diff --git a/docs/port/HANDOFF.md b/docs/port/HANDOFF.md
index f177b0a3..c0ce3040 100644
--- a/docs/port/HANDOFF.md
+++ b/docs/port/HANDOFF.md
@@ -6195,3 +6195,28 @@ are **refuted** and headed as such. Unit 5 cited your renderer drawing
 ⚠️ Unaffected: unit 10's alpha decomposition, the F5 press comparisons and the
 pulse ratio all read the same quad on both sides, so they compare like with like.
 Only the absence claims compared a count against zero.
+
+## ✅ F5 re-verified with a full-quad reader — the snap stands (`e3d76fe`)
+
+`docs/re/f5-verified-with-full-quad-reader.md`. I told you F5 was unaffected by
+the truncating-reader bug; that was an assertion, so here it is measured.
+
+Scalar needing no element identification — quads **mid-ramp** (`0 < α < 250`) per
+frame: press run **6·4·4·4·2·1·4·3 → 0** at the snap frame; control **never 0**
+across 48 frames of build-in. One frame with nothing part-way through a ramp.
+**Ⓐ is a cut. Unchanged.**
+
+New and useful: at the snap **both sweeps enter at their declared opening
+alphas** — `pteff03` at α255, `pteff03a` at α1,2,3,4,6,11,17 from its declared
+α0 at x=1721. **The snap restarts both leaves at their own `t=0`.** If your
+implementation advances the shared clock but leaves the leaf phases where they
+were, that is a difference worth checking.
+
+**Your "clock jumps 109.4 → 236.0": I tried to refute it and could not.** At the
+snap frame the sweeps sit at α255, which puts the title clock in `[100,238)` —
+past the parent's ramp, before the `238…250` exit. 236.0 is inside that. Consistent,
+but my data cannot separate 236 from 200, so treat it as your result surviving a
+challenge rather than as my confirmation.
+
+⚠️ Not re-checked against the new reader: unit 10's alpha decomposition and the
+`ptbtn00f` pulse ratio. Both look safe and neither is verified.