1 Introduction

The Hipparcos mission, the first space experiment dedicated to astrometry, was active from November 1989 to March 1993 and gave very precise astrometric data for about 118000 stars. 44% belong to the Hipparcos Survey (Turon et al. 1992) which is complete from V=7.3 to 9.0 as a function of the galactic latitude and the spectral type. The results are published in The Hipparcos Calalogue (ESA, 1997). For the Hipparcos magnitude below Hp=9, the median precision of positions, parallaxes and proper motions are better than 1mas and 1masyr^-1 respectively while the nominal precision was 2mas. In addition the satellite measured solved or suspected double systems as well as photometric variables.

One of the aims of the Hipparcos mission was the study of stellar and galactic dynamics. To compute the components of the spatial velocities, the radial velocities are necessary. Their precision should be consistent with the precision of the tangential velocities $V_\mathrm{t}$ computed from Hipparcos data: the error $\epsilon_{V_\mathrm{t}}$ is 2.1, 2.2 and 3.0 kms^-1 for the distances of 100, 200 and 500pc respectively, with the mean value $V_\mathrm{t} =\,15\,{\rm km\,s}^{-1}$ and an error on the absolute magnitude M_V=0.3.

As early as 1982 Ch. Fehrenbach decided to begin a radial velocities survey for early-type stars observed by Hipparcos. Three instruments were then available and used: the Objective Prism mounted on the Schmidt telescope (SPO), the Petit Prisme Objectif (PPO), and the Marly spectrograph. At that time the Input Catalogue had still not been defined and it was necessary to use a very preliminary version. The aim was to complete the Coravel observations which give radial velocities for spectral types later than F5. So we selected stars from B5 to F5. With the SPO (field $4\times4\deg^{2}$) the regions rich in early-type stars up to the magnitude B=12 were observed, with PPO (same field) regions rich with bright stars up to the magnitude B=8.5 were observed while with the Marly isolated stars were measured up to B=9. The priority was given to Bp-Ap and A-type stars belonging to the Hipparcos Survey, taking into account the radial velocities already published. The results of the SPO are published by Fehrenbach et al. (1990, 1992), those of the PPO by Duflot et al. (1990, 1992, 1995a) and Fehrenbach et al. (1987b, 1997). The Marly observations run from 1983 to 1995 for about 3000 stars. First results are in Fehrenbach et al. (1987a) named PaperI. Since that paper, many measurements were added, the calibration method was improved and the visual spectral classification was revised. So the first published radial velocities were recomputed and are given here for the sake of homogeneity.

The Hipparcos North Survey contains 12286 B5-F5 stars, 33% of which have a published radial velocity in Barbier-Brossat & Figon (1997) or in WEB (Duflot et al. 1995b), the number of measurements being $\geq 2$ without taking into account the above quoted papers based on SPO and PPO observations. For the stars included in the Hipparcos North Survey 2.6% radial velocities have been added by the SPO, 8.4% by the PPO and 18% by the Marly (this paper). Finally 62% of the B5-F5 stars of the North Survey have a measured radial velocity. In addition, some B2-B4 stars, some stars not in the Hipparcos Survey as well as some stars not in the Hipparcos Catalogue (in 1982 the Hipparcos Input Catalogue had not been determined) were observed with the Marly. Some stars with Coravel radial velocities were added to calibrate our measurements. Moreover a certain number of selected F5 stars were seen in fact F6. Finally about 3040 stars were observed with the Marly, 80% not having up to now reliable radial velocity (published value with an error lower than 6.0kms^-1 and a number of measurements $\geq 2$).