The speed with which supernovas are moving away from us alerted scientists to the existence of dark energy, one of the most significant discoveries of recent decades. However, a new analysis of 1,048 of these mighty explosions hints at something even stranger, causing unexplained changes to the universe’s expansion rate. The astronomers who made this announcement stress their work is not yet conclusive – it could turn out to be an artifact of instrumental error or sampling bias. If the pattern is real, however, it means our understanding of physics is missing something big.
Type Ia supernovas are especially valued by astronomers for their consistency in the amount of light they release. By comparing a supernova’s apparent brightness with our estimate of its true luminosity, we can estimate its distance. The speed with which the supernova’s galaxy is moving away from us can be estimated using the redshift of its light. Combined, these allow us to calculate the “Hubble constant” (H0), the ratio of speed to distance with which the universe is expanding.
However, attempts to measure H0 have run into trouble, with different values produced using different methods. Supernovas have failed to settle the question, perhaps because the samples used have been too small, or biased in some way. Dr Maria Dainotti of the National Astronomical Observatory of Japan set out to investigate using the so-called Pantheon sample of 1,048 supernovae.
Dainotti and co-authors conclude in The Astrophysical Journal (preprint on ArXiv.org) this sample is consistent with H0 having grown as the universe has aged. If true, we’d need to fundamentally rewrite physics – possibly by adding some currently unknown forces – to explain this.
Dainotti’s large sample size doesn’t fit well with the idea changes to H0 are a product of insufficient data points. Nevertheless, she isn’t ready to demand the standard model of cosmological physics be overturned just yet. The paper acknowledges a much less dramatic explanation for these results would be if there is a bias to the supernovas we are detecting. No matter how large your sample is, it will create a distorted picture if it is unrepresentative. Early in the universe, stars had less metal in them, and it is possible Type Ia supernovas from metal-poor stars are not as bright, causing us to miss some and miscalculate the distance to those we do see.
As a result, the quest to find fainter supernovas we might currently be overlooking continues, as does efforts to control for the composition of host galaxies. It’s possible, perhaps even probable, more powerful telescopes will yield a better sample that will make this discrepancy go away.
Nevertheless, Dainotti’s data adds to the suspicion something very big is going on in the universe we have not yet got our heads around. Given the recent evidence that new physics may be required on the scale of the very small, the laws of science look to be a long way from finalized.