diff --git a/crates/lance-csr/src/path_value.rs b/crates/lance-csr/src/path_value.rs
index 6f0ee44..0060bf7 100644
--- a/crates/lance-csr/src/path_value.rs
+++ b/crates/lance-csr/src/path_value.rs
@@ -13,10 +13,10 @@ use arrow::compute::take;
 use arrow_schema::{DataType, Field, Fields};
 use datafusion::common::Result;
 
-pub(crate) const PROPERTIES: &str = "__lance_graph_properties";
-pub(crate) const IDENTITY: &str = "__lance_graph_identity";
-pub(crate) const START: &str = "__lance_graph_start";
-pub(crate) const END: &str = "__lance_graph_end";
+pub(crate) const PROPERTIES: &str = "__lance_csr_properties";
+pub(crate) const IDENTITY: &str = "__lance_csr_identity";
+pub(crate) const START: &str = "__lance_csr_start";
+pub(crate) const END: &str = "__lance_csr_end";
 const ENTITY: &str = "lance_csr.entity";
 
 pub(crate) fn entity_kind(data_type: &DataType) -> Option<&str> {
diff --git a/docs/cypher.md b/docs/cypher.md
index 7aed051..cd2443f 100644
--- a/docs/cypher.md
+++ b/docs/cypher.md
@@ -348,9 +348,9 @@ traversed that relationship backwards. Property access, `keys()` and node `label
 A zero-hop path has one node and an empty relationship list. All path functions return null for a null path.
 
 `RETURN path` produces an Arrow struct with `length`, `nodes`, `node_records` and `relationships` fields.
-Entity records store user properties in `__lance_graph_properties` and `{label, id}` identity metadata
-in `__lance_graph_identity`. Relationship records also carry the node records `__lance_graph_start` and
-`__lance_graph_end`. These metadata fields are distinct from user properties, even if a property has the
+Entity records store user properties in `__lance_csr_properties` and `{label, id}` identity metadata
+in `__lance_csr_identity`. Relationship records also carry the node records `__lance_csr_start` and
+`__lance_csr_end`. These metadata fields are distinct from user properties, even if a property has the
 same name. `keys(entity)` lists only non-null user properties; `keys(map)` includes null-valued keys.
 Prefer the Cypher property and endpoint functions when writing queries. Relationship identities identify actual occurrences within this query's inventory, not persistent
 IDs that can be compared across queries or refreshes. Rich paths support the same signed integer and
diff --git a/python/python/tests/test_path_values.py b/python/python/tests/test_path_values.py
index 2d5c329..41bee22 100644
--- a/python/python/tests/test_path_values.py
+++ b/python/python/tests/test_path_values.py
@@ -81,7 +81,7 @@ def test_zero_path_records(engine, shortest):
     assert rows[0]["length"] == 0
     assert rows[0]["ids"] == [3]
     assert rows[0]["edges"] == []
-    assert rows[0]["p"]["node_records"][0]["__lance_graph_properties"]["name"] == "c"
+    assert rows[0]["p"]["node_records"][0]["__lance_csr_properties"]["name"] == "c"
 
 
 def test_named_trails_do_not_reuse_a_relationship(engine):
@@ -194,8 +194,8 @@ def test_identical_parallel_occurrences_are_distinct():
     ).to_pylist()
     relationships = [r["p"]["relationships"][0] for r in rows]
     assert len(relationships) == 2
-    assert len({r["__lance_graph_identity"]["id"] for r in relationships}) == 2
-    assert [r["__lance_graph_properties"]["weight"] for r in relationships] == [5, 5]
+    assert len({r["__lance_csr_identity"]["id"] for r in relationships}) == 2
+    assert [r["__lance_csr_properties"]["weight"] for r in relationships] == [5, 5]
 
 
 @pytest.mark.parametrize("bad_type", [pa.string(), pa.uint64()])
@@ -316,18 +316,18 @@ def test_entity_metadata_does_not_shadow_properties():
             "Person": pa.table(
                 {"id": [1, 2], "nullable": pa.array([None, None], pa.string())}
             ),
-            "R": pa.table({"src": [1], "dst": [2], "__lance_graph_identity": [42]}),
+            "R": pa.table({"src": [1], "dst": [2], "__lance_csr_identity": [42]}),
         },
     )
     rows = engine.execute(
         "MATCH p = (a:Person)-[:R]->(b:Person) UNWIND relationships(p) AS r "
-        "RETURN r.__lance_graph_identity AS user_value, keys(r) AS properties, "
+        "RETURN r.__lance_csr_identity AS user_value, keys(r) AS properties, "
         "keys(startNode(r)) AS node_properties, keys({nullable: null}) AS map_keys"
     ).to_pylist()
     assert rows == [
         {
             "user_value": 42,
-            "properties": ["__lance_graph_identity", "dst", "src"],
+            "properties": ["__lance_csr_identity", "dst", "src"],
             "node_properties": ["id"],
             "map_keys": ["nullable"],
         }
@@ -344,8 +344,8 @@ def test_multiple_path_operators_share_occurrence_identity(engine):
     ids = {}
     for row in rows:
         for edge in row.values():
-            weight = edge["__lance_graph_properties"]["weight"]
-            identity = edge["__lance_graph_identity"]
+            weight = edge["__lance_csr_properties"]["weight"]
+            identity = edge["__lance_csr_identity"]
             assert ids.setdefault(weight, identity) == identity
     assert ids[5] != ids[7]
 
@@ -460,7 +460,7 @@ def test_path_alias_retains_actual_relationship_values(engine, suffix):
     rows = engine.execute(query).to_pylist()
     edges = [row.get("edges", row.get("path", {}).get("relationships")) for row in rows]
     assert sorted(
-        [r["__lance_graph_properties"]["weight"] for r in path] for path in edges
+        [r["__lance_csr_properties"]["weight"] for r in path] for path in edges
     ) == [
         [5, 11],
         [7, 11],