The photograph that confirmed the scheme also made its worst problem harder: three identical quarks were sitting in one state, which the Pauli principle forbids outright.
🎯 Why this matters
Colour was invented as bookkeeping and ended up as a charge. A label added to satisfy an exclusion rule became the source of the strong interaction, which is the largest promotion any quantum number gets in this book.§4.7 got mesons for free. A quark and an antiquark, nine places, nine particles — and the only subtlety was which two of the nine sat at the centre. Baryons are three quarks, and three is where combinatorics starts to bite.
Here is the problem in one line. The quark content does not name the
particle. uud is the proton, but it is also the Δ⁺. uds is the Λ⁰, and the
Σ⁰, and the Σ⁰(1385), and the Λ(1405) — four particles, one bag of quarks.
Something other than which quarks must be distinguishing them, and that
something is how the state behaves when you swap two of the quarks over.
⚙️ Engineer’s bridge — the symmetry label is part of the type
You already know the two-element version of this. Any function of two variables splits uniquely into an even part and an odd part, , and those two halves never mix under anything that treats the variables alike. Symmetric and antisymmetric are not descriptions you apply afterwards; they are invariant subspaces, and any operation that does not care which variable is which acts inside them separately.
Three variables give the same story with more room. The 27 orderings of three
quarks decompose into blocks that a flavour-blind interaction can never mix, and
a particle is one vector in one block. So the “type” of a baryon is
(quark content, symmetry block) — the multiset alone under-determines it,
exactly as a struct and a union with the same fields are different types.
Where the analogy breaks: with three or more slots the permutation group is non-abelian, so besides “symmetric” and “antisymmetric” there are blocks that are neither — the mixed mixed symmetry (m,s and m,a) a three-body state symmetric (M,S) or antisymmetric (M,A) under the exchange of a chosen pair but not under all of them. The two mixed octets of 3 ⊗ 3 ⊗ 3 combine with the two mixed spin doublets to make one totally symmetric 1/2⁺ octet. defined in §4.8 — open in glossary ones. There is no two-variable intuition for those, and they turn out to be where the proton lives.
The tenth member of the decuplet
Start with the easy case: three identical quarks. There is no ambiguity at
all — uuu can only be the baryon Δ⁺⁺(1232)
m = 1.232 GeV · Q = +2 · JP = 3/2+ content uuu τ / Γ = 118 ± 2 MeV
open in the particle explorer
, ddd only the Δ⁻. And that
raises a question the decuplet decuplet the ten J^P = 3/2⁺ baryons, 10_S of 3 ⊗ 3 ⊗ 3, with equal mass spacing of ≈145 MeV per unit of strangeness. Its tenth member, the Ω⁻, was predicted before it was found.
defined in §4.2 — open in glossary
of
§4.2 leaves hanging: what about sss?
SU(3)_f says it must exist. A hyperon hyperon a baryon heavier than the nucleon carrying non-zero strangeness; the metastable ones are the Λ, three Σ and two Ξ, all spin ½, all decaying weakly in about 100 ps except the Σ⁰. defined in §2.1-2.2 — open in glossary with strangeness , isospin zero (because the s quark has none), sitting at the bottom vertex of the 3/2⁺ triangle. Gell-Mann called it the Ω⁻, and he could predict its mass, because the decuplet is evenly spaced.
Predict a particle that has never been seen
import numpy as np
# the 3/2+ decuplet as Fig. 4.23 rounds it (MeV)
rows = [("Delta ", 0, 1236), ("Sigma*", -1, 1385), ("Xi* ", -2, 1530)]
print("rung S mass step")
prev = None
for name, S, m in rows:
step = " --" if prev is None else f"{m - prev:+4.0f}"
print(f"{name} {S:2d} {m:4.0f} {step}")
prev = m
steps = [rows[i + 1][2] - rows[i][2] for i in range(len(rows) - 1)]
print(f"\nsteps {steps} MeV; Bettini quotes 'about 145'")
for label, d in [("book's 145", 145), ("measured mean", np.mean(steps))]:
print(f" {label:14s}: m(Omega-) = 1530 + {d:.0f} = {1530 + d:.0f} MeV")
print(" measured : m(Omega-) = 1672 MeV")
thresh = 1315 + 494 # Xi0 + K-, the lightest S = -3, B = 1 state
print(f"\nlightest S=-3 final state Xi0 K- = {thresh} MeV")
print(f" {thresh - 1672} MeV ABOVE the Omega-: the strong decay is closed, so it must be weak")
print(f" ctau = {299.792458 * 82e-3:.1f} mm -> a visible track, not a resonance")rung S mass step Delta 0 1236 -- Sigma* -1 1385 +149 Xi* -2 1530 +145 steps [149, 145] MeV; Bettini quotes 'about 145' book's 145 : m(Omega-) = 1530 + 145 = 1675 MeV measured mean : m(Omega-) = 1530 + 147 = 1677 MeV measured : m(Omega-) = 1672 MeV lightest S=-3 final state Xi0 K- = 1809 MeV 137 MeV ABOVE the Omega-: the strong decay is closed, so it must be weak ctau = 24.6 mm -> a visible track, not a resonance
💡 What this really says — the prediction has two halves, and the second one is the bold one
Predicting a mass from a linear extrapolation is not, by itself, brave. Three points on a line, ask for the fourth: any curve-fitter does that.
