Spin Angular Momentum

Frankfurt, 1922. Otto Stern and Walther Gerlach fire a thin beam of silver atoms through a magnet built deliberately lopsided — one pole a knife edge, the other a flat face — so that the field tugs harder on one side of an atom than the other. Each silver atom is a tiny magnet; classical physics says the atoms' magnetic axes point every which way, so the beam should smear into a vertical band. The photographic plate says otherwise: two crisp spots, nothing between. The atomic compass needle, asked which way it points along the field, gives only two answers — ever. What the plate recorded was the end of classical angular momentum. The two-ness was the scandal. Orbital angular momentum, quantized though it is, always offers an odd number of orientations ( $2\ell + 1$ ). An even count demands a half-integer — and no wavefunction on a sphere can supply one. The resolution is a new kind of angular momentum that no spatial motion carries: spin, an intrinsic, fixed-magnitude angular momentum the electron simply has, the way it has charge and mass. What makes spin the perfect subject is that it runs on algebra alone: three commutation relations — the same ones orbital angular momentum obeys — generate the allowed magnitudes, the $2s+1$ orientations, the Pauli matrices, the precession in fields. Master this and you hold the working language of MRI, atomic clocks, and every qubit built to date.

Key takeaways

Spin is intrinsic angular momentum: a particle carries it while at rest and pointlike — it is not orbital motion, and it has no expression in $x$ and $p$ .
The spin operators obey exactly the orbital algebra, $[S_i, S_j] = i\hbar\,\epsilon_{ijk}S_k$ ; the ladder-operator machinery applies unchanged.
Freed from the single-valuedness that constrains spatial wavefunctions, the algebra allows $s = 0, \tfrac12, 1, \tfrac32, \dots$ — half-integers included, each with $2s+1$ projections $m_s$ .
Electrons, protons, neutrons, and quarks all have $s = \tfrac12$ : a two-dimensional state space, spanned by up and down, run by the Pauli matrices.
Spin talks to the world through its magnetic moment $\boldsymbol{\mu} = g(q/2m)\mathbf{S}$ , precessing at the Larmor frequency — the physics of Stern–Gerlach, NMR, and MRI; a $2\pi$ rotation multiplies a spin- $\tfrac12$ state by $-1$ , verified by neutron interferometry.
Key idea: spin is angular momentum defined by algebra alone, living in an internal space — and the algebra permits the half-integers that experiment demanded.

The core insight

Spin is angular momentum defined purely by the algebra $[S_i, S_j] = i\hbar\,\epsilon_{ijk}S_k$ acting on an internal finite-dimensional space; freed from the single-valuedness of spatial wavefunctions, it admits half-integer quantum numbers, with $s = \tfrac12$ the case realized by electrons and nucleons.

Where this lives today

Spin is the most monetized abstraction in quantum mechanics. Every MRI scan is Larmor precession run in reverse: hydrogen nuclei precess at $42.6\ \mathrm{MHz/T}$ , are tipped by a resonant pulse, and betray their tissue through relaxation times — Derivation B with a radiologist attached. Hard drives read bits with spin-valve magnetoresistance; MRAM stores them in spin orientation; the field even has a name, spintronics. Qubits are spin- $\tfrac12$ mathematics wherever they are built — transmon, trapped ion, nitrogen-vacancy center — two-level systems on the Bloch sphere whose gates are rotation operators; an " $X$ gate" means $\sigma_x$ , whatever the hardware. And the electron's $g$ -factor — the coupling in equation (5) — is now measured to thirteen digits and matched by QED: the most precisely verified prediction in science, descended in a straight line from two spots on a photographic plate.

