Some remarks on the statistical model of heavy ion collisions