The second half is what makes this famous. Once the mass is ≈1675 MeV, look at what the particle is allowed to decay into. A strong decay must conserve strangeness, so the products must carry between them, and the lightest way to do that is a Ξ⁰ plus a K⁻ — 1809 MeV. The Ω⁻ cannot reach it. Every other member of the decuplet is a resonance, gone in 10⁻²³ s, detectable only as a bump in a mass distribution. This one had to be metastable metastable particle one that is stable against strong decay and so decays weakly or electromagnetically, living long enough to leave a measurable flight path. defined in §2.1-2.2 — open in glossary : it would decay weakly, live ~10⁻¹⁰ s, and leave a track in a bubble chamber.
So SU(3)_f did not just say “a particle exists near 1675 MeV”. It said: it exists, it is qualitatively unlike its nine siblings, here is how it decays, and here is what the photograph will look like. That is a falsifiable prediction with no adjustable parameters left.
Aside — the napkin
At the 1962 International Conference on High-Energy Physics at CERN, Gell-Mann publicly urged experimenters to look for the particle. At lunch, with Nicholas Samios and Jack Leitner of Brookhaven, he sketched on a napkin the decay cascade they should search for. They took the napkin home, showed it to the laboratory director, and got priority beam time on the AGS. Two years later they had it.
The napkin mattered because the prediction was operational: not “a state at 1675 MeV” but “a track that ends in a specific chain of vees”. You can hand that to a scanning table.
The discovery: one photograph is enough
🔬 Experiment card — Barnes et al., Brookhaven 1964
Apparatus
The BNL 80-inch hydrogen bubble chamber (§1.13b) in a magnetic field, exposed to a 5.0 GeV/c K⁻ beam from the AGS. Hydrogen so that the target is a free proton, and a chamber rather than a counter experiment because what is wanted is not a rate but a picture: the full topology of a decay chain, vertex by vertex.What is measured
The momentum of every charged track, from its curvature in the field, plus ionisation density to separate protons from pions. Nothing else. Neutral particles leave no track at all and are never measured — they are reconstructed.The result
A single event, Fig. 4.21, with a short track that ends in a cascade of four vertices. Step through it below.What it proved
MeV and ps, against a prediction of ≈1675 MeV and “metastable”. The tenth member of the decuplet existed with the properties SU(3) demanded — and with that, the classification stopped being numerology.🪜 Fig. 4.21 — the first Ω⁻, one vertex at a time
Step 1 of 8 — a K⁻ beam
Why you may do this: You cannot make strangeness −3 out of nothing: the strong interaction conserves S, so the beam has to bring some in. A 5 GeV/c K⁻ carries S = −1, and the target proton carries none.
The chamber is liquid hydrogen, so the target is a free proton — no nuclear corrections to unfold.
Redrawn from the Barnes et al. (1964) bubble-chamber photograph; the track numbers are the book's. Neutral particles are dashed because they leave nothing — every dashed line on this diagram is an inference, not a measurement.
The production reaction and the chain that follows are the book’s Eqs. (4.49)–(4.53):
One production and four decays. Three of the decays are weak, one is electromagnetic, and nothing here is strong.
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.
📏 The sign of ΔS, and why this site writes |ΔS|
Bettini writes for each of the three weak decays above. With the usual convention they are all : the Ω⁻ goes from to , the Ξ⁰ from to , the Λ from to . The book is tracking the magnitude coming down.
Nothing physical hangs on it, and to avoid a sign the reader has to keep re-deriving, this site writes — which is the actual selection rule: the weak interaction changes strangeness by one unit at a time, never two. That rule is doing real work here. It is why the cascade has three steps instead of one, and why nobody expected Ω⁻ → Λπ⁻ () even though it is kinematically allowed.
