Geometric Meaning of the Derivative

From Difference Quotients to Secant Slopes

In “Definition of the Derivative,” differentiability at an interior point $a$ was expressed as the existence of a limit of difference quotients. Each quotient also has a direct geometric meaning: it is the slope of a secant line through two points on the graph. When the second point approaches the first, the limiting slope describes the tangent line, if that limit exists and is finite.

Let $I$ be an interval, let $f:I\to\mathbb{R}$, and let $a$ be an interior point of $I$. The point on the graph above $a$ is $P=(a,f(a))$. For a sufficiently small nonzero increment $h$, a second point on the graph is $Q_h=(a+h,f(a+h))$. The line through $P$ and $Q_h$ is called a secant line.

\text{slope of the secant line }PQ_h =\frac{f(a+h)-f(a)}{(a+h)-a} =\frac{f(a+h)-f(a)}{h}.

The denominator is the horizontal change between the two points, and the numerator is their vertical change. Thus the difference quotient is not merely an algebraic expression: it records the secant line’s slope. As $h\to0$, $Q_h$ approaches $P$. If the secant slopes approach a finite number, that number is the slope of the tangent line in the usual finite-slope sense.

Definition: If $f$ is differentiable at an interior point $a$, the tangent line to the graph of $f$ at $(a,f(a))$ is the line through $(a,f(a))$ with slope $f'(a)$. Its equation is $$ y=f(a)+f'(a)(x-a). $$

The equation follows from the point-slope form of a line: a line of slope $m$ through $(a,f(a))$ has equation $y-f(a)=m(x-a)$. The derivative supplies $m$. This definition describes the tangent by the limiting slope of nearby secants; it does not say that the tangent line must meet the graph only at the point of tangency.

Secant Slopes and the Tangent Line

Theorem (Secant-Slope Characterization of the Tangent): Let $f:I\to\mathbb{R}$, where $I$ is an interval, and let $a$ be an interior point of $I$. The secant slopes $$ \frac{f(a+h)-f(a)}{h} $$ approach a finite limit $m$ as $h\to0$, $h\ne0$, if and only if $f$ is differentiable at $a$. In that case $m=f'(a)$, and the limiting tangent line is $$ y=f(a)+m(x-a). $$

Proof. By the definition of the derivative, $f$ is differentiable at $a$ precisely when the displayed difference quotient has a finite limit as $h\to0$, $h\ne0$. When that limit exists, its value is, by definition, $f'(a)$. The secant line through $P=(a,f(a))$ and $Q_h=(a+h,f(a+h))$ has slope equal to that quotient, so its equation is

y=f(a)+\frac{f(a+h)-f(a)}{h}(x-a).

As $h\to0$, the coefficient of $x-a$ tends to $m=f'(a)$, while the line continues to pass through $P$. Its limiting line is therefore $y=f(a)+m(x-a)$, as claimed. Conversely, if the secant slopes approach a finite limit $m$, then the defining difference quotient tends to $m$. Hence $f$ is differentiable at $a$ and $f'(a)=m$. $\square$

This result makes the geometric interpretation precise. A derivative exists exactly when the nearby secant slopes settle to one finite value. The tangent line is not chosen by visual inspection: its slope is fixed by the limit. The argument uses the pointwise definition from the previous tutorial, viewed through the geometry of the graph.

Theorem (Uniqueness of the Finite-Slope Tangent): If $f$ is differentiable at $a$, there is exactly one line through $(a,f(a))$ whose slope is the limit of the secant slopes. Its slope is $f'(a)$.

Proof. By the secant-slope characterization, the secant slopes tend to $f'(a)$, so the line through $(a,f(a))$ with that slope is such a line. Now suppose a line through $(a,f(a))$ has slope $M$ and is the limiting line of the secants. Its slope must equal the limit of their slopes. Since those slopes tend to $f'(a)$, it follows that $M=f'(a)$. A line through a fixed point is determined by its slope, so this line is unique. $\square$

Worked Examples: Finding Tangents from Slopes

Worked Example: A Tangent Line to a Cubic

Let $f(x)=x^3-2x$, and find the tangent line at $a=1$. First, $f(1)=1-2=-1$. For $h\ne0$, the secant slope is

\frac{f(1+h)-f(1)}{h} =\frac{(1+h)^3-2(1+h)-(-1)}{h}.

Expanding the numerator gives $1+3h+3h^2+h^3-2-2h+1=h+3h^2+h^3$. Therefore

\frac{f(1+h)-f(1)}{h} =\frac{h+3h^2+h^3}{h} =1+3h+h^2.

