astronomy is wrong, because Stonehenge's alignments are not accurate enough to detect a quarter-day drift in the first place; and the Egyptian analogy has nothing behind it. A fourth point matters too: the count of thirty uprights is itself partly a reconstruction, because the circle is incomplete. Ruggles and Chadburn published a further rebuttal in 2024 in the Journal of Astronomical History and Heritage, on a different ground: the scheme only works if you combine parts of the monument that were not standing at the same time. The calendar reading is not absurd. It is simply not something the stones can be made to prove, and on the current state of the argument the rebuttals are the stronger case.
The moon, and a test that has just been run
The four Station Stones form a rectangle set inside the earthwork, and the long-standing suggestion is that its sides point to moonrise and moonset at the major lunar standstill, the extreme of the moon's swing along the horizon that recurs roughly every 18.6 years. Unlike most claims of this kind, this one has just been put to a direct test rather than argued from a plan. Across the standstill of 2024 and 2025 a team led by Fabio Silva at Bournemouth University, with Amanda Chadburn and Erica Ellingson, working with English Heritage and the Royal Astronomical Society, systematically observed and recorded moonrise and moonset against the stones through the whole period. The results were still being analysed as this was written. That makes the lunar question genuinely open: not established, not dismissed, and about to be answered by measurement rather than by assertion, which is the best position an old argument can be in.
What is still open
Who organised the work is unknown. The builders were the Neolithic and early Bronze Age communities of southern Britain, and that much is not in doubt. Whether a single political or religious authority directed the construction, or whether it was a collaborative undertaking spread across communities and generations, is not something the evidence currently settles, and the two possibilities describe very different societies.
Large parts of the interior have never been excavated with modern techniques. William Hawley's campaigns of roughly 1919 to 1926 covered a great deal of ground and were poorly recorded by the standards now expected; the Stonehenge Riverside Project of the 2000s and early 2010s did much to repair the picture but did not dig everything. The number, arrangement and dating of individual stone settings across the phases are still being revised, and some stone positions are known only from empty sockets in the ground.
And the healing hypothesis remains in play. Timothy Darvill and Geoffrey Wainwright have argued that the bluestones were brought here for properties believed to be curative, and that the monument functioned partly as a place people travelled to in order to be healed. The isotopic evidence for long-distance travel among those buried at the site is consistent with it. Consistent is not the same as demonstrated, and this is a mainstream hypothesis under discussion rather than a settled account of what the place was for.
What our engines found
We ran the same test here as everywhere else, with the settings fixed in advance and identical at every one of the fifty sites we cover: a 200 km search radius, at least 12 km between sampled sites, 15 sites, a 2 per cent tolerance, and 2,000 Monte Carlo runs against a null model resampled from real catalogued sites in the same region. The settings are frozen because the answer moves if you tune them, and tuning them after seeing the answer is exactly the trick we object to elsewhere. Stonehenge returned 44 golden-ratio pairs where chance predicts 33.3, giving p = 0.1469.
That is a little more than chance would predict, and it is not significant. The control points the same way: the arbitrary ratio sqrt(2) scored 1.44 times its expected count against phi's 1.32, so nothing here singles out the golden ratio.
We publish the misses because a tool that only reports its hits is not measuring anything. Every figure above is reproducible from the same coordinates, the same settings and the same seed.
One limit worth stating plainly, because it is the one most often glossed over. This test compares positions between monuments. It says nothing about any building's own proportions, its height, base, slope or perimeter. That is a separate question, and at several sites a genuinely open one. It also has nothing to say about the solstice axis or the Station Stones rectangle, which are questions about sightlines from one monument rather than distances between many.
Explore it yourself
Open the living atlas to see Stonehenge among its neighbours on Salisbury Plain, run the alignment and ley line engines on the real coordinates, and set the sky over the monument to a midwinter sunset in 2500 BC. And if you want the honest version on a shirt, the Stonehenge Site-Emblem T-Shirt carries the trilithon artwork, the site geometry and the site's true coordinates.
Sources worth reading
Clarke, Kirkland and colleagues, "Geological origins of the Altar Stone of Stonehenge", Nature (2024). Nash and colleagues on the sarsen provenance, Science Advances (2020). Parker Pearson and colleagues on the Carn Goedog and Craig Rhos-y-felin quarries, Antiquity (2019), and on Waun Mawn, Antiquity (2021). Darvill, "Keeping time at Stonehenge", Antiquity (2022), with the rebuttals by Magli and Belmonte in Antiquity and by Ruggles and Chadburn in the Journal of Astronomical History and Heritage (2024). Ruggles and Chadburn, Stonehenge: Sighting the Sun (Liverpool University Press, 2024), the current standard reference on the astronomy. Bevins and colleagues on the Newall boulder, Journal of Archaeological Science: Reports (2025). John, The Holocene (2024).