The Lamb Shift

Pay every relativistic correction hydrogen's Dirac equation demands — spin-orbit, relativistic mass, the Darwin contact term — and two levels of the $n=2$ shell come to rest at exactly the same energy: $2S_{1/2}$ and $2P_{1/2}$ , which share $n=2$ and $j=\tfrac12$ . The exact Dirac–Coulomb spectrum depends on $n$ and $j$ alone, never on $\ell$ , so the coincidence is not an approximation waiting to be improved — it is degenerate to every order in $\alpha$ . Then the levels are found split, by $1057.8$ megahertz. Quantum mechanics builds the reason: aq-qm-lamb-shift derives the bound electron's self-energy, the zero-point vacuum fluctuations of the quantized field, and the mass renormalization that turns a divergent integral into a finite number; here we do not rebuild that theory, we recall it by name and own the other half. The other half is spectroscopy. Optics owns the 1947 experiment that found a splitting where relativity forbade one, the microwave method that resolved a $10^{-6}$ -relative interval when optical spectrographs could not, and the precision chain that grew out of it — the $1S$ – $2S$ two-photon standard, the optical frequency comb, and the muonic-hydrogen measurement that now weighs the proton. The lever throughout is one atomic accident: the $2S$ state cannot decay the easy way, so it lives an eighth of a second while its $2P$ neighbour dies in nanoseconds — the eight-order-of-magnitude gulf that let Lamb and Retherford fly a beam of metastable atoms, quench it with radio waves, and read the vacuum's fingerprint off a detector current. The theory says the vacuum is not empty; the experiment is how we know.

Key takeaways

The exact Dirac–Coulomb spectrum depends only on $n$ and $j$ , so $2S_{1/2}$ and $2P_{1/2}$ must be degenerate; the measured Lamb shift puts $2S_{1/2}$ about $1057.845$ MHz ( $4.375\ \mu$ eV, $0.0353\ \mathrm{cm}^{-1}$ ) above $2P_{1/2}$ .
The mechanism is quantum-electrodynamic and is owned by aq-qm-lamb-shift — the electron self-energy, vacuum fluctuations, and the mass renormalization that makes the shift finite; here we recall it and evaluate its spectroscopic consequences.
Optics owns the experiment: Lamb and Retherford (1947) prepared a metastable $2S$ beam, drove $2S\to 2P$ with microwaves, and read the transition as a quench dip in a beam that survives only while excited.
The enabling lever is metastability — $\tau(2S)\approx0.12$ s against $\tau(2P)=1.6$ ns, a ratio near $10^8$ — plus Zeeman tuning to walk a fixed-frequency resonance down to zero field.
The shift anchors metrology: the $1S$ shift is $8172$ MHz, the $1S$ – $2S$ two-photon transition ( $2.4661$ PHz, $243.1$ nm) fixes the Rydberg constant, and in muonic hydrogen the contact factor grows $6.4\times10^{6}$ -fold to weigh the proton.
Key idea: the Lamb shift is the vacuum's fingerprint on a spectrum — a splitting the Dirac equation forbids, found by driving a forbidden microwave interval inside a beam that announces its own survival.

The core insight

The Dirac–Coulomb spectrum leaves $2S_{1/2}$ and $2P_{1/2}$ exactly degenerate because it depends only on $n$ and $j$ . Coupling the bound electron to the quantized radiation field — the self-energy and vacuum fluctuations owned by aq-qm-lamb-shift — lifts $2S_{1/2}$ by $1057.845$ MHz, and because that shift is carried by the contact density $|\psi(0)|^2$ , only penetrating $S$ -states move.

Where this lives today

The Lamb shift is now an instrument of metrology rather than a curiosity of hydrogen. Frequency-comb spectroscopy of the $1S$ – $2S$ line at $10^{-15}$ relative precision tests bound-state QED exactly as well as the $1S$ Lamb shift — $8172$ MHz, computed through two-loop and recoil terms — is known; inverting the same equations extracts the Rydberg constant and the proton charge radius. The muonic-hydrogen programme made that inversion dramatic: with the contact factor amplified $6.4\times10^{6}$ -fold, the 2010 measurement returned a proton radius several standard deviations from the accepted value, and a decade of re-measurement ended with the muon vindicated and the recommended radius revised. At the strong-field frontier, the $1S$ Lamb shift in hydrogen-like uranium reaches hundreds of eV and is measured against all-order QED in heavy-ion storage rings, where the $Z\alpha$ expansion itself breaks down.

