Parametric Curves: When x and y Both Depend on t

25.1 The Trouble With $y = f(x)$

For twenty-four chapters, almost every curve we studied was the graph of a function: a rule $y = f(x)$ that assigns exactly one height $y$ to each input $x$. That restriction served us well, but it quietly ruled out most of the interesting curves in the plane.

Consider the simplest closed curve of all, a circle. Its equation $x^2 + y^2 = R^2$ fails the vertical-line test: above $x = 0$ sit two points, $(0, R)$ and $(0, -R)$. No single function $f$ can produce both. We are forced to break the circle into an upper half $y = \sqrt{R^2 - x^2}$ and a lower half $y = -\sqrt{R^2 - x^2}$ — an ugly, unnatural surgery on an object that could not be more symmetric.

Now consider a baseball in flight, or a point painted on the rim of a rolling wheel, or the path a robot arm sweeps through a factory. These are not graphs at all; they are trajectories — they double back, loop, and cross themselves. The question "what is $y$ when $x = 3$?" may have zero answers, or two, or infinitely many. The graph paradigm has run out of road.

The fix is to stop thinking of a curve as a graph and start thinking of it as a path traced by a moving point. Introduce a third variable $t$ — most often time — and let both coordinates depend on it:

$$x = x(t), \qquad y = y(t).$$

As $t$ sweeps through an interval, the point $\big(x(t), y(t)\big)$ moves through the plane, and the trail it leaves is the curve. These are parametric equations, and $t$ is the parameter.

The Key Insight. A parametric curve is a path, not a graph. Freeing both $x$ and $y$ to depend on a common parameter lets us describe any motion in the plane — loops, crossings, and all — and recovers the function viewpoint as the special case $x = t$.

This chapter, the opener of Part V, develops the full calculus of such paths: how to sketch them, how to recover a Cartesian equation when one exists, how to find slopes and concavity with the chain rule, how to measure their length and the area of the surfaces they sweep, and how the most famous curve in the history of calculus — the cycloid — falls out of these ideas. Along the way we treat projectile motion, the parametric curve every physics student meets first.

Two pointers before we begin. Chapter 26 treats polar coordinates, which you will see is a special, highly structured kind of parametrization. Chapter 28 rewrites everything here in the language of vector-valued functions, $\mathbf{r}(t) = \langle x(t), y(t)\rangle$, and pushes it into three dimensions; the velocity, speed, and curvature you may have heard of live there. This chapter stays in the plane and keeps the coordinates separate, which is the gentlest way in.

25.2 Parametric Equations and Orientation

A parametric curve is given by two functions and an interval:

$$x = x(t), \qquad y = y(t), \qquad t \in [a, b].$$

As $t$ increases from $a$ to $b$, the point moves along the curve in a definite direction. That direction is the orientation of the curve, and it is genuinely new information — the same set of points can be traced clockwise or counterclockwise, once or many times. A graph $y = f(x)$ has no orientation; a path always does.

A first catalogue of curves

Here are the parametrizations we will return to throughout Part V. Read each one as a recipe for a moving point.

Circle of radius $R$ centered at the origin: $$x = R\cos t, \qquad y = R\sin t, \qquad t \in [0, 2\pi].$$ At $t = 0$ the point sits at $(R, 0)$; as $t$ grows it sweeps counterclockwise. Replace $t$ by $-t$ and it goes clockwise instead — same circle, opposite orientation.

Ellipse with semi-axes $a$ and $b$: $$x = a\cos t, \qquad y = b\sin t.$$ The same angular parameter, scaled differently in each direction.

Line through $(x_0, y_0)$ in the direction $(\Delta x, \Delta y)$: $$x = x_0 + t\,\Delta x, \qquad y = y_0 + t\,\Delta y.$$ This is the parametric form of a line, and it generalizes verbatim to three dimensions in Chapter 28 — one of the first payoffs of the path viewpoint.

Cycloid, the path of a point on the rim of a wheel of radius $r$ rolling along the ground: $$x = r(t - \sin t), \qquad y = r(1 - \cos t).$$ We will earn this formula and mine it for surprises in §25.7.

Astroid, a four-cusped star: $$x = a\cos^3 t, \qquad y = a\sin^3 t.$$

Geometric Intuition. Think of $t$ as the reading on a clock and $\big(x(t), y(t)\big)$ as the position of a firefly. The parametric equations tell you where the firefly is at every instant. The curve is the glowing trail left behind in a long-exposure photograph — but the photograph loses the timing and the direction. Parametric form keeps both. That extra information (when the firefly is where, and which way it is heading) is exactly what graphs throw away and trajectories cannot afford to lose.

Check Your Understanding. Describe the curve traced by $x = \cos(2t)$, $y = \sin(2t)$ for $t \in [0, 2\pi]$, and compare it to $x = \cos t$, $y = \sin t$ on the same interval.

