1914 Geological Survey sketch map showing the Sarclet Dome and the structural relationships of the Old Red Sandstone units (conglomerates, sandstones and flagstones) along the East Caithness coast between Wick and Lybster (Crampton & Carruthers 1914)
Review of Clarke et al. (2026) “From Highlands to Henge” and Daw (2026) Geochemical Screening — With Full Data Tables and Reproducibility Notes
Executive Summary
Clarke et al. (2026) is a valuable intra-basin refinement of the Orcadian Basin provenance established in Clarke et al. (2024). It uses openly published detrital zircon U-Pb data and a two-sample Kolmogorov–Smirnov (KS) test to rank sampled localities. Sarclet is claimed as the strongest statistical match (claimed p = 0.96), with Braemore, Kirtomy and Portskerra also compatible.
Independent replication confirms the broader claim (four of five localities are statistically indistinguishable) but does not reproduce Sarclet as uniquely strongest. Braemore matches comparably or better. More critically, a bedrock cross-check against the BGS Geology 625k map reveals a stratigraphic mismatch: Sarclet and Braemore (strongest zircon matches) are mapped as Lower Old Red Sandstone, while the Altar Stone is established as Middle Old Red Sandstone. Kirtomy (confirmed Middle ORS) has the weakest passing p-value. This tension is not addressed in Clarke et al. (2026).
Daw (2026) — “The Stonehenge Altar Stone: Screening the Orcadian Basin” — provides independent, convergent support via a basin-wide stream-sediment Ba/Rb geochemical screen + bedrock verification. It identifies a 42.5 km² district-scale anomaly in East Caithness (~10 km southwest of the exact Sarclet Harbour zircon sample point) that is 98.2% genuine Middle Old Red Sandstone. This improves the Sarclet claim by using an entirely independent method, explicitly handling the Lower vs Middle ORS nuance, ranking the East Caithness cluster highest, and providing full reproducibility with clear limitations stated.
Overall assessment: The convergence of two independent methods on the same short stretch of East Caithness coast strengthens the case for prioritising this area. However, both papers operate at formation/age scale rather than confirmed facies scale. Neither establishes a specific quarry or direct match. The claim remains a high-priority triage/ranking result requiring targeted field petrography, mineralogy and new zircon work. Daw (2026) is the more robust and cautious of the two contributions on this specific point.
1. Review of Clarke et al. (2026) – Data, Methods and Claims
1.1 Core Claim Audited
Among the Orcadian Basin localities tested from Strachan et al. (2021), Sarclet returns the strongest detrital zircon U-Pb age-spectrum match to the Altar Stone (KS p = 0.96). Braemore, Kirtomy and Portskerra are also statistically compatible (p > 0.05); New Aberdour is weaker (p = 0.08).
1.2 Independent KS Replication Results
Full replication was performed using the same openly published grain-level datasets (Clarke et al. 2024 Nature SI; Strachan et al. 2021 JGS SI). The Altar Stone filter (≤10% discordance on Concordia Age) yields exactly n = 56 grains. Strachan grains were filtered to Group ∈ {S, Y} using the authors’ own Preferred Age column.
| Locality | n (grains) | KS D | Replication p-value | Clarke et al. (2026) reported |
|---|---|---|---|---|
| Sarclet | 44 | 0.114 | 0.865 | p = 0.96 (strongest claimed) |
| Braemore | 42 | 0.113 | 0.889 | p > 0.05 (not quantified) |
| Kirtomy | 46 | 0.192 | 0.266 | p > 0.05 |
| Portskerra | 46 | 0.187 | 0.293 | p > 0.05 |
| Watch Hill | 34 | 0.234 | 0.163 | Not mentioned in Clarke et al. |
Note: Braemore matches as well as or better than Sarclet across discordance thresholds and resampling scenarios. The exact p = 0.96 for Sarclet is not reproduced under standard point-estimate KS (likely difference in error-resampling implementation). Watch Hill (available in Strachan data but omitted from discussion) also passes the compatibility threshold in replication.
