A Modern Search for Wolf-Rayet Stars in the Magellanic Clouds. III. A Third Year of Discoveries

TL;DR

This paper reports the discovery of new Wolf-Rayet stars, including a unique WN3/O3 class, in the Magellanic Clouds, expanding the known population and providing insights into their spectral properties and variability.

Contribution

It introduces a new class of Wolf-Rayet stars, WN3/O3, and reports additional discoveries, enhancing understanding of WR star diversity in the Magellanic Clouds.

Findings

Discovered 10 WN3/O3 stars, constituting 6.5% of the LMC's WR population.

Identified spectral variability in a low-mass X-ray binary, RX J0513.9-6951.

Classified three unusual Of-type stars and two rapidly rotating Oef stars.

Abstract

For the past three years we have been conducting a survey for WR stars in the Large and Small Magellanic Clouds (LMC, SMC). Our previous work has resulted in the discovery of a new type of WR star in the LMC, which we are calling WN3/O3. These stars have the emission-line properties of a WN3 star (strong N V but no N IV), plus the absorption-line properties of an O3 star (Balmer hydrogen plus Pickering He II but no He I). Yet these stars are 15x fainter than an O3 V star would be by itself, ruling out these being WN3+O3 binaries. Here we report the discovery of two more members of this class, bringing the total number of these objects to 10, 6.5% of the LMC's total WR population. The optical spectra of nine of these WN3/O3s are virtually indistinguishable from each other, but one of the newly found stars is significantly different, showing a lower excitation emission and absorption spectrum (WN4/O4-ish). In addition, we have newly classified three unusual Of-type stars, including one with a strong C III 4650 line, and two rapidly rotating "Oef" stars. We also "rediscovered" a low mass x-ray binary, RX J0513.9-6951, and demonstrate its spectral variability. Finally, we discuss the spectra of ten low priority WR candidates that turned out not to have He II emission. These include both a Be star and a B[e] star.

Tables

Table 1. Table 1: Newly Found Wolf-Rayet Stars

IDaaDesignation from the current survey. We have denoted the e2v fields with a small “e” to distinguish them from our numbering system from Paper I, i.e., LMCe159 is distinct from LMC159. We plan to impose less idiosyncratic designations once our survey is complete.	${\alpha }_{2000}$	${\delta }_{2000}$	$V$bbPhotometry from Zaritsky et al. 2004.	$B-V$bbPhotometry from Zaritsky et al. 2004.	CT	WN - CT			He ii $\lambda$4686		$M_V$ccWe assume an apparent distance modulus of 18.9 for the LMC, corresponding to a distance of 50 kpc (van den Bergh, 2000) and an average extinction of $A_V=0.40$ (Massey et al., 1995; van den Bergh, 2000).	Sp. Type	Comment
						mag	$\sigma$		$\log$(-EWdd“EW” is the equivalent width, measured in Å.	FWHM (Å)
LMCe078-3	05 41 17.50	$-69$ 06 56.2	17.03	+0.03	17.2	$-0.27$	12.0		1.2	16	$-2.2$eeAssumes an extra 0.3 mag of extinction at $V$ given its $B-V$ color.	WN3/O3
LMCe055-1	04 56 48.72	$-69$ 36 40.3	16.15	$-$0.10	16.4	$-0.17$	8.4		0.8	17	$-2.8$	WN4/O4	OGLE LMC-ECL-3548

