0709.0097/aa.tex
1: \documentclass[oldversion]{aa}  
2: %\documentclass[oldversion,referee]{aa}
3: %
4: \usepackage{graphicx}
5: \usepackage{txfonts}
6: %
7: \def\tmap{\textsc{\bfseries tmap}}
8: \def\pro2{\textsc{\bfseries pro2}}
9: \def\lte2{\textsc{\bfseries lte2}}
10: \def\atoms2{\textsc{\bfseries atoms2}}
11: \def\setf2{\textsc{\bfseries setf2}}
12: \def\line1prof{\mbox{\textsc{\bfseries line1}\raisebox{-0.5ex}{\bfseries--}\textsc{\bfseries prof}}}
13: \def\ironic{\textrm {\bfseries IrOnIc}}
14: \def\etal{{et\,al.}\ }
15: \def\elf{PG\,1159$-$035}
16: \def\keins{K1-16}
17: \def\hh{H1504$+$65}
18: \def\rxj{RX\,J2117.1$+$3412}
19: \def\ngc{NGC\,246}
20: \def\lovier{Longmore\,4}
21: \def\kpd{KPD\,0005+5106}
22: \def\sand{Sand~3}
23: \newcommand{\Teff}{$T\mathrm{\hspace*{-0.4ex}_{eff}}$\,}
24: \newcommand{\logg}{$\log\,g$\hspace*{0.5ex}}
25: \newcommand{\Msol}{M\hspace*{0.2ex}$_\odot$}
26: \newcommand{\Lsol}{L\hspace*{0.2ex}$_\odot$}
27: \newcommand{\gppr}{\stackrel{>}{\scriptstyle \sim}}
28: \newcommand{\gappr}{\raisebox{-0.4ex}{$\gppr $}}
29: \newcommand{\lppr}{\stackrel{<}{\scriptstyle \sim}}
30: \newcommand{\lappr}{\raisebox{-0.4ex}{$\lppr $}}
31: %\newcommand{\Ion}[2]{\mbox{#1\,{\scriptsize #2}}}
32: %
33: \begin{document}
34: %
35: \title{Identification of \ion{Ne}{viii} lines in  H-deficient (pre-)
36:   white dwarfs: a new tool to constrain the temperature of the hottest
37:   stars\thanks{Based on observations made with the NASA-CNES-CSA Far
38:   Ultraviolet Spectroscopic  Explorer. FUSE is operated for NASA by the
39:   Johns Hopkins University under NASA contract NAS5-32985.}  \thanks{
40:   Some of the data presented in this paper were obtained from the
41:   Multimission Archive at the Space Telescope Science Institute
42:   (MAST). STScI is operated by the Association of Universities for
43:   Research in Astronomy, Inc., under NASA contract NAS5-26555. Support
44:   for MAST for non-HST data is provided by the NASA Office of Space
45:   Science via grant NAG5-7584 and by other grants and contracts.} }
46: 
47: \author{K\@. Werner\inst{1}
48:    \and T\@. Rauch\inst{1}
49:    \and J.~W\@. Kruk\inst{2}}
50: 
51: \institute{Kepler Center for Astro and Particle Physics, 
52: Institut f\"ur Astronomie und Astrophysik, Universit\"at T\"ubingen, Sand 1, 72076 T\"ubingen, Germany
53:    \and Department of Physics and Astronomy, Johns Hopkins University, Baltimore, MD 21218, USA}
54: 
55: \offprints{K\@. Werner\\ \email{werner@astro.uni-tuebingen.de}}
56: 
57: 
58: \date{Received; accepted}
59: 
60: \authorrunning{K. Werner \etal}
61: \titlerunning{\ion{Ne}{viii} in the hottest hydrogen-deficient (pre-) white dwarfs}
62: 
63: \abstract{For the first time, we have identified  \ion{Ne}{viii}
64: absorption lines in far-UV spectra of the hottest known
65: (\Teff$\gappr$150\,000~K) hydrogen-deficient (pre-) white dwarfs of
66: spectral type PG1159. They are of photospheric origin and can be matched
67: by synthetic non-LTE line profiles. We also show that a number of UV and
68: optical emission lines in these stars can be explained as being
69: photospheric \ion{Ne}{viii} features and not, as hitherto suspected, as
70: ultrahigh ionised \ion{O}{viii} lines created along shock-zones in the
71: stellar wind. Consequently, we argue that the long-standing
72: identification of the same emission lines in hot [WR]-type central stars
73: as being due to ultrahigh-ionised species (\ion{O}{vii-viii}, \ion{C}{v-vi})
74: must be revised. These lines can be entirely attributed to thermally excited
75: species (\ion{Ne}{vii-viii}, \ion{N}{v}, \ion{O}{vi}).  Photospheric
76: \ion{Ne}{viii} lines are also identified in the hottest known He-rich
77: white dwarf (\kpd), some of which were also attributed to \ion{O}{viii}
78: previously.  This is a surprise because it must be concluded that \kpd\ is
79: much hotter (\Teff$\approx$200\,000~K) than hitherto assumed
80: (\Teff$\approx$120\,000~K). This is confirmed by a re-assessment of the
81: \ion{He}{ii} line spectrum. We speculate that the temperature is high
82: enough to explain the mysterious, hard X-ray emission (1~keV)
83: as being of photospheric origin.}
84: 
85: \keywords{Stars: abundances -- 
86:           Stars: atmospheres -- 
87:           Stars: evolution  -- 
88:           Stars: AGB and post-AGB --
89:           White dwarfs}
90: 
91: \maketitle
92: %
93: %________________________________________________________________
94: 
95: \begin{figure*}[tbp]
96: \begin{center}
97: \includegraphics[width=0.85\textwidth]{fig_all.ps}
98:   \caption[]{\emph{Left panel}: Identification of \ion{Ne}{viii} lines in the FUSE spectra
99:     of PG1159 stars and the DO white dwarf \kpd. Several lines of the
100:     $n=5 \rightarrow 6$ transition are detected, as labeled in detail at
101:     the uppermost spectrum. Overplotted are computed profiles with
102:     \Teff\ and $\log\,g$, as indicated. \emph{Middle and right
103:     panels}: Optical spectral regions where we identified the \ion{Ne}{viii}
104:     $n=8 \rightarrow 9$ and $n=9 \rightarrow 10$ transitions. All observed
105:     spectra are shifted such that the photospheric lines appear at their
106:     rest wavelengths.}
107:   \label{fig_all}
108: \end{center}
109: \end{figure*}
110: 
111: \begin{figure*}[tbp]
112: \begin{center}
113: \includegraphics[width=0.85\textwidth]{fig_rxj2117_details.ps}
114:   \caption[]{Details of other spectral regions of the PG1159-type
115:     central star \rxj\ and the DO white dwarf \kpd\ displaying  \ion{Ne}{viii} lines. \emph{Panels a)--c)} show
116:     $n=5 \rightarrow 6$ and $n=6 \rightarrow 8$ transitions (the blending
117:     \ion{C}{iv} line in panel \emph{a)} is not included in the model). \emph{Panel d)}
118:     shows a barely detectable $n=6 \rightarrow 7$ emission feature in the HST/FOS spectrum of \kpd, which
119:     was previously assigned to \ion{O}{viii}. \emph{Panel e)} displays HST/GHRS
120:     spectra with a \ion{Ne}{viii}  $n=7 \rightarrow 8$ emission line that
121:     was also thought to stem from \ion{O}{viii}. The adjacent emission
122:     feature in \kpd\ is from nitrogen; it is not included in the model.
