The number of light neutrino generations

There are three copies of the fundamental fermions — three charged leptons, three up-type quarks, three down-type quarks, three neutrinos — and nothing in the Standard Model explains why. The number of generations is a free parameter, a fact of nature the theory takes as input. So the count must be settled by experiment, and the sharpest measurement comes not from finding a new particle but from weighing a resonance. The neutral Z boson couples to every fermion light enough to be produced in its decay, including the neutrinos that leave no track and no energy in any detector. Those decays are literally invisible — yet not silent, because every open channel adds to the Z's total decay width $\Gamma_Z$ , and $\Gamma_Z$ controls the height and breadth of the resonance peak that four experiments scanned to exquisite precision at LEP. The trick is subtraction. Measure the total width from the lineshape, measure every visible width — hadrons and charged leptons — from the event rates, and whatever is left over must be the neutrinos escaping unseen. This leftover is the invisible width $\Gamma_{\text{inv}}$ , and dividing it by the Standard Model prediction for one species counts the species: $N_\nu=\Gamma_{\text{inv}}/\Gamma_{\nu\bar\nu}$ . The LEP combination gives $N_\nu=2.9840\pm0.0082$ — three, to a fraction of a percent. This single number closes the fermion generations at three light active families and forbids a fourth ordinary neutrino below $45$ GeV. This article takes the Breit-Wigner machinery of the previous section and turns it into a census of the neutrino sector.

Key takeaways

The number of generations is not predicted by the Standard Model — it is measured, and the decisive measurement is the invisible width of the Z resonance at LEP.
The Z couples to all light fermions; neutrino decays $Z\to\nu\bar\nu$ escape detection but still add to the total width $\Gamma_Z$ .
Extract $\Gamma_{\text{inv}}=\Gamma_Z-\Gamma_{\text{had}}-3\Gamma_{\ell\ell}$ from the measured widths, then $N_\nu=\Gamma_{\text{inv}}/\Gamma_{\nu\bar\nu}$ .
Adding a species widens $\Gamma_Z$ and lowers the peak $\sigma^0_{\text{had}}$ , so the peak height alone discriminates $N_\nu=2$ , $3$ , $4$ .
The LEP result is $N_\nu=2.9840\pm0.0082$ : three light active neutrino families, ruling out a fourth ordinary generation.
Key idea: the Z couples to every light fermion, so the slice of its precisely-measured decay width unaccounted for by visible channels — the invisible width — divided by the single-species prediction directly counts the light neutrino generations, giving $N_\nu=3$ .

The core insight

The Z couples to every fermion, so the part of its precisely-measured total decay width that is not accounted for by visible (hadronic and charged-lepton) channels — the invisible width $\Gamma_{\text{inv}}$ — divided by the Standard Model width for one neutrino species counts the number of light neutrino generations: $N_\nu=\Gamma_{\text{inv}}/\Gamma_{\nu\bar\nu}=3$ .

Where this lives today

The three-family count is a fixed landmark, but the physics around it stays alive. The number constrains anything with the quantum numbers of a light neutrino: a fourth "sequential" generation is dead at the electroweak scale, and any new neutral state coupling to the Z must be heavier than $45$ GeV or sterile. That has channelled the search for extra neutrinos toward heavy neutral leptons and sterile neutrinos — states LEP is blind to — hunted now at the LHC, in beam-dump experiments, and in reactor and short-baseline oscillation anomalies. The cosmological census is meanwhile a precision measurement in its own right: Planck reads the effective relativistic species from the microwave background as $N_{\text{eff}}=2.99\pm0.17$ , consistent with the Standard Model's $3.044$ , and any deviation would signal extra light degrees of freedom — a neutrino, an axion, a dark-sector particle. And the LEP data are still reanalysed: a 2020 re-evaluation of the Bhabha luminosity nudged $N_\nu$ toward $2.996\pm0.007$ , easing a mild tension with exactly three. Collider, cosmos, and the early-universe furnace all report the same headcount.

The physical picture

The Z boson is the great democrat of the Standard Model: it couples to every fermion, quark and lepton alike, with a strength fixed by that fermion's weak isospin and electric charge. When it decays it partitions its total width among every channel that is kinematically open — every pair $f\bar f$ with $m_f<m_Z/2$ . Most decays are visible: quark pairs materialise as jets of hadrons, and the charged-lepton pairs $e^+e^-$ , $\mu^+\mu^-$ , $\tau^+\tau^-$ leave clean tracks. But three channels are invisible. A neutrino carries no charge and no colour and sails out of the detector without a trace. Those $Z\to\nu\bar\nu$ decays are missing energy — yet they are counted, because the width they carry off is part of $\Gamma_Z$ , which sets the shape of the resonance. The first figure shows where the width goes.

