WOH G64's First Close-Up Doesn't Show the Star

Chapters

1. 0:00 A star in another galaxy, up close
2. 0:38 The Large Magellanic Cloud, 50 kpc
3. 2:13 0.32 milliarcseconds wide
4. 3:07 The interferometer
5. 3:42 What it actually returns
6. 4:33 Two nights in December 2020
7. 5:34 Before: two telescopes and a model
8. 6:08 From numbers to a picture
9. 7:06 The star does not appear
10. 7:45 New hot dust
11. 8:45 Where 2,000 Suns comes from
12. 10:12 The 2014 change
13. 11:04 Still a red supergiant, for now
14. 11:59 The size that was measured

Transcript

A star in another galaxy, up close 0:00

This is the first close up image of a star in another galaxy. The star is said to be two thousand times wider than the Sun.

But the star is not in the picture. Not as a disc, and not even as a point. The paper behind the image says so in its own results section. What you are looking at is something around the star.

The instrument that made it never recorded a picture at all. It recorded pairs of numbers. This episode follows those numbers to the image, and ends on the one size in this story that was actually measured.

The Large Magellanic Cloud, 50 kpc 0:38

The star is called WOH G64. It sits in the Large Magellanic Cloud, a small galaxy that orbits our own, visible from the southern hemisphere as a faint smudge beside the Milky Way.

The distance is one of the best known in astronomy: about fifty thousand parsecs, or one hundred sixty thousand light years. It was measured to two per cent, using pairs of giant stars that eclipse each other.

That is the reason to look so far away. Red supergiants in our own galaxy have uncertain distances. Betelgeuse's runs from one hundred thirty to two hundred twenty parsecs, depending on the paper. For WOH G64, the distance is the solid part.

That is roughly three hundred times farther than Betelgeuse, the star of the last four episodes. Every number about WOH G64 inherits that two per cent, and nothing else in this story is known nearly as well.

WOH G64 has long been called the most extreme red supergiant in that galaxy. It is buried in its own dust, it sheds mass at one of the highest rates ever measured for a red supergiant, and its light used to swing on a cycle of about eight hundred fifty days.

Its wind carries away about one ten thousandth of a solar mass every year. At that rate, a whole Sun's worth of gas leaves the star in ten thousand years.

0.32 milliarcseconds wide 2:13

Here is the problem of scale. Take the radius the image paper adopts, one thousand seven hundred and thirty Suns, and put it fifty thousand parsecs away. The star would be zero point three two milliarcseconds wide.

Betelgeuse, much closer, is about forty four milliarcseconds across. WOH G64 is more than a hundred times smaller on the sky.

Zero point three two milliarcseconds is the size of a beach ball, sixty centimetres wide, sitting on the surface of the Moon, seen from Earth.

One of the largest telescopes on Earth, eight metres across, blurs anything smaller than about sixty seven milliarcseconds in this light. To that mirror, the star is two hundred times too small to have any shape at all.

The interferometer 3:07

So the light was collected at the Very Large Telescope Interferometer in Chile, which combines telescopes that stand far apart.

Not the four big ones. For WOH G64, the team used four of the small auxiliary telescopes, each one point eight metres across, moved onto stations up to one hundred twenty nine metres apart.

Their light travels through tunnels into an instrument called GRAVITY, which combines all four beams at once, in near infrared light around two point two microns.

What it actually returns 3:42

And here is what it actually returns. Four telescopes make six pairs. For each pair, one number: the visibility, how strongly the light from the two telescopes interferes. For each triangle of telescopes, a closure phase.

A visibility of one means the object looks like a point to that pair. Lower means it is resolved. That is all a pair of telescopes knows: one contrast, along one direction, at one separation.

The atmosphere scrambles the timing of the light at every telescope. But add the phases around a closed triangle, and the scrambling cancels. What is left belongs to the object. That is why the closure phase survives the sky.

Two nights in December 2020 4:33

The observations took two nights, the fifteenth and twenty sixth of December, twenty twenty. The first night used the stations far apart, the second close together, to fill in more directions.

Eleven sets of exposures, each built from frames of thirty seconds. Then nearly four years of calibration and reconstruction before the image was released, in November twenty twenty four.

It was not easy light to work with. At magnitude fifteen in visible light, WOH G64 is right at the limit of the adaptive optics on those small telescopes. The paper says so in its appendix.

To know what a perfect point looks like, the team also observed two ordinary stars, each about zero point one seven milliarcseconds wide. Half the apparent size of WOH G64, and treated by the instrument as points.

Before: two telescopes and a model 5:34

This was not the first time. In two thousand five and two thousand seven, the same team observed it with an older instrument that combined only two telescopes. Two telescopes cannot make an image, so they fitted a model: a thick torus of dust.