The physical picture

Figure

The Stern–Gerlach experiment — the question with only two answers

An inhomogeneous field is essential: a uniform field only torques a magnetic moment, but a field stronger near the knife edge exerts a net force, $F_z = \mu_z\,\partial B_z/\partial z$ , sorting atoms by $\mu_z$ . Classically $\mu_z$ ranges continuously from $-\mu$ to $+\mu$ : a smear. Quantum mechanically the silver atom's lone valence electron offers exactly two values of $S_z$ — two forces, two spots. The measurement does not reveal a pre-existing direction; it forces one of two answers to "up or down along this axis?", and the axis is yours to choose. The apparatus — pole shapes, oven arithmetic, chained magnets — has its own article, The Stern-Gerlach Experiment; here the two spots are the evidence that forced spin into physics. So what is spinning? Nothing. Model the electron as a classical ball of the largest radius experiment allows and ask how fast its equator must move to store $\hbar/2$ : well over a hundred times the speed of light — and the electron is pointlike as far as we can tell. Spin is not motion through space; it is an internal quantum label that happens to transform like an angular momentum. The right mental image is a fixed-length arrow, $|\mathbf{S}| = \hbar\sqrt{s(s+1)}$ , of which you may know the length and one component, never all three: the components refuse to commute, exactly like $L_x, L_y, L_z$ . That refusal is the engine of everything strange here: a particle definitely spin-up along $z$ is a 50/50 coin along $x$ ; measure $S_x$ and you scramble $S_z$ . No state has all three components sharp — the algebra forbids it. The Stern–Gerlach cascade turns that prohibition into beam statistics: chain three magnets — $S_z$ , then $S_x$ , then $S_z$ again — and a beam purified to 100% spin-up regrows a spin-down quarter from nothing. The chain's arithmetic ( $1/4$ selected, $1/2$ unread, $0$ removed) is owned by Measurement and Wavefunction Collapse; the beams are run at The Stern-Gerlach Experiment.

The math: an algebra and its ladder

How do you define an angular momentum with no orbit, no $\mathbf{r}\times\mathbf{p}$ , no wavefunction of angles to differentiate? You keep the only thing left: the algebra. Angular Momentum Commutation Relations made exactly this move — the abstraction pivot: angular momentum is any Hermitian triple obeying the bracket, whatever space it acts on. Spin is that definition cashed in — three operators commuting exactly the way $L_x, L_y, L_z$ do, on an internal space with no coordinates at all:

[S_x, S_y] = i\hbar S_z, \qquad [S_y, S_z] = i\hbar S_x, \qquad [S_z, S_x] = i\hbar S_y \qquad (1)

where: $S_i$ — spin component operators; compactly, $[S_i, S_j] = i\hbar\,\epsilon_{ijk}S_k$ . The nonzero right-hand sides are the physics: no two components can be sharp at once, which is why the fixed-length arrow of the physical picture can never be fully pinned down. Which questions can be answered together? $S^2 = S_x^2 + S_y^2 + S_z^2$ commutes with every component — the Casimir property proved in the algebra article — so it sharpens alongside any one of them, conventionally $S_z$ , and the simultaneous eigenstates carry two labels:

S^2\,|s, m_s\rangle = \hbar^2 s(s+1)\,|s, m_s\rangle, \qquad S_z\,|s, m_s\rangle = \hbar m_s\,|s, m_s\rangle \qquad (2)

where: $s$ — spin quantum number; $m_s$ — projection, running from $-s$ to $s$ in unit steps. These eigenvalue shapes are not assumptions: Ladder Operators and the Angular Momentum Spectrum derived them from the bracket alone, consumed here wholesale. Read the odd-looking $s(s+1)$ : the length $\hbar\sqrt{s(s+1)}$ always exceeds the largest projection $\hbar s$ , so $\mathbf{S}$ never points exactly along any axis — a tilted cone, never a settled needle. The ladder machinery moves between the rungs with the standard matrix elements, Condon–Shortley phases included:

S_\pm\,|s, m_s\rangle = \hbar\sqrt{s(s+1) - m_s(m_s \pm 1)}\;|s, m_s \pm 1\rangle \qquad (3)

where: $S_\pm = S_x \pm iS_y$ , with the same real non-negative coefficients as ever. Nothing in the formula knows whether it steps an orbital $m$ or a spin $m_s$ — that indifference is the point of defining angular momentum algebraically.

Derivation AWhy half-integers are legal for spin and illegal for orbital motion

The goal is to establish which values of $s$ the algebra permits, and why spin gets to use values orbital angular momentum must refuse. The strategy is not to re-derive the spectrum — Ladder Operators and the Angular Momentum Spectrum owns that theorem; we import its result — but to walk the one step that differs: where the orbital problem imposes an extra, spatial constraint, spin has no space to impose it on. Watch what plays the boundary-condition role in each case; the answers differ in kind.

