I'm not often corresponding author, but I distinctly remember one paper where, at the proofs stage (after acceptance), the copy for approval came back with one real word systematically replaced with another that wasn't in the dictionary (English or American). So even the tampering is not out of the question. (In hindsight, my suspicious mind wonders if this isn't a way of introducing subtle new copyrightable aspects.)
Biting the hand that feeds IT
