A one-line note 'the margin is too small', and a 358-year chase
3² + 4² = 5². Can you build such a perfect equation with cubes?
Proving that something does not exist becomes mathematics' longest journey.
Experiment
Hands-on experiment
🎯 Find whole numbers with xⁿ + yⁿ = zⁿ
Pick n, then move x and y. If z lands on a whole number, you win.
3² + 5² = 34
So close! Nearest is 6² = 36 — off by -2. z did not land on a whole number.
💡 Hint: at n=2, try x=3, y=4. (5,12) works too.
📖 Read more — why it exists · insights · common mistakes · formulasExpand ▾
Why
Why does this exist?
The Pythagorean equation x²+y²=z² has endless whole-number solutions: (3,4,5), (5,12,13), and on and on. So what happens if you raise the exponent to 3?
Around 1637, Fermat wrote in a book margin: no solutions exist for 3 and above, and he had found a marvelous proof the margin was too small to hold. That note opened the longest chase in the history of mathematics.
Proving 'there is one' ends with a single find. Proving 'there is none' must rule out infinitely many candidates. This asymmetry made the problem a 358-year one.
Insight
Insights from the video
“'There is one' takes a single find; 'there is none' takes everything.”
An existence proof ends with one example, but a nonexistence proof must handle all of infinity. The gap between you finding a solution at n=2 and mathematicians spending 358 years on n=3 is exactly this.
“Missing by a hair is not the same as not existing.”
6³+8³=728 falls exactly 1 short of 9³. Countless near misses like this exist, yet not a single exact hit. Wiles's proof settled that 'not a single one' over all of infinity.
Misconception
Common misconceptions
Checking up to a huge number by computer counts as a proof.
However far you check, the 'next number' always remains. Covering infinity takes a structural proof, not verification. That is why it took 358 years.
No solution has appeared because we haven't found it yet.
For n ≥ 3, solutions are not 'unfound' — they cannot exist. Wiles's proof ruled out every candidate at once.
Formula
Writing it as math
Write down the success you had at n=2 and the near misses at n=3, and you get the sentence Fermat left in the margin.
The Pythagorean door (n=2)
With squares, whole-number solutions are endless: (5,12,13), (8,15,17), and more. These are the solutions you found with the sliders.
Fermat's Last Theorem
Raise the exponent to just 3 and not a single solution exists. A conjecture from around 1637 became a theorem with Wiles's 1995 proof.
The agonizing near miss
Near-integer hits abound, but they always miss. No pile of 'almost right' ever adds up to one 'right'.
In Real Life
Where you meet it in real life
Why checking is not proving
Running software tests forever still cannot guarantee there are no bugs — the same structure. Where exhaustive checking is impossible, structural proof is required.
The byproduct that became the main act
Chasing this problem gave birth to entire fields like algebraic number theory, whose tools now underpin cryptography. One problem redrew the map of mathematics.
Wiles's seven years (true story)
Wiles worked in secret in his attic study for seven years and announced a proof in 1993. A flaw surfaced; he fought one more year, closed it in 1994, and published in 1995 — the dream of a boy who met this problem at age ten.
The power of one counterexample
In 1769 Euler made a Fermat-like conjecture. In 1966 a single counterexample (27⁵+84⁵+110⁵+133⁵=144⁵) demolished it. One counterexample ends a mathematical statement.
Practice
Practice — conquer the types
✏️ 4 practice problems — solve to conquerTap to solve ▾
What does Fermat's Last Theorem say?
With n=2, x=3, y=4, what is z?
Why is proving 'no solution exists' harder than proving 'a solution exists'?
A computer checks up to 10¹⁸ and finds no solution for n=3. Is the theorem proved?
Watch
Related video
Connection
Concepts connect
Previous concept
Russell's Paradox
Mathematics with rebuilt foundations now departs on a long journey to prove a 'none'.
← Russell's Paradox labLeads to next
The Riemann Hypothesis
After the problem solved in 358 years comes the million-dollar problem no one has solved yet.
Go to the The Riemann Hypothesis lab →Related
Labs worth exploring together
Related lab
Pythagorean Theorem
The n=2 door — the world where solutions pour out endlessly starts with this theorem.
Go to the Pythagorean Theorem lab →Related lab
Prime Numbers
The number-theory tools born chasing this problem meet the world of primes.
Go to the Prime Numbers lab →