Spin-Statistics and the CPT Theorem

Nonrelativistic quantum mechanics hands you the connection between spin and statistics as a gift you never earn: electrons are declared fermions, photons bosons, and the Pauli principle is written down as a postulate. It works, but it explains nothing. Why must a particle of half-integer spin refuse to share a state? The relativistic framework you have spent this chapter building — the Klein-Gordon and Dirac equations, negative-energy solutions, charge conjugation, the light cone — finally answers the question. Insist on three things any decent theory must have, and the answer is forced. The three demands are Lorentz invariance, locality (nothing propagates outside the light cone), and a ground state whose energy is bounded below. Under those constraints the statistics of a field stops being a free choice: quantize a spin-0 field with anticommutators and its fields fail to commute at spacelike separation — a signal outside the light cone; quantize a spin-½ field with commutators and its energy tumbles to −∞-\infty with every antiparticle you create. The only consistent choice in a local relativistic quantum field theory with a stable vacuum is Bose statistics for integer-spin fields and Fermi for half-integer. This is the spin-statistics theorem, and its logic is the load-bearing beam under the periodic table, the stability of matter, and the pressure that holds up a white dwarf. Bound to it is the CPT theorem: the combined operation of charge conjugation, parity, and time reversal is an exact symmetry of any local, Lorentz-invariant, Hermitian field theory — even when C, P, and CP shatter, as they do in the weak interaction. CPT welds particle to antiparticle: identical mass, identical lifetime, opposite charge. It is tested to a part in 101810^{18} and has never failed. This is the last article of the track, and it is the right one to end on: everything from the infinite square well to the Dirac sea converges here, into two theorems that tell you the quantum world could not have been built any other way.

Key takeaways
  • Lorentz invariance + locality (microcausality) + a stable vacuum together force the spin-statistics connection: integer spin → bosons (commutators), half-integer spin → fermions (anticommutators).
  • The proof is a contradiction for the wrong choice: a scalar quantized with anticommutators has {ϕ(x),ϕ(y)}≠0\{\phi(x),\phi(y)\}\neq0 at spacelike separation (acausal); a spinor quantized with commutators has energy unbounded below (no vacuum).
  • The exchange sign is (−1)2s(-1)^{2s}: s=0⇒+1s=0\Rightarrow+1 (symmetric, Bose), s=12⇒−1s=\tfrac12\Rightarrow-1 (antisymmetric, Fermi) — the Pauli exclusion principle read off the spinor's Lorentz structure.
  • The CPT theorem: any local, Lorentz-invariant, Hermitian field theory is invariant under the antiunitary product Θ=CPT\Theta=CPT, even when C, P, T fail individually.
  • CPT forces particle and antiparticle to share exactly equal mass and lifetime; the neutral-kaon system bounds any violation to ∣mK0−mKˉ0∣/m<10−18|m_{K^0}-m_{\bar K^0}|/m<10^{-18}.
  • Both theorems rest on the same three assumptions — the deepest reason they are among the most robust predictions in all of physics.
  • Key idea: relativity plus locality plus a stable vacuum simultaneously fix statistics from spin and guarantee CPT invariance — the quantum world's two non-negotiable structural laws.
The core insight

Lorentz invariance, locality, and a bounded-below energy together force integer-spin fields to be bosons and half-integer-spin fields to be fermions (spin-statistics), and make every such theory invariant under the combined CPT operation — even when C, P, and T are separately broken.

Where this lives today

These two theorems are still doing frontline work. On the spin-statistics side, the discovery and control of anyons in two-dimensional systems — quasiparticles with fractional exchange statistics 0<θ<π0<\theta<\pi — is one of the most active areas in condensed-matter physics, because the theorem's escape route (lower dimensionality) is exactly what topological quantum computers hope to exploit: braiding non-Abelian anyons is a fault-tolerant way to process quantum information. On the CPT side, entire experimental programs exist to look for the tiniest crack. The ALPHA and BASE collaborations at CERN trap antihydrogen and antiprotons to compare their spectra and magnetic moments with matter to parts in 10910^{9}–101210^{12}; neutral-meson factories push the kaon and BB-meson mass and lifetime comparisons ever tighter. A confirmed CPT violation would be a discovery of the first magnitude — direct evidence that Lorentz invariance or locality fails, a doorway to physics beyond the Standard Model. That the search keeps coming up empty is a measure of how deep these two theorems run.

