MA3K4: Introduction to group theory

Lecture notes
2
Contents

1 An introduction to group theory 5

1.1 Basic definitions and examples . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Orders of elements in groups . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3 Subgroups and cosets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.4 Normal subgroups and quotient groups . . . . . . . . . . . . . . . . . . . . . . 14
1.5 Homomorphisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

2 Permutation groups and group actions 19

2.1 Permutation groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2 Group Actions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.3 Fixed points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

3 The Sylow theorems 31

3.1 Wielandt’s Proof of Sylow Theorems 1 and 4 . . . . . . . . . . . . . . . . . . 32
3.2 Proof of Sylow Theorems 2 and 3 . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3 Consequences of Sylow’s theorems . . . . . . . . . . . . . . . . . . . . . . . . 36
3.4 Applications of Sylow’s theorems . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.5 Simplicity of An , n ≥ 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.6 An interesting example of a Sylow p-subgroup, and an application (non-examinable) 41

4 Classifying groups of small order 45

4.1 Semidirect products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.2 Semidirect products of abelian and cyclic groups, and a lemma of Fitting . . 48
4.3 Abelian groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.4 Groups of order p, p2 , or 2p, for p prime . . . . . . . . . . . . . . . . . . . . . 50
4.5 Groups of order 2p2 , where p is an odd prime . . . . . . . . . . . . . . . . . . 50
4.6 Groups of order pq, where p < q are primes with p ∤ q − 1 . . . . . . . . . . . 51
4.7 Groups of order 8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.8 Groups of order 12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.9 Unique simple group of order 60 . . . . . . . . . . . . . . . . . . . . . . . . . 55

5 Soluble groups and the Jordan–Hölder theorem 57

5.1 Composition Series and the Jordan-Hölder Theorem . . . . . . . . . . . . . . 57
5.2 Commutators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.3 Some important examples of soluble groups . . . . . . . . . . . . . . . . . . . 64
5.4 Nilpotent groups (non-examinable) . . . . . . . . . . . . . . . . . . . . . . . . 65

3
Appendices 69

A 3 tools from Chapters 1 and 2 for analysing element orders in finite groups 69

B The Sylow game 73

4
Chapter 1

An introduction to group theory

The purpose of this chapter is to revise some of the theorems and definitions that we saw in
Algebra II. Our main aim will be to define a group; see some examples and non-examples of
groups; and begin our studies of the three most important aspects of group theory: elements
(Section 1.2); subgroups (Section 1.3 and 1.4); and homomorphisms (Section 1.5).

1.1 Basic definitions and examples

We begin in the obvious way – by defining a group!
Definition 1.1. A group is a pair (G, ◦) where G is a set, and ◦ : G × G → G is a binary
operation on G that satisfies the following three properties:
(i) Associativity. For all g, h, k ∈ G, (g ◦ h) ◦ k = g ◦ (h ◦ k).

(ii) Identity. G has an element, which we call 1G , satisfying g ◦ 1G = 1G ◦ g = g for all

g ∈ G. We call 1G an identity element in (G, ◦).

(iii) Inverse. For all g ∈ G, there exists an element in G, which we call g −1 , satisfying
g ◦ g −1 = g −1 ◦ g = 1G , where 1G is as in (ii). We call g −1 an inverse of g in (G, ◦).
Remark. There are a few remarks to be made about Definition 1.1.
(1) Implicit in part (ii) of Definition 1.1 is that an identity element for an associative binary
operation is unique. This is why we write it as 1G , and from now on, call it the identity
element in (G, ◦).

(2) Similarly, the inverse of an element g in an associative binary operation is unique. This
is why we write it as g −1 , and from now on call it the inverse of g in (G, ◦).
Example. Let G = Z. If we define ◦ : G × G → G by g ◦ h := g + h, then we know that
G is a group (the binary operation is clearly associative; the identity element is 0; while the
inverse of g ∈ G is −g).
If we define ◦ : G × G → G by g ◦ h := gh, then G is not a group (the binary operation
is clearly associative and there is an identity element (namely 1), but if g ∈ G \ {1}, then g
does not have an inverse).
Exercise. Can you think of other “natural” binary operations that do not yield a group?

5
Remark. Note the following.
(1) When you are checking whether or not a given pair (G, ◦) is a group, note that you
first need to check that ◦ is a binary operation. This is a kind of “hidden” axiom, so be
careful!
(2) If (G, ◦) is a group, then we sometimes call ◦ the group operation.
(3) When the context is clear, we will streamline the notation for a group (G, ◦) by omitting
◦. Thus, (when the context is clear), we simply say that “G is a group” and we will
write gh instead of g ◦ h. We will call gh the “product” of g and h.
(4) The number of elements in G is called the order of G and is denoted by |G|. This may
be finite or infinite.
Next, notice that all the group (Z, +) satisfies gh = hg for all g, h ∈ G. Groups with this
property have a special name.
Definition 1.2. A group G is called abelian if gh = hg for all g, h ∈ G. More generally, two
elements g and h of G are said to commute if gh = hg.
Most groups we encounter in this course will not be abelian! Our first two (important)
examples of such groups are as follows.
Symmetric groups
Let X be a finite set. We write Sym(X) for the set of all bijections f : X → X. If we define
f ◦ g to be the usual composition of functions, then it is clear that ◦ : Sym(X) × Sym(X) →
Sym(X) is a binary operation. We now check the group theoretic operations:
(i) Let f, g, h ∈ Sym(X). We need to prove that (f ◦ g) ◦ h = f ◦ (g ◦ h). Since two
maps are equal if and only they agree on every element, it suffices to fix x ∈ X and
prove that ((f ◦ g) ◦ h)(x) = (f ◦ (g ◦ h))(x). Now, the left hand side of this equation
is (by definition of composition (f ◦ g)(h(x)) = f (g(h(x))). The right hand side is
f ((g ◦ h)(x)) = f (g(h(x))). Associativity follows.
(ii) The identity bijection idX : X → X defined by idX (x) := x for x ∈ X is clearly an
identity element for Sym(X).
(iii) If f ∈ Sym(X), then define f −1 : X → X by f −1 (y) = x, where y is the unique element
of x satisfying f (x) = y. One can then check very quickly that f ◦ f −1 = f −1 ◦ f = idX .
Thus, we have shown that for any set X, Sym(X) is a group (with group operation the usual
composition of functions).
Example. Suppose that X = {1, 2, 3}, and let Sym(X) be as above. Then by listing all of
the possibilities for f : X → X, we can see that Sym(X) := {f1 , f2 , f3 , f4 , f5 , f6 } where
f1 : 1, 2, 3 −→ 1, 2, 3
f2 : 1, 2, 3 −→ 2, 1, 3
f3 : 1, 2, 3 −→ 3, 2, 1
f4 : 1, 2, 3 −→ 1, 3, 2
f5 : 1, 2, 3 −→ 3, 1, 2
f6 : 1, 2, 3 −→ 2, 3, 1

6
 respectively. One can quickly check that f2 ◦ f3 ̸= f3 ◦ f2 , so Sym(X) is a non-abelian group!

We can now define symmetric groups.

Definition 1.3. Let X be any set, and let Sym(X) denote the set of bijections from X to
itself. Then Sym(X) is a group under composition of maps: it is called the symmetric group
on X. Elements of Sym(X) are called permutations.

Recall that Sym(X) is indeed a group (we proved this above).

Now, it is a little bit tedious to write down permutations of a given finite set, so we usually
use shorthand notation, called cyclic notation. You already saw this in Algebra II, but we
will now remind ourselves of this notation. We begin by defining cycles in Sym(X).

Definition 1.4. Let X be a set.

(i) If a1 , . . . , ar are distinct elements of X, then (a1 , a2 , . . . , ar ) denotes the permutation

f ∈ Sym(X) with

(a) f (ai ) = ai+1 for 1 ≤ i < r.

(b) f (ar ) = a1 , and
(c) f (b) = b for b ∈ X \ {a1 , a2 , . . . , ar }.

If f ∈ Sym(X) is equal to (a1 , . . . , ar ) for some distinct elements a1 , . . . , ar ∈ X, then

f is called a cycle in Sym(X).

(ii) Two cycles (a1 , . . . , ar ), (b1 , . . . , bs ) ∈ Sym(X) are called disjoint if {a1 , . . . , ar }∩{b1 , . . . , bs } =
∅.
(iii) The empty cycle, written (), is defined as () := IdX .

Remark. (Basic but important points about the symmetric groups)

(1) As part (iii) of the definition above suggests, we think of IdX as a cycle. Recall also
that IdX = 1G is the identity element for G = Sym(X).

(2) Recall that the group operation in Sym(X) is composition of functions, so if f =

(a1 , . . . , ar ) and g = (b1 , . . . , bs ) are two cycles in Sym(X), then f g = f ◦ g : X → X
means “apply g first, then apply f ”.

(3) An important point to note is that cycle notation is not unique. That is, two ordered
t-tuples of distinct elements of a set X may define the same cycle. For example, (1, 2, 3)
and (3, 1, 2) are different ordered 3-tuples in 1, 2, 3, but they define the same permutation
in Sym({1, 2, 3}).

Example. Let X = {1, 2, 3}, and consider the cycles f := (1, 2, 3) and g := (1, 2) in Sym(X).
Since f g is a bijection, to determine f g we need to determine the image of every element of
X under the map f g. To this end, we have

• (f g)(1) = f (g(1)) = f (2) = 3;

• (f g)(2) = f (g(2)) = f (1) = 2; and

7
 • (f g)(3) = f (g(3)) = f (3) = 1.
Thus, h := f g is the permutation defined by h(1) = 3; h(2) := 2 and h(3) := 1. Hence,
h = f g is the cycle (1, 3).
Theorem 1.5. Let X be a finite set. Then
(i) |Sym(X)| = |X|!.
(ii) Every element of Sym(X) can be written as a product of disjoint cycles (of length greater
than 1). Moreover, this product is unique in the following sense: If f ∈ Sym(X) has the
form f = f1 . . . fm = g1 . . . gt where the fi are disjoint cycles of length greater than 1;
and the gj are disjoint cycles of length greater than 1, then m = t and {f1 , . . . , fm } =
{g1 , . . . , gt }.
Proof. (Non-examinable)
(i) We simply need to count the number of bijections from X to itself. Write X = {x1 , . . . , xn },
where |X| = n, and let f ∈ Sym(X). Then the number of choices for f (x1 ) is n; the number
of choices remaining for f (x2 ) is n − 1; and so on. Thus, |Sym(X)| = n!, as needed.
(ii) Let f be an element of Sym(X). If f = IdX , then f = () is the empty cycle, so f is indeed
a product of disjoint cycles.
So assume that f ̸= IdX , and let Y := {x ∈ X : f (x) ̸= x}. Fix an element a1 of Y with
f (a1 ) ̸= a1 . Since Sym(X) has finite order, the element f has finite order. That is, there
exists n ∈ N such that f n = IdX . Set
m1 := min{m ∈ N : (f m )(a1 ) = a1 .
Then, for 2 ≤ i ≤ m1 , define ai ∈ X by ai := f (ai−1 ).
If Y = {a1 , . . . , am1 } (i.e. if a1 , . . . , am1 are the only elements in Y ), then we have
f = (a1 , . . . , am1 ) by definition of cycles in Sym(X). Thus, f is a product of disjoint cycles,
as needed. So assume that Y \ {a1 , . . . , am1 } is non-empty, and let am1 +1 be an element of
Y \ {a1 , . . . , am1 }. Set
m2 := min{m ∈ N : (f m )(am1 +1 ) = am1 +1 .
Then, for m1 + 2 ≤ i ≤ m2 , define ai ∈ X by ai := f (ai−1 ). Notice that since f is injective,
the sets {a1 , . . . , am1 } and {am1 +1 , . . . , am2 } have empty intersection.
If Y = {a1 , . . . , am2 }, then we have f = (a1 , . . . , am1 )(am1 +1 , . . . , am2 ) by definition of
cycles in Sym(X). Thus, f is a product of disjoint cycles, as needed. So we may assume that
Y \ {a1 , . . . , am2 } is non-empty, and we work as in the previous two paragraphs.
Since X is finite, the above process must eventually terminate. When it does, we will have
constructed f as a product of disjoint cycles, as required. The uniqueness follows because if
f = f1 . . . fm is a product of disjoint cycles, then the fi completely determine f .

Remark. We will study deep properties of the symmetric groups in Chapter 2.

General linear groups
Definition 1.6. Let F be a field, and let n be a positive integer. We define
GLn (F) := {A : A an invertible (n × n) − matrix with entries in F}
With the usual multiplication of matrices as the group operation, GLn (F) is a group, called
the general linear group of dimension n over F.

8
Remark. (Useful facts from Algebra II for working with general linear groups).

1. A field F has an additive identity element and a multiplicative identity element. Since
we need to talk about fields when studying general linear groups, we will write these
elements as 0F 1F , respectively (or 0 and 1, if the context is clear).

2. An (n × n)-matrix A over a field F is invertible if and only if det A ̸= 0 if and only if

the rows of A form a linearly independent set if and only if the columns of A form a
linearly independent set.

3. If F is a finite field, then there exists a prime p and a positive integer f such that
|F| = pf . The prime p is called the characteristic of F, and satisfies

pα = |α + .{z
. . + α} = 0.
p times

Remark. Before introducing our two classes of examples (the symmetric groups and the
general linear groups), I remarked that we are about to see examples of nonabelian groups.
This is almost true! A symmetric group Sym(X) is abelian if and only if |X| ≤ 2; while a
general linear group GLn (F) is abelian if and only if n = 1.

Exercise. Prove that a symmetric group Sym(X) is abelian if and only if |X| ≤ 2; while a
general linear group GLn (F) is abelian if and only if n = 1.

Theorem 1.7. Let F be a finite field, and write |F| = q. Then

n
n
|GLn (F)| = q ( 2 )
Y
(q n − 1).
i=1

Proof. Non-examinable, but see Exercise Sheet 1!

1.2 Orders of elements in groups

Now that we have two infinite classes of examples of groups, we can start to discuss important
and interesting properties of groups!
We begin with the order of an element in a group. First note that for a group G, an
element g ∈ G, and a positive integer n, we write g n for the product of g with itself n
times. That is, g n = g . . . g . (Note that we really mean g n = g ◦ . . . ◦ g , where ◦ is the group
| {z } | {z }
n n
operation, but as explained above, we are omitting the group operation in our work when the
context is clear!) If n = 0, then g n is defined to be 1G . If n is a negative integer, then g n is
defined as g n := (g −1 )−n .

Definition 1.8. Let G be a group, and let g ∈ G. The order of g, denoted by |g|, is defined
as follows: (
∞ if there is no integer n satisfying g n = 1G ;
|g| :=
min{n ∈ N : g n = 1G } otherwise.

9
Remark. Some important remarks about orders of elements in a group G:

• The order of the identity element 1G is 1. Moreover, 1G is the unique element of order
1 in G.

• |g −1 | = |g| for all g ∈ G.

• If n is the order of an element g ∈ G and n | ℓ, then ℓ = rn for some r ∈ Z, so

g ℓ = (g n )r = (1G )r = 1G .

Example. We will determine the orders of the elements of some of the groups we have already
seen:

• Consider the group (Z, +) from above. As mentioned, the identity element 1G = 0 has
order 1, but what about the other elements? Well, here the group operation is addition,
so for g ∈ Z we have g n = g + . . . + g = ng. Since ng = 0 if and only if either n or g is
| {z }
n
0, we deduce that |g| = ∞ for all g ∈ Z \ {0}.

• Let X be a set, and let f := (a1 , . . . , ar ) be a cycle of length r in Sym(X). Then, by

definition, f e (ai ) = aj , where j ≡ i + e (mod r). It follows that |f | = r.

• Let F be a finite field, of order q say, and let α, β be elements of the multiplicative group
F× . Then
   n 
α 0 n α 0
A := ∈ GLn (F), andA = for all n ∈ N.
0 β 0 βn

Thus, |A| equals the least common multiple of the orders of α and β as elements of F×
q .

We will see more examples like this later when we build up our theory a little more.
In the following lemma, we use some notation that you will have seen before: For an
element g in a group G, define ⟨g⟩ := {g i : i ∈ Z}.

Lemma 1.9. Let G be a group, and let a and b be two elements of G of finite order.

(i) If ℓ ∈ N, then aℓ = 1G if and only if |a| divides ℓ.

(ii) If m ∈ N, then |am | = |a|/gcd(|a|, m).

(iii) Suppose that a and b commute (i.e. ab = ba). Then |ab| divides lcm(|a|, |b|).

(iv) Suppose that a and b commute (i.e. ab = ba). If ⟨a⟩∩⟨b⟩ = {1G }, then |ab| = lcm(|a|, |b|).

Proof. We first prove (i). We have already noted that if |a| divides ℓ, then aℓ = 1G . For
the opposite direction, suppose that aℓ = 1G . By The Euclidean algorithm, we can write
ℓ = q|a| + r, where q ∈ Z and 0 ≤ r < |a|. Then 1G = aℓ = aq|a| ar = ar . Since r < |a| and
ar = 1G , we must have r = 0 by the definition of the order of an element.
We now prove (ii). Assume first that m divides |a|. Then by part (i),

amℓ = 1 =⇒ |a| divides mℓ

=⇒ |a|/m divides ℓ.

10
Thus, the order of am is the least positive integer ℓ such that |a|/m divides ℓ. Thus, |am | =
|a|/m. We have therefore proved that

|am | = |a|/m if m divides |a|. (1.1)

Let us now consider the general case. Let k = gcd(|a|, m), and write m = ks, for some s ∈ Z.
Then m|a|/k = s|a|, so (am )|a|/k = (a|a| )s = 1G . Thus, |am | divides |a|/k by part (i).
On the other hand, we have k = s|a| + tm for some s, t ∈ Z, by the Euclidean algorithm.
m
Hence, atm = ak , since as|a| = 1G . It follows, in particular, that |ak | = |atm |. But atm|a | =
m
(am|a| )t = 1G , so |atm | = |ak | divides |am | by (i). Also, |ak | = |a|/k by (1.1). It follows that
|a|/k divides |am |. It then follows from the paragraph above that |a|/k = |am |, as required.
Next, we prove (iii). Let ℓ := lcm(|a|, |b|). Since ab = ba, we have (ab)ℓ = aℓ bℓ = 1G .
Thus, ab has finite order, and |ab| divides ℓ, by part (i).
Finally, we prove (iv). So assume that ⟨a⟩ ∩ ⟨b⟩ = {1G }, and let n = |ab|. Then 1G =
(ab)n = an bn , so an = b−n . Thus, an ∈ ⟨b⟩. But then an ∈ ⟨a⟩ ∩ ⟨b⟩ = {1G }, so an = 1G . In
the same way, bn = 1G . Thus, |a|, |b| | n by part (i), so lcm(|a|, |b|) ≤ n. It follows from part
(iii) that lcm(|a|, |b|) = |ab|.

Exercise. Can you find an example of a group G and elements a, b of G with ab = ba such
that |ab| =
̸ lcm(|a|, |b|)?

1.3 Subgroups and cosets

Informally speaking, a subset H of a group G is called a subgroup if it forms a group under
the same operation as that of G. More precisely

Definition 1.10. A subset H of a group G is called a subgroup of G if

(i) h1 , h2 ∈ H =⇒ h1 h2 ∈ H;

(ii) 1G ∈ H; and

(iii) h ∈ H =⇒ h−1 ∈ H.

In this case, we write H ≤ G.

Our first important example of a subgroup is a special kind of abelian subgroup.

Definition 1.11. For an element g in a group G, recall that ⟨g⟩ := {g i : i ∈ Z}. We call ⟨g⟩
the cyclic subgroup of G generated by g. A group G is called cyclic if G = ⟨g⟩ for some g ∈ G.
In this case, the element g is called a generator of G.

Remark. If G is cyclic and g is a generator (so G = ⟨g⟩ in the language above), then |G| = |g|.

Example. Some examples and non-examples of cyclic groups:

• (Z, +) is a cyclic group, with a generator g := 1.

• If X = {1, 2}, then Sym(X) is a cyclic group, with a generator g := (1, 2).

11
 • If X = {1, 2, 3}, then the group Sym(X) is not cyclic. Why? Many reasons! Suppose
that Sym(X) is cyclic with G = ⟨g⟩. Then |G| = 6 = |g|. But we can check from our
example at teh bottom of page 2 above that all elements in Sym(X) have order 1, 2 or
3. Alternatively, as we showed in the example at the bottom of page 3, Sym(X) is not
abelian in this case!

Exercise. Let G be a group and g ∈ G. Using either the definition or the criteria in Lemma
1.12 below, you should check that the cyclic subgroup generated by g is indeed a subgroup.

The following gives a very useful way of checking whether or not a non-empty subset of a
given group is a subgroup.

Lemma 1.12. Let H be a non-empty subset of a group G. Then H is a subgroup of G if

and only if h1 h−1
2 ∈ H whenever h1 , h2 ∈ H.

Proof. (=⇒) Suppose first that H is a subgroup of G, and let h1 , h2 ∈ H. Then since h2 ∈ H,
condition (iii) in Definition 1.10 guarantees that h−1
2 ∈ H. Condition (i) then guarantees that
−1
h1 h2 ∈ H, as needed.
(⇐=) For the opposite direction, suppose that h1 h−1 2 ∈ H whenever h1 , h2 ∈ H. We need
to prove that H is a subgroup of G, that is, that conditions (i), (ii), and (iii) in Definition
1.10 hold. For condition (ii), we may choose an element h of H, since H is non-empty.
Then our condition guarantees that hh−1 = 1G ∈ H. Thus, condition (ii) in Definition
1.10 holds. Condition (iii) follows, since if h ∈ H, then 1G , h ∈ H, so h−1 = 1G h−1 ∈ H.
Condition (i) then also follows, since if h1 , h2 ∈ H, then h−1
2 ∈ H (as we just showed), so
h1 h2 = h1 (h−1
2 )−1 ∈ H, as needed.

Remark. (Notation) We will write H ⊆ G to mean that H is a subset of G; and H ≤ G

to mean that that H is a subgroup of G.

Let us look at some examples of subgroups which are not cyclic.

