epiphany/spec/PASS13_CANDIDATES.md

Pass 13 — candidate ledger

Ambiguities and cross-cut inconsistencies found since Pass 12 closed, filed per the house rule (a batch pass opens at ≥3 candidates; this file opened when P13-D1/D2 joined P13-K1). Each entry names the owning DECISIONS record; this file is the index, not the analysis.

Batch 1 — CLOSED (2026-07-08)

All four candidates are resolved (worked down in order): P13-D3 and P13-K1 by the user's ratified calls ("fix the mint only" / "reject the introduction"), P13-D1 and P13-D2 as correctness fixes with convergence-locked / execute-then-fix regressions.

Id	One-line statement	Filed in	Status
P13-K1	The K3 verdict for a system pitch introduced by a ModifyEvent replacement value differs across a snapshot cut (in-session TargetMissing vs post-snapshot SystemDerivedContentImmutable); really "may ModifyEvent introduce never-minted pitch ids"	crates/epiphany-ops/DECISIONS.md (Pass-12 G-pass code tranche)	resolved (Pass 13: reject the introduction — modify_event refuses a never-minted system-derived pitch, verdict now snapshot-cut-invariant; user "reject the introduction")
P13-D1	Undo-driven event tombstones run graph-side re-anchor/cascade but never ledger-side reanchor_for_tombstone: structures leave the graph while staying Live, no RepairRecord — Ch6's same-step recording MUST is unmet for undo-driven tombstones (pre-existing class: slurs/spanners; repeats now too)	crates/epiphany-ops/DECISIONS.md (Schema major 2, Phase D)	resolved (Pass 13: tombstone_undo_targets runs the ledger re-anchor per event target; liveness guard; convergence-locked)
P13-D2	Cue-cascade recursion re-anchors against the triggering event before its tombstone lands in objects: a structure anchored on {X, cue-of-X} can record Reanchored{to: X} then CascadeDeleted in one effect (contradictory repair trail; plausible by code trace, unexecuted)	crates/epiphany-ops/DECISIONS.md (Schema major 2, Phase D)	resolved (Pass 13: delete_event tombstones before the graph delete, matching cascade_cue/undo; repro executed then fixed)
P13-D3	CreateCrossCutting validates only event endpoints (CrossCuttingValue::endpoints()), so a SPANNER anchored to a missing region/measure mints dangling past anchor_target_exists; and non-event referent tombstones (DeleteRegion under a region-anchored spanner/repeat) re-anchor nothing — "every referenced endpoint is live" is events-only as implemented	crates/epiphany-ops/DECISIONS.md (Phase D follow-up)	resolved (Pass 13: mint fixed via anchor_object_refs; non-event referent re-anchoring ratified events-only, user "fix the mint only")

Batch 2 — CLOSED (2026-07-09)

Three candidates accumulated while the Standard-tier solver track closed and the notation-quality pass (stems, slurs) landed. All three were parked as they were found, each in crates/epiphany-layout-ir/DECISIONS.md, and reaching three reopened the pass per the house rule. None was a live incorrectness in shipped output; each was a place where the code, the spec, and the data disagreed about what is true.

All three resolved, worked down in order. Two grew when examined: P13-I1 was filed as two elided fields and was three, the third (diagnostics) named nowhere in core_spec; P13-I2's fix had to reach editor-core as well as the projection, or a click on a bass staff would have resolved its pitch as treble. Zero golden churn across all three. No open Pass-13 candidates remain; a future ≥3-candidate batch reopens the pass.

Id	One-line statement	Filed in	Status
P13-I1	Chapter 7's ConstrainedLayoutIR listing elides three fields the code carries: break_origins: Vec<BreakOrigin> (named by req:layoutir:break-origin-attribution, its own shape unlisted), catalog: GlyphCatalogIdentity (its type specified, the field unlisted), and diagnostics: Vec<LayoutDiagnostic> — which appears nowhere in core_spec, though it is how the projection's honesty rule manifests: an unspellable pitch or an unbundled glyph is placed as a fallback and recorded, never silently guessed	crates/epiphany-layout-ir/DECISIONS.md ("the ConstrainedLayoutIR listing is still abridged")	resolved (Pass 13: listing gains all three fields; BreakOrigin and LayoutDiagnostic shapes added; new req:layoutir:coverage-diagnostics ratifies as-implemented that an unengravable object is recorded AND still placed — never guessed, never dropped)
P13-I2	Staff::default_clef is never consulted: to_constrained takes the active clef from the staff instance's clef_sequence and falls back to Clef::default() (treble), so a bass-clef staff that declares its clef only on the Staff engraves as treble. The field is decorative in the projection — is it the fallback, or should it not exist?	crates/epiphany-layout-ir/DECISIONS.md ("Staff::default_clef is never consulted")	resolved (Pass 13: it IS the fallback. StaffContent carries it, active_clef_or resolves against it, and editor-core's hit-test reads the same function — else a click on a bass staff would resolve its pitch as treble. Removal was rejected: the field is named for its purpose, is encoded on the wire, and dropping it is schema-major)
P13-I3	BRAVURA_METRICS' NOTEHEAD_ANCHORS are hand-written, unconsumed, and doubly suspect: they name stemUpNW/stemDownSE — the corners a normal notehead's stems do not attach to, and a pair Bravura's noteheadBlack does not define — and their x of 1180 reads like 1.18 staff spaces written in thousandths rather than the table's 1/1024 units (1.18 sp = 1208). They enter only metrics_hash, so any correction moves the GlyphCatalogIdentity every conformance claim declares. The font is not vendored, so the values cannot be verified in-tree	crates/epiphany-layout-ir/DECISIONS.md ("the notehead stem anchors are unusable as written")	resolved (Pass 13: deleted. Unverifiable in-tree, unconsumed, and hand-derived where every neighbouring number is machine-extracted. extract_bravura_outlines.py --anchors now emits them from the pinned bravura_metadata.json, so the table regains them generated rather than remembered; the metadata's SHA-256 is deliberately unpinned and the script refuses until an operator with the font pins it. GlyphCatalogIdentity moves once, now, while no conformance claim declares the old one; user "delete them + teach the extractor")

Batch 3 — OPEN (2026-07-09)

Three candidates, so the pass reopens per the house rule. P13-S3 arrived as a live incorrectness — a Push-4a follow-up audit found that the spelling-set chain introduced to make TransposeInterval undoable had only one writer, so undo erased an ordinary RespellPitch on either side of it. It is resolved. S1, S2, and S3 are resolved. S5 and S6 were filed 2026-07-22 by verifying the two Push-4a audit claims that epiphany-core/DECISIONS.md had carried as unverified through two passes; both were real, and both were Chapter 4 defects standing in front of Push 4b. Both are now resolved, closing the spec half of Push 4b.

Filed 2026-07-22, second round. S7 is a bookkeeping repair: the id was minted in spec/PLAN_PUSH4B_TUNING.md ("new; file it"), ratified as that plan's Ruling C, and implemented — but never entered this ledger, so an id existed outside the index that is supposed to be the index. S8 and S9 are new. S9 is the third instance of one pattern in a single day and, notably, S4 is itself an instance of it, which is what moved it from incident to candidate.

Filed 2026-07-23, third round. S10–S12 are three underdetermined Chapter 4 types standing in front of Push 4b tranche 3 (the schema-major-3 wire freeze) — the same shape as S5/S7 before tranche 1. Scoping the accidental subtree that accidental_extensions drags onto the wire surfaced a raw f64 that the byte layer cannot encode, a leaf type referenced but never defined, and a version type whose obvious representation orders SMuFL's real release history backwards. All three are ratified; each freezes forever once tranche 3b lands, so they were resolved before dispatch, not after.

S4, S8, and S9 remain open; S10–S12 are resolved on ratification; S13 is resolved by the genesis tranche, not by this pass — it is one field of an eight-field gap, and a tuning-specific fix would have been the wrong shape (spec/ANALYSIS_GENESIS_PERSISTENCE.md). The pass stays open. (S13 was filed 2026-07-24, after S10–S12's wire freeze landed: it is the gap that freeze made visible — a value with a wire form and no canonical carrier to reach it.)

Filed 2026-07-31, fourth round — and two older rows corrected against the tree rather than against memory. S25 and S26 are new. Both came out of reconnaissance run before drafting, not out of implementation, which is the change in method worth recording: three read-only sweeps over S22, S16 and S8 were dispatched to establish current facts, and each returned something the ledger had wrong. S8's row carried a false mutation claim — it said flipping is_none_or to is_some_and killed no test; three tests kill it, executed and confirmed. The original was reached by reading only invariants.rs's own tests, which assert fires(...) and survive the flip; the kill sites are in another crate. S16's row had drifted in every code citation, some by hundreds of lines, and never named the test that makes its fix expensive. Neither row's conclusion changed. S26 is a P13-S9 instance and the sharpest one yet: a doc comment that claims a specification repair which never landed, where the claiming side is grep-guarded and the stale side is not. The pattern S9 names — fixing the loud one without reading the target converts a caught defect into an uncaught one — is why S26 forbids repairing its own evidence in isolation.

