Lepton universality

Fermi's 1934 theory of beta decay came with a single number, the coupling $G_F$ , that set the strength of the weak interaction. When the muon was found to decay by the same weak force a decade later, and the tau three decades after that, a sharp question came with them: does the weak force pull on an electron, a muon, and a tau with exactly the same hand, or does each flavour get its own coupling? Lepton universality is the answer the Standard Model insists on and experiment confirms to a fraction of a percent — the charged-current weak interaction couples to all three generations with one and the same strength $g$ , so there is a single $G_F$ , not a separate $G_F^{(e)}$ , $G_F^{(\mu)}$ , $G_F^{(\tau)}$ . The reason this is testable, rather than a matter of definition, is that a purely leptonic decay rate is calculable. A muon or a tau decays through two weak vertices and an internal $W$ , and the rate scales in a rigid, known way: $\Gamma\propto G_F^2\,m^5$ , the Sargent rule. Since the mass dependence is fixed by phase space and the coupling dependence by the vertices, a decay rate and a mass read off the coupling. Comparing the muon and tau — after correcting for the tau's many channels through branching ratios — isolates ratios like $g_\tau/g_\mu$ that come out equal to one within errors. The same test repeats at the $Z$ pole and in $B$ -meson decays, where a handful of anomalies keeps the question alive.

Key takeaways

The charged-current weak interaction couples with one universal strength $g$ to all three lepton flavours, so a single Fermi constant $G_F$ describes $e$ , $\mu$ , and $\tau$ .
Leptonic decay rates obey the Sargent rule $\Gamma=G_F^2 m^5/192\pi^3$ , so a measured rate plus a mass reads out the coupling.
The muon decays essentially only to $e\bar\nu_e\nu_\mu$ , so $\tau_\mu$ directly fixes $G_F$ ; the tau decays many ways, so you must fold in branching ratios.
Ratios of leptonic widths cancel the shared vertex and isolate $g_\tau/g_\mu$ , $g_\mu/g_e$ , $g_\tau/g_e$ — all equal to 1 to about 0.1 %.
The same universality is confirmed at the $Z$ pole and probed today in $B$ -meson decays, where $R(D^{*})$ still shows tension.
Key idea: decay rates scale as $G_F^2 m^5$ , so comparing the measured leptonic decay rates of the muon and tau — corrected for the heavy lepton's many channels via branching ratios — extracts ratios of the weak couplings directly, and they all come out equal.

The core insight

Decay rates scale as $G_F^2 m^5$ , so by comparing the measured leptonic decay rates (corrected for the heavy lepton's many decay channels via branching ratios) one can directly extract ratios of the weak couplings — and they all come out equal.

Where this lives today

Lepton universality is both a triumph and a live frontier. On the triumph side, it is one of the most precisely verified predictions of the Standard Model: the leptonic decays of the muon, tau, and pion, and the $Z$ 's decays into charged-lepton pairs at LEP, all confirm equal couplings at the $0.1$ – $0.2\%$ level, and $G_F$ from muon decay is the electroweak sector's most precise input after $\alpha$ and $M_Z$ . The pion is a second clean laboratory: the ratio $\Gamma(\pi\to e\nu)/\Gamma(\pi\to\mu\nu)$ is helicity-suppressed but calculable to $0.01\%$ , and tests $g_e/g_\mu$ sharply. On the frontier side, the deepest current probe is in $B$ -meson decays, where the ratios $R(K)=B(B\to K\mu^+\mu^-)/B(B\to K e^+e^-)$ and $R(D^{(*)})=B(B\to D^{(*)}\tau\nu)/B(B\to D^{(*)}\ell\nu)$ test whether the muon-versus-electron, or the third generation, behaves anomalously. LHCb's 2022 re-analysis brought $R(K)$ back into agreement with 1, retiring an earlier $3\sigma$ hint. But $R(D)$ and $R(D^{*})$ , measured by BaBar, Belle, Belle II and LHCb, still sit about $3\sigma$ above the Standard Model — a persistent excess of $b\to c\,\tau\nu$ that, if it survives, would be the first crack in Puppi's triangle. Leptoquarks, charged Higgs bosons, and non-universal $Z'$ gauge bosons are constrained precisely by how much lepton-universality violation they can hide.

