The Repository Absolute Magnitude Space

The Star That Shrank by a Third (and nobody took a new photo)

August 30, 2026 · Runtime 13:12 · Watch on YouTube

UY Scuti is the largest star you have ever been told about: 1,708 times the radius of the Sun. That number was never measured. What was measured is an angle.

Chapters

Transcript

Put a marble on the table 0:00

Put a marble on the table. Sixteen millimetres across. That is the Earth. At that scale, the Moon is a peppercorn, about four millimetres, and you set it down half a metre away. The Sun is not on the table. The Sun is a sphere one point seven five metres across, and to put it in the right place you have to carry it a hundred and eighty eight metres down the street. That is the solar system, to scale, and that is the easy part.

A measurement multiplied by a guess 0:31

This episode ends on a number, and you are going to know exactly how much of that number we measured and how much of it we assumed. Because the most famous size in astronomy, the largest known star, is a measurement multiplied by a guess. And when the guess got better, the star lost a third of itself. Nobody took a new photograph. Nobody found an error. The star did not change at all.

Jupiter, eleven Earths across 0:59

Back to the marble. Jupiter is eleven Earths across. On our table it is a ball a hundred and seventy six millimetres wide, about the size of a honeydew melon. Its real equatorial radius is sixty nine thousand nine hundred and eleven kilometres, and we know that to the metre, because spacecraft have flown past it and radio signals have been timed going in and out. This is not an estimate. This is surveying.

The Sun, and the end of what we measured directly 1:28

The Sun is a hundred and nine Earths across. The International Astronomical Union fixed its nominal radius by decree in twenty fifteen: six hundred and ninety five thousand seven hundred kilometres, exactly, so that everyone doing stellar physics divides by the same number. So far every step on this ladder was measured directly. Radar off Venus. Spacecraft timing. A committee agreeing on a definition. Then you leave the solar system, and the method changes completely.

You cannot photograph a star 2:03

Here is the thing almost nobody says out loud. You cannot photograph a star. Not the Sun, the Sun is next door. Any other star. Through the largest telescope on Earth, a star is a point. It stays a point at every magnification, because the disc is smaller than the blur the atmosphere and the optics put around it. There is no picture of a star's surface. There never has been, for almost all of them. So where does a stellar size come from?

Interferometry measures an angle, not a size 2:37

It comes from interferometry. You take two telescopes a hundred metres apart, combine the light, and read the interference pattern. The pattern is sensitive to something a single telescope cannot see: how wide the source is, as an ANGLE. Not how big it is. How wide it looks. That distinction is the whole episode, so hold on to it. An angle is what the instrument returns. A size is what we want.

5.48 milliarcseconds 3:09

In twenty thirteen a team led by Belen Arroyo-Torres pointed the Very Large Telescope Interferometer at UY Scuti, a red supergiant in the constellation Scutum. They measured its angular diameter: five point four eight milliarcseconds, plus or minus zero point one zero. A milliarcsecond is a three hundred and sixtieth of a degree, divided by a thousand, divided by another thousand. At that angle, a one euro coin would have to sit eight hundred and seventy five kilometres away. That is a coin in Berlin, read from Paris. One point eight per cent uncertainty. That is a good measurement.

The star has no edge 3:52

And there is a complication in that measurement that makes the rest of this episode almost inevitable. The same paper found that UY Scuti's apparent size depends on which colour you look in. In the water vapour band it measures ten to twenty five per cent wider. At the carbon monoxide bands, twenty to thirty five per cent wider. The star does not have a crisp edge. It has an atmosphere that keeps going, and the models the team compared against could not stretch far enough to match what they saw. So even the angle is the angle of a chosen surface.

The distance was adopted, not measured 4:29

And it is still not a size. An angle only becomes a length when you multiply it by a distance. Hold your thumb up: it covers the Moon and it covers a coin, and the difference between the thumb, the coin and the Moon is entirely how far away they are. So to turn five point four eight milliarcseconds into kilometres, the team needed to know how far away UY Scuti is. They did not measure that. They adopted it.

Where 2,900 parsecs came from 4:59

The paper is completely open about this, and the sentence is right there in section four. For UY Scuti they took the average of a distance published in nineteen ninety eight, and a distance obtained from a rule of thumb from nineteen eighty nine, which estimates how far away a star is from how much interstellar dust has reddened its light. The answer was two thousand nine hundred parsecs, plus or minus three hundred and seventeen. About nine thousand five hundred light years.

