The distribution of binary-system mass ratios: An extended, less-biassed sample
During the past 30 years, the astronomical literature has included orbits by Roger Griffin and colleagues for at least 297 single-lined and 97 double-lined spectroscopic binaries derived from data gathered wholly or partially with the Cambridge Radial Velocity Spectrometer and its close relatives. These systems have, on average, considerably longer periods than SB systems in other compilations and considerably smaller velocity amplitudes (except in comparison to samples of stars with planets), with medians near 700 days and 9.5 km s_1 for the SB1s and the extremes rising to several decades and falling to less than 2 km s-1. From these orbits, I here extract a distribution of mass ratios, using long-established but imperfect methods. A number of correction factors might be adopted: for over-representation of SB2s, observational selection against long periods, small amplitudes (especially when combined with large rotation speeds and/or near-equal brightness of the stars), large eccentricities, and very possibly other facts of life. Some combination of such corrections can undoubtedly produce whatever distribution your favourite models for binary-star formation predict; but the unvarnished graph does not look much like any such model.