1: %\documentclass[manuscript]{aastex}
2: \documentclass{emulateapj}
3: \usepackage{apjfonts}
4: \usepackage{lscape}
5:
6: \input{epsf}
7: \slugcomment{Submitted to the Astronomical Journal}
8:
9:
10: \shorttitle{Clumping in O-star winds}
11: \shortauthors{L\'epine \& Moffat}
12:
13: \begin{document}
14:
15: \title{Direct spectroscopic observations of clumping in O-star
16: winds.}
17:
18: \author{S\'ebastien L\'epine\altaffilmark{1,2} \& Anthony
19: F. J. Moffat\altaffilmark{1,3}}
20:
21: \altaffiltext{1}{Visiting astronomer, Cerro Tololo Inter-American
22: Observatory, National Optical Astronomy Observatories, operated by
23: the Association of Universities for Research in Astronomy, Inc.,
24: under cooperative agreement with the National Science Foundation.}
25:
26: \altaffiltext{2}{Department of Astrophysics, Division of Physical Sciences,
27: American Museum of Natural History, Central Park West at 79th Street,
28: New York, NY 10024, USA}
29:
30: \altaffiltext{3}{D\'epartement de Physique, Universit\'e de
31: Montr\'eal, Montr\'eal, Qc, Canada, moffat@astro.umontreal.ca}
32:
33: \begin{abstract}
34: We report the detection and monitoring of transient substructures in
35: the radiation-driven winds of five massive, hot stars in different
36: evolutionary stages. Clumping in the winds of these stars shows up as
37: variable, narrow subpeaks superposed on their wide, wind-broadened
38: (optical) emission lines. Similar patterns of emission-line profile
39: variations are detected in the Of stars $\zeta$ Puppis and HD93129A,
40: in the more evolved hydrogen-rich, luminous, Of-like WN stars HD93131
41: and HD93162, and in the more mass-depleted WC star in
42: $\gamma^2$Velorum. These observations strongly suggest that stochastic
43: wind clumping is a universal phenomenon in the radiation-driven, hot
44: winds from all massive stars, with similar clumping factors in all
45: stages of mass depletion.
46: \end{abstract}
47:
48: \keywords{line:profiles \--- stars: emission-line, Be \--- stars:
49: mass-loss \--- stars: winds, outflows \--- stars: Wolf-Rayet}
50:
51: %=...=%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
52: \section{Introduction}
53:
54: Evidence has been accumulating in the last two decades that hot
55: stellar winds are far from being the smooth flows of escaping matter
56: that were often conveniently assumed. Current observations are of
57: sufficient quality to show that the strong, hot stellar winds of
58: bright, population I Wolf-Rayet (WR) stars are systematically pervaded
59: by inhomogeneities on different scales. Variability on time scales of
60: minutes to hours is so far best described by a fully clumped wind with
61: no smooth component \citep{LM99,LEM99}. This clumping may very well
62: be related to small-scale radiative instabilities leading to strong
63: shocks in the wind \citep{GO95}.
64:
65: Is this phenomenon unique to the extremely dense winds of WR stars, or
66: do other hot stars also have similar globally clumped winds which have
67: been difficult to detect due to their weaker emission lines?
68: Subsequent to our WR studies, we found that the Of star $\zeta$ Puppis
69: shows identical structure in its wind, as found (statistically) in WR
70: stars \citep{ELM98}: $\zeta$ Pup's optical He II $\lambda$ 4686
71: emission line varies in exactly the same manner as the isolated HeII
72: 5411 line in WN stars or CIII 5696 in WC stars. The variable emission
73: component observed at very high S/N in $\zeta$ Pup is also the {\it
74: same fraction} of the total emission-line flux as in WR spectra,
75: whose absolute amplitude is significantly less in O stars. This may
76: explain why it had remained virtually unnoticed before in O stars. If
77: this one O star is typical, then we expect to find the same stochastic
78: emission-line variability in other O-star winds.
79:
80: Other types of hot stars also exhibit clumpy structure in their winds.
81: Examples include the [WC] nuclei of planetary nuclei \citep{GMA03},
82: novae \citep{LSLZ99}, supernovae \citep{M00} and possibly
83: others that have not been examined properly yet.
