Consider a group of order $200 = 2^3 \cdot 5^2$. How many subgroups does it have? Can it be simple? Does it have a normal subgroup of order $25$? Without any structural theory, these questions seem impenetrable — you would need to classify all possible multiplication tables, a task that quickly becomes combinatorially hopeless even for groups of moderate size. What Sylow's theorems give you is a complete answer to the first crude question about subgroup structure: for every prime power dividing the group order, a subgroup of that size exists, and the number of such subgroups is sharply constrained. They are one of the most powerful elementary tools in finite group theory. The key insight is that the prime factorization of $|G|$ tells you far more about $G$ than it might seem. Lagrange's theorem says that subgroup orders divide $|G|$; the Sylow theorems are a partial converse. If $p^k$ divides $|G|$, then $G$ has a subgroup of order $p^k$ — and the maximal such subgroup (the Sylow $p$-subgroup) enjoys special properties: any two of them are conjugate, and their number satisfies a rigid congruence condition. [example: A Group Where Sylow Theory Prevents Simplicity] Take $G$ a group of order $|G| = 30 = 2 \cdot 3 \cdot 5$. Suppose $G$ were simple. The number $n_5$ of Sylow $5$-subgroups must satisfy $n_5 \equiv 1 \pmod{5}$ and $n_5 \mid 6$, so $n_5 \in \{1, 6\}$. If $n_5 = 1$, the unique Sylow $5$-subgroup is normal, contradicting simplicity. So $n_5 = 6$. Similarly $n_3 \equiv 1 \pmod{3}$ and $n_3 \mid 10$, giving $n_3 \in \{1, 10\}$; simplicity forces $n_3 = 10$. Count elements: $6$ Sylow $5$-subgroups of order $5$, each pair intersecting only at $e$, contribute $6 \cdot 4 = 24$ non-identity elements. Ten Sylow $3$-subgroups contribute $10 \cdot 2 = 20$ non-identity elements. That is $44$ non-identity elements in a group of order $30$. This is impossible. Therefore no group of order $30$ is simple — every such group has a normal Sylow subgroup. [/example] This kind of argument — counting elements in Sylow subgroups to force a contradiction — is a template that works for dozens of specific orders. The theorems turn structural questions into arithmetic, and arithmetic into group theory. ## Definition Before stating the theorems, we need to isolate the objects they are about. To appreciate what Sylow theory gives us, consider what happens without it. Suppose you want to know whether a group of order $12$ has a subgroup of order $4$. Lagrange's theorem says subgroup orders divide $|G|$, so $4$ is a legal order — but it says nothing about existence. You might try to construct such a subgroup by hand, but there is no guarantee from first principles that a subgroup of every legal order exists. In fact, for groups of order $12$, such subgroups do exist — but the proof requires exactly the machinery we are building. The failure mode is this: for $A_4$ of order $12$, there is no subgroup of order $6$ even though $6 \mid 12$. Sylow theory explains this precisely: a subgroup of order $6$ would have index $2$, hence be normal, but $A_4$'s only normal subgroup is the Klein four-group $V_4 = \{e, (12)(34), (13)(24), (14)(23)\}$ of order $4$ (a direct check of all $\binom{12}{6} = 924$ six-element subsets of $A_4$ shows that none is closed under multiplication). The Sylow theorems work for prime-power orders, not arbitrary divisors, and this is not a weakness — it is the honest statement of what is true. What is special about a group of prime power order? Such groups, called $p$-groups, are unusually constrained: their centres are nontrivial, they have normal subgroups of every order dividing $|G|$, and they are nilpotent. They are the basic atoms of finite group theory from the prime-power