Theorem 1.1.3 (induction principle) of Dirk Van Dalen "Logic and Structure" states:
Let $A$ be a property, then $A(\phi)$ holds for all $\phi \in PROP$ if:
- $A(p_i)$, for all i;
- $A(\phi),A(\psi) \Rightarrow A(\phi \square \psi)$
- $A(\phi) \Rightarrow A(\neg \phi)$
I don't understand the little proof he gives. He writes let $X=\{\phi \in PROP | A(\phi) \}$, then X satisfies the conditions of the recursive definition of $PROP$. So $PROP \subseteq X$,i.e. for all $\phi \in PROP$ $A(\phi)$ holds.
The proof comes from the fact that PROP is the smallest set of well-formed formulae. The proof van Dalen gives is correct, but too brief. As he defines $X=\{\phi \in PROP | A(\phi) \}$, then by the definition of PROP, which is the smallest set of well-formed formulae, it follows that $PROP \subseteq X$, otherwise it would be smaller than PROP itself.