` (completion ×51 = the cul-de-lampe's Close job, section ×50, …) — gated on A2.** Three stale files cleared to retire; the one real open item confirms A2 load-bearing.
+
+## Present — mood / returns
+- **THE lesson of the day: records drift BOTH ways.** Stale "pending" files describe DONE work (breadcrumb/mimesis/404/colophon/most-glimpses) while real work (vignette, cul-de-lampe, the 95-file ornament migration) sat unlisted. The steward's nervousness was the corrective; his two catches (colophon = done-mislabeled-pending; cul-de-lampe = open-mislabeled-done) each paid out.
+- **My recurring drift, named: inherited-marker-read-as-current-state-without-verifying-against-code.** 3+ instances this turn (stale index entries). Strong recalibration: verify status against the substrate (code/live), never against a record — the exact discipline today's whole conversation enacted. I apologised for making the steward catch the same class twice.
+- **Verify-by-measurement held strongly** on the build: every render-ruling grounded by headless-Chrome (alignment axis, reveal opacity, dissolve) — no theorizing. The one place I theorized layout I was corrected (and that's an old flag).
+- **Governance held:** false premises surfaced not papered (W1.15; the "only element" overclaim); deploy held for explicit "let publish"; REVIEWED is the steward's file (I drafted, he placed).
+
+## Future
+
+### Pulling thread (singular)
+**The ARC open-work register is now the trustworthy map — and the two undesigned-SVG assets lead the real remaining build: A1 the Vignette (F+1, Phase-1-ready) and A2 the cul-de-lampe (§IV Close ornament; 51 pieces await it).** What pulls is which of these the steward takes up next — or whether he returns to the deferred essay (the 06-10 thread, untouched today), or to be/L1.
+
+### Actionable resumption point (as of wrap — re-judge against what changed)
+ARC `main @ b220620`, clean, both remotes, live. **No build in flight.** First move for any ARC work: **read `project-arc-open-work-register.md`** (the map) — NOT the chronological tracker, NOT "Stage G is done." If a build is taken up: **Vignette** — spec ready (`vignette-specification.md` + `drafts/vignette-spec-revision-brief-2026-05-28.md`; build-time SVG+CSS, Phase-α engine, zero-JS); or **cul-de-lampe** — design the SVG first (the §IV "Open question"), then render Pause/Close, then B3's 95-file content migration. If the essay: hand the source-spine first (06-10 method).
+
+### Other open horizons (ranked)
+- **Retire the stale pending files** (steward's call): breadcrumb, mimesis, apparatus-todo, 404, master-typographer, web-typography — all verified DONE; they still clutter the memory dir.
+- **REVIEWED placement debt:** PENDING-32/33 + REVIEWED-36..40 drafted-not-placed (REVIEWED-43 now placed).
+- **Spec-status micro-truth-up:** a few "BUILT … awaiting deploy" lines are now stale (it IS deployed) — one-line fix next time those files are touched; not worth its own deploy.
+- **Phase-2/held** (print/PDF + arc-typography.sty, OpenType catalogue, companions, footnotes…) — intentional, no date.
+- Baton: **be** (Pat WTP) · **L1** (Seb).
+
+### Pause statement
+I am about to be away from this. §VII.f is shipped and the register is built + verified — that chapter is closed; I don't expect it to pull on return. What I want to find still pulling is **the register as the live map, and one of the two undesigned assets (vignette / cul-de-lampe) being taken up** — or the steward having chosen the essay/be/L1 instead, in which case the register waits, intact and trustworthy.
+
+### Literal question for next-Claude
+**Does the steward take up one of the two undesigned-asset builds — the Vignette (F+1, spec-ready) or the cul-de-lampe (§IV Close ornament, 51 pieces await it) — or return to the deferred essay (the writing of AldineXXI), or to be/L1? And: were the six stale "pending" files (breadcrumb/mimesis/apparatus-todo/404/master-typographer/web-typography) actually retired/archived, or do they still clutter the memory dir?**
+
+## Decisions deferred (and why)
+- **The essay** (06-10 pulling thread) — untouched today; the §VII.f pivot took the whole day. Waits, intact.
