The hook. In one of the evening digests, a line flashed by that an engineer accustomed to thinking about things as "assembly parts" can't pass: "The proton spin crisis has lasted 38 years, quarks explain only ~30% of the spin, the rest is gluons and orbital motion." At first glance — so what, physics, some quarks, who needs it. But if you dig a little deeper, it turns out that behind this line stands one of the longest-running mysteries of fundamental physics, an unsolved problem officially recognized as such, a story in which are intertwined the quark model of the '60s, the London muon beam of 1987, a Japanese underground detector with 50 kilotons of ultrapure water, and the electron-ion collider under construction now in Brookhaven for $1.7–2.8 billion, where physicists plan to close this question definitively half a century later. The proton — a particle every schoolchild knows and which makes up 99% of visible matter in the Universe — turned out to be misunderstood in the most fundamental sense: we still don't know where its spin of 1/2 comes from.
In any physics textbook, the proton is described extremely simply: three valence quarks (two u, one d), held together by gluons. Each quark has spin 1/2, the proton has 1/2. Arithmetic: take two quarks with spin "up" and one with spin "down" — you get 1/2 + 1/2 − 1/2 = 1/2. Elementary, Watson. This exact picture — the naive quark model — has been mainstream since 1964, when Gell-Mann and Zweig independently proposed the quark structure of hadrons. All textbooks are tailored to it, all estimates of magnetic moments, all mass calculations. And it burst in 1987.
In November 1987, the European Muon Collaboration (EMC) at CERN conducted a series of measurements on a polarized target: a beam of polarized muons with 200 GeV energy scattered off polarized protons. From the angular distribution of scattering products, one can extract the spin structure function g₁(x) — how much the quarks "inside" the proton are polarized along its spin. If the naive model is correct, then the integral over all Bjorken-x gives close to unity. EMC in 1988 published a result in Physics Letters B: ΔΣ = 0.12 ± 0.09 ± 0.14. That is, quarks carry from 4% to 24% of the proton spin, with a central estimate of about 12%, and this number is statistically consistent with zero.
The world shuddered. This literally meant that 80–96% of the proton spin has nothing to do with quark spins. The effect was dubbed the "proton spin crisis" — spin crisis, and later, when it became clear that resolution was dragging on, the "proton spin puzzle". The name over 38 years has become official, entered the lists of the most important unsolved problems in physics, and around it has grown an entire subdiscipline — spin physics, with its own journals, conferences and theoretical schools.
Modern measurements, including polarized data with more accurate coverage of small x and perturbative QCD corrections, give ΔΣ ≈ 0.30 — that is, quarks carry about a third of the spin, and the errors in this number no longer cover unity. This is less than 12% in 1987, but still significantly less than one. The crisis has ended — the puzzle remains.
According to the current state (Keh-Fei Liu, AAPPS Bulletin, 2022, and review arXiv:2607.20761 from July 22, 2026), the proton spin decomposes into four contributions:
1/2 = 1/2 · ΔΣ + ΔG + Lq + Lg
where:
This last line is the most surprising. Gluons, which in the textbook picture "just glue quarks together," can actually carry most of the proton spin — and not through their intrinsic spin, but through orbital motion. It's as if the retaining cable turned out to be the main load-bearing structure.
The EMC experiment was not the first to measure the spin structure of the nucleon. Before it were SLAC E80/E130 (1970s–80s) and EMC-first results 1985, which gave ΔΣ around 0.20–0.30 — that is, already hinting at a problem, but with large errors. However, the 1987 session specifically with a high-energy muon beam (200 GeV instead of the previous 30–50) and a polarized ammonia target (protons in NH₃, polarized in a strong magnetic field at ultra-low temperature) gave statistics that didn't allow looking away. The number ΔΣ = 0.12 ± 0.09 ± 0.14 — with a confidence interval extending almost to zero from above, irrefutably said: the quark picture is insufficient.
Immediately after the 1988 publication in Physics Letters B began a "spin golden age" — an avalanche of theoretical work trying to explain where the missing spin went. Main candidates: gluon contribution ΔG (by analogy with deep inelastic scattering on gluons), orbital angular momentum L (relativistic corrections in the quark wave function), and sea quarks (quark-antiquark pairs constantly being born and annihilating in the proton). In 1989, EMC published refined data in Nuclear Physics B, confirming the result, and in the 1990s CERN, SLAC, DESY and Brookhaven launched a whole series of polarized experiments to resolve the puzzle.