The physical picture

Hold the "there should be nothing here" note firmly first. The fine-structure ladder of Fine Structure and Spin-Orbit Coupling was checked against the exact Dirac energy, whose formula contains $n$ and $j+\tfrac12$ and nothing else. Expand it to any order in $\alpha$ and $2S_{1/2}$ , $2P_{1/2}$ ride together — not the fragile degeneracy of the Bohr levels, but an algebraic identity. Finding a splitting there is finding daylight between two sides of an equation, so the equation must be missing physics: the quantized field. Quantize the electromagnetic field and every mode keeps a zero-point half-quantum, even in empty space. A bound electron is bathed in these fluctuations and jitters over a tiny cloud, a fraction of $10^{-5}$ of a Bohr radius. A smeared charge feels a diluted Coulomb pit; the dilution acts only where the potential is singular — at the nucleus — so only $S$ -states, the ones with amplitude at the origin, are moved, and moved up. That whole chain — the smearing, the contact evaluation, the mass renormalization that makes it finite — is derived in aq-qm-lamb-shift. What the spectrum shows is the outcome: a level lifted off a perch relativity swore was permanent.

Figure

Dirac ties 2S₁/₂ and 2P₁/₂ together; QED lifts 2S₁/₂ by 1057.8 MHz — about a tenth of the 10 949 MHz fine structure

Now the second figure is the experiment itself. The whole 1947 design turns on a beam that reports its own excitation, a resonance read as a dip in a detector current, and a magnetic field used not as a contaminant to be shielded away but as a dial. Ride the resonance across several fields and extrapolate to zero, and the intercept is the field-free interval.

Figure

The Lamb–Retherford method: a metastable beam, a quench dip, and the resonance extrapolated to B = 0 → 1057.8 MHz

The core equations

The backdrop is the exact relativistic hydrogen energy, recalled from Fine Structure and Spin-Orbit Coupling. Its structure is the entire reason a Lamb shift is a surprise: two quantum numbers enter, and $\ell$ is not one of them.

E_{n,j} = -\frac{\mathrm{Ry}}{n^{2}}\left[\,1 + \frac{\alpha^{2}}{n^{2}}\left(\frac{n}{\,j+\tfrac12\,} - \frac34\right)\,\right] + \dots \qquad (1)

where: $n$ — principal quantum number; $j$ — total angular momentum; $\alpha\approx1/137.036$ — fine-structure constant; $\mathrm{Ry}=13.6057\ \mathrm{eV}$ . Read it as a filter: any two states with the same $(n,j)$ are handed identical energies, so $2S_{1/2}$ and $2P_{1/2}$ coincide to all orders — no relativistic correction inside this formula can pull them apart. The measurement disagrees, and the disagreement is the whole subject. Expressed four ways, the field-free interval is a single small number that a spectroscopist must hit.

\Delta E(2S_{1/2}-2P_{1/2}) = h\times 1057.845\ \mathrm{MHz} = 4.375\ \mu\mathrm{eV} = 0.0353\ \mathrm{cm}^{-1} \qquad (2)

where: $2S_{1/2}$ lies above $2P_{1/2}$ ; the interval is one part in $10^{6}$ of the $n=2$ binding energy, about $9.66\%$ of the $10\,949$ MHz $2P$ fine structure, and a $28$ -centimetre microwave photon — hardware territory, which is exactly how it was first reached. Why it is nonzero belongs to QED, and aq-qm-lamb-shift owns that derivation. What a spectrum-writer needs is its anatomy: a dominant electron self-energy and a smaller vacuum-polarization term of opposite sign.

\Delta E \;\approx\; \underbrace{+1085}_{\text{self-energy}} \;\underbrace{-27}_{\text{vacuum pol.}} \;+\;\dots\;\approx\; 1058\ \mathrm{MHz}, \qquad \Delta E \sim \frac{Z^{4}\alpha^{5}}{n^{3}}\,m_ec^{2}\ln\frac{1}{(Z\alpha)^{2}} \qquad (3)

where: the self-energy (electron emitting and reabsorbing virtual photons) is the large positive piece; the vacuum-polarization Uehling term screens the nucleus with opposite sign; both mechanisms, and the mass renormalization that renders the sum finite, are derived in aq-qm-lamb-shift. The scaling carries one power of $\alpha$ beyond fine structure's $\alpha^{4}$ — a single photon loop — plus a logarithm and a steep $Z^{4}$ growth that makes heavy hydrogen-like ions strong-field QED laboratories. Because the shift is weighted by the contact density $|\psi_{ns}(0)|^{2}\propto n^{-3}$ , the lower $S$ -state is moved more. This fixes the $1S$ shift and the ratio that a naive $1/n^{3}$ would miss.

