MT2001

Integration

David G. Dritschel

November 27, 2007

2

Course Content
• Aim to improve integration techniques and learn how to compute double (area) and triple (volume)
integrals.
• Indefinite integrals and techniques of integration.
• Definite integrals.
• Double integrals over rectangular base.
• Double integrals over base bounded by lines and curves.
• Double integrals over base bounded by circles or parts of circles.
• Triple integrals over rectangular volumes.
• Triple integrals over volumes bounded by oblique planes and simple surfaces.
• Triple integrals over volumes with circular cross sections.
• Triple integrals over volumes which are portions of a sphere.
Chapter 0

Essentials

Some essential facts from earlier modules that you really should know. They are going to be useful in
this part of the course.

0.1 Standard Integrals

You should know the standard integrals listed below. If you can’t write the answers in immediately, go
and learn them. They are essential! In the following integrals a, b and n are constants.

1. eax dx =
R

R
2. sin axdx =
R
3. cos axdx =
R n
4. n 6= −1, (ax + b) dx =
R 1
5. ax+b dx =
R
6. tan(ax)dx =

7. sec2 axdx =
R

R 1
8. a2 +x 2 dx =

9. √a21−x2 dx =
R

0.2 Definition of an Integral

There are many ways to think of an integral. In the mathematical sense, it is defined in terms of a limit
as Z b Xn
I= f (x)dx = lim f (xi )δx,
a n→∞
i=1

3
4 CHAPTER 0. ESSENTIALS

where  
b−a 1
δx = , and xi = a + i − δx.
n 2
An alternative, but equivalent definition is
Z b n
X
I= f (x)dx = lim f (xi )δx.
a δx→0
i=1

This definition allows us to determine the limits of double and triple integrals by adding up the contri-
butions f (xi )dx to the integral. Visualising the integral as a summation can help determine the limits
of the integration, as we will see later.

0.3 Polar Coordinates

We are used to working with Cartesian coordinates but there are many situations where it is simpler to
use (and indeed it is the only way to proceed) where we need to use other coordinate systems. You have
already seen polar coordinates for two-dimensional space, where the point described by the Cartesian
coordinates (x, y) can be represented by the polar coordinates (R, φ). The relationship between Cartesian
and polar coordinates (shown in Figure 0.3 is obtained by simple trigonometry and gives

x = R cos φ, (1)
y = R sin φ, (2)
p
R = x2 + y 2 , (3)
y
φ = tan−1 , (4)
x
If you understand these, you will understand cylindrical coordinates later on. In addition, spherical
coordinates will also appear easier. Note that the notation used here is not that standard and is used
specifically to ease use of other coordinates later. Normally, (r, θ) are used for polar coordinates. It is
a fact of mathematical life that you need to use different notation in different situations.

0.4 Partial Differentiation

The ideas of partial differentiation (described in the functions of two variables part of the course) are
extremely useful. Given the definitions in (1) and (2), we have the following partial derivatives.

∂x
= cos φ
∂R
∂x
= −R sin φ
∂φ
∂y
= sin φ
∂R
∂y
= R cos φ
∂φ
0.5. CURVE SKETCHING 5

Figure 1: The relationship between Cartesian and polar coordinates.

A quantity used in coordinate transformations is the Jacobian and this is defined, in this example, as
∂x ∂y ∂x ∂y
− = R cos2 φ + R sin2 φ = R
∂R ∂φ ∂φ ∂R

0.5 Curve Sketching

Remember all the ideas involved in curve sketching, e.g. locate turning points, determine how a function
behaves for x = 0 and x large in negative and positive sense etc. We will need to sketch functions of
two and three variables but a key feature is the ability to draw standard functions of one variable.
Can you sketch, for example, the following functions?
• ex

• log x
• 1/(1 + x2 )
6 CHAPTER 0. ESSENTIALS
Chapter 1

Simple Integrals

1.1 Indefinite Integrals

R
The integral of a function f (x) is denoted by f (x)dx. It is called the indefinite integral because the
limits are not specified.
What does it mean? What does it tells us?
Some people like to think of it as the area under a curve y = f (x) but this is not always helpful.
For example, f (x) could be negative! We will come back to this later. An indefinite integral is itself a
function of x, denoted by F (x), where F satisfies

dF
= f (x).
dx

So, the indefinite integral is the function F (x) whose derivative is the integrand, f (x). This is why F (x)
is sometimes called the anti-derivative. Note that if you can find a function F (x) satisfying dF dx = f (x),
then F (x) + C, where C is any constant, also satisfies d(F + C)/dx = f (x) because dC/dx = 0. F is,
therefore, only known to within an arbitrary constant.
Sometimes, in say some physical application, we are given additional information that allows us to
choose the constant C uniquely.
Why is integration an art? F (x) is not known for every f (x) because some integrals are just too
tough. Not in this course, however! On the other hand, you can take the derivative of every (reasonable)
function F (x) to get f (x). For these f (x), F (x) is obviously known.
There is no systematic procedure to integrate analytically (we can always do definite integrals nu-
merically to whatever accuracy we desire), but there are a finite number of methods to try when faced
with a new integral.

1.2 Techniques of Integration

Here are a few standard techniques of integration that you have met either at school or in MT1001.

7
8 CHAPTER 1. SIMPLE INTEGRALS

1.2.1 Integration by parts

We have already seen this in the Fourier Series section and so this will be brief. We express

dv(x)
f (x) = u(x) ,
dx
and use
dv du
Z Z
u dx = uv − vdx. (1.1)
dx dx
By suitable choices of u and v (usually based on experience that comes from practice), the integral on
the r.h.s. can be made into a simpler integral. This is particularly effective when u is an integer power of
x and dv/dx is either a trigonometric function or an exponential (including hyperbolic) function. Many
examples were worked out in the Fourier Series part of the course.

1.2.2 Partial Fractions

This is useful when f (x) is the ratio of two polynomials, for example,
a1 x + a0
f (x) = ,
b2 x2+ b1 x + b0
and when the degree of the numerator is less than the degree of the denominator, namely 1 and 2
respectively in this example. The idea is to factorise the denominator in linear factors. This can always
be done provided we allow for complex factors. Then, we rewrite f (x) as a sum of fractions where the
denominators only contain polynomials of degree 1.
Let’s illustrate these ideas with a couple of examples.
Example 1.1
Integrate
1
f (x) = .
(x − 1)(x + 1)

Solution 1.1
Here we split the expression into the two fractions
1 A B
= + ,
(x − 1)(x + 1) x−1 x+1

where we need to find the values for A and B. The cover-up rule can be used but I prefer to use the
fail-safe method of recombining the fractions with a common denominator (basically the product of all
the linear, and possible quadratic, factors). Thus, we express

A B A(x + 1) + B(x − 1)
+ = .
x−1 x+1 (x − 1)(x + 1)

Comparing with the original function, we must have

A(x + 1) + B(x − 1) = 1, for all x.

1.2. TECHNIQUES OF INTEGRATION 9

Since this is true for all x, it must be true for x = 1 and x = −1. Choosing x = 1 we have
1
2A = 1, ⇒ A= .
2
Choosing x = −1, we have
1
−2B = 1, ⇒ B=− .
2
Thus, we have

1 1 1 1 1 1 1 1 1
Z Z Z Z
dx = − dx = dx − dx.
(x − 1)(x + 1) 2x−1 2x+1 2 x−1 2 x+1

Finally, the integrals give logarithms and so

 1/2
1 1 1 1 |x − 1| |x − 1|
Z
dx = log | x − 1 | − log |x + 1| + C = log + C = log +C
(x − 1)(x + 1) 2 2 2 |x + 1| |x + 1|

Example End

Example 1.2
Integrate
1
f (x) = .
x2 − 3x + 2

Solution 1.2
First of all, we need to factorise the denominator. Thus,

x2 − 3x + 2 = (x − 1)(x − 2).
2
If this
√is not obvious, then you obtain the roots of the quadratic x − 3x + 2 = 0. Hence, the roots are
(3 ± 9 − 8)/2 = 2 or 1. Now we use partial fractions

1 A B A(x − 2) + B(x − 1)
= + = ⇒ A(x − 2) + B(x − 1) = 1.
x2 − 3x + 2 x−1 x−2 (x − 1)(x − 2)

As this must be true for all x, we set x = 1 to get A = −1 and x = 2 to get B = 1. Thus,
 
1 1 1 |x−2|
Z Z
dx = − + dx = − log | x − 1 | + log | x + 2 | +C = log + C.
x2 − 3x + 2 x−1 x−2 |x−1|

Example End

Probably, it is worth reminding you that the integral of 1/x is log | x |. Obviously we can drop the
modulus signs when x is positive.
10 CHAPTER 1. SIMPLE INTEGRALS

1.2.3 Substitution
This is one of the most effective methods and will be used later on in this course. The general idea is to
make a change of variable or substitution of the form s = g(x) (or maybe x = g(s)), for some cleverly
chosen function g, which reduces Z Z
f (x)dx to h(s)ds,

where h(s) is simpler to integrate. How do you know what function to choose for the substitution? Well
this really comes down to knowing the standard integrals. If the integral is not a standard integral, then
you need to choose a substitution to put it into a standard form. We illustrate the idea through a few
examples.
Example 1.3
Integrate f (x) = x/(x2 − 1) with respect to x. Note, I am going to emphasise which variable we are
integrating with respect to now. This is important when using substitutions. Thus, we wish to evaluate
x
Z
I= 2
dx.
x −1

Solution 1.3
We could do this by partial fractions as above but instead we try the substitution
ds
s = x2 − 1, ⇒ = 2x, ⇒ ds = 2xdx.
dx
Thus, we replace x2 − 1 by s and dx by ds/2 to obtain

x ds/2 1 1 1
Z Z Z
dx = = ds = log | s | +C.
x2 − 1 s 2 s 2

Finally, we replace s by x2 − 1 to get the answer in terms of x. Hence,

x 1
Z
2
dx = log | x2 − 1 | +C.
x −1 2
Note that when we did the substitution, all x’s were replaced by s’s. Thus the integration was done
with respect to s. This is important. You cannot substitute only some of the integrand in terms of s
and leave some in terms of x. That just does not work.
Example End

Example 1.4
Integrate
1
Z
dx.
x2 + 1
This looks like the last example but the simple substitution s = x2 + 1 does not help here. Instead
we need some inspiration, or better recognise that this form can always be done using a trigonometric
substitution.
1.2. TECHNIQUES OF INTEGRATION 11

Solution 1.4
First of all, we remind you of a few trigonometric identities.

sin2 φ + cos2 φ = 1,
divide both sides by cos2 φ
tan2 φ + 1 = sec2 φ.