Figure

Where the Z width goes: hadrons take about 70 percent and three charged-lepton pairs about 10 percent, leaving an invisible remainder of 499 MeV; dividing that by the single-species prediction 167 MeV gives 499/167 = 2.98, three neutrino families

The invisible width is not a small correction — it is a full fifth of the entire decay rate of the Z, hiding in plain sight as missing energy. The genius of the LEP measurement is that you never detect a single neutrino to count them: the count is written into the shape of the resonance. Because $\Gamma_Z$ appears in the denominator of the peak cross section, every extra species pumps more width into $\Gamma_Z$ , which broadens the peak and pulls its height down — so scanning the peak precisely tells you how many neutrinos are draining it. The second figure is the iconic result: the same hadronic cross section drawn for two, three, and four families, with the real data landing squarely on the three-family curve.

Figure

The neutrino-counting plot: the Born-level e⁺e⁻ → hadrons peak falls as species are added — 47.6 nb for N = 2, 41.5 nb for N = 3, 36.4 nb for N = 4 — because each extra Z → ν ν̄ channel widens Γ_Z; the LEP data sit on the three-family curve

The core equations

Everything rests on the partial width of the Z into a single fermion pair, computed once from the neutral-current coupling. The rate depends on the vector and axial couplings $g_V^f,g_A^f$ , the colour multiplicity $N_c^f$ (three for quarks, one for leptons), and the mass and Fermi constant that set the electroweak scale.

\Gamma(Z\to f\bar f)=\frac{N_c^f\,G_F\,m_Z^3}{6\sqrt{2}\,\pi}\left[(g_V^f)^2+(g_A^f)^2\right],\qquad g_V^f=T_3^f-2Q_f\sin^2\theta_W,\quad g_A^f=T_3^f \qquad (1)

where: $G_F=1.166\times10^{-5}\ \mathrm{GeV^{-2}}$ — Fermi constant; $m_Z=91.19$ GeV; $T_3^f$ — weak isospin ( $+\tfrac12$ for up-type quarks and neutrinos, $-\tfrac12$ for down-type quarks and charged leptons); $Q_f$ — electric charge; $\sin^2\theta_W\approx0.231$ . Reading it: a neutrino has $(g_V)^2+(g_A)^2=\tfrac12$ exactly, while a charged lepton's near-cancellation $g_V\approx-0.04$ makes its width about half the neutrino's. The total width is the sum over every open channel — visible hadronic and charged-lepton pieces plus invisible neutrino pieces — with universality writing the three charged leptons as $3\Gamma_{\ell\ell}$ and the neutrinos as $N_\nu\Gamma_{\nu\bar\nu}$ .

\Gamma_Z=\Gamma_{\text{had}}+3\,\Gamma_{\ell\ell}+N_\nu\,\Gamma_{\nu\bar\nu},\qquad \Gamma_{\text{inv}}\equiv N_\nu\,\Gamma_{\nu\bar\nu}=\Gamma_Z-\Gamma_{\text{had}}-3\,\Gamma_{\ell\ell} \qquad (2)

where: $\Gamma_{\text{had}}$ — width into all five accessible quark flavours (u, d, s, c, b; the top is far too heavy); $\Gamma_{\ell\ell}$ — width into one charged-lepton pair; $\Gamma_{\text{inv}}$ — the invisible width, the visible-subtracted remainder. Reading it: $\Gamma_{\text{inv}}$ is measured not by seeing neutrinos but by subtracting everything visible from the total. Dividing the invisible width by the single-species prediction counts the species. In practice one divides by the well-measured charged-lepton width and multiplies by the clean Standard Model ratio, which is far more precisely known than either absolute width.

