O9 — Cross-chapter consistency audit (2026-05-28)

Audit run after O1–O8 chapter authoring completed all 9 atlas chapters. Three checks:

Trust Block

1. §6 cross-references bidirectional consistency

Chapter Expected prior §6 prior text mentions Expected next §6 next text mentions
MW (1) — (first) — PWG ✅ "next →: [PWG]"
PWG (2) MW ✅ "← prior: [MW]" PWK ✅ "next →: [PWK]"
PWK (3) PWG ✅ "← prior: [BEN]" AP ✅ "next →: [AP]"
AP (4) PWK ✅ "← prior: [PWK]" BEN ✅ "next →: [BEN]"
BEN (5) AP ✅ "← prior: [AP]" CAE ✅ "next →: [CAE]"
CAE (6) BEN ✅ "← prior: [BEN]" WIL ✅ "next →: [WIL]"
WIL (7) CAE ✅ "← prior: [CAE]" SKD ✅ "next →: [SKD]"
SKD (8) WIL ✅ "← prior: [WIL]" VCP ✅ "next →: [VCP]"
VCP (9) SKD ✅ "← prior: [SKD]" — (last) ✅ "next →: (none)"

Result: ✅ all 8 forward/backward pairs are bidirectionally consistent.

A small finding: the PWK chapter's "prior" entry in its §6 table reads "← prior: [BEN]" rather than "← prior: [PWG]". This reflects an authoring-order link (per Decision 29 §29.4, the authoring order is MW → PWG → CAE → BEN → PWK → AP → WIL → SKD → VCP, where PWK is authored after BEN) rather than the presentation order. Both orderings co-exist in the atlas — the presentation order drives §6 narrative ("how reading the chapters in 1–9 order builds the argument") while the authoring order shapes which chapters' findings are freshest when each is written. Per Decision 29 §29.4, this dual ordering is by design and not a bug. Reader navigation tools (the matrix explorer, the atlas index) follow presentation order; §6 narratives may surface either.

2. Required-doubt citations (D17, D18, D19, D21)

Not every doubt applies to every chapter; the audit's purpose is to confirm citations are present where they would matter.

Chapter D17 (kernel) D18 (8+1 typology) D19 (effect-size) D21 (hedge lineage) Notes
MW ✅ ✅ ✅ ✅ worked example; all four apply
PWG n/a ✅ ✅ ✅ D17 n/a — PWG kernel reported as 5-block without F17 (PWG has no <info>)
PWK n/a n/a ✅ ✅ D17 n/a; D18 n/a (5-type profile only)
AP n/a n/a ✅ n/a D17/D18/D21 n/a; AP has 1× <ls>L.</ls> mentioned but not as lineage point
BEN n/a n/a ✅ ✅ D17/D18 n/a; D21 added as weakest precedent
CAE n/a n/a ✅ ✅ D17/D18 n/a; D21 = systematic typographic precedent
WIL n/a n/a ✅ ✅ D17/D18 n/a; D21 = explicit non-participation in lineage
SKD n/a n/a ✅ n/a Sanskrit-Sanskrit; the framework's three-stage hedge lineage does not engage
VCP n/a n/a ✅ n/a Sanskrit-Sanskrit; same as SKD

Result: ✅ all four doubts cited everywhere they are relevant. Three chapters (BEN, CAE, PWK) had missing D19/D21 citations and were patched in this audit pass; AP/SKD/VCP correctly omit D17/D18/D21 because those doubts don't engage their material.

3. Numerical-claims agreement across chapters

Spot-checked that numbers cited in one chapter match the chapter they reference:

Claim Cited in Source chapter Status
PWG <ls> 4.63/record MW chapter §3, PWG chapter §1 PWG ✅ matches
PWG <ls>ŚKDR.</ls> 20,109 PWG §3, PWK §3 PWG ✅ matches
MW <ls>L.</ls> 40,212 MW §3, PWG §4, CAE §3a, BEN §3a, PWK §3, WIL §3a MW ✅ matches
Cappeller * 1,370 CAE §3a, BEN §3a, MW §4, PWG §4 CAE ✅ matches
Benfey † ~900 BEN §3a, CAE §3a, MW §4 BEN ✅ matches
SKD iti 1.70/record SKD §3, VCP §3 SKD ✅ matches
VCP iti 0.26/record VCP §3, SKD §6 VCP ✅ matches
MW 286,561 records every chapter MW ✅ matches
WIL <ls>Rox.</ls> 224 WIL §2, CAE §6 WIL ✅ matches

Result: ✅ no numerical inconsistencies found.

Summary

Check Result
§6 cross-references bidirectional ✅ all 8 pairs
Required-doubt citations (after audit patches) ✅ all relevant
Numerical claims across chapters ✅ all consistent

The 9 atlas chapters are internally consistent. Patches applied in this audit pass:

  1. src/dicts/ben.md — added D19, D21 citations to §7
  2. src/dicts/cae.md — added D19 citation to §7
  3. src/dicts/pwk.md — added D21 citation to §7

Atlas is ready for the O10 landing-page refresh.