On Edwards' criterion of seasonality and a non-parametric alternative.