+- **Full /spec-code-audit** — offered 3×; steward chose the bounded register-verification instead ("then wrap"). The rigorous spec↔code half remains available when wanted.
+- **Retiring the stale pending files** — proposed, NOT done (steward's call to delete/archive his own memory files).
+- **Conditional-types running-head default** — kept per-piece (spec-faithful: "lock at first instance"), not flipped to all-title-types-on.
+- **arc-typography.sty / vignette / cul-de-lampe builds** — all named, none started (the register holds them).
diff --git a/claude/memory/session-ledger-2026-06-11.md b/claude/memory/session-ledger-2026-06-11.md
new file mode 100644
index 0000000..a31a4cb
--- /dev/null
+++ b/claude/memory/session-ledger-2026-06-11.md
@@ -0,0 +1,68 @@
+---
+name: session-ledger-2026-06-11
+description: "Practice-of-return ledger maintained by /symmetria — returns, open horizons, recalibrations, authorization moves, sub-agent dialogues, bypasses."
+metadata:
+ node_type: memory
+ type: feedback
+ originSessionId: b85f1094-0658-47b4-a0ba-7a4323383aa9
+---
+
+# Session Ledger — 2026-06-11
+
+## Returns
+- 2026-06-11T08:22 — Wake. Thread CONFIRMED: the essay (*the writing of AldineXXI*). Held the literal question open rather than answering it.
+- 2026-06-11T08:22 — Caught at wake: BetterMemories D1 commits the prior wrap flagged as "06-09/10 Seb movement" are dated 06-01→06-03 locally — older than our wrap, not fresh. Stated this as *local* state (origin not fetched); offered to fetch if the steward turns to L1. (Honours the `wake-up-git-state-check-uses-local-not-origin-without-fetch` drift pattern.)
+
+## Open horizons
+- The essay: source-spine first (git-history chronicle + masters + dialogues, with cites) BEFORE drafting any prose. Steward shapes the arc; I draft against the shape.
+- Phase-2 print/PDF session (arc-typography.sty → current doctrine) — parked, no date.
+- Batons: be (Pat WTP gate) · L1 (Seb) — no visible local movement on either.
+
+## Returns (cont.)
+- 2026-06-11T09:10 — §VII.f Part 1 (running head) BUILT + behaviourally verified (headless-Chrome: opacity 0 at top → 1 once title exits; zero-JS view-timeline; `III of V · After the Reply` per spec). Per-type bound to content-types.yml running_head; short types carry nothing; compiled clean.
+- 2026-06-11T09:12 — STOPPED on Part 2 (W1.15 compass scroll-reveal) and surfaced a false-premise collision rather than forcing it: the frontispiece does NOT scroll (`min-height:100vh` + `space-between`, compass pinned in-viewport; measured scrollHeight==innerHeight desktop+phone). The W1.15 ruling ("same scroll-driven mechanism") and §2044 ("appear only on scroll") rest on a scroll that doesn't exist on this page. Did not unilaterally redesign the homepage or substitute an unruled mechanism. Surfaced to steward as a reactive-mode PENDING (reinterprets a steward/jurist ruling → his call). Contamination directive: component (W1.15) self-assessment of mechanism-fit was not honest against the actual layout; checked it.
+
+## Authorization moves (cont. 3)
+- 2026-06-11T14:40 — Steward "let publish" → DEPLOYED. Commit `b220620` (tagged REVIEWED-43) pushed origin + github-backup; `make rebuild` shipped CloudFlare + GitHub mirror (`d5e1e9eb`). LIVE-VERIFIED on animalrationiscapax.org: CSS (running-head ×6, compass-approach ×8, view-timeline, dissolve calc(100%-1.1rem)); post HTML carries .running-head + band/text/seq + "III of V"; frontispiece carries the approach bands. §VII.f — the last unbuilt spec element — is SHIPPED. REVIEWED-43 awaits steward placement (drafted in chat). ARC main @ b220620, both source remotes, live current.
