For a few uneasy months, the universe seemed to be slowing down. Late in 2025, a team of astronomers announced that the cosmic acceleration discovered nearly three decades earlier, the finding that won a Nobel Prize and rewrote the story of everything, might have been an illusion all along. The culprit, they argued, was hiding in the exploding stars astronomers had trusted to measure the heavens. If they were right, dark energy, the mysterious push driving galaxies apart, would quietly evaporate from the textbooks.
It did not. A new study led by the University of Southampton has gone back over the same data and found the cosmos behaving exactly as everyone thought. The expansion is still accelerating, dark energy is still there, and the crisis, it turns out, was a misunderstanding rather than a flaw in the universe itself.
The stakes were not small. The original discovery, made in the late 1990s by Adam Riess, Brian Schmidt and Saul Perlmutter, earned the 2011 Nobel Prize in Physics and underpins very nearly everything modern cosmology assumes about the fate of the cosmos. The 2025 claim, had it held, would have dismantled all of it. So when Riess and Schmidt joined the Southampton-led rebuttal, they brought a familiar standard with them. “Extraordinary claims require especially careful testing,” said Riess. The phrase has a long pedigree, and here it was being turned on a challenge to his own life’s work.
To understand what went wrong, you have to understand the tool. Astronomers measure cosmic distances using Type Ia supernovae, the violent, luminous detonations of white dwarf stars.
The trouble with exploding stars
These explosions are reliable enough to act as “standard candles”: because they peak at roughly the same brightness, their apparent dimness tells you how far away they are. The 2025 study, by Son and colleagues, argued that this reliability decays over cosmic time. As the universe aged, they claimed, the stars that produced these supernovae changed, shifting their peak brightness and tricking astronomers into reading acceleration where there was none. A subtle bias, but a fatal one if true.
The Southampton team, with lead author Dr Phil Wiseman, found the error sat somewhere more mundane: in how the age of the exploding stars had been estimated. The earlier work had assumed that the age of a galaxy was the same as the age of the star that blew up inside it. It is not, and the difference matters enormously.
Here is the awkward bit. A supernova progenitor is almost always far younger than the galaxy that hosts it, because of something cosmologists call the delay-time distribution: the probability that a given population of stars will produce a Type Ia explosion as time passes. That distribution is heavily weighted toward short delays, which means most of these supernovae, at every point in cosmic history, come from relatively young stellar systems regardless of how venerable their host galaxy happens to be. By treating galaxy age and progenitor age as interchangeable, the 2025 analysis inflated the supposed age difference between nearby and distant supernovae to roughly 5.3 billion years. Wiseman’s team puts the real figure closer to 1.9 billion. Overstated, in other words, by a factor of three to five.
There was a second omission, and arguably a more telling one. Modern supernova cosmology routinely applies a correction based on the stellar mass of the host galaxy, a standard adjustment that quietly absorbs most of these environmental quirks. The 2025 study left it out. Put it back in, the Southampton team found, and the brightness-versus-age relationship the earlier paper had leaned on simply vanishes.
Back to the real mystery
They checked it more than one way, which is what you’d hope. Supernovae in old, quiescent galaxies were compared against those in younger, star-forming ones of the same mass; no significant difference in brightness emerged, contrary to what the 2025 model demanded. Data from the Dark Energy Survey, which tracks how that mass correction shifts across cosmic distances, showed the kind of redshift evolution the rival theory required to be essentially zero. Folding it into the calculation nudged the dark energy measurement by less than 0.01, a rounding error in a field that obsesses over decimal places.
None of this means the underlying puzzle has gone away. Quite the opposite. “Thankfully we have averted this crisis, but the mystery about why the universe is still accelerating in size remains,” Wiseman said. The point of the exercise was never to close the book on dark energy. It was to confirm that there is still a book worth reading. “By proving our measurements are correct, we can get back to trying to understand what dark energy actually is, rather than wondering if it exists at all,” he added.
For all that, the challenge was not wasted effort, and the Southampton team is unusually generous about saying so. Mark Sullivan, also at Southampton, framed the whole affair as science doing exactly what it should. “This is how progress is made,” he said. “Although this idea did not turn out correct, it has opened up new ways of thinking about how supernovae explode and how we can measure dark energy more accurately.” Co-author Dr Brodie Popovic put it more plainly: the episode was a chance to go back and interrogate every assumption baked into the measurements, and the assumptions held.
The timing is fortunate. A new generation of instruments is about to bury cosmologists in supernovae: the Vera C. Rubin Observatory, the Nancy Grace Roman Space Telescope, and surveys that will catalogue these explosions by the tens of thousands rather than the hundreds. With that much data, even tiny systematic errors could masquerade as new physics, or hide it. Knowing the old tools still work, that the candles still burn true, is what lets the next discovery, whatever it turns out to be, actually count.
DOI / Source: 10.1093/mnras/stag797
Frequently Asked Questions
Wait, did dark energy almost get cancelled?
For a few months in late 2025 it looked possible. A study suggested that the exploding stars astronomers use to measure the universe became systematically misleading over cosmic time, which would have erased the evidence for accelerating expansion. The new Southampton-led analysis found the claim rested on a measurement error, so dark energy stays firmly on the books, even if nobody yet knows what it is.
How can a star be younger than the galaxy it lives in?
The galaxy is the whole stellar city; the supernova comes from one particular white dwarf system within it. Because most Type Ia supernovae explode within a relatively short window after their stars form, the typical progenitor is far younger than the ancient galaxy surrounding it. Conflating the two ages was the core mistake in the 2025 claim, and untangling them is what restored the standard picture.
Why does it matter whether the measurements were off by such tiny amounts?
In cosmology the difference between an accelerating universe and a decelerating one can hide in the third decimal place. The disputed correction would have shifted the key dark energy figure by less than 0.01, which sounds trivial but is exactly the scale at which real discoveries and false alarms both live. Getting it right is what separates genuine new physics from an artefact.
Is this the end of challenges to dark energy?
Almost certainly not, and the researchers seem glad of that. They were careful to credit the rejected idea with sharpening how supernovae and their host galaxies are understood. With vast new surveys like the Rubin Observatory coming online, the methods will be stress-tested far harder than this.
