The Sylow Theorems

How many subgroups of a given order must a finite group possess? How tightly are they linked? This question drove much of nineteenth-century group theory. The Sylow theorems answer it as completely as one could hope for prime-power orders. If GG is a finite group and pkp^k is the largest power of a prime pp dividing ∣G∣|G|, the Sylow theorems guarantee a subgroup of that exact order. They show all such subgroups are conjugate. They also pin down the number of them through sharp divisibility and congruence constraints. The result matters because it converts a purely numerical fact — the prime factorization of ∣G∣|G| — into deep structural information about GG itself. Where Cauchy's Theorem only promises an element of prime order pp, the Sylow theorems promise an entire subgroup. That subgroup realizes the full pp-part of ∣G∣|G|. They also explain exactly how many such subgroups exist up to the action of GG by conjugation. This is often the first tool reached for when trying to prove a group of a specific order is not simple. It is also useful when classifying groups of small order. The big picture is that Sylow subgroups are the local, pp-adic building blocks of a finite group. This is analogous to how prime factorization organizes the integers. Group actions — GG acting on sets of subgroups or cosets by conjugation and translation — are the engine behind every proof here. p-Groups and Their Properties supplies the combinatorial facts about orbits under pp-groups. These facts make the counting arguments work. This article proves existence, conjugacy, and the counting theorem in full, with careful hypothesis tracking. It closes with worked examples pinning down Sylow subgroups of S4S_4 and a group of order 6060. It also gives a counterexample showing why finiteness of GG cannot be dropped. Finally, it looks toward how Sylow analysis feeds into the study of Simple Groups.

Key takeaways
  • For ∣G∣=pkm|G|=p^k m with p∤mp\nmid m, a subgroup of order pkp^k (a Sylow pp-subgroup) always exists.
  • Any two Sylow pp-subgroups of GG are conjugate to each other.
  • Every pp-subgroup of GG lies inside some Sylow pp-subgroup.
  • The number npn_p of Sylow pp-subgroups satisfies np≡1(modp)n_p \equiv 1 \pmod p and np∣mn_p \mid m.
  • Proofs use group actions on cosets and on the set of Sylow subgroups by conjugation, plus the Orbit-Stabilizer Theorem.
  • Key idea: conjugation turns "how many Sylow subgroups" into a counting problem solvable via pp-group orbit arithmetic.

Where this lives today

Sylow theory is the backbone of the classification of finite simple groups: nearly every argument ruling out simplicity of a group of a specific order (as previewed in the ∣G∣=60|G|=60 example) proceeds by pinning down npn_p for well-chosen primes and deriving a normal subgroup or a contradiction. In computational group theory and computer algebra systems (GAP, Magma), Sylow subgroup algorithms are core primitives used to compute group structure, solve membership and conjugacy problems, and analyze permutation groups arising in combinatorics and coding theory. In number theory, Sylow-type analysis of Galois groups helps determine the structure of decomposition and inertia groups at primes, connecting to ramification theory. In cryptography, understanding the subgroup structure of groups used in Diffie–Hellman-type protocols (including elliptic curve groups) relies on Sylow-flavored reasoning about subgroup orders to assess security against subgroup-confinement attacks. Physics and crystallography use finite group classification techniques descended from Sylow-style order arguments when enumerating possible point groups and space groups compatible with a given lattice symmetry.

Ask the tutor: How do we combine Sylow constraints on two different primes simultaneously to force a group of a given order to have a nontrivial normal subgroup?

DefinitionSylow p-subgroup

Let GG be a finite group and let pp be a prime dividing ∣G∣|G|. Write

∣G∣=pkm,p∤m,k≥1.|G| = p^k m, \qquad p \nmid m, \qquad k \ge 1.

A subgroup P≤GP \le G is called a Sylow pp-subgroup of GG if ∣P∣=pk|P| = p^k. Equivalently, PP is a Sylow pp-subgroup if PP is a pp-subgroup of GG (every element has order a power of pp) that is maximal among pp-subgroups of GG under inclusion — this equivalence is a consequence of the theorems below, not part of the definition. We write Sylp(G)\mathrm{Syl}_p(G) for the set of all Sylow pp-subgroups of GG, and np=∣Sylp(G)∣n_p = |\mathrm{Syl}_p(G)|.

