An amorphous set is an infinite set that cannot be partitioned into 2 infinite subsets. An amorphous cardinality is the cardinality of an amorphous set. Working in $\sf ZF$, it is consistent that amorphous sets exist.

Amorphous sets necessarily don't have a lot of structure. For example, for an amorphous set $A$, there can't be an infinite well-ordered chain of subsets of $A$, since the differences of consecutive elements in such a chain can be used to partition the set. However, if we are instead given a chain of cardinalities below $A$, $(\mathfrak c_i)_{i \in \mathbb N}$ such that $\mathfrak c_i < \mathfrak c_j < |A|$ for $i < j$, this argument does not work, because we can't uniformly choose representatives (and injections) for each cardinality.

To me, the existence of "long" chains of cardinalities below a set implies that it is "big". If there is such a sequence of length $\kappa$, then even if there isn't a literal subset of $A$ with cardinality $\kappa$, $A$ does in a sense contain at least $\kappa$ elements. I am interested in knowing whether amorphous sets can be arbitrarily "big" according to this intuition.

I have 3 questions of increasing strength:

A monotonic sequence of cardinalities of length $\alpha$ is a function $s : \alpha \to \sf Card$ such that $s(\alpha) < s(\beta)$ for $\alpha < \beta$.

  1. Given an ordinal $\alpha$, is it consistent that there exists a monotonic sequence of amorphous cardinalities of length $\alpha$?

  2. Is it consistent that for every ordinal $\alpha$ there exists a monotonic sequence of amorphous cardinalities of length $\alpha$?

  3. Is it consistent that there is a monotonic class sequence $F : \sf Ord \to \sf Card$ of amorphous cardinalities?

  • $\begingroup$ You’re more likely to get a positive answer here if you use surjective comparison $<^*$ instead. $\endgroup$ Mar 28 at 12:04
  • $\begingroup$ If you replace amorphous with merely infinite Dedekind finite, then you can hope for a positive answer. The argument of my answer doesn't work at all in this case, and really used the amorphous attribute. $\endgroup$ Mar 28 at 15:50
  • $\begingroup$ @JoelDavidHamkins Is there an obvious reason to keep asking about cardinalities instead of sets themselves if we are talking about dedekind finite sets? I don't see a reason a dedekind finite set can't have a long chain of subsets. $\endgroup$
    – Ynir Paz
    Mar 28 at 16:28
  • $\begingroup$ Yes, it definitely can. For example, there can be infinite Dedekind finite sets of reals, which have a linear order, and consequently many cuts into proper subsets. These are all different cardinalities, and so also a chain of cardinalities. But also subsets can have incomparable cardinalities. $\endgroup$ Mar 28 at 18:02
  • $\begingroup$ Whereas below an amorphous cardinal, the cardinalities are linearly ordered, of type $\omega+\omega^*$ as I explain in my answer. $\endgroup$ Mar 28 at 21:01

1 Answer 1


If $A$ is amorphous, then every subset of $A$ is either finite or cofinite in $A$. Since every cardinal below $A$ is determined by a subset of $A$, it follows from this that the cardinals below $A$ are exactly the finite cardinals and the cardinalities of the cofinite-in-$A$ sets. These cardinalities are exhaustive and distinct (since otherwise we would form a countably infinite subset of $A$), and thus form a chain of length $\omega+\omega^*$. This order type has $\alpha$-sequences for exactly those ordinals $\alpha$ at most $\omega+n$ for some finite $n$; that is, $\alpha$ below $\omega\cdot 2$. So the answer to question (1) is no.

The answer to questions (2) and (3) are similarly no, since there can be no $\omega\cdot2$th amorphous set in any chain of cardinalities.

The longest well-ordered chain of amorphous cardinalities has order type $\omega+\omega$, since if $A$ is amorphous, then so is $A+n$ for any finite $n$. So we can start with an amorphous set and systematically add increasing finite sets to it to make the second copy of $\omega$ — but the limit will not be amorphous. Every amorphous cardinality is thus part of a maximal chain of amorphous cardinalities of order type $\omega+\omega^*+\omega$, which is to say, $\omega+\mathbb{Z}$. And furthermore, every chain of amorphous cardinalities is a subchain of such a chain.


Your Answer

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

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