The Paschen-Back and Anomalous Zeeman Regimes

Turn up the field on Zeeman's own sodium lines and the tidy triplet refuses to appear: the $D$ lines break into four and six components at rational-fraction spacings that classical theory forbade — the "anomalous" Zeeman effect that embarrassed atomic physics for a quarter century. Push the field higher still and the multiplets simplify, collapsing back toward a triplet — the Paschen-Back effect. Both are one competition read at opposite ends. Quantum mechanics supplies the machinery: the weak-field fan $E=g_J\mu_B B\,m_J$ is derived in aq-qm-weak-field-zeeman-effect, the projection-theorem Landé factor $g_J$ in aq-qm-lande-g-factor, and the strong-field grid $E=\mu_B B(m_\ell+2m_s)$ in aq-qm-paschen-back-effect. This article owns what a spectroscopist reads off the plate: the observed component patterns, the Preston and Runge bookkeeping rules, the crossover fields in tesla, and the vector-model geometry that makes every count guessable. The whole story turns on one number being different for every level. Because the spin $g$ -factor is $\approx2$ while the orbital one is $1$ , a level's magnetic moment is not parallel to its total angular momentum, and only its projection along $\mathbf J$ survives — the Landé $g_J$ . Upper and lower levels of a transition fan at different $g_J$ , so the line splits into a multiplet whose displacements are fractions of $\mu_B B$ : Runge's 1907 rule, now a two-line consequence of $g_J$ arithmetic. Sodium's crossover sits at an unreachable $36.8$ T, which is exactly why its $D$ lines are the textbook anomalous showcase; hydrogen's $n=2$ crosses at a bench-top $0.78$ T and lithium's $2p$ at $0.72$ T. Its companion, The Zeeman Effect, owns the normal-triplet limit this article generalizes.

Key takeaways

Which regime holds is set by one ratio, $\mu_B B/\Delta E_{fs}$ : weak field keeps $L$ and $S$ locked into $J$ (anomalous Zeeman); strong field tears them apart (Paschen-Back).
Weak-field fan (recalled from aq-qm-weak-field-zeeman-effect): each level splits into $2J+1$ rungs spaced $g_J\mu_B B$ , with $g_J$ the Landé factor of aq-qm-lande-g-factor — $g(^2S_{1/2})=2$ , $g(^2P_{1/2})=\tfrac23$ , $g(^2P_{3/2})=\tfrac43$ .
Because upper and lower $g_J$ differ, sodium's $D_1$ splits into 4 components ( $\pm\tfrac23,\pm\tfrac43$ ) and $D_2$ into 6 ( $\pm\tfrac13,\pm1,\pm\tfrac53$ ), in units of $\mu_B B/h$ — the anomalous fingerprints.
Strong field (recalled from aq-qm-paschen-back-effect): $L$ and $S$ decouple, $E=\mu_B B(m_\ell+2m_s)$ collapses to the ladder $\{-2,-1,0,0,+1,+2\}$ , and transitions regroup into a normal-looking triplet.
Crossover $B_c=\Delta E_{fs}/\mu_B$ : Na $36.8$ T (unreachable), H $n=2$ $0.78$ T, Li $2p$ $0.72$ T — light atoms reach Paschen-Back on a bench.
Key idea: whether spin and orbit stay locked into $J$ (Landé fan, generally not a triplet) or decouple into independent $m_\ell,m_s$ ladders (triplet restored) is decided entirely by $\mu_B B/\Delta E_{fs}$ — the anomalous and Paschen-Back regimes are the two ends of one dial.

The core insight

The character of a magnetic splitting is set by the ratio $\mu_B B/\Delta E_{fs}$ . Weak field: $L$ and $S$ stay locked into $J$ , each level fans with its own Landé $g_J$ , and unequal $g_J$ on the two levels makes the line anomalous — four, six, or more components. Strong field: $L$ and $S$ decouple, $E=\mu_B B(m_\ell+2m_s)$ , and the pattern collapses back to a triplet.