Intuition

Think of ∣G∣|G| as a number with a prime factorization, and think of each prime pp as contributing a "layer" of structure to GG. The Sylow pp-subgroup is the largest possible pp-group hiding inside GG — it uses up the entire pp-part of the order and none of the rest. Existence says this layer is never empty: no matter how complicated GG is, the full pp-power always gets realized as an actual subgroup, not just as an element (which is all Cauchy's theorem gives you). Conjugacy says that although there may be many Sylow pp-subgroups, they are all "the same shape" sitting differently inside GG: GG acts transitively on them by conjugation, so no Sylow subgroup is structurally distinguished from another. Finally, the counting theorem says the number of these copies cannot be arbitrary — it must divide the "co-pp" part mm of ∣G∣|G| and must leave remainder 11 upon division by pp. This rigidity is what makes Sylow theory such a powerful obstruction tool: often the congruence and divisibility constraints leave only np=1n_p = 1 as a possibility, which then forces that Sylow subgroup to be normal.

Figure
Divisors of m=15m=15 compatible with np≡1(mod5)n_p\equiv 1\pmod 5, illustrating why n5∈{1}n_5\in\{1\} for ∣G∣=60=22⋅3⋅5|G|=60=2^2\cdot3\cdot5
Candidates for n₅ dividing m = 15, with n₅ ≡ 1 (mod 5) 1 3 5 15 Only 1 ≡ 1 (mod 5) survives among divisors of 15 satisfying n₅≡1(5): n₅=1
TheoremThe Sylow Theorems

Let GG be a finite group and pp a prime dividing ∣G∣|G|. Write ∣G∣=pkm|G| = p^k m with p∤mp \nmid m and k≥1k \ge 1. Then:

(Sylow I — Existence). GG has a subgroup of order pkp^k; that is, Sylp(G)≠∅\mathrm{Syl}_p(G) \ne \varnothing.

(Sylow II — Conjugacy and covering). If PP is a Sylow pp-subgroup of GG and H≤GH \le G is any pp-subgroup, then there exists g∈Gg \in G with H≤gPg−1H \le gPg^{-1}. In particular, any two Sylow pp-subgroups of GG are conjugate in GG.

(Sylow III — Counting). Let np=∣Sylp(G)∣n_p = |\mathrm{Syl}_p(G)|. Then

np≡1(modp),np∣m.n_p \equiv 1 \pmod p, \qquad n_p \mid m.

Moreover, if P∈Sylp(G)P \in \mathrm{Syl}_p(G), then np=[G:NG(P)]n_p = [G : N_G(P)], where NG(P)N_G(P) is the normalizer of PP in GG.

LemmaOrbit-counting for pp-group actions

Let PP be a finite pp-group acting on a finite set XX. Then

∣X∣≡∣XP∣(modp),|X| \equiv |X^P| \pmod p,

where XP={x∈X:gx=x for all g∈P}X^P = \{x \in X : gx = x \text{ for all } g \in P\} is the set of fixed points.

Proof

This is standard from p-Groups and Their Properties; we include the short argument since everything else depends on it. Partition XX into PP-orbits. By the Orbit-Stabilizer Theorem, each orbit has size [P:StabP(x)][P : \mathrm{Stab}_P(x)], which divides ∣P∣=pa|P| = p^a for some aa, hence each orbit size is a power of pp (possibly p0=1p^0=1). An orbit has size 11 exactly when its unique element is fixed by all of PP, i.e. exactly the orbits contained in XPX^P. Summing orbit sizes,

∣X∣=∣XP∣+∑orbits of size>1(orbit size),|X| = |X^P| + \sum_{\text{orbits of size} > 1} (\text{orbit size}),

and every term in the sum is divisible by pp (being paip^{a_i} with ai≥1a_i \ge 1). Reducing mod pp gives ∣X∣≡∣XP∣(modp)|X| \equiv |X^P| \pmod p. ■\blacksquare

∎
Proof

We now prove the three Sylow theorems in turn. Throughout, ∣G∣=pkm|G| = p^k m, p∤mp \nmid m.

