Motivation
An integer is built by multiplying primes, while a finite group can be built by successive extensions of simple groups. If different normal-subgroup steps are chosen, do we get the same simple building blocks? A chain exhibits one construction; uniqueness needs a comparison of different choices.
The Jordan–Hölder theorem preserves the blocks and their multiplicities. It does not determine their assembly: and the Klein four group already have the same simple factors. We prove uniqueness with a subgroup diamond and induction, rather than quote a general refinement theorem as an unexplained dependency.
All groups here are finite. Quotient isomorphism and correspondence theorems from the first chapter on groups are prior background. We also use the solvability results and proved simplicity of from the chapter on derived series, solvability and . Neither finite-simple-group classification nor an infinite-length theorem is assumed.
Definitions
Definition 9.1.1
(Subnormal and composition series)
A subnormal series is a finite chain
where each lower group is normal in its predecessor. We use strict inclusions when counting nontrivial factors. A composition series has every quotient nontrivial and simple; these quotients are its composition factors. The trivial group has an empty series and no factors.
A simple group is nontrivial and has no normal subgroups besides itself and one. By correspondence, a simple-quotient step is exactly a maximal proper normal-subgroup step.
Normality is required only in the preceding group, not necessarily in all of . Maximal proper normal is also different from maximal proper subgroup. A nonabelian simple group can have many nontrivial proper subgroups, but no proper nontrivial normal ones. Inserting an arbitrary subgroup can make a proposed quotient undefined.
Theorems
Theorem 9.1.2
(Finite Jordan–Hölder theorem)
Every finite group has a composition series. Any two such series of the same group have the same multiset of simple factor isomorphism types. Their ordering may differ, but their length and the multiplicity of each type do not.
Every finite strict subnormal series refines to a composition series by inserting subgroups. Its composition-factor multiset is the union, with multiplicity, of the composition-factor multisets of its original quotient groups.
Theorem 9.1.3
(Extensions, length, and solvability)
Write for the composition length. If , the factor multiset of is the union of those of and . In particular,
A finite group is solvable exactly when all its composition factors are cyclic of prime order. Also , with equality exactly for groups of two-power order, including the trivial group.
Proof sketch or proof idea
Existence and refinement by finite descent
Proof
(Maximal normal descent and inverse-image refinement)
For nontrivial finite , the set of proper normal subgroups is finite and nonempty, containing one. Choose a maximal member . Its quotient is simple by correspondence. Repeat inside if it is nontrivial. Strictly decreasing positive orders force eventual arrival at one, producing a composition series.
For a step in a strict subnormal series, choose a composition series of . Inverse images under insert a chain from to . Adjacent normality follows from inverse-image normality, and the third isomorphism theorem identifies the new factors with those in the quotient series. Do this for each original step. The result is a composition refinement; it does not require every inserted term normal in the original group.
The finite hypothesis cannot be dropped from existence
Counterexample mode
An infinite group without a composition series
The infinite cyclic group has no finite composition series. Every nonzero subgroup is an infinite cyclic group , and it contains the proper nonzero normal subgroup . Thus the last nonzero term of any proposed finite series could not be simple. Infinite groups can have composition series in some cases, but finite generation alone does not supply one. Our existence proof uses decreasing finite orders; that is the termination invariant, rather than an assumption that every descent through subgroups must eventually stop.
The diamond compares different first choices
Proof
(Induction through the diamond )
Induct on to prove uniqueness. The trivial group is immediate. In two composition series of nontrivial , let the first proper terms be , both maximal proper normal subgroups of . If , their top factors agree and induction inside compares the tails.
If , neither contains the other: such containment would contradict maximality. The product is normal in and properly contains , so . Set , normal in and in both subgroups. The second isomorphism theorem gives
Both quotients are simple. Choose one composition series of , with its empty series allowed if . Prefixing yields a composition series of beginning ; prefixing does the same for .
Since , induction compares each original tail with its new series. Denote the factor multiset of the selected series of by . The full series through therefore has multiset
The full series through has the same multiset, with the first two types exchanged. This proves uniqueness for . Only strictly smaller groups are used in the inductive comparisons; uniqueness for has not been assumed to construct the diamond.
Finally, the inverse-image refinement of a general subnormal series has exactly the factors of its quotient-group series. The uniqueness just proved makes their union independent of all choices and completes the first theorem.
Proof X-ray
An intersection synchronizes two routes
Different maximal normal choices give different top quotients. Their product is the whole group; their intersection supplies a common lower group. The quotient theorem exchanges the two top types on opposite sides of the diamond, while the same series is used below the intersection. Induction replaces the original tails by these synchronized routes. Uniqueness concerns factor types, not equality of intermediate subgroups.
Extensions concatenate factor lists
Proof
(Concatenation, the order product and the two-power bound)
Lift a composition series of to a chain from to , then append a composition series of . This is a composition series of with precisely those two lists. Jordan–Hölder makes its length intrinsic, proving additivity. Multiplying cancels consecutive orders, proving the order formula.
