1 Introduction

Study of Wolf-Rayet (WR) star populations are important not only for our knowledge of the nature of the WR stars themselves but also for the invaluable information they give about their host galaxies and about galactic and stellar evolution in general. The technological advances of the recent years allow to search much deeper in the absorbed regions of our own Galaxy as well as in an increasing number of external galaxies (Massey 1996 and references therein, Testor et al. 1996).

With a few exceptions (Corso 1975; Armandroff & Massey 1985), the various surveys conducted so far consist of two-step procedures. The first one is a search for WR candidates by comparison of images obtained through various filters (Schild & Testor 1991; Shara & Moffat 1982; Shara et al. 1991) or by slitless spectroscopy (Azzopardi & Breysacher 1979ab; Morgan & Good 1985; Westerlund et al. 1983; Testor et al. 1996). The second is a spectroscopic study to precise the types and subtypes of the candidates. However, spatial crowding rules out spectroscopic analysis in the densest regions of clusters and remote galaxies. Clearly, these important fields of research are not compatible with a spectroscopic follow-up.

Using narrow-band filters, Corso (1975) was the first to achieve a differentiation between WN and WC spectral types (Massey & Conti 1983a). Armandroff & Massey (1985) improved the choice of narrow filters and used them to detect WR stars in Local Group galaxies and to separate the WN stars from the WC stars. One of their figures indicates that their choice of filters also allows some rough selectivity between WNL and WNE but no selectivity among the WC subtypes.

In this paper we present the first results of a five-filter photometric system whose physical content is high enough to allow meaningful comparisons with population modeling in the sense that it allows not only a separation between WN and WC spectral types, but also a discrimination at the level of subtypes, mainly for the WN subtypes.