Unknotting Number Is Not Additive: The 2025 Proof
Quick answer: The unknotting number of a knot is the fewest times you must pass the string through itself to untie it. For about 90 years, mathematicians conjectured that combining two knots end to end adds their unknotting numbers. In June 2025, Mark Brittenham and Susan Hermiller disproved this: the (2,7)-torus knot has unknotting number 3, but combined with its mirror image it can be untied with at most five crossing changes, fewer than the six additivity predicts.
Tie a knot in an extension cord, then plug the ends together so it cannot be slipped off. How many times would you have to cut the cord, pass it through itself and rejoin it to make it a plain loop? That number is the knot’s unknotting number, and it is one of the simplest-sounding and hardest-to-compute quantities in knot theory.
What Is the Unknotting Number?
A mathematical knot is a closed loop in space. A crossing change lets one strand pass through another at a crossing. The unknotting number \(u(K)\) is the minimum number of crossing changes, over all ways of drawing the knot, needed to turn \(K\) into the unknot (a simple circle).
| Knot | Crossings | Unknotting number |
|---|---|---|
| Unknot | 0 | 0 |
| Trefoil | 3 | 1 |
| Figure-eight knot | 4 | 1 |
| (2,5)-torus knot | 5 | 2 |
| (2,7)-torus knot | 7 | 3 |
What Was the Additivity Conjecture?
The connected sum \(K_1 \# K_2\) of two knots is made by cutting each loop open and joining the ends, so the two knots sit side by side on one string. The idea goes back to work of Hilmar Wendt in 1937 (the first explicit formulation is hard to pin down), and it seemed obvious that untying them separately is the best you can do:
\[ u(K_1 \# K_2) = u(K_1) + u(K_2). \]
It is easy to see that the left side is at most the right side: just untie each part separately. The conjecture said you can never do better. It was known to hold in many special cases.
The 2025 Counterexample
On June 30, 2025, Mark Brittenham and Susan Hermiller of the University of Nebraska–Lincoln posted “Unknotting number is not additive under connected sum.” Using a large computer search with the program SnapPy, they looked for clever sequences of crossing changes in complicated diagrams. They found that for \(K\) the (2,7)-torus knot, with \(u(K) = 3\), and its mirror image \(\bar K\):
\[ u(K \# \bar K) \le 5 < 6 = u(K) + u(\bar K). \]
Their construction starts from a 20-crossing diagram of the combined knot, not the simplest picture of each half. Intuitively, in a clever diagram one crossing change can help with both halves at once. The authors give explicit diagrams and verification code so anyone can check the sequence, and an independent paper by Chao Wang and Yimu Zhang verified the inequality directly. Note that this proves the unknotting number is at most five; it does not determine the exact minimum.
The original paper already gives infinitely many counterexamples. A January 2026 preprint by the same authors adds more, built from the figure-eight knot and the (2,5)-torus knot; the trefoil case remains only a candidate.
Why It Matters
- It overturns a long-held belief that guided knot theory for nearly a century.
- It removes a shortcut: knowing the unknotting numbers of two knots does not in general determine the unknotting number of their connected sum.
- It is a model of computer-assisted discovery: a SnapPy search found the example, and explicit diagrams let humans check it.
Common Misconceptions
“Knot theory is just about rope.” It is used in DNA biology, chemistry and physics, wherever long strands tangle.
“The counterexample needs a huge knot.” Each half has just 7 crossings in its simplest form, though the proof uses a more elaborate 20-crossing diagram of the combination.
Watch: The Unknotting Number in Video
If you learn better by watching, these videos are worth your time.
Check Your Understanding
Five quick questions to see how much of this has stuck. Pick an answer and press Submit; you will see a short explanation either way.
Answer each question and press Submit to see your result right away.
Frequently Asked Questions
What is the unknotting number of a knot?
It is the minimum number of crossing changes, passing one strand through another, needed to turn the knot into a simple unknotted loop, minimized over all diagrams of the knot.
Is the unknotting number additive?
No. In June 2025, Mark Brittenham and Susan Hermiller showed that the (2,7)-torus knot, with unknotting number 3, joined to its mirror image can be untied with at most five crossing changes, fewer than six. Their paper gives infinitely many such counterexamples.
Who disproved the unknotting number conjecture?
Mark Brittenham and Susan Hermiller of the University of Nebraska–Lincoln, in a paper posted on June 30, 2025.
How long had the additivity conjecture been open?
It dates back to the 1930s, so it stood for nearly 90 years.
Keep Exploring
- Einstein Tile Explained: Hat, Spectre and 3D Chair44
- Dudeney’s Dissection: Triangle to Square in 4 Pieces
- Lean Theorem Prover: Lean 4 and Mathlib Explained
Sources and Further Reading
Related to This Article
More math articles
- Reconstruction Amendments
- Top 10 Tips You MUST Know to Retake the CBEST Math
- Top 10 Tips to ACE the ISEE Math Test
- Representative samples
- Qualitative, Content, and Existing-Record Methods
- Revolution in China and the Korean War
- Descriptive and substantive representation
- Fifth-Grade Math in Texas (STAAR): 49 Free Printable Worksheets, One Concept Per Page
- The Nervous System
- Ageism and Population Aging


What people say about "Unknotting Number Is Not Additive: The 2025 Proof - Effortless Math"?
No one replied yet.