24 Boo b orbits at 39 solar radii. Its star once swelled past that.
Old giant stars should have eaten their close planets. Three still have one. The clearest case is 24 Boo b: by every model tested, its star once grew to roughly 100 solar radii or more.
The planet is at 39. The star reached 100 or more. Which one is wrong?
A swelling star should eat the planets close to it.
Stars like the Sun get huge late in life. They puff up into red giants.
Picture a balloon that keeps growing. Anything near it ends up inside.
The rule of thumb: a planet within about 1 au of a Sun-like star gets swallowed. One au is the distance from Earth to the Sun.
Surveys agree so far. Giant stars show few close giant planets, and earlier modelling found nearly all known planets around clump giants sit outside the predicted survival limit.
What if one survived?
Source: PDF p1 (abstract, intro); Kunitomo et al., arXiv 1106.2251
A star swells. A planet close by is in trouble.
Schematic of the old rule. Not to scale.
Before this paper, one planet broke the rule.
Astronomers had a hard job. A giant that is still swelling looks a lot like one that already swelled and shrank.
Starlight alone could not reliably tell them apart. So a close planet around a giant was hard to judge.
Only one case had been caught: 8 Ursae Minoris b, reported in 2023 in the journal Nature.
How do you tell which stars already swelled?
Source: PDF p1; Hon et al., Nature 618, 917 (2023), doi 10.1038/s41586-023-06029-0
Close planets found around stars that already swelled.
Counts as the paper states them. Units: number of systems.
Stars ring like bells. The ring shows what is inside.
A star wobbles, and its brightness wobbles with it. Different insides make different wobble patterns.
The team read that pattern in data from the TESS space telescope. All three stars match the red clump pattern: giants that already did their biggest swelling and shrank back.
The key number is the spacing of the pattern. It is 318.0 vs 337.0 vs 338.8 seconds for the three stars.
And how do you find a planet you cannot see?
Source: PDF p2, p5 (Table 1), Fig. 1 on p3
A star rings. Its insides change the sound.
Illustration of the idea, not data. The three spacing values are from Table 1 (p5), in seconds.
A planet tugs its star. The tug gives away the orbit.
A planet pulls its star in a small circle. From Earth, the star seems to move toward us and away from us in a steady rhythm.
The team refit older measurements from four instruments with one method. Orbits came out at 0.1827 vs 0.4885 vs 0.5157 au.
The periods are 30.337 vs 101.630 vs 132.93 days. The 24 Boo planet is the closest in.
So how big was each star at its biggest?
Source: PDF Table 1 (p5), Fig. 2a to c (p4), Methods (p7)
How far each planet orbits, in au.
Units: au. Table 1, p5.
- La Palma, Spain: HARPS-N and GIANO
- Okayama, Japan: HIDES (24 Boo)
- Bohyunsan, South Korea: BOES (HD 12648)
The star once swelled well past the planet. The planet is still there.
Now the key test. The team asked stellar models how big each star got at its biggest, trying different settings for how heat moves inside the star.
The 24 Boo planet orbits at 39.3 solar radii. The star's old size runs roughly 100 to 130 in the tested models.
TYC 4282-605-1 b also orbits below every tested curve, at 105.0 solar radii.
The old size comes from models, not from a measurement. That is why this stamp says Inferred. The 100 to 130 range is read off a chart in the paper, so it is approximate.
What does that change?
Source: PDF p2, Fig. 3a (p6), Table 1 (p5). Orbit in solar radii is computed: 0.1827 au x 215.032.
24 Boo: distance from the star, in solar radii.
Hatched bar: model range, read off Fig. 3a, so approximate. Orbit is computed. Star today is from Table 1.
The rule now has exceptions we can count.
Before, one odd case could be waved away. Now there are three systems with the same puzzle.
The authors say a common process may be at work that current theory does not include. With more such planets, they say, scientists can test which process matters.
Our takeaway, not the paper's: when you read "impossible," ask what was measured and what was modelled. Here the orbits were measured. The old star size was modelled.
So how sure is the result?
Source: PDF p1 (Significance box), p4; Hon et al. 2023
From one odd case to three more.
(Nature)
(this paper)
Dates: Hon et al., Nature 618 (2023); arXiv 2611.04544v1 (3 Nov 2026).
One of the three is a close call.
HD 12648 b is the soft spot. Its star's old size shifts a lot with one model setting.
At low settings the chance that the orbit sat inside the star is nearly 100%. At setting 2.2 it falls to 43.10%.
So "should not have survived" is firm for two planets and soft for one.
A second wrinkle: for 24 Boo, the measured mass and gravity differ from a direct size measurement by about 3.1 and 4.6 sigma. Sigma counts how far apart two numbers are, in units of their uncertainty. The star's radius and brightness agree.
And if the planets did survive, how?
Source: PDF p2 (engulfment probability, interferometry comparison)
HD 12648 b: chance the orbit was inside the star.
Units: percent. "Nearly unity" at low settings is the paper's wording, so the top bar is soft.
Nobody has shown how these planets survived.
The paper lists ideas. The star may have had a partner or merged with one.
A swallowed planet may have left a disk that pushed another planet inward. A planet may have dipped in and come back out.
The authors call these qualitative possibilities, not demonstrated mechanisms. None has been tested.
Also, the planet masses are minimums: 0.762 vs 14.278 vs 2.539 Jupiter masses.
What would settle it? More planets like these.
Source: PDF p2, p3 to p4, Table 1 (p5)
Four numbers to keep.
systems with a close planet around a star that already swelled.
PDF p1how far 24 Boo b orbits. A 30-day year.
Table 1, p5solar radii: the same orbit, computed. The star's old size was roughly 100 to 130.
Computed; Fig. 3a, p6HD 12648 b: the soft spot, at setting 2.2.
PDF p2