Goldbach's conjecture

Some numbers can only be divided evenly by 1 and themselves: 2, 3, 5, 7, 11, and so on. These are the prime numbers, the building blocks of all the others. In 1742 a mathematician named Christian Goldbach noticed that every even number bigger than 2 seems to be the sum of two primes. 4 is 2 plus 2, 10 is 3 plus 7, 100 is 3 plus 97. Try any even number you like, and it works.

That is Goldbach's conjecture, and the shocking part is that nobody has ever proved it. Computers have checked every even number up to four quintillion, a 4 followed by eighteen zeros, and it holds every time. But there are infinitely many even numbers, and checking is not proving.

Here is the twist. You would think a rule this simple would be easy to prove, yet it has defeated the best mathematicians for nearly three hundred years. And it gets stranger: the bigger an even number is, the more ways there usually are to write it as two primes. The rule works with room to spare, more and more as numbers grow, and even that is not enough to prove no giant even number could ever break it.

In a 1742 letter to Leonhard Euler, Christian Goldbach floated a simple idea about prime numbers, the numbers like 2, 3, 5, and 7 with no divisors except 1 and themselves. Euler restated it in the form we use today: every even number greater than 2 is the sum of two primes. So 8 is 3 plus 5 and 28 is 11 plus 17, and larger numbers usually have many such pairs. Using distributed computing, mathematicians have confirmed it for every even integer up to four quintillion without exception, but a verified range, however vast, is not a proof.

There is a close cousin, the weak or ternary Goldbach conjecture: every odd number greater than 5 is the sum of three primes. This one is no longer open. Ivan Vinogradov proved it for all sufficiently large odd numbers in 1937, and in 2013 Harald Helfgott closed the remaining gap, making it a theorem. The dividing line is startlingly thin: going from three primes down to two flips the problem from solved to one of the oldest open questions in mathematics.

Why is the famous version so much harder? Primes are creatures of multiplication, defined by how numbers factor, and the tools that describe them are all multiplicative. Goldbach instead asks an additive question, about sums, and there is no natural bridge from how a number factors to how it splits into a sum. Crossing from the multiplicative world of primes to the additive world of Goldbach is exactly the gap no one has managed.

The deepest surprise is that the conjecture is not barely true. The number of ways to write an even number as two primes generally grows as the number gets larger, so a counterexample would have to be an even number that, against an enormous and rising supply of prime pairs, somehow has none at all. The evidence does not whisper that Goldbach is true; it shouts it. And still, no one can rule out that one silent exception.

State the conjecture and it is deceptively elementary: every even integer \(n\) greater than 2 is the sum of two primes. Goldbach's original 1742 version, written when 1 still counted as prime, used three primes; Euler reduced it to the binary form, which implies the ternary one. Its difficulty is the central obstacle of additive number theory: primes are defined multiplicatively, and additive questions get almost no purchase on that definition.

Schnirelmann made the first unconditional dent. In 1930 Lev Schnirelmann proved, using what is now called Schnirelmann density, that the primes form an additive basis of finite order: some constant C makes every integer greater than 1 a sum of at most C primes. The original constant was enormous, but the result was decisive in showing that a bounded number of primes always suffices, even though two could not be reached. Later work cut the bound to a small handful.

The ternary problem yielded to the circle method. Using the Hardy-Littlewood circle method, Vinogradov proved in 1937 that every sufficiently large odd integer is a sum of three primes, estimating exponential sums over primes on the minor arcs while the major arcs give the main term. The method left an astronomically large threshold, and after decades of reductions Helfgott in 2013 lowered it far enough to check the rest by computer, completing the ternary conjecture for every odd number greater than 5.

Chen's theorem is the closest approach to the binary conjecture. In 1973 Chen Jingrun proved, by intricate sieve methods, that every sufficiently large even integer is the sum of a prime and a semiprime, a number that is prime or a product of two primes. It cannot be pushed the final step, because sieves meet the parity problem, Selberg's intrinsic barrier whereby they cannot distinguish numbers with an even number of prime factors from those with an odd number, exactly the distinction needed to force the second summand prime.

Heuristics predict not just truth but abundance. The Hardy-Littlewood conjecture gives an asymptotic for the number of ways to write an even \(n\) as an ordered sum of two primes. The skeleton is simple: a number near \(n\) is prime with density about \(1/\ln n\), so two numbers summing to \(n\) are both prime with likelihood about \(1/(\ln n)^2\), and over the roughly \(n/2\) splittings this gives on the order of \(n/(\ln n)^2\) representations, times a constant from the twin prime constant (about 0.6601) and a correction over the odd prime factors of \(n\). Crucially the count grows without bound, so large even numbers have more representations, not fewer.

The numerical and visual evidence is overwhelming. The conjecture has been verified for every even integer up to \(4 \times 10^{18}\) with no counterexample. Plotting the number of representations against \(n\) produces the Goldbach comet, a scatter that rises and fans out and splits into bands set by the small prime factors of \(n\), just as the Hardy-Littlewood correction predicts.

And here is what makes Goldbach so maddening. The evidence is not borderline; it is lavish, and it grows. A counterexample would be an even number that, against a supply of prime pairs increasing without limit, holds exactly zero, an outcome the heuristics call impossible yet no argument forbids. The obstruction is structural: every method either describes primes multiplicatively, where Goldbach's additive demand is invisible, or leans on sieves, which the parity problem stops one step short. The weak conjecture, needing three primes, has fallen; the strong, needing two, has not, and the whole distance between a celebrated theorem and a famous mystery is that one missing prime. The primes are the atoms of multiplication, and Goldbach asks whether they are also the bricks of addition. Three centuries of evidence say yes, overwhelmingly, while three centuries of effort have failed to say why.