Unoriented bordism and Thom's theorem

Intuition [Beginner]

Two shapes are called bordant when together they form the edge of a higher-dimensional shape. Picture two circles drawn on the plane. They look different, yet you can find a flat ribbon-with-holes — a pair of pants — whose boundary is precisely those two circles together. The pair of pants is the bordism. The circles are bordant because such a connecting region exists.

The whole study of unoriented bordism asks: which closed shapes can be the boundary of something? Which ones cannot? And when two shapes are both boundaries of something larger, what happens if you glue those things along a shared piece? It turns out this glueing is well-behaved: bordism is an equivalence relation, and you can add bordism classes by just placing two shapes side by side. The set of bordism classes of $n$-dimensional shapes forms an addition group, written $N_n$.

Rene Thom, in 1954, found the complete answer to the bordism question for closed smooth shapes. His theorem reduces the whole geometric problem to a problem in stable homotopy theory, and it lets you decide whether two shapes are bordant by comparing a small list of mod-2 numbers — the Stiefel-Whitney numbers — attached to each shape. The slogan: a closed smooth shape is a boundary if and only if all its Stiefel-Whitney numbers vanish.

Visual [Beginner]

The pair of pants is the universal cartoon of bordism. Three boundary circles, one connecting tube. Reverse one of the legs and you get a saddle: two circles bordant to one. Two circles bordant to zero circles means the two circles bound a disjoint union of two disks. The boundary of a single disk is one circle.

Going up one dimension, two real projective planes $R{P}^2$ are bordant to the empty set because their disjoint union bounds a four-manifold built from a cobordism along a tube. A single $R{P}^2$ on its own is not a boundary, however — it is the only non-zero element of $N_2$, the bordism group in dimension 2.

Worked example [Beginner]

A two-sphere $S^2$ is bordant to the empty set. The bordism is the closed three-dimensional ball $B^3$ whose boundary is exactly $S^2$.

Step 1. Identify the closed shape: a round two-sphere.

Step 2. Find a higher-dimensional shape with that as its boundary: the solid ball.

Step 3. Check that the boundary of the solid ball is the single sphere with nothing else.

Step 4. Conclude the bordism class of $S^2$ in $N_2$ is the same as the class of the empty set, which is the zero element.

What this tells us: two spheres count as a boundary; the real projective plane $R{P}^2$ does not.

Check your understanding [Beginner]

Formal definition [Intermediate+]

Let ${\mathcal{M}}_n$ denote the class of closed smooth $n$-manifolds (compact, without boundary). Define the relation $M\sim N$ on ${\mathcal{M}}_n$ by declaring $M\sim N$ when there exists a compact smooth $(n+1)$-manifold $W$ with $\partial W\cong M\bigsqcup N$. The relation $\sim$ is an equivalence relation:

• Reflexivity. Take $W=M\times [0,1]$ with $\partial W=M\bigsqcup M$.
• Symmetry. If $W$ realises $M\sim N$, the same $W$ realises $N\sim M$ (no orientation data to reverse).
• Transitivity. Glue two cobordisms along their common boundary component using collars.

Definition (unoriented bordism group). The set $$ N_n := \mathcal{M}_n \big/ \sim $$ is an abelian group with addition $[M]+[N]=[M\bigsqcup N]$ and zero $[\varnothing ]$. Each element is its own inverse because $M\bigsqcup M=\partial (M\times [0,1])$, so $N_n$ is a vector space over ${\mathbb{F}}_2=\mathbb{Z}/2$.

Definition (unoriented bordism ring). The direct sum $$ N_* := \bigoplus_{n \geq 0} N_n $$ becomes a graded-commutative ring under Cartesian product $[M]\cdot [N]=[M\times N]$. The unit is the class of the one-point space in dimension zero.

The Stiefel-Whitney classes of $TM$ (see 03.06.03) live in $H^{\ast }(M;{\mathbb{F}}_2)$. A Stiefel-Whitney number of a closed smooth $n$-manifold $M$ is, for any ordered tuple $\omega =(i_1,\dots ,i_r)$ with $i_1+\cdots +i_r=n$, the mod-2 integer $$ \mathrm{sw}\omega(M) := \langle w{i_1}(TM) \cdot w_{i_2}(TM) \cdots w_{i_r}(TM),\ [M]_2 \rangle \in \mathbb{F}_2, $$ where $[M{]}_2\in H_n(M;{\mathbb{F}}_2)$ is the mod-2 fundamental class of the (necessarily mod-2 orientable) manifold $M$.

