Bosons and Fermions

Every particle in the universe belongs to one of two tribes. Bosons live in the totally symmetric sector of the exchange operator, pile into a single mode without limit, and build lasers and condensates. Fermions live in the antisymmetric sector, refuse to share a state, and build atoms, chemistry, and the pressure that holds up dead stars. Symmetric and Antisymmetric States constructed the two-particle members of each tribe; this article takes the census — which species is which, and why. The membership rule is spin, through the spin-statistics theorem: integer spin means boson, half-integer spin means fermion. We state that dictionary and hand the reason to Spin-Statistics Theorem, its rightful owner. What this article owns is the machinery downstream of the dictionary. From the symmetric construction we read off the occupancy split — bosons stack in one mode legally, fermions annihilate on contact — and exhibit the $N!$ enhancement that makes identical bosons bunch. And we prove the one theorem that lets you classify composite objects from their parts: exchanging two identical composites, each made of $n_f$ fermions and any number of bosons, multiplies the state by $(-1)^{n_f}$ . That composite rule has a punchline you can weigh in a cryostat. Helium-4 is a boson (six fermionic constituents, even); helium-3 is a fermion (five, odd). One extra neutron flips the statistics, and with it the superfluid transition temperature drops by nearly three orders of magnitude — from $2.17$ K to about $2.5$ mK, where fermions must first pair up before they can flow. We verify the constituent count and the sign with an explicit calculation, catalog the elementary and composite tribes, and mark honestly where the thermal occupation formulas — the Bose-Einstein and Fermi-Dirac distributions — hand off to statistical mechanics.

Key takeaways

Identical particles in three dimensions come in exactly two kinds: bosons (totally symmetric, exchange eigenvalue $+1$ ) and fermions (totally antisymmetric, $-1$ ) — the two sectors built in Symmetric and Antisymmetric States.
The dictionary is spin: integer-spin particles are bosons, half-integer-spin fermions. This is the content of the spin-statistics theorem; its proof is owned by Spin-Statistics Theorem.
Occupancy splits: $N$ bosons stack into one mode legally (a symmetric product), while two fermions in one state give $\psi_-\equiv0$ — the Pauli seed, owned downstream.
The owned result is the composite-particle theorem: swapping two identical composites multiplies the state by $(-1)^{n_f}$ , where $n_f$ is the number of fermionic constituents — an even count makes a boson, an odd count a fermion.
The punchline: $^4$ He (6 fermions) is a boson, $^3$ He (5 fermions) a fermion; that one-neutron difference moves the superfluid transition from $2.17$ K to $\sim2.5$ mK, three orders of magnitude.
Identical bosons in one mode carry an $N!$ amplitude enhancement (the permanent-of-ones), the seed of bunching and stimulated emission; the general permanent machinery is owned elsewhere.
Key idea: statistics is fixed by spin — integer spin bosons pile up, half-integer spin fermions exclude — and for composites the sign is just the parity of the fermion count, $(-1)^{n_f}$ , which is why one neutron turns $^4$ He from a superfluid at $2.17$ K into $^3$ He needing $2.5$ mK.

The core insight

Identical particles in 3D are bosons (symmetric, integer spin, no occupancy limit) or fermions (antisymmetric, half-integer spin, Pauli exclusion), with statistics fixed by spin. For composite objects the exchange sign is the parity of the fermion count, $(-1)^{n_f}$ — the rule that makes $^4$ He a boson and $^3$ He a fermion.

Where this lives today

The two tribes define two of the largest research frontiers in physics. Bosonic quantum gases — Bose-Einstein condensates of $^{87}$ Rb, $^{23}$ Na, and dozens of other species — are the platform for atom interferometry, quantum simulation of superfluidity, and matter-wave optics. Their fermionic mirror, degenerate Fermi gases of $^6$ Li and $^{40}$ K, realize tunable models of high-temperature superconductivity through the BEC-BCS crossover, where fermion pairs continuously turn into composite bosons and condense — the composite rule of this article playing out in real time as pairs form. Molecular BECs, made by binding two fermionic atoms into a bosonic molecule, are a direct experimental demonstration of the $(-1)^{n_f}$ theorem: two fermions bound together condense, exactly as the parity count predicts. On the elementary side, the 2012 discovery of the Higgs boson completed the census of the Standard Model's fundamental particles, sorting each into its tribe by spin.