123: }
124:   \label{fig_rxj2117_details}
125: \end{center}
126: \end{figure*}
127: 
128: 
129: 
130: \section{Introduction}
131: \label{intro}
132: 
133: Observations of extremely hot post-AGB stars (\Teff$\gappr 100\,000$~K)
134: with the \emph{Far Ultraviolet Spectroscopic Explorer} (FUSE) have led
135: to the discovery of many unidentified spectral lines. Their
136: identification is difficult, mostly because of the lack of accurate
137: atomic data for highly ionized elements. On the other hand, their
138: identification is rewarding because they are often the only features by
139: which particular species are accessible for abundance
140: determinations. Such species can be used to test stellar evolutionary
141: models. For example, several lines of fluorine (\ion{F}{v-vi}), neon
142: (\ion{Ne}{vii}), silicon (\ion{Si}{v}) and argon (\ion{Ar}{vii}) were
143: identified for the first time (Jahn \etal 2007; Werner \etal 2004a, 2005,
144: 2007). In particular, the \ion{Ne}{vii} $\lambda~973$~\AA\ line occasionally
145: displays a prominent P~Cygni profile, which is an important
146: tool for stellar wind analyses (Herald \etal 2005).
147: In this paper, we report the identification of a number of
148: \ion{Ne}{viii} lines in the FUSE spectra of the hottest known
149: hydrogen-deficient (pre-) white dwarfs. We show that the mere occurrence
150: of such highly ionized neon lines puts a strict lower limit on the
151: effective temperature ($\approx 150\,000$~K).
152: 
153: \ion{Ne}{viii} is a one-valence electron, lithium-like ion with a
154: relatively simple  energy-term structure, however, a detailed
155: compilation of accurate level energies and line wavelengths, which is a
156: prerequisite of any quantitative work, became available only recently
157: (Kramida \& Buchet-Poulizac 2006). Our model atmosphere and spectrum
158: synthesis calculations predict the presence of \ion{Ne}{viii} lines in
159: other wavelength regions, too. As a surprising result, we find that all
160: previous identifications of ultra-high ionization (i.e., non-thermally
161: excited) oxygen lines (\ion{O}{viii}) in UV and optical spectra of
162: PG1159 stars, and the hottest known DO white dwarf, are wrong. We find
163: that, instead, these lines are due to photospheric \ion{Ne}{viii}. The
164: earlier identification of these lines as \ion{O}{viii} was motivated by
165: the supposed occurrence of the same lines in early-type Wolf-Rayet
166: central stars (i.e., spectral type [WCE]) and it was argued that they
167: are formed in shocked wind regions. The results presented here provide
168: evidence that also in [WCE] stars these, as well as other lines that were
169: assigned to ultrahigh-ionized C and O, do probably stem from thermally
170: excited neon.
171: 
172: We present our observations and line identifications in
173: Sect.\,\ref{observations} and describe our modeling in
174: Sect.\,\ref{modeling}. The results from line profile fits to individual
175: objects are presented in Sect.\,\ref{results}. Implications for [WCE]
176: stars are discussed in Sect.\,\ref{sect_wc} and we conclude with
177: Sect.\,\ref{conclusions}.
178: 
179: \begin{table}
180: \begin{center}
181: \caption{Effective temperature and surface gravity of the program stars as taken from the
182: literature. As described in the text, the discovery of \ion{Ne}{viii} lines
183: suggests that \Teff\ for \kpd\ and \lovier\ must be significantly higher. \label{tab_stars}}
184: \begin{tabular}{lcccc}
185:       \hline
186:       \hline
187:       \noalign{\smallskip}
188: Object  & Spectral &\Teff & \logg & Reference\\
189:         & Type     & [K]  & (cgs) &    \\
190:       \noalign{\smallskip}
191:       \hline
192:       \noalign{\smallskip}
193: \hh      & PG1159  & 200\,000 & 8.0 & A\\
194: \rxj     & PG1159  & 170\,000 & 6.0 & B\\
195: \ngc     & PG1159  & 150\,000 & 5.7 & C\\
196: \keins   & PG1159  & 140\,000 & 6.4 & D\\  
197: NGC\,2371& [WCE]   & 135\,000 & 5.5 & E\\
198: \lovier  & PG1159  & 120\,000 & 5.5 & F\\
199: \kpd     & DO      & 120\,000 & 7.0 & G\\
200:  \noalign{\smallskip}
201:       \hline
202:      \end{tabular}
203: \end{center}
204: References: 
205: A: Werner \etal 2004b,
206: B: Werner \etal 1996b,
207: C: Rauch \& Werner 1997,
208: D: Werner \etal 1992,
209: E: Herald \& Bianchi 2004,
210: F: Rauch \& Werner 1997,
211: G: Werner \etal 1994
212: \end{table}
213: 
214: \section{Observations and \ion{Ne}{viii} line identifications}
215: \label{observations}
216: 
217: FUSE observations and data reduction for most of our program stars
218: were described in our previous work (Werner \etal 2004a, 2004b, 2005,
219: 2007). Table~\ref{tab_stars} lists our program stars with references to
220: results from previous analyses. The FUSE instrument consists of four
221: coaligned telescopes, each with a prime-focus spectrograph. Descriptions
222: of the FUSE instrument, and channel alignment and wavelength calibration
223: issues, are given by Moos \etal (2000) and Sahnow \etal (2000). The FUSE
224: spectra cover the wavelength range from the Lyman edge up to 1187~\AA\
225: with a spectral resolution of 0.05~\AA. For spectra that are too faint
226: to permit co-alignment of individual exposures, the spectral resolution
227: may be degraded to 0.1~\AA.  The S/N ratio in the LiF2a spectra in the
228: vicinity of the \ion{Ne}{viii} lines being studied here ranges from a
229: low of 17:1 per 0.05~\AA\ resolution element for \lovier\ to 70:1 per
230: resolution element for \rxj; it was greater than 45:1 for the other
231: objects.  The S/N ratio in the LiF1b spectra at these wavelengths was
232: typically $\sim$30\% lower.  UV spectra of \kpd\ and \rxj,
233: taken with the FOS and GHRS spectrographs aboard the \emph{Hubble Space
234: Telescope}, were retrieved from the MAST archive.  High-resolution
235: optical spectra of \keins, \rxj, \kpd, and \hh\ were obtained at the
236: \emph{Keck} observatory and the HIRES spectrograph. For details on data
237: reduction see Zuckerman \& Reid (1998). Spectra of \ngc\ and \lovier\
238: were obtained with the ESO \emph{Very Large Telescope} and the UVES
239: spectrograph, in the framework of the SPY project (Napiwotzki \etal 2003).