84:
85: Turbulent clumping in hot-star winds has a number of very important
86: consequences, e.g. among the most important are: (1) Most current
87: estimates of mass-loss rates, which are sensitive to the square of the
88: density, have to be revised downwards {\em by about a factor three}
89: with important implications for stellar evolution \citep{MR94}. (2)
90: Since clumps follow the mean flow, on average, they are turning out to
91: be extremely useful in constraining the empirical wind velocity law,
92: v(r), that is quite fundamental, but poorly known at present in
93: hot-star winds; in particular, the convenient, popular interpolation
94: $\beta$-law appears to be only a rough approximation \citep{LM99}.
95: (3) Wind clumps may ultimately provide the necessary means of
96: compression and shielding required to form dust in some WR and nova
97: winds \citep{Wetal87,CT95,B95,MM07}. This might be relevant for dust
98: formation in any type of star. The presumed associated decrease in the
99: mass-loss rates could have a major effect on their evolution, slowing
100: down the shedding of their envelope and delaying their eventual
101: transfer to a WR stage.
102:
103: \begin{deluxetable*}{llcccll}
104: \tabletypesize{\scriptsize}
105: \tablecolumns{7}
106: \tablecaption{Stars observed}
107: \tablehead{HD number & Other name & RA(2000) & Dec(2000) & V magnitude & Spectral type & Comment}
108: \startdata
109: HD 66811 & $\zeta$ Puppis & 08 03 35.04 & -40 00 11.3 & 2.21 & O4I(n)f & single runaway \\
110: HD 93129A & \nodata & 10 43 57.46 & -59 32 51.3 & 6.97 & O3If* & visual binary\tablenotemark{1} \\
111: HD 93131 & \nodata & 10 43 52.25 & -60 07 04.0 & 6.48 & WN6ha & single \\
112: HD 93162 & \nodata & 10 44 10.33 & -59 43 11.4 & 8.11 & WN6ha & spectroscopic binary\tablenotemark{2} \\
113: HD 68273 & $\gamma^2$ Velorum & 08 09 31.95 & -47 20 11.7 & 1.81 & WC8+O7.5III-V & spectroscopic binary\tablenotemark{3} \\
114: \enddata
115: \tablenotetext{1}{Companion 0.05\arcsec from the primary \citep{Netal04}.}
116: \tablenotetext{2}{Orbital period 207.7 days \citep{Getal06}.}
117: \tablenotetext{3}{Orbital period 78.5 days \citep{DM00}.}
118: \end{deluxetable*}
119:
120: One way out, however, is via the still poorly-understood, intermediate,
121: luminous blue variable (LBV) stage, which may provide the necessary
122: strong episodic mass-loss to ultimately allow O stars to pass to WR
123: stars \citep{S07}. In the meantime, massive-star models have been
124: vastly improved. Including rotation in the models, which is generally
125: quite rapid for O stars, will tend to feed more H-rich fuel into the
126: core region and increase the luminosity for a given mass, thereby
127: allowing more O stars to become WR, and relaxing the requirement for
128: unexplainably high mass-loss rates in the HRD isochrone fits to
129: observations \citep{MM00}. Therefore, it appears particularly urgent
130: to examine the clumping question in O-star winds of all types, so that
131: evolutionary models for massive stars on/near the main sequence can be
132: checked/constrained. This will increase our confidence in the age
133: determinations of clusters and starbursts, which are turning out to be
134: crucial tracers of star formation in the early Universe. It is also
135: relevant in the question of the progenitors of the slow gamma-ray
136: bursts, believed to arise in rapidly rotating WR stars
137: \citep{H07}. The rotation rate will be higher if the progenitor loses
138: less mass on the way to becoming a WR star. In this context, it is
139: interesting to note that clumping is still as strong in SMC WR stars
140: at Z$_{\odot}/5$ as at Z$_{\odot}$ in the Galaxy \citep{Metal07}.
141:
142: Very recent studies using the indirect approach have revealed
143: decreased mass-loss rates in O-star winds by factors of 3 or more
144: \citep{BLH05} to 10 or more \citep{FMP06}. These are based on weak
145: resonance and other lines in the far UV with the FUSE satellite. In
146: the UV, the emission lines are stronger, but often saturated and in
147: any case, it is difficult to obtain sufficiently high S/N. A recent
148: complementary approach in the UV has been successful, however, in
149: revealing variable features in the strong P Cygni absorption edges in
150: the FUSE UV spectrum of the single WC8 star WR135 by \citet{Metal06},
151: linking the presence of overdense clumps with observed shocks.
