The Repository Absolute Magnitude Space

Betelgeuse's Six-Year Cycle May Not Be Its Companion

October 7, 2026 · Runtime 12:53 · Watch on YouTube

Betelgeuse's companion was predicted from a six-year cycle in the star's light, and a faint point turned up where it was predicted. Three months earlier, a Gaia study of 224 giants with the same kind of cycle had found that most of them behave like single stars.

Transcript

A point where the cycle said 0:00

This faint point beside Betelgeuse was found where a six year cycle in the star's light said a companion should be. It looked like a clean success.

224 other stars, most with no companion 0:10

But three months before it was announced, a paper had tested the same idea on two hundred and twenty four other stars with the same kind of cycle. One of its four authors had helped predict Betelgeuse's companion. And the paper concluded that most of those stars have no companion at all.

The test was run with Gaia, a spacecraft built to measure where stars are. It never weighed a companion, and it never saw one. For each star it returned one number, and this episode ends on what that number actually was.

The long secondary period 0:45

Red giants breathe. They swell and shrink over weeks to years, and their light rises and falls with them. But about a third of the luminous ones carry a second, slower rhythm on top, several times longer, and it has resisted explanation for almost a century.

It is called a long secondary period. Betelgeuse has both kinds. Its pulsation runs about four hundred days. Its long secondary period runs about twenty one hundred, close to six years. The slow one is the cycle that pointed to the companion.

The slow cycle has been blamed on almost everything. A pulsation mode. Dust. Giant convection cells, like the ones ALMA filmed churning on the red giant R Doradus over a single month in twenty twenty three.

The companion idea, and three predictions 1:40

And a companion. In that version, a small star or a brown dwarf circles inside the giant's outskirts once per cycle. It drags a wake of dust that dims the giant on schedule, the way ALMA has mapped a companion's spiral wake around another giant, LL Pegasi.

Written as a chain, the idea makes three predictions. The giant dims once per orbit. Its velocity toward us rocks once per orbit. And the giant itself is pulled back and forth across the sky, once per orbit. The first two had been seen many times. The third is what Gaia can check.

The rocking, measured since 2009 2:22

The rocking was measured long ago. In two thousand nine, a team used a spectrograph on the Very Large Telescope to follow fifty eight of these stars. Their velocities swung by one to two kilometres per second, and none of them by less, although the instrument could have seen it.

And in twenty twenty one the idea got its strongest support. In mid infrared light, many of these giants show a second, shallower dip, locked to the same slow cycle. That is what you would expect if a dusty cloud trailing a companion passed behind the star.

Gaia's 9,614 velocity curves 2:59

Gaia went much bigger. From a point in space beyond the Moon's orbit it scanned the whole sky for years, and besides positions it measured velocities. In October twenty twenty three it released velocity curves for nine thousand six hundred and fourteen slowly varying red stars.

Six thousand and ninety three passed every quality check. The new paper kept the ones with a long secondary period, within one and a half kiloparsecs. That left two hundred and twenty four. The seventy three closest, within seven hundred and fifty parsecs, carry the test.

Their velocities cluster tightly, just as they did in two thousand nine. The median rocking is one point six three kilometres per second, and only twelve of the two hundred and twenty four rock by less than one. That is the measured part.

Take one of them, a red giant Gaia knows only by an eighteen digit number, six hundred and eighty three parsecs away. Its slow cycle lasts seven hundred and eleven days. Its velocity swings by one point four eight kilometres per second, in step with its light, a fifth of a cycle apart, just like the rest of the sample.

Betelgeuse sits right in the pack. Its own rocking, from a century of records, is about one and a half kilometres per second. Antares, another nearby supergiant with a slow cycle, rocks by about two point seven. Binaries that are known to be binaries rock by about ten.

Read as orbits: 0.084 suns 4:39

Now read the rocking as an orbit. Assume every giant weighs one Sun, and the companions come out remarkably uniform. The median is zero point zero eight four solar masses. That is right on the line between the smallest stars and brown dwarfs.

The brown dwarf desert 4:58

That uniformity is itself a problem. Around stars like the Sun, companions of exactly that mass, a few astronomical units out, are notoriously rare. Astronomers call the gap the brown dwarf desert.

Fewer than one in a hundred Sun-like stars has such a companion. Yet these giants are what Sun-like stars become, and about thirty in a hundred show the slow cycle. The companions would have to appear, or grow, after the star became a giant.

And they would have to survive. Giants end as white dwarfs, so a comparable fraction of white dwarfs should still carry a companion like that a few astronomical units away. Infrared surveys find brown dwarf companions around about one white dwarf in a hundred, or fewer.

The third prediction: a loop on the sky 5:50

So the paper went to the third prediction. A companion does not only rock a star along our line of sight. It also drags it around a small loop on the sky, and measuring positions is the one thing Gaia was built for.