The physical picture

Picture two identical particles and a swap. The state can pick up only a sign — and only a sign, because swapping twice returns to start. That sign, $+1$ or $-1$ , sorts every particle into the sociable tribe or the antisocial one. Bosons with the plus sign are gregarious: nothing stops many of them piling into the same mode, and the amplitude to do so is actually enhanced, which is why lasers and Bose-Einstein condensates exist. Fermions with the minus sign are exclusive: put two in the same state and the antisymmetric combination cancels itself to zero. That single cancellation is the reason atoms have shells, chemistry has a periodic table, and white dwarfs do not collapse. Which sign a species carries is not up to the species — it is dictated by its spin. The spin-statistics theorem, whose proof belongs to the next article, says integer spin goes with the symmetric sign and half-integer spin with the antisymmetric one. Here we take that as a working rule and follow its consequences.

Figure

The composite rule and its punchline: one neutron flips the tribe and drops the transition 868×

The composite rule in the figure is the engine of this classification. It says you do not need to look inside an atom's dynamics to know its tribe — you only count its fermions. An even count of fermionic constituents and the composite is a boson; an odd count and it is a fermion. Because protons, neutrons, and electrons are all fermions, a neutral atom's fermion count is $Z + N + Z = 2Z + N$ , so the parity is just the parity of the neutron number $N$ . That is why isotopes of the same element can belong to different tribes.

The dictionary and the occupancy split

The exchange eigenvalue is fixed, and spin fixes which value a species takes.

\hat P_{12}|\psi\rangle=\pm|\psi\rangle,\qquad s\in\{0,1,2,\dots\}\Rightarrow +\ (\text{boson}),\quad s\in\{\tfrac12,\tfrac32,\dots\}\Rightarrow -\ (\text{fermion}) \qquad (1)

where: $\hat P_{12}$ — exchange operator, $\hat P_{12}^2=\mathbb1$ ; $s$ — the particle's spin quantum number; the sign is the tribe. The physical read: the assignment of sign to spin is the content of the spin-statistics theorem, stated here and argued in Spin-Statistics Theorem; this article never tries to prove it. The consequences for occupation follow from the two-particle construction of Symmetric and Antisymmetric States, recalled here. For bosons the symmetric combination survives when both particles share a mode: $N$ bosons can occupy one single-particle state as the plainly-normalized symmetric product $\psi_a(1)\cdots\psi_a(N)$ . For fermions the antisymmetric combination vanishes when two share a mode, $\psi_-\equiv0$ — the double-occupancy collapse that is the seed of the Pauli exclusion principle, whose full consequences (shells, the periodic table, degeneracy pressure) are owned by Pauli Exclusion Principle. In one line: bosons pile, fermions exclude.

Figure

The occupancy split and the two thermal distributions: bosons pile up, fermions cap at one

There is a quantitative signature of the boson pile-up worth exhibiting before it becomes machinery. The amplitude for $N$ identical bosons to occupy the same mode carries a factor of $N!$ relative to distinguishable particles — the "permanent of ones."

Derivation AThe N! bunching enhancement for same-mode bosons

Show that the many-boson symmetric state of $N$ particles in one mode carries an amplitude enhancement of $\sqrt{N!}$ relative to distinguishable particles, so probabilities scale as $N!$ . The goal is to expose bunching as a counting fact — there are $N!$ ways to assign labels that all give the same physical configuration, and symmetrization adds their amplitudes in phase. We stop at $N=2,3$ explicitly; the general permanent constructor is owned by Permanents and Symmetrized Boson States.

Step 1. The normalized symmetric state of $N$ particles distributed over modes is built by summing over all $N!$ label permutations. When all $N$ particles occupy the same mode, every permutation gives the identical term, so the sum is $N!$ copies of one product.

Step 2. The overlap that counts the enhancement is the permanent of the all-ones $N\times N$ matrix, $\mathrm{perm}(J_N)=N!$ . For $N=2$ this is $2!=2$ ; for $N=3$ it is $3!=6$ — computed exactly.

