Twins like to be seen: Observational biases affecting spectroscopically selected binary stars

Massive binary stars undergo qualitatively different evolution when the two components are similar in mass ('twins'), and the abundance of twin binaries is therefore important to understanding a wide range of astrophysical phenomena. We reconsider the results of Pinsonneault & Stanek (2006), who argue that a large proportion of binary stars have nearly equal-mass components; we find that their data imply a relatively small number of such 'twins.' We argue that samples of double-lined spectroscopic binaries are biased towards systems with nearly equal-brightness components. We present a Monte-Carlo model of this bias, which simultaneously explains the abundance of twins in the unevolved binaries of Pinsonneault & Stanek (2006), and the lack of twins in their evolved systems. After accounting for the bias, we find that their observed mass ratios may be consistent with a variety of intrinsic distributions, including either a flat distribution or a Salpeter distribution. We conclude that the observed overabundance of twins in Pinsonneault & Stanek (2006) does not reflect the true population of binaries, and we briefly discuss the astrophysical implications of the lack of twins.