2.2.3. Beyond the Magellanic Clouds

The first WR stars to be discovered beyond the Magellanic Clouds were in the nearby spiral galaxy M33. In 1972, James Wray and George Corso pioneered the interference-filter method of searching for WRs by comparing images of M33 taken through an interference filter centered on the C III *λ*4650 and He II *λ*4686 emission complex with that of a continuum image [52]. WR candidates would stand out by being brighter in the on-band compared to non-WR stars in the field. Their paper contained spectroscopic confirmation of two of their 25 candidates (thanks to Roger Lynds); both stars were of WC-type, although Lindsey Smith is quoted as saying that the spectra were "not quite like any I have seen from either the Galaxy or the Magellanic Clouds." (This was probably more due to the poor quality of these early spectroscopic efforts on these faint objects, which pushed the limits of photographic

spectroscopy at that time.) Spectroscopy of three other candidates followed five years later by Alex Boksenberg, Allan Willis, and Leonard Searle using one of the first digital photon-counting systems [53]. A search using photographic "grism" imaging on the Kitt Peak 4-meter (a technique similar to objective prism survey but using a grating prism and a much larger telescope) carried out by Bohannan, Conti, and Massey revealed a host of H II regions in M33, but only five more WRs [54]. Spectroscopy of the stars in M33's H II regions by Conti and Massey in 1981 was more effective, identifying 14 more WRs [55]; some were in common with the nearly contemporaneous study of the stellar content of NGC 604, the largest H II region in M33, by Mike Rosa and Sandro D'Odorico [56,57]. The properties of some of these stars were highly unusual, with higher luminosities and more hydrogen than normal WR stars, similar to what would be eventually noted in the R136 cluster as mentioned above. A photographic search with the 3.6-meter Canada–France–Hawaii telescope with followup spectroscopy on the Kitt Peak 4-meter provided the first galaxy-wide survey, including 41 newly found WRs [58]. This 1983 Massey and Conti catalog included all previous known WRs, for a total of 79 WRs, and revealed a trend in the relative number of WCs to WNs as a function of galactocentric distance within M33. Quantitative analysis of the lines (measurements of line strengths and widths) and absolute magnitudes showed no gross differences between the M33 WRs and those of the Milky Way or Magellanic Clouds [58,59], refuting the Smith's first impression from the Lynds' earlier spectroscopy.

The first use of CCDs to survey for WRs was carried out by Taft Armandroff and Massey in 1985 using the newly implemented prime-focus CCD camera on the Cerro Tololo Blanco 4-meter telescope [60]. They had refined the interference-filter method to include a three-filter system, with one centered on C III *λ*4650, another on He II *λ*4686, and a third on neighboring continuum, and used these with a CCD to search for WRs in the dwarf galaxies IC 1613 and NGC 6822, as well as two M33 test fields. One WR star had been previously identified in IC 1613, a WC star (now considered a WO) discovered in an H II region by D'Odorico and Rosa in 1982 [61], and subsequently studied by Kris Davidson and Tom Kinman [62]. Similarly a WN-type WR had previously been found in NGC 6822 by Westerlund and coworkers using an objective prism [63]. These early CCDs were incredibly tiny compared to what are in use today, and multiple fields were needed to cover even these relatively small galaxies. These CCDs were also incredibly noisy (with read-noise of 100 e- compared to typically 3 e- today). Armandroff and Massey found 12 "statistically significant" WR candidates in NGC 6822 and 8 in IC 1613. However, only four of the NGC 6822 WR candidates proved to be real (including the one that was previously known), and the only IC 1613 WR candidate that checked out was the one already known [64].

A search for WR stars in the dwarf galaxy IC 10 proved the most surprising of any of these early studies. Despite its small size, 16 WR candidates were initially found by Massey, Armandroff, and Conti [65], 15 of which were quickly confirmed [66], causing the authors to recognize this as the nearest starburst galaxy. Despite the galaxy's low metallicity, the relative proportion of WC stars was very large. Was this suggestive of a top-heavy initial mass function as has been historically suggested for other starbursts [67], or is indicative that an even larger number of WRs (predominantly WN) remained to be discovered, as suggested by [68]? This issue is still not settled. The current count is 29 spectroscopically confirmed WRs [69], with additional candidates still under investigation.

The situation for M31 was probably the worst. Interference photography by Tony Moffat and Mike Shara identified a few of the strongest-lined WRs [70,71]; CCD imaging through interference filters by Massey and collaborators went much deeper but covered only a small portion of the galaxy [64,72].

These early studies culminated in the 1998 paper by Massey and Olivia Johnson [73], who identified additional M33 WR stars found using a larger format (and less noisy) CCD, and provided a catalog of all of the known extragalactic WR stars beyond the Magellanic Clouds. For the purposes of this review, we will consider that the end of the "early era" of WR searches. Although completeness indeed would prove to be a problem, the following facts had emerged:

• The WC/WN ratio appeared to be strongly correlated with metallicity, with the exception of the starburst galaxy IC 10.

