Two particles came out with the same mass, the same lifetime and opposite parities. Because the measurement had no free parameters left to blame, something else had to give.
🎯 Why this matters
A result with no adjustable parts is the only kind that can force a theory to move. Anything with a knob in it gets turned instead, and the crisis is postponed rather than resolved.§4.4 built a tool that turns a scatter plot into a spin–parity measurement. This section uses it three times, and the first use broke physics.
The same tool applied to the K meson produced a contradiction so sharp that resolving it required abandoning a symmetry everyone assumed was exact. Applied to the η it identified an electromagnetic decay hiding among hadrons. Applied to the ω it gave the first vector meson. Three Dalitz plots, three results, one of which is the reason Chapter 7 exists.
The θ–τ puzzle
By the early 1950s cosmic-ray experiments had found several decay topologies belonging to strange particles of similar mass and similar lifetime. They were named as though they were different particles: θ decaying to 2π, τ decaying to 3π, and K with an assortment of other modes.
🔬 Experiment card — the G-stack, 1954
Apparatus
A 15-litre stack of nuclear emulsion — 63 kg of it, proposed by M. Merlin in 1953 — flown to 27 000 m on a balloon over the Po valley. The point of the size is containment: in a large enough volume, the complete set of tracks from a decay stays inside the emulsion.What is measured
Range, and therefore energy, for every charged decay product — the emulsion technique of §1.13a. With all daughters contained, the parent’s mass follows from energy–momentum conservation with no missing pieces.The result
The different decay modes were cleanly separated and their parent masses measured accurately. They agreed. Around the same time the Bevatron delivered the first K⁺ beams, and emulsion exposures put the θ and τ masses equal to within a few parts per thousand.What it proved
θ and τ are one particle — the K meson. Which immediately created the problem, because the two decay modes demand opposite parities.Fig. 4.14 — the 220 'τ events', K⁺ → π⁺π⁺π⁻
I = 1, J^P = 0⁻
M ∝ constant (or E₃)
Vanishing density: none — which is exactly why this is the assignment the τ and η data support
The events are weighted by |M|², so the empty regions are computed, not drawn. Only zeros are predicted — the overall normalisation and any slowly varying form factor are not, which is why the analysis looks for depleted regions rather than fitting the whole density.
220 simulated events at the K⁺'s own Q value of 75.0 MeV, weighted by the I = 1, J^P = 0⁻ matrix element — which is a constant, so this is what the published plot looks like: no depletion anywhere. Orear, Harris and Taylor's figure is FOLDED about the vertical axis because two of the three pions are identical; the fold is a presentation choice and changes no physics. Step the selector through the other five assignments and every one of them carves a hole this data does not have.
🪜 Why one particle cannot have two parities
Step 1 of 4 — The 3π mode says 0⁻(4.33)
Why you may do this: Compare Fig. 4.14 with the six patterns of §4.4. Only I = 1 with J^P = 0⁻ admits a constant matrix element, and only a constant produces no depleted region anywhere. Every other assignment is excluded by the absence of a hole.
This is the measurement Dalitz invented the plot for, in 1956.
Bettini pp. 148–149. Four steps, each individually unobjectionable, ending in a contradiction.
💡 What this really says — a proof by contradiction with four premises, three of them structural
Read the argument as a proof by contradiction with four premises, three of which are structural:
- kinematics and Bose statistics — used to build the six Dalitz patterns. Not negotiable;
- angular momentum — used to get and hence for two pions. Not negotiable;
- the two parents are one particle — an experimental fact, established well enough that denying it was worse than the alternative;
- parity is conserved in the decay — an assumption nobody had ever thought to question, because it had never failed.
Only the last one can go, and dropping it costs nothing anywhere else provided the failure is confined to weak decays. That is precisely Lee and Yang’s proposal, and it is why §7.2–7.3 is about the weak interaction specifically.
There is a second, quieter violation in the same decay that the book flags: changes isospin by . The K has and the 3π state has , so isospin is not conserved either — which is the ordinary signature of a weak decay and, unlike the parity failure, surprised nobody.
⚙️ Engineer’s bridge — when a measurement contradicts itself, audit the invariants
The shape of this episode is one you meet whenever two independent measurements of the same quantity disagree beyond their errors. The question is never “which measurement is wrong” first — it is which shared assumption both of them depend on.
Here the two “measurements” of used almost disjoint machinery: one a scatter plot and Bose statistics, the other a two-line angular-momentum argument. Almost — but not quite. Both assumed that the parity of the initial state equals the parity of the final state, and that assumption was the only thing they had in common besides the particle itself.