Where this lives today

The two regimes are now working tools rather than puzzles. Anomalous $g_J$ -weighted splittings are read in reverse to assign atomic term symbols and to fingerprint which term produced a stellar line — spectral-line catalogs list $g_J$ beside wavelength, and solar magnetography deliberately picks large- $g_J$ lines for sensitivity and $g_J\approx0$ lines as wavelength references. The Paschen-Back regime is a design choice in cold-atom physics: experiments work above the relevant crossover so that spin flips can be driven with clean, well-separated resonances, and the hyperfine-floor version (Back-Goudsmit) is standard for Feshbach-resonance control at hundreds of gauss. In astrophysics the strong-field regime is a magnetometer at the extreme: white-dwarf photospheres at $10^2$ – $10^5$ T sit deep in Paschen-Back territory, where hydrogen lines are identifiable only through eigenvalue-versus-field fans. The zoo that embarrassed atomic theory for a generation is now a routine label on every spectral line.

The physical picture

Picture spin and orbit as two coupled gyroscopes. In a weak field their mutual spin-orbit coupling wins: they lock into a rigid top $\mathbf J=\mathbf L+\mathbf S$ that leans slowly in the field, and the magnetic moment — tilted off $\mathbf J$ because spin is paid double — has only its $\mathbf J$ -projection survive the fast internal whirl. That surviving fraction is the Landé $g_J$ . Crank the field up and the magnetic torque overwhelms the coupling: $\mathbf L$ and $\mathbf S$ break their lock and precess independently about $\mathbf B$ . The vector model is that cartoon, and although it is only a mnemonic for the projection theorem, it predicts every $g_J$ correctly.

Figure

The vector model in two regimes: in a weak field L and S stay locked into J and only the J-projection of the tilted moment survives, giving g_J; in a strong field L and S decouple and precess independently about B

Now put two fans under one transition. The photon energy is the line center plus the upper rung minus the lower rung; if both fans shared a spacing the differences would collapse to three values — the normal triplet. But the upper and lower $g_J$ differ, so the difference table refuses to collapse and the line spreads into a multiplet. As the field grows past the crossover the two vectors decouple, the moment becomes simply $-(m_\ell+2m_s)\mu_B$ , and the multiplet snaps back toward a triplet. The punchline figure is this whole morph, drawn for the sodium doublet.

Figure

The sodium D lines from anomalous to Paschen-Back, in units of μ_B B/h: D1 is a quartet at ±2/3, ±4/3 and D2 a sextet at ±1/3, ±1, ±5/3; as B climbs past the 36.8 T crossover both collapse to the normal triplet at −1, 0, +1

The core equations

The Landé factor is recalled, not rebuilt: aq-qm-lande-g-factor proves it by projecting the tilted moment onto $\mathbf J$ with the projection theorem, and we quote the closed form and evaluate it for the levels this article needs.

g_J = 1 + \frac{J(J+1)+S(S+1)-L(L+1)}{2J(J+1)} \qquad (1)

where: $L,S,J$ are the orbital, spin and total angular-momentum quantum numbers of the term $^{2S+1}L_J$ . Physical reading: the leading "1" is the orbital baseline; the fraction is spin's excess wage, positive when spin aligns with $\mathbf J$ and negative when it opposes — the sign that sends $g(^2P_{1/2})$ below one to $\tfrac23$ . Each level then fans by first-order perturbation theory, the weak-field result owned by aq-qm-weak-field-zeeman-effect.

