Results of the TODCOR algorithm applied to two objects. Top panels: RV of the donor and accretor shown on the x- and y-axes, respectively. Markers show the RVs of both components determined for each epoch with different methods and in various wavelength windows, with each marker displaying a different epoch. The lines indicate the linear best-fit to the RVs, as the relationship between donor and accretor RVs is linear. Grey diamonds show the donor RVs from ceres (Brahm et al. 2017) on the x-axis, with corresponding accretor RVs on the y-axis. The accretor RVs were computed using the ceres donor RVs, the mass ratio from EB+22, centre-of-mass velocity from Gaia, and Equation (2). Coloured squares and lines show the results from TODCOR for a number of wavelength windows centered on different lines. Black squares show the unweighted mean for each epoch velocity over the different wavelength windows, offset to higher v_A by a small amount for clarity. The fit labelled ‘Joint’ is then performed to these mean values. From the slope and intercept of the best-fit line, the mass ratio and centre-of-mass velocity can be computed, see Equation (3). Bottom panels: Plot of the mass ratio vs the centre-of-mass velocity as determined from each set of data in the top panel, coloured accordingly. The left panel shows target G-2966, while on the right we see object G-5536, with a grey background to highlight the difficulty of the analysis for this object. For G-5336, we see that while there is a big discrepancy between the mass ratio from TODCOR and the one found by EB+22, the quality of the linear fit with TODCOR is not good, calling this result into question.