\frac{\Delta E(1S)}{\Delta E(2S)} = \frac{8172\ \mathrm{MHz}}{1057.8\ \mathrm{MHz}} = 7.7 \qquad (4)

where: the naive geometric factor is $2^{3}=8$ ; the measured ratio is $7.7$ because the state-dependent Bethe logarithm softens the pure $n^{-3}$ law. The $1S$ shift, $8172$ MHz, is the number modern comb spectroscopy actually tests. Finally the interval that turned the Lamb shift from a curiosity into a metrological anchor: the two-photon $1S$ – $2S$ resonance, driven Doppler-free at a wavelength half its own.

\nu(1S\text{–}2S) = 2.4661\ \mathrm{PHz}, \qquad \lambda_{\text{drive}} = \frac{c}{\nu/2} = 243.1\ \mathrm{nm} \qquad (5)

where: $\nu$ — the $1S$ – $2S$ transition frequency, known to $10^{-15}$ ; $\lambda_{\text{drive}}$ — the wavelength of each of two counter-propagating photons. Two photons at $243.1$ nm add to the transition; counter-propagation cancels first-order Doppler shift, and the $1S$ Lamb shift is what the extracted Rydberg and proton radius must be corrected for.

Derivation AWhy the 2S state lives long enough to fly

The entire 1947 measurement rests on one atomic accident: the $2S_{1/2}$ state cannot decay the fast way. Electric-dipole radiation, the only fast channel at optical energies, is forbidden between $2S$ and $1S$ , so a $2S$ atom finds a far slower escape and survives long enough to be launched down a metre of vacuum and interrogated. This card assembles the lifetimes and the flight arithmetic that make a beam of excited atoms possible at all — the precondition for everything else.

Step 1. The forbidden channel. One-photon electric-dipole (E1) decay requires $\Delta\ell=\pm1$ ; the transition $2S\to1S$ has $\Delta\ell=0$ and is E1-forbidden, while higher multipoles between these particular states are vanishingly weak. Why this matters: the fast optical route that empties the $2P$ state in nanoseconds is simply closed to $2S$ — the dipole selection rule $\Delta\ell=\pm1$ leaves it stranded.

Step 2. The slow escape. The $2S$ state decays instead by emitting two photons at once, a second-order process with rate $\approx 8.23\ \mathrm{s}^{-1}$ , so $\tau(2S)\approx 0.12\ \mathrm{s}$ — about an eighth of a second. Why it is so slow: two-photon emission is suppressed by extra powers of $\alpha$ and by phase space, which is precisely what buys the experiment its time.

Step 3. The contrast. The $2P$ state decays by E1 to $1S$ (the Lyman- $\alpha$ line) with rate $6.26\times10^{8}\ \mathrm{s}^{-1}$ , so $\tau(2P)=1.6\ \mathrm{ns}$ . The ratio of lifetimes is $\tau(2S)/\tau(2P)\approx 7.6\times10^{7}$ — near $10^{8}$ . Why it is the design: any $2P$ contamination self-cleans in millimetres, while $2S$ persists, so the surviving beam is purely metastable.

Step 4. The flight. A $2500$ K oven gives $v\approx\sqrt{3k_BT/m}\approx 7.9\ \mathrm{km/s}$ ; a $10$ cm flight takes $13\ \mu\mathrm{s}$ , about $8000$ $2P$ -lifetimes but only $10^{-4}$ of a $2S$ -lifetime. Why the numbers close: the $2P$ population dies $e^{-8000}$ while $2S$ loses one part in $10^{4}$ , so the detector sees only atoms still carrying $10.2$ eV of excitation. *Verified in verify_aq-optics-lamb-shift.py.*

Go deeper · Derivation B · 2 minThe microwave method: a quench dip ridden to zero field

Derivation BThe microwave method: a quench dip ridden to zero field

The obstacle in 1947 was resolution, not sensitivity: the splitting is a fraction of the Doppler width of any optical hydrogen line, so a spectrograph could never separate $2S_{1/2}$ from $2P_{1/2}$ inside a Balmer or Lyman line. The breakthrough was to abandon light as the carrier, drive the roughly one-gigahertz interval directly with microwaves, and read the transition through a detector that counts excitation rather than photons. This card sets out why optics fails, why microwaves win, and how a magnetic field became the tuning dial.

