Capacitance

Slide your finger across a phone screen and nothing mechanical moves — no button, no spring, no contact. What the glass actually detects is that your fingertip, a bag of conductive salt water, has crept close to a grid of tiny electrodes and changed a number by a few tenths of a picofarad. That number is capacitance: the measure of how much charge a pair of conductors swallows for every volt you place between them. It is the first genuinely new quantity of circuit theory, and it is hiding in plain sight in everything you have already done — every conductor pair you solved in electrostatics had a capacitance; we just never gave the ratio a name. You are unusually well armed for this article. You know that a conductor in equilibrium is an equipotential, that fields near charged surfaces follow $\sigma/\varepsilon_0$ , and that boundary-value problems have unique answers. Capacitance is where that machinery starts paying rent: the uniqueness theorem is why charge and voltage are strictly proportional, and Gauss's law is how you compute the constant for real geometries in three lines. Mastering this page buys you the entire capacitance cluster — series and parallel networks, stored energy, dielectrics, RC timing — and a working vocabulary for lab electronics, where "how many picofarads?" is asked about cables, sensors, and qubits alike. Two conductors, one number. Let's define it, understand why it exists, and compute it for the three geometries every physicist carries in their head.

Key takeaways

Capacitance is defined as $C = Q/V$ : the charge on the positive conductor per volt of potential difference between the pair.
$C$ is a pure geometry number (times $\varepsilon_0$ , or $\kappa\varepsilon_0$ with a dielectric): it does not depend on $Q$ or $V$ individually — that is a theorem, not an assumption.
The three canonical results: parallel plate $C=\varepsilon_0 A/d$ , concentric spheres $C=4\pi\varepsilon_0\,ab/(b-a)$ , coaxial cylinders $C/L=2\pi\varepsilon_0/\ln(b/a)$ .
An isolated sphere has $C=4\pi\varepsilon_0 R$ — the "second conductor" is at infinity. Even the whole Earth is only about $710\ \mu\mathrm{F}$ .
One farad is enormous: a vacuum parallel-plate capacitor with a 1 mm gap needs plates about $10.6\ \mathrm{km}$ on a side. Real electronics lives in pF–µF.
KEY IDEA: between any two conductors, charge transferred and voltage are strictly proportional; the ratio $C=Q/V$ is set by shape and separation alone, so you compute it once from electrostatics and use it forever as a circuit element.

The core insight

Capacitance $C=Q/V$ is the geometry-fixed proportionality between the charge $\pm Q$ held on a pair of conductors and the potential difference $V$ between them — the first circuit element, computed entirely from electrostatics.

Where this lives today

Capacitance in 2026 is less a component than a sensing modality. Every phone screen, trackpad, and car dashboard is a matrix of drive and sense electrodes resolving femtofarad-scale disturbances at kilohertz rates; MEMS accelerometers read out proof-mass motion as $\Delta C$ between interdigitated fingers; and each of the several billion cells in a DRAM chip stores its bit as roughly $25\ \mathrm{fF}$ of charge that must be refreshed dozens of times per second because it leaks — your laptop's memory is a field of tiny capacitors racing their own discharge. At the exotic end, the transmon qubit that anchors superconducting quantum computing is, electrically, a capacitor of order $70\ \mathrm{fF}$ shunting a Josephson junction: its transition frequency is set by the same $C$ you learned to compute in this article, and qubit designers still begin with electrostatic boundary-value solves of exactly the kind this chapter's prerequisites taught. From the Leyden jar to the quantum processor, the object never changed — two conductors and the space between.

The physical picture

Take any two isolated conductors — two plates, two nested shells, a wire and the chassis of your laptop — and move charge $Q$ from one to the other, leaving them at $+Q$ and $-Q$ . A field appears in the space between; a potential difference $V$ appears across that field. Now double the transferred charge. By superposition, the field everywhere doubles, so the line integral that gives $V$ doubles too. The ratio never moves. That ratio is the capacitance, and the picture to hold is a container: $V$ is how hard you have pushed, $Q$ is how much has gone in, and $C$ is the width of the vessel. A wide vessel (big plates, small gap) takes a lot of charge with little rise in "level"; a narrow one (small, well-separated conductors) fills to high voltage on a whisper of charge.

