Making a Paradigmatic Convention Normal: Entrenching Means and Variances as Statistics