Step 3 result
The demonstrated problem is the connection between figure labels and physical surfaces. Its propagation into the published counting records or dating inputs has not been established. This audit identifies which comparisons depend on that missing connection and corrects one unqualified link in our own comparison.
The review now covers the names TH2, TH3 and TH4: the central patch changes from TH3 to TH2 between the matched photographs, while the neighboring patch changes from TH4 to TH3. The TH5 strip retains its label. That agreement is a useful control within these images, not validation of every TH5 record. [WS-PHOTO-001]
Record: WS-SCOPE-002. Review date: October 28, 2026. The original WS-SCOPE-001 still records the separate problem of differing count assemblages. The photographic match and stratigraphic audit remain unchanged; WS-SURFACE-001 is unresolved.
Follow-up disposition: the focused field/sample search remains inconclusive. Use “LF-5 — horizon assignment unresolved.”1 WS-SURFACE-001 is parked pending new primary evidence; local track relationships are the next research priority.
What remains usable, and what is blocked
| Evidence or proposed use | What remains usable | What the label issue prevents |
|---|---|---|
| Matched photograph and local geometry | The same groove, footprint arrangement and excavation details can be examined under their source identifiers | Treating the two versions as observations of different ancient activity periods |
| 2025 LF-5 description | The detailed source assigns it to TH3; local alignment and deformation remain documented | Giving it a 2026 TH3 age solely because both records say “TH3” |
| 2021 BORDaR inventory | Thirteen keyed prints, original measurements, trackway assignments and model sharing remain source records | Renaming prints or joining them to dated 2026 surfaces without a print-level crosswalk |
| 2026 Table S1 | Counts, areas, taxa and arithmetic density remain attributable to the published rows | Assigning those counting windows to the disputed patch without matching polygons and track IDs |
| Sediment and preservation descriptions | Dolomite casts, excavatable infill and overprinting can be studied within documented source contexts | Treating their contrast as a measured preservation correction between two resolved dated surfaces |
| Laboratory assays and dated sample groups | Lab IDs, measured values, calibration/correction records and published sample ordering remain intact | Assuming that a particular photographed groove has a secure relationship to those samples |
| Published model-4 horizon outputs | The named outputs remain available as source-qualified estimates | Reading a named output as a direct date for LF-5 or another unmatched print |
| Climate and movement comparisons | Regional environmental context and local path questions can still be investigated | A dated change in route use, traffic, human–animal overlap or population across the disputed surfaces |
Even a definitive label correction would not make the movement comparison ready: registered trackway geometry, comparable exposure/preservation masks and recording duration remain separate gaps. The differing 2021, 2025 and 2026 count assemblages also still require a print-level crosswalk.
These limits do not make a groove disappear or invalidate every White Sands result. They also do not prove that only figure lettering is affected. The broad human-presence and Last Glacial Maximum claims are not overturned by this identified discrepancy; they were not independently revalidated in this step. The travois explanation retains its existing provisional status. Other dating and arithmetic review issues remain separate. [WS-C01–WS-C03]
Three possible scopes of correction
| If subsequent evidence establishes… | Appropriate action | What is currently missing |
|---|---|---|
| An annotation error confined to figures | Correct the documented figure-to-surface crosswalk; retain independently verified counts and model inputs | A source-supported correction identifying the intended layers |
| Incorrect membership of prints or exposure polygons in a counting row | Reconstruct affected inventories and denominators from track IDs and mapped boundaries | Matching excavation polygons and a print-by-print inventory |
| An incorrect physical relationship between a surface and dated samples | Reassess the relevant stratigraphic constraints and any dependent age modeling | Registered field contacts, sample positions and surface identifiers |
These are conditional cases, not three findings. Swapping whole TH2/TH3 rows is unjustified: the disputed photograph may cover only part of a counting window. A uniform numerical shift is also incompatible with the unchanged TH5 annotation.
Numerical consequences without assigning a new age
The following are unchanged 2026 source rows, not new identifications of the photographed patches. Density is calculated from Table S1; ages are published model-4 outputs from Table S9d, in cal ka BP ±2σ. [S5 · Springer et al. (October 14, 2026)]
| Published row | Prints / area (m²) | Calculated prints/m² | Modeled age ±2σ |
|---|---|---|---|
| TH2 | 45 / 3.82 | 11.78 | 22.66 ± 0.10 |
| TH3 | 19 / 1.35 | 14.07 | 22.61 ± 0.09 |
| TH4 | 37 / 6.70 | 5.52 | 22.57 ± 0.08 |
The TH3/TH2 density ratio remains 1.195, an arithmetic property of those rows. It supplies neither an adjusted traffic rate nor a population trend.
If the only question were which of the existing TH2 or TH3 outputs to associate with the central patch, their nominal ages differ by 50 years. The TH3/TH4 choice for the neighboring patch differs by 40 years. These are conditional differences between published estimates, not elapsed times between walking events, estimates of the labeling error, or bounds on all possible dating errors. The outputs share model inputs; their uncertainties cannot be treated as independent. Overlap alone does not erase stratigraphic order, and a difference of means does not establish its duration.
A more discriminating field question
The source code identifies a useful reference level: the seed determinations USGS-1808.1, USGS-1808, USGS-1830 and USGS-1810, represented as repository group G05. The deposition model puts this combined level just above TH2 and below TH3. The Sequence version instead places TH2 and those assays in the same phase, with TH3 above it. This distinction is preserved; the two formulations are not treated as identical field observations. [S5, supplement model-4 code, PDF pp. 6 and 12]
The useful question is therefore: where is the original LF-5 formation surface relative to that sampled level, traced through the actual sediment contacts? The depth of deformation below the groove does not answer it. Four determinations assigned to a shared level are not four independent surface correlations.
A resolving record should identify both photographed patches, connect them to the field layer/sample scheme, and state whether the clarification changes only annotations, also counting membership, or actual dating constraints. No such clarification has been obtained and no author has been contacted.
Repository changes and research that can continue
- Added review links for TH2–TH4 in the horizon context, counting display and dating ledger. The links concern surface identity across publications; they do not mark the numerical source values as corrected.
- Replaced our unqualified statement connecting Fig. S12’s groove direction to the TH2 row with an explicit unresolved-assignment note. The prior wording and replacement are recorded in
label-scope-audit.json. - Preserved all original feature IDs, track measurements, count/area values, sediment/proxy observations, assays and modeled ages. Canonical 2026 assignments for the source features and archive prints remain null.
- Retained two productive local questions: examine the two named trackways in their shared model T038S, and assess groove/footprint relationships within a documented patch. Either can proceed in source coordinates without claiming a calendar age, simultaneous travel or a horizon-wide behavioral trend. The archive contains selected close-up models, not a complete site map. [S320 · Bennett, M. (2021)]
Evidence trail
The machine-readable audit links the existing source records, the matched figure annotations, the stratigraphic review, unchanged numerical rows and the corrected repository wording. Relevant primary sources are S5, Springer et al. 2026, S6, Bennett et al. 2025, and S320, the BORDaR archive. This step is an assessment of dependencies and consequences, not new excavation, a rerun of OxCal or an author-confirmed correction.
Footnotes
-
The 2025 detailed account assigns LF-5 to TH3; the 2026 figure labels the same photographed patch TH2. No verified field connection to the dated seed level or correction defining annotation/inventory/sample effects was located. The canonical horizon and feature age remain unassigned. The final check records inspected sources and inaccessible release files. ↩