1.2.1 Replication Methods Summary
The two-sample Kolmogorov–Smirnov tests were implemented in Python using scipy.stats.ks_2samp (two-sided). Altar Stone grains were filtered exactly as described in Clarke et al. (2026): ≤10% discordance on the Concordia Age column, yielding n = 56 grains across the three analytical blocks (MS3, 2010K.240, FN593). For the Strachan et al. (2021) localities, grains were retained where the source table classified them as Group S or Y (concordant detrital) and the Preferred Age column was used; no additional percentage-discordance filter was imposed beyond the authors’ own classification. Point-estimate KS D-statistics and p-values were calculated directly on the filtered age lists. Sensitivity to discordance threshold was checked at 5%, 10%, 15% and 20%. A full Monte Carlo error-resampling implementation matching the precise parameters used by Clarke et al. (2026) could not be replicated because those parameters are not fully specified in the published text; the point-estimate results reported here are therefore conservative and directly reproducible from the cited public supplementary tables.
1.3 Bedrock Stratigraphic Cross-Check (BGS Geology 625k)
Every locality coordinate was tested by point-in-polygon spatial join against the national BGS 625k bedrock layer (LEX_D formation name, RCS_D lithology, MAX/MIN_PERIOD chronostratigraphy). A cell is genuine Old Red Sandstone if age includes Devonian and lithology is sedimentary (excluding igneous/metamorphic).
| Locality | Exact point result | Nearest ORS polygon | Distance | Period | Note |
|---|---|---|---|---|---|
| Sarclet | No polygon (harbour gap) | LOWER OLD RED SANDSTONE | 2.6 m | Devonian | Basal Sarclet Group (Lower ORS) |
| Braemore | LOWER OLD RED SANDSTONE | LOWER OLD RED SANDSTONE | 0.0 m | Devonian | Lower ORS |
| Kirtomy | MIDDLE ORS (undifferentiated) | MIDDLE ORS (undifferentiated) | 0.0 m | Devonian | Confirmed Middle ORS |
| Portskerra | LOWER OLD RED SANDSTONE | LOWER OLD RED SANDSTONE | 0.0 m | Devonian | Lower ORS |
| Watch Hill | Lewisian Complex (basement) | OLD RED SANDSTONE SUPERGROUP | 301.6 m | Sil-Dev | Basement, not genuine ORS |
Critical finding: The two strongest zircon matches (Sarclet & Braemore) are independently mapped as Lower Old Red Sandstone. Kirtomy (the only locality confirmed as Middle ORS in the tested set) has the weakest statistical compatibility. The Altar Stone itself is established as Middle ORS. This stratigraphic tension is not discussed in Clarke et al. (2026).
1.4 Strengths and Limitations of Clarke et al. (2026)
- Strengths: First quantitative intra-basin ranking using open grain-level data; identifies Sarclet as a priority; ice-flow modelling correctly concludes no viable direct glacial pathway (treated as speculative).
- Limitations: Does not reproduce Sarclet as uniquely strongest under independent standard KS; sparse sampling (only 5 localities in ~10,000 km² basin); no explicit stratigraphic cross-check against national bedrock mapping; exact p = 0.96 source remains unresolved.
2. Review of Daw (2026) Geochemical Screen – How It Improves the Claim
2.1 Method Summary
Daw (2026) performs a fully independent, basin-wide, reproducible screen using BGS G-BASE 500 m stream-sediment Ba and Rb grids. Composite condition: Ba ≥ 1025 ppm (from Bevins et al. 2023 Altar Stone pXRF) AND Ba/Rb ratio ≥ basin P95 (13.76). Cells are clustered (8-connected, ≥0.75 km²), then every passing cell is verified pixel-by-pixel against BGS 625k bedrock (genuine ORS = Devonian age + sedimentary lithology, excluding igneous/metamorphic). Core grid-screening was independently replicated by Grok (xAI) in Appendix B of the paper (matches P95 threshold, cell count, area, mean Ba and ratio to high precision).