Example. Let F be a field, and let G := GL2 (F). Let

  
α 0 ×
H := : α, β ∈ F .
0 β

Then H is a subgroup of G, often called the diagonal subgroup of G. Indeed,

α1 α2−1
     
α1 0 α2 0 −1 0
h1 := , h2 := ∈ H ⇒ h1 h2 = ∈ H.
0 β1 0 β2 0 β1 β2−1

Thus, H is a subgroup by Lemma 1.12. The subgroup H is not the equal to ⟨g⟩ for any g ∈ G.
Exercise: why?

Our next important example of a subgroup is a generalisation of a cyclic subgroup.

Definition 1.13. Let G be a group, and let A be a non-empty subset of G. The subgroup of
G generated by A, written ⟨A⟩, is defined to be

⟨A⟩ := {aϵ11 . . . aϵmm : m ∈ N, ai ∈ A, ϵi ∈ {±1}}.

12
 Note that ⟨A⟩ is indeed a subgroup of G. More precisely: It is non-empty, since A is
non-empty. Now, let g, h ∈ ⟨A⟩. Then by definition of ⟨A⟩, we have g = aϵ11 . . . aϵmm and
h = bδ11 . . . bδnn , for some ai , bj ∈ A, and ϵi , δj ∈ {±1}. Thus, h−1 = b−δ n . . . b−δ1 . It follows
n 1
−1 γ1 γm+n
that gh = c1 . . . cm+n , where ci := ai and γi := ϵi for 1 ≤ i ≤ m; and cm+i := bn−i+1 and
γi := −δn−i+1 ∈ {±1} for 1 ≤ i ≤ n. Thus, ⟨A⟩ is a group by Lemma 1.12.

Remark. If A = {a1 , . . . , ad }, then we sometimes write ⟨A⟩ = ⟨a1 , . . . , ad ⟩ (i.e. we don’t

bother with the set brackets).

We now discuss cosets of a subgroup in a group.

Definition 1.14. Let G be a group, H ≤ G, and g ∈ G. Then the left coset gH is the subset
{gh | h ∈ H} of G. (Similarly, the right coset Hg is the subset {hg | h ∈ H}, but we will
deal mainly with left cosets in this course).

Lemma 1.15. Let G be a group. The following are equivalent for g, k ∈ G:

(i) k ∈ gH;

(ii) gH = kH;

(iii) g −1 k ∈ H.

Proof. Notice first that hH = H whenever h ∈ H, since H is a subgroup.

The proof is now straightforward. If k ∈ gH, then k = gh for some h ∈ H. Thus,
kH = ghH = gH. So (i) implies (ii).
If kH = gH, then k = k1G ∈ kH, so k = gh, for some h ∈ H. Thus, g −1 k = h ∈ H. So
(ii) implies (iii).
Finally, if g −1 k = h ∈ H, then k = gh ∈ gH. So (iii) implies (i).

Corollary 1.16. Let G be a group, and let H be a subgroup of G. For g1 , g2 ∈ G, say that
g1 ∼H g2 if g1 H = g2 H. Then ∼H is an equivalence relation on G.

Proof. We need to prove that ∼H is (1) reflexive; (2) symmetric; and (3) transitive.
Since gH = gH for all g ∈ G, reflexivity of ∼H is trivial. Symmetry is also trivial, so we
just need to prove transitivity. So let g1 , g2 , g3 ∈ G with g1 ∼H g2 and g2 ∼H g3 . We need
to prove that g1 ∼H g3 . We have g1 H = g2 H and g2 H = g3 H. But then g1 H = g3 H, so
g1 ∼H g3 , as needed.

Corollary 1.17. Let G be a group and let H be a subgroup of G. Then

(i) If g1 , g2 ∈ G, then either g1 H = g2 H, or g1 H ∩ g2 H = ∅.

(ii) Let {gi H : Fi ∈ I} be the set of ∼H -equivalence classes in G (I here is an index set).
Then G = i∈I gi H is the disjoint union of the sets gi H, for i ∈ I.

Proof. Since ∼H is an equivalence relation, the distinct ∼H -equivalence classes are pairwise
disjoint, and form am partition of G. Both parts follow.

Theorem 1.18. (Lagrange’s Theorem) Let G be a finite group and H a subgroup of G. Then
the order of H divides the order of G.

13
 F
Proof. Let I beP
as in Corollary
P1.17(ii). Then I is finite, since G is finite. Since G = i∈I gi H,
we have |G| = i∈I |gi H| = i∈I |H| = |I||H|. Thus, |H| divides |G|, as needed.

The size of the set I from Corollary 1.17 is therefore either ∞, or the integer |G|/|H|. We
formalise this in the next definition.

Definition 1.19. Let G be a group, and let H be a subgroup of G.

(i) The set of left cosets of H in G is denoted by G/H. That is, G/H := {gH : g ∈ G}.

(ii) The number of distinct left cosets of H in G (i.e. the cardinality of the set G/H) is called
the index of H in G and is written as |G : H|. If G is finite, then |G : H| = |G|/|H|.

We close this section with the some corollaries of Lagrange’s theorem on orders of elements
in finite groups.

Corollary 1.20. Let G be a finite group, and let g ∈ G. Then |g| divides |G|.

Proof. The subgroup H := ⟨g⟩ has order |g|. The result now follows immediately from
Lagrange’s theorem.

1.4 Normal subgroups and quotient groups

In this section, we focus on conjugates of subgroups in G.

Lemma 1.21. Let G be a group, let H ≤ G, and let g ∈ G. Then g H = gHg −1 is a subgroup
of G.

Proof. Let g1 , g2 ∈ gHg −1 . By Lemma 1.12, it suffices to prove that g1 g2−1 ∈ gHg −1 , since
gHg −1 is certainly non-empty.
To this end, by definition we may write g1 = gh1 g −1 and g2 = gh2 g −1 , for some h1 , h2 ∈ G.
Then g2−1 = gh−1 −1 −1 −1 −1
2 g , so g1 g2 = gh1 h2 g ∈ gHg −1 .

The following two definitions are very important.

Definition 1.22. Let G be a group, and let H be a subgroup of G.

(a) H is called normal in G if gHg −1 = H for all g ∈ G. In this case, we write H ⊴ G.

(b) The normalizer of H in G, denoted NG (H), is defined as

NG (H) = {g ∈ G : gHg −1 = H}.

Remark. (1) The following is a subtle but very useful point: If G is a group and H ≤ G,
then H is normal in G if and only if xHx−1 ⊂ H for all x ∈ G. Let’s prove that
this is true: Clearly, if H is normal in G then xHx−1 = H for all x ∈ G, so it is certainly
true that xHx−1 ⊆ H for all x ∈ G.
For the opposite direction, suppose that xHx−1 ⊆ H for all x ∈ G, and let g ∈ G. Then
gHg −1 ⊆ H. But also, g −1 Hg = (g −1 )H(g −1 )−1 ⊆ H. Multiplying across by g on the
left and g −1 on the right, we get H ⊆ gHg −1 . Thus, H = gHg −1 . Thus, we have shown
that H = gHg −1 for all g ∈ G, so H is normal in G, as claimed.

14
 (2) Let H be a subgroup of G. Then NG (H) is a subgroup of G, and H is contained as a
subgroup of NG (H). In fact, H ⊴ NG (H). We will prove these facts in Homework 1.

(3) Then {1G } ⊴ G and G ⊴ G.

The extreme case in (3) above has a special name:
Definition 1.23. A group G is called simple if G ̸= {1G } and the only normal subgroups of
G are {1G } and G.
Remark. For a group G and subsets A and B of G, we will write AB := {ab : a ∈ A, b ∈ B}
for the product of A and B. Note that this is just a subset of G, not a subgroup in general
(even if both A and B are subgroups)!
Lemma 1.24. Let N be a normal subgroup of a group G, and let g, h ∈ G. Then (gN )(hN ) =
ghN .
Proof. We first prove that (gN )(hN ) ⊆ ghN . So let x = gn1 ∈ gN and y = hn2 ∈ hN
(n1 , n2 ∈ N ). Then
xy = gn1 hn2 = gh(h−1 n1 h)n2 ∈ ghN
since h−1 n1 h ∈ h−1 N h = N .
For the opposite containment, let x = ghn ∈ ghN , with n ∈ N . Then x = (g1G )(hn) ∈
(gN )(hN ). Thus, x ∈ (gN )(hN ), so (gN )(hN ) = ghN .

The point of the previous lemma is that if N is a normal subgroup of a group G, then
we have a natural binary operation on the set G/N = {gN : g ∈ G} of left cosets of N in G,
given by ◦ : G/N × G/N → G/N , (gN ) ◦ (hN ) = ghN .
Definition 1.25. Let G be a group, and let N ⊴ G.
(i) The binary operation ◦ : G/N × G/N → G/N defined by (gN ) ◦ (hN ) = ghN is called
the natural binary operation of G/N .

(ii) With the binary operation ◦ above, (G/N, ◦) is a group called the quotient (or factor)
group of G by N .
Remark. Let N be a normal subgroup of a group G. We have not proved that (G/N, ◦) is
a group, where ◦ is the natural binary operation, but this follows almost immediately from
Lemma 1.24: Indeed, Lemma 1.24 already shows that the natural binary operation is indeed
a binary operation on G/N , so we just need to check the group theoretic axioms:
(i) Associativity. Let gN, hN, kN ∈ G/N (with g, h, k ∈ G). Then ((gN )(hN ))kN =
((gh)N )(kN ) = (gh)kN by two applications of Lemma 1.24. By associativity in G, we
then have

((gN )(hN ))kN = ((gh)N )(kN ) = (gh)kN = g(hk)N = (gN )((hk)N ) = gN ((hN )(kN )),

again applying Lemma 1.24 twice. This proves associativity of G/N under the natural
binary operation.

(ii) Identity. Let gN ∈ G/N , with g ∈ G. Then (gN )(1G N ) = (1G N )(gN ) = gN by
Lemma 1.24, so 1G N is an identity element in G/N under the natural binary operation.

15
(iii) Inverse. Let gN ∈ G/N , with g ∈ G. Then (gN )(g −1 N ) = (gg −1 N ) = 1G N using
Lemma 1.24. Thus, g −1 N is an inverse of gN under the natural binary operation.

This completes the proof.

Recall that if G is finite, then |G/N | = |G : N | = |G|/|N |.

1.5 Homomorphisms
Definition 1.26. Let G and H be a groups.

(i) A map ϕ : G → H is called a homomorphism provided ϕ(g1 g2 ) = ϕ(g1 )ϕ(g2 ) for all
g1 , g2 ∈ G.

(ii) If ϕ : G → H is a homomorphism and is a bijection, ϕ is called an isomorphism, and

we say that the groups G and H are isomorphic and write G ∼= H.

(iii) If ϕ : G → H is a homomorphism, then the kernel and the image of ϕ are defined
respectively as

ker(ϕ) := {g ∈ G | ϕ(g) = 1G } ⊴ G; and

Im(ϕ) := {ϕ(g) | g ∈ G} ≤ H.

Definition 1.27. Let G be a group and N ⊴ G. Then the map π : G → G/N defined by
π(g) = gN is a surjective homomorphism (this follows immediately from Lemma 1.24). It is
is called the natural homomorphism from G to G/N .

Theorem 1.28. (First Isomorphism Theorem) Let G and H be groups, and ϕ : G → H a

homomorphism. Then the following conditions hold:

(i) ker(ϕ) := {g ∈ G | ϕ(g) = 1H } is a normal subgroup of G.

(ii) Im(ϕ) = {h ∈ H | h = ϕ(g) for some g ∈ G} is a subgroup of H.

(iii) The map g ker(ϕ) → ϕ(g) defines an isomorphism G/ ker(ϕ) → Im(ϕ).

So, G/ ker(ϕ) ∼
= Im(ϕ).
Proof. You saw the proof of this in Algebra II, so we won’t repeat it here.

Example. Let F be a field, and let n be a positive integer. Then det : GLn (F) → F× is a
homomorphism, with kernel

SLn (F) := {A : A ∈ GLn (F) with det A = 1F }.

The group SLn (F) is called the special linear group of dimension n over F.

Theorem 1.29. (Second Isomorphism Theorem) Let G be a group, H ≤ G and N ⊴G. Then
the following conditions hold:

(i) N H = HN ≤ G.

(ii) H ∩ N is a normal subgroup of H.

16
(iii) H/(H ∩ N ) ∼
= N H/N .

Proof. This is included as an exercise in Homework 1.

Theorem 1.30. (Third Isomorphism Theorem) Let G be a group, N ⊴ G and K ⊴ G, and

suppose that N ⊆ K ⊆ G. Then K/N ⊴ G/N , and (G/N )/(K/N ) ∼
= G/K.

Proof. This is included as an exercise in Homework 1.

Theorem 1.31. (Correspondence Theorem) Let G be a group and N ⊴ G. Then there is

one-to-one correspondence between the subgroups of G/N and the subgroups of G containing
N . More precisely, the map

f : {J ≤ G | N ≤ J} → {X | X ≤ G/N }, J → J/N,

is a bijection.

Proof. Let A = {J ≤ G | N ≤ J}, B := {X | X ≤ G/N }, and let f : A → B be as in the

statement of the theorem. Let J ∈ A. Then J ≤ G with N ≤ J, and so as N ⊴ G, N ⊴ J.
Thus we may consider J/N . Clearly J/N ∈ B. Thus, the map f is well-defined. We will now
show that f is a bijection.
If J1 , J2 ∈ A with J1 ̸= J2 , then without loss of generality we may assume that there
exists x ∈ J1 with x ̸∈ J2 . Then xN ̸∈ J2 /N . Hence, f (J1 ) ̸= f (J2 ) and so f is injective.
Next, we prove that f is surjective. Take any X ∈ B. Consider Y := {g ∈ G | gN ∈ X}.
We claim that Y is a subgroup of G. If we prove this, then since f (Y ) = X is clear, we will
have shown that f is surjective. That will of course complete the proof.
So let us prove that Y is a subgroup of G. Clearly, Y ̸= ∅. Let y1 , y2 ∈ Y . Then
y1 N, y2 N ∈ X, so (y1 N )(y2 N )−1 = (y1 N )(y2−1 N ) = y1 y2−1 N ∈ X, since X is a subgroup of
G/N . Thus, y1 y2−1 ∈ Y by definition of Y . Hence, Y is a subgroup by Lemma 1.12.

17
18
Chapter 2

Permutation groups and group

actions

2.1 Permutation groups

In this section, we look at a very important class of groups called permutation groups. We
begin with the definition.

Definition 2.1. Let X be any set. A subgroup of Sym(X) is called a permutation group on
X.

So for example, Sym(X) itself is a permutation group on X.

Next, we would like to define two important classes of examples of permutation groups
(the dihedral and alternating groups), but first we need some preparatory results about per-
mutations.

Definition 2.2. Let X be a set.

(i) For g ∈ Sym(X), the support of g is the set

supp(g) := {x ∈ X : g(x) ̸= x} ⊆ X.

(ii) Let G be a permutation group on X. The support of G is the set

supp(G) := {x ∈ X : g(x) ̸= x for some g ∈ G} ⊆ X.

Remark. If G = ⟨g⟩ ≤ Sym(X), then supp(⟨g⟩) = supp(g). Note also that if

g = (a1 , . . . , am1 ) . . . (amt−1 +1 , . . . , amt )

is a product of disjoint cycles, then we can read off supp(g): we have

supp(g) = {a1 , . . . , am1 , am1 +1 , . . . , amt−1 +1 , . . . , amt }.

Exercise. Prove that if G is a permutation group on X, and H ≤ G, then supp(H) ≤

supp(G). Deduce that if G1 and G2 are permutation groups on X and supp(G1 )∩supp(G2 ) =
∅, then G1 ∩ G2 = {1G }.

19
Theorem 2.3. Let X be a finite set.

(i) Disjoint cycles in Sym(X) commute.

(ii) If f = (a1 , . . . , ar ) ∈ Sym(X) is a cycle of length r, then |f | = r. More generally, if

f ∈ Sym(X) and f = f1 . . . fm is a decomposition of f into a product of disjoint cycles,
then |f | = lcm(|f1 |, . . . , |fm |).

(iii) Let f = (a1 , . . . , ar ) be a cycle in Sym(X), and let g ∈ Sym(X). Then

g
f = (g(a1 ), . . . , g(ar )).

Proof. Non-examinable
(i) Suppose that f = (a1 , . . . , ar ) and g = (b1 , . . . , bs ) are disjoint cycles in Sym(X). We just
need to prove that (f ◦ g)(x) = (g ◦ f )(x) for all x ∈ X. So let x ∈ X. Suppose first that
x ∈ {b1 , . . . , bs }. Then g(x) ∈ {b1 , . . . , bs } by definition, so g(x) ̸∈ {a1 , . . . , ar } since the cycles
are disjoint. Thus, (f ◦ g)(x) = f (g(x)) = g(x). Also, x ̸∈ {a1 , . . . , as }, so f (x) = x. Hence,
(g ◦ f )(x) = g(f (x)) = g(x).
Similarly, if x ̸∈ {b1 , . . . , bs }, then the same argument gives (f ◦ g)(x) = f (g(x)) = f (x) =
(g ◦ f )(x), as needed.
(ii) We already noted in Chapter 1 that a cycle that if f = (a1 , . . . , ar ) is a cycle of length r,
then |f | = r.
Next, let us consider the general case. Write f = f1 . . . fm , where the fi are disjoint cycles.
We will prove that |f | = lcm(|f1 |, . . . , |fm |) by induction on m. If m = 1, then |f | = |f1 | by
the case above.
So assume that m > 1, and that the result holds for smaller values of m. Set g :=
f1 . . . fm−1 . We have |g| = lcm(|f1 |, . . . , |fm−1 |) by the inductive hypothesis. Also, if we write
fi = (ai,1 , . . . , ai,ti ), then supp(g) = {ai,j : 1 ≤ i ≤ m − 1, 1 ≤ j ≤ ti }, while supp(fm ) =
{am,1 , . . . , am,t }. Thus, ⟨g⟩ ∩ ⟨fm ⟩ = {1G } by the exercise above. We deduce from Lemma
1.9(iv) and (i) above that

|f | = lcm(|g|, |fm |) = lcm(lcm(|f1 |, . . . , |fm−1 |), |fm |) = lcm(|f1 |, . . . , |fm |)

as needed.
(iii) Let f = (a1 , . . . , ar ) be a cycle in Sym(X), where a1 , . . . , ar are distinct elements in X.
Let g be another element of Sym(X). We need to prove that
g
f = (g(a1 ), . . . , g(ar )).

For 1 ≤ i ≤ r, define bi := g(ai ). Then, if 1 ≤ i ≤ r − 1, we have

(g f )(bi ) = (gf g −1 )(bi )

= g(f (g −1 (g(ai ))))
= g(f (ai ))
= g(ai+1 ) = bi+1 .

In the same way, (g f )(br ) = b1 . Finally, suppose that b ∈ X \ {b1 , . . . , br }, and let c ∈ X
such that g(c) = b. If f (c) ̸= c, then c lies in {a1 , . . . , ar }, by definition of a cycle. But then

20
b = g(c) = g(ai ) = bi for some i, contradicting our choice of b. Thus, we must have f (c) = c.
It follows that:
(g f )(a) = (gf g −1 )(a)
= g(f (g −1 (g(c))))
= g(f (c))
= g(c) = b.
Thus, g f is the cycle (b1 , . . . , br ), as claimed.

We can now define the dihedral groups.

Definition 2.4. Let n ∈ N with n ≥ 3, and set X := {1, . . . , n}. Define the permutations
σ, τ ∈ Sym(X) by
σ := (1, . . . , n) and
⌊n/2⌋
Y
τ := (i, n − i + 1).
i=1

The dihedral group of order 2n, written D2n , is the subgroup of Sym(X) defined by
D2n := ⟨{σ, τ }⟩.
For example, if n = 8, then
D2n := ⟨{(1, 2, 3, 4, 5, 6, 7, 8), (1, 8)(2, 7)(3, 6)(4, 5)}⟩;
while if n = 7, then
D2n := ⟨{(1, 2, 3, 4, 5, 6, 7), (1, 7)(2, 6)(3, 5)}⟩.
Theorem 2.5. Let n ≥ 3, and let D2n = ⟨σ, τ ⟩ be as defined in Definition 2.4. Then
(i) |D2n | = 2n;
(ii) ⟨σ⟩ is a normal subgroup of D2n of order n. In particular, D2n is not simple.
Proof. Part (i) is proved in Homework 2, so we just prove (ii). By Theorem 2.3(iii), τ σ =
(n, n − 1, . . . , 1) = σ −1 . Thus, τ lies in the normaliser of ⟨σ⟩. Since ⟨σ⟩ ≤ ND2n (⟨σ⟩) by
Homework 1, we have G = ⟨σ, τ ⟩ ≤ ND2n (⟨σ⟩). Thus, G = ND2n (⟨σ⟩), that is, ⟨σ⟩ ⊴ G. Since
|σ| = n by Theorem 2.3(ii), the result follows.

Remark. In the proof of Theorem 2.5, we implicitly used the following fact, which is fre-
quently used in group theory: if H and K are two subgroups of a group G, with H = ⟨A⟩
finite and K = ⟨B⟩, for some subsets A and B of G, then K lies in NG (H) if and only if
b a ∈ H for all a ∈ A and all b ∈ B. This is because of two things:

(1) g ∈ A means that g = aϵ11 . . . aϵmm for some ai ∈ A, ϵi ∈ {±1}, and n ∈ N. Thus,
b
g =b (aϵ11 . . . aϵmm ) = (b a1 )ϵ1 . . . (b am )ϵm
by Q13 on Homework 1. Thus, for k ∈ K, we have khk −1 ∈ H for all h ∈ H if and
only if kak −1 ∈ H for all a ∈ A. (Then, since H is finite, kHk −1 ⊆ H if and only if
kHk −1 = H.)

21
 (2) If L is a subgroup of G, then B ⊂ L if and only if ⟨B⟩ ≤ L (see Question 4 on Homework
2).

Next, we have the alternating groups.

Definition 2.6. Let X be a finite set.