Why the track lengths are the whole argument
🌿 The cascade of Fig. 4.21, with each lifetime turned into a flight length
- Ω⁻βγcτ = 29.4 mm
- └Ξ⁰23.6%βγcτ = 137 mm
- └Λ99.5%βγcτ = 143 mm
- └p63.9%— stable, or never seen
- └π⁻63.9%βγcτ = 29.2 m
- └π⁰99.5%βγcτ = 102 nm
- └γ₁98.8%— stable, or never seen
- └γ₂98.8%— stable, or never seen
- └π⁻23.6%βγcτ = 47.3 m
Ω⁻ — Metastable because Ξ⁰K⁻ at 1809 MeV is out of reach. This is the prediction that made the search possible.
Ξ⁰ — Neutral: no track. Reconstructed from the two vertices it connects.
Λ — Neutral too — but its charged decay gives the visible V.
p — The heavily ionising track.
π⁰ — cτ = 25 nm. On any scale a bubble chamber can draw, the π⁰ decays exactly where it was made.
Two-body steps use exact forward kinematics, plab = γ(p* + βE*). Steps with three or more daughters share the parent's momentum by the declared fraction — indicative, not exact.
Drag the momentum slider. The three weak decays give centimetre-scale flights that a bubble chamber resolves; the π⁰ never moves. Had the Ω⁻ been able to decay strongly, its bar would sit thirteen orders of magnitude to the left and there would be no photograph to analyse.
🔢 Worked example — could you have done this with a counter experiment?
No, and the reason is worth stating. The Ω⁻ is produced in a reaction with a cross-section of order a microbarn, and the branching chain above costs another factor 0.236 × 0.995 × 0.639 ≈ 0.15. What makes one event sufficient is not statistics; it is that a bubble chamber measures everything.
Count the constraints. Each of the four vertices imposes energy–momentum conservation: 4 equations. That is 16 equations. The unknowns are the momenta of the two neutrals that are never seen (Ξ⁰, Λ — 3 components each), the π⁰ and the two photons where they are unmeasured, and the mass of the Ω⁻ itself. The system comes out over-determined, which is the crucial word: the fit does not merely solve for the answer, it tests whether the hypothesis is consistent at all. A wrong assignment fails to converge.
That is why the paper’s title could claim a discovery from a single photograph, and why the same result from a counter experiment would have needed thousands of events and still not have been as convincing.
The paradox the discovery made worse
Now the trouble. Look at the three corners of the decuplet: Δ⁺⁺ = uuu,
Δ⁻ = ddd, and the brand-new Ω⁻ = sss. Each is three identical quarks.
📐 Physics you need first — the Pauli principle pauli principle the requirement that the total wave function of identical fermions change sign under the exchange of any two of them. Since a hadron's wave function is a product of a space, spin, flavour and colour factor, the constraint binds only the product — which is the loophole colour exploits and, through Eq. (4.56), the reason only two families of ground-state baryons exist. defined in §4.8 — open in glossary is a statement about swapping
The version taught in chemistry — “no two electrons in the same state” — is a consequence, not the principle. The principle is this.
If two particles are genuinely identical, no measurement can tell which is which, so the physics cannot depend on the labels you gave them. The state can therefore change by at most a phase when you swap them: . Swap twice and you are back where you started, so and the phase is . Only two kinds of particle exist:
- bosons, — the state is unchanged by the swap (photons, pions, the W);
- fermions, — the state changes sign (electrons, quarks, protons).
“No two electrons in the same state” follows immediately: if two fermions were in the identical state, swapping them would change nothing, so , so . There is no such state.
The part that matters here: the antisymmetry applies to the whole wave function, and a hadron’s wave function is a product of several factors. So the requirement is not that any one factor be antisymmetric — it is that the product be. That leaves room for a bargain, and §4.8 is where the bargain gets struck.
For an engineer: this is a parity constraint on the entire state vector under a permutation operator — one global assertion, not a rule about each field separately.
Apply it. All three particles are ground states, so the quarks have zero
orbital angular momentum and is symmetric. All three have
, which for three spin-1/2 quarks means all three spins aligned, so
is symmetric. And the flavour part is uuu — as symmetric
as it is possible to be.
Symmetric × symmetric × symmetric = symmetric. Three identical fermions in a completely symmetric state. The Pauli principle forbids it, and yet there the Ω⁻ is, 2.5 cm of it, on a photograph.