When the companion is huge, the loop is easy to see. In twenty twenty four, Gaia caught a star about two thousand light years away looping around something invisible. The dark partner turned out to be a black hole of thirty three Suns.

What Gaia returns: RUWE 6:22

But Gaia does not return loops. Each time it sweeps past a star, it records one position along its scan. Then a model of a single star is fitted to all of them: a position, a parallax, a steady motion. And Gaia reports how badly that fit went.

That report is a number called the renormalised unit weight error. It is scaled so that a well behaved single star scores about one. A star being tugged around a loop leaves residuals the single star model cannot absorb, and the number climbs.

Here, the loops are tiny. At one and a half kiloparsecs, these companions would swing a giant by at most about two tenths of a thousandth of an arc second. That is a coin seen from twenty five thousand kilometres. Betelgeuse's disc, for comparison, is forty four.

Simulated Gaia: 58 of 73 7:20

So the team built a test. For every star, they gave it the companion its own velocity implied, fifty times over with random orientations. They flew a simulated Gaia past it on the real scanning schedule, and computed the number Gaia should have reported.

Then they compared it with the number Gaia did report. Within seven hundred and fifty parsecs, the prediction came out higher than the measurement for fifty eight of the seventy three stars. Four in five. The loops the companions should have drawn are not in the data.

The control: real binaries show up 7:56

Is the method simply blind? The paper checks it on giants that do have partners: ellipsoidal variables, stretched into eggs by a close companion. They rock at about ten kilometres per second, with companions near two thirds of a solar mass.

For the nine within seven hundred and fifty parsecs, the simulation predicted an elevated number for all nine. Gaia measured an elevated number for seven. The method that finds no partner around the slow cycle stars finds the partners that are there.

There is a second comparison. Red giants that pulsate but have no slow cycle make a natural reference. If companions caused the slow cycle, the stars that have it should look less single than the ones that do not. They look slightly more single.

Can dust fake or hide it? 8:49

Could dust fake the signal, or hide it? A dark cloud covering one side of a giant shifts the centre of its light. The team injected that into the simulation too.

For the nearest star, at three hundred and ninety one parsecs, one sided dust could shift the light by up to about six tenths of a thousandth of an arc second. Random jumps nine times smaller would already have raised the number. Smooth, slow drifts can partly hide in the fit, so dust is not fully ruled out.

Measured versus assumed 9:23

So separate what was measured from what was assumed. Measured: the velocities, the periods, the distances, and the fit residual. Assumed: that each giant weighs one Sun, that the orbits are circular, and that the companion adds no light. The authors varied the orbits and the viewing angles. The answer held.

Their conclusion is careful but blunt. Low mass companions are not the dominant cause of long secondary periods. Some of these stars may be binaries. Most behave like single stars, and the cause, for most of them, is probably inside the star.

U Monocerotis: a cycle that is a binary 10:07

Binaries that make slow cycles do exist. U Monocerotis is a pulsating giant with a partner, inside a disc of dust. Every six and a half years the disc partly covers the giant, and the light dips on schedule. It is just not what most of these stars are.

Betelgeuse: typical, and not 10:27

Which brings it back to Betelgeuse. Its rocking, its brightness swings and its period all look typical of the population that failed the test. So the paper says its slow cycle may likewise come from something other than an orbit.

But in one way Betelgeuse is not typical at all. It weighs sixteen and a half to nineteen Suns, and the giants in the test were modelled at one. The authors say supergiants could be a different regime, and that Betelgeuse may be the special case.

And the direct test cannot be run. Gaia cannot measure Betelgeuse at all, because it is far too bright, and its disc is ten times wider than its own parallax. Antares, with a slow cycle of its own, is out of reach for the same reason.

Both sides at once 11:17

So two things are true at once. A companion was predicted from the cycle, and a point of light turned up where it was predicted. And in the stars Gaia can measure, that same kind of cycle mostly is not made by companions. Morgan MacLeod is an author on both sides.

December 2026 and November 2027 11:37

Two dates can move this. In December, Gaia's fourth data release covers sixty six months instead of thirty four, and the paper expects the gap to widen. And in November twenty twenty seven, the companion should show up on the far side of Betelgeuse.

If it does, the orbit is real. Whether that orbit is what drives the six year cycle is a separate question, and it is the one Gaia just made harder.

RUWE 1.03 12:06

So the episode ends on the number Gaia returned. Not a mass, not a companion, not a verdict. A fit residual. For the nearby stars with the slow cycle, the median is one point zero three. For ordinary pulsating giants, one point one two. A perfect single star scores one.

The stars that were supposed to have partners look more alone than the ones nobody suspected. One point zero three.

