The astonishing claim
The previous rung handed you Cauchy's integral theorem: a holomorphic function integrates to zero around any closed loop with a hole-free interior. That tells you when an integral vanishes. Cauchy's integral formula asks a bolder question — what if the integrand is engineered to NOT vanish, but to leave behind exactly one value? The answer is the centerpiece of the whole subject: if f is holomorphic on and inside a simple closed loop C, then for any point z_0 trapped inside, the value f(z_0) is recovered by a single integral around C.
f(z_0) = (1 / (2 pi i)) * integral over C of f(z) / (z - z_0) dz
Read that formula slowly. On the right side, z_0 appears only as a fixed parameter inside the kernel; everything actually integrated — f(z) and z — lives on the curve C, never in the interior. Yet the output is f at an interior point you may never have visited. This is the meaning of determination by boundary values: the values of f on a one-dimensional rim secretly contain the values of f throughout the two-dimensional region it bounds. Nothing in your real-variable experience prepares you for this.
Where the formula comes from
The derivation is short and worth carrying in your head, because every later consequence is really this one idea reused. Consider the function g(z) = f(z) / (z - z_0). It is holomorphic everywhere inside C except at the single bad point z = z_0, where the denominator vanishes. That one trapped singularity is exactly the obstruction Cauchy's theorem cannot ignore — and we exploit it deliberately rather than avoid it.
- By deforming the contour in the region where g stays holomorphic, replace the big loop C by a tiny circle of radius r centered at z_0. The integral does not change.
- On that tiny circle parametrize z = z_0 + r e^(i theta), so dz = i r e^(i theta) d theta and z - z_0 = r e^(i theta). The kernel 1 / (z - z_0) and the dz partly cancel, leaving the clean factor i d theta.
- The integral becomes i times the integral of f(z_0 + r e^(i theta)) d theta over theta from 0 to 2 pi. As r shrinks to zero, continuity of f makes the integrand approach the constant f(z_0).
- That limit gives i times 2 pi times f(z_0) = 2 pi i f(z_0). Divide by 2 pi i and the formula stands. The whole result is squeezed out of one trapped singularity and the geometry of a shrinking circle.
Reading the formula as a weighted average
There is a way to feel the formula in your bones. The factor 1 / (z - z_0) is the Cauchy kernel, and it acts like a spotlight. When z on the boundary is close to z_0, the denominator is small, so that part of the rim shines brightly and contributes a lot; when z is far away, the kernel dims and that stretch of boundary contributes little. So f(z_0) is a weighted blend of the boundary values, with nearby boundary points weighted most heavily. The Cauchy kernel is precisely the recipe that tells you how to mix the rim values to reproduce any interior value.
This viewpoint also explains why the winding number sneaks into the bookkeeping. The clean formula assumes C wraps z_0 exactly once, counterclockwise. If your contour circles z_0 twice, you collect 2 times 2 pi i f(z_0); if it never encloses z_0, you collect zero, recovering Cauchy's theorem. The honest general statement multiplies the right side by the winding number of C about z_0, so the formula keeps faithful count of how many loops the spotlight makes around the point.
A tiny worked example
Let C be the unit circle traversed counterclockwise, and evaluate the integral of e^z / (z - 0) dz around C. Match it to the formula: here f(z) = e^z, which is entire and so certainly holomorphic on and inside C, and the singled-out point is z_0 = 0, comfortably inside. The formula says the integral equals 2 pi i times f(0) = 2 pi i times e^0 = 2 pi i. No antiderivative hunt, no parametrization grind — the formula reads the answer straight off the value of e^z at the center.
Change one thing and watch the machinery respond. Take the integral of e^z / (z - 3) around the same unit circle. Now the singled-out point z_0 = 3 sits OUTSIDE the unit circle, so e^z / (z - 3) is holomorphic on and inside C, with no trapped bad point. Cauchy's theorem applies directly and the integral is zero. The location of z_0 relative to the loop, not the formula for f, is what flips the answer between 2 pi i f(z_0) and 0.
Why this is the gateway
Cauchy's integral formula is not just a slick way to evaluate integrals; it is the engine that drives the next four guides. The crucial structural fact is that on the right side z_0 enters only through the smooth kernel 1 / (z - z_0). You can differentiate that kernel with respect to z_0 as many times as you like, right under the integral sign, while the boundary values f(z) just sit there as fixed data. Each differentiation produces a new formula — and that is the secret behind infinite differentiability, the headline of the very next guide.
From that one observation the whole rung unfolds. Differentiating the formula yields the Cauchy estimates, which bound the derivatives by the size of f on the boundary; those bounds force Liouville's theorem (a bounded entire function must be constant) and with it the fundamental theorem of algebra; running the logic backward gives Morera's converse; and pushing the mean value property to its extreme gives the maximum modulus principle. Be honest with yourself about how special this is: one complex derivative, fed through this integral, manufactures infinitely many — a rigidity that has no counterpart anywhere in real-variable calculus.