I'm curious if there is a finite measure on the $\sigma$-algebra of subsets of $[0,1]$ with the Property of Baire, whose null sets are exactly the meagre sets.

I'd also be interested how "nice" such a measure can be like can it be Radon(when restricted to Borel sets) for example.


2 Answers 2


The answer is no. Assume that such a measure $\mu$ exists.

First, since every singleton in $[0,1]$ is closed with empty interior, $\mu(\{x\}) = 0$ for all $x \in [0,1]$. Write $B_{x,\epsilon}$ for the open ball around $x$ of radius $\epsilon$ with respect to the standard metric on $[0,1]$. By countable additivity, for all $x \in [0,1]$, $\mu(B_{x,2^{-n}}) \to 0$. If we take an enumeration of the rationals $(q_i)_{i \in \mathbb{N}}$, for each $i$ there exists an $n_i$ such that $\mu(B(q_i,2^{-n_i})) < 2^{-i}$. So $D_1 = \bigcup_{i=1}^\infty B(q_i,2^{-n_i})$ is a dense open set with $\mu(D_1) \leq 1$.

By re-doing the previous construction, picking $\mu(B(q_i,2^{-n_i})) < 2^{-(i+k)}$, we can define dense open sets $D_k$ with $\mu(D_k) \leq 2^{-k}$. Now, by countable additivity $N = \bigcap_{k=1}^\infty D_k$ has measure zero. The set $[0,1]\setminus N$ is a union of closed sets with empty interior, i.e. a meagre set, so $\mu([0,1] \setminus N) = 0$ as well, so $\mu([0,1]) = 0$.

I see that Nate Eldredge was a bit quicker than I was, so I'll add that it is possible to find a finitely-additive probability measure whose null sets are exactly the meagre sets -- this is most easily done using the isomorphism between the Baire property algebra modulo meagre sets and the algebra of regular open sets.


No. For any finite Borel measure $\mu$ on $[0,1]$, there is a comeager Borel set of $\mu$-measure zero.

First note that $\mu$ has at most countably many atoms, so it will be possible to find a countable dense set $D \subset [0,1]$ containing no atoms, i.e. $\mu(D) = 0$. Now any finite Borel measure on a metric space is outer regular, so for any $n$ there is an open set $U_n$ containing $D$ and with $\mu(U_n) < 1/n$. Setting $G = \bigcap_n U_n$, we see that $G$ is a dense $G_\delta$ (hence comeager) and $\mu(G) = 0$.

Relevant to your second question, the answers in the question linked above also mention that any finite Borel measure on a Polish space is Radon.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.