The physical picture

Every charged-current weak process is built from one elementary joint: a lepton, its neutrino, and a $W$ boson meeting at a vertex whose strength is the coupling $g$ . Universality is the claim that this joint is identical whether the lepton is an electron, a muon, or a tau — the $W$ does not care which generation it touches. Muon decay is the cleanest place to see it, because the muon is too light to decay into anything but leptons: it converts to its own neutrino by emitting a virtual $W^-$ , and that $W^-$ immediately materializes an electron and an anti-electron-neutrino. Two vertices, two powers of $g$ , one internal $W$ — and a rate that a first course in field theory can compute exactly.

Figure

Muon decay μ⁻ → e⁻ ν̄ₑ ν_μ through a virtual W⁻. The amplitude carries one factor of the coupling g at the muon vertex and one at the electron vertex — universality is the statement that this same g appears for e, μ, and τ, so a single Fermi constant G_F describes all three

The payoff is quantitative. Because the coupling dependence and the mass dependence of a leptonic width are both fixed, a single measured lifetime pins the coupling — and the ratios of couplings extracted from the muon and tau come out equal to one to a fraction of a percent. The second figure is the punchline of half a century of measurement: the three independent coupling ratios, each drawn from purely leptonic decays, plotted against the Standard-Model line at exactly 1.

Figure

The experimental verdict — three independent ratios of charged-current couplings from leptonic decays, each consistent with the Standard-Model value of exactly 1: g_μ/g_e = 1.0018 ± 0.0014, g_τ/g_μ = 1.0010 ± 0.0014, and g_τ/g_e = 1.0029 ± 0.0014 (HFLAV averages)

The core equations

The engine of every universality test is the leptonic decay rate. In the four-fermion (Fermi) description that survives at energies far below $M_W$ , muon decay is a contact interaction of two currents, and the phase-space integral over the three-body final state gives a rate with a fixed numerical coefficient. Taking the electron massless (an excellent approximation, since $m_e/m_\mu\approx1/207$ ) yields the Sargent rule.

\Gamma(\mu^-\to e^-\bar\nu_e\nu_\mu)=\frac{G_F^2\,m_\mu^5}{192\pi^3} \qquad (1)

where: $G_F$ — Fermi constant, $1.1663787\times10^{-5}\ \mathrm{GeV^{-2}}$ ; $m_\mu$ — muon mass, $105.6584\ \mathrm{MeV}$ ; the factor $192\pi^3$ is the exact phase-space coefficient for a $V\!-\!A$ contact interaction. Reading: the rate is set entirely by the coupling squared and the fifth power of the decaying mass — this steep $m^5$ is why the tau, $17$ times heavier than the muon, lives a million times shorter. The Fermi constant is not fundamental; it is the low-energy shadow of $W$ exchange. Contracting the $W$ propagator to a point when the momentum transfer is tiny compared with $M_W$ matches the four-fermion strength to the gauge coupling $g$ . Universality lives in this equation: the same $g$ appears for every generation.

\frac{G_F}{\sqrt{2}}=\frac{g^2}{8\,M_W^2} \qquad (2)

where: $g$ — the $SU(2)_L$ weak coupling, common to all leptons if universality holds; $M_W$ — $W$ -boson mass, $80.4\ \mathrm{GeV}$ . Reading: $G_F$ packages two vertex factors of $g$ and one heavy propagator into one measured number; a flavour-dependent $g_\ell$ would turn $G_F$ into a flavour-dependent $G_F^{(\ell)}$ . For the tau and for the muon's sub-percent corrections one must restore the daughter-lepton mass. The exact tree-level phase-space factor multiplies the Sargent rule.

\Gamma(\ell^-\to\ell'^-\bar\nu_{\ell'}\nu_\ell)=\frac{G_F^2\,m_\ell^5}{192\pi^3}\,f(x),\qquad f(x)=1-8x+8x^3-x^4-12x^2\ln x,\quad x=\frac{m_{\ell'}^2}{m_\ell^2} \qquad (3)

where: $f(x)\to1$ as $x\to0$ ; $f=0.99981$ for $\mu\to e$ , but $f=0.9726$ for $\tau\to\mu$ because the muon daughter is not light. Reading: $f(x)$ is the only place the daughter mass enters, and for the electron channels it is negligibly close to 1. Because the tau has many open channels, its partial width into a given leptonic mode is the total width times a branching ratio — this is the bookkeeping step the muon never needs.