Every nontrivial simple factor has order at least two, hence . Equality forces every factor order two, so is a power of two. Conversely, if , every factor is a two-group. A nontrivial simple finite -group is : its center is nontrivial by the class equation; simplicity makes the group central and hence abelian; an abelian simple group is prime cyclic. Thus all factors here are , and their order product gives length . For the trivial group the empty product and zero length give equality as well.
Solvability excludes exactly the nonabelian simple types
Proof
(Solvable simple groups are cyclic of prime order)
A finite simple solvable group is abelian. Its derived subgroup is normal, so is either one or the whole group. The second case would leave every derived term unchanged, contradicting solvability. An abelian simple finite group is prime cyclic: a nonidentity element generates the group, and composite cyclic order would admit a proper nontrivial subgroup.
If is solvable, its subgroup quotients in a composition series are solvable by the chapter on derived series and solvability, so every simple factor has this prime cyclic form. Conversely, if all composition factors are prime cyclic, start at the bottom and repeatedly apply the proved solvable-extension theorem to the successive abelian quotients. Every group in the chain, ending at , is solvable. This does not assert that the prime divisors of the order alone decide solvability: nonabelian simple types can have composite order.
Worked examples
Worked example 9.1.4
(A composition series of has a merely subnormal step)
The chain has factors , , , . The sign kernel gives the first normality; conjugation permutes the three double transpositions, giving normality of ; its order-two subgroup is normal because is abelian. The factor orders prove simplicity. The order-two subgroup is not normal in , since a three-cycle conjugates its generator to a different double transposition. Adjacent normality is nevertheless sufficient. Thus the length is four, and the factors confirm solvability.
Worked example 9.1.5
(A larger group can have fewer composition steps)
is a composition series: the first quotient is , and is simple by the full class-count proof in the chapter on derived series and . Thus the factors are and the nonabelian simple ; the length is two and is not solvable. Their orders multiply to . The prime factors of sixty are not extra composition factors of . Simple-group factors need not have prime order, and subdividing their order numerically does not subdivide their normal-subgroup structure.
Worked example 9.1.6
(The same factors in a different order)
For , the series has factors , , . Another series has factors , , . Product projections and the sign quotient identify the factors; each lower group is normal in its predecessor. The subgroups differ, but their simple multisets agree. Both order products are thirty, and both lengths are three.
Counterexample mode
Factor uniqueness does not reconstruct a group
and share the factor multiset , . They are not isomorphic: the first has an element of order four, while the second is killed by two. Composition factors forget extension data. Length is not even monotone for arbitrary subgroups: the point-stabilizer lies in , but their lengths are four and two. Additivity belongs to normal-subgroup quotients, not arbitrary inclusions.
Common mistakes
Common mistake
(Simple concerns normal subgroups only)
Simple excludes proper nontrivial normal subgroups, not all proper nontrivial subgroups. Composition series require maximal normal steps, not arbitrary maximal-subgroup steps. Requiring every term normal in the full group would exclude valid series.
Common mistake
(A prime divisor need not give a cyclic factor)
A prime dividing the group order need not give a cyclic composition factor. Cauchy's theorem finds a subgroup of that prime order, but does not place it as a quotient in a composition series. Jordan–Hölder preserves simple isomorphism types with multiplicity, not their ordering, the chain itself or the extension structure.
Summary
Finite descent produces composition series, and quotient inverse images refine any finite subnormal chain. The maximal-normal-subgroup diamond and induction prove simple-factor uniqueness. Normal extensions concatenate lists; solvability means all factors are prime cyclic. These are the building blocks, with further extension data needed to recover the group.
Exercises
Quick check
(Prove a group of order has exactly composition factors, all .)
Solution · Answer
Every factor has -power order, since subgroup and quotient orders divide the original -power order. A simple nontrivial -group has nontrivial center, so simplicity makes it abelian and thus . The product of factor orders is , so exactly occur. At the list is empty.
Quick check
(Construct a series proving normal-extension length additivity. Characterize equality in , including the trivial group.)
Solution · Answer
Lift a composition series of to a chain ending at , then append a series of . Correspondence and the third isomorphism theorem identify the factors; uniqueness proves the common length is the sum. Equality in the logarithmic bound forces every factor order two, hence a two-power group order. Conversely the first solution at gives equality for every two-group. The trivial group's length is zero and order one, so it is included.
Quick check
(Is “ divides the order” equivalent to “ occurs as a composition factor”? Give a general counterexample and prove the corrected multiplicity statement for finite solvable groups.)
Solution · Answer
A factor implies divides the group order by the product formula. The converse fails for : two, three and five divide sixty, but its only composition factor is . If is solvable, all factors are prime cyclic. Their order product shows that occurs exactly times. This uses solvability rather than Cauchy's theorem alone.
Quick check
(Determine the factor multiset and length of the order- dihedral group, , from the prime factorization of . Evaluate at .)
Solution · Answer
The rotation subgroup is normal with quotient . A series of divides its order by one prime at a time using its unique cyclic subgroups. For , the factors are copies of plus one extra , and the length is . At , there are three factors and one , so length four. Sharing this multiset with does not imply isomorphism.