Proof of Existence (Sylow I). We use induction on ∣G∣|G| together with the class equation, a standard combination for pp-group existence arguments. Base case: if ∣G∣=1|G| = 1 there is nothing to prove (take k=0k=0, P={e}P = \{e\}); more substantively, if k=0k = 0 (i.e. p∤∣G∣p \nmid |G|) the trivial subgroup already has order p0=1=pkp^0=1=p^k. So assume k≥1k \ge 1. Case 1: p∣∣Z(G)∣p \mid |Z(G)|, where Z(G)Z(G) is the center of GG. By Cauchy's Theorem applied to the abelian group Z(G)Z(G), there is an element z∈Z(G)z \in Z(G) of order pp. Let N=⟨z⟩N = \langle z \rangle; since zz is central, N⊴GN \trianglelefteq G. Consider Gˉ=G/N\bar G = G/N, which has order pk−1mp^{k-1}m. By induction (this is a smaller group, since ∣N∣=p>1|N|=p>1), Gˉ\bar G has a subgroup Pˉ\bar P of order pk−1p^{k-1}. Let PP be the preimage of Pˉ\bar P under the quotient map G→GˉG \to \bar G. Then PP is a subgroup of GG with ∣P∣=∣Pˉ∣⋅∣N∣=pk−1⋅p=pk|P| = |\bar P|\cdot|N| = p^{k-1}\cdot p = p^k, as required. Case 2: p∤∣Z(G)∣p \nmid |Z(G)|. Apply the class equation:

∣G∣=∣Z(G)∣+∑i[G:CG(xi)],|G| = |Z(G)| + \sum_{i} [G : C_G(x_i)],

where the sum runs over representatives xix_i of the non-central conjugacy classes, and CG(xi)C_G(x_i) is the centralizer of xix_i. Since p∣∣G∣p \mid |G| but p∤∣Z(G)∣p \nmid |Z(G)|, there must exist at least one term [G:CG(xi)][G:C_G(x_i)] in the sum with p∤[G:CG(xi)]p \nmid [G:C_G(x_i)] — otherwise every term on the right side except possibly ∣Z(G)∣|Z(G)| would be divisible by pp, forcing p∣∣Z(G)∣p \mid |Z(G)| (since pp divides the total ∣G∣|G| and all other class-sizes), a contradiction. Fix such an xix_i and set H=CG(xi)H = C_G(x_i). Since [G:H][G:H] is not divisible by pp and ∣G∣=∣H∣⋅[G:H]=pkm|G| = |H|\cdot[G:H] = p^k m, and p∤[G:H]p \nmid [G:H], we must have pk∣∣H∣p^k \mid |H|, so HH is a proper subgroup (as xi∉Z(G)x_i \notin Z(G) means H≠GH \ne G) whose order is still divisible by pkp^k. By induction (applied to the smaller group HH), HH has a subgroup of order pkp^k, and this subgroup is also a subgroup of GG of order pkp^k. In both cases we produce a subgroup of order pkp^k, completing the induction. ■\blacksquare

Proof of Conjugacy and Covering (Sylow II). Fix a Sylow pp-subgroup PP (which exists by Sylow I) and let H≤GH \le G be any pp-subgroup. Let X=G/PX = G/P denote the set of left cosets of PP in GG, so ∣X∣=[G:P]=m|X| = [G:P] = m, which is not divisible by pp by hypothesis. Let HH act on XX by left multiplication: h⋅(gP)=(hg)Ph \cdot (gP) = (hg)P. By the orbit-counting lemma applied to the pp-group HH acting on XX,

∣X∣≡∣XH∣(modp).|X| \equiv |X^H| \pmod p.

Since p∤∣X∣=mp \nmid |X| = m, we conclude ∣XH∣≢0(modp)|X^H| \not\equiv 0 \pmod p; in particular XH≠∅X^H \ne \varnothing, so there exists a fixed coset gP∈XHgP \in X^H. This means

h(gP)=gPfor all h∈H,h(gP) = gP \quad \text{for all } h \in H,

i.e. g−1hg∈Pg^{-1}hg \in P for all h∈Hh \in H, i.e. g−1Hg≤Pg^{-1}Hg \le P, equivalently H≤gPg−1H \le gPg^{-1}. This proves the covering statement. For conjugacy of two Sylow pp-subgroups P,QP, Q: apply the above with H=QH = Q (a pp-subgroup since Q∈Sylp(G)Q \in \mathrm{Syl}_p(G)) to get Q≤gPg−1Q \le gPg^{-1} for some g∈Gg \in G. But ∣gPg−1∣=∣P∣=pk=∣Q∣|gPg^{-1}| = |P| = p^k = |Q|, so Q≤gPg−1Q \le gPg^{-1} together with equal finite cardinalities forces Q=gPg−1Q = gPg^{-1}. Hence PP and QQ are conjugate. ■\blacksquare

