The Repository Absolute Magnitude Space

Nobody Has Ever Measured the Distance to a Galaxy

August 30, 2026 · Runtime 12:48 · Watch on YouTube

Every distance beyond about three thousand parsecs was calculated, not measured. There is exactly one rung of the ladder that is pure geometry, and it is very short.

Chapters

Transcript

Nobody has measured it 0:00

Nobody has ever measured the distance to a galaxy. Not one. Not Andromeda, not the galaxy in the Hubble Deep Field, not the one in the press release you read last week. Every distance in that catalogue was calculated, and every calculation started somewhere else, on a number that came from somewhere else again. There is exactly one rung in astronomy where somebody measured a distance directly. It is the bottom one, and it is very short.

An error at the bottom does not stay at the bottom 0:30

This episode ends on a number, and by then you will know which rung of the ladder it sits on and what is stacked on top of it. The ladder is real, it is published, and nothing about it is hidden. But it has a property that headlines never mention. Each rung is calibrated on the rung below, so an error at the bottom does not stay at the bottom. It rides all the way up.

The only honest rung: a triangle 0:56

Start with the only honest one. Hold up a finger and look at it with one eye, then the other. It jumps against the far wall. The jump is an angle, your two eyes are a known baseline, and from those two things alone you get a distance. No physics, no assumption about what your finger is made of. Just a triangle.

The baseline is the Earth's orbit 1:19

Astronomy does the same thing with a bigger head. Observe a star in January, observe it again in July, and the baseline between the two viewpoints is the diameter of the Earth's orbit. Two astronomical units. Two hundred and ninety nine million kilometres. That is the widest ruler anyone has ever had, and it is still not wide enough for the angles to be comfortable.

Bessel, 1838, 61 Cygni 1:45

It took until eighteen thirty eight for anyone to succeed. Friedrich Bessel spent a year measuring a faint double star in Cygnus and found that it moved back and forth by about three tenths of one second of arc. That is the width of a coin seen from thirteen kilometres away. It was the first time a human being knew the distance to anything outside the solar system.

Two centuries of not finding it 2:10

And the failure before that is the better story. For two centuries the absence of parallax was used as evidence against the Earth moving at all. If we really swung around the Sun, the stars should shift, and they did not. The stars were not shifting because the stars are absurdly far away, and nobody was willing to believe how far until the angle finally showed up.

No star has a parallax of one arcsecond 2:35

That angle is so central that the unit of distance is defined by it. A star whose parallax is one second of arc is one parsec away, which is three point two six light years. And here is the thing nobody says out loud: no star is that close. Not one. The nearest, Proxima Centauri, sits at zero point seven six eight seconds of arc. The record holder for the biggest angle in the sky is already under one.

The ruler had to go to orbit 3:06

Angles that small do not survive the atmosphere, so the ruler had to go to orbit. Hipparcos flew in nineteen eighty nine and measured about a hundred and eighteen thousand stars to roughly one thousandth of a second of arc. That sounds like a triumph, and it was. It also means ten per cent accuracy ran out after a few hundred light years, which is the neighbourhood, not the galaxy.

Gaia, and millionths of a second of arc 3:31

Then Gaia. Nearly two billion sources, with parallaxes for stars fainter than Hipparcos could see at all, and precision measured in millionths of a second of arc. One millionth of a second of arc is a human hair seen from fourteen thousand kilometres away, which is roughly the diameter of the Earth. This is the best geometry our species has ever done.

A tenth of one galaxy 3:55

So how far does the best geometry our species has ever done actually get you? Ten per cent accuracy takes Gaia out to a few thousand parsecs. Call it ten thousand light years, generously. The Milky Way is about a hundred thousand light years across. The bottom rung, the only rung that is pure measurement, covers a tenth of one galaxy, and the observable universe holds hundreds of billions more.

Where geometry stops 4:24

That is where geometry stops. Everything beyond it, every distance to every galaxy you have ever seen a number for, is not surveyed. It is inferred. The method has a name that sounds humble and is not: standard candles. You find an object whose true brightness you believe you know, you measure how bright it looks, and the difference is the distance.

How do you know a true brightness? 4:49

Which raises the obvious problem. How do you know the true brightness of anything without already knowing how far away it is? That is not a rhetorical question. It is the central difficulty of observational astronomy, and it was solved in nineteen twelve by someone who was not allowed to operate a telescope.

