On the growth of linear perturbations