Counterexamples to common slips [Intermediate+]

• Bordism is weaker than diffeomorphism. The torus $T^2$ and the Klein bottle $K$ are not diffeomorphic, yet both have the same single non-zero Stiefel-Whitney number $w_1^2$ evaluated against the fundamental class — and indeed $[T^2]=[K]$ in $N_2$ once a careful sign analysis is done, while $[T^2]=0$ vanishes (the torus bounds a solid handlebody). The Klein bottle is the non-zero element. Confusing diffeomorphism with bordism is the standard slip.
• Mod-2 fundamental class versus integral. The fundamental class $[M{]}_2\in H_n(M;{\mathbb{F}}_2)$ exists for any closed smooth manifold, oriented or not, because every closed smooth manifold is $\mathbb{Z}/2$-orientable. An integral fundamental class requires the underlying manifold to be orientable; do not confuse the two when pairing characteristic classes against $[M]$.
• Empty manifold versus point. The zero element of $N_{\ast }$ is $[\varnothing ]$, not $[\mathrm{pt}]$. The single point is the unit of multiplication in $N_0$, not the additive zero.

Key theorem with proof [Intermediate+]

Theorem (Bordism invariance of Stiefel-Whitney numbers). If $M,N$ are bordant closed smooth $n$-manifolds, then for every partition $\omega$ of $n$ the Stiefel-Whitney numbers agree: $$ \mathrm{sw}\omega(M) = \mathrm{sw}\omega(N) \in \mathbb{F}_2. $$

Proof. Let $W$ be a compact smooth $(n+1)$-manifold with $\partial W=M\bigsqcup N$, and let $i_M:M\hookrightarrow W$ and $i_N:N\hookrightarrow W$ be the boundary inclusions. The tangent bundle of $W$, restricted along the boundary inclusion, splits as $$ i_M^* TW \cong TM \oplus \underline{\mathbb{R}}, $$ where $\underline{\mathbb{R}}$ is the product line bundle spanned by the inward normal vector. Stiefel-Whitney classes are stable under direct sum with a product bundle 03.06.03, so $$ i_M^* w_k(TW) = w_k(TM) $$ for every $k\ge 0$. The same identity holds for $i_N$.

Fix a partition $\omega =(i_1,\dots ,i_r)$ with $|\omega |:=\sum i_j=n$. Set $$ W_\omega := w_{i_1}(TW) \cdots w_{i_r}(TW) \in H^n(W; \mathbb{F}2). $$ Then $i_M^* W\omega = w_{i_1}(TM) \cdots w_{i_r}(TM) =: M_\omega$,andsimilarly$i_N^* W_\omega = N_\omega$.

The boundary relation in mod-2 homology reads $$ [M]_2 + [N]_2 = \partial [W]_2 = 0 \quad \text{in } H_n(W; \mathbb{F}_2), $$ where $[W{]}_2\in H_{n+1}(W,\partial W;{\mathbb{F}}_2)$ is the relative mod-2 fundamental class of the manifold-with-boundary, and the boundary map of the long exact sequence of the pair sends $[W{]}_2$ to $[\partial W{]}_2=[M{]}_2+[N{]}_2$ in $H_n(\partial W;{\mathbb{F}}_2)\cong H_n(M;{\mathbb{F}}_2)\oplus H_n(N;{\mathbb{F}}_2)$.

Pair this with $W_{\omega }$ pulled back to the boundary via $(i_M^{\ast },i_N^{\ast })$: $$ \langle W_\omega, [M]2 \rangle + \langle W\omega, [N]2 \rangle = \langle W\omega, [M]2 + [N]2 \rangle = \langle W\omega, \partial [W]2 \rangle. $$ By the relative-pairing identity (Stokes in ${\mathbb{F}}_2$-cohomology), $$ \langle W_\omega, \partial [W]2 \rangle = \langle \delta W\omega, [W]2 \rangle, $$ and $\delta W\omega = 0$since$W\omega$ is already a cocycle (Stiefel-Whitney classes are pulled back from the classifying space and are absolute, not relative). Therefore $$ \langle M\omega, [M]2 \rangle + \langle N\omega, [N]2 \rangle = 0 \in \mathbb{F}2, $$ i.e. $\mathrm{sw}\omega(M) = \mathrm{sw}\omega(N)$.$\square$

Bridge. The bordism-invariance lemma is the foundational reason that $N_n$ admits a well-defined homomorphism into the dual of the space of Stiefel-Whitney numbers; this is exactly the calculation Thom 1954 promotes to a complete invariant. The argument generalises in two directions: appears again in 03.06.13 (oriented bordism, where the Pontryagin numbers play the role of Stiefel-Whitney numbers and the relevant ring is ${\Omega }_{\ast }^{SO}$ tensored with $\mathbb{Q}$), and builds toward the Pontryagin-Thom construction, which identifies the entire bordism ring with the homotopy groups of the Thom spectrum $MO$. The central insight is that any characteristic-class evaluation against the fundamental class is automatically a bordism invariant, because the pairing kills boundary classes — a stably abelian shadow of Stokes' theorem.

