Good measurements of visual binary stars (position angle and angular separation) have been made for nearly 200 years. Radial-velocity observers have exhibited less patience; when the orbital periods of late-type stars in the catalogue published in 1978 are sorted into bins half a logarithmic unit wide, the modal bin is the one with periods between 3 and 10 days. The same treatment of the writer's orbits shows the modal bin to be the one between 1000 and 3000 days. Of course the spectroscopists cannot quickly catch up the 200 years that the visual observers have been going, but many spectroscopic orbits with periods of decades, and a few of the order of a century, have been published. Technical developments have also been made in ‘visual’ orbit determination, and orbits with periods of only a few days have been determined for certain ‘visual’ binaries. In principle, therefore, the time domains of visual and spectroscopic binaries now largely overlap. Overlap is essential, as it is only by combining both techniques that orbits can be determined in three dimensions, as is necessary for the important objective of determining stellar masses accurately. Nevertheless the actual overlap—objects with accurate measurements by both techniques—remains disappointingly small. There have, however, been unforeseen benefits from the observation of spectroscopic binaries that have unconventionally long orbital periods, not a few of which have proved to be interesting and significant objects in their own right. It has also been shown that binary membership is more common than was once thought (orbits have even been determined for some of the IAU standard radial-velocity stars!); a recent study of the radial velocities of K giants that had been monitored for 45 years found a binary incidence of 30%, whereas a figure of 13.7% was given as recently as 2005 for a similar group.