The Physical Pendulum

In 1673 Christiaan Huygens faced a problem that sounds trivial and is not. His pendulum clocks — the best timekeepers on Earth — did not swing an idealized point mass on a massless string. They swung a brass rod carrying a heavy bob, with every gram of metal at a different distance from the pivot. So what, exactly, is "the length" of a real pendulum? Where along the rod does the textbook formula $T = 2\pi\sqrt{\ell/g}$ apply? Huygens answered it completely, for any rigid body whatsoever, more than a decade before Newton published the laws of motion. The answer is the physical pendulum: any rigid body swinging under gravity about a fixed horizontal axis. Two numbers replace the string length — the moment of inertia $I$ about the pivot, and the distance $d$ from pivot to center of mass — and they combine into one effective parameter, the equivalent length $\ell_{\mathrm{eq}} = I/(md)$ . Every rigid object that swings is secretly a simple pendulum of that length: a door on its hinges, a swinging leg, a cricket bat, a church bell, a gymnast on a high bar. Mastering this topic buys you three things. A worked template for turning rotational dynamics ( $I\ddot\theta = \tau$ ) into simple harmonic motion. A beautiful hidden symmetry — pivot and center of oscillation are interchangeable — that became the most precise way to measure $g$ for a century. And the answer to a question you have been living your whole life: why your natural walking pace is set by physics, not by choice.

Key takeaways

A physical pendulum is any rigid body free to swing about a horizontal pivot axis that misses its center of mass; gravity supplies the restoring torque $-mgd\sin\theta$ .
Small oscillations have period $T = 2\pi\sqrt{I/(mgd)}$ , where $I$ is the moment of inertia about the pivot and $d$ is the pivot-to-center-of-mass distance.
Every rigid body swings exactly like a simple pendulum of the equivalent length $\ell_{\mathrm{eq}} = I/(md) = k^2/d + d$ , where $k$ is the radius of gyration about the center of mass.
Pivot and center of oscillation (the point a distance $\ell_{\mathrm{eq}}$ below the pivot) are interchangeable: swing the body from either and the period is identical, because $d\,d' = k^2$ is symmetric.
The period is shortest when $d = k$ , giving $T_{\min} = 2\pi\sqrt{2k/g}$ ; it diverges both as $d \to 0$ (no torque) and $d \to \infty$ .
Finite amplitude stretches the period by the same factor as the simple pendulum, $T \approx T_0(1 + \theta_0^2/16)$ — the correction knows nothing about the body's shape.
Key idea: treat the whole rigid body as rotating about the pivot — the period $T = 2\pi\sqrt{I/(mgd)}$ uses the moment of inertia about the pivot (not the center of mass), and the body is equivalent to a simple pendulum of length $\ell_{\mathrm{eq}} = I/(md)$ .

The core insight

Treat the whole rigid body as rotating about the pivot: the period $T = 2\pi\sqrt{I/(mgd)}$ uses the moment of inertia about the pivot (not about the center of mass) and the pivot-to-cm distance $d$ . The body is equivalent to a simple pendulum of length $\ell_{\mathrm{eq}} = I/(md)$ .

Where this lives today

The physical pendulum's most consequential modern job is inside you. A human leg swinging from the hip is, to a good first approximation, a uniform rod pivoted at one end: for $L \approx 0.9\ \mathrm{m}$ , $T = 2\pi\sqrt{2L/3g} \approx 1.55\ \mathrm{s}$ , so a free swing-through takes half a period, about $0.78\ \mathrm{s}$ — roughly 77 steps per minute, right at the relaxed human cadence. Walking at that rate lets gravity swing the leg for free; walking much faster means paying muscular torque to beat the pendulum clock. This single estimate anchors gait analysis, explains why tall people stroll with slower cadence, and drives the design of passive-dynamic walking robots and energy-efficient humanoid gaits, where 2026-era bipeds recover a large fraction of their locomotion energy by letting the legs run at their natural pendulum frequency. Prosthetic legs are tuned the same way — mass added or removed to move $I/(md)$ until the artificial limb's swing time matches the intact one. The second modern home is inertia metrology. Invert equation (3) as $I = mgd\,T^2/4\pi^2$ and a stopwatch becomes a moment-of-inertia instrument: hang the object from a knife edge, time small swings, done. Sports engineering measures bat and racket "swing weight" this way; automotive and aerospace labs measure connecting rods, crankshafts, and whole spacecraft on pendulum and trifilar rigs before flight, because a wrong $I$ means a wrong attitude-control model. And the conjugate-point physics survives in sport as the "sweet spot": the center of percussion of a bat — the impact point that sends no jolt to the hands — coincides with the center of oscillation conjugate to the grip.

