Multiple star systems are commonly assumed to form coevally ; they thus provide the anchor for most calibrations of stellar evolutionary models .
In this paper we study the binary population of the Taurus-Auriga association , using the component positions in an HR diagram in order to quantify the frequency and degree of coevality in young binary systems .
After identifying and rejecting the systems that are known to be affected by systematic errors ( due to further multiplicity or obscuration by circumstellar material ) , we find that the relative binary ages , | \Delta \log \tau| , have an overall dispersion \sigma _ { | \Delta \log \tau| } \sim 0.40 dex .
Random pairs of Taurus members are coeval only to within \sigma _ { | \Delta \log \tau| } \sim 0.58 dex , indicating that Taurus binaries are indeed more coeval than the association as a whole .
However , the distribution of | \Delta \log \tau| suggests two populations , with \sim 2/3 of the sample appearing coeval to within the errors ( \sigma _ { | \Delta \log \tau| } \sim 0.16 dex ) and the other \sim 1/3 distributed in an extended tail reaching | \Delta \log \tau| \sim 0.4-0.9 dex .
To explain the finding of a multi-peaked distribution , we suggest that the tail of the differential age distribution includes unrecognized hierarchical multiples , stars seen in scattered light , or stars with disk contamination ; additional followup is required to rule out or correct for these explanations .
The relative coevality of binary systems does not depend significantly on the system mass , mass ratio , or separation .
Indeed , any pair of Taurus members wider than \sim 10′ ( \sim 0.7 pc ) shows the full age spread of the association .