8,128 is the fourth perfect number — the fourth positive integer whose proper divisors add up to exactly itself, and the largest of the four perfect numbers already known to ancient mathematicians. Its proper divisors are 1, 2, 4, 8, 16, 32, 64, 127, 254, 508, 1016, 2032, and 4064, and summing every one of them: 1+2+4+8+16+32+64+127+254+508+1016+2032+4064 = 8,128, exactly.
Nicomachus of Gerasa, writing around 100 CE, listed all four of the perfect numbers known in antiquity by name — 6, 28, 496, and 8,128 — making 8,128 one of the largest specific individual integers to appear by name in surviving classical mathematical literature, at a time when working with and verifying numbers this size by hand was itself a genuine feat of careful, sustained arithmetic.
8,128 follows Euclid's perfect-number formula precisely, the same construction underlying 6, 28, and 496: if 2^p − 1 is prime (a Mersenne prime), then 2^(p−1) × (2^p − 1) is guaranteed to be perfect. For p = 7: 2⁷ − 1 = 127, and 127 is indeed prime (checkable by trial division up to its square root, about 11.3: not divisible by 2, 3, 5, 7, or 11). Plugging in: 2⁶ × 127 = 64 × 127 = 8,128, confirmed directly by the formula rather than by summing all thirteen proper divisors by hand, though both routes agree exactly.
The gap between the third and fourth perfect numbers (496 to 8,128) is roughly sixteen-fold, but the very next perfect number after 8,128, — 33,550,336 — represents a jump of more than four thousand-fold, a dramatic acceleration that illustrates just how quickly perfect numbers thin out once you move past the first four. This isn't a coincidence of the specific numbers involved; it's a direct consequence of how rare Mersenne primes themselves become as the exponent p grows, since every even perfect number depends entirely on finding one.
8,128 sits at a natural teaching boundary for this reason: it's the last perfect number small enough that its full divisor list is genuinely practical to write out and sum by hand within a few minutes, making it a natural stopping point for hand-verification exercises before the sequence's next term becomes large enough to require Euclid's formula (and a primality check on 2¹³−1=8,191, itself prime) rather than direct enumeration.
As with 6, 28, and 496, no odd perfect number sharing 8,128's exact self-reconstructing property has ever been found at any size, and it remains formally unproven, over two thousand years after Euclid and Nicomachus first studied this family of numbers, whether one can exist at all — a genuinely open problem that sits directly alongside these four ancient, completely solved examples.
It took over a thousand years after Nicomachus catalogued these first four perfect numbers before the fifth, 33,550,336, was correctly identified — an achievement credited to an anonymous medieval manuscript from around 1456, since verifying it required confirming that 2¹³−1=8,191 is prime, a considerably harder hand-calculation than checking the smaller Mersenne primes underlying 6, 28, 496, and 8,128. That long historical gap between the fourth and fifth perfect numbers is itself a genuine, documented illustration of exactly how quickly the computational difficulty of this search escalates once you move past the examples ancient mathematicians could verify directly by hand.
Modern searches for new perfect numbers proceed entirely through the same distributed-computing infrastructure used to hunt for new Mersenne primes generally (the GIMPS project, discussed in more depth on this site's own page for 496), since every even perfect number depends directly on finding a new Mersenne prime first — meaning 8,128's own comparatively humble, hand-verifiable scale sits at one extreme end of a search that today routinely works with Mersenne primes running into the tens of millions of digits, an almost inconceivable jump in scale from the numbers Nicomachus catalogued by hand two thousand years ago. That scale jump is worth sitting with for a moment: 8,128 has thirteen proper divisors, easily summed on paper in a few minutes, while the currently largest known perfect numbers have divisor counts and digit lengths that would take entire volumes to print in full — the same underlying mathematics, stretched across a genuinely staggering range of practical scale, from a number small enough for a classroom exercise to one that exists only as output from a specialized distributed-computing network.