Exercises [Intermediate+]

Lean formalization [Intermediate+]

lean_status: partial — the module Codex.Modern.CharClasses.UnorientedBordismThom declares the central definitions and the bordism-invariance lemma as theorems with sorry-bodies, pending Mathlib smooth-manifold and Thom-spectrum infrastructure.

-- File: lean/Codex/Modern/CharClasses/UnorientedBordismThom.lean

structure ClosedSmoothManifold (n : ℕ) where
  carrier : Type
  isClosed : True

def IsBordant {n : ℕ} (_M _N : ClosedSmoothManifold n) : Prop := True

theorem isBordant_refl  : ∀ {n} M, IsBordant (n := n) M M := …
theorem isBordant_symm  : …
theorem isBordant_trans : …

def UnorientedBordismGroup (n : ℕ) : Type := Unit
instance : AddCommGroup (UnorientedBordismGroup n) := …

-- Central lemma: bordism invariance of Stiefel-Whitney numbers.
theorem sw_number_bordism_invariant {n : ℕ}
    {M N : ClosedSmoothManifold n}
    (_h : IsBordant M N) (ω : List ℕ) :
    stiefelWhitneyNumber M ω = stiefelWhitneyNumber N ω := …

-- Thom's main theorem (statement).
theorem thom_main_theorem (n : ℕ) :
    UnorientedBordismGroup n ≃+ MOHomotopy n := …

The full proof bodies depend on three Mathlib gaps: (1) smooth manifolds with boundary plus the collaring theorem; (2) the Stiefel-Whitney class API on real vector bundles; (3) the stable-homotopy category and the Thom-spectrum construction. The Lean module is shipped as a target API, not a finished formalisation.

Advanced results [Master]

Bordism as equivalence relation and the ring $N_{\ast }$

Pontryagin (1947) introduced the bordism relation for closed smooth manifolds in the course of his work on framed cobordism and homotopy groups of spheres ^[Pontryagin]. The unoriented version $N_{\ast }$ — where no orientation, framing, or stable structure is imposed — was singled out by Thom in 1954 ^[Thom] as the first cobordism theory admitting a complete computation.

The group operation $[M]+[N]=[M\bigsqcup N]$ is well-defined because disjoint union respects boundaries: $(M\bigsqcup N)\bigsqcup (M^{\prime }\bigsqcup N^{\prime })=\partial (W\bigsqcup W^{\prime })$ whenever $M\bigsqcup N=\partial W$. The ring structure $[M]\cdot [N]=[M\times N]$ is well-defined for the same reason combined with $\partial (W\times N)=\partial W\times N=(M\bigsqcup M^{\prime })\times N$. Graded-commutativity holds without sign because everything is in characteristic $2$. The identity element of the ring is $[\mathrm{pt}]\in N_0={\mathbb{F}}_2$, since $\mathrm{pt}\times M=M$.

The first low-dimensional values, as computed in Milnor-Stasheff §17–§18 ^{[Milnor-Stasheff]}: $$ N_0 = \mathbb{F}_2,\quad N_1 = 0,\quad N_2 = \mathbb{F}_2 \langle [RP^2] \rangle,\quad N_3 = 0, $$ $$ N_4 = \mathbb{F}_2 \langle [RP^4],\ [RP^2 \times RP^2] \rangle,\quad N_5 = \mathbb{F}_2 \langle [x_5] \rangle, $$ where $x_5$ is the Dold manifold $P(1,2)=(S^1\times \mathbb{C}{P}^2)/\tau$ with $\tau$ the antipodal-and-conjugation involution. The pattern is governed by the polynomial-generator theorem stated below.

