The group axioms: why these three?

Associativity, identity, inverses -- textbooks present these axioms as if they were carved in stone. But why not two? Why not four? By systematically adding and removing axioms, we discover the engineering brilliance behind the definition of a group.

Folio Official

February 27, 2026

Open any algebra textbook and you will find the definition of a group stated with quiet authority, as though it were the only possible choice:

Definition 1.

A pair

(G, \cdot)

consisting of a set

G

and a binary operation

\cdot : G \times G \to G

is a group if the following three conditions hold:

Associativity: $(a \cdot b) \cdot c = a \cdot (b \cdot c)$ for all $a, b, c \in G$ .
Identity: there exists $e \in G$ such that $e \cdot a = a \cdot e = a$ for all $a \in G$ .
Inverses: for each $a \in G$ there exists $b \in G$ such that $a \cdot b = b \cdot a = e$ .

What the textbook does not say – and what is almost never discussed – is why these three, and not some other list. Would two axioms suffice? Would a fourth make things better? The axioms did not descend from heaven; they are a deliberate design choice, refined over two centuries. To appreciate just how well-calibrated that choice is, let us experiment: first by stripping axioms away, then by trying to add one more.

1 Stripping axioms away

What happens when we remove the axioms one at a time, starting from the bottom?

Drop inverses: monoids. If we keep associativity and the identity element but drop the requirement that every element have an inverse, we obtain a monoid.

Example 2.

The natural numbers

(N, +)

form the prototypical monoid. Addition is associative, and

0

serves as the identity. But there is no natural number

x

satisfying

3 + x = 0

. You can add freely, but you can never undo addition – subtraction is simply unavailable.

Example 3.

The set

M_{n} (R)

of all

n \times n

real matrices forms a monoid under multiplication, with the identity matrix

I

playing the role of the identity element. Singular matrices (those with

det = 0

), however, have no inverses. Multiplying by such a matrix is a one-way operation: information is destroyed, and there is no getting it back.

The cost of losing inverses is clear: in a monoid, the equation $a x = b$ need not have a solution. The inverse axiom is precisely what guarantees that equations of this form can always be solved.

Drop the identity as well: semigroups. Keep only associativity, and the resulting structure is called a semigroup.

Example 4.

The positive even numbers

{2, 4, 6, \dots}

form a semigroup under ordinary addition. Since

0

does not belong to this set, there is no identity element.

In a semigroup, you cannot even express the idea of "doing nothing." There is no trivial element, no idle operation – the algebraic structure has no resting state.

Drop everything: magmas. Remove associativity as well, and all that remains is a set equipped with a binary operation – a magma. At this level of generality, almost nothing can be proved. Because $(a \cdot b) \cdot c$ and $a \cdot (b \cdot c)$ may differ, even the expression $a \cdot b \cdot c$ is ambiguous without parentheses. Notation itself begins to break down.

2 The hierarchy

Let us organize these structures into a chain, where each step adds one axiom:

Magma + associativity Semigroup + identity Monoid + inverses Group

At each level, we gain a concrete new capability:

Associativity lets us write $ab c$ without worrying about parentheses.
An identity element gives us a way to express "do nothing."
Inverses guarantee that every operation can be undone.

Remove any single axiom from the definition of a group and one of these capabilities vanishes.

3 What about a fourth axiom?

Having seen that removing axioms costs too much, let us try the opposite: what if we imposed one more? The most natural candidate is commutativity– the requirement that $ab = ba$ for all elements.

Definition 5.

A group

(G, \cdot)

satisfying

a \cdot b = b \cdot a

for all

a, b \in G

is called an abelian group (or commutative group).

The integers under addition, $(Z, +)$ , are abelian, and so are many other familiar groups. So why not build commutativity into the definition from the start? Because doing so would gut the theory of its most important examples.

Example 6.

In the symmetric group

S_{3}

, consider the two permutations

σ = (122133), τ = (112332)

Computing both compositions explicitly:

σ τ = (122331), τ σ = (132132)

we see that

σ τ \neq = τ σ

Requiring commutativity would exclude $S_{n}$ for $n \geq 3$ , the general linear group $G L_{n} (R)$ for $n \geq 2$ , the Rubik's Cube group, and a vast menagerie of other structures. These are precisely the groups that describe symmetry in the real world – the very objects that motivated the invention of group theory in the first place. Insisting on commutativity would amount to designing a language for symmetry that cannot express most symmetries.

Remark 7.

The omission of commutativity from the axioms was not an oversight or a sign of laziness. It was a conscious design decision: impose enough structure to yield a powerful and coherent theory, but not so much that the theory excludes its most important applications.

4 The sweet spot

The three group axioms occupy a sweet spot between too little and too much:

Minimal. Drop any one of the three and you lose the guarantee that the equation $a x = b$ always has a unique solution.
Powerful. From these three axioms alone, one can derive the cancellation law, Lagrange's theorem, the Sylow theorems, and the entire edifice of group theory.
Inclusive. By not demanding commutativity, the definition accommodates symmetric groups, matrix groups, transformation groups, and a wealth of non-abelian structures where the deepest phenomena reside.

To illustrate just how much follows from so little, consider the cancellation law:

Theorem 8 (Cancellation Law).

In a group

G

, if

ab = a c

then

b = c

(left cancellation). Similarly, if

ba = c a

then

b = c

(right cancellation).

Proof.

