On October 31, 1903, a Columbia professor walked up to a blackboard at the American Mathematical Society, filled it with numbers for an hour, and sat back down without saying a single word. The room gave him a standing ovation.
The talk with no words
The professor was Frank Nelson Cole. On one side of the board, he worked out 2 multiplied by itself 67 times, then subtracted one. The result was a 21-digit monster: 147,573,952,589,676,412,927.
Then he moved to the other side of the board and wrote two much smaller numbers: 193,707,721 and 761,838,257,287. He multiplied them by hand, digit by digit, in front of everyone.
The answer matched exactly. Cole returned to his seat. He had just shown that a number long suspected of hiding something could be split into two factors, and he never needed to explain it. The numbers did all the talking.
A famous friar’s list that was mostly wrong
To see why the audience stood up, you have to go back to 1644. That year, the French friar Marin Mersenne published a list of numbers of the form “a power of two, minus one” that he claimed were prime, meaning divisible only by 1 and themselves.
Mersenne was no amateur. He corresponded with Descartes and Galileo, and his network of scholars was so vast that he has been called “the post-box of Europe.” His list carried weight.
The problem is that it was wrong in five places. Mersenne included two numbers that are not prime (one built from 67, one from 257) and missed three that are (built from 61, 89 and 107). He also left almost no clue about how he had come up with his list. Checking it properly took roughly three centuries.
Knowing a lock exists without finding the key
In 1876, the French mathematician Édouard Lucas proved that the number built from 67 could not be prime. It had to break down into smaller factors.
But his proof had a frustrating gap. It showed that factors existed without revealing a single one of them. For 27 years, mathematicians knew the number was composite and had no idea how to split it.
Cole found the answer the slow way. He later admitted that hunting down the two factors had taken him “three years of Sundays.” Long before computers, it was just patience and arithmetic by hand. On a site called FactoringCalc, it is hard to imagine a better patron saint.
From desk work to thousands of graphics cards
That same year, Lucas also proved that the number built from 127 really is prime, as Mersenne had claimed. With 39 digits, it stayed the largest known prime for 75 years.
Manual hunters kept going after that. Horace Scudder Uhler, a retired Yale physics professor, tested candidates by hand and found nothing. He was unlucky: the next exponent that works is 521, more than four times larger than 127.
Machines changed everything. At 10 p.m. on January 30, 1952, a computer at UCLA, running a program written by R. M. Robinson under the direction of D. H. Lehmer, found the prime built from 521, the first new one in 38 years. Less than two hours later, it found another.
Since 1997, every new one has come from the Great Internet Mersenne Prime Search, where volunteers lend their computers. In 2018, one turned up on a machine in the offices of a church in Tennessee. In October 2024, former Nvidia employee Luke Durant found the current record, a prime with 41,024,320 digits. He spent about a year and roughly 2 million dollars running thousands of graphics cards in the cloud.
Why these primes still matter
These numbers are not just trophies. Each one produces a perfect number, and in binary every one of them is written with nothing but ones. They even turn up in Sophie Germain’s primes, which can reveal a factor of some of them instantly: 23 divides the number built from 11.
And yet the most basic question is still open. Nobody knows whether there are infinitely many of them. Fifty-two are known today. Somewhere out there, the 53rd is waiting, and someone, with a graphics card farm or three years of Sundays, will be the first to see it.

