Mean Value Theorem — proof, Rolle's theorem: Foundations

This page assumes you have seen nothing. We will name every squiggle used by the topic, draw the picture it stands for, and say why the theorem cannot live without it.

1. Numbers on a line, and the interval $[a, b]$

Two flavours of interval matter, and the whole topic hinges on the difference:

Figure — Mean Value Theorem — proof, Rolle's theorem

Why the topic needs both. Look at the figure. The theorem asks for continuity on the closed $[a, b]$ (so the endpoints count — we compare heights at $a$ and $b$ ) but for the slope to exist only on the open $(a, b)$ (the interior). The special interior point $c$ the theorem produces always lives in the open $(a, b)$ — strictly inside, never at an endpoint. So a common trap is to report $c = a$ or $c = b$ : that is never allowed, because == $c$ can never be an endpoint==.

Recall Why is

$c$ from the open interval? Why not closed? ::: The tangent (slope) is only guaranteed to exist strictly inside, so the conclusion is only claimed there.

2. A function $f$ and its graph

The letters you will meet:

$f (a)$ — the height at the left endpoint (start of the walk).
$f (b)$ — the height at the right endpoint (end of the walk).
$f (a) = f (b)$ — start and end are at the same height (the special case that powers Rolle's theorem).

3. Continuity — "no gaps, no jumps"

Why the topic needs it. The proof of Rolle first calls the Extreme Value Theorem (stated in the next box), which only promises a highest and lowest point if the curve is continuous on a closed bounded interval. A curve with a gap could sneak off to infinity or skip its own peak — no guaranteed top, no flat tangent to find. Continuity is what pins the curve down. See Continuity for the formal limit definition.

4. Slope — steepness as a number

5. The secant line and its slope $\frac{f ( b ) - f ( a )}{b - a}$

Its slope is the average steepness of the whole journey:

Why the topic needs it. This is the entire right-hand side of the MVT equation — the "overall rate" from the core idea at the top. Thinking of a drive: total distance $\div$ total time $=$ average speed. When $f (a) = f (b)$ the rise is $0$ , so the secant slope is $0$ — the secant is horizontal — and MVT collapses into Rolle. That is why Rolle is the flat-secant special case.

Recall What does the secant slope represent physically?

If $f (x)$ is distance and $x$ is time, the secant slope is... ::: the average speed over the whole trip.

6. The tangent line and the derivative $f^{'} (c)$

Let us unpack every piece of this, because the Rolle proof uses it directly:

$h$ — a tiny step to the right ( $h > 0$ ) or left ( $h < 0$ ) of $c$ .
$f (c + h) - f (c)$ — the rise over that tiny step.
$\frac{f ( c + h ) - f ( c )}{h}$ — the slope of a very short secant from $c$ to $c + h$ .
$h \to 0 lim$ — "the value this ratio homes in on as $h$ becomes vanishingly small." It is the answer to the question: what steepness are these short secants approaching?

Why both sides must agree (this is the engine of Rolle). At an interior peak, stepping right gives short secants with slope $\leq 0$ , stepping left gives short secants with slope $\geq 0$ . If $f$ is differentiable both must equal the same $f^{'} (c)$ , so $f^{'} (c) \leq 0$ and $f^{'} (c) \geq 0$ — forcing $f^{'} (c) = 0$ . That squeeze is Fermat's Theorem (interior extrema), and it is why the corner of $∣ x ∣$ (where the two sides disagree) breaks Rolle.

7. The symbols $M$ , $m$ , $c$ , and $\exists$

8. How every foundation feeds the theorem

The diagram below is a dependency flow: follow the arrows to see which idea must be built before which. Ideas at the top are the raw materials; the two theorems at the bottom are the finished products.

Read it top-down: number lines let us name a function; a function has continuity (which feeds the Extreme Value Theorem to guarantee a peak $M$ and valley $m$ ) and slope (which, through the limit, gives the derivative, which feeds Fermat's zero-derivative fact). Extremes plus derivative-zero give Rolle; Rolle plus the secant slope give the MVT.

Equipment checklist

Test yourself — say each answer aloud before revealing it.

I can state the difference between $[a, b]$ and $(a, b)$ ::: closed includes endpoints; open excludes them, and $c$ always lives in the open one.
I can describe what $f (x)$ means as a picture ::: the height of the graph above the ground-position $x$ .
I can say what "continuous" forbids ::: gaps, jumps, holes — you draw it without lifting your pen.
I can state what "smooth / differentiable" adds on top of continuous ::: a single well-defined tangent direction at every point — no sharp corners.
I can state the Extreme Value Theorem in words ::: a continuous function on a closed bounded interval actually reaches its highest and lowest values.
I can compute a secant slope ::: $\frac{f ( b ) - f ( a )}{b - a}$ = rise over run = average steepness.
I can explain why the derivative needs a limit ::: because at a single point the run is $0$ , so we watch short secants as $h \to 0$ .
I can write the derivative definition ::: $f^{'} (c) = lim_{h \to 0} \frac{f ( c + h ) - f ( c )}{h}$ .
I can say why a corner (like $∣ x ∣$ ) is not differentiable ::: the left and right limits of the slope disagree, so there is no single tangent.
I know what $M$ , $m$ , and $\exists c$ stand for ::: max value, min value, and "at least one interior point $c$ ."

Once every reveal feels obvious, return to Mean Value Theorem — proof, Rolle's theorem and the single core sentence at the top will read like plain English. Related deeper roads: Increasing and Decreasing Functions, Cauchy's Mean Value Theorem, Taylor's Theorem, Lipschitz Continuity.