Start with 3² + 4² = 5²: nine plus sixteen equals twenty-five. Change the little twos to threes, and the equality breaks. Could different numbers make it work?
Fermat’s Last Theorem states that an + bn = cn has no solutions in positive integers a, b, and c when n is an integer greater than 2. A short sentence, but ruling out every possible choice took centuries. The eventual answer came from mathematics that barely resembled the original equation.
What Does Fermat’s Last Theorem Actually Say?
Positive integers are the counting numbers: 1, 2, 3, and so on. No zero, negative numbers, or fractions. The small raised number, called an exponent, tells you how many copies of a number to multiply. Thus 3² means 3 × 3, while 3³ means 3 × 3 × 3. These are called a square and a cube.
For squares, the equation works with 3, 4, and 5. It also works with 5, 12, and 13: 25 + 144 = 169. These groups are Pythagorean triples, associated with the side lengths of right-angled triangles.
But 3³ + 4³ = 27 + 64 = 91, whereas 5³ = 125. That failed attempt is not a proof. It only eliminates one choice. Fermat’s claim rules out every positive-integer triple for cubes, fourth powers, fifth powers, and every larger integer exponent.
The restriction to integers matters. Allow real numbers and 1³ + 1³ = (∛2)³ works perfectly: both sides equal 2. Here ∛2 means the positive number whose cube is 2, approximately 1.26. It is not an integer, so this is not a counterexample.
Allowing zero would change things too: 0³ + 2³ = 2³. The theorem deliberately excludes that shortcut. Its conditions are part of the claim, not technical small print.
The Note in the Margin That Started a Centuries-Long Puzzle
Pierre de Fermat was a French lawyer and government official who also pursued mathematics. In his copy of Diophantus’s Arithmetica, he wrote that higher powers could not be split into two powers of the same kind. He claimed a proof but said the margin was too small to hold it.
The note is conventionally dated around 1637, not to its publication. Fermat died in 1665; his son Samuel published the annotated edition in 1670. The MacTutor biography documents that posthumous publication. The exact writing date should not be treated as a securely dated journal entry.
“Last” does not mean the last theorem Fermat wrote. It refers to the assertion that remained unproved after his other celebrated claims had been settled. The Abel Prize’s account of the margin note explains the name and reproduces the original Latin.
No valid general proof by Fermat is known. That does not establish that he lied, nor does the eventual truth of the theorem establish that his reasoning was sound. A correct conclusion and a correct proof are different things.
Why Was Such a Simple Equation So Hard to Prove?
Checking an example asks whether one calculation balances. Proving the theorem asks whether any calculation of the permitted kind could ever balance. There is no largest positive integer at which the search can stop.
Even proving the entire case n = 3 would not automatically settle n = 5. A method might depend on a property of cubes that fifth powers do not share. Earlier mathematicians nevertheless made substantial progress on particular exponents and broader families of cases. Centuries of work were not centuries of standing still.
There are useful shortcuts between exponents. For instance, a hypothetical solution a⁶ + b⁶ = c⁶ would also give (a²)³ + (b²)³ = (c²)³. Since the squares of positive integers remain positive integers, it would produce a solution for cubes. Ruling out cubes therefore rules out sixth powers too. Such reductions shrink the task, but this example alone does not cover every exponent.
A computer testing enormous numbers still faces the same logical gap unless a separate argument proves its search exhaustive. Computers can assist rigorous proofs; finite trial alone does not establish an unbounded claim. Finding one counterexample would disprove the theorem, but failing to find one is a different kind of evidence.
Andrew Wiles and the Long Road to a Proof
Andrew Wiles encountered the problem at ten through E. T. Bell’s The Last Problem in a public library. Oxford’s account of his Abel Prize acceptance records that recollection. As an adult, he developed the specialist knowledge needed to approach it through elliptic curves and modular forms.
After years of largely private research at Princeton, Wiles announced a proof in Cambridge in June 1993. Scrutiny uncovered a gap. The announcement was therefore not the end of the verification process.
Wiles worked with Richard Taylor on the difficulty, and a corrected argument emerged in 1994. Publication followed in 1995: Wiles’s main paper and a companion paper jointly written with Taylor appeared in Annals of Mathematics. Taylor’s contribution was part of the mathematics that made the repair work, not merely a final proofreading pass.
This distinction between announcement, correction, and publication explains why different dates appear in accounts of the proof. It also shows why checking matters. A difficult argument becomes reliable through its reasoning and examination, not the reputation of the person announcing it.
How Did the Proof Work—Without the Advanced Math?
Two Mathematical Worlds Became Connected
An elliptic curve is a mathematical object that can be described by a cubic equation in two variables, subject to conditions that exclude singularities such as cusps. Despite the name, it is not simply an ellipse. One question is which points on it have coordinates that are fractions of integers.