Id	One-line statement	Filed in	Status
P13-S1	169 of core_spec's 207 requirement blocks carry no \label, so no conformance claim can cite them. Chapter 4 (Tuning Systems and Pitch Spaces) is 9/9 unlabeled and Chapter 11 (Determinism Contract) is 15/15, but the gap is universal, not local: Semantic Operations 24/27, The Score Graph 22/28, Pitch 10/13. The requirements are normative and are implemented; they simply cannot be named. Every req:* label the repo cites was added ad hoc by the pass that needed it	this file	resolved (all 207 core_spec requirements labelled, suite 277/277; and the pass found that labelling alone was insufficient — no document numbered its requirements, so a \label bound to the enclosing section and 61 of 207 shared a rendered number, one shared by six. All six documents now carry a real counter. Locked by requirement_labels.rs)
P13-S2	cmn-24 was declared in the built-in pitch-space table as CMN with quarter-tone accidentals while PitchSpacePosition::Cmn.alteration, transposition, and CmnChromatic were all specified in fixed semitones. The specification now denominates both CMN alteration sites and interval chromatic steps in the enclosing space's chromatic layer, with cmn-24's map fixed explicitly. Until Push 4b provides structural registry resolution, the core fails closed outside provable built-in cmn-12 instead of applying guessed 12-chromatic arithmetic	spec/PLAN_P13S2_CMN24.md; crates/epiphany-core/DECISIONS.md	resolved (schema and wire layout unchanged; PitchSpaceUnavailable maps to existing PitchSpaceMismatch 6; Operation Catalog 0.9.0; registry-backed generalization remains a Push 4b implementation blocker)
P13-S3	The engraved_spelling_chain introduced with TransposeInterval's undo had a single writer. RespellPitch mutates the same graph attachments but recorded only on respell_chain, so (a) undoing a transposed transaction after a prior respell restored the pitch and erased the respell, and (b) a respell landing canonically after the transaction was invisible to the chain, so a StrictInverse undo reported Applied and wiped the newer authoring instead of refusing as superseded. A BestEffort undo could also restore the pre-transpose pitch while leaving a spelling authored against the transposed one	crates/epiphany-ops/DECISIONS.md (P13-S3)	resolved (both operations now record on the shared key; pitch value + spelling set undo as one unit; new req:opcat:spelling-set-chain. The chain stays physically separate from respell_chain, which is RespellPitch's LWW conflict state — folding transposes in would make a concurrent respell conflict with a transpose and move the canonical bytes of every existing history)
P13-S4	No labelled requirement governs vertical-band heights. Pass 12 twice recorded a behavioural fix to the inter-staff solve — realizing an InterStaffGap band's declared preferred_height rather than a constructor default — and both times cited req:layoutir:vertical-bands, which never existed. The two real band requirements (req:layoutir:primitive-band-ownership, req:layoutir:resolved-band-ownership) govern ownership: which band a primitive belongs to, and that a resolved primitive retains it. Nothing states what a band's height means or that the solver must realize it, so the shipped behaviour is unspecified and the log invented a name for the gap	this file	open (found in P13-S1 review: an agent 'corrected' the dangling citations to the two ownership requirements, which replaced a visibly broken pointer with a silently wrong one)
P13-S5	The JI prime basis is specified at two lengths. req:pitch:ji-vector-basis states that the built-in JI pitch spaces "order primes ascending starting with 2" and that components.len() MUST equal the basis size; the built-in pitch-space table calls ji-5limit "Two-dimensional (prime axes 3, 5)", ji-7limit three-dimensional and ji-11limit four-dimensional — each exactly one short, consistently, because the table is octave-reduced and the requirement is full-register. req:tuning:builtin-tuning-catalog makes the table normative, so a 5-limit vector is required to be both length 2 and length 3. The requirement's octave-reduction clause does not reconcile them: it normalizes the first component to a canonical range, it does not remove it	this file (verified 2026-07-22 against core_spec.tex:1005-1008 and :3496-3501)	resolved (Push 4b Wave 1a, 5dcaa58: ratified full register. The three table rows gain prime 2 and become three-, four- and five-dimensional; req:pitch:ji-vector-basis is untouched, being the side that was already correct. A JiVector is an absolute position: without the prime-2 exponent, ji-5limit cannot distinguish C4 from C5)
P13-S6	No built-in tuning system's resolution is pinned to a versioned definition, and 14 of 20 have no definition at all. req:tuning:builtin-tuning-catalog makes all 20 identifiers MUST-resolve with normative semantics, but only the six tet-* entries are actually specified — by TuningResolution::EqualTemperament's structural rule. The other 14 are names: 3 meantone variants, werckmeister-iii/iv, vallotti, kirnberger-ii/iii, young-ii, pythagorean (the 3:2 ratio is named but not the fifth-chain construction or wolf placement), 3 ji-static-5limit-*, and ji-adaptive-5limit. TuningResolution::Function delegates them to a TuningFunctionId, which Chapter 10 lists as an extension point; no built-in is mapped to a function id and no function id is pinned. Against req:tuning:tuning-resolution-determinism — determinism MUST hold across platforms — two conforming implementations may each choose a different published Werckmeister III and both pass	spec/DRAFT_P13S6_TEMPERAMENTS.md; spec/CONTRACT_P13S6_TEMPERAMENTS.md; spec/CONTRACT_P13S6_PROMOTION.md	resolved (5e465a1: all 20 pinned by construction, not by cents table. The ten temperaments carry which fifths are tempered by what fraction of which comma, wolf placement where the name does not force it, exact ratios, and a closure sum that makes the section self-verifying. The static 5-limit systems are the lattice block \{3^a5^b \mid a\in[-1,2], b\in[-1,1]\} — a 4×3 rectangle filling all twelve chromatic positions with nothing discarded — at three anchors, req:tuning:ji-static-construction. ji-adaptive-5limit is "default-v1", the key-anchored static scale, with the anchor pinned to a staff and a moment (req:tuning:adaptive-default-version, req:tuning:adaptive-anchor-derivation). Sourcing alone was insufficient: a first draft produced two arithmetically impossible temperaments and one false ambiguity, every one properly cited, and only the closure invariant caught them — the contract gained a permanent "check the arithmetic, not just the source" section as a result)
P13-S7	ScalePosition referenced a score pitch-space registry that does not exist. The listing comment said the space "References an entry in the score's pitch-space registry", and req:pitch:default-pitch-space said every score MUST define at least one pitch space and MAY define more. No such registry exists anywhere: Score has thirteen fields and none is one, and ScoreTuningContext carries ids plus accidental extensions — never space or system definitions. So the specification required an act (defining a pitch space) that the data model gives a score no way to perform	spec/PLAN_PUSH4B_TUNING.md (Ruling C); spec/CONTRACT_PUSH4B_SPEC1.md	resolved (Wave 1a, 5dcaa58: ratified the catalog is closed. Both "define"s become select, from the built-in catalog; the ScalePosition comment names the built-in catalog; score-local definition is deferred to a later schema major, because the Chapter 4 type surface has never had a consumer and req:binfmt:frozen-layout would freeze it permanently on first encode. No struct gained a registry field — that was the point of the ruling. Filed here retroactively: the id was minted in the plan, ratified, and implemented without ever entering this ledger)
P13-S8	Canonical state admits two encodings of one musical fact, and nothing normalizes them. TempoShape::Constant (core_spec.tex:2230) says end_tempo MUST be None or equal to start_tempo — explicitly legalizing two spellings of an identical constant-tempo segment. TempoSegment encodes end_tempo positionally through struct_codec! (codec.rs:1512) and TempoMap is canonical state on Score (graph.rs:1808, codec.rs:1519), so the two forms produce different canonical bytes: measured at 336 vs 363 for otherwise identical scores. invariants.rs:1399 merely permits both; no normalization runs anywhere. The project has ruled the other way everywhere else it looked — req:determinism:unicode-canonicalization applies NFC precisely so that "two texts whose NFC byte representations are equal are canonically equal", and req:determinism:canonical-collection-order pins iteration to a total order — but tempo has no equivalent, so two musically identical scores hash differently	this file (measured 2026-07-22 by encoding both forms)	open. Corrected 2026-07-31; two of this row's own claims were false. (1) "Found sideways: the branch is untested" — it is not. Flipping is_none_or to is_some_and at invariants.rs:1399 was executed against the working tree at a4d7847 and three tests kill it: set_tempo_segment_materializes_in_score_and_region_scope (epiphany-ops/tests/graph_reduction.rs:3083), undo_restores_the_pre_transaction_writes_for_replaced_keys (same file), and convergence::tests::graph_sessions_converge_on_the_real_score (testkit/src/convergence.rs:224, reporting the witness verbatim). The mutation was restored by hand. The original claim was reached by reading only invariants.rs's own tests, which assert fires(...) and therefore survive the flip — the kill sites are in another crate. The genuinely unconstructed spelling is the opposite one: Constant + Some(equal) has zero construction sites in the workspace; the sole production constructor, valuegen::tempo_segment (ops/src/valuegen.rs:569), emits None. (2) The row's framing that a normalization point may not exist — tempo is op-authored: OperationKind::SetTempoSegment, wire and tag discriminant 26, reduced at reduce.rs:5811, with PreconditionFailureReason::TempoMapMalformed already gating shape well-formedness. Ratification, restated — normalize-on-encode is NOT the default and should not be read as one. Accepting two byte forms and folding one violates req:binfmt:decode-vectors (binary_format.tex:3404–:3413): "A value decoded from an accept vector MUST re-encode to exactly that vector's bytes… Canonical decode is injective: distinct byte strings denote distinct values, which is what content-addressing rests on." The ratified precedent for exactly this shape is req:binfmt:compression-none-parameter (:1909–:1913), which requires a decoder to reject rather than normalize, on the stated rationale that leniency made the codec non-injective. A future repair must therefore either reject one spelling (a stored-form invariant plus strict decode, the pattern req:format:rationaltime-encoding already uses for reduced rationals) or explicitly retain both as distinct canonical values and accept divergent hashes for identical music. A third correction: the permissive rule is triplicated, so a repair touching only the invariant leaves two behind — invariants.rs:1399, reduce.rs:965 (tempo_segment_shape_well_formed), and tempo.rs:452 (the wall-clock integrator). Note also the types live in crates/epiphany-core/src/tempo.rs (:98, :109, :119); graph.rs holds only Score::tempo_map (:1808) and the second, unmentioned reach into canonical state, Region::local_tempo_map (:794)
P13-S9	The citation checker enforces cited→defined, never cited→relevant. requirement_labels.rs proves every req:* string in the repo resolves to a real label and nothing further, so a citation that resolves cleanly but does not support the sentence it is attached to is invisible to a fully green gate. Three instances in one day: a Push-4b spec agent justified "resolved layout is non-canonical" with req:layoutir:staff-space-coordinates, which mandates staff-space units and f32 and says nothing about canonicality; a dispatch contract asserted "Chapter 10 defines KeySignature" when the Score Graph is Chapter 5 (caught only because the agent used \ref{ch:graph} and reported the discrepancy rather than hardcoding either number); and P13-S4 is itself an instance — an agent "corrected" two dangling citations to the ownership requirements, replacing a visibly broken pointer with a silently wrong one	this file	open (relevance is not mechanically checkable, so the ratification is what discipline replaces a check: e.g. requiring the quoted sentence alongside any newly-added citation so review compares text to text, or treating agent-authored citations as unverified until a human reads the cited requirement. Note the asymmetry that makes this dangerous — the broken-pointer case is loud and the wrong-pointer case is silent, so fixing the loud one without reading the target converts a caught defect into an uncaught one)
P13-S10	PitchSpaceModification::Cents(f64) puts a raw f64 in canonical state, which the byte layer cannot encode. core_spec.tex:3112 declares Cents(f64); PitchSpaceModification is reached from canonical score state through AccidentalDefinition → ScoreAccidentalExtensions → ScoreTuningContext. req:determinism:canonical-floating-point requires canonical stored floats to be finite IEEE 754 binary64, and the byte layer enforces it with no escape hatch: serialize.rs:110 decodes floats only through CanonicalF64::from_le_bytes → NonFiniteFloat, so there is no Codec for f64. A raw f64 is therefore not merely risky in canonical state — it is unencodable without inventing a new unvalidated codec. The same chapter already resolved this exact tension forty lines later: EngravingBoundingBox (:3150) carries SpaceUnit edges with a rationale (:3191) invoking the same requirement	this file (verified 2026-07-23 against core_spec.tex:3112, :3150-3197, and crates/epiphany-determinism/src/serialize.rs:110)	resolved (ratified: Cents(CanonicalF64). A one-line spec correction, the only shape consistent with the existing byte layer — the identical maneuver Ruling D applied to the bounding box. Freezes in tranche 3b)
P13-S11	AnchorPoint, referenced by AccidentalEngraving.anchor, is defined nowhere. core_spec.tex:3166 names pub anchor: AnchorPoint; no struct or enum of that name exists in the spec, and epiphany-core (whose Cargo.toml has no epiphany-layout-ir dependency) cannot borrow any layout-ir type even if one shared the name — the same core-native requirement that forced EngravingBoundingBox. An undefined leaf frozen onto the wire is the KeyContext-shaped gap. Compounding it: the bounding box is documented "relative to the glyph's anchor point" (:3160), so freezing an anchor with no defined coordinate frame freezes a point with an undefined origin	this file (verified 2026-07-23; AnchorPoint appears once in core_spec.tex, defined nowhere; epiphany-core/Cargo.toml lists no layout-ir)	resolved (ratified: core-native AnchorPoint { x: SpaceUnit, y: SpaceUnit }, over the same SpaceUnit as advance_width and EngravingBoundingBox. Plus one normative sentence pinning the frame: x/y in canonical space units, y-up, relative to the glyph's coordinate origin — matching the repo's existing Bravura outline convention — so the anchor and the box it anchors share an unambiguous origin. Freezes in tranche 3b)