(i) Let f ∈ Sym(X), and write f = f1 . . . fr where the fi are disjoint cycles. We say that
f is even if the number of cycles of even length in {f1 , . . . , fr } is even (this includes the
trivial element (), since it has 0 cycles of even length). Otherwise, f is said to be odd.

(ii) The set Alt(X) := {f : f is even} is called the alternating group of degree n.

As the name suggests, the alternating group is in fact a group. We are going to prove this
in Homework 2, but we will state the result here.

Proposition 2.7. Let X be a finite set. Then Alt(X) is a subgroup of Sym(X), and it has
order |X|!/2. (In other words, |Sym(X) : Alt(X)| = 2.

Proof. See Homework 2.

Proposition 2.8. Let X and Y be finite sets. If |X| = |Y |, then the groups Sym(X) and
Sym(Y ) are isomorphic.

Proof. Let ϕ : Y → X be a bijection. Then it is an easy exercise to show that the map
θ : Sym(X) → Sym(Y ) defined by θ(f ) = ϕ−1 ◦ f ◦ ϕ is an isomorphism.

We finish this section by noting that the notation Sn is used to denote the group Sym({1, . . . , n}).
(So by Proposition 2.8, Sn is isomorphic to Sym(X) whenever |X| = n.)

2.2 Group Actions

When groups were first defined by Galois in the nineteenth century, the motivation was to have
algebraic objects which measured the symmetry of various different mathematical structures.
In this section, we will make this idea rigorous, by defining what it means for a group to “act
(as symmetries)” on a set X.

Definition 2.9. Let G be a group and X a set. An action of G on X is a map · : G×X → X,

which satisfies the properties:

(i) (gh) · x = g · (h · x) for all g, h ∈ G, x ∈ X;

(ii) 1G · x = x for all x ∈ X.

If this case, we also say that G acts on X, and that X is a G-set.

Example 2.10. Three important group actions, which we will frequently consider in this
course, are as follows.

(1) Action of a group on itself by left multiplication. Let G be a group, let X = G

and define the action . : G × X → X by g.x = gx. Then

22
 (i) (gh).x = ghx and g.(h.x) = g.(hx) = ghx, so (gh).x = g.(h.x) for all g, h ∈ G,
x ∈ X.
(ii) 1G .x = 1G x = x for all x ∈ X.

Thus, . is a group action!

(2) Action of a group on itself by conjugation. Let G be a group, let X = G and

define the action . : G × X → X by g.x = g x = gxg −1 . Then

(i) (gh).x = ghx(gh)−1 = ghxh−1 g −1 and g.(h.x) = g.(hxh−1 ) = ghxh−1 g −1 , so

(gh).x = g.(h.x) for all g, h ∈ G, x ∈ X.
(ii) 1G .x = 1G x1−1
G = 1G x1G = x for all x ∈ X.

So again, . is a group action. The orbit of x ∈ G under this action is written G x, and
is called the conjugacy class of x in G.

(3) Action of a group on the set of left cosets of a subgroup. Let G be a group,
let H be a subgroup of G, and let X = G/H = {gH : g ∈ G} be the set of left cosets
of H in G. Define the action . : G × X → X by g.xH = gxH. Then arguing as in the
previous examples, we see that . is a group action.
Remark. We have actually defined a left action of G on X. A right action can be defined
analogously as a map X × G → X.
There is a strong link between group actions and the symmetric groups Sym(X), which
is encoded in the following proposition.
Proposition 2.11. Let · be an action of the group G on the set X. For g ∈ G, define the
map ϕ(g) : X → X by ϕ(g)(x) = g · x. Then ϕ(g) ∈ Sym(X), and ϕ : G → Sym(X) is a
homomorphism.
Proof. Let g, h ∈ g, and x ∈ X. Then

ϕ(gh)(x) = (gh).x = g.(h.x) = ϕ(g)(h.x) = (ϕ(g) ◦ ϕ(h))(x)

so ϕ(gh) = ϕ(g) ◦ ϕ(h).

This leads naturally to the definition of the kernel and image of an action.
Definition 2.12. Let · be an action of a group G on a set X.
(i) The kernel of the action ·, denoted ker(G, X, ·), is defined to be the kernel ker(ϕ) of the
homomorphism ϕ : G → Sym(X) defined in Proposition 2.11. So

ker(G, X, ·) = { g ∈ G | g · x = x for all x ∈ X }.

(ii) The image of the action ·, denoted Im(G, X, ·), is defined to be the image Im(ϕ) of the
homomorphism ϕ : G → Sym(X) defined in Proposition 2.11. So

Im(G, X, ·) = {ϕ(g) | g ∈ G}.

(iii) The action · is said to be faithful if ker(G, X, ·) = {1G }; and trivial if ker(G, X, ·) = G.

23
Example. (1) If G acts on itself by left multiplication, then it is easy to see that ker(G, X, ·) =
{1G }.
(2) If G acts on itself by conjugation, then the action is trivial if and only if gxg −1 = x for
all g, x ∈ G, i.e. if and only if G is abelian.
(3) If G acts on G/H for a subgroup H ≤ G, then the action is trivial if and only if
gH = H for all g ∈ G, i.e. if and only if G = H. So if H is a proper subgroup of G,
then ker(G, G/H, ·) is a proper normal subgroup of G. This point will be very useful!
If · : G × X → X is a faithful action of a group G on a set X, then the First Isomorphism
Theorem implies that G ∼ = G/ ker(G, X, ·) ∼ = Im(G, X, ·) ≤ Sym(X). We state this as a
proposition:
Proposition 2.13. If · is a faithful action of G on X, then G is isomorphic to a subgroup of
Sym(X).
We now come to two very important definitions: the orbit of a point in X; and the
stabiliser of a point in X.
Definition 2.14. Let · be an action of a group G on a set X. Let x ∈ X.
(i) The orbit of x, which is denoted by by OrbG (x) is
OrbG (x) := { g · x | g ∈ G }.

(ii) The stabiliser of x in G, which is denoted StabG (x) is StabG (x) := { g ∈ G | g · x = x }.

Proposition 2.15. Let G act on X and x ∈ X. Then
(i) StabG (x) is a subgroup of G;
(ii) ∩x∈X StabG (x) is the kernel of the action of G on X.
Proof. See Homework 2, Question 8.

The next theorem is a very fundamental result in group theory.

Theorem 2.16. (The Orbit-Stabiliser theorem) Let the finite group G act on X, and
let x ∈ X. Then |G| = |OrbG (x)| |StabG (x)|.
Proof. Let Y := G/StabG (x) be the set of left cosets of StabG (x) in G. Define ϕ : Y →
OrbG (x) by gStabG (x) → g.x. We claim that ϕ is a bijection.
To see this, we first need to prove that ϕ is well-defined. So assume that g1 StabG (x) =
g2 StabG (x). Then g2 = g1 s, for some s ∈ StabG (x) by Lemma 1.15. Thus,
ϕ(g2 StabG (x)) = g2 .x = (g1 s).x = g1 .(s.x) = g1 .x = ϕ(g1 StabG (x)).
so ϕ is well-defined.
Finally, we will show that ϕ is a bijection. Suppose that ϕ(g1 StabG (x)) = ϕ(g2 StabG (x)).
Then g1 .x = g2 .x, so (g2−1 g1 ).x = x. Hence, g2−1 g1 ∈ StabG (x), so StabG (x) = g2−1 g1 StabG (x).
Hence, g2 StabG (x) = g1 StabG (x). This proves that ϕ is injective.
To see that ϕ is surjective, let y ∈ OrbG (x). Then by definition, y = g.x for some g ∈ G.
Then ϕ(gStabG (x)) = g.x. Thus, ϕ is surjective, so ϕ is bijective.
Since |Y | = |G|/|StabG (x)|, the proof is complete.

24
 Let us compute the orbits and stabilisers of the group actions from Examples 2.10.

Example. (1) Let G act on X := G by left multiplication, and let x ∈ X. Then, if y ∈ X,

we have y = (yx−1 ).x, so y ∈ OrbG (x). Thus, OrbG (x) = X for all x ∈ X. Also,
for g ∈ G, g.x = x if and only if gx = x, which is true if and only if g = 1G . Thus,
StabG (x) = {1G } for all x ∈ G.

(2) Let G act on X := G by conjugation, and let x ∈ X. Then OrbG (x) = {gxg −1 : g ∈
G}, which is called the conjugacy class of x in G, and written G x. Also, for g ∈ G,
g.x = x if and only if gxg −1 = x, which is true if and only if g lies in the centraliser
CG (x) := {g ∈ G : gxg −1 = x} of x in G. Thus, StabG (x) = CG (x) for all x ∈ G. We
also remark here that ker(G, X, ·) is the centre of G, written Z(G) (so Z(G) := {g ∈
G : gxg −1 = x for all x ∈ G}).

(3) Let H be a subgroup of G, and let G act on X := G/H by left multiplication. Then
StabG (xH) is the set

{g ∈ G : gxH = xH} = {g ∈ G : x−1 gxH = H} = {g ∈ G : x−1 gx ∈ H} = xHx−1 = x H.

Also, if x1 H, x2 H ∈ X, then (x1 x−1

2 ).x2 H = x1 H, so OrbG (xH) = X for all xH ∈ X.

The following is arguably the most useful consequence of the Orbit-Stabiliser theorem.

Corollary 2.17. Let G be a finite group acting on a set X, and let x ∈ X. Then
(i) Either OrbG (x) = OrbG (y) or OrbG (x) ∩ OrbG (y) = ∅.
(ii) The set {OrbG (x) : x ∈ X} forms a partition of X.
(iii) |OrbG (x)| divides |G|.

Proof. Define a relation ∼ on X by x ∼ y if and only if y = g.x for some g ∈ G. It is

easy to see that ∼ is an equivalence relation on X, and that the equivalence classes are
{OrbG (x) : x ∈ X}. Parts (i) and (ii) follow.
Part (iii) follows immediately from the Orbit-Stabiliser theorem and Lagrange’s theorem
since |OrbG (x)| = |G|/|StabG (x)|.

This last consequence (Corollary 2.17(iii)) is highly non-obvious! In particular, it shows

that the size of a conjugacy class |G x| in a finite group G (x ∈ G) divides the order of G. It
gives a number of pretty amazing applications, which we now discuss.

Applications of the Orbit-Stabiliser theorem from Example 2.10(1)

Here, we will consider the action of G on itslef by left multiplication.

Theorem 2.18. (Cayley’s Theorem) Every finite group G is isomorphic to a subgroup of

Sym(X) for some set X.

Proof. Let X = G. The kernel of the action of G on X by left multiplication is the set
{g ∈ G : gx = x for all x ∈ X}. But if g is an element of g with gx = x for all x ∈ X, then
g1G = 1G , so g = 1G . Thus, the kernel is trivial. The result then follows from Proposition
2.13.

25
Applications of the Orbit-Stabiliser theorem from Example 2.10(2)
In this part of Section 2.2, we will consider the action of a group G on itself by conjugation.
Theorem 2.19. Let p be a prime, let n ∈ N, and let G be a finite group with |G| = pn . Then
|Z(G)| > 1.
Proof. Corollary 2.17 shows that |G x| is a power of p for all x ∈ G. Also, by definition,
Z(G) = {x ∈ G : |G x| = 1}. Suppose that |Z(G)| = 1. Then Corollary 2.17(ii) implies that

pn = |G| = 1 + pa1 + . . . + pak

for some ai ∈ N. This is a contradiction however, since the left hand side of the equation
above is divisible by p (i.e. congruent to 0 modulo p), while the right hand side is congruent
to 1 modulo p.

Corollary 2.20. Let p be a prime, let n ∈ N, and let G be a finite group with |G| = pn .
(i) If n = 2, i.e. |G| = p2 , then G is abelian.

(ii) If n = 3, i.e. |G| = p3 , then either G is abelian, or |Z(G)| = p.

Proof. We first prove (i). So assume that |G| = p2 . In Homework 2, Question 1, we proved
that if G/Z(G) is cyclic, then G is abelian. Now, by Lagrange’s theorem and Theorem 2.19,
the centre of G has order either p or p2 . But since every group of prime order is cyclic,
|Z(G)| = p would imply that G/Z(G) is cyclic, so G is abelian by the result mentioned above.
If |Z(G)| = p2 then G = Z(G), so again, G is abelian.
Part (ii) is similar: By Lagrange’s theorem and Theorem 2.19, the centre of G has order
either p, p2 , or p3 . Suppose that G is non-abelian. Then by arguing as in part (i), we see
that Z(G) cannot have order p2 or p3 . Thus, |Z(G)| = p, and this completes the proof.

We finish with the following:

Theorem 2.21. (Cauchy’s theorem) Let G be a finite group, and let p be a prime divisor
of |G|. Then G contains an element of order p. In fact, the number of elements of G of order
p is congruent to −1 modulo p.
Proof. Define
X := {(x1 , . . . , xp ) : xi ∈ G and x1 . . . xp = 1G }.
Note that

(x1 , . . . , xp ) ∈ X ⇒ x1 . . . xp = 1G
⇒ x2 x3 . . . xp x1 = x−1 −1
1 (x1 . . . xp )x1 = x1 1G x1 = 1G
⇒ (x2 , x3 , . . . xp , x1 ) ∈ X.

Repeatedly applying the above implies that if (x1 , . . . , xp ) is in X, then

(xi , . . . , xp , x1 . . . , xi−1 ) ∈ X

for all i ≤ p.
Now, let C := ⟨(1, . . . , p)⟩ ≤ Sp be the subgroup of the symmetric group Sp generated by
the p-cycle (1, . . . , p). Define an action . : C × X → X by f.(x1 , . . . , xp ) := (xf (1) , . . . , xf (p) )

26
for f ∈ C, (x1 , . . . , xp ) ∈ X. By the paragraph above and the fact that (f g)(i) = f (g(i)) for
all f, g ∈ Sp , i ∈ {1, . . . , p}, we have that . : C × X → X is a well-defined action.
The idea of the proof is now to use Corollary 2.17 to partition our set X into a disjoint
union of C-orbits. To do this, we will split our orbits into three types:
(1) The p-tuple (1G , . . . , 1G ) of course lies in X. Moreover, f := (1, 2 . . . , p) lies in the
| {z }
p
stabiliser of C of (1G , . . . , 1G ). Since C := ⟨f ⟩ and stabilisers are subgroups, we deduce
| {z }
p
that StabG ((1G , . . . , 1G )) = C. Hence, setting R := OrbC ((1G , . . . , 1G )), we have |R| =
1 by the orbit stabiliser theorem.

(2) Let x be an element of G of order p. Then (x, . . . , x) lies in X, and arguing as in (1)
| {z }
p
above, we have Sx := OrbC ((x, . . . , x)) has size 1. We claim that if S any C-orbit in
X of size 1, then either S = R as in (1), or S = Sx for some x ∈ G with |x| = p.
To see this, suppose that S is such an orbit, and let (x1 , . . . , xp ) ∈ X. Then the
Orbit-Stabiliser theorem implies that StabC ((x1 , . . . , xp )) = C. Thus, recalling that
f = (1, 2, . . . , p) ∈ C, we have

(x1 , . . . , xp ) = f.(x1 , . . . , xp ) = (xp , x2 , . . . , xp−1 ).

That is, x1 = . . . = xp . Thus, xp1 = 1G , so |x1 | ∈ {1, p} by Lemma 1.9(i). Our claim
follows.

(3) Let T1 , . . . , Tn be the distinct orbits in X of size greater than 1. Since |C| = p, Corollary
2.17 implies that |Ti | = p for all i.
We have now listed all of the X-orbits. In order to complete the proof, we need to know
something about |X|. We claim that

|X| = |G|p−1 . (2.1)

To see this, we define a map ϕ : X → Gp−1 by ϕ((x1 , . . . , xp )) := (x1 , . . . , xp−1 ). Then ϕ is

injective, since
(x1 , . . . , xp ) ∈ X ⇒ xp = (x1 . . . xp−1 )−1 .
Furthermore, ϕ is surjective, since

(x1 , . . . , xp−1 ) ∈ Gp−1 ⇒ (x1 , . . . , xp−1 , y) ∈ X,

where y := (x1 . . . xp )−1 . Thus, |X| = |Gp−1 | = |G|p−1 , as claimed at (2.1).

We can now complete the proof. Let P := {x ∈ G : |x| = p}. Then Corollary 2.17 and
(2.1) implies that
X n
X
p−1
|G| = |X| = |R| + |Sx | + |Ti |
x∈P i=1
= 1 + |P| + np.

Since p | |G|, we deduce that |P| ≡ −1 (mod p). In particular, |P| > 0, that is, G contains
an element of order p. This completes the proof.

27
Applications of the Orbit-Stabiliser theorem from Example 2.10(3)
In this part of Section 2.2, we will consider the action of G by left multiplication on the set
of left cosets of a subgroup. We will see two very useful applications of the Orbit-Stabiliser
theorem coming from this action. The first is as follows. Recall that the product HK :=
{hk : h ∈ H, k ∈ K} of two subgroups in a group is not necessarily a subgroup. However:

Proposition 2.22. Let G be a finite group, and let H and K be subgroups of G. Then

|K||H|
|HK| = |KH| =
|H ∩ K|

Proof. Let X := G/H be the set of left cosets of H in G, and consider the action of K on X
given by k.(gH) := kgH. This is the restriction of the action of G on X by left multiplication,
so it is indeed an action. We claim first that
|KH|
is equal to the size of the K-orbit OrbK (H) of H ∈ X. (2.2)
|H|

To see this, note that by Corollary 1.17, if k1 , k2 ∈ K, then k1 H = k2 H or k1 H ∩ k2 H = ∅.

Thus, if we let k1 H, . . . , ks H be the distinct elements of the set {kH : k ∈ K}, then we have
s
G
KH = ki H.
i=1

Thus, |KH| = si=1 |ki H| = si=1 |H| = s|H|, so |KH|/|H| = s.

P P

On the other hand, if gH ∈ X, then gH lies in the K-orbit of H if and only if gH = kH

for some k ∈ K. Thus, OrbK (H) = {kH : k ∈ K} = {k1 H, . . . , ks H}. Hence, |OrbK (H)| =
s = |KH|/|H|, as claimed at (2.2).
Finally, we will prove that StabK (H) = H ∩ K. To see this, note that if k ∈ K then

k ∈ StabK (H) ⇔ kH = H ⇔ k ∈ H.

Thus, StabK (H) = H ∩ K. The result now follows from the Orbit-Stabiliser theorem.

Proposition 2.23. Let G be a finite group, and let H and K be subgroups of G. Then
|G : H ∩ K| ≤ |G : H||G : K|.

Proof. Let X1 := G/H and X2 := G/K be the set of left cosets of H in G and K in G
respectively. Let X := X1 × X2 be the cartesian product of X1 and X2 . Then G acts on X1
and X2 via g.(y1 H, y2 K) := (gy1 H, gy2 K). Let x := (H, K) ∈ X. For g ∈ g, we have

g ∈ StabG (x) ⇔ gH = H and gK = K and

g ∈ StabG (x) ⇔ g ∈ H and g ∈ K and

by Lemma 1.15. Thus, StabG (x) = H ∩ K. It follows from the Orbit-Stabiliser theorem that

|G : H ∩ K| = |OrbG (x)| ≤ |X| = |G : H||G : K|

This completes the proof.

28
2.3 Fixed points
We close this section with an interesting application of group actions. First, we need a
definition.

Definition 2.24. Let G be a group acting on a set X, and let g ∈ G.

1. We say that x ∈ X is a fixed point, or that g fixes x, if g.x = x. The set of all fixed
points for g ∈ G is written fixX (g) := {x ∈ X : g.x = x}.

2. We say that g ∈ G is fixed point free if fixX (g) = ∅.

We also need a lemma. This is sometimes referred to in the literature as “not Burnside’s
lemma”.

Lemma 2.25. (the “not Burnside” lemma) Let G be a finite group acting on a finite
set X, and let r := |{OrbG (x) : x ∈ X}| be the number of orbits of G. Then

1 X
r= |fixX (g)|.
|G|
g∈G

Proof. The proof uses a “double counting” argument. Indeed, we will determine the cardi-
nality of the set
Λ := {(g, x) ∈ G × X : g.x = x}

in two different ways.

P P
First, |Λ| = g∈G |fixX (g)|. Secondly, |Λ| = x∈X |StabG (x)|. Now, write OrbG (x1 ), . . . , OrbG (xr )
for the distinct orbits in X. Then by Corollary 2.17, we have X = OrbG (x1 ) ⊔ . . . ⊔ OrbG (xr ).
Also, |StabG (xi )| = |StabG (x)| whenever x ∈ OrbG (xi ) by the Orbit-Stabiliser theorem. Thus
X X
|fixX (g)| = |StabG (x)|
g∈G x∈X
X
= |StabG (x)|
x∈X
r
X
= |StabG (xi )||OrbG (xi )|
i=1
r
X
= |G| = r|G|.
i=1

The result follows.

One of the nicest consequences of Lemma 2.25 is the following.

Corollary 2.26. Let G be a finite group acting on a finite set X with |X| > 1, and suppose
that G has precisely 1 orbit (G is said to be transitive in this case). Then G contains a fixed
point free element.

29
Proof. Suppose that G contains no fixed point free elements. Then |fixX (g)| ≥ 1 for all g ∈ G.
Thus, since fixX (1G ) = |X|, Lemma 2.25 implies that
X X
|G| = |fixX (g)| = |X| + |fixX (g)| ≥ |X| + |G| − 1.
g∈G g∈G\{1G }

This is a contradiction, since |X| > 1. Thus, there must exist some g ∈ G with |fixX (g)| =
0.

Remark. If we delete the words “finite” from Corollary 2.26, then the result no longer holds.
See Homework 3 for more details..

30
Chapter 3

The Sylow theorems

Here, and in Algebra II, we studied Lagrange’s theorem, which states that if H is a subgroup
of a finite group G, then the order of H divides the order of G. Whenever one has a theorem
like this in mathematics, the natural follow up question is: Does the converse hold? That is,
if G is a finite group and r is a divisor of |G|, then does G contain a subgroup H of order r?
This is true when the group is cyclic (see Homework 2!), but is is true in general?
No! For example, if G is a non-abelian finite simple group, then G has no subgroup of order
|G|/2. (Such a subgroup H would have index 2, and hence would be a proper normal subgroup
of G. But |G| > 2, since G is non-abelian, so 1 < |H| < |G|, contradicting the simplicity of
G).

