Abstract. Radial velocities have been determined for a sample of 2930 B2-F5 stars, 95% observed by the Hipparcos satellite in the north hemisphere and 80% without reliable radial velocity up to now. Observations were obtained at the Observatoire de Haute Provence with a dispersion of 80Å mm −1 with the aim of studying stellar and galactic dynamics. Radial velocities have been measured by correlation with templates of the same spectral class. The mean obtained precision is 3.0 km s −1 with three observations. A new MK spectral classification is estimated for all stars.

The Gaia mission is designed as a Galaxy explorer, and will measure simultaneously, in a survey mode, the five or six phase-space parameters of all stars brighter than 20th magnitude, as well as providing a description of their astrophysical characteristics. These measurements are obtained by combining an astrometric instrument with micro-arcsecond capabilities, a photometric system giving the magnitudes and colours in 15 bands and a medium-resolution spectrograph named the Radial Velocity Spectrometer (RVS). The latter instrument will produce spectra in the 848-to 874-nm wavelength range, with a resolving power R = 11 500, from which radial velocities, rotational velocities, atmospheric parameters and abundances can be derived. A companion paper has presented the characteristics of the RVS and its performance. The present paper details the outstanding scientific impact of this important part of the Gaia satellite on some key open questions in present-day astrophysics. The unbiased and simultaneous acquisition of multi-epoch radial velocities and individual abundances of key elements in parallel with the astrometric parameters is essential for the determination of the dynamical state and formation history of our Galaxy. Moreover, for stars brighter than V 15, the resolving power of the RVS will give information about most of the effects that influence the position of a star in the Hertzsprung-Russell diagram, placing unprecedented constraints on the age, internal structure and evolution of stars of all types. Finally, the RVS multi-epoch observations are ideally suited C 2005 RAS Gaia Radial Velocity Spectrometer 1307 to the identification, classification and characterization of the many types of double, multiple and variable stars.

The definition and optimization studies for the Gaia satellite spectrograph, the ‘radial velocity spectrometer’ (RVS), converged in late 2002 with the adoption of the instrument baseline. This paper reviews the characteristics of the selected configuration and presents its expected performance. The RVS is a 2.0 × 1.6 degree integral field spectrograph, dispersing the light of all sources entering its field of view with a resolving power R=λ/Δλ= 11 500 over the wavelength range [848, 874] nm. The RVS will continuously and repeatedly scan the sky during the 5‐yr Gaia mission. On average, each source will be observed 102 times over this period. The RVS will collect the spectra of about 100–150 million stars up to magnitude V≃ 17–18. At the end of the mission, the RVS will provide radial velocities with precisions of ∼2 km s−1 at V= 15 and ∼15–20 km s−1 at V= 17, for a solar‐metallicity G5 dwarf. The RVS will also provide rotational velocities, with precisions (at the end of the mission) for late‐type stars of σvsin i≃ 5 km s−1 at V≃ 15 as well as atmospheric parameters up to V≃ 14–15. The individual abundances of elements such as silicon and magnesium, vital for the understanding of Galactic evolution, will be obtained up to V≃ 12–13. Finally, the presence of the 862.0‐nm diffuse interstellar band (DIB) in the RVS wavelength range will make it possible to derive the three‐dimensional structure of the interstellar reddening.