P13-S12	SmuflVersion is undefined, and its obvious representation orders SMuFL's real history backwards. SmuflVersionRequirement (core_spec.tex:3269) carries minimum/authored_against of type SmuflVersion, which Chapter 4 references but does not define (it exists only as a Chapter 7 / layout-ir type, glyph.rs:29, with literal-minor encoding — see the resolution). Ordering is load-bearing (SmuflVersionRequirement.minimum, :3271, gates the fallback of req:tuning:smufl-version-fallback), and the type is dual-purpose — it also anchors Chapter 9's GlyphCatalog::smufl_version() / GlyphCatalogIdentity (:10420, :10460), so this freeze touches layout-conformance identity, not just tuning. The trap: SMuFL versions are decimal fractions — 1.12 (2015), 1.18, 1.20 (2016), 1.3 (2019), 1.4 (2021) — that succeed by fraction (0.12 < 0.18 < 0.20 < 0.30 < 0.40). A { major: u16, minor: u16 } storing literal digits with derived Ord orders the minors 3 < 4 < 12 < 18 < 20, placing 1.3 and 1.4 before 1.12; old fonts declaring 1.18-era versions exist forever	this file (verified 2026-07-23 against core_spec.tex:3269-3274 and SMuFL's published release history)	resolved (ratified shape SmuflVersion { major: u16, minor_centi: u16 }, literal-minor storage rejected: the minor is stored fraction-normalized to hundredths — 1.12→(1,12), 1.18→(1,18), 1.20→(1,20), 1.3→(1,30), 1.4→(1,40), rule normative, release table a note. Derived Ord is then correct across the whole real history and collapses the 1.2/1.20 ambiguity (both → (1,20)). Correction on filing: the leaf is NOT undefined — it exists in epiphany-layout-ir/src/glyph.rs:29 as SmuflVersion { major, minor } with LITERAL minor ({1,4}) and derived Ord, so the backwards-ordering bug is LIVE there today (1.3 < 1.12), and it is a direct field of GlyphCatalogIdentity — conformance identity. epiphany-core cannot depend on layout-ir, and SmuflVersionRequirement is core, so the type MUST be defined in core and layout-ir must reuse it — a unification, not a fresh definition, which moves GlyphCatalogIdentity ({1,4}→minor_centi 40). Tranche 3a defines core::SmuflVersion for the tuning use and leaves layout-ir's alone (a bounded, core-invisible homonym since core can't import layout-ir's); tranche 3b performs the unification and the deliberate GlyphCatalogIdentity move. Correction (2026-07-23, tranche 3b-ii): "with golden/vector regen" above was verified false before that tranche's dispatch. No golden, baseline, or vector is pinned to the catalog identity — every assertion on ResolvedLayoutIR::canonical_bytes() is relative (stability, determinism, and a metrics_hash[0] ^= 1 sensitivity check that never touches smufl_version), and the committed SVG/PNG goldens do not embed it. The move changes the catalog's emitted bytes (encode_catalog's minor field: 04 00 → 28 00) in value, with nothing to regenerate. The hundredths scale is blocking for 3b, free to adjust in 3a)
P13-S13	The score tuning context has no canonical persistence path: it can be saved, but never authored, replicated, or merged. Schema major 3 (Push 4b tranche 3b-i) put smufl and overrides on the wire, but the only persisted carrier that embeds a ScoreTuningContext is the acceleration snapshot, which Chapter 8 makes explicitly non-canonical and regenerable and which the bundle may discard and rebuild at will. No canonical carrier embeds it at all: no operation authors it (epiphany-ops has no tuning-context payload anywhere — the vocabulary's only tuning references are the per-pitch TuningReference::Inherit), and MaterializedState carries effects, conflicts, anomalies, objects, spellings, breaks, page-breaks, and pending, but no tuning context. So a user who selects a tuning system or sets a per-voice override has authored something the format cannot durably represent, and req:tuning:tuning-resolution-order's scopes 2–4 (overrides) are in practice unreachable from any document a replica could exchange	this file (found 2026-07-24 while scoping text-projection parity; verified by searching the whole epiphany-ops operation vocabulary and MaterializedState's field list)	CLOSED by genesis tranche G2b (spec/CONTRACT_GENESIS_G2B_TUNING.md). SetTuningContext — kind/tag 34, schema major 3, minor epoch 10 — makes an operation author the tuning context, and the op log is canonical. The closure argument is the metadata precedent, not the canonical base: the base embeds no graph values for any field, including metadata, which has been op-authored since M2d and is durable purely through its operations. The rung paid the accept-set raise this entry predicted (OperationEnvelopeBlock 2→3), charged to this one surface because minimal stamping is a pure function of each payload's value and ScoreTuningContext is the only genesis payload born at v3. The payload carries epiphany_core::TuningContextSettings, a five-field subset, not the full graph type — accidental_extensions is staged off the wire, so a full-value payload would have diverged between a live session and the same document reloaded, and canonical_value! could not have caught it (it compares bytes, never the originating value). Reduction leaves that field untouched. Original filing and disposition follow. — spec/RULING_GENESIS_PERSISTENCE.md (ratified 2026-07-24, 011c68a) reverses Pass-12 K8 and absorbs genesis into the operation set, naming SetTuningContext in its §2 settings table. This was not a tuning-specific defect and did not get a tuning-specific fix: the operation arrives as one of nine surfaces in a single coordinated tranche, and the wire layout is untouched. The disposition, and the evidence that produced it: spec/ANALYSIS_GENESIS_PERSISTENCE.md maps the same gap across the whole Score: eight fields have no operation that can produce them — canvas.layout_defaults, instruments, staff_groups, parts, tuning_context (this entry, its field 10), spelling_precedence, analysis_layers, views, plus identity — and one more (decomposition_attachments) can only be pruned back, never authored. Independently re-verified against the working tree: each field's sole reduce.rs mention is a read-only base-seed read. Of the four dispositions weighed there (a canonical genesis block; closing the op-coverage gap; promoting the canonical base to carry graph values; scope-limiting), the second was taken — every mutable Score field becomes operation-authored — so S13 closes when that tranche lands and is tracked there, not here. One cost this entry contributes as evidence: the per-field operation route is not the free schema-minor it appears. Blocks stamp minimally, at the lowest major whose layouts decode them, and bundle.rs's max_supported_major currently caps OperationEnvelopeBlock at 2 on the explicit ground that no operation payload embeds the tuning context. A SetTuningContext-style operation would make its blocks stamp v3, dragging a role accept-set raise along with the kind append — for one field of eight. That is a concrete instance of the analysis's closing constraint, that any option enumerating fields must be re-audited against its table whenever Score gains a field. This cost is now paid deliberately rather than avoided — but NOT amortised, as first written. The ruling's "one accept-set raise, spent once" implied a single batch landing every new kind at OperationEnvelopeBlock major 3, spreading the raise across nine surfaces. Re-derived 2026-07-28 against the working tree, that is wrong: minimal stamping is a pure function of each payload's value, so the other eight surfaces stamp at major 0, 2, or 2 and never reach the raised bound at all. SetTuningContext is the sole payload born at v3, so the raise is charged to exactly this surface after all — which is why spec/PLAN_GENESIS_OPS.md §4 splits G2 and isolates it in G2b. And the closure argument is not the canonical base: the base is role-bound to major 0 (mis_stamped_canonical_base) and is a MaterializedState, which embeds no graph values for any field — including metadata, op-authored since M2d and durable purely through its operations. S13 closes on that precedent: the op log is canonical, and G2b makes an operation author the tuning context. Consequence to carry forward: once G2b lands, pruning would discard authored genesis state rather than merely re-derivable state, so the standing prohibition on pruning (blocked on disposition C) gains real teeth. Note bundle.rs documents the cap of 2 with the tuning-context rationale in prose, so that comment must move with the cap. Note the wire layouts themselves are already frozen and correct under every disposition: this was never about how the tuning context encodes, only about which carrier embeds it. Not a regression — nothing ever worked; major 3 made the gap visible by giving the value a wire form and no way to reach it)
P13-S14	The chunk schema minor is a normative MUST that no writer has ever honoured, so every appended operation discriminant is unattributable to version skew. binary_format.tex:2330 requires a writer to raise the chunk schema minor when it emits any discriminant appended after the minor it otherwise declares, with a stated rationale: so that a decode failure on an unknown appended discriminant is attributable to a version skew rather than to corruption. SchemaVersion::for_major (bundle/src/ids.rs:204) maps a major to a fixed constant — V0 is {0, 1}, not {0, 0} (ids.rs:173), and V1/V2/V3 carry minor 0 — and, decisively, takes only a major, so no per-kind additive minor can reach it — and both writer-side staging paths derive nothing but the major (testkit/src/bundle_harness.rs:25, textproj/src/serialize.rs:183, each mapping max(OperationEnvelope::schema_major) through for_major). So every kind appended after the initial vocabulary has no additive-version record: 24–27 (Phase-3 first tranche), 28–29 (the schema-major-2 repeat revision), 30 (Push 4a TransposeInterval), and 31 (genesis G1 CreateInstrument). The requirement's own failure mode is exactly what the gap produces: a reader meeting kind 31 from a newer writer cannot distinguish "my vocabulary is stale" from "these bytes are damaged"	this file (found 2026-07-28 during G2a contract review; verified against the working tree — for_major's signature admits no minor, and neither staging path computes one)	RESOLVED at ff9bd0f (the G-minor rung, 2026-07-28). Epoch ladder ratified as minors 2–9, one per additive event, complete against spec/AUDIT_GMINOR_VOCABULARIES.md and monotonic in the introducing commits' order. introduced_minor() landed on the five vocabularies with post-baseline variants, exhaustive and wildcard-free; envelope minor = max over emitted, block = max over envelopes, major and minor derived independently; baselines not normalised. Two things the filing did not anticipate, both from the audit that gated the rung: the scope was never just OperationKind (OperationPayload 3, ReanchorReason 6 and PreconditionFailureReason 10–15 also append), and the manifest reaches OperationKindTag through edit_barriers with no operation envelope in it — so the "no companion bump" note below is superseded: TextDocument carries the manifest SchemaVersion it cannot derive, and the companion moved 0.9.0 → 0.10.0. Op-block stamping did remain projection-invisible, exactly as scoped. Gate [7f] guards the manifest aggregate with an exact-equality oracle. Original filing follows. Ruled 2026-07-28: policy ratified 2026-07-28 as a global additive epoch with content-minimal stamping (spec/PLAN_GMINOR_SCHEMA_MINOR.md §4); the fix is its own rung, sequenced after G2a and before G2b — G2b appends kind 34, so running it first would either grow this sweep mid-flight or ship 34 with the very defect the rung retires. G2a appends kinds 32/33 and takes the violation from eight kinds to ten, knowingly — the debt is already eight deep, the +2 is immaterial against it, and one retroactive sweep over 24–33 is strictly less work than 24–31 now and 32–33 later. What the rung owes: a minor-assignment policy (which minor each past append belongs to — this is a retroactive judgement, not a derivation), a per-kind schema_minor() or equivalent, block minor = max over payloads, a for_major replacement that accepts one, and both staging paths. Scoped 2026-07-28: the minor IS in the chunk content-address preimage (chunk_content_hash, chunk.rs:177), so raising it moves every affected ChunkId and the manifest body naming it — real address churn, though not a req:format:manifest-id problem, since that promise is conditional on the same manifest body. It does not reach the text projection: op-block schemas are discarded during projection (textproj/src/project.rs:424), so no companion bump. And the problem is not per-OperationKind: an envelope also emits the outer OperationPayload discriminant (ResolveEquivocation = 3 carries no kind at all), so the gating work is an audit of every append-only vocabulary reachable from an affected payload. Note this is orthogonal to the major accept-set: OperationEnvelopeBlock stays at 2 through G2a and rises to 3 only at G2b
P13-S15	The OperationKind wire golden lock stops at discriminant 29, so kinds 30–39 — ten of them, every kind appended since the lock was written — have no byte-level lock at all. operation_kind_wire_discriminants_are_golden (crates/epiphany-ops/src/payload.rs:2189) declares let table: [(OperationKind, u8); 30] (:2233), covering 0..=29 and asserting for each that kind.discriminant() has not moved and that the discriminant byte truly leads the canonical encoding. Absent from that table: TransposeInterval (30, Push 4a), CreateInstrument (31, G1), SetCanvasLayoutDefaults (32) and SetSpellingPrecedence (33, G2a), SetTuningContext (34, G2b), CreateStaffGroup (35), CreatePartDefinition (36), CreateAnalysisLayer (37) and CreateView (38, G3a), and CreateMeasure (39, G3b). Their wire discriminants are unlocked: a hand edit to OperationKind::discriminant()'s match — itself the site Push 4a got wrong — would move any of the ten silently, and the encoding-leads-with-the-byte assertion never runs for them. The shape is the failure this project keeps rediscovering: a hand-maintained parallel table that quietly stopped being extended, exactly like the four literal sites found at Push 4a and the six found during G2a — and the golden lock is the one guard specifically meant to catch that class, so its own staleness is the worst place for it. The gap widened from four kinds to ten while the entry sat open, which is itself the argument for closing it now that the genesis ladder has stopped appending	this file (found 2026-07-28 by the G-minor implementation while building s1's coverage; scope corrected 2026-07-30, after the genesis ladder closed at G3b — the original text scoped the repair to kinds 30–33 and an array length of ; 34, both of which were accurate when written and are now six kinds short. Re-verified against the working tree at 4f141ce: the length literal is still 30, and no arm for 30–39 appears in the table)	RESOLVED 2026-07-30 (its own rung, no rider). Never a live incorrectness: every one of the ten discriminants was correct throughout, and binary_format.tex:1443ff (kind table rows 30–39) and :1548–:1552 (tag table rows 30–39) carry them normatively — the gap is the absence of a guard, not a wrong byte. The repair, as landed: operation_kind_wire_discriminants_are_golden's table goes from 30 to 40 entries, locking kinds 30–39 individually — each row asserting both that kind.discriminant() has not moved and that the discriminant byte leads the canonical encoding. No wire, schema-version, or specification change: this adds a guard over assignments that were already normative. Mutation evidence: OperationKind::discriminant()'s SetTuningContext arm was edited 34 → 44 and the extended lock failed ("wire discriminant for SetTuningContext moved — canonical encodings are append-only"); with the same mutation still applied, restricting the loop to &table[..30] — the exact pre-repair coverage — made it pass, which is P13-S15 itself reproduced in the tree rather than argued for. Both edits restored by hand. On the sibling: the tag half is total by construction and needs no extension. the_tag_vocabulary_is_complete (payload.rs:2652) is derived, not hand-written — it computes first_unknown_discriminant() from PAYLOAD_FREE's maximum and asserts the payload-free discriminants are exactly 0..unknown minus Registered's, dense and each decoding — so it already covers 30–39, and operation_kind_tag_vocabulary! makes a tag without a discriminant a compile error rather than a silent omission. One residue, scoped out and stated rather than papered over: density plus round-trip does not pin which tag holds which byte, so a permutation inside the dense range (swapping two assigned discriminants) survives both that test and phase3_tag_discriminants_are_golden (payload.rs:2728), which pins literals for 24–29 only. The same permutation on the kind side is caught by this rung's extended table. The tag-side gap is not part of this rung and is filed as P13-S22 rather than left as a closing remark — a residue recorded only inside a resolved row is a residue that gets lost