So the converse to Lagrange’s theorem is false in general! However, some amazing theorems
of Sylow show that if r is a certain prime power divisor of G, then G does have a subgroup
of order r. These theorems will be the focus of this section.
Throughout this chapter, let G denote a finite group and p a prime.

Notation : We will use |G|p to denote the highest power of p that divides |G|. That is if
|G| = pn · m where gcd(p, m) = 1, then |G|p = pn .

Definition 3.1. Let H ≤ G.

(i) We say that H is a p-subgroup of G, if |H| is a power of p.

(ii) Let P ≤ G, and suppose that |P | = |G|p . Then P is called a Sylow p-subgroup of G.

(iii) We write Sylp (G) := {P ≤ G : P a Sylow p-subgroup of G} for the set of Sylow p-
subgroups of G.

Remark. By definition, if p ∤ |G|, then {1G } is a Sylow p-subgroup of G. All of our appli-
cations will concern the case p | |G|, but for some proofs, it is useful to bare this (somewhat
uninteresting) fact in mind.

Example. The following are some examples of p-subgroups and Sylow p-subgroups.

(1) Take G = S4 . Then |G| = 4! = 23 · 3, so |G|2 = 23 and |G|3 = 3. We therefore have:

(a) {1, (123), (321)} is a Sylow 3-subgroup of G.

31
 (b) V4 := {1, (12)(34), (13)(14), (14)(23)} is a 2-subgroup of G, but not a Sylow 2-
subgroup of G.
(c) D8 = ⟨σ, τ ⟩, where σ := (1, 2, 3, 4) and τ := (1, 4)(2, 3) (see Definition 2.5) is a
Sylow 2-subgroup of G.
(d) A4 is not a p-subgroup of G for any prime p; while {1G } is a p-subgroup of G for
all primes p.

(2) Let G = Cpn ·m where p is a prime and gcd(p, m) = 1. From Algebra 2 (see also HW2)
we know that since pn divides |G|, there exists a unique subgroup H ≤ G of order pn .
By definition, H is a Sylow p-subgroup of G. Thus in this example, G has a unique
Sylow p-subgroup.

The following result is due to Ludwig Sylow and dates back to 1872. It is fundamental to
finite group theory.

Theorem 3.2. (The Sylow theorems) Let G be a finite group, and let p be a non-trivial
prime divisor of |G|. Then the following conditions hold:

1. G has at least one Sylow p-subgroup.

2. All the Sylow p-subgroups of G are conjugate.

3. Any p-subgroup of G is contained in a Sylow p-subgroup of G.

4. The number of Sylow p-subgroups of G is congruent to 1 modulo p.

There are various proofs of this result. You can find one of them in Professor Holt’s notes
for Algebra II. In this course we will give two proofs.

3.1 Wielandt’s Proof of Sylow Theorems 1 and 4

Our first proof is due to Wielandt. Before we begin, we require the following number theoretic
lemma.

Lemma 3.3. Let p be prime, and let n and m be integers with gcd(m, p) = 1. Then

(i) p divides pi for all 1 ≤ i ≤ p − 1; and

n

(ii) ppnm ≡ m (mod p).

Proof. (i) By definition,  

p p(p − 1) . . . p − (i − 1)
= .
i i(i − 1) . . . 1
Let a := (p − 1) . . . p − (i − 1) and b := i(i − 1) . . . 1 = i!, so that pi = pa/b. Since b is a

product of integers less than p, all prime divisors of b are less than p. Thus, p ∤ a. Thus,

since
b pi = pa; pi is an integer; and p does not divide b, we must have that p divides pi .

32
(ii) Let F be the field Z/pZ. Then we may think of F as F = {0, 1 . . . , p−1}, where addition in
F is defined as x + y := z, where z is the unique element of {0, . . . , p − 1} satisfying x + y ≡ z
(mod p); and multiplication is defined in the same way: xy := w, where w is the unique
element of {0, . . . , p − 1} satisfying xy ≡ w (mod p).
Consider the associated polynomial ring F[X] (where X is an indeterminate), and the
polynomial (1 + X)p ∈ F[X]. By the binomial theorem, we have
p  
X
p p
(1 + X) = Xi
i
i=0
   
p 0 p
= X + Xp
0 p
= 1 + X p, (3.1)

where the penultimate equality above follows from (i), since multiples of p are 0 in the field
F.
Applying (3.1) again with X replaced by X p , we get
2 2
(1 + X)p = ((1 + X)p )p = (1 + X p )p = 1 + (X p )p = 1 + X p .

Repeatedly applying the above then yields

n n
(1 + X)p = 1 + X p .

Hence,
nm n
(1 + X)p = (1 + X p )m . (3.2)

Using the binomial theorem to expand the left and right hand sides of (3.2), we get
nm 
pX m  
pn m

i
X m n
X = Xp j. (3.3)
i j
i=0 j=0

n
Comparing the coefficients of X p j on the left and right hand sides above, we get
 n   
p m m
n
= in the field F whenever 0 ≤ j ≤ m.
p j j

In other words,
pn m
   
m
n
≡ (mod p) whenever 0 ≤ j ≤ m.
p j j
Applying the above with j = 1 then gives the result.

We are now ready to prove parts 1 and 4 of Sylow’s theorem.

Proposition 3.4. Let G be a finite group and p a prime divisor of |G|. Write |G| = pn m,
where gcd(p, m) = 1. Recall that Sylp (G) is the set of Sylow p-subgroups of G (i.e. the set
of subgroups of G of order pn ). Then we have |Sylp (G)| ≡ 1 (mod p). In particular, parts 1
and 4 of Sylow’s theorem hold.

33
Proof. We remark first that if we can prove that |Sylp (G)| ≡ 1 (mod p), then parts 1 and 4
of Sylow’s theorem immeidately follow. So it suffices to prove that |Sylp (G)| ≡ 1 (mod p).
So let us do this. Let
X = {S ⊆ G | |S| = pn }.
Then
pn · m
 
|X| = ,
pn
so |X| ≡ m (mod p) by Lemma 3.3(ii). In particular,

p ∤ |X|. (3.4)

Consider the action of G on X by left multiplication: if S ∈ X, then define

g · S := gS = {gs : s ∈ S}.

(Check that this really does define a group action – this is Q12 in Homework 2.)
Now, by Corollary 2.18(ii), the set {OrbG (S) : S ∈ X} is a partition of X. That is, there
exists S1 , . . . , St ∈ X such that

X = OrbG (S1 ) ⊔ . . . ⊔ OrbG (St ).

In particular,
t
X
|X| = |OrbG (Si )|. (3.5)
i=1

We want to study |X| modulo p, so we will now analyse the orbits OrbG (Si ) with the property
that p does not divide |OrbG (Si )|. Note first that by the equation above and (3.4), at least
one of the orbits OrbG (Si ) has the property that p ∤ |OrbG (Si )|. Let r be the number of orbits
OrbG (Si ) with the property that p ∤ |OrbG (Si )|, so that 1 ≤ r ≤ t. To make our notation a
little easier, we will assume, without loss of generality, that p ∤ |OrbG (Si )| for 1 ≤ i ≤ r; and
p | |OrbG (Si )| for r < i ≤ t.

We now prove two claims:

Claim 1: Fix 1 ≤ i ≤ r. Then there exists xi ∈ G such that StabG (xi Si ) = xi Si . In

particular, StabG (xi Si ) ∈ Sylp (G) for each 1 ≤ i ≤ r.
Proof: Fix 1 ≤ i ≤ r. First, fix si ∈ Si , and set xi := s−1
i and Ti := xi Si . Then 1G ∈ Ti . We
claim that StabG (Ti ) = Ti .
Note first that by Corollary 2.18(ii), we have OrbG (Ti ) = OrbG (Si ), since Ti = xi Si =
xi · Si ∈ OrbG (Si ) by definition of orbits.
Next, by the Orbit-Stabiliser theorem, we have p ∤ |OrbG (Si )| = |OrbG (Ti )| = |G :
StabG (Ti )|. Thus, since |G| = pn m = |G : StabG (Ti )||StabG (Ti )|, we must have that pn
divides |StabG (Ti )|. It follows in particular that pn ≤ |StabG (Ti )|.
On the other hand, if g ∈ StabG (Ti ), then gTi = Ti by definition. Since 1G ∈ Ti , we have
g = g1G ∈ gTi = Ti . Thus, StabG (Ti ) ⊆ Ti . Since |Ti | = pn ≤ |StabG (Ti )| by the previous
paragraph, we deduce that StabG (Ti ) = Ti , as required.

34
Claim 2: r = |Sylp (G)|.
Proof: Let Ti = xi Si , where xi is as in Claim 1.
Define a map f : {OrbG (Ti ) : 1 ≤ i ≤ r} → Sylp (G) by f (OrbG (Ti )) = Ti . By Claim 1,
StabG (Ti ) = Ti is a subgroup of G of order pn , so this map is well-defined. We will now prove
that f is a bijection.
First, we prove that f is one-to-one. Let 1 ≤ i, j ≤ r, and suppose that Ti = Tj . Then
Ti = Tj , so OrbG (Si ) = OrbG (Ti ) = OrbG (Tj ) = OrbG (Sj ), so Si = Sj , since we chose
S1 , . . . , St so that the orbits OrbG (S1 ), . . . , OrbG (St ) are distinct. Thus, f is one-to-one.
Finally, let P ∈ Sylp (G). Then P ∈ X, and StabG (P ) = P by Lemma 1.15. It follows
that p does not divide m = |G : P | = |G : StabG (P )| = |OrbG (P )|, so OrbG (P ) = OrbG (Ti )
for some i. In fact, since 1G ∈ Ti and OrbG (Ti ) = {xTi : x ∈ G}, Ti is the unique element of
OrbG (Ti ) containing 1G . Thus, we must have P = Ti = f (OrbG (Ti )). Thus, f is onto.
This prove that f is a bijection, so the proof of our claim is complete.
Finally, we have |X| ≡ m modulo p by Lemma 3.3(ii). Also, we have
r
X
|X| ≡ |OrbG (Ti )| (mod p).
i=1

Since |OrbG (Ti )| = |G : StabG (Ti )| = |G|/|Ti | = m by Claim 1; and r = |Sylp (G)| by Claim
2, we deduce that
m ≡ mSylp (G) (mod p).
Since p does not divide m, m has an inverse, say u, modulo p. Multiplying both sides of the
above equation by u then completes the proof.

3.2 Proof of Sylow Theorems 2 and 3

We are now going to prove the remaining parts of Sylow Theorems. In order to do that, we
will prove the following statement.

Proposition 3.5. Let G be a group and p a prime divisor of |G|. Let H ≤ G and let P be a
Sylow p-subgroup of G. Then there exists g ∈ G such that H ∩ gP g −1 is a Sylow p-subgroup
of H.

Proof. Consider the set X := G/P = {xP | x ∈ G} of left cosets of P in G. Then G acts on
X by left multiplication (Example 2.11(3)) and so H acts on X. Let us study the action of
H on X. Take any xP ∈ X. Then

StabH (xP ) = {h ∈ H | h(xP ) = xP } = {h ∈ H | x−1 hxP = P } =

= {h ∈ H | x−1 hx ∈ P } = {h ∈ H | h ∈ xP x−1 } = H ∩ xP x−1

Since P is a Sylow p-subgroup of G, |X| = |G : P | is co-prime to p. Let OrbH (x1 P ), . . . , OrbH (xs P )
denote the distinct orbits of H in its action on X. Then by Corollary 2.18(ii), we have
s
X
|X| = |OrbH (xi P )|.
i=1

35
If every such orbit has length divisible by p, then |X| would be divisible by p, a contradiction.
Hence, there exists xi ∈ G such that |OrbH (xi P )| is not-divisible by p.
Let Q := StabH (xi P ) = H ∩ xi P x−1i . Then Q is a p-subgroup, since it is a subgroup of
xi P x−1
i . Also, |H : Q| = |Orb (x
H i P )| is not divisible by p. Thus, Q is a Sylow p-subgroup of
H. This completes the proof.
Corollary 3.6. Parts 2 and 3 of Sylow’s theorem hold.
Proof. Proof of part 2 of Sylow’s theorem: Let H and P be Sylow p-subgroups of G. By
Proposition 3.5, there exists g ∈ G such that H ∩ gP g −1 is a Sylow p-subgroup of H. But
|H| is a power of p, so Sylp (H) = H. Thus, H = H ∩ gP g −1 . Since |H| = |gP g −1 |, it follows
that H = gP g −1 . This completes the proof of part 2 of Sylow’s theorem.
Proof of part 3 of Sylow’s theorem: Let H be a p-subgroup of G, and let P be Sylow p-
subgroup of G. By Proposition 3.5, there exists g ∈ G such that H ∩ gP g −1 is a Sylow
p-subgroup of H. But |H| is a power of p, so Sylp (H) = H. Thus, H = H ∩ gP g −1 ≤ gP g −1 ,
and so this completes the proof of part 3 of Sylow’s theorem.

3.3 Consequences of Sylow’s theorems

Let us introduce some notation that we will use from now on.

Notation : By Sylow Theorem 2, G acts on Sylp (G) by conjugation and for any P ∈ Sylp (G),
OrbG (P ) = Sylp (G). The stabiliser of P under this action is of course NG (P ), i.e.,
StabG (P ) = NG (P ) = {g ∈ G | gP g −1 = P }

Corollary 3.7. Let G be a finite group, p a prime divisor of |G|, and P ∈ Sylp (G). Then
the following conditions hold:
(i) |Sylp (G)| = |G : NG (P )|.
(ii) |Sylp (G)| divides |G|/|G|p .
(iii) P ⊴ G if and only if |Sylp (G)| = 1.
Proof. (i) By the Orbit-Stabiliser Theorem, |Sylp (G)| = |OrbG (P )| = |G : NG (P )|.
(ii) Since P ≤ NG (P ), by Lagrange’s Theorem, |NG (P )| = |P | · |NG (P ) : P |. Hence,
|Sylp (G)| = |G : NG (P )| = |NG|G| |G| |G| |G|
(P )| = |P |·|NG (P ):P | which divides |P | = |G|p .

(iii) Finally, P ⊴ G if and only if G = NG (P ) if and only if |Sylp (G)| = 1.

Corollary 3.8. Let G be a finite group, and let p be a prime divisor of |G|. Let
Fp (G) := {x ∈ G : x ̸= 1, and |x| is a power of p}.
S
(i) Fp (G) = P ∈Sylp (G) (P \ {1G }).

(ii) |Fp (G)| ≥ |G|p − 1, with equality if and only if |Sylp (G)| = 1 (i.e. if and only if G has a
normal Sylow p-subgroup).
(iii) If |G|p = p, then |Fp (G)| = |Sylp (G)|(p − 1).
Proof. See Homework 3, Questions 10 and 11.

36
3.4 Applications of Sylow’s theorems
We will now consider two nice applications of the Sylow theorems. We will call these applica-
tions the Sylow game in the lectures. For more examples, and more details of the strategies
for dealing with these applications, see Appendix B!

Application 1: Proving groups of a particular order are not simple (Version 1 of

the Sylow game)

1. Let G be a group of order 20. Can G be simple?

|G| = 22 · 5. By Sylow Theorem 4, |Syl5 (G)| ≡ 1 (mod 5). By Corollary 3.7(ii),
|Syl5 (G)| | 20
5 = 4 and so |Syl5 (G)| = 1. Using 3.7(iii), we conclude that G has a normal
Sylow 5-subgroup P of order 5. Hence, G has a proper normal subgroup (1 < |P | =
5 < 20), and so G cannot be simple.

2. Let G be a group of order 48. Can G be simple?

|G| = 24 · 3. By Sylow Theorem 4, |Syl2 (G)| ≡ 1 (mod 2). By Corollary 3.7(ii),
48
|Syl2 (G)| | 16 = 3. Therefore either |Syl2 (G)| = 1 or |Syl2 (G)| = 3. If |Syl2 (G)| = 1,
by Corollary 3.7(iii), G has a unique normal Sylow 2-subgroup P of order 16. Hence,
G has a proper normal subgroup (1 < |P | = 16 < 48), and so G cannot be simple.
If |Syl2 (G)| = 3, then Syl2 (G) has cardinality 3 and G acts on Syl2 (G) by conjugation
(non-trivially!). Therefore by Proposition 1.14, there exists a non-trivial homomorphism
of G into S3 , ϕ : G → S3 . By the First Isomorphism Theorem, G/ ker(ϕ) ∼ = Im(ϕ). Using
this together with Lagrange’s Theorem we obtain that |G|/| ker(ϕ)| = |Im(ϕ)|, and so
| ker(ϕ)| = |G|/|Im(ϕ)|. Now, Im(ϕ) is a subgroup of S3 whose order is greater than 1
(ϕ is non-trivial!). Therefore 1 < |Im(ϕ)| ≤ 6, and so 48/6 ≤ | ker(ϕ)| < 48. Therefore
ker(ϕ) is a non-trivial normal subgroup of G. Therefore G is not simple.

Application 2: Proving that a particular group is simple (Version 2 of the Sylow

game)
We can also use Sylow’s theoprems to prove that various groups are simple. The example
we will give is a fact that you proved in Algebra II. We will now prove it using the Sylow
theorems. First, we need the following easy consequence of Sylow’s theorems.

Corollary 3.9. Let G be a finite group, and let p be a prime divisor of |G|. Let

Fp (G) := {x ∈ G : x ̸= 1, and |x| is a power of p}.

Let N be a normal subgroup of G.

(i) If x ∈ N , then G x ⊆ N .

(ii) Suppose that p does not divide |G : N |. Then

(a) Sylp (N ) = Sylp (G); and

(b) Fp (G) = Fp (N ).

37
Proof. Part (i) follows since x ∈ N ⇒ g x ∈ g N = N for all g ∈ G.
For (ii), suppose that p does not divide |G : N |. Then all Sylow p-subgroups of G are
contained in N by Homework 3, Question 8(ii). Also, each Sylow p-subgroup of N is contained
in a Sylow p-subgroup of G, by Sylow’s theorem 3. Hence, Sylp (G) = Sylp (N ), so (a) holds.
Finally, part (b) follows immediately from part (a) and Corollary 3.8(i).

Proposition 3.10. Let G = A5 . Then G is a simple group.

Proof. Assume the contrary. Then there exists {1G } ≠ N ◁ G with N ̸= G. By Lagrange’s
Theorem |N | divides |G| = 60. Hence the prime divisors of |N | are among 2, 3 and 5.
Note first three facts about A5 :

(1) G has 24 elements of order 5 (to see this, just count the possible 5-cycles, which are the
only elements of order 5).

(2) G has 20 elements of order 3 (to see this, just count the possible 3-cycles, which are the
only elements of order 3).

(3) The only elements of order 2 in A5 are those of the form (a, b)(c, d), for a, b, c, d distinct
elements of {1, 2, 3, 4, 5}. There are 15 of them, and they are all conjugate in A5 (use
Theorem 2.3(iii) to see this).

We now distinguish our arguments into two cases.

Case 1: 3 divides |N | or 5 divides |N |. Let p ∈ {3, 5}, and suppose that p | |N |. Then
p ∤ |G : N |, so we can use Corollary 3.9(ii): We have Fp (G) = Fp (N ).
Suppose that p = 5. Then Fp (N ) = 24, so |N | ≥ 25. Since |N | divides 60 and is less than
60, it follows that |N | = 30.
Suppose that p = 3. Then Fp (N ) = 20, so |N | ≥ 21. Since |N | divides 60 and is less than
60, it follows again that |N | = 30.
We have therefore shown that if either 3 or 5 divide N , then both 3 and 5 divide N ;
|N | = 30; and F3 (N ) = 20, F5 (N ) = 24. But then |N | = 30 > 20 + 24, a contradiction.
Case 2: Neither 3 nor 5 divides |N |. Then |N | divides 4 by Lagrange’s theorem. By
Cauchy’s theorem, we can choose x ∈ N of order 2. We then use Corollary 3.9(i) and Fact
(3) above: 4 = |N | ≥ |G x| = 15, another contradiction.
This completes the proof.

Remark. For more information about the thinking and strategies behind the Sylow game,
see Appendix B of these notes, and the videos of Lectures 16 and 17, on Moodle.

3.5 Simplicity of An , n ≥ 5
In this section we are going to prove the following result that was first proved by C. Jordan
in 1875.

Theorem 3.11. For n ≥ 5, the group An is simple.

38
 We proved the result for n = 5 in Proposition 3.10, and so we will prove Theorem 3.11 by
induction on n. But first, we will need to recall/prove few facts.
First of all, recall that Sn acts naturally on the set Xn := {1, 2, ..., n} (as was defined in
Algebra II), and as An ≤ Sn , so does An .
We now state and prove the following result.
Lemma 3.12. The following conditions hold:
(i) Let n ≥ 3, and let X be the set of 3-cycles in Sn . Note that X ⊆ An by definition of
An . We have An = ⟨X⟩.

(ii) Suppose that n ≥ 5. Then any two 3-cycles are conjugate in An .

Proof. (i) In Homework 2, Question 7, we proved that every element of An is a product of an
even number of transpositions (i.e. cycles of length 2). Thus to prove part (i) of the lemma,
it is enough to show that a product of a pair of transpositions can be written as a product of
3-cycles.
Consider a product (a, b)(c, d) of two transpositions, with a, b, c, d ∈ {1, . . . , n} and a ̸= b,
c ̸= d. We distinguish three possible cases according to the size of {a, b} ∩ {c, d}:
1. Suppose that {a, b} ∩ {c, d} is empty. Then by definition, (a, b) and (c, d) are disjoint
cycles. By direct calculation, we have

(a, b)(c, d) = (a, b, c)(b, c, d).

which gives us what we need.

2. Suppose that |{a, b}∩{c, d}| = 1. Without loss of generality, we may assume that a = c.
Then (a, b)(c, d) = (a, b)(a, d) = (a, d, b), again giving us what we need.

