The Hubble Tension

Summary

There are two good ways of working out how fast the universe is expanding. They have been refined for decades by careful people using excellent instruments.

They give different answers, and the answers do not overlap, and the gap between them is not closing.

It is getting worse. As both measurements have improved and their uncertainties have shrunk, the disagreement has become more significant, not less. That is the opposite of what a mistake does.

This is, on the registry's assessment, the most serious genuinely unsolved anomaly in the entire archive, and unlike almost everything else in it, the people working on it think it might mean the standard model of cosmology is wrong.

What is documented

The quantity. The Hubble constant, written H-nought, is the present-day rate of expansion of the universe. It is measured in kilometres per second per megaparsec.

The first method: the distance ladder. Measure the distances to things directly, and their recession speeds, and divide.

The ladder has rungs. Nearby stars give you geometric parallax. That calibrates Cepheid variable stars, whose brightness relates to their pulsation period. Cepheids in nearby galaxies calibrate Type Ia supernovae, which are bright enough to be seen very far away. Supernovae give you the far distances.

The SH0ES team, led by Adam Riess, obtains a value of around 73 kilometres per second per megaparsec.

The second method: the cosmic microwave background. Measure the afterglow of the Big Bang in exquisite detail, fit the standard cosmological model to it, and extrapolate that model forward thirteen point eight billion years to today.

The Planck satellite gives a value of around 67.4.

The gap. Roughly eight to nine per cent. In statistical terms, the disagreement now stands at around five standard deviations, which is the conventional threshold at which physicists declare a discovery.

It is not converging. This is the important part. Over two decades, both methods have improved enormously, and their error bars have shrunk, and the central values have not moved toward each other.

A systematic error usually reveals itself as the measurement gets better. This one has not.

JWST. The obvious worry was that the Cepheid measurements were contaminated: that the Hubble Space Telescope, at its resolution, was blending Cepheids with neighbouring stars and getting them wrong.

The James Webb Space Telescope has far better resolution. It was used to re-examine the same Cepheids, and it confirmed the earlier photometry. The crowding explanation does not appear to work.

And there is disagreement within the late-universe camp too. Wendy Freedman and colleagues, using the tip of the red giant branch as a distance indicator rather than Cepheids, obtain values closer to the CMB figure.

The registry records this honestly. The tension is real, and the late-universe measurements do not all agree with each other either.

Leading explanations

A systematic error somewhere. The default, and it has been the default for twenty years, and nobody has found it. JWST has removed the most obvious candidate.

New physics, and this is where it gets interesting.

Here is the crucial thing that is almost always stated wrongly, including in serious coverage.

The CMB value is not a measurement of the expansion rate. Planck did not measure how fast the universe is expanding today. Planck measured the universe as it was three hundred and eighty thousand years after the Big Bang, and then the standard cosmological model was used to predict what the expansion rate should be now.

It is not an observation. It is a prediction, conditional on the model being right.

So the tension may not be a disagreement between two measurements at all. It may be a disagreement between a measurement and a model.

If it is, then the distance ladder is telling the truth about the expansion, and the model is failing to get us from the early universe to here, and something is missing from our account of the intervening thirteen billion years.

Candidates for the missing thing. Early dark energy: an additional component acting briefly in the early universe. A dark energy that changes with time rather than being a constant. Additional relativistic species. Modifications to gravity.

The 2024 and 2025 results from the Dark Energy Spectroscopic Instrument, which have been read as hinting that dark energy may be evolving rather than constant, bear directly on this, and the situation is moving.

Current status

Unsolved, and the registry wants to say why this record matters to the archive as a whole.

Almost every case in this collection is a mystery that dissolved when somebody checked, or that never existed, or that was manufactured. This one is different, and it is worth having, because it demonstrates what a real anomaly looks like from the inside.

It is boring. It is a number. It is being pursued by hundreds of people with expensive instruments who very badly want it to go away and cannot make it. It gets harder as the data gets better, and it has been sitting there for twenty years being wrong.

That is what an actual unexplained thing looks like, and it looks nothing at all like a mystery.

Last reviewed: July 2026. Records are provisional. Where the evidence changes, the entry changes. Found an error? Tell us.