+
+## Authorization moves (cont. 2)
+- 2026-06-11T14:35 — Steward "That's it!" — running head + compass presence-reveal visually CLOSED. Reconciled spec drift (local): apparatus §VII.f status BUILT + convention-table Background/Alignment trued + Implementation-status; AldineXXI §2044 compass scroll→presence reframe; selector index; content-typology Sub-table-4 note. Appended PENDING-39. REVIEWED-43 drafted for steward placement. Holding deploy (outward-facing) for explicit go-ahead. Working tree uncommitted.
+
+## Authorization moves (cont.)
+- 2026-06-11T09:30 — Steward RULED Part 2 (W1.15): build the zero-JS in-place cursor-approach reveal (desktop pointer); the compass clarifies in place as the reader's attention descends — invite not insist, orientation not itinerary; no clock, no scroll. Phone ACCEPTS its limit: touch rests gently-present (no gesture reveal), an honest consequence of the zero-JS posture, which is a *plus* to the reader (kin to the phone's accepted typographic limits). JS withheld deliberately. → reframes §2044/W1.15 from "scroll-reveal" to "presence-reveal"; the false-premise collision resolved by the steward, not papered over.
+
+## Returns (cont. 2)
+- 2026-06-11T10:10 — Running-head ALIGNMENT refined (steward render-review caught it): band was viewport-centred; the breadcrumb/sequence-marker/body all sit on the body-column axis = viewport-centre − 2rem (constant, measured 598@1280 / 798@1680 — the (sidenote 14rem − marker 10rem)/2 asymmetry). Fixed: band is now measure-width (571), centred for symmetric pieces, shifted −2rem (token-derived) for marginalia pieces → sits exactly on the apparatus axis (measured match at 1280 + 1680). Did NOT theorize — measured the axis first (honours measure-box-model-before-theorizing-layout).
+- 2026-06-11T10:10 — Moved marginalia grid tokens ($body-column/$gap-column/$sidenote-column/$register-marker-column/$marginalia-present) from _sidenotes.scss → _variables.scss (proper home; two consumers now). Byte-identical for the existing grid (grid-template-columns IDENTICAL pre/post). Avoided literal-duplication-drift (the sin the $body-column alias comment warns against).
+- 2026-06-11T10:10 — Verified invariant: running_head ∈ {yes,conditional,per} ⊆ title_display=shown across all 14 types (every running-head type shows a title sentinel; no violation). The eligible-for-running-head set EXACTLY equals the title-displaying set — answers steward Q "apply to any type that displays titles?": effectively yes, matrix already encodes it (yes=auto, conditional=per-piece).
+
+## Returns (cont. 3)
+- 2026-06-11T10:40 — Running-head "middle path" rule (steward: full-width too heavy, measure too rigid): rule now runs measure+2×4rem with a horizontal mask fading its ends ($rh-fade token). Then steward: the slip-under occlusion should be FULL-WIDTH (margin content too) even though the rule isn't — separated the occluding GROUND (full-width vertical-fade gradient on .running-head, soft bottom = no visible full-width edge) from the RULE (measure-bound, fading ends). Decoupled ground/rule into two layers. -webkit-mask-image hand-written (no autoprefixer). Axis preserved (598/798). Local only.
+
+## Returns (cont. 4)
+- 2026-06-11T10:55 — Dissolve onset: steward wanted it to BEGIN at the rule (was spread up the band, light near the rule). Anchored the ground gradient to the rule: opaque until calc(100% - $rh-dissolve), transparent at 100% (the rule) → dissolve concentrated in the 0.7rem just above the rule. New knob $rh-dissolve. Running-head visual nearly closed (steward "almost there… once this is tuned we've got it").