240: 
241: The starting point of our analysis was the identification of two strong
242: and broad absorption features of \ion{Ne}{viii} at
243: $\lambda=1162-1165$~\AA\ in the FUSE spectra of five PG1159 stars and
244: the DO white dwarf \kpd\ (left panel of Fig.\,\ref{fig_all}). They are
245: blends of several  $n=5\rightarrow 6$ lines with high angular quantum
246: number $l$. \ion{Ne}{viii} lines are not detected in the FUSE spectra of
247: any other PG1159 star, indicating that the minimum \Teff\ for exhibiting
248: these lines is around 150\,000~K (this limit will be assessed more closely
249: below). In addition to these lines, several weaker \ion{Ne}{viii} lines
250: are detectable in the FUSE spectra of some objects, particularly in
251: \rxj\ (panels \emph{a)--c)} in Fig.\,\ref{fig_rxj2117_details}). These
252: are low-$l$ $n=5\rightarrow 6$ lines and an isolated
253: $n=6\rightarrow 8$ line. The strongest, high-$l$,  $n=6\rightarrow 7$
254: transitions are located at $\lambda~1932$~\AA. We found only one
255: archival spectrum covering this wavelength position, namely an HST/FOS
256: observation of \kpd. An emission feature is barely detectable at this
257: location (panel \emph{d)} of Fig.\,\ref{fig_rxj2117_details}), 
258: formerly attributed to \ion{O}{viii} $n=6\rightarrow 7$ (Sion \&
259: Downes 1992). We note that the line positions for \ion{O}{viii} and
260: \ion{Ne}{viii} between levels with these high principal quantum numbers
261: become indistinguishable. Both ions have the same core charge but their
262: valence electrons are single $1s$ and $2s$ electrons,
263: respectively. Based on our model computations described below, we
264: attribute the previously discovered emission features at
265: $\lambda\lambda~2977, 4340$, and 6068~\AA\ (Figs.\,\ref{fig_rxj2117_details}
266: and \ref{fig_all}) to the $n=7\rightarrow 8$, $n=8\rightarrow 9$, and
267: $n=9\rightarrow 10$ transitions of \ion{Ne}{viii}, respectively. They,
268: too, were previously thought to stem from \ion{O}{viii} (Werner \& Heber
269: 1992; Werner \etal 1996). The $\lambda~6068$~\AA\ emission line was also
270: detected in the very hot PG1159 stars Longmore~3 and HE\,1429$-$1209
271: (Werner \etal 1994, 2004a).
272: 
273: Table~\ref{tab} summarizes all \ion{Ne}{viii} lines that were detected in
274: any of the examined objects. The NIST\footnote{http://physics.nist.gov/
275: PhysRefData/ASD/index.html} wavelengths coincide with the photospheric
276: rest wavelengths, with the exception of the $5p\rightarrow 6d$,
277: $5f\rightarrow 6d$, and $6p\rightarrow 8d$ transitions. Their observed
278: wavelengths are smaller by 0.16, 0.09, and 0.07~\AA, respectively. This
279: is corrected for in panels \emph{b)} and \emph{c)} of
280: Fig.\,\ref{fig_rxj2117_details} and in Table~\ref{tab}.
281: 
282: We note that we have not detected any  \ion{Ne}{viii} lines in the
283: hottest (\Teff$\approx$100\,000~K) hydrogen-rich central stars or DA
284: white dwarfs from which FUSE spectra are available. The reason is that
285: their temperature is not sufficiently high (all have
286: \Teff$<$150\,000~K). The same holds for all hot DO white dwarfs besides
287: \kpd. The possible presence of \ion{Ne}{viii} lines in the FUSE spectra
288: of [WCE] stars is discussed below (Sect.\,\ref{sect_wc}).
289: 
290: \begin{table}
291: \begin{center}
292: \caption{\ion{Ne}{viii} lines identified in our program stars. 