The physical picture

Picture a quantum field as a continuum of oscillators, one per mode p\mathbf p. You must choose how to quantize each: with commutators (a boson, whose mode can hold any number of quanta) or anticommutators (a fermion, whose mode holds zero or one). Nonrelativistically that choice is yours to make. Relativity takes it away. The theorem says: for each spin, exactly one choice survives, and the other choice does not merely give a different theory — it gives a broken one, where either energy has no floor or signals leap outside the light cone. The hinge is microcausality, and the tool is a single Lorentz-invariant function. When you (anti)commute two fields you get, up to a sign, the same object: the Pauli-Jordan function Δ(x−y)\Delta(x-y), an odd solution of the Klein-Gordon equation that vanishes outside the light cone. For a scalar, only the commutator is proportional to Δ\Delta (odd, vanishing); the anticommutator is a sum of Wightman functions (even, nonvanishing). For a spinor the extra minus sign hidden in the spinor structure flips this — only the anticommutator vanishes at spacelike separation. The Lorentz transformation properties of the field determine which bilinear can satisfy microcausality, thereby fixing the allowed statistics.

Figure
Wrong statistics breaks the theory in two different ways
Two forks, two disasters — only the diagonal survives spin 0 (scalar) spin ½ (spinor) commutators ✓ [φ(x),φ(y)] = iΔ(x−y) = 0 for spacelike x−y causal, H ≥ 0 → BOSE commutators ✗ H = Σ E(b†b − d d†) antiparticles carry −E E → −∞, no vacuum anticommutators ✗ {φ(x),φ(y)} = Δ₊(x−y)+Δ₊(y−x) even sum, ≠ 0 spacelike acausal, signals leak anticommutators ✓ {b,b†}, {d,d†} → H ≥ 0 exclusion (b†)² = 0 causal, H ≥ 0 → FERMI

CPT can be pictured heuristically as a total spacetime reflection combined with charge conjugation. A rotation by an imaginary angle, available in any Lorentz-invariant theory, continues a spatial rotation into a full reversal of one space axis and time; combined with the particle↔antiparticle exchange of charge conjugation it becomes the total spacetime reflection x→−xx\to-x. Because that hidden rotation lives inside the Lorentz group itself, no local relativistic theory can escape it. The slogan that makes it vivid is the Feynman-Stückelberg reading: a particle moving forward in time is an antiparticle moving backward. CPT makes that exact, welding a particle to its antiparticle with identical mass and lifetime.

Figure
CPT as the total spacetime reflection x → −x welding particle to antiparticle
Θ = CPT : Ô(x) → Ô†(−x), [H, Θ] = 0 ⇒ m = m̄, τ = τ̄ t x particle, forward in t = antiparticle, backward in t CPT forces: • mass m(particle) = m(antiparticle) • lifetime τ = τ̄ • |charge|, |μ| equal, opposite sign tested on K⁰: |m(K⁰)−m(K̄⁰)|/m under 10⁻¹⁸ the most precise mass equality in physics

The core equations

Everything below is a free-field statement: the fields are expanded in creation and annihilation operators, and the whole theorem is read off two computed quantities — the energy and the spacelike (anti)commutator. We work in natural units ℏ=c=1\hbar=c=1 with the mostly-minus metric x2=t2−x2x^2=t^2-\mathbf x^2 (so x2<0x^2<0 is spacelike), and the on-shell energy is Ep=p2+m2E_{\mathbf p}=\sqrt{\mathbf p^2+m^2}. Start with the free real scalar field, the spin-0 case. It is a superposition of plane-wave modes, each with an annihilation and a creation operator.

ϕ(x)=∫d3p(2π)3 2Ep(ap e−ip⋅x+ap† e+ip⋅x)(1)\phi(x)=\int\frac{\mathrm d^3p}{(2\pi)^3\,2E_{\mathbf p}}\Big(a_{\mathbf p}\,e^{-ip\cdot x}+a^{\dagger}_{\mathbf p}\,e^{+ip\cdot x}\Big) \qquad (1)

where: ap,ap†a_{\mathbf p},a^{\dagger}_{\mathbf p} — annihilate / create a spin-0 quantum of momentum p\mathbf p; p⋅x=Ept−p⋅xp\cdot x=E_{\mathbf p}t-\mathbf p\cdot\mathbf x is the Lorentz-invariant phase. Reading it physically: the field is Hermitian (ϕ=ϕ†\phi=\phi^\dagger) because a real scalar is its own antiparticle; the measure d3p/2Ep\mathrm d^3p/2E_{\mathbf p} is the Lorentz-invariant one, which is what makes every statement below frame-independent. The correct spin-0 quantization uses commutators; the correct spin-½ quantization uses anticommutators. Pin both, because the theorem is precisely the statement that swapping them breaks the theory.