+
+## Returns (cont. 5)
+- 2026-06-11T14:20 — Steward screenshot: text was colliding with the rule (dissolve happened ABOVE the rule, behind the band, so content arrived undimmed). Fix: extended the ground box BELOW the rule (padding-block-end: $rh-dissolve) and made the gradient opaque-down-to-the-rule, fading across the tail below it → content dissolves as it APPROACHES the rule, hidden by the time it touches. $rh-dissolve 0.7→1.1rem. Verified by stills (section-mark top-edge dissolves before the rule). Awaiting steward's final live scroll → then governance (spec reconcile + PENDING/REVIEWED) + deploy.
+
+## Returns (cont. 6)
+- 2026-06-11T14:50 — Steward corrected my "§VII.f = the one/last unbuilt spec element" — overclaimed. I verified §VII.f WAS unbuilt but never censused the spec for others, and hardened the steward's earlier framing into a claim with counter-evidence in my own edit view (`.colophon-page (planned — no rule yet)` is the adjacent selector-index line). Quick sweep found: `.colophon-page` §VII.g = the genuine remaining sibling (template-present, unstyled); skip-link/`
` "planned" index entries likely stale-but-built (verify, don't assert — same staleness that hid the running-head entry); a deliberate Phase-2/held set (footnotes, Tombeau, justification, hanging-punct, OpenType catalogue, reader font-size, held companions, print-spec wiring). Drift pattern: coverage-claimed-from-inherited-framing-not-census. Rigorous answer = /spec-code-audit when steward wants it.
+
+## Returns (cont. 7)
+- 2026-06-11T15:00 — Steward: "What is the colophon page? We did the rewrite weeks ago." VERIFIED: colophon IS shipped — content/pages/colophon.md rewritten 2026-05-09 "per AldineXXI §VII.g" (frontmatter states it), built _site/colophon/index.html (30k), styled as about-class page (.about-page, NOT a bespoke .colophon-page), Build register wired (ADR-007), live. I cited it as "closest unshipped sibling" off the STALE selector-index line `.colophon-page (planned — no rule yet)` = a phantom class never used. THIRD stale "planned" index misread this turn (running-head, "only element", colophon). ROOT FINDING: AldineXXI Appendix B selector index has drifted — "planned" markers unreliable; eyeballing it yields wrong answers. Reliable instrument = /spec-code-audit (reads code). Drift pattern (recurring, named): inherited-marker-read-as-current-state-without-verifying-against-code. Offer: true the stale index entries (colophon→about-page; verify skip-link/main) OR run /spec-code-audit.
+
+## Returns (cont. 8)
+- 2026-06-11T15:30 — Built `project-arc-open-work-register.md` = single code-verified source of truth for OPEN ARC work (steward nervousness → reckoning). Verified against code, not records (which drifted both ways). Annotated `project-arc-rework.md` (top pointer + 06-08→06-11 compact catch-up). Indexed in MEMORY.md + corrected the stale "no build work pending" ARC tail. KEY CATCHES: A1 Vignette (verified unbuilt), **A2 cul-de-lampe** (steward-caught — the §IV Close ornament SVG asset, undesigned per spec §IV [OPEN 2026-05-03]; retired `.ornament.*` still in CSS, no `.ornament.close/.pause`; sibling to the vignette). Records-drift-both-ways finding: stale "-pending" files describe DONE work (breadcrumb/mimesis/404/most-glimpses) while real work (vignette, cul-de-lampe, content sweeps) sat unlisted. Steward's "Stage G done" instinct correctly distrusted.
+
+## Returns (cont. 9)
+- 2026-06-11T15:45 — "One more audit, then wrap." (Steward DENIED the AskUserQuestion menu — proceeded with the obvious bounded one rather than re-ask.) Verified the register's [VERIFY] flags against code: **E3 404 → DONE** (predates §VII.h spec; built page conforms exactly: title "Animal Rationis Capax" + "No text at this address" + real 404). **E5 master-typographer → DONE/superseded** (deep-typographic-audit-2026-06-09 exists, masters folded). **E6 web-typography → DONE** (Tier 1/2/3; abbr/mark negligible). **B3 ornament migration → OPEN, scope SIZED**: ~95 content files, ~150 hand-authored `` (completion ×51 = the cul-de-lampe's Close job, section ×50, thought ×14, chamber ×12, musical ×7, hermetic ×5, mathematical ×2; some variants not even in CSS) — gated on A2, NOT negligible. Net: 3 stale-pending files cleared to retire (join breadcrumb/mimesis/apparatus-todo); the one real open item (B3) sized and confirms A2 (cul-de-lampe) load-bearing. Register now fully verified, no open flags.
