Green's Theorem in the Plane

Green's theorem is the first genuinely two-dimensional fundamental theorem of calculus, and it is the hinge on which all of vector-integral analysis turns. It says something almost unreasonably simple: whatever a planar vector field does around the rim of a region is completely determined by what it does throughout the interior. Trace a closed loop $C$ and add up the field's push along your direction of travel — the circulation — and you have measured, exactly, the total curl bottled up inside. Read the same field the other way and the net flux leaking outward through $C$ equals the total divergence of the sources within. Boundary bookkeeping and interior bookkeeping are two names for one number. The reason is a cancellation so clean it feels like a trick. Pave the region with tiny cells, give each cell its own circulation, and add them up: every interior wall is shared by two neighbours who traverse it in opposite directions, so the interior erases itself and only the outer rim survives. That single idea — interiors cancel, boundaries survive — is the seed of Stokes' theorem, the divergence theorem, the residue calculus, and the one-line master formula $\int_M \mathrm{d}\omega = \int_{\partial M}\omega$ of differential forms. Green's theorem also does honest work: it measures areas by walking a boundary, governs incompressible flow and vorticity, and turns into Cauchy's theorem the instant the field obeys the Cauchy–Riemann equations. This article rebuilds it from the cell picture, proves it three ways, and follows it out to where it still earns its keep.

Key takeaways

Green's theorem converts a closed line integral around a planar curve $C$ into a double integral over the region $R$ that $C$ bounds — boundary data equals interior data.
Circulation form: $\oint_C (P\,\mathrm{d}x + Q\,\mathrm{d}y) = \iint_R (\partial_x Q - \partial_y P)\,\mathrm{d}A$ , with $C$ traversed counterclockwise so $R$ lies to the left.
The integrand $\partial_x Q - \partial_y P$ is the scalar curl, so Green's theorem is the flat restriction of Stokes' theorem; the dual flux form is the planar divergence theorem.
Setting $P=-y/2,\ Q=x/2$ collapses the right side to the area, giving $A=\tfrac12\oint_C(x\,\mathrm{d}y-y\,\mathrm{d}x)$ — the principle of the planimeter.
Validity needs $P,Q$ to be $C^1$ on all of $R$ : the field $(-y,x)/(x^2+y^2)$ is curl-free yet loops to $2\pi$ because it is singular at the enclosed origin.
Key idea: a closed line integral measures net boundary effect; the matching area integral sums the local rotation (curl) or source (divergence). Green's theorem says the two agree exactly, because interior contributions cancel pairwise and only the outer boundary survives.

The core insight

A closed line integral measures net boundary effect; the matching area integral sums the local rotation (curl) or source (divergence). Green's theorem says these two bookkeepings agree exactly, because interior contributions from adjacent cells cancel pairwise and only the outer boundary survives.

Where this lives today

Green's theorem quietly runs under a surprising amount of working technology. Every time a GIS or CAD system reports the area, centroid, or moment of a polygon, it evaluates the shoelace formula — $A=\tfrac12\oint_C(x\,\mathrm{d}y-y\,\mathrm{d}x)$ collapsed onto straight edges — the same identity that let a nineteenth-century mechanical planimeter measure an irregular indicator diagram by tracing its outline. In computational fluid dynamics the circulation form links the bound circulation around an airfoil to the vorticity in the flow, feeding the Kutta–Joukowski theorem that predicts lift; in computer graphics the winding-number rule that decides whether a pixel lies inside a self-intersecting outline is Derivation C made discrete. Green's identities remain the engine of the boundary element method — a workhorse of acoustics, electrostatics, and elastostatics that solves a partial differential equation by discretizing only a region's boundary — a direct dividend of the theorem's promise that boundary data determines interior behaviour.

The physical picture

Picture the region $R$ tiled by a mosaic of tiny square cells. Each cell has its own microscopic circulation — walk once around its edge, tallying the field's push along your path, and you collect an amount equal to the local curl times the cell's area. Now add up the circulations of every cell at once. Along any wall shared by two neighbouring cells, the two walks run in opposite directions: one cell sweeps that wall left-to-right, the other right-to-left, and their contributions annihilate. Every interior wall is shared, so the whole interior cancels itself out. What is left un-cancelled? Only the outermost walls — the boundary curve $C$ — where there is no neighbour on the far side. The grand total is therefore the boundary loop integral on one side and the summed curl on the other, and those must be equal.