Figure

Kater's reversible pendulum — reciprocity turned into a measuring instrument

The physical picture

Grab a meter stick at one end and let it swing. Gravity pulls on every centimeter of wood, but for torque purposes the entire weight might as well act at one point — the center of mass, a distance $d$ below the pivot. That concentrated pull, times its lever arm, is the restoring torque. Fighting it is the stick's reluctance to angular acceleration: the moment of inertia $I$ about the pivot. The period is set by this tug-of-war, exactly as a block's period is set by spring stiffness against mass — here gravity-through-geometry plays the spring, and rotational inertia plays the block.

Figure

A rigid body on a pivot and the simple pendulum it secretly is

Why does a swinging stick have a smaller time period than a simple pendulum of its own length? Because its mass is spread out: much of it rides high, near the pivot, contributing lever arm to the restoring torque budget ( $mgd$ with $d = L/2$ ) while costing relatively little inertia. The balance point of the two effects is the equivalent length $\ell_{\mathrm{eq}} = I/(md)$ — for an end-pivoted uniform stick, two-thirds of its length, so a meter stick swings like a 67 cm simple pendulum and runs ahead of a 100 cm one. The deeper structure appears when you ask how the period changes as you move the pivot. Drill pivot holes at different distances $d$ from the center of mass of the same body and the period traces a valley: infinite at $d = 0$ (pivot the body at its center of mass and gravity has no lever arm — it just sits), falling to a minimum at $d = k$ , the radius of gyration, then rising again like $\sqrt{d}$ as the body turns into an effective point mass on a long arm. And the valley hides a symmetry: for every pivot there is a conjugate point on the other side of the center of mass, at $d' = k^2/d$ , that gives exactly the same period. That conjugate point is the center of oscillation — hang the body from it, and the old pivot becomes the new center of oscillation.

Figure

The period valley — one body, every possible pivot, and the conjugate-point symmetry

Finally, shrink the body to a point. Concentrate all the mass at the center of mass, so $I \to md^2$ , and the machinery collapses: $\ell_{\mathrm{eq}} \to d$ and $T \to 2\pi\sqrt{d/g}$ . The simple pendulum was never a separate topic — it is the physical pendulum with the geometry switched off.

From torque to period

Everything starts from the rotational form of Newton's second law taken about the fixed pivot — the one axis in the problem whose reaction forces have no lever arm and therefore no say. Gravity, acting effectively at the center of mass a distance $d$ from the pivot, supplies a torque with moment arm $d\sin\theta$ , and it always pulls the body back toward hanging straight down:

I\,\ddot\theta = -mgd\sin\theta \qquad (1)

where: $I$ — moment of inertia about the pivot axis; $\theta$ — angular displacement from the downward vertical; $m$ — total mass; $g$ — gravitational acceleration; $d$ — pivot-to-center-of-mass distance. This is exact at any angle. The minus sign is the physics: displace the body either way and the torque opposes you, which is what makes hanging-straight-down a stable equilibrium and oscillation possible at all.

Derivation APeriod of a physical pendulum from torque about the pivot

We are deriving the small-oscillation angular frequency and period of an arbitrary rigid body pivoted a distance $d$ from its center of mass, starting from the exact equation of motion (1). The strategy is the universal small-oscillations routine: establish the torque about the pivot (where constraint forces are invisible), linearize $\sin\theta \approx \theta$ about the stable equilibrium, and match the result to the standard SHM template $\ddot\theta + \omega^2\theta = 0$ . Two things deserve watching. The moment of inertia must be taken about the pivot, not the center of mass — the body genuinely rotates about the pivot, and using $I_{\mathrm{cm}}$ silently changes the physical problem. And the weight may be treated as acting wholly at the center of mass only because gravity is uniform — each mass element feels $g$ in the same direction, so the net gravitational torque is exactly $\mathbf{r}_{\mathrm{cm}}\times m\mathbf{g}$ .

Step 1. Compute the gravitational torque about the pivot. Summing $\mathbf{r}_i \times m_i\mathbf{g}$ over all mass elements factors into $\left(\sum_i m_i\mathbf{r}_i\right)\times\mathbf{g} = m\,\mathbf{r}_{\mathrm{cm}}\times\mathbf{g}$ , whose component along the pivot axis is

\tau = -mgd\sin\theta

Why the concentration at the cm is legitimate: uniform gravity is the special case — for any position-dependent force this shortcut would fail. The lever arm of the weight is the horizontal offset of the center of mass, $d\sin\theta$ , and the torque opposes the displacement.

Step 2. Apply the rotational second law about the fixed pivot, $I\ddot\theta = \tau$ :

I\,\ddot\theta = -mgd\sin\theta

Why the pivot is the right axis: the unknown reaction force at the hinge acts at zero distance from this axis and drops out of the torque budget entirely — choosing any other axis would drag it back in.