E_{m_J} = g_J\,\mu_B B\,m_J,\qquad m_J=-J,\dots,+J \qquad (2)

where: the level splits into $2J+1$ equally spaced rungs of spacing $g_J\mu_B B$ , centroid unmoved. Physical reading: a $j=\tfrac32$ level becomes a quartet spanning $3g_J\mu_B B$ , a $j=\tfrac12$ level a doublet — and because the spacing carries $g_J$ , different levels fan at different rates. A spectral line is the difference of two fans, pruned by the electric-dipole rule. This is where "anomalous" is born.

h\Delta\nu = (g_u m_u - g_l m_l)\,\mu_B B,\qquad m_u-m_l = 0,\pm1 \qquad (3)

where: $g_u,m_u$ and $g_l,m_l$ are the upper and lower level's factor and projection; $\Delta m=0$ gives $\pi$ components, $\Delta m=\pm1$ gives $\sigma^\pm$ . Physical reading: when $g_u\neq g_l$ the difference table does not telescope, so the line splits into more than three components — four for $D_1$ , six for $D_2$ . Above the crossover the field wins, $L$ and $S$ decouple, and the spectrum becomes the regular grid owned by aq-qm-paschen-back-effect.

E = \mu_B B\,(m_\ell + 2m_s) \qquad (4)

where: $m_\ell,m_s$ are now the good projections; the factor 2 is the spin $g$ -factor. Physical reading: for a $p$ electron this is the ladder $\{-2,-1,0,0,+1,+2\}$ , and with $\Delta m_s=0$ the transitions regroup into three $\mu_B B$ -spaced clusters — the normal triplet restored. The dial between the two regimes is a single field scale, and evaluating it is this article's practical skill.

B_c = \frac{\Delta E_{fs}}{\mu_B} \qquad (5)

where: $\Delta E_{fs}$ is the fine-structure splitting of the specific manifold. Physical reading: below $B_c$ the Landé fan holds; above it, Paschen-Back. The scale is manifold-specific — $0.78$ T for hydrogen $n=2$ , $36.8$ T for sodium's $3p$ — so "strong" is never about the magnet alone.

Derivation AThe vector model gives g_J, then the sodium D patterns

The goal is to turn the precessing-gyroscope cartoon into the Landé factor and then use it to count and place the components of the two classic anomalous lines, sodium's $D_1$ and $D_2$ . We recall the rigorous projection-theorem derivation of $g_J$ by name from aq-qm-lande-g-factor — that article proves the tilted moment acts as its $\mathbf J$ -projection — and here evaluate the factor for the levels involved, then build the observed multiplets from the difference rule (3). The census, not the theorem, is what this article owns.

Step 1. Read $g_J$ off equation (1) for the sodium doublet: $g(^2S_{1/2})=2$ ( $L=0$ , pure spin), $g(^2P_{1/2})=\tfrac23$ (spin anti-aligned, so below one), $g(^2P_{3/2})=\tfrac43$ (spin aligned). Why these three: $D_1$ connects $^2P_{1/2}\to{}^2S_{1/2}$ and $D_2$ connects $^2P_{3/2}\to{}^2S_{1/2}$ , so the ground $g=2$ pairs with each excited factor in turn.

Step 2. Build $D_1$ . Upper rungs $g_u m_u=\tfrac23(\pm\tfrac12)=\pm\tfrac13$ ; lower rungs $g_l m_l=2(\pm\tfrac12)=\pm1$ . All four $(m_u,m_l)$ pairs satisfy $|\Delta m|\le1$ , giving shifts $\{\pm\tfrac23,\pm\tfrac43\}$ — four components, no central line, the $\pi$ pair inside at $\pm\tfrac23$ and the $\sigma$ pair outside at $\pm\tfrac43$ . Why no unshifted line: a central component needs $g_u m_u=g_l m_l$ , impossible with unequal $g$ 's and half-integer $m$ 's — the tell that both levels carry spin.

Step 3. Build $D_2$ . Upper rungs $\tfrac43\{\pm\tfrac32,\pm\tfrac12\}=\{\pm2,\pm\tfrac23\}$ ; the rule kills $m_u=\pm\tfrac32\to m_l=\mp\tfrac12$ , leaving six shifts $\{\pm\tfrac13,\pm1,\pm\tfrac53\}$ — the $\pi$ pair at $\pm\tfrac13$ , a $\sigma$ quartet at $\pm1,\pm\tfrac53$ . Why six and not eight: two of the eight $(m_u,m_l)$ combinations demand $\Delta m=\pm2$ , forbidden for electric-dipole radiation.