Figure

Paving the region R with cells: each cell's circulation equals curl × area, interior walls cancel in opposite-going pairs, and only the outer boundary C survives — that is Green's theorem, ∮ (P dx + Q dy) = ∬ (∂Q/∂x − ∂P/∂y) dA

The same mosaic, read with the arrows turned outward instead of around, gives the flux version: tally the net outflow through each cell's walls, and the shared interior walls again carry equal-and-opposite flux and cancel, leaving only the outflow through $C$ — the accumulated interior source strength $\iint_R \nabla\cdot\mathbf{F}\,\mathrm{d}A$ . Because the result rests only on the boundary, deforming $C$ without crossing a singularity of $\mathbf{F}$ never changes the integral — the topological reflex that grows into the residue theorem. And the humblest special case measures area: pick a field whose curl is exactly one, and the interior sum collapses to $\iint_R 1\,\mathrm{d}A = A$ , computable by a single walk around the rim.

Figure

Area of an ellipse from a boundary walk: with x = a cos θ and y = b sin θ the element x dy − y dx = ab dθ, so A = ½ ∮ (x dy − y dx) = π a b — for a = 3 and b = 2 this gives A = 6π ≈ 18.85

The fundamental theorem of calculus hides inside all of this. Integrating $\partial_y P$ up a vertical fibre uses $\int_{g_1}^{g_2}\partial_y P\,\mathrm{d}y = P(x,g_2)-P(x,g_1)$ , trading an interior integral for boundary values one fibre at a time. Green's theorem is the ordinary FTC stacked sideways across a two-dimensional region and knitted together along the boundary.

The core equations

The primary statement is the circulation (tangential) form. It equates the counterclockwise loop integral of the tangential field component to the area integral of the scalar curl. Everything else in this article is either a re-reading, a specialization, or a consequence of this one line.

\oint_C \left(P\,\mathrm{d}x + Q\,\mathrm{d}y\right) = \iint_R \left(\frac{\partial Q}{\partial x} - \frac{\partial P}{\partial y}\right)\,\mathrm{d}A \qquad (1)

where: $C$ — piecewise-smooth, positively oriented (counterclockwise) simple closed curve; $R$ — closed region bounded by $C$ , lying to the left of the traversal; $P,Q$ — components of $\mathbf{F}=(P,Q)$ , with continuous first partials on $R$ ; $\mathrm{d}A=\mathrm{d}x\,\mathrm{d}y$ — area element. Read it as circulation around $C$ equals total curl inside $R$ . Rotate the field by ninety degrees and the same theorem reads as a statement about flux and divergence — the two-dimensional divergence (Gauss) theorem. Here the field is dotted with the outward normal rather than the tangent.

\oint_C \mathbf{F}\cdot\hat{\mathbf{n}}\,\mathrm{d}s = \iint_R \left(\frac{\partial F_x}{\partial x} + \frac{\partial F_y}{\partial y}\right)\,\mathrm{d}A \qquad (2)

where: $\hat{\mathbf{n}}$ — outward unit normal to $C$ ; $\mathrm{d}s$ — arc-length element; $F_x,F_y$ — Cartesian components of $\mathbf{F}$ . Read it as net outward flux through $C$ equals total divergence inside $R$ . Derivation B shows (1) and (2) are the same theorem under the swap $P=-F_y,\ Q=F_x$ . The integrand of the circulation form is not an accident of notation: it is the $z$ -component of the three-dimensional curl of $\mathbf{F}=(P,Q,0)$ . This is exactly why Green's theorem is the flat shadow of Stokes' theorem.

\left(\nabla\times\mathbf{F}\right)_z = \frac{\partial Q}{\partial x} - \frac{\partial P}{\partial y} \qquad (3)

where: $(\nabla\times\mathbf{F})_z$ — the $\hat{\mathbf{z}}$ -component of $\nabla\times(P,Q,0)$ . When this vanishes throughout a simply connected $R$ , every loop integral is zero and $\mathbf{F}$ is a gradient field. Choosing a field whose curl is identically one turns the area integral into the area itself, and the theorem becomes a formula for area as a boundary walk — the analytic heart of the planimeter and the shoelace formula.