3. Suppose finally that |{a, b} ∩ {c, d}| = 2, that is {a, b} = {c, d}. Then (a, b)(c, d) =
(a, b)2 = () = 1An = (a, b, e)3 , for any e ∈ {1, . . . , n} \ {a, b}. Again, we have what we
need. This completes the proof of (i).

(ii) By Theorem 2.3(iii), g x is a 3-cycle for all 3-cycles x ∈ An . Thus, An acts on the set X
by conjugation. That is, the map · : An × X → X given by g · x := g x for g ∈ An , x ∈ X, is
a group action (see also Example 2.11(3)).
Thus, the statement of (i) is equivalent to the statement that An has precisely one orbit
in its action on X. Equivalently,

X = OrbG (x) (3.6)

where x is the 3-cycle x := (1, 2, 3).

So let us prove (3.6). Let σ = (a, b, c) ∈ An be a 3-cycle. Again, we distinguish 3 cases:
1. Suppose that {a, b, c} ∩ {1, 2, 3} is empty. Then n ≥ 6, and (1, a), (2, b), (3, c) are
transpositions. Let g := (1, 2)(1, a)(2, b)(3, c). Then g ∈ An , and by Theorem 2.3(iii)
we have
g
(1, 2, 3) = (1,2) (a, b, c) = (a, b, c)
with the last equality above following since {a, b, c} and {1, 2} are disjoint (and disjoint
cycles commute).

39
 2. Suppose that |{a, b, c} ∩ {1, 2, 3}| = 1. Without loss of generality, we may assume that
a = 1. Then by direct calculation using Theorem 2.3(iii), we have
g
(1, 2, 3) = (a, b, c)

where g := (2, b)(3, c) ∈ An .

3. Suppose that |{a, b, c} ∩ {1, 2, 3}| = 2. Assume first that a = 1 and b = 2. Let
d ∈ {1, . . . , n} \ {1, 2, 3, c}. Then as in case 1 above we have
g
(1, 2, 3) = (a, b, c)

where g := (3, d)(3, c) ∈ An . The arguments with a = 1, b = 3; and a = 2, b = 3, etc

are identical.
4. Suppose finally that {a, b, c} = {1, 2, 3}. If (a, b, c) = (1, 2, 3), then 1An (1, 2, 3) = (a, b, c),
as needed. If (a, b, c) = (1, 3, 2), then g (1, 2, 3) = (a, b, c), where g := (2, 3)(4, 5) ∈ An .
The result follows.

Our proof of Theorem 3.11 below is going to follow the proof of D. Dummit and R. Foote’s
proof in their book “Abstract Algebra”.
Lemma 3.13. For n ≥ 5, any 1 ̸= σ ∈ An has a conjugate σ ′ such that σ ̸= σ ′ and σ(i) = σ ′ (i)
for some i ∈ {1, 2, ..., n}.
Proof. Take any 1 ̸= σ ∈ An . Let r be the longest length of a disjoint cycle in σ. Relabelling
if necessary, we may write
σ = (1, 2, ..., r)π
where (1, 2, ..., r) and π are disjoint permutations.
If r ≥ 3, let τ = (3, 4, 5) and σ ′ = τ στ −1 . Then σ(1) = 2 and σ ′ (1) = 2, while σ(2) = 3
and σ ′ (2) = 4. Hence, σ ̸= σ ′ while σ(1) = σ ′ (1).
If r = 2, then σ is a product of disjoint transpositions. If there are at least 3 disjoint
transpositions involved, then n ≥ 6 and after relabelling we can write σ = (1, 2)(3, 4)(5, 6)....
Let τ = (1, 2)(3, 5) and σ ′ = τ στ −1 . Then σ(1) = 2 = σ ′ (1) while σ(3) = 4 and σ ′ (3) = 6.
Hence, again σ ̸= σ ′ while σ(1) = σ ′ (1).
Finally, if r = 2 and σ is a product of two disjoint transpositions. Again relabelling if
necessary we may write σ = (1, 2)(3, 4). This time take τ = (1, 3, 2) and set σ ′ = τ στ −1 .
Then σ ′ = (1, 3)(2, 4). Clearly σ ̸= σ ′ and σ(5) = 5 = σ ′ (5).

Proof. (Proof of Theorem 3.11) By Proposition 3.10, A5 is simple. Thus we may suppose that
n ≥ 6. Recall that An acts on Xn in the natural way. For each i ∈ Xn , let Hi := StabAn (i) ∼ =
An−1 . By induction, Hi is a simple group. Observe that Hi contains a 3-cycle (based on three
points of Xn other than i).
Assume there exists a non-trivial proper N ◁ An . Take any 1 ̸= σ ∈ N . By the previous
lemma, there exists σ ′ ∈ An such that σ ̸= σ ′ , σ ′ is conjugate to σ in An , and σ(i) = σ ′ (i) for
some i ∈ Xn . Since N ◁ An , σ ′ ∈ N . Hence, σ −1 σ ′ ∈ N , σ −1 σ ′ ̸= 1An and σ −1 σ ′ (i) = i. Thus
σ −1 σ ′ ∈ Hi and so N ∩ Hi ̸= {1An }. Now, N ◁ An , and so N ∩ Hi ◁ Hi by Theorem 2.4. But
Hi is a simple group, and so N ∩ Hi = Hi . Therefore Hi ≤ N . But Hi contains a 3-cycle,
and so N contains a 3-cycle. As N ◁ An , by Lemma 3.12(ii), N contains all 3-cycles of An .
The result now follows by Lemma 3.12(i).

40
3.6 An interesting example of a Sylow p-subgroup, and an
application (non-examinable)
In this section, we will see an interesting example of a Sylow p-subgroup in a finite group,
and we will also see a nice application of this example, due to Serre.
The example requires a bit of preparation:

The group GL(n, Fp )

Let p be a prime, n ∈ N and Fp := Z/pZ, so that Fp is a finite field of order p.
Consider the set of invertible n × n matrices over the field Fp :

GL(n, Fp ) = {A = (aij )n×n | aij ∈ Fp , 1 ≤ i, j ≤ n, , det(A) ̸= 0}

It is easy to see that GL(n, Fp ) forms a group under matrix multiplication.

Remark. Let V = Fnp be the n-dimensional column vector space over Fp , and let

G := {T : V → V | T an invertible linear map}.

Then G is a group under composition of functions, and we saw in MA106 that G is isomorphic
to GL(n, Fp ). Indeed, for A ∈ GL(n, Fp ), define θ(A) : V → V by θ(A)(v) := Av. Then it
follows immediately from the definition of matrix multiplication that θ is a homomorphism.
If {e1 , . . . , en } is the standard basis for V , and T ∈ G, then define AT to be the matrix of
T with respect to {e1 , . . . , en }. Then the map φ : G → GL(n, Fp ) defined by φ(T ) := AT is
both a left and right inverse to θ.
Notice that GL(n, Fp ) is finite. Our next lemma uses group actions to derive the precise
order of GL(n, Fp ).
n Q
Lemma 3.14. |GL(n, Fp )| = p( 2 ) ni=1 (pi − 1).
Proof. Let V = Fnp be the n-dimensional column vector space over Fp . Let G be the group

G := {T : V → V | T an invertible linear map}

defined above. Since GL(n, Fp ) = ∼ G, it will suffice to prove that |G| = p(n2 ) Qn (pi − 1).
i=1
So let us do this. Let E := (e1 , . . . , en ) be an ordered basis of V . Then for any T ∈ G, we
have that (T (e1 ), ..., T (en )) is again an ordered basis of V , since T is invertible. Let

B = {E | E is an ordered basis of V }.

Then we may define a map

·:G×B →B
by (T, E) → T · E where T · E = (T (e1 ), ..., T (en )) for T ∈ G, E = (e1 , ..., en ) ∈ B. It is easy
to check that this map gives us an action of G on B.
Take any two ordered bases E = (e1 , ..., en ) and F = (f1 , ..., fn ) of V . Then from MA106
we know that there exists a unique linear map T : V → V with T (ei ) = fi , 1 ≤ i ≤ n. Since
{f1 , ..., fn } is a basis of V , this map is invertible. It follows immediately that

OrbG (E) = B.

41
Moreover, because of the uniqueness, the identity linear map IdV : V → V must be the only
linear map T with the property that
T.(e1 , . . . , en ) = (e1 , . . . , en ).
In our language, this means that
StabG (E) = {IdV }.
By the Orbit-Stabiliser theorem, it follows that
|G| = |OrbG (E)| · |StabG (E)| = |B| · 1 = |B|.
Counting the number of ordered bases (e1 , . . . , en ) of V , we see that can choose any of the
|V | − 1 = pn − 1 non-zero elements of V as e1 . We can then choose any of the |V | − p elements
of V \ {λe1 : λ ∈ Fp } for e2 , etc. Continuing in this, we obtain that
n(n−1) n(n−1)
|B| = (pn − 1)(pn − p)...(pn − pn−1 ) = p 2 (pn − 1)(pn−1 − 1)...(p − 1) = p 2 Πni=1 (pi − 1),
and so
n n
n(n−1) n
(pi − 1) = p( 2 )
Y Y
|GL(n, Fp )| = |G| = p 2 (pi − 1)
i=1 i=1
as needed.

Next, consider the subgroup

U (n, Fp ) = {A = (aij )n×n | aij = 0, 1 ≤ j < i ≤ n; aii = 1, 1 ≤ i ≤ n; aij ∈ Fp , 1 ≤ i < j ≤ n}
of GL(n, Fp ). In HW2, you showed that U (n, Fp ) ≤ GL(n, Fp ) is indeed a subgroup. If we
n(n−1)
count its cardinality, we obtain that |U (n, Fp )| = p 2 . Therefore U (n, Fp ) ∈ Sylp (GL(n, Fp )).
We have therefore proved the following statement.
Lemma 3.15. U (n, Fp ) is a Sylow p-subgroup of GL(n, Fp ).

An application: Serre’s proof of Sylow’s theorem part 1

We are now going to see a nice application of our work on GL(n, Fp ) above: namely,
Serre’s proof of part 1 of Sylow’s theorem.
The first result we need is a very useful general result in group theory.
Lemma 3.16. The group GL(n, Fp ) contains a subgroup isomorphic to Sn .
Proof. For σ ∈ Sn let Iσ be the n × n matrix over Fp with
(
1, if j = σ(i)
aij =
0, otherwise

Clearly, det(Iσ ) ̸= 0, and so Iσ ∈ GL(n, Fp ). In particular, we have a map

f : Sn → GL(n, Fp )
defined by f (σ) = Iσ . Since Iσ1 σ2 = Iσ1 Iσ2 for σ1 , σ2 ∈ Sn , f (σ1 σ2 ) = f (σ1 )f (σ2 ), and so f
is a group homomorphism. Now ker(f ) = {σ ∈ Sn | Iσ = In } = {id}. Therefore, by the First
Isomorphism Theorem, Sn /{id} ∼ = Im(f ).

42
 We are now going to apply this to prove the first Sylow Theorem.

Serre’s proof of Sylow theorem 1: Let G be a finite group of order pn · m where p is

a prime, n ∈ N and gcd(p, m) = 1. Let N := pn · m. Then by Cayley’s Theorem, G is
isomorphic to a subgroup of SN . Now by Lemma 3.16, there exists a subgroup of GL(N, Fp )
isomorphic to SN . Therefore there exists a subgroup G0 of GL(N, Fp ) isomorphic to G. By
Lemma 3.15, GL(N, Fp ) has a Sylow p-subgroup U (N, Fp ). Therefore by Proposition 3.5, G0
has a Sylow p-subgroup, and thus so does G.

43
44
Chapter 4

Classifying groups of small order

In this chapter, we are going to classify the groups of order up to 18 (apart from those of
order 16). In order to do so, we will need some results on special types of groups. We will
begin with groups called semidirect products. As the name suggests, they are a generalisation
of direct products.

4.1 Semidirect products

Let us start by defining the direct product of groups.

Definition 4.1. Let H and K be groups, and define a binary operation · : (H × K) × (H ×

K) → H × K by (h1 , k1 ) · (h2 , k2 ) := (h1 h2 , k1 k2 ). Then (H × K, ·) is a group, called the
direct product of H and K. We denote it by H × K.

Remark. Of course, one can extend this definition, and define the direct product of an
arbitrary number of groups G1 , . . . , Gn . To do that, we just define multiplication in G1 ×
. . . × Gn by (g1 , . . . , gn ) · (g1′ , . . . , gn′ ) = (g1 g1′ , . . . , gn gn′ ).

One can check that G1 × ... × Gn with the operation defined above is indeed a group with
1G = (1G1 , ..., 1Gn ) and (g1 , ..., gn )−1 = (g1−1 , ..., gn−1 ).

The following is a very useful way to tell if the group you are working with is isomorphic
to a direct product of two given subgroups.

Lemma 4.2. Let G have normal subgroups H, K ⊴ G such that G = HK and H ∩ K = 1.

Then

(i) x commutes with y for all x ∈ H, y ∈ K (so if H and K are abelian, then G is abelian);
and

(ii) G ∼
= H × K.
Proof. Take h ∈ H, k ∈ K. Consider g = hkh−1 k −1 . Since both H and K are normal in G
we have
g = hkh−1 k −1 = (hkh−1 )k −1 = h(kh−1 k −1 ) ∈ H ∩ K = 1,
so g = 1. Thus hk = kh, and all elements of H commute with all elements of K. This proves
(i).

45
 Now define φ : H × K → G by φ((h, k)) = hk. Then

φ((h1 , k1 )(h2 , k2 )) = φ((h1 h2 , k1 k2 )) = h1 h2 k1 k2 = h1 k1 h2 k2 = φ((h1 , k1 ))φ((h2 , k2 )),

so φ is a homomorphism.
Notice that φ is surjective because G = HK. Suppose that φ((h, k)) = 1. Then hk = 1
so h = k −1 ∈ H ∩ K = 1. Thus h = k −1 = 1 = k. So φ is injective (see HW1, Q11). Hence φ
is an isomorphism, and H × K ∼ = G.

Remark. In the situation of Lemma 4.2, we simply say G = H × K, rather than G ∼

= H × K.

We would now like to generalise the direct product of two finite groups. First, we need a
definition.

Definition 4.3. Let G be a group. An automorphism of G is an isomorphism ϕ : G → G.

The set {ϕ : ϕ is an automorphism of G} is a group under composition of functions. It is
called the automorphism group of G, and is denoted by Aut(G).

The following definition is a generalisation of the direct product of two groups.

Definition 4.4. Let H and K be groups, and let ϕ : H → Aut(K) be a homomorphism. Write
the image of h ∈ H under ϕ as ϕh , and define a binary operation · : (H×K)×(H×K) → H×K
by
(h1 , k1 ) · (h2 , k2 ) := (h1 h2 , ϕh−1 (k1 )k2 ).
2

Then (H × K, ·) is a group called the semidirect product of H and K with respect to ϕ. We

denote it by H ⋉ϕ K.

It can be tricky to imagine maps H → Aut(K) for given groups H and K, so let’s look
at some nice (and important!) examples.

Example. (3 important examples of homomorphisms H → Aut(K) for certain

groups H, K).

1. The trivial homomorphism Let H and K be any groups. Then the map ϕ : H →
Aut(K) given by ϕ(h) := ϕh := IdK for all k ∈ K (i.e. ϕh (k) := k for all h ∈ H, k ∈ K) is
a homomorphism. The resulting semidirect product is isomorphic to the direct product:

H ⋉ϕ K ∼
= H × K.

2. The inversion homomorphism Let H = ⟨x⟩ ∼ = C2 , and let K be any abelian group.
Then the map ϕ : H → Aut(K) given by ϕ(1H ) := IdK and ϕ(x) := ϕx : K → K
with ϕx (k) := k −1 for all k ∈ K, is a homomorphism. If K ∼ = Cn , then the resulting
semidirect product is isomorphic to the dihedral group:

H ⋉ϕ K ∼
= D2n .

That is, C2 ⋉ϕ Cn ∼= D2n , where ϕ : C2 → Aut(Cn ) is the inversion homomorphism.

(See the example following Lemma 4.5 for a proof).

46
3. The conjugation homomorphism Let H and K be subgroups of a group G, with
K normal in G. Then the map ϕ : H → Aut(K) given by ϕ(h) := ϕh : K → K,
with ϕh (k) := hkh−1 for all h ∈ H, k ∈ K, is a homomorphism.

The following result gives us a useful way to tell if a group is isomorphic to the semidirect
product of two subgroups.

Lemma 4.5. Let G be a group, and let K and H be subgroups of G, with K ⊴ G. Define

ϕ : H → Aut(K)

by ϕh (k) := hkh−1 for h ∈ H, k ∈ K. Assume that G = HK and that K ∩ H = {1G }. Then

G∼
= H ⋉ϕ K.

Proof. Define f : H ⋉ϕ K → G by f ((h, k)) := hk. Note that

−1
(h1 k1 )(h2 k2 ) = h1 h2 h2 (k1 )k2 ,

for h1 , h2 ∈ H, k1 , k2 ∈ K. We therefore have

f ((h1 , k1 )(h2 , k2 )) = f ((h1 h2 , ϕh−1 (k1 )k2 ))

2

= h1 h2 ϕh−1 (k1 )k2 = h1 h2 (h−1

2 k1 h2 )k2 = h1 k1 h2 k2 = f ((h1 , k1 ))f ((h2 , k2 )).
2

by definition of multiplication in H ⋉ϕ K. Since f is clearly surjective, all that remains is

to prove that f is injective. But if f ((h, k)) = 1G for h ∈ H, k ∈ K, then 1G = hk, so
h = k −1 ∈ {1G }. That is, h = k = 1G . Thus, ker(f ) = {1G }, i.e. f is injective (see HW1,
Q11).

Example. (Dihedral groups as semidirect products) Let n ∈ N with n ≥ 3, and

consider the dihedral group G := D2n = ⟨σ, τ ⟩ from Definition 2.4. Here σ and τ are the
elements

σ := (1, . . . , n) and
⌊n/2⌋
Y
τ := (i, n − i + 1)
i=1

of Sn . Let K := ⟨σ⟩ and H := ⟨τ ⟩, and recall that τ σ = σ −1 . It follows that τ k = k −1 for all
k ∈ K.
Since |τ | = 2, |σ| = n, and D2n = K ⊔ τ K, we have G = HK and H ∩ K = {1G }. Hence,
G∼ = H ⋉ϕ K by Lemma 4.5, where ϕ : H = {1G , τ } → Aut(K) is given by ϕ1G (k) := k and
ϕτ (k) := k −1 , for k ∈ K.
The example of the dihedral group (and some generalisations that we will see later) will
actually be very important in our upcoming classification of groups of small order. For this
reason, we record the following very useful lemma.

Lemma 4.6. Let G be a nonabelian finite group, and suppose that

(1) G has a cyclic subgroup K of order n := |G|/2;

47
 (2) G \ K contains an element of G of order 2; and
(3) If i ∈ {0, . . . , n − 1} satisfies i2 ≡ 1 (mod n), then i ≡ ±1 modulo n.
Then G ∼
= D2n .
Proof. Let G be a group satisfying (1), (2), and (3) in the statement of the lemma, with
K ≤ G cyclic of order n := |G|/2. Then |G : K| = 2, so K ⊴ G.
Let x be an element of G \ K of order 2, and set H := ⟨x⟩. Then H ∩ K = {1G }, so
|HK| = |H||K| by Proposition 2.22. It follows that |HK| = |G|, whence HK = G. Let
ϕ : H → K be the homomorphism defined by ϕh (k) := hkh−1 , for h ∈ H, k ∈ K. Thus, we
have
G∼ = H ⋉ϕ K
by Lemma 4.5.
By the example above, we are therefore done if we can prove that ϕx (k) = k −1 for all
k ∈ K. Write K = ⟨y⟩, where |y| = n. Then xyx−1 = y i , for some 0 ≤ i ≤ n − 1 so
2 2
y = x2 yx−2 = xy i x−1 = (xyx−1 )i = y i . It follows that y i −1 = 1G , so n = |y| divides i2 − 1
by Lemma 1.9(i). In other words, i2 − 1 ≡ 0 (mod n). It follows from hypothesis (3) that
i ≡ ±1 (mod n). Since 0 ≤ i ≤ n − 1, we deduce that i ∈ {1, n − 1}.
If i = 1, then we’d have xyx−1 = y 1 = y, so xy = yx. Then G is abelian – a contradiction.
So we must have xyx−1 = y n−1 = y −1 . Thus, the homomorphism ϕ is given by ϕ1 (k) = k,
and ϕx (k) := k −1 , for all k ∈ K. This prove that G ∼= H ⋉ϕ K ∼ = D2n , as needed.
Example. The following are some examples of positive integers n which satisfy Hypothesis
(3) in Lemma 4.6.
1. Let n = 6. Then 22 , 32 , 42 ≡ 4, 3, 4 (mod n) respectively, so i2 ≡ 1 (mod n) if and only
if i ≡ ±1 (mod 6).
2. Let n = p, where p is prime. Then since Z/pZ = {0, . . . , p − 1} is a field, it has no zero
divisors. That is, i2 − 1 = 0 in Z/pZ if and only if (i − 1)(i + 1) ≡ 0 (mod p), which is
true if and only if i ≡ ±1 (mod p).
3. Let n = p2 , where p is prime. Then Z/p2 Z is not a field, so we can’t use the argument
above. If p = 2, then since 22 ̸≡ 1 (mod 4), we have that i2 ≡ 1 (mod 4) if and only if
i ≡ ±1 (mod 4).
Assume now that p is odd, and let i be an element of {0, 1 . . . , p2 − 1} such that i2 ≡ 1
(mod p2 ). Then p2 divides (i − 1)(i + 1). Also, since p is odd, we see that either p does
not divide i − 1; or p does not divide i + 1 (if it divided both, then it would divide
i + 1 − (i − 1) = 2). Thus, since p2 divides i2 − 1, we must have that p2 divides i − 1 or p2
divides i + 1. Hence, since 0 ≤ i ≤ p2 − 1, the only possibilities are i = 1 or i = p2 − 1,
i.e. i ≡ ±1 (mod p2 ).