Remember that
sin φ 1
= tan φ and sec φ = .
cos φ cos φ
Finally, the last reminder before completing this example involves the derivatives of trigonometric func-
tions. These should be memorised!
d
sin φ = cos φ,
dφ
d
cos φ = − sin φ,
dφ
d
tan φ = sec2 φ,
dφ
d
sec φ = sec φ tan φ.
dφ
Now we are ready to progress! Set
dx
x = tan φ, ⇒ = sec2 φ ⇒ dx = sec2 φdφ.
dφ

Using tan2 φ + 1 = sec2 φ, we have x2 + 1 = sec2 φ. Therefore,

1 1
= = cos2 φ.
x2 +1 sec2 φ
Thus,
1
Z Z Z
2 2
dx = cos φ(sec φdφ) = dφ = φ + C.
x2 + 1
Hence, substituting back for x (using the fact that if x = tan φ, then φ = tan−1 x where the power
minus one means INVERSE FUNCTION and not reciprocal) we have

1
Z
dx = tan−1 x + C.
x2 + 1
The natural extension of this gives the standard integral
1 1 −1 x
Z  
dx = tan + C.
x2 + a 2 a a

Example End
12 CHAPTER 1. SIMPLE INTEGRALS

Example 1.5
This example uses the same idea of trigonometric substitutions. Consider
1
Z
√ dx.
a − x2
2

Solution 1.5 √
For expressions involving purely a2 − x2 (and powers of this square root) always try

x = a sin φ.
√
As an aside, note that the integrand x/ a2 − x2 could use the simpler substitution u = a2 − x2 . Since,

sin2 φ + cos2 φ = 1,
a2 sin2 φ + a2 cos2 φ = a2 ,
x2 + a2 cos2 φ = a2 ,
p
a2 cos2 φ = a2 − x2 ⇒ a2 − x2 = a cos φ.

So the denominator simplifies to a cos φ. What about the dx? Thus,

dx
x = a sin φ ⇒ = a cos φ ⇒ dx = a cos φdφ.
dφ
Substituting these into the original integral gives,
1 1
Z Z Z
√ dx = a cos φdφ = dφ = φ + C.
a 2 − x2 a cos φ

Finally, substituting back in terms of x gives

1 x
Z
√ dx = sin−1 + C.
a 2 − x2 a

Example End

Example 1.6
One last common trigonometric substitution involves integrals of the form
1
Z
√ dx.
x − a2
2

Note that we cannot use x = a sin φ because sin φ is less than or equal to one and so x is less than or
equal to a. Therefore, x2 − a2 is less than or equal to zero. Hence, we cannot take the square root. The
assumption must be that x is greater or equal to a. We need another substitution.

Solution 1.6
1.2. TECHNIQUES OF INTEGRATION 13

Instead we make use of

tan2 φ + 1 = sec2 φ.
Rearranging this gives
sec2 φ − 1 = tan2 φ.
Multiplying by a2 and comparing with the denominator suggests trying

x = a sec φ ⇒ x2 − a2 = a2 sec2 φ − a2 = a2 tan2 φ.

So far so good. Now we need to replace dx by the correct expression involving only φ and dφ. Thus,
dx
= a sec φ tan φ
dφ
dx = a sec φ tan φdφ.

Thus we have converted the original integral into

1 1 dφ
Z Z Z Z
√ dx = a sec φ tan φdφ = sec φdφ = .
2
x −a 2 a tan φ cos φ
This integral may appear simpler but still requires a bit of effort to get a final answer. This illustrates
that you may need to do several substitutions to get to the actual result.
To proceed, and I suspect you would never guess this, consider the two derivatives discussed above,
d
tan φ = sec2 φ,
dφ
d
sec φ = sec φ tan φ.
dφ
Add these equation together to get
d
(tan φ + sec φ) = sec2 φ + sec φ tan φ,
dφ
= sec φ (tan φ + sec φ) .

Hence, we have the result

1 dv
= sec φ,
v dφ
where
v = tan φ + sec φ.
Therefore,
1
Z Z
sec φdφ = dv = log | v | +C = log | tan φ + sec φ | +C.
v
This is not a result I would expect you know.
Example End

Example 1.7
14 CHAPTER 1. SIMPLE INTEGRALS

Consider
x
Z
√ dx.
x − a2
2

Solution 1.7
It is the xdx that appears in the numerator that should warn you that a simple trigonometric substi-
tution will not work. The correct substitution to use is
du 1
u = x2 − a 2 ⇒ = 2x ⇒ du = xdx.
dx 2
Hence, the original integral can be re-written as
x 1 1 1
Z Z Z
√ dx = √ du = u−1/2 du.
x2 − a 2 2 u 2

Therefore,
x
Z p
√ dx = u1/2 + C = x2 − a2 + C.
x2 − a 2
Note that you can always check that you have done the integration correctly by differentiating and
retrieving the original integrand. Remember
dF
Z
F (x) = f (x)dx ⇐⇒ = f (x).
dx

Example End

1.3 Definite Integrals

These refer to integrals with limits, i.e.
Z b
f (x)dx. (1.2)
a

Remember that the answer is a number if a and b are numbers. It is NOT a function of x. However,
we can use the indefinite integral, which is a function of x, as an intermediate step. Let
dF
Z
F (x) = f (x)dx, or equivalently = f (x),
dx
Then Z b
f (x)dx = [F (x)]ba = F (b) − F (a). (1.3)
a
Note we could think of the integral as being a function of the end points. This will be useful when the
integrals are double integrals and the limits in, say x, involve functions of y. More on this later.

Example 1.8
1.4. LINEARITY 15

Consider Z π/4
I= tan xdx.
0

What is the value of the number I?

Solution 1.8
First step is to find the indefinite integral.

sin x
Z Z
F (x) = tan xdx = dx.
cos x

Use the substitution u = cos x, so that

du
u = cos x ⇒ = − sin x ⇒ du = − sin xdx.
dx
Hence,
sin x 1
Z Z
F = dx = − du = − log | u | +C.
cos x u
Finally, we can replace u by cos x in the answer. Remember that the constant C is NOT important
when evaluating definite integrals. Thus,
Z π/4
I= tan xdx = [− log | cos x |]π/4
0 = − log | cos(π/4) | + log | cos(0) | .
0

Hence,
√
   −1
1 1 1
I = − log √ + log 1 = log √ + 0 = log 2= log 2 ≈ 0.346.
2 2 2

Example End

1.4 Linearity
This section contains some obvious (and perhaps not so obvious) properties of integrals.
1.
Z Z Z
(f (x) + g(x)) dx = f (x)dx + g(x)dx.

2.
Z Z
cf (x)dx = c f (x)dx

where c is a constant. This means that Integration is a vector space.

16 CHAPTER 1. SIMPLE INTEGRALS

3.
Z b Z c Z b
f (x)dx = f (x)dx + f (x)dx.
a a c
Thus, you can always split the integration range into smaller pieces and still get the same result.
In a sense, this comes from the fact the integration is the limit of a sum. The formal definition of
an integral is
Z Xn
f (x)dx = lim f (xi )δx,
δx→0
i=1
where x0 = a and xn = b, which means that nδx = b − a. Thus, as δx → 0, n → ∞.
4.
Z b Z a
f (x)dx = − f (x)dx.
a b

Hence, switching the limits switches the sign in front of the integral (but the value of I remains
the same). From 3 above with b = a, we have
Z a Z c Z a Z c Z a
f (x)dx = 0 = f (x)dx + f (x)dx, ⇒ f (x)dx = − f (x)dx,
a a c a c
as stated above.
Chopping the range of integration up can be very useful. One particular instance is when f (x) → ∞
at a point c, with a < c < b, but the integral still exists. It just depends on how ‘singular’f is at c. The
integral exists if limx→c {(x − c)f (x)} = 0.
Example 1.9
An example is
5
dx
Z
I= p .
0 |x−1|
Note that we are using the modulus of x − 1 instead the square root so that | x − 1 | is always positive
and we can always take the square root. The integrand is shown in Figure 1.1.
Solution 1.9
Thus, we split the range of integration at x = 1.
Z 1 Z 5
1 1
I = √ dx + √ dx
0 1 − x 1 x −1

1 − x for x ≤ 1,
since | x − 1 |=
x − 1 for x ≥ 1.
√
The first integral has an indefinite integral, F (x) = −2 1 − x√and so its contribution to I is F (1) −
F (0) = 0 − (−2) √ = 2. For the second integral, F (x) = +2 x − 1 and so its contribution to I is
F (5) − F (1) = 2 4 − 0 = 4. Hence,
I = 2 + 4 = 6.