1,708 solar radii 5:29

Multiply the angle by that distance and you get the number you have seen a hundred times. One thousand seven hundred and eight solar radii. Plus or minus one hundred and ninety two. That is the number in the thumbnails. That is the number in the size comparison videos. That is the number that made UY Scuti famous as the largest known star. Now look at the two error bars side by side, because they do not match.

The radius error was never the radius 6:00

The angle was good to one point eight per cent. The distance was good to ten point nine per cent. And the radius came out at eleven point two per cent. Read those three numbers again. The uncertainty on the size is not the uncertainty of the measurement. It is the uncertainty of the ASSUMPTION, wearing the measurement's name. Which raises an obvious question. What happens to the famous number if somebody finally measures the distance?

Gaia measures the parallax itself 6:30

Somebody did. The European Space Agency's Gaia spacecraft spent years measuring parallax: the tiny shift in a star's apparent position as the Earth swings from one side of its orbit to the other. It is the only direct way to measure a stellar distance, and it is pure geometry. No models. No dust rules. For UY Scuti, Gaia reports a parallax of zero point five one six six milliarcseconds, plus or minus zero point zero four nine four.

Is the Gaia solution trustworthy here? 7:04

And you should be suspicious of that, because Gaia is not supposed to be good at stars like this one. UY Scuti is enormous, it pulsates, and its surface has bright patches that move. All of that can drag the apparent centre of the star around and corrupt the astrometry. Gaia publishes a number that tells you when that has happened. It is called RUWE, and anything above about one point four means do not trust this solution. For UY Scuti, RUWE is one point zero four. The fit is clean.

1,867 parsecs, and no overlap 7:40

Turning a parallax into a distance is not simply one divided by the angle, and the careful version of that calculation was published by Coryn Bailer-Jones and colleagues in twenty twenty one. For UY Scuti it gives one thousand eight hundred and sixty seven parsecs, with a range from one thousand six hundred and ninety four to two thousand and eighty nine. The old adopted distance, two thousand nine hundred parsecs, is not inside that range. It is not even close to the edge of it.

The same chain, redone 8:11

So let us do the arithmetic again. Same star. Same instrument. The same five point four eight milliarcseconds, untouched. Only the distance is different, and now it is measured instead of assumed. The radius comes out at about one thousand one hundred solar radii. Between roughly one thousand and one thousand two hundred and thirty, if you carry the range through. UY Scuti just lost about a third of its size, and nothing happened to UY Scuti.

Brightness scales with distance squared 8:45

And the size is not the only casualty, because distance does not enter every quantity the same way. A radius scales with distance. Brightness scales with distance SQUARED, because the light spreads out over a sphere. The twenty thirteen paper puts UY Scuti at about three hundred and thirty seven thousand times the luminosity of the Sun. Move it to the Gaia distance and that falls to about a hundred and forty thousand. Not a correction. A factor of two point four. The ten point nine per cent on the distance became almost twenty two per cent on the light, and the star slid across the diagram astronomers use to classify it.

This is not an accusation 9:27

This is not a story about a mistake. The twenty thirteen paper printed the word adopted, printed the source of the distance, and printed the error bar. Everything needed to see this coming was on the page. What went wrong happened afterwards, in the retelling, where the plus or minus gets dropped because it is inconvenient, and a conditional result becomes a fact. The number was never a lie. It was a conclusion with a condition attached, and the condition got deleted.

Still absurd, and now honest 9:59

And do not feel sorry for the star. Eleven hundred solar radii is still five point one astronomical units. If you dropped it where the Sun is, its surface would sit just inside the orbit of Jupiter. Straight across, it is over ten astronomical units wide, which puts the far edge past the orbit of Saturn. Light takes almost an hour and a half to cross it. It shrank by a third and it is still one of the largest objects anyone has ever pointed a telescope at.

Back on the table 10:32

Put it back on our table, where the Earth is a sixteen millimetre marble and the Sun is a one point seven five metre sphere a hundred and eighty eight metres down the street. At that scale the old UY Scuti was a ball nearly three kilometres across. The new one is one point nine kilometres across. Either way it would not fit in the town. The difference between them is about a kilometre of star, and that kilometre was never in the sky. It was in the denominator.

Every superlative is a measurement times an assumption 11:03

And this is the shape of every superlative in astronomy. The most distant galaxy. The most massive black hole. The brightest thing ever recorded. Underneath each one there is something an instrument actually returned, and something a human being had to supply so the instrument's number could be converted into the unit you wanted to hear. The headline always quotes the second one. The confidence always belongs to the first.