Table 2. Table 2: Other Emission-Lined Stars

IDaaDesignation from the current survey. We have denoted the e2v fields with a small “e” to distinguish them from our numbering system from Paper I, i.e., LMCe159 is distinct from LMC159. We plan to impose less idiosyncratic designations once our survey is complete.	${\alpha }_{2000}$	${\delta }_{2000}$	$V$bbPhotometry from Zaritsky et al. 2004.	$B-V$bbPhotometry from Zaritsky et al. 2004.	CT	WN - CT			He ii $\lambda$4686		$M_V$ccWe assume an apparent distance modulus of 18.9 for the LMC, corresponding to a distance of 50 kpc (van den Bergh, 2000) and an average extinction of $A_V=0.40$ (Massey et al., 1995, 2007).	Sp. Type	Comment
						mag	$\sigma$		$\log$(-EWdd“EW” is the equivalent width, measured in Å.)	FWHM (Å)
LMCe078-1	05 37 29.63	-69 14 52.0	13.49	$-0.02$	13.8	$-0.07$	3.3		0.2	15	$-5.4$	O6 Ifc	O5.5 Iaf in Evans et al. (2015)
LMCe078-2	05 37 27.84	-69 23 52.9	13.38	$-0.16$	13.3	$-0.06$	3.2		-0.2	$\cdots$	$-5.5$	O5 nfp
LMCe113-1	04 49 28.01	-67 42 39.8	13.18	$-0.23$	13.2	$-0.09$	4.4		-0.9	$\cdots$	$-5.7$	O7.5nfp
LMCe058-1	05 13 50.80	$-69$ 51 47.6	16.71	$-0.05$	17.1	$-0.35$	15.3		0.7eeThe He ii $\lambda$4686 line consists of a broad component (WN star?) and a narrow component (disk?). The values given in the table are for the broad component. The narrow component has a $\log$(-EW) of 0.8 and a FWHM of 2.8Å. The total $\log$(-EW) of the line is 1.0.	14dd“EW” is the equivalent width, measured in Å.	$-2.2$	LMXB	RX J0513.9-6951

Table 3. Table 3: Stars without He  ii λ 𝜆 \lambda 4686 Emission

IDaaDesignation from the current survey. We have denoted the e2v fields with a small “e” to distinguish them from our numbering system from Paper I, i.e., LMCe159 is distinct from LMC159. We plan to impose less idiosyncratic designations once our survey is complete.	${\alpha }_{2000}$	${\delta }_{2000}$	$V$bbPhotometry from Zaritsky et al. 2002 for SMC members, and Zaritsky et al. 2004 for LMC members.	$B-V$bbPhotometry from Zaritsky et al. 2002 for SMC members, and Zaritsky et al. 2004 for LMC members.	$M_V$ccFor the SMC, we assume an apparent distance modulus of 19.1, corresponding to a distance of 59 kpc (van den Bergh, 2000), and an average extinction of $A_V=0.30$ (Massey et al., 1995, 2007). For the LMC, we assume an apparent distance modulus of 18.9, corresponding to a distance of 50 kpc (van den Bergh, 2000), and an average extinction of $A_V=0.40$ (Massey et al., 1995, 2007).	Sp. Type	Comment
SMCe033-1	0:45:28.28	-73:56:36.6	14.98	$-0.14$	$-4.1$	B0.5 III	[M2002] SMC 5611; B2 (II), Evans et al. 2004
SMCe055-1	0:29:46.91	-73:08:40.6	15.65	$-0.06$	$-3.5$	B1-B1.5 IIIe	Weak Si III, no Mg II. Balmer emission
LMCe005-1	5:31:31.76	-71:56:10.1	16.27	$-0.07$	$-2.6$	B2-V	Si III and Mg II
LMCe027-1	4:50:47.05	-70:37:58.7	15.31	$-0.10$	$-3.6$	B5 III	[M2002] LMC 8808
LMCe029-1	4:59:01.99	-70:48:53.0	15.37	$-0.03$	$-3.5$	B0.5 V[e]	[M2002] LMC 41535; Balmer and Fe II (?) emission
LMCe050-2	5:46:05.71	-69:59:57.2	16.65	$+0.20$	$-2.3$	A0 V:	No He I
LMCe117-1	5:06:13.53	-67:49:11.3	15.96	$+0.09$	$-2.9$	B2 V	Si III and Mg II
LMCe117-2	5:05:43.53	-67:53:53.0	16.85	$-0.05$	$-2.0$	B2 V	Si III and Mg II
LMCe141-1	4:51:19.42	-66:52:18.9	16.90	$+0.03$	$-2.0$	B8 III	Mg II stronger than He I
LMCe155-1	5:03:37.35	-66:33:19.5	16.42	$-0.18$	$-2.5$	B2 V-III	Weak MgII and Si III