Answer

Both trace the unit circle, since $x^2 + y^2 = 1$ in each case. But $x = \cos(2t)$, $y = \sin(2t)$ runs twice as fast: as $t$ goes from $0$ to $2\pi$, the angle $2t$ goes from $0$ to $4\pi$, so the point circles the origin twice. Same set of points, different parametrization — a first hint that a curve has many parametrizations but one shape.

Eliminating the parameter

Sometimes you can erase $t$ algebraically and recover a single equation in $x$ and $y$. This is called eliminating the parameter, and it connects the path viewpoint back to the Cartesian one.

Circle. From $x = R\cos t$ and $y = R\sin t$, the Pythagorean identity does the work: $$x^2 + y^2 = R^2\cos^2 t + R^2\sin^2 t = R^2(\cos^2 t + \sin^2 t) = R^2.$$ The parameter vanishes and the familiar Cartesian equation appears.

Line. Take $x = 1 + 2t$, $y = 3 - 5t$. Solve the first equation for the parameter: $t = (x - 1)/2$. Substitute into the second: $$y = 3 - 5\cdot\frac{x - 1}{2} \quad\Longrightarrow\quad 5x + 2y = 11.$$

Parabola. Take $x = t^2$, $y = t$. Then $t = y$, so $x = y^2$ — a parabola opening to the right. Notice this is not a function $y = f(x)$: each positive $x$ has two values of $y$. The parametric form sidestepped that difficulty effortlessly.

Common Pitfall. Eliminating the parameter can quietly change the curve by forgetting the range of $t$. The equations $x = \cos t$, $y = \cos t$ give $y = x$ — but only the segment $-1 \le x \le 1$, because $\cos t$ never leaves $[-1, 1]$. Writing "$y = x$" as the answer, with no restriction, claims the whole infinite line. Always carry the implied domain: after eliminating $t$, ask which part of the Cartesian curve the parameter actually reaches, and in what direction.

For curves that loop or cross themselves — the cycloid, a figure-eight, a knot — no clean Cartesian equation exists at all. That is not a defect; it is the whole point. Parametric form remains perfectly valid where the graph viewpoint simply cannot follow.

25.3 The Slope of a Parametric Curve

A moving point still traces a curve with tangent lines, and we still want the slope $dy/dx$. The obstacle is that we have no formula for $y$ in terms of $x$ — only both in terms of $t$. The chain rule bridges the gap.

Intuitive level

Over a tiny interval of time $dt$, the point moves $dx = \dot x\,dt$ horizontally and $dy = \dot y\,dt$ vertically. (We use the Newton dot, $\dot x = dx/dt$, for derivatives with respect to the parameter — a notation that signals "rate with respect to $t$.") The slope of the path is rise over run:

$$\text{slope} = \frac{dy}{dx} = \frac{\dot y\,dt}{\dot x\,dt} = \frac{\dot y}{\dot x}.$$

The $dt$'s cancel, and we are left with a beautifully simple rule.

Computational level

Slope of a parametric curve. If $x = x(t)$ and $y = y(t)$ are differentiable and $\dot x = dx/dt \neq 0$, then $$\frac{dy}{dx} = \frac{dy/dt}{dx/dt} = \frac{\dot y}{\dot x}.$$

This is nothing but the chain rule, $\dfrac{dy}{dt} = \dfrac{dy}{dx}\cdot\dfrac{dx}{dt}$, solved for $dy/dx$. The requirement $\dot x \neq 0$ is the condition that the point is actually moving horizontally so that "slope as a function of $x$" makes sense.

Worked Example 25.3.1 (a self-crossing curve). Let $x = t^2$, $y = t^3 - t$. Find the slope at $t = 1$.

Differentiate each coordinate with respect to $t$: $$\dot x = 2t, \qquad \dot y = 3t^2 - 1, \qquad \frac{dy}{dx} = \frac{3t^2 - 1}{2t}.$$ At $t = 1$ the point is $(x, y) = (1, 0)$ and the slope is $\dfrac{3 - 1}{2} = 1$. The tangent line is $y = x - 1$.

This curve, incidentally, crosses itself: $y = 0$ when $t^3 = t$, i.e. at $t = -1, 0, 1$, and $t = \pm 1$ both give $x = 1$. So the point $(1, 0)$ is visited twice, with two different slopes ($+1$ at $t=1$, $-1$ at $t=-1$). A single $(x, y)$ location, two tangent lines — something a graph could never do, and a vivid reminder of why we keep the parameter.