4.2 Semidirect products of abelian and cyclic groups, and a

lemma of Fitting
In this section, we consider the following specific case: Let G be a finite group with |G|/2 odd,
and suppose that G has an abelian normal subgroup K of order |G|/2. Then H ∈ Syl2 (G)
has order 2, so we may write H = ⟨x⟩.

48
 The following is a powerful result of Fitting.

Lemma 4.7. (Fitting’s lemma) For v ∈ K, define [v, x] := vxv −1 x−1 and [K, x] :=
⟨[v, x] : v ∈ K⟩. Then

(i) x a = a−1 for all a ∈ [K, x]; and

(ii) K = CK (x) × [K, x].

(iii) G ∼
= (H ⋉θ [K, x]) × CK (x), where θ : H → Aut([K, x]) is the inversion homomorphism.

Proof. (i) Note that since K is normal in G, the group [K, x] is indeed contained in K.
Moreover, since K (and hence [K, x]) is abelian, we have that x a = a−1 for all a ∈ [K, x]
if and only if x a = a−1 for all a in a generating set for [K, x]. So it suffices to prove that
x [v, x] = [v, x]−1 for all v ∈ K. To see this, fix v ∈ K. Then

x
[v, x] = x(vxv −1 x−1 )x = xvxv −1 = [v, x]−1

with the last equality above following since x = x−1 .

(ii) Define f : K → [K, x] by f (k) := [k, x], for k ∈ K. Then

f (k1 k2 ) = k1 k2 xk2−1 k1−1 x−1

= k1 k2 xk2−1 x−1 xk1−1 x−1
= k1 k2 (x k2−1 )(x k1−1 )
= k1 (x k1−1 )k2 (x k2−1 ) since K is abelian
= [k1 , x][k2 , x] = f (k1 )f (k2 ).

Thus, f is a homomorphism. Clearly, ker(f ) = CK (x). Also, since Im(f ) is a subgroup of

[K, x] and contains a generating set for [K, x], we have Im(f ) = [K, x]. Thus, by the first
isomorphism theorem, we have

|K| = | ker(f )||Im(f )| = |CK (x)||[K, x]|.

Finally, since x inverts every element of [K, x], and K has odd order (so the only element k
of K with k = k −1 is 1G ), we have [K, x] ∩ CK (x) = {1G }. It follows from Proposition 2.22
and the above that
|K| = |CK (x)||[K, x]| = |CK (x)[K, x]|.

Thus, K = CK (x)[K, x]. Since K is abelian (so every subgroup is normal), we deduce from
Lemma 4.2 that K = CK (x) × [K, x], as needed.
(iii) See Homework 4, Question 14.

Remark. In our situation above, condition (2) always implies condition (1) by Cauchy’s and
Lagrange’s theorems! That is, if G has an abelian subgroup K of order |G|/2, and |G|/2 is
odd, then G has an element of order 2 by Cauchy’s theorem, and this element cannot be in
K by Lagrange’s theorem.

49
4.3 Abelian groups
This is a very small section. We just want to state (but not prove) the Fundamental theorem
of finite abelian groups. This was proved in Algebra II.

Theorem 4.8. (The fundamental theorem of finite abelian groups) Let G be a finite
abelian group. Then there exists divisors d1 , . . . , dt of |G| such that d1 | d2 | . . . | dt , and

G∼
= (Z/d1 Z) × (Z/d2 Z) × . . . × (Z/dt Z).

4.4 Groups of order p, p2 , or 2p, for p prime

Lemma 4.9. If |G| = p is prime, then G ∼
= Cp .

Proof. Let 1 ̸= g ∈ G. By Lagrange’s Theorem, |g| divides p. Since g ̸= 1, |g| = p and so

G = ⟨g⟩.

Lemma 4.10. If |G| = p2 with p prime, then either G ∼

= Cp2 ; or G ∼
= Cp × Cp .

Proof. We already proved in Corollary 2.20 that all groups of order p2 are abelian. The result
then follow immediately from the Fundamental theorem of finite abelian groups.

Lemma 4.11. If |G| = 2p with p an odd prime, then G is either cyclic (G ∼

= C2p ) or dihedral
(G ∼ D
= 2p ).

Proof. Let G be a group of order 2p, with p prime. If G is abelian, then we have G ∼ = C2p by
the Fundamental theorem of finite abelian groups.
So assume that G is nonabelian. Let P ∈ Sylp (G). Since |Sylp (G)| | |G|/|P | = 2, and
|Sylp (G)| ≡ 1 (mod p), we must have |Sylp (G)| = 1, that is, P ⊴ G.
Since p is odd, it follows that all elements of G of order 2 lies in G \ P . Also, since Z/pZ
is a field, the only solutions to the equation x2 − 1 in Z/pZ are congruent to ±1 modulo p. It
then follows from Lemma 4.6 that G ∼ = D2p , as needed.

4.5 Groups of order 2p2 , where p is an odd prime

Definition 4.12. Let p be an odd prime, let H := C2 = ⟨x⟩ and let K := Cp × Cp . Write
H = ⟨h⟩ with |h| = 2. Let
θ : H → Aut(K)
i
be the inversion homomorphism. That is, θ is defined by θxi (k) := k (−1) for i ∈ {0, 1}. The
group H ⋉θ K is called the generalised dihedral group of order 2p2 , and denoted by GD2p2 .

Lemma 4.13. Let G be a group of order 2p2 . Then G is isomorphic to one of the following
groups:
C2p2 , Cp × C2p , Cp × D2p , D2p2 or GD2p2 .

50
Proof. Suppose first that G is abelian. Then by the fundamental theorem of finite abelian
groups, we see that G is isomorphic to one of:
C2p2 or C2p × Cp .
So let’s assume that G is nonabelian. Let K ∈ Sylp (G) and H = ⟨x⟩ ∈ Syl2 (H), with
|x| = 2. Then |K| = p2 , so |G : K| = 2. It follows that K ⊴ G. Also, K is isomorphic to one
of Cp2 or Cp × Cp by Lemma 4.10.
Case 1: Suppose that K ∼ = Cp2 . Then G satisfies all of the conditions of Lemma 4.6, so
∼
G = D2p2 (condition (3) in Lemma 4.6 holds here by Example 3 after the statement of
Lemma 4.6).
Case 2: Suppose that K ∼ = Cp × Cp . Then G satisfies all of the conditions of Lemma 4.7
(Fitting’s lemma), so we have
G∼
= (H ⋉θ [K, x]) × CK (x),
where θ : H → Aut([K, x]) is the inversion homomorphism. By Lagrange’s theorem, the
possibilities for |CK (x)| are 1, p, or p2 . However, since G = ⟨K ∪ {x}⟩, we see that |CK (x)|
cannot be p2 . Indeed, otherwise, G would be abelian, since K is abelian, and we are assuming
that G is nonabelian. Thus, we have
|CK (x)| ∈ {1, p}.
Note that |[K, x]| = |K|/|CK (x)| by part (ii) of Fitting’s lemma (so [K, x] = K if |CK (x)| = 1)
Case 2(a): |CK (x)| = 1
In this case, we have G ∼ = H ⋉θ K, since [K, x] = K and CK (x) = {1G }. Hence, since
K∼ = Cp × Cp , we have G ∼ = GD2p2 by definition of GD2p2 .
Case 2(b): |CK (x)| = p
In this case, we have |CK (x)| = p = |[K, x]|.
Thus, H ⋉θ [K, x] is a nonabelian group of order 2p (it is nonabelian since G is nonabelian,
and a direct product of abelian groups is itself abelian). Hence, H ⋉θ [K, x] ∼= D2p by Lemma
4.11. Since CK (x) ∼= Cp by Lemma 4.9, we deduce that G ∼ = D2p × Cp . This completes the
proof.

4.6 Groups of order pq, where p < q are primes with p ∤ q − 1

Lemma 4.14. Let p and q be primes with p < q, and p ∤ (q − 1). Let G be a group of order
pq. Then G ∼
= Cpq .
Proof. By Corollary 3.7(ii) and Sylow’s theorem 4, |Sylp (G)| divides q, and is congruent to 1
modulo p. If |Sylp (G)| = q, then it would follow that q ≡ 1 (mod p), i.e. that p divides q − 1.
This is a contradiction. Thus, |Sylp (G)| = 1.
Next, again using Corollary 3.7(ii) and Sylow’s theorem 4, |Sylq (G)| divides p, and is
congruent to 1 modulo q. Since q > p, it follows that |Sylq (G)| = 1.
We have therefore shown that G has a normal Sylow p-subgroup, say H, and a normal
Sylow q-subgroup, say K. We have H ∩K = {1G } by Lagrange’s theorem; and hence HK = G
by Proposition 2.22 (see also Homework 2, Question 10). We therefore have G ∼ = H × K by
Lemma 4.2. Let x and y be generators of H and K respectively. Then x and y commute, so
|xy| = |x||y| = pq by Lemma 1.9(iv). Thus, G = ⟨xy⟩.

51
 We have now classified (in particular) all groups of the following orders:

1, 2, 3, 4, 5, 6, 7, 9, 10, 11, 13, 14, 15, 17, 18.

Thus, all that remains is to classify groups of orders

8 and 12.

We distinguish these cases now.

4.7 Groups of order 8

We begin by defining a special group of order 8.
Definition 4.15. Let i, j, and k be indeterminates, and consider the set

Q8 := {±1, ±i, ±j, ±k} ⊂ R[i, j, k].

Define a binary operation · : Q8 × Q8 → Q8 on Q8 by

1 · g = g · 1 = g and (−1) · g = g · (−1) = −g for all g ∈ Q8 (1)

i · j = k; j · k = i; and k · i = j (2)
j · i = −k; k · j = −i; and i · k = −j (3)
2 2
(±1) = 1 and g = −1 for all g ∈ {±i, ±j, ±k} (4)

Under this binary operation, Q8 is a group, called the quaternion group. (Note that relation
(3) follows from relations (2) and (4) above).
Remark. The following are useful facts about the group G = Q8 :
(1) Z(G) = {±1};

(2) G has precisely 1 element of order 2 (namely −1), and 6 elements of order 4 (namely
±i, ±j and ±k).

(3) G = ⟨i, j⟩ = ⟨j, k⟩ = ⟨k, i⟩.

(4) The Cayley table of G can be completely determined by the relations (1)–(4) in the
definition.
Note finally that G is not isomorphic to D8 , since D8 has 5 elements of order 2, and 2 elements
of order 4.
Lemma 4.16. Let n = 8. Then G is isomorphic to one of the following groups:

C2 × C2 × C2 , C4 × C2 , C8 , D8 , Q8

Proof. Let G be a group of order 8. If G is abelian, then G is isomorphic to one of

C2 × C2 × C2 , C2 × C4 , C8

by the fundamental theorem of finite abelian groups.

52
 So assume that G is nonabelian. Then G ∼ ̸ C8 , so no element of G has order 8. Also, we
=
proved in our very first lecture that groups in which x2 = 1G for all x ∈ G are abelian. Thus,
we may choose some u ∈ G with |u| > 2. We then have |u| = 4 by Lagrange’s theorem.
Set K := ⟨u⟩, and let v ∈ G \ K, with |v| minimal. Then |v| ∈ {2, 4}.
Before distinguishing these two possibilities for |v|, we note that G = ⟨u, v⟩, since v ̸∈ K =
⟨u⟩, and |K| = 4. Also, K ⊴ G, since |G : K| = 2. It follows that v u ∈ K = {1G , u, u2 , u3 }.
If v u = u, then v commutes with u, so G is abelian, since G = ⟨u, v⟩. This is a contradiction.
So v u ∈ {1G , u2 , u3 }. Since u3 = u−1 is the only element of the latter set of order 4, and
|v u| = |u| = 4, we must therefore have v u = u−1 .
We now distinguish the two possibilities for |v|:
Case 1: |v| = 2
Suppose that |v| = 2. Then G satisfies the three hypotheses of Lemma 4.6 (for condition (3),
see Example 3 after the statement of Lemma 4.6). Thus, G ∼ = D8 .
Case 2: |v| = 4
Suppose that |v| = 4. Then since v ̸∈ K and |G : K| = 2, we must have G/K = {K, vK}.
Hence, G = K ⊔ vK. Since 4 = |v| is the minimal order of an element in G \ K, we deduce
that every element of vK = G \ K has order 4. Hence, G has precisely 6 elements of order 4
(namely vK ⊔ {u, u−1 }); and 1 element of order 2, (namely z := u2 ). It follows that

x2 = z for all x ∈ vN ⊔ {u, u−1 }.

Hence, z ∈ Z(G) and zx = x3 = x−1 for all x ∈ vN ⊔ {u, u−1 }. Let

w := uv = v v u = vu−1 ∈ vK.

Note that u±1 , v ±1 and w±1 are distinct elements of G of order 4. Indeed, u±1 ∈ N and
v ±1 , w±1 ∈ yN , so the only possibility is v ±1 = w = uv, which would imply that either
u = 1G or v 2 = u. These are both contradictions, since u ̸= 1G , and |v| = |u| = 4 (so
|v 2 | = 2).
It follows that

G = {1G , z, u±1 , v ±1 , w±1 } = {1G , z, u, zu, v, zv, w, zw}.

Note that
−1
uv = vv −1 uv = v v u = vu−1 = vu3 = vuz = zvu.

The identical argument with u and v replaced by any distinct pair of elements of {u, v, w}
yields
gh = zhg for all g, h ∈ {u, v, w}.

Since u2 = v 2 = w2 = z ∈ Z(G), it follows that

zg = gz for all g ∈ G (1)

2 2
uv = w; vw = v(uv) = zv u = u; and wu = (uv)u = zvu = v (2)
vu = zw; wv = zu; and uw = zv in the same way; and (3)
2 2
(±1) = 1 and g = z for all g ∈ {u, zu, v, zv, w, zw} (4)

53
Clearly, the Cayley table of G can be completely determined by the relations (1)–(4) above.
It follows easily that the map

ϕ :G → Q8 by
1G → 1
u→i
v→j
w→k
z → −1
zu → −i
zv → −j
zw → −k

is an homomorphism. Since it is surjective and injective by inspection, we have G ∼ = Q8 , as

needed.
Summing up, there are five isomorphism types of groups of order 8, C2 × C2 × C2 , C8 ,
C4 × C2 , D8 , and Q8 . The first three of these are abelian and the final two nonabelian.

We omit classification of groups of order 16. Here are some facts about the number of
isomorphism types of finite 2-groups.
|G| 2 22 23 24 25 26 27 28 29 210
no. of isomorphism types 1 2 5 14 57 267 2328 56092 10494213 49487365422

4.8 Groups of order 12

We have already seen two nonabelian groups of order 12: namely, A4 and D12 . We now define
another.
Definition 4.17. Let H := C4 and K := C3 , and write H = ⟨h⟩. Define ϕ : H → Aut(K) by
i
ϕhi (k) := k (−1) , for 0 ≤ i ≤ 3 and k ∈ K. The resulting group H ⋉ϕ K is called the dicyclic
group of order 12, and is denoted by Dic12 .
Lemma 4.18. Let G be a group of order 12. Then G is isomorphic to one of the groups:

C12 , C6 × C2 , Dic12 , D12 , A4

Proof. If G is abelian, then we have

G∼
= C12 or G ∼
= C6 × C2

by Theorem 4.8.
So assume that G is nonabelian. We will distinguish our cases according to whether or
not G contains an element of order 6.
Case 1: Suppose that G contains an element of order 6.
Let a be such an element, and let K := ⟨a⟩. Suppose first that G \ K has an element of order

54
2. Then G satisfies the three hypotheses of Lemma 4.6 (for condition (3), see Example 1 after
the statement of Lemma 4.6). Thus, G ∼ = D12 .
Assume now that no element of G \ K has order 2, and let H ∈ Syl2 (G). Then H ̸≤ K,
since |K|2 = 2. Also, |H| = 4. If every non-identity element of H has order 2, then H \ K
would consist of elements of order 2, contrary to our assumption that G \ K has no elements
of order 2. Thus, H must have an element of order 4. Hence, H ∼ = C4 .
2 2
Now, let K1 := ⟨a ⟩. Then |K1 | = |a | = 3, by Lemma 1.9. Thus, K1 ∈ Syl3 (G). Also,
K1 ⊴ G by Homework 3, Question 9. We now have

(1) H ≤ G, K1 ⊴ G;

(2) H ∩ K1 = {1G } by Lagrange’s theorem; and

(3) |HK1 | = |H||K1 | = |G| by (2) and Proposition 2.22. Thus, HK1 = G.

It then follows from Lemma 4.5 that G ∼ = H ⋉ϕ K1 , where ϕ is the map ϕ : H → Aut(K1 )
−1
given by ϕh (k) := hkh . Finally, note that if H = ⟨h⟩ and k ∈ K \ {1G }, then G = ⟨h, k⟩,
so hkh−1 ̸= k, since G is nonabelian. Thus, since hkh−1 ∈ K1 and |hkh−1 | = |k|, we must
have hkh−1 = k −1 . Thus,
i
hi kh−i = k (−1) .
Hence, G ∼
= Dic12 by definition of Dic12 .
Case 2: Suppose that G has no element of order 6.

By Cauchy’s theorem, we may choose an element x of G of order 3. Since x ∈ CG (x), and

G has no element of order 6, we see that 2 cannot divide |CG (x)|. Indeed, if it did, then we
could choose an element y of CG (x) of order 2, again by Cauchy’s theorem. But then |xy| = 6
by Lemma 1.9 – a contradiction.
Thus, we must have |CG (x)| = 3, by Lagrange’s theorem. Hence, |G x| = |G : CG (x)| = 4
by the Orbit–Stabiliser theorem. Since G has precisely |Syl3 (G)|(3 − 1) elements of order 3
by Corollary 3.8(iii), we deduce that |Syl3 (G)| > 1. In particular, H is not normal in G.
Finally, let H ∈ Syl3 (G), and consider the action of G on G/H = {gH : g ∈ G} by left
multiplication. Let K be the kernel of this action. Then K ≤ H by Homework 2, Question 9.
But H is not normal in G, since H ∈ Syl3 (G) and |Syl3 (G)| > 1. So we must have |K| = 1,
since |H| = 3. Thus,

G∼
= G/K is isomorphic to a subgroup of S4 .

But the only subgroup of S4 of order 12 is A4 by Homework 4, Question 5. Thus, G ∼

= A4 .
This completes the proof.

4.9 Unique simple group of order 60

In this section we are going to prove the following result.

Theorem 4.19. Let G be a finite simple group of order 60. Then G ∼

= A5 .

Proof. Let G be a finite simple group group with |G| = 60. We will subdivide the proof into
a series of claims.

55
 Claim 1 G does not contain a proper subgroup of index n ≤ 4.
proof Assume the contrary. Let H ≤ with 1 ̸= |G : H| ≤ 4. Then G acts non-trivially
by left multiplication on the set of left cosets G/H of H in G, and |G/H| = n for some
2 ≤ n ≤ 4. Since G is simple, the kernel of this action is trivial. Thus by Proposition 2.13,
G is isomorphic to a subgroup of Sn . Since 2 ≤ n ≤ 4, |Sn | ≤ 4! = 24. On the other hand
|G| = 60, a contradiction.
Claim 2 G contains a subgroup of index 5.
proof Assume the contrary. Let P ∈ Syl2 (G). By Sylow’s theorem 4 and Corollary 3.7(ii),
|Syl2 (G)| is odd and divides 15. Moreover, |Syl2 (G)| ≠ 1 since G is simple. Also, |Syl2 (G)| =
|G : NG (P )| by Corollary 3.7(i), so |Syl2 (G)| =
̸ 3 by Claim 1.
Thus,
Syl2 (G) ∈ {5, 15}.
If Syl2 (G) = 5, then we are done, since then |G : NG (P )| = 5 by Corollary 3.7(i). So assume
that |Syl2 (G)| = 15.
Let g ∈ G \ NG (P ), and set H = ⟨P ∪ g P ⟩. Then H contains at least two Sylow 2-
subgroups (namely, P and g P ). So m := |Syl2 (H)|, being an odd number by Sylow’s theorem
4, is at least 3. Since 4 divides |H|, we deduce that |H| is divisible by 4|Syl2 (H)|. By Claim
1, we have either H = G, or |H| < |G|/4 = 15. Thus, we must have either H ̸= G and
|Syl2 (H)| = 3, or H = G. If H ̸= G and |Syl2 (H)| = 3, then 12 divides |H|, as noted above,
so we must have |H| = 12 by Claim 1. Thus, |G : H| = 5, as needed.
So assume that for all g ∈ G \ NG (P ), we have G = ⟨P ∪ g P ⟩. If x ∈ P ∩ g P , then x is
centralised by both P and g P , since groups of order 4 are abelian. Thus, x is centralised by
G, since G = ⟨P ∪ g P ⟩. Since Z(G) = {1G } (As G is simple), we deduce that P ∩ g P = {1G }
for all g ∈ G. Now, [
F2 (G) = P \ {1G }
P ∈Syl2 (G)

by Corollary 3.8(i). By our work above, this is a disjoint union. Thus, |F2 (G)| = |Syl2 (G)|(4−
1) = 45.
Finally, by Sylow’s theorem 4 and Corollary 3.7(ii), we have |Syl5 (G)| is congruent to 1
modulo 5, and divides 12. Since |Syl5 (G)| > 1 (as G is simple), we deduce that |Syl5 (G)| = 6.
But then
|F5 (G)| = 6(5 − 1) = 24
by Corollary 3.8(ii)(b). So |G| > |F2 (G)| + F5 (G)| = 69, a contradiction. This completes the
proof of Claim 2.

Let H ≤ G be a subgroup of index 5. Then G acts non-trivially by left multiplication on

the set of left cosets G/H of H in G. Since G is simple, the kernel of this action is trivial, and
so by Proposition 4.4, G is isomorphic to a subgroup G1 of S5 . By Homework 4, Question 5,
the only subgroup of S5 of order 60 is isomorphic to A5 . This completes the proof.