Step 3. Physically, the joint detection amplitude for finding all $N$ bosons together is enhanced by $\sqrt{N!}$ over the distinguishable case, so the probability is enhanced by $N!$ . This is stimulated emission in embryo: a photon is more likely to join a mode already occupied, precisely by this bunching factor. Why it stops here: the general $N$ -mode, mixed-occupancy bookkeeping is the permanent, with its own normalization $1/\sqrt{N!\prod_k n_k!}$ ; that machinery is owned downstream. The Hong-Ou-Mandel two-photon coincidence dip is the $N=2$ version of this same symmetrization, recalled from Indistinguishability of Identical Particles. Verified: `verify_aq-qm-bosons-fermions.py` — perm(ones $2\times2)=2$ , perm(ones $3\times3)=6$ exact.

The composite-particle theorem

Now the owned result. Most of the particles we care about — atoms, nuclei, hadrons — are composite, and their statistics is not postulated but computed from their parts. The rule is remarkably clean: only the fermion count matters, and only its parity.

\text{swap of two identical composites}\ \Longrightarrow\ (-1)^{n_f},\qquad n_f=\text{number of fermionic constituents} \qquad (2)

where: $n_f$ — the number of fermions inside one composite; bosonic constituents contribute nothing to the sign. The physical read: exchanging two identical composites is bookkeeping-equivalent to exchanging their constituents pairwise; each fermion pair-swap costs a minus sign, each boson pair-swap costs nothing, so the net phase is $(-1)^{n_f}$ . Even $n_f$ makes a boson, odd makes a fermion.

Go deeper · Derivation B · 1 minThe composite-particle theorem: statistics from constituent parity

Derivation BThe composite-particle theorem: statistics from constituent parity

Derive that exchanging two identical composites multiplies the state by $(-1)^{n_f}$ , and confirm it with an explicit four-particle calculation. The goal is to show this is not a hand-wave but a parity count: swapping two composites is a specific permutation of all their constituents, and the sign of that permutation is fixed by how many fermion transpositions it contains. The payoff is a rule that classifies every atom and nucleus from a headcount.

Step 1. Two identical composites $A$ and $B$ , each with the same internal structure, occupy disjoint sets of constituents. Composite $A$ holds fermions $a_1,\dots,a_{n_f}$ and bosons; $B$ holds the matching $b_1,\dots,b_{n_f}$ and bosons.

Step 2. Exchanging $A$ with $B$ as whole objects is the permutation that maps each constituent of $A$ to the corresponding constituent of $B$ and vice versa. This factorizes into $n_f$ transpositions among the fermions ( $a_i\leftrightarrow b_i$ ) and some transpositions among the bosons.

Step 3. Each fermion transposition contributes a factor $-1$ to the total antisymmetric state; each boson transposition contributes $+1$ . The composites being identical, the internal wavefunctions match, so the only surviving phase is the product of these signs: $(-1)^{n_f}\cdot(+1)^{n_b}=(-1)^{n_f}$ .

Step 4. Machine confirmation. Build two "molecules," each an antisymmetrized pair of fermions on distinct modes (four fermions total, a Slater determinant over four orbitals). Apply the composite swap that exchanges the two molecules as blocks and read the eigenvalue: it is exactly $+1$ (two 2-fermion molecules make a boson), while a single internal fermion transposition gives exactly $-1$ as a control. Both are computed to machine precision. Why the boson constituents drop out: a pair-swap of two identical bosons multiplies by $+1$ , so any number of them leaves the sign untouched. The statistics of a composite is blind to its bosonic glue. Verified: `verify_aq-qm-bosons-fermions.py` — the $(-1)^{n_f}$ rule and the numpy 4-particle composite-swap eigenvalue $+1$ (control $-1$ ), exact.

The theorem turns statistics into arithmetic. A neutral atom has $Z$ protons, $N$ neutrons, and $Z$ electrons — all fermions — so $n_f=2Z+N$ and the tribe is set by the parity of $N$ . Helium-4 has $n_f=6$ (a boson); helium-3 has $n_f=5$ (a fermion). Rubidium-87 has $n_f=124$ (a boson, which is why it was the first Bose-Einstein condensate); lithium-6 has $n_f=9$ (a fermion, a workhorse of degenerate Fermi-gas experiments). The hydrogen atom ( $n_f=2$ ) and the deuteron ( $n_f=2$ ) are both bosons. One headcount, one bit of information.

