Sylow's Theorems

  • Sylow's theorems guarantee that if prime power pkp^k divides |G|, then G contains a subgroup of order pkp^k (extends Lagrange's theorem)
  • All maximal p-subgroups (Sylow p-subgroups) in a group are conjugate to each other, constraining structural diversity
  • Number of Sylow p-subgroups npn_p satisfies np≡1(modp)n_p \equiv 1 \pmod{p} and divides |G|/pnp^n, tightly limiting possibilities
  • Critical tool for classifying small finite groups and proving non-simplicity of groups with specific orders
  • Bridge between numerical properties of |G| and actual geometric/combinatorial group structure
  • Published 1872, foundational results moving group theory from studying examples to universal structural principles

Sylow's Theorems

Overview

Sylow's theorems are fundamental results in finite group theory that provide detailed information about the existence and structure of subgroups whose order is a prime power. Named after the Norwegian mathematician Ludwig Sylow, these theorems answer a natural question arising from Lagrange's theorem: while Lagrange tells us that the order of any subgroup divides the order of the group, the converse fails in general. Sylow's theorems give precise conditions under which subgroups of particular orders must exist, and they characterize how many such subgroups there can be. The power of Sylow's theorems lies in their ability to impose strong structural constraints on finite groups based solely on arithmetic properties of their order. They provide the primary tool for classifying groups of small order, for proving that groups of certain orders cannot be simple, and for understanding the internal architecture of arbitrary finite groups. The theorems establish a bridge between the purely numerical data encoded in ∣G∣|G| and the geometric and combinatorial structure of GG itself. Historically, these results marked a watershed moment in the development of group theory. Published in 1872, they represented one of the first deep theorems about the general structure of finite groups, moving beyond particular examples to establish universal principles. The Sylow theorems remain central to modern algebra, appearing in applications ranging from Galois theory to representation theory and even in combinatorial designs and graph theory.

💡 The Core Insight

If a prime power pkp^k divides the order of a finite group GG, then GG must contain a subgroup of order pkp^k. Moreover, all maximal such subgroups (Sylow pp-subgroups) are conjugate, and their number is tightly constrained by congruence conditions modulo pp.

Definition

Sylow's Theorems: Let GG be a finite group of order ∣G∣=pnm|G| = p^n m where pp is prime, n≥1n \geq 1, and gcd⁡(p,m)=1\gcd(p, m) = 1. A Sylow pp-subgroup of GG is a subgroup of order pnp^n (i.e., a subgroup whose order is the maximal power of pp dividing ∣G∣|G|). We denote by Sylp(G)\text{Syl}_p(G) the set of all Sylow pp-subgroups of GG, and by npn_p the cardinality ∣Sylp(G)∣|\text{Syl}_p(G)|.

Core Theory

TheoremMain Result

Sylow's Theorems: Let GG be a finite group of order ∣G∣=pnm|G| = p^n m where pp is prime and gcd⁡(p,m)=1\gcd(p,m) = 1. Then:

[1] (Existence) GG contains at least one Sylow pp-subgroup. Moreover, if pkp^k divides ∣G∣|G| for some k≤nk \leq n, then GG contains a subgroup of order pkp^k.

[2] (Conjugacy) All Sylow pp-subgroups of GG are conjugate to one another. That is, if P,Q∈Sylp(G)P, Q \in \text{Syl}_p(G), then there exists g∈Gg \in G such that Q=gPg−1Q = gPg^{-1}.

[3] (Counting) The number np=∣Sylp(G)∣n_p = |\text{Syl}_p(G)| satisfies:

  • np≡1(modp)n_p \equiv 1 \pmod{p}
  • np∣mn_p \mid m

In particular, np∈{d∣m:d≡1(modp)}n_p \in \{d \mid m : d \equiv 1 \pmod{p}\}.

Proof

We prove each part in turn.