56
Chapter 5

Soluble groups and the

Jordan–Hölder theorem

The purpose of this chapter is two-fold: First, we are going to state and prove the Jordan–
Hölder theorem (which shows us just why finite simple groups are so important in finite group
theory).
Secondly, we are going to learn about a very important class of groups: the soluble groups.
The soluble groups are the central object of the Abel-Ruffini theorem (one of the two main
results in Galois theory), which determines precisely when an integer polynomial is solvable
by radicals. We won’t get into the details of that here, but Galois theory allows one to prove
that such a polynomial is solvable by radicals if and only if the associated Galois group is
soluble.

5.1 Composition Series and the Jordan-Hölder Theorem

Throughout, we will use the notation H ⪇ G to mean that H is a proper subgroup of the
◁G
group G (while H ≤ G continues to just mean that H is a subgroup of G). Similarly H ⪇
means that H is a proper normal subgroup of G (while H ⊴ G continues to just mean that
H is a normal subgroup of G).
Definition 5.1. A composition series of a group G is a series
◁ G1 ⪇
{1G } = G0 ⪇ ◁ ··· ⪇
◁ Gr = G

such that for each 1 ≤ i ≤ r, the group Gi /Gi−1 is simple. Moreover, in this case, r is called
the length of the composition series.
Example. The following are examples of compositions series of some important groups we
have seen in the course.
1. Let p be an odd prime, and let G = D2p = ⟨σ, τ ⟩, where σ and τ are as in Definition
2.4). Let G0 := {1G }, G1 = ⟨σ⟩ and G2 = G. Then G0 ⪇◁ G1 and G1 ⪇ ◁ G2 . Moreover,
∼ ∼ ∼
G1 /G0 = G1 = Cp and G2 /G1 = C2 are simple. Hence, the series
◁ G1 ⪇
{1G } = G0 ⪇ ◁ G2 = G

is a composition series of G of length 2.

57
 ◁ G1
2. Let n ≥ 5, and let G = Sn . Let G0 = {1G }, G1 := An , and G2 := Sn . Then G0 ⪇
◁ ∼ ∼
and G1 ⪇ G2 . Moreover, G1 /G0 = G1 = An and G2 /G1 = C2 are simple. Hence, the
series
{1G } = G0 ◁
⪇ G1 ◁
⪇ G2 = G
is a composition series of G of length 2.

3. Let G = D8 = ⟨σ, τ ⟩ (again, where σ and τ are as in Definition 2.4). Let G0 := {1G },
G1 := ⟨σ 2 ⟩ ≤ Z(G), G2 := ⟨σ⟩ and G3 := G. Then G0 ⪇ ◁ G1 , G1 ⪇
◁ G2 , and G2 ⪇
◁ G3 .
Moreover, G1 /G0 ; G2 /G1 ; and G3 /G2 all have order 2, so each of these groups is
isomorphic to C2 . Hence, each is simple, so the series

◁ G1 ⪇
{1G } = G0 ⪇ ◁ G2 ⪇
◁ G3 = G

is a composition series of G of length 3.

Remark. For technical reasons, we assume if G is the trivial group (i.e. G = {1G }), then
the series
{1G } = G0 = G
is a composition series of G of length 0.

Theorem 5.2. Every finite group has a composition series.

Proof. We will prove the theorem by induction on |G|. As mentioned above, if |G| = 1, then
G has a composition series.
So assume that |G| > 1, and that every group of order less than |G| has a composition
◁ G1 := G is a composition series of G.
series. If G is simple, then {1G } =: G0 ⪇
So assume that G is not simple. Then G has a normal subgroup N with {1G } = ̸ N ̸= G.
Then |N | < |G| and |G/N | < |G|, so our inductive hypothesis implies that:

◁ N1 ⪇
1. N has a composition series {1G } = N0 ⪇ ◁ ... ⪇
◁ Nr = N ; and

◁ G1 ⪇
2. G/N has a composition series {1G } = G0 ⪇ ◁ ... ⪇
◁ Gs = G/N .

Fix 0 ≤ i ≤ s. By the Correspondence Theorem (Theorem 1.31), there exists a subgroup

Xi of G containing N such that Gi = Xi /N . Since Gi is a normal subgroup of Gi+1 , it
is a routine exercise to see that Xi is a normal subgroup of Xi+1 . Moreover, by the Third
isomorphism Theorem, we have

Gi+1 Xi+1 /N ∼ Xi+1

= = ,
Gi Xi /N Xi

so Xi+1 /Xi is simple.

Now, notice that Xs = G and Nr = N = X0 . Set Gi := Ni for 0 ≤ i ≤ r, and set
Gi := Xi−r for r + 1 ≤ i ≤ r + s. Then by our work above,

G0 := {1G } ◁
⪇ G1 ◁
⪇ ... ◁
⪇ Gr+s = G

is a composition series for G. This completes the induction, and the result follows.

58
Corollary 5.3. Let G be a finite group, and let N ⊴ G. Suppose that

◁ N1 ◁
{1G } =: N0 ⪇ ◁ Nr = N ; and
⪇ ... ⪇ (5.1)
X0 X X G
{1G/N } =: ◁ 1 ⪇
⪇ ◁ s = ;
◁ ... ⪇ (5.2)
N N N N
are composition series’ for N and G/N , respectively, where each Gi is a subgroup of G
containing N (in particular, X0 = N and Xs = G by the Correspondence theorem). For
0 ≤ i ≤ r, set Gi := Ni , and for 1 ≤ i ≤ s, set Gr+i := Xi . Then

◁ ... ⪇
{1G } := G0 ⪇ ◁ Gr+s = G

is a composition series for G.

Now, it is important to note that a finite group may have more than one composition
series. Indeed, as an example, let us consider the case G = D8 = ⟨σ, τ ⟩ above.

Example. We saw that by setting G0 := {1G }; G1 := ⟨σ 2 ⟩; G2 := ⟨σ⟩; and G3 := G, we get

a composition series
◁ G1 ⪇
{1G } = G0 ⪇ ◁ G2 = G
of G of length 3.
Next, let N0 := {1G }, N1 := ⟨σ 2 ⟩ ≤ Z(G), N2 := ⟨σ 2 , τ ⟩ and N3 := G. Then N0 ⪇ ◁ N1 ,
N1 ⪇◁ N2 , and N2 ⪇◁ N3 (each Ni has index 2 in Ni+1 ). Moreover, N1 /N0 ; N2 /N1 ; and N3 /N2
all have order 2, so each of these groups is isomorphic to C2 . Hence, each is simple, so the
series
{1G } = N0 ⪇◁ N1 ⪇
◁ N2 = G
is a composition series of G of length 3.
Finally, notice that N2 ̸= G2 , so these composition series are distinct.

Now, the composition series in the example above are different, but the lengths are the
same (both 3); and the simple factors are the same up to isomorphism (each of the three
factors are isomorphic to C2 , in each case). Composition series with these factors are thought
of as “the same” via the following definition.

Definition 5.4. We shall say that the two composition series of a group G

◁ A1 · · · ⪇
{1G } = A0 ⪇ ◁ Ar = G (I) and ◁ B1 · · · ⪇
{1G } = B0 ⪇ ◁ Bs = G (II)

are equivalent and write (I) ∼ (II) if r = s and there is a bijection

f : {Ai /Ai−1 : 1 ≤ i ≤ r} → {Bj /Bj−1 : 1 ≤ j ≤ s}

such that Ai /Ai−1 ∼

= f (Ai /Ai−1 ) for each 1 ≤ i ≤ r.

Theorem 5.5 (The Jordan-Hölder theorem). Let

{1G } = A0 ◁
⪇ A1 · · · ◁
⪇ Ar = G(I) and {1G } = B0 ◁
⪇ B1 · · · ◁
⪇ Bs = G (II)

be two composition series of a finite group G. Then (I) ∼ (II).

59
Proof. We may assume, without loss of generality, that r ≤ s. We will now prove the result
by induction on r. If r = 0, then G = {1G } and the result is clear. So assume that r > 0.
We now distinguish a number of cases.

Case 1 : Suppose first that Ar−1 = Bs−1 . Then

◁ A1 · · · ⪇
{1G } = A0 ⪇ ◁ Ar−1 and ◁ B1 · · · ⪇
{1G } = B0 ⪇ ◁ Bs−1

are two composition series of Ar−1 = Bs−1 of lengths r − 1 and s − 1, and so by induction
they are equivalent, and hence so are (I) and (II).
◁ G and Bs−1 ⪇
Case 2 : Suppose now that Ar−1 ̸= Bs−1 . Consider Ar−1 Bs−1 . Since Ar−1 ⪇ ◁
G, we have
Bs−1 ⪇ Ar−1 Bs−1 ⊴ G.
It follows from the Third isomorphism theorem that Ar−1 Bs−1 /Bs−1 is a normal subgroup
of G/Bs−1 . Since Bs−1 ⪇ Ar−1 Bs−1 , it is in fact a non-trivial normal subgroup, by the
Correspondence theorem. Thus since G/Bs−1 is simple, we have Ar−1 Bs−1 /Bs−1 = G/Bs−1 ,
so Ar−1 Bs−1 = G by the Correspondence theorem.
Let D = Ar−1 ∩ Bs−1 . By the Second Isomorphism Theorem,
G/Ar−1 = Ar−1 Bs−1 /Ar−1 ∼
= Bs−1 /(Ar−1 ∩ Bs−1 ) = Bs−1 /D
and as G/Ar−1 is simple, so is Bs−1 /D. Similarly, by the Second Isomorphism Theorem,
G/Bs−1 1 = Ar−1 Bs−1 /Bs−1 ∼
= Ar−1 /(Ar−1 ∩ Bs−1 ) = Ar−1 /D
and as G/Bs−1 is simple, so is Ar−1 /D.
Now, let
◁ ... ⪇
{1G } = D0 ⪇ ◁ Dt := D
be a composition series for D. Then
◁ ... ⪇
{1G } = D0 ⪇ ◁ Dt = D ⪇
◁ Ar−1 ⪇
◁ G (III) and ◁ ... ⪇
{1G } = D0 ⪇ ◁ Dt = D ⪇
◁ Bs−1 ⪇
◁ G(IV)

are also composition series for G.

Since we are assuming that r ≤ s, we may apply Case 1 to the series (I) and (III), and
deduce that (I) ∼ (III). In particular, we have r = t + 2, and hence series (IV) has length
t + 2 = r. Thus we may also apply Case 1 to conclude that (II) ∼ (IV).
Finally, we saw above that G/Ar−1 ∼= Bs−1 /D and Ar−1 /D ∼
= G/Bs−1 . Thus (III) ∼ (IV),
and hence (I) ∼ (II) as required.

Thus, we have proved that both the set of factors {Gi /Gi−1 : 1 ≤ i ≤ r} (up to isomor-
phism); and the length r of any composition series
◁ ... ⪇
1 =: G0 ⪇ ◁ Gr := G

of a finite group G are invariants of the group. Thus, we have:

Definition 5.6. Let
◁ G1 ⪇
{1G } = G0 ⪇ ◁ ··· ⪇
◁ Gr = G
be a composition series for a finite group G. The groups Gi /Gi−1 for 0 ≤ i ≤ r are called the
composition factors of G, while r is called the composition length of G.

60
Remark. When we write down the composition factors of a finite group, we usually just
write them up to isomorphism. Thus, for example, in the examples following Definition 5.1
above, we’d say that:
1. D2p (for p an odd prime) has composition length 2, with composition factors Cp , C2 .
2. Sn (for n ≥ 5) has composition length 2, with composition factors An , C2 .
3. D8 has composition length 3, with composition factors C2 , C2 , C2 .

Definition 5.7. A finite group is called soluble if it is either trivial, or its composition factors
are all cyclic groups of prime order.
Example. 1. Let G be an abelian group. Then every quotient of every subgroup of G is
of course abelian. It follows that the set of composition factors of G consists of abelian
simple groups, i.e. cyclic groups of prime order. Thus, all abelian groups are soluble.
2. Let n ≥ 5, and let G = An . Then G is a nonabelian simple group, so it has precisely
one composition factor: An itself (which is nonabelian). Thus, G is not soluble.
Being able to check whether or not a given finite group is soluble is very important!
Especially if you go on to do Galois theory. We will see two very useful techniques to do this.
The first is as follows (the second is the subject of the next section).
Lemma 5.8. Let G be a finite group, and let N be a normal subgroup of G. Then G is
soluble if and only if both N and G/N are soluble.
Proof. Write CF(G) for the multiset of composition factors of G. By Corollary 5.3 and the
Jordan–Hölder theorem, we have

CF(G) = CF(N ) ∪ CF(G/N ).

Thus, G is soluble if and only if both N and G/N are soluble.

Example. Let G = D2n = ⟨σ, τ (see Definition 2.4). Let N := ⟨σ⟩. Then N is abelian, and
|G/N | = 2, so G/N is abelian. Hence, N and G/N are both soluble by Example 1 above. So
G is soluble by Lemma 5.8.

5.2 Commutators
In this section, we will learn about another tool to check whether or not a given group is
cyclic. To state, we will need the language of commutators (we saw a special case of these
already when we studied Fitting’s lemma in Chapter 2).
Definition 5.9. For g, h ∈ G the commutator [g, h] of g and h is defined to be ghg −1 h−1 .

Example. Let G = A5 . Then

[(1, 2, 4), (1, 3, 5)] = (1, 2, 4)(1, 3, 5)(4, 2, 1)(5, 3, 1) = (1, 2, 3)(4)(5) = (1, 2, 3)

More generally, if {i, j, k, l, m} = {1, 2, 3, 4, 5}, then

[(i, j, l), (i, k, m)] = (i, j, l)(i, k, m)(l, j, i)(m, k, i) = (i, j, k)

61
Remark. Note that [g, h] = 1 if and only if gh = hg (i.e., g and h commute).

Definition 5.10. The commutator subgroup [G, G] of G is the subgroup of G generated by

all of its commutators:
[G, G] = ⟨[g, h] | g, h ∈ G⟩.

Remark. More generally, if H, K ≤ G, then we will write

[H, K] = ⟨[h, k] | h ∈ H, k ∈ K⟩.

We sometimes call [H, K] the commutator subgroup of H and K.

Remark 5.11. In general, the set {[g, h] : g, h ∈ G} is not a subgroup of G. So in general
the elements of [G, G] may be products of several commutators.

Example. The following are some examples of commutator subgroups in the groups we’ve
seen in the course so far:
1. Let G be an abelian group. Then gh = hg for all g, h ∈ G. Hence, [g, h] = 1 for all
g, h ∈ G, and so [G, G] = 1.

2. Let G = A5 . From the example after the definition above, we see that every 3-cycle in
A5 is a commutator. But A5 is generated by 3-cycles, and so [A5 , A5 ] = A5 .

Theorem 5.12. The following conditions hold:

(i) [G, G] ⊴ G.
(ii) G/[G, G] is abelian.
(iii) If N ⊴ G and G/N abelian, then [G, G] ≤ N .
Proof. (i) For all g, h, k ∈ G:

g[h, k]g −1 = ghkh−1 k −1 g −1

= (ghg −1 )(gkg −1 )(gh−1 g −1 )(gk −1 g −1 )
= [ghg −1 , gkg −1 ] ∈ [G, G].

So, for [h1 , k1 ][h2 , k2 ] · · · [hr , kr ] ∈ [G, G], we have

g[h1 , k1 ] · · · [hr , kr ]g −1 = (g[h1 , k1 ]g −1 ) · · · (g[hr , kr ]g −1 ) ∈ [G, G]

and hence, [G, G] ⊴ G.

(ii) Let g, h ∈ G. Then

ghg −1 h−1 ∈ [G, G] =⇒ [G, G]ghg −1 h−1 = [G, G]

=⇒ [G, G]gh = [G, G]hg
=⇒ [G, G]g[G, G]h = [G, G]h[G, G]g

So G/[G, G] is abelian.
(iii) If G/N is abelian, then N gN h = N hN g for all g, h ∈ G. Thus N (ghg −1 h−1 ) = N , and
so [g, h] ∈ N . Since N ≤ G, [G, G] ≤ N .

62
Corollary 5.13. A group G is abelian if and only if [G, G] = 1.

Example. Let n ≥ 3, and let

G = D2n = ⟨σ, τ ⟩
where σ and τ are as in Definition 2.4. Recall that τ σ = σ −1 . Then [τ, σ] = τ στ −1 σ −1 = σ −2 .
So σ 2 = (σ −2 )−1 ∈ [G, G]. Consider K = ⟨σ 2 ⟩ ≤ G. Then K ≤ [G, G]. Note also that K ⊴ G
by Homework 3, Question 9.
Case 1: n odd. Since n is odd, K = ⟨σ 2 ⟩ = ⟨σ⟩. Thus |G : K| = 2, and so G/K is abelian.
Hence, [G, G] ≤ K by Theorem 5.12(iii). Thus [G, G] = K = ⟨σ⟩ has order n.
Case 2: n even. Since n is even, K = ⟨σ 2 ⟩ has order n/2. Thus |G : K| = 4, and so G/K
is abelian by Corollary 2.19. Hence, [G, G] ≤ K by Theorem 5.12(iii). Thus, K = [G, G] has
order n/2.
Definition 5.14. For a group G, define G(0) = G, G(1) = [G, G] and, inductively, G(n+1) =
[G(n) , G(n) ] for integers n > 0. Then the descending series
G(0) ≥ G(1) ≥ G(2) ≥ · · · ≥ G(n) ≥ G(n+1) ≥ · · ·
is called the derived series of G.
Remark. By definition of the derived series for G, we have
1. (G(n) )(m) = G(n+m) ; and
2. H (n) ≤ G(n) .
for all non-negative integers n and m. (These assertions follow from easy inductive argu-
ments.)
We can now state our second tool for deciding whether or not a given finite group is
soluble.
Theorem 5.15. Let G be a finite group. Then G is soluble if and only if G(n) := {1G } for
some positive integer n.
Proof. Suppose first that G is soluble. We will prove, by induction on |G|, that G(n) = {1G }
for some positive integer n. Now, if G = {1G }, then this clearly holds, so assume that |G| > 1.
Let N := [G, G]. Then N is soluble by Corollary 5.3. Also, G has a composition series
{1G } =: G0 ◁
⪇ G1 ◁
⪇ ... ◁
⪇ Gr := G
where all of the groups Gi /Gi−1 are cyclic of prime order. This is by definition of solubility. It
follows that Gr−1 is a normal subgroup of G with Gr /Gr−1 = G/Gr−1 cyclic of prime order.
Thus, N ≤ Gr−1 by Theorem 5.12(3). In particular, |N | < |G|. So N (m) = {1G } for some
m ∈ N by the inductive hypothesis. Since G(n) = [G, G](n−1) by definition, it follows that
G(m+1) = {1G }, as needed.
Suppose next that G(n) = {1G } for some n ∈ N. We need to prove that G is soluble.
We will again do this by induction on |G|. This is clearly true if G = {1G }, so assume that
|G| > 1. Let N := [G, G]. If N = G, then we’d have G(n) = [G, G](n−1) = G(n−1) = . . . = G,
which contradicts our hypothesis. So we must have N ⪇ G. Since N (n−1) = G(n) = {1G },
we have that N is soluble by the inductive hypothesis. Also, G/N = G/[G, G] is abelian by
Theorem 5.12. So both N and G/N are soluble. Hence, G is soluble by Corollary 5.3. This
completes induction, whence the proof.

63
5.3 Some important examples of soluble groups
In this section, we would like to say a bit more about soluble groups, and about what kinds
of groups are soluble.
Theorem 5.16. Let G be a finite soluble group, and let H ≤ G. Then H is soluble.
Proof. Since G is soluble, it follows from Theorem 5.15 that G(n) = {1G } for some n ∈ N.
Since H (n) ≤ G(n) by the remark above, we have H (n) = {1G }. Thus, another application of
Theorem 5.15 yields the result.

Theorem 5.17. Let p be prime, let n ∈ N, and let G be a group of order pn . Then G is
soluble, and all composition factors of G are isomorphic to Cp .
Proof. We prove the theorem by induction on |G|. If |G| = p then G is cyclic of prime order,
so G is soluble of composition length 1, and its composition factor is Cp of course.
Assume now that |G| > p, and that the result holds for groups of order less than |G|. By
Theorem 2.18, Z := Z(G) is non-trivial. Since Z is abelian, Z is soluble. Also, G/Z is soluble
by the inductive hypothesis. Thus, G is soluble by Lemma 5.8. Finally, as in the proof of
Lemma 5.8, we see that the composition factors of G are the composition factors of Z, together
with the composition factors of G/Z. These both consist precisely of of groups isomorphic to
Cp by the inductive hypothesis. This completes induction, whence the proof.

Example. Let G be a finite group of order 4pn , with p prime. We claim that G is soluble.
Indeed, let P be a Sylow p-subgroup of G, and set
\
x
K := P.
x∈G

Since K ≤ P and P is soluble by Theorem 5.17, we have that K is soluble by Theorem 5.16.
Now, K is the kernel of the action of G on G/P = {gP : g ∈ G}. Thus, G/K is isomorphic
to a subgroup of S4 by Proposition 2.10. It follows from Theorem 5.16 that G/K is soluble.
Since G/K and K are soluble, we deduce from Lemma 5.8 that G is soluble.
Theorem 5.18. Let G1 , G2 , . . . , Gt be finite soluble groups. Then G := G1 × . . . × Gt is
soluble.
Proof. We prove the claim by induction on t. If t = 1, then G = G1 and the result is trivial.
So assume that t > 1, and that the result holds for smaller values of t. Then

G = G1 × X

where X = G2 × . . . × Gt . Since X is soluble by the inductive hypothesis, it therefore suffices

to prove that the direct product of two finite soluble groups is soluble. This follows from
Homework 5, Question 6. (Recall that H × K ∼ = H ⋉ϕ K where ϕ : H → Aut(K) is the trivial
homomorphism (i.e. the one defined by ϕ(h) := IdK for all h ∈ H.))
Alternatively, consider the projection homomorphism π : G × X → X. We have N :=
ker(π) = G × {1X } ∼ = G, and Im(π) = X. Thus, N is soluble, and G/N ∼ = X is soluble
(applying the first isomorphism theorem). Lemma 5.8 now gives the result.