\Gamma(\tau\to\ell\,\nu\bar\nu)=\frac{B(\tau\to\ell\,\nu\bar\nu)}{\tau_\tau},\qquad \sum_i B_i=1 \qquad (4)

where: $B(\tau\to e\,\nu\bar\nu)=17.82\%$ , $B(\tau\to\mu\,\nu\bar\nu)=17.39\%$ , the rest hadronic; $\tau_\tau=290.3\ \mathrm{fs}$ . Reading: only a fifth of taus decay to any one leptonic mode, so a raw tau lifetime is not a partial width — the branching ratio converts it into one. Dividing the tau's leptonic width by the muon's makes the shared electron vertex and all common constants cancel, isolating one ratio of couplings — the master formula of universality tests.

\left(\frac{g_\tau}{g_\mu}\right)^2=\frac{\tau_\mu}{\tau_\tau}\,\frac{m_\mu^5}{m_\tau^5}\,B(\tau\to e\,\nu\bar\nu)\,\frac{f(m_e^2/m_\mu^2)}{f(m_e^2/m_\tau^2)} \qquad (5)

where: every quantity on the right is separately measured. Reading: this is a pure number that the Standard Model predicts to be exactly 1; any deviation would be new physics distinguishing the generations. The same test runs at high energy on the $Z$ resonance, where the partial widths into charged-lepton pairs measure the neutral-current couplings and confirm universality independently of the charged current.

\frac{\Gamma(Z\to\tau^+\tau^-)}{\Gamma(Z\to e^+e^-)}=\left(\frac{g_\tau}{g_e}\right)^2\Big|_{\rm NC}\;\xrightarrow{\ \rm SM\ }\;1 \qquad (6)

where: LEP measured these three leptonic $Z$ widths to be equal to $\sim0.2\%$ . Reading: universality is not a quirk of low-energy decays — it holds for both the charged and neutral currents and across four orders of magnitude in energy.

Derivation AWhy the leptonic decay rate scales as the coupling squared times mass to the fifth

The goal is to establish the Sargent rule $\Gamma=G_F^2 m^5/192\pi^3$ from the ground up, first fixing the powers $G_F^2$ and $m^5$ by dimensional analysis alone — which is all universality really needs — and then quoting where the exact coefficient $192\pi^3$ comes from in the three-body phase-space integral. This scaling is the reason a single lifetime measurement can be turned into a coupling, and it is the backbone of every argument that follows.

Step 1. In natural units ( $\hbar=c=1$ ) a decay rate has dimension of mass, $[\Gamma]=M^{+1}$ , because it is an inverse time. The Fermi constant has dimension $[G_F]=M^{-2}$ , since it multiplies two currents each of dimension $M^{+3}$ in an interaction Lagrangian of dimension $M^{+4}$ . Why: getting the dimensions straight is what lets pure counting fix the mass power without doing the integral.

Step 2. Muon decay is first order in the four-fermion interaction, so the amplitude is linear in $G_F$ and the rate is quadratic: $\Gamma\propto G_F^2$ , carrying dimension $M^{-4}$ . To make a rate of dimension $M^{+1}$ , the kinematics must supply $M^{+5}$ . The only mass scale available in $\mu\to e\bar\nu_e\nu_\mu$ once the daughters are treated as massless is the muon mass itself, so $\Gamma\propto G_F^2\,m_\mu^5$ . Why: this single line is the whole physics of the Sargent rule — the steep fifth power comes from nothing but dimensions and the masslessness of the light final state.

Step 3. The dimensionless coefficient is fixed by the spin-averaged matrix element and the Lorentz-invariant three-body phase space. For the $V\!-\!A$ current the squared, summed amplitude integrated over the Michel spectrum of the electron energy gives $\int_0^1 2\varepsilon^2(3-2\varepsilon)\,d\varepsilon=1$ after normalization, and assembling this with the $(2\pi)$ factors of three-body phase space produces the coefficient $1/192\pi^3$ . Why: this is the one place real field theory enters; the result is a clean rational-times- $\pi^3$ number precisely because the interaction is pointlike and the final state is massless.

Step 4. Restoring the electron mass multiplies the result by the phase-space function $f(x)=1-8x+8x^3-x^4-12x^2\ln x$ with $x=(m_e/m_\mu)^2=2.34\times10^{-5}$ , so $f=0.99981$ — a $0.02\%$ shift. The same formula with $x=(m_\mu/m_\tau)^2$ gives $f=0.9726$ for the tau's muonic channel, a $2.7\%$ correction that must be kept. Why: universality tests compare channels with different daughters, so the one honest place the daughter mass matters is $f(x)$ , and dropping it would fake a $2.7\%$ violation in the tau-to-muon mode. *Verified: $\Gamma_0=G_F^2 m_\mu^5/192\pi^3$ gives $\tau_\mu^{\rm LO}=2.187\ \mu s$ ; $f(m_e^2/m_\mu^2)=0.99981$ and $f(m_\mu^2/m_\tau^2)=0.9726$ reproduced numerically.*

