The validity of Drude's optical method of investigating transparent films on metals