A NOTE ON GREEN'S THEOREM

B. D. CRAVEN
(received 31 October 1963)

Green's theorem, for line integrals in the plane, is well known, but
proofs of it are often complicated. Verblunsky [1] and Potts [2] have
given elegant proofs, which depend on a lemma on the decomposition of
the interior of a closed rectifiable Jordan curve into a finite collection of
subregions of arbitrarily small diameter. The following proof, for the
case of Riemann integration, avoids this requirement by making a con-
struction closely analogous to Goursat's proof of Cauchy's theorem. The
integrability of Qx— Pv is assumed, where P(x, y) and Q(x, y) are the
functions involved, but not the integrability of the individual partial
derivatives Qx and P,; this latter assumption being made by other authors.
However, P and Q are assumed differentiable, at points interior to the curve.

THEOREM. Let C be a closed rectifiable Jordan curve, enclosing a plane

region R. Let the functions P(x, y) and Q(x, y) be differentiable at all points
of R, and continuous on W — C+R. Let Qx—Pv be Riemann-integrable
on R. Then
(1) J p (Pdx+Qdy) = jjR (Qm-Pv)dxdy.
PROOF. Let C have finite positive length L. Then there is a square A
of area L*, with sides parallel to the axes, which contains W.
Choose any positive s. In what follows, a neighbourhood of a point
shall denote a square neighbourhood, with the point as its centre, and
with sides parallel to the axes. Then, from the hypotheses, every point
(x0, y0) of R has a neighbourhood N(x0, y0) such that, for every point
(x, y) which lies in both N and W,
(2) \P(x,y)-P(xo,yo)\<lelL
(3) \Q(x,y)-Q(xo,yo)\<hlL
(4) P(x, y) = p*+P*(x-Xo)+P*(y-yo)+S
(5) <?(*, V) = Q*+Q:(x-xo)+Q*{y-yo)+t,
where P*. P*, P*, Q*. Q*. Q* denote the values of P, Px, Pv, Q, Qm,
Qt at (ar0, y0), a n d
289

https://doi.org/10.1017/S1446788700024058 Published online by Cambridge University Press

 290 B. D. Craven [2]

(6) |f| < srjL*

(7) M < erIL*
where

Divide A into four squares of side \L, by lines parallel to the axes.
Repeat this procedure for each of the four squares, and so on indefinitely.
Denote by F the family of closed squares, with sides tending to zero, so
obtained. Denote by F{6) the subset of F consisting of squares each of
side <S, for d = \L, \L, • • •. In this notation, Lemma 2 of Potts [2] states
that the number of squares of F(d) necessary to cover C is less than 4(L/d)+4.
This follows, since an arc of C of length less than d can have points in
common with at most four such squares.
There exists, for some <5, a finite collection Ft of squares At of F,
disjoint except for common boundaries, such that every point of W lies
in some At, and such that if At lies wholly interior to R, and {x0, y0) is
its centre point, then every point (x, y) of At satisfies (2) to (7), whereas
if At contains points of C, then At belongs to F(d), and (2) and (3) hold
for any two points (x0, y0) and (x, y) which lie in both A( and W. For if
not, some region of W requires an infinite collection of squares. Then
successive subdivision of this region produces a nested sequence of squares,
to each of which the same statement applies. Since W is compact, the
nested sequence defines a limit point in W, at which P or Q is discontinuous
or not differentiable, contrary to hypothesis.
Let i 7 ' denote any finite collection of squares of F, obtained by further
subdividing the squares of Fx. The relation between F' and Fl will be
written F' < Ft. Then if Ft has the property stated in the previous
paragraph, the same statement applies to any F' < F j .
Denote by R* the union of those squares of Ft which contain points
of W. Let
*(*. V) = <?*(*> y ) - P , ( * . V) *or (*, y) in W
= 0 elsewhere.
Then

Since by hypothesis, this Riemann integral exists, there exists, for some
d < e/(jL+l), a finite collection F2 of squares of F(8), with F 2 < F1,
such that

(8) <B

https://doi.org/10.1017/S1446788700024058 Published online by Cambridge University Press

 [3] A note on Green's theorem 291

where Q* and P* refer to the centre point of At, and \Af\ is the area of A{.
The summation includes all squares A't of F2 which lie wholly within R,
and some (possibly all or none) of those squares A" which include points
of C. Let B denote an upper bound to 4>, implied by its Riemann-integrability.
Then

(9)
By Pott's Lemma,
(10) 2 \A"\ ^ «52[4(I/<5)+4] < 4e
supposing e < 1. Therefore

(11) \jjR (Qx-Py)dxdy-2 ifX~Pi)\A"\\

Denote by /><
' the boundary of A'(, and by p't' the boundary of that part
of A" which lies in W. Then if \p'{'\ denotes the length of p",

(12) 2 \p','\
again applying Potts' Lemma. Now

(13) J c (Pdx+Qdy) = 2 jp, (Pdx+Qdy) + 2 £„ (Pdx+Qdy)

with all paths traversed in the positive direction. From (4) and (5),

(14) J , (Pdx+Qdy) = ( t f - i * ) M,| + / , (£dx+r,dy),

since for a square,

- f ,y*r = M and \ xdy = \A\\.

By (6) and (7),

(15) 2 I f , iSdx+vdif) ^ 2 («/^2) * 4V2 \A',\ < 6e.

And from (2) and (3), and (12),

(16) 2 I L (Pdx+Qdy) < (ejL) • 17(L + 1) = B'e, say.

\Jpt

Combining (11), (13), (14), (15), and (16),

\jjR(Qx-Pv)dxdy- jc(Pdx+Qdy)\
Since e is arbitrary, (1) is proved.

https://doi.org/10.1017/S1446788700024058 Published online by Cambridge University Press

 292 B. D. Craven [4]

Acknowledgement
Thanks are due to Professor E. R. Love, whose suggestion it was that
Goursat's method might be applicable to Green's Theorem, and to the
referees for several suggestions.

References
[1] Verblunsky, J., On Green's formula, J. Lond. Math. Soc., 24 (1949), 146-148.
[2] Potts, D. H., A note on Green's Theorem, J. Lond. Math. Soc., 26 (1981), 302-304.

Department of Mathematics,
University of Melbourne.

https://doi.org/10.1017/S1446788700024058 Published online by Cambridge University Press