Horizontal and vertical tangents

Reading the slope formula carefully tells us where the tangent line is flat or upright:

• Horizontal tangent where $\dot y = 0$ and $\dot x \neq 0$ (numerator zero: no vertical motion).
• Vertical tangent where $\dot x = 0$ and $\dot y \neq 0$ (denominator zero: no horizontal motion).
• When both $\dot x = 0$ and $\dot y = 0$, the point momentarily stops, and the curve may have a cusp — a sharp corner. We meet this at the top arch of the cycloid in §25.7.

Worked Example 25.3.2 (tangents on the ellipse). For $x = 3\cos t$, $y = 2\sin t$, find all horizontal and vertical tangents.

Here $\dot x = -3\sin t$ and $\dot y = 2\cos t$. Horizontal tangents need $\dot y = 0$, i.e. $\cos t = 0$, at $t = \pi/2$ and $t = 3\pi/2$ — the points $(0, 2)$ and $(0, -2)$, the top and bottom of the ellipse. Vertical tangents need $\dot x = 0$, i.e. $\sin t = 0$, at $t = 0$ and $t = \pi$ — the points $(3, 0)$ and $(-3, 0)$, the left and right ends. Exactly the four extreme points you would mark by eye, now located by calculus.

The second derivative

Concavity asks for $d^2y/dx^2$, and here a tempting mistake lurks.

Common Pitfall. The second derivative is not $\ddot y / \ddot x$. You cannot just differentiate $x$ and $y$ twice and divide. The correct rule treats $dy/dx$ as a new function of $t$ and runs the slope formula again: $$\frac{d^2y}{dx^2} = \frac{d}{dx}\!\left(\frac{dy}{dx}\right) = \frac{\dfrac{d}{dt}\!\left(\dfrac{dy}{dx}\right)}{\dfrac{dx}{dt}}.$$ The denominator is $\dot x$ — not $\ddot x$. Writing $\ddot y/\ddot x$ is the single most common error in this chapter, and it gives the wrong answer except by accident.

The logic is exactly the slope rule applied to the quantity $g(t) = dy/dx$ in place of $y$: to differentiate anything with respect to $x$, differentiate it with respect to $t$ and divide by $\dot x$.

Worked Example 25.3.3 (concavity). Continue with $x = t^2$, $y = t^3 - t$, where we found $dy/dx = (3t^2 - 1)/(2t)$. Simplify first: $\dfrac{dy}{dx} = \dfrac{3t}{2} - \dfrac{1}{2t}$. Differentiate with respect to $t$: $$\frac{d}{dt}\!\left(\frac{dy}{dx}\right) = \frac{3}{2} + \frac{1}{2t^2}.$$ Divide by $\dot x = 2t$: $$\frac{d^2y}{dx^2} = \frac{\tfrac{3}{2} + \tfrac{1}{2t^2}}{2t} = \frac{3t^2 + 1}{4t^3}.$$ At $t = 1$ this is $4/4 = 1 > 0$, so the curve is concave up there. For $t > 0$ the second derivative is positive (concave up); for $t < 0$ it is negative (concave down) — consistent with the curve's leftward loop bending the other way.

Check Your Understanding. For the circle $x = \cos t$, $y = \sin t$, find $dy/dx$ at $t = \pi/4$ and confirm it is the slope you expect.

Answer

$\dot x = -\sin t$, $\dot y = \cos t$, so $dy/dx = \cos t / (-\sin t) = -\cot t$. At $t = \pi/4$ this is $-\cot(\pi/4) = -1$. The point is $(\tfrac{\sqrt2}{2}, \tfrac{\sqrt2}{2})$, on the line $y = x$ through the origin; the tangent to a circle is perpendicular to that radius, hence slope $-1$. Calculus and geometry agree.

25.4 Arc Length of a Parametric Curve

How long is a parametric curve? We measure it the way we measure every length in calculus: chop the path into tiny pieces, approximate each piece by a straight segment, add them up, and pass to the limit.

Geometry first

Over a small time step $dt$, the point moves $dx = \dot x\,dt$ across and $dy = \dot y\,dt$ up. By the Pythagorean theorem, the little segment of path has length

$$ds = \sqrt{(dx)^2 + (dy)^2} = \sqrt{\dot x^2 + \dot y^2}\;dt.$$

The quantity $ds$ is the arc length element. Summing (integrating) it from $t = a$ to $t = b$ gives the total length.

Arc length of a parametric curve. If $x(t)$ and $y(t)$ have continuous derivatives and the curve is traced once on $[a, b]$, its arc length is $$L = \int_a^b \sqrt{\left(\frac{dx}{dt}\right)^2 + \left(\frac{dy}{dt}\right)^2}\;dt = \int_a^b \sqrt{\dot x^2 + \dot y^2}\;dt.$$

This single formula subsumes the Cartesian arc-length integral from Chapter 18. If the curve is a graph $y = f(x)$, parametrize it trivially by $x = t$, $y = f(t)$; then $\dot x = 1$, $\dot y = f'(t)$, and

$$L = \int_a^b \sqrt{1 + f'(t)^2}\;dt,$$

exactly the formula of §18 (Cartesian arc length). The parametric version is the general one; the Cartesian formula is the special case $x = t$. This is the first of several places in Part V where a familiar integral turns out to be a corner of a larger parametric picture.