Example End
1.4. LINEARITY 17

p
Figure 1.1: A singular function, f (x) = 1/ | x − 1 |, but it is integrable.
18 CHAPTER 1. SIMPLE INTEGRALS
Chapter 2

Double Integrals

A double integral is Z Z
I= f (x, y)dA,
A
where A is the region or domain or area of integration and dA is the infinitesimal ‘area’ element. In
simple integrals of one variable we had dx as the infinitesimal length in the x direction. Now we have
an area. In Cartesian coordinates (x, y), we have
dA = dxdy. (2.1)
This is illustrated in Figure 2.1
We will consider other coordinates later.
Notes:
1. I is a number when the area A is fixed. It is not a function of x and y. Like the definite integrals,
we have ‘integrated out’ the dependence on x and y. However, as before, indefinite integrals are
often used in intermediate steps when evaluating I.
2. f can be a function of other coordinates, such as polar coordinates (R, φ), defined over the region.
3. When f (x, y) = 1, I is simply the area of the region. Only in this case is
Z Z
Area = dA.
A
Otherwise it is not the area.
4. When f (x, y) is the height of a surface above the x-y plane, I is the volume under the surface.
However, do not always think in these terms. f (x, y) can be many other things. For example
f (x, y) could be the temperature at a location (x, y). Then if A is the whole of the UK, we could
calculate the mean temperature by integrating the temperature over A and dividing by the area.
RR
A
f (x, y)dA
Mean temperature = RR .
A
dA

In addition, note that f (x, y) can be locally positive or negative. (It would always have to be
positive if we only thought in terms of volume under a surface!!)

19
20 CHAPTER 2. DOUBLE INTEGRALS

Figure 2.1: (a) The infinitesimal line element dx. (b) The infinitesimal area element dA = dxdy.

2.1 Rectangular Base Areas

A rectangular base will have a form similar to Figure 2.3. These are the easiest, as they can always be
done as a composition of two single integrals. In this case the limits of integration are simple, namely
!
Z d Z b Z d Z b
I= f (x, y)dxdy = f (x, y)dx dy. (2.2)
y=c x=a y=c x=a

Notice that the x integration is done first, where the y in f (x, y) is treated as a constant and the y
integration is done second. There is now no x appearing as it has already been ‘integrated out’.
Consider the inner integral
Z b
f (x, y)dx ≡ g(y).
x=a
Remembering that y is held constant at present. Then, if the indefinite integral of f (x, y) with respect
to x is F (x, y), then
Z b
g(y) = F (b, y) − F (a, y) = f (x, y)dx.
x=a
This is exactly the same as in single integrals. Since f depends on x and y, we use partial derivatives
to define F through
∂F
= f (x, y).
∂x
2.1. RECTANGULAR BASE AREAS 21

Figure 2.2: The height of the surface above the x-y plane is given by f (x, y).

The point is that, if we can find any function satisfying ∂F/∂x = f (x, y), then g(y) = F (b, y) − F (a, y)
is unique. The proof is that F is known only to within an arbitrary function of y, say h(y). Thus, if
∂F ∂(F + h(y))
= f (x, y), ⇒ = f (x, y)
∂x ∂x
as well. However, when we apply the limits, we have
g(y) = [F (b, y) + h(y)] − [F (a, y) + h(y)] = F (b, y) − F (a, y),
is independent of h(y).
Having completed this inner integral, we now know g(y). Then, our original integral has reduced to
Z d
g(y)dy,
c

and is now a definite integral over y. If G(y) is the indefinite integral, namely
dG
= g(y),
dy
then
I = G(d) − G(c) = [G(y)]dc .
This is the final answer.
Note that the order of the integration, namely x first and then y second or y first and x second, does
not matter. You get the same answer both ways. However, sometimes the integrals are simpler one way
than the other.
22 CHAPTER 2. DOUBLE INTEGRALS

Figure 2.3: A rectangular base, where x lies between a and b and where y lies between c and d.

Example 2.10 √
Find the integral of f (x, y) = xy over the rectangle 1 ≤ x ≤ 4 and 4 ≤ y ≤ 9. Thus, we must evaluate
9 4
√
Z Z
I= xydxdy.
y=4 x=1

Solution 2.10 √
Let us do the x integration first, as indicated above. Firstly, we integrate xy = x1/2 y 1/2 with respect
to x keeping y fixed. Thus,
Z 4  4    
1/2 1/2 2 3/2 1/2 2 3/2 1/2 2 3/2 1/2
x y dx = x y = 4 y − 1 y .
x=1 3 1 3 3
Hence, the first integral gives
4
2 1/2 14 1/2
Z
x1/2 y 1/2 dx = y (8 − 1) = y .
x=1 3 3
Now we can do the second integral over y, so that
Z 9 Z 4 9
14 9 1/2

√ 14 2 3/2
Z
I= xydxdy = y dy = y .
y=4 x=1 3 y=4 3 3 4

Hence,
28  3/2  28 541
I= 9 − 43/2 = (27 − 8) = .
9 9 9
2.1. RECTANGULAR BASE AREAS 23

In this example, reversing the order of integration is simple since f (x, y) is separable in x and y.
Hence, there is no advantage in changing the order of integration.
Example End

Example 2.11
In this example, the integrand is not separable.
y
Z Z
I= 2 2
dxdy, where A is the rectangle 0 ≤ x ≤ 1, 1 ≤ y ≤ 2.
A (x + y )

We will do this example twice. Firstly we do the x integration first and the y integration second.
Secondly, we will reverse the order of integration.
Solution 2.11
(a)
2 Z 1 
y
Z
I= dx dy.
y=1 x=0 (x + y 2 )2
2 2
We find F satisfying ∂F/∂x = y/(x + y ) . Thus,
y
F =− ,
(x + y 2 )
is a good choice as we can check by differentiating. Hence,
Z 1
y y 1
2 )2
dx = g(y) = F (1, y) − F (0, y) = − 2
+ .
x=0 (x + y 1 + y y
Next we find G satisfying dG/dy = g. Thus,
1
G = − log(1 + y 2 ) + log y,
2
will do, as we can again check by differentiating. If you cannot spot that G is the integral of g, then we
could use a substitution u = 1 + y 2 . Finally, we have
Z 2    2    
y 1 1 2 1 1
I= − + dy = − log(1 + y ) + log y = − log(5) + log 2 − − log 2 + log 1 .
y=1 1 + y2 y 2 1 2 2
Thus,  
1 3 1 1 1 8
I = − log 5 + log 2 = − log 5 + log 23 = log .
2 2 2 2 2 5
(b) Now we reverse the order, doing the y integration first and the x integration second.
Z 1 Z 2 
y
I= 2 2
dy dx.
x=0 y=1 (x + y )

Find F (x, y) satisfying ∂F/∂y = f (x, y).

1 1
F (x, y) = − ,
2 (x + y 2 )
24 CHAPTER 2. DOUBLE INTEGRALS

will do, as simple differentiation will verify. However, the substitution u = x + y 2 could also have been
used. Thus Z 2  2
y 1 1 1 1
2 2
dy = g(x) = − 2
=− + .
y=1 (x + y ) 2 x + y y=1 2(x + 4) 2(x + 1)
Next, we find G(x) such that dG/dx = g.

1 1
G = − log(x + 4) + log(x + 1)
2 2
is a good choice. Again we can verify this by differentiation. The final answer is
1
1 1
Z
I = − + dx
x=0 2(x + 4) 2(x + 1)
 1
1 1
= − log(x + 4) + log(x + 1)
2 2 0
1 1 1 1
= − log 5 + log 2 + log 4 − log 1
2   2 2 2
1 8
= log ,
2 5

as before.
Example End

2.2 Base Areas Bounded by Lines and Curves

RR
Now we move onto more interesting situations, when the base area, A, in A
f (x, y)dxdy is not rect-
angular but is formed from areas bounded by curves. Some careful thinking is required! In particular,
the order of the integration can be crucial.
The key steps every time (until you become expert) are

• Sketch the base area, A. If you can’t do this, you will struggle to get the correct limits of
integration.

• Decide on the order of integration from the form of f (x, y) and the shape of A.

• Work out the inner integral’s limits of integration. Warning: these limits will depend on the
outer variable. The outer integral’s limits are always constants. In a sense, this step is the crux of
the whole problem.

• Perform the integrations as usual in the correct order. This will only involve integrals and inte-
gration techniques you have seen before.

There is no automatic procedure: setting up the integrals and, in particular, the limits of integration
are the hardest parts. These depend on both the form of f (x, y) and the shape of the area, A.
We will learn by doing lots of examples.
2.2. BASE AREAS BOUNDED BY LINES AND CURVES 25

Example 2.12
Find Z Z Z Z
I= dA = dxdy,
A A
where A is the region in the first quadrant bounded by the line x + y = 1. So the area is contained by
the lines y = 1 − x, x = 0 and y = 0. This shown in Figure 2.4

Figure 2.4: The area contained by the x and y axes and the straight line y = 1 − x. The area of the
small square is dxdy and we add up the contributions from y = 0 up to y = 1 − x.