Go deeper · Derivation B · 2 minFrom W-boson exchange to a single Fermi constant

Derivation BFrom W-boson exchange to a single Fermi constant

The aim is to show precisely where universality is encoded in the Standard Model: that $G_F$ is the low-energy contraction of $W$ exchange, $G_F/\sqrt2=g^2/8M_W^2$ , and that the $SU(2)_L$ gauge structure forces the same coupling $g$ on all three generations — so a violation of universality is a violation of the gauge principle, not a free parameter one can tune. We track the $W$ propagator from full to contact form and identify the flavour-blind vertex.

Step 1. The charged-current interaction couples each left-handed lepton doublet to the $W$ with a vertex proportional to $g$ : schematically $\mathcal{L}\supset\frac{g}{\sqrt2}\,W_\mu^-\,\bar\ell\,\gamma^\mu P_L\,\nu_\ell+\text{h.c.}$ , written identically for $\ell=e,\mu,\tau$ . The single coupling $g$ appearing for every generation is universality, built into the gauge group. Why: in the SM $g$ is a property of the $SU(2)_L$ gauge field, not of the fermion it touches, so equal couplings are mandatory, not fitted — this is the theoretical content of the principle.

Step 2. Muon decay proceeds by $\mu^-\to\nu_\mu W^-$ at one vertex and $W^-\to e^-\bar\nu_e$ at the other, so the amplitude carries $g^2$ and one $W$ propagator $\dfrac{-g_{\mu\nu}+q_\mu q_\nu/M_W^2}{q^2-M_W^2}$ . Why: two vertices give two powers of $g$ ; the propagator carries the heavy scale that will become $G_F$ .

Step 3. The momentum transfer in muon decay is at most $m_\mu\approx0.1\ \mathrm{GeV}$ , utterly negligible against $M_W\approx80\ \mathrm{GeV}$ , so $q^2-M_W^2\to-M_W^2$ and the propagator collapses to $g_{\mu\nu}/M_W^2$ . The amplitude becomes a contact interaction of two currents with strength $g^2/M_W^2$ . Why: shrinking the propagator to a point is exactly the passage from the full electroweak theory to Fermi's 1934 four-fermion theory — the same physics at two resolutions.

Step 4. Matching the contact strength to the historical normalization of Fermi's constant gives $\dfrac{G_F}{\sqrt2}=\dfrac{g^2}{8M_W^2}$ . Because $g$ is universal, so is $G_F$ : measuring it in muon decay, in nuclear beta decay, or in tau decay must give the same number, up to calculable radiative corrections. Why: this equation is the bridge — it turns the abstract statement "one gauge coupling" into the testable statement "one measured $G_F$ ", which is what the experiments check. *Verified: with $g=0.653$ and $M_W=80.38\ \mathrm{GeV}$ , $g^2/8M_W^2\sqrt2^{-1}$ reproduces $G_F=1.166\times10^{-5}\ \mathrm{GeV^{-2}}$ .*

Go deeper · Derivation C · 1 minIsolating a ratio of couplings from measured lifetimes

Derivation CIsolating a ratio of couplings from measured lifetimes

The purpose here is to build equation (5) — the number an experimentalist actually computes — by dividing the tau's leptonic width by the muon's so that the common electron vertex, the phase-space coefficient, and the neutrino couplings all cancel, leaving a clean ratio $(g_\tau/g_\mu)^2$ that the Standard Model fixes at 1. This is the step that turns two masses, two lifetimes, and one branching ratio into a single test of the principle.

Step 1. Write both leptonic widths with generation-specific vertex couplings restored: $\Gamma(\mu\to e)\propto g_\mu^2 g_e^2\,m_\mu^5\,f(m_e^2/m_\mu^2)$ and $\Gamma(\tau\to e)\propto g_\tau^2 g_e^2\,m_\tau^5\,f(m_e^2/m_\tau^2)$ , with the same proportionality constant $1/(192\pi^3\times\text{propagator}^2)$ . Why: keeping the couplings explicit per vertex is what makes the cancellation visible — each decay has one heavy-lepton vertex and one electron vertex.

