Queensland: 2,884 Supernovae Cast Doubt on Dark Energy
The University of Queensland merged two samples into the Unite catalog of 2,884 supernovae: the deviation from the standard model is 2.5 to 3.1 sigma, a new hint that dark energy changes over time.

A team of scientists at the University of Queensland in Australia has published the most complete and consistently processed catalog of Type Ia supernovae to date, a set of 2,884 explosions used as distance markers across the cosmos. The paper, titled “Supernovae Unite: Combining Pantheon+ and DES-SN5YR,” was submitted to the arXiv preprint server on September 4, 2026, under the identifier 2609.05053, with the university’s press release and a writeup on Phys.org following on September 8. The work is led by Ryan Camilleri, a PhD candidate at the University of Queensland’s School of Mathematics and Physics, with 23 co-authors including astrophysicist Tamara Davis, David Rubin, Paul Sah, and Dan Scolnic. The result: the dataset deviates from the standard cosmological model by a strength of 2.5 to 3.1 sigma, a fresh hint that dark energy, the force driving the accelerating expansion of the universe, may be changing over time.
The dataset, named Unite, was built by merging two earlier samples. The first is Pantheon+, and the second is DES-SN5YR, part of the Dark Energy Survey research published in 2024. That DES sample was the first to hint that dark energy might be a variable rather than a constant, and the new compilation, according to Davis, confirms that hint from a slightly different angle of measurement. Separately from this study, the DESI instrument, in its own measurements of sound waves left over from the universe’s early history, has also found hints that dark energy changes over time, something Davis points to as two entirely independent measurements pointing in the same direction.
“We’ve pieced back together three decades of astronomical observations into one single, consistent framework,” Camilleri said in the university’s press release.
The catalog draws on roughly three decades of observations, with older data corrected using an improved understanding of how supernovae, cosmic dust, and the mass of host galaxies behave. “Over the years we’ve learned a lot more about how supernovae behave, so we were able to go back and apply that improved understanding to the older data,” Camilleri explained. He added that considerable work went into reconciling observations from different telescopes with different capabilities, and into accounting for cosmic dust and host galaxy mass, factors that can affect the light reaching us from a supernova.
To sharpen the cosmological constraints, the supernova data was also cross checked against measurements of radiation left over from the Big Bang and against maps of how galaxies are arranged in space. For a flat model with constant dark energy, derived from the Unite catalog alone, the matter density came out to 0.310, with an error of plus 0.012 and minus 0.011. When dark energy was modeled as a quantity that changes over time, using the parameters w0 and wa, the values obtained were w0 equal to minus 0.861 and wa equal to minus 0.60. The figure of merit for dark energy in that model is 315, meaning the uncertainty region for w0 and wa has narrowed by about 30 percent compared with earlier constraints.
Why a deviation of 2.5 to 3.1 sigma is not proof
The variable dark energy model’s edge over the standard model comes out to 2.5 sigma using one statistical method (maximum a posteriori estimation) and 3.1 sigma using another (maximum likelihood). Both are frequentist measures, and both fall short of the 5 sigma threshold physics conventionally requires before a discovery can be claimed. The authors also note that a Bayesian analysis, a separate statistical approach applied to the same data, shows only a weak preference for the variable dark energy model over the standard one. In other words, the Unite catalog by itself does not prove that dark energy is changing. It adds one more hint in that direction, on top of the one that had already come from the DESI instrument.
“Rather than confirming the standard cosmological model, which assumes dark energy is fixed and unchanging, we have further evidence that dark energy may be evolving over time,” Camilleri said.
“Our 2024 DES supernova data was the first to hint that dark energy might be evolving over time, and this new compilation also sees a deviation from the standard model,” Davis said. “So we have two completely independent measurements that have found hints of dark energy evolving over time, which challenges the standard model’s assumption that dark energy doesn’t change.”
Confirming or ruling out this hint will require more data. The Dark Energy Bedrock All-Sky Supernova program, DEBASS, is expected to add several hundred more supernovae to the catalog, drawn from closer to Earth than the DES sample reached. Until those measurements are folded into the existing catalog, the deviation from the standard model, even though it has now shown up in two independent measurements, remains a hint and nothing more.