Figure

The capacitor as a charge container: same voltage, different geometry, different charge

Why should the ratio be so rigidly fixed? Because of theorems you already own. Put charge $+Q$ and $-Q$ on the pair; the potential in the space between them solves Laplace's equation with the two conductor surfaces as equipotential boundaries, and the uniqueness theorem (see Uniqueness Theorems, which owns this argument) says that solution is the only one. Scale the charges by any factor $\lambda$ and $\lambda\,V(\mathbf r)$ is a solution with the new boundary data — so by uniqueness it is the solution. Voltage scales in lockstep with charge, always and exactly, for conductors in vacuum or in any linear dielectric. Nonlinearity never enters because the medium (vacuum) has no knobs; that is the entire content of "capacitance exists." The second picture worth carrying: the three geometries below are not three formulas but one recipe run three times — find $E$ between the conductors by symmetry, integrate to get $V$ , divide into $Q$ . Plane, sphere, cylinder: the trio you will use for the rest of your career.

Figure

The canonical trio: one recipe, three symmetries

Defining the object

Everything begins with the ratio. For two conductors carrying $+Q$ and $-Q$ with potential difference $V$ between them (positive conductor minus negative), we define

C \;=\; \frac{Q}{V}

where: $Q$ is the magnitude of the charge on either conductor (not the sum, which is zero), $V$ is the potential difference between the two equipotential bodies, and $C$ — measured in farads, $1\ \mathrm{F}=1\ \mathrm{C/V}$ — is a property of the pair. Read it as a sensitivity: a 1 F capacitor swallows a full coulomb for each volt. That is a colossal appetite. As we verify below, building 1 F from vacuum-gapped plates 1 mm apart takes plates ten kilometres on a side, which is why laboratory life is spent between picofarads ( $10^{-12}$ ) and microfarads ( $10^{-6}$ ), and why "supercapacitors" rating whole farads (they cheat with molecular-scale gaps — see Capacitor Energy and Dielectrics) were headline news. One subtlety the definition hides: a single conductor also gets a capacitance, by quietly placing the second conductor at infinity. "The capacitance of a sphere" means sphere-versus-infinite-shell. And one reassurance it should not hide: because $C$ is fixed by geometry, the electrostatics you compute once — in the lab frame, with no circuit attached — remains valid when the capacitor later sits in an RC network switching a thousand times a second. That transferability is the whole business model of circuit theory. For orientation, here is which tool computes $C$ when:

geometry	field source	result
two parallel plates, gap $\ll$ size	uniform sheet field $\sigma/\varepsilon_0$	$C=\varepsilon_0 A/d$
concentric spheres $a<b$	point-like $1/r^2$ field of inner sphere	$C=4\pi\varepsilon_0\,\dfrac{ab}{b-a}$
isolated sphere ( $b\to\infty$ )	same	$C=4\pi\varepsilon_0 R$
coaxial cylinders $a<b$	line-charge $1/s$ field	$C/L=\dfrac{2\pi\varepsilon_0}{\ln(b/a)}$
anything without symmetry	boundary-value solve (see Laplace's Equation: Solution Methods)	numeric $C$

Derivation AParallel-Plate Capacitance from the Sheet Field

We want $C$ for the workhorse geometry: two flat plates of area $A$ carrying $+Q$ and $-Q$ , separated by a gap $d$ small compared to the plate dimensions. The strategy is the standard recipe's simplest run: get the interior field from results we already own — the conductor-surface field $\sigma/\varepsilon_0$ established in Gauss's Law Applications and Boundary Conditions on E and V, which we recall rather than re-derive — then integrate across the gap for $V$ and divide. The thing to watch is what "small gap" is doing for us: it is licensing us to treat the charge as uniform sheets and to ignore the bulging field at the edges.

