By Zeeya Merali

It's another victory for Einstein -- albeit not a resounding one. General relativity has been confirmed at the largest scale yet. But the galactic tests used to put the theory through its paces cannot rule out all rival theories of gravity.

General relativity has been rigorously tested within the Solar System, where it explains the motion of planets with precision. But its reach between galaxies has been harder to verify and should not be taken for granted, says cosmologist Alexie Leauthaud, at the Lawrence Berkeley National Laboratory in California. "It's actually a tremendous extrapolation to assume that general relativity works on cosmic scales," she says.

If general relativity does break down at large scales, it could help cosmologists to explain away one of their biggest headaches: dark energy. In the 1990s, astronomers were surprised to discover that the expansion of the Universe is accelerating. That runs counter to the predictions of general relativity, which suggests that gravity's grip should be slowing the expansion. To explain this, cosmologists now invoke a 'dark energy', a force that makes up almost three-quarters of the matter and energy in the Universe and pushes it apart. But the origin of dark energy remains a mystery.

The accelerated expansion could be explained without dark energy, however, if general relativity is wrong and gravity weakens at cosmic scales. Several candidate 'modified gravity' theories take this line but, until now, no one has come up with a way to test them at large scales.

Relative success

Now, Reina Reyes at Princeton University in New Jersey and her colleagues have compared some of these models using data on the position, velocity and apparent shape of 70,000 distant galaxies mapped by the Sloan Digital Sky Survey1. The rival theories make different predictions about the degree to which light travelling to us from distant galaxies will be bent by the gravity of intermediate galaxies. This process, called 'gravitational lensing', distorts the apparent shape of the galaxies. The theories also make different predictions for both how fast galaxies grow and how they cluster together.

No single prediction can be used compare the theories directly, says Reyes. Because all the models include assumptions about whether dark matter -- the invisible substance thought to make up the bulk of matter in the Universe -- exists and if so, how it clumps together, relative to visible matter. Instead, the team had to combine all three measures -- gravitational lensing, growth and clustering -- into one ratio, called EG, in such a way that any uncertainty introduced by dark matter assumptions cancels out. The team found a value for EG of about 0.39 - a good match to the general relativistic prediction of around 0.4.

Scott Dodelson at the Fermilab Center for Particle Astrophysics in Batavia, Illinois, who was part of the team that proposed2 the EG test in 2007, says that the work is "exciting and important". "It's impressive that the team has shown we can rigorously test general relativity at large scales with data we have now," he says.

Open question

General relativity's rivals, however, returned mixed results. The team's test rules out a version of the tensor-vector-scalar, or TeVeS, model, proposed in 2004, which modifies gravity using a set of interacting fields to mimic dark matter. This model was already struggling to explain observations of galactic collisions that seem to show direct evidence of dark matter3, says Reyes. "Our test is another blow to TeVeS," says Reyes. However, she notes, more complicated versions of TeVeS could well pass the test.

The team also failed to rule out the set of so-called 'f(R) models' that slightly tweak the parameters of general relativity at large scales to explain away dark energy.

"The question of whether modified gravity or general relativity will ultimately prevail is still very much open," says Dodelson.

Glenn Starkman, a cosmologist at Case Western Reserve University in Cleveland, Ohio, says that EG "will be a useful addition to the battery of tests that modified gravity theories are already subjected to". Starkman and his colleagues have been testing the viability of a third modified gravity model, dubbed 'Einstein-Aether' theory, which uses a single new field to replace both dark matter and dark energy. Their analysis concludes that Einstein aether cannot explain the patterns seen in the cosmic microwave background4.

Leauthaud thinks that within the next two decades various planned experiments will have collected enough observational data to discriminate between general relativity and its competitors using EG. "Either we'll find general relativity is wrong, or we'll discover a new type of physics to explain dark energy," she says. "We're looking at a paradigm shift, either way."