A Radio Source in the Milky Way Has Been Blazing for 21 Years, Brightened Sixfold, Never Expanded and Emits No X-Rays. Caltech Astronomers Think Something Is Feeding a Hidden Black Hole or Neutron Star.

Most things that flare in the radio sky follow a script. They brighten suddenly, whether from a stellar explosion, a neutron star's outburst or a white dwarf gorging on a companion, and then they fade over days, months or years. VT J1906+0849 has not read the script. First flagged by the Very Large Array Sky Survey on Oct. 27, 2017, as the single brightest transient in the survey, it turns out to have been shining for more than two decades, and a new analysis by a Caltech-led team argues that it belongs to no known class of galactic radio source.

The team, led by Jessie M. Miller of the California Institute of Technology, went back through the archives and found the source already present in the MAGPIS survey in 2005, at about 35 millijansky at 4.86 gigahertz. It then climbed, peaking above 200 millijansky around 2014, a brightening of at least a factor of six over roughly 21 years. It faded after that and then, in late 2025, began brightening again. The paper, posted to the arXiv preprint server on Aug. 21, combines that archival record with new observations across radio, infrared, optical and X-ray wavelengths.

Two findings anchor the puzzle. Very Long Baseline Array observations in 2010, 2022 and 2025 show the emitting region is extremely compact and has not expanded, even though spectral lines in the near-infrared reveal gas moving at thousands of kilometers per second. And a Swift X-ray observation found nothing at all. For a source this luminous, at an estimated distance of 15 to 32 kiloparsecs, roughly 49,000 to 104,000 light-years, that silence is hard to explain.

The brightness temperature, a measure of how intense the radio emission is per unit of emitting area, comes out between 100 million and 10 billion kelvin. No thermal process, such as gas heated by star formation, can reach that. The emission has to be nonthermal, almost certainly synchrotron radiation from electrons spiraling at near light speed through strong magnetic fields, which points to energetic activity around a compact object. Flare stars, magnetically active binaries, novae and cataclysmic variables are all far too faint. A young pulsar could supply the energy, but pulsar wind nebulae evolve over centuries, not the few years over which this source has changed.

That leaves accretion. "These properties favor sustained accretion onto a compact object rather than an impulsive explosion or passively evolving nebula," the team writes. In their picture, a black hole or neutron star is pulling in material and launching a jet of relativistic particles and magnetic field. If that jet is plowing into a dense wind blowing off the accretion disk, the wind would confine the radio-emitting plasma, explaining why the source stays compact. The near-infrared lines support this: one component of the line emission is blueshifted, moving toward Earth, consistent with a fast, lopsided disk wind.

The missing X-rays fit the same story. "The approaching side of the wind may produce the persistent blueshifted near-infrared line, while a denser inner component suppresses direct X-ray emission and the redshifted component of the disk wind," the authors write. In other words, the central engine may be buried inside its own outflow. The closest analog they can find is SS 433, the famous microquasar in Aquila whose accreting compact object is wrapped in a thick wind that interacts with its precessing jets, though VT J1906+0849 would be a far milder version.

None of this is settled. High-resolution infrared spectroscopy could confirm the disk wind, and a deep hard X-ray observation could see through the obscuring material to whatever is powering the source. If those checks hold up, the team suggests, VT J1906+0849 could be the first known member of a population of accreting compact objects in our own galaxy that are nearly invisible in X-rays but extraordinarily loud in radio, a class that X-ray-led surveys would have missed entirely.