Why Can't Astronomers Agree on How Fast the Universe Is Expanding? The Hubble Tension Explained

Space & Astronomy · Published 2026-10-23 · Sources checked 2026-10-22

Short answer: because two different ways of measuring the universe's expansion rate, called the Hubble constant, keep giving different answers. Working forward from the early universe, the Planck satellite's map of the cosmic microwave background implies about 67.4 kilometres per second per megaparsec. Measuring nearby galaxies directly, the SH0ES team using the Hubble Space Telescope found about 73.0. The gap is larger than either team's stated uncertainty. The James Webb Space Telescope has since tested one suspected error in the nearby measurements and ruled it out, while another JWST-based team finds a value closer to 70 that does not require new physics. The disagreement, known as the Hubble tension, is not settled.

This page explains what the Hubble constant is, how each method works, what the latest measurements report, and what the possible explanations are. All values come from the papers listed at the end.

What the Hubble constant measures

The universe is expanding, so distant galaxies move away from us faster than nearby ones. The Hubble constant, written H₀, says how much faster: for every megaparsec of distance, how many kilometres per second a galaxy's recession speed increases. A megaparsec is about 3.26 million light years (our conversion). A value of 70 km/s/Mpc means a galaxy 10 megaparsecs away recedes at about 700 km/s because of expansion.

The number matters beyond astronomy trivia. It sets the scale of the universe and, through the standard model of cosmology, ties into its age and contents. If two careful measurements truly disagree, either one of them contains an unrecognised error, or the standard model is missing something.

Method 1: predict it from the early universe

The Planck satellite mapped the cosmic microwave background, the faint glow left from when the universe was young. The final Planck cosmology paper, published in Astronomy & Astrophysics, reports that the data fit the standard "base ΛCDM" model, a six-parameter model of a flat universe with dark energy and cold dark matter, and that the angular scale of the patterns in the map is measured to 0.03% precision.

The Hubble constant is not read directly off the map. It is inferred by assuming that model and calculating forward to today. On that basis Planck reports H₀ = 67.4 ± 0.5 km/s/Mpc. The paper also states that it finds "no compelling evidence for extensions to the base-ΛCDM model".

Method 2: measure it from nearby galaxies

The other approach builds a "distance ladder". Astronomers first measure distances to nearby objects geometrically, then use those to calibrate stars of known brightness, then use the stars to calibrate type Ia supernovae, which are bright enough to see far away. Comparing supernova distances with how fast their galaxies recede gives H₀.

The SH0ES team, led by Adam Riess, reported its main result in Astrophysical Journal Letters in 2022. Its main ingredients:

• Cepheid variable stars (stars whose pulsation period reveals their true brightness) observed with Hubble in the host galaxies of 42 type Ia supernovae, using more than 1,000 Hubble orbits.
• Three geometric anchors for the Cepheids: parallaxes from the Gaia mission (Early Data Release 3), water masers in the galaxy NGC 4258, and detached eclipsing binary stars in the Large Magellanic Cloud.
• About 70 variations of the analysis, testing choices such as dust, metallicity and which supernova surveys to include.

The baseline result is H₀ = 73.04 ± 1.04 km/s/Mpc, including systematic uncertainties. The paper reports a 5-sigma difference from the value predicted by Planck with the standard model. That is about 8% higher than Planck's 67.4 (our calculation).

Sources: Planck Collaboration, A&A 641, A6 (2020); Riess et al., ApJL 934, L7 (2022); Freedman et al., arXiv:2408.06153 (2024), published as ApJ 985, 203 (2025). CCHP values are from the preprint abstract; a correction to the journal version was published in November 2025, so check the journal version for final figures.

What the James Webb Space Telescope checked

One long-standing suspicion was crowding. In distant galaxies, Cepheids sit among many other stars, and Hubble's resolution may blend their light with neighbours, making them look brighter and nearer than they are. If that effect grew with distance, it could push H₀ upward.

JWST has sharper vision in infrared light, so it can separate those stars. In a 2024 paper in Astrophysical Journal Letters, Riess and colleagues reported JWST observations of more than 1,000 Cepheids in NGC 4258 and in five galaxies that hosted eight type Ia supernovae. They found no significant difference between distances measured with JWST and with Hubble: a formal difference of −0.01 ± 0.03 magnitudes. They rejected unrecognised crowding as the cause of the tension at 8.2 sigma, which they note is greater confidence than that of the tension itself.

