The Empty Set

The empty set is not a museum piece confined to introductory set theory; it is structurally load-bearing across mathematics and its applications. In foundations of arithmetic, the von Neumann construction defines $0 := \varnothing$, $1 := \{\varnothing\}$, $2 := \{\varnothing,\{\varnothing\}\}$, and so on, literally building the natural numbers as an infinite tower of sets generated from $\varnothing$ alone — the tower explored concretely in Worked Example 2. In topology, $\varnothing$ is required by definition to be both open and closed in every topological space (it vacuously satisfies the open-set axioms, exactly as $\varnothing \subseteq A$ was proved vacuously above), a fact essential to the definition of connectedness. In probability theory, $\varnothing$ represents the impossible event in a sample space, with $P(\varnothing) = 0$ built into the Kolmogorov axioms. In computer science and databases, empty sets and empty lists are the base case of essentially every recursive algorithm operating on collections — searching, sorting, and traversing all terminate their recursion at an empty structure. In formal logic and automated theorem proving, the empty set underlies the semantics of "vacuously true" universally quantified statements, which recur constantly in verifying edge cases of software and hardware correctness proofs. Even in everyday combinatorics, the empty set is the unique object counted by "$0$ choose $0$ equals $1$" reasoning (there is exactly one way to select nothing from anything: choose the empty subset), which anchors binomial coefficient identities and the base case of the power set construction, since $\varnothing$ is always among the subsets of any set $A$, and $|\mathcal{P}(A)| = 2^{|A|}$ starts from $\mathcal{P}(\varnothing) = \{\varnothing\}$, itself a set of size $1$.

Ask the tutor: Why does the empty set have to be unique, and what would go wrong in mathematics if it weren't?

Think of a set as a container and its elements as the contents. Almost every container you can imagine — a box of marbles, a set of even numbers, a set of students who passed an exam — has some description that could, for a particular collection of objects, be satisfied by nothing. "The set of students who scored above 105% on the exam" is a perfectly meaningful description; it just happens to describe a container with nothing inside. The empty set formalizes this: it is what you get whenever a legitimate defining condition turns out to be satisfiable by no object at all. The crucial conceptual jump is that the empty container is still a container — it is a genuine set, an object in its own right, distinguishable from "no set" or "undefined." A second important picture is the subset relation. Recall $A \subseteq B$ means "every element of $A$ is an element of $B$," formally $\forall x\,(x \in A \Rightarrow x \in B)$. When $A = \varnothing$, this conditional's hypothesis $x \in \varnothing$ is always false, so the conditional is always true — regardless of what $B$ is. This is the logical principle of vacuous truth: a universally quantified conditional with an unsatisfiable hypothesis is true by default. Visualize this with a nested-boxes diagram: $\varnothing$ sits inside every box because it has no contents that could ever stick out.

The right side of the figure records a common trap: $\varnothing$ has zero elements, but $\{\varnothing\}$ has exactly one element, namely $\varnothing$ itself. Conflating the two is a frequent and consequential beginner error, since it is the difference between "an empty box" and "a box containing an empty box."

Example 1: Specifying $\varnothing$ from an unusual set with an unusual condition. Let $A = \{2, 4, 6, 8, 10\}$, and consider

We verify directly, by checking every element of $A$, that $S = \varnothing$, and separately that $S$ equals the empty set defined abstractly, confirming the uniqueness theorem is not just formal but computationally checkable. Check each element of $A$ against the condition "$x$ is odd and $x > 3$":

• $x=2$: even, fails "$x$ is odd." Condition false.
• $x=4$: even. Condition false.
• $x=6$: even. Condition false.
• $x=8$: even. Condition false.
• $x=10$: even. Condition false.