P13-S16	StaffGroup.members and Staff.group may disagree in both directions, and nothing detects it. The two fields encode the same relation twice. Staff.group (core_spec.tex:5592, graph.rs:848) names the group a staff belongs to; StaffGroup.members (core_spec.tex:4242, graph.rs:1650) lists the staves in a group. The specification declares both and states neither an authority nor an agreement requirement. Graph invariant 10 checks that each side resolves — a staff's group is declared (invariants.rs:1188), a group's members are declared (invariants.rs:1196) — and never that the two agree, so both stale forms pass every check in the tree: a missing member (s.group == Some(g) while g.members omits s) and a spurious member (g.members contains s while s.group is None or names a different group). Genesis G3a ruled Staff.group the sole authority and StaffGroup.members a non-authoritative denormalized projection (spec/CONTRACT_GENESIS_G3A_ENTITIES.md §1.1, disposition B, ratified 2026-07-29) — a normative semantic ruling that makes the disagreement defined rather than merely undetected, but does not make it impossible: G3a stores the projection without maintaining it	spec/CONTRACT_GENESIS_G3A_ENTITIES.md §1.1 (found 2026-07-29 during G3a contract review; the authorship cycle is what surfaced it — with mints only, no authoring order produces an agreeing pair, since CreateStaff requires its group live at reduce.rs:4372 and CreateStaffGroup requires its members live)	open. Deliberately no code change in G3a. The fix is disposition A: maintain the projection under reduction — create_staff with group: Some(g) appends to g.members — plus a candidate graph invariant 21 enforcing agreement in both directions (20 exist; invariants.rs:149 and the count guard at :6064 both move), and a decision on whether a re-carried CreateStaffGroup compares against its carried members (empty) or the current derived state. Consumers must read Staff.group, never StaffGroup.members, for membership until that lands. The G3b sequencing note is now discharged — G3b landed, and invariant 20 is no longer competing. Locators corrected 2026-07-31; every code citation in this row had drifted, some by hundreds of lines, and spec/CONTRACT_GENESIS_G3A_ENTITIES.md carries two of the same stale pointers (corrected separately, citation-only). The concrete implementation cost, which this row did not state: t8b_both_permitted_stale_forms_hold (reduce.rs:16340) pins both stale forms as passing — asserting staff.group == Some(g) while g.members == vec![], and g2.members == vec![staff2] while staff2.group == None — and its own doc block (:16317–:16339) names disposition A's two maintenance rules as mutations that must break it. So the fix is precisely the mutation an existing test exists to detect: that test must be rewritten, not merely supplemented, and rewriting it is the rung's real cost. Complete participant set, none of it obvious from the two field declarations: create_staff (reduce.rs:4330) and create_staff_group (:4458); both byte-identical re-carry comparators (:4334, :4466), which under A must compare the carried value, never the derived one; both undo retain arms (:2967, :2977), where undoing a CreateStaff currently leaves its id in a live group's members with no guard — the reverse direction is guarded at :6736 and this one is not; base ingest (:1604, :1611), which can seed a disagreeing pair from a blob authored before the rule; valuegen::staff_group (valuegen.rs:376), which documents that it never normalizes; and the four G3a undo-repair tests (:16838, :16992, :17138, :17345) that each construct the missing-member form. No schema consequence: CreateStaffGroup is schema major 0 and the accept-set does not move — but if A normalizes at construction or decode rather than at reduction, the pinned literal-byte vector at ops/src/vectors.rs:829 and the text-projection golden at textproj/src/vectors.rs:198 both break on re-encode injectivity. Reduction-time refusal leaves both untouched — the design fork is resolved that way in the contract. Scoped 2026-07-31 as spec/CONTRACT_P13S16_PROJECTION.md (DRAFT), and immediately BLOCKED on P13-S27. Two findings from that scoping changed the rung's size in opposite directions. Cheaper than feared: the refusal needs no new PreconditionFailureReason and no schema-minor epoch — reduce.rs:1236's container_not_empty() helper already covers "a create carrying children" by its own doc, and three creates already call it for exactly this shape (create_region :4174, create_staff_instance :4246, create_voice :4310); create_staff_group is the sole outlier. More expensive than feared: this is a canonical reduction-semantics change — the same operation set now reduces to a different Score — so core_spec.tex:11614 applies: canonical bases materialized beforehand cannot be reused without rebuilding. That requirement is currently unenforceable (P13-S27), so the rung may not execute until S27 has a disposition. Also found while scoping: base ingest (reduce.rs:1619) reseeds staff_group_values from the maintained value, so a snapshot round trip launders derived members into the carried slot and a re-carry misverdicts only after a reload; and t6 (:16154), t7 (:16229) and t9 (:16454) each assert behaviour the refusal removes, alongside the already-known t8b. Blocking chain, 2026-07-31: S16 → P13-S27 → P13-S28. S16 does not become dispatchable when S27 lands; it additionally needs S27's pin-2a legacy-base disposition, which S28 owns
P13-S17	Binary Format revision history omitted genesis tranche G2b entirely, including the accept-set raise it performed. spec/binary_format.tex's Revision History chapter ran G2a 0.12.0 straight to G-minor 0.13.0 to G3a 0.14.0, with no row anywhere recording G2b (spec/CONTRACT_GENESIS_G2B_TUNING.md, kind/tag 34, SetTuningContext) — even though OperationEnvelopeBlock's accept-set raise 2→3, the first accept-set move since genesis tranche G2a explicitly recorded staying at 2, reached the normative tables (the per-chunk-role accept-set prose at binary_format.tex:2353–:2363, which names the OperationEnvelopeBlock role's maximum as 3 "as of genesis tranche G2b", and the OperationKind minor-additive bullet at :2395–:2418, which records "genesis tranche G2b took 34 (SetTuningContext)") and never the history. G2b's own contract touch table row 27 required "version, Revision History row" among the four-document ritual, and the rung was signed off without it — the gate did not catch a documentation MUST because nothing tests the revision history	spec/CONTRACT_GENESIS_G3A_UNDO_REPAIR.md §0 (found 2026-07-29 during the G3a undo-repair contract review, verified against the working tree: binary_format.tex:3599, :3628, :3643 before this rung's edit)	RESOLVED in this same commit. The chronology is restored: G2b lands as its own 0.14.0 row between G-minor and G3a, and G3a renumbers to 0.15.0 (spec/CONTRACT_GENESIS_G3A_UNDO_REPAIR.md Packet B, pins B2–B3). A new scoped guard in epiphany-testkit makes the omission recurrence-detectable — a distinct principal marker per standalone-row rung (G2a, G-minor, G2b, G3a — G1 is deliberately unguarded, having no standalone row of its own), strictly ordered, with G2b's content anchored within its own row segment — so the entry is filed and closed by the same packet that finds it, unlike P13-S15 and P13-S16, which stay open because their fixes are sequenced to later rungs
P13-S18	Graph invariant 20's agreement and boundary-consistency checks are partial, but only three of the checker's nine non-success paths are the real gap — not the whole checker, as first filed and over-sized. check_measure_meter_consistency (invariants.rs:2662) has nine non-success paths across its two clauses, enumerated and classified by both its own doc comment and spec/CONTRACT_P13S18_MATRIX.md's 18-cell outcome matrix (agreement: A1-A4; boundary: B1-B5). A1 (None declared signature) is inapplicable — nothing to disagree with. A3/B3 (Governing20::None, an empty governing candidate set) are vacuous (pin 6c case 1). A2/B2 (an unresolving declared or governing signature) are delegated to invariant 10's own resolution check (invariants.rs:1220ff, its per-measure and instance-local-grid arms) — verified, not merely asserted: the matrix rung's mutations M7/M8 delete each arm in turn and show the SAME condition go unreported by the ENTIRE workspace suite, not merely by invariant 20. B1 (the first measure) is the already-filed pickup/anacrusis deferral, P13-S19. Exactly three paths are genuine abstention: A4 and B4 (Governing20::Indeterminate — the comparable relation cannot place a candidate) and B5 (a decided governing selection with no computable musical delta — order without distance). Also corrected: the earlier "any Measure end anchor is incomparable" claim was wrong only in its "any" — same-id, same-position End<->End IS comparable under c2 when its offsets compare (measure20_offset_order, :2419, returns None only for Musical against WallClock); distinct-id End anchors genuinely are incomparable (c3 restricts to Start+Zero) and do reach A4/B4/B5; and the resolve_anchor Measure arm citation (now :503-:516, formerly miscited as :400ff — see the companion line-number repair to CONTRACT_GENESIS_G3B_MEASURE.md:206) names the missing duration machinery, not invariant 20's own execution path, which never calls resolve_anchor. And the WallClock-anchored measure start epiphany-ops::valuegen::measure (ops/src/valuegen.rs:447) actually emits does not uniformly abstain both clauses: against WallClock-anchored meter changes (c5) agreement DECIDES and only the boundary clause abstains (B5, since measure20_musical_delta never returns a WallClock delta); against Region-anchored meter changes both abstain (A4/B4), because WallClock against Region has no comparability arm at all	spec/CONTRACT_P13S18_MATRIX.md pin 8 (ratified 2026-07-31, a diagnostic and bookkeeping rung: test coverage over check_measure_meter_consistency's existing nine paths plus these three ledger repairs, no behaviour change)	open, at its true size. A4, B4 and B5 remain real, still gated on the deferred common-timeline machinery — but the capability that closes them is now filed as P13-S23 (below), not P11-C5: P11-C5 is a narrower re-anchoring proximity metric this entry previously over-cited as the gate; P13-S23 is the broader "place any anchor pair on a common timeline and measure musical distance" capability that actually owns A4, B4 and B5
P13-S19	A pickup's successor, not the pickup itself, is what CreateMeasure and invariant 20 refuse — pickup/anacrusis measures are authorable, but the rest of the instance becomes unauthorable once one exists. Genesis tranche G3b (spec/CONTRACT_GENESIS_G3B_MEASURE.md pin 9) makes create_measure's clauses 1 (ordering) and 3 (boundary distance) vacuous for a first measure, and invariant 20's boundary clause is symmetrically exempt at i == 0 — both for lack of a predecessor. The agreement clause (clause 2) is not predecessor-dependent and applies to a first measure like any other: a pickup declaring a signature that disagrees with the governing grid is itself refused (MeasureMeterMismatch) and flagged; only a pickup declaring None or a matching signature clears it, and only when its other preconditions (a live parent StaffInstance, resolving referents) are satisfied too. The real defect is one measure downstream: both rules compare delta(prev.start, m.start) against the governing signature's full measure_duration(), when the true distance is prev's own content duration — those coincide only for full measures. A pickup's successor is therefore refused (MeasureMeterMismatch) by create_measure and flagged by invariant 20, which is what actually makes the rest of the instance unauthorable once a pickup is minted. This applies to any partial measure, not only a first one: a mid-score partial measure enters successfully (nothing examines its own duration when it is created), while its successor is refused the identical way — the failure is always attributed to the measure after the partial one. Closing this needs a per-measure duration — the "partial measure" notion this entry already names — and that is deliberately not introduced by the correction that reworded this entry. See also P13-S24, the Chapter 3 derived-notation deferral that shares this same missing concept but touches neither invariant 20 nor CreateMeasure	spec/CONTRACT_GENESIS_G3B_MEASURE.md pin 9 (filed 2026-07-30 during the G3b contract's own drafting); corrected 2026-07-31 by spec/CONTRACT_P13S19_PARTIAL.md, which found the original wording true only of the pickup itself and, read as written, indistinguishable from "pickups work"	open, deliberately. A pickup/anacrusis authoring story — whatever shape it takes — is later schema-fill work, sequenced after the genesis ladder closes; it now depends on a per-measure duration concept, which does not yet exist
P13-S24	The Chapter 3 integer-grid metric splitter assumes the region origin falls on a barline, deferring anacrusis/pickup handling — a second, independent site carrying the same missing partial-measure-duration concept P13-S19 names for invariant 20 and CreateMeasure. core_spec.tex:2484 (Chapter 3, derived notation) states the assumption directly; the same deferral is mirrored at crates/epiphany-core/DECISIONS.md:340 ("region origin assumed to be a barline (anacrusis/pickup deferred)"). Not a duplicate of P13-S19: it predates genesis tranche G3b, lives in the derived-notation splitter rather than the graph model, and does not touch invariant 20 or CreateMeasure at all — a pickup measure can now be authored (P13-S19's corrected scope) with no effect whatsoever on this splitter's own barline assumption. The two entries are filed separately because they own different subsystems' work — merging them would conflate a graph-model rung with a derived-notation one — and both wait on the same underlying capability: a per-measure partial-duration notion neither rung introduces. See also P13-S19	core_spec.tex:2484 and crates/epiphany-core/DECISIONS.md:340 (mirrored deferral sites; filed 2026-07-31 during spec/CONTRACT_P13S19_PARTIAL.md's drafting, pin 6)	open. No code owed by this rung — filed so the deferral is discoverable on its own terms, independent of P13-S19, with which it shares a root cause but not a subsystem
P13-S20	decode.rs's precondition_reason decoder stopped at discriminant 13, so PreconditionFailureReason 14 (AcousticRealizationPinned) and 15 (TranspositionOutOfRange), both live since Push 4a, encoded but could not decode. A materialized effect carrying either reason failed canonical round-trip in production code, undetected because epiphany-testkit's precondition_failure_reason generator (generators.rs:417, then rng.below(14)) never drew past 13 despite its doc comment claiming "every core and registered variant"	spec/CONTRACT_GENESIS_G3B_MEASURE.md §0 / touch-table row 12a (found 2026-07-30 during the G3b contract's own drafting, verified against the working tree: decode.rs:205ff ended at 13, generators.rs:417 drew below(14))	RESOLVED in this rung (genesis tranche G3b packet 1, e64a4b7). decode.rs's precondition_reason now decodes discriminants 14 through 18 (the pre-existing 14/15 hole plus G3b's own 16–18), and generators.rs's precondition_failure_reason now draws below(19) with arms for all of them — the generator's doc comment claim is true again, and the decode hole this rung found already latent in the tree, not introduced by it, is closed alongside G3b's own additions
P13-S21	The wire specification never documented PreconditionFailureReason 14 and 15, so the decoder hole P13-S20 repaired had no normative text to be measured against. binary_format.tex's bounded reason table (\sectionsc{Bounded Enumerations}) ran \tablenums{13} straight to G3b's \tablenums{16}, and Push 4a's own Revision History row recorded only OperationKind \tablenums{30} (TransposeInterval) while saying nothing about the two reasons it appended in the same epoch. AcousticRealizationPinned and TranspositionOutOfRange appeared nowhere in the document, though the Operation Catalog documented both at 0.8.0 and effect.rs has carried both since Push 4a. This is P13-S20's specification-side twin: the same two variants, the same silence, and the reason a decoder that stopped at 13 could sit in the tree unchallenged — a conforming implementer reading only the wire specification would have built exactly that decoder and been right	Found 2026-07-30 while verifying genesis tranche G3b packet 3b (d58eee8), whose touch-table row 28 scoped the reason table to discriminants 16–18 only; the executing agent correctly declined to widen scope and reported the gap instead	RESOLVED in this rung (the G3b pre-push repair). The bounded table gains \tablenums{14} and \tablenums{15} with their refusal conditions, and Push 4a's Revision History row is amended to record both appends in its own epoch — no version bump and no new history row, because this documents an assignment that has been normative since Push 4a rather than making one. binary_format_history.rs gains a bounded regression test requiring both names in the reason table and in Push 4a's own row segment, so neither the G3b row's prose nor the Operation Catalog's coverage can satisfy it