[ap,aq†]=(2π)3 2Ep δ3(p−q),{bpr,bqs†}=(2π)3 δrs δ3(p−q)(2)[a_{\mathbf p},a^{\dagger}_{\mathbf q}]=(2\pi)^3\,2E_{\mathbf p}\,\delta^3(\mathbf p-\mathbf q),\qquad \{b^{r}_{\mathbf p},b^{s\dagger}_{\mathbf q}\}=(2\pi)^3\,\delta^{rs}\,\delta^3(\mathbf p-\mathbf q) \qquad (2)

where: the left relation is Bose (spin 0); the right is Fermi (spin ½, with r,sr,s the spin labels). Reading it physically: the anticommutator immediately gives (bpr†)2=12{bpr†,bpr†}=0(b^{r\dagger}_{\mathbf p})^2=\tfrac12\{b^{r\dagger}_{\mathbf p},b^{r\dagger}_{\mathbf p}\}=0 — you cannot create two quanta in the same mode. That single line is the Pauli exclusion principle, whose antisymmetric-state and Slater-determinant readings are developed in Pauli Exclusion Principle; here it falls out of the algebra. The whole causal content of the theory lives in one Lorentz-invariant function, the Pauli-Jordan function. It is the odd combination of the two positive-frequency Wightman functions.

Δ(x)=−i∫d3p(2π)3 2Ep(e−ip⋅x−e+ip⋅x),Δ(x)=0  for  x2<0(3)\Delta(x)=-i\int\frac{\mathrm d^3p}{(2\pi)^3\,2E_{\mathbf p}}\Big(e^{-ip\cdot x}-e^{+ip\cdot x}\Big),\qquad \Delta(x)=0\ \ \text{for}\ \ x^2<0 \qquad (3)

where: Δ(x)\Delta(x) is an odd (Δ(−x)=−Δ(x)\Delta(-x)=-\Delta(x)) solution of the Klein-Gordon equation. Reading it physically: Δ\Delta vanishing at spacelike separation is microcausality — the statement that measurements outside each other's light cones cannot interfere. For a free scalar field, the commutator satisfies [ϕ(x),ϕ(y)]=iΔ(x−y)[\phi(x),\phi(y)] = i\Delta(x-y), so microcausality follows directly from the vanishing of Δ(x−y)\Delta(x-y) at spacelike separation. The spin-statistics theorem determines which choice of (anti)commutator reproduces this causal structure. For spin-12\tfrac{1}{2} fields, the spinor structure changes which (anti)commutator is compatible with Lorentz covariance and microcausality. The field is built from Dirac spinors uu and vv, with distinct particle and antiparticle operators.

ψ(x)=∑s∫d3p(2π)312Ep(bps us(p) e−ip⋅x+dps† vs(p) e+ip⋅x)(4)\psi(x)=\sum_{s}\int\frac{\mathrm d^3p}{(2\pi)^3}\frac{1}{\sqrt{2E_{\mathbf p}}}\Big(b^{s}_{\mathbf p}\,u^{s}(p)\,e^{-ip\cdot x}+d^{s\dagger}_{\mathbf p}\,v^{s}(p)\,e^{+ip\cdot x}\Big) \qquad (4)

where: us,vsu^{s},v^{s} are the positive- and negative-energy spinor amplitudes; bpsb^{s}_{\mathbf p} annihilates a particle, dps†d^{s\dagger}_{\mathbf p} creates an antiparticle. Reading it physically: the sixteen bilinears ψˉΓψ\bar\psi\Gamma\psi built from these spinors — the scalar, vector, tensor, axial, pseudoscalar tabulated in Bilinear Covariants and Dirac Adjoint — are exactly the objects C, P, and T act on, which is why the CPT bookkeeping below is a statement about their Lorentz indices.

Derivation AWrong statistics for the scalar violates microcausality

Our goal is the sharp version of the spin-statistics theorem for spin 0: quantizing the Klein-Gordon field with commutators gives a spacelike commutator that vanishes (causal), while anticommutators give one that does not (acausal). The strategy is to compute both the commutator and the anticommutator of two free scalar fields in terms of the positive-frequency function Δ+\Delta_+, and then use the odd/even symmetry of the result together with the spacelike-vanishing of the Pauli-Jordan Δ\Delta. The one subtlety to watch is that Δ+\Delta_+ itself does not vanish at spacelike separation — only its antisymmetric combination does.