Step 1. Charge spreads uniformly. With $d$ tiny, each plate sees the other as an infinite plane, so by symmetry the charge has surface density $\sigma = Q/A$ on the facing surfaces (the outer surfaces carry none: their fields would have nowhere to terminate).

Step 2. Field in the gap. The field just outside a conductor is $E=\sigma/\varepsilon_0$ , directed from $+$ to $-$ ; between infinite sheets this value holds uniformly across the whole gap,

E \;=\; \frac{\sigma}{\varepsilon_0} \;=\; \frac{Q}{\varepsilon_0 A}.

Why not $\sigma/2\varepsilon_0$ ? That is the field of one isolated sheet. Here both sheets contribute, doubling the interior field and cancelling the exterior — equivalently, the conductor-surface rule $\sigma/\varepsilon_0$ already contains both contributions.

Step 3. Voltage across the gap. With $E$ uniform, the line integral is a multiplication:

V \;=\; E\,d \;=\; \frac{Qd}{\varepsilon_0 A}.

Step 4. Divide. $C = Q/V$ gives

C \;=\; \frac{\varepsilon_0 A}{d}

where: capacitance grows with plate area (more room for charge) and blows up as $d\to0$ (the same charge produces vanishing voltage). Note that $Q$ cancelled — the proportionality theorem in action. Why trust the "ignore the edges" move? The exact field bulges outward at the perimeter (fringing), which stores extra field energy and makes the true $C$ slightly larger than $\varepsilon_0A/d$ ; for a gap-to-size ratio of a few percent the correction is at the percent level, and every honest lab capacitor is guarded or measured, not trusted to the formula. *Verified: $\varepsilon_0 A/d$ with $A=0.1\ \mathrm{m^2}$ , $d=1\ \mathrm{mm}$ gives $C=0.885\ \mathrm{nF}$ ; at $V=100\ \mathrm{V}$ that is $Q=88.5\ \mathrm{nC}$ and $E=10^5\ \mathrm{V/m}$ , and recomputing $Q/V$ returns $0.885\ \mathrm{nF}$ exactly.*

Go deeper · Derivation B · 2 minThe Sphere Family: Isolated Sphere and Concentric Shells

Derivation BThe Sphere Family: Isolated Sphere and Concentric Shells

We now run the recipe on spherical symmetry, deriving both the isolated-sphere capacitance and the concentric-shell formula, then audit the two limits that tie the family together. The strategy: the field of the inner sphere is the point-charge field (Gauss's law for spheres — recalled from Gauss's Law Applications with Symmetry, not re-derived), so the voltage integral is the familiar $1/r$ potential difference. Watch how the isolated sphere emerges as the $b\to\infty$ member, and how the parallel plate hides inside the small-gap limit.

Step 1. Set up the pair: inner sphere radius $a$ with $+Q$ , outer shell radius $b$ with $-Q$ . Between them ( $a<r<b$ ) only the inner charge is enclosed, so

E(r) \;=\; \frac{Q}{4\pi\varepsilon_0 r^2}.

Step 2. Integrate from the negative to the positive conductor:

V \;=\; \int_a^b \frac{Q}{4\pi\varepsilon_0 r^2}\,dr \;=\; \frac{Q}{4\pi\varepsilon_0}\left(\frac1a-\frac1b\right) \;=\; \frac{Q}{4\pi\varepsilon_0}\,\frac{b-a}{ab}.

Step 3. Divide into $Q$ :

C \;=\; 4\pi\varepsilon_0\,\frac{ab}{\,b-a\,}

where: the numerator $ab$ rewards large radii, while the denominator $b-a$ — the gap — plays exactly the role $d$ played for plates. The formula is symmetric in the two radii, a reminder that capacitance belongs to the pair.