A = \frac{1}{2}\oint_C \left(x\,\mathrm{d}y - y\,\mathrm{d}x\right) = \oint_C x\,\mathrm{d}y = -\oint_C y\,\mathrm{d}x \qquad (4)

where: $A$ — area of the region enclosed by $C$ . The symmetric first form comes from $P=-y/2,\ Q=x/2$ ; the two one-sided forms from $P=0,\ Q=x$ and $P=-y,\ Q=0$ , each giving $\partial_x Q-\partial_y P=1$ . Feeding the flux form the special field $\mathbf{F}=\phi\,\nabla\psi$ produces Green's first identity — the planar workhorse of potential theory, and the origin of the man's name attached to the whole subject.

\oint_C \left(\phi\,\nabla\psi\right)\cdot\hat{\mathbf{n}}\,\mathrm{d}s = \iint_R \left(\phi\,\nabla^2\psi + \nabla\phi\cdot\nabla\psi\right)\,\mathrm{d}A \qquad (5)

where: $\phi,\psi$ — twice-differentiable scalar fields on $R$ ; $\nabla^2=\partial_x^2+\partial_y^2$ — planar Laplacian. Subtracting this identity from itself with $\phi$ and $\psi$ swapped gives Green's second identity, $\oint_C(\phi\,\nabla\psi-\psi\,\nabla\phi)\cdot\hat{\mathbf{n}}\,\mathrm{d}s=\iint_R(\phi\,\nabla^2\psi-\psi\,\nabla^2\phi)\,\mathrm{d}A$ . Finally, all of these are avatars of a single statement about differential forms, in which Green's theorem, Stokes' theorem, the divergence theorem, and even the ordinary FTC are one line read on regions of different dimension.

\int_M \mathrm{d}\omega = \int_{\partial M} \omega \qquad (6)

where: $\omega=P\,\mathrm{d}x+Q\,\mathrm{d}y$ — a differential 1-form; $\mathrm{d}\omega=(\partial_x Q-\partial_y P)\,\mathrm{d}x\wedge\mathrm{d}y$ — its exterior derivative; $M,\partial M$ — the region $R$ and its oriented boundary $C$ . The generalized Stokes' theorem is Green's cell-cancellation argument carried out in any dimension.

Derivation AProof on a vertically simple region

The goal is to prove the circulation form outright, not merely motivate it, by establishing the $P$ -half on a region caught between two graphs and then quoting the mirror argument for the $Q$ -half. The whole proof is the fundamental theorem of calculus applied one vertical fibre at a time and summed across the region — the cell-cancellation picture in the limit of infinitely fine paving. Take $R=\{(x,y): a\le x\le b,\ g_1(x)\le y\le g_2(x)\}$ , vertically simple with continuous $g_1\le g_2$ , and aim at the boxed identity $\oint_C P\,\mathrm{d}x = -\iint_R \partial_y P\,\mathrm{d}A$ .

Step 1. Reduce the area integral by the FTC: $\iint_R \partial_y P\,\mathrm{d}A = \int_a^b\!\int_{g_1(x)}^{g_2(x)} \partial_y P\,\mathrm{d}y\,\mathrm{d}x = \int_a^b\big[P(x,g_2(x))-P(x,g_1(x))\big]\,\mathrm{d}x$ . Why: the inner $y$ -integral of a pure $y$ -derivative is just the antiderivative evaluated at the top and bottom of the fibre — the one-dimensional FTC doing the real work.

Step 2. Write the boundary line integral by pieces: the counterclockwise curve runs along the bottom graph $y=g_1$ from $x=a$ to $x=b$ , then back along the top graph $y=g_2$ from $x=b$ to $x=a$ , so $\oint_C P\,\mathrm{d}x = \int_a^b P(x,g_1)\,\mathrm{d}x + \int_b^a P(x,g_2)\,\mathrm{d}x$ . Why: the two vertical sides contribute nothing because $\mathrm{d}x=0$ where $x$ is held fixed, so only the top and bottom graphs matter.

Step 3. Flip the reversed limit and compare: $\oint_C P\,\mathrm{d}x = \int_a^b\big[P(x,g_1)-P(x,g_2)\big]\,\mathrm{d}x = -\iint_R \partial_y P\,\mathrm{d}A$ , which is the boxed $P$ -part. Why: reversing $\int_b^a$ to $\int_a^b$ flips the sign, and the bracket is exactly the negative of the Step 1 result — the two independent computations lock together.

