In 1974, the J/ψ particle was discovered almost simultaneously at Brookhaven National Laboratory in New York and the SPEAR collider at Stanford. This discovery, known as the "November Revolution," demonstrated the validity of the quark model to the world. But that same year, another prediction was quietly taking shape among theoretical physicists: that gluons—the particles responsible for binding quarks together—might attract one another and form particles entirely on their own.
The basis for this prediction lies in the mathematical structure of quantum chromodynamics (QCD). Photons, which mediate the electromagnetic force, carry no charge, so photons do not interact directly with one another. Gluons, however, which mediate the strong force, carry a charge of their own called "color charge." QCD is a theory based on the non-Abelian gauge group SU(3), and this non-commutativity necessarily gives rise to gluon self-interaction. The asymptotic freedom discovered in 1973 by David Gross, Frank Wilczek, and David Politzer revealed the high-energy behavior arising from this non-Abelian structure, but at low energies the opposite phenomenon occurs. The gluon coupling constant grows large, and gluons are predicted to bind strongly to one another, forming bound states known as glueballs.
All known particles in nature contain matter particles (quarks or leptons). Glueballs were predicted to be the only particle species composed solely of force-carrying particles. This is a direct consequence of QCD's non-Abelian structure, and its experimental discovery would serve as a touchstone for verifying the theory in the low-energy regime.
A Half-Century Chain of "Unconfirmed" Candidates
The search for glueballs began in the mid-1970s. Mass predictions using lattice QCD—a technique that discretizes spacetime into a lattice to numerically solve QCD—began in earnest in the 1980s. In an anisotropic lattice simulation published in 1999, Colin Morningstar and Mike Peardon predicted the mass of the lightest glueball to be 1730 MeV for the scalar ($0^{++}$), 2400 MeV for the tensor ($2^{++}$), and 2590 MeV for the pseudoscalar ($0^{-+}$).
Despite the clarity of the theoretical prediction, experimental confirmation proved extremely difficult, for two reasons.
First, in the real world where QCD includes quarks, a pure glueball mixes with quark-antiquark states (mesons). Lattice QCD predictions give masses in a "pure gauge theory"—an idealized world without quarks—but in reality this mixing is unavoidable. The observed particle becomes a superposition of glueball and quark components, blurring the criteria for declaring definitively "this is a glueball."
Second, every time a candidate appeared, an alternative interpretation was proposed. In the 1980s, ι(1440) (iota 1440), found in the radiative decay of J/ψ, drew attention as the first glueball candidate, but it was later found to be a superposition of two distinct pseudoscalar mesons, η(1405) and η(1475). f₀(1500) and f₀(1710) were proposed as scalar glueball candidates, but no consensus has been reached on which contains a larger glueball component. The tensor glueball candidate ξ(2230) was not even confirmed to exist in antiproton-proton annihilation experiments.
Over fifty years, multiple candidates emerged, and all of them remained without decisive evidence. This state of affairs continued until August 2026.
X(2370) Appearing in the Radiative Decay of J/ψ
The turning point came in 2011. The BESIII detector, operating at the Beijing Electron-Positron Collider II (BEPC II, a circular collider with a circumference of 240 m) at the Institute of High Energy Physics (IHEP) of the Chinese Academy of Sciences, discovered a new particle with a mass of about 2370 MeV in the radiative decay of the J/ψ particle (). This particle, named X(2370), had a statistical significance exceeding 6.4σ, well above the threshold recognized as a discovery.
There is a reason why the radiative decay of J/ψ is considered the optimal path for glueball searches. J/ψ is a bound state of a charm quark and an anti-charm quark, and when the quark pair annihilates by emitting a photon during radiative decay, the remaining energy creates a gluon-rich state. This gluon-rich environment favors glueball production.
After the discovery, the BESIII Collaboration spent 13 years accumulating data. In 2024, a partial wave analysis using 10 billion (precisely, about 10.087 billion) J/ψ events determined for the first time that the spin-parity quantum numbers of X(2370) are $0^{-+}$. This result was published as a peer-reviewed paper in Physical Review Letters in May 2024 (PRL 132, 181901).
The agreement in mass and quantum numbers is consistent with the properties of the lightest pseudoscalar glueball predicted by lattice QCD. However, this alone was not sufficient—ordinary mesons with $0^{-+}$ quantum numbers (such as excited states of the η meson) could also carry the same quantum numbers.