N_\nu=\frac{\Gamma_{\text{inv}}}{\Gamma_{\nu\bar\nu}}=\frac{\Gamma_{\text{inv}}}{\Gamma_{\ell\ell}}\left(\frac{\Gamma_{\ell\ell}}{\Gamma_{\nu\bar\nu}}\right)_{\!\text{SM}},\qquad \left(\frac{\Gamma_{\nu\bar\nu}}{\Gamma_{\ell\ell}}\right)_{\!\text{SM}}=1.9912 \qquad (3)

where: the ratio $(\Gamma_{\ell\ell}/\Gamma_{\nu\bar\nu})_{\text{SM}}=1/1.9912=0.5022$ is fixed almost entirely by the couplings in equation (1), nearly free of the top-quark and Higgs-mass uncertainties that afflict the absolute widths. Reading it: measuring the ratio $\Gamma_{\text{inv}}/\Gamma_{\ell\ell}$ removes the common systematics, which is why this route gives the smallest uncertainty. The observable that delivers $\Gamma_{\text{inv}}/\Gamma_{\ell\ell}$ is the peak hadronic cross section, into which the Breit-Wigner formula folds the total width squared. Because $\Gamma_Z$ sits in the denominator, the peak height is a sensitive counter of the invisible channels.

\sigma^0_{\text{had}}=\frac{12\pi}{m_Z^2}\,\frac{\Gamma_{ee}\,\Gamma_{\text{had}}}{\Gamma_Z^2}\;\;\Longrightarrow\;\; \frac{\Gamma_{\text{inv}}}{\Gamma_{\ell\ell}}=\sqrt{\frac{12\pi\,R_\ell}{m_Z^2\,\sigma^0_{\text{had}}}}-R_\ell-3 \qquad (4)

where: $R_\ell\equiv\Gamma_{\text{had}}/\Gamma_{\ell\ell}$ — the measured hadron-to-lepton width ratio; the inversion uses $\Gamma_{ee}=\Gamma_{\ell\ell}$ and $\Gamma_Z=\Gamma_{\ell\ell}(3+R_\ell+\Gamma_{\text{inv}}/\Gamma_{\ell\ell})$ . Reading it: the lineshape observables $m_Z$ , $\sigma^0_{\text{had}}$ and $R_\ell$ fix $\Gamma_{\text{inv}}/\Gamma_{\ell\ell}$ , and equation (3) turns that into $N_\nu$ . There is also a fully independent method: produce the neutrinos above the Z peak and tag the event by a single photon radiated from the initial state, $e^+e^-\to\nu\bar\nu\gamma$ . This cross section grows in proportion to the number of species.

\sigma(e^+e^-\to\nu\bar\nu\gamma)\;\propto\;N_\nu \qquad (5)

where: the photon recoils against nothing visible, so the signature is a lone photon plus missing energy. Reading it: counting single-photon events at known luminosity measures $N_\nu$ directly, with different systematics from the lineshape — and it agrees, giving three.

Derivation AFrom the electroweak couplings to the Z partial widths and their ratio

The goal of this derivation is to show that the ratio in equation (3), the Standard Model $\Gamma_{\nu\bar\nu}/\Gamma_{\ell\ell}\approx2$ , is not a fudge factor but a direct consequence of the neutral-current couplings, and to compute the widths the neutrino count is calibrated against. We start from the universal partial-width formula, insert the isospin and charge of a neutrino and of a charged lepton, and read off the absolute widths and the ratio that makes the ratio method clean.

Step 1. Write the partial width in the form $\Gamma(Z\to f\bar f)=\dfrac{N_c^f G_F m_Z^3}{6\sqrt2\,\pi}[(g_V^f)^2+(g_A^f)^2]$ and evaluate the coupling prefactor $P=G_F m_Z^3/(6\sqrt2\,\pi)$ . With $G_F=1.16637\times10^{-5}\ \mathrm{GeV^{-2}}$ and $m_Z=91.1876$ GeV this gives $P=0.3318$ GeV $=331.8$ MeV. Why: $P$ carries all the dimensionful electroweak physics, so every partial width becomes just $P$ times a pure coupling number times the colour count.

Step 2. For a neutrino, $T_3=+\tfrac12$ and $Q=0$ , so $g_V=g_A=\tfrac12$ and $(g_V)^2+(g_A)^2=\tfrac12$ ; with $N_c=1$ the tree-level width is $\Gamma_{\nu\bar\nu}=P\times\tfrac12=165.9$ MeV. Why: the neutrino has no charge, so the mixing angle drops out entirely — its width is the cleanest prediction in the electroweak sector.

