A counterexample to the « maximal subgroup rule » for continuous crystalline transitions