Nature 433, 230–238 (2005)

In the legend to Figure 1, the experiment shown was wrongly attributed to Clauser. The legend should have read ‘Principle of Grangier, Roger and Aspect’s experiment...(ref. 10)’. In contrast, the Clauser experiment (ref. 4) involved one beam splitter on each side with detectors in each of the resulting four output ports. Four characteristic correlations were measured. In both the Clauser (ref. 4) and the Grangier, Roger and Aspect (ref. 10) experiments the observed correlations cannot be explained via classical light fields, but can easily be understood by assuming single photons that can only be detected once behind a beam splitter.