Step 3. Linearize about the equilibrium. For $|\theta| \ll 1$ , $\sin\theta \approx \theta$ (the error is $\theta^3/6$ , about 0.5% at $10°$ ):

\ddot\theta + \frac{mgd}{I}\,\theta = 0

Why this is the whole game: the equation is now the SHM template with the role of "stiffness over mass" played by $mgd/I$ — gravitational torque strength over rotational inertia.

Step 4. Read off frequency and period by matching to $\ddot\theta + \omega^2\theta = 0$ :

\omega = \sqrt{\frac{mgd}{I}}\,, \qquad T = \frac{2\pi}{\omega} = 2\pi\sqrt{\frac{I}{mgd}} \qquad (2,\ 3)

Why amplitude is absent: linear restoring force means amplitude-independent period — isochrony — which holds only as far as the linearization does; equation (9) below prices the correction. *Quick checks: for a point mass, $I = md^2$ gives $T = 2\pi\sqrt{d/g}$ , recovering the simple pendulum; dimensions confirm $\sqrt{\mathrm{kg\,m^2}/(\mathrm{kg}\cdot\mathrm{m\,s^{-2}}\cdot\mathrm{m})} = \mathrm{s}$ .*

The result deserves a moment of respect before we compress it further:

T = 2\pi\sqrt{\frac{I}{mgd}} \qquad (3)

where: $T$ — the small-oscillation period. Read it structurally: inertia upstairs, torque strength downstairs. Anything that redistributes mass away from the pivot at fixed $m$ and $d$ raises $I$ and slows the swing; anything that shifts the center of mass away from the pivot at fixed $I$ strengthens the torque and speeds it up. Note that $I \propto m$ , so the mass itself cancels — a heavy door and a light one of the same shape swing at the same rate, the rotational echo of Galileo's freely falling masses.

One length to rule the body

A rigid body has infinitely many shape details; the period, remarkably, cares about exactly one combination of them. Demand that a simple pendulum of length $\ell_{\mathrm{eq}}$ match the physical pendulum's period, $2\pi\sqrt{\ell_{\mathrm{eq}}/g} = 2\pi\sqrt{I/(mgd)}$ , and the length that does it is

\ell_{\mathrm{eq}} = \frac{I}{md} \qquad (4)

where: $\ell_{\mathrm{eq}}$ — the equivalent (simple-pendulum) length. The point a distance $\ell_{\mathrm{eq}}$ below the pivot, along the pivot-cm line, is called the center of oscillation: hang a point mass on a massless string of that length and it will swing stride for stride with the body. For real bodies $\ell_{\mathrm{eq}} > d$ always — the equivalent bob hangs below the center of mass. To see what $\ell_{\mathrm{eq}}$ is made of, split the moment of inertia into shape and placement with the parallel-axis theorem:

I = I_{\mathrm{cm}} + md^2 \qquad (5)

where: $I_{\mathrm{cm}}$ — the moment of inertia about a parallel axis through the center of mass. The theorem says moving the rotation axis away from the cm always adds inertia, in the amount a point mass would contribute. Encode the shape in a single length, the radius of gyration $k$ defined by $I_{\mathrm{cm}} = mk^2$ (the distance at which a point mass would mimic the body's cm-inertia), and the period becomes a function of two lengths only:

T = 2\pi\sqrt{\frac{k^2 + d^2}{gd}}\,, \qquad k^2 = \frac{I_{\mathrm{cm}}}{m} \qquad (6)

where: $k$ — radius of gyration about the cm ( $k = L/\sqrt{12} \approx 0.289L$ for a uniform rod, $R/\sqrt{2}$ for a uniform disk). Mass has cancelled outright; every physical pendulum is characterized by its $k$ and its $d$ . This is the equation graphed in the period-valley figure above — steep divergence at small $d$ , slow $\sqrt{d}$ growth at large $d$ , minimum in between.

Go deeper · Derivation B · 2 minEquivalent length and the interchangeability of pivot and center of oscillation

Derivation BEquivalent length and the interchangeability of pivot and center of oscillation

We are deriving the anatomy of the equivalent length, $\ell_{\mathrm{eq}} = k^2/d + d$ , and from it the physical pendulum's hidden symmetry: the conjugate-point relation $d\,d' = k^2$ , which says the pivot and the center of oscillation can swap roles without changing the period. The strategy is pure algebra on equation (4) — substitute the parallel-axis decomposition, define the conjugate distance $d'$ , and check that the map $d \mapsto d'$ leaves $\ell_{\mathrm{eq}}$ invariant. The payoff is far from academic: this reciprocity is the operating principle of Kater's reversible pendulum, which measured $g$ to parts per million without ever knowing $I$ , $m$ , or even where the center of mass sat. Watch how the symmetry emerges: $d\,d' = k^2$ treats the two distances identically, so of course swapping them changes nothing — the trick is noticing it.

