§ 01The Intuitive Idea
Imagine throwing a ball straight up into the air and catching it at exactly the same height you released it. At some point during the flight, the ball was neither going up nor going down, it was momentarily stationary at the peak. Its velocity was zero. Rolle's Theorem is the mathematical statement of exactly this observation.
More precisely: if a smooth, unbroken curve begins and ends at the same height, it cannot simply travel from one endpoint to the other without ever having a horizontal tangent. It must either rise to a peak or fall to a trough, or do both, and at any turning point the tangent is exactly horizontal.
This idea was first stated by Michel Rolle in 1691, making it one of the oldest theorems in calculus. Ironically, Rolle himself was a vocal critic of infinitesimal calculus at the time, and he presented the result using polynomial methods rather than derivatives. The modern formulation in terms of differentiability came later, as calculus was placed on rigorous foundations. Today, Rolle's Theorem is primarily studied as a stepping stone to the Mean Value Theorem: but it is also a powerful tool in its own right for proving that functions have at most one root in an interval, and for locating hidden critical points.
§ 02Theorem Statement — All Three Conditions
Rolle's Theorem has exactly three hypotheses. All three must hold before the conclusion can be claimed. The theorem is an existence theorem, it guarantees that at least one point c with the required property exists, but it does not say how many there are or where they are located.
- f is continuous on the closed interval [a, b]
- f is differentiable on the open interval (a, b)
- f(a) = f(b) (the function returns to its starting value at the right endpoint)
In geometric terms: there is at least one point on the curve between x = a and x = b where the tangent line is perfectly horizontal.
The three conditions each play a distinct role in forcing the conclusion. Notice the careful asymmetry: continuity is required on the closed interval [a, b] (including the endpoints), while differentiability is only required on the open interval (a, b) (excluding the endpoints). This is deliberate: the function is allowed to have corners or cusps at the endpoints a and b, but must be smooth throughout the interior. Differentiability at the endpoints themselves is not needed and is not assumed.
§ 03Geometric Interpretation
The geometric picture is as clear as a theorem can be. Draw any smooth curve from the point (a, k) to the point (b, k), both at the same height k. The curve must turn around at least once as it moves from one endpoint to the other. At every turning point, the tangent is horizontal. Rolle's Theorem guarantees at least one such turning point exists.
The theorem says at least one such c exists, but there may be many. For a function that oscillates several times between two equal endpoints, every local maximum and every local minimum in the interior provides a valid c. The theorem makes no claim about uniqueness, that is a separate question to investigate for each specific function.
§ 04Proof of Rolle's Theorem
The proof uses two foundational results: the Extreme Value Theorem (a continuous function on a closed bounded interval attains its absolute maximum and minimum) and Fermat's Theorem (if f has a local extremum at an interior point c and f′(c) exists, then f′(c) = 0).
Since f is continuous on [a, b], the Extreme Value Theorem guarantees that f attains its absolute maximum M and absolute minimum m on [a, b]. We consider two cases.
If both the maximum M and the minimum m are attained only at the endpoints a and b, then since f(a) = f(b) we have M = m = f(a) = f(b). This means f is constant on [a, b]: f(x) = f(a) for all x ∈ [a, b]. Therefore f′(x) = 0 for all x ∈ (a, b), and any c ∈ (a, b) works.
If either the maximum or the minimum is attained at some interior point c ∈ (a, b), then f has a local extremum at c. Since f is differentiable on (a, b), f′(c) exists. By Fermat's Theorem, f′(c) = 0. This c is the guaranteed point in the conclusion. ∎
The elegance of this proof lies in its case structure. The first case handles the trivial situation where the function is flat; the second case captures every non-trivial curve by using Fermat's observation that a smooth local extremum always has a horizontal tangent. Together, the two cases are exhaustive and the proof is complete.
§ 0510 Worked Examples
Each example follows the same four-step procedure: verify all three conditions, differentiate, solve f′(c) = 0, and confirm that every solution lies in the open interval (a, b).
Polynomial, continuous and differentiable everywhere. f(1) = 1 − 4 + 3 = 0. f(3) = 9 − 12 + 3 = 0. f(1) = f(3). ✓ All three conditions hold.
c = 2 ∈ (1, 3) ✓. This is the vertex of the parabola, sitting symmetrically between the two roots.
sin x is continuous and differentiable everywhere. f(0) = 0, f(π) = 0. ✓
c = π/2 ≈ 1.571 ∈ (0, π) ✓. This is the maximum of the sine curve on this interval.
Polynomial. f(−1) = −1 + 1 = 0. f(1) = 1 − 1 = 0. f(−1) = f(1) = 0. ✓
Both 1/√3 ≈ 0.577 and −1/√3 ≈ −0.577 lie in (−1, 1). ✓ Two valid values of c.
Polynomial. f(−√2) = 4 − 4 = 0. f(√2) = 4 − 4 = 0. ✓
f′ = 0 at x = −1, 0, 1.