P13-S22	OperationKindTag's normative tag-to-byte mapping is asserted for a minority of the vocabulary; the rest is defended only incidentally, by frozen byte artifacts that cannot say what broke. P13-S15 closed the OperationKind half with one table covering 0..=39. The tag half has no equivalent. Semantic locks — a named tag bound to a literal byte — exist only for: 24–29 (phase3_tag_discriminants_are_golden, payload.rs:2728), 34 (one assertion inside a G2b reduction test, reduce.rs:12744), 35–38 (t1_g3a_kinds_and_tags_are_35_to_38_in_both_spaces, reduce.rs:15931), 39 (payload.rs:2936); 16 (Registered, whose corpus row is emitted under its variant name at ops/src/vectors.rs:210 and whose committed literal leads with 0x10 at spec/vectors/decode_vectors.txt:80, so the drift comparison binds the association); and — incidentally — 1, spelled in a comment inside a golden blob (barrier.rs:1058, "len 1, DeleteEvent (tag 1)"). Tags 0, 2–15, 17–23, and 30–33 have no semantic lock at all. The derived tests do not supply one: the_tag_vocabulary_is_complete (payload.rs:2652) proves completeness, density, and round-trip — that the payload-free discriminants are exactly 0..unknown with no gaps and each decodes — which constrains the shape of the assignment, not which tag holds which byte, and operation_kind_tag_vocabulary! guarantees coverage rather than value. What actually catches a permutation is byte-level goldens that embed the tag by accident, and they diagnose it as a blob mismatch: three probes, three catches, none of them naming a tag (below)	this file (found 2026-07-30 while closing P13-S15, whose kind-side extension made the asymmetry visible; the entry was rewritten the same day after its own probes falsified its first draft. That draft claimed a permutation was invisible and named 32↔33 as the demonstration. Three permutations were then executed against the working tree at dcb28f0 and all three failed, so the claim was wrong and the severity is lower than filed: 32↔33 (SetCanvasLayoutDefaults/SetSpellingPrecedence) → caught by the_committed_corpus_matches_the_generator (testkit/src/vectors.rs:225); 1↔2 (DeleteEvent/ModifyEvent) → caught by edit_barriers_blob_bytes_are_golden (layout-ir/src/barrier.rs:1061); 2↔3 (ModifyEvent/RespellPitch) → caught by the corpus again. Each mutation was restored by hand and the suite returned to 1541/0)	RESOLVED 2026-07-31 (its own rung, no rider). Disposition A — a hand-written literal tag→byte table in crates/epiphany-ops/src/payload.rs (tag_wire_discriminants_are_golden) — ruled 2026-07-31. Disposition B (giving the numbered corpus rows the variant-naming property) was considered and is deferred to a separate ledger-only commit; it is no part of this rung. Corrected inventory, recounted from the tree at 17c1d67: semantic locks — a named variant bound to a literal byte — exist for 14 of 40 (1, 16, 24–29, 34, 35–38, 39); unlocked: 26 (0, 2–15, 17–23, 30–33). Not a live incorrectness, and — confirmed — not an undefended one either: every tag discriminant was correct throughout, and binary_format.tex:1548–:1552 carries the mapping normatively. The gap was intent and diagnosis, not exposure — verified at all six incidental-catch sites (spec/CONTRACT_P13S22_TAGLOCK.md §0.4). The corpus's tag coverage is deliberate, not accidental — ops/src/vectors.rs:201–:204 emits one row per tag straight from the vocabulary — but those rows lock byte→byte, not variant→byte: each is named tag_{discriminant} and carries [discriminant], both derived from the value alone, so tag_32 asserts that 0x20 round-trips — never that SetCanvasLayoutDefaults is 32. The signing mutation, executed: the coordinated 32↔33 permutation — swapping both the discriminant literals and the two declaration lines inside operation_kind_tag_vocabulary!, so PAYLOAD_FREE still emits ascending discriminants and every derived artifact stays byte-identical while the variant→byte association is reversed — run against the working tree at 17c1d67: cargo test --workspace → 1558 passed, 0 failed, byte-identical to the clean baseline. Two operations exchanged wire discriminants in complete silence. Restored by hand; git status verified clean. The repair, as landed: tag_wire_discriminants_are_golden (payload.rs), a hand-typed 40-row [(OperationKindTag, u8); 40] literal table transcribed independently from operation_kind_tag_vocabulary!'s invocation rather than derived from its output, so that moving a tag now fails by name. Two wording repairs to this row's earlier drafts, both now corrected: the table supersedes one fragment (phase3_tag_discriminants_are_golden, retired) — not "the six scattered fragments" as an earlier draft of this row claimed — and deliberately duplicates three more (payload.rs:3086, reduce.rs:12744, reduce.rs:15941) for the locality reason spec/CONTRACT_P13S22_TAGLOCK.md pin 5 states; and the mutation that signs the rung is the coordinated permutation above, not the literal-only inverse ("delete the proposed table and show that some permutation then passes") this row's original probe-design note proposed — that literal-only form is caught by the_committed_corpus_matches_the_generator and proves nothing. Its own rung, not a rider
P13-S23	No filed candidate owns "place any anchor pair on a common timeline and measure musical distance along it" — P13-S18 previously mis-cited a narrower capability as its gate. Two disjoint deficiencies, both owned by this candidate. (1) No ordering. The pair is not comparable under any of measure20_comparable_order's five shapes c1-c5 (invariants.rs:2457) at all — whether the failure is in the referent (distinct Event ids; distinct Measure ids outside c3's Start+Zero restriction), the variant or selector (Event against Measure, Measure against Region, differing pos/edge), or the clock (Musical against WallClock, including inside measure20_offset_order, :2419) — this is what invariant 20's A4 and B4 are made of. (2) Ordering without a usable delta. The pair IS comparable and still yields no musical distance: c3 supplies a vector index (an order, never a distance), and c5 compares two WallClocks, and measure20_musical_delta (:2522) never returns a WallClock delta (:2527) — this is what invariant 20's B5 is made of. Scoping this as merely "anchors of differing shapes" or "not directly comparable under c1-c5" would exclude B5 entirely — S5 (distinct-id Measure Start/Zero) is c3-comparable and S1 (WallClock measures, WallClock meter changes) is c5-comparable, and both still reach B5 — an earlier draft of this filing made exactly that narrower mistake. Explicitly broader than P11-C5: P11-C5 (PASS11_WORKLIST.md:159) is a re-anchoring proximity metric that resolves "when the graph-mutation phase tracks resolved positions", and covers narrowly the two-distinct-Events case (CONTRACT_GENESIS_G3B_MEASURE.md:223, effect.rs:139-:142's PositionOutsideRegion Reserved note); P13-S23 is the timeline itself, whatever positions get placed on it. Names its dependents: invariant 20's A4, B4 and B5, and PositionOutsideRegion's Reserved status	spec/CONTRACT_P13S18_MATRIX.md pin 10 (filed 2026-07-31 during the same rung that corrected P13-S18's over-narrow P11-C5 citation)	open. No code owed by this rung. Closing it needs the deferred common-timeline/duration machinery — once a Measure end, a distinct-id Measure/Event referent, or an Event position on a wall-clock-placed region can be placed on a common timeline with a musical distance, invariant 20's A4/B4/B5 residue and PositionOutsideRegion's Reserved status shrink together
P13-S25	The committed decode corpus's numbered tag rows lock byte→byte, not variant→byte — one row already has the property the other thirty-nine lack. ops/src/vectors.rs:206–:209 emits one row per tag as format!("tag_{:02}", tag.discriminant()) carrying [discriminant]: both the name and the payload derive from the value alone, so tag_32 asserts that 0x20 round-trips and never that SetCanvasLayoutDefaults is 32. The Registered row (:210–:217) is different — its name is the hard-coded string "registered" while its bytes are computed from the variant, so the frozen literal at spec/vectors/decode_vectors.txt:80 binds the association. Disposition B of P13-S22: give the numbered rows the same property. It does catch the coordinated permutation — by exactly the Registered mechanism, with the committed text serving as the independent statement — and it propagates the property to every implementation that reads the cross-impl corpus, which an in-crate Rust test cannot do	spec/CONTRACT_P13S22_TAGLOCK.md (disposition B, considered and deferred during the 2026-07-31 ruling; filed rather than left as a closing remark, per the same discipline that moved P13-S22 out of P13-S15's resolved row)	open. Complementary to P13-S22, not a replacement for it, and not a re-litigation of it. P13-S22 landed disposition A (tag_wire_discriminants_are_golden, payload.rs:2730), which fails by variant name inside the crate. B cannot supply that: its failure is still "spec/vectors/decode_vectors.txt is stale. Regenerate: …" (testkit/src/vectors.rs:224) — the misleading diagnosis P13-S22 was filed about — even though the diff text would now name variants. What B buys is cross-implementation reach; what it costs is churn in a committed artifact other implementations pin. Both are wanted; neither substitutes for the other. Sequencing note: run B's own signing mutation as the coordinated permutation (literals and declaration lines), since the literal-only form is caught today by row ordering and proves nothing
P13-S26	A doc comment in shipped code claims a specification repair that never landed, and the claim is guarded on the code side and nowhere on the specification side. crates/epiphany-core/src/invariants.rs:69–:71 enumerates invariant 10's four reference classes and states that “genesis tranche G3a repairs this prose to name what the check body already enforced”. It did not. core_spec.tex:6570–:6572, the normative enumeration item 10, still reads only “Every cross-cutting structure's references resolve to extant objects in the graph, except where explicit re-anchoring rules permit transient dangling states during edits” — naming neither a staff's declared instrument, a staff's group, a staff group's members, a part's staves, a view's active layers, nor any of the meter/time-signature references the Rust doc lists and the check body enforces. The repair landed in the Rust doc comment only. The asymmetry is the defect's sharp edge: the Rust doc block is protected by a grep-assert, t12_invariant_10_doc_comment_names_the_four_reference_classes (invariants.rs:4554, needles at :4562–:4566), so the side that is wrong about the other is the side that is locked, while the side that is actually stale is unguarded	this file (found 2026-07-31 during P13-S16 reconnaissance, while verifying that row's invariant-10 citations; no ledger entry covered it)	open. Not a live incorrectness — the check body is correct and enforces every class; only the normative prose under-describes it, and only the doc comment lies about that. A P13-S9 instance, and filed deliberately as one: the loud form (a dangling citation) is caught by requirement_labels.rs, and this quiet form — a true-sounding claim about another document's state — is caught by nothing. invariants.rs:69–:71 MUST NOT be “corrected” on its own. It is currently the only artifact in the tree pointing at the core_spec.tex gap; softening the Rust claim in isolation would make the specification defect invisible and convert a caught defect into an uncaught one — which is P13-S9's stated failure mode verbatim. Repair both sides in one rung, and consider whether the LaTeX enumeration deserves the grep-assert its Rust mirror already has
P13-S27	The reduction-algorithm-version machinery is self-referential, so the one check that would detect a canonical-semantics change necessarily passes. core_spec.tex:11614–:11617 is normative — "Snapshots produced under an earlier algorithm version cannot be used as canonical bases under a later one without rebuilding" — and :14369–:14372 states that replicas at differing versions "may produce different canonical states from the same operation set." The machinery to enforce it appears to exist: ReductionAlgorithmVersion (bundle/src/ids.rs:291) is a superblock wire field (bytes 68..72, superblock.rs:20); reduction_version_for (bundle.rs:989) sets a new superblock's value; and open (bundle.rs:396–:399) rejects a mismatch. But the writer sources the value from the canonical base's own self-report (mapping the base's reduction_algorithm_version through unwrap_or_default()), and the reader compares it only against the superblock that value seeded. Nothing compares either against the semantics the running implementation actually implements. The check is not vacuous — it catches a corrupt or tampered base whose version disagrees with its superblock — but it necessarily passes for a conformingly propagated stale base, which is precisely the case the requirement exists to prevent. Supporting: no constant or accessor anywhere names the implementation's current reduction semantics, and ids.rs:288–:289 states that "the algorithm catalog itself lives in epiphany-ops" while nothing of the kind exists in that crate — a second instance of P13-S26's pattern, a doc comment asserting a false fact about another module	spec/CONTRACT_P13S16_PROJECTION.md pin 0 (found 2026-07-31 while scoping P13-S16, which is a canonical reduction-semantics change and therefore the first rung to need this guarantee; filed in the same ledger edit as the row it blocks)	UNBLOCKED 2026-08-07 — the format-epoch rung landed; dispatchable, and still blocking P13-S16. (Was: open, BLOCKED on P13-S28.) Scoped 2026-07-31 as spec/CONTRACT_P13S27_REDUCTION_AUTHORITY.md (DRAFT, not dispatchable). Rulings taken: a typed BundleCapabilities required at both Bundle::open and Bundle::create and carried on the Bundle — no default, so every caller states the semantics it implements — and outright rejection on mismatch via a new CanonicalBaseRequiresRebuild error, not read-only and not an integrity anomaly. Storing the capability keeps all 57 commit sites unchanged; only open (57 sites) and create (32) move. The scoping also falsified this row's first reading that the writer path was test-only: epiphany-textproj's serialize_document (serialize.rs:119) and project.rs:936 are production paths that copy a base's reduction_algorithm_version verbatim into a fresh SnapshotRef, which commit_versioned then stamps into the superblock (bundle.rs:798) — so production mints self-consistent stale documents without ever calling open, and the capability must govern writers too. What blocks it: contract pin 2a. Baseline authority 0 does not preserve the corpus (serialize.rs:327 stamps 1 and round-trips it; vectors.rs:353/:363 likewise), and once P13-S16 moves the authority to 1, a pre-S27 base that happens to carry 1 is indistinguishable from a legitimately rebuilt one — a raw u32 carries no provenance. Four dispositions are recorded there; FORMAT_MINOR as a provenance carrier was proposed and rejected (the header never changes after creation, core_spec.tex:10799, so a legacy bundle committing a freshly validated base keeps its old minor forever; and a minor change may only append append-safe discriminants, :12258, not alter acceptance semantics). The surviving requirement — provenance must ride a container property old readers cannot silently accept and a later commit cannot inherit unchanged — is a format-epoch design, filed as P13-S28. Scope of the claim, deliberately narrow: this establishes that the current implementation has no detection mechanism. It does not establish that no reduction-semantics change in the project's history was ever detectable — that needs a history audit not yet done, and the stronger sentence is deliberately not written here. What closing it requires: an authority naming the semantics this build implements, and a rejection-or-rebuild path when a base disagrees with it. Until then any rung changing canonical reduction semantics can record its break in prose but cannot make stale bases unusable — which is why P13-S16's contract is complete, ratifiable as a plan, and not dispatchable. Method note: an earlier draft of S16's pin 0 claimed no writer path existed at all. That was false, and the way it was false is the point — the search behind it looked for ReductionAlgorithmVersion( constructor calls, which cannot find a path that propagates an existing value without constructing one. The instrument could not observe the thing it was used to rule out. UNBLOCKED 2026-08-07: the format-epoch rung landed and its pin 8 resolves pin 2a — reduction-version authority is meaningful only in major-1 containers, so legacy bases are refused by container epoch and never by version arithmetic. The collision pin 2a identified never has to be adjudicated: a pre-S27 base carrying 1 and a rebuilt S16 base