Everything this rung built, in one frame
You have arrived at the summit of the rung, so let us first take stock of the gear we are carrying. We learned the gamma function Gamma(s) as the factorial extended to the whole plane, with its reflection formula, its Weierstrass product, and its Stirling asymptotics. We met the Riemann zeta function zeta(s) = sum over n >= 1 of 1/n^s, alive only for Re s > 1 at first. We saw the Euler product secretly factor zeta over the primes, then watched analytic continuation and the functional equation carry zeta to every point of the plane except a single pole at s = 1. This final guide does not introduce a fifth gadget — it shows how the four you already have lock together into one machine.
Here is the one sentence the whole rung was secretly building toward. The prime numbers — 2, 3, 5, 7, 11, ... — are the multiplicative atoms of the integers, scattered with no visible pattern; yet how densely they are scattered is controlled, with surgical precision, by the location of the complex zeros of zeta. That is the punchline, and it should sound impossible the first time you hear it. A question about counting prime numbers, which feels purely arithmetic, turns out to be a question about where a holomorphic function vanishes in the complex plane. The bridge from one to the other is built entirely out of the tools of this rung.
From the Euler product to a sum over primes
Start where zeta first meets the primes. For Re s > 1 the Euler product says zeta(s) = product over primes p of 1/(1 - p^(-s)). This identity is unique factorisation written analytically: expand each factor as a geometric series 1/(1 - p^(-s)) = 1 + p^(-s) + p^(-2s) + ..., multiply over all primes, and every integer n = p_1^(a_1) p_2^(a_2) ... appears exactly once, rebuilding the sum 1 + 1/2^s + 1/3^s + ... . But a product is awkward to handle; we want a sum. The standard move — the same logarithmic-derivative trick you met for counting zeros and poles — is to take the logarithm.
Take logs of the Euler product and the product becomes a sum: log zeta(s) = sum over primes p of -log(1 - p^(-s)) = sum over p of (p^(-s) + (1/2) p^(-2s) + ...). The leading piece, sum over primes p of p^(-s), is already a Dirichlet series whose terms are indexed by the primes themselves — a faithful analytic encoding of the prime list. Differentiate once more and you get -zeta'(s)/zeta(s) = sum over prime powers of (log p) p^(-ks), a clean sum that puts a weight of log p on each prime and each of its powers. This object, the logarithmic derivative of zeta, is the workhorse that carries the primes into the world of complex analysis.
The explicit formula: zeros become waves over the primes
Now the two halves collide. On one side, -zeta'(s)/zeta(s) is a sum over primes. On the other side, by the continuation of zeta, this same function has poles exactly where zeta has a pole or a zero: a pole at s = 1 (from zeta's pole), and a pole at every zero of zeta. Riemann's idea was to integrate -zeta'(s)/zeta(s) times x^s/s around a large contour and read off the answer two ways — once as a sum over primes (the left side), once as a sum of residues over the poles (the right side). The residue theorem equates them. The result is the celebrated explicit formula.
EXPLICIT FORMULA (smoothed prime count psi(x), schematic)
psi(x) = x - sum over zeros rho of x^rho / rho - (small stuff)
^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
main term one oscillating wave PER zero
(from pole at s=1) (from each nontrivial zero rho)
a zero at rho = beta + i gamma contributes a term of size x^beta,
oscillating in x with frequency set by gamma
psi(x) = sum over prime powers p^k <= x of log p (counts primes, with weights)Read this formula slowly, because it is the heart of the subject. The smoothed prime count psi(x) is not a chaotic mess — it is a single clean trend line, the term x, plus a chorus of waves, one wave for every nontrivial zero of zeta. The zero at rho = beta + i gamma produces a wave of amplitude x^beta that oscillates in x. The primes are the trend; the zeros are the music written over them. This is the precise sense in which 'the zeros control the primes': move a zero and you change one of the waves, and the prime counting function wobbles in response. The mystery of the primes has become the geography of a set of points in the complex plane.