Suppose

ab = a c

. Left-multiply both sides by

a^{- 1}

: then

a^{- 1} (ab) = a^{- 1} (a c)

. By associativity,

(a^{- 1} a) b = (a^{- 1} a) c

, which gives

e b = ec

, and hence

b = c

. □

5 Takeaway

That the group axioms are "these three" is no accident. Associativity, the existence of an identity, and the existence of inverses together formalize a minimal requirement: that operations can be composed without ambiguity, that doing nothing is an available option, and that every operation is reversible. Any fewer axioms and the structure is too weak to support a meaningful theory; add commutativity and the scope narrows to a fraction of what matters. The definition of a group is the product of two centuries of mathematical refinement – an exquisitely balanced compromise between generality and power.

Group Theory Algebra Between the Lines

Folio Official

Mathematics "between the lines" — exploring the intuition textbooks leave out, written in LaTeX on Folio.

1 followers·107 articles

The group axioms: why these three?

Folio Official

February 27, 2026

Open any algebra textbook and you will find the definition of a group stated with quiet authority, as though it were the only possible choice:

Definition 1.

A pair

(G, \cdot)

consisting of a set

G

and a binary operation

\cdot : G \times G \to G

is a group if the following three conditions hold:

Associativity: $(a \cdot b) \cdot c = a \cdot (b \cdot c)$ for all $a, b, c \in G$ .
Identity: there exists $e \in G$ such that $e \cdot a = a \cdot e = a$ for all $a \in G$ .
Inverses: for each $a \in G$ there exists $b \in G$ such that $a \cdot b = b \cdot a = e$ .

1 Stripping axioms away

What happens when we remove the axioms one at a time, starting from the bottom?

Drop inverses: monoids. If we keep associativity and the identity element but drop the requirement that every element have an inverse, we obtain a monoid.

Example 2.

The natural numbers

(N, +)

form the prototypical monoid. Addition is associative, and

0

serves as the identity. But there is no natural number

x

satisfying

3 + x = 0

. You can add freely, but you can never undo addition – subtraction is simply unavailable.

Example 3.

The set

M_{n} (R)

of all

n \times n

real matrices forms a monoid under multiplication, with the identity matrix

I

playing the role of the identity element. Singular matrices (those with

det = 0

), however, have no inverses. Multiplying by such a matrix is a one-way operation: information is destroyed, and there is no getting it back.

The cost of losing inverses is clear: in a monoid, the equation $a x = b$ need not have a solution. The inverse axiom is precisely what guarantees that equations of this form can always be solved.

Drop the identity as well: semigroups. Keep only associativity, and the resulting structure is called a semigroup.

Example 4.

The positive even numbers

{2, 4, 6, \dots}

form a semigroup under ordinary addition. Since

0

does not belong to this set, there is no identity element.

In a semigroup, you cannot even express the idea of "doing nothing." There is no trivial element, no idle operation – the algebraic structure has no resting state.

2 The hierarchy

Let us organize these structures into a chain, where each step adds one axiom:

Magma + associativity Semigroup + identity Monoid + inverses Group

At each level, we gain a concrete new capability:

Associativity lets us write $ab c$ without worrying about parentheses.
An identity element gives us a way to express "do nothing."
Inverses guarantee that every operation can be undone.

Remove any single axiom from the definition of a group and one of these capabilities vanishes.

3 What about a fourth axiom?

Definition 5.

A group

(G, \cdot)

satisfying

a \cdot b = b \cdot a

for all

a, b \in G

is called an abelian group (or commutative group).

Example 6.

In the symmetric group

S_{3}

, consider the two permutations

σ = (122133), τ = (112332)

Computing both compositions explicitly:

σ τ = (122331), τ σ = (132132)

we see that

σ τ \neq = τ σ

Remark 7.

4 The sweet spot

The three group axioms occupy a sweet spot between too little and too much:

Minimal. Drop any one of the three and you lose the guarantee that the equation $a x = b$ always has a unique solution.
Powerful. From these three axioms alone, one can derive the cancellation law, Lagrange's theorem, the Sylow theorems, and the entire edifice of group theory.
Inclusive. By not demanding commutativity, the definition accommodates symmetric groups, matrix groups, transformation groups, and a wealth of non-abelian structures where the deepest phenomena reside.

To illustrate just how much follows from so little, consider the cancellation law:

Theorem 8 (Cancellation Law).

In a group

G

, if

ab = a c

then

b = c

(left cancellation). Similarly, if

ba = c a

then

b = c

(right cancellation).

Proof.

Suppose

ab = a c

. Left-multiply both sides by

a^{- 1}

: then

a^{- 1} (ab) = a^{- 1} (a c)

. By associativity,

(a^{- 1} a) b = (a^{- 1} a) c

, which gives

e b = ec

, and hence

b = c

. □

5 Takeaway

Group Theory Algebra Between the Lines

Folio Official

Mathematics "between the lines" — exploring the intuition textbooks leave out, written in LaTeX on Folio.

1 followers·107 articles

The group axioms: why these three?

1 Stripping axioms away

2 The hierarchy

3 What about a fourth axiom?

4 The sweet spot

5 Takeaway

Share your expertise with the world

More from Folio Official

Cosets and quotient groups: the art of controlled forgetting

Cosets and Lagrange's Theorem

Composition Series and the Jordan--Hölder Theorem

What is "dimension," really? The truth about degrees of freedom

The group axioms: why these three?

1 Stripping axioms away

2 The hierarchy

3 What about a fourth axiom?

4 The sweet spot

5 Takeaway

Share your expertise with the world

More from Folio Official

Cosets and quotient groups: the art of controlled forgetting

Cosets and Lagrange's Theorem

Composition Series and the Jordan--Hölder Theorem

What is "dimension," really? The truth about degrees of freedom