Proof of Counting (Sylow III). Fix a Sylow pp-subgroup PP. By Sylow II, GG acts transitively by conjugation on Sylp(G)\mathrm{Syl}_p(G) (transitivity is exactly the conjugacy statement just proved), so by the Orbit-Stabilizer Theorem,

np=∣Sylp(G)∣=[G:StabG(P)],n_p = |\mathrm{Syl}_p(G)| = [G : \mathrm{Stab}_G(P)],

where StabG(P)={g∈G:gPg−1=P}=NG(P)\mathrm{Stab}_G(P) = \{g \in G : gPg^{-1} = P\} = N_G(P), the normalizer of PP. This proves np=[G:NG(P)]n_p = [G:N_G(P)]. Since P≤NG(P)≤GP \le N_G(P) \le G, we have

m=[G:P]=[G:NG(P)]⋅[NG(P):P]=np⋅[NG(P):P],m = [G:P] = [G:N_G(P)]\cdot[N_G(P):P] = n_p \cdot [N_G(P):P],

so np∣mn_p \mid m. This proves the divisibility claim. For the congruence np≡1(modp)n_p \equiv 1 \pmod p: let PP act on Sylp(G)\mathrm{Syl}_p(G) by conjugation (restricting the GG-action to the subgroup PP). By the orbit-counting lemma,

np=∣Sylp(G)∣≡∣Sylp(G)P∣(modp),n_p = |\mathrm{Syl}_p(G)| \equiv |\mathrm{Syl}_p(G)^P| \pmod p,

where Sylp(G)P\mathrm{Syl}_p(G)^P is the set of Sylow pp-subgroups fixed by conjugation by every element of PP. We claim Sylp(G)P={P}\mathrm{Syl}_p(G)^P = \{P\}. Certainly PP itself is fixed (conjugation by any x∈Px \in P sends PP to xPx−1=PxPx^{-1}=P). Suppose Q∈Sylp(G)PQ \in \mathrm{Syl}_p(G)^P, i.e. xQx−1=QxQx^{-1}=Q for all x∈Px \in P; this says P≤NG(Q)P \le N_G(Q). Now both PP and QQ are Sylow pp-subgroups of the group NG(Q)N_G(Q) (they have the correct order pkp^k and lie inside NG(Q)N_G(Q); note Q⊴NG(Q)Q \trianglelefteq N_G(Q) by definition of normalizer). Since QQ is normal in NG(Q)N_G(Q), it is the unique Sylow pp-subgroup of NG(Q)N_G(Q): indeed by Sylow II applied inside NG(Q)N_G(Q), PP and QQ are conjugate in NG(Q)N_G(Q), but QQ is normal there, so conjugation fixes QQ, forcing P=QP = Q. Hence Sylp(G)P={P}\mathrm{Syl}_p(G)^P = \{P\}, so ∣Sylp(G)P∣=1|\mathrm{Syl}_p(G)^P| = 1, giving

np≡1(modp).n_p \equiv 1 \pmod p.

This completes the proof of all three Sylow theorems. ■\blacksquare

∎
CorollaryNormality criterion

A Sylow pp-subgroup PP of GG is normal in GG if and only if np=1n_p = 1.

Proof

If np=1n_p = 1 then the unique Sylow pp-subgroup PP satisfies gPg−1∈Sylp(G)={P}gPg^{-1} \in \mathrm{Syl}_p(G) = \{P\} for every g∈Gg\in G (since conjugates of Sylow subgroups are Sylow subgroups of the same order), so gPg−1=PgPg^{-1}=P for all gg, i.e. P⊴GP \trianglelefteq G. Conversely if P⊴GP \trianglelefteq G, then for every g∈Gg \in G, gPg−1=PgPg^{-1} = P, so the conjugacy class of PP under GG-action is {P}\{P\}; but by Sylow II this conjugacy class is all of Sylp(G)\mathrm{Syl}_p(G), forcing np=1n_p=1. ■\blacksquare

