The Hubble Tension Isn't in Webb's Distances

Chapters

1. 0:00 Two Hubble constants from one telescope
2. 0:39 What the Hubble constant is
3. 1:14 The Hubble tension
4. 2:14 A correction: 70.4 is not Webb alone
5. 2:40 The ladder and its anchor
6. 3:50 What Webb's camera records
7. 4:25 Three kinds of standard star
8. 5:09 Finding the tip of the red giant branch
9. 6:09 The supernova host galaxies
10. 7:22 Same galaxy, same distance?
11. 8:58 Where the two numbers separate
12. 10:53 What each team says
13. 12:11 The number that was measured

Transcript

Two Hubble constants from one telescope 0:00

This is the James Webb Space Telescope unfolding on its way to orbit. Two teams have since used it to measure how fast the universe is expanding. One published seventy point four. The other published seventy two point six.

Both are the same quantity, the Hubble constant, and both teams pointed Webb at the same kinds of stars in nearby galaxies. The gap between them is the size of an argument that has run for a decade. This episode goes down to what Webb actually recorded in those galaxies, and ends on the one number where the two teams agree.

What the Hubble constant is 0:39

Galaxies are moving apart, and the farther away a galaxy is, the faster it recedes. The Hubble constant is that ratio, in kilometres per second for every megaparsec of distance, a megaparsec being about three and a quarter million light years.

The speed half is the easy half. Light from a receding galaxy is stretched toward the red, and a spectrum measures that stretch directly, to high precision. The hard half is the distance. Nothing in a galaxy's light says how far it travelled.

The Hubble tension 1:14

Here is why the number matters. Light from the early universe, read by the Planck satellite and run through the standard model of cosmology, predicts sixty seven point four, plus or minus half a unit. Measured locally with Hubble, with Cepheid stars and supernovae, the answer was seventy three point two, plus or minus point nine.

In the brightness of a supernova, that gap is about eighteen hundredths of a magnitude. In distance, about nine per cent. Either the local distances carry an error no one has found, or the standard model is missing something.

Webb was the test. The suspected weak point was crowding. In a distant galaxy, a Cepheid blends with the stars around it, and blended light makes a star look brighter, and so closer, than it is. In the near infrared, Webb's images are about two and a half times sharper than Hubble's.

A correction: 70.4 is not Webb alone 2:14

A correction to the last episode, which said both numbers came from the same telescope. That was too tidy. The seventy point four uses twenty four supernovae. Eleven are calibrated with Webb, thirteen with Hubble alone. With Webb by itself, that team gets sixty eight point eight. The seventy two point six is Webb only.

The ladder and its anchor 2:40

Both answers come from a ladder with three rungs. At the bottom, one galaxy whose distance is known from geometry. In the middle, stars of known brightness, seen both in that galaxy and in galaxies that hosted a supernova. At the top, supernovae far enough away that their speed is mostly the expansion.

For Webb, both teams stand on the same bottom rung. NGC forty two fifty eight, a spiral galaxy with water masers orbiting its central black hole. Tracking those masers by radio gives a distance of seven point five seven six megaparsecs, about twenty five million light years, good to one and a half per cent.

The geometry is the reason it can be trusted. Radio telescopes spread across the Earth track the masers' positions, their speeds along the line of sight, and how fast those speeds change. The speeds give the size of the orbits in kilometres, the positions give it as an angle, and the ratio of the two is the distance. No star brightness enters.

What Webb's camera records 3:50

Now the first measured rung. Webb's near infrared camera does not record a star's distance, or even its brightness in any physical unit. Each pixel, thirty one thousandths of an arcsecond wide, records counts per second.

A calibration factor, found by observing standard stars, turns counts into flux. The flux becomes a magnitude on a scale tied to the star Vega. Every distance in this story starts as that: a star's apparent magnitude in one filter, in one image.

Three kinds of standard star 4:25

Three kinds of star serve as the middle rung. Cepheids, young supergiants that pulse, where the period of the pulse gives the brightness. The tip of the red giant branch. And the carbon stars of the asymptotic giant branch, called J A G B, which have nearly the same brightness in the near infrared.

Cepheids carry a catch. A pulse is found by visiting the same galaxy again and again in visible light, and that is still Hubble's job. Webb then measures how bright each known Cepheid is in the near infrared. The red giants and carbon stars need no repeat visits, only one deep image.

Finding the tip of the red giant branch 5:09

The tip of the red giant branch is a ceiling. Old red giants brighten as they age, until helium ignites in the core and they drop. So in a crowd of old stars, the count rises with brightness, then stops at an edge. That edge is the standard candle.