No element of $A$ satisfies the condition, so by Specification, $S$ has no elements: $\forall x\, (x \notin S)$. By the Uniqueness half of the theorem above, since $S$ satisfies the defining property of the empty set, $S = \varnothing$ — the very same object as $\{x \in A : x \ne x\}$ from the proof, even though the conditions defining them ("odd and greater than $3$" versus "$x \ne x$") look completely different and were checked against completely different sets $A$. This is the payoff of uniqueness: emptiness is a property of the resulting object, not of the syntactic description, so any two descriptions that produce no elements produce literally the same set. As a sanity check on the subset corollary: is $S \subseteq A$? Yes — vacuously, since $S = \varnothing$ and $\varnothing \subseteq A$ for every set $A$, in particular for this specific $A = \{2,4,6,8,10\}$.

Example 2: Distinguishing $\varnothing$, $\{\varnothing\}$, and $\{\{\varnothing\}\}$ via a cardinality and membership audit. Define three sets:

We show these are pairwise distinct by exhibiting, for each pair, an object that is a member of one but not the other (which by the contrapositive of Extensionality — if the sets were equal they'd have identical membership — proves inequality). Number of elements. $X_0$ has $0$ elements (it's $\varnothing$ by definition). $X_1$ has exactly $1$ element, namely $\varnothing$: indeed $\varnothing \in X_1$ since $X_1 = \{\varnothing\}$ literally names $\varnothing$ as its sole member, and nothing else is in $X_1$ because a set built by the pairing/singleton construction $\{a\}$ contains precisely $a$ and nothing more. $X_2$ likewise has exactly $1$ element, namely $\{\varnothing\} = X_1$. $X_0 \ne X_1$: We have $\varnothing \in X_1$ (shown above) but $\varnothing \notin X_0$, since $X_0 = \varnothing$ has no elements at all, so in particular $\varnothing$ (as a candidate element) is not in it. Since $X_1$ has a member that $X_0$ lacks, by Extensionality $X_0 \ne X_1$. $X_1 \ne X_2$: We ask whether $\varnothing \in X_1$ versus $\varnothing \in X_2$. We showed $\varnothing \in X_1$. Is $\varnothing \in X_2$? The single element of $X_2$ is $\{\varnothing\}$, i.e. $X_1$, not $\varnothing$ itself. Is $X_1 = \varnothing$? No — we just showed $X_1 \ne X_0 = \varnothing$. So the element of $X_2$ (namely $X_1$) is not equal to $\varnothing$, meaning $\varnothing \notin X_2$. Since $\varnothing \in X_1$ but $\varnothing \notin X_2$, Extensionality gives $X_1 \ne X_2$. $X_0 \ne X_2$: $X_2$ has one element ($X_1$) while $X_0$ has none, so trivially $X_0 \ne X_2$ by the same cardinality argument as before, or by transitivity-style reasoning ($X_0 \ne X_1$ and $X_1 \ne X_2$ don't by themselves force $X_0\ne X_2$ logically, so we check directly: $X_1 \in X_2$ but $X_1 \notin X_0=\varnothing$, giving inequality by Extensionality). This example demonstrates concretely that "nesting" the empty set inside braces generates an infinite tower of pairwise distinct sets $\varnothing, \{\varnothing\}, \{\{\varnothing\}\},\ldots$ — a fact used later to build the von Neumann construction of natural numbers out of nothing but $\varnothing$ and the operation "form the singleton" or "form the successor."

The empty set as a formal object crystallized during the birth of axiomatic set theory in the late nineteenth and early twentieth centuries, most directly through the work of Ernst Zermelo, who in 1908 gave the first widely accepted axiomatization of set theory (later refined by Abraham Fraenkel into what we now call ZF/ZFC). Zermelo's axioms needed a guaranteed starting object to get the theory off the ground without presupposing an infinite universe already populated with sets — the empty set served exactly that role, alongside pairing and union, as one of the minimal generative tools from which every other set is built. The notation $\varnothing$ itself (a slashed zero) is credited to the Bourbaki group of French mathematicians in the mid-twentieth century, who deliberately chose a symbol distinct from the numeral $0$ and from the Scandinavian letter Ø, to avoid exactly the confusions discussed in this article.