One of the tools used by people doing cosmology is the assumption that at a certain scale in the Universe, things become isotropic in distribution. In other words, if you look at a big enough sample of the Universe, one part of the Universe should be essentially identical to another.
When we look out at the local section of the Universe, the Milky Galaxy or even our local galactic supercluster, this clearly isn’t true. But, if you zoom out far enough, the local density anisotropy begins to disappear. It’s a matter of practice that when you’re looking at length scales of this magnitude, you are working in the realm where the Cosmological principle holds. Even the massive voids between superclusters seem to be evenly distributed.
Except there’s been a recent observational challenge apparently. New images of a Large Quasar Group (LQG) show a structure much larger than should be allowed:
“Based on the Cosmological Principle and the modern theory of cosmology, calculations suggest that astrophysicists should not be able to find a structure larger than 370 Mpc. Clowes’ newly discovered LQG however has a typical dimension of 500 Mpc. But because it is elongated, its longest dimension is 1200 Mpc (or 4 billion light years) – some 1600 times larger than the distance from the Milky Way to Andromeda.”
(A quasar is now believed to be an early form of the core of an active galaxy – we only see them at extreme distance from the Earth, and thus at a very early moment in the Universe’s history.)
I’m not current enough in the field to know whether or not this is a major challenge, or merely represents a data point that the models can be adjusted to include. Cosmological data has extremely large error built into as a result of the difficulty inherent in making the observations used to support the models.
As I read this, it’s an error of a little bit more than a factor of 2. That just doesn’t strike me as a sufficient motivation to overturn the principle. (It is a principle representing phenomenological experience, not a formal law after all.)
Neat result though. More data for the models.