perspective. In a general finite group $G$, the Sylow $p$-subgroups are the largest $p$-power subgroups — the ones at the top of the $p$-power hierarchy. [definition: $p$-Group] Let $p$ be a prime. A finite group $G$ is a **$p$-group** if $|G| = p^k$ for some $k \ge 0$. More generally, a subgroup $H \le G$ is a $p$-subgroup of $G$ if $|H|$ is a power of $p$. [/definition] The condition $k \ge 0$ includes the trivial group (when $k = 0$), which is a $p$-group for every prime. In practice we usually consider $k \ge 1$. A basic fact that underpins everything is Cauchy's theorem: if $p \mid |G|$, then $G$ contains an element of order $p$. The Sylow theorems massively strengthen this — not just one element of order $p$, but an entire subgroup of order $p^k$ where $p^k$ is the highest power of $p$ dividing $|G|$. [definition: Sylow $p$-Subgroup] Let $G$ be a finite group and $p$ a prime. Write $|G| = p^a \cdot m$ where $p \nmid m$ and $a \ge 1$. A **Sylow $p$-subgroup** of $G$ is a subgroup $P \le G$ with $|P| = p^a$. [/definition] Equivalently, a Sylow $p$-subgroup is a maximal $p$-subgroup of $G$: it is not properly contained in any larger $p$-subgroup. The equivalence between these two characterisations is itself a theorem — it is part of what makes the Sylow theory work. As we apply the theorems, we will need to count Sylow subgroups. The count turns out to be a key invariant: knowing whether there is exactly one Sylow $p$-subgroup or many of them determines whether $G$ carries a normal subgroup. It is worth giving this count a name so we can state the third Sylow theorem cleanly. [definition: $n_p(G)$] For a finite group $G$ and prime $p$ with $p \mid |G|$, define \begin{align*} n_p(G) &= \text{the number of Sylow } p\text{-subgroups of } G. \end{align*} When $G$ is clear from context, we write $n_p$. [/definition] Now we can state the three Sylow theorems. Together they assert existence, conjugacy, and a congruence condition on $n_p$. The remarkable thing is that pure divisibility arithmetic — no multiplication tables, no generators, no structure at all — forces these conclusions. [quotetheorem:3249] Existence is only the beginning. The second theorem says all Sylow $p$-subgroups are not just structurally similar but literally the same up to conjugation — they are all in the same orbit of the conjugation action of $G$ on its subgroups. [quotetheorem:3250] Conjugacy controls the shape; the third theorem controls the number. Both conditions together — a congruence and a divisibility — will typically leave only a handful of possibilities for $n_p$, often only $n_p = 1$. [quotetheorem:3251] The two conditions $n_p \equiv 1 \pmod{p}$ and $n_p \mid m$ are what we will use over and over. Before seeing the first application, it is worth extracting the most important special case: when does the third theorem force a normal subgroup? [remark: Sylow's Second Implies Normality Criterion] A Sylow $p$-subgroup $P \le G$ is normal in $G$ if and only if $n_p = 1$. This is immediate from the second theorem: $P \trianglelefteq G$ iff $P$ is closed under conjugation by all of $G$, iff $P$ is the only conjugate of itself, iff $n_p = 1$. [/remark] This criterion will be our primary tool in the applications that follow. But to appreciate why it is powerful, it helps to understand how the three theorems interlock. [explanation: Why These Three Conditions Together Are So Powerful] The three theorems interlock. The first guarantees existence — you do not have to search for Sylow subgroups, they are guaranteed to be there. The second says all Sylow $p$-subgroups are structurally identical (conjugate), so there is essentially one Sylow $p$-subgroup "up to conjugation." The third constrains how