Rolle's and Lagrange's Theorems Explained Simply: Hypotheses, Intuition and Worked Examples

With Rolle and Lagrange, students almost never get the statement wrong: they know it by heart. They get the exercise built on top of it wrong. In the fifth-year tests we mark, the typical question is "check whether $f$ satisfies the hypotheses of Lagrange's theorem on $[a,b]$ and, if so, find the points $c$". Three students out of four launch straight into calculating $c$, skipping the check on the hypotheses — which is exactly the part the teacher wanted to see. Here you'll find both theorems with the geometric idea behind them, the proof stripped to its essentials, worked examples with numbers, and the list of mistakes that cost marks.

What does Rolle's Theorem say?

Let $f$ be a function that is continuous on the closed interval $[a,b]$ and differentiable on the open interval $(a,b)$, with $f(a) = f(b)$. Then there is at least one point $c \in (a,b)$ such that

The idea is almost physical. If you start at a certain height and, after a stretch, find yourself back at the same height, with no jumps and no sharp corners, then somewhere along the way you must have stopped climbing and started descending (or the other way round). At that point the tangent is horizontal: the derivative is zero.

The proof is shorter than the textbook makes it look. By Weierstrass's theorem, $f$ continuous on a closed interval has a maximum $M$ and a minimum $m$. If $M = m$ the function is constant and $f'$ is zero everywhere: any $c$ will do. Otherwise, at least one of the two extreme values is attained at an interior point (if both were attained only at the endpoints, where $f$ takes the same value, we would have $M = m$). At an interior maximum or minimum point where $f$ is differentiable, Fermat's theorem says that $f'(c) = 0$. Done.

Worked example. $f(x) = x^2 - 4x + 3$ on $[1,3]$. It's a polynomial: continuous and differentiable everywhere, hypotheses satisfied. $f(1) = 1 - 4 + 3 = 0$ and $f(3) = 9 - 12 + 3 = 0$, so $f(1) = f(3)$. Let's find $c$:

and $2 \in (1,3)$. It's the vertex of the parabola, as you'd expect.

When it doesn't work. $f(x) = |x|$ on $[-1,1]$: continuous, $f(-1) = f(1) = 1$, but not differentiable at $x = 0$, which lies inside the interval. Indeed $f'(x)$ is $-1$ to the left and $+1$ to the right, and never vanishes. One sharp corner is enough to break the theorem — and the conclusion along with it.

What does Lagrange's Theorem (or the Mean Value Theorem) say?

Let $f$ be continuous on $[a,b]$ and differentiable on $(a,b)$. Then there is at least one $c \in (a,b)$ such that

On the right is the difference quotient: the slope of the chord joining the points $(a, f(a))$ and $(b, f(b))$. On the left, the slope of the tangent at $c$. The theorem guarantees that somewhere between $a$ and $b$, the tangent is parallel to the chord. Compared with Rolle, the hypothesis $f(a) = f(b)$ has been dropped: the chord is no longer horizontal, but tilted.

The name "mean value" comes from the kinematic reading: if $f(t)$ is the distance travelled at time $t$, the difference quotient is the average speed between $a$ and $b$, and $f'(c)$ is the instantaneous speed. The theorem says that at some instant you travelled at exactly your average speed. It's the principle the highway Speed Tutor relies on, and if you're curious about the application, we cover it in our article on Lagrange's Theorem and speed checks.

Worked example. $f(x) = x^3$ on $[0,2]$. Polynomial: hypotheses satisfied. The difference quotient is

Set $f'(x) = 3x^2 = 4$, so $x^2 = \frac{4}{3}$ and $x = \pm\frac{2}{\sqrt{3}} = \pm\frac{2\sqrt{3}}{3} \approx \pm 1.155$. The negative solution is not in $(0,2)$ and must be discarded: the only Lagrange point is $c = \frac{2\sqrt{3}}{3}$. Writing both roots as the answer is a mistake we see in one test out of three on this type of exercise.

An example that tests the hypotheses. $f(x) = \sqrt{x}$ on $[0,4]$. The function is continuous on $[0,4]$, but at $x = 0$ it is not differentiable (the tangent is vertical). Does the theorem still apply? Yes: differentiability is only required on the open interval $(0,4)$, and $0$ is not in it. The difference quotient is $\frac{2 - 0}{4} = \frac{1}{2}$, and

How are Rolle's Theorem and Lagrange's Theorem related?

They are the same theorem seen from two angles. Rolle is Lagrange in the case $f(a) = f(b)$: the difference quotient becomes $\frac{0}{b-a} = 0$ and the conclusion reduces to $f'(c) = 0$. But the textbook proves Lagrange using Rolle, and this is where students get lost, because an "auxiliary function" appears out of nowhere with no one explaining where it comes from.

Here is where it comes from: take $f$ and subtract the chord from it. The chord is the line through $(a, f(a))$ with slope $\frac{f(b)-f(a)}{b-a}$, so the auxiliary function is

Now $g$ measures the vertical distance between the curve and the chord. At the endpoints, this distance is zero: $g(a) = f(a) - f(a) = 0$ and $g(b) = f(b) - f(a) - \big(f(b) - f(a)\big) = 0$. Moreover, $g$ is continuous and differentiable exactly where $f$ is, because we have only subtracted a straight line. So $g$ satisfies the hypotheses of Rolle's Theorem: there is a $c$ with $g'(c) = 0$, that is

which is exactly Lagrange's conclusion. Four lines. What you should take away is not the formula for $g$ but the idea: Lagrange is Rolle applied to "$f$ minus the chord". Anyone who remembers it this way can reconstruct the proof even in a surprise oral test. The same trick, with a slightly more elaborate auxiliary function, proves Cauchy's theorem, from which L'Hôpital's rule then follows.