Thom space MO and the Pontryagin-Thom construction

Let ${\gamma }^k\to BO(k)$ be the universal rank-$k$ real vector bundle. Its Thom space is the one-point compactification of the total space, equivalently $\mathrm{Th}({\gamma }^k):=D({\gamma }^k)/S({\gamma }^k)$ where $D,S$ are the closed disk and sphere bundles. Define $MO(k):=\mathrm{Th}({\gamma }^k)$. The structure maps $\Sigma MO(k)\to MO(k+1)$ — coming from the inclusion ${\gamma }^k\oplus \underline{\mathbb{R}}\hookrightarrow {\gamma }^{k+1}{|}_{BO(k)}$ — assemble these spaces into a spectrum, the unoriented Thom spectrum $MO$ ^{[Milnor-Stasheff §17]}.

The Pontryagin-Thom construction takes a closed smooth $n$-manifold $M$, embeds it into a sphere $S^{n+k}$ for $k$ large (Whitney's embedding theorem), takes a tubular neighbourhood ${\nu }_M\cong U\subset S^{n+k}$, and forms the collapse map $$ c_M : S^{n+k} \longrightarrow \mathrm{Th}(\nu_M) $$ sending the disk bundle of ${\nu }_M$ identically and crushing the complement of $U$ to the base point. Composing $c_M$ with the classifying map $\mathrm{Th}({\nu }_M)\to MO(k)$ — built from a bundle map ${\nu }_M\to {\gamma }^k$ — yields a map $S^{n+k}\to MO(k)$, i.e. an element of ${\pi }_{n+k}(MO(k))$. Passing to the stable colimit gives the Thom homomorphism $$ \Phi : N_n \longrightarrow \pi_n(MO). $$

Three properties of $\Phi$ require checking, all standard from Milnor-Stasheff §17 ^{[Milnor-Stasheff]}:

(i) Well-definedness. Two embeddings $f,g:M\hookrightarrow S^{n+k}$ become regularly homotopic for $k$ large, and the regular-homotopy classes match because the tubular neighbourhoods are unique up to isotopy.

(ii) Boundary kills class. If $M=\partial W$, the embedding $W\hookrightarrow D^{n+k+1}$ extending $M\hookrightarrow S^{n+k}$ produces a null-homotopy of the collapse map.

(iii) Multiplicativity. The product $M\times N$ embeds in $S^{m+n+k+\ell }=S^{m+k}\wedge S^{n+\ell }$, and the collapse map factors as $c_M\wedge c_N$, recovering the smash-product multiplication on the spectrum side.

Thom's main theorem $N_{\ast }\cong {\pi }_{\ast }(MO)$

Theorem (Thom 1954, Théorème IV.8). The Thom homomorphism $\Phi : N_ \to \pi_*(MO)$isanisomorphismofgraded$\mathbb{F}_2$-algebras.*

Injectivity comes from transversality: a null-homotopy of the collapse map produces, by transversality with the zero section of ${\gamma }^k$, a bordism $W$ between $M$ and the empty manifold. Surjectivity uses the same transversality argument applied to an arbitrary stable map $S^{n+k}\to MO(k)$: making the map transverse to $BO(k)\hookrightarrow MO(k)$ produces a closed manifold of dimension $n$ whose Pontryagin-Thom class is the given homotopy class.

The right-hand side ${\pi }_{\ast }(MO)$ is then computed via the Adams spectral sequence $$ E_2^{s,t} = \mathrm{Ext}_{\mathcal{A}}^{s,t}(H^(MO; \mathbb{F}_2), \mathbb{F}2) \Rightarrow \pi{t-s}(MO), $$ where $\mathcal{A}$ is the mod-2 Steenrod algebra. Thom's computation: $H^(MO; \mathbb{F}2)$isafree$\mathcal{A}$-modulewithgeneratorsindexedbynon-dyadicpartitions$\omega = (i_1, \ldots, i_r)$—sequencesofpositiveintegersnoneofwhichhastheform$2^s - 1$.Byfreenessthe$E_2$-pageconcentratesonthe$s = 0$ row, the spectral sequence collapses, and $$ \pi(MO) \cong \mathbb{F}_2[a_n : n \geq 2,\ n + 1 \neq 2^k] $$ as a graded ${\mathbb{F}}_2$-algebra, with $|a_n|=n$. Combined with $\Phi$, this is the polynomial-generator theorem for $N_$ ^[Stong].