Full text

galaxies: stellar content — galaxies: individual (LMC, SMC) — Local Group — stars: evolution — stars: Wolf-Rayet

1 Introduction

For many years, our knowledge of the Wolf-Rayet (WR) population of the Magellanic Clouds (MCs) was considered essentially complete: twelve WRs were known in the SMC (Massey et al., 2003) and 134 were known in the LMC (Breysacher et al., 1999). These stars had been found by a combination of general objective prism surveys, directed searches, and accidental discoveries by spectroscopy (see Massey, 2013; Massey et al., 2015b). However, over the years several additional WRs were found in the LMC, culminating in our own discovery of a very strong-lined WO-type (Neugent et al., 2012b), only the second known example of this rare type of WR in the LMC. This discovery prompted us to begin a multi-year survey of both the SMC and LMC in an effort to obtain a complete census of their Wolf-Rayet population. In part this was motivated in terms of finding a more accurate value for the relative number of WC- and WN-type WRs, as this provides a key test of the evolutionary models (see, e.g., Massey & Johnson 1998, Meynet & Maeder 2005, Neugent et al. 2012a). And, such a survey was timely, given our improved knowledge of the populations of other evolved massive stars in the Magellanic Clouds, such as yellow and red supergiants (Neugent et al., 2010, 2012c), and on-going improvements in massive star models (Ekström et al., 2012; Stanway et al., 2016).

It was entirely possible, of course, that our survey would fail to find much of interest. Instead, in the first year we discovered nine more WRs in the LMC (Massey et al., 2014, hereafter Paper I). More interesting than the numbers, however, were the type of WRs we found: six had spectra that were unlike those of any previously observed. They were also somewhat fainter in absolute visual magnitude ($M_{V}=-2.5$ to $-3.0$) than the previously known WRs. Their emission-line spectra were dominated by N v $\lambda\lambda$4603,19, $\lambda$4945 and He ii $\lambda$4686, and no trace of N iv $\lambda$4058, implying a WR class of WN3. However, they also showed He ii and Balmer absorption spectra with no trace of He i, characteristic of an O3 V star. Yet, these stars were 10$\times$ too faint to be WN3+O3 V binaries, as a typical O3 V star has $M_{V}\sim-5.5$ (Conti, 1988; Walborn et al., 2002). Our preliminary modeling demonstrated that we could reproduce both the emission and absorption lines with a single set of physical parameters (Paper I; Neugent et al., 2015), lending further credence to these being single objects. (Our limited number of repeat observations also failed to detect any radial velocity variations.) In our second year of the survey we detected two more of these stars (Massey et al., 2015a, hereafter Paper II), which we are calling WN3/O3s. In addition, our survey has found the third known WO-type star in the LMC and four other WNs (including two WN+O binaries, an O3.5If*/WN5 star, and a WN11), two rare Of?p stars (magnetically braked oblique rotators; see Walborn et al. 2010a)111See the recent followup study of these two stars plus other members of this class by Nazé et al. (2015), an Onfp star (see Walborn et al., 2010b), a possible B[e]+WN binary, four Of-type supergiants, and a peculiar emission-line star (the nature of which eludes us), and other, more normal, early-type stars (Papers I and II).

Here we report on the third year of discoveries, which include two additional members of the WN3/O3 class in the LMC, bringing the number of this class to 10, 6.5% of the LMC’s WR population, which now stands at a total of 154 stars. One of these WN3/O3 stars is not like the others, in that it shows somewhat lower excitation emission and absorption lines. We report the discovery of three Of stars (all unusual), and revisit a low mass x-ray source previously known to have a very broad He ii $\lambda$4686 component. We also classify 10 low ranked candidates that failed to have He ii $\lambda$4686, including one Be and one B[e] star.