Proof of [1] (Existence via counting): Let S\mathcal{S} denote the collection of all subsets of GG having exactly pnp^n elements. Then ∣S∣=(∣G∣pn)=(pnmpn)|\mathcal{S}| = \binom{|G|}{p^n} = \binom{p^n m}{p^n}. We analyze this binomial coefficient modulo pp. Writing (pnmpn)=(pnm)!(pn)!(pnm−pn)!=(pnm)!(pn)!(pn(m−1))!\binom{p^n m}{p^n} = \frac{(p^n m)!}{(p^n)!(p^n m - p^n)!} = \frac{(p^n m)!}{(p^n)!(p^n(m-1))!}, we use the fact that the highest power of pp dividing k!k! is ∑i=1∞⌊k/pi⌋\sum_{i=1}^{\infty} \lfloor k/p^i \rfloor. For the numerator: vp((pnm)!)=∑i=1∞⌊pnm/pi⌋=pn−1m+pn−2m+⋯+m+⌊m/p⌋+⋯v_p((p^n m)!) = \sum_{i=1}^{\infty} \lfloor p^n m / p^i \rfloor = p^{n-1}m + p^{n-2}m + \cdots + m + \lfloor m/p \rfloor + \cdots For the denominator: vp((pn)!)=pn−1+pn−2+⋯+1v_p((p^n)!) = p^{n-1} + p^{n-2} + \cdots + 1 and vp((pn(m−1))!)=pn−1(m−1)+⋯v_p((p^n(m-1))!) = p^{n-1}(m-1) + \cdots Computing vp((pnmpn))=vp((pnm)!)−vp((pn)!)−vp((pn(m−1))!)v_p(\binom{p^n m}{p^n}) = v_p((p^n m)!) - v_p((p^n)!) - v_p((p^n(m-1))!), the pn−1p^{n-1} terms give pn−1m−pn−1−pn−1(m−1)=0p^{n-1}m - p^{n-1} - p^{n-1}(m-1) = 0, and similarly for all higher powers of pp. The remaining terms involve only mm and m−1m-1, which are coprime to pp, so vp((pnmpn))=0v_p(\binom{p^n m}{p^n}) = 0. Thus p∤(pnmpn)p \nmid \binom{p^n m}{p^n}. Now let GG act on S\mathcal{S} by left multiplication: g⋅S={gs:s∈S}g \cdot S = \{gs : s \in S\}. Since p∤∣S∣p \nmid |\mathcal{S}|, there exists at least one orbit whose size is not divisible by pp. Let S0S_0 be a representative of such an orbit, and let H={g∈G:gS0=S0}H = \{g \in G : gS_0 = S_0\} be its stabilizer. By the orbit-stabilizer theorem, ∣G∣=∣Orb(S0)∣⋅∣H∣|G| = |\text{Orb}(S_0)| \cdot |H|, so pnm=∣Orb(S0)∣⋅∣H∣p^n m = |\text{Orb}(S_0)| \cdot |H|. Since p∤∣Orb(S0)∣p \nmid |\text{Orb}(S_0)|, we have pn∣∣H∣p^n \mid |H|. We claim ∣H∣≤pn|H| \leq p^n. Note that H⊆S0H \subseteq S_0 (since if h∈Hh \in H, then hS0=S0hS_0 = S_0, and e∈S0e \in S_0, so h=he∈hS0=S0h = he \in hS_0 = S_0). Thus ∣H∣≤∣S0∣=pn|H| \leq |S_0| = p^n. Therefore ∣H∣=pn|H| = p^n, and HH is a Sylow pp-subgroup. For the "moreover" clause, let PP be a Sylow pp-subgroup and pk∣∣G∣p^k \mid |G| with k≤nk \leq n. By Cauchy's theorem applied repeatedly, PP contains an element of order pp, hence a subgroup of order pp. Inductively, we can show PP contains a subgroup of order pkp^k for each k≤nk \leq n.

Proof of [2] (Conjugacy): Let P∈Sylp(G)P \in \text{Syl}_p(G) and let QQ be any pp-subgroup of GG (not necessarily maximal). We show that Q⊆gPg−1Q \subseteq gPg^{-1} for some g∈Gg \in G. Let QQ act on the set G/P={gP:g∈G}G/P = \{gP : g \in G\} of left cosets by left multiplication. Since ∣G/P∣=m|G/P| = m and gcd⁡(p,m)=1\gcd(p,m) = 1, and QQ is a pp-group, there must exist a fixed point: some coset gPgP such that qgP=gPqgP = gP for all q∈Qq \in Q. This means g−1qg∈Pg^{-1}qg \in P for all q∈Qq \in Q, so Q⊆gPg−1Q \subseteq gPg^{-1}. Now if QQ is also a Sylow pp-subgroup, then ∣Q∣=pn=∣P∣=∣gPg−1∣|Q| = p^n = |P| = |gPg^{-1}|, so Q=gPg−1Q = gPg^{-1}.