Step 1. Import the spectrum theorem. From the algebra (1) alone, the ladder analysis gives $S^2$ eigenvalues $\hbar^2 s(s+1)$ with $m_s$ running from $-s$ to $+s$ in unit steps, the ladder length a whole number of rungs:

m_{\max} - m_{\min} = 2s \in \{0, 1, 2, 3, \dots\} \quad\Longrightarrow\quad s = 0, \tfrac12, 1, \tfrac32, \dots

Why: positivity of norms terminates the ladder at both ends, an integer number of unit steps apart — the imported theorem, used as a licence, not re-proved.

Step 2. Ask what orbital angular momentum adds. There the operators are differential — $L_z = -i\hbar\,\partial_\phi$ — and eigenstates are functions $e^{im\phi}$ on real space; a half-integer $m$ makes $e^{im\phi}$ change sign at the same physical point after $\phi \to \phi + 2\pi$ . The single-valuedness argument and its fine print are owned by Eigenfunctions of L_z and the Azimuthal Quantization; the verdict is what matters: orbital $m$ , hence $\ell$ , must be integer. Half the ladders are struck off. Why: the constraint is spatial, not algebraic — a consistent wavefunction on the circle, a demand the bracket knows nothing about.

Step 3. Ask what spin adds: nothing. A spin state is a function of no angle — no $\langle\theta,\phi|s,m_s\rangle$ , no circle to walk, no point of space where a sign flip could contradict itself. The algebra's full solution set survives, half-integers included. Why: no representation on functions means no single-valuedness tax — spin's only "boundary condition" is positivity of the inner product, which the imported theorem already used.

Step 4. Let experiment choose. Silver's two spots demand $2s + 1 = 2$ : $s = \tfrac12$ , a value only spin can realize. And note the grammar: $s$ is a species label, fixed per particle type — an electron is $s = \tfrac12$ the way it is charge $-e$ — while only $m_s$ is dynamical. Why: here spin parts company with orbital motion most sharply — $\ell$ changes when an atom absorbs a photon; $s$ never changes at all. *Verified: the imported spectrum theorem re-checked numerically (ladder norms vanish exactly at the rungs $m = \pm s$ ; $S^2 = \tfrac34\hbar^2$ for $s = \tfrac12$ ); reproduces Sakurai §3.5 and Griffiths §4.3–4.4.*

Derivation A licenses the half-integers; nature's favourite is the smallest. For $s = \tfrac12$ the ladder has two rungs, the state space is two-dimensional, every operator a $2\times2$ matrix — and the electron's whole spin theory fits in three of them:

\mathbf{S} = \frac{\hbar}{2}\boldsymbol{\sigma}, \qquad \sigma_x = \begin{pmatrix} 0 & 1 \\ 1 & 0 \end{pmatrix},\; \sigma_y = \begin{pmatrix} 0 & -i \\ i & 0 \end{pmatrix},\; \sigma_z = \begin{pmatrix} 1 & 0 \\ 0 & -1 \end{pmatrix} \qquad (4)

where: $\boldsymbol{\sigma}$ — the Pauli matrices; the basis vectors $\binom{1}{0}, \binom{0}{1}$ are spin-up and spin-down along $z$ . This is a preview, not the anatomy — the product rule, completeness, and the why of these matrices are owned by Spin-1/2 and the Pauli Matrices, next in the chapter. Every spin- $\tfrac12$ state is $|\chi\rangle = \cos\tfrac{\theta}{2}|{\uparrow}\rangle + e^{i\phi}\sin\tfrac{\theta}{2}|{\downarrow}\rangle$ — a point on the Bloch sphere, spin-up along some axis; the construction and half-angle geometry are owned by Spinors and Two-Component Wavefunctions.

Figure

The spin ladders — one algebra, one ladder per s

Coupling to the world: the magnetic handle

Spin would be invisible if nothing coupled to it. The handle is magnetic: spin — despite being no motion at all — carries a magnetic moment, with a characteristic multiplier:

\boldsymbol{\mu} = g\,\frac{q}{2m}\,\mathbf{S} = \gamma\,\mathbf{S}, \qquad H = -\boldsymbol{\mu}\cdot\mathbf{B} \qquad (5)

where: $\gamma$ — the gyromagnetic ratio; $g$ — the dimensionless $g$ -factor, $\approx 2$ for electron spin (twice the orbital value; Dirac explained the 2, QED refined it to $2.00232$ ). The electron's natural moment scale is the Bohr magneton $\mu_B = e\hbar/2m_e = 9.274\times10^{-24}\ \mathrm{J/T}$ . The full dictionary of $g$ -factors across particles is owned by Magnetic Moment and the Gyromagnetic Ratio; this article needs only equation (5) and its most famous consequence, next.