Warning. The formula assumes the curve is traced exactly once on $[a, b]$. If the parametrization doubles back or repeats — like $x = \cos(2t)$, $y = \sin(2t)$ on $[0, 2\pi]$, which circles twice — the integral measures the total distance the point travels, not the geometric length of the curve. For the doubly-traced unit circle that integral gives $4\pi$, twice the circumference. Length is a property of the path-with-its-timing; pick a parameter interval that covers the shape once.

Worked Example 25.4.1 (a circle, as a sanity check). For $x = R\cos t$, $y = R\sin t$ on $[0, 2\pi]$: $$\dot x^2 + \dot y^2 = R^2\sin^2 t + R^2\cos^2 t = R^2, \qquad L = \int_0^{2\pi} R\;dt = 2\pi R.$$ The circumference $2\pi R$, recovered as it must be. When a new formula reproduces a known one, trust grows.

Worked Example 25.4.2 (an ellipse, and a famous obstruction). For $x = a\cos t$, $y = b\sin t$ with $a \neq b$: $$L = \int_0^{2\pi} \sqrt{a^2\sin^2 t + b^2\cos^2 t}\;dt.$$ This integral has no elementary antiderivative — it is an elliptic integral, and it is precisely the integral that gave that whole family of functions its name. The perimeter of an ellipse, despite the ellipse being one of the simplest curves imaginable, cannot be written with the functions of elementary calculus. We meet the same wall we met with $\int e^{-x^2}\,dx$ in Chapter 14: the antiderivative provably escapes elementary form (Liouville). We evaluate it numerically instead, which the Python at the end of this chapter does in two lines.

Historical Note. The intractable ellipse perimeter haunted mathematicians for centuries. Ramanujan, around 1914, gave the staggeringly accurate approximation $L \approx \pi\big[3(a+b) - \sqrt{(3a+b)(a+3b)}\big]$, correct to many digits for nearly-circular ellipses. That a problem so easy to state should resist closed-form solution — and yield instead to a genius's approximation — is a recurring lesson of calculus: stating a problem and solving it are different difficulties.

25.5 Surface Area of a Surface of Revolution

Spin a parametric curve about an axis and it sweeps out a surface. The area of that surface follows from the arc-length element by the same logic Chapter 18 used for Cartesian curves: a thin band of the surface is (approximately) the lateral surface of a cone frustum, with slant width $ds$ and circumference $2\pi(\text{radius})$.

Surface area of revolution (parametric). Revolving the curve $x = x(t)$, $y = y(t)$, $t \in [a, b]$, about the $x$-axis generates a surface of area $$S = \int_a^b 2\pi\, y\,\sqrt{\dot x^2 + \dot y^2}\;dt,$$ where $y \ge 0$ is the distance from the axis. Revolving about the $y$-axis replaces the radius $y$ by $x$: $$S = \int_a^b 2\pi\, x\,\sqrt{\dot x^2 + \dot y^2}\;dt.$$

The factor $2\pi y$ is the circumference of the circle each point traces; $\sqrt{\dot x^2 + \dot y^2}\,dt = ds$ is the width of the band. Their product, summed, is the area.

Worked Example 25.5.1 (the sphere). Revolve the upper semicircle $x = R\cos t$, $y = R\sin t$, $t \in [0, \pi]$, about the $x$-axis. We have $\sqrt{\dot x^2 + \dot y^2} = R$ and radius $y = R\sin t$, so $$S = \int_0^{\pi} 2\pi (R\sin t)(R)\;dt = 2\pi R^2 \int_0^\pi \sin t\;dt = 2\pi R^2\big[-\cos t\big]_0^\pi = 2\pi R^2(1 + 1) = 4\pi R^2.$$ The surface area of a sphere, $4\pi R^2$ — derived, not memorized. Archimedes proved this by hand and was so proud of it he asked for a sphere-in-a-cylinder to be carved on his tombstone. We got it in three lines because we stand on FTC (Chapter 14).

Real-World Application — Manufacturing and CAD. Surfaces of revolution are everywhere a part is turned on a lathe: bottles, lampshades, rocket nozzles, lenses, wheel rims. An engineer specifies the profile curve parametrically and the geometry kernel of the CAD system integrates the formula above to report surface area — needed for estimating paint, coating mass, heat transfer, or the cost of plating. The same integral that gives Archimedes' sphere prices the chrome on a hubcap.

