On the constancy of the evolutionary rate of cistrons