Solution 2.12
Here, we go through the steps outlined above.
• Sketch the area A. This is shown in Figure 2.4. It is clearly a triangle and so the area is simply
(half base times height) 1/2. However, we continue to illustrate the method and confirm that we
do indeed get the same answer.
• Decide on the order of integration. Here it is does not matter as f (x, y) = 1 in this case. I have
chosen to do the y integration first and the x integration second.
• In Figure 2.4, I have drawn a typical rectangle of area dxdy. I now add up all the contributions
from similar rectangles going from y = 0 up to the line where y = 1 − x. This gives me the limits
on the x integration (i.e the inner integration). Having got the area of this single strip, I now need
to add up all the area of each strip (as shown in Figure 2.5), going from x = 0 to x = 1. I should
have completely covered the base area only once. Thus,
Z 1 Z 1−x 
I= dy dx.
x=0 y=0
26 CHAPTER 2. DOUBLE INTEGRALS

• Now we actually do the integrals. Thus,

Z 1 Z 1
1−x
I= [y]y=0 dx = {(1 − x) − 0} dx.
x=0 x=0

Thus, we have
1 1
x2

1 1
Z
I= (1 − x) dx = x − =1− = .
x=0 2 0 2 2

Figure 2.5: Having calculated the area of each strip, we add them up from x = 0 to x = 1.

Finally, this example can be easily done in the other order. This time the limits of the integration
will give us, Z 1 Z 1−y 
I= dx dy.
y=0 x=0
If you cannot understand why the limits are as given, please see me as soon as possible.
Example End

Example 2.13
Evaluate,
1−y
Z Z
I= dA,
A 1−x
over the same area as above. This is a little more challenging. Note that f (x, y) 6= 1 but is f (x, y) =
(1 − y)/(1 − x). Hence, we are not finding the area this time.
Solution 2.13
Go through the same steps as before.
2.2. BASE AREAS BOUNDED BY LINES AND CURVES 27

• Well the area is exactly the same as shown in Figure 2.4.

• Decide on the order of integration. If we choose x first we will get logarithms. If we choose y first
we will get powers of y. The latter seems easier, so we will do the y integration first and the x
integration second.
• Decide on limits. As above we imagine drawing a small rectangle of area dA = dxdy and add all
these contribution up from y = 0 up to the line at y = 1 − x. Then we add up all the vertical
strips going from x = 0 to x = 1. Thus,
Z 1 Z 1−x 
1−y
dy dx.
x=0 y=0 1 − x

• Let’s do the y integration first, remembering that x is kept fixed.

1−x 1−x 
y − y 2 /2 1 − x − (1 − x)2 /2
 
1−y 1−x
Z
dy = = −0=1− .
y=0 1−x 1−x 0 1−x 2

Now we do the x integration

1 1
x2
  
1 x 1 1 1 3
Z
I= + dx = x+ = 1+ −0= .
x=0 2 2 2 2 0 2 2 4

Challenge: Can anyone do the integral by reversing the order of integration?

Example End

Example 2.14
Evaluate Z Z
2
I= e−y dA,
A
where A is the area bounded by the y axis, the line y = x and the line y = 1.
Solution 2.14
We follow the usual steps.
• Firstly, we draw the area A. This is shown in Figure 2.6.
• Decide on the order of integration. From your knowledge of standard integrals, you cannot inte-
2
grate e−y analytically (over a finite domain) no matter how imaginative you are. So the best bet
is to try the x integration first.
• Decide on the limits. From Figure 2.7, we can see that we must add up the contributions from the
infinitesimal areas from x = 0 to x = y. This gives us a strip. We then add up all the contributions
from all the strips from y = 0 to y = 1. This will completely cover the area once.
Z 1 Z y 
2
I= e−y dx dy.
y=0 x=0
28 CHAPTER 2. DOUBLE INTEGRALS

Figure 2.6: The area contained by the x and y axes and the straight line y = 1 − x.

• Do the integration. Well the first integral is straightforward, since we assume that y is fixed while
we integrate in the x direction.
Z y
2 y
2
h i 2
e−y dx = xe−y = ye−y .
x=0 0

That is very useful. The original integral has now reduced to

Z 1
2
I= ye−y dy.
y=0

This can be done either directly, if you can spot the indefinite integral, or by a substitution u = y 2 .
So du/dy = 2y and this means du/2 = ydy. Remembering to change the limits we have u = 0
when y = 0 and u = 1 when y = 1. Therefore,
1
1
Z
I = e−u du
2 u=0
1 1 1
= − e−u 0 = − e−1 − e0


2  2
1 1
= 1− .
2 e

Example End

Note that not all integrals are straightforward. Not all integrals can be done analytically. However,
even if the integrals need to be done numerically, the derivation of the limits always follows the same
pattern.
2.2. BASE AREAS BOUNDED BY LINES AND CURVES 29

Figure 2.7: The area contained by the x and y axes and the straight line y = 1 − x. The area of the
small square is dxdy and we add up the contributions from x = 0 up to x = 1 − y.

Example 2.15
Find the mean values of x and y over the area A bounded by the parabola y = y1 (x) = 4x(1 − x) and
the line y = y2 (x) = x.
Solution 2.15
First of all, we need to know what is meant by x̄, the mean of x. This is defined as
RR
xdxdy
x̄ = R RA . (2.3)
A
dxdy
Thus, we have two integrals to evaluate, namely the area of the base
Z Z
dxdy,
A

and Z Z
I= xdxdy.
A
Similarly, ȳ, the mean of y is defined by
RR
ydxdy
ȳ = R RA . (2.4)
A
dxdy
Again we need to evaluate two integrals but at least the area is the same. Thus, we need
Z Z
J= ydxdy.
A
30 CHAPTER 2. DOUBLE INTEGRALS

(x̄, ȳ) gives the centre of mass of a plate with uniform density.
Thus, there are three integrals to evaluate, I, J and the area, A. They are all straightforward once
they are correctly set up.

• Draw the area, A. This is shown in Figure 2.8. We see that the two curves intersect at two values

Figure 2.8: The area contained by the y = x and the parabola y = 4x(1 − x).

of x, namely x = 0 and x = b. These occur at the same values of y. Thus, setting y = x equal to
y = 4x(1 − x) we have

3
x = 4x(1 − x) ⇒ x = 0 or 1 = 4(1 − x) ⇒ x= .
4

Hence, b = 43 .

• Now we decide on the order of integration. Really there is only one straightforward choice, y first.

• Decide on the limits. Integrating in y first means we imagine adding up contributions from
infinitesimal boxes (of area dxdy) from y = x up to y = 4x(1 − x). This generates a strip. Then
we add up all the strips from x = 0 to x = 3/4 so that we cover the area exactly once.

• Do the integration. We have three integrals to evaluate. Let’s do them in turn.

1. Firstly, we evaluate the area. Thus,

!
Z 3/4 Z 4x(1−x)
A= dy dx.
x=0 y=x
2.2. BASE AREAS BOUNDED BY LINES AND CURVES 31

Doing the inner integral first, we get

Z 4x(1−x)
4x(1−x)
dy = [y]y=x = (4x(1 − x)) − x = 3x − 4x2 .
y=x

Now we can do the outer integral.

Z 3/4  3/4
3 2 4 3
3x − 4x2 dx =


A= x − x
x=0 2 3 0

Substituting in the limits we get

3 9 4 27 27 9 9
A= − = − = .
2 16 3 64 32 16 32
2. Next we evaluate, I. !
Z 3/4 Z 4x(1−x)
I= xdy dx.
x=0 y=x

The inner integral gives, since x is constant while integrating with respect to y,
Z 4x(1−x)
xdy = [xy]4x(1−x) = 4x2 (1 − x) − x2 = 3x2 − 4x3 .



y=x
y=x

Now we can do the outer integral.

Z 3/4
3/4 27 81
3x2 − 4x3 dx = x3 − x4 0 =


I= − .
0 64 256
This simplifies to
27
I= .
256
3. The final integral is J. Here
!
Z 3/4 Z 4x(1−x)
J= ydy dx.
x=0 y=x

The inner integral gives

Z 4x(1−x)  4x(1−x)
1 2 1 15 2
ydy = y = 8x2 (1 − x)2 − x2 = x − 16x3 + 8x4 .
y=x 2 y=x 2 2

Again, we are ready to evaluate the outer integral to get

Z 3/4    3/4
15 2 3 4 5 3 4 8 5
J= x − 16x + 8x dx = x − 4x + x .
x=0 2 2 5 0

Thus,
135 81 243 27
J= − + = .
128 64 640 160
32 CHAPTER 2. DOUBLE INTEGRALS

Now we can work out the mean of x and y as

27 32 3 27 32 3
x̄ = = , ȳ = = .
256 9 8 160 9 5
Thus, the centre of mass of plate with this shape is at ( 83 , 35 ).
Interestingly, if I was to throw a plate with this shape, it would travel in exactly the same manner as a
point mass located at the centre of mass (ignoring air resistance)!
Example End

2.3 Base areas of circular or part circular form

When the base area, A, is a circle or a sector of a circle, it is usually more effective to change coordinates
from Cartesian coordinates to polar coordinates when evaluating double integrals. This is particularly
useful when the integrand is a function of x2 + y 2 .
The first example will illustrate the basic ideas and importantly how to calculate the infinitesimal
area dA.
Example 2.16
Calculate Z Z p
I= x2 + y 2 dA,
A
where A is the sector of a circular disk of radius R = a between the angles π/6(= 30◦ ) and π/3(= 60◦ ).
Remember we must work in radians!
Solution 2.16
We follow the usual approach.
• Draw the area A. This is shown in Figure 2.9. This would be extremely complicated in Cartesian
coordinates. So we switch from (x, y) to (R, φ), where
x = R cos φ, (2.5)
y = R sin φ. (2.6)
Equations (2.5) and (2.6) give x and y in terms of R and φ. The inverse relationships are
p
R = x2 + y 2 , (2.7)
y
φ = tan−1 . (2.8)
x
We are nearly ready to tackle the integral I but we need to get an expression for dA. There are
several ways to obtaining dA and I will illustrate two.
Firstly, Figure 2.10 shows the area dA. The difference in radii is dR and the angle subtended is dφ.
The area is simply (‘length times height’). Now the distance around the circle is the appropriate
fraction of the circumference, which at a general radius, R is
dφ
× 2πR = Rdφ.
2π
2.3. BASE AREAS OF CIRCULAR OR PART CIRCULAR FORM 33

Figure 2.9: The area contained by the circle R = a and the lines y = x tan(π/6) and y = x tan(π/3).

