New analysis of 1,700 supernovae questions cosmic acceleration

A new study argues that the case for an accelerating universe, one of the most celebrated findings in modern cosmology, does not hold up once a suspected bias in the supernova data is corrected. The work was led by researchers at the Tata Institute of Fundamental Research in Mumbai together with Professor Emeritus Subir Sarkar of the University of Oxford, and appears in Monthly Notices of the Royal Astronomical Society. It revisits the Pantheon+ catalogue, a compilation of more than 1,700 exploding stars called Type Ia supernovae, which for over 25 years have been used to measure how fast the universe is expanding.

The original discovery that this expansion appears to be speeding up, rather than slowing under gravity as expected, won the 2011 Nobel Prize in Physics and led to the idea of 'dark energy', a mysterious force thought to arise from the quantum vacuum and push space apart. Sarkar and his colleagues, Animesh Sah and Mohamed Rameez, applied a recently proposed correction that accounts for the age of the stars that eventually explode as these supernovae. After applying it, they found the data no longer support a universe accelerating uniformly in every direction. Instead, their analysis suggests the expansion is, overall, slowing down.

Published in the same journal issue, a separate paper led by the University of Southampton with Associate Professor Maria Vincenzi of Oxford reaches the opposite conclusion, maintaining that the evidence still points to an accelerating universe. The two papers directly disagree, and the sources give no indication of which conclusion will prevail.

Type Ia supernovae are useful for measuring cosmic distances because their explosions are thought to reach a fairly consistent brightness, letting astronomers treat them as 'standard candles' whose apparent dimming reveals how far away they are. Sarkar explained that there is increasing evidence that this brightness actually depends on the age of the star system that produced the explosion. If that dependence is not accounted for, he said, it can create the appearance of accelerating expansion where none exists.

A related, independent line of evidence, published separately by a team at Yonsei University in South Korea, points in the same direction. Based on a sample of 300 host galaxies, that team reported a systematic effect at very high statistical confidence: supernovae from younger stellar populations appear systematically fainter than expected, while those from older populations appear brighter, even after standard corrections. When they corrected for this bias, the data no longer matched the standard cosmological model with a fixed dark energy, and instead fit better with a model in which dark energy weakens over time, a model derived from separate measurements by the Dark Energy Spectroscopic Instrument (DESI) project using sound-wave patterns left over from the early universe and the cosmic microwave background.

Sarkar's team also examined whether the apparent acceleration is the same in every direction, a property called isotropy that the standard cosmological model assumes to be true. They found that the inferred acceleration is directed mainly along the direction the Milky Way is moving locally, as marked by a hotspot in the cosmic microwave background, and that it fades with distance. Because an effect from the quantum vacuum, which is what dark energy is thought to be, cannot depend on direction, this pattern would rule out dark energy as the cause regardless of whether the age correction is applied. Applying the correction, they say, turns the remaining direction-independent signal into an actual deceleration, again ruling out dark energy.

This is not the first attempt to test type Ia supernova data for such directional effects. Earlier work using a modified model of the universe that allows for rotation and expansion together found no significant signal of anisotropy in a sample of a few hundred supernovae, though the resulting limits on any anisotropy were described as comparatively weak, meaning a small effect could not be excluded. A separate study using the Union2, Pantheon, and gamma-ray burst datasets examined a possible preferred direction in models of dark energy that vary with direction, and found the preferred axis in some model comparisons was comparable to results reported in earlier studies, though results differed depending on which datasets and models were combined.

Doubts about the strength of the acceleration evidence are not new. A study published in Nature examined 740 supernovae from the Joint Lightcurve Analysis catalogue using a statistically rigorous maximum-likelihood method, rather than the more common technique of minimising a chi-squared value, and found only marginal evidence, in the researchers' own words, for the claim that the universe's expansion is presently accelerating. That paper noted the data were 'still quite consistent with a constant rate of expansion' once the empirical corrections applied to supernova brightness were properly accounted for in the statistical model.

Taken together, the sources describe a body of evidence that has never been entirely settled, running from early statistical critiques through directional tests to the newest age-correction studies. What has changed is not agreement, but the number of independent groups now questioning the standard picture from different angles: one through the statistics of the original fits, one through the direction of the sky, and one through the age of the stars themselves.

Sarkar's team says its next step is to test the findings against a much larger dataset expected from the Vera C. Rubin Observatory's Legacy Survey of Space and Time, which is expected to measure hundreds of thousands of supernovae, roughly two orders of magnitude more than the current Pantheon+ catalogue. The Yonsei-led team is separately running what it calls an 'evolution-free test', using only supernovae from young, similarly aged host galaxies across the full range of measured redshifts, and reports that its first results support the team's main conclusion of a decelerating universe. A co-author on that study noted that within the next five years, Rubin's discovery of tens of thousands of new supernova host galaxies should allow far more precise measurements of stellar ages, providing a stronger test of the debate.

For now, the field is left with two peer-reviewed papers published side by side reaching opposite conclusions from related data, plus older independent work questioning both the strength and the direction of the original acceleration signal. If the age-correction result is confirmed, both the Oxford-led and the Yonsei-led teams say it would mark a major shift in cosmology, since the discovery of accelerating expansion and dark energy more than two decades ago. The sources are clear that this remains an open, contested question rather than a settled one.