2 Observations and Reductions

The imaging part of our survey is being conducted using the Swope 1-m telescope on Las Campanas. Complete details are given in Papers I and II; here we summarize our observing and reduction procedures. The current e2v camera provides a 29$\farcm$8(EW) $\times$ 29$\farcm$7(NS) field of view with 4110$\times$4096 15$\mu$m pixels, each subtending 0$\farcs$435. Our exposure times are 300 s through each of three interference filters, a WC filter centered on C iii $\lambda$4650, the strongest optical line in WC- and WO-type WRs, a WN filter centered on He ii $\lambda$4686, the strongest optical line in WN-type WRs, and a continuum filter (CT), centered at 4750Å. All three filters have a 50Å bandpass (full width at half maximum, FWHM). We were initially assigned ten nights on the Swope (UT) 2015 Nov 15-24. However, since some of that time was lost due to clouds, Carnegie was kind enough to assign us five additional nights, Dec 25-29, which had been unscheduled. Typical image quality on both runs ranged from 1$\farcs$3-1$\farcs$9. There were 93 fields observed (or reobserved) in the LMC, and 21 in the SMC in total during this third year of our survey. There is about a 1′ overlap between adjacent fields, and so the total new coverage is 19.9 deg2 in the LMC, and 4.5 deg2 in the SMC.

For calibration purposes, ten bias frames were taken daily. When conditions were clear, 3-5 sky flats were obtained in bright twilight through each filter, dithering the telescope between exposures in order to allow us to filter out any stars when we combined frames. Since these twilight exposures were short (a few seconds), it was necessary to correct each of the exposures for the iris pattern of the camera shutter. The details of this procedure are given in Paper II. Since the CCD is read out through four amplifiers, each quadrant is treated separately in the preliminary reductions. Overscan and bias structure (which is practically non-existent) were removed. After these additive corrections were made, the counts in each quadrant were corrected for slight non-linearities in the CCD amplifiers as discussed in Paper II. Each quadrant was flat-fielded, and then the four sections were recombined. Accurate (0$\farcs$5) coordinates are obtained through the use of the “astrometry.net” software (Lang et al., 2010).

The frames were then analyzed by first running aperture photometry on each image to identify stars that were significantly brighter in one of the two emission-line filters than in the continuum image. In addition, image subtraction was performed with the High Order Transform of Point-spread function ANd Template Subtraction (hotpants) software described by Becker et al. (2004). The resultant WC-CT and WN-CT images were examined by eye to identify WR candidates. The combination of the two techniques has proven very effective, as shown in Papers I and II.

This procedure was carried out under some time pressure as we had a single night (UT 11 Jan 2016) allocated on the Baade 6.5-m Magellan telescope for follow-up spectroscopy. Despite the short time between the December imaging and January spectroscopy time, we had successfully reduced and analyzed most of the fields, with 15 WR candidates at various significance levels.

The spectroscopic observations took place with the Magellan Echellette Spectrograph (MagE) mounted on a folded port of the Baade. The instrument is described in detail by Marshall et al. (2008). MagE provides complete coverage from the atmospheric cutoff ($\sim$3200Å) to $\sim$1$\mu$m. We used it with a 1″ slit, which then yielded a resolving power $R$ of 4100. The night was clear, and the seeing was 0$\farcs$6-0$\farcs$7. As we described in Paper II (and in more detail in Massey & Hanson 2013), the pixel-to-pixel sensitivity variations are quite small, and we have found we can achieve better results by not flat-fielding the data in the blue. The data from this night were flat-fielded in the red, primarily to remove fringing. After extraction, wavelength calibration, and fluxing, the individual orders were combined. Our typical signal-to-noise (S/N) in the blue classification region is 100-150 per spectral resolution element.