In the 1990s, several competing explanations emerged:
Hypothesis 1: gluon polarization. Gluon spin equals 1, the proton has three gluons, and they can be polarized along the proton spin, giving non-zero ΔG. This can be checked through polarized proton-proton collisions at RHIC, where gluon-gluon interactions dominate in pion and jet production. The PHENIX and STAR experiments at Brookhaven from the 2000s measured double spin asymmetries and concluded: ΔG is positive in the region x > 0.05, and its magnitude is possibly comparable to the quark contribution in this region. However, the full integral over all x remains uncertain, because the small-x region (x < 0.05) is experimentally inaccessible, and the small-x contribution could be significant.
Hypothesis 2: orbital momentum. In the naive non-relativistic quark model, quarks are in an s-wave (spherically symmetric ground state), they have no orbital momentum. But in the relativistic case (bag model, quark-diquark model, lattice calculations), quarks move at near-light speeds, and there arises a lower component of the Dirac wave function, which is interpreted as orbital angular momentum Lq. Lattice calculations in 2018 (Lattice QCD) showed that Lq can be the dominant contribution to proton spin — that is, relativistic corrections themselves explain half of what the quark spins lack.
Hypothesis 3: axial anomaly. In 1983, Altarelli et al. showed that in quantum chromodynamics there is an axial anomaly — a quantum effect that "eats" the spin contribution of quarks in favor of gluons. This anomaly depends on the renormalization scale Q², and at large Q² (deep inelastic scattering) ΔΣ is measured with the anomaly taken into account, and at small Q² — without. This introduced methodological confusion in interpretation, but ultimately resolved: depending on the renormalization scheme, ΔΣ varies from ~0.20 to ~0.60, and "the spin crisis" was largely an artifact of scheme choice. Now physicists use Jaffe-Manohar and Ji decompositions (two different Hamiltonian decompositions of spin), and both give consistent answers.
Modern consensus (2024–2026): proton spin is 30% quark spin + 15–20% gluon spin + 15% quark orbital momentum + 40% gluon orbital momentum. This picture is confirmed by global QCD fits (JAM Collaboration, Jefferson Lab), lattice calculations, and RHIC experiments. But the exact breakdown still has large errors, especially in small-x regions.
Measuring spin is a direct demonstration that physicists can extract incredibly subtle signals from noise. Proton spin is one quantum unit 1/2ℏ, and you need to know how this unit is divided between 3 valence quarks, an infinite sea of gluons, and a pair of sea quarks. There is no direct way to "look inside the proton" — we don't have a microscope with 10⁻¹⁵ m resolution. Instead, three indirect methods are used, each with its own strengths and weaknesses:
Method 1: deep inelastic scattering (DIS). A polarized lepton (muon or electron) scatters off a polarized proton target. From the angle and energy of the scattered lepton, Bjorken-x is reconstructed — the fraction of proton momentum carried by the parton on which the interaction occurred. From the difference in cross-sections for parallel and antiparallel spins, g₁(x) is extracted — the polarized structure function, from which ΔΣ is obtained by integration. This is the method of EMC, SLAC, HERMES (DESY), COMPASS (CERN), JLab. Modern measurements extended the range to Q² ≈ 10.7 GeV² and x down to ~0.001, which improved the ΔΣ estimate, but small x is still poorly covered.
Method 2: polarized pp-collisions at RHIC. RHIC is the only polarized proton collider in the world, operated from 2000 to 2025 (25-year program completed in spring 2025). The PHENIX and STAR detectors measured double spin asymmetries in pion, jet, and direct photon production. These processes are sensitive to Δg(x) — the polarized gluon distribution function. By 2024, RHIC data combined with DIS data established that ΔG in the region x > 0.05 is positive and significant. Global fits give ΔG(x=0.05–0.2) ≈ 0.15–0.20, but the full integral over all x still has a factor 2 uncertainty.
Method 3: lattice calculations (Lattice QCD). Quantum chromodynamics on a discrete spacetime lattice. Supercomputers "compute" vacuum expectation values of spin and orbital operators, directly within the Standard Model, without approximations. By 2018, several groups (LHPC, ETMC, χQCD) independently confirmed that quark orbital angular momentum Lq dominates in the full picture. This is a strong theoretical argument, but it depends on lattice size, pion mass (usually taken heavier than physical) and extrapolation to physical mass. So lattice results have their methodological caveats.
Latest news (2024–2026): in October 2024, Physical Review Letters published work by C. Cocuzza et al. "New Data-Driven Constraints on the Sign of Gluon Polarization in the Proton" (PRL 133, 161901), which, using new RHIC data and lattice QCD, established that negative gluon polarization (ΔG < 0) is incompatible with observations — that is, gluons in the proton "spin" in the same direction as the proton itself, not against. In March 2024, Hunt-Smith et al. in arXiv:2403.08117 confirmed this through independent analysis. This is an important step, because 30 years ago it was theoretically allowed that ΔG could be negative, and in that case gluon spin could "compensate" for the lack of quark spin. Now this hypothesis is closed.
