ON THE ORIGIN AND PERSISTENCE OF ERROR IN GEOGRAPHY