Step 1. Define the positive-frequency Wightman function Δ+(x−y)=∫d3p(2π)3 2Ep e−ip⋅(x−y)\Delta_+(x-y)=\int\frac{\mathrm d^3p}{(2\pi)^3\,2E_{\mathbf p}}\,e^{-ip\cdot(x-y)}, the vacuum two-point function ⟨0∣ϕ(x)ϕ(y)∣0⟩\langle0|\phi(x)\phi(y)|0\rangle. Why: inserting the mode expansion (1) into ⟨0∣ϕ(x)ϕ(y)∣0⟩\langle0|\phi(x)\phi(y)|0\rangle and keeping only a a†a\,a^\dagger terms (the vacuum kills the rest) leaves exactly this integral; it is nonzero but falls off outside the light cone.

Step 2. Commutator (the Bose choice): [ϕ(x),ϕ(y)]=Δ+(x−y)−Δ+(y−x)=i Δ(x−y)[\phi(x),\phi(y)]=\Delta_+(x-y)-\Delta_+(y-x)=i\,\Delta(x-y). Why: the commutator subtracts the two orderings, giving the odd combination — precisely the Pauli-Jordan function Δ\Delta of equation (3), which vanishes for (x−y)2<0(x-y)^2<0.

Step 3. Anticommutator (the wrong choice for spin 0): {ϕ(x),ϕ(y)}=Δ+(x−y)+Δ+(y−x)\{\phi(x),\phi(y)\}=\Delta_+(x-y)+\Delta_+(y-x). Why: the anticommutator adds the two orderings, giving the even combination Δ+(x−y)+Δ+(y−x)\Delta_+(x-y)+\Delta_+(y-x); being even and built from nonvanishing pieces, it does not vanish at spacelike separation.

Step 4. Read the verdict. At a spacelike interval the commutator is zero and the anticommutator is a strictly positive number: only the commutator quantization keeps observables from talking outside the light cone. Spin 0 must be a boson. Why: this is the causality test that The Feynman Propagator and Causality built — the vanishing of the invariant commutator function outside the light cone — now used as the selection rule: the even sum fails it, the odd difference passes it. One can check explicitly: evaluating Δ+(x)+Δ+(−x)\Delta_+(x)+\Delta_+(-x) and Δ+(x)−Δ+(−x)\Delta_+(x)-\Delta_+(-x) for a representative spacelike interval on a massive-scalar mode sum confirms that the even sum is nonzero while the odd difference cancels to zero.

Go deeper · Derivation B · 1 minWrong statistics for the spinor gives energy unbounded below
Derivation BWrong statistics for the spinor gives energy unbounded below

Here we prove the complementary half: quantizing the Dirac field with commutators drives the energy to −∞-\infty, while anticommutators restore a positive Hamiltonian. The plan is to mode-expand the Dirac Hamiltonian, isolate the antiparticle term's sign, and contrast what commutators versus anticommutators do to it. The subtlety is that the Dirac Hamiltonian arrives with a relative minus sign between particle and antiparticle contributions before any (anti)commutator is imposed — that pre-existing minus is what the statistics choice must repair.

Step 1. Insert (4) into H=∫d3x ψ†(−iα⋅∇+βm)ψH=\int\mathrm d^3x\,\psi^\dagger(-i\boldsymbol\alpha\cdot\nabla+\beta m)\psi and use spinor orthogonality to get H=∑s∫d3p(2π)3Ep(bps†bps−dpsdps†)H=\sum_s\int\frac{\mathrm d^3p}{(2\pi)^3}E_{\mathbf p}\big(b^{s\dagger}_{\mathbf p}b^{s}_{\mathbf p}-d^{s}_{\mathbf p}d^{s\dagger}_{\mathbf p}\big). Why: the negative-energy spinors vsv^{s} carry the antiparticle term with a minus sign — a direct fingerprint of the negative-energy branch catalogued in Negative-Energy Solutions and the Interpretive Crisis; the whole question is how to make −d d†-d\,d^\dagger positive.

Step 2. Try commutators: [dp,dp†]=(2π)3δ3(0)[d_{\mathbf p},d^{\dagger}_{\mathbf p}]=(2\pi)^3\delta^3(0) gives −dpdp†=−dp†dp−(positive const)-d_{\mathbf p}d^{\dagger}_{\mathbf p}=-d^{\dagger}_{\mathbf p}d_{\mathbf p}-\text{(positive const)}. Why: the antiparticle number operator d†dd^\dagger d now enters HH with a negative coefficient −Ep-E_{\mathbf p}; each antiparticle you create lowers the energy, so HH has no floor and there is no stable vacuum.