Step 1. The interval in spectroscopic units. $1057.845\ \mathrm{MHz} = 0.0353\ \mathrm{cm}^{-1} = 4.375\ \mu\mathrm{eV}$ , which is $9.66\%$ of the $2P$ fine-structure interval. Why the fraction matters: the Lamb shift is a tenth of a splitting that itself is only $10^{-5}$ of the binding energy, so it hides inside optical line structure.

Step 2. The optical wall. The Doppler full width of H- $\alpha$ at $300$ K is $\nu_0\sqrt{8\ln2\,k_BT/mc^{2}}\approx 5.6\ \mathrm{GHz}$ , five times the shift; resolving $1.06$ GHz on the H- $\alpha$ carrier needs a resolving power near $4\times10^{5}$ , and on Lyman- $\alpha$ near $2\times10^{6}$ , together with a sub-Doppler technique that did not exist. Why it was hopeless: no pre-laser spectrograph could both resolve one part in $10^{6}$ and beat the thermal broadening at once.

Step 3. The microwave key and the quench detector. Drive the $\sim1$ GHz interval directly; on resonance a $2S$ atom is promoted to $2P$ and decays in $1.6$ ns before reaching the tungsten plate, so the metastable current dips. Why the readout inverts the problem: instead of resolving a faint emission line, one counts atoms in a binary "still excited?" state, and resonance is a hole in a background.

Step 4. The Zeeman dial. The 1947 oscillators sat at fixed frequencies, so the levels were tuned instead by sweeping a magnetic field; plotting the resonant field for several oscillator settings and extrapolating to $B=0$ returned the field-free interval, roughly $1000$ MHz at first and $1057.77\pm0.10$ MHz by the 1953 series. Why extrapolation beats shielding: a measured field with a known level slope turns the stray-field contaminant into the controlled independent variable. *Verified in verify_aq-optics-lamb-shift.py.*

Go deeper · Derivation C · 1 minThe 1S–2S two-photon chain and the proton-radius payoff

Derivation CThe 1S–2S two-photon chain and the proton-radius payoff

The Lamb shift's second life is metrology, and the instrument is the $1S$ – $2S$ transition — narrow because $1S$ is the ground state and $2S$ metastable, and reachable Doppler-free because two counter-propagating photons share the energy. This card shows how first-order Doppler cancels, sizes the residual that comb spectroscopy fights, and explains why a muon converts the same physics into a caliper for the proton.

Step 1. Doppler cancellation by geometry. Absorbing one photon from each of two counter-propagating beams, an atom moving at velocity $v$ sees frequencies shifted by $+kv$ and $-kv$ ; the sum is independent of $v$ to first order, so the whole thermal ensemble contributes to one sharp line. Why it is the trick: the linear Doppler shift, the killer of Step 2 in the previous card, is arranged to vanish rather than be fought.

Step 2. The numbers. The transition sits at $\nu = 2.4661\ \mathrm{PHz}$ ; each of the two photons carries half, at $\lambda_{\text{drive}} = 243.1$ nm. Why the wavelength is convenient: a single ultraviolet source at $243$ nm, retro-reflected, drives the resonance with the Doppler geometry built in.

Step 3. The residual. Second-order Doppler survives at $\Delta\nu/\nu = v^{2}/2c^{2}$ ; at $v=300\ \mathrm{m/s}$ this is $5.0\times10^{-13}$ , an absolute $\approx1.2$ kHz on a $2.47$ PHz line. Why it sets the frontier: this relativistic time-dilation term, not the first-order shift, is now the leading kinematic systematic, tamed by cooling and by extrapolating to zero velocity.