Computation of $N_{\ast }$ via Stiefel-Whitney numbers and low-dimensional manifolds

The pairing $$ \mathrm{sw} : N_n \longrightarrow \mathrm{Hom}_{\mathbb{F}_2}(\mathcal{P}_n, \mathbb{F}2),\qquad [M] \longmapsto (\omega \mapsto \mathrm{sw}\omega(M)), $$ where ${\mathcal{P}}_n$ is the ${\mathbb{F}}_2$-span of partitions of $n$, is injective by Thom's bordism-invariance theorem combined with the dual computation that $H^{\ast }(MO;{\mathbb{F}}_2)$ is generated by Stiefel-Whitney classes of the universal bundle. The image has dimension equal to the number of polynomial monomials of total weight $n$ in the generators $x_m$, $m\ge 2$, $m+1\ne 2^k$. The image is not full: the Wu formula constrains which combinations of Stiefel-Whitney numbers occur for closed smooth manifolds (this is the content of 03.06.10 Stiefel-Whitney numbers and the Wu formula, sibling unit shipped in the same cycle).

Explicit calculations (Milnor-Stasheff §17):

• $N_0\cong {\mathbb{F}}_2$ with basis $[\mathrm{pt}]$.
• $N_1=0$: every closed smooth curve is $S^1\bigsqcup \cdots \bigsqcup S^1$, each $S^1=\partial D^2$.
• $N_2\cong {\mathbb{F}}_2$ with basis $[R{P}^2]$: closed surfaces are classified by genus and orientability; the Stiefel-Whitney number $w_1^2=\chi \mathrm{mod}2$ separates $[R{P}^2]$ from $0$.
• $N_3=0$: Rokhlin's theorem in unoriented form; every closed 3-manifold bounds a 4-manifold (proved geometrically via surgery; algebraically because $3+1=4=2^2$ is a power of $2$).
• $N_4\cong {\mathbb{F}}_2^2$ with basis $[R{P}^4],[R{P}^2\times R{P}^2]$.
• $N_5\cong {\mathbb{F}}_2$, generated by a Dold manifold $P(1,2)$ since $5+1=6$ is not a power of $2$.
• $N_n=0$ exactly when $n=1,3,7,15,31,\dots$ (Mersenne dimensions).

The Dold manifolds $P(m,n):=(S^m\times \mathbb{C}{P}^n)/\mathbb{Z}/2$, with $\mathbb{Z}/2$ acting by the antipodal map on $S^m$ and complex conjugation on $\mathbb{C}{P}^n$, were constructed by Dold 1956 ^[Dold] specifically to fill the odd-dimensional generators that real projective spaces alone cannot supply. They have dimension $m+2n$ and bordism class $[P(m,n)]$ realising the generator in any admissible dimension where $RP$ alone fails.

Synthesis. Thom's theorem is the foundational reason that bordism, originally a geometric equivalence relation invented by Pontryagin to compute homotopy groups of spheres, is exactly the homotopy of a Thom spectrum. The central insight is the Pontryagin-Thom collapse: a manifold embedded with normal bundle of dimension $k$ becomes a map from a high sphere into $MO(k)$, and the entire bordism class is captured by the stable homotopy class of that map. Putting these together with Thom's Steenrod-algebra computation, the unoriented bordism ring is identified with the polynomial ${\mathbb{F}}_2$-algebra on non-Mersenne-dimensional generators, and the bordism relation reduces to equality of finitely many Stiefel-Whitney numbers — this is exactly the route Wall 1960 ^[Wall] generalises to oriented bordism and Atiyah 1961 ^[Atiyah] generalises to the modern $K$-theoretic reformulation.

The bridge is between manifold topology and stable homotopy: the bordism invariance of characteristic numbers (Stokes in mod-2 cohomology), the Pontryagin-Thom collapse (transversality + tubular neighbourhood), and the Steenrod-algebra computation (Adams spectral sequence with collapsing $E_2$). This pattern recurs in every cobordism theory: oriented ${\Omega }_{\ast }^{SO}$ via $MSO$, complex ${\Omega }_{\ast }^U$ via $MU$, spin ${\Omega }_{\ast }^{\mathrm{Spin}}$ via $M\mathrm{Spin}$ — each computed by the same machine with a different universal bundle. The Stiefel-Whitney-number invariants generalise to Pontryagin numbers, Chern numbers, and $\hat{A}$-genera in the respective theories.

Full proof set [Master]

Proposition (Bordism is an equivalence relation). The relation $M\sim N$ defined by the existence of $W$ with $\partial W=M\bigsqcup N$ is reflexive, symmetric, and transitive on closed smooth $n$-manifolds.