Go deeper · Derivation B · 1 minLarmor precession — spin in a magnetic field

Derivation BLarmor precession — spin in a magnetic field

The goal is the equation of motion for a spin in a uniform field $\mathbf{B} = B_0\hat{z}$ : what happens, moment by moment, to the arrow Stern–Gerlach only sorts. The strategy is Heisenberg's — differentiate the operators and let the algebra (1) do all the work; no wavefunction, no representation, three lines. Watch the algebra convert a static Hamiltonian into circular motion: the transverse components feed each other with opposite signs — rotation in disguise.

Step 1. Write the Heisenberg equations. With $H = -\gamma B_0 S_z$ from (5),

\frac{dS_i}{dt} = \frac{1}{i\hbar}[S_i, H] = \frac{-\gamma B_0}{i\hbar}[S_i, S_z].

Why: operator dynamics comes straight from commutators — and we know every commutator by heart from (1).

Step 2. Insert the algebra. $[S_x, S_z] = -i\hbar S_y$ and $[S_y, S_z] = i\hbar S_x$ give

\dot{S}_x = -\gamma B_0\, S_y, \qquad \dot{S}_y = \gamma B_0\, S_x, \qquad \dot{S}_z = 0.

Why: the component along the field is conserved; the transverse pair feed each other — the signature of rotation.

Step 3. Take expectation values and solve:

\frac{d\langle\mathbf{S}\rangle}{dt} = \gamma\,\langle\mathbf{S}\rangle \times \mathbf{B}, \qquad \omega_L = |\gamma|B_0.

Why: the mean spin traces a cone about the field at $\omega_L$ — the classical gyroscopic-magnet equation, delivered by Ehrenfest's theorem; every magnetic-resonance machine is built on this line. The same precession run through the time-evolution operator is the worked exhibit of "The Time Evolution Operator" — an electron in one tesla turns at 28.02 GHz, the two routes agreeing exactly. The general- $j$ vector equation and the frequency dictionary across particles are owned by "Magnetic Moment and the Gyromagnetic Ratio". *Verified: the closed-form solution satisfies both Heisenberg equations symbolically (sympy), and the unitary-evolution route reproduces the cosine numerically for a proton at 1 T; matches Griffiths §4.4.2 and Sakurai §2.1.*

Go deeper · Derivation C · 1 minA full turn is not the identity — the spin-half sign

Derivation CA full turn is not the identity — the spin-half sign

The goal is the strangest fact in this article: rotate a spin- $\tfrac12$ state by $360°$ and it does not return to itself — it returns with a minus sign. The strategy is the cheapest possible: rotate about the quantization axis itself, where the operator is diagonal and the computation is two exponentials read off by eye — no half-angle identities, no matrix algebra. Watch the factor $\tfrac12$ in the exponent: it is the eigenvalue $m_s = \pm\tfrac12$ , and it halves what a full turn delivers.

Step 1. Write the rotation operator about $z$ : angular momentum generates rotations — the grammar established for $L_z$ in Orbital Angular Momentum Operators carries over verbatim — so $R_z(\theta) = \exp(-i\theta S_z/\hbar)$ . Why: the same exponential-of-generator construction as translations under momentum; nothing spin-specific yet.

Step 2. Act on the basis. On the eigenstates $|{\uparrow}\rangle, |{\downarrow}\rangle$ , $S_z$ is just its eigenvalue $\pm\hbar/2$ , so

R_z(\theta)\,|{\uparrow}\rangle = e^{-i\theta/2}\,|{\uparrow}\rangle, \qquad R_z(\theta)\,|{\downarrow}\rangle = e^{+i\theta/2}\,|{\downarrow}\rangle.

Why: diagonal operators exponentiate entry by entry — the whole calculation is here, and the half-angle is nothing but $m_s = \pm\tfrac12$ sitting in the exponent.