many there are. The conditions $n_p \equiv 1 \pmod{p}$ and $n_p \mid m$ together are often very restrictive. For $|G| = p^a \cdot m$, the divisors of $m$ that are $\equiv 1 \pmod{p}$ form a short list — often containing only $1$, forcing the Sylow subgroup to be normal. When $n_p = 1$, the Sylow $p$-subgroup is both unique and normal, and $G$ has a distinguished quotient $G/P$ of order $m$. This is the beginning of understanding how $G$ is built from $p$-groups and groups of smaller order. [/explanation] ## Conjugacy and the Orbit-Counting Argument The proofs of all three Sylow theorems use the same basic machine: let $G$ act on a carefully chosen set and apply the orbit-stabiliser theorem. Understanding this machine is worth the time — it explains not just the Sylow theorems but a whole family of counting arguments in finite group theory. The key setup for the second theorem is: let $G$ act on the set $\text{Syl}_p(G)$ of all Sylow $p$-subgroups by conjugation, \begin{align*} g \cdot P &= gPg^{-1}, \quad g \in G, \; P \in \text{Syl}_p(G). \end{align*} By the first theorem, $\text{Syl}_p(G)$ is nonempty. The second theorem says this action is transitive — there is only one orbit. The third theorem falls out of the same calculation. Fix any $P \in \text{Syl}_p(G)$ and let $P$ act on $\text{Syl}_p(G)$ by conjugation (restricting the $G$-action). The fixed points of this $P$-action are the Sylow $p$-subgroups $Q$ with $pQp^{-1} = Q$ for all $p \in P$, i.e., $P \le N_G(Q)$. Since $P$ and $Q$ are both Sylow $p$-subgroups and $P \le N_G(Q)$, a careful argument shows $P = Q$. So $P$ is the only fixed point. Now orbit sizes divide $|P| = p^a$, so every non-fixed orbit has size divisible by $p$. Counting modulo $p$: \begin{align*} n_p &= |\text{Syl}_p(G)| \equiv |\text{Fix}_P(\text{Syl}_p(G))| = 1 \pmod{p}. \end{align*} The condition $n_p \mid m$ comes from the orbit-stabiliser theorem applied to the transitive $G$-action on $\text{Syl}_p(G)$: the orbit size $n_p = [G : \text{Stab}_G(P)] = [G : N_G(P)]$. Since $P \le N_G(P)$ and $|P| = p^a$, we have $p^a \mid |N_G(P)|$, so $|N_G(P)| = p^a \cdot d$ for some $d \mid m$. Then $n_p = [G : N_G(P)] = p^a m / (p^a d) = m/d \mid m$. [example: Computing Sylow Subgroups in $S_4$] Let $G = S_4$, the symmetric group on $4$ elements, with $|S_4| = 24 = 2^3 \cdot 3$. **Sylow $3$-subgroups.** Here $p = 3$, $a = 1$, $m = 8$. A Sylow $3$-subgroup has order $3$, hence is generated by a $3$-cycle. The $3$-cycles in $S_4$ are: \begin{align*} (123), (132), (124), (142), (134), (143), (234), (243). \end{align*} Each generates a subgroup of order $3$, and each such subgroup contains two $3$-cycles plus the identity. So there are $8/2 = 4$ Sylow $3$-subgroups: \begin{align*} P_1 &= \langle (123) \rangle = \{e, (123), (132)\}, \\ P_2 &= \langle (124) \rangle = \{e, (124), (142)\}, \\ P_3 &= \langle (134) \rangle = \{e, (134), (143)\}, \\ P_4 &= \langle (234) \rangle = \{e, (234), (243)\}. \end{align*} Check: $n_3 = 4 \equiv 1 \pmod{3}$ and $4 \mid 8$. Both conditions are satisfied. **Sylow $2$-subgroups.** Here $p = 2$, $a = 3$, $m = 3$. A Sylow $2$-subgroup has order $8$. The constraints give $n_2 \equiv 1 \pmod{2}$ (so $n_2$ is odd) and $n_2 \mid 3$ (so $n_2 \in \{1, 3\}$). Since $S_4$ is not solvable with a normal Sylow $2$-subgroup of index $3$ — a normal subgroup of index $3$ in $S_4$ would give a homomorphism $S_4 \to S_3$ with kernel of order $8$, but $A_4$ has order $12$ so no order-$8$ normal subgroup exists — we get $n_2 = 3$. The three Sylow $2$-subgroups are each isomorphic to $D_8$ (dihedral group of order $8$, symmetries of a square). They correspond to the three ways of pairing the four elements $\{1,2,3,4\}$ into a "square" — that is, the