diff --git a/docs/re/structures/ui-paint-order-key.md b/docs/re/structures/ui-paint-order-key.md
index c5b8276d..224be167 100644
--- a/docs/re/structures/ui-paint-order-key.md
+++ b/docs/re/structures/ui-paint-order-key.md
@@ -326,11 +326,23 @@ X/Y, the `T8aD` header words, and `kind`.
 
 ## The corpus-wide census (2026-08-26)
 
-Everything above rests on twelve values from two screens. This walks **all 4 525
-`T8aD` sprites on the disc**, so the open question can be asked of the whole
-corpus. `tools/re-capture/paint_key_census.py`, output in
+Everything above rests on twelve values from two screens. This walks **4 525
+`T8aD` sprites on the disc**, so the open question can be asked of a corpus.
+`tools/re-capture/paint_key_census.py`, output in
 [`../data/paint-key-census.txt`](../data/paint-key-census.txt).
 
+> ⚠️ **Read the next paragraph before using these numbers.** The census counts
+> pak entries whose *own* first four bytes are `T8aD`. But the sprites this page
+> measures paint order on are **children of a `RATC` bundle** — `ui_layout.rs`
+> reaches them through `ratc::parse`, and a bundle entry's magic is `RATC`, so a
+> child `T8aD` is invisible to a top-level magic filter. The two populations may
+> overlap partly, wholly, or not at all, and **which one the 45 keys describe is
+> being measured, not assumed.** Until that lands, treat every count below as a
+> statement about top-level `T8aD` entries only.
+>
+> This is the same failure shape as the 37/45 note further down: a number
+> computed over a population I had not checked was the population in question.
+
 ### ✅ The field is a `u16` at `+0x0A`, not a word at `+0x08`
 
 The upper half of the 32-bit word this page reads is **zero in 4 525 / 4 525**