Where we are on the ladder
You arrive in this rung carrying the complex-geometry track: you know what a complex manifold is, an atlas whose transition maps are holomorphic rather than merely smooth, and you have seen the obstruction that decides when an almost-complex structure is genuinely complex. A Riemann surface is the smallest interesting case of all that — a complex manifold of complex dimension one. The promise of this rung is unusual and worth stating plainly: in this one dimension, topology, complex analysis, and algebra stop being separate subjects and become three faces of a single object.
An honesty note before we begin. These are graduate topics; we assume you are fluent in one-variable complex analysis (holomorphic functions, Laurent series, residues), point-set topology, and the smooth- and complex-manifold language from earlier rungs. We will say 'manifold', 'chart', 'covering space', 'genus' and lean back on Volume I and the complex-geometry guides rather than re-deriving them. Where a fact is a survey claim whose proof is a course in itself — the uniformization theorem is the prime example — we will say so honestly and point forward to the guide that develops it, rather than pretending a paragraph is a proof.
What a Riemann surface is, concretely
Strip the definition to its bones. A Riemann surface X is a connected Hausdorff topological space with an atlas of charts to open subsets of C, such that every transition map — the comparison of two overlapping charts — is holomorphic. That is the entire definition, and the only word that distinguishes it from a smooth surface is 'holomorphic'. Because every holomorphic map is in particular smooth and orientation-preserving, a Riemann surface is automatically a smooth, oriented, real two-dimensional surface. The complex structure is extra rigidity layered on top of that familiar topological skin.
Three examples carry almost all the intuition, so lead with them rather than with generality. The simplest non-trivial one is the Riemann sphere C union the single point at infinity, also written CP^1: two charts, the plane C and the plane near infinity in the coordinate w = 1/z, glued by the holomorphic transition z -> 1/z. Topologically it is the 2-sphere S^2. Next, a complex torus C / L, where L is a lattice like Z + Z*tau: the plane modulo translations, topologically the doughnut, and the home of elliptic curves. Third, any smooth projective algebraic curve — the zero set in CP^2 of a homogeneous polynomial with no singular points — is a compact Riemann surface, which is the bridge from analysis straight into algebraic geometry.
Meromorphic functions: zeros, poles, and their orders
On a Riemann surface, 'holomorphic function' is defined chart by chart: f is holomorphic near p if, expressed in a local coordinate z centred at p, it is an ordinary holomorphic function of z. Because the transition maps are holomorphic, this property does not depend on which chart you choose — that chart-independence is the whole point of demanding holomorphic gluing. A meromorphic function is the same idea allowing controlled blow-ups: f is meromorphic if near each point it is holomorphic except possibly for isolated poles, i.e. locally a ratio of holomorphic functions. The Riemann sphere's meromorphic functions are exactly the rational functions p(z)/q(z) — a fact worth holding onto.
The local invariant that makes all the bookkeeping work is the order of a zero or pole. Pick a local coordinate z with z(p) = 0; expand f in its Laurent series there. The smallest power that appears, call it n, is the order of f at p: if n > 0 the function vanishes to order n (a zero), if n < 0 it blows up to order |n| (a pole), and n = 0 means f is finite and non-zero there. The crucial check is that n does not depend on which local coordinate you chose: a change of coordinate w = w(z) is holomorphic with non-vanishing derivative at p, so it multiplies the leading term by a non-zero factor and leaves the leading power untouched. The order is an honest, coordinate-free integer attached to (f, p).
Local coordinate z with z(p) = 0. f(z) = c_n z^n + c_(n+1) z^(n+1) + ... ( c_n nonzero ) n > 0 -> zero of order n n = 0 -> finite, nonzero value n < 0 -> pole of order |n| ord_p(f g) = ord_p(f) + ord_p(g) ord_p(f/g) = ord_p(f) - ord_p(g) The integer n is independent of the chosen coordinate z.
