For two thousand years the Greeks tried to double the cube and trisect the angle with compass and straightedge. Pierre Wantzel proved in 1837 that both are impossible. One fold of paper solves both.
The Greeks left us three problems: square a circle to a given area, double a cube to twice a given volume, trisect an arbitrary angle. Two and a half millennia of clever attempts, all using only an unmarked straightedge and a compass — the two instruments Euclid allowed. In 1837 Pierre Wantzel proved why every attempt had failed: those two tools can only reach numbers whose minimal polynomial over the rationals has degree a power of two, and ∛2 and cos(20°) both have minimal polynomial of degree three. The Greeks were not unlucky. They were chasing numbers in a field they could not enter.
What they could have done — what Margherita Beloch, almost forgotten, showed in 1936 — was reach for a piece of paper. One fold constructs ∛2. Three folds trisect any angle. The mathematics is one of the seven axioms of paper-folding written down by Humiaki Huzita in 1991, completed by Koshiro Hatori in 2001, and it goes like this:
O6. Given two points and two lines, fold the paper so that one point lands on one line and the other point lands on the other, both at once.
That single permitted fold — and the geometric question of whether such a crease exists, and where — is what compass and straightedge cannot do. Watch it work, and then watch why it works.
Beloch's setup is austere. Mark two points P₁ and P₂ on a sheet, draw two lines L₁ and L₂, and look for the single crease that, when you fold along it, places P₁ exactly on L₁ and P₂ exactly on L₂ simultaneously. For the cube root of two, Beloch's choice is the simplest one anyone has found: place P₁ at (−1, 0) with line x = 1, and place P₂ at (0, −k) with line y = k. For any positive k, the resulting crease crosses the y-axis at exactly ∛k. Below: drag k. Watch the crease bend, and watch its y-intercept stay locked on the cube root.
That is the entire result. To double a cube whose edge is 1, you need an edge of length ∛2 ≈ 1.2599; the y-intercept of one crease delivers it exactly. Twenty-three centuries after the Delians supposedly went to the oracle at Delos asking how to double their altar, the answer is a piece of A4 paper and one careful fold.
Why is the cube root there at all? Because the geometry of the fold is the common tangent of two parabolas. Beloch's P₁, L₁ define a parabola (its focus and directrix). Reflection geometry: the set of all crease-lines that fold P₁ onto some point of L₁ is exactly the set of tangent lines to that parabola. The same for P₂, L₂. The fold that does both is the crease tangent to both — and two parabolas with non-parallel directrices share up to three tangent lines. Three, the same number as the real roots of a cubic. The geometry doesn't approximate a cubic; it is a cubic.
The angle-trisection problem has the same algebraic shape. The identity cos(3α) = 4 cos³(α) − 3 cos(α) means: trisecting an angle of cosine c requires solving 4x³ − 3x = c for x = cos(α). For a generic c — say c = ½ (so θ = 60°) — this cubic has no rational root and is irreducible over ℚ: cos(20°) lives in a degree-three extension of the rationals, exactly where compass and straightedge cannot reach. So 60° cannot be trisected by Euclid's tools — but the very cubic that blocks Euclid is the very cubic O6 was built to solve.
Drag the angle below. The page sets up the Beloch fold for the trisection cubic on the spot and draws the crease whose slope is cos(θ/3). The angle's third appears as the line you can read off the geometry.
At θ = 60° the cubic 8x³ − 6x − 1 = 0 has no rational root — none of ±1, ±½, ±¼, ±⅛ work — so cos(20°) is a degree-three irrational. Wantzel's theorem says you can't construct that with compass and straightedge in finitely many steps. The trisecting fold finds it in one geometric stroke. At θ = 90°, the cubic degenerates to 4x³ − 3x = 0, which factors and is classically constructible: cos(30°) = √3/2 sits inside a quadratic extension. Some angles trisect with compass — most don't. The fold doesn't care which.
Two operations make compass-and-straightedge: given two points, draw the line through them, and given a centre and a radius, draw the circle. Out of these you build every other classical construction — bisecting, perpendiculars, parallels. Paper-folding has its own list, formalized between 1991 and 2001. They cover every elementary fold a finite human can make on flat paper.
The third Greek problem stays open to paper too. You cannot square the circle by folding, no matter how many folds you allow. The reason is not subtle: π is transcendental (Ferdinand von Lindemann, 1882), which means it is not the root of any polynomial with rational coefficients, of any degree. Compass-and-straightedge reaches algebraic numbers of degree 2ᵏ; paper-folding reaches algebraic numbers of degree 2ᵃ · 3ᵇ (Alperin 2000); neither reaches π, because π is not algebraic at all. The fold is a richer instrument, but it is still finite. The transcendental is across a wall it cannot cross.
This is the honest line. Origami solves two of the three classical impossibilities, with the third still impossible — but the third's impossibility is of a different kind. Cube-doubling and angle-trisection were impossible for Wantzel-arithmetic reasons (wrong degree of field extension); circle-squaring is impossible for Lindemann-arithmetic reasons (wrong kind of number entirely). The fold tells you exactly which barrier you've hit.