Thus, the infinitesimal area is

dA = RdRdφ. (2.9)
Remember this!
The second method involves using the differences between the area of a sector of a circle of radius,
R + dR and one of radius R. The area of a sector is the appropriate fraction of the area of a circle.
Thus, the area of the larger sector is
dφ 1
A2 = π(R + dR)2 = dφ(R + dR)2 ,
2π 2
and the area of the smaller sector is
dφ 2 1
A1 = πR = dφR2 .
2π 2
Thus, the required area is
1  2
dφ R + 2RdR + dR2 − R2 .

dA = A2 − A1 =
2
Cancelling the terms in R2 and neglecting dR2 in comparison of the size of dR, we find that the
infinitesimal area in polar coordinates is again

dA = RdRdφ.

A third method, using the Jacobian, works for any change of 2D coordinates and is presented in
Section 2.4.
34 CHAPTER 2. DOUBLE INTEGRALS

Figure 2.10: The infinitesimal area, dA has sides of length dR and Rdφ.

Now return to our example.

Z Z Z Z Z Z
I= RdA = R(RdRdφ) = R2 dRdφ,
p
where we have used R = x2 + y 2 .
• Decide order of integration. Here it does not matter.
• Decide on limits. Again it does matter how we determine the limits. Assume we have our
infinitesimal area, dA, we add up all the contributions round the sector of the circle at the same
radius. Thus, we add up from φ = π/6 to φ = π/3. This creates a small curved strip located at a
certain radius. Now we add up all the strips from R = 0 to R = a. So,
Z a Z π/3 !
2
I= R dφ dR.
R=0 φ=π/6

• The inner integral is straightforward (since R is held fixed while integrating in the φ direction).
Thus, the inner integral gives,
Z π/3 π π  π
π/3
R2 dφ = R2 [φ]φ=π/6 = R2 − = R2 .
φ=π/6 3 6 6
Now we evaluate the outer integral to get
Z a a
π a3 πa3

π 2 π 1 3
I= R dR = R = = .
R=0 6 6 3 R=0 6 3 18
2.3. BASE AREAS OF CIRCULAR OR PART CIRCULAR FORM 35

Example End

Example 2.17
Obtain Z Z
I= x2 dA,
A
where A is the circular disk given by R ≤ a.

Solution 2.17
Use polar coordinates! It is easier. Remember,

x = R cos φ
y = R sin φ
2
R = x + y2
2
y
φ = tan−1
x
dA = RdRdφ.

So, Z Z Z Z
I= R2 cos2 φRdRdφ = R3 cos2 φdRdφ.
A A
The order does not matter and the limits are fairly obvious. We add up our infinitesimal area round
the circle so that φ goes from 0 to 2π. Then we add up all the strips in the radial direction from R = 0
to R = a. Thus, Z Z a 2π 
I= R3 cos2 φdφ dR.
R=0 φ=0

The inner integral had been done before in the Fourier Series section. Remember to express cos2 φ in
terms of multiple angles as cos2 φ = (1 + cos 2φ)/2. Thus, the inner integral gives,
2π 2π
R3

1 1
Z
R3 (1 + cos 2φ) dφ = φ + sin 2φ = πR3 .
φ=0 2 2 2 φ=0

Now the outer integral can be done,

a a
R4 πa4
Z 
3
I= πR dR = π = .
R=0 4 0 4

Example End

Example 2.18
Sometimes the order of integration does matter in polar coordinates. Consider finding the area
Z Z
I= dA
A
36 CHAPTER 2. DOUBLE INTEGRALS

where A is the area between the two spirals

R = R1 (φ) = aφ1/2 , R = R2 (φ) = bφ1/2 ,

with
0 < a < b, and 0 ≤ φ ≤ 2π.

Solution 2.18
We follow the usual steps.

• Draw the area A. This you may find difficult but you could always use MAPLE to help you. The
area is shown in Figure 2.11.

√ √
Figure 2.11: The area enclosed by two spirals r = a φ and r = b φ.

• Decide on the order of integration. Because we are given R = R1 (φ) and R = R2 (φ) it makes
sense to take R as the inner integral and φ as the outer integral.

• Determine the limits. We’ve just determined the order and the limits follow immediately. The
radial integral goes from R = R1 (φ) to R = R2 (φ). In the φ direction we have 0 ≤ φ ≤ 2π. Thus,

Z 2π Z b√φ !
I= √ RdR dφ.
φ=0 R=a φ
2.3. BASE AREAS OF CIRCULAR OR PART CIRCULAR FORM 37

• Do the inner integral first. Thus,

Z b√φ √
1  2 b φ 1 2 2


√ RdR = 2 R R=a√φ = 2 b φ − a φ
R=a φ

Thus, the outer integral can now be done.

2π
1 2 1 2
Z

 2π
b − a2 φdφ = b − a2 φ2 φ=0 .


I=
φ=0 2 4

This can be simplified to

1 2
b − a2 4π 2 = π 2 b2 − a2 .



I=
4

Example End

Example 2.19
Find the area of an ellipse, Z Z
I= dA,
A

where A is the area enclosed by

x2 y2
2
+ 2 = 1. (2.10)
a b
a and b are constants. If a = b, then we have a circle of radius a.

Solution 2.19
We follow the usual approach.

• Draw the area. This is shown in Figure 2.12. It is not obvious but ‘modified’ polar coordinates
are the sensible choice of coordinates. Given that any point lying inside the ellipse satisfies

x2 y2
2
+ 2 = R2 ≤ 1,
a b
we can try the following change of variables
x y
= R cos φ, = R sin φ.
a b
Then the area is covered by 0 ≤ φ ≤ 2π and 0 ≤ R ≤ 1.

• Decide on order of integration. The order of integration does not matter in this case.

• Choose the limits. As mentioned above, the changes of variables indicate that the limits are
0 ≤ φ ≤ 2π and 0 ≤ R ≤ 1
38 CHAPTER 2. DOUBLE INTEGRALS

Figure 2.12: The area enclosed by an ellipse.

• Do the integration. Now we must stop and think. We do not yet have the correct expression for
dA and this not quite as simple as the standard Cartesian and polar cases. A detailed derivation
is given in the next section. For now we take
dA = abRdRdφ.
Thus, Z 1 Z 2π 
I= abRdφ dR.
R=0 φ=0

The inner integral gives Z 2π

2π
abRdφ = abR [φ]φ=0 = 2πabR.
φ=0

The outer integral gives the area of the ellipse as

Z 1  1
1 2
I= 2πabRdR = 2πab R = πab.
R=0 2 R=0

Note that if a = b we get the area of a circle of radius a as πa2 . That is a useful check!
Example End

Example 2.20
Find Z Z
x2 + y 2 dA,


I=
A
2.3. BASE AREAS OF CIRCULAR OR PART CIRCULAR FORM 39

where A is the elliptical region in the previous example.

Solution 2.20
Here you might be tempted to use ordinary polar coordinates because the integrand has the ‘polar’
form x2 + y 2 . However, getting the region properly treated rather than the integrand is more important
in this example. Again we are going to use

x = aR cos φ y = bR sin φ

The order of the integration does not matter since the limits are fixed (0 ≤ φ ≤ 2π and 0 ≤ R ≤ 1) and
do not depend on the other variable. Thus, we must evaluate
Z 2π Z 1 h i 
2 2
I= (aR cos φ) + (bR sin φ) abRdR dφ.
φ=0 R=0

The inner integral can be expressed as

1 1 4 1
 
ab  2
Z
ab a2 cos2 φ + b2 sin2 φ R3 dR = ab a2 cos2 φ + b2 sin2 φ a cos2 φ + b2 sin2 φ
  
R =
R=0 4 R=0 4

The outer integral is

2π
ab  2
Z
a cos2 φ + b2 sin2 φ dφ.

I=
φ=0 4

We have done these types of integrals in the Fourier Series part of the course. However, remember the
trigonometric identities,

1
cos2 φ = (1 + cos 2φ)
2
1
sin2 φ = (1 − cos 2φ)
2
Thus,
Z 2π
cos2 φ dφ = π,
φ=0

and
Z 2π
sin2 φ dφ = π.
φ=0

Hence,
ab  2 πab 2
a π + b2 π = a + b2 .


I=
4 4
Example End
40 CHAPTER 2. DOUBLE INTEGRALS

2.4 Generating the area element dA for any 2D coordinate sys-

tem
If we have a change of coordinates so that the Cartesian coordinates are expressed in terms of the new
coordinates, denoted here by u and v, then the infinitesimal area

dA = dxdy

can be expressed as
dA = dxdy =| J | dudv, (2.11)
where J is the Jacobian. If we have the Cartesian coordinates expressed explicitly in terms of the new
coordinates as

x = x(u, v),
y = y(u, v),

then the following partial derivatives can be calculated

∂x ∂x
∂u ∂v
∂y ∂y
∂u ∂v
The Jacobian is given by evaluating the determinant,
 ∂x ∂x

 
 ∂u ∂v 
J =  
 ∂y ∂y


∂u ∂v

This is straightforward to obtain and reduces to

 
 ∂x ∂y ∂x ∂y 
|J| =  −  (2.12)
∂u ∂v ∂v ∂u 
Let’s try this out for the elliptic coordinates used in the last example of the last section. Here

x = aR cos φ, y = bR sin φ.

