Normal Subgroups: The Symmetry Behind Quotients

Learn to spot normal subgroups quickly, understand why they are exactly what makes quotient groups work, and connect normality to kernels, conjugation, and classification ideas like simple groups and extensions.

Cosets feel like they should multiply. You write $(gN)(hN)=(gh)N$ and move on. The catch is that this rule is not automatically well-defined, because $gN$ and $g'N$ can name the same coset while $ghN$ and $g'hN$ might not. Normal subgroups are precisely the condition that makes that ambiguity disappear.

The payoff is practical. Once you see normality as a symmetry condition inside the group, the standard tests stop feeling like unrelated tricks, and quotient groups become a tool you can trust instead of a definition you memorize. To get your hands on the issue, start by watching cosets misbehave.

When it works, you can slide elements of $N$ past group elements without changing the coset product. When it fails, the product depends on the representative you chose, so you cannot define a group operation on cosets consistently.

Three equivalent ways to recognize normality

A subgroup $N\le G$ is normal if it is stable under the group’s internal symmetries. Formally, normality can be stated in a few ways that look different but are the same condition.

Conjugation invariance

$N$ is normal in $G$ if for every $g\in G$, $gNg^{-1}=N$. Conjugation is the operation that compares the subgroup from different vantage points inside the group. Normal means every vantage point sees the same subset.

Left cosets equal right cosets

$N$ is normal in $G$ if for every $g\in G$, $gN=Ng$. This is the coset statement that directly fixes the quotient-group ambiguity.

Kernel form

$N$ is normal in $G$ if $N=\ker(\varphi)$ for some homomorphism $\varphi:G\to H$. This is often the easiest way to prove normality, because kernels come with built-in conjugation stability.

See how these equivalences line up across examples.

One move
To show $gNg^{-1}\subseteq N$, take $n\in N$ and rewrite $gng^{-1}$ as something you already know lies in $N$, often by using a homomorphism or a defining relation.

Fast recognition tests you can use immediately

When you are doing computations in concrete groups, you rarely want to chase the definition from scratch. A few patterns detect normality quickly.

• Index 2: If $[G:N]=2$, then $N\trianglelefteq G$ because there are only two left cosets, and the nontrivial one must also be the only nontrivial right coset.
• Abelian ambient group: If $G$ is abelian, every subgroup is normal since $gng^{-1}=n$.
• The center: The center $Z(G)$ is always normal, and any subgroup of $Z(G)$ is normal in $G$.
• Uniqueness by order or property: If $N$ is the unique subgroup of a given order (or the unique subgroup with a defining property preserved by automorphisms), then conjugation must send $N$ to itself, so $N$ is normal.

A common trap is assuming that being large makes a subgroup normal. High index or high order alone does not help unless it forces uniqueness or index 2.

Compare these sufficient conditions, including the typical places they fail.

Normality is not about size. It is about being fixed by conjugation, which is an action of $G$ on its own subgroups.

Normality as closure under internal symmetry

Conjugation does not just test normality, it generates it. If you start with a subset $S\subseteq G$, the group elements will produce all its conjugates $gsg^{-1}$. The smallest normal subgroup that contains $S$ has to contain all those conjugates, and it has to be a subgroup, so it also contains everything you can build from them by multiplication and inverses.

This motivates the normal closure of $S$ in $G$, often written $\langle S\rangle^G$. Conceptually, it is what you get after forcing $S$ to be compatible with every symmetry coming from elements of $G$.

Watch how adjoining conjugates expands a set into a normal subgroup.

Mental model
Normal closure is what you must add so that the statement contains $s$ becomes invariant under replacing $s$ by a conjugate.

Quotient groups depend on normality

A quotient group $G/N$ is the set of left cosets $\{gN:g\in G\}$ with multiplication defined by

$$(gN)(hN)=(gh)N.$$

This definition has one job. It must not depend on the choice of representatives.

To check well-definedness, assume $gN=g'N$ and $hN=h'N$. That means $g'=gn_1$ and $h'=hn_2$ for some $n_1,n_2\in N$. Then

$$g'h'=(gn_1)(hn_2)=g(n_1h)n_2.$$

For this to land in the same coset as $gh$, you need $n_1h$ to be rewritable as $h(\text{something in }N)$. That is exactly the condition $Nh=hN$ for all $h$, which is equivalent to $N\trianglelefteq G$.

If $N$ is not normal, you can still form the set of cosets, but you cannot turn it into a group using the naive multiplication rule. What fails is not associativity or identity, it is that the operation is not well-defined in the first place.

Where normal subgroups come from naturally

If normality feels like a special property you have to verify case-by-case, shift your viewpoint. Normal subgroups are produced by standard constructions.

The most important source is kernels. If $\varphi:G\to H$ is a homomorphism, then $\ker(\varphi)\trianglelefteq G$ because for any $k\in\ker(\varphi)$,

$$\varphi(gkg^{-1})=\varphi(g)\varphi(k)\varphi(g)^{-1}=\varphi(g)e\varphi(g)^{-1}=e.$$

Another canonical normal subgroup is the commutator subgroup $[G,G]$, generated by commutators $[g,h]=ghg^{-1}h^{-1}$. It measures how far $G$ is from abelian, and the quotient $G/[G,G]$ is the largest abelian quotient of $G$.

Normal subgroups also behave well under intersections. If $N_1,N_2\trianglelefteq G$, then $N_1\cap N_2\trianglelefteq G$ because conjugation preserves each $N_i$, hence preserves their intersection. Products are subtler. Even if $N_1,N_2$ are normal, $N_1N_2$ is a subgroup and is normal, but it need not behave like a direct product unless extra conditions hold.

Explore these closure properties and the fine print.

Use normal subgroups to classify groups

A normal subgroup is a handle. If $N\trianglelefteq G$, then $G$ can be studied via the pair $N$ and $G/N$, together with how they fit together. This is the idea behind group extensions. Many classification arguments are variations on the same move.

• If $G$ has no nontrivial proper normal subgroups, then $G$ is simple, and quotients do not break it apart.
• If you can find a nontrivial normal $N$ and understand $N$ and $G/N$, you often reduce a hard question about $G$ to easier questions about smaller groups.

A concrete next step is to pick one group you already know well, like $S_3$, $D_{2n}$, or a matrix group you have computed in, and list its normal subgroups using only the equivalences and quick tests above. The pattern you see is the extension structure peeking through.