152:
153: In this study, we look for direct evidence for wind clumping in
154: optical emission-lines generated in O-star and related winds. Clumping
155: is detected in the acceleration zone near the base of the wind as
156: transient, moving subpeaks on the emission-line profiles. Among O
157: stars, Of stars are the easiest to examine, since they have emission
158: lines in the optical, where it is possible to obtain high S/N spectra
159: of these normally weak lines.
160:
161: \begin{figure*}
162: \epsscale{1.15}
163: \plottwo{f1a.eps}{f1b.eps}
164: \caption{Nightly mean spectra of our five program stars, listed in
165: Table 1. For each star, two orders from the \'echelle spectrogram
166: are shown, with the wavelength regions around the broad, intense
167: emission lines of HeII 4686 (left) and HeII/H$\beta$ 4860
168: (right). The mean spectra for the nights of 2000 Jan 25, 26 and 27
169: are shown in red, green and blue, respectively.}
170: \end{figure*}
171:
172: \begin{figure}
173: \epsscale{1.0}
174: \plotone{f2.eps}
175: \caption{Time series of the residuals after subtraction of the mean
176: profile, for the Wolf-Rayet star $\gamma^2$ Velorum.}
177: \end{figure}
178:
179: %=...=%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
180: \section{Observations}
181:
182: We observed two Of (one a repeat of previous observations of $\zeta$
183: Pup) and two late-type, H-rich, weak-line, Galactic WN (super Of-like)
184: stars intensively during three nights, 2000 January 25/26-27/28, at
185: the 4m telescope at CTIO. Apart from $\zeta$ Pup, the other three
186: stars are all located in the Carina Nebula. Parallel observations of
187: the bright WR star $\gamma^2$Vel (WC8 + O7.5III-V) were also observed
188: as a check star for known clumps \citep{LEM99}. Basic data on our
189: five targets are found in Table 1.
190:
191: We used the bench-mounted \'echelle spectrograph giving an inverse
192: dispersion of 0.06 \AA\ per pixel and (2.7-pixel) resolving power R
193: $\sim$30,000 over the $\lambda \lambda$ 4200 - 5700 \AA\ range. The
194: slit was 1$\arcsec$ wide. Typical elementary exposures of 10-1000 sec,
195: depending on the stellar brightness, were combined from the CCD
196: detector into one spectrum every $\sim$1-2 hours for each star,
197: yielding net S/N $\approx$ 500 per pixel (no binning) after wavelet
198: filtering (see below). Wavelengths were calibrated using Th-Ar
199: spectra. An internal quartz-lamp was used for flat-fielding.
200:
201: The bright \'echelle B3V flux standard $\eta$ Hya \citep{Hetal92} was
202: observed in an attempt to calibrate and splice together the \'echelle
203: orders; however, its standard flux values turned out to be of too low
204: resolution to be useful, so its use was abandoned. This forced us to
205: treat each \'echelle order separately. In any case, this was not a
206: serious impediment, as we chose to focus on the strongest, most
207: isolated optical emission lines of HeII 4686 and HeII/H$\beta$ 4860 in
208: each star. After some experimentation, we obtained final extracted
209: spectra of good quality by dividing the spectra in order 9 by those
210: in order 8 for HeII 4686 and order 7 by order 6 for HeII/H$\beta$
211: 4860. In all the stars, orders 8 and 6 were essentially devoid of
212: emission lines and thus could be used as continuum reference
213: spectra. This division by neighboring orders allowed us to eliminate
214: the strong order sensitivity-function, and rectify the spectra without
215: polluting the target emission lines significantly.
216:
217: Nightly mean spectra for each of our 5 targets, are displayed in
218: Figure 1 for the \'echelle orders covering the HeII 4686 and
219: HeII/H$\beta$ 4860 atomic lines. All stars show those atomic lines
220: strongly in emission, with large equivalent widths and very broad
221: profiles which are typical of Wolf-Rayet and Of stars with strong,
222: radiation-driven winds. Profiles from the Wolf-Rayet stars are
223: dominated by the emission component, while profiles from the two Of
224: stars show a slightly blue-shifted dip which is due to P Cygni type
225: absorption from the stellar wind.