Our initial spectrum of LMCe058-1 was intriguing but quite noisy, and we were fortunate to be able to squeeze in additional MagE observations of the source on three additional nights throughout the following months, as described further below. Analysis of a few remaining fields not fully analyzed before our January night revealed one other high-significance WR candidate (LMCe113-1), which was observed with MagE on UT 29 March 2016 at the start of the night.

3 Discoveries

Our spectroscopy identified six stars among our candidates that have He ii $\lambda$4686 emission. Ten other candidates proved to be (mainly) B-type stars, not of immediate interest to us, but discussed further below. This success rate is similar to our second year (Paper II). We discuss our “winners” and “losers” below.

3.1 Newly Found Wolf-Rayet Stars

Our most exciting discovery was identifying two new members of the class of WRs that we are calling WN3/O3s (Papers I and II). Other members of this class show the strong N v $\lambda\lambda$4603,19 doublet, N  v $\lambda$4945 and He ii $\lambda$4686 emission lines characteristic of a WN3 star, and a He ii and Balmer line absorption spectrum typical of an O3 star. A composite spectrum can be immediately ruled out as these stars have $M_{V}=-2.3$ to $-3.0$, 15$\times$ fainter than an O3 V would be by itself ($M_{V}\sim-5.5$, Conti 1988) and even slightly fainter than a normal WN3 star would be by itself ($M_{V}\sim-3.8$, Hainich et al. 2014).

We list the properties of these two WN3/O3 stars in Table 1 and illustrate their spectra in Figs. 1 and 2. One thing we have been strongly struck by is how similar all the previous members of this group have been; their spectra are nearly indistinguishable; see, e.g., Fig. 7 in Paper I and Fig. 3 in Paper II. LMCe078-3 is now the ninth example, with a spectrum nearly identical to one of the prototypes of this class, LMC170-2, also illustrated here in Figs. 1 and 2.

However, the spectrum of our newly discovered tenth member of the WN3/O3 class, LMCe055-1, is substantially different. Both the emission and the absorption are indicative of a lower excitation temperature. The spectrum shows N iv $\lambda$4058 emission, which is completely missing from the other members of this group. Furthermore, weak He i $\lambda$4471 is visible in absorption. The N  v $\lambda\lambda$4603,19 doublet lines have P Cygni profiles, with a strong absorption component to the blue side of the emission. We would more properly call this star a WN4/O4 rather than an WN3/O3!

Further complicating the picture is the fact that this star was identified as an eclipsing binary by OGLE with a 2.159074 day period (Graczyk et al., 2011). Might this star simply be a normal WN4+O4 V binary? We can rule this out immediately using the same argument as for “normal” WN3/O3 stars: as shown in Table 1 has an absolute visual magnitude of only $M_{V}=-2.8$, while an O4 V star is expected to have an $M_{V}\sim-5.5$. (Conti, 1988). Furthermore, the OGLE light curve itself is inconsistent with the presence of a massive companion. As shown in Fig. 3 there is no sign of the ellipsoidal variations one sees in massive binaries with similarly short periods, such as DH Cep (Lines et al., 1986)222We are indebted to our colleague Dr. Laura Penny for commenting on the OGLE light curve, and pointing out the implications of the comparison with DH Cep.. We are continuing to investigate this star, including a radial velocity study.

3.2 Other Emission Line Stars

3.2.1 Of-type stars

Our survey is sufficiently sensitive that we detect many Of-type stars. Although such stars are significantly brighter than our WN3/O3s, the equivalent widths (EWs) of their He ii $\lambda$4686 emission are sufficiently small that these are among the hardest stars for us to detect (see, e.g., Fig. 9 in Paper II). Here we discuss the discovery of three of these stars.

The spectrum of LMCe078-1 is shown in Fig. 4. We classify the star as O6 Ifc, where the “c” is required given the strong presence of C iii $\lambda$4650 emission (Walborn et al., 2010a) in addition to the strong N iii $\lambda\lambda$4634,42 and He ii $\lambda$4686 lines that result in the luminosity “If” classification (. Subsequent to our spectroscopy, but prior to this publication, Evans et al. (2015) also classified this star (their star #271) as an O5.5 Iaf, in substantial agreement with our classification here, although the presence of the C iii emission went unnoticed. (The line is only marginally visible in the on-line version of their spectrum, due to their lower S/N.)

