For years, exomoons felt like one of those astronomical discoveries that should arrive almost automatically once telescopes became sensitive enough.
We had found thousands of planets around other stars. Our own Solar System contains more moons than planets. Surely the first moon beyond it was mostly a matter of waiting.
I had absorbed that assumption without examining it closely. Then I read the paper behind the July 2026 report from the CD-35 2722 system, and the long delay began to make more sense.
The object reported by Kevin Hoy and 12 colleagues is not a tidy counterpart to our Moon. It has a minimum mass close to Jupiter’s, it circles a brown dwarf rather than a planet, and nobody has photographed it directly.
What the team found was a rhythm in the brown dwarf’s motion.
The peer-reviewed paper in Nature describes this as evidence for an exosatellite, not an uncontested exomoon. That distinction is central to the story.
This is one paper, and the candidate still needs further observation. It may be the first plausible detection of its kind, but it is not the simple confirmation astronomers have been waiting for.
Why finding a moon is harder than finding a planet
A planet gives astronomers several ways to notice it. It may dim its star when it passes in front, pull the star back and forth through gravity, glow in infrared light, or bend the light of a more distant star.
A moon is usually smaller, fainter and attached observationally to an object that is already difficult to separate from its star.
If a moon transits, its dip in starlight can be tiny. Its position changes from one planetary transit to the next, so the signal does not repeat as neatly as a planet’s. The moon can also alter when and how long the planet transits, but stellar activity, other planets and the details of the data can produce confusing alternatives.
Direct imaging presents a different version of the same problem. Even a young giant planet or brown dwarf is faint beside its star. A moon beside that companion is fainter again and sits extremely close to it on the sky.
This helps explain why earlier candidates remained disputed. The absence of a confirmed exomoon was not evidence that other planetary systems lack satellites. It was evidence that the signal is easy to imitate and hard to repeat.
The useful question was therefore not simply whether telescopes could see faint things. It was whether astronomers could isolate a moon’s particular influence from every other movement and fluctuation in the system.
The brown dwarf became the thing that wobbled
CD-35 2722 is a young system in which a star of roughly half the Sun’s mass is orbited by a brown dwarf called CD-35 2722 B.
Brown dwarfs occupy the awkward territory between planets and stars. They are more massive than ordinary giant planets but cannot sustain hydrogen fusion in the way a star can. In this case, the brown dwarf is estimated at about 37 times Jupiter’s mass.
Hoy’s team observed its infrared spectrum with the CRIRES+ instrument on the European Southern Observatory’s Very Large Telescope in Chile. Their monitoring ran from October 2023 to February 2026, according to ESO’s account of the work.
The radial-velocity method looks for regular shifts in the wavelengths of light. As an unseen companion pulls an object slightly towards and away from us, lines in its spectrum move by small amounts through the Doppler effect.
This technique is famous for revealing planets through the wobble they induce in stars. The clever step here was to measure the spectrum of the directly imaged brown dwarf itself and ask whether a still smaller companion was tugging on it.
The best-fitting model in the final Nature paper contains at least one satellite with a minimum mass of about 0.9 Jupiter masses and an orbital period of roughly 170 days.
“Minimum” matters because radial velocity measures motion along our line of sight. Without knowing the orbit’s exact inclination, the researchers cannot turn that signal into one perfectly determined mass. The true object could be heavier.
The result changed between the early manuscript and the final paper
There is a useful lesson in comparing the first public manuscript with the paper that emerged from peer review.
The initial version posted to arXiv described a strong candidate of at least 0.743 Jupiter masses on a 169-day orbit and a less certain second candidate of 0.277 Jupiter masses on an 87-day orbit.
The authors attached an unusually direct warning to that version, asking readers to wait for the final Nature paper before drawing specific conclusions.
The published account is more conservative. It centres on at least one object at about 0.9 Jupiter masses and says models containing two satellites are highly unstable.
That does not make the result weaker in any embarrassing sense. It shows why peer review and continued modelling matter, especially when a claim sits at the edge of what an instrument can establish.
A periodic wobble is evidence of an orbiting mass. It is not a photograph, a spectrum of the candidate itself, or a complete history of how the system formed.
Further measurements should test whether the 170-day pattern continues at the predicted phase. A real orbital signal should keep time. Instrumental effects or unrecognised behaviour in the brown dwarf may not.
It behaves like a moon but has the mass of a planet
The naming problem is not cosmetic.
In our Solar System, the hierarchy feels obvious. Stars are at the centre, planets orbit stars, and moons orbit planets. CD-35 2722 gives us a star, then a brown dwarf, then an object massive enough to be called a planet if it orbited the star directly.
The third object is a satellite in the literal sense that it orbits something else. “Exosatellite” is therefore the least controversial term.
Calling it an exomoon is harder because there is no formally accepted definition that neatly covers a Jupiter-mass body orbiting a brown dwarf. The Nature paper explicitly leaves open whether this object meets the presently undefined criteria.
The mass ratio offers some perspective. A body near Jupiter’s mass sounds enormous, but its host is about 37 Jupiters. The smaller object is still only a few per cent of the brown dwarf’s mass.
Even so, this is not the kind of moon astronomers ultimately hope to detect around an ordinary exoplanet. It is a gaseous, planetary-mass body in a system that blurs the categories built from our local examples.
I find that ambiguity more useful than forcing a clean label. Nature does not have to organise distant systems according to the vocabulary we developed while looking at one star and its planets.
The method had been waiting for a sufficiently bright target
The idea of detecting moons through the Doppler wobble of their hosts did not appear in 2026.
A 2018 study led by Andrew Vanderburg calculated that a massive moon around a directly imaged planet could produce a radial-velocity signal large enough for existing or next-generation instruments to measure.
In 2023, Jason Wang and colleagues tested the approach on observations of another brown dwarf and published detection limits for satellites around self-luminous companions. Their analysis showed that current instruments were sensitive mainly to very large satellites, while future telescopes could reach much smaller mass ratios.
That sensitivity bias explains the strangeness of the first plausible result. Astronomers did not begin with a small moon around a familiar planet because that was the easier target. They began with a planet-mass satellite around a luminous brown dwarf because its gravitational pull was large enough to measure.
There was also an earlier 2026 hint in the HD 206893 system. A team led by Quentin Kral reported an astrometric wobble around a substellar companion, but described the possible exomoon interpretation as tentative.
The CD-35 2722 signal is stronger, yet the sequence is a reminder that discoveries rarely arrive as a single clean moment. Methods improve, candidates accumulate, alternative explanations are tested, and the word “first” keeps changing meaning.
The real milestone is a new way to search
If the candidate survives follow-up, its importance will not depend entirely on whether a committee eventually calls it a moon.
The work demonstrates that astronomers can take the radial-velocity method down one level in a system’s hierarchy. Instead of using a star’s light to infer a planet, they can use the light of a directly imaged substellar object to infer something orbiting it.
That is still a demanding measurement. Smaller planets are fainter than brown dwarfs, and smaller moons produce weaker wobbles. Reaching analogues of Europa, Titan or even Earth’s Moon will require more sensitive instruments, suitable nearby targets and long observing campaigns.
ESO argues that the 39-metre Extremely Large Telescope should extend the method to smaller exomoons. That is a prospect, not a detection schedule.
What changed in July 2026 was not that the universe finally supplied a moon. It was that one moon-like gravitational signal became strong enough to survive a serious analysis, while remaining strange enough to expose the limits of the word itself.
We expected technology to make exomoons obvious. Instead, the first plausible candidate arrived as a wobble inside a hierarchy we do not yet know how to name.
