I describe a series of processes, including hierarchical fragmentation, gravitational scattering, Kozai cycles within triple systems, tidal friction and magnetic braking, that I believe are responsible for producing the modest but significant fraction of stars that are observed as contact binaries. I also discuss further processes, namely heat transport, mass transport, nuclear evolution, thermal relaxation oscillations, and further magnetic braking with tidal friction, that influence the evolution during contact. The endpoint, for contact, is that the two components merge into a single star, as recently was observed in the remarkable system V1309 Sco. The single star probably throws off some mass and rotates rapidly at first, and then slows by magnetic braking to become a rather inconspicuous but normal dwarf or subgiant. If however the contact binary was part of a triple system originally–as I suggested above was rather likely–then the result could be a widish binary with apparently non-coeval components. There are several such known.

Binary companions are often invoked to explain the axial and point symmetry seen in the majority of planetary nebulae and proto-planetary nebulae (PPNs). To explore this hypothesis, we have undertaken a long-term (20 year) study of light and velocity variations in PPNs. From the photometric study of 24 PPNs, we find that all vary in brightness, and from a subset of 12 carbon-rich PPNs of F-G spectral type we find periods of 35-155 days, with the cooler having the longer pe\-riods. The variations are seen to be due to pulsation; no photometric evidence for binarity is seen. A radial velocity study of a sub-sample of seven of the brightest of these shows that they all vary with the pulsation periods. Only one shows evidence of a longer-term variation that we tentatively identify as being due to a binary companion. We conclude that the present evidence for the binary nature of these PPNs is meager and that any undetected companions of these PPNs must be of low mass (< 0.25 M ) or long period (> 30 years).

There are observational difficulties determining dynamical masses of binary star components in the upper HR diagram both due to the scarcity of massive binary systems and spectral and photometric contamination produced by the strong wind outflows in these systems. We discuss how variable X-ray emission in these systems produced by wind-wind col\-lisions in massive binaries can be used to constrain the system parameters, with application to two important massive binaries, Eta Carinae and WR 140.