Calculus › Section P · Vector analysis › Topic P.8

Stokes's theorem

Circulation around a closed curve equals the flux of the curl through any surface that spans it — the last boundary-for-interior trade, and the one that unites the whole family.

The statement

Stokes's theorem is Green's circulation form promoted to three dimensions. If \(C\) is a positively oriented, piecewise-smooth closed curve that bounds an oriented surface \(S\), and \(\mathbf{F} = \langle M, N, P \rangle\) is smooth on a region containing \(S\), then

\oint_C \mathbf{F} \cdot d\mathbf{r} = \iint_S (\operatorname{curl} \mathbf{F}) \cdot d\mathbf{S}.

Circulation around the boundary equals curl flux through the surface. Green's theorem already said this for a flat region in the plane: \(\oint M \, dx + N \, dy = \iint (N_x - M_y) \, dA\), and \(N_x - M_y\) is precisely the \(k\)-component of the curl. Stokes keeps the same trade and lets the region curve, tilt, and bubble out of the plane.

Orientation is the whole game

The two sides of the theorem must be oriented to match, and the matching rule is physical. Walk along \(C\) in its positive direction with your head pointing the way the normal \(\mathbf{n}\) points; the surface must lie on your left. Equivalently, curl the fingers of your right hand along \(C\) and your thumb gives \(\mathbf{n}\). Flip the normal and the positive direction of \(C\) flips with it — both sides of the equation negate, and the theorem still holds. But mix conventions (normal up, curve clockwise as seen from above) and the answer arrives with the wrong sign, the single most common Stokes error. The line-integrals topic's convention — counterclockwise viewed from above for curves in the \(xy\)-plane — corresponds to the upward normal.

Why it is true: a sketch

Tile \(S\) with many tiny patches. Each patch is a tiny loop, and its circulation is approximately \((\operatorname{curl} \mathbf{F}) \cdot \mathbf{n}\) times its area — curl measures spin per unit of area, about the normal direction, exactly as the vector-fields topic defined it. Now add the circulations of all patches. Every interior edge is shared by two neighbouring patches and traversed once in each direction, so the contributions cancel in pairs; only edges on the boundary curve \(C\) survive, and their total is \(\oint_C \mathbf{F} \cdot d\mathbf{r}\). The sum on the other side is a Riemann sum for \(\iint_S (\operatorname{curl} \mathbf{F}) \cdot d\mathbf{S}\). It is the same cancellation that powered Green's theorem in tiles and the divergence theorem in boxes — the family signature.

Worked example: the swirl, one dimension up

Take \(\mathbf{F} = \langle -y, x, 0 \rangle\) and \(C\) the circle of radius \(R\) in the \(xy\)-plane, counterclockwise viewed from above. Direct route, from the line-integrals topic: \(\mathbf{r}(t) = \langle R \cos t, R \sin t, 0 \rangle\), so \(\mathbf{F} \cdot \mathbf{r}'(t) = R^2\), and the work is \(\int_0^{2\pi} R^2 \, dt = 2\pi R^2\). Stokes route: \(\operatorname{curl} \mathbf{F} = \langle 0, 0, 2 \rangle\) (computed in the vector-fields topic), and \(C\) bounds the flat disk with upward normal \(\mathbf{k}\), so the flux is \(2 \cdot \pi R^2 = 2\pi R^2\). Same answer — and the upper hemisphere spans \(C\) too: the curl is constant, and its flux through the hemisphere equals \(2\) times the hemisphere's projected area \(\pi R^2\), again \(2\pi R^2\).

Worked example: a hairy curve made easy

Compute the circulation of \(\mathbf{F} = \langle z, x, y \rangle\) around the unit circle in the \(xy\)-plane, counterclockwise viewed from above. Stokes first: \(\operatorname{curl} \mathbf{F} = \langle 1, 1, 1 \rangle\) — component by component, \(\partial y/\partial y - \partial x/\partial z = 1\), \(\partial z/\partial z - \partial y/\partial x = 1\), \(\partial x/\partial x - \partial z/\partial y = 1\). Span the circle with the flat unit disk, normal \(\mathbf{k}\): only the third component of the curl contributes, so the flux is \(1 \cdot \pi = \pi\). Direct audit: on the curve \(z = 0\), so \(\mathbf{F} = \langle 0, \cos t, \sin t \rangle\) and \(\mathbf{F} \cdot \mathbf{r}'(t) = \cos^2 t\), whose integral over \([0, 2\pi]\) is \(\pi\). Both routes agree — and the disk route took three lines.

