The Three-Body Problem

Why the three-body problem matters

For two bodies, gravity is a solved subject. Given the masses and a snapshot of positions and velocities, Newton's inverse-square law lets you write down the entire past and future in one formula: an ellipse, parabola, or hyperbola fixed by the conserved energy and angular momentum. Add a third gravitating mass and that certainty evaporates. There is no equivalent formula for where the bodies will be a thousand orbits from now — you must integrate the equations of motion numerically, step by step, and even then rounding and measurement error eventually swamp the answer. The three-body problem is where deterministic physics first collided with genuine unpredictability, three-quarters of a century before "chaos theory" had a name.

• Birthplace of chaos theory. Poincaré's 1890 study of the restricted problem revealed homoclinic tangles — the geometric fingerprint of chaos — long before Lorenz's weather model.
• Solar System stability. The Sun–Earth–Moon and Sun–Jupiter–asteroid systems are three-body problems; their long-term fate is a live research question.
• Spacecraft navigation. Lagrange points and low-energy "interplanetary superhighway" transfers come straight from the restricted three-body problem.
• Exoplanets and star clusters. Circumbinary planets, hierarchical triple stars, and gravitational encounters in dense clusters are all governed by three-body dynamics.
• Limits of prediction. It is the cleanest demonstration that "deterministic" and "predictable" are not the same thing.

How it works, step by step

The setup is disarmingly simple. Three masses m₁, m₂, m₃ sit at positions r₁, r₂, r₃. Each pulls on the others through gravity, so the acceleration of body i is the vector sum of the pulls from the other two:

a_i = Σ_j≠i G m_j (r_j − r_i) / |r_j − r_i|³

Here G = 6.674 × 10⁻¹¹ m³ kg⁻¹ s⁻² is Newton's gravitational constant, m_j the mass of the other body in kilograms, and (r_j − r_i) the separation vector in metres. That is three coupled second-order vector equations — eighteen scalar variables in three dimensions.

1. Count the conserved quantities. The system conserves total energy (1), total linear momentum (3), total angular momentum (3), and the uniform drift of the barycenter (3) — ten classical integrals. To solve by "quadrature" (reduction to integrals) you would need enough independent integrals to pin down all eighteen variables; ten is not enough, and the missing ones do not exist as analytic functions.
2. Reduce what you can. Fixing the barycenter and using conservation removes several dimensions, but you are still left with an irreducible chaotic core.
3. Integrate numerically. With no formula, you advance the state in tiny time steps using a scheme like Runge–Kutta or a symplectic integrator (which conserves energy over long runs). Modern regularized, high-order integrators can follow a triple system for thousands of orbits.
4. Watch errors explode. Two integrations started a nanometre apart stay close for a while, then peel away exponentially. Past a few Lyapunov times the two futures share nothing but their conserved totals.
5. Expect ejections. The generic long-term outcome of an equal-mass triple is disruption: two bodies pair into a tight binary and the third is flung away, often at high speed, carrying off energy and angular momentum.

The special solutions that do exist

The general problem has no formula, but specific highly symmetric configurations do. These are the "central configurations," where the gravitational forces produce accelerations proportional to positions, so the whole pattern can rotate or scale rigidly.

Solution	Discoverer & year	Geometry	Stability
Collinear (L1, L2, L3)	Euler, 1767	Three bodies in a rotating straight line	Unstable (saddle)
Equilateral (L4, L5)	Lagrange, 1772	Three bodies at corners of a rotating equilateral triangle	Stable if M/m > 24.96
Figure-eight	Moore 1993; Chenciner & Montgomery 2000	Three equal masses on one ∞-shaped path	Stable to small perturbations
Sundman series	Sundman, 1912	Convergent power series (any non-collinear start)	Correct but converges impossibly slowly

The Lagrange points are the most consequential. In the restricted three-body problem — two massive bodies on a circular orbit plus one massless test body — five points let the test body co-rotate in a fixed relationship with the pair. The three collinear points L1, L2, L3 are saddle-unstable, so a spacecraft parked there needs regular station-keeping burns. The two triangular points L4 and L5, leading and trailing the smaller mass by 60°, are genuinely stable provided the mass ratio of the primaries exceeds 24.96 to 1. The Sun–Jupiter ratio is roughly 1047, so Jupiter's L4 and L5 trap thousands of Trojan asteroids; the Sun–Earth L2 point, 1.5 million km beyond Earth, hosts the James Webb Space Telescope.

Deterministic chaos and sensitive dependence

The heart of the problem is that determinism does not imply predictability. Formally, if two nearby trajectories start a distance δ₀ apart in phase space, their separation grows as

δ(t) ≈ δ₀ · e^λt

where λ is the largest Lyapunov exponent. When λ > 0 the system is chaotic; its reciprocal 1/λ is the Lyapunov time, the horizon beyond which prediction fails. For a typical bound triple this can be a few orbital periods; for our own Solar System, Laskar's simulations give a Lyapunov time of about 5–10 million years. Poincaré discovered this while competing for a prize honouring King Oscar II of Sweden in 1889: correcting an error in his own submission, he found that the stable and unstable manifolds of a periodic orbit could intersect in an infinitely folded "homoclinic tangle" so intricate he declined to even draw it. That tangle is the mechanism of three-body chaos.

A worked example: the Earth–Moon Lagrange criterion

The stability boundary for the triangular points is one of the few clean numbers in the whole subject. The equilateral solutions L4 and L5 are linearly stable exactly when

(m₁ + m₂)² ≥ 27 · m₁ m₂

Solving for the mass ratio gives the critical value M/m ≈ 24.9599, often rounded to 24.96. Check the real systems: the Sun–Jupiter ratio (≈ 1047) and the Sun–Earth ratio (≈ 333,000) both clear it comfortably, which is why Trojan populations survive for billions of years. The Earth–Moon ratio is about 81.3 — also above 24.96 — so the Earth–Moon L4 and L5 points are stable too, and faint concentrations of dust (the disputed Kordylewski clouds) may linger there. A hypothetical binary with a mass ratio below 24.96, however, would have unstable triangular points and could not hold Trojans at all.

Common misconceptions

• "The three-body problem has no solution." It has no general closed-form solution; Sundman gave a convergent series in 1912, and numerical integration solves any specific case to any desired accuracy for a finite time.
• "Chaos means randomness." The dynamics are perfectly deterministic. Chaos is exponential sensitivity to initial conditions, not the absence of rules.
• "Newton solved it." Newton solved the two-body problem and only approximated the Sun–Earth–Moon system; he never claimed a general three-body solution.
• "Lagrange points are just gravity balance points." They balance gravity plus the centrifugal term in the rotating frame; L1–L3 sit where those combine to zero and are unstable, not points of stable equilibrium.
• "More bodies are always harder." Statistically true, but the many-body problem of a smooth galaxy or star cluster is often easier to model with statistical mechanics than a lone chaotic triple.
• "The figure-eight is common in nature." It is a razor-thin set of initial conditions; no confirmed astronomical system follows it.