Circulation: adding up the push around a loop
Start with a picture you already own from the last guides in this rung. A vector field F fills the plane with arrows — think of the surface of a slow river seen from above. Now drop a closed loop into that flow, a little rubber band riding on the water, and ask one question: as you travel once all the way around the band, how much does the current help you along? At each step you take the component of F that points along your direction of travel and add it up. That running total is the circulation of F around the loop, and it is nothing more than the line integral of F around a closed curve, written with a circle on the integral sign to mean 'once around'.
Recall from the conservative-fields guide that this same closed-loop integral was the litmus test for a conservative field: if F was a pure slope (the gradient of some potential), the circulation around every loop came out exactly zero — all the help on one arc was paid back on another. Circulation, then, measures the field's failure to be conservative — the leftover net push that a potential cannot account for. Writing F = (P, Q) with components P and Q, the circulation is the work integral of P dx + Q dy taken once around the loop. The whole of Green's theorem is the answer to a single follow-up question: where does that leftover push come from?
Green's theorem: the boundary equals the inside
Here is the statement, and it is one of the most beautiful equalities in all of calculus. [[greens-theorem|Green's theorem]] says that for a region D in the plane bounded by a simple closed curve C, the circulation of F = (P, Q) around C equals the double integral over D of a single quantity, dQ/dx − dP/dy: the loop integral of (P dx + Q dy) around C equals the area integral of (dQ/dx − dP/dy) over D. The left side lives only on the one-dimensional edge; the right side lives on the two-dimensional inside. The theorem trades a walk around the rim for a sum over everything enclosed.
That mysterious integrand dQ/dx − dP/dy is not mysterious at all once you recognize it: it is exactly the curl of a two-dimensional field — the curl from guide one, with only the out-of-the-plane component surviving in flat 2D. So Green's theorem is really saying, in plain words: the total circulation around the boundary equals the total amount of local spin inside. Every little paddlewheel turning in the interior contributes its bit of swirl, and Green's theorem promises that when you add up all those microscopic spins over the whole region, the sum is precisely what you feel as net push when you stroll once around the edge.
Why should that be true? The honest one-line reason is cancellation, and it is worth carrying in your head. Chop the region D into a fine grid of tiny cells. Apply the loop integral to each cell separately and add them all up. Wherever two cells share an interior edge, you walk that edge once forward (for one cell) and once backward (for its neighbour); the two contributions are equal and opposite, so they cancel exactly. Every interior edge is shared, so every interior contribution cancels — and the only edges left uncancelled are the ones on the true outer boundary C. The sum of all the tiny loops collapses to the one big loop, and the sum of all the tiny insides is the double integral. That mass cancellation is the engine behind all three great theorems of this rung.
Orientation: which way is 'positive'?
Green's theorem is only true with a sign convention attached, and getting it wrong flips the answer's sign — so this is not pedantry, it is part of the theorem. The boundary C must be traversed in the positive sense: counterclockwise for the outer edge, so that the enclosed region stays on your left as you walk. If your region has a hole in it (a washer, an annulus), the inner boundary must be traversed clockwise, again keeping the material on your left. This is the same orientation bookkeeping you met when setting up flux integrals and oriented surfaces in the previous guide; the theorems of this rung simply will not balance unless the boundary's direction and the inside agree.
This sign convention is also what makes Green's theorem a clever area meter. Choose the field F = (−y/2, x/2); then dQ/dx − dP/dy = 1/2 + 1/2 = 1, so the double integral on the right is just the area of D. Green's theorem then says the area equals a single loop integral around the boundary — you can measure the area of any region by walking once around its rim, never stepping inside. That is exactly the principle of the planimeter, the little wheeled instrument draughtsmen once rolled around a map's coastline to read off the enclosed area. The boundary really does remember everything inside.
Stokes' theorem: lifting Green into space
Now take the loop off the flat tabletop and bend it into three dimensions — make it the wire rim of a soap film, a curve C in space that bounds some curved surface S draped across it. Everything that worked in the plane should still work, only now the 'inside' is a two-dimensional surface floating in 3D, and the 'spin' has all three components of the full curl, not just the one that survived in 2D. That generalization is [[stokes-theorem|Stokes' theorem]]: the circulation of F around the boundary curve C equals the surface integral over S of the curl of F dotted with the surface's normal — the flux of nabla cross F through S.