Another Of-type star, LMCe-078-2, is shown in Fig. 5. It too is unusual: He ii $\lambda$4686 emission shows a central reversal (absorption), and the absorption lines are extremely broad, with $v\sin{i}\sim$ 385 km s*-1*. These are the classic characteristics of “Oef” stars (Conti & Leep, 1974), also known as “Onfp” stars (Walborn, 1973; Walborn et al., 2010b; Sota et al., 2011). Based upon the relative He ii and He i absorption line strengths, we classify the star as O5nfp. Given the weakness of the emission, it is rather remarkable that we detected this star in the first place. Fortunately the N iii $\lambda\lambda$4634,42 emission doublet is nicely within the bandpass of the WC filter, and so this star was considered of high significance as it was brighter in both emission-line filters relative to the continuum. In addition, Onfp stars are known to have variable He ii strengths, and possibly we detected this star when the line was stronger. We note that this star too may show signs of C iii $\lambda$4650 as well, but the broadness of the lines precludes a definitive assessment.

Our third Of-type star, LMCe113-1, is yet another Onfp star, which we classify as O7.5nfp. As shown in Fig. 6, He ii $\lambda 4686$ has a central reversal. The emission here is quite weak, we are again surprised (but pleased) we detected it; without the N iii $\lambda\lambda$4634,42 emission we probably would not have. The absorption lines are quite broad, with $v\sin{i}\sim$ 300 km s*-1*. The star is identified in the Sanduleak (1970) catalog as Sk$-67^{\circ}$3 according to Brian Skiff’s on-line spectral catalog333http://cdsbib.u-strasbg.fr/cgi-bin/cdsbib?2014yCat….1.2023S, and was classified as B0.5 in Rousseau et al. (1978) based upon low resolution objective prism plates.

3.2.2 RX J0513.9-6951 Revisited

One of the more interesting spectra we’ve encountered is that of LMCe058-1, shown in Fig. 7. The star is a known to be a super-soft x-ray source, RX J0513.9-6951, and its optical spectrum has been previously described by Cowley et al. (1993). The star is variable both in the x-ray region (Cowley et al., 1993) as well as in the optical (Leavitt, 1908; Cowley et al., 1993; Alcock et al., 2003), with a Harvard Variable designation, HV 5682. Cowley et al. (1993) describe the spectrum as typical of that of a low-mass X-ray binary (LMXB), with narrow emission components of He ii and the Balmer lines, along with N iii and C iii and O vi (see also Pakull et al., 1993). The He ii $\lambda$4686 profile clearly has two components, a narrow component and a broad component. Cowley et al. (1993) attributes the broad component to the high velocities of the inner part of the accretion disk. Cowley et al. (1993) and Pakull et al. (1993) both note that the optical spectrum is very similar to that of Cal 83, the prototype of the LMXBs (Crampton et al., 1987).

We reobserved this star as we were intrigued by the description of a broad He ii $\lambda$4686 emission component. Might this be a previously unrecognized Wolf-Rayet star with a disk? Alas, the early descriptions are an accurate match to our optical spectrum; although He ii $\lambda$4686 has a broad component, there are no other WR signatures present. Qualitatively, the optical spectrum is very similar to that described over twenty years ago; compare Fig. 2 in Cowley et al. (1993) to our MagE spectrum in Fig. 7. We obtained four spectra of this star throughout a four month period and found dramatic changes in the emission line equivalent width; for instance, the EW of He ii ranged from -5Å to -25Å. (Cowley et al. 1993 reported an EW of -16Å.) Is this variability due to changes in the emission line fluxes or just to the change in the continuum level affecting the EWs? To answer this, we turned to the fluxed versions of our spectra. As shown in Fig. 8, it is clear that although the continuum flux varies, the emission line fluxes vary even more; the two appear to be correlated, in that the emission lines are their strongest (both in terms of EWs and fluxes) when the star is brightest.