Proof of [3] (Counting): Let P∈Sylp(G)P \in \text{Syl}_p(G). By (2), GG acts transitively on Sylp(G)\text{Syl}_p(G) by conjugation. By the orbit-stabilizer theorem, np=[G:NG(P)]n_p = [G : N_G(P)] where NG(P)={g∈G:gPg−1=P}N_G(P) = \{g \in G : gPg^{-1} = P\} is the normalizer of PP. Since P⊆NG(P)P \subseteq N_G(P), we have ∣NG(P)∣=pn⋅k|N_G(P)| = p^n \cdot k for some integer kk, so np=pnmpnk=mkn_p = \frac{p^n m}{p^n k} = \frac{m}{k}. Thus np∣mn_p \mid m. For the congruence, let PP act on Sylp(G)\text{Syl}_p(G) by conjugation. Since PP is a pp-group, each orbit has size a power of pp. The orbit of PP itself is {P}\{P\} (size 1) since pPp−1=PpPp^{-1} = P for all p∈Pp \in P. If Q∈Sylp(G)Q \in \text{Syl}_p(G) with Q≠PQ \neq P, we claim the orbit of QQ under PP-conjugation has size divisible by pp. The stabilizer is {p∈P:pQp−1=Q}=P∩NG(Q)\{p \in P : pQp^{-1} = Q\} = P \cap N_G(Q). If this equals PP, then P⊆NG(Q)P \subseteq N_G(Q), so P,Q⊆NG(Q)P, Q \subseteq N_G(Q). But PP and QQ are both Sylow pp-subgroups of NG(Q)N_G(Q), so by conjugacy within NG(Q)N_G(Q), there exists n∈NG(Q)n \in N_G(Q) with Q=nPn−1Q = nPn^{-1}. But n∈NG(Q)n \in N_G(Q) means nQn−1=QnQn^{-1} = Q, so nPn−1=QnPn^{-1} = Q gives P=QP = Q, contradiction. Thus P∩NG(Q)⊊PP \cap N_G(Q) \subsetneq P, so ∣P∩NG(Q)∣≤pn−1|P \cap N_G(Q)| \leq p^{n-1}, and the orbit size [P:P∩NG(Q)]≥p[P : P \cap N_G(Q)] \geq p. Therefore np=1+(sum of orbit sizes≥p)≡1(modp)n_p = 1 + (\text{sum of orbit sizes} \geq p) \equiv 1 \pmod{p}. □

∎
⚠️ Common mistake

Common Error: Do not confuse "subgroups of order a power of pp" with "Sylow pp-subgroups." A Sylow pp-subgroup has order equal to the maximal power of pp dividing ∣G∣|G|. For example, in a group of order 12 = 22⋅32^2 \cdot 3, a subgroup of order 2 is a pp-subgroup but not a Sylow 2-subgroup; only subgroups of order 4 are Sylow 2-subgroups. The conjugacy theorem applies only to Sylow pp-subgroups, not to all pp-subgroups.

mathematician

Ludwig Sylow (1872) — Proved the three fundamental theorems on the existence, conjugacy, and enumeration of maximal prime-power subgroups in his landmark paper "Théorèmes sur les groupes de substitutions."

visualization

Sylow 2-subgroups of D₆ (order 12) G = D₆, |G| = 12 = 2² · 3 P₁ ⟨r², s⟩ P₂ ⟨r², sr⟩ P₃ ⟨r², sr²⟩ n₂ = 3 ≡ 1 (mod 2), and 3 | 3 ✓

