The Fundamental Theorem of Curves

From Curve to Numbers, and Back Again

So far in this rung the arrow has always pointed one way. You started with a curve, parametrized it, walked along it at unit speed, and extracted numbers from it: the curvature kappa telling you how sharply it bends, and the torsion tau telling you how fast it twists out of its current plane. Geometry flowed from the curve to the numbers. The fundamental theorem of curves dares to reverse that flow. It claims the numbers come first: give me kappa and tau as functions of arc length, and I will hand you back the curve they describe.

Stated carefully: let kappa(s) > 0 and tau(s) be any smooth functions defined on an interval of arc length s. Then there exists a curve, parametrized by arc length, whose curvature is exactly that kappa(s) and whose torsion is exactly that tau(s). Moreover the curve is unique up to a rigid motion — any two curves with the same kappa and tau differ only by a translation and a rotation. You can slide the answer across space and spin it around, but its intrinsic shape is locked. That phrase "up to a rigid motion" is the whole heart of the theorem, so hold onto it.

Why It Works: The Frenet Equations Are a Recipe

The previous guide handed you the Frenet-Serret formulas, and they are the engine that makes this theorem run. Recall what they say: as you walk along the curve, the three frame vectors — the tangent T, the normal N, and the binormal B — keep turning, and the formulas tell you exactly how their turning rate depends on kappa and tau. Read the right way, those equations are not just a description. They are a set of instructions for building the frame step by step.

T' =        kappa N
N' = -kappa T          + tau B
B' =          -tau N

then recover the curve:   r(s) = integral of T(s) ds

Here is the logic, stripped bare. The three formulas express the derivatives T', N', B' purely in terms of the current T, N, B and the known numbers kappa(s), tau(s). That is exactly the form of a linear system of differential equations. A foundational fact about such systems — the existence-and-uniqueness theorem for ordinary differential equations — guarantees that once you fix the frame at a single starting point, the whole frame is determined for all later s, and determined uniquely. So kappa and tau pin down the frame; the frame's tangent T points the way; and integrating T traces out the position. The curve has nowhere left to hide.

What the Two Freedoms Mean

When the theorem says "unique up to a rigid motion," it is being precise about exactly two freedoms, and they correspond to two choices you made when you started integrating. The first freedom is where you begin: you had to drop the starting point r(0) somewhere in space, and a different starting point just slides the whole curve over without changing its shape — that is a translation. The second freedom is how the starting frame is oriented: you had to aim the initial T, N, B in some directions, and a different aiming spins the whole curve about that point — that is a rotation.

A translation plus a rotation is precisely what we call a rigid motion: it moves an object around without bending, stretching, or reflecting it. So the leftover ambiguity in the theorem is not vagueness — it is the only ambiguity geometry could possibly allow. Two helices with identical kappa and tau really are the same helix, just photographed from a different spot with the camera held at a different angle. Their kappa and tau, being rates of bending and twisting, never noticed where the curve sat or which way it faced.

Reading Shapes Straight Off Kappa and Tau

The real delight of the theorem is that you can often skip the integration entirely and read the curve's character straight from its two functions. Constant kappa with zero tau? A curve that bends at a fixed rate and never leaves its plane — that is a circle, of radius 1/kappa. Both kappa and tau constant and nonzero? Steady bending plus steady twisting gives the circular helix, the spring-shaped curve, and its radius and pitch are fixed by the two constants. The simplest case of all, kappa = 0 everywhere, means no bending at all — a perfectly straight line.

1. kappa = 0 (so tau is irrelevant): no bending — a straight line.
2. kappa = constant > 0 and tau = 0: bends steadily, stays flat — a circle of radius 1/kappa.
3. kappa = constant > 0 and tau = constant != 0: bends and twists steadily — a circular helix, its handedness set by the sign of tau.
4. tau = 0 everywhere (but kappa may vary): the curve never twists out of one plane — a plane curve.

That fourth line deserves a star: tau = 0 everywhere is the exact test for a curve to be flat, meaning it lives entirely in a single plane. It makes perfect sense once you remember what torsion measures — the rate at which the curve twists out of its current osculating plane. If that rate is zero forever, the osculating plane never changes, and the whole curve is trapped inside it. This is the kind of clean equivalence the fundamental theorem makes available: a geometric property (being planar) becomes a one-line condition on a single function (tau = 0).

Honest Caveats, and Where This Leads

A few honest fine-print items, because the theorem is clean but not magic. First, it asks for kappa(s) > 0 — strictly positive. Where the curvature drops to zero the normal N is not well defined (there is no preferred direction to bend toward), so the standard Frenet frame stutters and the clean statement needs patching at those points. Second, "unique up to a rigid motion" really does leave that ambiguity in place: the theorem will never tell you where the curve sits, only what shape it has. And third, the existence half rests on the existence theorem for differential equations, whose full proof belongs to a later analysis course — we are taking that engine on faith here, honestly and on purpose.

Step back and feel how big this idea is. We have a complete classification of curves up to rigid motion, encoded in two simple functions — and the same spirit will return, far more powerfully, when you reach surfaces. There, a single number per point will not be enough; the bending of a surface needs a richer bookkeeping, and the question of which data determine a surface up to rigid motion becomes the deep fundamental theorem of surfaces. The curve case you have just met is the clean, fully-understood rehearsal for that grander story.

There is one last specialization worth previewing, and it is the subject of the next and final guide in this rung. Drop to the flat plane, where tau is always zero, and the lone surviving descriptor is the signed curvature — curvature that remembers whether you are turning left or right. A single signed function then determines a plane curve up to a rigid motion of the plane, and integrating that turning around a closed loop produces a whole number, the turning number, with a beautiful theorem attached. The fundamental theorem of curves is what makes all of that possible.