In the Standard Model, the proton is strictly stable — this follows from baryon number conservation. But in any grand unified theory (GUT), which unifies strong, weak and electromagnetic interactions into one symmetry group (SU(5), SO(10), E₆), baryon number is no longer conserved: heavy gauge bosons (X-bosons in SU(5), mass ~10¹⁵–10¹⁶ GeV) can turn a quark into a lepton, and thus the proton can decay with a lifetime of 10³⁰–10³⁶ years depending on the model.
Predicted decay modes are diverse: p → e⁺ + π⁰ (cleanest signal, B-L charge violation), p → e⁺ + η (with η-meson production), p → μ⁺ + K⁰, p → ν̄ + K⁺ (strangeness violation). If we could detect even one such decay, it would be the first direct observation of physics beyond the Standard Model — without any colliders, without accelerators, only a giant tank of water and patience.
Super-Kamiokande in Japan is 50,000 tons of ultrapure water in an underground mine at 1000 m depth (to cut out cosmic rays), surrounded by 13,000 photomultipliers. Each proton in water — that's about 10³⁴ protons in 50 kilotons. If the proton lives ~10³⁴ years, then about one proton should decay per year in the detector. Super-Kamiokande has been operating since 1996 and searches for characteristic Cherenkov light from decay products. And finds nothing.
The latest result — the Super-Kamiokande collaboration in Physical Review D 110, 112011 (December 2024) (arXiv:2409.19633) published analysis of 0.373 Mton·year exposure (6050 live days!) for channels p → e⁺ + η and p → μ⁺ + η. Result: not a single candidate event. Lower limits on lifetime:
This is 1.5 times better than previous limits. The proton lives more than 10³⁴ years — for comparison, the age of the Universe ~1.4 × 10¹⁰ years, that is, the proton lives 10²⁴ times longer than our Universe has existed.
This places tight constraints on GUT models. Minimal SU(5) of Gell-Mann predicted τ(p → e⁺π⁰) ~ 10³⁰–10³¹ years — and it has long been ruled out by this data. Supersymmetric SU(5) predicts ~10³⁴–10³⁵ years and is still in the zone, but in the recent work of Evans & Shigekami (2025, arXiv:2409.06239) it's shown that for compatibility with experiment, specially tuned messenger-scale parameters are needed. Minimal GUT models are either already dead or require fine-tuning.
In coming years, Hyper-Kamiokande (Japan, launch 2027) will increase detector mass by 8 times and statistics by another order of magnitude, and DUNE (USA, Sanford Underground Research Facility) and JUNO (China) will add sensitivity to other channels. If the proton does decay and they catch it — this will be a scientific revolution on the level of the discovery of neutrino oscillations. If not — this will close another class of theories and push toward ideas that the baryon asymmetry of the Universe is provided not by GUT mechanisms, but by baryogenesis through leptogenesis (through massive neutrinos).
Electron-Ion Collider (EIC) is a collider now under construction at Brookhaven, in which polarized electrons will collide with polarized ions (from protons to uranium). Project budget is estimated at $1.7–2.8 billion depending on source; the first batch of equipment (electrical substations) was delivered in April 2026, main construction begins summer 2026, launch — early 2030s.
EIC will be the only polarized electron-ion collider in the world and the only machine capable of directly measuring g₁(x) in the region x ~ 10⁻⁴ — that is, in that small-x region where today most of the ΔG uncertainty hides. In addition, EIC will allow measuring deeply virtual Compton scattering (DVCS) and exclusive processes, through which generalized parton distributions (GPD) are reconstructed — and from GPD by Ji decomposition the orbital angular momentum of quarks and gluons is directly calculated. This is what in 2020 in the Nature Reviews Physics review (What we know and what we don't know about the proton spin after 30 years) Xiangdong Ji called the main unsolved problem.
Essentially, the 1987 spin crisis created an entire experimental infrastructure costing billions of dollars: RHIC (1985–2025, 25 years of operation), EIC (under construction), Hyper-Kamiokande (launch 2027), DUNE (2030s), JLab 12 GeV upgrade (2017), COMPASS/CERN, and all this for the sake of answering a question that a schoolchild can ask in 5 seconds: where does the proton get its spin of 1/2?
There exists a second parallel puzzle of the proton, no less famous in particle physics: the "proton radius puzzle" — discrepancy in measurements of the proton charge radius between ordinary hydrogen atom and muonic hydrogen (where the electron is replaced by a muon). The muon is 207 times heavier than the electron, so its orbit is 207 times closer to the proton, and it is much more sensitive to charge distribution inside the proton. Measurements of muonic hydrogen (CREMA experiments at PSI, Switzerland) in 2010 gave 0.84184(67) fm, which is 4% (5 standard deviations!) less than averaged measurements of ordinary hydrogen (~0.877 fm).
