About 50% of the B stars in the SMC cluster NGC 330 were found to be Be stars by Feast (1972). This high fraction, compared to 10 to 20% in the Milky Way, was confirmed by subsequent studies done by Grebel et al. (1992 , 1996), Mazzali et al. (1996), and Keller et al. (1999). A pronounced difference between the Be star content in the Magellanic Cloud clusters and in the Milky Way was also found by Grebel et al. (1994).
The Be phenomenon is closely related to fast rotation. Suggestions were made in the past that Be stars occur near the end of the main sequence (MS) phase; however, extended studies in clusters have shown the occurrence of Be stars from the zero-age sequence to the end of the MS (Mermilliod 1982). Mermilliod has also found that the fraction of Be stars with respect to normal B stars is low in very young clusters, that it reaches a maximum for clusters with turnoff in the range of O9 to B3 and that it declines after that. Evidences for an age effect are also well shown by Fig. 5 from Grebel (1997).
Be stars are generally slightly shifted to the red of the MS band (Grebel et al. 1996; cf. also Fig. 1 below). This effect is consistent with the effect of rotational reddening found by Maeder & Peytremann (1970) and also with the effect of a circumstellar envelope, which contributes to a reddening as well as to the formation of the emission lines (Gehrz et al. 1974, Coté & Waters 1987). A physical model of the radiation-driven wind in rotating stars has been proposed by Bjorkman & Cassinelli (1993). This model, the so-called wind compressed disk (WCD) model, predicts that the isotropic wind particles travel along trajectories that go through the equatorial plane, where shocks occur. The gas is thus confined there and forms a dense equatorial disk. Owocki et al. (1998) have shown that gravity darkening, as predicted by the von Zeipel theorem, leads to a very different wind morphology, where the disk is more difficult to form. The application of the radiative wind theory to rotating stars leads to an expression (cf. Maeder 1999a) for the mass loss rates as a function of the colatitude , this expression shows that there are two main effects influencing the anisotropic mass loss: a) the ""-effect, i.e. the higher gravity and at the pole of a rotating star (due to von Zeipel's theorem) enhances the polar mass loss; b) the "opacity-effect", i.e. the higher opacity, and the higher force multipliers, due to the lower temperature at the equator favour an equatorial ejection and the formation of a ring. The B[e] stars show both polar and equatorial ejections (Zickgraf 1998). The equatorial ejection is strongly favoured by the so-called bi-stability effect (Lamers 1997), i.e. a jump of the opacity near 20'000o K, i.e. close to spectral type B2 where a maximum of the fraction of Be stars is observed.
Models of rotating stars, including hydrostatic distorsion, shear mixing, meridional circulation, enhanced mass loss rates, loss of angular momentum etc. have been constructed recently (Meynet 1999; Maeder 1999b). These models show the high influence of rotation on massive star evolution. The main reason is that shear mixing, which is the most efficient mixing process, is favoured by a high thermal diffusivity as observed in massive stars (Maeder 1997); also, other effects such as meridional circulation (Maeder & Zahn 1998) and anisotropic mass loss are quite significant in massive stars. In view of the large consequences of rotation, it is particularly useful to examine whether the Be fraction is systematically higher in clusters of lower metallicities, a possibility also suspected by Grebel et al. (1992), Grebel (1997) and Mazzali et al. (1996). Sect. 2 establishes the number data of Be stars in an ensemble of clusters of the Galaxy, LMC and SMC. The relation of the results with metallicity is examined in Sect. 3, together with abundance indications related to rotation. Sect. 4 gives the conclusions.
© European Southern Observatory (ESO) 1999
Online publication: May 21, 1999