We have studied the properties of the winds of about 40 O and B supergiants in the northern emisphere, through the observation and the analysis of their H_alpha lines.

The observations were made using the echelle spectrograph of the Catania Astrophysical Observatory. The observed profiles were fitted using a code that models the emission from the star and the wind itself. For each star in the sample we derived the mass loss rate and the wind velocity field.

We then analyzed the correlations between the wind properties and the fundamental physical parameters of the stars (i.e. luminosity, gravity etc.) to obtain information on the mechanisms which are responsible for the acceleration of the stellar outflow. Our results can be summarized as follows:

1) In O and B supergiants, the mass loss rate depends only on the bolometric luminosity of the star. Considering also stars in different evolutive stages (i.e. giants or main sequence stars), we find that there must be an additional parameter that influences the mass loss rate, as indicated by the existence of parallel sequences in the log{dot{M}} - log{L} diagram for different luminosity class stars.
2) The wind velocity structure is strongly correlated with the radiation field of the star. In particular, the wind initial velocity increases with the effective temperature and with the escape velocity. The kinetic energy transferred by the radiation field to the wind material increases linearly with the gravitational potential of the star so that the fraction of the gravitational potential needed to accelerate the wind is the same regardless of the spectral type.
3) Single and/or multiple scattering of UV radiation by (mostly) resonance lines is able to provide the wind with the impulse necessary to its acceleration. In particular, single scattering alone can carry out this task for stars whose temperature is less than 25000 K. On the other hand, as temperature increases, multiple scattering plays a role more and more important, becoming the dominant acceleration mechanism for temperatures higher than 40000 K.

A subsample of the stars from our H_alpha survey was also observed at radio wavelengths using the VLA. The rationale for this additional effort is that a comparison of the results obtained for a sizeable sample of stars observed both in H_alpha and in radio continuum can provide a fundamental consistency check for the two methods. This is particularly important for an ideal use of the H_alpha method every time the radio cannot give useful information (e.g. strong non thermal contribution, or flux density too low because of either a low mass loss rate or a far distance). Furthermore, since radio and H_alpha emission originate from different parts of the wind, the two independent determinations of the mass loss can provide valuable information about the detailed structure and the symmetry of the expanding envelope.

This approach has proved to be very successful. In fact, out of 19 O and B type supergiants selected and observed, we detected 14 sources, 9 of them for the first time. The main results of our radio study can be summarized as follows:

1) The radio emission turned out to be of thermal origin for all objects detected at least at two frequencies but one (8 out of 9 stars).
2) The comparison of the radio derived mass loss rates with the ones derived through the analysis of the H_alpha line showed a good agreement between the two methods, with the H_alpha method giving more accurate values of the mass loss rate.
3) The radio emission spectral index turned out to be equal to the canonical value of 0.6 only for one object. All the other objects had a higher spectral indexes, whose mean value is 0.82 with a rms dispersion of ± 0.13. Deceleration at large distances or the presence of a temperature gradient may explain this behaviour.
4) The relationship dot{M} - L for O and B supergiants obtained combining the H_alpha and radio results turns out to be appreciably flatter than commonly reported, i.e. log{dot{M}} = (1.1± 0.1) log{L}.