T. Jake R. Ciborowski, David J. Nash,
Defining similarity: An arithmetic method for archaeological source provenance targeting using geochemical data,
https://doi.org/10.1016/j.jasrep.2025.105513.
(https://www.sciencedirect.com/science/article/pii/S2352409X25005462)
The provenance of Stonehenge’s sarsen megaliths continues to
stimulate scholarly debate, particularly as increasingly precise geochemical
datasets expose the methodological challenges of lithic sourcing. In a
significant contribution published online on 2 December 2025 in Journal of
Archaeological Science: Reports (69:105513), T. Jake R. Ciborowski and
David J. Nash introduce a new arithmetic framework for quantifying geochemical
similarity between archaeological artefacts and potential source outcrops.
Arising from the contested interpretations of the Phillips’ Core dataset (Nash
et al., 2020; Hancock et al., 2024; Nash & Ciborowski, 2025), their
open-access study reaffirms the West Woods provenance for Stonehenge’s
principal sarsens and proposes a generalisable method for lithic provenancing
across diverse geological contexts.
Methodological Innovation: From Ratios to Ranked
Similarity
Ciborowski and Nash’s central innovation is to formalise a
transparent, petrologically grounded arithmetic approach that overcomes
limitations in both visual inspection and conventional multivariate statistics.
These traditional methods can obscure key geological processes—especially the
variable silicification that characterises silcrete formation—and may introduce
subjectivity when applied to complex datasets.
Building on the immobile trace-element/Zr ratio approach
used in Nash et al. (2020), the authors propose a simple but powerful measure
of pairwise geochemical similarity. Equation 1 calculates the percentage
difference (ΔEi/Zr %) between any trace element/Zr ratio in an artefact and a
prospective source:
By taking the geometric mean of Δ values across many
elements—21 immobile trace elements in the case of the silcrete dataset—the
method yields a single, scale-independent similarity score that can be used to
rank potential source outcrops objectively.
This formulation avoids the known pitfalls of relying on raw
element concentrations, which may vary widely due to silicification,
hydrodynamic sorting in the host sediments, or weathering. The authors
explicitly contrast their approach with that of Hancock et al. (2024), who used
concentration data and unusually wide tolerances (–50% to +100%), a strategy
the present authors argue is incompatible with silcrete petrogenesis.
Applied to the Stonehenge dataset—comprising ICP-MS analyses
from the Phillips’ Core and samples from 20 southern British sarsen
outcrops—the method ranks West Woods (Outcrop 6) unequivocally as the
most similar source, with a geometric mean Δ value near 29%. Outcrops proposed
by Hancock et al. (2024), including Clatford Bottom (Outcrop 3) and Piggledene
(Outcrop 4), rank only 7th and 8th respectively.
Strikingly, comparisons among the three subsamples of the
Phillips’ Core itself yield similarity scores of 12–20%. In several cases, West
Woods samples are more similar to individual core subsamples than those
subsamples are to each other—a result that strongly reinforces the West
Woods connection and highlights the natural variability within a single
silcrete block.
The authors demonstrate the method’s generality through
multiple “worked examples” involving igneous lithologies—obsidian, basalt,
andesite, and dolerite—and show that the arithmetic framework performs well
across both high-precision ICP-MS datasets and lower-precision, non-destructive
pXRF data.
Implications for Stonehenge Provenance
Within Stonehenge research, this study consolidates the case
for West Woods as the principal source of the sarsen megaliths,
including the trilithon uprights. Rather than relying on binary
“match/no-match” interpretations, the arithmetic framework quantifies
similarity as a continuous measure. This is particularly valuable for silcrete,
where substantial intra-outcrop and intra-stone variability is expected.
While the present paper does not analyse other sarsen stones
directly, the authors note that this method is especially well suited for
evaluating sarsen outliers identified in earlier surveys—such as Stone 26 or
lintel Stone 160—where geochemical affinities differ from the main cluster.
They also demonstrate how similarity scores can be mapped spatially (“source
vectoring”) to identify promising areas for further field sampling (Fig. 13).
Taken together, these results support an interpretation of
deliberate, targeted extraction rather than glacial agency, consistent with
broader archaeological evidence for complex quarrying and transport networks in
the Late Neolithic.
A Note on Bluestone Dolerites: Scope and Clarification
A particularly informative worked example in the paper
applies the arithmetic method to Preseli dolerites, using the dataset
published by Pearce et al. (2022), which includes pXRF measurements from Stone
62, a core extracted from it, and seven potential source outcrops.
This case study demonstrates both the utility and the nuance of the ΔEi/Zr %
approach for igneous rocks. As expected, Stone 62 is most similar to its own
core, validating the method’s internal consistency. When compared against
regional outcrops, Carn Goedog emerges as the closest match (geometric
mean Δ ≈ 20–25%), followed by Carn Ddafad-las and Garn Ddu Fach
(both ≈ 25–30%) . Intriguingly, the Garn Ddu Fach sample appears slightly
more similar to Stone 62 than the Stone 62 core itself, highlighting
natural intra-monolith variation and illustrating how the arithmetic framework
can refine interpretations previously based solely on cluster analyses.
Although restricted to one monolith, this example shows how the method
complements ongoing Preseli quarry research, offering a transparent and
effective way to interrogate fine-grained geochemical differences within a
dolerite suite.
Broader Scholarly Significance
Beyond Stonehenge, the authors argue persuasively that their
arithmetic approach fills a methodological gap between subjective visual
comparisons and statistically opaque clustering or discriminant analyses. By
emphasising petrological reasoning—immobile elements for silcretes,
incompatible elements for igneous suite discrimination, compatible elements for
intra-suite differentiation—the method offers a clear and reproducible
framework for geochemical provenance work.
Limitations are candidly acknowledged:
- No
universal exclusion threshold yet exists for ΔEi/Zr values.
- Element
choice must be petrologically justified for each lithology.
- Arithmetic
similarity measures should complement, not replace, petrographic and
archaeological evidence.
Despite these caveats, the paper represents a measured and
substantial methodological advance, providing a transparent and adaptable tool
for archaeologists working with diverse lithic materials.
Conclusion
Ciborowski and Nash (2026) offers a rigorous,
process-aware approach to geochemical provenancing and provides the clearest
quantitative support yet for a West Woods origin of Stonehenge’s principal
sarsens. The authors’ arithmetic framework—simple in formulation but powerful
in application—bridges geochemical precision and archaeological interpretation.
Its demonstrated utility across silcrete, basalt, and obsidian artefacts
positions it as a promising standard for future provenance studies, both within
and beyond Stonehenge research.
