A topological manifold is a Hausdorff space with a countable basis where every point has a neighborhood homeomorphic to an open ball in ℝⁿ. Manifolds are the natural spaces for calculus and geometry. Understanding manifold topology is foundational for differential topology and differential geometry.
You already know that a Hausdorff space is one where distinct points can be separated by disjoint open sets — a minimal sanity condition that prevents points from being "too close to distinguish." You also know that a space with a countable basis (second-countable) has a topology fully described by countably many open sets, which gives it manageable size and good analytic properties. A topological manifold combines both conditions with one more: every point has a neighborhood that looks exactly like an open subset of Euclidean space ℝⁿ. This last condition, called local Euclidean structure, is what makes manifolds the right setting for geometry and calculus.
The dimension n is the key parameter: a 1-manifold locally looks like an open interval, a 2-manifold locally looks like an open disk, a 3-manifold locally looks like an open ball in ℝ³. The surface of the Earth is the canonical 2-manifold: at any given location, a small enough neighborhood looks like a flat patch of the plane (the reason flat maps work locally). A circle S¹ is a 1-manifold: any small arc looks like an open interval. The sphere S² and the torus are both 2-manifolds, even though globally they are very different — but locally, they are both indistinguishable from flat patches.
The precise condition is the existence of a homeomorphism (a continuous bijection with continuous inverse) between a neighborhood of each point and an open ball in ℝⁿ. These local homeomorphisms are called coordinate charts, and a collection of charts covering the whole manifold is called an atlas. The chart converts the abstract topological space into a concrete coordinate system where you can do calculations. The Hausdorff condition is needed to ensure that charts behave sensibly — without it, two different points might have identical local behavior and be impossible to distinguish. The second-countability condition ensures the manifold can be covered by countably many charts, which enables partition-of-unity arguments and other analytic tools.
Why impose these conditions rather than just working in ℝⁿ directly? Because many natural spaces in mathematics and physics have the local structure of Euclidean space without being globally flat. The surface of a sphere cannot be given a consistent Euclidean geometry (you cannot flatten a globe without distortion), yet locally it is flat. Spacetime in general relativity is a 4-manifold that curves globally but looks Euclidean at small scales. The solution sets of systems of equations — called varieties in algebraic geometry — are often manifolds. The manifold concept captures exactly the class of spaces where calculus makes sense locally, even when the global geometry is rich and non-Euclidean.
The step from topological manifolds to differential manifolds requires asking when the overlapping charts in an atlas are compatible in a smooth sense — this becomes the subject of differential topology and differential geometry, your next destinations. The topological manifold concept is the foundation: you need to understand the local-Euclidean, Hausdorff, second-countable conditions before asking about smoothness. Everything that follows — tangent spaces, vector fields, integration on manifolds, curvature — rests on the basic fact that you can work in local coordinates on each chart.