A Discussion Opened by L.J. Savage at the Meeting of the Joint Statistics Seminar, Birkbeck and Imperial Colleges, in the University of London