Step 4. Limit audit one: send the outer shell to infinity, $b\to\infty$ . Then $ab/(b-a)\to a$ and

C \;=\; 4\pi\varepsilon_0\,R

for an isolated sphere of radius $R$ — the "second conductor at infinity" made literal. A 1 cm lab sphere has $C=1.11\ \mathrm{pF}$ ; the entire Earth ( $R=6371\ \mathrm{km}$ ) manages only $C=709\ \mu\mathrm{F}$ . Sit with that: a fingernail-sized electrolytic from a junk drawer out-stores the planet, because the planet's "gap" to infinity is maximally unhelpful while the electrolytic's gap is micrometres. Why is big- $R$ big- $C$ ? A larger sphere holds the same $Q$ at larger radius, hence lower surface potential $Q/4\pi\varepsilon_0R$ — a wider vessel filling to a lower level.

Step 5. Limit audit two: bring the shells close, $b=a+d$ with $d\ll a$ . Then $ab\approx a^2$ and

C \;\approx\; \frac{\varepsilon_0\,(4\pi a^2)}{d} \;=\; \frac{\varepsilon_0 A}{d},

the parallel-plate formula with $A$ the sphere's area. Better still, no approximation is needed at all: since $4\pi\varepsilon_0\,ab/(b-a) = \varepsilon_0(4\pi ab)/d$ identically, the concentric-sphere capacitor is a parallel-plate capacitor whose effective area $4\pi ab$ uses the geometric mean radius $\sqrt{ab}$ . Curvature never spoils the plate picture; it just tells you which radius to use. Verified: $a=5\ \mathrm{cm}$ , $b=20\ \mathrm{cm}$ gives $C=4\pi\varepsilon_0(0.05)(0.20)/0.15=7.4\ \mathrm{pF}$ — matching the boundary-value solution of this same capacitor in Laplace's Equation (their Worked Example harvests $q=0.74\ \mathrm{nC}$ at $100\ \mathrm{V}$ ), a satisfying cross-check between the field route and the BVP route.

Go deeper · Derivation C · 1 minCoaxial Cylinders and the Logarithm

Derivation CCoaxial Cylinders and the Logarithm

Third symmetry, last member of the trio: an inner cylinder of radius $a$ inside a coaxial shell of radius $b$ , the geometry of every signal cable in your lab. The plan is identical — cylindrical Gauss field, integrate, divide — but the payoff is different in character: a logarithm appears, and it is worth understanding physically why that makes coax capacitance so stubbornly insensitive to geometry. We work per unit length $L$ , since the ends of a long cable contribute negligibly.

Step 1. Give the inner conductor charge per length $\lambda = Q/L$ . A coaxial Gaussian cylinder of radius $s$ ( $a<s<b$ ) encloses $\lambda L$ , and symmetry puts the field radially outward with magnitude

E(s) \;=\; \frac{\lambda}{2\pi\varepsilon_0\,s}.

Why $1/s$ and not $1/s^2$ ? Flux spreads over a cylinder area $2\pi s L$ that grows only linearly with $s$ — one power of distance, not two. This is the line-charge field recalled from the Gauss's-law catalogue.

Step 2. Integrate across the gap:

V \;=\; \int_a^b \frac{\lambda}{2\pi\varepsilon_0\,s}\,ds \;=\; \frac{\lambda}{2\pi\varepsilon_0}\,\ln\!\frac{b}{a}.

Step 3. Divide, and quote per unit length:

\frac{C}{L} \;=\; \frac{2\pi\varepsilon_0}{\ln(b/a)}

where: only the ratio $b/a$ enters, and only through a logarithm. Double the cable's overall scale and $C/L$ does not change at all; even swapping $b/a=2$ for $b/a=10$ moves $C/L$ by barely a factor of 3.3. The logarithm is geometry's great compressor — which is why nearly all practical coax clusters within a factor of two of the same capacitance per metre.

