The antiquary John Leland counted four great stones standing in a row on the western edge of Boroughbridge in the middle of the sixteenth century. By the time William Camden reached the same field at the end of it, one of them was lying flat. John Aubrey, visiting in 1687, recorded that local people had broken a stone up and apparently worked part of it into Peg Bridge over the River Tutt.

Three stones are still upright, and they are the tallest surviving standing stone row in Britain. Only the Rudston Monolith stands taller than the southern one. Their heritage listing says the stones came from Plumpton Rocks, 15 kilometres west. The mineral grains lifted off the monument say otherwise. They point instead to Brimham Rocks, about 18 kilometres west, in a provenance study in Proceedings of the Royal Society A dated 9 September.

A tenth of a gram of grit

Sawing a sample off a scheduled monument is not on offer, so Anthony Clarke and Chris Kirkland of Curtin University, working with Jim Leary of the University of York, used what their paper calls a peel method. Lengths of adhesive tape 20 centimetres long were pressed onto dry, inconspicuous patches of each stone, chosen for carrying as little lichen and moss as possible. After five minutes the tape came away carrying whatever mineral grains had stuck to it.

Historic England, on behalf of the Secretary of State, granted consent on 23 October 2024. The glue was dissolved in mineral turpentine, the residue washed off in an ultrasonic bath, the magnetic minerals drawn away with a hand-held magnet and the quartz and feldspar floated out in a dense liquid. What was left came to no more than a tenth of a gram, and it was enough.

The clock inside a sand grain

Zircon is the useful mineral in a job like this. It survives being weathered out of one rock and buried in another, and the uranium and lead locked inside each crystal record when it first formed. A sandstone therefore carries grains inherited from whatever older terrain fed the river that laid it down. Rivers draining different country delivered different mixtures, which is what tells apart two gritstones that look identical in the hand.

The three Arrows yielded 67, 64 and 105 datable zircon grains, from the southern, central and northern stones in turn. The three age spectra showed no significant difference from one another at the 95 percent confidence level, so the team pooled them into a single population of 236 ages. Those run from 3,743 to 377 million years, dominated by a Caledonian peak at roughly 460 to 400 million years, with older clusters near 1,300 to 1,000 and 1,800 to 1,500 million years.

The stones are Millstone Grit, a group of resistant coarse sandstones laid down in the Namurian about 326 to 313 million years ago. The ground beneath the monument is Triassic Sherwood Sandstone, younger and lithologically distinct from the grit, and carrying a different zircon population. The nearest Millstone Grit of any kind crops out about five kilometres west.

The match with Brimham Rocks

Two candidate sources were sampled for comparison, about a kilogram of rock from each. Brimham Rocks produced 75 datable zircon grains, spanning 3,544 to 386 million years. Plumpton Rocks, near Knaresborough, produced 76, spanning 3,471 to 397 million years but weighted more heavily towards Palaeozoic grains and carrying fewer of the old Precambrian components.

Plumpton has been the traditional candidate for a reason: its fluted, weathered blocks closely resemble the surface textures of the megaliths. The paper places it about 13 kilometres southwest of the monument. The Historic England listing gives 15 kilometres to the west, so even the distance to the favoured source is not agreed.

Brimham is a close statistical match for the Arrows and Plumpton is not. A two-sample Kolmogorov-Smirnov test against Brimham returned a p value of 0.77, far above the 0.05 threshold at which two distributions would be called different. The same test against Plumpton returned 0.000067. A multidimensional scaling analysis, which arranges datasets by how dissimilar their age spectra are, placed Brimham next to the Arrows and Plumpton at a distance.

Apatite, a second and much scarcer mineral, pointed the same way. Most of the apatite grains, the paper judges, are reworked fragments of fossil phosphate rather than igneous crystals. The minority read as ordinary detrital grains share populations with Brimham at about 480 million years and near 1,300 and 2,600 million. Plumpton’s apatite has a more restricted population at about 480 million years, though it too carries a prominent component near 2,600 million.

Vale of York ice ran northwest to southeast

A long-standing rival explanation says nobody hauled anything. R. S. Thorpe and O. Williams-Thorpe argued in 1991, in a paper titled The myth of long-distance megalith transport, that blocks like these reached the lowlands as glacial erratics. On that account they were carried south from gritstone outcrops near Northallerton and then stood up roughly where they lay. Aubrey Burl replied in the same volume, and the question has stayed open since.

Reconstructions of Late Devensian ice flow have the glaciers of the Vale of York advancing from the northwest towards the southeast, moving material down-valley to the south rather than eastward off the Pennine uplands. Brimham Rocks sits west of Boroughbridge on high ground and outside the principal reconstructed flow path, which the authors read as making the entrainment and eastward carriage of large blocks from there improbable.

