Searching for the origin of the Stonehenge Altar Stone

A paper now published in the Journal of the Geological Society reopens the question of where the Stonehenge Altar Stone came from. Pirrie, Bevins, Ixer and ten further co-authors give the most detailed description of the stone to date: sedimentology, petrography, heavy minerals, clays, geochemistry and isotopes. Their conclusion is that no outcrop of Old Red Sandstone (ORS) examined so far matches it on every count, and that its provenance "has yet to be determined with confidence".

Paper: Pirrie, D., Bevins, R.E., Ixer, R.A., Pearce, N.J.G., Turner, P., Power, M.R. et al. 2026. Searching for the origin of the Stonehenge Altar Stone: new sedimentological, petrographical, mineralogical, geochemical and isotopic constraints. Journal of the Geological Society 183, jgs2026-090. doi:10.1144/jgs2026-090
Data: figshare doi:10.6084/m9.figshare.33993488.v1

Fig. 1. (a) Map of the UK showing the location of Stonehenge and the distribution of non- marine sediments of the Old Red Sandstone. (b) Plan showing the position of the Altar Stone within the Stonehenge monument (c) Photograph looking towards the SW; the Altar Stone (arrowed AS) is lying mostly underneath the two large sarsen stones. (c)
Source: photograph by Adam Stanford.

Where the question stood

The Altar Stone is a block of about six tonnes (4.9 × 1 × 0.5 m) lying flat beneath fallen sarsens 55b and 156. For a century it was assigned to the Cosheston beds of West Wales, following H.H. Thomas in 1923. Bevins and colleagues refuted that in 2020, and in 2023 they excluded the Anglo-Welsh Basin as a whole: of 58 samples, all but four failed on geochemistry and the remaining four on mineralogy. In 2024 Clarke et al. (Nature) dated zircon, apatite and rutile grains from Altar Stone debitage and proposed the Orcadian Basin of NE Scotland. Clarke et al. (2026) later suggested Caithness or the Inverness–Black Isle area as the most likely sources.

What the new paper adds

The study combines previously published in-situ pXRF on the three exposed areas of the stone with new laboratory work on three museum fragments (MS1–MS3): automated SEM-EDS mineralogy, X-ray powder diffraction of bulk rock and clay fractions, cathodoluminescence, C–O isotopes and colour. Sr–Nd isotopes were measured on MS1 and MS2. The characteristics that emerge are:

• A bedding plane. Low-angle illumination in 2024 showed climbing ripples and slumped foresets on the exposed surface. If it is a bedding plane, the bed was at least 0.5 m thick and extended at least 4.9 m. There is no cleavage, and no petrographic sign of palaeosols or near-surface soil formation.
• Chrome spinel. The chrome spinel:zircon index (CZi) is 35.9–38.6 in the three fragments. In ORS comparison data it averages 1.3 on Orkney (n = 35), 1.4 on the Tarbat peninsula, Inner Moray Firth (n = 42), 0.4 on Foula and Shetland (n = 13) and 0.2 in the Lower ORS of Pembrokeshire (n = 32).
• Clay minerals. The clay fraction contains illite, mixed-layer illite/smectite, kaolinite, chlorite and tosudite. The illite/smectite is about 20% expandable in all three fragments. Expandability falls with burial and heating, and the ORS of Caithness and Orkney is below 10%, which excludes them. The clays alone would not exclude parts of the Moray Firth.
• Barium. Dispersed baryte cement gives pXRF Ba of about 1000–5500 ppm (mean 2758 ppm), linearly related to Sr. This separates the stone from ORS analyses in Wales and Orkney.
• Sr–Nd isotopes. ⁸⁷Sr/⁸⁶Sr is 0.7163–0.7169 and ¹⁴³Nd/¹⁴⁴Nd 0.5119–0.5121. Published Carboniferous sandstones (Leng et al. 1999) have either higher Nd or higher Sr isotope ratios, so there is little overlap. The Nd range of the ORS overlaps closely, but no ORS Sr isotope data exist, so the comparison cannot be taken further. The authors conclude that an ORS source is more likely than a Carboniferous one.
• Diagenesis. Pore-filling kaolinite pre-dates significant compaction; baryte post-dates compaction; both pre-date the dominant poikilotopic burial calcite cement.