25.6 Projectile Motion: A Parametric Curve You Already Know

Newton's analysis of a thrown object is the purest example of a parametric curve, because here the parameter $t$ genuinely is time, and the two coordinate functions encode two independent physical facts.

A projectile launched from the origin with speed $v_0$ at angle $\theta$ above the horizontal, under constant gravitational acceleration $g$ and ignoring air resistance, obeys:

$$x(t) = (v_0\cos\theta)\,t, \qquad y(t) = (v_0\sin\theta)\,t - \tfrac{1}{2}g t^2.$$

The horizontal equation says the across-motion is uniform — no horizontal force, so constant velocity $v_0\cos\theta$. The vertical equation says the up-motion is uniformly decelerated by gravity. These are two independent stories, married by the shared clock $t$. That is the power of the parametric viewpoint: it keeps the horizontal and vertical motions cleanly separate while describing a single curving path. This is the mathematics-of-change theme in its first physics costume — each coordinate is governed by its own rate law.

Range and optimal angle

The projectile lands when $y(t) = 0$ again: $$(v_0\sin\theta)\,t - \tfrac{1}{2}g t^2 = t\left(v_0\sin\theta - \tfrac12 g t\right) = 0 \;\Longrightarrow\; t = 0 \text{ or } t = \frac{2v_0\sin\theta}{g}.$$ The launch is $t = 0$; the landing is $t = 2v_0\sin\theta/g$. The horizontal distance covered — the range — is

$$R = x\!\left(\frac{2v_0\sin\theta}{g}\right) = (v_0\cos\theta)\cdot\frac{2v_0\sin\theta}{g} = \frac{v_0^2\,(2\sin\theta\cos\theta)}{g} = \frac{v_0^2\sin(2\theta)}{g},$$

using the identity $2\sin\theta\cos\theta = \sin(2\theta)$. Now optimize over the launch angle: $R$ is largest when $\sin(2\theta) = 1$, i.e. $2\theta = 90^\circ$, so

$$\boxed{\theta = 45^\circ}, \qquad R_{\max} = \frac{v_0^2}{g}.$$

The optimal launch angle for maximum range, in a vacuum, is exactly $45^\circ$. This is a genuine optimization result (Chapter 10) emerging from a parametric model — a small triumph that every shot-putter, golfer, and artillery officer half-knows by instinct.

Eliminating the parameter: the path is a parabola

Solve the horizontal equation for time, $t = x/(v_0\cos\theta)$, and substitute into the vertical equation:

$$y = (v_0\sin\theta)\cdot\frac{x}{v_0\cos\theta} - \frac{g}{2}\left(\frac{x}{v_0\cos\theta}\right)^2 = x\tan\theta - \frac{g}{2v_0^2\cos^2\theta}\,x^2.$$

This is $y = (\text{const})\,x - (\text{const})\,x^2$ — a parabola opening downward. Newton's celebrated conclusion: a projectile under uniform gravity follows a parabolic arc. Eliminating $t$ converted the trajectory in time into the shape in space, and the shape is a conic section (Chapter 27).

Real-World Application — Ballistics, sports, and game engines. This model underlies artillery firing tables, the flight of a baseball or a basketball's free throw, the optimal takeoff angle in the long jump (closer to $20^\circ$ in practice, because a jumper cannot generate the same speed at $45^\circ$), and the physics engine of essentially every video game with a thrown grenade or a leaping character. Real ballistics adds air resistance — a drag force proportional to speed or its square — which makes the path no longer exactly parabolic and pushes the optimal angle below $45^\circ$. But the parametric skeleton, two coupled rate laws sharing a clock, never changes.

Check Your Understanding. A ball is thrown at $v_0 = 20$ m/s at $\theta = 30^\circ$, with $g = 9.8\ \text{m/s}^2$. Find its range.

Answer

$R = \dfrac{v_0^2\sin(2\theta)}{g} = \dfrac{400\cdot\sin 60^\circ}{9.8} = \dfrac{400\cdot 0.866}{9.8} \approx 35.3$ m. Notice $30^\circ$ and $60^\circ$ give the same range, since $\sin(2\theta) = \sin(180^\circ - 2\theta)$ — a complementary pair, both falling short of the $45^\circ$ maximum of $400/9.8 \approx 40.8$ m.

25.7 The Cycloid: The Curve That Made Calculus Famous

No curve has a richer history than the cycloid, the path traced by a point fixed on the rim of a wheel as the wheel rolls without slipping along a straight line. Galileo named it and tried to find its area by weighing paper cutouts. Pascal, Descartes, Roberval, Huygens, and both Bernoullis fought over it. It earned the nickname "the Helen of geometers" — beautiful, and the cause of endless quarrels.