Step 3. For a charged lepton, $T_3=-\tfrac12$ and $Q=-1$ , so $g_A=-\tfrac12$ and $g_V=-\tfrac12+2\sin^2\theta_W\approx-0.038$ ; then $(g_V)^2+(g_A)^2\approx0.2514$ and $\Gamma_{\ell\ell}=P\times0.2514=83.4$ MeV. Why: the vector coupling nearly vanishes because $\sin^2\theta_W$ is close to $\tfrac14$ , and that cancellation is exactly what makes $\Gamma_{\ell\ell}\approx\tfrac12\Gamma_{\nu\bar\nu}$ .

Step 4. Form the ratio $\Gamma_{\nu\bar\nu}/\Gamma_{\ell\ell}=0.5/0.2514=1.989$ at tree level; electroweak corrections shift it to the precise $1.9912$ used in equation (3), and the absolute widths to $\Gamma_{\nu\bar\nu}=167.2$ MeV, $\Gamma_{\ell\ell}=84.0$ MeV. Why: in the ratio the large common corrections cancel, so $1.9912$ is known to one part in $10^3$ even though each absolute width carries a larger error — which is why the count is quoted as a ratio. *Verified: $P=331.8$ MeV; tree $\Gamma_{\nu\bar\nu}=165.9$ MeV, $\Gamma_{\ell\ell}=83.4$ MeV, ratio $1.989$ ; five-flavour $\Gamma_{\text{had}}^{\text{tree}}=1675$ MeV.*

Go deeper · Derivation B · 1 minExtracting the invisible width and counting species from the lineshape

Derivation BExtracting the invisible width and counting species from the lineshape

The aim here is to turn three numbers a collider actually measures — the peak position, the peak height, and the hadron-to-lepton ratio — into the invisible width and then into $N_\nu$ , without detecting a neutrino. The engine is the Breit-Wigner peak cross section, in which the total width appears squared in the denominator; we invert it to isolate the invisible slice, then divide by the clean theoretical ratio. This is the exact chain LEP followed to reach three families.

Step 1. Start from the peak height $\sigma^0_{\text{had}}=\dfrac{12\pi}{m_Z^2}\dfrac{\Gamma_{ee}\Gamma_{\text{had}}}{\Gamma_Z^2}$ and use universality $\Gamma_{ee}=\Gamma_{\ell\ell}$ with $R_\ell=\Gamma_{\text{had}}/\Gamma_{\ell\ell}$ to write the numerator as $\Gamma_{ee}\Gamma_{\text{had}}=\Gamma_{\ell\ell}^2 R_\ell$ . Why: working in units of $\Gamma_{\ell\ell}$ lets the unknown absolute lepton width cancel from the final ratio.

Step 2. Write the total width the same way, $\Gamma_Z=\Gamma_{\ell\ell}\left(R_\ell+3+\Gamma_{\text{inv}}/\Gamma_{\ell\ell}\right)$ , and substitute into the peak formula. The $\Gamma_{\ell\ell}^2$ cancels, leaving $\sigma^0_{\text{had}}=\dfrac{12\pi}{m_Z^2}\dfrac{R_\ell}{\left(R_\ell+3+\Gamma_{\text{inv}}/\Gamma_{\ell\ell}\right)^2}$ . Why: the only unknown left is the dimensionless $\Gamma_{\text{inv}}/\Gamma_{\ell\ell}$ — the peak height has become a pure counter of the invisible channels.

Step 3. Solve for that ratio: $\left(R_\ell+3+\dfrac{\Gamma_{\text{inv}}}{\Gamma_{\ell\ell}}\right)^2=\dfrac{12\pi R_\ell}{m_Z^2\,\sigma^0_{\text{had}}}$ , hence $\dfrac{\Gamma_{\text{inv}}}{\Gamma_{\ell\ell}}=\sqrt{\dfrac{12\pi R_\ell}{m_Z^2\,\sigma^0_{\text{had}}}}-R_\ell-3$ . Why: this is the master relation of the measurement — every quantity on the right is measured from the lineshape, so the invisible width is now pinned by data alone.