Next: a star in another galaxy 12:39

Next time, we leave the Milky Way. The first close up image of a star in another galaxy, two thousand times wider than the Sun, and what the interferometer recorded before it became a picture.

Description and sources

About a third of luminous red giants carry a slow brightness cycle, the long secondary period, on top of their pulsation. The leading explanation is a low-mass companion. It predicts three things: the giant dims once per orbit, its velocity rocks once per orbit, and its position loops on the sky once per orbit. Gaia's velocity curves for 224 of these stars within 1.5 kpc rock at a median of 1.63 km/s, which, read as orbits around a 1-Sun giant, implies companions of about 0.084 solar masses -- squarely in the brown dwarf desert. Shariat, El-Badry, MacLeod and Leiner then simulated what Gaia should have measured for each star and compared it with RUWE, the residual of Gaia's single-star fit. Within 750 pc the prediction exceeds the measurement for 58 of 73 stars, while known ellipsoidal binaries show the predicted signal in 7 of 9. The median observed RUWE is 1.03 for the slow-cycle giants and 1.12 for ordinary pulsating giants. Betelgeuse itself is too bright for Gaia, weighs 16.5 to 19 suns rather than one, and may be the special case; the far side of its companion's orbit is due around November 2027.

PRINT-READY, FROM THIS CHANNEL

The Conversion Chain, Vol. 01 — the method, and four numbers taken apart with it

https://therepository.gumroad.com/l/the-conversion-chain

PRIMARY SOURCES

Shariat, C., El-Badry, K., MacLeod, M. & Leiner, E. (2026). Gaia astrometry disfavors a binary origin for long secondary periods. arXiv:2604.09767 (10 April 2026) -- 224 LSP stars within 1.5 kpc; median K 1.63 km/s; implied M2 0.084 Msun for M1 = 1 Msun; RUWE predicted above observed for 58 of 73 within 750 pc; ellipsoidal control 7 of 9; median RUWE 1.03 vs 1.12; nearest star 391 pc, dust 0.646 mas, jitter 0.07 mas.
Gaia Collaboration, Trabucchi, M. et al. (2023). Gaia Focused Product Release: Radial velocity time series of long-period variables. A&A 680, A36 (10 October 2023) -- 9,614 sources, 6,093 top-quality.
Nicholls, C. P. et al. (2009). MNRAS 399, 2063 -- 58 LSP giants with VLT FLAMES/GIRAFFE.
Soszynski, I. et al. (2021). Binarity as the origin of long secondary periods in red giant stars. ApJL 911, L22.
Grether, D. & Lineweaver, C. H. (2006). How Dry is the Brown Dwarf Desert? ApJ 640, 1051.
Lindegren, L. (2018). Re-normalising the astrometric chi-square in Gaia DR2. GAIA-C3-TN-LU-LL-124 (RUWE).
Panuzzo, P. et al. (2024), A&A; ESO eso2408 -- Gaia BH3, 33 Msun, ~2,000 light years.
Vlemmings, W. et al. (2024), Nature; ESO eso2412 -- R Doradus with ALMA, July-August 2023.
Vega, L. D. et al. (2021). ApJ 909, 138; NASA Goddard -- U Mon, binary inside a dusty disc, ~6.5 years.
MacLeod et al. 2025, ApJ 978, 50 (Betelgeuse K ~ 1.5 +/- 0.34 km/s); Montarges et al. 2026, A&A 711, L12 (16.5-19 Msun); Howell et al. 2025, ApJ 988, L47 (elongation 26 Nov 2027); Harper et al. 2017, AJ 154, 11 (4.51 mas; 'too bright for GAIA').
ESA, Gaia data release scenario -- DR3 (13 June 2022, 34 months); DR4 planned 2 December 2026, 66 months.
Arithmetic shown on screen, recomputed: 58/73 = 79.5%; 7/9 = 78%; 12/224 = 5.4%; 0.2 mas = 9.70e-10 rad, a 24.26 mm coin at 25,020 km; 0.646/0.07 = 9.2; 2,100 d = 5.75 yr.

ARCHIVE PLATES — public-domain documents, with author, date, archive and licence.

Zooming in on Betelgeuse’s companion | ESO | cc-by-4.0

... and 22 more plates, each credited on screen in the video and listed in this project's manifest.

Absolute Magnitude takes one number the internet repeats and asks what an instrument actually returned. Every episode ends on a measured figure with its source on screen, and says plainly which part of it was converted afterwards.

No AI-generated imagery is used on this channel. Every frame of archive footage is real, publicly released material, credited on screen with its licence; every diagram is drawn in code from the numbers in the cited papers.

Corrections are pinned, not quietly edited.

#Betelgeuse #Gaia #Astronomy #ESA #AbsoluteMagnitude