Step 2. Form the ratio $\dfrac{\Gamma(\tau\to e)}{\Gamma(\mu\to e)}=\dfrac{g_\tau^2}{g_\mu^2}\,\dfrac{m_\tau^5}{m_\mu^5}\,\dfrac{f(m_e^2/m_\tau^2)}{f(m_e^2/m_\mu^2)}$ . The shared electron coupling $g_e^2$ and the overall constant cancel identically. Why: the whole point of taking a ratio is to delete everything universality does not test, leaving only the two heavy-lepton couplings being compared.

Step 3. Replace the two widths by measured quantities. The muon decays essentially $100\%$ to $e\bar\nu_e\nu_\mu$ , so $\Gamma(\mu\to e)=1/\tau_\mu$ . The tau does not, so $\Gamma(\tau\to e)=B(\tau\to e\,\nu\bar\nu)/\tau_\tau$ by equation (4). Why: this is where the branching ratio is unavoidable — using $1/\tau_\tau$ in place of the partial width would overcount by a factor of five and destroy the test.

Step 4. Solve for the coupling ratio: $\left(\dfrac{g_\tau}{g_\mu}\right)^2=\dfrac{\tau_\mu}{\tau_\tau}\,\dfrac{m_\mu^5}{m_\tau^5}\,B(\tau\to e\,\nu\bar\nu)\,\dfrac{f(m_e^2/m_\mu^2)}{f(m_e^2/m_\tau^2)}$ , which is equation (5). Every factor is independently measured; the SM predicts it equals 1. Why: this closes the loop — the abstract "same $g$ " is now a calculation on the PDG's own numbers. *Verified: substituting $\tau_\mu=2.1970\ \mu s$ , $\tau_\tau=290.3\ \mathrm{fs}$ , $m_\mu/m_\tau=0.05946$ , $B=0.1782$ gives $(g_\tau/g_\mu)^2=1.0024$ , i.e. $g_\tau/g_\mu=1.0012$ .*

Limiting cases

Case	Result	Meaning
Massless daughter ( $m_{\ell'}\to0$ )	$f(x)\to1$ , $\Gamma\to G_F^2 m_\ell^5/192\pi^3$	pure Sargent rule; exact for $\mu,\tau\to e$ to $0.02\%$
Electron channel of the tau	$f\approx1$ , $\Gamma\propto m_\tau^5$	cleanest universality mode — no daughter-mass ambiguity
Muon channel of the tau	$f(m_\mu^2/m_\tau^2)=0.9726$	the $2.7\%$ phase-space suppression must be kept, or universality looks broken
$q^2\ll M_W^2$ (all lepton decays)	propagator $\to1/M_W^2$ , contact $G_F$	Fermi theory valid; $G_F$ well-defined and flavour-independent
$q^2\sim M_W^2$ ( $Z/W$ resonance)	full propagator; on-shell $Z\to\ell\ell$	universality re-tested at high energy, independent of low-energy decays
Universality violated ( $g_\ell\neq g_{\ell'}$ )	$G_F^{(\ell)}$ splits; ratios $\neq1$	signals new physics — extra $W'/Z'$ , leptoquarks, charged Higgs

Common trapthe misconception

The tempting shortcut is to compare the muon and tau lifetimes directly, as if $\tau_\mu/\tau_\tau$ already measured a ratio of couplings. It does not. The muon has essentially one decay channel, so its lifetime is its inverse leptonic width. The tau has dozens of channels — into electrons, muons, and a whole zoo of hadronic final states — so only about $18\%$ of taus decay to any single leptonic mode. The tau lifetime is set by the total width, which is dominated by hadrons and knows nothing clean about $g_\tau$ alone. To recover the leptonic partial width you must multiply the total rate $1/\tau_\tau$ by the branching ratio $B(\tau\to\ell\,\nu\bar\nu)$ ; only then does equation (5) hold. Skipping the branching ratio inflates the extracted $\Gamma(\tau\to e)$ by roughly a factor of five and would fake a gross violation of universality where there is none.

Worked examples

1 · Predicting the muon lifetime from a single Fermi constant (fully worked)

Take $G_F=1.1663787\times10^{-5}\ \mathrm{GeV^{-2}}$ and $m_\mu=105.6584\ \mathrm{MeV}$ as given. Predict the muon lifetime, including the phase-space and leading QED radiative corrections, and compare with the measured $\tau_\mu=2.196981\ \mu s$ .

Step 1. Compute the leading Sargent-rule width. With $m_\mu=0.1056584\ \mathrm{GeV}$ , $m_\mu^5=1.3171\times10^{-5}\ \mathrm{GeV^5}$ and $G_F^2=1.3604\times10^{-10}\ \mathrm{GeV^{-4}}$ , so $\Gamma_0=\dfrac{G_F^2 m_\mu^5}{192\pi^3}=\dfrac{1.3604\times10^{-10}\times1.3171\times10^{-5}}{5953.1}=\mathbf{3.010\times10^{-19}\ \mathrm{GeV}}$ .

