# Scientific review sequence and corrections

This is the public account of the completed Cohler–Soon audit, DOI 10.53234/scc202612/05. Sanitized copies of the original scientific reviews accompany it. Removed material concerns local execution paths and private coordination; scientific conclusions, qualifications and exposure statements remain. [Provenance](PROVENANCE.json) identifies both original-record and public-copy hashes. Historical statements such as “pending reconciliation” inside first reports describe their original review stage, not the final status.

## Separate first reviews

The [mathematical reviewer](reviews/MATHEMATICAL-REVIEW.txt) read the full target, derived exact isotope balances and tested the strongest net-provenance interpretation. Its first report was frozen before reading the manager’s empirical results or the other reviewer’s findings. It inherited the commission and intake context. The [carbon-source reviewer](reviews/CARBON-SOURCE-REVIEW.txt) independently read the full target and supporting sources before seeing the mathematical review or manager’s fits. It exchanged acquisition/status information with the editor. These are distinct same-provider AI assessments, not external human or different-provider review.

The [mathematical addendum](reviews/MATHEMATICAL-ADDENDUM-WH.txt) followed the editor’s notification of the W-H printed-equation omission and modern intercept agreement. It is a targeted nonblind check, not an independent initial discovery. The source factor issue and reconstruction choice remain disclosed.

## Separate challenge

The [challenger](reviews/INDEPENDENT-CHALLENGE.txt) inherited context and read the reconciled report, earlier reviews, primary passages and actual scripts/results. It independently checked exact rational isotope balances, constructed a finite-duration version and a positive-natural-growth variant, directly recovered four monthly station fits, and recalculated the ramp and two-box response. It did not rerun every empirical sensitivity or produce a calibrated atmospheric model.

Its three required corrections were applied and [checked in a subsequent read](reviews/CHALLENGE-RESOLUTION-CHECK.txt):

1. **Historical curve scaling.** The report and article now say that the atmospheric isotope curve was scaled to the shallow-water sponge record before digitization. Exact handling/reversal remains unresolved. This does not itself invalidate the regression or the sponge measurements.
2. **GCB terminology.** Historical annual values are called estimates rather than direct observations. The assumed 2025 extension remains explicit.
3. **Package availability.** The author package was not located in the checked public locations; the audit does not claim it does not exist.

The original challenge report is preserved alongside the correction check. No substantive dissent remained about the narrowed finding that the claimed general mathematical falsification is not established. The corrections did not require another scientific execution.

## What agreement does and does not mean

The reviewers retain the modern intercept, conditional mixing arithmetic, several CO₂ model comparisons and conditional stock calculations as successes. Signed residual identification, unsupported constancy and turnover-to-causal-response inference remain the principal objections. The −0.45 isotope residual discrepancy and missing historical digitization remain unresolved. Mathematical counterexamples do not replace those empirical replications or measure actual fossil attribution.

The original reviews are not retroactively rewritten to sound blind or unanimous. Their recorded prior exposure, limits and subsequent corrections remain visible. All reviewers used AI from the same provider, under shared commissioning and organizational control. The publication-stage editorial and sanitization check is an additional targeted check of the staged copies, not a new scientific audit or external accreditation.