# Documentation review — 24 October 2026

Scope: internal accuracy, consistency and clarity of the supplied pass-4 package. No new literature search, raw-source audit, statistical model or independent human review was performed.

## Changes

- Replaced three competing start-here instructions with one current-file index.
- Distinguished 280 measurements, 237 raw labels, 94 longevity-quality candidate labels and the 117-measurement/91-label screening queue.
- Explained the inherited historical count of 95 in pass-2 JSON and the precedence of audited crosswalk fields.
- Marked pass-2 and pass-3 progress notes as historical. Their pending items may have been addressed in later passes.
- Reconciled the older peak-selection rule with pass 3: measured thresholds under recorded conditions are the working target. Selection across conditions remains unresolved; geometric aggregation is conditional on comparability.
- Qualified raw AnAge, tree and software-environment claims as earlier findings rather than newly performed checks.
- Added complete reproduction commands and distinguished self-contained pass 4 from the externally dependent full pipeline.
- Made the platypus preservation statement explicitly historical.

## Validation

Recounted the current measurement table, queue and crosswalk; confirmed all 280 eligibility fields remain not_finalized. Regenerated the four pass-4 outputs in a temporary directory and compared them byte-for-byte successfully. All pre-existing CSV, JSON and Python files, plus tree checks, remain byte-identical to the supplied ZIP. Only documentation and checksums changed; this review note was added. Refreshed package checksums and checked ZIP integrity.

## Still unresolved

Primary-source exact values and conditions; consistent statistic/condition selection; organism and tree matching; final eligibility; human review; registration of remaining analyses; final modelling and public-release updates. No candidate count is presented as a finished analysis sample.