Step 4. Combine with the dual: the mirror argument on a horizontally simple region (bounded left and right by graphs $x=h(y)$ ) gives $\oint_C Q\,\mathrm{d}y = +\iint_R \partial_x Q\,\mathrm{d}A$ ; adding the two halves yields the full circulation form. Why: a region that is both vertically and horizontally simple satisfies both halves at once, and a general region is stitched from such pieces with interior cuts that cancel. *Verified: both halves reduce to the 1-D FTC applied fibrewise; consistent with Stone & Goldbart §11 and Riley–Hobson–Bence §11.3.*

Go deeper · Derivation B · 1 minEquivalence of the circulation and flux forms

Derivation BEquivalence of the circulation and flux forms

The aim is to show that the tangential (curl) form and the normal (flux) form are not two theorems but one theorem in two coordinate frames, related by a rigid ninety-degree rotation of the field. This matters because the planar divergence theorem then needs no separate proof, and the sign conventions — which way the normal points, which way the curve turns — get fixed once and for all by the tangent-to-normal rotation. We start from the circulation form (1) and feed it a field rotated by the substitution $P=-F_y,\ Q=F_x$ .

Step 1. Transform the integrand: $\partial_x Q - \partial_y P = \partial_x F_x + \partial_y F_y = \nabla\cdot\mathbf{F}$ . Why: the ninety-degree relabelling turns the scalar curl of the rotated field into the divergence of the original field, so the right-hand area integrand becomes the divergence exactly.

Step 2. Transform the differential: $P\,\mathrm{d}x + Q\,\mathrm{d}y = -F_y\,\mathrm{d}x + F_x\,\mathrm{d}y$ , and on $C$ write $\mathrm{d}x = T_x\,\mathrm{d}s,\ \mathrm{d}y = T_y\,\mathrm{d}s$ with unit tangent $\hat{\mathbf{T}}=(T_x,T_y)$ . Why: parametrizing by arc length exposes the tangent vector, which is the object that must be rotated to produce the normal.

Step 3. Rotate tangent to normal: for a counterclockwise curve the outward normal is the tangent turned clockwise by ninety degrees, $\hat{\mathbf{n}}=(T_y,-T_x)$ , so $-F_y T_x + F_x T_y = F_x T_y - F_y T_x = \mathbf{F}\cdot\hat{\mathbf{n}}$ . Why: this is the single geometric fact that converts the tangential integrand into the normal one; the sign is fixed by demanding that $\hat{\mathbf{n}}$ point out of the region on the left of the traversal.

Step 4. Reassemble: substituting Steps 1–3 into (1) turns $\oint_C(P\,\mathrm{d}x+Q\,\mathrm{d}y)=\iint_R(\partial_x Q-\partial_y P)\,\mathrm{d}A$ into $\oint_C \mathbf{F}\cdot\hat{\mathbf{n}}\,\mathrm{d}s=\iint_R \nabla\cdot\mathbf{F}\,\mathrm{d}A$ , which is the flux form (2). Why: nothing was assumed beyond the rotation, so the flux form inherits its proof entirely from the circulation form — one theorem, two faces. *Verified: the rotation $\hat{\mathbf{n}}=(T_y,-T_x)$ and the identity $\mathbf{F}\cdot\hat{\mathbf{n}}=F_x T_y-F_y T_x$ reproduce the planar divergence theorem; consistent with Arfken & Weber §3.5.*

Go deeper · Derivation C · 2 minExtension to a region with a hole and the winding number

Derivation CExtension to a region with a hole and the winding number

The purpose is to push Green's theorem past simply connected regions to a region with a hole, and to harvest the payoff: a curl-free field can wrap the hole and return a nonzero loop integral, the germ of winding numbers and residues. The device is a pair of thin bridge-cuts that slice the annular region into simply connected pieces, each bridge walked twice in opposite directions and therefore self-erasing — the same cancellation that killed the interior of a solid region. Let $R$ lie between an outer curve $C_\text{out}$ and an inner hole boundary $C_\text{in}$ , aiming at $\oint_{C_\text{out}}+\oint_{C_\text{in}} = \iint_R(\partial_x Q-\partial_y P)\,\mathrm{d}A$ with each piece positively oriented.

