Is the massive young cluster Westerlund I bound?

Context. Westerlund I is the richest young cluster currently known in our Galaxy, making it one of the most massive clusters for which we can resolve the individual stars even in the crowded centre. This makes it an ideal target to assess whether massive clusters formed currently will remain bound or will disperse and contribute significantly to the stellar field population. Aims. We measure the radial velocity dispersion of Westerlund I to explore whether the cluster is either currently in virial equilibrium, in the process of collapse, or expanding and dispersing into the field. Methods. We obtained MIKE/Magellan high resolution optical spectra of 22 post main-sequence stars in Westerlund I for 2 or 3 epochs with a maximum baseline of about one year. Radial velocities variations between these spectra are measured by means of cross-correlation. Results. We calculate the velocity dispersion from the cross-correlation of five yellow hypergiants and one luminous blue variable, that show little radial velocity variations between epochs and have many spectral features in common. After taking into account the effect of small number statistics and undetected binaries, we estimate the velocity dispersion for the massive stars in Westerlund I to be 2.1(-2.1)(+3.3) km s(-1). For several different assumptions concerning possible mass segregation and the elongation of the cluster, we find that Westerlund I is subvirial at the 90% confidence level. Conclusions. We can rule out that the cluster is significantly supervirial at the 97% confidence level, indicating that Westerlund I is currently bound. This implies that Westerlund I has survived past the point where any gas expulsion has taken place and is expected to survive for billions of years.