diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
index 9f59ddb..5a7645a 100644
--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -51,10 +51,14 @@ left over from a previous run.
 
 Before computing deltas, the library detects field layout changes by comparing
 each table's stored fields in the STATE file against the current config's
-`ordered_field_names()`. Tables whose layout changed are recorded in the block
-as a `TableChange` with no delta (`delta: None`), signaling that patch
-consolidation should use a full state snapshot for that table instead of
-attempting to merge incompatible deltas.
+canonical field list (primary keys first, then subsidiaries; each group sorted
+lexicographically by name).
+
+Because tuple identity is canonical, reordering fields in `tables.toml` does not
+register as a layout change. Adding, removing, or renaming a field does. Tables
+whose layout changed are recorded in the block as a `TableChange` with no delta
+(`delta: None`), signaling that patch consolidation should use a full state
+snapshot for that table instead of attempting to merge incompatible deltas.
 
 All table changes are bundled into a block together with a parent hash and a
 timestamp, SHA-1 hashed, and stored as a file named by its hash. The `HEAD`
diff --git a/README.md b/README.md
index 8b9af0d..e35a364 100644
--- a/README.md
+++ b/README.md
@@ -65,11 +65,13 @@ Config can be `config.toml` or `config.json`.
 
 - Each table must have at least one field marked `primary-key = true`
 - Field names within a table must be unique
+- When `header = false` (the default), CSV columns are mapped to config fields by
+  position — the first column maps to the first field, etc.
 - When `header = true`, the first row of the CSV is treated as a header. Each
   config field is matched to a CSV column by name, so columns may appear in any
   order. Every config field name must be present in the header; extra CSV columns
-  are ignored. When `header = false` (the default), CSV columns are mapped to
-  config fields by position — the first column maps to the first field, etc.
+  are ignored. In this mode, the order in which fields are declared under the
+  table is cosmetic — reordering them does not invalidate existing state.
 - The type field controls how values are quoted in generated SQL. These are not
   database column types — your database may use any compatible type (e.g.
   `INTEGER`, `FLOAT`, `TIMESTAMP`). It is your responsibility to ensure the
@@ -114,9 +116,9 @@ fields = [
 ]
 ```
 
-`BOOLEAN` fields can override the strings recognised as true and false. When
+`BOOLEAN` fields can override the strings recognized as true and false. When
 either override is set, only the configured strings are accepted — the
-defaults are not honoured alongside them. The `true`, `false`, and `null`
+defaults are not honored alongside them. The `true`, `false`, and `null`
 sentinels on a single field must all differ.
 
 ```toml
diff --git a/tests/accept_field_reorder.rs b/tests/accept_field_reorder.rs
new file mode 100644
index 0000000..23694e1
--- /dev/null
+++ b/tests/accept_field_reorder.rs
@@ -0,0 +1,67 @@
+mod common;
+
+use leech2::block::Block;
+use leech2::config::Config;
+use leech2::patch::Patch;
+use leech2::sql;
+
+/// Reordering fields in `tables.toml` between two blocks must not produce a
+/// spurious delta when the underlying CSV data is unchanged. Tuple identity
+/// is canonical (lexicographic by field name) so that the field declaration
+/// order in the config is cosmetic.
+#[test]
+fn test_field_reorder_in_config_produces_no_delta() {
+    common::init_logging();
+    let tmp = tempfile::tempdir().unwrap();
+    let work_dir = tmp.path();
+
+    common::write_csv(
+        work_dir,
+        "users.csv",
+        "id,name,email\n1,Alice,a@example.com\n2,Bob,b@example.com\n",
+    );
+
+    common::write_config(
+        work_dir,
+        "config.toml",
+        r#"
+[tables.users]
+source = "users.csv"
+header = true
+fields = [
+    { name = "id", type = "NUMBER", primary-key = true },
+    { name = "name", type = "TEXT" },
+    { name = "email", type = "TEXT" },
+]
+"#,
+    );
+    let hash1 = Block::create(&Config::load(work_dir).unwrap()).unwrap();
+
+    // Reorder fields (id is still the only primary key, but its position in
+    // the declared field list moves; subsidiaries are also reordered).
+    common::write_config(
+        work_dir,
+        "config.toml",
+        r#"
+[tables.users]
+source = "users.csv"
+header = true
+fields = [
+    { name = "email", type = "TEXT" },
+    { name = "id", type = "NUMBER", primary-key = true },
+    { name = "name", type = "TEXT" },
+]
+"#,
+    );
+    let config = Config::load(work_dir).unwrap();
+    let _hash2 = Block::create(&config).unwrap();
+
+    // Patch from hash1 should be empty: same data, just a different
+    // declaration order.
+    let patch = Patch::create(&config, &hash1).unwrap();
+    if let Some(s) = sql::patch_to_sql(&config, &patch).unwrap() {
+        assert_eq!(s.trim(), "BEGIN;\nCOMMIT;");
+    }
+
+    common::assert_wire_roundtrip(&config, &patch);
+}