Deriving the parametrization

Roll a wheel of radius $r$ to the right, and let $t$ be the angle (in radians) through which it has turned. Rolling without slipping means the center has moved forward by the arc length unrolled, $rt$. The tracked point started at the bottom; after turning by $t$ it sits at angle $t$ around the rim relative to the center. Adding the center's position to the point's offset from the center gives

$$x = r(t - \sin t), \qquad y = r(1 - \cos t).$$

At $t = 0$ the point is on the ground ($y = 0$); at $t = \pi$ it is at the top of the wheel ($y = 2r$); at $t = 2\pi$ it touches the ground again, completing one arch. Then the pattern repeats.

Arc length of one arch — a beautiful number

Compute the derivatives and the arc-length element: $$\dot x = r(1 - \cos t), \qquad \dot y = r\sin t.$$ $$\dot x^2 + \dot y^2 = r^2(1 - \cos t)^2 + r^2\sin^2 t = r^2\big[1 - 2\cos t + \cos^2 t + \sin^2 t\big] = r^2(2 - 2\cos t).$$ Now apply the half-angle identity $1 - \cos t = 2\sin^2(t/2)$, so $2 - 2\cos t = 4\sin^2(t/2)$: $$\sqrt{\dot x^2 + \dot y^2} = \sqrt{4r^2\sin^2(t/2)} = 2r\,\big|\sin(t/2)\big| = 2r\sin(t/2) \quad\text{for } t \in [0, 2\pi],$$ since $\sin(t/2) \ge 0$ on that interval. Integrate over one arch: $$L = \int_0^{2\pi} 2r\sin(t/2)\;dt = 2r\Big[-2\cos(t/2)\Big]_0^{2\pi} = -4r\big(\cos\pi - \cos 0\big) = -4r(-1 - 1) = 8r.$$

One arch of a cycloid has length exactly $8r$ — eight times the radius of the wheel, a clean integer multiple with no $\pi$ in sight. For all the transcendence lurking in the circle ($2\pi r$), the rolling-point path delivers a perfectly rational answer. Christopher Wren proved this in 1658, one of the first arc-length computations in history, before the calculus that makes it routine even existed.

Area under one arch

The area under one arch (using the parametric area integral $A = \int y\,dx = \int y\,\dot x\,dt$): $$A = \int_0^{2\pi} r(1 - \cos t)\cdot r(1 - \cos t)\;dt = r^2\int_0^{2\pi}(1 - \cos t)^2\,dt.$$ Expand $(1-\cos t)^2 = 1 - 2\cos t + \cos^2 t$ and integrate over $[0, 2\pi]$: the $1$ contributes $2\pi$, the $-2\cos t$ contributes $0$, and $\cos^2 t$ averages to $\tfrac12$ and contributes $\pi$. So $$A = r^2(2\pi + 0 + \pi) = 3\pi r^2.$$ The area under one arch is three times the area of the rolling wheel ($\pi r^2$). This is the result Galileo guessed (correctly) from his paper-weighing and that Roberval first proved.

The brachistochrone and the tautochrone

The cycloid's fame rests on two astonishing physical properties, both upside-down (arch hanging below, like a valley).

The brachistochrone problem, posed by Johann Bernoulli in 1696 as a public challenge: of all curves joining a high point to a lower point not directly beneath it, down which does a frictionless bead slide in the least time? The intuitive guess — a straight line, the shortest path — is wrong. The answer is an inverted cycloid. A steeper initial drop builds speed early and wins the race despite the longer path. Newton, the story goes, received the challenge one evening, solved it before breakfast, and submitted anonymously; Bernoulli recognized the author "as one knows the lion by his claw."

The tautochrone (or isochrone) property, proved by Huygens in 1659: a bead released from rest anywhere on an inverted cycloid reaches the bottom in the same amount of time, regardless of its starting height. A bead from high up travels farther but moves faster, and the two effects cancel exactly. Huygens tried to exploit this to build a pendulum clock immune to the amplitude errors of an ordinary circular pendulum — the cycloid as the engineer's dream of a perfect timekeeper.

Geometric Intuition — why "least time" beats "least distance." Picture the bead choosing its path. On a straight ramp it accelerates gently the whole way. On the cycloid it plunges near the start, banking speed when it is cheap to gain, then spends that speed cruising across the gentler lower portion. Time is distance divided by speed; the cycloid sacrifices a little extra distance to buy a lot of early speed, and the trade pays. This is a calculus-of-variations result — optimization not over a number but over an entire curve — and it opened a whole branch of mathematics. We can only point at it here; the full machinery lies beyond a first course.

