diff --git a/docs/re/METHOD.md b/docs/re/METHOD.md
index 15c0965c..d7a1cd2f 100644
--- a/docs/re/METHOD.md
+++ b/docs/re/METHOD.md
@@ -2340,3 +2340,16 @@ detector can discard a candidate, it must say how many it discarded, or its clea
 run is unfalsifiable. Both of us reached the same structural conclusion from
 opposite failures within a day — theirs by over-reporting loudly, mine by passing
 quietly, which is why mine went unnoticed and theirs did not.
+
+### The suppressed set, read — and it is clean
+
+The 7 suppressed mentions (8 before a dedup fix; two registered claims can be
+substrings of one line, which printed it twice) were read individually rather than
+left as a number. **All 7 are genuine correction contexts** — an *"An earlier
+version of this bullet said"*, a *"Withdrawing …"*, an explicit *"does **not**
+revive …"*, two bullets under *"🔴 What this retracts"*, a *"supersedes the …
+banner"*, and one *"recorded as … It was on the disc all along"*. **Zero live
+revivals.**
+
+So the register's clean run is now backed by a reading. That is the point: the
+number was worth nothing until someone looked, and looking took one pass.
diff --git a/tools/re-capture/check_refuted.py b/tools/re-capture/check_refuted.py
index 664f007e..9f7eb8b2 100755
--- a/tools/re-capture/check_refuted.py
+++ b/tools/re-capture/check_refuted.py
@@ -53,6 +53,7 @@ for line in ref.read_text().splitlines():
 print(f"{len(claims)} quoted claims in REFUTED.md\n")
 hits = 0
 suppressed = []
+seen_marked = set()
 for c in claims:
     needle = c.strip()
     for f in root.rglob("*.md"):
@@ -76,7 +77,11 @@ for c in claims:
                 near = "\n".join(lines[lo:hi]) + "\n" + hdr
                 marked = any(m.lower() in near.lower() for m in MARKERS)
                 if marked:
-                    suppressed.append((f, i + 1, needle, l.strip()))
+                    # dedup like the reported path: two registered claims can be
+                    # substrings of one line, which printed it twice.
+                    if (f, i) not in seen_marked:
+                        seen_marked.add((f, i))
+                        suppressed.append((f, i + 1, needle, l.strip()))
                     continue
                 if (f, i) not in seen_report:
                     seen_report.add((f, i))
diff --git a/tools/re-capture/focus_persistence.sh b/tools/re-capture/focus_persistence.sh
new file mode 100755
index 00000000..ff75eb02
--- /dev/null
+++ b/tools/re-capture/focus_persistence.sh
@@ -0,0 +1,58 @@
+#!/usr/bin/env bash
+# Does the main menu REMEMBER its cursor across menu -> title -> menu?
+#
+# menu-navigation-semantics.md carries this 🟡: "Initial focus is reproducible
+# but not established as invariant. Both of my boots opened on TUTORIAL, and
+# both used boot_menu.sh." Two runs through one harness are not two samples --
+# the same shape as two captures through one gate (plate-pulse-phase-lock.md).
+#
+# And the sources DISAGREE. boot_menu.sh's own final line says "cursor on NEW
+# GAME"; the page says TUTORIAL 2/2; menu-state-in-memory.md reaches EXTRAS in
+# four downs, which only counts from NEW GAME. Two say NEW GAME, one says
+# TUTORIAL.
+#
+# ONE BOOT settles both, and the round trip is the part no run has done:
+#   F1  focus when the menu first appears           <- re-measures initial focus
+#   F2  focus after 2x DOWN                         <- CONTROL for the reader
+#   F3  focus after B (to title) then A (back)      <- persist or reset?
+#
+# F3 == F2 => the menu restores where you were; F3 == F1 => it resets.
+# If F2 is not two items below F1 the reader is not tracking the cursor and
+# NOTHING below it may be read. Run as ONE BLOCKING FOREGROUND call.
+set -u
+export HOME=/sylph-home/re SDL_AUDIODRIVER=dummy DISPLAY=:98
+SD="$(cd "$(dirname "$0")" && pwd)"
+OUT="${OUT:-/sylph-home/re/focuspersist}"; mkdir -p "$OUT"
+NAMES=(NEW_GAME LOAD_GAME TUTORIAL OPTIONS EXTRAS)
+
+. "$SD/ensure_single_emulator.sh"
+
+echo "── EFFECTIVE CONFIG ──────────────────────────────"
+echo "   out dir   = $OUT"
+echo "   harness   = boot_menu.sh (the one under suspicion)"
+echo "   sequence  = F1 -> 2x DOWN -> F2 -> B -> plate pulse -> A -> F3"
+
+BOOT_MENU_LOG="$OUT/boot.stdout" "$SD/boot_menu.sh" fp >"$OUT/boot.log" 2>&1 \
+  || { echo "BOOT FAILED"; tail -5 "$OUT/boot.log"; exit 1; }
+tail -2 "$OUT/boot.log"
+
+shot(){ screenshot "$OUT/$1.png" >/dev/null 2>&1; }
+focus(){ python3 "$SD/menu_focus.py" "$OUT/$1.png" | tail -1; }
+
+shot f1; echo "F1 $(focus f1)"
+
+python3 "$SD/pad.py" dpad down 0.06; sleep 1
+python3 "$SD/pad.py" dpad down 0.06; sleep 2
+shot f2; echo "F2 $(focus f2)"
+
+echo "-- B to the title --"
+python3 "$SD/pad.py" tap B 0.25
+python3 "$SD/wait_plate_pulse.py" 120 >"$OUT/pulse.log" 2>&1 \
+  && echo "   title settled (plate pulse)" || echo "   ⚠️ NO PLATE PULSE in 120 s"
+shot title
+
+echo "-- A back to the menu --"
+python3 "$SD/pad.py" tap A 0.25; sleep 12
+shot f3; echo "F3 $(focus f3)"
+
+echo "FOCUS PERSISTENCE RUN DONE"