Step 4. Convert with equation (3): $N_\nu=(\Gamma_{\text{inv}}/\Gamma_{\ell\ell})\,(\Gamma_{\ell\ell}/\Gamma_{\nu\bar\nu})_{\text{SM}}$ . The LEP averages return $\Gamma_{\text{inv}}/\Gamma_{\ell\ell}=5.943\pm0.016$ and $N_\nu=2.984\pm0.008$ . Why: the final step is one multiplication by a theory number known to $0.06\%$ , so the remaining uncertainty is experimental — and it lands decisively on three. *Verified: the inversion $\Gamma_{\text{inv}}/\Gamma_{\ell\ell}=\sqrt{12\pi R_\ell/(m_Z^2\sigma^0_{\text{had}})}-R_\ell-3=5.94$ from $\sigma^0_{\text{had}}=41.54$ nb, $R_\ell=20.767$ ; $N_\nu=5.94/1.9912=2.984$ .*

Go deeper · Derivation C · 1 minWhy an extra species lowers the peak — the sensitivity of the height

Derivation CWhy an extra species lowers the peak — the sensitivity of the height

This short derivation makes precise the claim behind the three-curve figure: the peak hadronic cross section falls monotonically with the number of neutrino species, so the height alone distinguishes two, three and four families. We hold the visible widths fixed, vary only $N_\nu$ , and evaluate the peak in each case to show the curves separate well beyond the experimental error.

Step 1. With the visible widths fixed, the only $N_\nu$ -dependence in $\sigma^0_{\text{had}}=\dfrac{12\pi}{m_Z^2}\dfrac{\Gamma_{ee}\Gamma_{\text{had}}}{\Gamma_Z^2}$ enters through $\Gamma_Z=\Gamma_{\text{had}}+3\Gamma_{\ell\ell}+N_\nu\Gamma_{\nu\bar\nu}$ in the denominator, so $\sigma^0_{\text{had}}\propto\Gamma_Z^{-2}$ falls as $N_\nu$ grows. Why: adding a channel widens the resonance and, the numerator fixed, must lower its peak — width and height trade off through the fixed area under a Breit-Wigner.

Step 2. Take the peak ratio relative to $N_\nu=3$ : $\dfrac{\sigma^0(N_\nu)}{\sigma^0(3)}=\left(\dfrac{\Gamma_Z(3)}{\Gamma_Z(N_\nu)}\right)^2$ . Each step in $N_\nu$ shifts $\Gamma_Z$ by $167$ MeV out of $2495$ — a $6.7\%$ width shift, hence roughly a $13\%$ peak shift. Why: the squared dependence doubles the sensitivity, so a percent-level width change becomes a several-percent height change a precision scan resolves easily.

Step 3. Evaluate: $N_\nu=2$ gives $\Gamma_Z=2.33$ GeV and $\sigma^0=47.6$ nb; $N_\nu=3$ gives $2.495$ GeV and $41.5$ nb; $N_\nu=4$ gives $2.67$ GeV and $36.4$ nb. The measured $41.54\pm0.04$ nb sits on the three-family curve, many standard deviations from both neighbours. Why: the $\sim5$ nb gaps dwarf the $0.04$ nb error, so the count is unambiguous — a fourth ordinary neutrino would have depressed the peak far below what was seen. *Verified: $\sigma^0(2,3,4)=47.6,\,41.5,\,36.4$ nb; each species shifts $\Gamma_Z$ by $\approx167$ MeV and the peak by $\approx5$ nb, versus a $0.04$ nb error.*

Limiting cases

Case	Result	Meaning
Exactly 3 massless active $\nu$	$\Gamma_{\text{inv}}=3\Gamma_{\nu\bar\nu}\approx499$ MeV	the Standard Model — matches the data, $N_\nu=2.98$
Hypothetical $N_\nu=4$	$\Gamma_Z\to2.67$ GeV, $\sigma^0_{\text{had}}\to36.4$ nb	peak depressed by $\sim5$ nb — excluded by many $\sigma$
Hypothetical $N_\nu=2$	$\Gamma_Z\to2.33$ GeV, $\sigma^0_{\text{had}}\to47.6$ nb	peak too high — also excluded
Fourth neutrino with $m_\nu>m_Z/2$	channel closed, $\Gamma_{\text{inv}}$ unchanged	not counted: too heavy to be produced in $Z$ decay
Sterile / right-handed $\nu$	no $Z$ coupling, $\Gamma_{\nu\bar\nu}=0$	invisible to the width — LEP simply cannot see it

Common trap“N_ν = 3 proves there are exactly three neutrinos in the universe.”