The partial derivatives (where u = R and v = φ in the above expressions) are

∂x ∂x
= a cos φ, = −aR sin φ
∂R ∂φ
and
∂y ∂y
= b sin φ = bR cos φ.
∂R ∂φ
Therefore, the modulus of the Jacobian is

|J| = |(a cos φbR cos φ) − (−aR sin φb sin φ)| = abR cos2 φ + sin2 φ = abR.


2.4. GENERATING THE AREA ELEMENT DA FOR ANY 2D COORDINATE SYSTEM 41

Hence, the infinitesimal area is

dA = dxdy = abRdRdφ
as used above.
You can check that the change of coordinates to polar coordinates does indeed give the correct form
for the infinitesimal area. The same idea can be extended to volume integrals where we will need to
evaluate the determinant of a 3 × 3 matrix based on all the partial derivatives.
42 CHAPTER 2. DOUBLE INTEGRALS
Chapter 3

Triple Integrals

We now move onto triple integrals. While these are generally more work than double integrals, the
procedure is exactly the same: draw the diagram, decide on the order, decide on the limits and then do
the integrals in turn.
A triple integral is the expression
Z Z Z
I= f (x, y, z)dV, (3.1)
V

where V is the region or domain or volume of integration and dV is an infinitesimal volume element.
In Cartesian coordinates,
dV = dxdydz. (3.2)
Of course, V is only volume if x, y, z are spatial coordinates; they could be other quantities. For example,
the total mass, M , crossing a surface in a certain time interval could be given by
Z Z Z
M= ρ(x, y, t)v(x, y, t)dV,

where ρ is the mass density, v is the speed of the material, x and y are spatial coordinates and t is
time. For this case, dV = dxdydt. The only point to take from this is that triple integrals involve three
integrals.
Notes
1. I is just a number, not a function of x, y, z.
2. f can be expressed in other coordinates if convenient, for example cylindrical polars (R, φ, z).
3. When f (x, y, z) = 1, I is the volume,
Z Z Z
Volume = I = dV. (3.3)
V

Otherwise, it is NOT.
4. When f is the density (or sometimes called the mass density) of a material within V , then I is
the total mass. We will consider such an example later.

43
44 CHAPTER 3. TRIPLE INTEGRALS

3.1 Rectangular Volumes

This follows closely Section 2.1 for double integration – here you just have one more integral to evaluate.
The limits of integration are all fixed in this Section and independent of each other. In general, the
order of integration is not important (you can do them in any order). A general integral of this type
can be written
Z Z Z Z g Z d Z b ! !
I= f (x, y, z)dxdydz = f (x, y, z)dx dy dz.
V z=e y=c x=a

As the order of integration does not really matter (do the simplest first), this expression for I can be
written in many other ways (but also equivalent in the sense of giving the same answer). As an example,
I can also be written as
Z Z Z Z b Z d Z g  !
I= f (x, y, z)dxdydz = f (x, y, z)dz dy dx.
V x=a y=c z=e

Note how important it is that you indicate clearly which variable has which limits. Just one example is
given for rectangular volumes (as they really are so similar to the double integrals in rectangular areas).

Example 3.21
Suppose f (x, y, z) = (x + y + z))−5/2 and the volume is

0 ≤ x ≤ 1,
0 ≤ y ≤ 2.
0 ≤ z ≤ 3.

Solution 3.21
Then, the integral is Z 3 Z 2 Z 1  
−5/2
I= {x + y + z} dx dy dz
z=0 y=0 x=0

During the x integration, y and z are held fixed. During the y integration, z is held fixed (remember
x does not appear as we have already integrated it out). Finally, the z integration is a simple integral
only involving one variable. I am going to remove the brackets indicating the order of the integration.
It is assumed that we will always start with the innermost integral and work our way out to finally do
the outermost one.
Z 3 Z 2 Z 1
I = {x + y + z}−5/2 dxdydz
z=0 y=0 x=0
3 2  1
2
Z Z
−3/2
= − (x + y + z) dydz
z=0 y=0 3 x=0
2 3
Z Z 2 n o
= − (1 + y + z)−3/2 − (y + z)−3/2 dydz
3 z=0 y=0
3.2. VOLUMES BOUNDED BY OBLIQUE PLANES AND SIMPLE SURFACES 45

3 h 
2 i2 i2
Z h
−1/2 −1/2
= − −2 (1 + y + z) + 2 (y + z) dz
3 z=0 y=0 y=0
Z 3
 
4 −1 1 1 1
= − + + − dz
3 z=0 (3 + z)1/2 (1 + z)1/2 (2 + z)1/2 z 1/2
8h i3
= − −(3 + z)1/2 + (1 + z)1/2 + (2 + z)1/2 − z 1/2
3 z=0
8 n √ √ √ √   √ √ o
= − − 6+ 4+ 5− 3 − − 3+1+ 2−0
3
8n √ √ √ o
= − 1+ 5− 6− 2
3
This cannot be simplified any further but it is approximately 1.67.
Example End

3.2 Volumes bounded by Oblique Planes and Simple Surfaces

For all but rectangular volumes, we need to work out the order of integration and the limits. Doing the
integrals themselves is no harder than any integrals you have seen before.
The key steps are the same as for double integrals:

1. Sketch the region. This is likely going to be the hardest part! It is hard to visualise regions in 3D
and some practice is needed. If you really find this difficult, you should use MAPLE initially to
help you.

2. Decide on the order of integration. This can sometimes be crucial and you should always reconsider
the order of integration if your first approach runs into difficulties.

3. Decide on the limits. In a sense, this is the key to triple integrals and, once correctly chosen,
everything else falls into place. Since we are dealing with functions of three variables, the inner-
most limits can depend on BOTH of the outer two variables (not the innermost as that is the
one we are integrating over). The middle limits can depend ONLY on the outer variable (cannot
depend on either the middle variable or the inner variable). The outer limits MUST be constants.
Evaluating the triple integral results in a CONSTANT.

4. Do the integration, starting from the innermost integration, through the middle integration to end
with the outer integration.

We learn how to do triple integrals by examples.

Example 3.22
Find the centre, x̄ of the volume, V defined by the region in the first OCTANT bounded by the plane
x + y + z = 1.
The centre is defined by RRR
xdV
x̄ = R R RV . (3.4)
V
dV
46 CHAPTER 3. TRIPLE INTEGRALS

So we have to evaluate two integrals. However, both are easy once they are set up properly. This
particular volume has been presented in ‘Function of Two Variables’ part of the course.
Solution 3.22
• Sketch the volume V . This is a little tricky to show clearly and an attempt is shown in Figure 3.1

Figure 3.1: The volume enclosed by the plane x + y + z = 1 and lying in the first octant.

• Decide on the order. Here there are six possible orders for doing the integrations but luckily all
are equally simple. Let us choose z as the outer variable, y as the middle one and x as the inner
variable.
• We imagine an infinitesimal volume given by dV = dxdydz placed at an arbitrary location (x, y, z).
Now, if we are integrating in x first, then we must add up all the contributions from all the
infinitesimal boxes from x = 0 to x = 1 − y − z. This give the innermost limits. This is shown
in Figure 3.2. Now we have generated a strip (in exactly the same way as we did with double
integrals). To integrate in y, we add up all the strips from y = 0 to y = 1 − z.
At this stage, we have a plane located at a height z and to complete the integration, we add up
all the planes from z = 0 to z = 1. Thus, we have added up all the infinitesimal boxes so that we
cover the volume exactly once. So the integrals we need to evaluate are
Z 1 Z 1−z Z 1−y−z  
I= xdx dy dz
z=0 y=0 x=0
3.2. VOLUMES BOUNDED BY OBLIQUE PLANES AND SIMPLE SURFACES 47

Figure 3.2: The volume viewed from above. The outer triangle represents the base at z = 0. The thick
line is located at an arbitrary height z (in this case z = 0.4). The contribution to the integrals at this
height is given by adding up the infinitesimal volumes from x = 0 to x = 1 − z − y (the bold line), as
shown in the Figure.

and Z 1 Z 1−z Z 1−y−z  

V = dx dy dz.
z=0 y=0 x=0

It is much neater if I leave out the brackets and write the integrals as,
Z 1 Z 1−z Z 1−y−z
I= xdxdydz
z=0 y=0 x=0

and Z 1 Z 1−z Z 1−y−z

V = dxdydz.
z=0 y=0 x=0

Remember, we start with the innermost integrals and work our way out.

• Do the integration. This is now relatively easy and we do them in turn. To evaluate I, we start
with the x integration
1−y−z
1  2 1−y−z 1
Z
2
xdx = x x=0 = (1 − y − z) .
x=0 2 2
48 CHAPTER 3. TRIPLE INTEGRALS

Now the y integration gives

1−z i1−z
1 11h 1n
Z o
2 3 3
(1 − y − z) dy = − (1 − y − z) =− 0 − (1 − z) .
y=0 2 23 y=0 6

Finally, we can do the z integration to get I

1 i1
1 1 h 1 1
Z
I= (1 − z)3 dz = − (1 − z)4 = − {0 − 1) = .
z=0 6 24 z=0 24 24

Now we repeat the integration to determine the volume. Firstly the x integration gives
Z 1−y−z
dx = (1 − y − z) .
x=0

next the y integration gives

1−z i1−z
1h 1
Z
(1 − y − z) dy = − (1 − y − z)2 = (1 − z)2
y=0 2 y=0 2

Finally, the z integration gives the volume as

1
1 1h i1 1
Z
2 3
V = (1 − z) dz = − (1 − z) = .
z=0 2 6 z=0 6

Hence, the centre x̄ is

1 6 1
x̄ = = .
24 1 4

It is fairly obvious, by symmetry, that

1
x̄ = ȳ = z̄ = .
4
Note that (x̄, ȳ, z̄) is called the centre of mass of the volume V if it is made up of a material of
uniform density.