Henrietta Leavitt, Circular 173 5:10

Henrietta Leavitt worked at Harvard measuring stars on photographic plates. In Harvard College Observatory Circular one seventy three, she published the periods of twenty five variable stars in the Small Magellanic Cloud, and noted a simple relation: the brighter ones took longer to pulse. One sentence, and the whole ladder above the bottom rung rests on it.

Why the Magellanic Cloud mattered 5:34

Look at why the Magellanic Cloud mattered. The stars in it are all at essentially the same distance from us, so distance drops out of the comparison entirely. Whatever is making one star look brighter than another in that cloud is real, not an artefact of one being closer. She could not have told you how far the cloud was. She did not need to.

A Cepheid is not flickering, it is breathing 5:58

The stars she was measuring are Cepheids, and they are not flickering. They are breathing. The star's outer layer traps radiation, heats, expands, becomes transparent, releases the energy, cools and falls back in. A physical cycle, days to months long, and the bigger and brighter the star, the longer it takes. That is the whole relation, and it has a mechanism underneath it.

The law only ranks them 6:26

But Leavitt's law only ranks them. It says this Cepheid is four times as luminous as that one. It does not say how luminous either one is. To convert the ranking into distances, you need the true brightness of at least one Cepheid, which means you need one close enough to have a parallax. The second rung has to reach down and hold hands with the first.

The handshake happens through dust 6:51

And the handshake happens through dust. The Milky Way is full of it, and it makes stars look fainter and redder than they are. Fainter reads as further away. So every Cepheid calibration carries a correction for something you cannot see directly, only infer from the colour. Get the dust wrong and the star moves, without anybody touching the telescope.

One star at a time, in another galaxy 7:16

With that done, the second rung is astonishing. Hubble and now Webb can pick out individual Cepheids in galaxies tens of millions of parsecs away, one star at a time, in another galaxy. That is a factor of ten thousand further than the parallax rung it stands on. And it is still not far enough to measure the expansion of the universe.

The third rung is a bomb 7:40

For that you need the third rung, and the third rung is a bomb. A type one a supernova is a white dwarf that pulls matter off a companion until it detonates. Because the trigger is roughly the same mass every time, the explosion is roughly the same brightness every time. Roughly is doing a lot of work in that sentence.

The Phillips relation, 1993 8:02

Raw, they scatter by about eight tenths of a magnitude, which is a factor of two in brightness and useless for measuring anything. In nineteen ninety three Mark Phillips showed that the ones that fade slower are brighter, in a predictable way. Correct for that and the scatter drops to about fifteen hundredths of a magnitude. That is roughly seven per cent in distance, per supernova.

A few dozen calibrating galaxies 8:27

Now the awkward part. To use a supernova as a ruler you need to know the true brightness of at least one, which means finding one in a galaxy close enough that Hubble can also resolve Cepheids in it. Those galaxies are rare. The entire third rung of the ladder is calibrated on a few dozen of them.

Three anchors at the bottom 8:48

And the whole structure is nailed down at the bottom by three anchors. Parallaxes of Cepheids in our own galaxy. Twenty eclipsing binary stars in the Large Magellanic Cloud, which give forty nine thousand five hundred and ninety parsecs to one per cent. And water masers orbiting the black hole in N G C four two five eight, which give seven point five seven six megaparsecs geometrically.

73.0 against 67.4 9:16

Run the ladder and you get the expansion rate of the universe: seventy three point zero, plus or minus one, kilometres per second per megaparsec. Measure the same quantity from the early universe, from the cosmic microwave background, and you get sixty seven point four, plus or minus zero point five. Those two do not overlap. They miss each other by about five sigma.

And a third number, with Webb 9:43

And it is not a clean fight between two camps. A third group, led by Wendy Freedman, rebuilt the second rung out of a different kind of star with Webb and got sixty nine point nine six, plus or minus one point five three. That sits between the other two and overlaps both. The disagreement is not settled, and pretending otherwise would be the same sin as the headline.

The spacecraft measured its own error 10:09

Which sends everyone back down to the bottom rung, and to something Gaia did that I find genuinely beautiful. The team pointed it at quasars. Quasars are so far away that their parallax is zero to any instrument we can build, so the catalogue should have returned zero. It did not. The median came back at minus seventeen millionths of a second of arc.

Minus 17, and it is not one number 10:34

That is the spacecraft measuring its own error, and it is subtracted from every parallax in the catalogue. It is not one number. It varies by about ten millionths of a second of arc with brightness, colour and where in the sky you look. And it is least well determined for bright stars, which is exactly what the Milky Way Cepheids are.

