diff --git a/restoration/source410/BT_L4/GAUGE-BLOCK.NOTES.md b/restoration/source410/BT_L4/GAUGE-BLOCK.NOTES.md
index ad27453a..7aca7c0d 100644
--- a/restoration/source410/BT_L4/GAUGE-BLOCK.NOTES.md
+++ b/restoration/source410/BT_L4/GAUGE-BLOCK.NOTES.md
@@ -722,3 +722,59 @@ NEXT ON THIS HEAD-BY-HEAD LADDER: the three MFD quadrant heads (Mfd1/2/3),
 which now hold essentially all of the remaining difference -- ~22.6K missing
 and ~12K extra between them, against ~0 on the engineering heads.  Those are
 the SubsystemCluster quadrant panels ([cluster] port=2/4/6 in the map).
+
+--------------------------------------------------------------------------------
+5.3.32b -- THE WEAPON PANEL: a measured work list for the MFD heads
+--------------------------------------------------------------------------------
+With the engineering heads exact, the whole remaining cockpit difference is
+the three MFD quadrant heads (Mfd1/2/3: ~22.6K missing, ~12K extra).  Those
+are the SubsystemCluster panels, and they are NOT CFG widgets -- the map
+shows port 2/4/6 carry only a bgBitMap, because every panel is built by the
+`vehicleSubSystems` widget (port 8) walking the mech's aux screens.  So this
+is our code end to end.
+
+Region breakdown of Mfd1 (planes.py bands), shipped vs ours:
+
+    top/bottom soft-button strips      0 missing /    0 extra   <- exact
+    upper title band                 553        /  553
+    upper panel bodies              6305        / 2033          <- the work
+    mid soft-button strip            386        /  235
+    lower panel bodies              1405        /  890
+
+The title band's missing==extra is NOT a placement error.  Correlating our
+extra pixels against the shipped missing set over a +-90px window peaks at
+only 30% for dy=-3, which is the signature of two copies of the SAME bitmap
+already overlapping, with equal-sized fringes where they disagree by a pixel
+or two.  Confirmed by experiment: 0xb6 and 0xb0 both measured WORSE than the
+0xb3 in the tree (Mfd1 extra 4063 -> 5010 and -> 6023), so 0xb3 stands as the
+measured optimum and the banner is at most 1-2px off.  Do not "fix" it again
+without the rig open.
+
+WHAT THE PANEL BODIES ARE ACTUALLY MISSING (shipped vs ours, per weapon panel):
+
+  1. THE AMMO DISC + RAYS ARE ABSENT ON THE AMMO WEAPON.  Shipped draws every
+     weapon panel with a filled disc ringed by radial ticks; the STREAK 6
+     panel carries "0024" inside it -- exactly the 24 rounds our own
+     [proj] log reports for SRM6_1.  We draw disc+rays fine on the two
+     ENERGY weapons (PPC, ERMED LASER) and draw NEITHER on STREAK 6.  So the
+     ammo-bearing path diverges from the energy path -- that is the thread to
+     pull, not the disc drawing itself.
+
+  2. THE AMMO DIGITS RENDER AS EMPTY CELLS.  Ours draws four hollow boxes at
+     exactly the shipped digits' x/y (verified by crop overlay) -- right
+     widget, right place, no glyphs.  Same failure class as the pilot-name
+     bitmap: the cell draws, the raster lookup returns nothing.
+
+  3. THE MECH SILHOUETTE IS MISPLACED.  Shipped centres the dotted mech
+     figure in the panel; ours puts it low-left, clipped by the panel edge.
+     Same bitmap, wrong origin -- likely the largest single pixel
+     contributor in the body count.
+
+  4. THE STATE BOX READS "OFF" WHERE SHIPPED SHOWS A DIGIT (4 / 2 / 1).
+     This is the lamp-state-frame item already on the list; the letter box
+     beside it (A / B / B) already matches.
+
+Everything in this list came from the per-head instrument in one sitting;
+none of it was visible in the whole-frame score, which moved less than a
+point across the entire session even as Eng1/2/3 went to exact.  Score the
+heads, not the frame.