293: Wavelengths are computed from NIST level energies (except for the $5p\rightarrow
294: 6d$, $5f\rightarrow 6d$, and $6p\rightarrow 8d$ transitions, see text)
295: and given in vacuum or air for $\lambda$ smaller or larger than
296: 3000~\AA, respectively. The presence of the feature at
297: $\lambda~1932$~\AA\ is uncertain.}
298: \label{tab} 
299: \begin{tabular}{ll}  
300: \hline 
301: \hline 
302: \noalign{\smallskip}
303: Wavelength / \AA &  Transition \\ \hline 
304: \noalign{\smallskip}
305: 1060.36, 1061.03          &$ 5s\rightarrow 6p$\\
306: 1129.02, 1130.12, 1130.35 &$ 5p\rightarrow 6d$ \\
307: 1150.12, 1150.81, 1150.92 &$ 6p\rightarrow 8d$ \\
308: 1162.24, 1162.67          &$ 5d\rightarrow 6f$ \\
309: 1164.54, 1164.75          &$ 5f\rightarrow 6g$ \\
310: 1164.88                   &$ 5g\rightarrow 6h$ \\
311: 1165.94, 1166.15, 1166.18 &$ 5f\rightarrow 6d$ \\
312: 1170.04, 1170.29          &$ 6d\rightarrow 8f$ \\
313: (1932.04)                 &$ 6h\rightarrow 7i$ etc.\\
314: 2976.75                   &$ 7i\rightarrow 8k$ etc.\\
315: 4340.77                   &$ 8k\rightarrow 9l$ etc.\\
316: 6068.63                   &$ 9l\rightarrow 10m$ etc.\\
317: \noalign{\smallskip} \hline
318: \end{tabular} 
319: \end{center}
320: \end{table}
321: 
322: \section{Model atmospheres and neon line formation calculations}
323: \label{modeling}
324: 
325: We use a grid of line-blanketed non-LTE model atmospheres, which is
326: described in detail in Werner \etal (2004a). In essence, the models
327: include the main photospheric constituents, namely, He, C, O, and
328: Ne. The neon model atom consists of the ionization stages
329: \ion{Ne}{iv-ix}. Of particular importance for the investigations
330: presented in this paper is the \ion{Ne}{viii} ion. It consists of 5 NLTE
331: levels and 6 lines, which is sufficient for the atmospheric model
332: structure computation, but considerable extensions were required in order
333: to be able to compute profiles of the identified highly excited
334: lines. Since we aimed at the computation of lines involving the $n=10$
335: levels (the $n=9\rightarrow 10$ emission at 6068~\AA), we included all
336: 77 levels up to $n=12$ with all 511 line transitions between them. In
337: addition, levels up to $n=14$ are included as LTE levels. With this
338: extended \ion{Ne}{viii} model ion, we performed line-formation iterations
339: for neon, i.e., keeping the atmospheric structure fixed. For the final
340: spectrum synthesis, fine-structure splitting of levels and lines was
341: considered, if possible. Energies for fine-structure splitting are not
342: available for all levels.  In particular, they are only partly available
343: for levels involved in the $n=5\rightarrow 6$ absorption lines at
344: $\lambda\ 1162-1165$~\AA. As a consequence, the detailed structure of
345: the computed line profile cores within these absorption troughs cannot
346: exactly match the observations and, indeed, a close inspection of
347: Fig.\,\ref{fig_all} reveals this problem. Level energies were taken from
348: NIST (Kramida \& Buchet-Poulizac 2006). Oscillator strengths were taken
349: from the Opacity (Seaton \etal 1994) and IRON (Hummer \etal 1993)
350: Projects databases (TIPTOPbase\footnote{
351: http://vizier.u-strasbg.fr/topbase/}), which are, however, not complete
352: for all transitions of our model atom. For lines involving levels with
353: $n=11$ and 12, we extrapolated oscillator strengths from transitions to
354: lower quantum numbers. For a lithium-like ion this appears reasonably
355: accurate; the estimates are probably within a 10\%
356: error. Photoionization cross-sections are taken from the Opacity Project
357: database when available or, otherwise, computed in hydrogen-like
358: approximation. Electron collisional rates were calculated with usual
359: approximation formulae. The Ne model atoms that were used for this
360: analysis have been developed in the framework of the German
361: Astrophysical Virtual Observatory
362: (\emph{GAVO}\footnote{http://www.g-vo.org}) project and are provided
363: within the T\"ubingen Model-Atom Database
364: \emph{TMAD}\footnote{http://astro.uni-tuebingen.de/\raisebox{.2em}{\tiny
365: $\sim$}rauch/TMAD/TMAD.html}.
366: 
367: The treatment of Stark broadening of \ion{Ne}{viii} lines poses a severe
368: problem. We employ an approximate formula to compute Stark widths, which
369: we use routinely to compute profiles of lines between highly excited
370: levels of lithium-like ions (\ion{C}{iv}, \ion{O}{vi}; Werner \etal
371: 1991), and which accounts for linear Stark broadening. While this
372: formula gave reasonable results for the other Li-isoelectronic ions in
373: our past work, we found that the $n=5\rightarrow 6$ absorption lines in
374: the FUSE spectral  range came out much too narrow (by about a factor of
375: three) compared to all of  the observations. Interestingly, a similar
376: notorious problem is encountered in the interpretation of \ion{Ne}{viii}
377: line widths of laboratory spectra. Measurements of the $3s\rightarrow
378: 3p$ transition at 2821~\AA\ (Glenzer \etal 1992) display line profiles
379: that are roughly a factor of two wider than predicted by theory, i.e.,
380: the $Z^{-2}$ scaling of the line width, as predicted by the impact theory,
381: significantly deviates from observations. Even today, this discrepancy
382: is not resolved, despite of intense efforts with both improved
383: experiments and quantum-mechanical calculations (Hegazy \etal 2003;
384: Griem \& Ralchenko 2006). Our finding from stellar spectra indirectly
385: suggests that the line widths measured in laboratory are in fact correct,
386: and that instead, the line-broadening theory is still inaccurate. It is
387: necessary to note, however, that we observe different lines of \ion{Ne}{viii}.
388: 
389: To cope with this problem we formally reduced the ionic core charge of
390: \ion{Ne}{viii} in our line broadening formula from $Z=8$ to $Z=2$. This
391: yields widths of the $n=5\rightarrow 6$ lines that coincide with the
392: observations. Of course, this procedure is rather unsatisfactory and
393: would prohibit an accurate abundance determination from these lines. In addition,
394: a close inspection of the computed profiles for the optical lines (i.e.,
395: the higher-$n$ transitions, Fig.\,\ref{fig_all}) reveals that they
396: become too broad with our crude $Z$-reduction. Fortunately, the neon
397: abundance of all PG1159 stars discussed here is known to be of the order
398: 2\% (Werner \etal 2004a). The focus of our paper is on the extreme
399: temperature sensitivity of the \ion{Ne}{viii} lines and line broadening
400: is a subordinate effect only.
401: 
402: \begin{figure}[tbp]
403: \begin{center}
404: \includegraphics[width=0.95\columnwidth]{fig_teff_variation.ps}
405:   \caption[]{The \ion{Ne}{viii} lines are strongly sensitive to
406:     \Teff\ and $\log\,g$. Generally, they become stronger with increasing \Teff\ and
407:     decreasing $\log\,g$.
408: \emph{Top panel}: \Teff-sensitivity in DO models with \logg=6.5. The
409: lines disappear if \Teff\ is below 160\,000~K.
410: \emph{Middle panel}: \Teff-sensitivity in PG1159 models with lower
411: gravity (\logg=5.7). In this case, the lines are detectable down to \Teff=140\,000~K.