Colour
The resolution is a fourth factor nobody had asked for. Each quark carries a new charge with three values — red, green, blue — and the baryon wave function is
Bettini p. 159. Four factors, and the Pauli principle constrains only their product.
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.
The colour wave function of every baryon. It is the same expression the book uses for the SU(3)_f singlet on p. 160 — the antisymmetric combination of three distinguishable things — which is why the algebra of the next two sections is already familiar by the time colour needs it.
Now run the Pauli requirement through Eq. (4.55). The colour factor is antisymmetric, and is symmetric because these are ground states. For the product of all four to be antisymmetric, what is left must be symmetric:
Bettini p. 159. The line that turns a rescue into a prediction.
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 — colour pays for itself
An extra quantum number invented to fix one embarrassment would be a bad trade. This one is not, and the reason is Eq. (4.56).
Colour was introduced to let uuu exist. But once the colour factor is
antisymmetric for every baryon, Eq. (4.56) applies to every baryon — and it is
a restriction. Of the four flavour multiplets in and the
three spin multiplets in , only some pairings survive it.
The quark model with no colour predicts all twelve pairings exist. Nature shows two families of ground-state baryons: a 3/2⁺ decuplet and a 1/2⁺ octet. The survivors of Eq. (4.56) are exactly those two. A fix that only fixed the thing it was invented for would explain nothing; this one explains the shape of the entire baryon spectrum, and it does so by forbidding things.
Three spins: 2 ⊗ 2 ⊗ 2
Building up to that, in the book’s order. First the spins, which are SU(2) doublets and therefore the case you can do by hand.
Bettini p. 159. Two spin-1/2 make a singlet and a triplet; adding a third splits the triplet again.
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.
Notice what “mixed” means, because it is the crux and it has no two-body analogue. The M,A doublet is antisymmetric under the exchange of quarks 1 and 2 — the pair whose spins were coupled first — and under the other two exchanges it is neither symmetric nor antisymmetric. It is not an eigenstate of them at all. Same for M,S with a plus sign.
Tab ① below is Table 4.2, rebuilt. Every ±1 badge is obtained by applying the transposition to the state you are looking at and comparing amplitude by amplitude — nothing is a lookup.
exchange symmetry of three quarks · every label below is computed, not looked up
| J | sym. | Jz = +3/2 | Jz = +1/2 | Jz = −1/2 | Jz = −3/2 |
|---|---|---|---|---|---|
| 3/2 | S | ||||
| 1/2 | M,A | — | — | ||
| 1/2 | M,S | — | — |
Jz = +1/2. The badges come from swapping the two quarks in every basis ket and comparing amplitude by amplitude. Notice that M,A and M,S are only ±1 under P₁₂ — they are not eigenstates of the other two exchanges at all, which is exactly what "mixed" means and why neither can carry a baryon by itself.
Erratum — the M,S row of Table 4.2
The printed table has the M,S states as
in the and columns. Neither is a eigenstate: the first term of each has the third arrow pointing the wrong way, so the cell mixes into and vice versa. Flipping that one arrow repairs the entry (up to an overall sign, which is free). The entry needs one more fix — a relative sign — before it is orthogonal to the state directly above it, which any state must be.
The widget above shows the repaired row. The snippet below diagnoses both faults rather than asserting them, using the only two properties that define these states: be a eigenstate, and be orthogonal to the state in the same column.