Step 4. Notice the rational pattern. Every shift in Steps 2–3 is an integer multiple of $\tfrac13\mu_B B$ — Runge's 1907 rule, the mysterious "rational fractions of the normal splitting" that catalogued the patterns before anyone knew what $J$ was. The denominator 3 is $2J(J+1)$ -arithmetic from the two $g$ 's. Why this demystifies a 20-year puzzle: the fractions are nothing but $g_u m_u-g_l m_l$ with $g$ 's of denominator 3. *Verified in verify_aq-optics-paschen-back-anomalous-zeeman.py: $g_J$ by sympy for $^2S_{1/2},{}^2P_{1/2},{}^2P_{3/2}$ ; the $D_1$ sweep giving $\{\pm\tfrac23,\pm\tfrac43\}$ and $D_2$ giving $\{\pm\tfrac13,\pm1,\pm\tfrac53\}$ ; every shift a multiple of $\tfrac13$ .*

Go deeper · Derivation B · 1 minThe strong field restores the triplet

Derivation BThe strong field restores the triplet

The aim is to show why cranking the field up simplifies the spectrum — the backwards result Paschen and Back reported in 1912 — by decoupling $L$ and $S$ and letting the grid $\mu_B B(m_\ell+2m_s)$ take over. The decoupling and the grid energy are owned by aq-qm-paschen-back-effect; what this article owns is the Preston/Runge-style bookkeeping that turns the grid into three observed lines. We evaluate the $p$ -electron ladder and prune it with the dipole rule.

Step 1. In the strong-field basis $|m_\ell,m_s\rangle$ the Zeeman term is diagonal, $E=\mu_B B(m_\ell+2m_s)$ . For a $p$ electron ( $m_\ell\in\{-1,0,1\}$ , $m_s=\pm\tfrac12$ ) this is the ladder $\{-2,-1,0,0,+1,+2\}$ — five distinct rungs, with two states sharing the middle. Why the middle is doubled: $(m_\ell,m_s)=(-1,+\tfrac12)$ and $(+1,-\tfrac12)$ both give $m_\ell+2m_s=0$ , a degeneracy the leading grid cannot lift.

Step 2. Transitions obey $\Delta m_\ell=0,\pm1$ with $\Delta m_s=0$ , because the photon's electric field does not touch the spin. The photon shift is therefore $\mu_B B\,\Delta m_\ell$ , taking only the values $\{0,\pm1\}\mu_B B$ . Why spin drops out: with $m_s$ fixed across the transition, the $2m_s$ term cancels in the difference — only the orbital step survives.

Step 3. Three clusters emerge at $0,\pm\mu_B B$ — a triplet indistinguishable, at this order, from the normal Zeeman effect of the companion article. The residual fine structure $\propto m_\ell m_s$ splits each cluster into a fine substructure, but the gross pattern is three lines. Why "anomalous" becomes "normal" again: the field has bought out the coupling that made $g_J\neq1$ , so every effective $g$ returns to 1. *Verified in verify_aq-optics-paschen-back-anomalous-zeeman.py: the ladder $\{-2,-1,0,0,+1,+2\}$ from $m_\ell+2m_s$ ; the pruned photon-shift set $\{0,\pm1\}$ under $\Delta m_\ell=0,\pm1$ , $\Delta m_s=0$ .*

Go deeper · Derivation C · 1 minThe crossover field, and why sodium never gets there

Derivation CThe crossover field, and why sodium never gets there

The goal is to convert the regime ratio (5) into tesla for three manifolds and draw the lesson that "strong field" is a property of the atom, not the magnet. Sodium's fine structure is so large that its crossover is beyond any static laboratory field, keeping its $D$ lines permanently anomalous, while hydrogen and lithium cross over on a bench. The exact interpolation across the crossover is owned by aq-qm-paschen-back-effect; we compute only the boundary field.

