Remarks on the conservatism of sigma(0 -- E)2/E in survival data.