carrying 1 are indistinguishable as numbers but can never meet, because the former exists only in a major-0 container, refused at the epoch boundary before any version is compared. The u32 never has to carry provenance because the container does. S27 now additionally owes three inherited items (both interim refusals converted to validation, M8's deferred laundering demonstration, pin 3c's two suspended conformance assertions), recorded in its contract as required tests. RATIFIED 2026-08-07 after review round 1 — RATIFICATION WITHDRAWN 2026-08-07, see round 2 below. RATIFIED 2026-08-08 on the repository owner's authority, after NINETEEN adversarial review rounds the last of which returned zero findings. PINS FROZEN — executed, not edited; a defect found during execution is reported, not patched in place. DISPATCHED for execution, with the work to be left STAGED and the execution report subject to INDEPENDENT REVIEW before completion is accepted, covering in particular M7's three observations and its control. Not settled by ratification: M7's authority/base leg is unverifiable until this rung is implemented, and every gate, test and mutation is specified while none has been run. Prior status through the review: NOT RATIFIED, NOT DISPATCHABLE, pins NOT frozen, awaiting the next independent review round. Which rounds have closed and the running tally live in the contract's own history table and are deliberately NOT restated here — this clause carried a count through two rounds and went stale in both, and the round list went stale the same way; no execution work may begin. Round 1 was run because this contract had reached "dispatchable" with zero ratification rounds on record, against the standing rule that contracts go through adversarial review before dispatch. Round 1 returned nine findings, four blocking, all now carried in the contract. Correcting this row's own preceding clause: the three inherited items were not all "recorded as required tests" — obligation 2, M8's laundering demonstration, appeared in neither the contract's test section nor its mutation plan, while that section's preamble claimed all of them were tests. It is now M7, and ruled a mutation rather than a capability restoration: the format rung's text refusal is permanent, COMPANION_VERSION stays 0.14.0, and the text-projection corpus keeps canonical_bases reach 0. The other blocking three: §0.4's commit-site count counted a same-named method in epiphany-editor-core, a crate with no epiphany-bundle dependency at all — the third instrument failure recorded in that one section; three independent stale list-counts (the test-section header, gate 1, and three report items) all naming figures the document had outgrown; and testkit/tests/requirement_labels.rs missing from the touch table while pin 9 may move CORE_REQUIREMENT_COUNT 213 → 214 — the escapee CLAUDE.md names by name, which also escaped the format-epoch rung. Non-blocking: locator drift since 381c498 (bc06706 grew bundle.rs by 338 lines; pin 5's own :396–:399 confirmed unmoved), pin 2a's corpus evidence superseded by the 2 → 0 rebuild, Bundle::open( 57 → 60, gate 6a's scope widened to epiphany-testkit, and a missing commit-side positive test now added as test 8. REVIEW ROUND 2, 2026-08-07, against the frozen contract: six further findings, four blocking — and round 1's ratification was therefore premature. (1) The call-site correction had been applied to §0.4 only, leaving the "Rung type" paragraph at 57 and touch row 2 claiming bundle.rs has 35 opens — a figure that was never bundle.rs alone and is stale besides — which made the required reconciliation impossible. (2) §0.4 called project.rs:936 a production bundle writer; #[cfg(test)] starts at :630 and every Bundle call in that file is below it, so the writer-path correction stands on serialize.rs alone. (3) M5 was unexecutable: serialize_document refuses bases at serialize.rs:151, so its output is base-free, and pin 5 with test 4 require base-free bundles to open at any authority — split into M5a/M5b. (4) M6's second half was unexecutable: open rejects a stale base, create rejects a base-bearing manifest (bundle.rs:234), and commit validates what it emits, so no caller can hold an open Bundle with a stale inherited base — replaced by broadening rather than narrowing, with the unreachability itself reported as the stronger result. Non-blocking: pin 3a's justification ("production code mints a self-consistent stale document") is false in-tree — zero production paths stage a base — so it now rests on guarding the public commit_versioned API; and serialize.rs:157 is dead code orphaned by the :151 guard, recorded and explicitly not repaired here. Two of round 2's findings were introduced by round 1 — ruling M7's refusal permanent is what made M5 unexecutable, and test 8 was added without re-deriving M6 against the same reachability. Method note: an amendment is a change to the system, not a patch to a line; a round must re-derive every mutation against every ruling the previous round made. Round 3 is warranted before dispatch — the defect rate has not fallen (9, then 6), and "dispatchable" is a claim requiring evidence of convergence rather than a status reached by running out of findings. REVIEW ROUND 3, 2026-08-07, INDEPENDENT, against b842975: six findings, four blocking — and every blocking finding was a defect in text rounds 1 and 2 wrote. (1) Pin 3a still carried the rationale round 2 retracted — §0.4 states there is no in-tree production base writer while pin 3a still said "§0.4 shows production code minting a stale document", so the contract asserted a claim and its negation; the third occurrence of fix-one-site-leave-the-others. Rewritten onto the footing that survives: commit/commit_versioned are public API and guard out-of-tree callers, not an in-tree path. (2) M5a had no observation mechanism — pin 3 required the capability be stored and nothing exposed it; Bundle carries 17 public accessors and none for capabilities, so no epiphany-textproj test could inspect it. Bundle::capabilities() is now pinned — new scope, flagged for round 4. (3) M5b could not fail: if the supplied capability and the base version both derive from CURRENT_REDUCTION_ALGORITHM_VERSION — the natural implementation, since roundtrip.rs:367 currently hardcodes ReductionAlgorithmVersion(0) — both operands move together and the comparison passes for every value. This is §0.1's own tautology reproduced inside the mutation built to detect it. The base version must now come from a source that does not track the authority (persisted artifact or deliberate literal), with both operands' provenance reported. (4) M6's replacement named a scenario with no test — test 6 stops at opening, so nothing asserted that an unrelated commit succeeds, and an implementation rejecting every post-base commit passed tests 2/5/6/8 while the broadening had nothing to break; test 9 added. Cleanup: touch row 7 listed generators.rs as "call sites, real authority" though it has zero Bundle::open/create calls and its rng.range(0, 8) versions are precisely the arbitrary wire values pin 3b assigns to synthetic capabilities — split to row 7a; and §7's call-site attribution credited round 1 alone where rounds 1 and 2 are both load-bearing. The pattern is now legible and it is not about counts: three separate mutations were unrunnable in three different ways — M5a could not observe, M5b could not fail, M6 had nothing to break. §7 item 4a now requires, for every mutation, the named test it breaks and the provenance of each operand. Defect rate across three rounds: 9, 6, 6 — not converging. The newest text (pin 3's accessor, M5a, M5b, test 9, row 7a) has had zero adversarial passes and was written by the same agent whose previous two attempts round 3 falsified. REVIEW ROUND 4, 2026-08-07, INDEPENDENT, against 53292f6: five findings, four blocking. It accepted pin 3's capabilities() accessor as bounded — the first new text any round has passed — and found the M5 pair defective a third time. (1) M5b cited the wrong value: roundtrip.rs:367 sits in assert_score_serialization_stable (:332) and versions an acceleration snapshot, not a canonical base, while assert_reduction_serialization_stable has no base at all because pin 3c suspended it — so the value round 3 warned the implementer not to touch was irrelevant to the authority check. The tautology diagnosis stands; only its evidence was wrong. (2) The instrument was left unchosen — round 3 said "the rung picks one" and offered two, one of which does not exist for the nominated crate, since craft_image_with_base is a private fn inside epiphany-bundle's #[cfg(test)] module (:1648). Now chosen: commit-then-reopen through public API only — build with synthetic_for_fixture(0), commit a base carrying the literal 0, reopen those bytes under the real constant. (3) No test could assert the error fields: assert_reduction_serialization_stable returns () and reopens with .expect (:292), so a mismatch panics rather than yielding a matchable CanonicalBaseRequiresRebuild { base, current }. Test 10b added. (4) M5a violated §7 item 4a, the rule round 3 added in the same edit — it named no test, and its natural assertion compares the constant with itself and cannot fail. Test 10a added, asserting against a deliberate literal. Round 3's error is the one to carry: it grepped ReductionAlgorithmVersion, saw a roundtrip.rs hit, and attributed it without resolving the enclosing item — the same shape as §0.4's .commit( miscount that round 1 had already recorded as a lesson. Recording a defect is not the same as not committing it. Both literals in tests 10a/10b are load-bearing as literals; §7 item 4b now requires confirming neither was tidied into the constant, a failure mode invisible to the suite. Defect rate: 9, 6, 6, 5 — still not converging after four rounds, and every blocking finding in rounds 3 and 4 was in text written to fix the previous round. REVIEW ROUND 5, 2026-08-08, INDEPENDENT, against df9e528: four findings, two blocking — the first round in which blocking findings fell below four. (1) The status history was numerically stale again — "amended three times … fifteen findings so far, eight blocking" were the round-2 figures, left standing through rounds 3 and 4 while the tables recording those very rounds sat directly below them. This is the fifth count-staleness defect in five rounds, and it was in the one block the author edited every round. Replaced with a table, so a round appends a row rather than requiring a number to be found and re-derived. (2) Test 10b could not make the two-field assertion M5b requires: §3 said only "assert it opens", and under mutation that yields a bare Err or a panic — a #[test] -> Result that returns Err asserts nothing about that error's fields, so M5b's required observation had no home in the test M5b names. Both Result arms are now pinned, plus a third for the wrong-error case, so the mutation run produces a verified observation rather than a stack trace. Smaller: M5b's "cannot be tidied" claim was false — keeping synthetic_for_fixture while passing CURRENT_REDUCTION_ALGORITHM_VERSION as both its argument and the base version preserves the fixture and fully restores the tautology, so the structure does not protect itself and the real protection is §7 item 4b; round 4 asserted a structural guarantee that undercut the procedural check actually doing the work, which is the same error as reasoning that a mutation would fail instead of running it. And §3's preamble still said the tests were "in epiphany-bundle" after round 4 added two that cannot be, since epiphany-bundle must not depend on epiphany-ops and reaching the real authority is their entire purpose — corrected, with each test's touch-table home named. Blocking findings by round: 4, 4, 4, 4, 2 — the first movement in four rounds and the first weak evidence of convergence, set against the fact that every round since the third has found blocking defects in text written to fix its predecessor. REVIEW ROUND 6, 2026-08-08, INDEPENDENT, against 03c85dd: three findings, ALL THREE BLOCKING, and all three in text round 5 wrote. (1) The amendment tally went stale inside the block round 5 restructured to prevent exactly that — round 5 turned the review totals into a table and left "amended five times … rounds 1–4" as prose immediately above it. The amendment count is now the number of rows, with no separate figure to go stale. (2) §3's test-home correction was itself false: round 5 wrote "tests 1–9 in epiphany-bundle", but test 7 is assert_reduction_serialization_stable, which the same section names as testkit/src/roundtrip.rs. Two wrong versions of that sentence, both written while fixing it; replaced with a per-crate table (1–6/8/9 bundle, 7 and 10b testkit, 10a textproj). (3) §7 item 4b protected one operand where test 10b has two — replacing both synthetic_for_fixture(0) and the committed base's ReductionAlgorithmVersion(0) with the constant keeps the synthetic call in place and fully restores the tautology, and test 10b's Err arm never executes in the unmutated run, so its literal cannot detect it. Item 4b now enumerates all three fixture operands individually and requires each quoted verbatim. All three findings are one defect in different clothes: a fix applied to the site named rather than to every site the claim covers — the sixth count-staleness defect in six rounds and the third range-correction that did not check its own range. The mechanism that works is structural, not vigilant: the review totals stopped going stale when they became a table, the amendment count did not because it stayed prose, and item 4b stopped being under-specified when it became a table. Demonstrated a seventh time inside round 6's own amendment, where the new table's Total row was first written "6 amendments" — a free-standing count, three paragraphs after the sentence declaring no such count exists, and already wrong at seven rows; caught before commit and replaced with "one amendment per row". Prose invites a number and a table does not; the defence must be the shape of the artifact, not the attention of the editor. REVIEW ROUND 7, 2026-08-08, INDEPENDENT, against c0d896c: three findings, all blocking, and all three the same defect — a claim living in two places and fixed in one. (1) §7 item 4a was unsatisfiable: it required every mutation to name "the test it breaks", while item 1 four paragraphs above states that M4 is observed to compile (no test is possible — that is why pin 3's prohibition is a review rule) and M7's expected outcome is success. A report obeying 4a literally could not be written, and the honest response would have been to invent a test. 4a is now a table of what each of the eight mutations owes, with M4 and M7 carved out explicitly. (2) Round 6's three-literal correction reached §7 and not §3 — §3 still said "both literals … tidying either", so the contract carried the fixed and the broken version of the same claim, reopening exactly the narrow-scope ambiguity round 6 existed to close. §3 no longer states the count at all; it points at item 4b. (3) "Rounds 3, 4 and 5 were independent" went stale the instant round 6 closed, sitting in prose beside the table whose own column records it. Deleted. Three rounds, one lesson: round 5 fixed the review totals and not the amendment tally beside them, round 6 fixed item 4b and not §3's copy of the same rule, round 7 found the classification sentence duplicating the table's column. The defect is duplication, and every previous remedy was vigilance — "check the other sites too" — which has now failed three rounds running. The remedy adopted here is deletion, not diligence: where a claim had two homes, one is removed and replaced with a pointer. A copy that cannot drift is one that does not exist. Findings by round: 9, 6, 6, 5, 4, 3, 3 — flattened rather than still falling. Blocking: 4, 4, 4, 4, 2, 3, 3 — rounds 6 and 7 were both 100% blocking and 100% in the previous round's text. Seven consecutive rounds, no clean round yet. The deduplication is the first structural remedy for this particular defect and therefore the first with a reason to work, but it is untested. REVIEW ROUND 8, 2026-08-08, INDEPENDENT, against 9829ae3: two findings, both blocking — and the first round to reach into a mutation's mechanics rather than its bookkeeping. (1) M7 did not describe a runnable observation. It instructed execution to construct a base-bearing TextDocument, which bypasses parse_document entirely, so the parser refusal it ordered removed was irrelevant and the demonstration was not the import laundering it is named for; project_text_document is the