What is excluded

• West Wales and the Anglo-Welsh Basin: excluded on earlier geochemistry and mineralogy, and in this paper by the absence of chrome spinel in West and South Wales and Somerset. Welsh Borderland tosudite occurrences are excluded on Ba.
• Orkney and Caithness: excluded on clay expandability. Chrome spinel is also scarce in the Orkney, Tarbat and Shetland samples; Caithness itself was not sampled for it.
• Moray Firth / Tarbat: compatible on clay, excluded on chrome spinel.
• Midland Valley ORS: excluded on abundant lithic grains (Phillips 2007). In the NE Midland Valley, the Stonehaven and Arbuthnott groups have too little I/S expandability, and the zircon populations of the Dunnottar–Crawton to Strathmore groups differ. The Strathmore Group clays are comparable, but its zircons and the texture of its tosudite are not.
• Southern Uplands: chrome spinel is common in the Lower Palaeozoic sandstones, but the Altar Stone contains kaolinite, which indicates it has not undergone low-grade metamorphism.
• Late Carboniferous–Triassic of south-central Britain: no Variscan-age zircons in the Altar Stone.

What remains open

• The ORS between the Midland Valley and the Welsh Borders, where modern heavy-mineral and zircon data are virtually absent.
• Areas with Ba anomalies in stream-sediment geochemistry, including the Lanark Basin and the Fintona Basin of Northern Ireland. The paper reports no chrome spinel, zircon or clay data for the Lanark Basin.
• Carboniferous sandstones recycled from Caledonide sources, such as the chrome-spinel-rich Clifton Rock of the Pennine Basin. Its zircon pattern differs (a large Early Palaeozoic peak, no Neoproterozoic group) and the Sr–Nd data favour the ORS, but the authors do not exclude it.

The diagnostic toggle map

These tests can be switched on and off in my Altar Stone diagnostic toggle map. Nothing to do with the paper or authors, I have just updated it with their data.  Each test is a layer: Ba/Rb screening, K-feldspar, zircon, clay and thermal maturity, strontium (L10, L11), and the chrome spinel (U1) and clay expandability (U2) results of Pirrie et al. The Pirrie layers start off; the preset "CZi + I/S ON (Pirrie et al. 2026)" switches both on. A place survives only if it fails none of the layers switched on, and untested is not the same as failed. Leave-one-out presets show how much each test contributes. The areas listed above as open, and the Sr–Nd result, are not scored on the map.

Relation to the NE Scotland proposal

The paper accepts the zircon, rutile and apatite ages of Clarke et al. and a Laurentian source for the heavy minerals. The argument is that the same age pattern can be shared by several basins, so ages alone cannot identify an outcrop, and that once chrome spinel, clays and barium are added the Orcadian candidates fail. Bevins and Ixer co-authored the 2024 Nature paper, so this is in part a team revising its earlier interpretation. The authors also regard the glauconite reported in one fragment by Clarke et al. (2026) as probably a misinterpretation of SEM-EDS spectra, and caution against relying on a single youngest zircon age.

Caveats

• Three fragments. The new analyses are on three small museum fragments, and the Sr–Nd data on two. Their attribution to the Altar Stone rests on earlier pXRF and petrographic comparison, because the stone itself cannot be sampled.
• Comparison data. The Pembrokeshire CZi value is from unpublished data, and the Orkney, Tarbat and Shetland values come from earlier compilations.
• Elimination, not identification. "No match found" covers only the ground that has been sampled. Large areas of ORS have no modern heavy-mineral or zircon data.
• Coarse Ba screening. National stream-sediment Ba maps are low resolution and can be distorted by historic mining.
• Sr/Ca. Surface pXRF Ca on the in-situ stone is lower than on debitage, which the authors attribute to weathering of the calcite cement. Sr/Ca from the surface pXRF data is therefore biased upward and should be treated with caution.

Why it matters

The origin of the Altar Stone determines the journey and the contacts it implies. A Moray Firth or Caithness source suggests a long coastal voyage; a Lanark Basin source, routes via the Clyde and the Irish Sea; Fintona, a crossing from Ulster. Until a source is identified, the origin is a variable in route modelling, not a fixed point.

The characteristics reported in the paper define the tests a candidate source must pass: high Ba with baryte cement, CZi of about 36–39, illite/smectite about 20% expandable with tosudite, plagioclase far exceeding K-feldspar, and an ORS-type Nd isotope signature. Those are the tests for the next field campaign.

Figures and short quotations are from Pirrie et al. (2026); see the paper and its supplementary tables for the data.