# Meridian source-comparison review / version 1

Fictional qualitative review of a bounded source set. No statistical inference, source authentication or independent scientific review is performed.

1. Preserve each report edition, account and scalar file by its exact digest.
2. Ask whether the selected sources establish a common physical source. Keep timing, geometry, emitter identity and observer recollection separate.
3. Read each complete review-note section as an attributed rationale. Keep the initial claim, the same team's revision and the other reviewer's alternative distinct. Only the revised shared-source claim supersedes the initial one; the independent explanation is not a revision or the winner.
4. Compare the exact input sets. Report A is replaced by report B as an input to the revised review; the original remains available. Observer A's follow-up and recording specifications are added. The first accounts, scalar files and their generator remain byte-identical.
5. Retain an inconclusive outcome. Neither a hypothesis nor a resolved reference establishes physical association. Do not create a probability or overall scientific grade.

The short labels, input contribution notes, limits and follow-up prompts in catalog.json are authored reading aids. They do not add weighting or a normative supporting/challenging relation to the claim contract. Full review rationales are taken verbatim from reader-r2/sources/reviews.md, with exact line ranges in the application projection. Canonical provenance retains the source digest; no unsupported line locator is invented in the standard. Source-edition inputs use exact edition/page citations.

These dated reviews describe the sourcebook at its recorded stage. Later instrument captures, image arrays and material studies elsewhere in the showcase are not silently added as inputs. The earlier short-example history remains unchanged; this expanded review history is separately identified and does not supersede across snapshots.
