The Tangent Problem
Textbook Reference
| Primary source | OpenStax Calculus Volume 1, Section 2.1: “A Preview of Calculus” |
| Direct link | https://openstax.org/books/calculus-volume-1/pages/2-1-a-preview-of-calculus |
| Supplementary | OpenStax Calculus Volume 1, Section 3.1: “Defining the Derivative” |
| Supplementary link | https://openstax.org/books/calculus-volume-1/pages/3-1-defining-the-derivative |
| Textbook used in class | Stewart, Calculus, Section 1.4: “The Tangent and Velocity Problems” (Examples 1, 2) |
Both OpenStax sources are free and openly licensed.
Key idea
The tangent problem asks a question that seems impossible at first: what is the slope of a curve at a single point? Slope needs two points, and a single point gives you only one. The resolution of that puzzle is the idea that launches all of differential calculus.
A tangent line is the line that just touches a curve at a point and heads in the same direction the curve is heading there. You can see it, but you cannot compute its slope directly, because you only know one point on it. The trick is to sneak up on it. Pick a second, nearby point on the curve and draw the line through both. That line is a secant line, and its slope is easy to compute from the two points. Now slide the second point closer and closer to the first. The secant lines pivot, and they settle toward a limiting line: the tangent. The slope of the tangent is the limit of the secant slopes.
That is the whole idea, and it explains why the tangent problem comes right after limits. Computing the slope of a secant is just the slope formula, $\frac{\text{rise}}{\text{run}}$. Taking the limit as the two points merge is a limit problem, which is why limits had to come first. The slope of the tangent line, defined this way, is the derivative, the central object of Chapter 2. The derivative first appears in its original disguise: as the answer to “how steep is this curve right here?”
Prerequisite Check
Make sure you can do all of the following first:
If reading “what number is this list approaching” from a table feels uncertain, practice that first. The tangent slope is found exactly that way, as the value the secant slopes close in on.
Quick Reference
Secant line. A line through two points on a curve. For $y = f(x)$, the secant through $P(a, f(a))$ and $Q(x, f(x))$ has slope \[ m_{PQ} = \frac{f(x) - f(a)}{x - a}. \]
Tangent line. The line that touches the curve at $P$ and matches its direction there. Its slope is the limit of the secant slopes as $Q$ approaches $P$: \[ m = \lim_{x \to a}\frac{f(x) - f(a)}{x - a}. \]
The procedure.
- Pick the point of tangency $P(a, f(a))$.
- Write the secant slope $m_{PQ}$ using a nearby point $Q(x, f(x))$.
- Take the limit as $x \to a$ (or build a table of $m_{PQ}$ for $x$ near $a$).
- The limit is the tangent slope $m$; write the line with point-slope form.
Key fact. The tangent slope, defined as this limit of secant slopes, is the derivative of $f$ at $a$.
Key Concepts
1. Why a Single Point Is Not Enough
The slope formula needs two points. A tangent line touches the curve at one known point, so naively there is not enough information to find its slope. The way around this is to approximate the tangent with secant lines, which do use two points, and then improve the approximation.
A secant line cuts the curve at two points; a tangent line touches at one. If you fix the point of tangency $P$ and let a second point $Q$ slide along the curve toward $P$, the secant line $PQ$ rotates about $P$ and approaches the tangent. So the tangent is the limiting position of the secants, and its slope is the limit of their slopes.
2. The Parabola Example: Secant Slopes Closing In
The cleanest first example is the parabola $y = x^2$ at the point $(1, 1)$.
Example 1. Find the slope of the tangent line to $y = x^2$ at $P(1, 1)$. (This is Stewart 1.4, Example 1.)
Goal. Write the secant slope through a nearby point $Q(x, x^2)$, then see what it approaches as $x \to 1$.
The secant through $P(1, 1)$ and $Q(x, x^2)$ has slope \[ m_{PQ} = \frac{x^2 - 1}{x - 1}. \] For example, with $Q(1.5, 2.25)$, $m_{PQ} = \dfrac{2.25 - 1}{1.5 - 1} = \dfrac{1.25}{0.5} = 2.5$.
