In a paper by J.-L. Krivine, [*Modèles de ZF+AC dans lesquels tout ensemble de réels définissable en termes d'ordinaux est mesurable-Lebesgue*](http://gallica.bnf.fr/ark:/12148/bpt6k4802973/f547.image) [C. R. Acad. Sci. Paris Sér. A-B 269 (1969), A549–A552, [MR0253894](http://www.ams.org/mathscinet-getitem?mr=253894)], he presents a model of ZFC in which every set of reals definable from ordinals is Lebesgue measurable.

 My question is:

 1. He seems not to assume the existence of an inaccessible cardinal in the ground model, is this correct? 

 2. In Theorem 3 of that paper, he concludes that every set of reals definable from ordinals is Lebesgue measurable. Does this continue to hold if we replace definable from ordinals by definable from a countable sequence of ordinals?

 3. Does Theorem 3 of that paper still hold if we replace "Lebesgue measurable" by "having the property of Baire"?