Example End

Example 3.23
Find Z Z Z
I= zdV,
V

where V is the volume within a conical ‘mountain’ of width 2a and height h, above the plane z = 0.

Solution 3.23
3.2. VOLUMES BOUNDED BY OBLIQUE PLANES AND SIMPLE SURFACES 49

• Draw the volume V . A cone can be generated as a surface of revolution about the z axis. At any
height between z = 0 and z = h, the edge of the cone appears circular. The radius of the circle is
Re = a at z = 0 and Re = 0 at z = h. These points can be joined by a straight line to give
 z
Re = a 1 − .
h
The equation of a circle (and hence the equation of the surface of the cone) is
 z 2
x2 + y 2 = Re2 = a2 1 − .
h
Thus, we have a description of the volume within the cone, namely that
x2 + y 2 ≤ Re2 .
The volume is shown in Figure 3.3

Figure 3.3: The volume defined by an inverted cone.

• Decide on the order of integration. It seems plausible that z should be the outer variable, i.e. do
the z integration last. What do we choose for the inner variable, x or y? Actually it does not
matter. What are we calculating with the integrals over x and y? Remember that z is held fixed,
during these integrations, and so we have
Z h Z Z 
I= z dxdy ,
z=0 A
50 CHAPTER 3. TRIPLE INTEGRALS

where A is the area of a circle of radius Re . This is, of course, just πRe2 and so
h
z 2
Z 
I= zπa2 1 − dz.
z=0 h

• Decide on the limits. While I have just said what the integral will turn out to be, it is worthwhile
getting the limits for the x, y and z integrals as practice. Thus, we imagine that we have an
infinitesimal box of volume dV = dxdydz at a general point (x, y, z). Now we add up all the
contributions in the x direction. Remember that the equation describing the edge of the circle (at
height z) is x2 + y 2 = Re2 , the x integration will be from
p p
x = − Re2 − y 2 , to x = + Re2 − y 2 .

This gives a horizontal strip. Now we add up the contributions from all the strips in the y direction
to generate the contribution from the circle at height z. Thus, the y integration goes from

y = −Re to y = +Re .

Finally, we add up all the circle contributions from z = 0 to z = h. Thus,

Z h Z Z √ 2 2
Re
! !
Re −y
I= zdx dy dz.
z=0 y=−Re x=−sqrtRe2 −y 2

• Do the integration in the correct order. We start with the x integration. Thus, we have
Z √Re2 −y2 p
2 2
√ 2 2 zdx = 2z Re − y .
x=− Re −y

The y integral becomes

Z Re p
2z Re2 − y 2 dy.
−Re

Now the y integration, requires the substitution y = Re sin φ. Thus, dy = Re cos φdφ. The y limits
now give y = −Re implies φ = −π/2 and y = Re implies φ = π/2. Therefore,
Z Re p Z π/2
2 2
2z Re − y dy = 2zRe2 cos2 φdφ = πRe2 z,
y=−Re φ=−π/2

as predicted above. Finally, the z integration gives

h h
z 2
Z Z 
I= πzRe2 dz = πza2 1 − dz.
z=0 z=0 h

Expanding the brackets gives

h
z2
 
z
Z
2
I = πa z 1 − 2 + 2 dz
z=0 h h
3.2. VOLUMES BOUNDED BY OBLIQUE PLANES AND SIMPLE SURFACES 51

h
z2 2 z3 1 z4

= πa2 − +
2 3 h 4 h2 z=0
 2 
2 h 2 2 1 2
= πa − h + h
2 3 4
1
= πa2 h2
12

Similarly, it is easy to show that the volume of this cone is

1 2
V = πa h.
3
So the mean height of the cone (or centre) is
RRR
zdV 1
z̄ = R R R = h.
dV 4

Example End

The above example illustrates a tremendous simplification to triple integrals when

1. V is a volume of revolution – obtained by taking a curve in the y-z plane and rotating it about
the z axis and

2. f (x, y, z) = f (z) only.

The cone is just one of many examples. All we are doing is stacking circles or discs of radius Re (z),
where Re is the radius to the edge of the shape and whose value depends on the height z, and thickness
dz. Thus, Z Z Z Z z2
f (z)dV = π f (z)Re (z)2 dz,
V z1

over the range of z, namely z1 and z2 that defines the volume V .

Let us do one example of a volume of revolution.

Example 3.24
Find, Z Z Z
I= e−z dz,
V

where V is the volume enclosed by a paraboloid,

az = x2 + y 2 ,

and the plane z = b. Here, a is a constant.

Solution 3.24
52 CHAPTER 3. TRIPLE INTEGRALS

• Draw V . The paraboloid only exists for z > 0. Along y = 0, az = x2 and this is just a parabola.
Along x = 0, az = y 2 and this again a parabola. Since
R 2 = x2 + y 2 ,
in polar coordinates, we can define
Re2 (z) = az.
Note that the cross-sectional area, πRe2 just becomes
πRe2 = πaz,
for a paraboloid at a fixed height z. The paraboloid is shown in Figure 3.4

Figure 3.4: The volume of a paraboloid is illustrated.

• Limits. These are simply z = 0 and z = b. Hence, the integrals become

Z Z Z Z b Z b
I= e−z dz = π e−z Re2 (z)dz = πa ze−z dz.
V z=0 z=0

• Do the integration. This integration will require integration by parts (as in Fourier Series part of
the course). We choose,
dv
u = z, = e−z
dz
du
= 1, v = −e−z .
dz
3.3. VOLUMES WITH CIRCULAR OR PART CIRCULAR CROSS SECTIONS 53

Hence, ( )
Z b Z b
−z b
 
I = πa ze dz = πa −ze
−z
z=0
+ e−z dz .
z=0 z=0

Thus,
 b
I = −πa (z + 1)e−z z=0


= −πa (b + 1)e−b − 1


= πa 1 − (b + 1)e−b > 0.

Example End

3.3 Volumes with Circular or Part Circular Cross Sections

Briefly, we consider a generalisation of the two examples just given above, namely integrals of the form
Z Z Z
I= f (x, y, z)dV,
V

where V is circular (at least in part) in a cross-section. To keep things simple, we assume that the
circular cross-section is always in the plane z =constant. Some typical examples are
1. R = constant: the cylinder.
2. R = az: the cone.
√
3. R = a z: the paraboloid.
√
4. R = a2 − z 2 : the sphere. This is probably more easily recognised in the form x2 + y 2 + z 2 = a2 .
√
5. R = a2 + z 2 : the hyperboloid.
In all of these examples,
R 2 = x2 + y 2 .
Unlike the examples of Section 3.2, we now allow the integrand to be a function of all three spatial
coordinates, namely f (x, y, z). Now we cannot simplify the triple integral so easily as above. Care is
needed in deciding the order of integration.
We illustrate the idea through two examples.
Example 3.25
Obtain Z Z Z
I= x2 dV,
V
where V is the volume enclosed by the inverted cone considered previously, namely
 z
R = Re (z) = a 1 − ,
h
for 0 ≤ z ≤ h and the plane z = 0.
Solution 3.25
54 CHAPTER 3. TRIPLE INTEGRALS

• Draw V . This has been done before in Figure 3.3.

• Decide on order. Since the integrand in independent of z, we can always do this one at any time.
Hence, take z as the outer variable. Thus, we are looking at the integral in the form
Z Z Z 
2
I= x dA dz,
A

where A is a circular area of radius Re (z). This is exactly the same as the double integrals with
circular cross-section that we saw in Section 2.3. We need to use polar coordinates, with
x = R cos φ,
y = R sin φ,
dA = RdRdφ.
We can do either R or φ first. It does not matter. Choose R as the inner variable and φ as the
middle one. Thus, Z Z Z  
2 2
I= R cos φRdR dφ dz.

• Decide on limits. These are fairly straightforward. For R we have

0 ≤ R ≤ Re (z).
For φ we have
0 ≤ φ ≤ 2π.
Finally for z we have
0 ≤ z ≤ h.
Hence, the integral is
! !
Z h Z 2π Z Re (z)
2 2
I= R cos φRdR dφ dz.
z=0 φ=0 R=0

• Do the integrals. Starting with the R integral, remembering that φ is constant during this inte-
gration so that the cos2 φ term can be brought outside the integral sign, we have
Z Re
1  4 Re 1  z 4
R3 dR = R R=0 = a4 1 − .
R=0 4 4 h
The φ integration can now be done. Remember the Fourier Series results where we integrated
terms like cos2 φ, this gives,
Z 2π 2π
1 2π

1 1
Z
cos2 φdφ = (1 + cos 2φ) dφ = φ + sin 2φ = π.
φ=0 2 φ=0 2 2 φ=0

Hence, the final integration gives

Z h  h
π 4 z 4 π 4  z 5
I= a 1− dz = − a h 1 − .
z=0 4 h 20 h z=0
3.3. VOLUMES WITH CIRCULAR OR PART CIRCULAR CROSS SECTIONS 55

This can be evaluated to give

1
I= πa4 h.
20
Example End

Example 3.26
Find Z Z Z
I= xyzdV,
V

where V is the portion of the sphere of radius a in the first octant, i.e. x, y, z > 0.