Step 4. Numbers. A vacuum-spaced line with RG-58 proportions ( $b/a\approx3.34$ , $\ln(b/a)=1.204$ ) gives $C/L = 2\pi\varepsilon_0/1.204 = 46\ \mathrm{pF/m}$ — precisely the number the relaxation-method solve in Laplace's Equation: Solution Methods extracts for this cross-section by harvesting $\sigma=-\varepsilon_0\,\partial V/\partial n$ , and the commercial cable's quoted $\sim100\ \mathrm{pF/m}$ is this vacuum value boosted by its polyethylene filling ( $\kappa\approx2.25$ ) — the dielectric story taken up in Capacitor Energy and Dielectrics. Verified: $2\pi\times8.854\times10^{-12}/\ln(3.335) = 4.62\times10^{-11}\ \mathrm{F/m} = 46.2\ \mathrm{pF/m}$ ; one metre of vacuum RG-58 is a 46 pF capacitor.

Limiting cases

case	result	meaning
$b\to\infty$ (spheres)	$C\to 4\pi\varepsilon_0 a$	isolated conductor: the far shell drops out; "capacitance of one object" defined
$b-a=d\ll a$ (spheres)	$C=\varepsilon_0(4\pi ab)/d \to \varepsilon_0A/d$	any thin-gap capacitor is locally parallel-plate; curvature only picks the effective area
$d\to0$ (plates)	$C\to\infty$	vanishing gap: any charge produces almost no voltage — and breakdown arrives first in practice
$A\to\infty$ at fixed $d$	$C\to\infty$ linearly	capacitance is extensive in plate area; parallel connection will exploit exactly this
cylinder $b/a\to1$	$C/L\to\infty$ like $1/\ln(b/a)$	thin-gap divergence again, now logarithmically slow
fill gap with dielectric $\kappa$	$C\to\kappa C_{\mathrm{vac}}$	preview of the dielectric chapter: same geometry, $\varepsilon_0\to\kappa\varepsilon_0$

Pitfall

Common trapDoubling the charge doubles the capacitance

It does not — and if you catch yourself computing "the capacitance at 5 volts," stop. Doubling $Q$ doubles $V$ and leaves $C$ untouched: capacitance is the slope of the strictly linear $Q$ – $V$ line, pinned by the uniqueness theorem, not a reading that drifts with operating point. What does change $C$ is geometry or medium: move the plates, change the overlap, bring a dielectric (or a fingertip) near. Every capacitive sensor ever built runs on this distinction — the touchscreen detects a capacitance change because nothing about the applied voltage could have produced one. A related trap: "the capacitance of a sphere" is not a counterexample to "capacitance belongs to a pair"; the partner is a shell at infinity, and forgetting that makes the isolated formula seem to contradict the concentric one.

Worked examples

Example 1 — the keyboard key (confidence builder). A laptop key sits on a capacitive sensor: two plates of area $A=1\ \mathrm{cm^2}$ , gap $d=0.5\ \mathrm{mm}$ of air. Pressing the key squeezes the gap to $0.3\ \mathrm{mm}$ . What capacitance change must the controller detect?

Step 1. Unpressed: $C_1 = \varepsilon_0 A/d_1 = (8.854\times10^{-12})(10^{-4})/(5\times10^{-4}) = 1.77\ \mathrm{pF}$ .

Step 2. Pressed: $C_2 = (8.854\times10^{-12})(10^{-4})/(3\times10^{-4}) = 2.95\ \mathrm{pF}$ .

Step 3. The signature of a keystroke is $\Delta C = C_2-C_1 = \boxed{1.18\ \mathrm{pF}}$ — about a 67% rise. Your keyboard controller resolves picofarads hundreds of times per second, which is routine for a switched-capacitor readout chip; the physics content is simply $C\propto1/d$ .

Example 2 — design audit: sphere versus plate recipe (exam level). An instrument calls for a $\sim45\ \mathrm{pF}$ vacuum capacitor built as concentric spheres with inner radius $a=2.0\ \mathrm{cm}$ and gap $d=1.0\ \mathrm{mm}$ ( $b=2.1\ \mathrm{cm}$ ). Compute $C$ exactly, then with the naive parallel-plate recipe using the inner sphere's area, and explain the discrepancy. Compare with one metre of vacuum RG-58-proportioned coax.