Math Major Sidebar — calculus of variations. The brachistochrone asks you to minimize the functional $T[y] = \int \sqrt{\dfrac{1 + y'^2}{2g\,y}}\;dx$ over all curves $y(x)$ joining the two endpoints — a minimization not over points in $\mathbb{R}^n$ but over an infinite-dimensional space of functions. The tool is the Euler–Lagrange equation, which converts "the functional is stationary" into a differential equation whose solution is the cycloid. Johann Bernoulli's own 1696 method was slicker: he reinterpreted the bead as a ray of light obeying Snell's law through a medium of continuously varying density, turning a mechanics problem into optics. The calculus of variations he and his brother Jakob built from this single problem now underlies Lagrangian mechanics, optimal control, general relativity's geodesics, and the training objectives of machine learning. One rolling wheel; an entire field.

25.8 A Gallery of Classical Curves

The cycloid has a large and beautiful family. Each is a parametric curve, each arises from rolling circles, and each rewards the calculus of this chapter.

Astroid ($x = a\cos^3 t$, $y = a\sin^3 t$). A four-cusped star, traced by a point on a small circle of radius $a/4$ rolling inside a fixed circle of radius $a$ — a hypocycloid. Eliminating the parameter gives the elegant Cartesian form $x^{2/3} + y^{2/3} = a^{2/3}$. Its arc length is $6a$ and the area it encloses is $\tfrac{3}{8}\pi a^2$, both worth computing as exercises. (Note the four cusps at $t = 0, \tfrac\pi2, \pi, \tfrac{3\pi}{2}$, where $\dot x = \dot y = 0$ — the sharp-point condition from §25.3.)

Cardioid ($x = a(2\cos t - \cos 2t)$, $y = a(2\sin t - \sin 2t)$). A heart shape, traced by a point on a circle rolling outside an equal fixed circle — an epicycloid with ratio $1\!:\!1$. We will meet it again in polar coordinates in Chapter 26, where its equation $r = a(1 + \cos\theta)$ is far simpler.

Trochoid ($x = rt - d\sin t$, $y = r - d\cos t$). The path of a point at distance $d$ from the center of a rolling wheel of radius $r$. When $d = r$ the point is on the rim and we recover the cycloid; when $d < r$ (a point inside the wheel) the curve is smooth and wavy (a curtate trochoid, like a reflector on a bicycle wheel); when $d > r$ (a point on an extended flange, as on a train wheel) the curve develops backward-looping kinks (a prolate trochoid). Train wheels really do momentarily move backward relative to the ground at the bottom of each loop — a fact this parametrization makes precise.

Lissajous figures ($x = \sin(at)$, $y = \sin(bt + \phi)$). When $a$ and $b$ are different integers, these trace intricate woven patterns. They were the glowing shapes on every analog oscilloscope screen, used to compare two signal frequencies: a stable closed figure means the frequencies are in the integer ratio $a\!:\!b$.

Real-World Application — gears, cams, and clockwork. Cycloidal and epicycloidal curves are not museum pieces. The teeth of many gears are cut to a cycloidal or involute profile precisely because that shape makes contacting teeth roll rather than scrape, minimizing wear. Cam profiles in engines — the lobes that open and close valves — are designed as parametric curves so that a follower rises and falls on a prescribed schedule. The mathematics Huygens developed chasing a perfect clock is the mathematics a mechanical engineer still uses to cut a gear.

25.9 Computation: Plotting and Measuring Parametric Curves

Python turns these abstract paths into pictures and turns the intractable integrals into numbers. The three-tier pattern — pose the curve, reason by hand, then verify by machine — is our standard from the continuity guide.

# Plot one arch of a cycloid and confirm its arc length is exactly 8r.
import numpy as np
import matplotlib.pyplot as plt
from scipy.integrate import quad

r = 1.0
t = np.linspace(0.0, 2 * np.pi, 400)
x = r * (t - np.sin(t))           # cycloid x(t)
y = r * (1 - np.cos(t))           # cycloid y(t)

plt.figure(figsize=(7, 3))
plt.plot(x, y, lw=2)
plt.title("One arch of a cycloid  (r = 1)")
plt.axis("equal"); plt.grid(True)
plt.savefig("cycloid.png", dpi=120)   # Figure 25.1

# Arc length L = ∫₀^{2π} sqrt(x'^2 + y'^2) dt, with x' = r(1-cos t), y' = r sin t
def speed(s: float) -> float:
    return np.sqrt((r * (1 - np.cos(s)))**2 + (r * np.sin(s))**2)

L, _ = quad(speed, 0.0, 2 * np.pi)
print(f"numerical arc length = {L:.6f}")   # numerical arc length = 8.000000
print(f"hand result 8r       = {8 * r:.6f}")  # hand result 8r       = 8.000000

Figure 25.1 shows the characteristic arch — rising to height $2r$, falling back to the ground, with a sharp cusp where it touches down. The numerical integral confirms the hand result $L = 8r$ to six digits: calculus by hand and scipy.integrate.quad shaking hands.