Diagnose and repair Table 4.2
import numpy as np
from itertools import product
basis = ["".join(b) for b in product("ud", repeat=3)] # u = spin up, d = spin down
idx = {b: i for i, b in enumerate(basis)}
def ket(*terms):
v = np.zeros(8)
for c, s in terms: v[idx[s]] += c
return v
r6, r23, r3 = 1/np.sqrt(6), np.sqrt(2/3), 1/np.sqrt(3)
Jz = np.diag([sum(0.5 if q == "u" else -0.5 for q in b) for b in basis])
S12 = ket((r3,"uud"), (r3,"udu"), (r3,"duu")) # the J = 3/2, Jz = +1/2 state
Sm12 = ket((r3,"ddu"), (r3,"dud"), (r3,"udd"))
def report(name, v, partner):
jz = v @ Jz @ v
spread = np.sqrt(max(v @ Jz @ Jz @ v - jz**2, 0))
ov = v @ partner
print(f" {name:26s} <Jz> = {jz:+.4f} spread = {spread:.4f} "
f"<3/2|psi> = {0.0 if abs(ov) < 1e-12 else ov:+.4f}")
print("Table 4.2, the M,S row -- AS PRINTED:")
report("Jz = +1/2 as printed", ket((r6,"udd"), (r6,"dud"), (-r23,"uud")), S12)
report("Jz = -1/2 as printed", ket((-r6,"udu"), (-r6,"duu"), (-r23,"ddu")), Sm12)
print(" neither is a Jz eigenstate: a spread of 0.47 means the cell mixes Jz = +1/2 and -1/2\n")
print("third-quark arrow corrected:")
report("Jz = +1/2 arrow fixed", ket((r6,"udu"), (r6,"duu"), (-r23,"uud")), S12)
report("Jz = -1/2 arrow fixed", ket((-r6,"udd"), (-r6,"dud"), (-r23,"ddu")), Sm12)
print(" the +1/2 cell is now right; the -1/2 cell is still not orthogonal to J = 3/2\n")
print("the repaired row:")
report("Jz = +1/2 repaired", ket((r23,"uud"), (-r6,"udu"), (-r6,"duu")), S12)
report("Jz = -1/2 repaired", ket((r6,"udd"), (r6,"dud"), (-r23,"ddu")), Sm12)Table 4.2, the M,S row -- AS PRINTED: Jz = +1/2 as printed <Jz> = +0.1667 spread = 0.4714 <3/2|psi> = -0.4714 Jz = -1/2 as printed <Jz> = -0.1667 spread = 0.4714 <3/2|psi> = -0.4714 neither is a Jz eigenstate: a spread of 0.47 means the cell mixes Jz = +1/2 and -1/2 third-quark arrow corrected: Jz = +1/2 arrow fixed <Jz> = +0.5000 spread = 0.0000 <3/2|psi> = +0.0000 Jz = -1/2 arrow fixed <Jz> = -0.5000 spread = 0.0000 <3/2|psi> = -0.9428 the +1/2 cell is now right; the -1/2 cell is still not orthogonal to J = 3/2 the repaired row: Jz = +1/2 repaired <Jz> = +0.5000 spread = 0.0000 <3/2|psi> = +0.0000 Jz = -1/2 repaired <Jz> = -0.5000 spread = 0.0000 <3/2|psi> = +0.0000
Three flavours: 3 ⊗ 3 ⊗ 3
The same exercise with three-valued objects instead of two-valued ones. Bettini declines to derive it — “we do not assume the reader to have enough knowledge of the group” — and gives the answer:
Bettini p. 160. Twenty-seven states, four blocks — and every block can be counted by hand.
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.
Tab ② does the counting live. Pick a quark content and the widget reports how many orderings it spans and how that slice splits, then totals the ten contents to 10 + 8 + 8 + 1 = 27.
exchange symmetry of three quarks · every label below is computed, not looked up
uud has 3 distinct orderings — duu, udu, uud — so it spans a 3-dimensional slice of the 27. That slice splits into 1 S, 1 M,S, 1 M,A, 0 A.
None. Antisymmetrising uud gives exactly zero — two of the quarks are identical, so the two terms that should cancel do cancel.
Summed over all ten contents: 10 S + 8 M,S + 8 M,A + 1 A = 27 = 3³. That is Eq. (4.59), 10S ⊕ 8M,S ⊕ 8M,A ⊕ 1A, obtained by counting rather than by group theory.
Count 3 x 3 x 3 with a projector, one content at a time
from itertools import permutations, product
# the six permutations of three slots, with their signatures
SIX = [((0,1,2),1), ((1,0,2),-1), ((2,1,0),-1), ((0,2,1),-1), ((1,2,0),1), ((2,0,1),1)]
tot = dict(S=0, MS=0, MA=0, A=0)
print("content orderings S M,S M,A A")
for content in sorted({"".join(sorted(c)) for c in product("uds", repeat=3)}):
orders = sorted(set("".join(p) for p in permutations(content)))
# dim of a subspace = trace of its projector = (1/6) sum over permutations
nS = sum(1 for p, _ in SIX for k in orders if "".join(k[i] for i in p) == k) // 6
nA = sum(s for p, s in SIX for k in orders if "".join(k[i] for i in p) == k) // 6
nM = len(orders) - nS - nA
tot["S"] += nS; tot["A"] += nA; tot["MS"] += nM // 2; tot["MA"] += nM // 2
print(f" {content} {len(orders):2d} {nS:2d} {nM//2:2d} {nM//2:2d} {nA:2d}")
print(f" TOTAL 27 {tot['S']:2d} {tot['MS']:2d} {tot['MA']:2d} {tot['A']:2d}"
f" -> 3x3x3 = 10 + 8 + 8 + 1")content orderings S M,S M,A A ddd 1 1 0 0 0 dds 3 1 1 1 0 ddu 3 1 1 1 0 dss 3 1 1 1 0 dsu 6 1 2 2 1 duu 3 1 1 1 0 sss 1 1 0 0 0 ssu 3 1 1 1 0 suu 3 1 1 1 0 uuu 1 1 0 0 0 TOTAL 27 10 8 8 1 -> 3x3x3 = 10 + 8 + 8 + 1
💡 What this really says — the four particles made of uds
The row that answers the question this page opened with is dsu. Six orderings,
and they split 1 + 2 + 2 + 1. Those six states are, in the ground-state spectrum:
- the S state → the Σ⁰(1385), in the decuplet;
- the two mixed states → the Σ⁰ and the Λ⁰, in the octet, distinguished by whether the ud pair is in isospin 1 or isospin 0;
- the A state → the SU(3) singlet, which shows up as the Λ(1405).