The census of everything

Tribe	Elementary examples	Composite examples	Occupancy
Bosons (integer spin)	photon $s=1$ , gluon, $W/Z$ , Higgs $s=0$	$^4$ He, $^{87}$ Rb, H atom, deuteron, mesons (quark-antiquark)	unlimited per mode
Fermions (half-integer spin)	electron, quark, neutrino $s=\tfrac12$	$^3$ He, $^6$ Li, proton, neutron, baryons (three quarks)	one per mode (Pauli)

Two threads of honesty belong here. First, the composite rule holds only while the internal structure is frozen — while the binding energy is much larger than the thermal energy that might excite the composite. Cold rubidium atoms behave as elementary bosons because every internal scale dwarfs the cloud: even the ground-state hyperfine splitting corresponds to $0.33$ K — a factor of $3\times10^6$ above a 100 nK condensate — and the first electronic excitation to $1.8\times10^4$ K. A hot plasma that ionizes the atom exposes its fermionic electrons, and the composite statistics no longer applies. Second, the thermal distributions — how the two tribes populate energy levels at temperature $T$ — are the Bose-Einstein and Fermi-Dirac occupancy formulas, and their derivation is owned by statistical mechanics.

\langle n_k\rangle=\frac{1}{e^{(\epsilon_k-\mu)/k_BT}\mp1}\qquad(-\ \text{Bose-Einstein},\ +\ \text{Fermi-Dirac}) \qquad (3)

where: $\epsilon_k$ — single-particle energy of mode $k$ ; $\mu$ — chemical potential; the $\mp$ is the tribe. The physical read: the Fermi-Dirac form ( $+$ ) is bounded between $0$ and $1$ — one fermion per state at most — while the Bose-Einstein form ( $-$ ) can diverge as $\mu\to\epsilon_k$ , the mathematical face of condensation. We state these shapes; their thermodynamics lives in the statistical-mechanics track.

Limiting cases

Case	Result	Meaning
Dilute / non-degenerate, $n\lambda_{\text{th}}^3\ll1$	both statistics → Maxwell-Boltzmann	when packets rarely overlap, the tribe is invisible; classical counting works
Degenerate fermions, $T\to0$	fill to the Fermi energy, degeneracy pressure	exclusion stacks momenta; supports white dwarfs and metals (owned by Pauli Exclusion Principle)
Degenerate bosons, $n\lambda_{\text{th}}^3\gtrsim2.612$	macroscopic ground-state occupation	Bose-Einstein condensation; the onset constant is $\zeta(3/2)\approx2.612$ (stat-mech owns the derivation)
Composite, binding $\gg k_BT$	$(-1)^{n_f}$ rule valid	internal structure frozen; the atom acts as one particle of definite tribe
Two spatial dimensions	exchange phase $e^{i\theta}$ , any $\theta$	the $\pm1$ dichotomy is a 3D fact; anyons own the 2D loophole (Leinaas-Myrheim 1977)

Worked examples

Example 1 — the isotope census. Classify $^4$ He, $^3$ He, $^{87}$ Rb, $^6$ Li, and H $_2$ by counting fermionic constituents, and give the statistics verdict for each.

Step 1. Count fermions. A neutral atom has $Z$ protons $+ N$ neutrons $+ Z$ electrons, all fermions, so $n_f=2Z+N$ . For $^4$ He: $2(2)+2=\mathbf{6}$ . For $^3$ He: $2(2)+1=\mathbf{5}$ . For $^{87}$ Rb ( $Z=37,N=50$ ): $2(37)+50=\mathbf{124}$ . For $^6$ Li ( $Z=3,N=3$ ): $2(3)+3=\mathbf{9}$ . For the H $_2$ molecule (two H atoms, each $1p+1e=2$ fermions): $\mathbf{4}$ .

Step 2. Read parity. Even $n_f$ is a boson, odd is a fermion: $^4$ He (6) boson, $^3$ He (5) fermion, $^{87}$ Rb (124) boson, $^6$ Li (9) fermion, H $_2$ (4) boson.

Step 3. Boundary check — the physical consequence. The two bosonic alkali/inert species ( $^{87}$ Rb, $^4$ He) can Bose-condense; the two fermionic species ( $^3$ He, $^6$ Li) cannot condense directly and instead form degenerate Fermi gases or (with pairing) BCS-type superfluids. The 1995 BEC used $^{87}$ Rb precisely because it is a composite boson; $^6$ Li is a standard choice for degenerate Fermi gases for the mirror reason. The census predicts which experiment each atom is fit for.

