I'm reading, on Jech's book Set Theory, 3rd millenium edition, the proof of Solovay's theorem for measurability of projective sets. The proof goes on using some properties of Boolean algebras.
In particular, let $M$ be the $\operatorname{ZFC}$-model $M[G]$, the generic extension carried out by the Lévy collapse $Col(\aleph_0,< \kappa)$, $\kappa \in M$ inaccessible, and $B$ the relating Boolean algebra. If $s\in M[G]$ is a countable sequence of ordinals of $M$, there exists a complete subalgebra $D\subseteq B$ such that $M[s]=M[D\cap G]$ (this happens every time i have a model between $M$ and $M[G]$, and this is the Lemma 15.43 of the book).
What I can't see (in the Lemma 26.16 in the proof of the theorem) are the following:
- Since $B$ is $\kappa$-saturated, there exists a subalgebra $D\subseteq B$ such that $|D| < \kappa$, and $M[s]=M[D\cap G]$.
and, immediately,
- It follows that $\kappa$ is inaccessible in $M[s]$.
I'd say these follow only by basic properties of Boolean algebras (and the construction of $D$ in Lemma 15.43) and by the definition of inaccessible cardinals, but I can't see why they hold.