UC San Diego test pits Planck's detectors against each other to vet a 0.37-degree twist in the oldest light

A rotation of 0.37 degrees with an uncertainty of 0.12 degrees sits about three uncertainties away from zero ^[1]. In my view that is enough to take seriously. The UC San Diego team still says plainly that its result does not confirm cosmic birefringence ^[8].

The light being tested is the cosmic microwave background. It first appeared some 380,000 years after the Big Bang, and a small share of it is polarized ^[11]. Recent observations suggest that polarization turned by a fraction of a degree on its way to us ^[1]. A real rotation could point to physics beyond the Standard Model and bear on questions such as dark matter ^[2]. The same signal could also come from a small error in how a telescope's polarization detectors are oriented ^[3].

The design is a comparison. Anto I. Lonappan and colleagues took maps made by different detectors, reasoning that a genuine cosmic rotation would show up in every one ^[5]. Their method cancels whatever rotation the maps share, so what survives most likely reflects differences in how each detector's polarization angle was calibrated ^[5]. It is a relative calibration check, measuring detectors against one another ^[4]. The Discover account of the paper does not explain how a tilt shared equally by every detector is told apart from a cosmic one.

For a test run, the team applied the estimator to eight maps from the European Space Agency's Planck satellite and set the output beside the Minami-Komatsu analysis, the standard method for separating birefringence from calibration error ^[6][16]. The two agreed on 0.37 degrees, plus or minus 0.12 ^[7]. The researchers describe what they have as an independent estimator and modeling route ^[8].

The thing this comparison doesn't tell you is whether the eight maps are clean. Both methods read the same Planck data, so a flaw built into those maps would reach both ^[6]. The agreement shows the new estimator reproduces the established one on identical inputs.

"The signal we are looking for is incredibly small, so we have to be certain that we are seeing the universe and not our instrument," Lonappan, a postdoctoral fellow at UC San Diego, said in a statement ^[9]. "Before interpreting a tiny rotation as new physics, we want to know that the calibration itself can be trusted," he said ^[10].

The same calibration question applies to B modes, curl-like polarization patterns that gravitational lensing produces and that primordial gravitational waves from the very early universe may also produce ^[12]. The primordial kind has not been detected, and it could carry clues about cosmic inflation ^[12]. The team hopes the method helps avoid errors in telling B modes from E modes ^[13].

"The detection of primordial B modes would transform our understanding of the early universe, so the measurement must survive rigorous calibration checks," co-author Brian Keating, a Chancellor's Distinguished Professor of Physics, said in a statement ^[14]. "This is ultimately about knowing when we can trust a measurement," he said ^[15].