Instantaneous Rate of Change: From Secant Lines to the Tangent Idea

Why “instantaneous” is different from “average”

You already know how to measure how fast something changes over an interval: pick two input values, find the change in output, and divide by the change in input. That gives an average rate of change over that interval. The new challenge is this: many real situations ask for the rate of change at a single moment or at a single input value. For example, a car’s speedometer reports speed “right now,” not averaged over the last mile. A population model might need the growth rate at year 5, not between years 0 and 10.

But a single point on a graph does not give a slope by itself. Slope needs two points. The key idea is to use two points that are very close together, compute the slope of the line through them (a secant line), and then see what happens as the second point moves closer and closer to the first. The limiting value of those secant slopes is what we mean by the instantaneous rate of change at the point, and geometrically it corresponds to the slope of the tangent line.

Secant lines: the “two-point” slopes that approach a point

Fix a function f and a specific input value a where you want the instantaneous rate of change. Choose a nearby input value a+h (where h can be positive or negative, but not zero). The two points on the graph are (a, f(a)) and (a+h, f(a+h)). The line through them is a secant line, and its slope is

(f(a+h) - f(a)) / ((a+h) - a) = (f(a+h) - f(a)) / h

This expression is called a difference quotient. It measures the average rate of change from a to a+h. The instantaneous rate of change at a is the value these slopes approach as h gets closer to 0.

What you should picture

• Start with a point on the curve at x=a.
• Pick a second point on the curve at x=a+h.
• Draw the secant line through the two points.
• Slide the second point toward the first (h → 0). The secant line rotates and “settles” into a limiting position.
• That limiting line is the tangent line at x=a, and its slope is the instantaneous rate of change.

Even if you are not drawing, the computation mirrors this motion: you compute secant slopes for smaller and smaller values of h and look for a stable value.

Continue in our app.

You can listen to the audiobook with the screen off, receive a free certificate for this course, and also have access to 5,000 other free online courses.

Or continue reading below...

Download the app

Instantaneous rate of change as a limit of secant slopes

We define the instantaneous rate of change of f at x=a (when it exists) as

lim_{h→0} (f(a+h) - f(a)) / h

This is the central object behind derivatives, but you do not need to memorize rules yet. The focus here is the meaning: it is a limit of average rates of change over shrinking intervals. If that limit exists and is finite, it gives a single number that describes how f is changing at exactly x=a.

Units matter

If x is measured in seconds and f(x) is measured in meters, then (f(a+h) - f(a))/h has units of meters per second. The instantaneous rate of change has the same units. Thinking in units helps you interpret the result: a slope of 3 means “about 3 meters of change in output per 1 second of change in input” at that moment.

Practical step-by-step: estimating instantaneous rate from a table or calculator

Often you will not have a formula you can simplify, or you may be working from measured data. You can still estimate instantaneous rate by computing secant slopes with small h values.

Step-by-step method (numerical secant slopes)

• Choose the point a where you want the instantaneous rate.
• Pick several small h values (for example 0.1, 0.01, 0.001) and also negative ones (−0.1, −0.01, −0.001) if possible.
• Compute the secant slope (f(a+h) − f(a))/h for each h.
• Look for the values to stabilize as h gets closer to 0 from both sides.
• If left-side and right-side values approach the same number, that number is your estimate of the instantaneous rate.

Using both positive and negative h is important because it checks whether the slope approaches the same value from the left and from the right. If they approach different values, the instantaneous rate at that point does not exist (at least not as a single slope).

Example: f(x)=x^2 at a=3 (numerical approach)

We know the graph is smooth, so we expect a single tangent slope. Let’s estimate it using secant slopes.

Compute f(3)=9. Now compute slopes for several h values.

h = 0.1:  (f(3.1)-f(3))/0.1 = (9.61-9)/0.1 = 6.1  (since 3.1^2=9.61)  h = 0.01: (f(3.01)-f(3))/0.01 = (9.0601-9)/0.01 = 6.01  h = 0.001:(f(3.001)-f(3))/0.001 = (9.006001-9)/0.001 = 6.001  h = -0.1: (f(2.9)-f(3))/(-0.1) = (8.41-9)/(-0.1) = 5.9  h = -0.01:(f(2.99)-f(3))/(-0.01) = (8.9401-9)/(-0.01) = 5.99  h = -0.001:(f(2.999)-f(3))/(-0.001) = (8.994001-9)/(-0.001) = 5.999

From the right, the slopes are 6.1, 6.01, 6.001. From the left, they are 5.9, 5.99, 5.999. Both sides are closing in on 6. So the instantaneous rate of change at x=3 is 6, and the tangent line there has slope 6.

Practical step-by-step: finding the tangent slope from an explicit formula

When you do have a formula, you can often simplify the difference quotient before taking the limit. This is not about heavy algebra tricks; it is about removing the “0/0 obstacle” that appears when you substitute h=0 too early.

Step-by-step method (symbolic difference quotient)

• Write the difference quotient: (f(a+h) − f(a))/h.
• Substitute the function expression into f(a+h) and f(a).
• Simplify the numerator so that a factor of h appears.
• Cancel the factor of h (allowed because h is not 0 during simplification).
• Now evaluate the limit as h → 0 by substituting h=0 into the simplified expression.

Example: f(x)=x^2 at a=3 (symbolic approach)

Start with the difference quotient.

(f(3+h) - f(3)) / h = ((3+h)^2 - 3^2) / h

Expand and simplify the numerator.

((9 + 6h + h^2) - 9) / h = (6h + h^2) / h

Factor out h and cancel.

h(6 + h) / h = 6 + h

Now take the limit as h → 0.

lim_{h→0} (6 + h) = 6

This matches the numerical estimate and shows how the tangent slope emerges from the secant slopes as the interval shrinks.

