Posted by The exoplanet definition sounds like the easiest thing in astronomy to write down: a planet that orbits a star other than the Sun. It holds up until you try to make it official. What the field actually has is a working definition, a short statement adopted by an International Astronomical Union working group in 2003, revised once since, never voted into anything binding, and argued over more or less continuously by the people who use it every day.
The trouble is that the universe did not sort its objects into tidy bins. Mass runs in an unbroken sequence from pebbles to stars, with no convenient gap where a committee could plant a fence. Somewhere along that sequence an object stops being a planet and becomes something else. Choosing where is a human decision, not a natural one.
It helps to know what the working exoplanet definition is not. The famous 2006 vote that demoted Pluto applies only inside our own Solar System. Its third criterion, that a planet must have cleared its orbital neighbourhood of debris, cannot be checked for a world 300 light years away. We cannot see the debris. Exoplanets live under a separate rule with different logic.
Thirteen Jupiters, Give or Take
The upper boundary comes from nuclear physics. Pack enough hydrogen and helium into a ball and the core eventually gets hot and dense enough to fuse deuterium, the heavy isotope of hydrogen. That threshold sits near thirteen times the mass of Jupiter. Below it an object never ignites anything at all and just cools, slowly, forever. Above it deuterium burns for a few million years and the object gets called a brown dwarf. Keep going, to roughly seventy five or eighty Jupiter masses, and ordinary hydrogen fusion starts, at which point you have a star.
A threshold set by physics rather than by taste sounds tidier than it is. The deuterium limit is not one number. It shifts with the object’s helium fraction and its abundance of heavier elements, and detailed models place it anywhere between roughly eleven and sixteen Jupiter masses. Thirteen is an average of a blurry line. An object of twelve and a half Jupiter masses might sit above the limit for its own composition, and we rarely know the composition well enough to say.
Which means the border is drawn in pencil.
What the Object Was Born As
Plenty of astronomers think mass is the wrong criterion altogether. Planets are supposed to build up inside a disk of gas and dust around a young star, grain by grain and then by runaway gas accretion, while brown dwarfs collapse straight out of a molecular cloud the way stars do, only failing to gather enough material. A definition based on origin would be more honest about what these objects are.
It would also be unusable. You cannot run the film backwards. By the time we detect a companion it has sat there for hundreds of millions of years and carries almost no memory of how it assembled. The working exoplanet definition sidesteps formation because formation is not observable, and pays for that with cases nobody is happy about.
The four giant planets imaged around HR 8799 in 2008 are a good example. Nobody weighed them. Their masses come from measured brightness fed through a model of how such objects cool, which requires an assumed age for the system. Change the assumed age and the masses move. Move them far enough and a planet slides over the deuterium line into brown dwarf territory without anything about the object itself having changed. Then there is 2M1207 b, the first planetary mass companion ever imaged directly, back in 2004. It is a few Jupiter masses. It orbits a brown dwarf rather than a star, which the 2003 wording did not really anticipate.
A 2018 revision from the IAU’s exoplanet commission tried to firm things up: an object now has to be massive enough for its own gravity to pull it round, and light enough relative to what it orbits that near equal pairs count as binaries. Useful, and still not a law.
The Mass You Measure Is Not the Mass It Has
There is a further complication that catalogues note in small print. The radial velocity method, the technique that found 51 Pegasi b in 1995 and put Michel Mayor and Didier Queloz on the path to a Nobel Prize, measures the star’s wobble along our line of sight. It cannot tell how the orbit is tilted. What comes out is a minimum mass, the true mass multiplied by the sine of an unknown angle.
For a randomly oriented orbit the minimum is usually close enough. But an orbit seen nearly face on hides most of its motion, and a companion listed at eight Jupiter masses could in principle be twenty. Proxima b, announced in 2016 around the nearest star to the Sun, has never been caught transiting, so its listed mass of a little over one Earth is a floor rather than a value. Astrometry from Gaia has started pinning down inclinations for the wider orbits, and a handful of objects have quietly moved from the planet column to the brown dwarf column as a result.
Worlds With Nothing to Orbit
The 2003 statement says a planet orbits a star or a stellar remnant. That last clause exists because of Aleksander Wolszczan and Dale Frail, who in 1992 found the first confirmed planets anywhere outside our system, orbiting a millisecond pulsar. Nobody wanted to exclude them on a technicality.
Free floating objects are harder. Gravitational microlensing surveys such as OGLE and MOA keep catching brief brightenings of background stars caused by something planet sized drifting past with no star attached. Each event happens once and never repeats, which makes ordinary confirmation impossible. In 2023, JWST imaged something stranger still in the Orion Nebula: dozens of Jupiter mass objects, many of them paired off and orbiting each other, drifting free of any star. Nothing in either formation story predicts pairs like that.
Under the strict reading these are not planets at all. The IAU statement calls free floating objects below the deuterium limit sub brown dwarfs, whatever their history. That satisfies almost nobody: an object that formed in a disk and was later flung out by a gravitational encounter is a planet by biography and a sub brown dwarf by definition, and no observation can tell it apart from something that condensed alone. The Nancy Grace Roman Space Telescope should turn this from a scattering of oddities into a census, which will make the naming problem worse before it makes it better.
Candidate Is Not Confirmed
Every large survey produces candidates first. Kepler called them Kepler Objects of Interest, TESS calls them TOIs, and the label means only that the light curve of a star dipped in a way consistent with a planet crossing in front. Plenty of things imitate that. A pair of eclipsing stars in the background, blended into the same photometric aperture, produces a shallow periodic dip, and so does a binary grazing the edge of its companion. A large starspot rotating in and out of view can fake a signal for months.
Confirmation means an independent line of evidence. The cleanest is a radial velocity measurement of the same star showing a wobble at the same period, which delivers a mass and rules out most impostors. Where the star is too faint for that, and Kepler’s stars often were, astronomers use transit timing variations: in a system with several planets, the pull between them makes each transit run slightly early or late, and the pattern of those shifts yields masses. All seven TRAPPIST-1 planets, announced in 2017, were weighed that way.
There is a third route that troubles some researchers. Statistical validation calculates the odds that any known false positive scenario could produce the observed signal, and if those odds are small enough the planet is declared real without anyone ever detecting it a second time. In May 2016 more than a thousand Kepler candidates were validated in a single announcement using that approach. Kepler-186f, the first Earth sized world found in its star’s habitable zone, is validated rather than confirmed by mass. Its density, composition and whether it is rocky at all remain unmeasured.
Sometimes the Catalogue Takes It Back
Peter van de Kamp spent decades reporting planets around Barnard’s Star from photographic plates. The wobble turned out to be in his telescope, appearing whenever the lens assembly had been serviced. When 51 Pegasi b was announced, David Gray argued the signal came from oscillations in the star itself, and it took further data to settle the question in Mayor and Queloz’s favour. Alpha Centauri Bb, announced in 2012 as the closest exoplanet known, was shown three years later to be an artefact of how the observations were spaced in time. It no longer exists.
This is also why two respectable exoplanet catalogues will give you different totals on the same afternoon. They apply different upper mass cuts and different standards of evidence, so the count depends on whose exoplanet definition you are using. Neither is wrong. The number is just softer than a number ought to look.
None of this is a failure of the field. A definition that gets revised, contested and occasionally embarrassed is a definition doing its job, and the borderline objects are usually the interesting ones. If you want to follow where those arguments go next, and what the next generation of instruments does to the boundaries, SETIworld is a good place to keep reading.