That model did real work. Seen nearly pole on, the dust torus explained the infrared light, and it lowered the star's estimated power to about two hundred eighty thousand Suns, a star that began life with roughly twenty five solar masses.

From numbers to a picture 6:08

To turn visibilities and closure phases into a picture, you search for an image whose own visibilities match the measured ones. Many images match. The choice between them is made by a rule you pick, called a regularization.

The team ran two independent programs, IRBis and MiRA, with different rules. Both gave the same compact, elongated shape. That agreement is the evidence that the shape is in the data and not in the software.

The finished image resolves about one milliarcsecond. That is about three times wider than the star the image is named after.

What it shows is a glow about four by three milliarcseconds, tilted at about fifty six degrees. The press release called it an egg shaped cocoon. In stellar radii, it reaches about thirteen in one direction and nine in the other.

The star does not appear 7:06

And then the sentence that matters. The central star does not clearly appear as a point source in our reconstructed image. It may be very faint, or entirely hidden.

There is also a faint ring, under three per cent of the peak brightness. The authors decline to interpret it, because it falls exactly where the reconstruction tends to leave artifacts.

The old torus model made a prediction for this light: the star should be at least four hundred times brighter, patch for patch, than the dust. The new data say the star contributes far less than that.

New hot dust 7:45

Something changed in between. Infrared spectra taken before two thousand three show a cool star. Spectra from two thousand sixteen on show a rising glow of hot dust, and the change likely began after December two thousand nine.

In longer infrared light, nothing changed. A spectrum taken in twenty twenty two matches the ones from before two thousand seven. So the new dust is hot, and it sits close to the star, which is what the compact glow shows.

So the first close up picture of a star in another galaxy is a picture of fresh dust, forming close to the star and hiding it.

Why stretched? The authors offer two readings and choose neither. Gas flowing out along two opposite poles, or the pull of a companion nobody has seen. The glow fits inside the hole of the old torus model, but they say that is not proof.

Where 2,000 Suns comes from 8:45

Then where does two thousand times the Sun come from? Not from this image. Nobody has measured the size of this star directly. Its radius is a conversion.

Measure how much light arrives, correct for the dust, scale by the distance, and you get the total power, the luminosity. Fit the spectrum, and you get a temperature. A star of that power at that temperature needs a certain surface. That surface is the radius.

And the radius has been published four ways. Seventeen hundred thirty Suns from the dust torus model in two thousand eight. Fifteen hundred forty from a fit to the spectrum in two thousand nine. Roughly two thousand in the press release. And about eight hundred in a paper from this year.

Put each one where the Sun is. At eight hundred Suns, the surface would sit beyond the asteroid belt. At seventeen hundred thirty, past the orbit of Jupiter. At two thousand, close to Saturn. Same star, same light.

The difference is mostly temperature. The radius goes with the inverse square of it. Two hundred degrees, from thirty two hundred to thirty four hundred kelvin, moves the radius by thirteen per cent, with the same luminosity.

The 2014 change 10:12

And the temperature may not hold still. In two thousand fourteen, within about a year, WOH G64 brightened in visible light by two magnitudes while its near infrared light held steady. Its regular cycle gave way to irregular flickers.

A team in Athens read the new spectrum as a star more than a thousand degrees hotter, around forty eight hundred kelvin. A yellow hypergiant, like this one in our own galaxy. At that temperature, the same power needs a star about half the size.

That would matter. A red supergiant turning hot before the end is one of the paths to a supernova. The last naked eye supernova, in nineteen eighty seven, came from this same galaxy, from a star that had probably already left the red phase.

Still a red supergiant, for now 11:04

But a paper published this year, by van Loon and Ohnaka, took spectra from November twenty twenty four to December twenty twenty five. Every one shows bands of titanium oxide, the signature of a cool red supergiant.

Their title says it is still a red supergiant, for now. Meanwhile, in twenty twenty five, the star faded by two magnitudes in ten months, its faintest state on record in that band.

The image paper says what would settle the first question. At the longest baselines, the visibilities flatten, which a point source would explain. Confirming that star needs telescopes even farther apart.

So in eleven years the star got hidden behind new dust, may or may not have changed its temperature, and had its size published at anywhere from eight hundred to two thousand Suns.

The size that was measured 11:59

Which leaves the number this episode ends on, the one size in this story that an instrument measured. The glow in the image is about four by three milliarcseconds. The paper gives it with a tilde, and no error bar.

Four by three milliarcseconds of dust, and inside it a star a third of a milliarcsecond wide that the picture does not show. Four by three. Measured.