The one question you can ask tonight 11:32

There is a practical version of this, and you can use it tonight. When you read that something is the largest, the oldest or the brightest, ask one question: what did the instrument actually return? Almost always the answer is an angle, a brightness, a redshift or a time delay. Those are the honest numbers. Everything after them is a conversion, and every conversion carries an assumption that somebody chose. You are not being asked to distrust the science. You are being asked to read the sentence the headline deleted.

The measured number 12:08

So here is the measured number this episode ends on, and it is deliberately not the size. Five point four eight milliarcseconds, plus or minus zero point one zero. That is the angular diameter of UY Scuti, and it is the only quantity in this entire episode that came out of an instrument pointed at that star. Everything else, including both of the famous radii, is that angle multiplied by something we believed about how far away it is.

Next: where the distance itself comes from 12:39

You just watched a star lose a third of its size because one number in the denominator got better. So the obvious next question is where that number comes from. Next time: the distance ladder. Parallax, then Cepheid variables, then supernovae. Each rung is calibrated on the rung below it, which means every error bar gets multiplied all the way up. Absolute Magnitude Space. Every episode ends in a measured number.

Description and sources

In 2013 a team led by Belen Arroyo-Torres pointed the Very Large Telescope Interferometer at UY Scuti and measured its angular diameter: 5.48 plus or minus 0.10 milliarcseconds, an error of 1.8 per cent. To turn an angle into a size you need a distance, and the distance was not measured. It was adopted -- 2,900 parsecs, the average of a 1998 paper and a 1989 rule of thumb that estimates distance from how much interstellar dust has reddened a star's light. That number carries 10.9 per cent. The published radius carries 11.2 per cent, and that error was never the radius: it is the distance error wearing the radius's name. When Gaia measured the star's own parallax, the distance fell to about 1,867 parsecs -- an interval that does not contain 2,900 at all. Same angle, measured distance, and the largest known star loses a third of itself. The star did not change. Our distance did.

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

Arroyo-Torres, Wittkowski, Marcaide & Hauschildt (2013). The atmospheric structure and fundamental parameters of the red supergiants AH Scorpii, UY Scuti, and KW Sagittarii. A&A 554, A76. arXiv:1305.6179 -- angular diameter 5.48 +/- 0.10 mas, adopted distance 2,900 +/- 317 pc, R = 1,708 +/- 192 Rsun, Teff = 3,365 +/- 134 K (Table 3).
Gaia Collaboration (2022). Gaia Data Release 3. source_id 4152993273702130432 -- parallax 0.5166 +/- 0.0494 mas, RUWE 1.039, G = 6.747.
Bailer-Jones, Rybizki, Fouesneau, Demleitner & Andrae (2021). Estimating distances from parallaxes V. AJ 161, 147. VizieR I/352/gedr3dis -- geometric distance 1,867 pc, 1,694-2,089 pc at the 16th-84th percentiles.
Sylvester, Skinner & Barlow (1998). Silicate and hydrocarbon emission from galactic M supergiants. MNRAS 301, 1083 -- one of the two inputs to the adopted distance.
Jura & Kleinmann (1989). Dust-enshrouded asymptotic giant branch stars in the solar neighborhood. ApJ 341, 359 -- the extinction relation behind the other input.
IAU 2015 Resolution B3 -- nominal solar radius 6.957 x 10^8 m.
IAU 2012 Resolution B2 -- astronomical unit 1.495 978 707 x 10^11 m, exact.
NASA Planetary Fact Sheet -- Jupiter equatorial radius 6.9911 x 10^7 m, semi-major axis 5.204 au; Saturn 9.583 au.
IUGG / GRS 80 -- Earth mean radius 6.371 x 10^6 m.

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

High-Resolution True Color Earth Imagery Simulated by GEOS During the | NASA's Scientific Visualization Studio | public-domain
SPHEREx All Sky Map 2025 | NASA's Scientific Visualization Studio | public-domain
Roman Galactic Plane Survey | NASA's Scientific Visualization Studio | public-domain
Four Days of Solar Dynamics in 16 Minutes | NASA's Scientific Visualization Studio | public-domain
Mapping Stellar ‘Polka Dots’ | NASA's Scientific Visualization Studio | public-domain
Scientists Build a Detailed Image of U Mon Binary | NASA's Scientific Visualization Studio | public-domain
Pandora Spacecraft Animations | NASA's Scientific Visualization Studio | public-domain
NASA's Fleet of Active Satellites (July 2025) | NASA's Scientific Visualization Studio | public-domain

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 was measured and which part was assumed.

Narration is synthesised speech. All footage is NASA public-domain material, credited on screen and listed above; every diagram is drawn in code from the figures in the cited papers, not generated by an image model.

#UYScuti #Astronomy #Space #Gaia #StarSize