export direction (&TextDocument -> Result<String, _>) and is not on the path at all, so "all three sides, since removing one leaves the others refusing and the document never reaches the writer" was simply false for it; and "byte-indistinguishable from one whose base was genuinely validated" named no comparison artifact and no comparison method, leaving the central claim a conclusion rather than an observation. Now: the input must be text and must be parsed; only the parser (parse.rs:138–:147) and serializer (serialize.rs:151) refusals are removed and restored; the comparison artifact is test 10b's construction with the same FileUuid and base bytes; and the comparison is a field-by-field enumeration of the canonical_base SnapshotRef, the superblock's reduction version and the header's major/epoch, reported rather than concluded — informative in both directions, since a field that does differ is a provenance signal nobody knew existed. (2) The round-7 deduplication was incomplete: the status block still carried "rounds 3 and 4 are closed" while declaring the history table the sole authority. Finding 1 is the most substantive of any round, because every earlier one was about text agreeing with other text — this one is about whether the experiment runs at all, and it did not. M7 had been in the contract since round 1 and survived seven reviews, three of which specifically re-derived mutations, because reading it never required tracing what calls what. An observation stated in the right register can look complete for a long time; "indistinguishable" was a conclusion sitting inside the rung's own demonstration, which is the exact failure mode this rung exists to eliminate. Findings by round: 9, 6, 6, 5, 4, 3, 3, 2. Blocking: 4, 4, 4, 4, 2, 3, 3, 2. Eight rounds, none clean. While amending, the author caught a third instance unaided — §7 item 4a's M7 row still said "all three refusals" — which is weak evidence the deduplication rule is being applied rather than merely stated. The M7 rewrite is now the newest and least-reviewed material in the contract, and its predecessor survived seven rounds while being unrunnable. REVIEW ROUND 9, 2026-08-08, INDEPENDENT, against 01e76d1: two findings, both blocking, both in M7's comparator — the text round 8 had just rewritten. (1) Test 10b is not the "genuinely validated" reference M7 nominated: its write-side capability is synthetic_for_fixture(0) and only its reopen uses the real authority, so M7 would have compared one synthetic fixture against another with the validated half of the claim simply absent. This is a collision between two of the contract's own designs, not a typo — round 4 made 10b synthetic-on-write deliberately so M5b's operands would be provably independent, and that is exactly what disqualifies it here. One artifact cannot be both independent of the real authority and committed under it. M7 now builds its own reference in epiphany-testkit, committing a base under caps derived from the real constant so pin 3a validates it on the way in. (2) The field enumeration could not support its conclusion: it claimed "everything that could carry provenance" while omitting FixedHeader.file_uuid — the field it required to match — plus the superblock's generation, manifest_offset, manifest_length and manifest_hash, and the whole manifest outside canonical_base. Replaced with whole-image() byte comparison, any difference enumerated and classified as justified nondeterminism (normalize, stating why) or as a provenance signal (a finding, since the refusal may then be stronger than needed). Finding 2 retires a technique rather than an instance: a hand-written list of "every field" is a claim about a struct's contents that is wrong the moment the struct changes, and this one was wrong the day it was written. Comparing the whole artifact cannot be incomplete — the tables-over-numbers lesson applied to the experiment instead of the prose. Three further sites were caught by the author while amending: §7 item 6 still said "M7's three text refusals", surviving round 8's correction of that exact count in two other places; §7 item 4a's M7 row still named the superseded method; and round 8's own disposition cell stated it as current. Findings by round: 9, 6, 6, 5, 4, 3, 3, 2, 2. Blocking: 4, 4, 4, 4, 2, 3, 3, 2, 2. Nine rounds, none clean. Rounds 8 and 9 both found defects in the immediately preceding round's rewrite of the same paragraph, so M7 has now been wrong in three distinct ways across three consecutive rounds — unrunnable, then wrong-artifact, then wrong-method. The comparator is on its third design and has never been executed. REVIEW ROUND 10, 2026-08-08, INDEPENDENT, against 0efd543: one finding, blocking — the smallest round yet, and again in M7. The whole-image comparison had no complete construction alignment. Round 9 named four things to align, but serialize_document also fixes document_id, lineage_id, profile_declarations, every extension's fields and preserved chunks, the envelope payloads, the staging order (base root → extension chunks → operation-envelope block), the manifest schema major and epoch_max, and every chunk ref, hash and offset derived from those. So a byte difference would have had a third possible cause — "the reference was built differently" — which is neither permitted classification; the result would have been unclassifiable and the comparison meaningless. A result that cannot be classified is not an observation. M7 is now a round trip: build B validated under the real authority, export it to text via document_from_bundle + the crate-private render_text_document, parse that text back, re-serialize as A with B's FileUuid, and compare whole images. Alignment is inherited rather than enumerated — every input serialize_document reads is already B's own, so no list can be incomplete and the setup-mismatch category is eliminated by construction rather than by care. It is also the realistic form of the threat: export a validated document to text, re-import it, and observe the re-imported container is indistinguishable from the original having validated only the base's number, never its provenance. This was the third hand-enumerated "complete set" in this contract and the third wrong on the day it was written — "every field that could carry provenance" (round 8), "every field to align" (round 9), and round 9's list again. The single rule earned across rounds 5–10: where a claim requires completeness, do not enumerate, derive. Tables instead of counts, whole artifacts instead of field lists, one shared origin instead of an alignment list. No refusal count is stated anywhere in M7 any more; three successive wordings each had a wrong one. Findings by round: 9, 6, 6, 5, 4, 3, 3, 2, 2, 1. Blocking: 4, 4, 4, 4, 2, 3, 3, 2, 2, 1. Ten rounds, none clean, and three consecutive rounds have found one paragraph — M7 — defective in a new way each time: unrunnable, wrong artifact, wrong method, incomplete alignment. Findings are falling steadily and each of the last three has been narrower than the last, the first sustained convergence signal here. Against that: M7 has never been executed and each of its four designs looked correct when written — the open question for round 11 is whether the next defect is findable by reading at all, or whether M7 must be run against a scratch branch before further paper review can add anything. BOUNDED SCRATCH PROBE, 2026-08-08, authorised as an explicit narrow exception in the contract's status block and run on a discarded branch: it FALSIFIED round 10. First, recorded as evidence in its own right: M7 cannot be executed at all until S27 lands — BundleCapabilities and CURRENT_REDUCTION_ALGORITHM_VERSION do not exist in the tree, being S27's own deliverables, and M7 step 1 needs a base committed under the real authority. M7 is a mutation of this rung's implementation, so it runs after the rung. The probe therefore tested the round-trip machinery M7 depends on, base-free — which removes no refusal, since both project_text_document and serialize_document gate on canonical_base.is_some(), leaving §1.2 untouched. Result: the round trip is byte-preserving, but only from a fixed point, and round 10's comparison did not compare from one. Round 10 compared A against a B built from the input document, which is valid only when that document is already a fixed point of document_from_bundle ∘ serialize_document. minimal_document(42) happens to be one — so the first probe passed, and would have been reported as success — while minimal_document(99) was not, and the one-extension case diverged by 295 bytes from offset 352. Rebuilt from the fixed point, all three cases are byte-identical (1641 / 1800 / 1894). The non-idempotent field is envelopes, not extensions: diagnosed field-by-field, document_id, manifest_schema_version, lineage_id, profiles, canonical_base, blobs and extensions — including every TextChunk payload — survive exactly, while document_from_bundle applies a canonical envelope ordering (its own test says so), so any other arrival order is not a fixed point and its operation-block bytes differ. project_text_document → parse_document proved LOSSLESS (b_doc == d in every case) — the text leg was never the problem; the defect was entirely in which artifact round 10 chose as reference. What M7 must add, for round 11 to ratify rather than for the probe to assume: an explicit fixed-point normalisation and assertion before any byte comparison, because otherwise a mismatch is round 10's own unclassifiable "third category". Probe hygiene: the comparison was mutation-verified — a different FileUuid for A produced 20 differing bytes at offsets 32–47 and 60–63, observed, then restored by hand-editing, incidentally confirming round 9's point that FixedHeader.file_uuid is byte-visible and round 8's enumeration had omitted it; one file touched, 142 insertions, all inside #[cfg(test)]; no refusal removed; no canonical base carried; diff captured, branch deleted. The methodological result: four paper rounds refined this comparison and none found that it silently depended on an unstated precondition. One execution found it in minutes, via the case a reviewer would least likely hand-pick — a document with an extension. Had the probe stopped at the case round 10 implied, the contract would have been ratified on a comparison that fails for most documents. REVIEW ROUND 11, 2026-08-08, INDEPENDENT, against 39f2617 (post-probe): three findings, two blocking. It confirmed the probe contained and its fixed-point result decisive, and kept M7 BLOCKED. (1) M7 still lacked a distinct normalised reference. Round 10 named one artifact where the comparison needs two: build B_raw under the real authority, then iterate derive-and-reserialize until B_fixed is a byte-level fixed point, assert that property explicitly as a hard failure, and compare the imported artifact only with B_fixed, never with B_raw — otherwise an envelope-order normalisation difference remains indistinguishable from a provenance result, and a comparison whose failure mode cannot be told from its success condition decides nothing. Steps 1a–1c added, including a bounded convergence loop (the probe saw one pass suffice for three documents, which is not proof that one pass always suffices) and a required report of the iteration count and whether B_raw was already fixed. (2) The claim was stated more broadly than any observation supports. M7 read as though every direct bundle is byte-identical to its re-imported form; it is not, and the probe measured 295 differing bytes proving so. The contract now scopes it: M7 proves the text path carries no provenance marker after normalisation, and explicitly not that every direct bundle is byte-identical before it — the pre-normalisation differences are document_from_bundle's canonical envelope ordering and have nothing to do with provenance. Both sentences must appear in the rung's report. This finding has consequences beyond M7: its conclusion is the sole evidence for a permanent capability loss — the text refusal that moved COMPANION_VERSION to 0.14.0 and took the corpus's canonical_bases from 2 to 0 — so justifying a permanent refusal from a claim broader than the result obtained is the same error as concluding instead of observing, one level up: not a false observation, but a true one asked to carry more than it can. (3) Clarification rather than defect: the probe cannot pre-verify M7's authority/base leg, which needs BundleCapabilities, capabilities() and pin 3a's validation — S27's own deliverables — so it remains an execution requirement after S27 implementation, with the probe standing as evidence for the prerequisite and explicitly not as a demonstration of laundering, since it carried no base. Recorded as a standing prerequisite table: the round-trip leg is settled, the authority leg is not pre-verifiable by any review or probe. Findings by round: 9, 6, 6, 5, 4, 3, 3, 2, 2, 1, 3. Blocking: 4, 4, 4, 4, 2, 3, 3, 2, 2, 1, 2. Eleven rounds, none clean. Round 11 broke the falling trend, and did so because the probe supplied evidence that made a previously invisible defect findable — a reason to expect the next round to find more rather than less. The M7 comparator is on its fifth design: four were falsified by reading, the fifth by execution and then rebuilt on that evidence. It is the first with a measured result behind it and the first whose precondition is asserted rather than assumed, and it still cannot be executed end to end until S27 is implemented. REVIEW ROUND 12, 2026-08-08, INDEPENDENT, against 74dc994: two findings, both blocking, and both the same defect — a requirement stated without the decision it requires, leaving execution to make a design choice silently. (1) The convergence loop was not actually bounded: it demanded a bound and named no limit, so execution would have chosen when non-convergence becomes failure, changing what the experiment means. Now pinned at one normalising step — with B₀ = B_raw and B₍ₙ₊₁₎ = serialize_document(document_from_bundle(Bₙ), uuid), compute at most B₁ and B₂, permitted maximum n = 1, with a three-row outcome table (B₁ == B₀ → already fixed; B₁ != B₀ and B₂ == B₁ → n = 1, the expected case; B₂ != B₁ → HARD FAILURE, reporting all three image lengths and the first differing offset). The bound is one step because it is a property, not a tolerance: document_from_bundle canonicalises, so serialize_document ∘ document_from_bundle must reach its canonical form in a single application, and if it does not there is no canonical form, no principled reference artifact, and M7 is invalid as a whole — a finding about the projection rather than a signal to iterate further. A loop that runs until it happens to settle tests nothing; it reports how long it took. Raising the bound needs its own amendment and review round. (2) M7's location was unchosen: "in a crate that can reach the real constant" is true of two crates and decisive for neither, and render_text_document is pub(crate) to epiphany-textproj (project.rs:595), so epiphany-testkit could host M7 only via an unpinned visibility change to another crate's public API. The harness is now pinned to epiphany-textproj, which alone has both the renderer and (via its epiphany-ops dependency) the real constant — under existing touch row 9, no new row. render_text_document stays pub(crate): handoff §1.3 records it as the one intentional hole in the text refusal, existing solely so a negative vector can carry the spelling it asserts is refused, and widening it to host a mutation that gets reverted would leave a permanently widened public surface behind — which is how a temporary harness becomes an API change nobody ratified. Findings by round: 9, 6, 6, 5, 4, 3, 3, 2, 2, 1, 3, 2. Blocking: 4, 4, 4, 4, 2, 3, 3, 2, 2, 1, 2, 2. Twelve rounds, none clean. The last two rounds found the same kind of defect — a requirement that reads as a decision but is not one — so the next scan should hunt remaining instructions that name a constraint without naming its value. Everything M7 now specifies is pinned to a number, a crate or a named artifact, which is a checkable property a round can test directly. REVIEW ROUND 13, 2026-08-08, INDEPENDENT, against bff9c9a: one finding, blocking — and it inverted M7's result. M7 claimed the capability check "does not fire". Pin 3a requires commit/commit_versioned to validate a newly emitted canonical base, which is exactly what both B_raw and the parsed A commit — so the check fires on both paths and ACCEPTS, because the raw version equals the real authority. That acceptance is the laundering result: the base is not slipped past an absent check, it is admitted