three ways to impose a cyclic order on four points. Concretely: \begin{align*} P_1 &= \{e, (1234), (13)(24), (1432), (13), (24), (12)(34), (14)(23)\} \cong D_8, \\ P_2 &= \{e, (1324), (12)(34), (1423), (12), (34), (13)(24), (14)(23)\} \cong D_8, \\ P_3 &= \{e, (1243), (14)(23), (1342), (14), (23), (12)(34), (13)(24)\} \cong D_8. \end{align*} Each has order $8$, and one can verify these are pairwise distinct subgroups of $S_4$. The three double-transpositions $(12)(34)$, $(13)(24)$, $(14)(23)$ each appear in two of the three Sylow $2$-subgroups — a sign that they share significant overlap, consistent with the fact that all three are conjugate. This confirms $n_2 = 3$. Check: $n_2 = 3 \equiv 1 \pmod{2}$ (odd) and $3 \mid 3$. Both conditions hold. [/example] The example shows that the third theorem's constraints, while necessary, may not uniquely determine $n_p$. In $S_4$ both $n_2 = 1$ and $n_2 = 3$ are consistent with the arithmetic; the actual value requires deeper analysis (or explicit construction). ## Applications to Group Structure The real power of the Sylow theorems is not in finding Sylow subgroups but in using them to prove that certain groups cannot be simple, or to classify groups of small order. Let us develop this systematically. The fundamental strategy: if the Sylow constraints force $n_p = 1$ for some prime $p \mid |G|$, then $G$ has a proper normal subgroup (the unique Sylow $p$-subgroup), so $G$ is not simple. If every $n_p > 1$, then the element count may still yield a contradiction. [quotetheorem:3252] The cleanness of this result is striking: once we know the prime factorization $pq$ with $p < q$, pure arithmetic determines either the full structure (cyclic) or reduces the problem to semidirect products classified by $\operatorname{Aut}(\mathbb{Z}/q\mathbb{Z})$. Let us unpack the argument. [explanation: Why This Classification Works] For the normal Sylow $q$-subgroup: $n_q \equiv 1 \pmod{q}$ and $n_q \mid p$. Since $p < q$, the divisors of $p$ are $1$ and $p$. But $p < q$ means $p \not\equiv 1 \pmod{q}$ (since $1 \le p < q$ and the only integer in $\{1, \ldots, q-1\}$ that is $\equiv 1 \pmod{q}$ is $1$ itself). So $n_q = 1$, giving a normal Sylow $q$-subgroup $Q \trianglelefteq G$. For the Sylow $p$-subgroup $P$ of order $p$: $n_p \mid q$ and $n_p \equiv 1 \pmod{p}$, so $n_p \in \{1, q\}$. If $q \not\equiv 1 \pmod{p}$, i.e., $p \nmid (q-1)$, then $n_p = 1$, and both Sylow subgroups are normal. Having normal subgroups $P$ and $Q$ of coprime orders with $PQ = G$ and $P \cap Q = \{e\}$ forces $G \cong P \times Q \cong \mathbb{Z}/p\mathbb{Z} \times \mathbb{Z}/q\mathbb{Z} \cong \mathbb{Z}/pq\mathbb{Z}$ (by the Chinese Remainder Theorem). If $p \mid (q-1)$, then $n_p$ may equal $q$. In this case $G$ is a semidirect product $\mathbb{Z}/q\mathbb{Z} \rtimes_\varphi \mathbb{Z}/p\mathbb{Z}$ where $\varphi: \mathbb{Z}/p\mathbb{Z} \to \operatorname{Aut}(\mathbb{Z}/q\mathbb{Z}) \cong \mathbb{Z}/(q-1)\mathbb{Z}$ is a non-trivial homomorphism. Since $p \mid (q-1)$, such a non-trivial $\varphi$ exists (take a generator of the unique subgroup of order $p$ in $\operatorname{Aut}(\mathbb{Z}/q\mathbb{Z})$). All non-trivial choices of $\varphi$ give isomorphic groups (conjugate subgroups of $\operatorname{Aut}(\mathbb{Z}/q\mathbb{Z})$), so there is exactly one non-abelian group of order $pq$ up to isomorphism. [/explanation] The simplest illustration of the theorem is the case $|G| = 15$, where the arithmetic is clean enough to follow every step by hand. [example: Groups of Order $15$] $|G| = 15 = 3 \cdot 5$. Here $p = 3$, $q = 5$, $p < q$, and $p \nmid (q-1) = 4$ (since $3 \nmid 4$). The theorem gives $G \cong \mathbb{Z}/15\mathbb{Z}$. Let us verify directly. $n_5 \equiv 1 \pmod{5}$ and $n_5 \mid 3$, so $n_5 = 1$. Unique Sylow $5$-subgroup $Q \cong \mathbb{Z}/5\mathbb{Z}$, $Q \trianglelefteq G$. $n_3 \equiv 1 \pmod{3}$ and $n_3 \mid 5$, so $n_3 \in \{1, 5\}$. The semidirect product structure determines which holds. Since $Q \trianglelefteq G$, we have $G \cong Q \rtimes_\varphi P$ where $\varphi: P \to \operatorname{Aut}(Q) \cong \mathbb{Z}/4\mathbb{Z}$. For a non-trivial homomorphism to exist, $|P| = 3$ must divide $|\operatorname{Aut}(Q)| = 4$. But $3 \nmid 4$, so $\varphi$ must be trivial, giving $G \cong Q \times P \cong \mathbb{Z}/5\mathbb{Z} \times \mathbb{Z}/3\mathbb{Z} \cong \mathbb{Z}/15\mathbb{Z}$, and $n_3 = 1$. [/example] The order-$15$ case required almost no work: the arithmetic left only one option. The next example, order $200$, is similarly decisive — but the argument collapses to a single computation, which is itself worth observing as a testament to how tight the Sylow constraints can be. [example: No Simple Group of Order $200$] $|G| = 200 = 2^3 \cdot 5^2$. We show $G$ cannot be simple. **Sylow $5$-subgroups.** $n_5 \equiv 1 \pmod{5}$ and $n_5 \mid 8$. The divisors of $8$ are $\{1, 2, 4, 8\}$; among these, only $1 \equiv 1 \pmod{5}$ (since $2 \equiv 2$, $4 \equiv 4$, and $8 \equiv 3$). So $n_5 = 1$. The unique Sylow $5$-subgroup $P \le G$ has order $25$. Since $P$ is unique, it is fixed by all conjugations: for any $g \in G$, the conjugate $gPg^{-1}$ is another Sylow $5$-subgroup, hence equals $P$. Therefore $P \trianglelefteq G$. **What this means.** $P$ is a proper nontrivial normal subgroup of $G$ ($1 < 25 < 200$). A simple group has no such subgroup, so $G$ is not simple. Moreover, $G$ has a quotient $G/P$ of order $8$, and a normal subgroup $P$ isomorphic to either $\mathbb{Z}/25\mathbb{Z}$ or $\mathbb{Z}/5\mathbb{Z} \times \mathbb{Z}/5\mathbb{Z}$ (the two abelian groups of order $25$). Sylow theory has reduced the structure of a group of order $200$ to understanding groups of order $8$ and groups of order $25$ and how they interact — a much more tractable problem. [/example] ## The Normaliser and the Number $n_p$ In every calculation so far, we obtained $n_p$ from the formula $n_p = [G : N_G(P)]$ — we divided the group order by the size of the normaliser of a Sylow subgroup. The normaliser is therefore the invisible hand controlling $n_p$. But what is the normaliser, and why should we care about it beyond this formula? The normaliser of a subgroup $H$ in $G$ answers a natural question: how much of $G$ sees $H$ as a normal subgroup? Every element of $H$ preserves $H$ under conjugation (since $hHh^{-1} = H$ for $h \in H$), and elements outside $H$ may or may not preserve $H$ as a set. The elements that do form a subgroup — the largest subgroup of $G$ in which $H$ is normal. That is the normaliser. The formula $n_p = [G : N_G(P)]$ tells us that controlling $n_p$ is equivalent to controlling how large $N_G(P)$ is. When $N_G(P) = G$, we have $n_p = 1$ and $P \trianglelefteq G$. When $N_G(P) = P$ (the smallest possible normaliser), we have $n_p = [G : P] = m$, the largest possible value consistent with $n_p \mid m$. This is the key to understanding when the congruence condition is tight. [definition: Normaliser of a Subgroup] Let $G$ be a group and $H \le G$. The **normaliser** of $H$ in $G$ is \begin{align*} N_G(H) &= \{g \in G : gHg^{-1} = H\}. \end{align*} It is the largest subgroup of $G$ in which $H$ is normal: $H \trianglelefteq N_G(H) \le G$. [/definition] The normaliser is sandwiched between $H$ and $G$: always $H \le N_G(H) \le G$. The two extremes are illuminating: $N_G(H) = H$ means $H$ is self-normalising (no element outside $H$ conjugates $H$ to itself), giving $n_p = m$ the maximum; $N_G(H) = G$ means $H$ is