Read the words slowly: circulation around the boundary equals curl-flux through the inside. The local spin of guide one — the curl nabla cross F, an arrow at each point — is here measured not by how much it spreads but by how much of it pierces the surface, exactly the flux idea from the surface-integral guide. Stokes' theorem says the boundary loop integral collects, in one number, the total curl poking through the membrane stretched across it. Green's theorem is just the special case where the surface S happens to be a flat patch of the xy-plane, the normal points straight up, and only the dQ/dx − dP/dy component of the curl survives. Same theorem, one dimension richer.
The most startling consequence is one Stokes' theorem hands you for free: the surface does not matter, only its rim. The left side of the equation mentions only the boundary curve C; the right side never had to commit to which surface S you stretched across it. So any two surfaces sharing the same boundary loop must give the same curl-flux — bulge the soap film into a dome or pinch it into a saddle, the flux through it is identical as long as the wire rim is unchanged. (This works because the divergence of a curl is always zero, one of the vector identities from guide one: nothing leaks out the sides of the closed bubble that two such surfaces form together.) The boundary alone fixes the answer.
Why curl is the swirl per unit area
Stokes' theorem also finally explains what curl really means — it gives the local quantity its honest definition. Shrink the loop down to a tiny circle of area A around a single point. Stokes' theorem says the circulation around that little loop equals the curl-flux through the little disk, which for a vanishingly small patch is approximately (component of curl along the normal) times A. Divide by A and let the loop shrink to nothing, and you get: the normal component of the curl at a point equals the circulation per unit area of the smallest loops around that point. Curl is not just a formula of partial derivatives; it is circulation density.
GREEN -> STOKES (boundary integral = inside integral, via curl)
Green (flat plane, region D, boundary curve C, ccw):
loop_C ( P dx + Q dy ) = area_D ( dQ/dx - dP/dy ) dA
\__ circulation around C __/ \__ curl-flux through D __/
Stokes (curved surface S in 3D, boundary curve C):
loop_C ( F . dr ) = surf_S ( nabla x F ) . n dS
\_ circulation _/ \__ flux of the curl through S __/
Green = Stokes with S a flat patch of the xy-plane, n = +z,
so only the z-component (dQ/dx - dP/dy) of the curl survives.
Local meaning (shrink the loop, area A -> 0):
(nabla x F) . n = lim ( circulation around loop / A )
i.e. curl = circulation per unit area.These theorems are not bookkeeping curiosities — they are the literal grammar of physics, just as the rung's title promised. Stokes' theorem in the field F equal to the magnetic field is Ampere's law: the magnetic circulation around a wire equals the current threading through any surface the loop bounds. In Faraday's law it ties a changing magnetic flux to the voltage circulating around the loop that encloses it. Stripped of the field names, both are the same sentence: what circulates on the edge is fed by what swirls through the inside. The same equality underlies vortices in fluids, lift on a wing, and the conservation laws that make those subjects work.
One pattern, and what comes next
Step back and notice that you have now seen the same shape three times. The gradient theorem from guide two said: the change of a potential between two endpoints equals the integral of its gradient along a path between them — the boundary was the two endpoints. Green's and Stokes' theorems say: the circulation along a boundary curve equals the integral of the curl over the surface it bounds — the boundary is now a loop. In every case, an integral of a derivative over a region equals an integral of the original thing over the region's boundary. The boundary keeps showing up because the boundary is where the cancellation stops.
- Spot the structure: pick the field F = (P, Q) (or (P, Q, R) in space) and the closed boundary curve C, then identify the region D (flat) or surface S (curved) that C bounds.
- Fix the orientation: walk C so the region stays on your left (counterclockwise outer rim), and choose the surface normal n by the right-hand rule to match.
- Trade hard for easy: replace the loop integral by the inside integral (Green) or the curl-flux through S (Stokes) — or go the other way if the boundary walk is the simpler computation.
- Check for singularities: if F blows up inside the region, excise a small loop around the bad point and apply the theorem to the punctured region instead.
The next guide completes the trio with the divergence theorem, which trades a flux through a closed surface for the divergence summed inside — the spreading counterpart to this circulation story. And further along this volume, the differential-forms guides will reveal that the gradient theorem, Green, Stokes, and the divergence theorem are not four cousins but a single statement in disguise, the generalized Stokes' theorem: the integral of d-omega over a region equals the integral of omega over its boundary. Everything you met here is one line of that one equation, read in three dimensions.