All three (x = −1, 0, 1) lie in (−√2, √2) ≈ (−1.414, 1.414). ✓ Three valid c values.
cos(2x) is continuous and differentiable everywhere. f(0) = cos 0 = 1. f(π) = cos 2π = 1. f(0) = f(π). ✓
x = 0, π/2, π. Only x = π/2 is strictly inside (0, π).
Polynomial. f(2) = 4 − 12 + 8 = 0. f(4) = 16 − 24 + 8 = 0. ✓
c = 3 ∈ (2, 4) ✓. The midpoint of the interval, as always with a symmetric quadratic.
Continuous and differentiable (composition of exponential and polynomial). f(−1) = e¹ − e = 0. f(1) = e¹ − e = 0. ✓
c = 0 ∈ (−1, 1) ✓. By even symmetry this was expected.
Polynomial. f(0) = 0. f(2) = 2(1)(0) = 0. f(0) = f(2). ✓
c₁ = 1 − 1/√3 ≈ 0.423 ∈ (0, 2) ✓
c₂ = 1 + 1/√3 ≈ 1.577 ∈ (0, 2) ✓
f(x) = sin²x − cos²x = −cos(2x) [by double-angle identity]. So f(0) = −cos 0 = −1. f(π) = −cos(2π) = −1. f(0) = f(π). ✓
Only x = π/2 is strictly interior.
Polynomial. f(1) = 0 · 4 = 0. f(3) = 2 · 0 = 0. f(1) = f(3). ✓
x = 3 (endpoint, excluded) or x = 5/3.
c = 5/3 ≈ 1.667 ∈ (1, 3) ✓. x = 3 is excluded because it is an endpoint, not in the open interval.
§ 06Counterexamples — When a Condition Fails
The conditions of Rolle's Theorem are not decorative, each one is genuinely necessary. When any single condition is violated, it is possible to construct a function on [a, b] with f(a) = f(b) and yet no c where f′(c) = 0. The counterexamples below illustrate exactly what can go wrong.
§ 07Relationship to the Mean Value Theorem
Rolle's Theorem and the Mean Value Theorem are not two separate results, they are the same geometric idea at two levels of generality. Understanding this connection clarifies both theorems.
| Feature | Rolle's Theorem | Mean Value Theorem |
|---|---|---|
| Conditions on f | Continuous on [a,b], differentiable on (a,b) | Continuous on [a,b], differentiable on (a,b) |
| Endpoint condition | f(a) = f(b) (equal heights) | No restriction on f(a), f(b) |
| Conclusion | ∃ c ∈ (a,b) with f′(c) = 0 | ∃ c ∈ (a,b) with f′(c) = [f(b)−f(a)]/(b−a) |
| Geometric meaning | Tangent is horizontal somewhere | Tangent is parallel to the secant somewhere |
| Logical relationship | Special case of the MVT (when average slope = 0) | Proved by reducing to Rolle's Theorem |
The proof of the MVT literally uses Rolle's Theorem as its key step: construct the auxiliary function h(x) = f(x) − L(x), where L(x) is the secant line. Then h(a) = h(b) = 0, so Rolle's Theorem applies to h, giving a c where h′(c) = 0. Computing h′(c) then yields the MVT conclusion. Rolle's Theorem is therefore not just historically prior: it is logically prior, serving as the foundation on which the MVT is built.
So f′(c) = 0 — which is exactly Rolle's conclusion.
Rolle's Theorem is the Mean Value Theorem with a zero numerator.
§ 08Using Rolle's Theorem to Count Roots
One of the most powerful practical applications of Rolle's Theorem is proving that a function has exactly one root in an interval: or more generally, bounding the number of roots. The argument is a classic combination of the Intermediate Value Theorem (to prove existence) and Rolle's Theorem (to prove uniqueness).
To show f has at most one root in [a, b]: Suppose f had two distinct roots r₁ < r₂. Then f(r₁) = f(r₂) = 0. By Rolle's Theorem, f′(c) = 0 for some c ∈ (r₁, r₂). If you can prove f′(x) ≠ 0 on the relevant interval, you have a contradiction: so f cannot have two roots.
To show f has exactly one root: Use the Intermediate Value Theorem to prove existence (find x₁ where f < 0 and x₂ where f > 0), then use the Rolle's argument above to prove uniqueness.
f(0) = −1 < 0 and f(1) = 1 + 3 − 1 = 3 > 0. By the Intermediate Value Theorem, f has at least one root in (0, 1).
Suppose f has two distinct roots r₁ < r₂. Then f(r₁) = f(r₂) = 0. By Rolle's Theorem, there exists c ∈ (r₁, r₂) with f′(c) = 0.
f′(x) = 3x² + 3 = 3(x² + 1) ≥ 3 > 0 for all xf′ is strictly positive everywhere: it can never be zero. This contradicts f′(c) = 0. Therefore f cannot have two roots.
§ 09Common Mistakes
§ 1010-Question Quiz
Questions cover all three conditions, finding c, counterexamples, the root-counting technique, and the relationship to the MVT.