The Jordan Curve Theorem: The Obvious Claim That Stumped Mathematicians for Decades

Draw a wobbly, looping circle on a napkin — nothing fancy, just a single closed line that never crosses itself. Now ask the question a child could ask: is this dot I just made inside the loop, or outside it? You already know the answer just by looking. It seems too obvious to be worth proving. And yet this "obvious" fact defeated some of the sharpest minds in mathematics for the better part of a century, spawned a proof so long a computer needed 44,000 lines of code and 20 million logical steps to fully check it, and quietly underpins everything from the GPS app that tells you when you've left a delivery zone to the software that renders video game worlds.

This is the story of the Jordan curve theorem — a statement so self-evident that stating it feels almost silly, and so hard to prove rigorously that its own discoverer got it wrong.

The Concept

The Jordan curve theorem says this: any simple closed curve in a flat plane — a loop that starts and ends at the same point and never crosses itself — divides the plane into exactly two regions. One is the "inside," a bounded region you could fence in. The other is the "outside," which stretches away to infinity. And critically, the curve itself is the complete boundary between them: to get from a point inside to a point outside, you have to cross the curve.

Picture a rubber band dropped on a table, twisted into any shape you like — a circle, a five-pointed star, a scribbled blob with a hundred wiggles — as long as it never crosses itself. The theorem guarantees that "inside" and "outside" are well-defined, unambiguous concepts for that shape, no matter how tangled it looks.

Why would anyone need to prove that? Because in mathematics, "obviously true" and "provably true" are different things, and the gap between them is exactly where interesting ideas live. A circle or a square is easy: you can see the inside and outside at a glance. But the theorem has to hold for any simple closed curve — including curves so jagged and convoluted that no human could trace them by eye, and even curves that are continuous but nowhere smooth, with infinitely many wiggles packed into a finite space. For those, "just look at it" stops working, and you need an actual argument.

Why It Matters

The theorem was first stated by the French mathematician Camille Jordan in the third volume of his textbook Cours d'analyse, published in 1887. Jordan — full name Marie Ennemond Camille Jordan, born in Lyon in 1838 — was already a towering figure in mathematics, largely for his foundational work in group theory. But his proof of the curve theorem that now bears his name was, for over a century, considered incomplete. Critics argued he'd glossed over the hardest case: proving the result for simple polygons, the building block the rest of his argument depended on.

It took until 1905 for the American mathematician Oswald Veblen to publish what was long regarded as the first fully rigorous proof, in the Annals of Mathematics. Interestingly, more recent historical scholarship has pushed back on the "Jordan got it wrong" narrative — several mathematicians who've re-examined the original Cours d'analyse argue Jordan's proof was more complete than he's been given credit for, and that the century of criticism was itself a little too casual with the facts. Either way, the theorem's reputation as deceptively hard to prove stuck, and it became a favorite example in mathematics of a statement where intuition sprints ahead of rigor.

That reputation got a dramatic exclamation point in January 2005, when mathematician Thomas Hales completed a fully machine-verified proof of the Jordan curve theorem using the HOL Light theorem-proving system. The formal proof ran to 138 definitions, 1,381 lemmas, and roughly 59,000 lines of code, compiling down to about 20 million primitive logical inferences — every one of them checked by computer, with zero appeals to human intuition about what "inside" a loop obviously means. Later that same year, an independent team using a different formal system called Mizar completed their own machine-checked version. Two independent computer-verified proofs of a fact any toddler with a crayon would call obvious, arriving in the same year — a nice reminder that "obvious" and "trivial" are not the same thing.

The Details

So what actually makes the proof hard? The difficulty is almost entirely about how wild the word "curve" is allowed to be. The theorem doesn't just apply to nice, smooth loops — it applies to any curve that's merely continuous, meaning it can be drawn without lifting your pen, but with no requirement that it be smooth, differentiable, or made of straight segments. Mathematicians have constructed curves that are continuous everywhere but have infinitely many corners packed into every stretch, no matter how small you zoom in. The Koch snowflake is one gentle relative of this idea — infinitely jagged, yet still a single unbroken loop enclosing a well-defined area. For curves like that, the human eye's confident "that's clearly inside" stops being a proof; you need machinery robust enough to handle infinite pathology.

The standard modern proofs lean on tools like topology's Brouwer degree, or clever combinatorial arguments that first nail down the result for polygons (curves made of straight segments, which are easier to reason about) and then approximate wilder curves by polygons with more and more sides, taking a limit. That "start simple, then approximate the general case" strategy is a recurring trick throughout analysis — but making the approximation argument airtight, so that no edge case slips through, is exactly the kind of bookkeeping that swallowed Veblen's career-defining paper and, a century later, Hales's tens of thousands of lines of formal code.

The theorem also turns out to be a load-bearing wall for other results you might not expect. It's a close cousin of the Brouwer fixed-point theorem (which guarantees, among other things, that if you stir a cup of coffee, some point in the liquid ends up exactly where it started). It's essential to making rigorous sense of contour integration in complex analysis, where mathematicians integrate functions "around" a closed loop and need to know, without ambiguity, which side of the loop counts as the interior. And a discrete cousin of the idea — the Jordan curve theorem for graphs drawn on a grid — shows up in the guts of computer science.

Which brings us to the applications that touch daily life. Every time a mapping app determines whether your GPS coordinates fall inside a school district, a flood zone, or a food-delivery radius, it's running some version of a point-in-polygon algorithm — commonly a "ray casting" test, where the software draws an imaginary ray from your location out to infinity and counts how many times it crosses the boundary. An odd number of crossings means you're inside; an even number means you're outside. That simple crossing-parity trick is a direct, algorithmic descendant of the Jordan curve theorem, and it's the same logic — just applied on a computer instead of a napkin — quietly running behind geofencing, congressional-district lookups, video game collision detection, and the "clip this region" tools in image-editing and CAD software.

There's something satisfying about that arc: a claim so obvious it took over a century of arguing (and eventually a computer running twenty million logical steps) to fully nail down, ending up as one of the more practical, invisible workhorses of modern software.

Takeaways

• The Jordan curve theorem states that any simple closed curve in the plane splits it into exactly two regions — an inside and an outside — with the curve itself as the complete boundary between them.
• It was first stated by Camille Jordan in 1887; his proof was long considered flawed, and Oswald Veblen supplied the first widely accepted rigorous proof in 1905 — though modern scholars have partly rehabilitated Jordan's original argument.
• The difficulty isn't about circles or squares — it's about curves so jagged and pathological that "just look at it" stops being a valid argument, forcing mathematicians toward formal, limit-based reasoning.
• In January 2005, Thomas Hales produced a full machine-checked proof in HOL Light (roughly 20 million logical inferences); an independent Mizar-system proof followed the same year.
• The theorem's descendants are everywhere in software: point-in-polygon and ray-casting algorithms used in GPS geofencing, GIS, video games, and image processing all lean on the same "inside vs. outside" guarantee.

Resources: - Jordan curve theorem — Wikipedia - Camille Jordan — MacTutor History of Mathematics - Thomas Hales, "The Jordan Curve Theorem, Formally and Informally"