Step 2. Convert to a lifetime with $\hbar=6.58212\times10^{-25}\ \mathrm{GeV\,s}$ : $\tau_0=\hbar/\Gamma_0=6.58212\times10^{-25}/3.010\times10^{-19}=\mathbf{2.1873\ \mu s}$ . Already within half a percent of experiment — the raw $G_F^2 m^5$ law does most of the work.

Step 3. Apply the phase-space factor. $x=(m_e/m_\mu)^2=2.339\times10^{-5}$ gives $f(x)=0.999813$ , which lowers the rate and raises the lifetime to $\tau=\tau_0/f=2.1877\ \mu s$ — a $0.02\%$ nudge.

Step 4. Apply the leading radiative correction. The one-loop QED factor is $1+\Delta q$ with $\Delta q=\dfrac{\alpha}{2\pi}\left(\dfrac{25}{4}-\pi^2\right)=-4.20\times10^{-3}$ , which lowers the rate a further $0.42\%$ : $\tau=2.1877/(1+\Delta q)=2.1877/0.99580=\mathbf{2.1970\ \mu s}$ .

Step 5. Compare. The prediction $2.1970\ \mu s$ matches the measured $2.196981\ \mu s$ to $\sim0.01\%$ . A single universal $G_F$ from muon decay reproduces the lifetime to four significant figures — the same $G_F$ that beta decay and tau decay must share.

2 · Testing $g_\tau/g_\mu$ from the tau branching ratio (fully worked)

Using $\tau_\mu=2.196981\ \mu s$ , $\tau_\tau=290.3\ \mathrm{fs}$ , $m_\mu=105.6584\ \mathrm{MeV}$ , $m_\tau=1776.86\ \mathrm{MeV}$ , and $B(\tau\to e\,\nu\bar\nu)=0.1782$ , evaluate equation (5) and state whether universality holds.

Step 1. Form the lifetime ratio: $\tau_\mu/\tau_\tau=2.196981\times10^{-6}/2.903\times10^{-13}=\mathbf{7.568\times10^{6}}$ . The muon lives about seven million times longer than the tau.

Step 2. Form the mass ratio to the fifth power: $m_\mu/m_\tau=0.059463$ , so $(m_\mu/m_\tau)^5=\mathbf{7.435\times10^{-7}}$ . This tiny factor is the tau's $m^5$ advantage in decay rate — the reason it lives so briefly.

Step 3. The phase-space factors for both electron channels are $\approx1$ ( $f_{\mu e}/f_{\tau e}=0.99981$ ), so they barely matter here. Multiply everything: $(g_\tau/g_\mu)^2=7.568\times10^{6}\times7.435\times10^{-7}\times0.1782\times0.99981$ .

Step 4. Evaluate: $7.568\times10^{6}\times7.435\times10^{-7}=5.627$ ; $\times0.1782=1.0027$ ; $\times0.99981=\mathbf{1.0025}$ . Taking the square root, $g_\tau/g_\mu=\mathbf{1.0012}$ .

Step 5. Interpret. The raw arithmetic lands at $1.001$ , and the dominant uncertainties — the tau lifetime and branching ratio, each near $0.3\%$ , entering $g$ with a square root — put the result squarely on 1. The full HFLAV average, with precise inputs and radiative corrections, is $g_\tau/g_\mu=1.0010\pm0.0014$ . The tau couples to the $W$ with the same strength as the muon to better than two parts in a thousand — a clean confirmation of universality across two generations.

3 · The tau's leptonic branching ratio (estimate)

Why is $B(\tau\to e\,\nu\bar\nu)\approx18\%$ and not, say, $50\%$ ? The tau decays to $e\nu\nu$ , $\mu\nu\nu$ , or into hadrons via $W^-\to\bar ud$ (and $\bar us$ ). Counting each open channel by its degrees of freedom, the two leptonic channels count $1$ each and the hadronic channel counts $\approx N_c=3$ for the three quark colours. The naive split $1:1:3$ gives each leptonic mode $B\approx1/5=20\%$ . The measured $17.8\%$ sits just below this, the deficit being phase-space and QCD corrections — and that colour counting works here is itself a check that the tau's charged current is the universal one.