Example 2 — the $N!$ bunching numbers. Compute the same-mode occupation enhancement for two and three identical bosons versus distinguishable particles, and tie it to the Hong-Ou-Mandel effect.

Step 1. Two bosons. The permanent of the $2\times2$ all-ones matrix is $\mathrm{perm}(J_2)=2!=\mathbf{2}$ , so the joint amplitude to find both bosons in one mode is enhanced by $\sqrt{2}$ and the probability by a factor $2$ over distinguishable particles.

Step 2. Three bosons. $\mathrm{perm}(J_3)=3!=\mathbf{6}$ : the probability enhancement is a factor $6$ . The bunching grows factorially — the mathematical root of stimulated emission and lasing.

Step 3. The HOM tie. For two identical photons meeting at a beamsplitter, the same symmetrization forces them to leave together (a coincidence dip): the two-particle amplitude for one-here-one-there cancels. This is the $N=2$ face of the bunching computed above, and the indistinguishability that drives it is owned by Indistinguishability of Identical Particles.

Origins · 1 minSatyendra Nath Bose and how this was discovered

Satyendra Nath Bose

1894–1974 · Calcutta, Dhaka

In 1924, teaching in Dhaka, Satyendra Nath Bose found a way to derive Planck's radiation law by counting photon states with a new statistics — treating indistinguishable light quanta as genuinely interchangeable rather than labelled. His paper was rejected by a British journal, so he sent it to Einstein, who recognized its importance, translated it into German himself, and submitted it to Zeitschrift für Physik (26, 178). Einstein then extended the idea from photons to material atoms and predicted the condensation that now bears both their names. Bose spent his career between Calcutta and Dhaka, elected a Fellow of the Royal Society but never awarded a Nobel Prize despite the statistics — and the entire tribe of "bosons" — carrying his name. The naming itself came later: Dirac is generally credited with popularizing "boson" as the counterpart of "fermion," honoring Bose's founding calculation.

History

1924

— Bose derives Planck's law from photon-state counting; Einstein translates and submits the paper (Z. Phys. 26, 178)

1925

— Einstein's second quantum-gas paper extends the statistics to atoms and predicts condensation

1926

— Fermi and Dirac formulate the antisymmetric (exclusion-obeying) statistics for electrons

1938

— Kapitza, and Allen and Misener, discover superfluidity in liquid

^4

He at the 2.17 K λ-point

1972

— Osheroff, Richardson, and Lee find superfluid

^3

He near 2.5 mK, requiring fermion pairing

1995

— Bose-Einstein condensation achieved in dilute

^{87}

Rb and

^{23}

Na gases

Going further

References. Griffiths, Introduction to Quantum Mechanics, 3rd ed., Ch. 5 · Sakurai & Napolitano, Modern Quantum Mechanics, 2nd ed., Ch. 7 · Shankar, Principles of Quantum Mechanics, 2nd ed., Ch. 10 · Cohen-Tannoudji, Diu & Laloë, Quantum Mechanics, Vol. II, Ch. XIV · Townsend, A Modern Approach to Quantum Mechanics, Ch. 12

Provenance. All structural claims standard across Griffiths Ch. 5, Sakurai Ch. 7, Shankar Ch. 10, Cohen-Tannoudji Vol. II Ch. XIV. A verifier should: (1) confirm $\hat P_{12}^2=\mathbb 1\Rightarrow$ eigenvalues $\pm1$ ; (2) check that setting $\psi_a=\psi_b$ in the antisymmetric state gives zero; (3) verify the exchange-energy result $E_\pm=J\pm K$ with the stated direct/exchange integrals; (4) confirm the BEC critical value $\zeta(3/2)\approx2.612$ ; (5) verify composite parity rule on He-4 (boson) vs He-3 (fermion). No fabricated numerical constants used. — Editorial note for this rebuild: the exchange-energy structure $E_\pm=J\pm K$ is honored as a pointer only (owned by Exchange Forces and Exchange Energy and applied in Singlet and Triplet States); the Bose-Einstein/Fermi-Dirac distribution derivations and the $\zeta(3/2)$ onset are owned by the statistical-mechanics track and stated here prose-only. Every number added by this rebuild — the isotope fermion counts and verdicts, the $N!$ permanents, $\zeta(3/2)$ , and the $2.17$ K / $2.5$ mK λ-points — is recomputed in the committed verification script, including an explicit numpy 4-particle composite-swap eigenvalue.