It proves something narrower and more precise. The LEP number counts only neutrinos that are both light enough to be produced ( $m_\nu<m_Z/2\approx45.6$ GeV) and coupled to the Z with standard strength ( $T_3=+\tfrac12$ ). A fourth-generation neutrino heavier than $45.6$ GeV leaves $\Gamma_{\text{inv}}$ untouched — the channel is closed; a sterile or right-handed neutrino has no Z coupling and contributes nothing, however light. So $N_\nu=2.984$ reads as: exactly three light, weakly-interacting, "active" species. It rules out a fourth ordinary generation, but does not forbid heavy neutral leptons, sterile states, or the small neutrino masses we know exist from oscillations. A second confusion is with the cosmological $N_{\text{eff}}$ : that counts relativistic degrees of freedom in the early universe and equals $3.044$ (not exactly 3, from non-instantaneous decoupling) — a related but distinct observable.

Worked examples

1 · Counting neutrinos from the lineshape (fully worked)

The LEP combination gives the peak hadronic cross section $\sigma^0_{\text{had}}=41.540$ nb, the hadron-to-lepton width ratio $R_\ell=20.767$ , and $m_Z=91.1876$ GeV. Using $(\Gamma_{\nu\bar\nu}/\Gamma_{\ell\ell})_{\text{SM}}=1.9912$ , find $N_\nu$ .

Step 1 — put the cross section in natural units. With $1\ \mathrm{GeV^{-2}}=3.894\times10^{5}$ nb,

\sigma^0_{\text{had}}=\frac{41.540}{3.894\times10^{5}}\ \mathrm{GeV^{-2}}=1.0668\times10^{-4}\ \mathrm{GeV^{-2}}.

Step 2 — evaluate the master ratio. The argument of the square root in equation (4) is

\frac{12\pi R_\ell}{m_Z^2\,\sigma^0_{\text{had}}}=\frac{12\pi\times20.767}{(91.1876)^2\times1.0668\times10^{-4}}=\frac{782.9}{0.8871}=882.6,

so $\sqrt{882.6}=29.709$ .

Step 3 — isolate the invisible width. Subtract the visible bookkeeping:

\frac{\Gamma_{\text{inv}}}{\Gamma_{\ell\ell}}=29.709-R_\ell-3=29.709-20.767-3=\mathbf{5.942}.

The invisible channels carry almost six lepton-widths — three neutrinos, each worth about two lepton units.

Step 4 — convert to a species count. Using equation (3),

N_\nu=\frac{\Gamma_{\text{inv}}}{\Gamma_{\ell\ell}}\left(\frac{\Gamma_{\ell\ell}}{\Gamma_{\nu\bar\nu}}\right)_{\!\text{SM}}=\frac{5.942}{1.9912}=\mathbf{2.984}.

Restoring experimental errors gives $N_\nu=2.984\pm0.008$ . Three light neutrino families, from a peak height and a width ratio, without detecting a neutrino.

2 · The direct subtraction cross-check (fully worked)

From the lineshape fits, $\Gamma_Z=2495.2$ MeV, $\Gamma_{\text{had}}=1744.4$ MeV, $\Gamma_{\ell\ell}=83.98$ MeV per flavour. Using the Standard Model $\Gamma_{\nu\bar\nu}=167.2$ MeV, count the neutrinos by direct subtraction.

Step 1 — total up the visible width. The three charged-lepton pairs contribute $3\Gamma_{\ell\ell}=3\times83.98=251.9$ MeV, so $\Gamma_{\text{vis}}=\Gamma_{\text{had}}+3\Gamma_{\ell\ell}=1744.4+251.9=1996.3$ MeV.

Step 2 — subtract to get the invisible width. Equation (2) gives

\Gamma_{\text{inv}}=\Gamma_Z-\Gamma_{\text{vis}}=2495.2-1996.3=\mathbf{498.9\ MeV}.

Step 3 — divide by the single-species width.

N_\nu=\frac{\Gamma_{\text{inv}}}{\Gamma_{\nu\bar\nu}}=\frac{498.9}{167.2}=\mathbf{2.984}.

The two independent routes — the lineshape ratio and the direct subtraction — agree to the last digit, a strong consistency check.

Step 4 — sanity-check the divisor. The tree-level neutrino width from equation (1) is $\Gamma_{\nu\bar\nu}=P\times\tfrac12=165.9$ MeV, raised by corrections to $167.2$ MeV. Had you wrongly used the charged-lepton width $84$ MeV you would find $N_\nu\approx5.9$ — which is $\Gamma_{\text{inv}}/\Gamma_{\ell\ell}$ , not a species count. The invisible $499$ MeV is three neutrino species, not six leptons — the factor of two matters.