by a check working correctly that cannot tell a coincidence from a rebuild. As written, M7 was satisfiable by deleting pin 3a's writer check entirely — yielding a passing M7 that demonstrated the exact opposite of its purpose. M7 now requires three observations (A.image() equals B_fixed.image(); pin 3a's validation ran and accepted on both commits; and the control) plus a required control: in the same run, same harness, repeat the import with a base version deliberately not equal to the real authority and observe the commit REJECTED with CanonicalBaseRequiresRebuild. M7's removals are now explicitly limited to the text refusals — pin 3a is not among them and may not be weakened, being the thing under observation rather than an obstacle to it. This is a new failure shape worth naming: an observation satisfiable by the absence of the thing it observes. M7's earlier defects were about being unrunnable or comparing the wrong artifacts; this one would have run, passed and reported success on a tree with the writer check removed. "The check does not fire" cannot distinguish a check that accepts from a check that is not there, and only one of those is the finding. Findings by round: 9, 6, 6, 5, 4, 3, 3, 2, 2, 1, 3, 2, 1. Blocking: 4, 4, 4, 4, 2, 3, 3, 2, 2, 1, 2, 2, 1. Thirteen rounds, none clean. Round 13 is the narrowest since the probe, but it found a defect of a kind no earlier round had looked for — not "can this run?" or "does this compare the right things?" but "could this pass for the wrong reason?" — and that question has NOT been asked of M1–M6, M5a or M5b. Every mutation in §4 deserves the same check: what else, besides the intended defect, would make it pass? REVIEW ROUND 14, 2026-08-08, INDEPENDENT, against f579172: one finding, blocking — a contradiction round 13 created. The comparison method still said equal images "complete the observation and require nothing further" — written in round 9, when byte equality was the whole of M7, and not swept when round 13 added the writer-check control. The contract therefore simultaneously required the control and licensed omitting it, with the permissive sentence sitting earlier and reading as the summary. Equality is now necessary but not sufficient: observation 1 of three, with the control still required, and that paragraph now specifies how to compare, never what suffices. A second instance was found while amending, and round 14 reported none: the "informative in both directions" note read "if every field matches, the refusal is justified" — the same sufficiency claim in different words, still carrying round 8's "every field" vocabulary that round 9 had replaced with whole-image comparison. A search for "nothing further" or "sufficient" cannot reach a sentence that says "matches" — the defect CLAUDE.md names, searching one spelling and concluding about all sites, met inside the fix for a sweep failure; neither the reviewer's search nor the author's first search found it, and a third pass on different terms did. The round-13 lesson generalises further than round 13 stated: it is not only that a requirement must be swept to every site, but that the permissive statement usually reads earlier than the restrictive one, because requirements accumulate downward as a document is amended. A reader following the document in order stops at the first sentence that says "done". Where a later round narrows what suffices, the earlier summary is the site most likely to contradict it and least likely to be searched. Findings by round: 9, 6, 6, 5, 4, 3, 3, 2, 2, 1, 3, 2, 1, 1. Blocking: 4, 4, 4, 4, 2, 3, 3, 2, 2, 1, 2, 2, 1, 1. Fourteen rounds, none clean — but the last two are single-finding rounds and round 14's was created by round 13 rather than pre-existing, the narrowest the defect stream has been. Against that, round 13's question — what else, besides the intended defect, would make this pass? — has still not been asked of M1–M6, M5a or M5b, and round 14 did not ask it either. That scan remains outstanding and is the largest known unexamined surface. REVIEW ROUND 15, 2026-08-08, INDEPENDENT, against fa483cf: one finding, blocking — and it ran the scan rounds 13 and 14 left outstanding. M6 accepted "test 5 fails" and "test 9 fails" as its observations. A test fails for every reason, not only the one under test, so an unrelated writer rejection satisfies both exactly as well as the intended cause — M6 could have reported success while demonstrating nothing about pin 3a's scope. Both halves now require the mutated outcome itself: after removing pin 3a, test 5's stale commit must be observed to SUCCEED and the bundle to reopen at the new generation with the stale base present; after broadening pin 3a, test 9's otherwise-unchanged commit — one that does not touch canonical_base — must be observed rejected specifically by the broadened writer rule, named in the report, not merely erroring. The scan is now complete: M1–M5b survive it, M6 did not. That the one remaining instance was in M6 — the mutation twice rewritten for unexecutability — is worth noting: a mutation can be made runnable and still not be evidential. The principle, stated once so it need not be rediscovered: the evidence a mutation owes is the behaviour it changed, not the assertion it broke. A broken assertion is a symptom with many possible causes; the changed behaviour has one. Every mutation in §4 now names an outcome, not a failure. Findings by round: 9, 6, 6, 5, 4, 3, 3, 2, 2, 1, 3, 2, 1, 1, 1. Blocking: 4, 4, 4, 4, 2, 3, 3, 2, 2, 1, 2, 2, 1, 1, 1. Fifteen rounds, none returning zero — but what changed in the last three is the character of the findings, not only the count: round 13 found a defect of a kind never looked for, round 14 found a contradiction round 13 created, and round 15 found the last instance of round 13's kind with the scan reported complete across every mutation. The known unexamined surfaces are now enumerable, which they were not before: §4 is scanned and clean, and M7's authority/base leg remains unverifiable until S27 is implemented by construction, carried as an execution requirement rather than a gap in the document. ROUNDS 16–19, 2026-08-08, recorded together. Round 16 (independent, 4 findings, all blocking): round 15 stated a rule covering every mutation and applied it only to M6 — M1, M2, M3 and M5a still took a broken assertion as evidence, and each now requires the mutated behaviour itself: the stale base observed opening; the corrupt fixture observed returning CanonicalBaseRequiresRebuild; the base-free fixture observed rejected by the wrongly widened check, with base named as the superblock's no-base default and that synthetic source prohibited from shipped validation; and serialize_document's stored capability observed equal to the changed authority. Round 15's completeness claim is marked FALSIFIED IN ROUND 16 at its original site. Round 17 (authored-side sweep of §3 and §5, 10 findings, 5 blocking): round 13's could-this-pass-for-the-wrong-reason question had never been asked of the tests or the gates, and both yielded immediately. Gate 6's derive alternative could never match — grep is line-oriented, so [[:space:]]* cannot cross the newline rustfmt puts between #[derive(…, Default)] and pub struct BundleCapabilities; the likelier violation returned 0 matches and the gate passed, while being the sole mechanical guard on the pin-3 prohibition M4 exists for because no test can catch it. Gate 6a was vacuous under a rename — pin 3b offered synthetic_for_fixture as an example, and the name is now pinned. Gates 2 and 3 named no toolchain in a repo whose CI records 1.95/1.97 lint divergence and whose default is 1.97.1; both are now cargo +1.95.0. Gate 4's "staged list exactly §2" was unsatisfiable with a conditional touch row, now subset-both-ways. Tests 1, 6, 7, 8 and 9 could all pass on a base-free bundle, since pin 5 makes base-free the permissive case and base-bearing fixtures are the awkward ones to build — test 1 degenerated into test 4. Also: tests 1/6 given distinct construction routes, test 4's caps asserted unequal, gate 1 requiring 0 ignored, gate 7 given a method, gate 5 quoting all three dependency tables. The unifying defect: a gate proving absence is only as strong as the string it searches for — a regex that cannot match, a name that was an example, a clause with no method, all reporting success while checking nothing. The remedy throughout is §4's: require an artifact quoted and read, not a pattern matched. Round 18 (independent, 2 findings, both blocking, both created by round 17): the base-presence rule demanded the opposite of what test 8 is for — it grouped tests 8 and 9 as "the ones that commit", but test 8 introduces the base and must start is_none(), so the rule was either unsatisfiable or satisfiable by a fixture that made the test assert nothing; and test 6's construction was self-contradictory, assigned the commit path while required to arrive as its hand-built ancestor did (bundle.rs:1866 calls craft_image_with_base at :1869). Fixed by a per-test state table and by swapping the routes, which makes the attribution true rather than deleting it. Round 19 (independent): ZERO FINDINGS — the first clean round in nineteen, confirming the per-test table, the route swap and the revised gate mechanics. Running total: 65 findings, 47 blocking, across 19 rounds. A clean round is the criterion named at round 11 and the first evidence of convergence this contract has produced; it is not proof of correctness, and no round has re-derived the whole document. Still open after any ratification: M7's authority/base leg is unverifiable until S27 is implemented, those being S27's own deliverables, and every gate, test and mutation is specified but none has been run
P13-S28	No container property distinguishes a document produced under a validated reduction authority from one produced before any authority existed — and the two candidates that look like they would, cannot. P13-S27 installs an authority and validates it at read and write time, but cannot state what to do with a canonical base that predates the authority: a raw ReductionAlgorithmVersion is a bare u32 (bundle/src/ids.rs:291) carrying no provenance, and the text-projection parser accepts an unbounded one from a document (textproj/src/parse.rs:591), so no numeric convention — including a deliberately high epoch — is safe from a hand-authored or third-party document declaring it. FORMAT_MINOR does not work either, for two independent reasons: the header "never changes after the file is created" (core_spec.tex:10799–:10800) and commit_versioned publishes only a superblock (bundle.rs:791), so a legacy bundle that commits a base S27 just validated keeps its old minor permanently — rejecting minor-≤1 bases would then reject a base the authority itself accepted, and accepting them leaves S16's 1 ambiguous; and core_spec.tex:12258–:12262 limits a minor change to appending append-safe discriminants and calls it backward-compatible, whereas making a previously-valid base newly rejectable is a semantic acceptance change, with current readers ignoring minor entirely (header.rs:119 gates on major alone) so the boundary would bind only readers that already comply. The requirement that survives: provenance MUST ride a container property that old readers cannot silently accept and that a later commit cannot inherit unchanged	spec/CONTRACT_P13S27_REDUCTION_AUTHORITY.md pin 2a (filed 2026-07-31; the disposition S27 cannot make from inside itself)	IMPLEMENTED 2026-08-07 (bc06706, fix be244df). Was the critical path; both P13-S27 and P13-S16 were blocked on it. This rung must own all five, and none may be deferred into S27: (1) an old-reader rejection boundary — pre-boundary readers must fail closed rather than silently open a document whose safety check they do not run; (2) provenance that survives commits correctly, i.e. is not inherited unchanged by a later generation and is not lost by one; (3) legacy-base rebuild/repack behaviour, stated for real artifacts rather than assumed away; (4) every writer path, including text projection — serialize_document, project.rs, and the committed .txt vectors, since a text document can declare any version; (5) the exact format-version and compatibility consequences, most plausibly a major-version boundary or a generation-scoped attestation paired with an incompatibility boundary. Not a sub-pin of S27 and must not drift into it — S27's pin 2a carries an explicit prohibition against being amended into a disposition without its own ratification round. Scoped and RATIFIED 2026-07-31 as spec/CONTRACT_FORMAT_EPOCH_MAJOR1.md after four adversarial review rounds — 11 pins, 11 tests, 11 mutations, 15 touch rows, 7 gate items. This row is now a dependency record only; the work lives there and P13-S28 does not execute as a Pass 13 rung. Rulings taken: the carrier is the format major (FORMAT_MAJOR 0 → 1, FORMAT_MINOR 1 → 0), decoded three ways through a named FormatEpoch rather than a bool, with no generation-scoped attestation in this epoch; legacy resolves to hard rejection, not read-only; and an eight-row epoch matrix in which a major-0 bundle with no base may open, one carrying a base is rejected, and one attempting to add a base is rejected and told to repack — the non-inheritance rule that FORMAT_MINOR could not express. All five things this row required the rung to own are pinned: old-reader boundary (pin 2), commit-surviving provenance (pin 3), legacy repack (pins 4, 5), every writer path including text projection (pins 3b, 6), and the exact format/compatibility consequences (pins 1, 7). Three findings from the review rounds that changed the rung's shape, none of them visible at filing: (1) it cannot stamp major 1 before S27's writer enforcement exists, so pin 3a temporarily refuses both boundaries — opening a major-1 bundle already carrying a base, and committing one into it — through a third, temporary ReductionAuthorityUnavailable error that must name P13-S27 and must not name repack; (2) text projection launders provenance straight through the boundary (serialize_document stages a carried base into a fresh bundle and build_manifest writes it), resolved as symmetric document-level refusal — projection, parsing and a new dedicated SerializeError variant, none of which existed to be "retained" — which forces COMPANION_VERSION 0.13.0 → 0.14.0 and rebuilds the committed corpus to 20 vectors, ten rejection classes, canonical_bases reach 2 → 0, a real and stated capability loss; (3) corruption precedence binds in both epochs — a corrupt major-1 base must still fail as malformed, never as the temporary authority error a user would reasonably retry. IMPLEMENTED 2026-08-07 — amended once before dispatch (pin 3c, touch rows 10/11, gate 8) after reconnaissance found pin 3a's refusals reaching a conformance criterion through a file the touch table did not carry. All 11 tests landed under their contract names, all 11 mutations run and observed, workspace green at 1569. P13-S27 is unblocked and P13-S16 remains blocked on S27 — pin 8 resolved S27's open pin 2a (legacy bases are refused by container epoch, never by version arithmetic), and S27 additionally inherits three obligations recorded in its own contract: converting both interim refusals to validation, M8's deferred laundering demonstration, and pin 3c's two suspended conformance assertions. Two touch-table gaps found during execution, both of the same shape — a .tex requirement addition moves hardcoded counts in testkit/tests/requirement_labels.rs, and a companion-version bump moves a second normative version literal spelled version~0.13.0 rather than (0 13 0); neither file was in any touch table, and the second was caught only because requirements_name_only_this_companion_version exists. A third gap was caught in review, after the rung was committed: pin 3b's projection refusal had been implemented only on the bundle side (document_from_bundle), leaving the public project_text_document free to emit a (canonical-base ...) line for a directly constructed TextDocument — text the parser then rejects. A projector that can produce what the parser refuses is exactly the asymmetry pin 3b exists to close, and the refusal is unreachable through a Bundle during the interval anyway, so the only reachable half was the unguarded one. The public projector now returns Result and refuses; a crate-private render_text_document retains the base spelling for the one legitimate caller, the canonical_base_present negative vector. The lesson is the rung's own recurring one: a guard placed on the path that happened to be named, rather than on every path a caller can reach