Step 1. Sodium $3p$ . The $D$ -doublet wavelengths $588.995/589.592$ nm give $\Delta E_{fs}=2.13\times10^{-3}$ eV, so $B_c=\Delta E_{fs}/\mu_B=36.8$ T. Why unreachable: $36.8$ T exceeds the strongest continuous magnets, so every tabletop sodium experiment is deeply weak-field — which is why $D_1/D_2$ are the canonical anomalous showcase.

Step 2. Hydrogen $n=2$ . The $2p_{3/2}$ – $2p_{1/2}$ gap is $\Delta E_{fs}=4.53\times10^{-5}$ eV, giving $B_c=0.78$ T. Why so low: fine structure scales as $\alpha^2\mathrm{Ry}\sim10^{-4}$ of the gross spectrum and falls with $n$ , so hydrogen's internal field is easy for a solenoid to beat.

Step 3. Lithium $2p$ . With a measured fine-structure splitting of $0.337$ cm $^{-1}$ , $B_c=0.72$ T — comparable to hydrogen, again bench-top. Why light alkalis join hydrogen: small $Z$ means small spin-orbit coupling, so the field wins early.

Step 4. Between the limits neither $(J,m_J)$ nor $(m_\ell,m_s)$ is exact; only $m_J=m_\ell+m_s$ survives, and same- $m_J$ branches bend away from each other — avoided crossings whose closed-form, Breit-Rabi-like shape is worked in aq-qm-paschen-back-effect. Why the honest region needs matrices: the two competing couplings share no basis, so the crossover is a genuine diagonalization, not either limiting formula. *Verified in verify_aq-optics-paschen-back-anomalous-zeeman.py: $B_c(\mathrm{Na}\,3p)=36.8$ T from the $D$ wavelengths; $B_c(\mathrm{H}\,n{=}2)=0.782$ T; $B_c(\mathrm{Li}\,2p)=0.72$ T from $0.337$ cm $^{-1}$ .*

Limiting cases

case	result	meaning
$\mu_B B\ll\Delta E_{fs}$	$E=g_J\mu_B B\,m_J$	weak field: $J$ good, anomalous fan set by Landé $g_J$
$\mu_B B\gg\Delta E_{fs}$	$E=\mu_B B(m_\ell+2m_s)$	strong field: $L,S$ decoupled, triplet restored (Paschen-Back)
$S=0$	$g_J=1$ , all fields	singlet: the normal triplet of The Zeeman Effect at any field
$g_J=0$ (e.g. $^4D_{1/2}$ )	no first-order shift	magnetically silent level — immune to a weak field
$\mu_B B\sim\Delta E_{fs}$	avoided crossings	intermediate regime; exact shape owned by aq-qm-paschen-back-effect
$\mu_B B\sim A_{hf}$	nuclear decoupling	hyperfine analogue (Back-Goudsmit); teased in Hyperfine Structure

Common trapThe anomalous Zeeman effect is a rare, exotic special case

It is the exact opposite — the anomalous pattern is the generic case and the normal triplet is the exception. Any level with unpaired spin has $g_J\neq1$ , so its lines split into four, six, or more components; that describes hydrogen, every alkali, and most of the periodic table. The "normal" triplet needs a spin-singlet ( $S=0$ ) transition, which only paired-spin configurations supply. The word "anomalous" is a fossil of the years 1898–1925, when the multiplets were an unexplained scandal and the triplet was all classical theory could produce. Two further traps: the vector model is a mnemonic for the projection theorem of aq-qm-lande-g-factor, not a literal spinning gyroscope; and a magnetic field being "strong" is manifold-relative — $2$ T is strong for hydrogen $n=2$ but weak for sodium, so always compute $B_c=\Delta E_{fs}/\mu_B$ before choosing a formula.

Worked examples

Example 1 — resolving the sodium D₁ anomalous pattern at 1 tesla. Place the $D_1$ line ( $589.6$ nm, $^2P_{1/2}\to{}^2S_{1/2}$ ) in a $1$ T field. Where are its four components, and what étalon resolves them?