226:
227: %=...=%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
228: \section{Analysis and results}
229:
230: We first examine whether there are significant global line variations
231: from night to night in the P Cygni component, as seen before in the
232: nightly spectra of $\zeta$ Pup \citep{MM81,ELM98}. The nightly means from
233: each of the three nights are shown in Fig.1 in different
234: colors. Significant night-to-night variations are observed in the WC
235: star $\gamma^2$Vel and in the two Of stars $\zeta$Pup and HD
236: 93129A. Variations are also apparent in the HeII 4686 line of the two
237: WNha stars HD 93131 and HD 93162. The HeII 4686 line in $\zeta$ Pup is
238: found to display the most pronounced variations, which suggest that
239: the intense line-profile variations discovered by \citet{ELM98} might
240: be an exceptional case.
241:
242: Variations on timescales of hours turn out to be much more interesting
243: in all the stars. For each spectrum collected during the observing
244: run, we subtracted the mean profile for the run. Despite all the
245: calibration steps described in \S2, many instrumental artifacts remain
246: in the spectra at the $\lesssim$1\% level. While these artifacts have
247: little effect on the mean spectra, they do show up prominently in the
248: residuals. Most artifacts are either of low or high frequency in
249: wavelength/pixel space. We thus filtered all of these out below 0.5
250: \AA\ and above $\sim$8 \AA\ using wavelet transforms following the
251: method of \citet{LEM99}. This is fortunate, given that most of the
252: intrinsic variations in Wolf-Rayet and Of star lines occur as narrow
253: emission line features (subpeaks) on a 1-2 \AA\ scale.
254:
255: \begin{figure*}
256: \epsscale{1.0}
257: \plottwo{f3a.eps}{f3b.eps}
258: \caption{Time series of the residuals after subtraction of the mean
259: profile, for the H-rich WN stars HD93131 (left) and HD93162 (right).}
260: \end{figure*}
261:
262: \begin{figure*}
263: \epsscale{1.0}
264: \plottwo{f4a.eps}{f4b.eps}
265: \caption{Time series of the residuals after subtraction of the mean
266: profile, for the Of stars $\zeta$ Puppis (left) and HD93129A (right).}
267: \end{figure*}
268:
269: In Figs. 2, 3, and
270: 4, variability in the emission lines of HeII 4686 (left) and
271: HeII/H$\beta$ 4860 (right) are shown for each of our target stars. The
272: mean profile for each line is displayed in the bottom panel. The
273: middle panel plots the dispersion at each pixel for the entire
274: observing run. The top panel is a two dimensional plot of the residuals
275: (i.e. the instantaneous deviation from the mean profile) as a function
276: of time and wavelength. Time runs upward on the vertical axis. The
277: horizontal axis is labeled in units of radial velocity, calculated
278: based on the redshift/blueshift of the emission line under
279: consideration.
280:
281: Significant variations are observed in all the lines examined. With
282: variability occurring on timescales of hours, it is clear that at least
283: some of this variation will average out in the nightly mean. The
284: night-to-night variations noted in Figure 1 thus underestimate the
285: variability in the line profiles. Conversely, the mean from all three
286: nights can be confidently used as a base profile, and the residuals
287: from this means (as shown in Figs.2, 3, and 4) will provide a good
288: representation of the variable elements, be they either excess
289: emission subpeaks or excess absorption troughs.
290:
291: In all the target stars, both the HeII 4686 and HeII/H$\beta$
292: 4860 lines show variability which is extremely similar to that
293: previously observed in WR spectra. The same general pattern of
294: propagation outwards from the line center is apparent in all the stars
295: for at least one of the nights. These indicate outwards acceleration
296: in the winds of all these stars. According to the phenomenological
297: models of \citet{LM99}, emission subpeaks in the line profile do not
298: necessarily trace the acceleration of single, overdensities (clumps)
299: in the wind. Rather, the emission subpeaks are likely the combined
300: signature of numerous clumps, so the outwards motion of subpeak
301: elements across the line may not be a measure of the true wind
302: acceleration. In any case, in the one star where the acceleration is
303: most apparent (HD 93131) subpeak profiles on the blue and red edges
304: are observed to move by about 400 km/s over the course of one night (8
305: hours). This suggests an outward wind-acceleration rate on the order
306: of 14 m s$^{-2}$, which is consistent with acceleration rates
307: measured in other Wolf-Rayet stars \citep{LM99}.