Step 3. Set $\theta = 2\pi$ : both phases are $e^{\mp i\pi} = -1$ , so $R_z(2\pi) = -\mathbb{1}$ on every spin- $\tfrac12$ state; only $\theta = 4\pi$ gives $+\mathbb{1}$ . Why: the sign is invisible on a lone particle — a global phase — but physically real: rotate half of a split neutron beam relative to the other half and the fringes shift, restoring only after $720°$ (done in the 1970s, sign confirmed). Why a minus sign is representation-legal, the general theorem $R(2\pi) = (-1)^{2j}$ , and the double-cover anatomy are owned by "Rotation Operators and SU(2)" — this card establishes only the spin-half fact. *Verified: $e^{-i2\pi S_z/\hbar} = -\mathbb{1}$ and $e^{-i4\pi S_z/\hbar} = +\mathbb{1}$ computed exactly on the $2\times2$ matrices; diagonal phases $e^{\mp i\theta/2}$ checked at arbitrary $\theta$ .*

Limiting cases

Case	Result	Meaning
$s = 0$	one state, $\dim = 1$	a scalar under rotations — Higgs boson, pions; no internal arrow at all
$s = \tfrac12$	$\mathbf{S} = \tfrac{\hbar}{2}\boldsymbol{\sigma}$ , $\dim = 2$	every fundamental matter particle; the canonical two-level system, the qubit
$s = 1$	$\dim = 3$	photon (two helicities survive masslessness), W/Z, deuteron
$s \to \infty$	$\Delta m_s/s \to 0$	rungs blur into a continuum: the classical fixed-length vector returns
$\hbar \to 0$ at fixed $\|\mathbf{S}\|$	quantization invisible	discreteness hides when the rung spacing vanishes

Common trap“The electron is a tiny ball, spinning”

The name is a fossil; the picture is wrong twice over. Kinematically: give the electron the largest radius scattering experiments allow and demand its surface store $\hbar/2$ — you need speeds well over a hundred times $c$ . Structurally: spin states are not functions of angles — there is no $\psi(\theta, \phi)$ whose wiggles encode $m_s = \pm\tfrac12$ ; single-valuedness of such a function is exactly what forbids half-integers for orbital motion. Spin lives in its own internal vector space, attached to the particle like a sidecar; the full wavefunction is the two-component object $(\psi_\uparrow(\mathbf{r}), \psi_\downarrow(\mathbf{r}))$ . Keep the cartoon only for what it is good for — a charge with angular momentum has a magnetic moment — and even there it fails: the electron's $g \approx 2$ is twice the classical value, a factor only the Dirac equation explains.

Worked examples

1 · The doubled spectral line (confidence builder)

Sodium's famous yellow emission is not one line but two: $589.0\ \mathrm{nm}$ and $589.6\ \mathrm{nm}$ , the D doublet. The transition is $3p \to 3s$ . What does the doubling say about the electron, and what energy splitting does it measure?

Step 1. Count states without spin. The $3p$ level has $\ell = 1$ , the $3s$ level $\ell = 0$ : orbital quantum numbers alone predict a single transition energy. The observed pair means $3p$ is itself split in two.

Step 2. Add spin. The electron's $s = \tfrac12$ can orient two ways relative to its own orbital motion; the weak coupling shifts the level differently for each, splitting $3p$ into a close pair. The mechanism belongs to the fine-structure articles — here spin does exactly one job: doubling the state count.

Step 3. Convert wavelengths to energies with $hc = 1239.84\ \mathrm{eV\,nm}$ : $E_2 = 1239.84/589.0 = 2.1050\ \mathrm{eV}$ and $E_1 = 1239.84/589.6 = 2.1028\ \mathrm{eV}$ .

Step 4. Subtract: $\Delta E = \mathbf{2.1\ meV}$ — one part in a thousand of the transition energy, resolvable on a bench spectrometer. A $0.6\ \mathrm{nm}$ wavelength gap is direct spectroscopy of the electron's fourth quantum number: the D doublet is spin, photographed through a prism.

2 · Clocking the precession (exam level)

A proton ( $\gamma_p = 2.675\times10^{8}\ \mathrm{s^{-1}\,T^{-1}}$ ) sits in a $1\ \mathrm{T}$ field along $\hat z$ , prepared spin-up along $\hat x$ . (a) How long does one full precession cone take? (b) What is $\langle S_x \rangle$ after $100\ \mathrm{ns}$ ? (c) Why does the electron in the same field precess about 660 times faster?

Step 1. Larmor frequency from Derivation B: $\omega_L = \gamma_p B_0 = 2.675\times10^{8}\ \mathrm{rad/s}$ , i.e. $f_L = \omega_L/2\pi = 42.577\ \mathrm{MHz}$ . One full cone takes $T = 1/f_L = \mathbf{23.5\ ns}$ .