412: \emph{Bottom panel}: \logg\ sensitivity in PG1159 models at
413: \Teff=160\,000~K. The gravity must be sufficiently low, otherwise the
414: \ion{Ne}{viii} lines are not detectable. The numbers in the lower right
415: corner of the panels give the model abundances of He/C/O/Ne in \% mass fraction.}
416:   \label{fig_teff_variation}
417: \end{center}
418: \end{figure}
419: 
420: \begin{figure*}[tbp]
421: \begin{center}
422: \includegraphics[width=0.85\textwidth]{fig_kpd0005_fit.ps}
423:   \caption[]{The \ion{He}{ii} lines in \kpd\ are compared with two different
424:     models. \emph{Left panel}: \Teff=200\,000~K, \logg=6.5. \emph{Right panel}:
425:     \Teff=120\,000~K, \logg=7.0. The hotter model yields a much better
426:     fit. In particular, only this model matches the emission cores in the $\lambda\lambda~6560$ and $4686$~\AA\
427:     lines. The emission line at $\lambda~4340$~\AA\ is due to \ion{Ne}{viii}
428:     and is only exhibited in the hotter model.}
429:   \label{fig_kpd0005_fit}
430: \end{center}
431: \end{figure*}
432: 
433: 
434: \section{Results}
435: \label{results}
436: 
437: We now compare our computed \ion{Ne}{viii} line profiles to the
438: observations (Fig.\,\ref{fig_all}). It is instructive to look at
439: Fig.\,\ref{fig_teff_variation}, in which we demonstrate that the
440: strengths of the $\lambda\lambda~1162-1165$~\AA\ absorption lines
441: strongly depend on \Teff\ and $\log\,g$.  The minimum \Teff\ for a line
442: detection is a strong function of gravity, ranging from about 160\,000~K
443: at \logg=6.5 to 140\,000~K at \logg=5.7.  The mere detection of these
444: \ion{Ne}{viii} lines thus provides a strict lower limit to $T\mathrm{\hspace*{-0.4ex}_{eff}}$.
445: Generally, the overall characteristics of the \ion{Ne}{viii} lines are
446: reproduced by the models. We find that the UV lines are in absorption,
447: while the optical lines are in emission.
448: 
449: Let us discuss in some detail each of the six objects in which we
450: detected the UV \ion{Ne}{viii} lines, particularly those cases in which
451: contradictions to \Teff\ determinations from previous work appear. This
452: mainly affects the DO white dwarf \kpd, whereas for the PG1159 stars the
453: \ion{Ne}{viii} line features can be matched with model parameters that
454: are in good or acceptable agreement with previous results.
455: 
456: \subsection{PG1159 stars}
457: 
458: Models for individual objects were computed with element abundances as
459: given in Werner \etal (2005). As already mentioned, the neon abundance
460: is kept fixed at 2\%. We started with values for \Teff\ and \logg\ also
461: taken from this reference; typical uncertainties are 10\% and 0.5~dex,
462: respectively.
463: 
464: \paragraph{\keins} The previously determined parameters are
465: \Teff=140\,000~K and \logg=6.4. Clearly, a model with these values shows
466: no \ion{Ne}{viii} lines at all. We find an acceptable fit  only at
467: \Teff\ and \logg\ values that are slightly beyond the error bars, namely
468: \Teff=160\,000~K and \logg=5.8.
469: 
470: \paragraph{\rxj} We find a good fit to the \ion{Ne}{viii} lines with
471: models close to the literature values (\Teff=170\,000~K, \logg=6.0). At
472: these parameter values, the \ion{Ne}{viii} absorption lines are
473: strongest. This explains why we can identify a number of additional weak
474: \ion{Ne}{viii} lines in the FUSE spectrum
475: (Fig.\,\ref{fig_rxj2117_details}).
476: 
477: \paragraph{\ngc} A good fit to the \ion{Ne}{viii} lines is
478: obtained at the literature values (\Teff=150\,000~K, \logg=5.7). Note
479: that \ngc\ is a fast rotator and the computed profiles were broadened
480: with 70~km/s (Rauch \& Werner 1997).
481: 
482: \paragraph{\lovier} The strong \ion{Ne}{viii} lines in the FUSE spectrum
483: suggest that this star is as hot as \rxj. We find a good fit with a
484: \Teff=170\,000~K, \logg=6.0 model. Therefore, \lovier\ is significantly
485: hotter than previously thought (\Teff=120\,000~K, \logg=5.5). This
486: result is, however, not surprising because we have already found
487: independent hints that the temperature is at least about 150\,000~K
488: (Werner \etal 2004a). Note that \lovier\ exhibited much more prominent
489: optical \ion{Ne}{viii} lines during its observed ``outburst'' and change
490: of spectral type from PG1159 to [WCE] (Werner \etal 1992).
491: 
492: \paragraph{\hh} The \ion{Ne}{viii} lines in the FUSE spectrum are
493: shallow and broad. This is a consequence of the high gravity (8.0). The
494: profiles confirm the extremely high \Teff\ (200\,000~K).
495: 
496: \subsection{The DO white dwarf \kpd}
497: 
498: An analysis of optical and HST UV spectra of \kpd\ gave the result
499: \Teff=120\,000~K and \logg=7 (Werner \etal 1994). Although being the
500: hottest known DO white dwarf, the identification of \ion{Ne}{viii} lines
501: is a big surprise. Figure~\ref{fig_teff_variation} shows that at this
502: relatively high gravity the temperature must be around 180\,000~K. We
503: find that models with \Teff=180\,000--200\,000~K and \logg=6.5--7.0 give
504: the best fit to the \ion{Ne}{viii} lines. Interestingly, this is not in
505: contradiction with the \ion{He}{ii} lines. We even find that a model
506: with \Teff=200\,000~K and \logg=6.5 gives better \ion{He}{ii} line fits
507: than the cooler 120\,000~K model. This is shown in
508: Fig.\,\ref{fig_kpd0005_fit}. To be more specific, the hot model is able
509: to match the height of the central emissions reversals in the
510: $\lambda\lambda~4686$ and $6560~$\AA\ line cores. At the same time, the
511: $\lambda~1640$~\AA\ line is matched very well.  We conclude that \kpd\
512: is significantly hotter than previously thought.
513: 
514: Unlike for the PG1159 stars, the neon abundance in \kpd\ is not known
515: from previous analyses. The usual diagnostic lines are \ion{Ne}{vii}
516: $\lambda\lambda~973$ and $3644~$\AA. The FUSE spectrum of \kpd\ is
517: contaminated by interstellar H$_2$ absorption, making the detection of \ion{Ne}{vii}
518: $\lambda~973~$\AA\ impossible. In a medium-resolution optical spectrum
519: published by Werner \etal (1994), no line feature is detected at
520: $\lambda~3644~$\AA. We have verified that this is compatible with all
521: our models presented here in the \Teff=120\,000-200\,000~K range.
