Topic · Paradox

The Banach–Tarski Paradox

Cut a ball, end up with two.

A solid ball, split into a handful of pieces, can be reassembled into two solid balls each identical to the original — no stretching, no extra matter. The Explorer draws the engine behind the trick: the free group F₂ of two rotations, whose self-similar Cayley tree contains four shifted copies of itself. That branching structure is, almost literally, where the second ball comes from.

Discover the story → Open the Explorer

First encounter

Cut one ball. End up with two.

The big idea

Take a solid ball. Cut it into a handful of pieces. Move the pieces around — only rotations and translations, no stretching, no distorting. Reassemble them. You now have two solid balls, each the same size as the original. This is a theorem, not a trick. It is also, obviously, physically impossible.

A concrete example

Take a solid orange. Cut it — mathematically — into just five pieces. Not orange-slices: each piece is a dense, intertwined cloud of points threaded through the fruit. Now rotate each cloud rigidly in place, no stretching, no extra matter. The five pieces reassemble into two solid oranges, each the same size as the original and indistinguishable from it. The cut is exactly defined but physically unbuildable — no scalpel can trace it, because it lives only in the continuum.

Why it matters

The paradox is the cleanest signal that mathematical reality and physical reality are not the same thing. It depends on the Axiom of Choice — a quiet assumption most mathematicians use without thinking — and it forced a century of debate about which set-theory foundations are worth keeping. The pieces are not regions you could touch; they are infinite point-clouds with no defined volume.

Walk a free-group tree. Watch the duplication trick play out below.

Section 01 · Your intuition

Physics says no

Drop a cricket ball on the floor. Cut it cleanly. Whatever pieces fall out, their volumes have to add up to the volume of the ball. That is conservation of matter, and the world will not let you cheat it. Everything you have ever cut, ground, sliced, melted, or mixed has obeyed this rule. So when somebody tells you they can carve a ball into five pieces that reassemble into two identical balls, your gut answer is correct: nothing physical works that way. The surprise is that, in pure mathematics, something can.

Section 02 · The Axiom of Choice

Picking one of each, uncountably many times

The Axiom of Choice says: given any collection of non-empty sets, you can simultaneously pick one element from each — even if there are uncountably many sets and you have no rule for choosing. For finite or countable collections it is obvious. For uncountable collections it is an act of pure declaration: a chooser exists, full stop. Banach and Tarski use it to select one representative from each of uncountably many orbits of a rotation group acting on the sphere. From that single selection, the duplication follows.

Section 03 · Non-measurable sets

Pieces that have no volume

Lebesgue measure assigns a volume to every set you could reasonably draw. But Vitali showed in 1905 that, using the Axiom of Choice, you can construct a subset of the unit interval that admits no consistent length at all — assigning it any value, even zero, contradicts the rules of measure. The Banach–Tarski pieces are three-dimensional cousins of Vitali's set: dense point-clouds whose volume is genuinely undefined. That is why the rearrangement does not violate volume conservation — the input pieces never had a volume to conserve.

Section 04 · The free group F₂

Two rotations, no relations

Pick two rotations a and b of the unit sphere — for instance both through the angle θ = arccos(1/3), about two perpendicular axes. No matter how you compose them — aba⁻¹b, b²a⁻³, anything — you never get back to the identity except by trivially cancelling inverses. The group they generate is the free group F₂ on two letters: every reduced word in a, a⁻¹, b, b⁻¹ is a distinct rotation. F₂ has a paradoxical decomposition: split it into the four pieces of words starting with each generator. Shift the a⁻¹-piece by a and it turns into every word that does not start with a, that is the other three pieces plus the identity, so W(a) together with that shifted piece already fills all of F₂. Do the same with b and you fill F₂ a second time, so two of the four pieces rebuild the whole group and the other two rebuild it again. The sphere inherits the same property, and then the ball.

Cayley graph of F₂

current word

length · 0

Every reduced word in a, a⁻¹, b, b⁻¹ is a unique vertex, and every vertex has the same four neighbours — one per generator. Drawn with the identity at the root, the picture fans into four directions; at every further node three new branches open (the fourth would just retrace your last step). The whole infinite tree contains four shrunken copies of itself, and that self-similarity is the paradox.

Section 05 · The actual decomposition

Five pieces — and Robinson proved that is the minimum

Banach and Tarski's 1924 paper gives a construction with a finite number of pieces; later refinements brought it down to five. Raphael M. Robinson proved in 1947 that five is sharp — four pieces are not enough. One of the five is a single point at the centre, an artifact of how the rotations behave at the origin; the other four are the dense, non-measurable point-clouds carved out of the sphere by the F₂ action and pushed back into the ball. Each piece is rotated as a rigid body, with no stretching, and the result is two solid balls indistinguishable from the original.

Robinson · 1947

Minimum number of pieces. Four are dense non-measurable point-clouds carved by the F₂ action; the fifth is a single point at the centre, an artefact of rotations fixing the origin.

Section 06 · Why it isn't a contradiction

Different worlds, different rules

Physical pieces have measure: you can weigh them, measure their volume, count their atoms. Banach–Tarski pieces have no measure — they are uncountable, dense, unphysical sets. Volume conservation is a statement about measurable things, and the theorem never says the pieces are measurable; it says the opposite. Matter has atoms, set theory does not. The paradox is an exact, formal warning that the continuum has structure we cannot touch, and that the comforting equation 'volume in = volume out' is a property of nice sets, not all sets. That is the whole lesson.

See the engine

Open the Explorer.

The Explorer draws the F₂ Cayley tree explicitly: it lets you mark the four word-classes, watch the paradoxical decomposition assemble, and trace each piece back to a rotation. The maths above becomes a picture you can rotate.

→ Open the Explorer

Where you meet it

How and where this technique lives in the world today.

Set theory teaching

Banach–Tarski is the textbook example of why the Axiom of Choice is controversial — used in every graduate-level real analysis course.

Foundations of mathematics

It motivated 20th-century work on alternative set-theoretic foundations (constructivism, intuitionism) and influenced computer-verified proofs.

Philosophy of mathematics

Frequently invoked in discussions of mathematical realism, the meaning of 'infinity', and the limits of intuition.