I would mentioned the Duhem-Quine thesis here. The notion of a "critical experiment" is generally quite naive in the grand scheme of things. Parsimonious perhaps, but any falsifying experiment tests not merely the theory under consideration, but all human knowledge, as it were.