The wedge subjected to tractions: a paradox re-examined