A function is continuous if its graph has no sudden jumps, gaps, or breaks — informally, if you could trace it with a pen without ever lifting it off the paper. Made mathematically precise via the concept of a limit, continuity is a foundational property underlying two of the most useful theorems in analysis: the Intermediate Value Theorem and the Extreme Value Theorem.
Imagine graphing your height over the course of your life. It rises smoothly from birth through childhood and adolescence, eventually levelling off — but at no point does it suddenly jump, say, from 100cm to 150cm in an instant. That smooth, unbroken, jump-free quality is what mathematicians call continuity.
Continuity is the property that makes many intuitive assumptions about functions actually valid. If a function is continuous and takes a negative value at one point and a positive value at another, common sense says it must cross zero somewhere in between — and this is provably true, but only if the function is continuous. For a discontinuous function (like a sudden jump straight from negative to positive without passing through zero), this intuition can fail.
If the temperature outside was 5°C at 6am and 20°C at noon, and temperature changes continuously (no sudden jumps), then the temperature must have been exactly 12°C at some specific moment in between — you don't need to know precisely when, just that it must have happened somewhere along the way. This intuitive, obviously-true-sounding fact is a direct real-world instance of continuity's power.
A function f is continuous at a point a if three conditions all hold: (1) f(a) is defined, (2) the limit of f(x) as x approaches a exists (see the Limit entry), and (3) that limit actually equals f(a). In symbols: lim(x→a) f(x) = f(a). A function is continuous on an interval if it's continuous at every single point within that interval.
If f is continuous on [a,b] and N is any value between f(a) and f(b), then there exists some c in [a,b] with f(c)=N. This theorem underlies numerous practical applications, from proving that certain equations must have a solution, to justifying the bisection method for numerically approximating roots of equations on a computer.
If f is continuous on a closed, bounded interval [a,b], then f must attain both an absolute maximum value and an absolute minimum value somewhere on that interval. This theorem guarantees that optimization problems (finding the best possible outcome) posed on continuous functions over closed intervals genuinely have a solution — an essential guarantee underlying much of applied calculus and optimization theory (see the Optimization entry).
Every differentiable function is automatically continuous (see the Derivative entry) — but the reverse is false. The function f(x)=|x| is continuous everywhere but not differentiable at x=0 (there's a sharp corner, with no single well-defined tangent slope there).
Ordinary continuity is a "pointwise" property — the required δ (from the epsilon-delta definition) can depend on which point a you're considering. Uniform continuity demands a single δ that works simultaneously for every point in the domain, for any given ε. Every continuous function on a closed, bounded interval is automatically uniformly continuous (a consequence of the Heine-Cantor theorem), but this equivalence fails on unbounded or open domains — for instance, f(x)=x² is continuous everywhere on ℝ but not uniformly continuous, since the function's steepness grows without bound as x increases.
As introduced in the Topological Space entry, continuity generalises far beyond ordinary calculus: a function between topological spaces is continuous if the preimage of every open set is open. This definition requires no notion of distance whatsoever, and reduces exactly to the familiar epsilon-delta definition when applied to ordinary Euclidean space — demonstrating that continuity is fundamentally a much more general, topological concept, with the calculus version being merely a specific, familiar special case.
As mentioned in the Derivative entry, Karl Weierstrass (1872) constructed a function that is continuous at every point yet differentiable at no point whatsoever — an infinite sum of cosine waves at ever-increasing frequencies, creating an infinitely jagged curve with no smooth tangent anywhere. This deeply counterintuitive example shocked 19th-century mathematicians who had widely (and, it turned out, mistakenly) assumed continuous functions must be differentiable "almost everywhere," and it played a significant historical role in motivating the fully rigorous, modern epsilon-delta foundations of real analysis.
An even stronger condition than uniform continuity, Lipschitz continuity requires |f(x)−f(y)| ≤ L|x−y| for some fixed constant L and all x,y in the domain — bounding how fast the function can possibly change, in absolute terms. This condition is exactly what the Picard–Lindelöf existence-and-uniqueness theorem for differential equations (see that entry) requires, making Lipschitz continuity a key technical bridge connecting the theories of continuity and differential equations.
As covered in the Complex Number entry, complex-differentiable (holomorphic) functions are vastly more constrained than merely continuous ones — a striking illustration that "continuity" and its stronger relatives (differentiability, analyticity) can behave in dramatically different ways depending on whether the underlying number system is the real line or the complex plane.