Where the zeros are, and the prime number theorem
So the whole game is to locate the zeros. Three facts pin them down. First, the Euler product shows zeta has no zeros at all for Re s > 1, since none of the factors 1/(1 - p^(-s)) can vanish. Second, the functional equation xi(s) = xi(1 - s), built with the gamma factor pi^(-s/2) Gamma(s/2), reflects everything across the line Re s = 1/2; combined with the zero-free right half-plane it forces all the trivial zeros to the negative even integers s = -2, -4, -6, ... (where the factor sin(pi s/2) in the unsymmetric form vanishes). Third, by reflection, the only zeros left to worry about — the nontrivial ones — are trapped inside the critical strip 0 <= Re s <= 1.
Now watch the explicit formula do real work. The leading term x comes from the pole of zeta at s = 1. Every nontrivial zero rho = beta + i gamma adds a wave of size x^beta. If we could prove that no zero has beta = 1 — that zeta does not vanish on the edge line Re s = 1 — then every wave would be strictly smaller than x, the main term would dominate, and psi(x) ~ x. That single non-vanishing fact, proved by Hadamard and de la Vallee Poussin in 1896, is exactly what yields the prime number theorem: pi(x), the count of primes up to x, is asymptotic to x/log x. The primes thin out at precisely the logarithmic rate, and the proof is a statement about where a holomorphic function refuses to vanish.
The Riemann hypothesis, stated honestly
The functional equation's symmetry Re s -> 1 - Re s means the nontrivial zeros are arranged in mirror pairs straddling the central line Re s = 1/2, and complex conjugation pairs them across the real axis too. The fixed axis of that reflection is the critical line Re s = 1/2: a zero sitting exactly on it is its own mirror image. Riemann, staring at this symmetry in his 1859 memoir, conjectured the cleanest possible thing — that the symmetry is not just satisfied in pairs but saturated, with every nontrivial zero lying exactly on the line. This is the Riemann hypothesis: every nontrivial zero of zeta has real part exactly 1/2.
Translate it back through the explicit formula and you see why it matters so intensely. If every zero has beta = 1/2, then every oscillating wave over the primes has amplitude x^(1/2) — the smallest the symmetry allows — and the prime counting error pi(x) - Li(x) is bounded by about sqrt(x) log x, essentially the best conceivable. The Riemann hypothesis is, in plain terms, the assertion that the primes are distributed as regularly as they possibly can be, with no rogue zero hiding off the line to create unexpected clumping or thinning. It is not an arbitrary technical wish; it is the statement that the deepest possible order reigns among the primes.
How far the idea reaches
Zeta is the first member of a vast family, not a lonely curiosity. Replace 1/n^s by a_n/n^s with carefully chosen coefficients and you get a general Dirichlet series; the most important examples are the Dirichlet L-functions, which encode primes in arithmetic progressions and carry their own Euler products, functional equations, and critical lines. Each is conjectured to have all its nontrivial zeros on Re s = 1/2 — the Generalised Riemann Hypothesis — and the same machinery (gamma factors completing the function, a reflection s <-> 1 - s, zeros steering an explicit formula) repeats almost verbatim. The whole rung you just climbed was, in hindsight, basic training for an entire landscape of L-functions.
Step all the way back and admire the shape of the argument, because it is one of the most beautiful in mathematics. A question about the integers (how are the primes spread?) became, through the Euler product, a question about an analytic function; through analytic continuation and the functional equation, that function was tamed across the whole plane; through the residue theorem and an explicit formula, its complex zeros became audible as waves over the primes; and the Riemann hypothesis is simply the conjecture that those zeros are as orderly as the symmetry permits. Every single link in that chain is a tool from this rung. That is the deepest lesson of complex analysis: a holomorphic function, rigid and over-determined, can reach into a foreign subject and decide its facts.
One closing caution so the picture stays honest. We have spoken of 'reading the primes off the zeros' as if it were a finished computation; in truth the explicit formula requires real care to make rigorous — the contour must be justified, the arc contributions must be shown to vanish, and the sum over zeros only converges in a suitably smoothed sense. None of that is hand-waving away the depth; it is the depth. The miracle is not that the bridge is easy to cross, but that it exists at all: that the discrete, stubborn primes and the smooth, complex-analytic zeros of zeta are two views of one single object. Sit with that for a moment — it is, quite genuinely, one of the summits of human thought.