Examples and Applications

Example1

Classifying groups of order 15 Let ∣G∣=15=3⋅5|G| = 15 = 3 \cdot 5. We prove G≅Z15G \cong \mathbb{Z}_{15}. For Sylow 3-subgroups: n3∣5n_3 \mid 5 and n3≡1(mod3)n_3 \equiv 1 \pmod{3}. The divisors of 5 are {1, 5}. Since 5≡2(mod3)5 \equiv 2 \pmod{3}, we must have n3=1n_3 = 1. For Sylow 5-subgroups: n5∣3n_5 \mid 3 and n5≡1(mod5)n_5 \equiv 1 \pmod{5}. The divisors of 3 are {1, 3}. Since 3≡3(mod5)3 \equiv 3 \pmod{5}, we must have n5=1n_5 = 1. Let PP be the unique Sylow 3-subgroup and QQ be the unique Sylow 5-subgroup. Since they are unique, they are normal in GG (as np=1n_p = 1 implies gPg−1=PgPg^{-1} = P for all g∈Gg \in G). We have P∩Q={e}P \cap Q = \{e\} (since ∣P∩Q∣|P \cap Q| divides both 3 and 5, hence equals 1), and ∣PQ∣=∣P∣∣Q∣∣P∩Q∣=3⋅51=15=∣G∣|PQ| = \frac{|P||Q|}{|P \cap Q|} = \frac{3 \cdot 5}{1} = 15 = |G|, so G=PQG = PQ. By Cauchy's theorem, ∣P∣=3|P| = 3 (prime) implies P≅Z3P \cong \mathbb{Z}_3, and ∣Q∣=5|Q| = 5 implies Q≅Z5Q \cong \mathbb{Z}_5. Since P,QP, Q are both normal and P∩Q={e}P \cap Q = \{e\}, we have G≅P×Q≅Z3×Z5≅Z15G \cong P \times Q \cong \mathbb{Z}_3 \times \mathbb{Z}_5 \cong \mathbb{Z}_{15} by the Chinese Remainder Theorem.

Example2

Analyzing A5A_5 (the alternating group) Let G=A5G = A_5, the group of even permutations on 5 elements. Then ∣A5∣=60=22⋅3⋅5|A_5| = 60 = 2^2 \cdot 3 \cdot 5.

Sylow 5-subgroups: n5∣12n_5 \mid 12 and n5≡1(mod5)n_5 \equiv 1 \pmod{5}. The divisors of 12 congruent to 1 mod 5 are: {1, 6}. We claim n5=6n_5 = 6. A Sylow 5-subgroup has order 5, hence is cyclic, generated by a 5-cycle. The number of 5-cycles in A5A_5 is 5!5=24\frac{5!}{5} = 24 (since there are 5!/5=245!/5 = 24 distinct 5-cycles, all even permutations). Each cyclic group of order 5 contains ϕ(5)=4\phi(5) = 4 generators, so the number of Sylow 5-subgroups is 24/4=624/4 = 6. Thus n5=6n_5 = 6.

Sylow 3-subgroups: n3∣20n_3 \mid 20 and n3≡1(mod3)n_3 \equiv 1 \pmod{3}. The candidates are {1, 4, 10}. A Sylow 3-subgroup has order 3, generated by a 3-cycle. There are (53)⋅2=20\binom{5}{3} \cdot 2 = 20 distinct 3-cycles in A5A_5, each generating a different subgroup of order 3 (since 3-cycles have order 3). Each subgroup contains 2 non-identity elements, so n3=20/2=10n_3 = 20/2 = 10.

Sylow 2-subgroups: n2∣15n_2 \mid 15 and n2≡1(mod2)n_2 \equiv 1 \pmod{2}. So n2∈{1,3,5,15}n_2 \in \{1, 3, 5, 15\}. Since A5A_5 is simple (has no proper normal subgroups), n2≠1n_2 \neq 1. One can verify n2=5n_2 = 5 by explicit computation: Sylow 2-subgroups are isomorphic to the Klein four-group V4V_4, and there are exactly 5 of them. This analysis is crucial for proving A5A_5 is simple: if N◃A5N \triangleleft A_5 is normal and nontrivial, then by Sylow's second theorem (conjugacy), NN must contain entire conjugacy classes of Sylow subgroups, leading to contradictions with ∣N∣|N| dividing 60.