One bag of quarks, four particles, and what tells them apart is nothing but
exchange symmetry. Every other content is simpler because it has fewer orderings
to divide up — and uuu, with exactly one ordering, has no freedom at all. That
is precisely why it is the one that broke the Pauli principle.
The Pauli filter
Now put the two decompositions together under Eq. (4.56). There are four flavour blocks and three spin blocks: twelve conceivable pairings, of which the quark model without colour predicts that all twelve exist as baryons. Tab ③ multiplies each pair out and tests the product.
exchange symmetry of three quarks · every label below is computed, not looked up
ψcolour is antisymmetric and ψspace is symmetric (S wave), so Eq. (4.56) demands ψspinψSU3 be symmetric. Each cell below multiplies a real flavour state by a real spin state and tests the product.
| SU(3)f ↓ spin → | 4 (S) | 2 (M,S) | 2 (M,A) |
|---|---|---|---|
| 10 (S) | symmetric ✓ | mixed ✗ | mixed ✗ |
| 8 (M,S) | mixed ✗ | mixed ✗ | mixed ✗ |
| 8 (M,A) | mixed ✗ | mixed ✗ | mixed ✗ |
| 1 (A) | antisym ✗ | mixed ✗ | mixed ✗ |
The cell the grid cannot show: the two mixed products added together
(8M,S·2M,S + 8M,A·2M,A)/√2
Each product on its own is symmetric under P₁₂ and nothing else — that is why both mixed rows above read "mixed ✗". Their sum is symmetric under all three. So the Pauli principle admits exactly two baryon families: the spin-3/2 decuplet (10S, 4S) and one spin-1/2 octet built from the mixed pair — which is precisely what nature shows, and it is fewer multiplets than the quark model alone predicts. Note the whole 1A row is dark: there is no Λ₁.
Two survive, and they are the two nature shows:
| pairing | states | what it is | |
|---|---|---|---|
| 3/2⁺ | 10 | the decuplet of Fig. 4.23 — Δ, Σ, Ξ, Ω⁻ | |
| 1/2⁺ | 8 | the octet of Fig. 4.22 — N, Σ, Λ, Ξ | |
| everything with 1_A | — | 0 | |
| the other seven pairings | — | 0 |
The quark model <em>alone</em> would predict all four flavour multiplets of Eq. (4.59) as ground-state baryons. The observed restriction to an octet and a decuplet is not classification — it is dynamics, and it is the first fingerprint of QCD in this book.
Verify the pairings — including the one that only works as a sum
import numpy as np
from itertools import product
sb = ["".join(b) for b in product("ud", repeat=3)] # spin: u = up, d = down
fb = ["".join(b) for b in product("uds", repeat=3)] # flavour
si = {b: i for i, b in enumerate(sb)}
fi = {b: i for i, b in enumerate(fb)}
r2, r3, r6, r23 = 1/np.sqrt(2), 1/np.sqrt(3), 1/np.sqrt(6), np.sqrt(2/3)
def vec(index, *terms):
v = np.zeros(len(index))
for c, s in terms: v[index[s]] += c
return v
def swap(p, keys, index):
M = np.zeros((len(keys), len(keys)))
for k in keys: M[index["".join(k[i] for i in p)], index[k]] = 1.0
return M
PERM = [("12",(1,0,2)), ("13",(2,1,0)), ("23",(0,2,1))]
PROD = {n: np.kron(swap(p, fb, fi), swap(p, sb, si)) for n, p in PERM}
def tag(v):
out = []
for n, _ in PERM:
w = PROD[n] @ v
out.append("+1" if np.allclose(w, v) else "-1" if np.allclose(w, -v) else "??")