This discrepancy excited physics for 14 years. In 2019, a new series of ordinary hydrogen measurements (A. Beyer et al., Science) confirmed the old value, and in 2023 the FOOP team (PSI) refined the muonic one — and the discrepancy seems to have decreased, but didn't disappear completely. In the work arXiv:2406.18738 from September 2024, it's shown that proton spin polarizability (that same Δpol, related to spin structure g₁) introduces systematic uncertainty in the calculation of hyperfine splitting in muonic hydrogen, and new spin data reduce this uncertainty by half.
That is, two proton puzzles — spin and size — are connected. Spin structure g₁(x) determines polarizability, polarizability affects hyperfine splitting, hyperfine splitting is used to measure radius. Everything in one proton.
In 2022, NICER (Neutron star Interior Composition Explorer) results were published on measuring radii and masses of neutron stars. A neutron star is essentially a giant atomic nucleus, and the equation of state of its interior depends on how quarks and gluons interact in extremely dense matter. One of the key parameters is spin-orbit interaction in nuclear forces, which itself is a consequence of nucleon spin structure. That is, understanding proton spin helps understand what happens inside stellar-mass black holes — particle physics meets astrophysics.
The spin crisis was one of the first serious indications that the naive quark model is incomplete — long before this became obvious for other Standard Model problems. In a sense, 1987 is the moment when high-energy physics began to understand that QCD effects are deeper than seemed, and that the proton is not just "three quarks in a bag," but a complex relativistic bound state in which the gluon field plays an independent role. This is a path that in the 2020s leads to understanding gluon distributions in extreme conditions (LHC, RHIC, future EIC), to AdS/CFT-like dualities (Maldacena et al.), and ultimately to string theory and quantum gravity — that is, the 1987 spin crisis turned out to be one of the early pointers to that same "theory of everything" that physicists have been searching for 50 years.
2026: review arXiv:2607.20761 comes out (July 22, 2026) — modern summary of Jaffe-Manohar and Ji decompositions. RHIC completed operation in 2025, its 25-year program is finished. EIC began main construction at Brookhaven.
2027: launch of Hyper-Kamiokande in Japan — 8 times more mass than Super-K, which will give another order of magnitude better limits on proton lifetime.
2028–2030: commissioning of JLab 12 GeV at full power, new data on DVCS and g₁.
2030–2032: launch of EIC — first phase, physics begins. Capability to measure g₁ down to x ~ 10⁻⁴.
2030s: launch of DUNE (USA, Sanford) — sensitivity to modes p → K⁺ν̄, which are especially important for SUSY-GUT.
If the puzzle resolves by 2035 (which is realistic given nominal EIC and Hyper-K operation), this will be one of the longest scientific projects in physics history: 48 years from the first EMC surprise to the final answer.
Why this matters not only to physicists. The proton spin crisis is a rare case where one unsolved fundamental problem spawned an entire ecosystem of research: new colliders (RHIC, EIC), new detectors (Hyper-K, DUNE), new theoretical methods (lattice QCD, GPD, Ji decomposition), new applications (spin polarizability, proton radius puzzle, nuclear astrophysics). And all this started with the wrong answer to a simple question: "where does the proton get its spin?"
What I admire in this story. The textbook picture "three quarks with spins 1/2 give 1/2 of the proton" is not an outright error, but a simplification that works at 30%. The remaining 70% is gluon sea, orbital momenta, axial anomalies, relativistic corrections, and effects that depend on the energy scale at which you "look" at the proton. A simple building block of the Universe turned out to be a structure of unimaginable complexity — and it took us 38 years to even roughly understand how it's built.
What troubles me. EIC is a $2+ billion investment in a single question to which we already have partial answers. If EIC shows that ΔG at small x is huge, and gluon orbital momentum is indeed 40% — this will be a triumph. If EIC shows that we need to build EIC-2 with even greater luminosity to go further — this will be a serious blow to fundamental physics priorities. But either way, the spin crisis is a story in which science works as it should: poses a simple question, gets an uncomfortable answer, doesn't retreat, and over decades builds tools capable of nailing down that answer. In a world where scientific institutions increasingly demand "quick results" and "impact factors in the moment," the proton spin crisis is a reminder that real fundamental physics is measured in decades, not quarters.
Personal note. I'm an engineer, not a physicist. I like this story not because I understand all the QCD formalism (I don't understand it — 90% of the constituents I can't reproduce from memory), but because it shows the architectural honesty of physics as a discipline. The proton is an "assembly part" of everything we see in the Universe. And we have the courage to admit we don't understand how it's built, and are willing to spend 38 years and billions of dollars to find out. This is very human and very beautiful.