1 Introduction

Blue stragglers are stars that are bluer and more luminous than the main-sequence turnoff of the population to which they belong. Hence they appear to be either younger than the other stars or have main-sequence lifetimes that exceed those of stars of similar mass. Blue stragglers are easiest to distinguish in globular and open clusters but are also found among field halo stars. Formation scenarios that require anomalous evolution of a single star are considered less likely than explanations that involve the interaction of stars in a binary or in a multiple-star encounter. For example, a blue straggler may be formed when one star in a binary accretes a substantial amount of mass from its companion or when two stars merge either in a binary or in a collision. In the last case, blue stragglers carry information about the high-stellar density environments in which they are found. Reviews on blue stragglers are given by e.g. Stryker (1993) and Bailyn (1995).

The blue stragglers in M 67 have received frequent attention. Ten of them were monitored for radial-velocity variations for nearly twenty years; this revealed one short-period (4.2 day) eccentric binary and five circular and eccentric binaries with periods between 850 and 5000 days (Milone & Latham 1992; Latham & Milone 1996). Leonard (1996) and Hurley et al. (2001) concluded that not one of the proposed blue-straggler formation mechanisms alone can explain the properties of these binaries.

S 1082 is one of the remaining four blue stragglers without an orbital solution. Its spectrum shows the lines of an early F-type subgiant. These lines show only small radial-velocity variations ( $v_{\rm rad}<7$ km s^-1 peak-to-peak, Mathieu et al. 1986) and little rotational broadening ( $v_{\rm rot}\sin i=4$-11 km s^-1, van den Berg et al. 1999). This appears to be in contradiction to the eclipse light curve with a period of 1.07 days found by Goranskij et al. (1992). The relative velocity v of two stars in a 1.07 day binary is v=208 ( $M/M_{\odot})^{1/3}$ km s^-1 where M is the total mass of the binary; the stellar rotation in such a short-period binary is expected to be synchronised with the orbit, giving $v_{{\rm rot}}=(R/a)v$, with a the distance between the binary stars and R the stellar radius. The eclipse light curve excludes the low inclination as well as the extreme mass ratio required to bring the observed velocity limits in agreement with the predicted values for an F star in the binary. We therefore tentatively conclude that S 1082 is a (visual or physical) triple.

Evidence for a short orbital period is also furnished by the variation detected in a broad, shallow component (Mathys 1991) in e.g. the H$\alpha$ line on time scales of hours (van den Berg et al. 1999). Evidence for a wide orbital period of about 1000 days was found from the radial-velocity variations of the narrow lines of the F star (Milone 1991).

We have collected multiband photometry of S 1082 together with high-resolution spectra that sample the light curves at different phases. Our goal was to investigate the eclipses found by Goranskij et al., to monitor the variability of the second spectral component as function of photometric phase and to model the light curve together with the radial-velocity curves. Section 2 describes the observations and data reduction. The results of the spectroscopic and light curve analysis are presented in Sects. 3 and 4. In Sect. 5 we summarise our interpretation of the nature of this blue straggler.