From slope to the tangent line equation

Once you have the instantaneous rate of change at x=a, you have the slope m of the tangent line at that point. The tangent line touches the curve at (a, f(a)) and has slope m. Using point-slope form, the tangent line is

y - f(a) = m(x - a)

This line is a powerful local approximation: near x=a, the function behaves similarly to its tangent line. In many applications, you use the tangent line to estimate function values for inputs close to a.

Example: tangent line to f(x)=x^2 at a=3

We found m=6 and f(3)=9. The tangent line is

y - 9 = 6(x - 3)

So

y = 6x - 9

As a quick check of the “local approximation” idea, compare f(3.02)=3.02^2=9.1204 with the tangent estimate y=6(3.02)-9=9.12. The tangent line is very close for inputs near 3.

When instantaneous rate of change fails: corners, cusps, and vertical tangents

Not every graph has a well-defined tangent slope at every point. The secant slopes might approach different values from the left and right, or they might grow without bound. Understanding these cases is essential because it prevents you from assuming “every point has a slope.”

Corners: left and right slopes disagree

A classic example is f(x)=|x| at a=0. The graph has a sharp corner at the origin. Compute the secant slope:

(f(0+h) - f(0)) / h = (|h| - 0) / h = |h|/h

If h>0, then |h|/h=1. If h<0, then |h|/h=−1. The slopes approach 1 from the right and −1 from the left, so there is no single limiting slope. That means there is no instantaneous rate of change at x=0 for f(x)=|x|, and there is no single tangent line there.

Cusps: slopes blow up in opposite directions

Some graphs come to a sharp point where the slope becomes extremely steep, with left and right behavior not matching in a finite way. In such a case, secant slopes may become very large in magnitude as h→0, and the notion of a single finite instantaneous rate fails.

Vertical tangents: slopes blow up to infinity

Sometimes the graph is smooth but becomes vertical at a point. Then secant slopes can grow without bound and approach ±∞. In that situation, you may say the tangent line is vertical, but the instantaneous rate of change as a finite number does not exist. This matters in applications: a vertical tangent corresponds to an output changing extremely rapidly with respect to the input near that point.

One-sided instantaneous rates and why they matter

Even when the full instantaneous rate does not exist, one-sided versions can still be meaningful. You can look at the limit of secant slopes as h→0+ (from the right) or as h→0− (from the left):

Right-hand: lim_{h→0+} (f(a+h) - f(a)) / h  Left-hand:  lim_{h→0-} (f(a+h) - f(a)) / h

If these two one-sided limits exist but are different, the function has a corner-like behavior at a. In practical contexts, one-sided rates appear naturally: for example, if a process starts at time a and you can only look forward in time, you are effectively using a right-hand instantaneous rate.

Instantaneous rate as “sensitivity”

Another way to interpret instantaneous rate of change is sensitivity: how sensitive is the output to tiny changes in the input near a? If the instantaneous rate at a is large in magnitude, small input changes cause relatively large output changes near that point. If it is near zero, the output is relatively insensitive near that point.

Example: sensitivity in a simple model

Suppose a model gives the cost C(x) of producing x units. The instantaneous rate of change at x=a represents the marginal cost near a: approximately how much extra cost you incur for producing one more unit when you are already producing a units. Even without naming formal derivative rules, the secant-to-tangent idea explains why “marginal” is about shrinking the interval to focus on the local behavior.

Choosing h in computations: accuracy and pitfalls

When estimating instantaneous rate numerically, smaller h usually gives a better approximation, but there are practical limits. If you use a calculator or computer with limited precision, extremely small h can cause rounding issues because f(a+h) and f(a) may become so close that their difference loses accuracy.

Guidelines for good numerical estimates

• Use a sequence of h values that shrink by a factor of 10 (0.1, 0.01, 0.001) and watch for stabilization.
• Check both sides (positive and negative h) when possible.
• If values bounce around for very small h, try a slightly larger h where the computation is stable.
• Compare with a symmetric secant slope when appropriate: (f(a+h) − f(a−h)) / (2h). This often reduces numerical error because it balances left and right behavior.

Example: symmetric secant slope as a tangent estimate

For f(x)=x^2 at a=3 and h=0.01:

(f(3.01) - f(2.99)) / (0.02) = (9.0601 - 8.9401) / 0.02 = 0.12 / 0.02 = 6

Here it lands exactly on 6 because of the symmetry of the square function, but even for other functions it often provides a strong estimate.

Connecting the geometry to real motion

Think of x as time and f(x) as position along a line. A secant slope over [a, a+h] is average velocity over that time interval. As the interval shrinks, the average velocity approaches the instantaneous velocity at time a. The tangent slope is therefore the velocity “at that instant.”

This viewpoint also clarifies sign: if the tangent slope is positive, position is increasing at that moment (moving forward). If it is negative, position is decreasing (moving backward). If it is zero, the object is momentarily not changing position with respect to time (though it might still be about to change direction).

Practice workflow: from a graph to a tangent idea without formulas

Sometimes you only have a graph. You can still approximate the instantaneous rate of change at a point by drawing or imagining secant lines close to the point and estimating their slopes.

Step-by-step method (graph-based estimate)

• Mark the point of interest (a, f(a)) on the graph.
• Choose a nearby point to the right and estimate its coordinates from the grid; compute the secant slope.
• Choose a nearby point to the left and compute the secant slope.
• Move the nearby points closer and repeat until the slopes stabilize.
• Use the stabilized value as the tangent slope estimate.

When doing this by hand, the main skill is reading coordinates carefully and choosing points close enough to reflect local behavior but far enough apart that you can read them reliably from the grid.