See It: Slide the Secant Into the Tangent
Move the slider to bring the second point toward the first on the graph of $x^2$. The secant line rotates and its slope settles on one value. When the two points meet, the secant has become the tangent. Name the slope the secant slopes approached.
Tabulating $m_{PQ}$ for $x$ approaching $1$ from both sides:
| $x$ | $m_{PQ}$ | $x$ | $m_{PQ}$ | |
|---|---|---|---|---|
| $2$ | $3$ | $0$ | $1$ | |
| $1.5$ | $2.5$ | $0.5$ | $1.5$ | |
| $1.1$ | $2.1$ | $0.9$ | $1.9$ | |
| $1.01$ | $2.01$ | $0.99$ | $1.99$ | |
| $1.001$ | $2.001$ | $0.999$ | $1.999$ |
From both sides the secant slopes close in on $2$, so \[ m = \lim_{x\to 1}\frac{x^2 - 1}{x - 1} = 2. \] With slope $2$ through $(1, 1)$, the tangent line is \[ y - 1 = 2(x - 1), \quad \text{that is,} \quad y = 2x - 1. \]
Boxed answer: the tangent slope is $2$, and the tangent line is $y = 2x - 1$.
Recap. The slope is read off the table as the value both columns approach. It can also be found exactly by simplifying: $\dfrac{x^2 - 1}{x - 1} = x + 1$ for $x \neq 1$, and $x + 1 \to 2$ as $x \to 1$. The table and the algebra agree.
Important: the tangent slope is a limit, not a single secant slope. No individual secant equals the tangent; each is only an approximation. The slope $2$ is what the secant slopes approach as $Q$ merges with $P$. Reporting one secant slope (like $2.1$) as the answer mistakes an approximation for the limit.
3. Estimating a Tangent Slope From Data
Many real functions come as a table of measurements, not a formula. The same secant-slope idea estimates the tangent slope there too.
Example 2. A capacitor discharges through a laser. The charge $Q$ (in coulombs) at time $t$ (in seconds) is measured. Estimate the slope of the tangent line at $t = 0.04$. (This is Stewart 1.4, Example 2.)
| $t$ | $0$ | $0.02$ | $0.04$ | $0.06$ | $0.08$ | $0.1$ |
|---|---|---|---|---|---|---|
| $Q$ | $10$ | $8.187$ | $6.703$ | $5.488$ | $4.493$ | $3.676$ |
Goal. Compute secant slopes from $P(0.04, 6.703)$ to neighboring data points, then estimate the limiting slope.
For the point $R(0, 10)$: \[ m_{PR} = \frac{10 - 6.703}{0 - 0.04} = -82.425. \] The secant slopes from the two closest points (at $t = 0.02$ and $t = 0.06$) are $-74.20$ and $-60.75$. The tangent slope lies between them; averaging gives \[ \frac{1}{2}(-74.20 - 60.75) = -67.475 \approx -67.5. \]
Boxed answer: the tangent slope at $t = 0.04$ is about $-67.5$ (interpreted physically as the current, in amperes, flowing from the capacitor).
Recap. With data instead of a formula, you cannot take an exact limit, so you estimate: compute the secant slopes on each side of the point and average the two closest. The negative slope reflects the charge decreasing over time.
4. The Tangent Slope Is the Derivative
The limit you just computed has a name and a destiny.
The slope of the tangent line to $y = f(x)$ at $x = a$ is, by definition, \[ m = \lim_{x\to a}\frac{f(x) - f(a)}{x - a}. \] This limit is the derivative of $f$ at $a$, written $f'(a)$. Everything in Chapter 2 grows from it: rules for computing $f'(a)$ quickly, the interpretation of $f'$ as a rate of change, and the applications to motion, optimization, and graphing.
There is also a companion problem, the velocity problem, that produces the exact same limit in a different costume: average velocities over shrinking time intervals approach an instantaneous velocity, and that instantaneous velocity is the slope of the tangent to the position graph. The tangent problem and the velocity problem are two readings of one piece of mathematics.
the tangent line cannot be computed because you only have one point.