Proof. Reflexivity: $W=M\times [0,1]$ has $\partial W=M\bigsqcup M=M\bigsqcup M$. Symmetry: the witness $W$ is the same regardless of the order of $M$ and $N$ in the disjoint-union decomposition of $\partial W$. Transitivity: given $W_1$ with $\partial W_1=M\bigsqcup N$ and $W_2$ with $\partial W_2=N\bigsqcup P$, glue along the common boundary component $N$ using the collaring theorem (Milnor 1965 Topology from the differentiable viewpoint). The collaring theorem provides open collars $N\times [0,ϵ)\subset W_1$ and $N\times [0,ϵ)\subset W_2$ with the boundary embedded at $t=0$ in each; identifying these collars produces a smooth structure on $W_1{\cup }_N{W}_2$ whose boundary is $M\bigsqcup P$. $\square$

Proposition (Each element of $N_n$ has order $\le 2$). For every closed smooth $n$-manifold $M$, $2[M]=0$ in $N_n$.

Proof. The cylinder $W=M\times [0,1]$ has boundary $\partial W=M\bigsqcup M$. Therefore $[M]+[M]=[M\bigsqcup M]=[\partial W]=0$ in $N_n$. $\square$

Proposition (Bordism invariance of Stiefel-Whitney numbers). Established in the Key Theorem section.

Proposition (Multiplicativity of the bordism ring). The product $[M]\cdot [N]:=[M\times N]$ defines a well-defined commutative graded ring structure on $N_$withunit$[\mathrm{pt}] \in N_0$.*

Proof. Well-definedness: if $M^{\prime }\sim M$ via $W$, then $W\times N$ is a cobordism between $M\times N$ and $M^{\prime }\times N$ — the boundary $\partial (W\times N)=\partial W\times N=(M\bigsqcup M^{\prime })\times N=(M\times N)\bigsqcup (M^{\prime }\times N)$, using that $N$ is closed so $\partial (W\times N)=(\partial W)\times N$. Commutativity: the diffeomorphism $M\times N\cong N\times M$ swaps factors. Associativity: $(M\times N)\times P\cong M\times (N\times P)$. Distributivity: $M\times (N\bigsqcup N^{\prime })=(M\times N)\bigsqcup (M\times N^{\prime })$ at the level of manifolds. The unit follows from $\mathrm{pt}\times M=M$ as smooth manifolds. $\square$

Proposition (Pontryagin-Thom collapse is well-defined modulo bordism). Let $f_0,f_1:M\hookrightarrow S^{n+k}$ be two embeddings of a closed smooth $n$-manifold. For $k$ large enough, the collapse maps $c_{f_0},c_{f_1}:S^{n+k}\to MO(k)$ are stably homotopic. More generally, if $W$ is a compact $(n+1)$-manifold with $\partial W=M_0\bigsqcup M_1$ and $F:W\hookrightarrow S^{n+k}\times [0,1]$ extends $f_0\bigsqcup f_1$ on the boundary, then the collapse maps differ by a homotopy.

Proof. The relative tubular-neighbourhood theorem gives an open tubular neighbourhood $V\subset S^{n+k}\times [0,1]$ of $F(W)$ diffeomorphic to the normal bundle ${\nu }_F$, and this restricts to a tubular neighbourhood of $f_i(M_i)$ on the slice $S^{n+k}\times \{i\}$. The relative collapse map $$ S^{n+k} \times [0,1] \longrightarrow \mathrm{Th}(\nu_F) $$ quotients the complement of $V$ to the base point; composing with the classifying map $\mathrm{Th}({\nu }_F)\to MO(k)$ produces a homotopy from $c_{f_0}$ to $c_{f_1}$ in the slice direction. Two embeddings of the same $M$ are connected by the cylinder bordism $W=M\times [0,1]$, so any two collapse maps for $M$ are homotopic. $\square$

Proposition (Surjectivity of $\Phi$ via transversality). Every class $[\varphi ]\in {\pi }_{n+k}(MO(k))$ is the Pontryagin-Thom class of some closed smooth $n$-manifold $M$.

Proof. Choose a representative $\varphi :S^{n+k}\to MO(k)=\mathrm{Th}({\gamma }^k)$. By the transversality theorem (Thom's transversality version, available since the target is smooth away from the base point), perturb $\varphi$ to a smooth map transverse to the zero section $BO(k)\hookrightarrow \mathrm{Th}({\gamma }^k)$. The preimage $$ M := \phi^{-1}(BO(k)) \subset S^{n+k} $$ is a closed smooth submanifold of dimension $(n+k)-k=n$, with normal bundle ${\nu }_M={\varphi }^{\ast }{\gamma }^k$. The Pontryagin-Thom construction applied to the embedding $M\hookrightarrow S^{n+k}$ with this normal bundle recovers the original class $[\varphi ]$. Different transverse perturbations yield bordant manifolds (parametrised transversality applied to the cylinder $S^{n+k}\times [0,1]$), so $\Phi ([M])$ is uniquely defined as $[\varphi ]$. $\square$

Proposition (Computation of $N_n$ for $n\le 4$ from polynomial generators). With basis indexed by monomials in $x_2,x_4$ of total degree $n$: ${\dim }_{{\mathbb{F}}_2}{N}_n=1,0,1,0,2$ for $n=0,1,2,3,4$ respectively.

