On the symmetric Anderson model for a magnetic impurity and on Mattis' model