522: 
523: We assumed that the Ne abundance is determined
524: by radiative levitation. The hottest model presented by Chayer \etal
525: (1995) has \Teff=100\,000~K with \logg=7.5 and predicts $\log({\rm
526: Ne/He})=-4$ by number. It is difficult to extrapolate this to a significantly
527: higher \Teff\ but the tendency is that the Ne abundance increases with
528: $T\mathrm{\hspace*{-0.4ex}_{eff}}$. A further increase can be expected because of the smaller
529: gravity. Although uncertain, it is not unreasonable that we set Ne=1\%
530: (by mass), that is, $\log({\rm Ne/He})=-2.7$ by number. We stress that
531: the high \Teff\ derived from the \ion{Ne}{viii} lines is hardly affected
532: by this assumption. In addition, the \ion{He}{ii} line profiles are
533: essentially independent of the neon abundance.
534: 
535: \section{Implications for ultrahigh-ionisation emission line identifications in [WCE] stars}\label{sect_wc}
536: 
537: Since the supposed identification of several optical emission lines in
538:  the hot [WC] star \sand\ as being due to ultrahigh ionised
539:  (i.e. non-thermally excited) C and O (\ion{C}{v} and \ion{O}{vii-viii})
540:  (Barlow \etal 1980), a number of authors claimed the identification of
541:  these features in other early-type [WC] stars. In the light of
542:  our results, we propose that most, and probably all, of these features
543:  stem from thermally excited ionisation stages of neon
544:  (\ion{Ne}{vii-viii}), carbon and oxygen. Table~\ref{tab2} summarizes
545:  our proposed identifications.  We give qualitative arguments for this,
546:  but detailed NLTE modeling with expanding model atmospheres will be
547:  necessary for a quantitative confirmation.
548: 
549: A broad emission feature in \sand\ and other [WCE]s at
550: $\lambda~3893$~\AA\ was attributed to the $n=7 \rightarrow 8$ transition
551: of \ion{O}{vii}. We propose that it originates from a prominent
552: \ion{Ne}{vii} multiplet recently identified in the hottest PG1159
553: central stars (Werner \etal 2004a), which has its strongest component
554: located at $\lambda~3892$~\AA. In addition, the \ion{Ne}{vii} $n=7
555: \rightarrow 8$ transition contributes. The emission at
556: $\lambda~4340$~\AA\ is probably from \ion{Ne}{viii}, not
557: \ion{O}{viii}. Emission lines at $\lambda\lambda~4555$~\AA\ and
558: 5665~\AA\ were assigned to the \ion{O}{vii} transitions $n=9 \rightarrow
559: 11$ and $n=8 \rightarrow 9$, respectively. We suggest that they are
560: lines between high Rydberg states of \ion{Ne}{vii}, with the same
561: principal quantum numbers (level energies are given in NIST and Lapierre
562: \& Knystautas 1999). This is supported by our discovery of two high-$l$
563: $n=8\rightarrow 9$ emission lines of \ion{Ne}{vii} at
564: $\lambda~5665$~\AA\ in some of our program stars (Fig.\,\ref{ne7_5660}).
565: Finally, the strong emission feature at $\lambda~4945$~\AA\ was assigned
566: to \ion{C}{v} $n=7 \rightarrow 6$. We think that it is the respective
567: \ion{N}{v} transition, as in the case of \kpd. Nitrogen is definitely
568: present in \sand, because the $3s\rightarrow 3p$ emission doublet at
569: $\lambda\lambda~4602, 4620$~\AA\ is present, as well as a prominent
570: P~Cygni profile of the \ion{N}{V} resonance line. Thus, the entire
571: optical emission line spectrum of \sand\ and other [WCE]s can be
572: explained without invoking ultrahigh-ionisation features.
573: 
574: \begin{table}
575: \begin{center}
576: \caption{List of lines for which we propose photospheric identifications
577:     as opposed 
578:   to previously thought identifications as non-photospheric
579:   ultrahigh-ionisation features in PG1159 stars, [WCE] stars, and in the
580:   DO \kpd. The \ion{O}{vi} and  \ion{N}{v} emission lines were
581:   identified before but thought to be blended by the ultrahigh ionisation
582:   features.} 
583: \label{tab2} 
584: \begin{tabular}{lll}  
585: \hline 
586: \hline 
587: \noalign{\smallskip}
588: Wavelength / \AA & Old ultrahigh-    & New photospheric \\
589:                  & ionisation        & identification   \\
590:                  & identification    &                  \\ \hline 
591: \noalign{\smallskip}
592: 1932             & \ion{O}{viii} \quad $n=7\rightarrow  8$ & \ion{Ne}{viii} \quad $n=7\rightarrow 8$ \\
593: 2977             & \ion{O}{viii} \quad $n=6\rightarrow  7$ & \ion{Ne}{viii} \quad $n=6\rightarrow 7$ \\
594: 3893             & \ion{O}{vii}  \quad $n=7\rightarrow  8$ & \ion{Ne}{vii} 3p\,$^3$P$^{\rm o}\rightarrow$ 3d\,$^3$D\\
595:                  &                                   & plus \ion{Ne}{vii} \quad $n=7\rightarrow  8$ \\
596: 4340             & \ion{O}{viii} \quad $n=8\rightarrow  9$ & \ion{Ne}{viii} \quad $n=8\rightarrow 9$ \\
597: 4500             & \ion{C}{vi}   \quad $n=8\rightarrow 10$ & \ion{O}{vi}    \quad $n=8\rightarrow 10$ \\
598: 4555             & \ion{O}{vii}  \quad $n=9\rightarrow 11$ & \ion{Ne}{vii}  \quad $n=9\rightarrow 11$ \\
599: 4945             & \ion{C}{v}    \quad $n=6\rightarrow  7$ & \ion{N}{v}     \quad $n=6\rightarrow  7$ \\  
600: 5290             & \ion{C}{vi}   \quad $n=7\rightarrow  8$ & \ion{O}{vi}    \quad $n=7\rightarrow 8$ \\
601: 5665             & \ion{O}{vii}  \quad $n=8\rightarrow  9$ & \ion{Ne}{vii}  \quad $n=8\rightarrow 9$  \\
602: 6068             & \ion{O}{viii} \quad $n=9\rightarrow 10$ & \ion{Ne}{viii} \quad $n=9\rightarrow 10$ \\
603: \noalign{\smallskip} \hline
604: \end{tabular} 
605: \end{center}
606: \end{table}
607: 
608: Based on observations  with the \emph{International Ultraviolet
609: Explorer} (IUE), several papers appeared in the literature claiming the
610: presence of ultrahigh ionisation lines in the UV spectra of [WCE]
611: stars. In \sand, as well as in NGC~5315 and NGC~6905, Feibelman (1996a,
612: b) discovered the $\lambda\lambda~1932$~\AA\ and 2977~\AA\ emission
613: features that we have discussed above, and he also assigned them to
614: \ion{O}{viii}. As in \kpd\ and \rxj\
615: (Fig.\,\ref{fig_rxj2117_details}), they probably are from \ion{Ne}{viii}.
