SECTION 15.

2 ITERATED INTEGRALS |||| 961

(b) H ere we first integrate with respect to x :

x�3
2 3 2 3 2 x3
x 2 y dx dy � y dy
1 0 1 0
yy x 2 y dx dy � y �y � 1 3
� � x�0
y
2
2 y2 27
� M
1
� y 9y dy � 9
2 2
�
1

N otice that in E xample 1 we obtained the same answer whether we integrated with
respect to y or x first. In general, it turns out (see T heorem 4) that the two iterated integrals
in E quations 2 and 3 are always equal; that is, the order of integration does not matter.
(T his is similar to C lairaut’s T heorem on the equality of the mixed partial derivatives.)
T he following theorem gives a practical method for evaluating a double integral by
expressing it as an iterated integral (in either order).

N Theorem 4 is named after the Italian mathe- 4 FUBINI’S THEOREM If f is continuous on the rectangle
matician Guido Fubini (1879–1943), who proved R � ��x, y� � a  x  b, c  y  d �, then
a very general version of this theorem in 1907.
But the version for continuous functions was b d d b
known to the French mathematician Augustin- f �x, y� dx dy
a c c a
yy f �x, y� dA � y y f �x, y� dy dx � y y
Louis Cauchy almost a century earlier. R

M ore generally, this is true if we assume that f is bounded on R , f is discontin-

uous only on a finite number of smooth curves, and the iterated integrals exist.
z
T he proof of Fubini’s T heorem is too difficult to include in this book, but we can at least
give an intuitive indication of why it is true for the case where f �x, y�  0. R ecall that if
f is positive, then we can interpret the double integral xxR f �x, y� dA as the volume V of
0 A(x)
the solid S that lies above R and under the surface z � f �x, y�. B ut we have another for-
a mula that we used for volume in C hapter 6, namely,
x y
b b
x
a
V � y A �x� dx

FIGURE 1
where A �x� is the area of a cross-section of S in the plane through x perpendicular to the
TEC Visual 15.2 illustrates Fubini’s x-axis. From Figure 1 you can see that A �x� is the area under the curve C whose equation
Theorem by showing an animation of is z � f �x, y�, where x is held constant and c  y  d . T herefore
Figures 1 and 2.
d
z c
A �x� � y f �x, y� dy

and we have
b b d
f �x, y� dy dx
a a c
yy f �x, y� dA � V � y A �x� dx � y y
R
0 c y d
y A similar argument, using cross-sections perpendicular to the y-axis as in Figure 2, shows
x
that
d b
f �x, y� dx dy
c a
yy f �x, y� dA � y y
FIGURE 2 R