∎

Worked examples

Example 1: Sylow subgroups of S4S_4. We have ∣S4∣=24=23⋅3|S_4| = 24 = 2^3 \cdot 3. Take p=2p=2: Sylow 22-subgroups have order 88, and m=3m=3. By Sylow III, n2∣3n_2 \mid 3 and n2≡1(mod2)n_2 \equiv 1 \pmod 2; the divisors of 33 are 1,31,3, and both are odd, so a priori n2∈{1,3}n_2 \in \{1,3\}. We exhibit one Sylow 22-subgroup directly: the dihedral subgroup generated by the 44-cycle (1234)(1234) and the transposition (13)(13),

D={e, (1234), (13)(24), (1432), (13), (24), (12)(34), (14)(23)},D = \{e,\ (1234),\ (13)(24),\ (1432),\ (13),\ (24),\ (12)(34),\ (14)(23)\},

has order 88 (it is isomorphic to D4D_4, the symmetries of a square with vertices labeled 1,2,3,41,2,3,4 in order). One checks DD is not normal in S4S_4 (for instance, conjugating (13)∈D(13)\in D by (12)(12) gives (23)∉D(23)\notin D), so by the Normality criterion n2≠1n_2 \ne 1, forcing n2=3n_2 = 3. Indeed the three Sylow 22-subgroups correspond to the three ways of pairing {1,2,3,4}\{1,2,3,4\} into two disjoint transpositions determining a "square" structure: {12∣34}\{12|34\}, {13∣24}\{13|24\}, {14∣23}\{14|23\}. Take p=3p=3: Sylow 33-subgroups have order 33, and m=8m=8. By Sylow III, n3∣8n_3 \mid 8 and n3≡1(mod3)n_3\equiv1\pmod3; candidates among divisors of 88 are 1,41,4 (since 2,8≡2,2(mod3)2,8\equiv 2,2\pmod3 fail, and 1≡11\equiv1, 4≡14\equiv1 both pass). The Sylow 33-subgroups are generated by 33-cycles; there are 88 distinct 33-cycles in S4S_4 (namely (43)×2=8\binom{4}{3}\times 2 = 8, choosing 33 of the 44 symbols and one of the two cyclic orders), and each Sylow 33-subgroup ⟨σ⟩={e,σ,σ2}\langle \sigma\rangle=\{e,\sigma,\sigma^2\} contains exactly 22 of these 33-cycles. Hence n3=8/2=4n_3 = 8/2 = 4, consistent with the constraint (and ruling out n3=1n_3=1, confirming no Sylow 33-subgroup is normal in S4S_4).

Example 2: A group of order 6060 cannot be simple unless it satisfies strong constraints — Sylow 55-analysis. Let ∣G∣=60=22⋅3⋅5|G| = 60 = 2^2\cdot 3\cdot 5, and consider p=5p=5, so m=12m = 12. By Sylow III, n5∣12n_5 \mid 12 and n5≡1(mod5)n_5 \equiv 1 \pmod 5. The divisors of 1212 are 1,2,3,4,6,121,2,3,4,6,12; checking each mod 55: 1≡11\equiv1, 2≡22\equiv2, 3≡33\equiv3, 4≡44\equiv4, 6≡16\equiv1, 12≡212\equiv2. Only 11 and 66 satisfy the congruence, so

n5∈{1,6}.n_5 \in \{1, 6\}.

If n5=1n_5=1, the unique Sylow 55-subgroup is normal by the Normality criterion, and GG is not simple. Suppose instead n5=6n_5 = 6. Each Sylow 55-subgroup has order 55 (prime), so any two distinct Sylow 55-subgroups intersect trivially (a subgroup of prime order has no nontrivial proper subgroups, so if two order-55 subgroups shared a nonidentity element they would coincide). Counting nonidentity elements across the 66 Sylow 55-subgroups: each contributes 5−1=45-1=4 elements of order 55, and these sets are pairwise disjoint, giving

6×4=246 \times 4 = 24

distinct elements of order 55 in GG. This leaves only 60−24=3660 - 24 = 36 elements for everything else, including a Sylow 22-subgroup analysis: one shows (by a further Sylow argument on p=2p=2 or p=3p=3, or via the standard classification argument) that these constraints are exactly what force G≅A5G \cong A_5 to be the unique simple possibility of order 6060 — this is the beginning of the classical proof that A5A_5 is simple, previewing Simple Groups and Classification via Sylow Analysis. The key point illustrated here is purely computational: the congruence n5≡1(mod5)n_5\equiv1\pmod5 combined with divisibility n5∣12n_5\mid 12 crushed six candidate values down to two, and basic element-counting eliminated further structural possibilities.