The engineering habit that transfers: when two independent paths to a number disagree, enumerate what they share. A shared calibration, a shared clock, a shared assumption about linearity. The disagreement is usually located in the intersection, not in either path — and the intersection is often something so basic that nobody listed it as an assumption at all.
Where it breaks: in physics you can only drop an invariant if the replacement theory still explains everything the old one did. Lee and Yang’s proposal survived because parity violation could be confined to the weak interaction, leaving the strong and electromagnetic results of Chapters 3 and 4 untouched. An engineer who drops an assumption usually has to re-derive far less.
Where it breaks: “drop the assumption with the smallest blast radius” is good engineering and it is not how the θ–τ puzzle was actually resolved. Parity was not a small assumption — it was believed to be exact, and abandoning it was the most expensive option on the table, not the cheapest. What made it right was evidence, not blast radius: Wu measured it. The heuristic tells you where to look first; it is not a criterion for what to believe, and treating it as one would have kept parity and invented a second particle.
The η, and a decay that is not strong
- the 233-event sample
- phase space (the book’s dashed curve)
🔬 Experiment card — Pevsner, Block and collaborators, 1961
Apparatus
The 72″ Alvarez bubble chamber (§1.13b) filled with liquid deuterium, exposed to a π⁺ beam of 1.23 GeV/c momentum. Deuterium rather than hydrogen because the reaction needs a neutron target inside the deuteron.What is measured
The reaction , event by event, with the π⁰ reconstructed as a missing mass. Then the invariant mass of the three-pion system — production, exactly as §4.1 defined it.The result
A narrow peak at MeV with MeV, and no charged partner ever observed — so . Soon afterwards the decay η → 2γ was found to happen at a rate comparable to the 3π mode: today and .What it proved
A new pseudoscalar meson, and — through its Dalitz plot and its G-parity — the fact that its hadronic decay is not strong. The reasoning is below.Fig. 4.15(b) — the η → π⁺π⁻π⁰ Dalitz plot
I = 1, J^P = 0⁻
M ∝ constant (or E₃)
Vanishing density: none — which is exactly why this is the assignment the τ and η data support
The events are weighted by |M|², so the empty regions are computed, not drawn. Only zeros are predicted — the overall normalisation and any slowly varying form factor are not, which is why the analysis looks for depleted regions rather than fitting the whole density.
At the η's Q value of 133.7 MeV, weighted by the assignment the data supports. As for the τ events, the distribution is featureless — and among the six patterns of §4.4 only I = 1 with J^P = 0⁻ has no vanishing region. Try the others: each predicts a hole the data does not show.
The argument in one line. C(η) = +1 comes from the observed 2γ decay; I(η) = 0 from the absence of charged partners; and G(nπ) = (−1)ⁿ is Eq. (3.51).
Every symbol, one at a time
Hover or tap a symbol above — it lights up in the equation and its meaning, units and type appear here.
💡 What this really says — a hadronic decay that is not strong
An η decaying to three pions looks like the most ordinary hadronic process imaginable. G-parity says it cannot be strong, and the rate confirms it: a 28 % branching ratio into 3π sitting beside a 39.4 % branching ratio into two photons. A strong decay competing on level terms with an electromagnetic one is a contradiction in itself — strong decays are typically times faster.
The width is the clincher. At MeV the η lives s, which is a hundred times longer than a strong decay of that mass would take. Compare the ω below: same 3π final state, nearly the same phase space, and a width of 8.7 MeV because for the ω the decay is strong.
The book adds the neat closing observation. Why does the η not simply decay to two pions, which would conserve G? Because cannot have — Eq. (3.5) gives parity and , so and is not on the list. And four pions are forbidden by energy conservation. The η is stuck, and being stuck is why it is narrow.
🔢 Worked example — the η and the ω, side by side
The two mesons on the same plot in Fig. 4.15(a) have the same decay products and nearly the same available energy. Everything else about them differs, and the G-parity bookkeeping says why.