Step 1. Substitute $I = m(k^2 + d^2)$ into the equivalent length:

\ell_{\mathrm{eq}} = \frac{I}{md} = \frac{m(k^2 + d^2)}{md} = \frac{k^2}{d} + d \qquad (7)

Why this split is illuminating: the $d$ term is the point-mass part (the cm swinging on its arm), and $k^2/d$ is the shape surcharge — the extra length the spread-out mass adds. As $k \to 0$ the body collapses to a point and $\ell_{\mathrm{eq}} \to d$ .

Step 2. Locate the center of oscillation relative to the center of mass. It sits a distance $\ell_{\mathrm{eq}}$ from the pivot, on the far side of the cm, so its distance from the cm is

d' = \ell_{\mathrm{eq}} - d = \frac{k^2}{d} \quad\Longrightarrow\quad d\,d' = k^2

Why this relation is the heart of the matter: it is symmetric under $d \leftrightarrow d'$ . The geometry has no memory of which point was the pivot and which the center of oscillation.

Step 3. Re-pivot the body at the center of oscillation and recompute the equivalent length from the same formula (7), now with pivot distance $d'$ :

\ell'_{\mathrm{eq}} = \frac{k^2}{d'} + d' = d + d' = \ell_{\mathrm{eq}}

Why the middle equality holds: $k^2/d' = d$ follows directly from $dd' = k^2$ . The new equivalent length is the same sum of the same two distances — hence the same period, from either pivot.

Step 4. State the operational consequence. If two pivot points on opposite sides of the cm give equal periods, the distance between them is the equivalent length, and

g = \frac{4\pi^2\,\ell_{\mathrm{eq}}}{T^2}

Why this is a metrologist's dream: the two hard-to-measure quantities ( $I$ and the cm position) have vanished; only a length between two knife edges and a period remain, and both can be measured superbly well. *Cross-check: for the end-pivoted rod, $d = L/2$ and $k^2 = L^2/12$ give $d' = L/6$ and $\ell_{\mathrm{eq}} = 2L/3$ , consistent with the direct calculation $I/(md) = (mL^2/3)/(mL/2)$ .*

Go deeper · Derivation C · 1 minThe pivot that swings fastest

Derivation CThe pivot that swings fastest

We are deriving where to pivot a given body to minimize its period, and what that minimum is. The strategy is single-variable calculus on the period formula (6): at fixed shape ( $k$ constant), the period depends on the pivot placement only through $f(d) = k^2/d + d$ , so minimizing $T$ means minimizing $f$ . The interesting physics is in the shape of the answer: the two terms of $f$ pull in opposite directions — moving the pivot outward strengthens the torque lever arm but also inflates the point-mass inertia — and the optimum is exactly where the two effects trade at par. Note also what happens at the extremes, because both limits destroy the oscillation in different ways.

Step 1. Isolate what varies. From (6), $T(d) = 2\pi\sqrt{f(d)/g}$ with $f(d) = k^2/d + d$ ; $T$ is monotonic in $f$ , so extremizing $f$ suffices.

Step 2. Differentiate and set to zero:

\frac{\mathrm{d}f}{\mathrm{d}d} = -\frac{k^2}{d^2} + 1 = 0 \quad\Longrightarrow\quad d = k

Why the balance point sits at the radius of gyration: for $d < k$ the shape term $k^2/d$ dominates and lengthening the arm pays off; for $d > k$ the point-mass term $d$ dominates and it costs. At $d = k$ the marginal gain and loss cancel.

Step 3. Confirm a minimum and evaluate it. The second derivative $2k^2/d^3 > 0$ , and at $d = k$ the equivalent length is $\ell_{\mathrm{eq}} = k + k = 2k$ :

T_{\min} = 2\pi\sqrt{\frac{2k}{g}}

Why the extremes fail: as $d \to 0$ the restoring torque $mgd$ dies while $I \to mk^2$ stays finite — period diverges (pivot at the cm and the body is indifferent, neutral equilibrium); as $d \to \infty$ the body degenerates into a point mass on an ever-longer arm and $T \sim 2\pi\sqrt{d/g}$ grows without bound. *For example, a meter stick has $k = 0.289\ \mathrm{m}$ , so the fastest pivot sits 28.9 cm from the center and gives $T_{\min} = 2\pi\sqrt{2(0.2887)/9.81} = 1.52\ \mathrm{s}$ — quicker than the end pivot's 1.64 s, as the valley plot shows.*