Step 4. The muonic amplification. Replace the electron by a muon and the orbit shrinks with the reduced mass, $186\,m_e$ ; the contact density $|\psi(0)|^{2}$ grows by $(186)^{3}=6.4\times10^{6}$ , so the finite-nuclear-size term $\propto|\psi(0)|^{2}r_p^{2}$ , negligible in ordinary hydrogen, dominates the muonic $2S$ – $2P$ interval. Why it weighs the proton: measuring the muonic Lamb shift reads the proton charge radius directly, the measurement that opened the proton-radius puzzle. *Verified in verify_aq-optics-lamb-shift.py.*

Limiting cases

Case	Result	Meaning
exact point-Coulomb Dirac	$\Delta E = 0$	with a classical static Coulomb field the $(n,j)$ degeneracy is exact — the shift is entirely a quantized-field effect
$\ell\ge 1$ states	leading shift $\approx 0$	no amplitude at the origin, so the contact term passes them by; $2P$ barely moves
$n$ -dependence	$\Delta E \propto \lvert\psi_{ns}(0)\rvert^{2}\propto n^{-3}$	$1S$ : $8172$ MHz versus $2S$ : $1058$ MHz — ratio $7.7$ , the Bethe log softening the naive $8$
power of $\alpha$	$\alpha^{5}m_ec^{2}\ln(1/\alpha)$	one photon loop beyond fine structure's $\alpha^{4}$ — a single extra power of $\alpha$ , not three
$Z$ scaling	$\sim Z^{4}\alpha^{5}m_ec^{2}\ln(1/Z\alpha)/n^{3}$	hydrogen-like uranium: a $1S$ shift of hundreds of eV, tested in storage rings
electron $\to$ muon	contact factor $\times(186)^{3}\approx 6.4\times10^{6}$	the shift graduates from a QED test to a proton-size ruler

Common trapThe Lamb shift is a fine-structure correction you forgot

No — and three confusions live here. First, no improvement within the Dirac equation can produce it: equation (1) is $\ell$ -blind to all orders, so any "derivation" of the Lamb shift from single-particle relativistic corrections is wrong by construction. Second, the sign: students expect a "vacuum attraction" to bind harder, but the smearing dilutes the Coulomb singularity, so $S$ -states are less bound and $2S_{1/2}$ moves up. Third, the scaling is $\alpha^{5}m_ec^{2}\ln(1/\alpha)$ , one power of $\alpha$ beyond fine structure — not "order $\alpha^{3}$ beyond fine structure"; that $\alpha^{3}$ is the size relative to gross structure. And $1420$ MHz is hyperfine, not Lamb — the two land within $35\%$ by coincidence.

Worked examples

Example 1 — Why 1947 needed microwaves, not a spectrograph (the instrument number). Put the field-free interval on every spectroscopic ruler, then compare it to the Doppler width of an optical hydrogen line.

Step 1. Convert the anchor. $1057.845\ \mathrm{MHz}$ : dividing by $c$ gives $0.0353\ \mathrm{cm}^{-1}$ ; times $h$ gives $4.375\ \mu\mathrm{eV}$ ; against the $10\,949$ MHz $2P$ fine structure it is $1057.845/10949 = 9.66\%$ .

Step 2. Size the optical broadening. H- $\alpha$ sits at $\nu_0 = 4.568\times10^{14}$ Hz; its Doppler full width at $300$ K is $\nu_0\sqrt{8\ln2\,k_BT/mc^{2}} = \mathbf{5.6\ GHz}$ , more than five times the Lamb shift.

Step 3. Size the resolving power. To separate $1.06$ GHz on the H- $\alpha$ carrier needs $R=\nu_0/\Delta\nu = 4.3\times10^{5}$ ; on Lyman- $\alpha$ , $R=2.3\times10^{6}$ — and both demand a sub-Doppler method absent in 1947.

Step 4. Draw the conclusion. Driving the $\sim1$ GHz interval directly with microwaves needs no optical carrier and no Doppler-beating trick, so the microwave route was not merely convenient but the only one available. Answer: $\Delta E = 0.0353\ \mathrm{cm}^{-1} = 4.375\ \mu$ eV $= 9.66\%$ of the $2P$ fine structure; optical detection would need $R\sim10^{6}$ plus sub-Doppler resolution, so microwaves were the natural key.

Example 2 — The two-photon Doppler ledger. For the modern $1S$ – $2S$ standard, show that first-order Doppler cancels and estimate the surviving relativistic term.

Step 1. First order. Counter-propagating absorption gives shifts $+kv$ and $-kv$ ; their sum vanishes for every $v$ , so the first-order Doppler contribution is exactly zero.

Step 2. Second order. Time dilation leaves $\Delta\nu/\nu = v^{2}/2c^{2}$ . At $v=300\ \mathrm{m/s}$ (a cold thermal beam), $\Delta\nu/\nu = (300)^{2}/(2\times(2.998\times10^{8})^{2}) = 5.0\times10^{-13}$ .