return " ".join(f"P{n}={t}" for (n, _), t in zip(PERM, out)), all(t == "+1" for t in out)
# flavour states written with u <-> spin up, d <-> spin down, so the pairing is term by term
f10 = vec(fi, (r3,"uud"), (r3,"udu"), (r3,"duu"))
f8S = vec(fi, (r23,"uud"), (-r6,"udu"), (-r6,"duu"))
f8A = vec(fi, (r2,"udu"), (-r2,"duu"))
s4 = vec(si, (r3,"uud"), (r3,"udu"), (r3,"duu"))
s2S = vec(si, (r23,"uud"), (-r6,"udu"), (-r6,"duu"))
s2A = vec(si, (r2,"udu"), (-r2,"duu"))
for name, v in [("Delta+ (10_S x 4_S)", np.kron(f10, s4)),
("8_MS x 2_MS alone", np.kron(f8S, s2S)),
("8_MA x 2_MA alone", np.kron(f8A, s2A))]:
t, ok = tag(v)
print(f" {name:22s} {t} {'TOTALLY SYMMETRIC' if ok else 'mixed -- forbidden alone'}")
proton = (np.kron(f8S, s2S) + np.kron(f8A, s2A)) / np.sqrt(2)
t, ok = tag(proton)
print(f" {'proton (the SUM)':22s} {t} {'TOTALLY SYMMETRIC' if ok else 'mixed'}"
f" |psi| = {np.linalg.norm(proton):.4f}")
print(" the sum of two forbidden things is allowed: that is the whole content of Eq. (4.56)")Delta+ (10_S x 4_S) P12=+1 P13=+1 P23=+1 TOTALLY SYMMETRIC 8_MS x 2_MS alone P12=+1 P13=?? P23=?? mixed -- forbidden alone 8_MA x 2_MA alone P12=+1 P13=?? P23=?? mixed -- forbidden alone proton (the SUM) P12=+1 P13=+1 P23=+1 TOTALLY SYMMETRIC |psi| = 1.0000 the sum of two forbidden things is allowed: that is the whole content of Eq. (4.56)
⚙️ Engineer’s bridge — the proton does not factorise
The line to take away from that snippet is the third one. 8_MS × 2_MS fails,
8_MA × 2_MA fails, and their sum passes.
This is entanglement, in the plain linear-algebra sense you already use: a vector in a tensor-product space that cannot be written as for any and . The proton has no separate “flavour state” and “spin state”. Ask which flavour combination it is in and there is no answer that does not also mention the spins. The correlation between the two factors is the particle.
That is also why the counting works out to one octet rather than two. There are two mixed flavour octets and two mixed spin doublets, so a naive count expects four families; the symmetry requirement ties them together in one specific correlated combination, and four collapses to one.
Where it breaks: the factorisation fails here for a specific reason — the totally antisymmetric colour wavefunction correlates the spin and flavour parts — and it is worth not over-learning that. Factorisation is an excellent approximation in plenty of neighbouring problems: the colour wavefunction itself factorises out of every hadron mass calculation in ch06, and heavy-quark systems are treated non-relativistically precisely because the pieces do separate. The lesson is that a symmetry requirement can couple factors you expected to be independent, not that structure never factorises.
The two ground-state families
Fig. 4.22 — the baryon octet, J^P = 1/2⁺ · JP = 1/2⁺
Click any member for its quantum numbers — and for the Gell-Mann–Nishijima check Iz = Q − Y/2.
Two states sit at the centre, as they did in the meson nonet — but this time the reason is exchange symmetry rather than singlet–octet mixing, and both belong to the same octet. Press the antiparticle button: unlike a meson nonet, the baryon octet does NOT map onto itself. Antibaryons form a separate, disjoint octet, because B = +1 cannot become B = −1 by relabelling.
Fig. 4.23 — the baryon decuplet, J^P = 3/2⁺ · JP = 3/2⁺
Click any member for its quantum numbers — and for the Gell-Mann–Nishijima check Iz = Q − Y/2.
A triangle, not a hexagon: the decuplet is the completely symmetric block, so the number of states falls by one at each rung and the pattern closes at a single point. That point is the Ω⁻ — and the geometry of the diagram is what told Gell-Mann it had to be there. Masses are the book's Fig. 4.23 values.