Proof. The polynomial-generator theorem ^{[Thom; Milnor-Stasheff §18]} states $N_{\ast }={\mathbb{F}}_2[x_2,x_4,x_5,x_6,x_8,\dots ]$. Restricting to degree $n\le 4$, the only generators that can contribute are $x_2$ (degree 2) and $x_4$ (degree 4). The monomials are:

• Degree 0: $\{1\}$, dimension 1.
• Degree 1: $\varnothing$, dimension 0.
• Degree 2: $\{x_2\}$, dimension 1.
• Degree 3: $\varnothing$, dimension 0.
• Degree 4: $\{x_2^2,x_4\}$, dimension 2.

Realising the generators: $x_2=[R{P}^2]$ and $x_4=[R{P}^4]$ as established in the Advanced results section. $\square$

Connections [Master]

• Stiefel-Whitney classes 03.06.03. The bordism-invariance argument depends entirely on the Whitney sum behaviour and naturality of $w_k$. Stiefel-Whitney classes provide the cohomological invariants whose pairing against the fundamental class produces the bordism-detecting numbers; this unit shows those numbers are complete invariants of the bordism class.

• Stiefel-Whitney numbers and the Wu formula 03.06.10. Sibling unit shipping in the same cycle (mark pending). The Wu formula expresses Stiefel-Whitney numbers of $TM$ purely in terms of cohomology operations on $H^{\ast }(M;{\mathbb{F}}_2)$ via Steenrod squares, and constrains which collections of Stiefel-Whitney numbers can be realised by closed smooth manifolds. Together, 03.06.10 and the present unit identify both the full set of bordism invariants and the constraints they satisfy.

• Oriented bordism and Pontryagin-Thom 03.06.13. Generalises in: replacing $BO$ by $BSO$ produces the oriented Thom spectrum $MSO$, whose homotopy ring rationally equals $\mathbb{Q}[\mathbb{C}{P}^2,\mathbb{C}{P}^4,\dots ]$ — Wall 1960 computed the integral version. Pontryagin numbers play the role Stiefel-Whitney numbers play here. The orientation-forgetting map ${\Omega }_{\ast }^{SO}\to {\Omega }_{\ast }^O=N_{\ast }$ is the structural bridge: oriented bordism is the enriched unoriented theory, and the Pontryagin-Thom collapse of 03.06.13 reduces to the present unit's $MO$-construction after forgetting orientation. Anchor phrase: oriented bordism enriches unoriented bordism via ${\Omega }^{SO}\to {\Omega }^O$.

• Hirzebruch signature theorem 03.06.11. The downstream specialisation in: once the rational structure of oriented bordism is known, the signature is exposed as a particular ring homomorphism ${\Omega }_{\ast }^{SO}\otimes \mathbb{Q}\to \mathbb{Q}$, computed via the $L$-genus, which is the canonical multiplicative sequence in Pontryagin classes.

• Multiplicative sequences and the $L$-, $\hat{A}$-, $\mathrm{Td}$-genera 03.06.15. The chapter-closing synthesis appears in: Hirzebruch's multiplicative-sequence formalism reinterprets ring homomorphisms out of ${\Omega }_{\ast }^{SO}$ as formal power series in Pontryagin classes, and the present unit's $N_{\ast }\to {\mathbb{F}}_2$ characters are the mod-2 analogues — homomorphisms $N_{\ast }\to {\mathbb{F}}_2$ classified by Stiefel-Whitney numbers.

• Chern-Weil homomorphism 03.06.06. Provides the prerequisite framework for: characteristic-class representatives, which combined with the present unit's bordism invariance produce all the closed-form integral formulae for cobordism characters (e.g., the integral of $\hat{A}(p_{\ast })$ against the fundamental class is the spin-bordism signature in the parallel theory).

• Complex vector bundle 03.05.08. Builds toward: complex cobordism ${\Omega }_{\ast }^U$ via $MU$ is the universal complex-oriented cohomology theory; Quillen 1969 identified its formal group law with the universal formal group law, making the present unit's $N_{\ast }$-computation a degenerate ${\mathbb{F}}_2$-shadow of a much richer additive structure visible in chromatic stable homotopy theory.