Step 1. Cut and apply: two thin bridges split $R$ into simply connected pieces $R_1,R_2$ , on each of which the earlier proof holds, so $\oint_{\partial R_1}+\oint_{\partial R_2} = \iint_{R_1}+\iint_{R_2} = \iint_R(\partial_x Q-\partial_y P)\,\mathrm{d}A$ . Why: the area integrals simply add because $R_1$ and $R_2$ tile $R$ without overlap, and each piece is now a region the vertically-simple proof already covers.

Step 2. Cancel the bridges: every bridge edge appears once in $\partial R_1$ and once in $\partial R_2$ , traversed in opposite directions, so those line-integral contributions cancel in pairs. Why: this is the identical opposite-direction cancellation of the cell picture, now applied to the artificial cuts rather than to interior cell walls.

Step 3. Read off the surviving boundary: what remains is $C_\text{out}$ counterclockwise together with $C_\text{in}$ clockwise, which is the correct positive orientation of $\partial R$ (region always on the left). If the field is curl-free on $R$ , the area integral vanishes and $\oint_{C_\text{out}}\mathbf{F}\cdot\mathrm{d}\mathbf{r}=\oint_{C_\text{in}}\mathbf{F}\cdot\mathrm{d}\mathbf{r}$ . Why: with zero interior curl the outer and inner loops must agree, so a loop can be slid freely as long as it never crosses the hole — path-independence up to homotopy.

Step 4. Test on the vortex field $\mathbf{F}=(-y,x)/(x^2+y^2)$ : it is curl-free wherever defined, yet on any circle enclosing the origin $\oint \mathbf{F}\cdot\mathrm{d}\mathbf{r}=2\pi$ . Why: the origin is a genuine hole where $\mathbf{F}$ is undefined, so the invariant loop value counts how many times the curve winds the singularity — the $2\pi$ winding number that seeds the residue theorem. *Verified: bridge cancellation is the standard multiply-connected argument in Arfken & Weber §3.5; the $2\pi$ vortex value matches direct evaluation (see Worked example 3).*

Limiting cases

Green's theorem specializes and connects to its neighbours in several instructive limits.

Case	Result	Meaning
Conservative field	$\partial_x Q = \partial_y P \Rightarrow \oint_C(P\,\mathrm{d}x+Q\,\mathrm{d}y)=0$	Zero curl on a simply connected $R$ makes every closed loop vanish: the field is path-independent, a gradient of some potential.
Lift to 3-D (Stokes)	$\oint_C\mathbf{F}\cdot\mathrm{d}\mathbf{r}=\iint_S(\nabla\times\mathbf{F})\cdot\hat{\mathbf{n}}\,\mathrm{d}S$	With $S$ the flat region $R$ and $\hat{\mathbf{n}}=\hat{\mathbf{z}}$ , Stokes' theorem collapses exactly onto the circulation form.
Divergence (Gauss in 2-D)	$\oint_C\mathbf{F}\cdot\hat{\mathbf{n}}\,\mathrm{d}s=\iint_R\nabla\cdot\mathbf{F}\,\mathrm{d}A$	Trading tangential for outward-normal component gives the planar divergence theorem, Gauss's law in the plane.
Area functional	$\partial_x Q-\partial_y P=1 \Rightarrow \oint_C(x\,\mathrm{d}y-y\,\mathrm{d}x)=2A$	A unit-curl field makes the interior integral count area, turning a boundary walk into a measurement of enclosed area.
Singular field	$\displaystyle\oint_{\|\mathbf{r}\|=a}\frac{-y\,\mathrm{d}x+x\,\mathrm{d}y}{x^2+y^2}=2\pi$	With an enclosed singularity the naive area integral (zero) is wrong; the loop returns $2\pi$ , foreshadowing residues and winding numbers.