Physics of the portrait

The count was settled not by one discovery but by a convergence — a cosmological argument that predicted three before the accelerators confirmed it, and a precision machine that nailed the number to a fraction of a percent.

Origins · 1 minthe invisible width and how this was discovered

the invisible width

Γ_inv = Γ_Z − Γ_had − 3Γ_ℓℓ, the missing slice that counts neutrinos · electroweak precision at the Z pole

The idea of counting neutrino species without catching one is older than LEP. In 1977 Gary Steigman, David Schramm and James Gunn showed that the primordial abundance of helium-4, forged in the first minutes of the Big Bang, depends on how many light neutrino species sped up the cosmic expansion — more species, more helium — and turned the measured helium fraction into a bound of roughly $N_\nu\lesssim4$ . It already whispered "three." Two years later Karl Gaemers, Raymond Gastmans and François Renard proposed the collider counterpart: at an $e^+e^-$ machine the reaction $e^+e^-\to\nu\bar\nu\gamma$ tags invisible neutrino production by a single radiated photon, its rate scaling linearly with the number of species. When LEP switched on at CERN in 1989 — followed within weeks by the SLC at SLAC — four detectors, ALEPH, DELPHI, L3 and OPAL, scanned the Z resonance with a beam energy calibrated so finely it registered the tides in Lake Geneva and passing TGV trains. The invisible width emerged from the peak's shape: everything visible was subtracted from the total, and the leftover $\Gamma_{\text{inv}}\approx499$ MeV, divided by $167$ MeV per species, gave three. Over the following decade the experiments logged some seventeen million Z decays, and the final 2006 combination stands as a definitive number of the field: $N_\nu=2.9840\pm0.0082$ — a quantity measured precisely from the absence of a signal, and from that absence a count of the neutrinos of the universe.

History

1977

— Steigman, Schramm and Gunn bound the neutrino species count from primordial helium

1979

— Gaemers, Gastmans and Renard propose direct counting via e⁺e⁻ → ν ν̄ γ

1989

— LEP and the SLC begin the Z scan; the invisible width already points to three families

1990

— single-photon counting at the Z corroborates the lineshape result

2006

— the final LEP combination reports N_ν = 2.9840 ± 0.0082

Going further

Texts. Thomson, Modern Particle Physics, §16.1 (the Z resonance) and §16.1.1 (the Z production cross section, Eqs. 16.16–16.17) · Halzen & Martin, Quarks and Leptons, §15 (the electroweak theory and Z decays) · Griffiths, Introduction to Elementary Particles, 2nd ed., §10.6 · Peskin & Schroeder, An Introduction to Quantum Field Theory, §20.3 (Z couplings and widths) · Schwartz, Quantum Field Theory and the Standard Model, §29 (electroweak precision).

Provenance. The Breit-Wigner peak formula $\sigma^0_{\text{had}}=(12\pi/m_Z^2)\Gamma_{ee}\Gamma_{\text{had}}/\Gamma_Z^2$ is recalled from aq-pp-resonances-breit-wigner and aq-pp-z-resonance-lep; the neutral-current couplings $g_V^f,g_A^f$ that fix the partial widths are owned by aq-pp-neutral-current and aq-pp-z-decays. A verifier should confirm: (1) the tree-level widths $\Gamma_{\nu\bar\nu}=165.9$ MeV, $\Gamma_{\ell\ell}=83.4$ MeV and ratio $1.989$ from equation (1) with $\sin^2\theta_W=0.231$ ; (2) the lineshape inversion $\Gamma_{\text{inv}}/\Gamma_{\ell\ell}=\sqrt{12\pi R_\ell/(m_Z^2\sigma^0_{\text{had}})}-R_\ell-3=5.94$ from $\sigma^0_{\text{had}}=41.54$ nb, $R_\ell=20.767$ ; (3) the direct subtraction $\Gamma_{\text{inv}}=\Gamma_Z-\Gamma_{\text{had}}-3\Gamma_{\ell\ell}=499$ MeV; (4) both routes give $N_\nu=2.984$ . Standard values ( $m_Z$ , $\Gamma_Z$ , $\sigma^0_{\text{had}}$ , $R_\ell$ , $N_\nu$ ) follow the LEP Electroweak Working Group combination and the Particle Data Group.