Compactness changes everything: the field of functions
Here the global topology asserts itself. On a non-compact surface like the plane C there are vast oceans of holomorphic functions — every power series with a decent radius. But on a compact Riemann surface the only globally holomorphic functions are the constants. The reason is a one-line application of the maximum principle: |f| is continuous on a compact space so it attains a maximum, and a non-constant holomorphic function cannot have an interior maximum of modulus, so f must be constant. This is not a technicality; it is the structural fact that forces us to allow poles. If we want any non-constant functions at all on a compact surface, they must be meromorphic.
Now collect all meromorphic functions on a compact Riemann surface X. Sums, products, and quotients of meromorphic functions are meromorphic (poles are isolated and controllable), so they form a field — the field of meromorphic functions M(X). This single algebraic object is astonishingly powerful: for the sphere it is C(z), the rational functions in one variable; for a curve it is the function field studied in algebraic geometry; and a deep theorem says X is recovered, up to isomorphism, from M(X) alone. In other words the analytic geometry of the surface and the pure algebra of its function field are equivalent data — the first concrete payoff of the rung's promise that analysis and algebra fuse.
The genus: one integer that rules them all
Every compact orientable surface is, topologically, a sphere with some number g of handles attached, and the classification of surfaces says that integer g — the genus — is a complete topological invariant: g = 0 is the sphere, g = 1 the torus, g = 2 the two-holed doughnut, and so on. It is pinned down by the Euler characteristic through chi = 2 - 2g, and by the first homology, which has rank 2g. So 'how many holes' is not loose talk; it is a precise, computable number. A compact Riemann surface, being a compact orientable surface, has a perfectly definite genus.
The remarkable thing is that this purely topological g controls the analysis. The dimension of the space of globally holomorphic 1-forms (holomorphic differentials) on X turns out to equal exactly g — zero forms on the sphere, a one-dimensional space on the torus (the famous dz on C/L), and a g-dimensional space in general. The genus likewise sets the bar in the Riemann-Roch theorem, which counts how many meromorphic functions you may build with prescribed poles, and it appears in the Riemann-Hurwitz formula relating the genera of two surfaces joined by a branched map. One integer, read off from the holes, governs the entire function theory of the surface.
Why should topology dictate analysis so tightly? The deep answer is the uniformization theorem, the climax of guide 4: every simply connected Riemann surface is one of exactly three — the sphere CP^1, the plane C, or the unit disk — and the three genus regimes line up with these three geometries. Genus 0 is the sphere with positive curvature, genus 1 is the flat plane (a torus is C/L), and genus at least 2 is hyperbolic, modelled on the disk. Be honest about status: uniformization is stated and motivated here, not proved; its proof is the substance of a course, and we will only sketch its shape when we reach it.
Why this crossroads matters, and what comes next
Step back and see the three towers meeting. Topologically a compact Riemann surface is a g-holed surface; analytically it is the stage for meromorphic functions whose zeros and poles obey precise order arithmetic; algebraically it is a field M(X), equivalently a smooth projective algebraic curve. The rest of this rung is the dictionary between these three views. Guide 2 studies maps between surfaces — branched covers — and the Riemann-Hurwitz formula that tracks how the genus changes. Guide 3 introduces divisors, the canonical class, and the Riemann-Roch theorem, the central counting result that the genus governs.
Two honest cautions to carry up the ladder. First, almost everything that is clean and powerful here requires compactness — the maximum principle, the finiteness of holomorphic-differential dimensions, Riemann-Roch — so always check whether a result assumes a compact surface before you apply it. The non-compact theory (the plane, the disk, the punctured surfaces) is real and important but obeys different rules; do not import a compact theorem into a non-compact setting. Second, sign and normalization conventions vary across sources — some define the genus through chi = 2 - 2g, others normalize differentials differently, and the canonical class can appear with assorted signs. Pick a single textbook, write its conventions on the inside cover, and stay loyal.
A clean way to hold this guide in memory: a compact Riemann surface is one object with three names — a g-holed topological surface, a field of meromorphic functions, and a projective algebraic curve — and the single integer g is the hinge between them. Everything downstream, from the divisor bookkeeping of guide 3 to the Abel-Jacobi map and Jacobian of guide 5, is the unfolding of consequences from that one number and that triple identity. Hold the three examples — sphere, torus, smooth plane curve — in your hands, and the general theory will feel like organized common sense rather than abstraction for its own sake.