2.2 Key Results and Convergence with Zircon Evidence
| Cluster | Area (km²) | Genuine Middle ORS % | Mean Ba (ppm) | Mean Ba/Rb | Distance to Sarclet Harbour zircon point |
|---|---|---|---|---|---|
| East Caithness (near Sarclet) | 42.5 | 98.2% | 1453 | 18.2 | ~9.7 km (nearest cell ~6 km) |
| Shetland (Melby/Walls) | ~28 | 85% | 1285 | 18.6 | N/A (secondary priority) |
| Loch Duntelchaig | 14.75 | 40.7% | 1076 | 17.9 | N/A (poor facies match) |
| Nairn/Elgin & Helmsdale | Partial | <50% | — | — | Not priority |
2.3 How Daw (2026) Strengthens and Refines the Sarclet Claim
- Independent convergence: Identifies a district-scale geochemical anomaly on the same short stretch of East Caithness coast as the Clarke zircon hotspot, using completely different data (stream sediment chemistry + bedrock polygons) and no pre-existing sample locations.
- Better stratigraphic handling: Explicitly uses the GeoGuide, Strachan et al. (2021) and BGS 625k evidence to show that while the exact Sarclet Harbour GCR site is Lower ORS, the same coastal belt passes conformably into fine-grained lacustrine flagstone facies (Caithness Flagstone Group, Middle ORS) within a few kilometres. The geochemical hotspot sits in ground where this compatible facies is documented.
- Downplays other Clarke hotspots: Kirtomy and Portskerra do not produce equivalent strong, pure Middle ORS geochemical clusters at district scale. The method ranks the East Caithness anomaly highest by bedrock purity (98.2%) and Ba signature.
- Transparency and reproducibility: Full code + per-cell/per-cluster CSVs archived on GitHub; whole-UK robustness check (Appendix C) shows the East Caithness cluster survives under stricter national threshold.
- Clear limitations stated: Formation identity ≠ facies identity; method is triage/ranking for future field sampling, not confirmation; glacial till question left unresolved.
2.4 Limitations of Daw (2026)
Still operates at formation/age scale (from 625k polygons), not confirmed facies or quarry match. The ~10 km offset from the exact zircon sample point is appropriately framed as regional convergence. Requires targeted petrographic follow-up (BGS thin sections already exist near the cluster margins: S13937, S13938, S27114 etc.).
3. Overall Assessment and Recommendations
The Sarclet / East Caithness claim is improved but not proven. Clarke et al. (2026) provides the first quantitative intra-basin zircon ranking but contains an unresolved p-value discrepancy and an unaddressed Lower vs Middle ORS stratigraphic mismatch. Daw (2026) supplies independent, convergent, district-scale geochemical evidence on the same coastal stretch, with superior handling of stratigraphy, full reproducibility, and explicit caveats. It correctly downplays the other Clarke hotspots by showing they lack equivalent support.
The convergence of two independent methods (zircon KS + stream-sediment Ba/Rb + bedrock verification) on the same few kilometres of East Caithness coast is the strongest current evidence narrowing the search within the Orcadian Basin.
Remaining requirements for a robust source identification:
- Facies-level petrographic and mineralogical comparison (baryte cement, tosudite/kaolinite, K-feldspar absence, ripple-laminated texture) at the geochemical hotspot and nearby BGS thin-section localities.
- New detrital zircon (and apatite) sampling directly from the 42.5 km² East Caithness cluster and the Middle ORS flagstone facies along the same coast.
- Heavy-mineral and clay XRD work to test the diagnostic Altar Stone signature.
Data Availability and Reproducibility
This audit was performed entirely on publicly available data. No restricted or paywalled datasets were used. All primary sources are cited below with direct access links. The computational steps are described at a level sufficient for independent reconstruction.