Modular forms are functions of a complex variable that obey exceptionally restrictive symmetry rules. Rather than merely repeating a pattern, they transform in prescribed ways when their inputs undergo certain changes. Those rules make them powerful sources of arithmetic information.
The modularity conjecture proposed a precise connection between elliptic curves over the rational numbers and modular forms. As an analogy, imagine two different recording systems encoding matching information. That image suggests why a connection might help; it is not a definition or a proof of modularity.
If unexpected connections appeal to you, our article on Euler’s formula explores another meeting of mathematical languages. It requires more algebra, but it offers a separate example of why apparently unrelated ideas can belong together.
What If Fermat’s Equation Had a Counterexample?
Suppose, provisionally, that the theorem were false. Gerhard Frey proposed associating a special elliptic curve with a hypothetical counterexample. Kenneth Ribet, through a result connected to Jean-Pierre Serre’s work, established that the resulting curve could not be modular.
Wiles proved the needed modularity result for semistable elliptic curves over the rational numbers, a class containing the relevant Frey curves. The imagined curve would therefore have to be modular and not modular. That contradiction rules out the counterexample. The official Abel Prize citation identifies these contributions and the semistable scope.
This is a logical outline with major technical conditions and reductions omitted, not a compressed elementary proof. Wiles did not prove the entire modularity conjecture at that time. The narrower result was sufficient for Fermat’s theorem.
Notice the change in strategy: instead of checking candidates individually, the argument shows that any counterexample would force an impossible object to exist. The contradiction itself is straightforward to state. Establishing both sides of it required the deep work.
Why Does Fermat’s Last Theorem Matter?
The equation is famous, but the methods and connections outgrew the original question. The modularity techniques opened routes to further results in number theory. Solving the puzzle also changed what mathematicians could attempt next.
Elliptic curves have applications in cryptography, but that does not mean Fermat’s Last Theorem made internet encryption possible. A shared mathematical object is not the same thing as a direct technological cause.
Nor must every mathematical question justify itself with a product. There is value in discovering why a plausible pattern holds, and in learning that the explanation lives somewhere unexpected. Our guide to cubic roots and coefficients offers a more accessible algebraic example of extracting structure instead of repeatedly guessing answers.
Common Questions About Fermat’s Last Theorem
What is Fermat’s Last Theorem in simple terms?
No two positive whole numbers raised to the same integer power greater than 2 can add up to a third positive whole number raised to that power. All three powers must use the same exponent.
Who proved Fermat’s Last Theorem?
Andrew Wiles proved it, with a crucial contribution from Richard Taylor to the corrected argument. The proof also depends on earlier results connecting the problem to elliptic curves and modularity.
When was Fermat’s Last Theorem proved?
The corrected proof was completed in 1994 and published in 1995. Wiles’s initial 1993 announcement preceded the discovery and repair of a gap.
Did Fermat actually have a proof?
It is not known whether Fermat possessed a valid proof for the general case. His margin note claims one, but supplies no argument that can be checked.
Why doesn’t the theorem apply when n = 2?
Because squares do have positive-integer solutions, including 3² + 4² = 5². The condition n > 2 excludes a case in which the proposed impossibility is false.
Can a computer prove it by checking enough numbers?
Not by an arbitrarily large finite search alone. A rigorous proof must account for all permitted integers and exponents, whether its reasoning uses computers or not.
A Small Equation, a Much Bigger Idea
The numbers 3, 4, and 5 still fit together when squared. Replacing the exponent changes a classroom calculation into a question about all counting numbers. The answer did not come from finding the right enormous calculation, but from discovering a connection strong enough to rule out every counterexample.
Sources and Further Reading
University of St Andrews, MacTutor — Pierre de Fermat: biography, the margin note, and the 1670 publication.
The Abel Prize, Arne B. Sletsjøe — Hanc marginis exiguitas non caperet: the Latin note, conventional dating, and the theorem’s name.
University of Oxford, Mathematical Institute — Andrew Wiles presented with the Abel Prize in Oslo: Wiles’s recollection and the research timeline.
The Norwegian Academy of Science and Letters — 2016 Abel Prize citation for Sir Andrew J. Wiles: the Frey–Ribet connection, semistable modularity, and Taylor’s contribution.
Annals of Mathematics, Volume 141, Issue 3 (1995) — Andrew Wiles, Modular elliptic curves and Fermat’s Last Theorem, and Richard Taylor and Andrew Wiles, Ring-theoretic properties of certain Hecke algebras: the two publication records.
Last fact-checked: September 21, 2026. This introduction is independently written; it does not reproduce the technical proof.