Counterexample

Why we need pnp^n to be the maximal power Consider G=S3G = S_3 with ∣G∣=6=2⋅3|G| = 6 = 2 \cdot 3. There are three subgroups of order 2 (each generated by a transposition): H1=⟨(12)⟩H_1 = \langle (12) \rangle, H2=⟨(13)⟩H_2 = \langle (13) \rangle, H3=⟨(23)⟩H_3 = \langle (23) \rangle. These are not conjugate to each other in the sense that all three exist simultaneously—but this doesn't violate Sylow's second theorem because they are all conjugate! Indeed: (13)(12)(13)=(23)(13)(12)(13) = (23) and (12)(13)(12)=(23)(12)(13)(12) = (23). However, if we mistakenly tried to apply "conjugacy" to the non-maximal subgroup {e}\{e\} (order p0=1p^0 = 1), we'd trivially have only one such subgroup. The theorem's power is specifically for Sylow pp-subgroups (maximal prime-power order). A better counterexample: In Z4×Z2\mathbb{Z}_4 \times \mathbb{Z}_2 (order 8 = 232^3), consider subgroups of order 2. We have ⟨(2,0)⟩\langle (2,0) \rangle, ⟨(0,1)⟩\langle (0,1) \rangle, and ⟨(2,1)⟩\langle (2,1) \rangle—these are not all conjugate (the group is abelian!). But the unique Sylow 2-subgroup is the entire group, so there's no contradiction. Conjugacy applies only to Sylow subgroups, not arbitrary pp-power subgroups.

🧠 Intuition

Mental Model: Think of a finite group as a Crystal Structure. Sylow's theorems tell us that for each prime pp dividing the order, the "p-colored atoms" must cluster together in specific ways—forming maximal clumps (Sylow pp-subgroups) of size pnp^n. The conjugacy theorem says that from the group's "symmetry perspective," all these maximal pp-clumps look identical—they're related by the group's internal symmetries (conjugation). You can't distinguish between them using the group structure alone. The counting theorem provides a "crystallographic restriction": the number of such clumps is constrained by both divisibility (np∣mn_p \mid m, the "pp-free part") and by a periodicity condition (np≡1(modp)n_p \equiv 1 \pmod{p}). This is remarkably restrictive—often forcing np=1n_p = 1, which means that pp-clump is actually frozen in place (normal), completely constraining the group's structure.

Historical Development

Ludwig Sylow published his theorems in 1872 in the paper "Théorèmes sur les groupes de substitutions" in Mathematische Annalen. At the time, group theory was still in its infancy, with groups primarily understood as permutation groups (substitution groups). Sylow's work built on earlier results by Augustin-Louis Cauchy, who had proven in 1845 that if a prime pp divides ∣G∣|G|, then GG contains an element of order pp. The context of Sylow's work was the broader program of understanding when polynomial equations are solvable by radicals, initiated by Évariste Galois. Sylow's theorems provided powerful tools for analyzing the structure of Galois groups, which are finite groups in many classical cases. Initially, Sylow's proofs used detailed combinatorial arguments about permutation groups. The modern proof via group actions on coset spaces was developed later, particularly through the work of Ferdinand Georg Frobenius in the 1880s-1890s. Frobenius systematized the theory of group actions and the orbit-stabilizer theorem, providing the conceptual framework we use today. The notation and terminology evolved over time. The term "Sylow pp-subgroup" became standard in the early 20th century. William Burnside's 1897 treatise Theory of Groups of Finite Order was instrumental in disseminating these results and establishing them as foundational in finite group theory.

Connections to Other Areas

connection

Combinatorial Designs and Block Structures: The Sylow theorems have surprising applications in combinatorial design theory. For instance, if a Steiner system S(2,k,v)S(2, k, v) admits a group GG acting on it, Sylow's theorems constrain which groups can act. The condition np≡1(modp)n_p \equiv 1 \pmod{p} combined with np∣mn_p \mid m has been used to prove non-existence results for certain symmetric designs. More dramatically, the Feit-Thompson theorem (every group of odd order is solvable) relies fundamentally on character-theoretic refinements of Sylow theory, and this result has implications for which combinatorial configurations can exist with prescribed symmetry groups.

What You Need First

What This Opens Up

See Also

Normalizer · p-Groups · Hall Subgroups · Transfer Homomorphism · Frattini Argument