Two of history’s greatest mathematicians searching for a perfect pattern in prime numbers, only to find that nature is far more chaotic and computationally demanding than it appears.

Fermat

Pierre de Fermat, the same man behind the “Last Theorem,” was obsessed with finding a formula that would consistently generate prime numbers. He proposed the Fermat Numbers, defined by the form:

$$F_n = 2^{2^n} + 1$$

Fermat calculated the first five:

• $F_0 = 2^{2^0} + 1 = 3$
• $F_1 = 2^{2^1} + 1 = 5$
• $F_2 = 2^{2^2} + 1 = 17$
• $F_3 = 2^{2^3} + 1 = 257$
• $F_4 = 2^{2^4} + 1 = 65537$

All five are prime. Fermat famously conjectured that all numbers of this form were prime.^[1] It was a clean, beautiful, and efficient model of the universe.

The Breakdown: Nearly a century later, Leonhard Euler did the math (without a MacBook Pro or a VPS) and discovered that $F_5$ was composite^[2]:

$$F_5 = 2^{2^5} + 1 = 4,294,967,297 = 641 \times 6,700,417$$

In fact, to this day, we have not found a single other Fermat prime beyond $F_4$. It serves as a stark reminder for any innovator: A pattern that holds for five iterations is not a law of physics.

Mersenne Primes

While Fermat looked at $2^{2^n} + 1$, Marin Mersenne explored the “neighboring” form:

$$M_p = 2^p - 1$$

For $M_p$ to be prime, $p$ itself must be prime. These are the giants of the number world. Because they grow exponentially, they are the primary targets for the Great Internet Mersenne Prime Search (GIMPS)^[3], a massive, distributed computing project that essentially treats the internet as one giant supercomputer.

The connection to perfect numbers (numbers equal to the sum of their proper divisors) is where the “high-fidelity” design of math shows through. The Euclid-Euler Theorem states that every even perfect number is of the form:

$$2^{p-1}(2^p - 1)$$

…where $2^p - 1$ is a Mersenne prime.^[4]

The Computational Frontier

For those of us managing homelabs or building robotics databases, Mersenne primes represent the ultimate stress test. Finding a new Mersenne prime isn’t just about math; it’s about efficiency in work and raw CPU cycles.

We use the Lucas-Lehmer Test to verify them. The algorithm is remarkably elegant for a computer to execute:

1. Let $s_0 = 4$.
2. For $i$ from 1 to $p-2$:

   $$s_i = (s_{i-1}^2 - 2) \pmod{M_p}$$

3. If $s_{p-2} = 0$, then $M_p$ is prime.

The Innovation Takeaway

Fermat’s failure was a failure of edge-case testing. He assumed the system scaled because the initial results looked perfect. Mersenne’s success was creating a framework that later on invited collaboration and infinite scaling.

In business and in code, we often look for the “Fermat formula”—a shortcut that works every time. But the real “Mersenne” wins come from building systems that can handle the massive, unglamorous work of testing every prime until you find the one that doesn’t break.

References

---

1. Fermat, P. (1640). Letter to Marin Mersenne. The conjecture that all Fermat numbers are prime was stated without proof. ↩︎

2. Euler, L. (1732). Observations de theoremate quodam Fermatiano aliisque ad numeros primos spectantibus. Euler’s factorization of $F_5$ was the first counterexample. ↩︎

3. GIMPS. (2026). The Great Internet Mersenne Prime Search. Retrieved from mersenne.org. ↩︎

4. Burton, D. M. (2010). Elementary Number Theory. The proof linking Mersenne primes to even perfect numbers is a cornerstone of classical number theory. ↩︎