This is the limit-as-unreachable-barrier error. The tangent slope cannot be found by plugging in a single point directly -- that is true. But the limit process provides an exact answer by approaching the point from two sides. For $y = x^2$ at $x = 1$: the secant slopes $\frac{x^2 - 1}{x - 1} = x + 1$ approach $2$ as $x$ approaches $1$. The “one-point barrier” is bypassed by a limit, not by having a second point. The tangent slope $2$ is not an approximation; it is the exact limit of all the secant slopes.
a large function value at a point means a steep tangent there.
This is the height-vs-slope error. For $f(x) = 1000 - x^2$, the function value at $x = 0$ is $f(0) = 1000$ (very large) but the tangent at $x = 0$ is horizontal (slope $= 0$) because the parabola has its peak there. Conversely, $f(x) = x^3$ has $f(0) = 0$ (very small) but a nonzero tangent slope of $0$ there too -- and at $x = 2$, $f(2) = 8$ while $f'(2) = 12$. The height and slope are independent measurements. Predict the slope by asking how fast the function is rising or falling, not by looking at how high it is.
Common Errors Summary
| Error | Example | Correction |
|---|---|---|
| Reporting a secant slope as the tangent slope | answering $2.1$ for the parabola | The tangent slope is the limit, $2$, not any single secant |
| Using only one nearby point | estimating from one secant on data | Use points on both sides and average the closest two |
| Plugging $x = a$ into the secant slope formula | computing $\frac{0}{0}$ at $x = 1$ | Take the limit (simplify first); the formula is undefined exactly at $a$ |
| Forgetting point-slope form | giving only the slope | The tangent line needs the slope and the point: $y - y_1 = m(x - x_1)$ |
| Misreading the sign of a decreasing slope | reporting $+67.5$ for the discharging capacitor | The charge decreases, so the slope is negative |
Leveled Practice
Level 1 -- Direct Application
Problem 1. Write the secant slope $m_{PQ}$ for $y = x^2$ between $P(2, 4)$ and $Q(x, x^2)$, and simplify.
Show answer
\[ m_{PQ} = \frac{x^2 - 4}{x - 2} = \frac{(x-2)(x+2)}{x-2} = x + 2 \quad (x \neq 2). \]
Boxed answer: $m_{PQ} = x + 2$ for $x \neq 2$.
Problem 2. Using the result of Problem 1, find the slope of the tangent line to $y = x^2$ at $P(2, 4)$.
Show answer
Take the limit of the simplified secant slope as $x \to 2$: \[ m = \lim_{x\to 2}(x + 2) = 4. \]
Boxed answer: the tangent slope is $4$. (The tangent line is $y - 4 = 4(x - 2)$, i.e. $y = 4x - 4$.)
Problem 3. From the parabola table in Example 1, the secant slopes from the left ($x = 0.9, 0.99, 0.999$) are $1.9, 1.99, 1.999$. What tangent slope do these suggest, and is it consistent with the right-side values?
Show answer
The left-side slopes approach $2$. The right-side slopes ($3, 2.5, 2.1, 2.01, 2.001$) also approach $2$. Both sides agree.
Boxed answer: the tangent slope is $2$, confirmed from both sides. (Agreement of the one-sided behavior is what lets you trust the limit.)
Level 2 -- Multiple Steps
Problem 4. The point $P(2, -1)$ lies on the curve $y = \dfrac{1}{1 - x}$. Find the secant slope $m_{PQ}$ for $Q\left(x, \dfrac{1}{1-x}\right)$ and simplify.
Show answer
\[ m_{PQ} = \frac{\frac{1}{1-x} - (-1)}{x - 2} = \frac{\frac{1}{1-x} + 1}{x - 2}. \] Combine the numerator over the common denominator $1 - x$: \[ \frac{1}{1-x} + 1 = \frac{1 + (1-x)}{1-x} = \frac{2 - x}{1-x}. \] So \[ m_{PQ} = \frac{\frac{2-x}{1-x}}{x - 2} = \frac{2-x}{(1-x)(x-2)} = \frac{-(x-2)}{(1-x)(x-2)} = \frac{-1}{1-x} = \frac{1}{x-1}. \]
Boxed answer: $m_{PQ} = \dfrac{1}{x - 1}$ for $x \neq 2$. (This is Stewart 1.4, Exercise 3.)