Which formula answers which question

you want	use	notes
period of any pivoted rigid body	$T = 2\pi\sqrt{I/(mgd)}$ , eq. (3)	$I$ about the pivot
the "length" of the body as a pendulum	$\ell_{\mathrm{eq}} = I/(md) = k^2/d + d$ , eqs. (4), (7)	locates the center of oscillation
convert cm-inertia to pivot-inertia	$I = I_{\mathrm{cm}} + md^2$ , eq. (5)	always adds, never subtracts
period from shape + placement	$T = 2\pi\sqrt{(k^2+d^2)/(gd)}$ , eq. (6)	mass never enters
partner pivot with equal period	$d\,d' = k^2$	Kater's reversibility
fastest possible swing	$d = k$ , $T_{\min} = 2\pi\sqrt{2k/g}$	valley bottom of $T(d)$
measure $g$	$g = 4\pi^2\ell_{\mathrm{eq}}/T^2$	knife-edge distance + stopwatch

Energy and amplitude

The same physics can be told in energy language, which is what survives when the amplitude grows. Take the pivot as the axis and the lowest point of the cm as the reference height; rotating by $\theta$ lifts the center of mass by $d(1 - \cos\theta)$ :

E = \tfrac{1}{2}\,I\dot\theta^2 + mgd\,(1 - \cos\theta) \qquad (8)

where: $E$ — the conserved mechanical energy; the first term is rotational kinetic energy about the pivot, the second is gravitational potential energy of the cm. For small angles $1 - \cos\theta \approx \theta^2/2$ and (8) becomes the standard SHM energy $\tfrac12 I\dot\theta^2 + \tfrac12 (mgd)\theta^2$ — a "rotational spring" of stiffness $mgd$ . At larger angles the potential is softer than parabolic, so the body lingers near the turning points and the period stretches:

T(\theta_0) \approx T_0\left(1 + \frac{\theta_0^2}{16} + \cdots\right) \qquad (9)

where: $\theta_0$ — angular amplitude; $T_0$ — the small-amplitude period of eq. (3). The correction is identical to the simple pendulum's, and for a good reason: in the variable $\theta$ the two systems obey literally the same differential equation, $\ddot\theta = -\omega^2\sin\theta$ , with different constants inside $\omega$ . Shape sets the clock rate; amplitude physics is universal. At $\theta_0 = 20°$ the stretch is $\theta_0^2/16 = 0.76\%$ — the gateway to the anharmonic pendulum article downstream.

Limiting cases

Case	Result	Meaning
Point mass at distance $d$	$I = md^2 \Rightarrow T = 2\pi\sqrt{d/g}$	all mass at the bob recovers the simple pendulum; $\ell_{\mathrm{eq}} = d$
Uniform rod, pivot at end	$I = \tfrac13 mL^2,\ d = \tfrac{L}{2} \Rightarrow T = 2\pi\sqrt{2L/3g}$	swings like a simple pendulum of length $2L/3$ , faster than length $L$
Fastest pivot	$d = k \Rightarrow T_{\min} = 2\pi\sqrt{2k/g}$	valley bottom of $T(d)$ ; marginal torque gain = marginal inertia cost
Pivot at the center of mass	$d \to 0 \Rightarrow T \to \infty$	zero lever arm, zero restoring torque — neutral equilibrium, no oscillation
Pivot very far away	$d \gg k \Rightarrow T \approx 2\pi\sqrt{d/g}$	shape becomes irrelevant; the body is a point mass on a long arm
Finite amplitude $\theta_0$	$T \approx T_0(1 + \theta_0^2/16)$	same universal anharmonic stretch as the simple pendulum

Common trap“The I in the period formula is the moment of inertia about the center of mass”

It is not — and this is the single most common physical-pendulum error. The body rotates about the pivot, so $T = 2\pi\sqrt{I/(mgd)}$ needs the moment of inertia about the pivot axis. Given the tabulated $I_{\mathrm{cm}}$ , you must first apply the parallel-axis theorem, $I = I_{\mathrm{cm}} + md^2$ . Skipping it underestimates the period, and by a lot when $d \gtrsim k$ : for an end-pivoted rod the mistake predicts $T$ too small by a factor $\sqrt{I_{\mathrm{cm}}/I} = 1/2$ . Two adjacent traps: the distance $d$ runs from pivot to center of mass (not to the end of the body, not to the center of oscillation); and the equivalent length $\ell_{\mathrm{eq}} = I/(md)$ is generally larger than $d$ — if your computed "pendulum length" comes out shorter than the cm distance, a term went missing.

Worked examples

1 · The swinging meter stick (confidence builder)

A uniform meter stick ( $L = 1.000\ \mathrm{m}$ ) is pivoted at one end and released with small amplitude. Find its period, and compare with a simple pendulum of the same length.

Step 1. Inertia about the pivot. From the table value $I_{\mathrm{cm}} = \tfrac{1}{12}mL^2$ and the parallel-axis theorem with $d = L/2$ : $I = \tfrac{1}{12}mL^2 + m(L/2)^2 = \tfrac13 mL^2$ .