Step 3. Try anticommutators: {dp,dp†}=(2π)3δ3(0)\{d_{\mathbf p},d^{\dagger}_{\mathbf p}\}=(2\pi)^3\delta^3(0) gives −dpdp†=+dp†dp−(const)-d_{\mathbf p}d^{\dagger}_{\mathbf p}=+d^{\dagger}_{\mathbf p}d_{\mathbf p}-\text{(const)}. Why: the anticommutator flips the sign, turning −d d†-d\,d^\dagger into +d†d+d^\dagger d; after dropping the infinite constant by normal ordering, the antiparticle term is now positive.

Step 4. Assemble H=∑s∫d3p(2π)3Ep(bps†bps+dps†dps)≥0H=\sum_s\int\frac{\mathrm d^3p}{(2\pi)^3}E_{\mathbf p}\big(b^{s\dagger}_{\mathbf p}b^{s}_{\mathbf p}+d^{s\dagger}_{\mathbf p}d^{s}_{\mathbf p}\big)\ge0. Only Fermi statistics gives a bounded-below energy. Why: this is the positivity test built in Field Quantization of the Klein-Gordon and Dirac Fields — that canonical quantization must produce H≥0H\ge0 — used here as a selection rule that rejects Bose statistics for spin ½. This can be seen concretely on a single mode with a 2×22\times2 operator model: commutators give a −E-E eigenvalue in the spectrum, while anticommutators give +E+E; the anticommutator also directly confirms (b†)2=0(b^\dagger)^2=0.

Go deeper · Derivation C · 1 minCPT invariance from a Lorentz-scalar Hermitian Lagrangian
Derivation CCPT invariance from a Lorentz-scalar Hermitian Lagrangian

Finally the CPT half: any Hermitian, Lorentz-invariant, local Lagrangian density is automatically invariant under the antiunitary product Θ=CPT\Theta=CPT. The route is an index count — each Lorentz index a tensor carries picks up a factor −1-1 under the combined operation, and a scalar has an even number of contracted indices — closed off by using Hermiticity to absorb the antiunitary conjugation and locality to place every factor at the single point −x-x. We do not attempt the rigorous distribution-theoretic proof (that is Jost's theorem, stated below); this is the operator-level bookkeeping that makes the result transparent.

Step 1. Scalar bilinear: Θ ψˉψ(x) Θ−1=ψˉψ(−x)\Theta\,\bar\psi\psi(x)\,\Theta^{-1}=\bar\psi\psi(-x). Why: the Dirac scalar carries zero free Lorentz indices, so it returns to itself at the reflected point −x-x; the complex conjugation forced by the antiunitary Θ\Theta is undone by the term's Hermiticity.

Step 2. Vector bilinear: Θ ψˉγμψ(x) Θ−1=−ψˉγμψ(−x)\Theta\,\bar\psi\gamma^\mu\psi(x)\,\Theta^{-1}=-\bar\psi\gamma^\mu\psi(-x), one index → one factor −1-1. Why: a current couples to a field AμA_\mu that also flips sign under the same operation, so the physical coupling AμψˉγμψA_\mu\bar\psi\gamma^\mu\psi is invariant; the sixteen bilinears of Bilinear Covariants and Dirac Adjoint each transform with their own (−1)(index count)(-1)^{(\text{index count})}, and charge conjugation supplies the particle↔antiparticle exchange developed in Charge Conjugation and the Positron.

Step 3. General tensor: Tμ1⋯μn(x)→(−1)n Tμ1⋯μn(−x)T^{\mu_1\cdots\mu_n}(x)\to(-1)^n\,T^{\mu_1\cdots\mu_n}(-x). Why: each free Lorentz index contributes a factor of −1-1 under spacetime inversion. In a Lorentz scalar all Lorentz indices are completely contracted, so the total number of free indices is even and the overall factor is (−1)n=+1(-1)^n=+1.