How do you check whether the theorem applies?

This is the standard exercise, and the marks are almost always split between checking the hypotheses and calculating $c$. The protocol we have our students follow has four steps, always in the same order:

1. Domain. Is the whole interval $[a,b]$ inside the domain of $f$? If not, stop: the theorem does not apply.
2. Continuity on $[a,b]$, closed. For polynomials, exponentials, sine and cosine, it's enough to state it; for piecewise functions you have to check the join point using the left- and right-hand limits.
3. Differentiability on $(a,b)$, open. Here too, in piecewise functions the critical point is the join: the derivatives of the two expressions must match there.
4. For Rolle only: $f(a) = f(b)$. Then solve $f'(c) = 0$ or $f'(c) = \frac{f(b)-f(a)}{b-a}$ and keep only the solutions that lie inside the interval.

If calculating the derivative slows you down, our guide to differentiation rules with worked examples covers every case you need here.

Worked example with a piecewise function. Let's check Lagrange on $[0,2]$ for

Continuity at $x = 1$: from the left $1^2 = 1$, from the right $2 \cdot 1 - 1 = 1$, and $f(1) = 1$. Continuous. Differentiability at $x = 1$: from the left $f'(x) = 2x \to 2$, from the right $f'(x) = 2$. They match, so $f$ is differentiable throughout $(0,2)$. Hypotheses satisfied. Difference quotient: $f(2) = 3$, $f(0) = 0$, so $\frac{3}{2}$. Let's look for $c$ piece by piece: on $(0,1)$ we solve $2x = \frac{3}{2}$, giving $x = \frac{3}{4}$, which lies in the right interval; on $(1,2)$ we would need $2 = \frac{3}{2}$, impossible. The only point is $c = \frac{3}{4}$.

An example where it does not apply. $f(x) = \frac{1}{x}$ on $[-1,2]$. Zero lies inside the interval and $f$ isn't even defined there: continuity fails and the theorem does not apply. A student who skips the check calculates the difference quotient, $\frac{\frac{1}{2} - (-1)}{3} = \frac{1}{2}$, sets $-\frac{1}{c^2} = \frac{1}{2}$, finds no solutions and writes "no $c$ exists". The conclusion happens to be true, but the answer is wrong: the question was about the hypotheses, and the exercise is worth zero.

What are the most common mistakes with Rolle and Lagrange?

Marking tests and Maturità mock papers, the mistakes cluster around five points:

1. Hypotheses stated, not checked. "$f$ is continuous and differentiable" written under a piecewise function without looking at the join. For a polynomial that sentence is enough; for a piecewise function, an absolute value or a root, it is not.
2. Closed and open swapped. Continuity is needed on the closed interval $[a,b]$, differentiability only on the open interval $(a,b)$. Anyone who says that $\sqrt{x}$ on $[0,4]$ "doesn't satisfy the hypotheses because it isn't differentiable at $0$" has mixed up the brackets: $0$ is an endpoint, and differentiability is not required at the endpoints.
3. Rolle applied without $f(a) = f(b)$, or Lagrange rejected because $f(a) \ne f(b)$. These are two different theorems, with one extra hypothesis on one side: the check on the endpoint values must always be done, and written down.
4. Out-of-interval solutions kept. In $x^3$ on $[0,2]$, $c = -\frac{2\sqrt{3}}{3}$ must be discarded. The line "$c \in (a,b)$?" is the last step of the process and the one most often forgotten.
5. "There is a point" read as "there is only one point". The theorem says at least one. $f(x) = x^3 - 3x$ on $[-2,2]$ has difference quotient $\frac{2 - (-2)}{4} = 1$, and $3x^2 - 3 = 1$ gives $x = \pm\frac{2\sqrt{3}}{3} \approx \pm 1.155$: both interior, both must be reported.

What are these theorems actually good for?

They hold up half of the fifth-year syllabus. The criterion you use every day in curve sketching — "if $f' > 0$ then $f$ is increasing" — is not self-evident, it's a consequence of Lagrange: take $x_1 < x_2$, there is a $c$ between them with $f(x_2) - f(x_1) = f'(c)\,(x_2 - x_1)$, and if $f'(c) > 0$ the difference is positive. In the same way, "zero derivative on an interval implies a constant function" comes from here, and it is the foundation of integral calculus.

There's also a use as a hand calculator, which comes up in Maturità questions. With $f(x) = \sqrt{x}$ on $[25,26]$, Lagrange gives $\sqrt{26} - 5 = \frac{1}{2\sqrt{c}}$ for some $c$ between $25$ and $26$. Since $\sqrt{c}$ lies between $5$ and $\sqrt{26}$, the fraction is between $\frac{1}{2\sqrt{26}} \approx 0.098$ and $\frac{1}{10} = 0.1$, so

The true value is $5.099$: an estimate accurate to two decimal places without a calculator, using one theorem and one line of arithmetic.

Rolle and Lagrange don't end with the school-leaving exam: the proof of Lagrange via Rolle is among those that come up most regularly in first-year Calculus 1 oral exams at university, and anyone who arrives there able to reconstruct it, not just recite it, has a real advantage. If you're already thinking about university admission, our preparation paths start precisely by consolidating these foundations.

If instead the problem lies further back — derivatives come out wrong, piecewise functions cause trouble, checking the hypotheses feels like a meaningless ritual — a dedicated tutor can put the pieces back in order within a few weeks. Tell us about your situation and book your first lesson: we'll start from an exercise like the ones above and see where it gets stuck.

FAQ