Prerequisites. The Z Resonance at LEP · Breit-Wigner Resonances · Z Decays and Partial Widths · The Weak Neutral Current. Leads to. Neutrino Flavours · Open Questions of the Standard Model · Electroweak Radiative Corrections.

Ask the tutor: Ask me to walk the Z lineshape from a measured peak height and width ratio all the way to the neutrino count, or to show why a sterile neutrino stays invisible to the invisible width.

How exactly did LEP count neutrinos without detecting any?

By subtraction. The total width $\Gamma_Z$ is measured from the breadth of the resonance, and every visible channel — hadrons and the three charged-lepton pairs — from its own event rate. Whatever is left over cannot be going anywhere the detector sees, so it must be the Z decaying to neutrino pairs. That leftover is the invisible width, $\Gamma_{\text{inv}}\approx499$ MeV; dividing by the calculable $\Gamma_{\nu\bar\nu}\approx167$ MeV per species gives $N_\nu\approx3$ . The neutrinos announce themselves only through the width they carry off, so the count is read from the shape of a curve, not from any neutrino hitting the apparatus.

Does N_ν = 3 rule out a fourth generation entirely?

It rules out a fourth ordinary generation with a light neutrino. A conventional fourth family would bring a fourth active neutrino below about $45$ GeV, which would have widened $\Gamma_Z$ and depressed the peak far beyond what was seen — so a sequential fourth generation is excluded. But the result is silent about heavier or differently-coupled states: a neutrino above $m_Z/2$ cannot be produced in Z decay, and a sterile or right-handed neutrino does not couple to the Z at all. So $N_\nu=3$ closes the door on light active neutrinos while leaving open the searches for heavy neutral leptons and sterile states.

What is the difference between the LEP N_ν and the cosmological N_eff = 3.044?

They sound identical but measure different things. The LEP $N_\nu$ counts neutrino species that couple to the Z and are lighter than $m_Z/2$ — a laboratory measurement of the electroweak neutrino sector. The cosmological $N_{\text{eff}}$ counts the effective relativistic degrees of freedom in the early universe, inferred from how fast the cosmos expanded during nucleosynthesis and recombination. Its Standard Model value is $3.044$ , slightly above three because neutrinos did not decouple instantaneously and picked up a little extra energy from electron-positron annihilation. The two are consistent with three ordinary neutrinos but are not the same observable: one is a collider width, the other a count of anything relativistic in the primordial plasma.

Why is the invisible width literally invisible?

Because neutrinos interact only weakly, and the weak force is feeble at these energies. A neutrino produced in a Z decay flies out of a detector without ionising an atom, depositing energy in any calorimeter, or curving in the magnetic field — neutral, colourless, it passes through the apparatus (and the Earth) as if it were not there. So the $Z\to\nu\bar\nu$ decays register as nothing: the event simply has energy missing. The only trace they leave is collective — the extra width they add to the resonance — which is exactly what makes the invisible width measurable though no individual neutrino is ever seen.

Could there be neutrinos LEP simply cannot see?

Yes, and this is the crucial caveat. LEP's method is blind to any neutral fermion that fails one of two conditions: it must couple to the Z with standard strength, and it must be lighter than half the Z mass. A sterile neutrino carries no weak isospin and contributes zero to $\Gamma_{\text{inv}}$ however light it is; a heavy neutral lepton above $45$ GeV is kinematically forbidden in Z decay. We now know from oscillations that neutrinos do have small masses and mix — physics beyond what LEP measured. So $N_\nu=3$ is a precise statement about light active species, not a claim that no other neutral leptons exist.

What sets the single-species width Γ_νν ≈ 167 MeV?

It comes straight from the electroweak coupling of the Z to a neutrino. The partial width is $\Gamma(Z\to\nu\bar\nu)=(G_F m_Z^3/6\sqrt2\,\pi)[(g_V)^2+(g_A)^2]$ , and for a neutrino $g_V=g_A=\tfrac12$ because it has weak isospin $+\tfrac12$ and zero charge — the charge-dependent term drops out. That gives $(g_V)^2+(g_A)^2=\tfrac12$ and a tree-level width of about $166$ MeV, lifted by radiative corrections to $167.2$ MeV. Because the neutrino carries no charge, this is one of the cleanest predictions in the Standard Model — no mixing-angle uncertainty — which is why the invisible width converts to a species count with so little theoretical ambiguity.