Step 1. Exact: $C = 4\pi\varepsilon_0\,ab/(b-a) = (1.113\times10^{-10})\times(0.020\times0.021)/0.001 = \boxed{46.7\ \mathrm{pF}}$ .

Step 2. Naive plate recipe with $A=4\pi a^2$ : $C_{\mathrm{plate}} = \varepsilon_0(4\pi a^2)/d = 44.5\ \mathrm{pF}$ — low by 5.0%.

Step 3. Diagnose: the exact formula is the plate formula with area $4\pi ab$ , not $4\pi a^2$ . The ratio is $b/a = 1.05$ , exactly the 5% gap. Using the geometric-mean radius $\sqrt{ab}=2.05\ \mathrm{cm}$ reproduces the exact answer to this precision — the limit audit from the derivation, working for a living.

Step 4. The coax alternative: $C = (2\pi\varepsilon_0/\ln 3.335)\times(1\ \mathrm{m}) = 46.2\ \mathrm{pF}$ . One metre of cable equals the precision sphere — which is why real designers reach for cable or plates, and why stray cable capacitance haunts every fast circuit you will ever wire.

History

Origins · 1 minHenry Cavendish and how this was discovered

Henry Cavendish

1731–1810 · Clapham, London

The silent millionaire who measured capacitance a century early — and filed it in a drawer. Cavendish, whose null experiment on the inverse-square law appears earlier in this track, ran between 1771 and 1781 a complete research program on what he called the "charge" of bodies: comparing coated plates, spheres, and Leyden jars against a calibrated sphere, he effectively measured capacitance ratios — using his own body as the electrometer, grading shocks from "perceptible" to "severe." He established that capacity is a stable geometric property, found the plate-capacitor dependence on area and spacing, and even measured the factor by which glass and wax boost capacity: the dielectric constant, sixty-six years before Faraday's celebrated 1837 measurements. Then he published almost none of it. The notebooks surfaced only when Maxwell, in his final years, edited The Electrical Researches of the Honourable Henry Cavendish (1879), performing every experiment himself with Victorian instruments and finding Cavendish right, again and again. Capacitance is the rare quantity that was measured before it was named — twice.

History

1745

— The Leyden jar stuns Europe: Musschenbroek in Leiden (and Kleist in Pomerania) bottle "electric fire" in a water-filled jar, the first capacitor, and half the continent lines up to take the shock

1771

— Cavendish begins his unpublished decade of capacity measurements, comparing plates, spheres, and jars and anticipating the dielectric constant

1837

— Faraday, using concentric-sphere capacitors, shows the insulating medium itself sets the capacity: specific inductive capacity, the door to the dielectric chapter

1861

— The British Association committee on electrical standards begins the unit system that will name the practical unit of capacity the farad, in Faraday's honour

1879

— Maxwell publishes Cavendish's electrical researches, revealing the whole subject had a secret founder

Going further

References. Halliday, Resnick & Walker, Fundamentals of Physics, ch. 25 (capacitance, the canonical trio); Purcell & Morin, Electricity and Magnetism, §3.5–3.6 (conductor pairs, coefficients of capacitance); Griffiths, Introduction to Electrodynamics, §2.5.4 (capacitance from electrostatics); Maxwell (ed.), The Electrical Researches of the Honourable Henry Cavendish (1879) for the secret prehistory.

Provenance. Authored fresh for the SuperSTEM E&M track's circuits chapter (recon-§d gap heal); sources: v3 circuit pilots (rewritten to UG register) + legacy wiki circuit articles (narrative) — no AllQFT base exists for this topic. This article supplies the "Capacitance and Capacitors" / "Capacitance and Stored Energy" prerequisite that earlier articles in the electrostatics chapter reference.

Prerequisites. Conductors in Electrostatic Equilibrium · The Electrostatic Potential · Gauss's Law Applications with Symmetry · Uniqueness Theorems.