Common trapGreen's theorem still applies as long as the curl is zero everywhere on the curve

The single most common error is to apply Green's theorem when the field has a singularity inside the loop. The theorem requires $P$ and $Q$ and their first partials to be continuous on the entire closed region $R$ , interior included — not merely on the curve $C$ . The textbook trap is $\mathbf{F}=(-y,x)/(x^2+y^2)$ : a quick check gives $\partial_x Q-\partial_y P=0$ everywhere the field is defined, so a careless application predicts $\oint_C\mathbf{F}\cdot\mathrm{d}\mathbf{r}=0$ . But the actual loop integral around the origin is $2\pi$ , because the field blows up at the enclosed origin and the hypothesis fails there. The fix is to excise the singularity with a small inner circle and apply the multiply-connected form; the surviving inner loop carries the $2\pi$ . A second, cheaper mistake is orientation: the theorem is stated for counterclockwise $C$ with $R$ on the left, and a clockwise traversal flips the sign of the whole line integral.

Worked examples

1 · A line integral over the unit square (confidence builder)

Evaluate $\oint_C (x^2 y\,\mathrm{d}x + x\,\mathrm{d}y)$ counterclockwise around the unit square $[0,1]\times[0,1]$ .

Step 1. Identify the components and form the curl integrand: $P=x^2 y$ and $Q=x$ , so $\partial_x Q - \partial_y P = 1 - x^2$ . The direct four-segment line integral is avoidable — Green's theorem replaces it with one area integral.

Step 2. Integrate over the square, doing the $y$ -integral first (the integrand has no $y$ -dependence, so it just multiplies by the unit height): $\iint_R (1-x^2)\,\mathrm{d}A = \int_0^1\!\int_0^1 (1-x^2)\,\mathrm{d}y\,\mathrm{d}x = \int_0^1 (1-x^2)\,\mathrm{d}x$ .

Step 3. Evaluate the remaining integral: $\int_0^1(1-x^2)\,\mathrm{d}x = 1 - \tfrac13 = \tfrac23$ . The loop integral is therefore $\mathbf{2/3}$ , obtained without ever parametrizing a single edge of the square.

2 · Area of an ellipse by a boundary walk

Use $A=\tfrac12\oint_C(x\,\mathrm{d}y - y\,\mathrm{d}x)$ to find the area enclosed by the ellipse $x=a\cos\theta,\ y=b\sin\theta$ , and evaluate it for $a=3,\ b=2$ .

Step 1. Differentiate the parametrization: $\mathrm{d}x=-a\sin\theta\,\mathrm{d}\theta$ and $\mathrm{d}y=b\cos\theta\,\mathrm{d}\theta$ , with $\theta$ running $0\to2\pi$ counterclockwise.

Step 2. Assemble the integrand: $x\,\mathrm{d}y - y\,\mathrm{d}x = (a\cos\theta)(b\cos\theta)\,\mathrm{d}\theta - (b\sin\theta)(-a\sin\theta)\,\mathrm{d}\theta = ab(\cos^2\theta+\sin^2\theta)\,\mathrm{d}\theta = ab\,\mathrm{d}\theta$ . The Pythagorean identity collapses the messy boundary integrand to a constant.

Step 3. Integrate and evaluate: $A=\tfrac12\int_0^{2\pi} ab\,\mathrm{d}\theta = \tfrac12\,ab\,(2\pi) = \pi a b$ . For $a=3,\ b=2$ this gives $A=\mathbf{6\pi\approx 18.85}$ , the exact area of the ellipse — recovered by walking its rim, with the disk area $\pi r^2$ as the special case $a=b=r$ .

3 · The vortex field and why the hypothesis matters

Evaluate $\oint_C\mathbf{F}\cdot\mathrm{d}\mathbf{r}$ for $\mathbf{F}=(-y,x)/(x^2+y^2)$ around the unit circle, and reconcile it with $\partial_x Q-\partial_y P=0$ .

Step 1. Check the curl: with $P=-y/(x^2+y^2)$ and $Q=x/(x^2+y^2)$ , a short computation gives $\partial_x Q = \partial_y P = (y^2-x^2)/(x^2+y^2)^2$ , so $\partial_x Q-\partial_y P=0$ everywhere the field is defined. Green's theorem would seem to force the loop integral to zero.

Step 2. Compute the loop directly on the unit circle $x=\cos\theta,\ y=\sin\theta$ , where $x^2+y^2=1$ : the integrand becomes $-y\,\mathrm{d}x + x\,\mathrm{d}y = \sin^2\theta\,\mathrm{d}\theta + \cos^2\theta\,\mathrm{d}\theta = \mathrm{d}\theta$ , so $\oint_C\mathbf{F}\cdot\mathrm{d}\mathbf{r} = \int_0^{2\pi}\mathrm{d}\theta = \mathbf{2\pi}$ .