308:
309:
310: \begin{figure*}
311: \epsscale{1.15}
312: \plottwo{f5a.eps}{f5b.eps}
313: \caption{
314: Comparison between the normalized mean profiles of the emission lines
315: and the observed dispersion in the line profile, as a function of
316: wavelength. Mean (thick) lines and dispersion (thin) lines are
317: superposed on each plot, with the amplitude of the dispersion
318: multiplied by a factor of 35 (hence a dispersion level of 0.35 on the
319: plot actually corresponds to a dispersion amplitude of 0.01 in
320: normalized flux units).
321: }
322: \end{figure*}
323:
324: In Figure 5, we condense the variable information into plots of
325: standard deviation per pixel across each line profile for all spectra
326: collected over the three nights. Standard deviation profiles are all
327: multiplied by a factor of 35, and compared with the mean profiles for
328: each star. On this scale, the deviation profiles and mean
329: emission-line profiles have similar amplitudes, which suggest that all
330: lines show variability at the $\sim$3\% level. This level of
331: variability is on a par with previous data for 9 WR stars
332: \citep{LM99}, the Of star $\zeta$ Pup \citep{ELM98}, and $\gamma^2$
333: Vel \citep{LEM99}. The only exception is the He II $\lambda4860$
334: profile in HD93162, whose variability is below the 1\% level.
335:
336: %=...=%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
337: \section{Discussion and conclusions}
338:
339: Our spectroscopic time-series show systematic variability on a
340: timescale of hours in bright emission-line profiles from hot, massive
341: stars with moderately strong winds. The emission lines examined
342: here include lines which are significantly fainter, relative to the
343: continuum, than the bright lines from Wolf-Rayet stars, where the
344: phenomenon was first examined \citep{LM99}. We find that the amplitude
345: of the variations are at the 1\%-3\% level of the line emission. This
346: suggests that variability is universal in all emission lines, but that
347: it is only apparent in the strongest of the lines.
348:
349: While the variability profiles ($\sigma (\lambda)$), after rescaling,
350: generally follow the underlying line profiles ($I(\lambda)$), they do
351: not match exactly on close examination. There are two possible causes
352: for this: (1) the variability is strictly proportional to line
353: emission, but the profile is distorted by an underlying
354: absorption component, and/or (2) there exists an additional source of
355: variability in the line profile, e.g. because of variations in an
356: underlying absorption component, as in P Cygni absorption edges
357: \citep{R92}. The effects from distorted line profiles are most
358: apparent in the lines of the Of stars, which are flanked by a blueward
359: absorption component, and are strongly asymmetric relative to the rest
360: wavelength. Despite this, however, we clearly observe (Fig.4) that the
361: variability profile is roughly symmetric about the rest wavelength,
362: which confirms our assertion that the variability is associated mainly
363: with the {\em emission} component.
364:
365: Overall, it appears that the clumps in the winds that cause the
366: variable subpeaks across the emission lines, do indeed trace the wind.
367: This would suggest that, within their turbulent speeds, the clumps
368: obey essentially the same acceleration and velocity laws as the
369: average wind itself. Indeed, the clumps {\it are} the wind! In fact,
370: using extremely simple assumptions, \citet{LM99} were able
371: to simulate strong WR emission lines {\it and their observed
372: variability} solely using a superposition of a very large number of
373: discrete wind-emitting elements, without recourse to any kind of
374: continuous background wind. This may not be proof of concept, but at
375: least it is consistent with other astrophysically turbulent media such
376: as the ISM \citep{S93}. We would thus claim that hot-star
377: winds are the (supersonic compressible) turbulent result of some kind
378: of energy-input driving. Whether this occurs predominantly as a process
379: of cascading dissipation from large to small scales \citep{H94} or of
380: shock-merging aggregation from small radiative instabilities on
381: sub-Sobolev \citep{O94} to large scales \citep{F05}, remains to
382: be demonstrated observationally.