Flavor Singlet Nature: Showing "No Trace of Quarks"
In July 2026, the BESIII Collaboration submitted a comprehensive paper to arXiv (arXiv:2607.20366), and on August 5 formally announced the results in a plenary session at the International Conference on High Energy Physics (ICHEP 2026) held in Brazil. The chain of evidence presented in this paper consists of the following elements.
| Verification Item | Measured Value for X(2370) | Consistency with Glueball Prediction |
|---|---|---|
| Mass | 2376 MeV/ | Consistent with lattice QCD prediction (2.2–2.6 GeV/) |
| Spin-parity | $0^{-+}$ | Consistent with prediction for lightest pseudoscalar glueball |
| Production rate in J/ψ radiative decay | High () | Glueballs are preferentially produced in gluon-rich environments |
| Decay pattern | Similar to | Glueballs exhibit flavor-independent decay |
| Flavor singlet nature | decay suppressed (upper limit , 90% C.L.) | Direct evidence of lacking quark flavor structure |
| Radiative decay | Strongly suppressed | Glueballs show suppressed radiative decay to flavor-specific vector mesons |
Particularly decisive is the confirmation of flavor singlet nature. If X(2370) were an ordinary quark-antiquark meson (an excited state of η or η'), theory predicts its decay to should appear with a partial width of 15–200 MeV. BESIII's measurement found the statistical significance of this decay mode to be only 0.1σ, with an upper limit on the partial width below 2 MeV—more than an order of magnitude smaller than even the lower bound of the prediction (15 MeV).
This suppression constitutes direct evidence that X(2370) lacks a specific quark flavor structure—that is, that it is a flavor singlet. This marks the first observation of a flavor-singlet light hadron with a mass above 1 GeV/.
What "Glueball as the Dominant Component" Means
It is important to accurately understand the claim made by the BESIII Collaboration. IHEP's official announcement states that "the dominant constituent of the X(2370) is a pseudoscalar glueball." It does not say "X(2370) is a pure glueball." The expression "the dominant constituent is a glueball" indicates the fact that X(2370) is a mixed state of glueball and quark-onium components, in which the glueball component dominates.
This caution is physically justified. Since quarks exist in real QCD, a strictly pure glueball state does not exist. Lattice QCD predictions are calculations in a pure gauge theory (without quarks), and mixing is unavoidable in the real world. BESIII's claim is that a dominant glueball component is the most natural interpretation capable of simultaneously explaining all the observed properties.
Indeed, the paper states that "other interpretations at present are disfavored," showing that alternative interpretations such as η-η' excited states cannot simultaneously explain the flavor singlet nature, the narrow partial width, and the suppression of radiative decay.
Significance as a Low-Energy Test of QCD
The significance of this discovery extends beyond the mere fact that a new particle has been found. The non-Abelian gauge structure of QCD—the property that gluons themselves carry color charge and interact with one another—has been verified at high energies through asymptotic freedom. Behind the 2004 Nobel Prize in Physics awarded to Gross, Wilczek, and Politzer for this discovery lay precise verification through high-energy experiments.
However, in the low-energy regime, the large coupling constant renders perturbation theory inapplicable, and verification of QCD has relied on numerical lattice QCD calculations. The existence of glueballs was one of the most direct predictions of what concrete physical phenomena this non-Abelian structure produces at low energies. The confirmation of X(2370) means that this prediction has now been experimentally realized.
| Verification Regime | Verification Method | Time of Establishment |
|---|---|---|
| High energy (short distance) | Asymptotic freedom, jet events, deep inelastic scattering | Theory in 1973, experiments from the 1970s |
| Medium energy | Hadron spectroscopy, quark model | From the 1960s |
| Low energy (long distance) | Lattice QCD numerical calculations, confinement, glueballs | Theory from the 1980s, experimental confirmation in 2026 |
Remaining Questions
The glueball interpretation of X(2370) still leaves several unresolved issues.
First, there is the quantification of the mixing ratio. It has been shown that the glueball component is "dominant," but its precise proportion (figures such as 73.8% have been given for the scalar glueball candidate f₀(1710)) has not yet been established for X(2370).
Second, there is the matter of peer review. While the determination of spin-parity (PRL 132, 181901) has been peer-reviewed, the paper presenting the glueball interpretation systematically (arXiv:2607.20366) remains at the preprint stage. The ICHEP presentation was a communication to the community, and has not yet undergone peer review verification.
Third, there has been no independent confirmation by another experiment. BESIII is currently the only experimental facility capable of producing J/ψ in bulk in this energy range, and there is no prospect of another facility reproducing the same measurement. Future experiments such as GlueX (Jefferson Lab) and PANDA (FAIR) may conduct related measurements, but these would be difficult to call direct confirmation.
Fourth, the candidate for the scalar glueball ($0^{++}$) remains unresolved. Whether f₀(1500) or f₀(1710) contains the larger glueball component, or whether both are mixed states, is a question that remains unsettled even after the confirmation of X(2370).
The first solid step in a search that has continued for fifty years has finally been taken. But that step is not "the discovery of a pure glueball" but rather "the identification of a particle in which the glueball component dominates," and the verification of the full glueball spectrum predicted by QCD has only just begun.