On The Rationale Of Maximum Entropy Methods