Based upon observations obtained at the INT and the European Southern Observatory, La Silla, Chile. The INT is operated on the island of La Palma by the ING in the Spanish Observatorio de El Roque de los Muchachos of the Instituto de Astrofísica de Canarias.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

Appendix A in only available in electronic form at http://www.edpsciences.org

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... evolution

Note that, e.g., for O-stars the product of typical mass-loss rate times hydrogen burning life-time is a significant fraction of total mass.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... model-grid

Comprising six supergiants and six dwarfs between 30 000 and 50 000 K, see also http://www.usm.uni-muenchen.de/people/adi/Models/Model.html

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...Walborn (1972)

For a more thorough discussion concerning the problem of distances and magnitudes, we refer the reader to Markova et al. (2003).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... constant

Except for objects which lie close to the Eddington-limit, where the actual value of $R_{\ast }$ has a direct impact on the photospheric structure.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... effect

In order to account for the effects of line-blanketing incorporated in the present work, we have used $T_{\rm rad} \approx 0.9$ $T_{\rm eff}$ , cf. Sect. 7.2.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... $T_{\rm eff}$

This result roughly holds for all spectral types considered here.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...

The reason that model 1 has the lowest emergent flux is given by the fact that for this model the He II continuum becomes optically thick already in the wind, since the He II population is larger there, compared to the other models.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... $T_{\rm rad}$

Actually, this expression needs to be modified by a factor containing certain branching ratios with respect to ordinary and metastable levels, which in the following is of no concern.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... higher

Because of the usual effect that for ground states and close to the surface, $J_\nu > B_\nu(T)$ , and the additional increase of $J_\nu$ due to back-scattered photons from above in case of the blanketed model 1.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... $\log g$

Note that the values of $\log g$ from Paper I include an approximate correction for wind-effects.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... stars

Note that this analysis has been performed with the same code as applied by us, i.e., the results are at least in a differential sense of equal quality.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... simulations

For the "outliers'' around $\log (L/L_{\odot})$ $\approx 5.8$ , the deduced factor might be too large, indicating that these stars are affected by a smaller clumped wind volume than the rest.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.