A second JWST team gets a lower number

The Chicago-Carnegie Hubble Program (CCHP), led by Wendy Freedman, uses JWST to calibrate the distance ladder with three independent kinds of stars: the tip of the red giant branch (TRGB), J-region asymptotic giant branch (JAGB) stars, and Cepheids. Its 2024 status report covered ten nearby galaxies hosting eleven type Ia supernovae, plus NGC 4258 as the geometric anchor.

Its highest-precision estimate, from the TRGB method with 24 supernova calibrators from Hubble and JWST, is H₀ = 70.39, with uncertainties listed in the table above. Using JWST data alone, the TRGB and JAGB methods give 68.81 and 67.80. The team reports that TRGB and JAGB distances agree to better than 1% on average and agree with SH0ES Cepheid distances "at just over the 1% level", and concludes that its results are consistent with the standard ΛCDM model "without the need for the inclusion of additional new physics".

So the nearby-universe camp is not unanimous. Both teams use JWST and both find their own measurements robust, yet their headline values differ.

What could explain the gap

The papers point to two broad possibilities.

1. A measurement problem in one route. The SH0ES papers argue that the main known error sources on the distance-ladder side, including crowding, have been tested and do not close the gap. The CCHP results suggest that with different calibrating stars the nearby value moves closer to Planck's.
2. Something missing from the standard model. Planck's 67.4 depends on assuming base ΛCDM. If the early-universe physics differs from that model, the predicted value would change. The Planck paper itself, however, found no compelling evidence for extensions of the model in its own data.

The CCHP report ends by saying future JWST data will be needed to increase the precision and accuracy of the local distance scale. That is the realistic state of play: more data from more independent methods.

Common questions

When might the question be settled?

No date can be given. The CCHP team writes that future JWST data will be required to increase the precision and accuracy of the local distance scale, and independent methods on both sides will have to converge before the tension can be called resolved.

Why are the units so strange?

Kilometres per second per megaparsec means the speed added for each megaparsec of distance. One megaparsec is about 3.26 million light years (our conversion).

How big is the disagreement?

SH0ES's 73.04 is about 8% higher than Planck's 67.4 (our calculation), and SH0ES reports the difference as 5 sigma, meaning it is very unlikely to be a statistical fluke if both uncertainties are correct.

Did JWST solve the Hubble tension?

Not yet. JWST ruled out Cepheid crowding as the cause, according to the SH0ES team, while the CCHP team's JWST measurements give lower values consistent with the standard model. The two results point in different directions.

What does "sigma" mean here?

It is a measure of statistical confidence. Larger sigma values mean a difference is less likely to be due to chance, assuming the stated uncertainties capture all the errors.

Sources and how to check

• Planck Collaboration. "Planck 2018 results. VI. Cosmological parameters." Astronomy & Astrophysics 641, A6 (2020). DOI: 10.1051/0004-6361/201833910 · arXiv: 1807.06209
• Riess A. G. et al. "A Comprehensive Measurement of the Local Value of the Hubble Constant with 1 km/s/Mpc Uncertainty from the Hubble Space Telescope and the SH0ES Team." Astrophysical Journal Letters 934, L7 (2022). DOI: 10.3847/2041-8213/ac5c5b · arXiv: 2112.04510
• Riess A. G. et al. "JWST Observations Reject Unrecognized Crowding of Cepheid Photometry as an Explanation for the Hubble Tension at 8σ Confidence." Astrophysical Journal Letters 962, L17 (2024). DOI: 10.3847/2041-8213/ad1ddd · arXiv: 2401.04773
• Freedman W. L. et al. "Status Report on the Chicago-Carnegie Hubble Program (CCHP): Measurement of the Hubble Constant Using the Hubble and James Webb Space Telescopes." arXiv: 2408.06153 (2024); Astrophysical Journal 985, 203 (2025), DOI: 10.3847/1538-4357/adce78; erratum ApJ 993, 252 (2025), DOI: 10.3847/1538-4357/ae146d
• Abstracts checked on arXiv and journal details on Crossref on 22 October 2026. The megaparsec conversion and the 8% difference are our calculations.