Step 2. Apply the period formula (3): $T = 2\pi\sqrt{\dfrac{\tfrac13 mL^2}{mg(L/2)}} = 2\pi\sqrt{\dfrac{2L}{3g}} = 2\pi\sqrt{\dfrac{2(1.000)}{3(9.81)}} = \boxed{1.64\ \mathrm{s}}$ .

Step 3. Read it as an equivalent length: $\ell_{\mathrm{eq}} = \tfrac{2}{3}L = 66.7\ \mathrm{cm}$ . A simple pendulum of the stick's full length would take $T = 2\pi\sqrt{1.000/9.81} = 2.01\ \mathrm{s}$ — the stick beats it because its mass rides high, keeping the inertia cheap relative to the torque.

Step 4. Sanity-check the center of oscillation: it sits $66.7$ cm from the pivot, i.e. $d' = 16.7\ \mathrm{cm}$ beyond the cm, and indeed $d\,d' = (0.500)(0.167) = 0.0833\ \mathrm{m^2} = L^2/12 = k^2$ . ✓

2 · A disk on a nail, its twin pivot, and its fastest pivot (exam level)

A uniform disk of radius $R = 0.250\ \mathrm{m}$ hangs on a nail through a point on its rim. Find (a) its period, (b) the second pivot point with the same period, (c) the pivot placement that minimizes the period, and the minimum period.

Step 1. (a) Inertia about the rim: $I = I_{\mathrm{cm}} + mR^2 = \tfrac12 mR^2 + mR^2 = \tfrac32 mR^2$ , with $d = R$ . Then $T = 2\pi\sqrt{\dfrac{\tfrac32 mR^2}{mgR}} = 2\pi\sqrt{\dfrac{3R}{2g}} = 2\pi\sqrt{\dfrac{3(0.250)}{2(9.81)}} = \boxed{1.23\ \mathrm{s}}$ .

Step 2. (b) The conjugate pivot. The radius of gyration: $k^2 = I_{\mathrm{cm}}/m = R^2/2 = 0.03125\ \mathrm{m^2}$ . The twin sits at $d' = k^2/d = R/2 = 12.5\ \mathrm{cm}$ from the center, on the opposite side of the cm along the same line. Check: $\ell_{\mathrm{eq}} = k^2/d + d = \tfrac32 R = 37.5\ \mathrm{cm}$ , and the two pivots are exactly $\ell_{\mathrm{eq}}$ apart: $25.0 + 12.5 = 37.5\ \mathrm{cm}$ . ✓ Drill a hole halfway out from the center and the disk swings with the same 1.23 s period as from the rim.

Step 3. (c) Fastest pivot at $d = k = R/\sqrt2 = 17.7\ \mathrm{cm}$ from the center:

T_{\min} = 2\pi\sqrt{\frac{2k}{g}} = 2\pi\sqrt{\frac{2(0.177)}{9.81}} = \boxed{1.19\ \mathrm{s}}

Step 4. Assemble the valley: $T(\text{rim}) = 1.23\ \mathrm{s}$ , $T(\text{half-radius}) = 1.23\ \mathrm{s}$ , $T(0.177\ \mathrm{m}) = 1.19\ \mathrm{s}$ , and $T \to \infty$ at the exact center. The two equal periods straddle the minimum, as the conjugate-point symmetry demands.

3 · Measuring g the nineteenth-century way (applied)

A Kater reversible pendulum has knife edges $\ell = 0.9940\ \mathrm{m}$ apart. With the trim mass adjusted until the periods about both edges agree, the common period is $T = 2.0000\ \mathrm{s}$ . Find $g$ .

Step 1. Equal periods about two pivots on opposite sides of the cm mean the knife edges are conjugate points, so the distance between them is the equivalent length: $\ell_{\mathrm{eq}} = 0.9940\ \mathrm{m}$ — no moment of inertia, mass, or cm position needed.

Step 2. Invert the simple-pendulum relation: $g = \dfrac{4\pi^2\,\ell_{\mathrm{eq}}}{T^2} = \dfrac{4\pi^2 (0.9940)}{(2.0000)^2} = \boxed{9.810\ \mathrm{m/s^2}}$ .

Step 3. Appreciate the error budget. A length known to $0.1\ \mathrm{mm}$ contributes $10^{-4}$ relative error; timing 1000 swings pushes the period error to $\sim 10^{-6}$ . This is how $g$ was mapped worldwide to part-per-million precision a century before free-fall laser gravimeters — every term that is hard to know about a rigid body cancelled out of the physics.