Leads to. Capacitors in Series and Parallel · Capacitor Energy and Dielectrics · RC Circuits.

Is capacitance defined for the pair, or does each conductor have its own?

Fundamentally for the pair (or, in full generality, for a whole system of conductors via a matrix of coefficients — Maxwell's formulation). The everyday shortcut "capacitance of a sphere" designates a specific pair: the sphere and a shell pushed to infinity. That is why the isolated formula $4\pi\varepsilon_0R$ is the $b\to\infty$ limit of the two-conductor formula rather than a separate law. In circuits the distinction resurfaces as stray capacitance: your "single" wire always has some partner — the ground plane, the chassis, your hand.

Why exactly is Q proportional to V — could a weird shape break the linearity?

No shape can. Fix the geometry and solve the boundary-value problem for charges $\pm Q$ ; the uniqueness theorem guarantees one solution. Scale all charges by $\lambda$ : the scaled potential satisfies Laplace's equation with the scaled boundary data, so uniqueness forces it to be the solution, and $V$ scales by $\lambda$ too. Linearity fails only if the medium responds nonlinearly (ferroelectrics near saturation, reverse-biased junction "varactors" where the effective gap moves with voltage) — never because of geometry.

Where does the charge actually sit on the plates?

On the two facing surfaces, as the derivation's Step 1 argued: field lines must leave $+\sigma$ and land on $-\sigma$ across the gap, and any charge on the outer faces would generate field with nowhere consistent to go (for an isolated, symmetric pair). This is also why capacitance is unchanged if you thicken the plates: the working surfaces and the gap are all that matter. In the real device, fringing puts a small ribbon of charge near the rims — the percent-level correction flagged in the derivation.

Does the field between the plates depend on the gap distance?

For fixed charge, no: $E=\sigma/\varepsilon_0$ knows only the surface density. Pull the plates of an isolated charged capacitor apart and $E$ stays constant while $V=Ed$ grows linearly — you did work against the plates' attraction, and the voltmeter shows it. For fixed voltage (battery attached), widening the gap thins the field, $E=V/d$ , and charge flows off the plates. Which quantity is held fixed changes every answer in this subject; the energy chapter turns that distinction into a systematic ledger.

Is the capacitance of a real parallel-plate capacitor bigger or smaller than ε₀A/d?

Slightly bigger. The fringing field at the edges is extra field in extra volume: for given $V$ the system holds a bit more charge than the ideal formula accounts for. The correction scales like the perimeter-to-area ratio times the gap — sub-percent for a capacitor whose gap is a hundredth of its width — and precision metrology sidesteps it with guard rings that push the fringes outside the measured region.

Why is one farad so absurdly large, and how do shops sell 3000 F "supercapacitors"?

The farad is the coulomb — itself a monster unit — per volt, and vacuum-gap geometry fights you: $\varepsilon_0\approx8.9\times10^{-12}\ \mathrm{F/m}$ sets the scale, so metre-sized plates with millimetre gaps yield nanofarads. Supercapacitors abandon the plate-gap paradigm entirely: charge sits in an electrochemical double layer about a nanometre thick over an electrode with sponge-like internal area (thousands of square metres per gram). Tiny $d$ , colossal $A$ — the same formula, driven to extremes. The price is a low voltage rating (~2.7 V per cell), which is the energy chapter's problem.

How do I measure a capacitance in the lab without a fancy meter?

Use time or ratio. Time: charge the unknown through a known resistor and watch the voltage rise on a scope — the RC time constant hands you $C=\tau/R$ (this becomes precise in RC Circuits). Ratio: put the unknown in series with a known capacitor across a known AC or stepped voltage and read the voltage division, which goes inversely with capacitance (the series rule from the next article). Cavendish's method was the ratio method, with a calibrated sphere as the standard and his nervous system as the detector; yours can be a 10 MΩ scope probe.

Capacitance

Where this lives today

The physical picture

Defining the object

Limiting cases

Pitfall

Worked examples

History

Going further

Before this

Where this leads