Step 3. Absolute size. On the $2.4661$ PHz line this is $2.4661\times10^{15}\times5.0\times10^{-13} = \mathbf{1.2\ kHz}$ , the leading kinematic systematic that cooling and zero-velocity extrapolation must remove. Answer: first-order Doppler cancels; the residual second-order shift is $5.0\times10^{-13}$ , about $1.2$ kHz.

Physics of the portrait

The Lamb shift carries three people inside it: the theorist who saw the anomaly in old data, the experimentalists who built an instrument out of the atom's own asymmetries, and the theorist who tamed the infinity on a train ride home.

Origins · 1 minlamb shift and how this was discovered

lamb shift

the microwave anomaly that founded QED · Pasternack 1938 → Lamb and Retherford 1947 → Bethe 1947 → Hänsch's comb

Simon Pasternack read the shift first, in 1938, teasing a $\sim0.03\ \mathrm{cm}^{-1}$ displacement of $2S$ out of scattered optical fine-structure data — an anomaly with no mechanism, easy to dismiss. Willis Lamb and his student Robert Retherford supplied the mechanism's evidence in 1947 by refusing to look at light at all. Their instrument was built on the atom's asymmetries: the $2S$ state's eighth-of-a-second stubbornness against the $2P$ 's nanosecond death, the $10.2$ eV of stored energy that let a metastable atom announce itself by kicking an electron off a tungsten plate, and the Zeeman fan that let a fixed-frequency oscillator walk the resonance. They assembled these into the quench-dip method, then spent six years sharpening it — chasing stray fields, hyperfine overlaps, and line-shape asymmetries down to $1057.77\pm0.10$ MHz. Within weeks of the 1947 result, Hans Bethe did the theory on the train back from the Shelter Island conference: he computed the electron's self-energy, recognised that its divergence is the same one hiding inside the measured electron mass, subtracted it, and got $\approx1040$ MHz from what survived — quantum electrodynamics' first quantitative triumph, and the birth of renormalization. The style Lamb built — drive a forbidden interval directly, detect state survival, extrapolate the field away — became the template for precision atomic physics, running today in Theodor Hänsch's frequency-comb spectroscopy of $1S$ – $2S$ and the muonic-hydrogen experiments that made the shift a proton caliper.

History

1938

— Pasternack reads a 2S anomaly out of hydrogen fine-structure data

1947

— Lamb and Retherford drive the 2S–2P interval with microwaves: about 1000 MHz where Dirac demands zero

1947

— Bethe's mass-renormalized self-energy estimate lands near 1040 MHz

1947

— Kusch and Foley measure the electron's anomalous moment, the shift's radiative cousin

1953

— the Lamb–Retherford campaign reaches 1057.77 ± 0.10 MHz

1965

— Nobel to Tomonaga, Schwinger, and Feynman for quantum electrodynamics

2005

— Nobel to Hänsch and Hall for the optical frequency comb

2010

— the muonic-hydrogen Lamb shift opens the proton-radius puzzle

Going further

Texts. Foot, Atomic Physics, §2.3 and Ch. 7 (Lamb shift and QED corrections) · Demtröder, Laser Spectroscopy, Vol. 1, §2.7 (precision hydrogen spectroscopy) · Bransden and Joachain, Physics of Atoms and Molecules, Ch. 5 · Loudon, The Quantum Theory of Light (vacuum fluctuations and radiative shifts).

Provenance. The QED mechanism — self-energy, vacuum fluctuations, mass renormalization, the Welton smearing, and why only $S$ -states are shifted — is recalled by name from aq-qm-lamb-shift and is not re-derived here; this article owns the experiment, the spectroscopic evaluation, and the precision chain. Base provenance preserved verbatim: measured $2S_{1/2}$ – $2P_{1/2}$ interval $\approx1058$ MHz and $1S$ shift $\approx8172$ MHz are standard textbook/CODATA values (Foot, Demtröder); Bethe's nonrelativistic estimate $\approx1040$ MHz — Bethe, Phys. Rev. 72, 339 (1947); the measurement — Lamb and Retherford, Phys. Rev. 72, 241 (1947). Constants: $\alpha\approx1/137.036$ , $m_ec^{2}=0.511$ MeV, $a_0=0.529\times10^{-10}$ m. A verifier should confirm the $2S$ / $1S$ conversions, the metastability ratio, the Doppler and resolving-power estimates, the two-photon Doppler cancellation, and the muonic contact factor — all executed in verify_aq-optics-lamb-shift.py. Edition note (base repair): the dry base's limiting-cases table labelled the Lamb shift "order $\alpha^{3}$ beyond fine structure"; the shift is $\alpha^{5}m_ec^{2}\ln(1/\alpha)$ , i.e. one power of $\alpha$ beyond fine structure ( $\alpha^{4}$ ) and $\alpha^{3}$ beyond gross structure — corrected here and flagged in the cheat sheet.

