1. FUV DETECTION OF ETA CARINAE A K. Davidson (University of Minnesota) & N. Smith (University of Colorado) AAS meeting 207 -- January 2006 Summary: A recently reported FUV detection of  Car’s hypothetical companion star almost certainly represents, instead, the primary star’s wind. There’s no clear proof that any observed radiation comes from the elusive secondary object. The same data have at least two positive implications: (1) They imply that the 2003 spectroscopic event was a mass-ejection episode rather than an eclipse. (2) They’re consistent with the polar wind scenario, wherein the pseudo-photosphere is hottest near the equator. Acknowledgments: This work is part of the Hubble Treasury Program for Eta Carinae, supported by funding from STScI. N.S. was supported by a NASA Hubble Fellowship. We are grateful to J.C. Martin and R.M. Humphreys for valuable discussions. The relevance of the symbol in the upper right corner of this poster is hinted at http://www.sixofone.org.uk/. http://www.sixofone.org.uk/ 1. The current problem Recent studies of  Car have concentrated on its yet-unexplained 5.5-year spectroscopic cycle. Two high-priority questions are:  Is this object really a 5.5-year binary system? That seems likely, but proof has not yet materialized. Contrary to many published assertions, so far the observations appear consistent with a single-star thermal / rotational recovery cycle (see refs. below).  “Spectroscopic events” occur at 5.5-year intervals, most recently in mid-2003. If  Car is indeed a binary, do these phenomena signify eclipses, or are they more fundamental? This question is important for astrophysics, because a non-eclipse interpretation implies an undiagnosed instability in the most massive well-studied star. For more information see http://etacar.umn.edu http://etacar.umn.edu and refs. there. * * * Iping et al. [1] recently described FUSE observations of  Car near  1100 Å. They regard these as a direct detection of a hot companion star. Here we note, however, that – (1) much of the detected emission probably originates in the primary star’s wind; (2) there is no clear proof that any of it comes from another star; and (3) the FUSE data are more useful for the second question above. In this poster we mention several concepts that appear quite likely, based on available evidence. In order to show that FUSE detected a second star, one must prove that all the models sketched here are wrong. Fig. 2. Likely intrinsic UV continua of both stars, not corrected for extinction and circumstellar / interstellar absorption lines. “Primary star” means the primary wind, see Fig. 1. In this model more than 60% of the flux near 1100 Å comes from the primary star. 2. The primary star is expected to account for much of the FUV flux As shown by Smith et al. (2003), the polar wind of “  Car A” probably dominates the observed spectrum, and the effective photosphere is located in the wind. ( continued in next column ) …The photosphere is therefore prolate as sketched in Fig. 1, with T eff  14000 K near the poles but T eff > 20000 K near the equator. Some details may eventually prove inaccurate, but this is currently the “best bet” picture in view of existing evidence. (Caveat: The axial ratio shown here is plausible but has not been measured.) As Smith et al. emphasized, the hot equatorial region can produce a substantial far-ultraviolet luminosity. Fig. 2 shows, roughly, the expected UV continua of both stars. Since no realistic wind models are available (see below*), here the primary continuum is a composite of near-zero-gravity Kurucz models with T eff = 14000 K and 20000 K. The resulting energy distribution is good enough for our purposes, since worse uncertainties arise from other causes [2, 3]. The hypothetical secondary continuum in Fig. 2 is a relatively unevolved 35000 K O-type star, normalized so that 5% of the total luminosity occurs at ionizing wavelengths  < 912 Å -- consistent with the relatively weak high-excitation emission lines in the ejecta. * { Spherical wind models ( e.g., [4] ) are unsuitable for the FUV flux, because the radiative transfer problem is very different. Almost any spherical model adjusted to the longer-wavelength spectrum will underestimate the FUV. Even a composite of two or more spherical models shouldn’t be trusted. One can make a rough analogy with gas dynamics, 1-d vs. 2-d. }... The main point of Fig. 2 : Near  1100 Å, we should expect substantial radiation from the primary star, most likely exceeding the hypothetical companion. At least some of the flux detected with FUSE represents “  Car A,” not a companion star. 3. Spectral characteristics in the FUSE data don’t match the expected secondary The FUSE data show a strong and variable N II 1085 emission feature, which Iping et al. attribute to “  Car B”, the hypothetical companion star. That interpretation requires an unusually dense, probably nitrogen-rich wind with lower speeds than expected for the companion. As Ebbets et al. [5] noted, the primary star’s wind is a more likely place to generate the observed N II 1085. The FUSE data indicate wind speeds < 1100 km/s, but the secondary star needs a wind speed close to 3000 km/s to explain the X-rays [6]. Therefore: Even without any other information, the character of the spectrum seen with FUSE makes a secondary-star interpretation very doubtful. 4. A successful prediction The FUV flux practically disappeared during the 2003 spectroscopic event. Based on an eclipse scenario, Iping et al. [1] regard this as proof that the FUV did not come from the primary star. However: The FUV disappearance matched a prediction made long ago by Zanella et al., with no reference to eclipses or a second star [7]. Those authors proposed that  Car’s spectroscopic events are mass-ejection episodes which temporarily quench the far UV. This idea works even better with a polar wind [2], and a companion star might trigger the instability. HST/STIS spectroscopy strongly favors this type of model rather than an eclipse [8]. At least part of the detected emission came from the primary (Fig. 2). In a normal eclipse model, that fraction of the FUV should have remained visible during the event. But it disappeared; the simplest interpretation is a mass-ejection model. This, not a detection of the secondary star, is the most significant result of the FUSE observations. 5. Unorthodox possibilities Detection of the secondary star is highly desirable because that would eliminate single-star models. Unfortunately, as we explained in sections 2—4 above, there is no proof that any emission seen with FUSE came from a second star. It’s not hard to imagine a single-star model. The equatorial photosphere (Fig. 1) may be hot enough to produce all the ionizing photons, and the primary is known to have produced ejecta fast enough to account for the observed X-rays [9]. The 5.5-year period may be a thermal / rotational recovery cycle [10]. We do not strongly advocate such a model, but it remains quite possible. Obviously the topic needs more work! A realistic theoretical analysis may reveal that the HST/STIS and other data are incompatible with a single-star model, but so far this has not been proven. Refs. [1] Iping et al. (2005) Ap.J. 633, L37. [6] Pittard & Corcoran (2002) A&A 383, 636. [2] Smith et al. (2003) Ap.J. 586, 432. [7] Zanella et al. (1984) A&A 137, 79. [3] Davidson et al. (1995) Astron.J. 109, 1784. [8] Martin et al. (2006) Ap.J. in press. [4] Hillier et al. (2001) Ap.J. 553, 837; (2005) preprint. [9] Smith & Morse (2004) Ap.J. 605, 854. [5] Ebbets et al. (1997) Ap.J. 489, L161. [10] Davidson (2005) ASP Conf. 332, p. 101; and (1999) ASP Conf. 179, pp. 304 & 374. In general, see also http://etacar.umn.edu/. http://etacar.umn.edu/.