## Fig. 3

Analysis of the empirical uncertainty for short-term repeat observations of RV using filter HR15N. The solid line in plot **a) **shows the variation of *E*_{RV} × (*S*/*N*)/(1 + ( [ *v*sin*i* ] /*C*)^{2})^{3/4} with S/N. The horizontal line indicates the value of parameter *B* in Eq. (1) fitted to the full dataset. Blue crosses show the estimated values *B* as a function of S/N corrected for the measured variation of *B* with *T*_{eff} (see Sect. 4.1 and Table 3). Plot **b) **shows the variation of *E*_{RV} × *S*/*N* with *v*sin*i*. The solid line show the relationship predicted using the theoretical value of *C* and the value of *B* from plot **a)**. The dashed line shows a curve of similar functional form using parameters *B* and *C* fitted to the binned data. In plots **a) **and **b) **the *y*-axis shows an estimate of the standard deviation based on the MAD divided by 0.72 (see Sect. 3.2). Plot **c) **shows the cumulative probability distribution (CDF) of the normalised uncertainty in RV for short-term repeats. The red solid line shows results for measured data, the dashed line shows the cumulative distribution of a Gaussian with unit dispersion, and the diamond symbols show the cumulative distribution function for a Student’s t-distribution with *ν* = 6.