Reproduce it
hbar = 6.582119569e-22 # MeV s
mpi, mpi0 = 139.57039, 134.9768
print("quantum numbers, G = C (-1)^I")
for nm, m, G, C, I in (("eta", 547.862, 1.310, +1, 0), ("omega", 782.66, 8.680, -1, 0)):
g = C * (-1)**I
verdict = "match -> strong" if g == -1 else "MISMATCH -> not strong"
print(f" {nm:6s} I={I} C={C:+d} -> G = {g:+d} 3pi has G = (-1)^3 = -1 {verdict}")
print("consequences")
for nm, m, G in (("eta", 547.862, 1.310), ("omega", 782.66, 8.680)):
print(f" {nm:6s} Gamma = {G:7.3f} MeV -> tau = {hbar/G:.2e} s "
f"Q(3pi) = {m - 2*mpi - mpi0:5.1f} MeV")
print(f" the omega is {8.680/1.310:.1f}x wider despite a similar final state")
print("why the eta cannot take the easy way out")
print(" 2pi from a 0- parent: two spinless bosons have J = l and P = (-1)^l,")
print(" so J^P = 0+, 1-, 2+, ... and 0- is not available")
print(f" 4pi: needs 4 x {mpi:.1f} = {4*mpi:.1f} MeV > m(eta) = {547.862:.1f} MeV, "
f"forbidden by {4*mpi - 547.862:.1f} MeV")quantum numbers, G = C (-1)^I eta I=0 C=+1 -> G = +1 3pi has G = (-1)^3 = -1 MISMATCH -> not strong omega I=0 C=-1 -> G = -1 3pi has G = (-1)^3 = -1 match -> strong consequences eta Gamma = 1.310 MeV -> tau = 5.02e-22 s Q(3pi) = 133.7 MeV omega Gamma = 8.680 MeV -> tau = 7.58e-23 s Q(3pi) = 368.5 MeV the omega is 6.6x wider despite a similar final state why the eta cannot take the easy way out 2pi from a 0- parent: two spinless bosons have J = l and P = (-1)^l, so J^P = 0+, 1-, 2+, ... and 0- is not available 4pi: needs 4 x 139.6 = 558.3 MeV > m(eta) = 547.9 MeV, forbidden by 10.4 MeV
Note what is doing the work. The η is not narrow because of a small coupling — it is narrow because every faster channel is closed by a selection rule or by energy, and what is left has to go through the electromagnetic interaction. The same pattern will explain the φ at the end of this page and the J/ψ in §4.9.
The η′, briefly
A second isosinglet pseudoscalar exists, with the same quantum numbers as the η: MeV, MeV, and important electromagnetic decays (2γ, ργ, ωγ). The book flags one thing about it and moves on: its mass is enormous compared with the pions, in a way no naive picture explains. §4.7 is where that becomes a real problem, and §6.8 is where it is solved.
The ω, and the first vector meson
| total charge of the triplet | combinations per event | which ones | what the mass distribution shows |
|---|---|---|---|
| 0 | 4 | π⁺ᵢ π⁻ⱼ π⁰ — two choices of each charged pion | A narrow peak at 782 MeV |
| ±1 | 4 | π⁺π⁺π⁻ and π⁻π⁻π⁺, two of each sign | No peak — smooth phase space |
| ±2 | 2 | π⁺π⁺π⁰ and π⁻π⁻π⁰ | No peak |
Ten triplets, and only the neutral ones resonate. <strong>That single fact fixes the isospin at I = 0</strong> — an I = 1 state would have charged partners appearing in the charge-±1 combinations, and an I = 2 state would show up at ±2 as well. It is the same logic as the η's missing charged states, made quantitative by having all the charge combinations inside one event sample.
🔬 Experiment card — Maglić, Alvarez and collaborators, 1961
Apparatus
The same 72″ Alvarez hydrogen bubble chamber, exposed to the Bevatron’s antiproton beam — the beam whose construction was the subject of §2.6–2.7.What is measured
The invariant mass of all ten three-pion combinations in every event, sorted by the triplet’s charge — and then, for events in the peak, the Dalitz plot of the neutral triplet.The result
MeV, MeV, a peak in the neutral combinations only. The Dalitz plot’s radial distribution — events counted in zones of equal area — matches plus a flat background.What it proved
from the charge states and from the Dalitz plot: the first vector meson. And since follows from , its G-parity is , matching the 3π final state — so unlike the η, this decay is strong.Fig. 4.16(b) — the ω → π⁺π⁻π⁰ Dalitz plot
I = 0, J^P = 1⁻
M ∝ q = p₁ × p₂
Vanishing density: the whole boundary — there the three momenta are collinear and the normal vanishes
The events are weighted by |M|², so the empty regions are computed, not drawn. Only zeros are predicted — the overall normalisation and any slowly varying form factor are not, which is why the analysis looks for depleted regions rather than fitting the whole density.
Weighted by the I = 0, J^P = 1⁻ matrix element M ∝ q = p₁ × p₂, which vanishes wherever the three momenta become collinear — so the density falls to zero all along the boundary and the events pile up in the middle. That is exactly what the book's radial distribution, counted in zones of equal area, is measuring. Switch to I = 1, J^P = 0⁻ and the plot fills its rim: the two are trivially distinguishable with a few hundred events.