CounterexampleWhy finiteness of GG is essential

Drop the hypothesis that GG is finite, keeping pp a prime "dividing ∣G∣|G|" in the sense that GG has elements of order pp but ∣G∣=∞|G|=\infty. Consider

G=⨁i=1∞Z/pZ,G = \bigoplus_{i=1}^{\infty} \mathbb{Z}/p\mathbb{Z},

the direct sum (not product) of countably many copies of Z/pZ\mathbb{Z}/p\mathbb{Z}, an infinite abelian group in which every nonidentity element has order pp. Here there is no well-defined "pp-part" pkp^k of ∣G∣|G| since ∣G∣=∞|G|=\infty, so the very statement of Sylow I becomes meaningless as phrased — but one can still ask whether there is a maximal pp-subgroup, and in this abelian example GG itself is a pp-group, so the question trivializes. A sharper failure: let G=S∞G = S_\infty, the group of all permutations of N\mathbb{N} that move only finitely many points is one option, but to see genuine failure of the counting theorem take GG to be any infinite group with infinitely many conjugate copies of a fixed finite pp-subgroup PP that is maximal as a pp-subgroup. For instance let

G=(⨁i=1∞Z/pZ)⋊Z/pZ,G = \left(\bigoplus_{i=1}^\infty \mathbb{Z}/p\mathbb{Z}\right) \rtimes \mathbb{Z}/p\mathbb{Z},

with the last factor acting by a shift-like automorphism permuting the direct summands nontrivially and infinitely. Here maximal finite pp-subgroups exist but there are infinitely many of them, and no finite index or congruence relation like np≡1 (mod p)n_p \equiv 1\ (\mathrm{mod}\ p) constrains their count, since "npn_p" is infinite and the quotient [G:NG(P)][G:N_G(P)] may itself be infinite, making the orbit-counting lemma (which crucially used finiteness of X=Sylp(G)X=\mathrm{Syl}_p(G) to reduce mod pp) inapplicable. The entire proof of Sylow III rests on partitioning a finite set XX into pp-power-sized orbits and reducing mod pp; with XX infinite this argument gives no information at all. This shows finiteness of GG (or at least of the relevant coset spaces) is not a cosmetic hypothesis but the mechanism the proof depends on.

Scope, edge cases & common traps

SituationWhat holdsWhat can fail / must be checked
p∤∣G∣p \nmid |G|Sylow pp-subgroup is trivial, np=1n_p=1 vacuouslyStatement is trivial but still technically covered (k=0k=0)
GG abelianSylow pp-subgroup is unique and normal, np=1n_p=1 alwaysDo not assume the converse (np=1⇒n_p=1 \Rightarrow abelian) — false in general
∣G∣=∞|G|=\inftyDefinition of "pkp^k-part of ∣G∣|G|" breaks downExistence/counting proofs use finite coset counting; do not port them naively
Multiple primes dividing ∣G∣|G|Sylow theorem applies separately to each primenpn_p for different primes are independent constraints, not automatically compatible with a single group unless verified simultaneously
np=1n_p=1P⊴GP\trianglelefteq GDo not conclude G/PG/P or GG itself is simple or abelian from this alone

The mathematician behind it

Origins · 1 minPeter Ludwig Mejdell Sylow and how this was discovered
PL
Peter Ludwig Mejdell Sylow
1832–1918 · Norwegian mathematician, gymnasium teacher

Sylow spent much of his career teaching secondary school mathematics in Norway while working on group theory in his spare time, encouraged by contact with fellow Norwegian mathematician Sophus Lie. His 1872 paper, barely six pages, proved the three theorems that now bear his name — a landmark example of deep structural mathematics discovered outside a conventional academic research post, later expanded through his collaboration with Lie on editing the collected works of Niels Henrik Abel.