normal in all of $G$, giving $n_p = 1$. [remark: Normaliser vs Centraliser] The normaliser $N_G(H)$ requires $gHg^{-1} = H$ (conjugation preserves $H$ as a set). The centraliser $C_G(H) = \{g \in G : gh = hg \text{ for all } h \in H\}$ requires conjugation to fix every element of $H$. Always $C_G(H) \trianglelefteq N_G(H)$, with equality iff $H \le Z(G)$. [/remark] The interplay between $N_G(P)$ and $P$ itself is the site of many structural results. For instance, the Frattini argument (which often appears as an exercise immediately after the Sylow theorems) uses the fact that conjugates of $P$ fill out $G$ to build a factorisation $G = N_G(P) \cdot K$ for any normal subgroup $K$ that contains a Sylow subgroup of itself. [quotetheorem:3253] The proof is a beautiful one-line application of the second Sylow theorem. The key idea is that conjugating a Sylow subgroup of $K$ by an element of $G$ gives another Sylow subgroup of $K$, and by conjugacy they must be the same up to a correction from $K$. [explanation: Why the Frattini Argument Works] Take any $g \in G$. Since $K \trianglelefteq G$, we have $gKg^{-1} = K$, so $gPg^{-1} \le K$. But $gPg^{-1}$ is a Sylow $p$-subgroup of $K$ (it has the same order $p^a$ as $P$, and $p^a$ is the full $p$-power in $|K|$). By the second Sylow theorem applied inside $K$, all Sylow $p$-subgroups of $K$ are conjugate: there exists $k \in K$ with $gPg^{-1} = kPk^{-1}$. Then $k^{-1}g \in N_G(P)$, so $g = k \cdot (k^{-1}g) \in K \cdot N_G(P)$. [/explanation] ## Sylow Theory and Simple Groups The most celebrated use of the Sylow theorems is in proving non-existence of simple groups of certain orders. The argument is always some variant of: count the Sylow subgroups, show they fill up too many elements, or show $n_p = 1$ for some $p$. There is a complementary direction: Sylow theory can identify the only possible simple groups of a given order. If for every prime $p \mid |G|$ we can find multiple Sylow $p$-subgroups, then the element counts are consistent, and we must work harder (or simply verify that a known simple group has the right order). [quotetheorem:3254] <!-- illustration-needed: lattice of subgroups for a group of order p^2 q showing the Sylow p-subgroups, their conjugates when n_p > 1, and the normal Sylow q-subgroup when it exists --> The argument is a crisp one-liner once you observe the size constraint forces $n_p = 1$. [explanation: The Counting Argument for $p^2 q$] $n_p \equiv 1 \pmod{p}$ and $n_p \mid q$. Since $q$ is prime, $n_p \in \{1, q\}$. If $n_p = q$, then $q \equiv 1 \pmod{p}$, i.e., $p \mid (q - 1)$. But $q - 1 < q < p^2$, and for $p \mid (q-1)$ we would need $q \ge p + 1$, hence $q > p$. This contradicts $p > q$. So $n_p = 1$. [/explanation] The classification of groups of order $12$ shows Sylow theory at its most systematic: we branch on $n_3$, exhaust the cases, and identify each branch with a known group. [example: All Groups of Order $12$ Are Classified by Sylow Theory] $|G| = 12 = 2^2 \cdot 3$. **Sylow $3$-subgroups:** $n_3 \equiv 1 \pmod 3$ and $n_3 \mid 4$, so $n_3 \in \{1, 4\}$. **Sylow $2$-subgroups:** $n_2 \equiv 1 \pmod 2$ (so $n_2$ is odd) and $n_2 \mid 3$, so $n_2 \in \{1, 3\}$. **Case 1: $n_3 = 1$.** The unique Sylow $3$-subgroup $Q \cong \mathbb{Z}/3\mathbb{Z}$ is normal. The Sylow $2$-subgroup $P$ has order $4$, so $P \cong \mathbb{Z}/4\mathbb{Z}$ or $P \cong \mathbb{Z}/2\mathbb{Z} \times \mathbb{Z}/2\mathbb{Z}$. The group $G$ is a semidirect product $Q \rtimes P$ (or $P \times Q$ if the action is trivial). Working through the homomorphisms $P \to \operatorname{Aut}(Q) \cong \mathbb{Z}/2\mathbb{Z}$, this yields three groups: $\mathbb{Z}/12\mathbb{Z}$, $\mathbb{Z}/2\mathbb{Z} \times \mathbb{Z}/6\mathbb{Z}$, and $D_{12}$ (dihedral of order $12$). **Case 2: $n_3 = 4$.