Step 4. Conclude Θ L(x) Θ−1=L(−x)\Theta\,\mathcal L(x)\,\Theta^{-1}=\mathcal L(-x), hence Θ S Θ−1=S\Theta\,S\,\Theta^{-1}=S with S=∫d4x LS=\int\mathrm d^4x\,\mathcal L. Why: Hermiticity absorbs the antiunitary conjugation and locality lets every factor sit at −x-x; the action is invariant even when C, P, and T each fail separately — the content of the CPT theorem. The index parity rule (−1)n=+1(-1)^n=+1 iff nn is even holds across all five bilinear types — scalar, vector, tensor, axial, and pseudoscalar — as each carries a definite number of free Lorentz indices.

Limiting cases

caseresultmeaning
spin 0, commutators[ϕ(x),ϕ(y)]=iΔ(x−y)=0[\phi(x),\phi(y)]=i\Delta(x-y)=0 for x2<0x^2<0causal, H≥0H\ge0 — the allowed Bose quantization
spin ½, anticommutators{ψa(x),ψˉb(y)}=(iγμ∂μ+m)abΔ(x−y)\{\psi_a(x),\bar\psi_b(y)\}=(i\gamma^\mu\partial_\mu+m)_{ab}\Delta(x-y)H≥0H\ge0, exclusion (b†)2=0(b^\dagger)^2=0 — the allowed Fermi quantization
wrong statisticsscalar {ϕ,ϕ}≠0\{\phi,\phi\}\neq0 spacelike; spinor [ψ,ψ]⇒E→−∞[\psi,\psi]\Rightarrow E\to-\inftyacausality or no stable vacuum — the theory is broken
2+1 dimensionsψ(x)ψ(y)=eiθψ(y)ψ(x)\psi(x)\psi(y)=e^{i\theta}\psi(y)\psi(x), 0<θ<π0<\theta<\pianyons interpolate Bose↔Fermi; the 3+1 theorem does not apply
Lorentz violation / nonlocalityCPT may breakCPT is exact only with Lorentz invariance + locality; a target of quantum-gravity searches

Worked examples

Example 1 — the exchange sign is (−1)2s(-1)^{2s}. Show that the statistics selected by the theorem is exactly the sign a two-particle state picks up under exchange, and that it is fixed by spin.

Step 1. For a field of spin ss, the theorem assigns commutators (Bose) to integer ss and anticommutators (Fermi) to half-integer ss. The two-quantum state built from creation operators is ∣p,q⟩=cp†cq†∣0⟩|\mathbf p,\mathbf q\rangle=c^\dagger_{\mathbf p}c^\dagger_{\mathbf q}|0\rangle.

Step 2. Swapping the two operators uses the (anti)commutator at different momenta, where the delta-function term is absent: cp†cq†=η cq†cp†c^\dagger_{\mathbf p}c^\dagger_{\mathbf q}=\eta\,c^\dagger_{\mathbf q}c^\dagger_{\mathbf p} with η=+1\eta=+1 for commutators and η=−1\eta=-1 for anticommutators.

Step 3. Match to spin: η=+1\eta=+1 for integer spin, η=−1\eta=-1 for half-integer spin. Both cases are captured by the single formula η=(−1)2s\eta=(-1)^{2s}, since 2s2s is even for integer ss and odd for half-integer ss.

Step 4. Evaluate the anchor cases. For s=0s=0: (−1)0=+1(-1)^{0}=\mathbf{+1} (symmetric, Bose). For s=12s=\tfrac12: (−1)1=−1(-1)^{1}=\mathbf{-1} (antisymmetric, Fermi). The antisymmetry η=−1\eta=-1 is the exclusion principle: setting q=p\mathbf q=\mathbf p gives cp†cp†=−cp†cp†c^\dagger_{\mathbf p}c^\dagger_{\mathbf p}=-c^\dagger_{\mathbf p}c^\dagger_{\mathbf p}, so the doubly occupied state is zero.

Example 2 — the neutral-kaon CPT mass bound. CPT forces mK0=mKˉ0m_{K^0}=m_{\bar K^0} exactly; the neutral-kaon system provides the tightest experimental test. Turn the measured bound into the fractional mass-difference limit.

Step 1. The observable is the fractional mass difference ∣mK0−mKˉ0∣/mK0|m_{K^0}-m_{\bar K^0}|/m_{K^0}, which CPT predicts to be exactly zero. The kaon mass is mK0=497.6 MeVm_{K^0}=497.6\ \text{MeV}.

Step 2. The neutral-kaon system is uniquely sensitive because the K0K^0–Kˉ0\bar K^0 mass difference would show up in the tiny oscillation frequency of the two-state system, whose scale is set by ΔmK≈3.48×10−12 MeV\Delta m_{K}\approx3.48\times10^{-12}\ \text{MeV} (the KLK_L–KSK_S difference) — an interferometer running at the 10−12 MeV10^{-12}\ \text{MeV} level on a 500 MeV500\ \text{MeV} mass.