Physics of the portrait

Universality was not discovered in one experiment but recognized as a pattern: three different weak processes, measured independently, turned out to share one coupling. The person who drew that pattern into a single picture was an Italian physicist working in the late 1940s, and the picture he drew still hangs over the subject.

Origins · 1 minGiampietro Puppi and how this was discovered

Giampietro Puppi

1917–2006, Italian physicist · the universality of the weak interaction

In 1948, working in Bologna, Giampietro Puppi noticed that three seemingly unrelated processes shared a common coupling strength: nuclear beta decay ( $n\to p\,e\,\bar\nu$ ), muon decay ( $\mu\to e\,\nu\bar\nu$ ), and muon capture ( $\mu\,p\to n\,\nu$ ). Fermi's constant $G_F$ , extracted from each, came out the same to within the crude precision of the day. Puppi expressed this by placing the three processes at the corners of a triangle — the "Puppi triangle" — whose sides represented the pairs of currents (nucleon, electron, muon) coupled with equal strength. The same universality was proposed independently by Oskar Klein, by Tiomno and Wheeler, and by Lee, Rosenbluth and Yang around 1949, but the triangle became the enduring image — a bold claim on thin data that one force reached into nuclei, decaying muons, and captured muons with a single strength, blind to what it acted on. The V $-$ A theory of Feynman, Gell-Mann, Sudarshan and Marshak (1957) gave it a Lorentz structure; the electroweak theory of Glashow, Weinberg and Salam gave it a gauge origin, making $g$ a property of the $SU(2)_L$ field itself. The 1975 discovery of the tau extended the triangle to a third generation, and every measurement since has kept Puppi's corners equal.

History

1934

— Fermi's theory of beta decay introduces the coupling G_F

1947

— Pontecorvo notes muon capture and beta decay share a strength, hinting at μ–e universality

1948

— Puppi draws the triangle linking beta decay, muon decay, and muon capture with one coupling

1957

— the V−A theory fixes the Lorentz structure of the universal charged current

1975

— Perl's group at SLAC discovers the tau, extending universality to a third generation

2000

— LEP confirms equal Z decays to ee, μμ, ττ to about two parts in a thousand

Going further

Texts. Thomson, Modern Particle Physics, §11–12 (weak interactions and lepton universality) · Halzen & Martin, Quarks and Leptons, Ch. 12 (charged-current weak interactions) · Griffiths, Introduction to Elementary Particles, 2nd ed., §9.2 (muon and tau decay) · Peskin & Schroeder, An Introduction to Quantum Field Theory, §5.3 and problem set (muon decay rate) · Schwartz, Quantum Field Theory and the Standard Model, Ch. 29 (weak decays).

Provenance. The Sargent rule $\Gamma=G_F^2 m^5/192\pi^3$ and the phase-space factor $f(x)$ follow Thomson Eqs. (11.16) and (12.1)–(12.5) and are re-derived here from dimensional analysis plus the Michel-spectrum integral. The matching $G_F/\sqrt2=g^2/8M_W^2$ is standard electroweak theory (Halzen–Martin §12). Numerical inputs ( $m_\mu$ , $m_\tau$ , $\tau_\mu$ , $\tau_\tau$ , branching ratios, $G_F$ ) are 2024 PDG values; the coupling ratios $g_\mu/g_e=1.0018\pm0.0014$ , $g_\tau/g_\mu=1.0010\pm0.0014$ , $g_\tau/g_e=1.0029\pm0.0014$ are HFLAV tau-decay averages. A verifier should confirm: (1) $\tau_\mu^{\rm LO}=2.187\ \mu s$ from $G_F^2 m_\mu^5/192\pi^3$ ; (2) the radiative-corrected $2.1970\ \mu s$ ; (3) $f(m_\mu^2/m_\tau^2)=0.9726$ ; (4) equation (5) yields $(g_\tau/g_\mu)^2=1.002$ .

Prerequisites. Charged-Current Weak Interaction · Muon Decay · The Fermi Constant · Particle Decay Rates and Branching Ratios. Leads to. Neutrino–Nucleon Scattering · Z Decays · W-Boson Properties.

Ask the tutor: Ask me why the tau's leptonic branching ratio is near 1/5, or how the $R(D^{*})$ anomaly could break Puppi's triangle.

Is lepton universality an assumption of the Standard Model or a prediction?

It is a prediction, forced by the gauge structure. In the Standard Model the charged current couples through the $SU(2)_L$ gauge field, whose coupling $g$ is a property of the field, not of the particle it touches — so all three generations must share it. Nothing was tuned to make the couplings equal; equality is mandatory once you write down a single gauge group acting identically on each lepton doublet. That is exactly what makes universality a powerful test: a measured deviation cannot be absorbed by adjusting a Standard-Model parameter, so it would unambiguously signal physics beyond it, such as a new boson that distinguishes the generations.