Origins · 1 minChristiaan Huygens and how this was discovered

Christiaan Huygens

1629–1695 · The Hague and Paris — Horologium Oscillatorium

Huygens is the physicist who made time itself a precision instrument. In 1656 he attached a pendulum to clockwork and improved timekeeping accuracy roughly a hundredfold overnight; navigation, astronomy, and physics all changed downstream. But real clock pendulums were rods and bobs, not points on strings, and finding the "length" of such a compound pendulum was a famous open problem — Mersenne had circulated it decades earlier. In Horologium Oscillatorium (1673), a book many rank second only to Newton's Principia in the century, Huygens solved it: he introduced what we now call the center of oscillation and computed it by a bold new principle — that the center of gravity of the swinging system rises exactly as high as it fell, an early conservation-of-energy argument, stated twelve years before Newton's laws were available to derive it from torque. In the same book he unveiled the cycloidal pendulum (exactly isochronous at any amplitude), the theory of evolutes, and the formula for centripetal force. Modern mechanics texts derive his center of oscillation in three lines from $I\ddot\theta = \tau$ ; Huygens got there first with none of that machinery.

History

1656

— Huygens builds the first pendulum clock, turning the pendulum from Galileo's curiosity into the world's timekeeping standard.

1673

— Horologium Oscillatorium solves the compound pendulum: center of oscillation, equivalent length, and the cycloid's perfect isochrony.

1687

— Newton's Principia grounds Huygens' results in the laws of motion; the physical pendulum becomes a standard application of torque about a fixed axis.

1817

— Henry Kater turns pivot–center-of-oscillation reciprocity into the reversible pendulum, measuring g in London to five significant figures.

1826

— Bessel's error analysis (knife edges, air buoyancy, amplitude corrections) makes pendulum gravimetry a mature part-per-million science that maps g worldwide for the next century.

Going further

References. Morin, Introduction to Classical Mechanics, Ch. 9 (Angular Momentum / Rigid Bodies) · Goldstein, Classical Mechanics, 3rd ed., §1.6 and oscillation chapters · Landau & Lifshitz, Mechanics (Vol. 1), §21 (Small Oscillations) · Taylor, Classical Mechanics, Ch. 5 (Oscillations) · Huygens, Horologium Oscillatorium (1673), Part IV, for the original center-of-oscillation theory. The key results in this article are: (1) the period $T = 2\pi\sqrt{I/mgd}$ , obtained by linearizing $I\ddot\theta = -mgd\sin\theta$ ; (2) the equivalent length $\ell_{\mathrm{eq}} = I/(md) = k^2/d + d$ and the conjugate-point relation $dd' = k^2$ ; (3) the uniform-rod result $T = 2\pi\sqrt{2L/3g}$ from $I = mL^2/3$ , $d = L/2$ ; (4) the disk-on-rim period $2\pi\sqrt{3R/2g}$ from $I = \tfrac{3}{2}mR^2$ ; (5) the minimum-period condition $d = k$ giving $T_{\min} = 2\pi\sqrt{2k/g}$ ; (6) the finite-amplitude correction $T \approx T_0(1 + \theta_0^2/16)$ , identical to the simple pendulum series.

Builds on: The Simple Pendulum · Small Oscillations About Equilibrium · Simple Harmonic Motion. Leads to: Anharmonic Pendulum and Amplitude Dependence · Damped Harmonic Oscillator · Energy in Simple Harmonic Motion.

Why can I pretend all of gravity acts at the center of mass when computing the torque?

Because gravity here is uniform — every mass element feels the same $\mathbf{g}$ . The total gravitational torque about the pivot is $\sum_i \mathbf{r}_i \times m_i\mathbf{g}$ , and since $\mathbf{g}$ factors out of the sum, this collapses to $(\sum_i m_i\mathbf{r}_i)\times\mathbf{g} = m\,\mathbf{r}_{\mathrm{cm}}\times\mathbf{g}$ — exactly the torque of the full weight applied at the center of mass. This is a special privilege of uniform fields: in a tidal (non-uniform) field the shortcut fails and the effective point of action (the "center of gravity") can differ from the center of mass. For laboratory pendulums the distinction is utterly negligible; for a satellite-sized body in orbit it is the origin of gravity-gradient torques.

If the parallel-axis theorem already adds the md² term, why isn't the equivalent length just d?

Because the two ingredients of the period do different jobs and only one of them grows with the shape. The torque strength is $mgd$ — it cares only about where the cm hangs. The inertia is $I = m(k^2 + d^2)$ — it carries the extra $mk^2$ that the spread-out mass costs. The ratio that sets the period is $\ell_{\mathrm{eq}} = I/(md) = k^2/d + d$ , which exceeds $d$ by precisely the shape surcharge $k^2/d$ . Only when $k \to 0$ (all mass shrunk to a point at the cm) does $\ell_{\mathrm{eq}} \to d$ . A useful check on any homework answer: if your computed equivalent length comes out shorter than the cm distance, you have dropped the $I_{\mathrm{cm}}$ term.