Step 3. The Particle Data Group bound on CPT-violating mass splitting is ∣mK0−mKˉ0∣/mK0<10−18|m_{K^0}-m_{\bar K^0}|/m_{K^0}<\mathbf{10^{-18}} (more precisely <6×10−19<6\times10^{-19}). This is one of the most precise symmetry tests in physics — equivalent to comparing two 500 MeV500\ \text{MeV} masses and finding them equal to within about 5×10−16 MeV5\times10^{-16}\ \text{MeV}.

Step 4. Sanity-check the reach. A fractional precision of 10−1810^{-18} on a 497.6 MeV497.6\ \text{MeV} mass probes an absolute splitting of ∼5×10−16 MeV\sim5\times10^{-16}\ \text{MeV} — some four orders of magnitude below the already tiny KLK_L–KSK_S difference. No violation has ever been seen, exactly as the CPT theorem demands.

Origins · 1 mincpt theorem and how this was discovered
CT
cpt theorem
1940–1957 · Pauli's spin-statistics theorem (1940) and the CPT theorem of Lüders, Pauli, Bell and Jost (1954–1957) — the two structural theorems of relativistic quantum theory

The spin-statistics connection was proved for free fields by Markus Fierz in 1939 and given its definitive form by Wolfgang Pauli in 1940 (Phys. Rev. 58, 716), turning the empirical Pauli principle into a consequence of relativity. The CPT theorem came later and from several hands: Gerhart Lüders (1954) and Pauli (1955) established the operator-level result, John Bell reached it independently in his 1955 thesis, and Res Jost (1957) gave the rigorous proof within the Wightman axioms, using analytic continuation of correlation functions to complex "Jost points." Streater and Wightman's monograph PCT, Spin and Statistics, and All That (1964) welded both theorems into a single axiomatic structure — the reason they are always taught together, as two faces of Lorentz invariance plus locality.

History
1939
— Fierz proves the spin-statistics connection for free fields of arbitrary spin
1940
— Pauli publishes the general spin-statistics theorem (Phys. Rev. 58, 716)
1954
— Lüders establishes the CPT theorem at the operator level
1955
— Pauli gives his form of CPT; Bell independently derives it in his thesis
1957
— Jost proves CPT rigorously within the Wightman axioms via complex Jost points
1964
— Streater and Wightman publish PCT, Spin and Statistics, and All That

Going further

This capstone gathers the whole relativistic chapter into two theorems. The spin-statistics argument uses the causality test of The Feynman Propagator and Causality and the positivity requirement of Field Quantization of the Klein-Gordon and Dirac Fields as its two selection rules; it cashes out the exclusion principle of Pauli Exclusion Principle as an algebraic identity, and reads the negative-energy antiparticle term of Negative-Energy Solutions and the Interpretive Crisis as the very thing Fermi statistics repairs. The CPT bookkeeping acts on the sixteen bilinears of Bilinear Covariants and Dirac Adjoint and uses the particle↔antiparticle map of Charge Conjugation and the Positron. Beyond this article, the rigorous proofs — Jost's CPT theorem and the Streater-Wightman treatment of both — open the door to axiomatic quantum field theory; the interacting theory that follows (quantum electrodynamics and the full Standard Model) inherits these two theorems intact. The key results — the spacelike-commutator contrast, the Dirac Hamiltonian sign flip, the exchange-sign (−1)2s(-1)^{2s}, the (−1)n(-1)^n index parity, and the kaon CPT bound — are each independently verifiable from the derivations above. The kaon figure (∼10−18\sim10^{-18}) is the PDG order-of-magnitude bound, quoted as such with no fabricated precise digits.

Prerequisites. The Klein-Gordon Equation · The Dirac Equation · Negative-Energy Solutions and the Interpretive Crisis · Charge Conjugation and the Positron · Bilinear Covariants and Dirac Adjoint · Field Quantization of the Klein-Gordon and Dirac Fields. Leads to. Quantum field theory (canonical quantization) · the Wightman axioms and axiomatic QFT · quantum electrodynamics and the Standard Model.

Is the spin-statistics theorem really impossible to prove without relativity?

