Seventeen MeV: An Anomaly and Its Fences

Somewhere in the debris of three light nuclei there is a bump that should not be there, and ten years of work have neither made it go away nor made it solid. When an excited beryllium-8 nucleus settles to its ground state, it sometimes sheds its energy as an electron and a positron instead of a photon, and in 2016 the ATOMKI group in Debrecen reported that those pairs pile up near an opening angle of 140 degrees in the 18.15 MeV transition far more often than the known pair-creation process allows: a 6.8-sigma excess, readable as a new particle of about 17 MeV decaying on the spot into the pair. The same reading, at the angles the same mass predicts, was then reported in helium-4 and carbon-12. The particle was named X17 and given a shape: a vector boson coupling to neutrons and electrons at the milli-charge level and barely to protons at all — protophobic — because an older pion-decay search had already forbidden it to touch protons. This paper audits the anomaly's standing on the record as it exists in late 2026 and prices every fence around it. The findings, printed plainly: no independent experiment has confirmed the bump at discovery strength, and none has killed it. The one direct replication attempt, MEG II, saw nothing — but its limit sits at twice the claimed signal size, so its silence wounds the claim without closing it. The one fully independent accelerator test, PADME, did not return a clean null: scanning positron annihilation across the mass, blind, it found its largest excess — 2.5 sigma local, about 1.8 sigma after the look-elsewhere correction — at 16.90 MeV, the address the nuclei had already named. Beam-dump and pion-decay fences have cut the particle's allowed coupling window by roughly two-thirds of its original width and forced its proton coupling below a fourteenth of its neutron coupling. A nuclear-structure explanation, steelmanned at full strength here, must reproduce one mass-like scaling across three nuclei, quiet controls in every run, a replication on a different spectrometer, and an accelerator excess with no nucleus in it. Both doors are still open. Both are expensive. The claim of this paper is exactly that standing, no larger: Grade: **ANOMALY STANDS — NEITHER CONFIRMED NOR DEAD, EVERY FENCE PRICED.** This paper stands outside my work on the vacuum catastrophe; it draws nothing from that work and repairs nothing in it, and it touches no claim of that program.