Retraction to: Nature https://doi.org/10.1038/s41586-020-2801-z Published online 14 October 2020

The editors of Nature wish to retract this paper. Following publication, questions were raised regarding the manner in which the data in this paper have been processed and analysed, which the authors and Nature have been working to resolve.

We have now established that some key data processing steps—namely, the background subtractions applied to the raw data used to generate the magnetic susceptibility plots in Fig. 2a and Extended Data Fig. 7d—used a non-standard, user-defined procedure. The details of the procedure were not specified in the paper and the validity of the background subtraction has subsequently been called into question.

The authors maintain that the raw data provide strong support for the main claims of the original paper. Nevertheless, we are of the opinion that these processing issues undermine confidence in the published magnetic susceptibility data as a whole, and we are accordingly retracting the paper. All authors disagree with this decision. An earlier version of this note stated that not all authors expressed their opinion, but the editors have since been contacted by the missing authors.