Correct — and this trips up nearly everyone. In nonrelativistic quantum mechanics, indistinguishability alone only tells you the wavefunction is symmetric or antisymmetric under exchange (a one-dimensional representation of the permutation group); it does not say which spin goes with which. Both choices give consistent theories. Feynman famously admitted he could not find an elementary argument. The connection needs the full relativistic package: Lorentz invariance ties the field's transformation to its spin, microcausality demands vanishing spacelike (anti)commutators, and positive energy rules out the other branch. Remove any one leg and the theorem falls.

What exactly is the Pauli-Jordan function Δ and why is it central?

Δ(x)\Delta(x) is the odd, Lorentz-invariant solution of the Klein-Gordon equation, equal to the commutator [ϕ(x),ϕ(0)][\phi(x),\phi(0)] up to a factor ii. Its defining property is that it vanishes for spacelike arguments x2<0x^2<0. That vanishing is microcausality made quantitative: two field measurements outside each other's light cones commute, so they cannot influence one another. The spin-statistics argument is entirely about which (anti)commutator reproduces Δ\Delta: for a scalar the commutator is the odd combination (= iΔi\Delta, vanishing), the anticommutator the even one (nonvanishing); the spinor's extra sign flips which is which.

If normal ordering just throws away an infinite constant, is the positivity argument trustworthy?

Yes, because the disease in the wrong-statistics Dirac Hamiltonian is not the infinite constant — it is the sign of the number operator. With commutators, −d d†-d\,d^\dagger becomes −d†d-d^\dagger d plus a constant: the operator part is negative, so each real antiparticle you add lowers the energy by EpE_{\mathbf p}, giving a spectrum with no floor. Normal ordering cannot fix a genuine operator with negative eigenvalues. Anticommutators flip −d d†-d\,d^\dagger to +d†d+d^\dagger d minus a constant: now the operator part is positive, and only then is the leftover constant an irrelevant (infinite) shift you discard. The argument turns on the operator sign, not the constant.

How is CPT related to the Feynman-Stückelberg "antiparticle = particle going backward in time" picture?

They are the same statement seen from two angles. CPT sends a field operator O(x)\mathcal O(x) to O†(−x)\mathcal O^\dagger(-x): it reverses all four spacetime coordinates (P reverses space, T time) and conjugates charge (C). A process and its total spacetime reflection therefore have equal amplitudes. Reflecting a particle's worldline through the origin turns a forward-in-time particle into a backward-in-time antiparticle — exactly the Feynman-Stückelberg reading. CPT is what makes that heuristic exact and forces the reflected object (the antiparticle) to share the original's mass and lifetime.

Why is the neutral-kaon system such a good CPT test rather than, say, the electron?

Because the K0K^0–Kˉ0\bar K^0 pair is a naturally occurring two-state quantum system whose time evolution is governed by a tiny energy (mass) difference, so any CPT-violating mass splitting appears as a measurable shift in the oscillation between the two states. You are running an interferometer whose fringe spacing is set by Δm/m\Delta m/m, and the kaon's oscillation frequency lets you resolve fractional mass differences down to ∼10−18\sim10^{-18}. Electron–positron comparisons (via gg-factors in Penning traps) and antiproton spectroscopy are also superb tests, but the kaon's mass equality remains the single most stringent number, precisely because oscillation amplifies a minuscule energy difference into an observable phase.

Does spin-statistics apply to composite particles like atoms and nuclei?

Yes, and it follows from the constituents. A composite of an even number of fermions (plus any number of bosons) has integer total spin and behaves as a boson; an odd number of fermions gives half-integer spin and fermionic behavior. Helium-4 (two protons, two neutrons, two electrons — six fermions) is a boson and can Bose-condense into a superfluid; helium-3 (one fewer neutron — five fermions) is a fermion and only becomes superfluid by first forming Cooper pairs. The rule is exact as long as the composite is probed at energies too low to resolve its internal structure — below that scale, its exchange sign is the product of its constituents' signs.

Could CPT ever be violated, and what would that mean?

Within standard quantum field theory, no — CPT is a theorem. Violating it requires breaking one of its premises: Lorentz invariance or locality. Some speculative quantum-gravity and string-inspired scenarios predict tiny Lorentz-violating, and hence CPT-violating, effects at energies near the Planck scale. That is why the experimental bounds matter so much: a confirmed CPT violation, even at the 10−1810^{-18} level in the kaon mass or in an antihydrogen spectral line, would be unambiguous evidence that spacetime is not exactly Lorentz-invariant — a genuine crack in the foundations, and one of the few laboratory windows onto physics beyond the Standard Model.

Before this