History
- 1832: Peter Ludwig Mejdell Sylow born in Christiania (Oslo), Norway.
- 1862: Sylow begins substituting for Sophus Lie's teacher and later becomes acquainted with him, sparking mathematical collaboration.
- 1872: Sylow publishes "Théorèmes sur les groupes de substitutions" in Mathematische Annalen, proving existence, conjugacy, and the counting theorem for what are now called Sylow subgroups.
- 1881–1902: Sylow collaborates with Lie on editing and publishing the complete works of Niels Henrik Abel.
- 1918: Sylow dies in Kristiania, having spent decades as a respected teacher whose theorems became foundational to modern finite group theory.
Why does Sylow I not just follow from Cauchy's theorem applied repeatedly?

Cauchy's theorem only guarantees an element (hence a cyclic subgroup) of prime order pp; iterating it does not automatically build a subgroup of order pkp^k for k>1k>1, because elements of order pp found this way need not generate a subgroup of the right order or even a pp-group at all. Sylow I requires a genuinely different induction, using the class equation and normal subgroups from the center, or quotienting by a central subgroup of order pp to reduce to a smaller case. The subtlety is ensuring the subgroup constructed has exactly order pkp^k, using all of the pp-part of ∣G∣|G| at once, which is a strictly stronger existence claim than Cauchy's single-element guarantee.

Is every maximal p-subgroup automatically a Sylow p-subgroup?

Yes, and this is a consequence of Sylow II (the covering property), not part of the definition. If HH is a pp-subgroup that is maximal among pp-subgroups but ∣H∣<pk|H|<p^k, then Sylow II says HH is contained in some conjugate gPg−1gPg^{-1} of an actual Sylow pp-subgroup PP, which is itself a pp-subgroup strictly larger than HH (since ∣gPg−1∣=pk>∣H∣|gPg^{-1}|=p^k>|H|). This contradicts maximality of HH, so ∣H∣=pk|H|=p^k after all. Thus "maximal pp-subgroup" and "Sylow pp-subgroup" describe the same objects, which is why some textbooks use maximality as the defining property instead of the explicit order formula.

Can npn_p be computed without knowing the full group structure of GG?

Often only partially: Sylow III gives necessary constraints (np≡1(modp)n_p\equiv1\pmod p, np∣mn_p\mid m) that narrow down candidates, but pinning down the exact value of npn_p generally requires more information, such as explicit subgroup constructions or further combinatorial arguments like element counting (as in the order-6060 example). In some cases the constraints alone force a unique value, immediately determining npn_p; in other cases several values remain possible and additional structural analysis, sometimes using multiple primes simultaneously, is needed to resolve the ambiguity.

Why is the normalizer NG(P)N_G(P) the stabilizer of PP under conjugation, and why does that matter?

By definition, NG(P)={g∈G:gPg−1=P}N_G(P)=\{g\in G: gPg^{-1}=P\}, which is exactly the stabilizer of PP under the conjugation action of GG on subsets of GG. This matters because the Orbit-Stabilizer Theorem then directly gives np=[G:NG(P)]n_p=[G:N_G(P)], converting the abstract count of Sylow subgroups into an index computation. It also explains the Normality criterion: P⊴GP\trianglelefteq G exactly when NG(P)=GN_G(P)=G, i.e. when the orbit of PP under conjugation is a single point, i.e. np=1n_p=1.

Does Sylow theory tell us anything when pk=∣G∣p^k = |G|, i.e. GG itself is a p-group?

In that degenerate case m=1m=1, so the only Sylow pp-subgroup is GG itself, and indeed np=1n_p=1 trivially satisfies both np≡1(modp)n_p\equiv1\pmod p and np∣1n_p\mid 1. The theorems add no new information here since there is only one subgroup of full order, namely GG. The theorems become substantive precisely when GG has "extra" structure beyond its pp-part, i.e. when m>1m>1 and there is room for multiple copies of a maximal pp-subgroup to sit differently inside GG.

How does Sylow theory connect to proving simplicity or non-simplicity of specific finite groups?

The standard strategy is to compute the constraints on npn_p for a well-chosen prime pp dividing ∣G∣|G|, and show that either np=1n_p=1 (giving an immediate normal subgroup, hence non-simplicity unless GG itself has prime order) or that any other allowed value of npn_p forces enough elements of a certain order to overcrowd the group, again yielding a normal subgroup by further counting. This is exactly the technique previewed in the order-6060 worked example, and it generalizes into the systematic order-by-order elimination process central to Classification via Sylow Analysis and the broader study of Simple Groups.

Before this

Where this leads