💡 What this really says — why a radial histogram is the right summary
The ω’s pattern is the only one of the six that is rotationally symmetric about the centre of the plot: depends on how far a configuration is from the collinear limit, and nothing else. So the whole two-dimensional distribution can be compressed, with no loss, into a one-dimensional radial histogram — which is precisely what Fig. 4.16(b) plots, in zones of equal area so that a flat background stays flat.
That is a genuinely good piece of analysis design. Two hundred events spread over a two-dimensional plot are sparse; the same events in ten radial bins are not. Choosing the projection along which the signal survives and the background stays featureless is what makes a small sample decisive.
Contrast the η and the τ, where the prediction was “no structure at all”. There, no projection helps and you simply have to look at the plot — which is also why those two results needed the argument from absence that §4.4’s bridge described.
The rest of the vector nonet nonet the nine mesons of one J^P, filling an SU(3)_f octet plus a singlet, since 3 ⊗ 3̄ = 1 ⊕ 8. One nonet is pseudoscalar (0⁻), one vector (1⁻). defined in §4.5 — open in glossary
There are nine vector mesons, in isospin multiplets identical to those of the pseudoscalars. The book lists them; their quantum numbers are the argument of §4.7.
| meson | m (MeV)↕ | ↕ | principal decay | |
|---|---|---|---|---|
| 1⁺ | 775 | 2π — hence G = +1, and the ρ⁰ has C = −1 | ||
| 1/2 | 892 | 51 MeV | Kπ; two charge states each for K* and its antiparticle | |
| 0⁻ | 783 | 8.7 MeV | 3π, strongly — the subject of this page | |
| 0⁻ | 1019 | 4.25 MeV |
Compare the widths: ρ 149 MeV, K* 51, ω 8.7, φ 4.25 — a factor of 35 across four particles of similar mass, all decaying strongly. <strong>Something is suppressing the heavier two, and phase space is not it.</strong>
📏 The φ anomaly, quantified
The φ has essentially every advantage in the 3π channel and does not use it:
| channel | Q value | branching ratio |
|---|---|---|
| φ → 3π | 605 MeV | 15.6 % |
| φ → K⁺K⁻ | 32 MeV | 83 % (both KK̄ modes) |
Phase space alone would favour 3π by orders of magnitude — a 32 MeV Q value is barely enough to make the kaons at all, and it is why the φ’s width is only 4.25 MeV, small “by strong interaction standards”.
So the KK̄ preference is a dynamical suppression of the 3π mode, not a kinematic preference for the kaons. The book names the two ingredients and defers both: the φ’s quark content is almost purely ss̄ (§4.7), and there is a QCD rule that penalises decays in which the initial quarks have to annihilate (§6.5). The same mechanism, one flavour up, is why the J/ψ of §4.9 is a thousand times narrower still.
🔑 If you remember only three things
-
One tool, three uses, three kinds of answer. A contradiction that broke a conservation law, a rate that disagreed with its own appearance, and a particle nobody had predicted.
-
A decay that looks strong and is not is the most informative kind. Three pions from a hadron is the most ordinary process imaginable, and here it is forbidden to the strong interaction.
-
The same chamber found both new mesons inside a year. One instrument and two beams: by 1961 the limiting resource was what you could shoot, not what you could photograph.
Where this goes next
- §4.6–4.7 explains the pattern behind all nine mesons of each spin–parity: SU(3), the nonet, and the mixing that makes the ω non-strange and the φ strange.
- Chapter 7 is the resolution of the θ–τ puzzle — parity violation in the weak interaction, and the experiments that confirmed it within a year of the proposal.
- §6.5 supplies the dynamical suppression that keeps the φ from decaying to pions, and §6.8–6.9 explains the η′‘s embarrassing mass.
- §4.9 repeats the whole φ story at four times the mass, where the narrowness becomes so extreme that it announced a new quark.
✅ Check yourself — pseudoscalar and vector mesons
0/5 answered · 0 correct
1.The θ–τ puzzle rests on four premises. Which one did Lee and Yang propose dropping, and why that one?
2.Both the τ events and the η give featureless Dalitz plots. Why is a plot with no structure a measurement rather than a failure?
3.The η and the ω both decay to π⁺π⁻π⁰, with widths of 1.3 MeV and 8.7 MeV. What distinguishes them?
4.In the ω's Dalitz plot the density falls to zero all along the boundary. What does that tell you, and why is a radial histogram the right way to summarise it?
5.The φ has 605 MeV of phase space for 3π and 32 MeV for K⁺K⁻, and takes the kaons 83 % of the time. What does that imply?