Step 1. Component offsets from Derivation A: $\{\pm\tfrac23,\pm\tfrac43\}$ in units of $\mu_B B/h=13.996$ GHz. So the $\pi$ pair sits at $\pm9.33$ GHz and the $\sigma$ pair at $\pm18.66$ GHz.

Step 2. Convert to wavelength at $589.6$ nm, $\Delta\lambda=\lambda^2\Delta\nu/c$ : the $\pi$ components fall at $\pm10.8$ pm and the $\sigma$ components at $\pm21.6$ pm from line center.

Step 3. The components sit at $-21.6,-10.8,+10.8,+21.6$ pm, so the tightest gap between adjacent lines (a $\sigma$ and a $\pi$ ) is $\Delta\lambda_{\min}=10.8$ pm, needing $R=\lambda/\Delta\lambda_{\min}=589.6/0.0108\approx5.5\times10^4$ . Four components at $\pm10.8$ and $\pm21.6$ pm ( $\pm9.33$ and $\pm18.66$ GHz), no central line — a Fabry-Pérot étalon with $R\gtrsim5.5\times10^4$ separates them, and the missing central line immediately signals that both levels carry spin.

Example 2 — a level a magnetic field cannot touch. Compare the responses of a $^2P_{3/2}$ level and a $^4D_{1/2}$ level at $1$ T.

Step 1. $^2P_{3/2}$ ( $L=1,S=\tfrac12,J=\tfrac32$ ): $g_J=\tfrac43$ , so four rungs at $g_J\mu_B B\,m_J=\pm\tfrac43\times57.88\ \mu\mathrm{eV}\times\{\tfrac12,\tfrac32\}=\pm38.6,\pm115.8$ $\mu$ eV — a clean quartet.

Step 2. $^4D_{1/2}$ ( $L=2,S=\tfrac32,J=\tfrac12$ ): $g_J=1+\dfrac{\tfrac34+\tfrac{15}{4}-6}{2\cdot\tfrac34}=1+\dfrac{-\tfrac32}{\tfrac32}=0$ exactly. Both $m_J=\pm\tfrac12$ states shift by $g_J\mu_B B\,m_J=0$ : the level ignores the field at first order.

Step 3. The vanishing is not absolute — with the QED spin factor $g_s=2.0023$ (recalled from aq-qm-lande-g-factor) the zero moves to $g_J=2-g_s=-0.0023$ , a residual $575$ times below the $^2P_{3/2}$ rung. The $^2P_{3/2}$ quartet fans at $\pm38.6,\pm115.8$ $\mu$ eV while the $^4D_{1/2}$ level shows no first-order splitting at all — a magnetically silent level, the kind clock builders hunt for as a field-insensitive transition.

Physics of the portrait

The vector model is a picture, not a law — two tops precessing about their resultant — but for twenty years it was the only thing that organized the anomalous zoo, and it delivered the exact $g_J$ formula years before quantum mechanics could justify it.

Origins · 2 minvector model and how this was discovered

vector model

the precessing-tops picture behind the Landé factor · Preston → Runge → Landé → Paschen & Back → the 1925 spin resolution

The vector model grew out of a crisis. In 1898 Thomas Preston, in Dublin, photographed sodium lines splitting into four and six components — patterns the Lorentz triplet theory could not produce at any setting — and stated the rule that lines of one spectral series split alike. In 1907 Carl Runge added that every displacement was a rational fraction of the normal splitting, organizing the zoo without explaining a line of it. The decisive figure was Alfred Landé at Tübingen, who in 1921 assigned each level an empirical factor $g$ and found that a single formula — with half-integer quantum numbers, scandalous at the time — reproduced every catalogued pattern; the model pictured $\mathbf L$ and $\mathbf S$ as vectors precessing about their sum $\mathbf J$ , with $g_J$ the projection. From the other end, Friedrich Paschen and Ernst Back pushed spectroscopy to several tesla and in 1912 reported the backwards result that the anomalous multiplets simplified toward a triplet at high field — fine structure was a coupling that could be broken — mapping the "transformation" systematically by 1921. Pauli's Hamburg lament, that one could not look happy while thinking about the anomalous Zeeman effect, dates from these years. The resolution came in 1925, when Goudsmit and Uhlenbeck's spinning electron with $g_s=2$ turned Landé's fitted fractions into projection arithmetic, and in 1926 Heisenberg and Jordan derived the patterns from the new quantum mechanics. The vector model survived as a mnemonic for the projection theorem — a gyroscope that is not really spinning, yet computes the right answer every time.

