sciencebriefs
13:00in productionCh. 1 · What a mathematical knot actually is/ 13:00 · ceiling 15 min
Mathematics · Physics

Knot theory

A wrong theory that atoms were knotted loops of ether pushed nineteenth-century mathematicians to catalogue every possible knot — the classification survived even after the physics it was built to explain was abandoned.

Knot theory studies closed loops in space and when two of them can be manipulated into each other without cutting, a question first given rigorous mathematical footing when Carl Friedrich Gauss defined the linking integral in the nineteenth century. The field's biggest early push came from a mistaken physical theory: Lord Kelvin's proposal that atoms were knots in an all-pervading ether, which motivated Peter Guthrie Tait to build detailed tables classifying knots by their number of crossings. Kelvin's atomic theory was abandoned, but the classification project it inspired became a genuine and still-active branch of mathematics, later connected unexpectedly to statistical mechanics, quantum field theory and the study of DNA.

Chapters & takeaways6
  1. 0:08
    What a mathematical knot actually is

    A mathematical knot is a closed loop that cannot be undone without cutting, and two knots count as equivalent only if one can be manipulated into the other.

  2. 2:10
    Gauss's linking integral

    Carl Friedrich Gauss gave the field rigorous grounding in the nineteenth century by defining an integral that measures how two loops are linked.

  3. 4:20
    Kelvin's wrong idea, Tait's real tables

    Lord Kelvin's mistaken theory that atoms were knots in the ether motivated Peter Guthrie Tait's systematic tabulation of knots by crossing number from the 1860s.

  4. 6:30
    The theory that outlived its motivation

    Kelvin's vortex-atom theory was abandoned, but Tait's classification project continued and grew into an independent field of mathematics.

  5. 8:40
    From tables to invariants

    Twentieth-century mathematicians built tools like the Alexander and Jones polynomials to tell knots apart more reliably than crossing counts alone.

  6. 10:50
    Where it still matters

    Knot theory now describes how enzymes act on DNA, molecular chirality, and structures in theoretical physics, well outside its original, mistaken motivation.

Worth your time?

Yes. Study the whole thing.

4/ 5
What works
  • makes the wrong physical motivation, Kelvin's knotted atoms, as interesting as the mathematics it produced
  • traces a clear line from Gauss's linking integral to modern invariants without losing the thread
  • shows a real, still-open problem rather than presenting the field as fully solved
What does not
  • does not go deep into the mathematics of the Jones polynomial or hyperbolic invariants
  • gives only a brief account of Gauss's specific contribution relative to later figures
Study it if
  • anyone who enjoys a mathematical field with a genuinely wrong origin story
  • readers curious how a tabulation project driven by bad physics became real mathematics
  • anyone interested in how DNA topology connects back to nineteenth-century knot tables
Skip it if
  • readers wanting the technical detail of specific knot invariants like the Jones polynomial
  • anyone looking for Gauss's own knot-theory notebooks rather than a summary of his contribution
The written brief3 min read

What a mathematical knot actually is

A mathematical knot is a closed loop in three-dimensional space, distinguished from an everyday knotted shoelace by the fact that its ends are joined, so it cannot simply be pulled undone. Two knots are considered the same, or equivalent, if one can be deformed into the other through continuous manipulation in space without ever cutting the loop or passing it through itself, a relationship mathematicians call ambient isotopy. The claim knot theory makes is that this apparently informal, hands-on idea of tangling and untangling can be turned into precise mathematics: that knots can be classified, counted, and distinguished from one another using rigorous tools, rather than relying on the unreliable intuition of simply looking at a tangled diagram and guessing whether it can be simplified.

Gauss’s linking integral

Early steps toward this rigour came well before the field had a settled name. Alexandre-Théophile Vandermonde noted in 1771 that a knot’s essential features were topological rather than about exact measurements or shapes. The more decisive step came from Carl Friedrich Gauss, who in the nineteenth century defined what is now called the linking integral, a way of calculating, from the geometry of two closed loops, an integer that measures how many times they wind around each other. This gave knot theory something it had lacked: a computable, mathematically defined quantity attached to a specific knot or link, rather than only a visual impression, and it set a precedent that later invariants, tools for telling knots apart, would follow.

Kelvin’s wrong idea, Tait’s real tables

The field’s decisive push toward systematic classification, though, came from a theory that turned out to be entirely wrong. In the 1860s, Lord Kelvin proposed that atoms were knots tied in a pervasive, undetectable substance called the ether, an idea that would have made the physical properties of different elements a direct consequence of which knot each atom’s ether-loop was tied in. Taking this seriously as a research programme, Peter Guthrie Tait, working with John Kirkman and Charles Little, set out to catalogue every distinct knot by its number of crossings, publishing tables extending through ten crossings by 1885 and formulating a set of conjectures about knot structure that guided the field for a century afterward.

The theory that outlived its motivation

Kelvin’s vortex-atom theory did not survive; atoms are not knots tied in a luminiferous ether, and the concept was abandoned once atomic physics developed along entirely different lines. What did survive, and grew independently of the physical theory that had motivated it, was Tait’s tabulation project itself, which mathematicians continued refining well into the twentieth century. Errors were found and corrected along the way, including a duplicate entry in the historical tables identified by Kenneth Perko in 1974, now known as the Perko pair. By the late 1990s, mathematicians had tabulated every knot through sixteen crossings, and by 2020 all prime knots up to nineteen crossings had been catalogued, an ongoing project entirely detached from the disproven physics that first justified it.

From tables to invariants

Classification by crossing number alone eventually proved too blunt a tool, since different knots can share the same crossing count without being equivalent, so twentieth-century mathematicians developed sharper invariants to distinguish them. The Alexander polynomial, developed alongside the knot group approach in the early twentieth century, was followed in 1984 by Vaughan Jones’s discovery of the Jones polynomial, which revealed unexpected connections between knot theory and statistical mechanics and quantum field theory. Later work by figures including Edward Witten and Maxim Kontsevich extended these connections further. A basic algorithmic question, whether any given tangled diagram represents the unknot, remains only partly resolved; a candidate quasi-polynomial-time algorithm was proposed by Marc Lackenby in 2021, but the underlying computational difficulty of the general problem is still not fully settled.

Where it still matters

The appeal of this story is that it inverts the usual order of scientific progress: normally a correct theory produces a lasting mathematical framework, but here a false theory, Kelvin’s knotted atoms, produced a mathematical framework that turned out to be genuinely valuable once detached from the physics that inspired it. That framework now does real work well outside its origin, describing how the enzyme topoisomerase manages the tangling of circular DNA during replication, explaining molecular chirality in chemistry, and surfacing in the mathematics of topological quantum field theory. Anyone drawn to how mathematics can outlive and outgrow its motivating idea, rather than simply confirming it, will find this a genuinely satisfying case, even without following the technical detail of the invariants themselves.

Same field · Mathematics4 of 15
Up next in Science

Konrad Lorenz

· 11:26

Imprinting isn’t love—it’s a hardwired species-recognition switch flipped in the first hours after hatching.

11:26