[FOM] Hahn Banach and the Baire Property

[D.R. MacIver]

Sorry, managed to find a reference on my own after all. 

The existence of a finitely additive probability measure on P(N) which 
vanishes on finite sets is sufficient to give a subset of {0, 1}^N which 
lacks the baire property. Specifically (identifying {0, 1}^N with P(N) in 
the obvious way) the set { a \in {0, 1}^N : u(a) = 0 } lacks the baire 
property.

If you want more details I can either reproduce them here, or you can check 
out the reference HAF (page 810, ``Pincus's Pathology'' ).

Thanks,
David MacIver

On Mar 5 2005, D.R. MacIver wrote:

> Does anyone happen to know (and ideally have a cite for) whether it is 
> known if the Hahn Banach theorem implies the existence of a subset of R 
> without the Baire Property (in ZF say)? I'm presuming it does, as it 
> proves the existence of a non-measurable subset of R, but I can't seem to 
> find a reference one way or the other.
> 
> (Sorry for the rather analysis-like question, but I think it really is 
> still FOM. Especially as it's entirely possible that the only way to 
> answer it is to construct a certain model of set theory!)
> 
> Thanks,
> David R. MacIver