On the gravitational stability of a disk of stars