diff --git a/CHANGELOG.md b/CHANGELOG.md
index 01597056..d15593a7 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -16,6 +16,35 @@ true until the next version shipped.
 
 ## [Unreleased]
 
+### Fixed
+
+- One geometry arm accepted a comparison the property does not describe (#1081).
+
+  #1081 replaced a count of `entry->fileOffset != rg->fileOffset` with a per-field
+  membership loop, so that dropping any one of the four compared fields reddens by
+  name. The loop matched `entry->$_f != rg->$_f` OR `entry->$_f != natts` for every
+  field, because `natts` is the one field compared against the scan's own column
+  count rather than against the row group.
+
+  Harmless today -- `entry->firstRowNumber != natts` appears nowhere -- and still
+  wrong as a claim: each arm would accept a comparison its own name denies. Three
+  fields now match only the `rg` form, and `natts` has its own arm and its own name.
+
+  Reported by @jdatcmd on #1081's review.
+
+  All four verified by removal on the real suite, each reddening only its own arm:
+
+      firstRowNumber removed   FAIL  a hit re-checks the group's firstRowNumber ...
+      rowCount removed         FAIL  a hit re-checks the group's rowCount ...
+      fileOffset removed       FAIL  a hit re-checks the group's fileOffset ...
+      natts removed            FAIL  a hit re-checks the group's natts against the
+                                     scan's column count
+      restored                 26 passed, source byte-identical
+
+  `firstRowNumber` is the one #1081 shipped unproven. Its mutation did not apply --
+  that line starts `(entry->` rather than `entry->`, so the pattern missed -- and the
+  harness's applied-assertion reported it rather than counting a clean run as a pass.
+
 ### Added
 
 - Six more guards counted their callers instead of pinning their property (#1078's
diff --git a/test/native_fetch_cache.sh b/test/native_fetch_cache.sh
index 950f7983..e936e050 100755
--- a/test/native_fetch_cache.sh
+++ b/test/native_fetch_cache.sh
@@ -237,13 +237,25 @@ check "an entry from an earlier command is rejected" \
 
 # The recheck compares four fields. Counting one of them asserted nothing about
 # the other three, and a reflow that dropped one would have passed.
-for _f in firstRowNumber rowCount fileOffset natts; do
+#
+# THREE COMPARE AGAINST THE ROW GROUP AND ONE DOES NOT. `natts` is compared against
+# the scan's own column count rather than against `rg`, so it needs its own pattern.
+# Reported by @jdatcmd: the first version put `!= natts` in the alternation for ALL
+# four, which is harmless -- `entry->firstRowNumber != natts` appears nowhere -- and
+# is still wrong as a claim, because each arm would then accept a comparison the
+# property does not describe.
+for _f in firstRowNumber rowCount fileOffset; do
 	check "a hit re-checks the group's $_f against the row group" \
 		"$(case "$_nfc_row" in \
-			*"entry->$_f != rg->$_f"*|*"entry->$_f != natts"*) echo yes ;; \
+			*"entry->$_f != rg->$_f"*) echo yes ;; \
 			*) echo no ;; esac)" "yes"
 done
 
+check "a hit re-checks the group's natts against the scan's column count" \
+	"$(case "$_nfc_row" in \
+		*"entry->natts != natts"*) echo yes ;; \
+		*) echo no ;; esac)" "yes"
+
 # Named call sites rather than a count of 2: a third honest caller is not a defect,
 # and losing either of these two is.
 #