Some remarks on the method of minimal residues