📏 Masses on these two figures are the book’s rounded values
Fig. 4.23 writes the Δ as 1236 MeV and the Ξ* as 1530 MeV; current PDG values are 1232 and 1533. The differences are within the widths of these states — the Δ is 117 MeV wide, so quoting its mass to the MeV is meaningless — and the site keeps the book’s numbers so that the arithmetic on this page matches the arithmetic on the page you are reading beside it. The 1/2⁺ octet masses, by contrast, are metastable particles with widths below an eV, and there the book’s figures agree with the PDG to the MeV.
A closing observation: where the mass actually is
Bettini ends the section by looking back at the quark masses of Table 4.1, and the point he makes is easy to skim past and hard to overstate.
Add up the quarks and compare with the hadron
mq = dict(u=2.16, d=4.67, s=93.4) # Table 4.1, MeV
for had, content, m in [("proton", "uud", 938.3),
("Lambda", "uds", 1115.7),
("Omega-", "sss", 1672.5)]:
q = sum(mq[c] for c in content)
print(f" {had:7s} {content} quarks = {q:6.2f} MeV hadron = {m:7.1f} MeV"
f" quarks are {100*q/m:5.2f} %")
print(" the other 83-99 % is energy of the colour field (Chapter 6)")proton uud quarks = 8.99 MeV hadron = 938.3 MeV quarks are 0.96 % Lambda uds quarks = 100.23 MeV hadron = 1115.7 MeV quarks are 8.98 % Omega- sss quarks = 280.20 MeV hadron = 1672.5 MeV quarks are 16.75 % the other 83-99 % is energy of the colour field (Chapter 6)
💡 What this really says — a composite that weighs 100× its parts
For every other composite system you have met, binding makes the whole lighter than its parts. A deuteron is 2.2 MeV lighter than a free proton plus a free neutron; a hydrogen atom is 13.6 eV lighter than a free proton plus a free electron. The mass defect is small and negative.
A proton is 100 times heavier than its three quarks. Its mass is not the mass of its constituents at all — it is the energy stored in the colour field that confines them, and the quark masses are a rounding error on top.
Two things follow immediately, and both are the point of the section:
- Isospin is a good symmetry even though the d quark weighs twice the u, because a factor of two on 2 MeV is invisible next to 938. The colour force does not look at flavour, so u and d are interchangeable to the accuracy that their mass difference is negligible — about 1 %, which is exactly the accuracy isospin has.
- SU(3)_f is a worse symmetry for exactly the same reason read the other way: 93 MeV is not negligible next to a hadron mass, so multiplets spread over hundreds of MeV. That is why the decuplet has 145 MeV rungs at all.
So both symmetries are accidental: they hold because certain masses happen to be small compared with the confinement scale, not because anything in the theory requires them. Which does not make them less useful — the Ω⁻ was found with them — but it does mean they are not fundamental, and Chapter 6 will not derive them.
🔑 If you remember only three things
-
A confirmation and a crisis arrived in the same instant. The tenth member completed the pattern and simultaneously demanded three identical quarks share one state.
-
The fix was cheap to state and expensive to believe. One new three-valued label, and nothing in physics had ever needed one before.
-
The catalogue of ground-state baryons is an output, not an input. Spin, flavour and colour are combined and a single exclusion requirement selects whatever survives.
Where this goes next
- §4.9–4.11 repeat this chapter’s whole arc for the three heavy quarks: a prediction, a discovery, a spectrum.
- Chapter 6 is colour taken seriously — where the antisymmetric factor of Eq. (4.55) comes from, why only colourless combinations exist, and where a proton keeps the other 99 % of its mass.
- §6.4 turns the number of colours into a measurement: the R ratio in e⁺e⁻ annihilation counts them, and gets three.
- The exchange-symmetry algebra of this page is reused whenever identical fermions appear together — the two neutrinos of double beta decay in Chapter 10 are the next place it decides what can happen.
✅ Check yourself — baryons, exchange symmetry and colour
0/5 answered · 0 correct
1.Which half of the Ω⁻ prediction was the risky one?
2.In the widget's tab ①, the M,A state shows P₁₂ = −1 but 'neither' for P₁₃ and P₂₃. What does that mean?
3.Why is colour more than an ad hoc rescue of the Pauli principle?
4.The verification snippet shows 8_MS × 2_MS failing, 8_MA × 2_MA failing, and their normalised sum passing. What does that tell you about the proton?
5.A proton weighs about 100 times the sum of its quark masses. Which conclusion of the section does that support?