Solution 3.26
• Draw V . This is shown in Figure 3.5.

Figure 3.5: The sphere is the first octant.

We will use polar coordinates for x and y or more correctly we switch from Cartesian coordinates
(x, y, z) to cylindrical coordinates (R, φ, z) as shown in Figure 3.6. This is sensible since the cross-
section at fixed height z is part of a circle. In any horizontal plane the cross-section of the volume
is shown in Figure 3.6. In cylindrical coordinates, R varies from 0 to Re (z) where

R2 = x2 + y 2 = a2 − z 2 = Re (z)2 .
56 CHAPTER 3. TRIPLE INTEGRALS

z ez
e
R φ

r
e
R
x
φ y

Figure 3.6: Cylindrical coordinates (R, φ, z).

√
Thus, Re (z) = a2 − z 2 . φ varies from 0 to 2π and z varies from 0 to a.
Using the same approach as in Section 3.3, we have

dV = dxdydz = dAdz = RdRdφdz.

• Decide on order of integration. Clearly z should be the outer variable and done last. R and φ are
the inner and middle variables, since the R limits in particular depend on z through Re (z).

• Decide on limits. These have more or less been worked out already.

0 ≤ z≤h
0 ≤ φ ≤ π/2
p
0 ≤ R ≤ Re (z) = a2 − z 2 .

Thus,
Z h Z π/2 Z Re (z)
I= R cos φ R sin φ zRdRdφdz,
z=0 φ=0 R=0

where we have replaced x and y in the integrand by R cos φ and R sin φ respectively and dV by
RdRdφdz.

• Do the integrals. First we integrate with respect to R. Thus,

Re (z) √
1  4  a2 −z2 1 2
Z

2
R3 dR = R R=0 = a − z2 .
R=0 4 4

Next, integration with respect to φ (using the substitution u = sin φ) gives

π/2  π/2
1 1
Z
cos φ sin φ dφ = sin2 φ = .
φ=0 2 φ=0 2
3.4. VOLUMES WHICH ARE PORTIONS OF SPHERES 57

Finally, bringing all the results together and doing the z integration gives
Z a
1
2
I= z a2 − z 2 dz.
z=0 8
We can do this integral by expanding the brackets (with care) and integrating term by term.
However, we can also do this using a substitution, u = a2 − z 2 . Thus,

u = a2 − z 2 , ⇒ du = −2zdz.

Thus,
0
1 1 1  3 a2 a6
Z
I= − u2 du = u u=0 = .
u=a2 8 2 48 48

Example End
This example brings us to spherical coordinates which often prove useful for evaluating integrals over
spherical volumes, as discussed in the next section.

3.4 Volumes which are Portions of Spheres

Sometimes, and it occurs more often than you might imagine, integrals arise over spherical volumes.
For example, in meteorology finding the mean temperature of the Earth and in Solar Physics finding
the mass of the Sun, requires the evaluation of integrals over a spherical volume.
z

er

θ e
φ
r
eθ
x y
φ

Figure 3.7: Spherical coordinates, (r, θ, φ).

Spherical coordinates consist of the three variables (r, θ, φ), where r is the radius, θ is the co-latitude
and is measured from the North pole θ = 0 down to the South pole θ = π and φ is the longitude. This
is shown in Figure 3.7. We have,

x = r sin θ cos φ (3.5)

y = r sin θ sin φ (3.6)
z = r cos θ (3.7)
58 CHAPTER 3. TRIPLE INTEGRALS

Important notes: We can readily calculate the infinitesimal volume element by a similar method to
that introduced in Section 2.4.

3.4.1 Volume Element

We take the partial derivatives and form the matrix
 ∂x ∂x ∂x

∂r ∂θ ∂φ
 ∂y ∂y ∂y 
 ∂r ∂θ ∂φ 
∂z ∂z ∂z
∂r ∂θ ∂φ

This gives  
cos φ sin θ r cos φ cos θ −r sin φ sin θ
 sin φ sin θ r sin φ cos θ r cos φ sin θ 
cos θ −r sin θ 0
Now we calculate the determinant of this 3 × 3 matrix. It is not too difficult if we expand it by the first
row. This gives the Jacobian, J, as
   
 r sin φ cos θ r cos φ sin θ 
 − r sin φ sin θ  sin φ sin θ r cos φ sin θ 
 
J = cos φ sin θ 
−r sin θ 0   cos θ 0 
 
 sin φ sin θ r cos φ sin θ 
−r sin φ sin θ  
cos θ −r sin θ 

This can be simplified to

J = r 2 cos2 φ sin3 θ + r 2 cos2 φ cos2 θ sin θ + r 2 sin2 φ sin θ(sin2 θ + cos2 θ)

= r 2 sin θ(cos2 φ sin2 θ + cos2 φ cos2 θ + sin2 φ)
= r 2 sin θ(cos2 φ + sin2 φ) = r 2 sin θ

Therefore, the volume element is given by

dV = dxdydz = r 2 sin θdrdθdφ (3.8)

3.4.2 Examples
Now consider integrals of the form
Z Z Z
I= f (r, θ, φ)dV, (3.9)
V

where V is a sphere (or part of a sphere). We are nearly ready to start the integration process but first
we need to recall from above the volume element dV in terms of spherical coordinates.
Let us consider two examples.
Example 3.27
Find the volume of a sphere of radius a.
Solution 3.27
3.4. VOLUMES WHICH ARE PORTIONS OF SPHERES 59

You probably know the answer, but do you know how to get it by integration? I’ll solve the problem
in a way which will uncover a few other famous formulas you have come across. In this example, our
function f = 1, so we don’t have too much work to do. The volume of the sphere is found from
Z Z Z Z ∞ Z π Z 2π
I= dV = r 2 sin θdφdθdr.
V r=0 θ=0 φ=0

The integration order does not matter, but I’ll write the integral in the following way to reveal a few
surprises:
Z ∞ Z π Z 2π
I= dr rdθ Rdφ
r=0 θ=0 φ=0

where R = r sin θ in the inner integral. Note, R does not depend on φ, so it could be shifted to the
middle integral; however, in the above integral, I have split the volume element dV into the infinitesimal
distances Rdφ, rdθ and dr in the φ, θ and r directions respectively. Why you ask?! Well, the innermost
integral then is
Z 2π
Rdφ = 2πR
φ=0

— simply the circumference of a circle of radius R. The middle integral then is

Z π Z π
2
rdθ(2πR) = 2πr sin θdθ = 4πr 2
θ=0 θ=0

— simply the surface area of a sphere of radius r. Finally, the outermost integral then is
Z ∞
4
dr(4πr 2 ) = πa3
r=0 3

— the volume of a sphere of radius a. So, each integration reveals an interesting result: the circumference
of a circle, the surface area of a sphere and the volume of a sphere.
Example End

Example 3.28
Find the total mass of the Earth’s atmosphere, taking for simplicity the density of the air to vary like

ρ = f (r) = ρ0 e−(r−a)/h ,

where ρ0 is the density of the air at ground level, denoted by r = a. h is the scale height. Typical values
for these constants are
a = 6.4 106 m, ρ0 = 1.3kg/m3 , h = 7.4 103 m
However, although we know these values, don’t insert them until all the integration has been performed.
Why? Well if you want to repeat the example for a different planet you can use the same integration
and just change the values of the constants in the final answer.

Solution 3.28
60 CHAPTER 3. TRIPLE INTEGRALS

The mass is given by integrating the density over the volume of the atmosphere,
Z Z Z
I = ρ(r)dV
Z Z ZV
= ρ(r)r 2 sin θdrdθdφ
V

This is not so hard as the density is only a function of radius. Hence, the order of integration is not
really important. What about the limits? To cover the volume of atmosphere above ground level, we
must integrate φ from 0 to 2π and θ from 0 to π (see Figure 3.7). The radial integration goes from
ground level, r = a to infinity. Thus,
Z ∞ Z π Z 2π
I= ρ0 e−(r−a)/h r 2 sin θdφdθdr.
r=a θ=0 φ=0

The φ integration is easy

Z 2π
ρ0 e−(r−a)/h r 2 cos θdφ = 2πρ0 e−(r−a)/h r 2 sin θ
φ=0

Now the integration in θ gives,

Z π
2πρ0 e−(r−a)/h r 2 sin θdθ = 2πρ0 e−(r−a)/h r 2 × 2
θ=0

Finally, the r integration, after expanding the brackets in the exponential, gives
Z
I = 4πρ0 r 2 e−r/h ea/h dr.
r=a

This requires integration by parts (twice). We choose u = r 2 so that du/dr = 2r and dv/dr = e−r/h so
that v = −he−r/h . Hence.
Z ∞
a/h
I = 4πρ0 e r 2 e−r/h dr
r=a
h i∞ Z ∞ 
a/h 2 −r/h
= 4πρ0 e −hr e +h 2re−r/h
dr
a r=a
 h i∞ Z ∞ 
a/h 2 −a/h
= 4πρ0 e h a e + 2 −hre −r/h
+ 2h e−r/h
dr
a
n or=a
= 4πρ0 ea/h h a2 e−a/h + 2hae−a/h + 2h2 e−a/h .

This simplifies to give

I = 4πρ0 h a2 + 2ha + 2h2 .

Putting in the numbers for the Earth, this comes out to 5.0 1018 kg. This should be approximately equal
to the mean surface pressure 1.0 105 kg m−1 s−2 divided by gravity 9.8 m s−2 times the surface area of
the Earth 4πa2 . This comes out to 5.3 1018 kg.
Example End