None of this is hidden 10:57

None of this is hidden and none of it is sloppy. The correction is published, the recipe is published, and the European Space Agency publishes the code that applies it. The teams on both sides of the disagreement know all of it and argue about it in public, in detail, with each other. That is what a healthy measurement looks like from the inside.

Ask which rung it came from 11:20

Here is the practical version, and it works on any distance you will ever be shown. Ask which rung it came from. Under one thousand parsecs, you are being told a geometric measurement. Between there and about thirty million, you are being told a Cepheid, which is a parallax plus a dust model. Past that, you are being told a supernova, which is all of the above plus a handful of calibrating galaxies.

The measured number 11:48

So here is the measured number this episode ends on, and like last time it is deliberately not the famous one. Minus seventeen millionths of a second of arc. That is what Gaia read when it looked at objects that should have read zero, and it is the correction sitting underneath every rung above it. The five sigma disagreement about how fast the universe expands is standing on that.

Next: 12,500 red giants, ringing 12:13

You just watched the entire distance scale of the universe come down to a triangle with a very short base and a correction of a few millionths of a second of arc. So the obvious next question is whether there is any way to check it that does not use the ladder at all. There is one, and it works by listening. Next time: twelve thousand five hundred red giant stars, ringing like bells, used to audit the number this episode just ended on. Absolute Magnitude Space. Every episode ends in a measured number.

Description and sources

Parallax is the only distance in astronomy that comes out of a triangle: a known baseline, the diameter of the Earth's orbit, and a measured angle. Bessel got the first one in 1838 -- 0.314 arcseconds for 61 Cygni -- and no star in the sky reaches a full arcsecond. Gaia now measures angles in millionths of a second of arc for 1.8 billion sources, and ten per cent accuracy still runs out around three thousand parsecs, which is a tenth of one galaxy. Everything past that is a standard candle: Cepheids calibrated on parallaxes, supernovae calibrated on Cepheids, each rung inheriting the error of the one below it. Run the chain and you get 73.0 +/- 1.0 for the expansion rate of the universe; measure the same quantity from the cosmic microwave background and you get 67.4 +/- 0.5. Those miss each other by about five sigma. So this episode goes back to the bottom rung, where Gaia pointed at quasars -- objects far enough away that their parallax is zero -- and the catalogue returned a median of minus 17 microarcseconds instead. That correction is subtracted from every parallax in the catalogue, it varies with magnitude, colour and position, and it is least well determined for exactly the bright stars the ladder is anchored on. None of it is hidden. It is just never printed next to the headline.

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

Lindegren et al. (2021). Gaia EDR3: parallax bias versus magnitude, colour and position. A&A 649, A4; astrometric solution, A&A 649, A2. arXiv:2012.01742 -- quasars give a median parallax of -17 uas, varying by about 10 uas. ESA publishes the correction code at gitlab.com/icc-ub/public/gaiadr3_zeropoint.
Riess et al. (2022). ApJL 934, L7 -- H0 = 73.0 +/- 1.0 km/s/Mpc from 42 Type Ia supernovae calibrated by Cepheids.
Planck Collaboration (2020). A&A 641, A6 -- H0 = 67.4 +/- 0.5 km/s/Mpc.
Freedman et al. (2025). Chicago-Carnegie Hubble Program with JWST, ApJ -- H0 = 69.96 +/- 1.53 km/s/Mpc.
Pietrzynski et al. (2019). Nature 567, 200 -- LMC at 49,590 +/- 90 (stat) +/- 540 (sys) pc from 20 eclipsing binaries.
Reid, Pesce & Riess (2019). ApJL 886, L27. arXiv:1908.05625 -- NGC 4258 at 7.576 +/- 0.082 (stat) +/- 0.076 (sys) Mpc from water masers.
Leavitt & Pickering (1912). Harvard College Observatory Circular 173 -- periods of 25 variable stars in the Small Magellanic Cloud.
Phillips (1993). ApJL 413, L105 -- the decline-rate relation, scatter from about 0.8 to about 0.15 magnitudes.
Bessel (1838). MNRAS 4, 152 -- 61 Cygni at 0.3136 arcsec, the first stellar distance.
ESA (1997). Hipparcos and Tycho Catalogues, SP-1200 -- 118,218 stars at roughly 1 mas.
Gaia Collaboration (2023). A&A 674, A1 -- Gaia DR3, 1.8 billion sources.
GRAVITY Collaboration (2019). A&A 625, L10 -- Galactic Centre at 8,178 pc.

ARCHIVE PLATES: all 11 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 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.

#DistanceLadder #HubbleTension #Gaia #Astronomy #Space