The edge is not a star, it is a feature in a histogram. The Chicago Carnegie team smoothed the count of stars by a tenth of a magnitude, then ran an edge detector over it, the Sobel filter from image processing. In the anchor galaxy, six hundred twenty four giants sat close to the edge. The tip was placed to two hundredths of a magnitude.

And the measurement was blind. A random offset was added to the data, so no one knew what Hubble constant a choice would produce until the analysis was frozen. The carbon stars were measured the same way, from the peak of a smoothed count.

The supernova host galaxies 6:09

Then the same measurement in the galaxies that hosted supernovae. The Chicago Carnegie program observed ten. Messier one hundred one, the Pinwheel, home of supernova two thousand eleven F E. NGC thirteen sixty five in the Fornax cluster. The Antennae, a pair of colliding galaxies. NGC fifty six forty three, with two supernovae.

The supernovae themselves were measured long before Webb launched. Supernova two thousand eleven F E was followed by telescopes on the ground for months, in two thousand eleven. Webb arrived more than a decade later, to measure the ordinary stars of the galaxy it exploded in.

The other team, SH0ES, led by Adam Riess, observed five hosts of eight supernovae, and combined its data with the Chicago Carnegie galaxies to reach sixteen supernovae closer than about twenty five megaparsecs. From Cepheids, seventy three point four. From carbon stars, seventy two point two. From the tip, seventy two point one.

Same galaxy, same distance? 7:22

So here is the test the episode promised. When both teams measure the same galaxy, do they get the same distance?

Distances on this rung are written as a distance modulus. For NGC fifty six forty three, from Webb Cepheids, the Chicago Carnegie team has thirty point five one, and SH0ES has thirty point five two. For the Pinwheel, twenty nine point one four and twenty nine point one two. For the Antennae, thirty one point seven oh and thirty one point six seven.

A hundredth of a magnitude is half a per cent in distance. The Hubble tension is eighteen hundredths.

And Webb against Hubble. For eight galaxies chosen by SH0ES, Webb's Cepheid distances differ from Hubble's by minus two hundredths of a magnitude, plus or minus three. The Chicago Carnegie team ran its own check, red giant tip against the SH0ES Cepheids in seventeen galaxies, and found two and a half hundredths, plus or minus two.

In these galaxies, blending is not hiding a nine per cent error. Measured star by star, the first rung agrees between telescopes, between methods, and between teams, at around one per cent. The Chicago Carnegie team adds a caveat: these galaxies are nearby, where crowding is mildest. So where do two different Hubble constants come from?

Where the two numbers separate 8:58

From which supernovae. Each galaxy calibrates the brightness of the supernova it hosted, and supernovae are not identical. Run the ten Chicago Carnegie galaxies through Hubble's own Cepheid distances, and their supernovae come out four hundredths of a magnitude brighter than the average of all forty two.

On its own, that choice of galaxies predicts seventy one point two, before any Webb measurement enters. The SH0ES paper says the gap comes simply from the choice of hosts and anchor, not from the distances.

The carbon star result goes lower still. That analysis dropped three of the ten galaxies, where the count of stars would not settle on a peak. Those three, SH0ES notes, would have pulled the answer up. And using the peak of the count, not its average, takes off about one more unit.

Then the top rung. The Chicago Carnegie team ties its calibration to supernovae from the Carnegie Supernova Project. The same team showed in two thousand twenty five that switching to the larger Pantheon plus catalogue moves its answer up by about two.

And the anchor. Webb has only NGC forty two fifty eight. Of the four anchors Hubble's ladder can use, that galaxy gives the lowest value, by about seven tenths. The Chicago Carnegie paper says the accuracy of its result is dominated by the distance to that one galaxy.

Add those up. Which galaxies, which method for the carbon stars, which supernova catalogue, which anchor. Each is a choice made after the photons were counted, and each is defensible. Together they span the space between sixty eight and seventy three.

What each team says 10:53

The Chicago Carnegie team reads its result as consistent with the standard model, with no need for new physics, and says a five sigma tension may be premature. SH0ES answers that a smaller sample only inflates the uncertainty, it does not explain the gap.

Both have published since. In early two thousand twenty five, the Chicago Carnegie team found sixty eight point four to sixty nine point six, depending on which set of supernova measurements is used. A month later, a SH0ES paper found that the eleven galaxies not picked for the Chicago Carnegie sample give seventy four point one on their own.

And in September, SH0ES combined Webb Cepheids in nineteen galaxies with the full Hubble sample, for seventy three point four nine, plus or minus point nine three. The disagreement is still about the sample. Not about the telescope.

What would settle it is more of the bottom and middle rungs. More galaxies with both a supernova and Webb photometry, and more anchors measured by geometry. Every new supernova that goes off in a nearby galaxy can become a new calibrator.