History

1898

— Preston photographs the sodium quartets and sextets and states his rule

1907

— Runge systematizes the patterns: every displacement is a rational fraction of the normal splitting

1912

— Paschen and Back reach strong fields and watch the anomalous multiplets collapse toward a triplet

1921

— Landé fits his g-factor formula from Zeeman data, half-integer quantum numbers and all

1921

— Paschen and Back publish their systematic strong-field study of the transformation

1925

— Uhlenbeck and Goudsmit's spinning electron (g_s = 2) makes the Landé g_J physical

1926

— Heisenberg and Jordan derive the anomalous patterns from the new quantum mechanics with spin

Going further

Texts. Foot, Atomic Physics, Ch. 5 (weak and strong field, the crossover) · Demtröder, Laser Spectroscopy, Vol. 1, Zeeman and Paschen-Back sections · Bransden & Joachain, Physics of Atoms and Molecules, Ch. on atoms in fields · Condon & Shortley, The Theory of Atomic Spectra, Ch. 16 (the classic pattern catalogs, Runge denominators and all).

Provenance. The Landé $g_J$ formula and its projection-theorem derivation are recalled by name from aq-qm-lande-g-factor; the weak-field fan $E=g_J\mu_B B m_J$ from aq-qm-weak-field-zeeman-effect; the strong-field grid $E=\mu_B B(m_\ell+2m_s)$ and the exact avoided-crossing interpolation from aq-qm-paschen-back-effect. A verifier should confirm: (1) $g_J$ for $^2S_{1/2}=2$ , $^2P_{1/2}=\tfrac23$ , $^2P_{3/2}=\tfrac43$ , $^3P_1=\tfrac32$ , $^1P_1=1$ , $^4D_{1/2}=0$ ; (2) the $D_1$ census $\{\pm\tfrac23,\pm\tfrac43\}$ and $D_2$ census $\{\pm\tfrac13,\pm1,\pm\tfrac53\}$ ; (3) the crossover fields Na $36.8$ T, H $n=2$ $0.78$ T, Li $2p$ $0.72$ T; (4) the strong-field ladder $\{-2,-1,0,+1,+2\}$ . A base-article correction is noted below: the dry base wrote its all-fields crossover formula as $-\tfrac{A}{4}\pm\tfrac{A}{2}\sqrt{1+2x+x^2}$ , but $\sqrt{1+2x+x^2}=1+x$ identically, collapsing the very anticrossing it describes into two straight lines — the correct nonlinear Breit-Rabi-like form (with an $m_J$ -weighted radicand) is recalled by name from aq-qm-paschen-back-effect rather than reproduced.

Prerequisites. The Zeeman Effect · Fine Structure and Spin-Orbit Coupling · Electron Spin and the Spin Magnetic Moment. Leads to. Hyperfine Structure · The Stark Effect.

Why is it called "anomalous" if it is actually the normal state of affairs?

Pure historical irony. In the years after 1898 the only Zeeman pattern classical theory could explain was the three-line triplet Lorentz derived for a spinless oscillator. Real lines with unpaired spin split into four, six, or more components at spacings the theory forbade, so they were labeled "anomalous" — a polite word for "we have no idea." When electron spin arrived in 1925 the fractions turned out to be routine $g_J$ arithmetic, and the "anomalous" case was revealed as the generic one: any level with spin has $g_J\neq1$ . By then the name had stuck. So "anomalous Zeeman effect" means the ordinary magnetic splitting of spin-carrying levels, and "normal" means the rare singlet triplet — the terminology is exactly backwards from the physics.

