On the equivalence of the weighted least squares and the generalised least squares estimators