Search for a command to run...
Deriving the distribution of binary parameters for a particular class of stars over the full range of orbital separations usually requires the combination of results from many different observing techniques (radial velocities, interferometry, astrometry, photometry, direct imaging), each with selection biases. However, Cepheids---cool, evolved stars of $\sim$$5\, M_\odot$---are a special case because ultraviolet spectra will immediately reveal any companion star hotter than early type A, {\it regardless of the orbital separation}. We have used {\it International Ultraviolet Explorer} (\IUE) UV spectra of a complete sample of all 76 Cepheids brighter than V=8 to create a list of {\it all 18} Cepheids with companions more massive than $2.0\, M_\odot$. Orbital periods of many of these binaries are available from radial-velocity studies, or can be estimated for longer-period systems from detected velocity variability. In an imaging survey with the {\it Hubble Space Telescope} Wide Field Camera 3, we resolved three of the companions (those of $\eta$ Aql, S Nor, and V659 Cen), allowing us to make estimates of the periods out to the long-period end of the distribution. Combining these separations with orbital data in the literature, we derive an unbiased distribution of binary separations, orbital periods, and mass ratios. The distribution of orbital periods shows that the $5\, M_\odot$ binaries have systematically shorter periods than do $1\, M_\odot$ stars. Our data also suggest that the distribution of mass ratios depends both on binary separation and system multiplicity. The distribution of mass ratios as a function of orbital separation, however, does not depend on whether a system is a binary or a triple.