Embedding classical in minimal implicational logic

Consider the problem which set V of propositional variables suffices for StabV `i A whenever `c A, where StabV := {¬¬P → P | P ∈ V }, and `c and `i denote derivability in classical and intuitionistic implicational logic, respectively. We give a direct proof that stability for the final propositional variable of the (implicational) formula A is sufficient; as a corollary one obtains Glivenko’s theorem. Conversely, using Glivenko’s theorem one can give an alternative proof of our result. As an alternative to stability we then consider the Peirce formula PeirceQ,P := ((Q → P ) → Q) → Q. It is an easy consequence of the result above that adding a single instance of the Peirce formula suffices to move from classical to intuitionistic derivability. Finally we consider the question whether one could do the same for minimal logic. Given a classical derivation of a propositional formula not involving ⊥, which instances of the Peirce formula suffice as additional premises to ensure derivability in minimal logic? We define a set of such Peirce formulas, and show that in general an unbounded number of them is necessary.