Noticed during #442 that there are no methods at all for squin's noise statements even though the number of sites is usually known.