The Banach-Tarski Paradox is a genuine, rigorously proven mathematical theorem stating that a solid ball can be decomposed into a finite number of pieces (as few as five) and reassembled, using only ordinary rotations and translations, into two solid balls — each one exactly the same size as the original. Proven by Stefan Banach and Alfred Tarski in 1924, it is not a trick or an approximation, but a genuine, logically valid consequence of the Axiom of Choice (see the ZFC Axioms entry) — and a striking illustration of how that axiom's power comes at the cost of some deeply counterintuitive consequences.
Imagine taking a solid ball, cutting it into a small number of oddly-shaped pieces, and then simply moving those pieces around (only rotating and sliding them, never stretching or adding anything) — and ending up with two complete solid balls, each exactly the same size as the one you started with. This sounds like it must violate some basic law of physics or conservation of matter. It doesn't (physical matter genuinely can't be duplicated this way) — but as a purely mathematical statement about abstract points in space, the Banach-Tarski Paradox proves this is genuinely, rigorously true.
The "pieces" the theorem requires are not solid chunks you could cut out with a knife — they are bizarre, infinitely intricate, and (crucially) non-measurable sets of points, so wildly and infinitely scattered and detailed that they don't have any well-defined notion of volume at all. Ordinary physical matter, made of a finite number of atoms, simply cannot be divided into pieces of this kind — which is precisely why you can't actually perform this trick with a real orange or a real cannonball, however sharp your knife.
The Axiom of Choice (see the ZFC Axioms entry) is enormously useful throughout ordinary mathematics, and virtually all mathematicians accept it as part of the standard axioms. Banach-Tarski is the most famous illustration of the genuinely strange consequences that come bundled together with this convenience — a vivid, concrete demonstration that accepting a seemingly reasonable axiom can commit you to some deeply weird conclusions elsewhere.
Ordinary geometric solids have a well-defined volume — a number describing how much space they occupy, satisfying natural properties (like: if you cut a shape into pieces and reassemble them without overlap, the total volume doesn't change). A non-measurable set is a set of points so pathologically irregular that no consistent notion of volume can be assigned to it at all, even in principle, without breaking these natural, seemingly essential properties. The pieces used in the Banach-Tarski decomposition are all non-measurable in exactly this sense.
Constructing the specific non-measurable pieces required for the Banach-Tarski decomposition requires making infinitely many arbitrary selections — choosing one specific point from each of infinitely many sets, with no explicit rule available for making each individual choice. This is precisely what the Axiom of Choice (see the ZFC Axioms entry) guarantees is possible. Without the Axiom of Choice, it's actually consistent (compatible with the remaining ZF axioms) that every set of real numbers IS measurable — meaning the Banach-Tarski construction genuinely requires the Axiom of Choice specifically, and cannot be carried out using only the more "modest" ZF axioms alone.
Banach and Tarski's original 1924 proof didn't specify a minimum number of pieces required. It was later shown that five pieces suffice to double a solid ball, and (remarkably) that four pieces are never enough — five is the precise, proven minimum. Even more strikingly, it's possible to decompose a ball into just finitely many pieces and reassemble them into a solid object the size of the Sun, or, going the other direction, reassemble a pea-sized ball into an object the size of the Sun using the same essential technique — sometimes summarised (slightly informally) as the "pea and the Sun" paradox.
The Banach-Tarski paradox specifically requires at least three spatial dimensions. In one or two dimensions, no analogous "doubling" decomposition using only finitely many pieces and rigid motions (rotations/translations) is possible — a fact related to a mathematical property (called amenability) that the rotation group in 3D dimensions notably lacks, but that the corresponding groups of rigid motions in 1D and 2D do possess.
Felix Hausdorff discovered a closely related, slightly less dramatic paradoxical decomposition in 1914, a full decade before Banach and Tarski's own 1924 result — involving a sphere (rather than a solid ball) with a small, countable set of "bad" points removed to avoid certain technical complications. Banach and Tarski's key further contribution was substantially strengthening Hausdorff's original construction, extending it to the full solid ball (with no points needing to be excluded) and establishing the full, maximally dramatic "doubling" result in its now-famous form.
The proof exploits a special, exotic property of the group of rotations in three-dimensional space (see the Group entry): it contains a specific free subgroup (a subgroup with no unexpected relations among its generators, other than the bare minimum forced by group axioms) generated by two independent rotations. This free-group structure is precisely what allows points on the sphere to be partitioned into a small number of pieces that can be independently rotated and reassembled to create the extra "copy" — the underlying group-theoretic freedom is the essential engine driving the entire paradoxical construction.
Robert Solovay proved in 1970 that, assuming the existence of a certain kind of large cardinal (an additional consistency assumption beyond bare ZF), there exists a mathematically consistent model of set theory in which every set of real numbers is measurable, and Banach-Tarski-style decompositions are consequently impossible. This result (building on Cohen's forcing technique — see the Continuum Hypothesis entry) rigorously confirms that the Axiom of Choice is not merely convenient for the Banach-Tarski construction, but is a genuinely necessary ingredient — dropping it (in Solovay's specific alternative model) removes the paradox entirely.
The Banach-Tarski paradox has often been invoked, both by working mathematicians and in more popular accounts, as the single most vivid and startling illustration of the genuinely non-constructive character of the Axiom of Choice — it guarantees decompositions exist without providing any way to actually describe, visualise, or construct them explicitly, since the required non-measurable sets are, in a precise sense, fundamentally beyond any possible explicit description. Some mathematicians and philosophers have cited results of this kind as at least a mild motivation for exploring set theories without full Choice, though the overwhelming mainstream mathematical consensus continues to accept full ZFC, treating results like Banach-Tarski as a fascinating and instructive curiosity about the nature of infinity, rather than as any genuine reason for concern about the axiom's soundness.