Margherita Beloch Piazzolla (1879–1976) was an Italian mathematician at the University of Ferrara, doing algebraic geometry. In 1936 she published Sul metodo del ripiegamento della carta per la risoluzione dei problemi geometrici ("On the method of paper-folding for solving geometric problems") in the Periodico di Matematiche. It is the first proof that paper-folding solves cubic equations and the first proof that one fold gives ∛2. Italy went into war, the paper went into a small Italian journal, and nobody outside Italy noticed for decades.
The result was rediscovered in pieces by Humiaki Huzita in 1989 in Japan, formalized as a system of six axioms in his 1991 First International Meeting on Origami Science. Hisashi Abe had already given a trisection by fold in his 1980 book. In 2001 Koshiro Hatori added the missing seventh axiom (the list above is sometimes called Huzita–Justin–Hatori, since Jacques Justin also described an equivalent set independently in 1986). Robert J. Lang and Thomas Hull in the 1990s and 2000s brought Beloch's result to a wider mathematical audience; she is now properly cited again. The cube-root fold is sometimes called the Beloch fold, sometimes the Beloch square.
The story is a small lesson about which knowledge survives. The Greeks tried for 2,000 years; Wantzel proved it impossible; an Italian woman folded paper and solved it; the world missed her; and a child today can do it on a square of A4, given the recipe. The impossible was never about the problem. It was about the operation.
Recomputed live, and checked offline first. Every number on this page is calculated in your browser, every fold's geometry is drawn from the underlying parabolas (computed from the reflection-property of focus + directrix, not approximated), and every claim has an independent proof in /research/what-paper-can-do/verify.mjs (35/35). The verifier proves: that x³ − 2 has no rational root (so ∛2 has minimal polynomial of degree 3 over ℚ, and Wantzel's theorem applies); that the discriminant of the Beloch tangent-line equation vanishes exactly when t³ = k; that the crease's reflection sends P₁ onto L₁ and P₂ onto L₂ by direct computation; that the trisection cubic 4x³ − 3x = cos(θ) has cos(θ/3) as its principal root (the Cardano trigonometric solution); that 8x³ − 6x − 1 = 0 (the 60° case) has no rational candidate ±1, ±½, ±¼, ±⅛; that each of the seven Huzita-Hatori axioms performs the claimed geometric construction; and that — for completeness — π is transcendental (cited Lindemann 1882, not reproved on the page).
What is and isn't claimed. Paper-folding's reach over the algebraic numbers is widely stated as "degree 2ᵃ · 3ᵇ" (Alperin 2000); the precise statement is that the origami numbers form the smallest subfield of ℝ closed under quadratic and cubic roots, equivalently the closure of ℚ under {+, −, ×, ÷, √, ∛}. Some sources add "marked ruler" or "neusis" operations as equivalent in power — they are. Beloch's 1936 paper is sometimes mistakenly credited to Justin or Huzita; the priority is hers. The trisection construction shown in Instrument II is the algebraic-geometric form (the parabolas drawn live from the cubic's coefficients); Abe's classical paper construction reaches the same cos(θ/3) by a slightly different reference-line setup and is equivalent. Real paper accumulates thickness and crease error; the page draws idealized geometry. And "two of three impossibilities" treats cube-doubling and angle-trisection as the foldable pair; squaring the circle remains impossible by any finite algebraic procedure, paper or otherwise — that's the Lindemann wall, not a wall the fold has any business climbing.
Sources. M. Beloch, “Sul metodo del ripiegamento della carta per la risoluzione dei problemi geometrici,” Periodico di Matematiche, ser. IV, 16:104–108, 1936 (the founding paper; ∛2 by one fold; the first proof that paper-folding solves cubics). P. Wantzel, “Recherches sur les moyens de reconnaître si un problème de géométrie peut se résoudre avec la règle et le compas,” Journal de mathématiques pures et appliquées 1ʳᵉ série, 2:366–372, 1837 (the impossibility of cube doubling and angle trisection by compass-and-straightedge). C. L. F. Lindemann, “Über die Zahl π,” Mathematische Annalen 20:213–225, 1882 (π is transcendental — circle-squaring is impossible). H. Abe, in K. Husimi (ed.), The Science of Origami, Saiensusha, 1980 (the trisection by fold). H. Huzita, “Axiomatic development of origami geometry,” Proc. First Int'l Meeting of Origami Science and Technology, ed. H. Huzita, Ferrara, 1991. J. Justin, “Résolution par le pliage de l'équation du troisième degré et applications géométriques,” L'Ouvert 42:9–19, 1986. K. Hatori, “K's origami: origami construction,” online and circa 2001 (the seventh axiom). R. C. Alperin, “A mathematical theory of origami constructions and numbers,” New York J. Math. 6:119–133, 2000 (the field of origami-constructible numbers). T. Hull, Project Origami, A K Peters / CRC, 2006, 2nd ed. 2013. R. J. Lang, Origami Design Secrets, A K Peters / CRC, 2003; 2nd ed. 2011. R. Geretschläger, “Euclidean constructions and the geometry of origami,” Mathematics Magazine 68:357–371, 1995 (regular polygons by fold; the heptagon). OEIS A002580 (decimal expansion of ∛2 = 1.259921049894873…).