Is the vector model a real physical picture or just a memory aid?

It is a mnemonic for a theorem. Nothing is literally precessing in an energy eigenstate; " $\mathbf L$ and $\mathbf S$ whirl about $\mathbf J$ and only the projection survives" is a cartoon. The honest statement is the projection theorem of aq-qm-lande-g-factor: within a fixed- $J$ level, every vector operator's matrix elements are proportional to those of $\mathbf J$ , and the proportionality constant is exactly what the cartoon computes. The reason to keep the picture is that it gets $g_J$ right every time and lets you guess a pattern in ten seconds. Use the gyroscopes to remember; cite the projection theorem when someone asks why it works.

Why does a strong field make the spectrum simpler instead of more complex?

Because a strong field buys out the coupling that caused the complexity. The anomalous richness comes from spin and orbit being locked into $\mathbf J$ with a moment that mis-tracks it; once $\mu_B B$ exceeds the spin-orbit energy, the field tears $\mathbf L$ and $\mathbf S$ apart and each simply aligns to the field. The energy becomes the regular grid $\mu_B B(m_\ell+2m_s)$ , and since the photon does not flip spin ( $\Delta m_s=0$ ), the transitions regroup into three clusters spaced by $\mu_B B$ — a triplet. Paschen and Back found this collapse in 1912 and it looked backwards, but it is the cleanest proof that fine structure is a breakable coupling, not a fixed property of the level.

Why can we reach hydrogen's Paschen-Back regime but not sodium's?

Because the crossover field is set by the atom's own fine structure, which differs enormously between them. Hydrogen's $n=2$ fine-structure splitting is only $4.5\times10^{-5}$ eV, so $B_c=\Delta E_{fs}/\mu_B=0.78$ T — a superconducting solenoid clears it. Sodium's $3p$ splitting is fifty times larger ( $2.1\times10^{-3}$ eV), giving $B_c=36.8$ T, beyond any static laboratory magnet. Fine structure grows steeply with nuclear charge (roughly as $Z^4$ for a given series), so light atoms cross over on a bench while heavier ones stay weak-field to absurd fields. "Strong" always means strong relative to this manifold's internal magnetism.

What happens right at the crossover, where neither limit applies?

Neither $(J,m_J)$ nor $(m_\ell,m_s)$ is an exact label; the only survivor is $m_J=m_\ell+m_s$ , so the problem block-diagonalizes by $m_J$ into small matrices that must be solved exactly. The result is a set of energy-versus-field curves in which branches of the same $m_J$ bend away from one another — avoided crossings — smoothly connecting the low-field Landé fan to the high-field grid. The closed-form, Breit-Rabi-like expression for these curves is worked in aq-qm-paschen-back-effect; this article only locates the field where you must switch from one limiting formula to the honest diagonalization.

Can the Landé g-factor really be zero or negative?

Yes, and it is not a curiosity of one term. Solving $g_J=0$ in the Landé formula gives an exact family — $^4D_{1/2}$ , $^5F_1$ , $^6G_{3/2}$ , and so on — where anti-aligned spin's double wage exactly cancels the orbital contribution, leaving a magnetically silent level that a weak field does not split at first order. Push the spin further against a small $J$ and $g_J$ goes negative ( $^6F_{1/2}=-\tfrac23$ ), inverting the fan. The folklore that $g_J$ lives between $\tfrac23$ and $2$ holds only for one-electron doublets; the full catalogue, including the zeros and negatives, is anatomized in aq-qm-lande-g-factor. Silent levels are precisely what clock and qubit designers want for field-insensitive transitions. The same competition also replays a thousand times weaker at the hyperfine scale — a $g_F$ fan decoupling into the Back-Goudsmit regime at millitesla — owned by Hyperfine Structure.