** There are $4$ Sylow $3$-subgroups, contributing $4 \cdot 2 = 8$ elements of order $3$. Combined with the identity, that is $9$ elements accounted for. The remaining $3$ elements must form the unique Sylow $2$-subgroup (since they cannot have order $3$ or $1$), so $n_2 = 1$: the Sylow $2$-subgroup is normal. But now the group must be $A_4$, the alternating group of order $12$, which indeed has $n_3 = 4$ and $n_2 = 1$. The five groups of order $12$ up to isomorphism are therefore: $\mathbb{Z}/12\mathbb{Z}$, $\mathbb{Z}/2\mathbb{Z} \times \mathbb{Z}/6\mathbb{Z}$, $D_{12}$, $A_4$, and the dicyclic group $\text{Dic}_3$ (equivalently, the semidirect product $\mathbb{Z}/3\mathbb{Z} \rtimes \mathbb{Z}/4\mathbb{Z}$ via the unique non-trivial action). [/example] The example shows both the power and the limits of Sylow theory: it narrows the possibilities drastically, but it does not by itself complete the classification — you still need to enumerate semidirect products and check which are distinct. ## $p$-Groups and the Centre Sylow theory would be incomplete without an understanding of $p$-groups themselves. The reason Sylow $p$-subgroups behave so nicely is that $p$-groups have rich internal structure. The most fundamental fact is that every nontrivial $p$-group has a nontrivial centre. [quotetheorem:799] This follows from the class equation. If $G$ is a group, the conjugacy classes partition $G$: \begin{align*} |G| &= |Z(G)| + \sum_{\substack{x \notin Z(G) \\ \text{one per class}}} [G : C_G(x)]. \end{align*} Every term $[G : C_G(x)]$ for $x \notin Z(G)$ is greater than $1$ and divides $|G|$. If $|G| = p^k$, every such term is divisible by $p$. Since $p \mid |G|$, we conclude $p \mid |Z(G)|$, so $Z(G)$ is nontrivial. This result has a striking consequence for groups of order $p^2$: when a $p$-group is so small, its centre must take up the whole group. [quotetheorem:801] The argument is a short but elegant loop: a $p$-group has nontrivial centre, which forces the quotient by the centre to be cyclic if the centre is proper, but a cyclic quotient by the centre forces the group to be abelian — contradicting the assumption that the centre was proper. [explanation: Why $|G| = p^2$ Forces Abelian] By the theorem on $p$-group centres, $|Z(G)|$ is a nontrivial divisor of $p^2$, so $|Z(G)| \in \{p, p^2\}$. If $|Z(G)| = p^2$, then $Z(G) = G$ and $G$ is abelian. If $|Z(G)| = p$, then $G/Z(G)$ has order $p$, hence $G/Z(G)$ is cyclic. But if $G/Z(G)$ is cyclic, then $G$ is abelian: take any $g, h \in G$. Since $G/Z(G) = \langle xZ(G) \rangle$ for some $x$, we can write $g = x^a z_1$ and $h = x^b z_2$ with $z_1, z_2 \in Z(G)$. Then $gh = x^a z_1 x^b z_2 = x^{a+b} z_1 z_2 = x^b z_2 x^a z_1 = hg$ (since $z_1, z_2$ commute with everything). So $G$ is abelian, contradicting $|Z(G)| = p < p^2 = |G|$. Hence $|Z(G)| = p^2$ and $G$ is abelian. [/explanation] This result is useful in Sylow theory because Sylow $p$-subgroups of order $p^2$ are always abelian. It also marks the boundary of the general theory: groups of order $p^3$ need not be abelian (the quaternion group $Q_8$ has order $8 = 2^3$ and is non-abelian), so $p^2$ is exactly the threshold. It simplifies the analysis of groups whose Sylow subgroups are small. [example: Structure of Sylow $p$-Subgroups of $GL_2(\mathbb{F}_p)$] Let $p$ be an odd prime and $G = GL_2(\mathbb{F}_p)$, the group of invertible $2 \times 2$ matrices over the field with $p$ elements. The order is \begin{align*} |GL_2(\mathbb{F}_p)| &= (p^2 - 1)(p^2 - p) = p(p-1)^2(p+1). \end{align*} The Sylow $p$-subgroup has order $p$. The upper triangular unipotent matrices \begin{align*} U &= \left\{ \begin{pmatrix} 1 & a \\ 0 & 1 \end{pmatrix} : a \in \mathbb{F}_p \right\} \end{align*} form a subgroup of order $p$, and since $p \nmid (p-1)^2(p+1)$ (as $p$ is a prime factor of $|G|$ occurring exactly once), $U$ is a Sylow $p$-subgroup. The number of Sylow $p$-subgroups is $n_p = [G : N_G(U)]$. The normaliser $N_G(U)$ is the Borel subgroup of upper triangular matrices, which has order $p(p-1)^2$ (choose the diagonal entries from $\mathbb{F}_p^\times$ and the upper-right entry freely). So \begin{align*} n_p &= \frac{p(p-1)^2(p+1)}{p(p-1)^2} = p + 1. \end{align*} Check: $p + 1 \equiv 1 \pmod{p}$ and $(p+1) \mid (p-1)^2(p+1)$. Both hold. [/example] ## Sylow Theory in Context: Towards the Schur–Zassenhaus Theorem The Sylow theorems say: Sylow subgroups exist and are conjugate. A natural follow-up question is about complementation: if $N \trianglelefteq G$ and $\gcd(|N|, [G:N]) = 1$, does $N$ have a complement — a subgroup $H \le G$ with $G = NH$ and $N \cap H = \{e\}$? The answer is yes, and this is the content of the Schur–Zassenhaus theorem. [quotetheorem:3255] The connection to what we have already done is direct: the Sylow theorems are the case where $N$ is a Sylow $p$-subgroup and $[G:N]$ is the complementary $p'$-part of $|G|$. Schur–Zassenhaus lifts this to arbitrary normal subgroups of coprime order. [explanation: Connection to Sylow Theory] The Schur–Zassenhaus theorem is a vast generalisation of the Sylow theorems. When $N$ is a Sylow $p$-subgroup $P$ of $G$, then $|N| = p^a$ and $[G:N] = m$ with $\gcd(p^a, m) = 1$, so the coprimality condition is satisfied. The theorem then guarantees a complement to $P$ in $G$ — a subgroup of order $m$ — and that all such complements are conjugate. The second Sylow theorem (conjugacy of Sylow subgroups) is a special case of this. The proof of Schur–Zassenhaus uses cohomological methods or, in the split case, group cohomology $H^1(G/N, N)$. The vanishing of this cohomology group — which holds when $|N|$ and $[G:N]$ are coprime, by a transfer argument — gives both existence and uniqueness of complements up to conjugacy. [/explanation] The Sylow theorems thus stand as the first instance of a much deeper principle: when orders are coprime, extensions split. This is one of the central themes of finite group theory and leads eventually to the Feit–Thompson theorem and the classification of finite simple groups. [remark: Sylow Theory Over Infinite Groups] The three Sylow theorems are genuinely finite results. For infinite groups, even the existence of maximal $p$-subgroups can fail without additional hypotheses (such as local finiteness or profinite structure). In the profinite setting — inverse limits of finite groups — one studies *pro-$p$ Sylow subgroups*, and the conjugacy theorem has an analogue: any two maximal pro-$p$ subgroups are conjugate. But this requires entirely different methods. [/remark] ## References L. C. Grove, *Algebra* (1983). Chapter 3 contains a careful development of the Sylow theorems with full proofs via the orbit-counting method. I. N. Herstein, *Topics in Algebra* (2nd ed., 1975). A classic source for the Sylow theorems and their applications to groups of small order. D. S. Dummit and R. M. Foote, *Abstract Algebra* (3rd ed., 2004). Chapter 4 gives a thorough treatment of group actions, the class equation, and the Sylow theorems; Chapter 6 applies Sylow theory to classify groups of specific orders. J. Rotman, *An Introduction to the Theory of Groups* (4th ed., 1995). Includes the Schur–Zassenhaus theorem and its connection to group cohomology. M. Aschbacher, *Finite Group Theory* (2nd ed., 2000). The authoritative modern reference for the deeper uses of Sylow theory in the context of the classification of finite simple groups.