They add a second argument that does not depend on ice modelling. All three stones match one outcrop rather than presenting the mixed signature a random scatter of erratics from several gritstone exposures would produce.

The parts of the case that rest on similarity

A Kolmogorov-Smirnov p value of 0.77 is a failure to tell two distributions apart. That is weaker than a demonstration that they are the same rock, and it identifies a source population, not a quarry pit. The point cuts both ways, because Plumpton and Brimham sit in the same Namurian interval and are themselves distinguishable at p equals 0.0062, individual river channels having delivered sediment from different sources on a scale of kilometres. That heterogeneity is what gives the method its power, and it also leaves room for an unsampled exposure carrying a Brimham-like signature. Of the wider Pennine datasets compared, the least dissimilar one whose uncertainty ellipse overlaps the Arrows is a younger sandstone at Binchester Crags, about 70 kilometres north. The authors read that as the Arrows sharing the broad signature of sand carried by a Pennine river fed from the north.

The ice argument is firmer in the abstract than in the text. The abstract says regional ice-flow reconstructions exclude glacial transport from Brimham. The discussion concedes that a block need not follow one constant flow direction over thousands of years, that models integrating changing flow geometries are spatially coarse and sensitive to kilometre-scale shifts in ice divides, and that they cannot resolve or demonstrate a specific transport pathway across the 18 kilometres in question. Improbable and excluded are different words.

The same gap opens in a second place. The abstract lists shared surface weathering as further linking the Arrows to Brimham, while the introduction attributes the close resemblance of weathered surface textures to Plumpton and credits Brimham only with similar lithology and surface features. Whatever the weathering shows, it is not pointing only one way.

A third and smaller mismatch sits between the discussion and its own summary table, on the statistics separating the Arrows from the Sherwood Sandstone. The text gives a p value of 0.000035 where the table gives 0.00029. Both sit far below the threshold and the conclusion survives.

One number in the apatite results is worth setting aside entirely: a regression returning an age of 4 plus or minus 11 million years, which the authors flag as no geological event at all, only a property of lead-rich fossil phosphate.

The monument’s own age is still unknown. Archaeologists place it in the Late Neolithic to Early Bronze Age, and no recent excavation has narrowed that.

A battle cross with a compatible signature

The Battle Cross is a 5.5 metre stone cross commemorating the Battle of Boroughbridge, fought in 1322. It was probably carved in the fifteenth century, stood for centuries in Boroughbridge market square and was moved to Aldborough in 1852. The antiquary J. R. Walbran suggested in the nineteenth century that it had begun life as one of the Devil’s Arrows.

Zircon from the cross carries the dominant Caledonian and minor Neoarchean components that characterise the Arrows, and zircon taken from the masonry of Peg Bridge falls inside the Neo- to Mesoproterozoic population found in the stones. The paper says both support a shared provenance, and adds that no analysed grain from either is inconsistent with derivation from the same Millstone Grit source.

That final clause is doing real work, and this is the study’s most lightly evidenced limb. Compatible is not matched, and the paper reports no grain count in its main text for either the cross or the bridge. The reuse reading rests on three strands leaning together: Arrow-like dimensions, compatible ages, and the antiquarians writing down what they saw and heard. No one strand would carry it.

The choice of a further quarry

Sourcing stone from further away than necessary is a pattern rather than an oddity. Two of this paper’s authors helped trace the Stonehenge Altar Stone to northeast Scotland, at least 750 kilometres away, and Stonehenge’s bluestones travelled about 225 kilometres from southwest Wales.

Eighteen kilometres is modest beside that, and it was still a choice against convenience. The authors offer two reasons that sit comfortably together. Brimham Rocks is a landscape of weathered, sculptural tors that may have been read as an otherworldly place, and Late Neolithic to Bronze Age rock art lies just west of it. That includes a three-metre-long gritstone marked with wavy lines and around 21 cup-and-ring marks, described in the listing as about 21 cups with one of them ringed. Brimham is also a place where natural joints and bedding planes have already cut the rock into elongate blocks, so a monolith could be detached with less labour than a massive outcrop would demand.

The Arrows stand at the southern edge of the densest concentration of henge monuments in Neolithic Britain, seven of them between the Swale and the Ure.

What the row has never had is a date, and nobody can yet say how many stones it held when it was complete. Excavation could narrow the first of those, and the paper proposes none. The case that the alignment was once longer still rests on an accumulation rather than a proof: what the antiquaries wrote down, the postulated post holes running north towards the Thornborough henges, the possible stone socket excavated in 1993, two patches of magnetic disturbance. And now a cross in the next village, probably fifteenth-century, with the right minerals in it.