I am working through a theorem in Royden and Fitzpatrick's Real Analysis on necessary and sufficient conditions for a set to be measurable, and in their proof they use the following:

Let $E$ be measurable with outer measure $m^*(E) = \infty$. Then $E$ can be written as a disjoint union of a countable collection of measurable sets, each of which have finite outer measure.

How might one prove this. I could use some help on getting started.

  • 1
    $\begingroup$ Is this in the chapter about the Lebesgue outer measure, or in the chapter about the general theory of outer measure? Because the claim is not true without some condition of $\sigma$-finiteness on $m$. $\endgroup$ – user228113 Sep 5 '17 at 16:27
  • $\begingroup$ What is your measure space? (I.e., is this Lebesgue measure on $\mathbb{R}$ or not?) Do you believe $E$ can be written as the plain union of a countable collection of measurable sets, each of which have finite outer measure? $\endgroup$ – Eric Towers Sep 5 '17 at 16:28
  • $\begingroup$ @G.Sassatelli This is in chapter 2 called Lebesgue Measure, specifically in theorem 11 in section 2.4 (called Outer and Inner Approximation of Lebesgue Measurable Sets). $\endgroup$ – user193319 Sep 5 '17 at 16:31
  • $\begingroup$ @EricTowers Honestly, I am not sure: I haven't yet gotten comfortable with the material to make any 'intuitive' judgments or arguments. $\endgroup$ – user193319 Sep 5 '17 at 16:32
  • $\begingroup$ @user193319 That theorem deals with Lebesgue measure on the real line. There, we can chop a set up into pieces that are contained in intervals of unit length, each of which must have finite measure. $\endgroup$ – Xander Henderson Sep 5 '17 at 16:33

I am working out of the 4th edition. Near the top of p. 41, the authors write

Now consider the case that $m^{\ast}(E) = \infty$. Then $E$ may be expressed as the disjoint union of a countable collection $\{E_k\}_{k=1}^{\infty}$ of measurable sets, each of which has finite outer measure.

Note that this passage is from the chapter on the Lebesgue measure on $\mathbb{R}$, and that the particular quote above is part of the proof of a theorem which asserts several equivalent characterizations of the measurability of a set of real numbers $E$. Hence the theorem (and the quoted passage) is about sets of real numbers. This is important, as the real numbers are $\sigma$-finite—the result does not hold in complete generality.

As an example of such a decomposition into disjoint sets of finite measure, take $$ E_k = E \cap [k,k+1), $$ where $k$ ranges over the integers. Since each interval is measurable and $E$ is measurable (Royden and Fitzpatrick are seeking to show that each of four conditions is equivalent to measurability, hence $E$ is measurable by hypothesis), the intersection of each interval with $E$ is also measurable. Finiteness of the outer measure of $E_k$ follows from monotonicity: each set is contained in an interval of unit length.


Since the setting is Lebesgue measure on $\mathbb{R}$, we may break up the reals into unit intervals.

Let $F_n = E \cap [n,n+1)$ for $ n \in \mathbb{Z}$. Then $\{F_n \mid n \in \mathbb{Z}\}$ is a countable collection of measurable sets ($[n,n+1)$ is measurable, intersections of measurables are measurable) that are disjoint and whose union is $E$.

(I see that Xander Henderson commented to this effect while I was typing.)

  • $\begingroup$ Jinx! It seems we had the same thought at the same time. ;) $\endgroup$ – Xander Henderson Sep 5 '17 at 16:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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