616: 
617: Feibelman (1996b) even claimed the existence of an ``\ion{O}{viii}
618: sequence'' of planetary-nebulae nuclei, mainly based on the supposed
619: \ion{O}{vii-viii} identifications in PG1159 and [WCE] stars. Obviously,
620: that idea must now be discarded. This is further corroborated by the
621: fact that we detected the \ion{Ne}{viii} lines at
622: $\lambda~1162-1165$~\AA\ in a [WCE] star too, namely in NGC~2371
623: (Fig.\,\ref{wcstars}). The line strengths suggest \Teff$\approx
624: 150\,000$~K, which is, considering error limits, in good agreement with
625: the result of a detailed analysis of FUSE and IUE spectra (135\,000~K;
626: Herald \& Bianchi 2004). NGC~2371 was classified as [WC4] by Acker \&
627: Neiner (2003). These authors have based a new spectral classification
628: system for the hottest [WC]s on the occurrence of \ion{O}{vii-viii}
629: lines in the optical spectrum. Our identification as \ion{Ne}{vii-viii}
630: lines means that the empirical classification criteria remain
631: essentially correct because only the hottest [WC]s are able to show
632: these lines.
633: 
634: \begin{figure}[tbp]
635: \begin{center}
636: \includegraphics[width=0.95\columnwidth]{fig_ne7_5660.ps}
637:   \caption[]{First identification of two high-$l$ $n=8\rightarrow 9$ emission
638:     lines of \ion{Ne}{vii} (!). A strong and broad emission feature is seen
639:     in [WCE] stars at this location. We propose that it is due to these
640:     \ion{Ne}{vii} transitions and not due to ultrahigh ionized
641:     \ion{O}{vii}. The labels denote the theoretical line positions using
642:     energy levels that are not precisely known.}
643:   \label{ne7_5660}
644: \end{center}
645: \end{figure}
646: 
647: 
648: \begin{figure}[tbp]
649: \begin{center}
650: \includegraphics[width=0.95\columnwidth]{fig_wcstars.ps}
651:   \caption[]{Identification of \ion{Ne}{viii} lines in the [WCE] central
652:     star NGC~2371 (\emph{top}) and comparison to the PG1159-type central star
653:     \lovier\ (\emph{bottom}).}
654:   \label{wcstars}
655: \end{center}
656: \end{figure}
657: 
658: \section{Summary and conclusions}
659: \label{conclusions} 
660: 
661: We have identified \ion{Ne}{viii} absorption lines in FUSE spectra of
662: PG1159 stars and a DO white dwarf. Line profile fits  confirm that the
663: PG1159 stars are the very hottest members of their spectral class
664: (\Teff$\geq$150\,000\,K). We have shown that two well-known emission
665: lines in the optical spectra of these stars are also from \ion{Ne}{viii}
666: and not, as previously thought, from \ion{O}{viii}. While \ion{Ne}{viii}
667: is thermally excited in the hot photospheres, the existence of
668: \ion{O}{viii} would require temperatures of the order of one million~K,
669: thus, an unknown process (e.g., shock zones in the wind) was invoked to
670: explain these emission lines. This is no longer necessary.
671: 
672: We argue that these, and probably all, emission lines in early-type [WC]
673: central stars that are usually assigned to ultrahigh-ionisation stages
674: (\ion{O}{vii-viii}, \ion{C}{v}) originate from \ion{Ne}{vii-viii} and
675: other thermally excited species.
676: 
677: The discovery of \ion{Ne}{viii} lines in the central star \lovier\
678: suggests that its \Teff\ is distinctively higher than previously thought
679: (170\,000~K instead of 120\,000~K). To a smaller extent, the same
680: tendency is seen for the PG1159 star \keins\ and the [WCE] star
681: NGC\,2371. This calls for a re-analysis of the complete spectra of these
682: stars to see how the fit to other line features can be reconciled. The
683: computation of model grids with varying parameters $T\mathrm{\hspace*{-0.4ex}_{eff}}$, $\log\,g$, and
684: abundances of the most important elements (He, C, O, Ne) will be
685: necessary. Since the stars display prominent P~Cygni line profiles,
686: it could be advantageous to use expanding atmosphere models with the
687: available spectral data. 
688: As already mentioned, the neon abundance cannot
689: be determined from the newly discovered \ion{Ne}{viii} lines because of
690: the uncertainties with the line broadening theory. Instead, the strong
691: P~Cygni profile of the \ion{Ne}{vii} $\lambda~973~$\AA\ line exhibited
692: by these stars can be used as a sensitive tool (Bianchi \& Herald
693: 2007). Again, detailed parameter studies must be performed to find
694: which neon abundance and which mass-loss rate yield good fits to this
695: \ion{Ne}{vii} line profile, while at the same time determining the \Teff\ of the model that
696: is high enough to produce detectable \ion{Ne}{viii} lines.
697: 
698: A surprising result of our investigation is the identification of
699: \ion{Ne}{viii} lines in the hottest known DO white dwarf, \kpd. We
700: conclude that its temperature is close to \Teff=200\,000~K and, hence,
701: that it is significantly hotter than hitherto thought (120\,000~K). As
702: in the case of the PG1159 stars, the optical \ion{Ne}{viii} emission
703: lines were previously assigned to ultrahigh-ionized \ion{O}{viii}. The announcement of the discovery of a relatively \emph{soft}
704: X-ray emitting corona about \kpd\ seemed to support this assignment
705: (Fleming \etal 1993). However, the analysis of a \emph{Chandra} spectrum
706: exhibiting flux in the range 20--80~\AA\ proved that the soft X-rays
707: stem from the photosphere of \kpd\ (Drake \& Werner 2005). The
708: deposition of the corona is now in accordance with the deposition of the
709: ultrahigh-ionisation lines. A comprehensive
710: re-analysis of all available data (\emph{Chandra}, FUSE, HST,
711: \emph{Keck}) is required to tightly constrain the atmospheric parameters and metal
712: abundances.
713: 
714: It remains to be seen how the extremely high effective temperature
715: derived in our work relates to the observed \emph{hard} X-ray emission
716: at 1~keV (12~\AA) from \kpd\ (O'Dwyer \etal 2003). It might be that this
717: emission is also of photospheric origin. After all, the entire spectral
718: properties of \kpd\ could be deciphered as thermal photospheric
719: radiation. In any case, the star has turned out to be \emph{by far} the
720: hottest known DO white dwarf.