Step 2. The transverse component oscillates at the full Larmor rate: $\langle S_x\rangle(t) = \tfrac{\hbar}{2}\cos\omega_L t$ . After $100\ \mathrm{ns}$ the spin has turned $f_L t = 4.26$ revolutions — $0.26$ past whole — so $\langle S_x\rangle = \tfrac{\hbar}{2}\cos(2\pi\times0.26) \approx \mathbf{-0.03\,\hbar}$ : a quarter-turn short, the spin now almost entirely along $\pm y$ .

Step 3. The ratio. $\omega_L = |\gamma|B$ with $\gamma = gq/2m$ : the electron is 1836 times lighter (with a different $g$ ), netting $f_e/f_p = 28.02\ \mathrm{GHz/T}\,/\,42.58\ \mathrm{MHz/T} \approx \mathbf{658}$ — mass in the denominator is why NMR lives at radio frequencies and electron resonance in microwaves.

Step 4. Fix $B$ , measure $f$ — that is magnetic resonance. At a hospital scanner's $3\ \mathrm{T}$ , protons answer at $3 \times 42.58 = 127.7\ \mathrm{MHz}$ ; the frequency dial is equation (5).

Origins · 1 minGeorge Uhlenbeck and how this was discovered

George Uhlenbeck

1900–1988 · Leiden, then Ann Arbor

Spin was proposed by two graduate students who tried to retract it. In autumn 1925 George Uhlenbeck and Samuel Goudsmit, Ehrenfest's students at Leiden, saw that the alkali doublets and the anomalous Zeeman effect fell into place if the electron itself carried angular momentum $\hbar/2$ with a double-strength moment. Uhlenbeck took the idea to Lorentz — who gently computed that a spinning electron's surface would move far faster than light. Alarmed, the pair asked Ehrenfest to withdraw the paper; he refused, having already mailed it: "You are both young enough to be able to afford a stupidity." The stupidity was the mechanical picture, not the physics. Within months Thomas's relativistic factor repaired the last factor-of-two in the fine structure (a story told at Magnetic Moment and the Gyromagnetic Ratio), and by 1927 Pauli had built spin its matrix formalism. Uhlenbeck's later career was in statistical mechanics — the Ornstein–Uhlenbeck process is his — but the gamble remains the textbook case of trusting data over picture: the electron does not spin, and it has spin.

History

1922

— Stern and Gerlach split a silver beam in two: space quantization is real, and the count is mysteriously even.

1925

— Uhlenbeck and Goudsmit propose intrinsic electron spin ħ/2; Pauli's exclusion principle, published months earlier, quietly needed exactly this fourth quantum number.

1927

— Pauli builds the two-component spinor formalism and his three matrices: spin gets its algebra.

1928

— Dirac's relativistic equation delivers spin and g = 2 unbidden — spin is woven into relativity plus quantum mechanics.

1947

— Kusch measures g slightly above 2; the anomaly becomes the launchpad of quantum electrodynamics.

Going further

Texts. Griffiths & Schroeter, Introduction to Quantum Mechanics, 3rd ed., §4.4 · Sakurai & Napolitano, Modern Quantum Mechanics, Ch. 3 · Shankar, Principles of Quantum Mechanics, Ch. 14 · Cohen-Tannoudji, Diu & Laloë, Vol. 1, Ch. IX.

Provenance. Equations and spectrum cross-checked against Griffiths (3rd ed., Ch. 4.3–4.4), Sakurai & Napolitano (2nd ed., Ch. 3), and Shankar (2nd ed., Ch. 14). A verifier should: (1) confirm the ladder derivation yields $2s\in\mathbb Z_{\ge0}$ and eigenvalues $\hbar^2 s(s+1)$ , $\hbar m_s$ ; (2) verify the Pauli matrices satisfy $[\sigma_i,\sigma_j]=2i\epsilon_{ijk}\sigma_k$ and $\{\sigma_i,\sigma_j\}=2\delta_{ij}\mathbb1$ , so $\mathbf S=\tfrac\hbar2\boldsymbol\sigma$ obeys the spin algebra; (3) recompute the rotation operator using $(\hat{\mathbf n}\cdot\boldsymbol\sigma)^2=\mathbb1$ to get the half-angle form and $R(2\pi)=-\mathbb1$ ; (4) check the Bohr magneton value $\mu_B=e\hbar/2m_e\approx9.274\times10^{-24}\,\mathrm{J/T}$ (CODATA). No nonstandard numbers used; electron $g\approx2$ stated as the Dirac value (QED gives $g\approx2.00232$ ). Edition note (2026-07-08). This article folds the approved v3 pilot into the track as its narrative spine, with the AllQFT base grafted for depth; every spine number was machine re-verified before shipping (all clean — 480 m/s is the beam-weighted most-probable speed $\sqrt{3kT/m}$ at the stated 1000 K oven, and the deflection chain reproduces). The spine's two Stern–Gerlach worked examples are relocated to "The Stern-Gerlach Experiment" (QUANT.ANGM.STERN.0922), which owns the machinery; the spectrum derivation is recast to consume Ladder Operators and the Angular Momentum Spectrum, and the $2\pi$ -sign card is scoped to the spin-half fact, group anatomy deferred to Rotation Operators and SU(2).