Next time, a sound. The note a black hole plays in the Perseus cluster, fifty seven octaves below middle C, and what the X ray telescope actually measured before anyone called it music.

Description and sources

WOH G64 is the most extreme red supergiant in the Large Magellanic Cloud, about 50 kiloparsecs away -- a distance known to two per cent from eclipsing binaries. At that distance a star of 1,730 solar radii is 0.32 milliarcseconds wide, a beach ball on the Moon. In December 2020 four 1.8 m Auxiliary Telescopes of the VLTI, up to 129 m apart, fed the GRAVITY instrument at 2.0 to 2.45 microns. What GRAVITY returns is visibilities and closure phases, not an image. Two independent reconstruction programs, IRBis and MiRA, turned them into the same compact, elongated glow, about 4 by 3 milliarcseconds, at a resolution of about 1 milliarcsecond. The paper reads it as hot new dust formed after 2009, and says the central star does not clearly appear as a point source. The '2,000 times the Sun' is a conversion from luminosity and temperature, and has been published as 1,730, 1,540, roughly 2,000 and about 800 solar radii. After a dramatic change in 2014 one team reads the star as a yellow hypergiant; spectra from 2024 and 2025 still show titanium oxide, the mark of a cool red supergiant.

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

• Ohnaka, K., Hofmann, K.-H., Weigelt, G., van Loon, J. Th., Schertl, D. & Goldman, S. R. (2024). Imaging the innermost circumstellar environment of the red supergiant WOH G64 in the Large Magellanic Cloud. A&A 691, L15; arXiv:2412.01921 -- GRAVITY, 15 and 26 Dec 2020, four ATs, max projected baseline 129 m; IRBis and MiRA; resolution ~1 mas; emission ~4 x 3 mas, PA ~56 deg; stellar angular radius 0.16 mas (1,730 Rsun at 50 kpc); 'The central star does not clearly appear as a point source'; hot dust formed between Dec 2009 and 2016.
• ESO press release eso2417 (21 November 2024): 'Astronomers take first close-up picture of a star outside our galaxy' -- 'a size roughly 2000 times that of our Sun'.
• Ohnaka, K. et al. (2008). Spatially resolved dusty torus toward the red supergiant WOH G64 in the Large Magellanic Cloud. A&A 484, 371 -- MIDI 2005/2007; torus model; 1,730 Rsun; ~2.8e5 Lsun; ~3,200 K.
• Levesque, E. M., Massey, P., Plez, B. & Olsen, K. A. G. (2009). The physical properties of the red supergiant WOH G64: the largest star known? AJ 137, 4744 -- 3,400 +/- 25 K; log L/Lsun = 5.45 +/- 0.05; 1,540 Rsun.
• Munoz-Sanchez, G. et al. (2026). The dramatic transition of the extreme red supergiant WOH G64 to a yellow hypergiant. Nature Astronomy; arXiv:2411.19329 -- 2014 transition; Teff 4,800 +/- 300 K; ~800 Rsun, 'indicative rather than definitive'; 2 mag fading over 10 months in 2025.
• van Loon, J. Th. & Ohnaka, K. (2026). A phoenix rises from the ashes: WOH G64 is still a red supergiant, for now. MNRAS 546, stag012; arXiv:2601.02057 -- SALT spectra Nov 2024 to Dec 2025, TiO bands at all times.
• Pietrzynski, G. et al. (2013). An eclipsing-binary distance to the Large Magellanic Cloud accurate to two per cent. Nature 495, 76 -- 49.97 +/- 0.19 (stat) +/- 1.11 (sys) kpc.
• Goldman, S. R. et al. (2017). MNRAS 465, 403 -- mass-loss rate of order 1e-4 Msun/yr; OH maser at 23.8 km/s.
• GRAVITY Collaboration (2017). First light for GRAVITY. A&A 602, A94.
• Arithmetic shown on screen, recomputed: 1,730 x 6.957e5 km / 1.5428e18 km = 7.80e-10 rad = 0.161 mas radius; 1 mas at the Moon = 1.86 m, 0.32 mas = 0.60 m; 1.22 x 2.2 um / 8.2 m = 67 mas; R = sqrt(L)/(T/5772)^2 gives 1,722 (2.8e5, 3,200 K), 1,530 (10^5.45, 3,400 K), 768 (10^5.45, 4,800 K); (3400/3200)^2 = 1.13; 800/1,730/2,000 Rsun = 3.7/8.0/9.3 AU.

ARCHIVE PLATES: all 16 are 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.

#WOHG64 #ESO #Astronomy #LargeMagellanicCloud #AbsoluteMagnitude