721: 
722: \begin{acknowledgements}
723: We are grateful to the referee, Luciana Bianchi, for a careful reading
724: of the manuscript and for her constructive comments.
725: We thank Uli Heber and Ralf Napiwotzki for putting their \emph{Keck} and
726: \emph{VLT} spectra at our disposal.  We thank A.E.~Kramida, H.R.~Griem,
727: \mbox{H.-J.} Kunze, M.S.~Dimitrijevi{\'c}, and {\'E}. Knystautas for
728: discussions on atomic data and sending us reprints. T.R. is supported
729: by the \emph{German Astrophysical Virtual Observatory} project of the
730: German Federal Ministry of Education and Research under grant
731: 05\,AC6VTB. J.W.K. is supported by the FUSE project, funded by NASA
732: contract NAS5-32985.
733: \end{acknowledgements}
734: 
735: \begin{thebibliography}{}
736: 
737: \bibitem{ack03} Acker, A., \& Neiner, C. 2003, A\&A, 403, 659
738: 
739: \bibitem{ba80} Barlow, M.~J., Blades, J.~C., \& Hummer, D.~G. 1980, ApJ, 241, L27
740: 
741: \bibitem{bi07} Bianchi, L., \&  Herald, J.~E. 2007, Proceedings,
742:   IAU General Assembly 2006, Joint Discussion 04, in press
743: 
744: \bibitem{cha95} Chayer, P., Fontaine, G., \& Wesemael, F. 1995, ApJS, 99, 189
745: 
746: \bibitem{Dr05} Drake, J. J., \& Werner, K. 2005, ApJ, 625, 973
747: 
748: \bibitem{Fe96a} Feibelman, W.~A. 1996a, ApJ, 464, 910
749: 
750: \bibitem{Fe96b} Feibelman, W.~A. 1996b, PASP, 108, 664
751: 
752: \bibitem{Fl93} Fleming, T. A., Werner, K., \& Barstow, M. A. 1993, ApJ, 416, L79
753: 
754: \bibitem{gl92} Glenzer, S., Uzelac, N. I., \& Kunze, H.-J. 1992, Phys. Rev. A., 45, 8795
755: 
756: \bibitem{Gr06} Griem, H. R., \& Ralchenko, Y. 2006, in Spectral Line
757:   Shapes: 18$^{\rm th}$ International Conference, eds. E. Oks,
758:   M. Pindzola, American Institute of Physics, CP874, 14
759: 
760: \bibitem{he03} Hegazy, H., Seidel, S., Wrubel, T., \& Kunze, H.-J. 2003, JQSRT, 81, 221
761: 
762: \bibitem{hb04} Herald, J.~E., \& Bianchi, L. 2004, ApJ, 609, 378
763: 
764: \bibitem{h05}  Herald, J.~E., Bianchi, L., \& Hillier, D.~J. 2005, ApJ, 627, 424
765: 
766: \bibitem{hea1993} Hummer, D.~G., Berrington, K.~A., Eissner, W., \etal 1993, A\&A, 279, 298
767: 
768: \bibitem{ja07} Jahn, D., Rauch, T., Reiff, E., Werner, K., Kruk, J.~W., \& Herwig, F. 2007, A\&A, 462, 281
769: 
770: \bibitem{kr06} Kramida, A. E., \& Buchet-Poulizac, M.-C. 2006, Eur. Phys. J. D, 39, 173
771: 
772: \bibitem{la99} Lapierre, A., \& Knystautas, {\'E}.~J. 1999, Physica Scripta, 59, 426
773: 
774: \bibitem{Moo00} Moos, H.W., Cash, W.C., Cowie, L.L., \etal 2000, ApJL, 538, 1
775: 
776: \bibitem{na03} Napiwotzki, R., Christlieb, N., Drechsel, H., \etal 2003,
777:   The Messenger, 112, 25
778: 
779: \bibitem{od03} O'Dwyer, I.~J., {Chu}, Y.-H., {Gruendl}, R.~A.,
780: 	{Guerrero}, M.~A., \& {Webbink}, R.~F. 2003, AJ, 125, 2239
781: 
782: \bibitem{ra97} Rauch, T., \& Werner, K. 1997, 
783:         in The Third Conference on Faint Blue Stars, eds.\
784:         A.G.D.\,Philip, J.\,Liebert, R.A.\,Saffer, L.\,Davis Press, Schenectady, NY,
785: 	p.\,217
786: 
787: \bibitem{Sah00} Sahnow, D.J., Moos, H.W., \& Ake, T.B. 2000, ApJL, 538, 7
788: 
789: \bibitem{Sea94} Seaton, M.~J., Yan, Y., Mihalas, D., \& Pradhan, A.~K. 1994, MNRAS, 266, 805
790: 
791: \bibitem{si92} Sion, E.~M., \& Downes, R.~A. 1992, ApJ, 396, L79
792: 
793: \bibitem{w92a} Werner, K., \& Heber, U. 1992, 
794:         in  Atmospheres of Early-Type Stars, eds. U. Heber,  
795:         C.S. Jeffery, Lecture Notes in Physics 401, Springer, Berlin, 
796:         p.\,291
797: 
798: \bibitem{w91} Werner, K., Heber, U., \& Hunger, K. 1991, A\&A, 244, 437
799: 
800: \bibitem{w92b} Werner, K., Hamann, W.-R., Heber, U., Napiwotzki, R., Rauch, T., 
801:         \& Wessolowski, U. 1992, A\&A, 259, L69
802: 
803: \bibitem{w94} Werner, K., Heber, U., \& Fleming, T.~A. 1994, A\&A, 284, 907
804: 
805: \bibitem{w96} Werner, K., Dreizler, S., Heber, U., \etal 1996, A\&A, 307, 860
806: 
807: \bibitem{w04a} Werner, K., Rauch, T., Reiff, E., Kruk, J.~W., \& Napiwotzki, R. 2004a, \aap, 427, 685 
808: 
809: \bibitem{w04b} Werner, K., Rauch, T., Barstow, M.~A., \& Kruk, J.~W. 2004, A\&A, 421, 1169
810: 
811: \bibitem{w05} Werner, K., Rauch, T., \& Kruk, J.~W.\ 2005, \aap, 433, 641 
812: 
813: \bibitem{w07} Werner, K., Rauch, T., \& Kruk, J.~W.\ 2007, \aap, 466, 317
814: 
815: \bibitem{zu98} Zuckerman, B., \& Reid, I.~N. 1998, ApJ, 505, L143
816: 
817: \end{thebibliography}
818: 
819: \end{document}
820: