WHAT ARE THE MASS-LOSS RATES OF O STARS

Empirical mass-loss rates of 28 luminous galactic OB stars are derived from thermal radio emission and from Ha recombination radiation. The velocity fields of all stars had previously been analyzed in the IUE ultraviolet spectral region. The results of these ultraviolet line-profile analyses (shape of the velocity law and terminal velocity) are combined with the radio and Halpha data for a precise and consistent determination of the stellar mass-loss rates. It is found that radio and Halpha rates agree within the observational errors. This suggests that significant clumping in the wind is unlikely due to the very different radius of origin of radio and Halpha radiation within the wind. Comparison of the empirical mass-loss rates with values derived theoretically from the theory of radiatively driven winds demonstrates that theoretical mass-loss rates are lower on the average by 0.29 dex. Similarly, theoretical terminal velocities are higher by 40%. By combining the observed wind velocities with the mass-loss rates we show that the empirical momentum fluxes are also higher by 0.17 dex than the prediction of the theory. The theoretical momentum fluxes are independent of the stellar masses. Therefore the momentum (and also the mass-loss and terminal velocity) discrepancy cannot be explained by systematic errors in the adopted stellar masses. We find that the discrepancy between theoretical and empirical mass-loss rates increases with increasing wind density. By including a sample of WNL stars with weak winds in our sample, we show that the most luminous 0 stars form a natural extension of the least extreme WNL stars with respect to the momentum transfer rate to the stellar winds. It is suggested that the '' momentum problem '' observed in Wolf-Rayet stars is already present in a less severe form in luminous 0 stars. This can provide an explanation for the mass-loss and terminal velocity discrepancy in hot stars.