383:
384: The ratio $\sigma(\lambda) / I(\lambda)$ is observed to vary between
385: different lines, though only by a factor of a few at most for the same
386: line in different stellar winds. The more pronounced case here is the
387: star HD193162 in which HeII 4686 is more variable than HeII/H$\beta$
388: 4860 by about a factor of 3. This could be related to enhanced HeII
389: 4686 emission from the highly turbulent zone of wind-wind collision
390: between the components in this massive binary system. The variability
391: levels are however remarkably similar between the various stars
392: observed here. Since the intensity variations across optical
393: (recombination) spectral lines vary with the square of the density,
394: the clumping filling factor, which varies as the square root of this
395: quantity, must actually show much less dispersion compared to the
396: observed line-intensity variations, from one wind to another. From
397: this, we deduce that most hot-star winds (O, WR, LBV) require nearly
398: the same reduction factor in mass-loss rate of typically $\sim$3 (2-5
399: in the extreme) when the mass-loss rates are based on density-squared
400: mechanisms. This result is in accord with the consensus on the
401: ubiquity of the clumping phenomenon that came out of the recent
402: workshop devoted entirely to clumping in hot-star winds \citep{H08}.
403:
404: Our conjecture is that spectroscopic observations with even higher
405: signal-to-noise ratio should reveal similar variability patterns in
406: all emission-line profiles of stars with hot winds. Wind clumping is
407: the probable source of the variability. The timescale is set by the
408: time it takes for individual clumps to propagate through the layer in
409: the wind from which line emission occurs. Different lines will show
410: different patterns, depending on the height and depth of the layer
411: where emission occurs. Emission lines formed in overlapping layers
412: will have similar line-profile variability patterns. From a typical
413: timescale of $\sim$6 hours and a wind velocity speed of $\sim1000$km
414: s$^{-1}$, the typical depth of the emission layers is $\sim$2 10$^7$
415: km, or about 30 R$_{\sun}$, consistent with the idea that the lines
416: are formed relatively close to the hydrostatic surface. All of these
417: things should be tested with better future data.
418:
419: %==%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
420: \acknowledgments
421:
422: {\bf Acknowledgments}
423:
424: The authors thank the referee, Ken Gayley, for useful comments. AFJM
425: is grateful for financial assistance from NSERC (Canada). SL is
426: supported by the U.S. National Science Foundation, through grant
427: AST-0607757.
428:
429: %==%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
430: \begin{thebibliography}{}
431:
432: \bibitem[Bode(1995)]{B95}
433: Bode, M.F. 1995, IAU Symp. 163, 363
434:
435: \bibitem[Bouret {\it et al.}(2005)]{BLH05}
436: Bouret, J.-C., Lanz, T., \& Hillier, D.J. 2005
437:
438: \bibitem[Cherchneff {\it et al.}(1995)]{CT95}
439: Cherchneff, I., \& Tielens, A. G. G. M. 1995, IAU Symp. 163, 346
440:
441: \bibitem[De Marco {\it et al.}(2000)]{DM00}
442: De Marco, O., Schmutz, W., Crowther, P.A., Hillier, D.J., Dessart, L.,
443: de Koter, A., Schweickharddt, J. 2000, A\&A, 358, 187
444:
445: \bibitem[Eversberg {\it et al.}(1998)]{ELM98}
446: Eversberg, T., L\'epine, S., Moffat, A.F.J. 1998, ApJ, 494, 799
447:
448: \bibitem[Feldmeier (1995)]{F05}
449: Feldmeier, A. 1995, \aap, 299, 523
450:
451: \bibitem[Fullerton {\it et al.}(2006)]{FMP06}
452: Fullerton, A.W., Massa, D.L., Prinja, R.K. 2006, ApJ, 637, 1025
453:
454: \bibitem[Gamen {\it et al.}(2006)]{Getal06}
455: Gamen, R., Gosset, E., Morrell, N., Niemela, V.,
456: Sana, H., Naz\'e, Y., Rauw, G., Barb\'a , R.,
457: Solivella, G. 2007, A\&A, 460, 777
458:
459: \bibitem[Gayley \& Owocki(1995)]{GO95}
460: Gayley, K.G., Owocki, S.P. 1995, ApJ, 446, 801
461:
462: \bibitem[Grosdidier et al.(2003)]{GMA03}
463: Grosdidier, Y., Moffat, A.F.J., Acker, A. 2003, IAUS, 212, 192
464:
465: \bibitem[Hamann {\it et al.}(2008)]{H08}
466: Hamann, W.-R. {\it et al.} 2008, International Workshop on Clumping in
467: Hot-Star Winds, Potsdam, Germany, in press.