Prerequisites. Angular momentum commutation relations · ladder operators and the angular momentum spectrum · orbital angular momentum operators (for the contrast). Leads to. Spin-1/2 and the Pauli matrices · spinors and two-component wavefunctions · the Stern-Gerlach experiment · magnetic moment and the gyromagnetic ratio · rotation operators and SU(2).

If nothing is spinning, why call spin an angular momentum and not just a two-valued label?

Because it enters every conservation law as angular momentum: atomic-transition books balance only if $\mathbf{S}$ is counted alongside $\mathbf{L}$ , and the Einstein–de Haas effect makes it mechanical — flip the spins in an iron rod and the rod starts rotating the other way. Formally "angular momentum" means "generator of rotations," and spin rotates the internal state exactly as $\mathbf{L}$ rotates the wavefunction. Every credential except a spatial picture.

Is s = 1/2 something the electron could be excited out of, the way an atom changes ℓ?

No. $s$ is a species label, fixed like charge and mass: every electron in the universe is $s = \tfrac12$ , and no interaction changes it — an "electron with $s = \tfrac32$ " would be a different particle. Only $m_s$ is dynamical: fields rotate the spin, measurements collapse it, but always within the same two-rung ladder. That is why the spin Hilbert space is exactly two-dimensional and never grows.

Where exactly does the orbital case lose the half-integers if the algebra allows them?

At the representation, not the algebra. Orbital operators act on functions of real angles, so an eigenstate must be $e^{im\phi}$ on the circle — and a half-integer $m$ makes that function contradict itself at a single physical point ( $\phi$ and $\phi + 2\pi$ are the same place). The careful version — including why "just require $|\psi|^2$ single-valued" is not enough — is the business of Eigenfunctions of L_z and the Azimuthal Quantization. Spin states are functions of no angle, so the constraint has nothing to grab.

Does the −1 after a 360° rotation actually do anything, since overall phases are unobservable?

On an isolated particle, nothing — $-|\chi\rangle$ and $|\chi\rangle$ are the same ray. The sign becomes physical the moment it is relative: split a neutron beam, precess one arm through $2\pi$ , recombine — the interference minimum sits where naive periodicity predicts a maximum, recovering only at $4\pi$ (Rauch and Werner groups, 1975). Why half-integer representations may carry this sign and integer ones cannot is owned by Rotation Operators and SU(2).

The arrow picture says I can know |S| and one component. What stops me from measuring Sₓ right after S_z and combining the answers?

In sequence, yes — but the second measurement invalidates the first answer. After $S_z$ returns $+\hbar/2$ the state is $|{\uparrow_z}\rangle$ , an equal superposition of $S_x$ eigenstates; measuring $S_x$ collapses it to (say) $|{\uparrow_x}\rangle$ , itself an equal superposition of $S_z$ eigenstates — the original "up" no longer applies. The beam version is the cascade, its arithmetic worked in Measurement and Wavefunction Collapse.

Half the textbooks write the electron moment as negative. Which sign is right?

Both, if you keep the bookkeeping straight. The electron's charge is $q = -e$ , so $\gamma = gq/2m_e < 0$ : its moment points opposite its spin, and it precesses oppositely to a proton in the same field. Writing $\boldsymbol{\mu} = -g\mu_B\mathbf{S}/\hbar$ with $\mu_B > 0$ makes the sign explicit. Magnitudes — splittings, frequencies, deflections — never care; direction conventions (which beam is up, which way the cone turns) do.