Problem 5. Using the result of Problem 4, guess the slope of the tangent line to $y = \dfrac{1}{1-x}$ at $P(2, -1)$, then write the tangent line.
Show answer
Take the limit as $x \to 2$: \[ m = \lim_{x\to 2}\frac{1}{x - 1} = \frac{1}{2 - 1} = 1. \] Tangent line through $(2, -1)$ with slope $1$: \[ y - (-1) = 1\,(x - 2), \quad \text{i.e.} \quad y = x - 3. \]
Boxed answer: tangent slope $1$; tangent line $y = x - 3$. (This is Stewart 1.4, Exercise 3, parts (b) and (c).)
Problem 6. A tank drains over half an hour. The volume $V$ (gallons) remaining after $t$ minutes is given. Estimate the slope of the tangent line at $P(15, 250)$ by averaging two secant slopes.
| $t$ | $5$ | $10$ | $15$ | $20$ | $25$ | $30$ |
|---|---|---|---|---|---|---|
| $V$ | $694$ | $444$ | $250$ | $111$ | $28$ | $0$ |
Show answer
Compute the secant slopes from $P(15, 250)$ to the two closest points, at $t = 10$ and $t = 20$: \[ m_{P,\,t=10} = \frac{444 - 250}{10 - 15} = \frac{194}{-5} = -38.8, \] \[ m_{P,\,t=20} = \frac{111 - 250}{20 - 15} = \frac{-139}{5} = -27.8. \] Average the two: \[ \frac{1}{2}(-38.8 - 27.8) = -33.3. \]
Boxed answer: the tangent slope at $t = 15$ is about $-33.3$ gallons per minute (the drain rate at $15$ minutes). (This is Stewart 1.4, Exercise 1.)
Level 3 -- Deeper Problems
Problem 7. For $y = x^2$ at $P(a, a^2)$, show algebraically that the tangent slope is $2a$.
Show answer
The secant slope through $P(a, a^2)$ and $Q(x, x^2)$ is \[ m_{PQ} = \frac{x^2 - a^2}{x - a} = \frac{(x-a)(x+a)}{x-a} = x + a \quad (x \neq a). \] Take the limit as $x \to a$: \[ m = \lim_{x\to a}(x + a) = 2a. \]
Boxed answer: the tangent slope at $x = a$ is $2a$. This is the derivative of $x^2$, found before you have any differentiation rules. (At $a = 1$ it gives $2$, matching Example 1; at $a = 2$ it gives $4$, matching Problem 2.)
Problem 8. For the laser-discharge data in Example 2, the closest secant slopes to $t = 0.04$ are $-74.20$ (from $t = 0.02$) and $-60.75$ (from $t = 0.06$). Explain why averaging these gives a better estimate of the tangent slope than either one alone.
Show answer
A secant from a point to the left tends to be steeper or shallower than the tangent in a consistent direction, and a secant from the right errs in the opposite direction, because the curve bends between the points. The tangent slope sits between the left and right secant slopes.
Averaging the two closest secants cancels much of that opposite-direction error, landing closer to the true tangent slope than either secant by itself: \[ \frac{1}{2}(-74.20 - 60.75) = -67.475. \]
Conclusion: the average of the two closest secant slopes brackets and approximates the tangent slope better than a one-sided estimate. (This is the reasoning behind Stewart 1.4, Example 2. It is the same idea that makes a midpoint or centered estimate more accurate than a one-sided one.)
Problem 9. A pebble falls off a bridge $275$ feet above a river; its height after $t$ seconds is $y = 275 - 16t^2$. Find the average velocity over $[4, 4.1]$, $[4, 4.05]$, $[4, 4.01]$, and estimate the slope of the tangent line to the height graph at $t = 4$.
Show answer
The average velocity over $[4, 4 + h]$ is the secant slope $\dfrac{y(4+h) - y(4)}{h}$. With $y(4) = 275 - 16(16) = 19$:
- $[4, 4.1]$: $y(4.1) = 275 - 16(16.81) = 5.04$, slope $= \dfrac{5.04 - 19}{0.1} = -139.6$.
- $[4, 4.05]$: slope $= -128.8$.
- $[4, 4.01]$: slope $= -128.16$.
The slopes close in on $-128$.
Boxed answer: the tangent slope at $t = 4$ is $-128$ feet per second. (This is Stewart 1.4, Exercise 5. This is the velocity problem in disguise: the tangent slope of the height graph is the instantaneous velocity, here downward at $128$ ft/s.)
Mastery Checklist
Finding a tangent slope is mastered when you can do all of the following without referring to notes:
Mental Model
Think of finding a tangent slope as zooming in until the curve looks straight.
Pick a point on a smooth curve and zoom in on it. The more you magnify, the straighter the curve looks, until the small piece around your point is indistinguishable from a line. That line is the tangent, and its slope is the slope of the curve at the point. The secant-and-limit procedure is the algebraic version of that zoom: a far-away second point gives a rough secant, and sliding it in tightens the secant toward the tangent.
Three things follow from the picture:
- The tangent slope is a destination, not a single measurement. Each secant is one snapshot at a fixed zoom; the tangent slope is what those snapshots converge to as you zoom all the way in. That is why it is defined as a limit.
- A formula gives an exact answer; data gives an estimate. With a formula like $y = x^2$, you can simplify and take the limit exactly. With a table of measurements, you cannot zoom infinitely, so you average the closest secants instead.
- This single limit is the derivative. The slope of the tangent, the instantaneous rate of change, and the derivative $f'(a)$ are three names for the same number. Read as “the steepness of the curve,” it is differential calculus at its source.
Connections
Within Functions and Limits (Chapter 1)
- Evaluating functions and the difference quotient: The secant slope $\dfrac{f(x) - f(a)}{x - a}$ is a difference quotient, built by evaluating $f$ at two inputs. The tangent problem gives that algebraic object its geometric meaning as a slope.
- The velocity problem: The companion problem in the same section of Stewart. Average velocities over shrinking intervals approach an instantaneous velocity by the identical limiting process; the two problems are one idea.
- Limits: The tangent slope is a limit, which is why limits come first. The secant slope is undefined exactly at the point of tangency, so a limit is unavoidable.
Toward Calculus (MATH161)
- The derivative: In Chapter 2 the tangent-slope limit $\lim_{x\to a}\dfrac{f(x)-f(a)}{x-a}$ is named the derivative $f'(a)$. The tangent slope is the derivative before it has a name or any shortcut rules.
- Tangent-line slope and linear approximation: Once you can compute derivatives quickly, the tangent line becomes the best straight-line approximation to a curve near a point, the basis of linearization and of differentials.
- Rates of change everywhere: The tangent slope reads as a rate: current from a charge graph, velocity from a position graph, marginal cost from a cost graph. The derivative is the universal tool for instantaneous rate of change, and the tangent problem is where it first appears.
Audience Notes
For students who find math intimidating: You already know how to find the slope between two points. That is all a secant slope is. The only new step is to compute several secant slopes for points getting closer together and notice what number they approach. That number is the tangent slope. Build the table, watch the column settle, and report the value it settles on.
For career-focused students: Estimating a rate from a table of measurements (the laser-discharge current, the tank’s drain rate) is exactly how instruments and data pipelines compute instantaneous rates from sampled data. The averaging-the-two-closest-secants trick is a simple finite-difference derivative, the same idea used in numerical software to differentiate measured signals.
For gifted and curious students: Problem 7 shows that the tangent slope of $x^2$ at any point $a$ is $2a$, derived purely from the secant limit with no differentiation rules. Try the same for $x^3$ (expand $\dfrac{x^3 - a^3}{x - a} = x^2 + ax + a^2$ and take the limit) and watch the power rule $3a^2$ emerge from the algebra.
For PhD-track students: The “zoom in until it looks straight” picture is local linearity, and a function is differentiable at $a$ exactly when this limit exists, which is strictly stronger than continuity (a corner is continuous but has no single tangent slope, since the left and right secant limits differ). The careful study of when this limit exists is the entry point to differentiability and, later, to differentiable manifolds.
Back to Functions and Limits | Related: The Velocity Problem | Next: Tangent-Line Slope