The mysterious substance known as dark energy thought to be pushing the universe apart at ever greater speeds may be nothing more than an artefact of our acceleration through a local patch of the universe. That is the controversial claim of a group of physicists who reckon they have found flaws in the evidence underpinning the Nobel-prize winning discovery of cosmic acceleration. The dispute centres on exploded stars known as type Ia supernovae, which allow researchers to calculate cosmic distances and rates of expansion.
Type Ia supernovae are known as “standard candles” since they are generated by stars exploding with a very specific mass and therefore known absolute brightness. By observing these objects’ apparent brightness, astronomers can work out how far they are (in space and therefore time), and by combining that information with the red shift in their emitted light can then calculate how fast the universe was expanding at that point in time.
In 1998 two research groups – one led by Saul Perlmutter and the other by Brian Schmidt – announced that they had found evidence from some 50 distant supernovae that the expansion of the universe was not slowing down, as expected, but was in fact speeding up. They came to that conclusion after finding that the light from those supernovae was dimmer than anticipated. This extraordinary discovery implied that the universe was being pushed apart by (a still unknown) entity known as dark energy, and earned Perlmutter, Schmidt and Schmidt’s colleague Adam Riess the 2011 Nobel Prize for Physics.
This cosmic acceleration is now a central element of the Standard Model of cosmology and has been corroborated by other types of observational evidence, including data from the cosmic microwave background (CMB) – the cold and faint radiation generated shortly after the Big Bang.
However, in 2015 Oxford University physicist Subir Sarkar and two colleagues at the Niels Bohr Institute in Copenhagen uploaded a paper to the arXiv server claiming that the evidence for cosmic acceleration was not as watertight as generally supposed. Carrying out statistical tests on a sample of 740 type Ia supernovae, they investigated the empirical procedure used to adjust absolute brightness to account for variations in emission between supernovae as well as absorption of their light by intervening dust. They claimed to have found “only marginal” evidence for cosmic acceleration, calculating a statistical significance for such acceleration of less than 3σ. Normally, 5σ is seen as the gold standard for a discovery.
Their paper was published 16 months later in Scientific Reports and came in for much criticism from other scientists. Experts took aim at the statistical analysis itself, questioning Sarkar and colleagues’ assumptions about the properties of the supernovae involved, and criticized the group for considering the supernovae data in isolation. When combined with other data, including those from the CMB and the distribution of galaxies in the universe (known as baryon acoustic oscillations), critics argued that there could be no doubt about the reality of cosmic acceleration.
But Sarkar and colleagues at the Niels Bohr Institute and the Paris Institute of Astrophysics have now upped the ante by writing a second paper, accepted for publication in Astronomy and Astrophysics and uploaded to arXiv. In it they further downgrade the significance of the supernovae evidence, arguing in fact that cosmic acceleration probably does not exist – that what the Nobel prize-winning teams saw was simply the result of local motion in our particular corner of the universe.
The researchers came to this conclusion after scrutinizing publically-available supernova data for a “monopole”, which would cause all points in the universe to accelerate away from an observer in a particular rest frame (that of the CMB). But they also looked for a “dipole”, which would mean some points moving away from an observer while others would move towards them. This is in fact how we see the CMB. Thanks to our motion through space as part of a group of galaxies travelling at over 600 kms-1 relative to cosmic expansion – known as the “bulk flow” – measurements of temperature in opposite halves of the microwave sky differ by about 1 part in 1000.
It was when considering the effect of this bulk flow that Sarkar and colleagues noticed what seemed to be an anomaly within the supernovae red-shift data. These data had been adjusted to convert red-shifts from the Earth’s rest frame to what was thought to be the CMB frame. However, that conversion, says Sarkar, assumed that all local motion gets washed out when moving to scales above about 500 million light-years. In fact, he argues, numerous observations of galaxy motion show that such movement persists up to at least 1::billion light-years.
By re-converting the red-shift data back to their raw “heliocentric” form as best they could, and plugging the data into a model, Sarkar and colleagues found that the monopole component – the universal acceleration – yielded just a 1.4σ signal, while the dipole – presumably a local motion – was present at 3.9σ. What’s more, they found that this dipole lines up with the one in the CMB.
According to Riess, however, the supernovae data used by Sarkar’s group are out of date. He says that he and some colleagues, including D’Arcy Kenworthy of Johns Hopkins University, plugged data from a sample of about 1300 supernovae with lower systematic uncertainties into the model used in the latest work. The results, he says, were unambiguous, with the existence of a dipole rejected at more than 4σ and cosmic acceleration confirmed at over 6σ.
More importantly, says Riess, the objections against Sarkar and colleagues’ original statistical analysis still stand, as do the criticisms of neglecting other data. “The evidence for cosmic acceleration and dark energy are much broader than only the supernovae Ia sample, and any scientific case against cosmic acceleration needs to take those into account,” he says.
Even here, however, Sarkar insists the evidence is lacking. He claims that the data on baryon acoustic oscillations are too sparse to chose between models with and without cosmic acceleration, while dark energy would have been too weak to leave a significant imprint in the early universe.