468:
469: \bibitem[Hamuy {\it et al.}(1992)]{Hetal92}
470: Hamuy, M., Walker, A. R., Suntzeff, N. B., Gigoux, P., Heathcote,
471: S. R., Phillips, M. M. 1992, PASP, 104, 533
472:
473: \bibitem[e.g. Henriksen(1994)]{H94}
474: Henriksen, R. N. 1994, \apss, 221, 25
475:
476: \bibitem[van den Heuvel(2007)]{H07}
477: van den Heuvel, E. P. J. 2007, ASPC, 367, 549
478:
479: \bibitem[L\'epine \& Moffat(1999)]{LM99}
480: L\'epine, S., Moffat, A.F.J. 1999, ApJ, 514, 909
481:
482: \bibitem[L\'epine {\it et al.}(1999)]{LEM99}
483: L\'epine, S., Eversberg, T., Moffat, A.F.J. 1999, ApJ, 117, 1441
484:
485: \bibitem[L\'epine {\it et al.}(1999)]{LSLZ99}
486: L\'epine, S., {\it et al.} 1999, ApJ, 522, 121
487:
488: \bibitem[L\'epine {\it et al.}(2001)]{L01}
489: L\'epine, S., {\it et al.} 2001, AJ, 122, 3407
490:
491: \bibitem[Maeder \& Meynet(2000)]{MM00}
492: Maeder, A., Meynet, G. 2000, ARA\&A, 38, 143
493:
494: \bibitem[Marchenko {\it et al.}(2006)]{Metal06}
495: Marchenko, S.V., Moffat, A.F.J., St-Louis, N., Fullerton, A.W. 2006,
496: ApJ, 639, L75
497:
498: \bibitem[Marchenko \& Moffat(2007)]{MM07}
499: Marchenko, S.V., Moffat, A.F.J. 2007, in Proc. of Workshop ``Massive
500: Stars in Interacting Binaries'', eds. N. St-Louis \& A.F.J. Moffat,
501: ASPC, 367, 213
502:
503: \bibitem[Marchenko et al.(2007)]{Metal07}
504: Marchenko, S. V., Foellmi, C., Moffat, A. F. J.,
505: Martins, F., Bouret, J.-C., Depagne, E. 2007, ApJ, 656, 77
506:
507: \bibitem[Mathewson {\it et al.}(2000)]{M00}
508: Mathewson, T. {\it et al.} 2000, AJ, 120, 1499
509:
510: \bibitem[Moffat \& Michaud(1981)]{MM81}
511: Moffat, A.F.J., \& Michaud, G. 1981, ApJ, 251, 133
512:
513: \bibitem[Moffat \& Robert(1994)]{MR94}
514: Moffat, A.F.J., \& Robert, C. 1994, ApJ, 421, 310
515:
516: \bibitem[Nelan {\it et al.}(2004)]{Netal04}
517: Nelan, E.P., Walborn, N.R., Wallace, D.J., Moffat, A.F.J., Makidon,
518: R.B., Gies, D.R., Panagia, N. 2004, AJ, 128, 323
519:
520: \bibitem[Owocki(1994)]{O94}
521: Owocki, S. P. 1994, Ap\&SS, 221, 3
522:
523: \bibitem[e.g. Robert (1992)]{R92}
524: Robert, C. 1992, PhD thesis, Universit\'e de Montr\'eal
525:
526: \bibitem[Smith(2007)]{S07}
527: Smith, N. 2007, in Proc. of Workshop ``Massive Stars: Fund. Par. \&
528: Circumstellar Interactions'', eds. G. Bosch, C. Cappa, P. Benaglia, in
529: press
530:
531: \bibitem[e.g. Stutzki(1993)]{S93}
532: Stutzki, R. 1993, Rev. in Modern Astronomy, no.6, p. 209.
533:
534: \bibitem[Williams {\it et al.}(1987)]{Wetal87}
535: Williams, P.M., van der Hucht, K.A., Th\'e, P.S. 1987, A\&A, 182, 91
536:
537: \end{thebibliography}
538:
539: \end{document}
540:
541: