Email Record: On the estimation of the length of normal derivations /