Research/Space
A new eclipsing star whose tiny companion is twice its expected size
TESS shows that TYC 6340-23-1, a Sun-like star 111 parsecs away, is eclipsed every 2.33 days by a companion that Gaia weighs at 0.097 solar masses but that is 0.224 solar radii across, 1.9 times the size expected. A pre-registered check finds the system does not move with any known young star group, so the simplest explanation, youth, is now disfavoured.
Nov 20, 2026 · 7 min read

We found a new eclipsing binary, TYC 6340-23-1 (TIC 71945261), in which a very small star crosses a Sun-like star every 2.3287 days. Gaia's spectroscopic orbit puts the companion's mass at 0.097 solar masses, right at the boundary between stars and brown dwarfs. The eclipses in NASA's TESS data say it is 0.224 solar radii across, about 1.9 times the radius a normal star of that mass should have. A companion that oversized would be extreme, so we set out to rule out the ordinary explanations before calling it anything more. This week's pre-registered check removes the easiest one: the system does not move like a member of any known young star group.
2.3287 d
orbital period (TESS sector 92)
0.224 R☉
companion radius from the eclipses
1.9×
radius relative to a normal 0.097 M☉ star
0.4%
chance of membership in a known young group
How we found it
The star turned up in our search of Gaia DR3's single-lined spectroscopic orbits for the transits and eclipses those orbits predict (the same search behind our earlier finding that most of Gaia's close brown-dwarf orbits cannot be real). Here the orbit passed the test: TESS sector 92 shows a 6.1% deep primary eclipse at Gaia's period (Gaia gives 2.32893 ± 0.00014 days), odd and even eclipses of equal depth, and a 0.39 ± 0.10% secondary eclipse at phase 0.5. The secondary eclipse implies a companion temperature of about 2,950 K, as expected for a very cool, low-mass star.
Before claiming it, we checked the AAVSO VSX, the TESS eclipsing-binary catalogue, the TESS Ten Thousand catalogue, Gaia's own eclipsing-binary and variability tables, ASAS-SN, ExoFOP, NASA's list of TESS threshold-crossing events, SIMBAD and the literature. None lists the star as eclipsing or variable. SIMBAD knows it only as a spectroscopic binary from Gaia.
Why the size matters
The smallest stars are fully convective, and models of how big they should be are tested on exactly this kind of system: a low-mass star eclipsing a brighter Sun-like partner. Most such companions come out a few percent larger than models predict. A factor of 1.9 is far outside that range. The main explanations are that the system is young and the companion is still contracting, that Gaia's velocity amplitude (and with it the mass) is wrong, that the light comes from a blended triple, or that the companion really is anomalously large. The fit itself sets a floor on the size: even a perfectly central eclipse would need a companion of about 0.20 solar radii.
Companion radius: measured versus expected
The check we ran this week
We wrote the rules down before the job ran. A 0.1 solar-mass star is this large if it is only a few tens of millions of years old, so we tested whether TYC 6340-23-1 moves with any known young association using BANYAN Σ, the standard tool for that question, with Gaia's position, parallax, proper motion and the binary's systemic velocity (−17.57 ± 0.39 km/s). As a positive control, the same code had to place AU Mic in the β Pictoris moving group with at least 80% probability. It did (92.7%). For TYC 6340-23-1 the young-group probability is 0.41%, and 3.7% without the velocity. Both fall below the pre-registered 20% line, giving the reading FIELD_KINEMATIC. Its space velocity, (U, V, W) = (−19.6, −14.8, −6.3) km/s, is ordinary for the Galactic disk.
Probability of belonging to a known young group (BANYAN Σ)
The second planned test was to check Gaia's velocity amplitude against independent radial velocities from the RAVE, GALAH and APOGEE surveys. None of them observed the star, so that test has no verdict. One fact named in advance as ungraded points the same way as Gaia's orbit: the spread of Gaia's own 15 velocity measurements matches the orbit, not the twice-larger swing a normal-sized companion would require. Those measurements are the ones the orbit was fitted to, so they are not an independent check.
Velocity swing: what Gaia saw versus the two hypotheses
| Quantity | km/s |
|---|---|
| Gaia epoch-velocity range (15 epochs, outliers removed) | 32.54 |
| Expected from Gaia's orbit (2K) | 32.32 |
| Needed for a normal-size 0.21 M☉ companion (2K) | 64.8 |
A correction to how far our fitting code can be trusted
The radius comes from our own eclipse-fitting code, and that code has not passed validation. On 20 November a repaired version failed its pre-registered check against 14 published low-mass eclipsing binaries from the EBLM project: 6 of 14 passed where 80% was required (results). We had predicted it would pass. Its one large overestimate of a companion radius was a grazing eclipse (impact parameter 0.99). This system's eclipse is not grazing (0.67), and the star's radius inferred from the light curve matches Gaia's to within 1%. Every other radius miss in that check came out too small, never twice too large. That makes a fitting artefact an unlikely cause of the 1.9× size, but the radius still needs to be confirmed independently.
What it means
The easiest explanation, membership of a known young group, is now disfavoured. That does not prove the system is old, and Gaia's mass has not been checked independently. Under our own pre-registered rule we will not call the companion inflated until a spectrum does two things: shows whether lithium, a marker of youth, is present, and measures the velocity swing directly. The star is bright (V = 10.7) and well placed for southern telescopes, so one or two high-resolution spectra would settle it. Gaia DR4, due on 2 December 2026, should help before then. Gaia DR3's orbit used data only up to May 2017, so the DR4 velocities measured after that date are independent of it. We have pre-registered a test that uses only those later epochs to decide between Gaia's 16.2 km/s and the 32.4 km/s a normal-size companion would need. If Gaia's mass holds and the star is old, the companion would sit far outside the few-percent offsets usually measured for such stars. If the mass is wrong, the system is a clear example of a Gaia orbit that passes the eclipse test but gets the companion's weight wrong by a factor of two.