Primary Source Publications
- Clarke et al. (2026): From Highlands to Henge: Refining the Provenance and Transport Pathways of Stonehenge’s Altar Stone. Journal of Quaternary Science. https://doi.org/10.1002/jqs.70080
- Daw (2026): The Stonehenge Altar Stone: Screening the Orcadian Basin. https://doi.org/10.13140/RG.2.2.10365.12008 (CC BY 4.0). Full code and per-cell outputs: https://github.com/TimDaw37/Altar-Stone-Source-Screening
- Clarke et al. (2024): A Scottish provenance for the Altar Stone of Stonehenge. Nature. https://doi.org/10.1038/s41586-024-07652-1 (supplementary data contain the Altar Stone zircon U–Pb analyses).
- Strachan et al. (2021): Evidence from the U-Pb-Hf signatures of detrital zircons for a Baltican provenance for basal Old Red Sandstone successions, northern Scottish Caledonides. Journal of the Geological Society. https://doi.org/10.1144/jgs2020-241 (supplementary tables contain the Orcadian Basin zircon U–Pb analyses).
- Bevins et al. (2023): The Stonehenge Altar Stone was probably not sourced from the Old Red Sandstone of the Anglo-Welsh Basin. Journal of Archaeological Science: Reports. https://doi.org/10.1016/j.jasrep.2023.104215
Geological and Geochemical Datasets (Public)
- BGS G-BASE stream-sediment barium and rubidium grids (500 m resolution): British Geological Survey. barium grid and rubidium grid
- BGS Geology 625k bedrock, fault and superficial deposit GIS layers: British Geological Survey. https://www.bgs.ac.uk/download/bgs-geology-625k-gis-line-and-polygon-data-shapefile-format/
Replication Methods (Kolmogorov–Smirnov and Bedrock Verification)
The two-sample Kolmogorov–Smirnov tests were implemented in Python 3 using scipy.stats.ks_2samp (two-sided). Altar Stone grains were filtered exactly as described in Clarke et al. (2026): ≤10% discordance on the Concordia Age column, yielding n = 56 grains across the three analytical blocks (MS3, 2010K.240, FN593). For the Strachan et al. (2021) localities, grains were retained where the source table classified them as Group S or Y (concordant detrital) and the Preferred Age column was used; no additional percentage-discordance filter was imposed beyond the authors’ own classification. Point-estimate KS D-statistics and p-values were calculated directly on the filtered age lists. Sensitivity to discordance threshold was checked at 5%, 10%, 15% and 20%. A full Monte Carlo error-resampling implementation matching the precise parameters used by Clarke et al. (2026) could not be replicated because those parameters are not fully specified in the published text; the point-estimate results reported here are therefore conservative and directly reproducible from the cited public supplementary tables.
Bedrock verification used standard point-in-polygon spatial join against the BGS Geology 625k layer in British National Grid (OSGB36 / EPSG:27700) projection. For each locality coordinate the following fields were retrieved: LEX_D (formation name), RCS_D (lithology description), MAX_PERIOD and MIN_PERIOD (chronostratigraphic age). A locality was classified as genuine Old Red Sandstone only if the age field included Devonian and the lithology description did not contain igneous, metamorphic, lava, tuff, schist, ultramafite, pyroclastic, metabreccia, felsic-rock or gneiss keywords. The same filter was applied pixel-by-pixel to every 500 m grid cell inside the geochemical clusters reported by Daw (2026).
Geochemical Screening Code
The basin-wide Ba/Rb composite screen, P95 threshold derivation, 8-connected clustering, and per-cluster bedrock verification were performed exactly as described in Daw (2026) Sections 2.2–2.5 and Appendix A. The complete Python implementation, per-cell and per-cluster CSV outputs, and the independent replication of the core grid-screening component are archived at the GitHub repository linked above (CC BY 4.0).
Recommended citation: Grok (xAI). (2026). Independent Audit of the Sarclet Provenance Claim for the Stonehenge Altar Stone: Review of Clarke et al. (2026) and Daw (2026). 5 July 2026.
No comments:
Post a Comment
Comments welcome on fresh posts - you just need a Google account to do so.