7 Discussion and conclusion

 We have presented an up-to-date and presumably quite complete compilation of WR galaxies from the literature. The number of such objects has considerably increased in the last years and now totals 139 (Table 1). In addition to broad HeII 4686 HeII $\lambda$4686, the basic "classification line'' for WR galaxies, we include, for the first time, relevant information about the presence of various other broad emission lines. In particular the presence of CIV $\lambda$5808 emission originating essentially from WC stars is now detected in many objects.

A large fraction of WR galaxies also show nebular HeII 4686 HeII $\lambda$4686 emission, indicative of high excitation. Where available we also include information about this line. Conversely, few objects are known which show only nebular HeII emission, i.e. no apparent signs of broad stellar emission features. A list of these extra-galactic HII regions is provided in Table 3. This may in particular be used to investigate the possible link between the phenomenon of high excitation and the presence of WR stars as suggested e.g. by Schaerer (1996).

We have also compiled a list of objects suspected to harbour WR stars (Table 4). This could serve for future follow-up spectroscopy.

Most of the work on WR galaxies and related objects discussed above is based on spectroscopy in the visible. Are populations of WR stars also detectable at other wavelengths?

In the infrared no direct signature of WR populations have been detected to the best of our knowledge. The most likely explanations are that 1) many strong features of WR stars coincide with strong nebular lines (e.g. He I 2.06 $\mu$m, Br$\gamma$), and 2) WR features in the IR are strongly diluted by cool stars, which contribute the bulk of the emission at these wavelengths.

The strongest indicator of WR stars in UV is the presence of broad HeII $\lambda$1640 emission which is not seen in emission in other stars. Predictions of its strength and the expected line profile for integrated populations are given by Leitherer et al. (1995) and Schaerer & Vacca (1998). The use of this line is, however, not straightforward for several reasons. Potential difficulties specific to IUE spectra have been discussed by Leitherer et al. (1995). HeII $\lambda$1640 is detected in the average spectra of $\sim$ 20 starbursts (including several WR galaxies) of low and high metallicity of Heckman et al. (1998). In particular the spectra illustrate also the presence of multiple stellar and interstellar absorption lines in this wavelength range, which complicates the quantitative use of this line for diagnostics.

UV high resolution and high S/N spectra obtained with HST indeed show the presence of this emission line in known WR galaxies (e.g. NGC 4214, NGC 1741; Conti et al. 1996; Leitherer et al. 1996). Interestingly these objects show a close resemblance with recently discovered high redshift galaxies (cf. Steidel et al. 1996; Ebbels et al. 1996; Lowenthal et al. 1997). In addition to the strongest stellar UV lines (Si IV, C IV wind lines) the average spectrum of Lowenthal et al. of 11 objects with $z \sim$ 3 also shows the HeII $\lambda$1640 emission line! Quantitative studies of stellar populations including possibly WR stars should be possible in the future using these lines.

The numerous recent findings of massive stars in Seyfert 2 and LINERs, as well as detections of high redshift galaxies exhibiting signatures of massive stars further stress the need to deepen our understanding of the physical processes in "local objects'', and illustrate the interest of studies on massive stars and their interactions with the ISM, stellar populations, and starbursts in a wider context. It is the hope that our compilation will provide a useful basis for such undertakings.

Acknowledgements

We thank Rosa González-Delgado, Daniel Kunth, Cristal Martin, and Bill Vacca for useful communications and discussions, and Peter Conti, Daniel Kunth, Claus Leitherer and Yuri Izotov for comments on an earlier version of the manuscript. Yuri Izotov, Benoit Joguet, Bill Vacca, and Donna Weistrop, kindly provided results before publication. Some comments and corrections were made during the IAU Symposium 193 by several colleagues. We thank them here for their help. This research has made extensive use of the NASA Extragalactic Database (NED), the NASA Astrophysics Data System (ADS) Article Service, and SIMBAD, which is operated by the CDS in Strasbourg, France. DS acknowledges a grant from the Swiss National Foundation of Scientific Research.