Prerequisites. Fine Structure and Spin-Orbit Coupling · Solution of the Schrödinger Equation for Hydrogen · Hydrogenic Wavefunctions and Radial Distributions · Quantum Numbers and Orbital Angular Momentum. Leads to. Hyperfine Structure · The Stark Effect · The Zeeman Effect · Selection Rules · Rydberg Atoms.

Is the vacuum really "fluctuating," or is that a cartoon for the maths?

The operational content is real: the quantized field's ground state has nonzero $\langle E^{2}\rangle$ though $\langle E\rangle=0$ , and anything coupled to the field inherits that variance — the Lamb shift, spontaneous emission, and the Casimir effect are its measured consequences. The cartoon part is picturing classical fields flickering in time; the fluctuations are quantum variances, not histories. The Welton heuristic in aq-qm-lamb-shift treats them as a classical drive because, for this observable, the variance is all that enters.

Why doesn't fine structure already account for it — aren't both just corrections at n = 2?

They live in different theories. Fine structure (spin-orbit, relativistic kinetic, Darwin) is entirely inside the Dirac equation, whose energy depends only on $n$ and $j$ , so it splits states of different $j$ but leaves $2S_{1/2}$ and $2P_{1/2}$ together. The Lamb shift needs the quantized radiation field, which the Dirac equation does not contain — which is why no better basis, no higher $(v/c)$ order, and no exact Dirac solution can produce it. The missing ingredient is a whole field, not a missing term.

If the self-energy calculation diverges, why is the Lamb shift finite and predictable?

Because the divergence is not a property of the atom. A free electron's self-energy diverges identically, and that infinity is already absorbed into the measured electron mass. Subtracting the free-electron expression — mass renormalization — leaves the finite difference, how being bound changes the self-interaction, with the atomic physics entering through $|\psi(0)|^{2}$ and the excitation spectrum inside the logarithm. The Lamb shift was the first number to survive this surgery and match experiment; the surgery itself is derived in aq-qm-lamb-shift.

Why exactly is 2s metastable — what forbids 2s → 1s?

One-photon electric-dipole decay requires $\Delta\ell=\pm1$ ; $2s\to1s$ has $\Delta\ell=0$ and is dipole-forbidden, and the higher multipoles between these two states are suppressed to invisibility. The available escape is two-photon emission — a second-order process, extra powers of $\alpha$ , small phase space — giving $\tau\approx0.12$ s. That eight-order-of-magnitude gap against the $1.6$ ns $2p$ lifetime is not a nuisance; it is the entire experimental design, because it lets a beam of excited atoms exist.

The hyperfine 21 cm line is 1420 MHz and the Lamb shift is 1058 MHz — coincidence?

Complete coincidence, and a useful warning. Hyperfine structure scales as $(m_e/m_p)\alpha^{4}m_ec^{2}$ (nuclear-moment physics, living in $1S$ ), the Lamb shift as $\alpha^{5}m_ec^{2}\ln(1/\alpha)$ (radiative physics, splitting $2S$ from $2P$ ). Different mechanisms, quantum numbers, and states; hydrogen's constants merely land them within $35\%$ . One is a spin-flip within a level, the other the vacuum lifting a level.

Does the 2p₁/₂ state shift at all?

Slightly. It has no contact term, since $|\psi(0)|^{2}=0$ , but the electron's anomalous magnetic moment rescales its spin-orbit energy and small non-contact residues remain — tens of MHz against the $2s$ state's roughly $1$ GHz. The measured $1057.8$ MHz is the difference of the two shifts, dominated by $s$ -state physics, so " $2s$ moves, $2p$ barely does" is honest to a few percent.

The Lamb Shift

Where this lives today

The physical picture

The core equations

Limiting cases

Worked examples

Physics of the portrait

Going further

Before this

Where this leads