The radial motions of stars have been studied through spectroscopy since the year 1868 (Hearnshaw 1986). Recently, the accuracies realized in astrometry have enabled such determinations to be made also through purely geometric methods. Once sufficient accuracies are reached, this will enable an absolute calibration of the stellar velocity scale for stationary and variable stars, irrespective of any complexities in their spectra. Indeed, for several early-type stars (with complex spectra smeared by their rapid rotation) the radial velocities already now determined through astrometry are more accurate than has been possible to reach spectroscopically in the past.
The differences between these astrometric radial-velocity values and wavelength measurements of different spectral features may become a new diagnostic tool in probing the dynamic structure of stellar atmospheres. Already the present work has made available quite accurate astrometric radial velocities for stars of many more spectral types than those for which hydrodynamic model atmospheres have been developed (from which, e.g., convective and gravitational wavelength shifts in their spectra could have been predicted). For such stars, the limitations in understanding the differences between astrometric and spectroscopic radial velocities may now lie primarily with spectroscopy and atmospheric modelling, rather than in astrometry.
In this series of three papers, we started by exploring different types of fundamental possibilities of astrometrically determining radial velocities, identifying which methods could be applicable on existing data already today. Among the latter, the moving-cluster method was found capable of yielding astrophysically interesting, sub-km s-1 accuracies, and its mathematical methods were developed in Paper II. In the present paper, data from Hipparcos were used in applying the method to obtain solutions for more than 1000 stars in nearby clusters and associations. Although most of these do not reach the high accuracies realized for the Hyades, they hint at the future potential.
Quantitatively, we have obtained radial velocities with standard errors of
0.6 km s-1 for individual stars in the Hyades.
The accuracies reached begin to make visible the convective and gravitational
shifts expected in the spectra of F and G stars. For A stars and
earlier types, where the convective shifts cannot yet be reliably
predicted from theory, the spectra appear to be blueshifted by a
few km s-1 compared with the astrometrically determined motions
and expected gravitational redshifts.
This illustrates that astrometric radial velocities with
uncertainties even well in excess of 1 km s-1 could be
astrophysically interesting.
Such accuracies may also be sufficient to provide information about the expansion of OB associations, as illustrated by the results for the Sco OB2 complex. Even with the modest precision of existing spectroscopic velocities, we see indications of expansion in the OB associations Upper Centaurus Lupus and Lower Centaurus Crux (causing a bias in the astrometric radial velocities of 5-10 km s-1), while Upper Scorpius surprisingly shows no such indication. The limitations in the present understanding of these associations come not from astrometry but mainly from spectroscopy and theory.
From the same solution that gave astrometric radial velocities, we get kinematically improved parallaxes. These can be used to study in greater detail the spatial structures and the Hertzsprung-Russell diagrams of both clusters and OB associations. In case of the Hyades, Upper Centaurus Lupus and Lower Centaurus Crux, the better-defined main sequences can also be taken as proof of the validity of the kinematic solution, and hence of the astrometric radial velocities.
Hipparcos parallax measurements reached typical accuracies of about 1.5 mas, while our improved parallaxes reach 0.3 mas for the Hyades. Space astrometry missions in the near future are expected to improve this by more than an order of magnitude to about 0.05 mas (Horner et al. 1998), with another order-of-magnitude gain by the future GAIA to 0.004 mas (Perryman et al. 2001). As detailed in Paper I, such accuracies will enable also other methods than the moving-cluster one for determining radial velocities by purely geometric means. The future prospects for studying absolute radial velocities independent of spectroscopy look exciting indeed!
Acknowledgements
This project was supported by the Swedish National Space Board and the Swedish Natural Science Research Council. We want to thank Tim de Zeeuw for providing data on several nearby OB associations before publication, Floor van Leeuwen for providing data on the Pleiades and Praesepe clusters, and the referee, Anthony Brown, for valuable comments.
Copyright ESO 2002