The oldest light in the universe may arrive with a tiny twist. A reanalysis of data from the Atacama Cosmology Telescope in Chile finds that the polarization of the cosmic microwave background appears rotated by 0.215 degrees, give or take 0.074 degrees, an effect called cosmic birefringence that standard physics does not predict.
The study, by Patricia Diego-Palazuelos and Eiichiro Komatsu of the Max Planck Institute for Astrophysics, is posted on arXiv and was revised in April 2026. The institute says the result reaches a 99.6% confidence level, while the paper puts its statistical significance at 2.9 sigma, short of the 5-sigma bar physicists normally require for a discovery.
What the twist is
Light is a wave that vibrates in a particular direction, its polarization. The cosmic microwave background, emitted about 13.8 billion years ago, carries a polarization pattern that cosmologists can predict. If some unknown field filling space interacted with that light on its long trip to Earth, it could slowly turn the direction of vibration by a small, uniform angle across the whole sky.
That would break parity symmetry, the rule that physics behaves the same in a mirror image. In the standard model of particle physics, parity violation is confined to the weak force at atomic scales. Some theories propose that dark matter or dark energy could be made of a particle that breaks this symmetry; axions are the prime example named by the institute.
To put the angle in perspective, a full circle is 360 degrees, so a 0.215-degree turn is roughly 1/1,675 of a rotation, smaller than the width of a full Moon in the sky.
How the team measured it
The big problem is calibration. If a telescope’s detectors are slightly misaligned with the sky, the data look rotated even when nothing in space is doing it. Earlier work led by Komatsu using NASA’s WMAP and ESA’s Planck missions found an angle of 0.34 degrees, plus or minus 0.09, by calibrating against polarized light from dust in our own galaxy, a method that critics said depended on how that dust was modeled.
The new analysis avoids the dust. The team used a model of the telescope’s own optics to set prior expectations for instrumental misalignment, then used Bayesian statistics to separate that from any cosmic rotation in the Data Release 6 maps. That release, which the ACT collaboration published in March 2025, covers 90, 150 and 220 GHz and has about three times better sensitivity and five times sharper resolution than Planck.
Treating the ACT and WMAP-plus-Planck results as unrelated measurements, the institute reports a combined angle of 0.264 degrees, plus or minus 0.058. The paper calls it suggestive that separate data sets, analyzed with different methods, gave an angle with the same sign and a similar size.
Limits
The authors are cautious. Their abstract says “there remain systematics in the ACT data that are not understood” and that these do not allow strong cosmological conclusions. The combined figure also assumes the two measurements share no hidden errors.
The decisive tests will come from BICEP3, CLASS and the Simons Observatory, which are building artificial polarization sources to calibrate their detectors without relying on optics models or galactic dust. If they confirm the signal, the institute writes, it “will have a revolutionary impact on cosmology and fundamental physics”. For more on the dark side of the cosmos, see our stories on a model in which dark energy switched sign 9.5 billion years ago and the Euclid telescope’s first look at the dark universe.
Further reading
- Diego-Palazuelos and Komatsu, Cosmic Birefringence from the Atacama Cosmology Telescope Data Release 6 (arXiv)
- Wikipedia: Cosmic microwave background
- Wikipedia: Atacama Cosmology Telescope
Image: Till Niermann, CC BY-SA 3.0 via Wikimedia Commons
This article was produced with AI assistance and reviewed by our Editorial Desk. How we work.