Step 3. Reconcile the two: the results disagree because Green's theorem does not apply — the field is singular at the enclosed origin, so it is not $C^1$ on all of $R$ and the hypothesis fails. Excising the origin with a small circle and using the multiply-connected form (Derivation C) puts the surviving $2\pi$ on the inner loop. The number is the winding number times $2\pi$ , the exact bridge from Green's theorem to the residue calculus.

Physics of the portrait

The theorem is named for a man who taught himself the mathematics of his age in the attic of a windmill, published the single most influential pamphlet in the history of applied analysis at his own expense, and then went almost entirely unread until after his death.

Origins · 1 minGeorge Green and how this was discovered

George Green

1793–1841 · self-taught Nottingham miller who founded potential theory and the theory of Green's functions

George Green was the son of a Nottingham baker and had barely a year of formal schooling; he worked the family windmill and educated himself from borrowed books, most likely through the local subscription library. In 1828 he privately printed An Essay on the Application of Mathematical Analysis to the Theories of Electricity and Magnetism, a work of staggering originality that introduced the word "potential", the integral identities now called Green's first and second identities, and the device we call the Green's function — all built on the planar boundary-to-interior theorem that carries his name. Only a few dozen subscribers bought it, and it sank without trace. At the age of forty Green entered Gonville and Caius College, Cambridge, as an undergraduate, graduating fourth wrangler, but died in 1841 with his masterwork still obscure. Five years later William Thomson, the future Lord Kelvin, stumbled on a copy, grasped its importance at once, and had it republished — after which Green's methods became the backbone of electromagnetism, potential theory, and eventually quantum field theory, where the propagator is a Green's function in disguise.

History

1793

— George Green born in Nottingham, son of a baker and miller

1828

— Green privately prints his Essay, stating the theorem and introducing the term potential and the Green's function

1833

— at the age of forty Green enters Gonville and Caius College, Cambridge, as an undergraduate

1846

— William Thomson rediscovers the Essay and arranges its republication, five years after Green's death

1889

— Volterra and successors extend the boundary theorems toward the general Stokes formula for forms

Going further

Texts. Stone & Goldbart, Mathematics for Physics, §11 (vector calculus) · Arfken & Weber, Mathematical Methods for Physicists, §3.5 (Gauss's and Stokes's theorems) · Riley, Hobson & Bence, Mathematical Methods for Physics and Engineering, §11.3 · Hassani, Mathematical Physics, Ch. 15 · Spivak, Calculus on Manifolds, Ch. 5 (the generalized Stokes' theorem).

Provenance. Equations cross-checked against Stone & Goldbart §11, Arfken & Weber §3.5, and Riley–Hobson–Bence §11.3. A verifier should: (1) confirm the circulation-to-flux equivalence via the tangent-normal rotation $\hat{\mathbf{n}}=(T_y,-T_x)$ ; (2) re-derive the vertically-simple proof by the fundamental theorem of calculus applied fibrewise; (3) recompute the check integral $2/3$ , the ellipse area $\pi a b = 6\pi\approx 18.85$ , and the vortex value $2\pi$ ; (4) verify Green's first identity follows from the product rule $\nabla\cdot(\phi\nabla\psi)=\nabla\phi\cdot\nabla\psi+\phi\nabla^2\psi$ and the flux form. No experimental numbers are claimed; the $2\pi$ is the exact winding-number result.

Prerequisites. Line Integrals · Scalar and Vector Fields · Curl of a Vector Field · Divergence of a Vector Field. Leads to. Stokes' Theorem · The Divergence (Gauss) Theorem · The Laplacian Operator.

Ask the tutor: Ask me to walk through any of these — the cell-cancellation proof, the circulation-to-flux rotation, or why the vortex field sneaks a $2\pi$ past a zero curl — and we will build it up step by step.

Is Green's theorem the circulation statement or the flux statement?