Does a heavier body swing slower? The mass sits right there in T = 2π√(I/mgd)

It cancels. The moment of inertia is proportional to the mass ( $I = m(k^2 + d^2)$ ), so $T = 2\pi\sqrt{(k^2+d^2)/(gd)}$ contains no $m$ at all — a solid oak door and a hollow foam replica of identical shape swing at exactly the same rate. This is the rotational version of Galileo's result that free-fall acceleration is mass-independent, and it fails for the same reasons: once air drag or pivot friction enter (forces not proportional to mass), the heavier body holds its amplitude longer. What mass distribution does change is $k$ — move mass toward the pivot line and the body swings faster even at fixed total mass and fixed cm.

What does the infinite period at d = 0 actually look like physically?

Pivot a body exactly through its center of mass and gravity produces zero torque at every angle — the weight acts at the axis itself, with no lever arm. There is no restoring tendency and no oscillation to have a period: displace the body and it simply stays where you put it (neutral equilibrium), turning forever if you give it a spin. The divergence $T \to \infty$ as $d \to 0$ is the formula's way of announcing that limit: as the pivot creeps toward the cm, the restoring "stiffness" $mgd$ dies while the inertia survives at $mk^2$ , so each swing takes longer and longer. A pivot slightly off the cm makes a superb slow oscillator — this is exactly how balance-beam and torsion instruments get their enormous sensitivity.

How can the pivot and center of oscillation be interchangeable when the body is not symmetric about them?

The symmetry lives in the formula, not in the body's shape. The period depends on the pivot's location through one combination only: $\ell_{\mathrm{eq}} = k^2/d + d$ . The two points sit at cm-distances $d$ and $d' = k^2/d$ , and the map $d \leftrightarrow d'$ leaves $k^2/d + d$ invariant — swap them and the sum reproduces itself ( $k^2/d' = d$ , $d' = k^2/d$ ). No geometric mirror symmetry is required; an arbitrarily lopsided body obeys it, because all shape information enters through the single number $k$ . That is also why the theorem is experimentally so robust: Kater's method needs only equal periods about two pivots on opposite sides of the cm, never any statement about the body's form.

Is the sweet spot of a bat really the center of oscillation?

Essentially yes. The center of percussion relative to a grip point is the impact point for which the blow produces pure rotation about the grip — zero net reaction force ("sting") at the hands — and a short impulse-momentum calculation puts it at distance $I/(md)$ from the grip: exactly the center of oscillation conjugate to it. Hit there and the translational kick of the impulse and the rotational recoil cancel at your hands; hit closer or farther and the handle jerks forward or backward. Real bats complicate the story (vibrational nodes of the bat's bending modes also matter, and they sit near but not exactly at the percussion point), but the rigid-body sweet spot is genuinely conjugate-point physics — the same $dd' = k^2$ that measures $g$ .

My lab pendulum's period comes out a bit long compared with T = 2π√(I/mgd). What are the usual suspects?

Three, in the order worth checking. Amplitude: equation (9) says a $10°$ swing already runs $(\theta_0^2/16) \approx 0.19\%$ slow — time at the smallest amplitude you can read reliably, or extrapolate $T(\theta_0^2)$ to zero. Added inertia you forgot: the clamp, glue, or the knife-edge assembly itself contributes to $I$ but maybe not to your calculated $I_{\mathrm{cm}} + md^2$ ; air also loads the body slightly (buoyancy reduces the effective restoring torque). Pivot imperfection: a fat or dull knife edge means the contact line wanders, effectively shifting $d$ , and friction there both damps the swing and (weakly) biases the period. Kater-style measurements beat all three by design — which is why they held the precision crown for a century.

Huygens solved this in 1673, but Newton's laws arrived in 1687. What did Huygens actually use?

An energy principle, stated with complete confidence before anyone could derive it: the center of gravity of a swinging system rises exactly as high as it fell, and no arrangement of internal constraints can lift it higher — Huygens' hypothesis in Horologium Oscillatorium. From it he computed the center of oscillation of compound pendulums, effectively performing conservation-of-energy calculations twelve years before the Principia and more than a century before energy became an organizing concept of mechanics. Torricelli's principle (a system's center of gravity cannot rise on its own) was the seed; Huygens sharpened it into a quantitative tool. Today we derive his results in three lines from $I\ddot\theta = -mgd\sin\theta$ — but historically, energy reasoning got there first, which is a good thing to remember when choosing your own line of attack on a mechanics problem.

The Physical Pendulum

Where this lives today

The physical picture

From torque to period

One length to rule the body

Which formula answers which question

Energy and amplitude

Limiting cases

Worked examples

1 · The swinging meter stick (confidence builder)

2 · A disk on a nail, its twin pivot, and its fastest pivot (exam level)

3 · Measuring g the nineteenth-century way (applied)

Going further

Before this

Where this leads