The same quad handles the ellipse perimeter that has no closed form:

# Perimeter of the ellipse x = a cos t, y = b sin t — the elliptic integral.
a, b = 3.0, 2.0
peri, _ = quad(lambda s: np.sqrt((a*np.sin(s))**2 + (b*np.cos(s))**2), 0, 2*np.pi)
ram = np.pi * (3*(a+b) - np.sqrt((3*a + b)*(a + 3*b)))   # Ramanujan's approximation
print(f"numerical perimeter   = {peri:.6f}")   # numerical perimeter   = 15.865439
print(f"Ramanujan approx      = {ram:.6f}")    # Ramanujan approx      = 15.865439

Computational Note. scipy.integrate.quad does not need an antiderivative — it samples the integrand adaptively and converges to the true value regardless of whether a closed form exists. This is the practical answer to §25.4's elliptic-integral obstruction: hand computation builds the understanding (we know exactly why the perimeter resists a formula), but machine computation delivers the number to whatever precision an engineer needs. Ramanujan's approximation matching quad to six digits is a small thrill — a 1914 hand formula keeping pace with a modern numerical library.

25.10 Three Ways to Describe a Curve

Step back and survey the landscape. A plane curve can be specified three ways, and choosing well is half the battle.

The same circle is $y = \pm\sqrt{R^2 - x^2}$ (explicit, awkward), or $x^2 + y^2 = R^2$ (implicit, clean), or $x = R\cos t$, $y = R\sin t$ (parametric, oriented and ready for calculus of motion). Each form has a calculus: explicit curves use $f'(x)$ (Chapter 7), implicit curves use implicit differentiation (Chapter 8), and parametric curves use the chain-rule slope $\dot y/\dot x$ of this chapter. They are three dialects describing the same geometry, and a fluent mathematician translates freely among them. The geometry-and-algebra-are-inseparable theme of this book is nowhere more visible: every one of these curves is both a picture and a triple of formulas.

Add to Your Modeling Portfolio. Add a parametric trajectory to your model — a path traced in time, with its slope, length, and orientation. Biology: model a foraging animal's path or a cell's chemotaxis trajectory $\big(x(t), y(t)\big)$; compute the total distance traveled via the arc-length integral and compare it to the straight-line displacement (a measure of "path tortuosity"). Economics: trace a parametric phase path of two interacting quantities (e.g. price and inventory, or two competing firms' outputs) as time advances, and identify where the path has horizontal or vertical tangents — the turning points of each variable. Physics: model your projectile (or orbit, or oscillation) parametrically; find the range, the optimal launch angle by the $45^\circ$ argument, and the arc length of the flight path. Data Science: plot the parametric trajectory of gradient descent — the sequence of $(\theta_1(t), \theta_2(t))$ parameter values as training proceeds — over a contour plot of the loss surface; the arc length measures how far the optimizer wandered. (This anchors the gradient-descent example we will complete in Chapter 30.)

25.11 Looking Ahead

You can now treat any plane curve as a path: sketch it, orient it, eliminate the parameter when one exists, find its slope and concavity with the chain rule, locate its horizontal, vertical, and cusp tangents, and measure both its length and the area of the surface it sweeps. Projectile motion and the cycloid showed these tools at work on the two curves that, more than any others, taught the world what calculus was for.

Three roads lead onward. Chapter 26 specializes to polar coordinates, where the parameter is an angle and the parametrization $x = r(\theta)\cos\theta$, $y = r(\theta)\sin\theta$ is so structured that it deserves its own calculus of area and length. Chapter 27 studies the conic sections — the ellipse, parabola, and hyperbola we keep meeting — in full, including their parametric forms ($x = a\cosh t$, $y = b\sinh t$ for the hyperbola, using the identity $\cosh^2 t - \sinh^2 t = 1$). And Chapter 28 lifts everything here into vector-valued functions $\mathbf{r}(t)$, where velocity, speed, acceleration, and curvature live, and where the plane curve becomes a space curve in three dimensions. The path viewpoint you adopted in this chapter is the one that will carry you, unchanged in spirit, all the way through multivariable calculus.

Reflection

For most of this book a curve meant a graph — a static rule pairing each $x$ with a height. This chapter quietly overturned that. A curve is a journey: a point moving through the plane, leaving a trail. The change of viewpoint is small to state and enormous in consequence. It is what let Newton describe a planet's orbit, what lets a font designer shape a letter, what lets an animator move a character, and what let a rolling wheel — the cycloid — become the curve of fastest descent and the curve of equal time, two facts so surprising they launched the calculus of variations. The mathematics of change began as the study of how one quantity varies with another. Here it grows up into the study of motion itself. Turn the page, and we will let the parameter be an angle.

Continue to: Exercises · Quiz · Case Study 1 · Case Study 2 · Key Takeaways · Further Reading