MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I was thinking a bit more about the setting of my recent question about unions of chains of nowhere dense subsets of the reals and got stuck almost immediately on a follow-up question. I suspect that my difficulty is rooted in my profound ignorance of set theory, so I have abstracted the problem away from its original setting.

Suppose that $B$ is a family (set) of sets which is closed downward ($x \in B, y \subset x \Rightarrow y \in B$ - I don't expect this property to be essential) and closed under finite unions ($x,y \in B \Rightarrow x \cup y \in B$). Let $W$ be the family of all sets which are the union of some chain in $B$. I have two questions.

  1. If $x,y \in W$, does it follow that $x \cup y \in W$?
  2. If $C \subset W$ is a chain of sets, does it follow that $\bigcup C \in W$?

Some optional motivation: $B$ is intended to model the family of nowhere dense subsets of some topological space. If $M$ is the set of countable unions of sets from $B$ (ie. models the meagre subsets) then it is easy to see that $M$ inherits both of the abstracted closure properties of $B$ and is furthermore closed under countable unions. If we instead form the (larger) family $W$, I was wondering if something analogous holds. That is, I wondered if $W$ inherits both of the closure properties of $B$ (hence my 1st question - $W$ is obviously closed downward) and if it also picks up a new closure property reminiscent of the way in which we constructed it (hence my 2nd question).

I do not embarrass easily and am fully prepared to embrace the possibility that one or both of my queries has a laughably obvious answer! Thanks.

share|cite|improve this question
@Andre: Thanks, I can't say I surprised to see a negative answer for 2. This example seems optimally nice. – Michael Jan 18 '11 at 6:16
I think the first question also has a negative answer. Let $X$ be a copy of omega, and $Y$ a disjoint copy of omega_1, and let $B$ consist of all sets which intersect $X$ in a finite set and intersect $Y$ in a countable set. Both $X$ and $Y$ are in $W$, but $X\cup Y$ is not. Any chain whose union was $X\cup Y$ would have to be of uncountable cofinality; but then all of $X$ is included in some initial segment, impossible. – Justin Palumbo Jan 18 '11 at 6:42
@Justin: Hehe, I just posted the same construction. – Andrés E. Caicedo Jan 18 '11 at 6:49
@Justin, thanks you nailed it. – Michael Jan 18 '11 at 7:30
up vote 5 down vote accepted

Let $X$ be the disjoint union of a countable set $A$ and $\omega_1$, the first uncountable ordinal. Let $B$ be the family of sets of the form $a\cup c$ where $a$ is a finite subset of $A$ and $c$ is a countable subset of $\omega_1$. $B$ is closed under subsets and finite unions. The union of a chain in $B$ has the form $d\cup e$ where $d$ is a subset of $A$, $e$ is a subset of $\omega_1$, and (if $d$ is infinite then $e$ is countable).

Note that $A$ and $\omega_1$ belong to $W$, but their union $X$ does not. This proves that the answer to the first question is no.

(In general, we can get a counterexample by looking at the disjoint union of sets whose cardinalities have different cofinalities.)

Now, let $B$ be the family of finite subsets of $X$, where $X$ is uncountable. Then $B$ is closed downwards, and under finite unions. If $A$ is a chain in $B$, then the union of $A$ is countable. Conversely, any countable subset of $X$ is in $W$. Now, there is a chain of countable subsets of $X$ with uncountable union. This shows that the second question also has negative answer.

A natural question that remains is to see what happens if we assume now that the axiom of choice fails and all well-ordered cardinals have cofinality $\omega$, so the counterexample above to the first question does not work.

share|cite|improve this answer
Thanks. In retrospect, when my original naive attempt at a proof of 2 hit a stumbling point, it was because I had tripped over precisely this flavour of counterexample. Unfortunately I wasn't savvy enough to connect the dots. – Michael Jan 18 '11 at 7:42

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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