Why does the decay rate depend so steeply on mass, as m to the fifth?

Dimensional analysis fixes it. A decay rate has units of inverse time, i.e. mass in natural units. The rate is proportional to $G_F^2$ , which carries units of mass to the minus four. To build something with the units of a rate, the kinematics must supply five powers of mass — and in a purely leptonic decay with light daughters the only available scale is the decaying lepton's own mass. Hence $\Gamma\propto G_F^2 m^5$ , the Sargent rule. The steepness is dramatic: the tau is about $17$ times heavier than the muon, so $17^5\approx1.4$ million, which is roughly why the tau lives a million times more briefly even though its coupling is identical.

How does taking a ratio of decay rates remove the neutrino couplings?

Each leptonic decay has two charged-current vertices, and each vertex contributes a coupling $g$ and, on the neutrino side, the same neutrino field. When you divide $\Gamma(\tau\to e\nu\bar\nu)$ by $\Gamma(\mu\to e\nu\bar\nu)$ , both numerator and denominator contain one electron vertex and its associated $\nu_e$ , plus the universal phase-space coefficient $1/192\pi^3$ and the propagator factor. All of that cancels identically, leaving only the ratio of the heavy-lepton couplings, $(g_\tau/g_\mu)^2$ , times known mass and phase-space factors. The neutrinos never carry generation-splitting couplings in this comparison, so they simply drop out — which is why a ratio, not an absolute rate, is the clean observable.

What is the Puppi triangle, exactly?

It is Giampietro Puppi's 1948 device for displaying weak universality. He placed three "currents" — the nucleon pair $(n,p)$ , the electron pair $(e,\nu_e)$ , and the muon pair $(\mu,\nu_\mu)$ — at the corners of a triangle. Each side represented a weak process joining two currents: beta decay joins the nucleon and electron corners, muon decay joins the electron and muon corners, and muon capture joins the nucleon and muon corners. The striking empirical fact was that the coupling extracted along every side was the same $G_F$ . The triangle made "one coupling for three processes" a single picture, anticipating by a decade the idea of a weak interaction with universal strength — later grounded in the V $-$ A theory and the electroweak gauge group.

Do the tiny deviations like 1.0018 mean universality is slightly broken?

No. Those numbers are consistent with exactly 1 within their quoted uncertainties of about $\pm0.0014$ . A value of $1.0018\pm0.0014$ is roughly one standard deviation from unity — statistically indistinguishable from perfect universality. The precision is limited by how well the tau lifetime, the branching ratios, and the masses are measured, not by any resolved discrepancy. One should read the whole ensemble together: three independent ratios all sitting within one-to-two sigma of 1 is a strong, mutually reinforcing confirmation. Genuine tension only appears elsewhere — notably in the $R(D^{*})$ measurements — and even there it is at the $3\sigma$ level, not yet a discovery.

If universality holds, why bother measuring it ever more precisely?

Because the precision itself is the discovery tool. Many extensions of the Standard Model — leptoquarks, charged Higgs bosons, non-universal $Z'$ gauge bosons — would couple to the third generation, or to muons versus electrons, slightly differently, and their signatures are exactly small departures of these ratios from 1. Pushing the measurement from $1\%$ to $0.1\%$ to $0.01\%$ shrinks the space where such particles can hide. The current $B$ -meson anomalies are this game played at the edge of sensitivity: universality tests are quietly one of the most sensitive searches for new physics we have.

Lepton universality

Where this lives today

The physical picture

The core equations

Limiting cases

Worked examples

1 · Predicting the muon lifetime from a single Fermi constant (fully worked)

2 · Testing $g_\tau/g_\mu$ from the tau branching ratio (fully worked)

3 · The tau's leptonic branching ratio (estimate)

Physics of the portrait

Going further

Before this

Where this leads

Lepton universality

Where this lives today

The physical picture

The core equations

Limiting cases

Worked examples

1 · Predicting the muon lifetime from a single Fermi constant (fully worked)

2 · Testing gτ/gμg_\tau/g_\mugτ​/gμ​ from the tau branching ratio (fully worked)

3 · The tau's leptonic branching ratio (estimate)

Physics of the portrait

Going further

Before this

Where this leads

2 · Testing $g_\tau/g_\mu$ from the tau branching ratio (fully worked)