How fast is the universe expanding?
It sounds like a question astronomers should have settled long ago. Instead, it has become one of the biggest puzzles in modern physics.
Different ways of measuring the expansion rate give different answers, and the gap has refused to go away as the measurements have improved.
Now a team of astronomers has produced a new estimate from an unusual source: a single exploding star whose light reached Earth along several different paths.
Their result: about 65 kilometers per second per megaparsec, with an uncertainty of about 3%. They say it’s the most precise value yet obtained from a single gravitational lens.
Two families of measurements disagree about how fast space expands
The expansion rate of the universe today is called the Hubble constant, written H₀.
It’s measured in an unusual unit: kilometers per second per megaparsec. A megaparsec is about 3.26 million light-years. A value of 65 means that a galaxy one megaparsec away is moving away from us at about 65 km per second, and one twice as far at about 130 km per second, because space itself is stretching.
There are two main ways to measure it:
- From the early universe. Measurements of the faint glow left over from the Big Bang, combined with the standard model of cosmology, predict a value of around 67.
- From the nearby universe. Measuring distances to relatively nearby galaxies, using pulsating stars and a type of supernova as “standard candles,” gives values around 73.
That difference of about 9% may sound small. But both kinds of measurement have become precise enough that the gap can’t easily be explained as random error.
This disagreement is called the Hubble tension. Either something is wrong with one or both measurement methods, or something is missing from our understanding of the universe.
A galaxy cluster bent one supernova’s light into several images
The new measurement relies on a third, completely independent method, based on one of Einstein’s predictions: gravity bends light.
A massive object, such as a cluster of hundreds of galaxies, curves the space around it. Light from a more distant object passing nearby is bent, much as a glass lens bends light. This is called gravitational lensing.
When the alignment is right, light from a single distant object can reach us along several different paths around the cluster. We then see multiple images of the same object.
That’s what happened with SN Refsdal, a supernova first spotted in 2014 behind the galaxy cluster MACS J1149.5+2223. Astronomers saw four images of the exploding star arranged in a cross. Models of the cluster predicted that another image would appear later, and in 2015 it did.
The delays between the images carry the expansion rate
Here’s the key idea.
The different light paths have different lengths, and they pass through different depths of the cluster’s gravity. So the images don’t arrive at the same time. The same explosion is seen in one image, then, days, months or even a year later, in another.
Those time delays depend on two things:
- how mass is distributed in the lensing cluster, and
- the distances involved, which depend on how fast the universe is expanding.
If astronomers can measure the delays and build an accurate map of the cluster’s mass, they can work out the Hubble constant directly, without relying on the distance ladder used in the nearby-universe method or on the early-universe model.
The idea goes back to 1964, when the Norwegian astronomer Sjur Refsdal proposed it. The supernova is named after him.
A richer mass map gave 65 km/s/Mpc, with about 3% uncertainty
The hardest part is the mass map. The cluster contains galaxies, gas and dark matter, and only some of it can be seen.
The research team started from an earlier mass model of the cluster that used the positions of lensed images. Then they added much more information: the detailed shape and brightness of the supernova’s host galaxy, as it appears stretched and distorted by the lens.
They incorporated about 77 000 pixels of Hubble Space Telescope imaging into the model.
That extra detail halved the uncertainty in the Hubble constant compared with the earlier approach.
Their result:
- H₀ = 65.0 (+2.0 / −1.8) km/s/Mpc, an uncertainty of about 3%
They tested four different versions of the model to check how sensitive the result was to choices in how the data were weighted and how the host galaxy was reconstructed.
They also report that the uncertainties from the mass model and from the measured time delays are now roughly equal. Further improvement will require both.
An independent method matters more than the number
The headline number is interesting. But the method is more important than the number.
The Hubble tension is a disagreement between two families of measurements, each with its own assumptions and possible blind spots. Adding more measurements of the same kind can make a result more precise without making it more trustworthy.
A completely different technique, with different assumptions and different potential errors, is a better test.
If lensing measurements consistently agree with the early-universe value, that would suggest the problem may lie in the nearby-universe measurements. If they agree with the nearby value, the tension would point more strongly toward missing physics.
This result, at about 65, sits close to the early-universe value and well below the nearby-universe value. But one lens is not enough to decide.
The limits of the result
This is a preprint that has not yet been peer reviewed.
The value depends heavily on the mass model of the galaxy cluster. The team deliberately kept the overall form of that model the same as in earlier work, to make a fair comparison. Different modeling choices could shift the answer, and four model variations don’t cover every possibility.
The calculation also assumes the standard model of cosmology, in which space is flat and dominated by dark energy and dark matter.
And it is a single object. Galaxy clusters are complicated, and any single lens may carry quirks that aren’t captured in the model. Confidence will come only from many lensed supernovae and quasars, analyzed by different teams.
The universe has provided a third ruler
The question of how fast the universe expands is not only about one number. It’s a test of whether our current picture of the cosmos is complete.
This study shows that a single exploding star, seen several times over through a natural lens, can contribute a meaningful measurement to that test.
It doesn’t resolve the Hubble tension. What it does is add an independent witness.
When two rulers disagree, the answer is a third ruler, and the universe has provided one.