The number that was measured 12:11

Which leaves the number this episode ends on. Not seventy point four, and not seventy two point six. Both of those are conversions, built on choices. The one thing measured the same way by both telescopes, in the same galaxies, is the distance itself.

Webb minus Hubble, for the same stars: minus two hundredths of a magnitude, plus or minus three. Under one per cent in distance. The tension is nine. Minus zero point zero two, plus or minus zero point zero three.

Next time, a Hubble constant with no ladder at all. In two thousand seventeen, two neutron stars collided, and the gravitational wave itself gave the distance. Seventy, plus twelve, minus eight.

Description and sources

The Hubble constant needs a distance, and a distance needs a ladder. At the bottom, NGC 4258, whose water masers give 7.576 megaparsecs by geometry alone. In the middle, stars of known brightness -- Cepheids, the tip of the red giant branch, carbon stars -- measured in that galaxy and in galaxies that hosted a Type Ia supernova. What Webb's near-infrared camera records there is counts per second in each pixel, which a calibration turns into an apparent magnitude. Both teams did that. For NGC 5643, from Webb Cepheids, the Chicago Carnegie Hubble Program has a distance modulus of 30.51 and SH0ES has 30.52; for the Pinwheel, 29.14 and 29.12. Webb against Hubble, for eight SH0ES galaxies: -0.02 +/- 0.03 magnitudes. The Hubble tension is 0.18. The two numbers separate later. The 70.4 uses 24 supernovae, 11 calibrated with Webb and 13 with Hubble alone; Webb by itself gives that team 68.8. The CCHP host galaxies predict 71.2 with Hubble's own Cepheids, before any Webb data. The carbon-star analysis dropped three hosts and used the peak of the count. A different supernova catalogue moves the answer by about two. And Webb has one anchor, the lowest of four. Both teams have published since: 68.4 to 69.6, and 73.49 +/- 0.93.

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

• Freedman, W. L. et al. (2025). Status Report on the Chicago-Carnegie Hubble Program (CCHP). ApJ 985, 203; arXiv:2408.06153 -- H0 = 70.39 +/- 1.22 (stat) +/- 1.33 (sys) +/- 0.70 (SN), 24 SNe (11 JWST, 13 HST); JWST only: TRGB 68.81, JAGB 67.80; NIRCam F115W; NGC 4258 anchor; Sobel edge, 624 giants, +/- 0.02 mag; TRGB minus Cepheid 0.025 +/- 0.021 mag in 17 galaxies.
• Riess, A. G. et al. (2024). JWST Validates HST Distance Measurements: Selection of Supernova Subsample Explains Differences in JWST Estimates of Local H0. ApJ 977, 120; arXiv:2408.11770 -- 73.4, 72.2, 72.1, combined 72.6 +/- 2.0; JWST minus HST -0.02 +/- 0.03 mag; Table A2 distance moduli; CCHP hosts predict 71.2.
• Reid, M. J., Pesce, D. W. & Riess, A. G. (2019). An Improved Distance to NGC 4258 and Its Implications for the Hubble Constant. ApJL 886, L27; arXiv:1908.05625 -- 7.576 +/- 0.082 (stat) +/- 0.076 (sys) Mpc.
• Planck Collaboration (2020). Planck 2018 results VI. Cosmological parameters. A&A 641, A6; arXiv:1807.06209 -- H0 = 67.4 +/- 0.5 (base LCDM).
• Hoyt, T. J. et al. (2025). arXiv:2503.11769 -- 68.4 < H0 < 69.6 depending on the supernova magnitudes; Pantheon+ shifts +2.
• Li, S. et al. (2025). arXiv:2504.08921 -- the 11 hosts not selected for the CCHP compilation give 74.1.
• Riess, A. G. et al. (2025). The Perfect Host. arXiv:2509.01667 -- 73.49 +/- 0.93.
• STScI, JWST User Documentation: NIRCam detector overview and absolute flux calibration (0.031 arcsec per pixel; DN/s to MJy/sr via PHOTMJSR).
• Hubble, E. (1929). A Relation Between Distance and Radial Velocity Among Extra-Galactic Nebulae. PNAS 15, 168 -- pinned comment.
• LIGO Scientific Collaboration and Virgo Collaboration et al. (2017). A gravitational-wave standard siren measurement of the Hubble constant. Nature 551, 85; arXiv:1710.05835 -- for the next episode.

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

• Webb Telescope Nominal Deployment Sequence with Graphics | NASA's Scientific Visualization Studio | public-domain

... and 10 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. This episode opens one: the previous episode said both numbers came from the same telescope, and the 70.4 also uses Hubble-only calibrators.

#HubbleTension #JWST #Cosmology #Astronomy #AbsoluteMagnitude