As $h\to0$, this slope tends to $1$, so $f'(1)=1$. The tangent line passes through $(1,-1)$ and has slope $1$, giving

y-(-1)=1(x-1), \qquad\text{so}\qquad y=x-2.

The equation can be checked at the point of tangency: when $x=1$, the line gives $y=-1=f(1)$.

Worked Example: A Horizontal Tangent That Crosses the Graph

Consider $g(x)=x^3$ at $a=0$. Here $g(0)=0$, and for $h\ne0$,

\frac{g(0+h)-g(0)}{h} =\frac{h^3}{h} =h^2.

Since $h^2\to0$, the tangent slope is $g'(0)=0$. Its equation is $y=0$, the $x$-axis. For $x>0$, $x^3>0$; for $x<0$, $x^3<0$. Thus the graph lies on opposite sides of the tangent line on the two sides of the point. The tangent line crosses the graph at the origin, and the origin is not a local maximum or a local minimum.

Worked Example: A Corner with Two Different Secant Limits

Let $q(x)=|x-2|$, and examine the graph at $a=2$. Since $q(2)=0$, for $h>0$ we have $q(2+h)=h$, and the secant slope is $h/h=1$. For $h<0$, $q(2+h)=|h|=-h$, so the secant slope is $(-h)/h=-1$. The right-hand secant slopes tend to $1$, while the left-hand secant slopes tend to $-1$.

There is no single finite limit for the two-sided difference quotient, so $q$ is not differentiable at $2$. Geometrically, the graph has two distinct one-sided directions at the corner. A line of slope $1$ describes the right-hand direction, and a line of slope $-1$ describes the left-hand direction; neither is a two-sided tangent line determined by a derivative.

Worked Example: Secants Becoming Nearly Vertical

Consider $r(x)=\sqrt[3]{x}$ at $a=0$. For $h\ne0$, the secant slope is

\frac{r(h)-r(0)}{h} =\frac{\sqrt[3]{h}}{h} =\frac{1}{|h|^{2/3}}.

For the last equality, if $h>0$, then $\sqrt[3]{h}/h=h^{-2/3}=1/|h|^{2/3}$. If $h<0$, write $h=-|h|$; then $\sqrt[3]{h}=-|h|^{1/3}$, so the quotient is $(-|h|^{1/3})/(-|h|)=1/|h|^{2/3}$ as well. This quantity grows without bound as $h\to0$. There is no finite derivative at zero. The secants become increasingly steep, corresponding to a vertical direction, rather than approaching a line of finite slope.

What the Geometric Picture Does—and Does Not—Say

In the usual Cartesian picture, with the same scale on both axes, a line of slope $m$ makes an angle $\theta$ with the positive horizontal direction satisfying $m=\tan\theta$, when that angle is not vertical. The derivative is the slope, not the angle itself. A positive derivative means the nearby secant slopes approach a positive value; a negative derivative means they approach a negative value; and a zero derivative means they approach zero, giving a horizontal tangent.

A horizontal tangent alone does not establish a local maximum or minimum. The cubic example shows why: $g'(0)=0$, but the function values are negative just to the left of zero and positive just to the right. The derivative at one point records the limiting slope there; it does not, by itself, determine whether the function turns around.

Similarly, the word “tangent” should not be taken to mean that the graph cannot cross the line. A tangent is determined by the limiting secant slope, not by a rule that the curve must remain on one side. The graph of $x^3$ crosses its tangent $y=0$, while its derivative at zero exists.

The converse pitfall is to infer differentiability from a graph that merely looks smooth at a particular drawing scale. At a corner, as with $q(x)=|x-2|$, the one-sided slopes can disagree even if the graph appears nearly straight from far away. The two-sided limit in the derivative definition requires both sides to approach the same finite slope. A vertical-looking tangent is also distinct from a finite derivative: the cube-root example has secant slopes growing without bound, not approaching a real number.

The geometric interpretation is useful because it turns an abstract limit into a local question about slopes. Compute the secant slopes, check whether they approach one finite value from both sides, and then use that value with the point $(a,f(a))$ to write the tangent line. The derivative gives the tangent’s direction and slope at the point; further claims about the function nearby require additional reasoning.

Check Your Understanding

Use the secant-slope interpretation and the examples above to answer these questions.

Why is the difference quotient at $a$ the slope of the secant line through $(a,f(a))$ and $(a+h,f(a+h))$?
If $f(a)=3$ and $f'(a)=-2$, what is the equation of the tangent line at $a$?
Why does the graph of $x^3$ at zero show that a horizontal tangent need not occur at a local extremum?
For $q(x)=|x-2|$, what are the right-hand and left-hand limits of the secant slopes at $2$, and what do they imply?
Why does an unbounded secant slope, as in the cube-root example at zero, fail to give a finite derivative?