It is both — they are the same theorem written in two frames. The circulation (tangential) form equates the counterclockwise loop integral of the field's tangential component to the area integral of the scalar curl $\partial_x Q-\partial_y P$ . The flux (normal) form equates the outward flux through the curve to the area integral of the divergence. Rotating the field by ninety degrees, through the substitution $P=-F_y,\ Q=F_x$ , carries one exactly into the other, as Derivation B shows. So there is no need to memorize them as separate results: prove one, rotate, and you have the other for free. Which one you reach for depends only on whether your physics is about circulation and vorticity or about flux and sources.

Why does the counterclockwise orientation matter, and what happens if I go clockwise?

Orientation is baked into the sign of the area integrand. The theorem is stated so that as you walk along $C$ the region $R$ stays on your left, which for a simple loop means counterclockwise. This convention is what makes the right-hand side come out as $+(\partial_x Q-\partial_y P)$ rather than its negative, and it is the same convention that fixes the outward normal as the tangent rotated clockwise. If you traverse $C$ clockwise instead, every $\mathrm{d}x$ and $\mathrm{d}y$ reverses, so the whole line integral flips sign: you would get $-\iint_R(\partial_x Q-\partial_y P)\,\mathrm{d}A$ . Nothing is wrong with a clockwise walk as long as you carry the minus sign consistently.

What exactly goes wrong with the field (−y, x)/(x² + y²)?

Its curl $\partial_x Q-\partial_y P$ is genuinely zero at every point where the field is defined, which tempts you to conclude the loop integral vanishes. But the field is undefined at the origin — both components blow up there — so it is not continuously differentiable on the interior of any loop enclosing the origin, and Green's theorem simply does not apply. When you compute the loop directly you get $2\pi$ . The honest bookkeeping is to cut out the origin with a tiny circle and use the multiply-connected form: the annulus between your loop and the tiny circle is now singularity-free, the area integral there is zero, and the $2\pi$ lives on the inner circle. This is precisely the reasoning that becomes the residue theorem.

How can a boundary integral possibly know the area inside?

Because area is just the integral of the constant function one over the region, and Green's theorem lets you write that constant as a curl. Any field with $\partial_x Q-\partial_y P=1$ — for instance $P=-y/2,\ Q=x/2$ — turns the right-hand side into $\iint_R 1\,\mathrm{d}A=A$ , so the equal left-hand side $\tfrac12\oint_C(x\,\mathrm{d}y-y\,\mathrm{d}x)$ must also be the area. Physically, the boundary integral accumulates the signed swept area as your position vector sweeps around the curve, and by the time you return it has swept the whole enclosed region exactly once. A mechanical planimeter is this integral made of gears and a tracing arm; the shoelace formula is the same thing for a polygon.

How is Green's theorem related to Stokes' and the divergence theorems?

Green's theorem is the flat, two-dimensional ancestor of both. Take Stokes' theorem, $\oint_C\mathbf{F}\cdot\mathrm{d}\mathbf{r}=\iint_S(\nabla\times\mathbf{F})\cdot\hat{\mathbf{n}}\,\mathrm{d}S$ , and let the surface $S$ be a flat patch of the $xy$ -plane with upward normal $\hat{\mathbf{z}}$ : the surface integral becomes the area integral of $\partial_x Q-\partial_y P$ and you are back to the circulation form. The flux form is likewise the two-dimensional divergence theorem. All three, together with the ordinary fundamental theorem of calculus, are single instances of $\int_M\mathrm{d}\omega=\int_{\partial M}\omega$ , the generalized Stokes' theorem for differential forms, read on regions of dimension one, two, and three. Green's cell-cancellation argument is the proof of that master formula in miniature.

What is the connection to complex analysis and Cauchy's theorem?

It is remarkably direct. Write a complex function as $f=u+iv$ and expand $\oint_C f(z)\,\mathrm{d}z=\oint_C(u+iv)(\mathrm{d}x+i\,\mathrm{d}y)$ into its real and imaginary line integrals. Applying Green's theorem to each and imposing the Cauchy–Riemann equations $u_x=v_y,\ u_y=-v_x$ makes both area integrands vanish identically, so $\oint_C f(z)\,\mathrm{d}z=0$ for a function holomorphic on a simply connected region — that is Cauchy's integral theorem. Put a pole inside, and the multiply-connected version of Derivation C keeps a nonzero residue, giving Cauchy's integral formula and the whole residue calculus. Green's theorem is, in this sense, the real-variable skeleton underneath complex integration.