The progenitor masses of Wolf-Rayet stars and luminous blue variables determined from cluster turnoffs. II. Results from 12 Galactic clusters and OB associations

In a previous paper on the Magellanic Clouds, we demonstrated that coeval clusters provide a powerful tool for probing the progenitor masses of Wolf-Rayet (W-R) stars and luminous blue variables (LBVs). Here we extend this work to the higher metallicity regions of the Milky Way, studying 12 Galactic clusters. We present new spectral types for the unevolved stars and use these, plus data from the literature, to construct H-R diagrams. We find that all but two of the clusters are highly coeval, with the highest mass stars having formed over a period of less than 1 Myr. The turnoff masses show that at Milky Way metallicities some W-R stars (of early WN type) come from stars with masses as low as 20-25 M. Other early-type WN stars appears to have evolved from high masses, suggesting that a large range of masses evolve through an early WN stage. On the other hand, WN7 stars are found only in clusters with very high turnoff masses, over 120 M.. Similarly, the LBVs are only found in clusters with the highest turnoff masses, as we found in the Magellanic Clouds, providing very strong evidence that LBVs are a normal stage in the evolution of the most massive stars. Although clusters containing WN7 stars and LBVs can be as young as 1 Myr, we argue that these objects are evolved, and that the young age simply reflects the very high masses that characterize the progenitors of such stars. In particular, we show that the LBV eta Car appears to be coeval with the rest of the Trumpler 14/16 complex. Although the WC stars in the Magellanic Clouds were found in clusters with turnoff masses as low as 45 M., the three Galactic WC stars in our sample are all found in clusters with high turno+ masses (> 70 M.); whether this difference is significant or due to small number statistics remains to be seen. The bolometric corrections of Galactic W-R stars are hard to establish using the cluster turnoff method but are consistent with the standard model of Hillier.