∆υ_rad distributions for HR10 (thick black line) and HR21 (thin red line) observations triggering an SB2 detection. Left: earliest iDR4 analysis described in Merle et al. (2017); right: iDR4 re-analysis (this study). We are interested in the individual exposures, therefore an SB2 with more than one epoch will be counted more than once and may appear in different bins depending on the change of its components velocities from one epoch to another.