Ma · Mathematics

Does Russell’s Paradox Mean There Is No Universe of Sets?

Published By the 9to5Science editorial team

At a glance

  1. 1. Separate within a set

    The axiom forms a subset by selecting elements from a set that already exists.

  2. 2. Assume a set universe

    Suppose a set U contains every set, then apply separation to U.

  3. 3. Form the Russell set

    Select from U exactly those sets that are not members of themselves.

  4. 4. Derive the contradiction

    The resulting set belongs to U, so asking whether it belongs to itself yields a contradiction.

  5. 5. Distinguish set and class

    Ordinary set theory has a class of all sets, but that class is not itself a set.

Russell’s paradox rules out a set of all sets in standard set theory. It does not rule out speaking of the universe of sets altogether: in Zermelo–Fraenkel set theory, that universe is treated as a proper class, which is not itself a set.

The distinction matters because the axiom of specification does not let you gather objects into a set without limits. It only lets you select elements from a set that already exists.

What the axiom of specification says

The axiom of specification, also called the axiom of separation, says that if AA is a set and φ(x)\varphi(x) is a property, then the elements of AA that satisfy the property form a set:

{x∈A:φ(x)}.\{x\in A : \varphi(x)\}.

For example, if A={1,2,3,4}A=\{1,2,3,4\} and φ(x)\varphi(x) means “xx is even,” specification gives the set {2,4}\{2,4\}. The set AA is the bound on what can be collected. Specification does not say that {x:φ(x)}\{x : \varphi(x)\} is a set for any property, with no prior set to draw from.

In ZF and ZFC, specification is technically an axiom schema: there is an instance for each suitable formula φ\varphi, rather than one axiom that quantifies over every possible property.

How the Russell contradiction works

Suppose, for contradiction, that there is a set UU containing every set. Apply specification to UU using the property “xx is not a member of itself.” This produces the set

R={x∈U:x∉x}.R=\{x\in U : x\notin x\}.

Since UU contains every set, it contains RR. Now ask whether RR is a member of itself:

  • If R∈RR\in R, the definition of RR says that R∉RR\notin R.
  • If R∉RR\notin R, then RR satisfies the defining condition for membership in RR, so R∈RR\in R.

Either answer implies its opposite. In logical form, the definition gives R∈RR\in R if and only if R∉RR\notin R, which is impossible. Therefore the assumed set UU cannot exist in a theory with this form of specification.

The contradiction does not come from specification by itself. It comes from combining specification with the claim that there is a set containing every set. Without that universal set, there is no set UU from which specification can form RR.

So is there a universe of sets?

The word “universe” can mean different things. In ordinary mathematical usage, the universe of sets is the collection of all sets, often written VV. In ZF or ZFC, VV is a proper class, not a set. It is a way of talking about all sets, but it is not an object that can be placed inside another set.

The class of sets that are not members of themselves is also not paradoxical in standard ZF. With the axiom of foundation, no set is a member of itself, so this class is simply the class of all sets, VV. It is not a set. In class theories such as NBG, proper classes are not elements of classes, so the expression “is this class a member of itself?” does not work like set membership.

This conclusion depends on the axioms. Some alternative set theories restrict which properties may be used to form sets and can have a universal set. Russell’s argument shows that a universal set is incompatible with unrestricted separation of the kind used above; it does not show that every conceivable foundation for mathematics must forbid universal sets.

Common mistakes

  • Treating specification as unrestricted comprehension. Specification requires a previously given set AA. The unrestricted rule “every property defines a set” would allow the Russell construction without a bound and leads to contradiction.
  • Concluding that the collection of all sets cannot be discussed. It can be discussed as a proper class in standard set theory; it just cannot be a set.
  • Assuming the contradiction depends on self-membered sets existing. In ZF, foundation rules out x∈xx\in x for every set xx, but the Russell argument still shows that a set of all sets cannot exist. If such a set existed, its Russell subset would have to be a set and would produce the contradiction.
  • Assuming every theory must use the same rules. The result applies to standard set theories with the relevant separation principle. Alternative systems can change the rules, though their notions of set and membership may differ.

References

  1. Elements of Set Theory, Herbert B. Enderton
  2. Set Theory, Thomas Jech

Spotted an error? Send a correction. Read how we write and check articles in our editorial policy.