Remark. One could have also proved the previous theorem by showing that if G = G1 ×
. . . × Gt , then G(i) = (G1 )(i) × . . . × (Gt )(i) for all i.

64
Remark. In this chapter, we only defined finite soluble groups. More generally, you may see
in the literature that an arbitrary group G is defined to be soluble if G(n) = {1G } for some
n ∈ N. This agrees with Definition 5.7 on the set of finite groups, by Theorem 5.15.

5.4 Nilpotent groups (non-examinable)

In the final section of the course, we will study a special class of soluble groups, called nilpotent
groups.

Definition 5.19. For a group G, define γ1 (G) := G, and for i ∈ N, i ≥ 2, define γi (G) :=
[γi (G), G]. The descending series

G = γ1 (G) ≥ γ2 (G) ≥ · · · ≥ γn (G) ≥ γn+1 (G) ≥ · · ·

is called the lower central series of G.

Definition 5.20. A group G is said to be nilpotent if γn (G) = {1G } for some positive integer
n. If G is nilpotent and c is maximal such that γc (G) ̸= 1, then c is called the nilpotency
class of G.

The following exercise will be useful for our next proposition.

Exercise. Let G be a group, A ⊆ G, N ⊴ G. Show that ⟨A⟩N/N = ⟨B⟩, where B :=

{aN : a ∈ A}.

Proposition 5.21. Let G be a nilpotent group. Then

(i) Let H ≤ G. Then H is nilpotent.

(ii) Let N ⊴ G. Then γn (G/N ) = γn (G)N/N for all positive integers n. In particular, G/N
is nilpotent.

Proof. (i) It is clear from the definition that if H ≤ G, then γi (H) ⊆ γi (G) for all i. Thus, if
G is nilpotent and H ≤ G, then H is nilpotent.
(ii) Let N ⊴ G, and let g, h ∈ G. Then [gN, hN ] = [g, h]N . We will now prove, by induction
on n, that γn (G/N ) = γn (G)N/N . If n = 1, then GN = G, so γ1 (G/N ) = G/N = GN/N =
γ1 (G)N/N , as needed.
Assume now that n > 1, and that the claim holds for indices less than n. Then, applying
the inductive hypothesis to γn−1 (G/N ), we have

γn (G/N ) = ⟨[xN, gN ] : xN ∈ γn−1 (G/N ) = γn−1 (G)N/N, g ∈ G⟩

= ⟨[xN, gN ] : x ∈ γn−1 (G), g ∈ G⟩
= ⟨[x, g]N : x ∈ γn−1 (G), g ∈ G⟩ = γn (G)N/N

with the last inequality following from our exercise above. This completes our induction.
Finally, assume G is nilpotent. Then γn (G) = {1G } for some positive integer n,so
γn (G/N ) = γn (G)N/N = {1G }N/N = {1G/N }. Thus, G/N is nilpotent.

Corollary 5.22. Let G be a finite nilpotent group. Then G is soluble.

65
Proof. Consider the derived series

G = G(0) ≥ G(1) ≥ . . .

of G. By Theorem 5.15, it will suffice to prove that G(i) = {1G }. We will first prove, by
induction on n, that G(n) ≤ γn+1 (G), for all n ∈ N ∪ {0}. This follows from Definitions 5.14
and 5.19 if n = 0, so assume that n > 1, and that the claim holds for indices less than n.
Then G(n) = [G(n−1) , G(n−1) ] ≤ [γn (G), G] = γn+1 (G), using the inductive hypothesis. This
gives us what we need.

Remark. If we use the more general definition of soluble from the remark after Theorem 5.15
above, then our proof of Corollary 5.22 shows that an arbitrary (i.e. not necessarily finite)
nilpotent group G is always soluble

Proposition 5.23. Let G be a finite group of prime power order. Then G is nilpotent.

Proof. Write |G| = pn for p prime. We will prove that G is nilpotent by induction on n. If
n = 0, then G is the trivial group, so G is nilpotent.
So assume that n ≥ 1, and that the result holds for groups of order less than pn . By
Theorem 2.19, Z := Z(G) is not trivial. Thus, G/Z is a group of p-power and |G/Z| < |G|,
so the inductive hypothesis applies: G/Z is nilpotent. By definition, this implies that there
exists a positive integer n such that γn (G/Z) = {1G/Z }. By Proposition 5.21, we deduce
that γn (G)Z/Z = {1G/Z } = Z/Z, so γn (G)Z = Z by the Correspondence theorem. That is,
γn (G) ≤ Z. It follows that γn+1 (G) = [γn (G), G] ≤ [Z(G), G] = {1G }. Hence, γn+1 (G) =
{1G }, as needed.

Example. Let G = D2p , for p an odd prime. We claim that G is not nilpotent. Indeed,
write G = ⟨σ, τ ⟩ ≤ Sp as in Definition 2.4. Thus, σ = (1, 2, . . . , p) ∈ Sp and τ = (1, p)(2, p −
1) . . . ( p−1 p+1
2 , 2 ). In particular, τ στ
−1 = σ −1 (as we’ve noted many times before!).

As in the example prior to Definition 5.14, we deduce that γ2 (G) = N := ⟨σ⟩. Thus,
γ3 (G) = [γ2 (G), G] contains [σ, τ ] = σ −2 , so γ3 (G) contains ⟨σ −2 ⟩ = N = γ2 (G) (recall that
|σ| is odd). Hence, γ3 (G) = γ2 (G) = N , and it follows that γi (G) = N for all i. Hence, G is
not nilpotent.

Exercise. 1. Let G1 , . . . , Gt be finite groups. Prove that γi (G1 × . . . × Gt ) = γi (G1 ) ×

. . . × γi (Gt ). Deduce that if Gi is nilpotent for all 1 ≤ i ≤ t, then G1 × . . . × Gt is
nilpotent.

2. Let p1 , . . . , pt be primes, and let Pi be a group of pi -power order. By the previous

question and Proposition 5.23, P1 × . . . × Pt is nilpotent. Prove that a finite group G is
nilpotent if and only if G ∼ = P1 × . . . × Pt for some groups Pi , where each Pi has prime
power order. (Hint: Suppose that G is nilpotent. Using induction on the nilpotency
class, prove that |Sylp (G)| = 1 for all primes p dividing |G| (Homework 3, Question 8 is
very useful for this..)).

66
Appendices

67
Appendix A

3 tools from Chapters 1 and 2 for

analysing element orders in finite
groups

In Chapters 1 and 2, we have developed quite a few tools which will allow us to deduce deep
properties on the structure of a finite group. In particular, the following 3 tools are very
useful for deriving information on the orders of elements in a given finite group G:

Tool 1: If x ∈ G, then |G x| = |G : CG (x)| (reference: the Orbit-Stabiliser theorem).

Tool 2: If p is a prime divisor of |G|, then the number of elements of G of order p is congruent
to −1 modulo p (reference: Cauchy’s theorem).

Tool 3: If x, y ∈ G such that

(a) xy = yx; and

(b) ⟨x⟩ ∩ ⟨y⟩ = {1G },

then |xy| = lcm(|x|, |y|). In particular, if xy = yx and gcd(|x|, |y|) = 1, then |xy| =
|x||y|. (Reference: Lemma 1.9(iv)).

Sample questions (practice in applying the 5 tools above)

Question 1: Let p and q be distinct primes, and let G be a group of order pq. Prove that
either

(i) G is cyclic; or

(ii) Z(G) = {1G }.

Solution: Suppose first that G is abelian. By Tool 2, there exists x ∈ G of order p; and y ∈ G
of order q. Since G is abelian, we have xy = yx. Also, since p ̸= q, Lagrange’s theorem yields
⟨x⟩ ∩ ⟨y⟩ = {1G }. Thus, Tool 3 yields |xy| = lcm(|x|, |y|) = pq = |G|. Thus, G = ⟨xy⟩.
Assume next that G is nonabelian. Then Z(G) ̸= G, so Lagrange’s theorem implies that
|Z(G)| ∈ {1, p, q}. By HW 2, Q1, |G/Z(G)| = pq/|Z(G)| cannot be p or q (since all groups of
prime order are cyclic). Thus, |Z(G)| cannot be q or p. Hence, |Z(G)| = 1, i.e. Z(G) = {1G }.

69
Question 2: Let G be a finite group, and let m be odd. Then the number of elements of G
of order m is even.
Solution: Let X be the set of elements of G of order m. If X is empty, then we are done.
Otherwise, define a relation on X by x ∼ y if and only if y ∈ {x, x−1 }. It is easy to see that
∼ is an equivalence relation on X. Since X is a disjoint union of ∼-equivalence classes, and
(since m is odd) each equivalence class has size 2 (in fact has the form {x, x−1 }), we have
that |X| is even.
Question 3: Let G be a group of order 12, and suppose that G contains no elements of order
6. Prove that

(i) G has precisely 8 elements of order 3; and

(ii) G has precisely 3 elements of order 2.

Solution: For p ∈ {2, 3}, define

|Ep (G)| = {x ∈ G : |x| = p}.

So we need to prove that |E3 (G)| = 8 and |E2 (G)| = 3.

We will start with |E3 (G)|. By Tool 2 and Q2 above, the possibilities for |E3 (G)| are

|E3 (G)| ∈ {2, 8}.

(Tool 2 implies that the possibilities are |E3 (G)| ∈ {2, 5, 8, 11}, as these are the only integers
between 0 and |G| = 12 which are congruent to −1 modulo 3. Q2 above then rules out the
possibilities |E3 (G)| ∈ {5, 11}.) Suppose, in pursuit of a contradiction, that E3 (G) has size
2. Let x ∈ E3 (G). By Tool 1, and the fact that |g x| = |x| for all g ∈ G (the conjugate of a
group elements always has the same order as the element in a group – we saw this in Chapter
1), we have
|G : CG (x)| = |G x| ≤ |E3 (G)| = 2, so |G : CG (x)| ∈ {1, 2}.
Since |G| = 12 and |G : CG (x)| = 12/|CG (x)|, we deduce that |CG (x)| ∈ {6, 12}. In particular,
2 divides |CG (x)|, so applying Q2 above again implies that there exists y ∈ CG (x) of order 2.
Tool 3 then yields |xy| = 6, as in Question 1 above. This contradicts our assumption that G
contains no elements of order 6. Thus, we have proved that |E3 (G)| = 8.
Finally, since E2 (G) is contained in G \ (E3 (G) ⊔ {1G }), we have |E2 (G)| ≤ 3. It follows
from Tool 2 that |E2 (G)| ∈ {1, 3}. Suppose, in pursuit of a contraduiction, that |E2 (G)| = 1,
and let z ∈ E2 (G). Then since |z g | = |z| = 2 for all g ∈ G, and z is the unique element
of order 2, we have that z g = z for all g ∈ G. That is, z ∈ Z(G). Let x ∈ E3 (G). Then
xz = zx since z ∈ Z(G), so as in Question 1, we apply Tool 3 to get |xz| = 6. Again, this is
a contradiction. So |E2 (G)| = 3, as needed.
More sample questions (practice in applying the 5 tools above)
Try using the methods in Question 3 above to come up with your own solutions to the
following.

1. Let G be a group of order 24 such that G has no elements of order 6. Prove that G has
at least 8 elements of order 3; and at least 3 elements of order 2.

70
 2. Let G be a group of order 18 such that G has no elements of order 6. Prove that G
has precisely 9 elements of order 2. Hence or otherwise, prove that G has either 2 or 8
elements of order 3.

There will likely be a question exactly like one of these on the exam, so please take the
time to practice and understand the methods!

71
72
Appendix B

The Sylow game

In Lectures 16 and 17, we introduced the Sylow game. This is a game that has two versions.
Version 1: Proving that a group of a particular order cannot be simple
The first version of the game has the following form:
Let G be a group of order ∗. Prove that G cannot be simple.
One always starts with by assuming that G is simple. There are various ways to then find
a contradiction, but I usually have three strategies in mind:
Strategy 1: Fix a prime p dividing |G|, and use the two facts:

• |Sylp (G)| divides |G|/|P | (Corollary 3.7(ii)); and

• |Sylp (G)| ≡ 1 (mod p) (Sylow’s theorem 4)

to deduce that |Sylp (G)| = 1, i.e. that a Sylow p-subgroup (Say P ) of G is normal in G.
As long as G is not a p-group, we are done: P ⊴ G, and 1 ̸= P ̸= G, so G is not simple.

Strategy 2: Suppose that p is a prime with |G|p = p. Assume also that Strategy 1 did
not work: that is, we have failed to prove that |Sylp (G)| = 1 using Corollary 3.7(ii) &
Sylow’s theorem 4. In any case, Corollary 3.7(ii) & Sylow’s theorem 4 give us a list of
possibilities for |Sylp (G)|. Let Fp (G) be as in Corollary 3.8. Then by Corollary 3.8(iii),
we have
|Fp (G)| = |Sylp (G)|(p − 1).
Let q ̸= p be another prime dividing |G|. Then Fq (G) ⊆ G \ (Fp (G) ⊔ {1G }), so
|Fq (G)| ≤ |G| − |Sylp (G)|(p − 1) − 1. If this latter number is at most |G|q − 1, then
Corollary 3.8(ii) tells us that G has a normal Sylow q-subgroup, say Q. As long as G is
not a p-group, we are then done: we’ve shown that Q ⊴ G, and 1 ̸= Q ̸= G, so G is not
simple.

Strategy 3: Assume that Strategy 1 did not work: that is, we have failed to prove that
|Sylp (G)| = 1 for some prime p using Corollary 3.7(ii) & Sylow’s theorem 4. In any
case, Corollary 3.7(ii) & Sylow’s theorem 4 give us a list of possibilities for |Sylp (G)|.
Assume that G is simple. Then of course, |Sylp (G)| > 1. Let K be the kernel of the
action of G in its conjugation action on Sylp (G). Then, as noted in the Lecture 16, we
have
K ̸= G, since |Sylp (G)| = |G : NG (P )| > 1 and K ≤ NG (P ).

73
In other words, G acts non-trivially on Sylp (G). Thus, K = {1G }, since we are assuming
that G is simple.
So how do we do that? Well, suppose that |Sylp (G)| = n > 1. Then Proposition 2.13
tells us that G is isomorphic to a subgroup of Sn . Thus, Lagrange’s theorem implies
that |G| divides n!. If |G| does not divide n!, then we have our contradiction!

74
The examples

Prove that a group of order 20 cannot be simple. And

Prove that a group of order 48 cannot be simple.

are in the lecture notes, so let’s do another example.

Example: Prove that a group of order 36 cannot be simple.
Solution: Let G be a group of order 36 = 4 · 9, and assume that G is simple.
Trying Strategy 1: We have |Syl3 (G)| divides 4 by Corollary 3.7(ii), and |Syl3 (G)| ≡ 1
(mod 3) by Sylow’s theorem 4. Thus, |Syl3 (G)| ∈ {1, 4}. Also, |Syl2 (G)| divides 9 by Corollary
3.7(ii), and |Syl2 (G)| ≡ 1 (mod 2) by Sylow’s theorem 4. Thus, |Syl2 (G)| ∈ {1, 3, 9}.
So Strategy 1 did not work! Still, it gives us a list of possibilities for Syl3 (G), which we will
use. Note that Strategy 2 is not really effective here, since |G|p ̸= p for any p dividing |G|.
Thus, let’s try..
Trying Strategy 3: We have |Syl3 (G)| ∈ {1, 4} as noted above. Since G is simple, we must
have |Syl3 (G)| = 4. Let K be the kernel of the conjugation action of G on Syl3 (G). Then
K ̸= G, since |Syl3 (G)| > 1 (in the exam, it might be good to say here “K ̸= G by Sylow’s
theorem 2, since |Syl3 (G)| > 1”).
Thus, K = {1G }. We then have that G is isomorphic to a subgroup of S4 by Prop 2.13.
Since |S4 | = 24, and |G| = 36, we have our contradiction.
Example: Prove that a group of order 80 cannot be simple.
Solution: Let G be a group of order 80 = 16 · 5, and assume that G is simple.
Trying Strategy 1: We have |Syl5 (G)| divides 16 by Corollary 3.7(ii), and |Syl5 (G)| ≡ 1
(mod 5) by Sylow’s theorem 4. Thus, |Syl5 (G)| ∈ {1, 16}. Hence, |Syl5 (G)| = 16 since G is
simple. Also, |Syl2 (G)| divides 5 by Corollary 3.7(ii), and |Syl2 (G)| ≡ 1 (mod 2) by Sylow’s
theorem 4. Thus, |Syl2 (G)| ∈ {1, 5}. Hence, |Syl5 (G)| = 5 since G is simple.
So again, Strategy 1 did not work! As above, it still gives us very good information on
Syl5 (G), which we will use below. Note that |G|5 = 5 , so Strategy 2 might be good to try..
Trying Strategy 2: We have |Syl5 (G)| = 16. Then |F5 (G)| = 16 · 4 = 64 by Corollary
3.8(iii). It follows that |F2 (G)| ≤ 80 − 64 − 1 = 15 = |G|2 − 1. Thus, G has a normal Sylow
2-subgroup, say Q, by Corollary 3.8(ii). Since |Q| = 16, we have 1 ̸= Q ̸= G, so G is not
simple.

Note that Strategy 3 would not have worked with p = 5 above, since 80 does divide |S16 | =
16!. However, Strategy 3 would have worked with p = 2. Indeed, we have |Syl2 (G)| = 5. Since
80 does not divide 5! = 120, we could have argued in the usual way and got our contradiction.

75
We now look at the second version of the Sylow game..
Version 2: Proving that a particular group is simple
Note: Since we didn’t get time to do enough examples in the lectures, I won’t ask a question
on Version 2 in the exam. But the techniques are very useful, so please do have a look at the
ideas and strategies.
The second version of the Sylow game has the following form:
Let G be the group ∗. (Or, let G be a group satisfying a particular list of properties.) Prove
that G is simple.
The main tool that we use for this form of the game is Corollary 3.9. The strategy is
as follows: Suppose (in pursuit of a contradiction) that N is a normal subgroup of G with
1 ̸= N ̸= G.
Step A: Choose a prime p with |G|p = p, and suppose that p divides |N |. Then p does not
divide |G : N | = |G|/|N |, since |G|p = p. Thus,
• |Sylp (G)| = |Sylp (N )| (Corollary 3.9(ii)(a)); and
• |Fp (G)| = |Fp (N )| (Corollary 3.9(ii)(b)).
We now use Corollary 3.7(ii) & Sylow’s theorem 4 to list the possibilities for |Sylp (G)| =
|Sylp (N )|. By the above, this them gives us the possibilities for |Fp (N )| = |Fp (G)|. The
idea is to prove that if |Sylp (N )| = |Sylp (G)| is “small”, then |Fp (G)| = |Fp (N )| is
“small”, and this will usually contradict a property of the group G we are working with
(see the example of G = A5 in the lecture notes).
Thus, we conclude that |Fp (G)| = |Fp (N )| is “large”. Since |N | ≥ |Fp (N )| + 1, we then
have a lower bound on |N |.
Step B: Since |N | divides |G|, we use our lower bound from Step A to give us a list of
possibilities for |N | (see the example of G = A5 in the lecture notes.) Note that if q ̸= p
is prime, and we have shown that q must divide |N |, then N has an element of order q by
Cauchy’s theorem. Moreover, G x ⊆ N by Corollary 3.9(i), so |N | ≥ |G x| + 1 + |Fp (N )|.
Does this give us a contradiction?
If not, and |G|q = q, then we can repeat Step A with q in place of p, and get a list of
possibilities for |Fq (N )|. We then have |N | ≥ |Fq (N )| + 1 + |Fp (N )|. Does this give us
a contradiction?
Step C: Repeat Steps A and B for each prime p with |G|p = p. Hopefully we get a contra-
diction in each case using our approach above..
Step D: As we said, hopefully Steps A–C allow us to deduce that |N | can only be divisible
by primes p with |G|p ̸= p. The idea is then to use the properties of G to deduce a
contradiction. For example, suppose we have shown that 2 is the only prime dividing
|N |. Then we know that |N | divides |G|2 . If we know that |G x| ≥ |G|2 for all elements x
of G of order 2, then we have a contradiction, since G x ⊔ {1G } ⊆ N by Corollary 3.9(ii).

Since there are not many “small” positive integers n for which there exists a nonabelian
finite simple group of order n, we can’t really be more specific than the above, since the

76
methods can vary wildly! Indeed, by the Classification of Finite Simple Groups, the orders
of the smallest 5 nonabelian finite simple groups are 60, 168, 360, 504, and 660.
However, the approach above works very well if all but one of the prime divisors p of |G|
has the property that |G|p = p (for example, if |G| ∈ {60, 168, 660}). See the below for some
exercises in applying these ideas..
Exercises in practising the Sylow game:

Exercises for Version 1:

1. Let G be a group of order 24. Prove that G cannot be simple.

2. Let G be a group of order 28. Prove that G cannot be simple.

3. Let G be a group of order 72. Prove that G cannot be simple.

4. Let G be a group of order 112. Prove that G cannot be simple.

5. Let G be a group of order 160. Prove that G cannot be simple.

6. Let G be a group of order 196. Prove that G cannot be simple.

Exercises for Version 2:

1. Let G be a group of order 168, and assume that

(a) All elements of G of order 2 are conjugate, and if x is such an element, then
|G x| = 21.
(b) G has at least 7 elements of order 7;
(c) G has 56 elements of order 3.

Prove that G is simple.

2. Let G be a group of order 660, and assume that

(a) All elements of G of order 2 are conjugate, and if x is such an element, then
|G x| = 55.
(b) G has 120 elements of order 11;
(c) G has 264 elements of order 5.
(d) G has 110 elements of order 3.

Prove that G is simple.

We remark that in the last two questions above, the simple groups G := SL(3, 2) and
G := SL(2, 11)/Z(SL(2, 11)) satisfy the listed conditions, respectively.

77
78
Books and sources used in the
course:

While preparing this course I made use of the following works:

• MA249 and MA442 Notes written by Derek Holt.

• “Abstract Algebra” by D. S Dummit and R. M. Foote.

• “Finite Group Theory” by I .M. Isaacs.

• “Finite Groups: an Introduction” by J.-P. Serre.

79