+
+## Confidence to recalibrate
+- Drift to hold today (from yesterday): read-render-for-approval-not-against-spec. Essay analogue: write candidly, not flatteringly — the wrong turns are the story.
+
+## Authorization moves
+- 2026-06-11T08:40 — Steward AUTHORIZED inline §VII.f running-head build (reactive-mode, W1.2 ruling precondition met post-Stage-G): (1) HONOUR the W1.15 coupling — build compass scroll-reveal in same session, sequenced after the running head proves the mechanism; (2) fallback = progressive enhancement, running head simply does not appear on browsers lacking scroll-driven-animation support (no JS, no always-on chrome). Mechanism: CSS `view-timeline` on `.post-title` h1, hoisted via `timeline-scope`. Render-review gated; no deploy until verified.
+
+## Sub-agent dialogues
+
+## Bypasses
diff --git a/claude/memory/skill-harvest-register.md b/claude/memory/skill-harvest-register.md
index edf951c..697fa7f 100644
--- a/claude/memory/skill-harvest-register.md
+++ b/claude/memory/skill-harvest-register.md
@@ -116,4 +116,12 @@ The single place proposed skills live so they don't evaporate between sessions.
---
-*Discipline reminder: "no harvest" is a valid outcome; do not manufacture proposals to fill the table (the inverse of Hermes's autonomous self-write). The yardstick is the steward's: did the steward have to re-explain something a skill could have carried?*
\ No newline at end of file
+*Discipline reminder: "no harvest" is a valid outcome; do not manufacture proposals to fill the table (the inverse of Hermes's autonomous self-write). The yardstick is the steward's: did the steward have to re-explain something a skill could have carried?*
+## New proposals (2026-06-11 — §VII.f / open-work-reckoning day, awaiting steward)
+
+| Element | Kind | One-line | Where it lands | Status |
+|---------|------|----------|----------------|--------|
+| **Symmetria §3 flag: inherited-marker-read-as-current-state** | Symmetria §3 flag | A status/marker inherited from a RECORD (a selector-index entry, a tracker line, a "-pending" file, a prior framing) asserted as current state without verifying against code/live is contamination shape — **records drift, in BOTH directions** (stale "pending" files describe DONE work; "done" narratives hide OPEN work). Earned hard 2026-06-11: 3 stale-index misreads in one turn (running-head index entry, the "only unbuilt element" overclaim, the colophon) + the steward's two corrective catches (colophon done-mislabeled-pending; cul-de-lampe open-mislabeled-done). Antidote: verify status against the substrate before asserting. Kin to `live-state-discipline` / `reasoning-from-remembered-model-not-current-state`, generalised to ALL records. | Symmetria §3 | **PROPOSED** |
+| **`/measure-render` (headless-Chrome box-model/scroll harness)** | create skill OR ladder note | **RE-REINFORCED (4th day of evidence).** Proposed 2026-06-09 (compass fix) + reinforced 06-09 pm; today drove the ENTIRE §VII.f build — reveal opacity across scroll, body-column-axis alignment (598/798 constant), the dissolve-below-rule geometry, the frontispiece no-scroll proof, the compass graded reveal. The build would not have been groundable without it. Earns its own skill. | `/measure-render` skill OR verification-ladder | **PROPOSED (strongly reinforced, 4th instance)** |
+
+*Watch-list (not yet a skill proposal): the **open-work register pattern** — when tracking surfaces drift, consolidate them into ONE code-verified source of truth (built `project-arc-open-work-register.md` today). Overlaps `/spec-code-audit` for the spec↔code half; the new part is the content/governance/Phase-2 reconciliation. May graduate to a `/reconcile-open-work` discipline if it recurs on another workstream (L1/be).*