A sixteen-member team led by the University of Minnesota has used the James Webb Space Telescope to detect a single near-infrared emission line in the Hubble Ultra Deep Field where even the deepest images show no catalogued object. Published on 22 September in The Astrophysical Journal, the finding is a study in elimination: nine possible identifications across five scenario families, each with a quantified weakness.
- The signal: a 16,004-angstrom (1.6-micron) line with flux 2.5 × 10⁻¹⁸ erg/s/cm² and width below about 300 km/s, with no continuum or companion line from 0.97 to 1.89 microns.
- No flash: two epochs 88 days apart show statistically equal fluxes, arguing against a fast-moving solar-system body; JADES F150W imaging reaches 4.6 nJy at 5σ, yet the line alone should yield about 5.5 nJy.
- Blind search: MOCSS opened nearly all NIRSpec micro-shutters and closed only catalogued objects, yielding 42 candidates; seven appeared in both orientations, six with JADES hosts.
A Signal With No Imaging Counterpart
“Empty” means no catalogued object sits at that exact position in one of the most observed patches of sky. The signal is not a “flash”: its stability across 88 days argues against a fast solar-system source. “Cannot be artificial” in the abstract refers to an instrument artefact, not intelligent origin; no extraterrestrial claim appears in the paper.
The search used MOCSS, multi-object coronagraphy for slitless spectroscopy. Instead of opening only the micro-shutters of known galaxies, the team opened almost all and closed those over catalogued objects, turning NIRSpec into a slitless spectrograph with resolution near 1,000. Of 42 candidate lines, seven appeared in both orientations; six had JADES hosts, leaving the seventh.
The absence is awkward because the JADES F150W image includes the line’s wavelength and reaches 4.6 nJy at five sigma. Simulated through the filter, the line alone would give about 5.5 nJy—roughly seven-tenths of a magnitude above the limit—so it should have appeared. The maximum flux compatible with non-detection is less than half the measured value.
That implies an equivalent width above 7,620 angstroms if the source is persistent. If it was transient, the requirement drops to about 41 angstroms. The JADES image was taken in October 2022, a year before the spectroscopy, allowing a transient; the stable 88-day flux complicates it.
Nine Identifications, No Clean Fit
If the line is [O II], the object sits at redshift 3.29, but the required rest-frame equivalent width is unprecedented in such emitters, and the narrow line would demand a density that suppresses [O II]. If [O III], at redshift 2.20, a sister line should appear at 15,864 angstroms with signal-to-noise near 2.9; none is present. If Hα, at redshift 1.44, the equivalent width nears the maximum for a zero-age burst, and Hβ is missing.
Lyα is the most attractive option: a galaxy at redshift 12.16, where no other line would fall in range. It would need a rest-frame equivalent width above 580 angstroms, against a lower limit of about 40 for JADES-GS-z13-1-LA. At that redshift the intergalactic medium is more than 90% neutral, requiring an ionized bubble a weak source could not sustain alone. None of the three neighbours within one arcsecond has a compatible photometric redshift.
Transients remain possible. A flaring Milky Way star is ruled out because a second Brackett line is absent. A Type IIn supernova would need a visible Paschen line, though dust could attenuate it. The most comfortable option is a tidal disruption event: late-time emission can show strong Hα with a weak continuum, thick gas suppresses forbidden lines and dust weakens Hβ, explaining both the lone line and the blank earlier image. It does not explain the stable 88-day flux, though the authors note that “some TDEs evolve on longer timescales.” An active nucleus in a neighbour illuminating a separate gas cloud, as in voorwerps, still leaves Hα and [O III] requiring missing lines.
Resolution needs a redshift, and without distance there is no identification. The authors propose NIRSpec integral-field-unit mode, “particularly powerful, enabling simultaneous spectroscopy of both this line-emitter and nearby galaxies.” No approved programme is announced, only that “follow-up observations are essential.”
The paper is the opposite of the “astronomers have no idea” framing. Its deliberately flat presentation—a single-line emitter found in an unbiased search—contains a systematic elimination in which every hypothesis carries a numerical objection. The authors’ closing line is more honest than any headline: “Regardless of its true nature […] this object is highly unusual and probes extreme physical conditions.” They do not know what it is, but they know with precision what it cannot be. That, plus the new technique, is the real result—and it sets up a redshift measurement still waiting to be scheduled.