diff --git a/docs/re/ui-title-build-map.md b/docs/re/ui-title-build-map.md
index e6098cd6..bd873f57 100644
--- a/docs/re/ui-title-build-map.md
+++ b/docs/re/ui-title-build-map.md
@@ -829,4 +829,46 @@ vertex-colour capture — were built on "the skewed draw is the swoosh", an
 identification made by *elimination on one screen* and never checked against the
 draw's own coordinates. The eliminations themselves stand (they were measured
 against the capture, not against the identification), but the chain of reasoning
-that pointed at `ptlogo_back2*` did not.
\ No newline at end of file
+that pointed at `ptlogo_back2*` did not.
+
+---
+
+## 🟢 Refutation attempt, 2026-08-29 — "entries 6/9 are a three-button submenu". It SURVIVED.
+
+The port reported `GP_TITLE` entries 6 and 9 as a **three-button** submenu
+(`ptbtn11/12/13` at x = 532, y = 282 / 362 / 442). This page and HANDOFF call
+6/9 `EXTRAS` and say nothing about a button count, and 18 elements looked like
+too many for three buttons, so the claim was challenged.
+
+**The challenge was wrong and the claim stands.** `screen info --all --build 6`:
+
+```
+ptframe3.t32 ptframe4.t32 ptbtn11.rat ptbtn12.rat ptbtn13.rat ptmsg2.t32
+pteff20.t32 pteff21.t32 pteff22.t32 pteff23.t32 pttitle.t32 ptbase.t32
+pteff05.t32 pteff02.prm ptloop01.rat ptloop02.rat pteff10.t32 pteff00.prm
+```
+
+Entry 9 is identical. Three `ptbtn*` elements, and the other fifteen are frame,
+title, background and effect layers. Both statements hold at once: 6/9 **are**
+`EXTRAS` — our composite of entry 6 correlates **+0.944** whole-frame with the
+committed [`live-extras.png`](captures/title-builds/live-extras.png) — **and**
+`EXTRAS` is a three-button screen. The corpus had never recorded its button
+count; the port established it, and this page now does too.
+
+⚠️ The challenge was raised from an element *count* without listing the elements,
+when a one-line `screen info` was available. Recorded because the protocol's
+adversarial duty is worth nothing if only the successful challenges get written
+down.
+
+## ✅ Entries 5 and 8 cannot be told apart by layout — only by pixels
+
+Checked in the same pass, because it bears on which build is which screen state.
+Entries 5 and 8 have **identical element lists** (`pteff00.prm ptbase pteff05
+ptloop01 ptloop02 pteff02.prm ptframe1 ptframe2 pteff10 pteff12 ptbtn01…05
+ptmsg`) and identical button placements. The EN/JP difference between the two
+main-menu bundles lives in the **baked sprite pixels**, not in any layout field.
+
+So the 🟡 rule that "the English member of a pair is the one in the first half of
+the data segment" is not going to be replaced by a layout field for this pair —
+separating 5 from 8 needs either the sprite images or a capture. The same is true
+of 6 vs 9, whose element lists are also identical.