The family tree, complete

The fundamental theorem of calculus integrates a derivative over an interval and reads the answer off the boundary — two points. Green's theorem did it for flat regions of the plane, in two flavours: circulation (paired with \(N_x - M_y\)) and flux (paired with \(M_x + N_y\)). The divergence theorem is the flux flavour in three dimensions, trading boundary flux for \(\operatorname{div} \mathbf{F}\) over a solid. Stokes is the circulation flavour in three dimensions, trading boundary circulation for curl flux over a surface. Green is literally Stokes in disguise: a flat surface with normal \(\mathbf{k}\) turns \((\operatorname{curl} \mathbf{F}) \cdot \mathbf{k} = N_x - M_y\) into exactly the circulation form. Four theorems, one idea: derivative inside, boundary outside.

Choosing the surface

Any oriented surface whose boundary is \(C\) may be used — a disk, a hemisphere, a soap bubble — and the answer is the same. That freedom is tactical: when \(C\) is a wriggling space curve, span it with the flattest, simplest surface available and integrate the curl there. The freedom is also safe, and the reason is an old identity: take two surfaces \(S_1\) and \(S_2\) spanning \(C\); glued together along \(C\) they form a closed surface, and the flux of a curl through any closed surface is \(\iiint \operatorname{div}(\operatorname{curl} \mathbf{F}) \, dV = 0\), the vector-fields identity. So the two fluxes are equal. The fine print matches the divergence theorem's: \(\mathbf{F}\) must be smooth throughout the region between the two surfaces, or the closed-surface argument has a hole in it.

Common mistakes to avoid

First, mismatched orientation: normal up with curve clockwise, or the reverse — check the walking rule before computing anything. Second, wrong boundary: the surface's boundary must be exactly \(C\); if the surface has extra rim, the theorem counts that rim too. Third, trusting the surface-freedom across a singularity: \(\mathbf{F}\) must be smooth between the surfaces, or the \(\operatorname{div}(\operatorname{curl} \mathbf{F}) = 0\) argument breaks. Fourth, sign slips in the curl: the middle component is \(P_x - M_z\), a difference that punishes haste. Fifth, overworking: when the curve is a friendly circle, the direct line integral is a genuine audit — run it (the \(\pi\) example took three lines both ways) rather than trusting a single route. Sixth, confusing the theorems: Stokes pairs circulation with curl flux; the divergence theorem pairs flux with divergence — match the integral you have to the theorem that consumes it.

Key takeaways

Stokes's theorem trades boundary circulation for interior spin: \(\oint_C \mathbf{F} \cdot d\mathbf{r} = \iint_S (\operatorname{curl} \mathbf{F}) \cdot d\mathbf{S}\), with \(C\) the positively oriented boundary of the oriented \(S\). Orientation must match — walk \(C\) head-along-\(\mathbf{n}\) with the surface on your left. The proof is the family's cancellation argument: tiny patch circulations cancel along interior edges. Any surface spanning \(C\) gives the same answer, safe because \(\operatorname{div}(\operatorname{curl} \mathbf{F}) = 0\) — so choose the easiest one. Green's circulation form is Stokes on a flat surface. And with it the chapter closes: FTC, Green, Stokes, divergence — one theorem at four sizes, each trading interior derivatives for boundary data.

Flashcards

Flashcard

Stokes's theorem

∮ F·dr = ∬ (curl F)·dS over a spanning surface

Circulation around the boundary = curl flux through the surface.

Test your understanding

Question 1 of 5

Stokes's theorem equates the circulation of a smooth field around a closed curve C with which quantity?

Question 2 of 5

Find the circulation of F = ⟨z, x, y⟩ around the unit circle in the xy-plane, counterclockwise viewed from above.

Question 3 of 5

By the walking rule, when you stroll along C in its positive direction with your head pointing along the normal n, the surface S must lie on which side?

Question 4 of 5

What is the work done by F = ⟨−y, x, 0⟩ in one counterclockwise circuit of the circle of radius 2 in the xy-plane?

Question 5 of 5

Why may any surface spanning C be used on the flux side of Stokes's theorem?

Standards: Beyond AP

← Previous topic