Andrew Wiles (b. 1953) was a ten-year-old poking around his local library in Cambridge when he pulled a book off the shelf and met a puzzle that would follow him for the rest of his life: Fermat's Last Theorem. It looked so simple a child could understand it — and yet Fermat had left the world dangling for three centuries. Wiles later said he knew, right then, that he would never let it go. Most children who say that grow out of it. He did not.

Seven years in the attic

By the 1980s Wiles was a respected Princeton professor with a comfortable career in number theory — exactly the sort of person who is supposed to know better than to chase the most notorious unsolved problem in mathematics. He chased it anyway. Realising the theorem might finally be reachable through the modularity of elliptic curves, he did something almost unheard of for a modern mathematician: he went underground. For seven years he worked in near-total secrecy in his attic study, telling almost no one what he was really up to, quietly publishing small unrelated results now and then so nobody would guess.

In June 1993 he unveiled his work in a series of Cambridge lectures, ending — almost casually — by writing Fermat's Last Theorem on the board and saying, "I think I'll stop here." The room erupted. A 350-year-old problem looked, at last, closed.

The gap, and the year that nearly broke him

Then came the twist. As experts combed through the proof, a subtle gap surfaced — one stubborn piece that simply did not hold. For over a year Wiles wrestled with it in public view, the triumph curdling into a very slow, very visible failure. Just as he was ready to give up, in September 1994 he had a flash of insight: an approach he had abandoned years earlier could be welded to another to plug the hole. With his former student Richard Taylor he repaired the argument, and the completed proof was published in 1994. Fermat's margin note was finally answered.