Builds on: Indistinguishability of Identical Particles · Permutation Symmetry and the Exchange Operator · Symmetric and Antisymmetric States Leads to: Spin-Statistics Theorem · Pauli Exclusion Principle · Slater Determinants · The Helium Atom

Why does only the fermion count matter for a composite, and not the total number of particles?

Because the exchange sign is a product over constituent pair-swaps, and only fermion pair-swaps contribute a minus sign — boson pair-swaps contribute $+1$ . Swapping two identical composites is exactly $n_f$ fermion transpositions plus some boson transpositions, so the net phase is $(-1)^{n_f}$ . The bosonic glue is statistically invisible. A neutral atom's tribe is therefore fixed by the parity of $2Z+N$ , i.e. by the neutron number.

Can a composite boson ever "leak" its fermionic constituents?

Only if you excite it. The composite rule holds while the internal structure is frozen — while the binding energy far exceeds the thermal (or collision) energy that could excite or dissociate the composite. Cold $^{87}$ Rb behaves as a perfect boson because electronic binding ( $\sim$ eV) dwarfs microkelvin thermal energies. Heat it enough to ionize, and the fermionic electrons are exposed and the atom no longer acts as a single boson. In BEC-BCS crossover experiments this is deliberate: fermion pairs are loosely enough bound that one can tune continuously from tightly-bound composite bosons to overlapping Cooper pairs.

If bosons have no occupancy limit, why don't all the electrons in an atom fall into the 1s state?

They would if they were bosons — but electrons are fermions ( $s=\tfrac12$ ), so the antisymmetric requirement forbids two of them from sharing all quantum numbers. That exclusion is what builds shells and the periodic table. The contrast is the whole point of the two tribes: bosons pile up (condensates, lasers), fermions stack (atoms, white dwarfs). The consequences of the fermion side are owned by Pauli Exclusion Principle.

Where do the Bose-Einstein and Fermi-Dirac distributions come from, and why aren't they derived here?

They come from maximizing entropy (or the grand partition function) for indistinguishable particles obeying the two occupancy rules, at temperature $T$ and chemical potential $\mu$ . That is a statistical-mechanics calculation, and its natural home is the stat-mech track. This article owns the quantum input — which particles are which tribe and how they occupy modes — and states the distribution shapes (Eq. 3) as the handoff, without deriving their thermodynamics.

Is the hydrogen atom really a boson?

Yes — a hydrogen atom is one proton plus one electron, two fermions, so $(-1)^2=+1$ : a composite boson. Spin-polarized atomic hydrogen was in fact one of the earliest candidates for Bose-Einstein condensation, and hydrogen BEC was achieved in 1998. The deuteron (one proton, one neutron) is likewise a boson, while the deuterium atom (proton + neutron + electron, three fermions) is a fermion — a nice illustration that adding one fermion flips the tribe.

Why do the boson and fermion occupancy formulas differ only by a sign in the denominator?

The $\mp1$ in $\langle n_k\rangle=1/(e^{(\epsilon_k-\mu)/k_BT}\mp1)$ is the arithmetic residue of symmetric versus antisymmetric counting of states. For fermions the $+1$ keeps the occupancy between $0$ and $1$ — one particle per state — while for bosons the $-1$ lets the occupancy diverge as $\mu\to\epsilon_k$ , which is exactly the mathematical onset of Bose-Einstein condensation. The single sign that sorts the tribes at the two-particle level reappears here as the sign that shapes their thermodynamics.

Does the

2\pi

-rotation-gives-minus-sign fact prove the spin-statistics connection?

It is suggestive but not a proof. A single half-integer-spin particle rotated by $2\pi$ acquires a factor $-1$ , and exchanging two of them also gives $-1$ , so the two facts rhyme. Heuristic "belt-trick" arguments trade on this. But turning the rhyme into a theorem requires relativistic quantum field theory and its assumptions (locality, positive energy) — that is the content of Spin-Statistics Theorem, which also explains why the naive rotation argument is incomplete.