3.3 Non-Emission-Line Stars

In assigning observing priorities, we ranked our candidates on a 1-4 scale, with 1 being the most potentially interesting. All of the emission-lined stars discussed above had been classified as a 1 or a 2. All of our remaining candidates that were observed were ranked as either a 3 or a 4, and none of them proved to have He ii $\lambda$4686 emission. These priority 3-4 candidates were invariably B-type stars (including both a Be and B[e] star) plus one A0 I. Given the large number of stars on our frames, and our desire to be thorough, a few low-significance candidates without emission are expected just on statistical grounds. Although these stars are not of immediate interest to us, we list their properties in Table 3. These stars were classified using the relative strengths of Si iv $\lambda$4089, Si iii $\lambda$4553, Si ii $\lambda$4128. For stars later than B2, Mg ii $\lambda$4481 and He i $\lambda$4471 are secondary indicators. (For more details, see Massey et al. 2016.) In this, we made reference to the Walborn & Fitzpatrick (1990) atlas. A difficulty, particularly for the SMC B stars, is that the metal lines are almost undetectable, and thus our classifications particularly uncertain even at our S/N. We confess to being influenced by our knowledge of the absolute visual magnitudes in assigning luminosity classes.

4 Summary and Future Work

We report here the detection and spectroscopic observations of 16 WR candidates from our late 2015/early 2016 observing season, the third year of our survey. All six of the higher ranked candidates proved to have He ii $\lambda$4686 emission. Of these five, two are members of the WN3/O3 class, bringing to 10 the total number of this class. All ten are in the LMC, and they make up $>$6% of the LMC’s WR known population, which now numbers 154. The spectral features of nine of these WN3/O3 are remarkably homogeneous, but one of the newly found members here is not like the others, showing lower excitation emission spectrum (specifically, N iv $\lambda$4058), P Cygni profiles for the N v $\lambda\lambda$4503,19 emission, and weak He i absorption. Spectral modeling of all of these stars is in progress, with preliminary results reported by Neugent et al. (2015).

In addition, we have found three unusual Of-type stars, one of which is a member of the rare “Ifc” luminosity class showing relatively strong C iii at $\lambda$4650, while two others are rapidly rotating “Oef” type star (now often designated as “nfp.”) We also present modern spectra of the low-mass x-ray binary RX J0513.9-6951, and note that qualitatively they look very similar to the discovery spectra of Cowley et al. (1993) and Pakull et al. (1993). Our repeated observations demonstrate that there are significant changes in the emission-line intensities, unsurprising given the star’s large photometric variations.

Ten of the candidates showed B-type or early A-type spectra; these were all lower ranked candidates and were observed for completeness. One turned out to be a previously unrecognized Be star, and another a newly found B[e] star.

We are now prepared to begin the fourth year of our survey, with 26% of the SMC and 12% of the LMC left to observe. We do not know what our final year of discoveries will bring us, but if past performance is any indication of future results, we are sure to find something interesting!

We thank the referee, Dr. Nolan Walborn, for useful comments which improved this paper. We are grateful for the excellent support we always receive at Las Campanas Observatory, as well as the generosity of the Carnegie Observatory and Steward Observatory Arizona Time Allocation Committees. Support for this project was provided by the National Science Foundation through AST-1008020 and AST-1612874, and through Lowell Observatory. This research has made use of the VizieR catalogue access tool, CDS, Strasbourg, France. The original description of the VizieR service was published in A&AS 143, 23. The Catalog of Stellar Spectral Classification prepared by our colleague at Lowell Observatory, Brian Skiff, proved particularly useful, as always. We also made use of data products from the Two Micron All Sky Survey (2MASS), which is a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center/California Institute of Technology, funded by the National Aeronautics and Space Administration and the NSF. Facilities: Magellan: Baade (MagE spectrograph), Swope (e2v imaging CCD)