Historical & philosophical context [Master]

Pontryagin 1950 ^[Pontryagin] introduced the bordism relation as a tool for computing homotopy groups of spheres: he set up the Pontryagin-Thom collapse for framed manifolds and used it to compute ${\pi }_{n+1}(S^n)$ and ${\pi }_{n+2}(S^n)$ stably. Thom 1954 ^[Thom], in his Comment. Math. Helv. paper Quelques propriétés globales des variétés différentiables, abstracted the construction to arbitrary structure groups (orthogonal, special-orthogonal, unitary), introduced the Thom spectra $MO,MSO,MU$, and gave the first complete computation of any cobordism ring — the unoriented case $N_{\ast }$ — as a polynomial ${\mathbb{F}}_2$-algebra. The Steenrod-algebra calculation of $H^{\ast }(MO;{\mathbb{F}}_2)$ as a free $\mathcal{A}$-module on non-dyadic generators is Thom's central technical innovation; combined with the Adams spectral-sequence collapse it yields the closed-form polynomial generators.

Milnor-Stasheff Characteristic Classes ^{[Milnor-Stasheff]} (Princeton, 1974) gave the modern textbook exposition, splitting Thom's theorem into the Pontryagin-Thom construction (§17) and the algebraic computation (§18). Dold 1956 ^[Dold] supplied explicit polynomial generators in odd admissible dimensions where projective spaces fail (Dold manifolds). Stong's Notes on Cobordism Theory ^[Stong] (Princeton, 1968) remains the canonical reference for the full generalised-cobordism program.

The lineage continues in two directions. Wall 1960 ^[Wall] computed the oriented cobordism ring, leveraging Thom's framework with $BSO$ replacing $BO$. Atiyah 1961 ^[Atiyah] reformulated cobordism as a generalised cohomology theory $M{O}^{\ast }(X)$ on topological spaces, making the spectrum-level structure first-class and opening the door to the chromatic stable-homotopy program of Quillen, Morava, and Ravenel. Bordism remains the geometric prototype for every modern generalised cohomology theory; its computation by Thom is the first substantive example of the Adams spectral sequence collapsing, and the polynomial-generator answer is the historical model for every subsequent cobordism computation.

Bibliography [Master]

@article{Thom1954,
  author  = {Thom, Ren\'e},
  title   = {Quelques propri\'et\'es globales des vari\'et\'es diff\'erentiables},
  journal = {Commentarii Mathematici Helvetici},
  volume  = {28},
  year    = {1954},
  pages   = {17--86},
}

@book{MilnorStasheff1974,
  author    = {Milnor, John W. and Stasheff, James D.},
  title     = {Characteristic Classes},
  publisher = {Princeton University Press},
  year      = {1974},
  series    = {Annals of Mathematics Studies},
  number    = {76},
  note      = {Pontryagin-Thom construction in \S17, computation of $N_*$ in \S18.},
}

@book{Stong1968,
  author    = {Stong, Robert E.},
  title     = {Notes on Cobordism Theory},
  publisher = {Princeton University Press},
  year      = {1968},
  series    = {Mathematical Notes},
}

@article{Pontryagin1950,
  author  = {Pontryagin, L. S.},
  title   = {Smooth manifolds and their applications in homotopy theory},
  journal = {Trudy Mat. Inst. Steklov},
  volume  = {45},
  year    = {1955},
  note    = {Russian original 1950; English translation: AMS Translations Series 2, vol. 11 (1959).},
}

@article{Dold1956,
  author  = {Dold, Albrecht},
  title   = {Erzeugende der Thomschen Algebra $\mathfrak{N}$},
  journal = {Mathematische Zeitschrift},
  volume  = {65},
  year    = {1956},
  pages   = {25--35},
}

@article{Wall1960,
  author  = {Wall, C. T. C.},
  title   = {Determination of the cobordism ring},
  journal = {Annals of Mathematics},
  volume  = {72},
  year    = {1960},
  pages   = {292--311},
}

@article{Atiyah1961,
  author  = {Atiyah, M. F.},
  title   = {Bordism and cobordism},
  journal = {Proc. Cambridge Philos. Soc.},
  volume  = {57},
  year    = {1961},
  pages   = {200--208},
}

@book{Milnor1965,
  author    = {Milnor, John W.},
  title     = {Topology from the Differentiable Viewpoint},
  publisher = {University Press of Virginia},
  year      = {1965},
  note      = {Pontryagin-Thom for framed cobordism in geometric form; collaring theorem.},
}