1: \documentclass[12pt,preprint]{aastex}
2: %% TM corrected 211006
3: %% manuscript produces a one-column, double-spaced document:
4:
5: %\documentclass[manuscript]{aastex}
6:
7: %% preprint2 produces a double-column, single-spaced document:
8:
9: %\documentclass[preprint2]{aastex}
10:
11: %% Sometimes a paper's abstract is too long to fit on the
12: %% title page in preprint2 mode. When that is the case,
13: %% use the longabstract style option.
14:
15: %% \documentclass[preprint2,longabstract]{aastex}
16:
17: \newcommand{\vdag}{(v)^\dagger}
18: \newcommand{\myemail}{sergey.marchenko@wku.edu}
19:
20: \shorttitle{Wind-clumping in SMC WR stars}
21: \shortauthors{Marchenko et al.}
22:
23: \begin{document}
24:
25: \title{Spectroscopy of SMC Wolf-Rayet Stars Suggests that Wind-Clumping does not Depend on Ambient
26: Metallicity}
27:
28: \author{S.V. Marchenko\altaffilmark{1}, C. Foellmi\altaffilmark{2,3}, A.F.J. Moffat\altaffilmark{4},
29: F. Martins\altaffilmark{5}, J.-C. Bouret\altaffilmark{6}, E. Depagne\altaffilmark{3,7}}
30:
31: \altaffiltext{1}{Department of Physics and Astronomy, Western Kentucky
32: University, 1906 College Heights Blvd., 11077, Bowling Green, KY
33: 42101-1077; sergey.marchenko@wku.edu}
34: \altaffiltext{2}{Laboratoire d'Astrophysique, Observatoire de Grenoble, BP 53,
35: 38041 Grenoble Cedex 9, France; cfoellmi@eso.org}
36: \altaffiltext{3}{European Southern Observatory, 3107 Alonso de Cordova, Casilla 19001,
37: Vitacura, Santiago, Chile}
38: \altaffiltext{4}{D\'epartement de Physique and
39: Observatoire du Mont M\'egantic, Universit\'e de Montr\'eal, CP 6128,
40: Succursale Centre-Ville, Montr\'eal, QC H3C 3J7, Canada; moffat@astro.umontreal.ca}
41: \altaffiltext{5}{Max-Planck-Institut f\"{u}r extraterrestrische Physik, Postfach 1312,
42: D-85741 Garching, Germany; martins@mpe.mpg.de}
43: \altaffiltext{6}{Laboratoire d'Astrophysique de Marseille, Traverse du Siphon - BP 8,
44: 13376 Marseille Cedex 12, France; jean-claude.bouret@oamp.fr}
45: \altaffiltext{7}{Departamento de Astronomía y Astrof\'{i}sica, Pontificia Universidad
46: Cat\`{o}lica de Chile, Campus San Joaqu\'{i}n, Vicu\~{n}a Mackenna 4860 Casilla 306,
47: Santiago 22, Chile; edepagne@eso.org}
48:
49: \begin{abstract}
50: The mass-loss rates of hot, massive, luminous stars are considered a decisive
51: parameter in shaping the evolutionary tracks of such stars and influencing the interstellar
52: medium on galactic scales. The small-scale structures (clumps) omnipresent in
53: such winds may reduce empirical estimates of mass-loss rates by an evolutionarily significant factor
54: of $\geq 3$. So far, there has been no {\it {direct}} observational evidence that
55: wind-clumping may persist at the same level in environments with a low
56: ambient metallicity, where the wind-driving opacity is reduced. Here we report
57: the results of time-resolved spectroscopy of three presumably single Population I Wolf-Rayet
58: stars in the Small Magellanic Cloud, where the ambient metallicity is $\sim 1/5 Z_\sun$.
59: We detect numerous
60: small-scale emission peaks moving outwards in the accelerating parts of the stellar winds.
61: The general properties of the moving features, such as their velocity dispersions,
62: emissivities and average accelerations, closely match the corresponding characteristics
63: of small-scale inhomogeneities in the winds of Galactic Wolf-Rayet stars.
64:
65: \end{abstract}
66:
67: \keywords{stars: mass loss --- stars: winds, outflows --- stars: Wolf-Rayet}
68:
69: \section{Introduction}
70:
71: Wolf-Rayet (W-R) stars, as a final evolutionary phase of hot, massive stars
72: ($M_{init} \geq 25 M_{\sun} $ for a $Z\sim Z_{\sun} $ ambient metallicity), are especially sensitive
73: to all phenomena which may influence their mass-loss rates. The fast, dense
74: winds of W-R stars are driven by radiation pressure exerted on multiple lines of mainly heavy
75: elements. Hence, W-R mass-loss rates should be sensitive to the
76: ambient metallicity content, as well as to the locally enhanced chemical composition of the wind.
77: The former sensitivity, though suggested from general principles
78: (Lamers \& Cassinelli 1999), has escaped detection until recently (see Crowther 2006,
79: Gr{\"a}fener \& Hamann 2006, and references therein).
80:
81: Line-driven winds are inherently unstable (Lucy \& Solomon 1970), being
82: constantly fragmented by numerous embedded shocks (Dessart
83: \& Owocki 2005). Such fragmentation may
84: change the local ionization balance and create a local non-monotonic velocity field,
85: thus providing a strong feedback to the driving force.
86: In the past, wind diagnostics
87: relied on models of smooth, homogeneous winds, until numerous spectroscopic
88: observations of Galactic W-R and OB stars
89: (Moffat et al. 1988; Robert 1992; L\'epine \& Moffat 1999; Eversberg et al. 1998;
90: Bouret et al. 2005; Fullerton et al. 2006 and references therein) demonstrated
91: the omnipresence of wind-embedded clumps, usually taking on the form of discrete
92: density enhancements outmoving with an accelerating wind. As an immediate
93: and long-lasting impact, the relatively simplistic treatment of structures in the
94: otherwise highly elaborate models of hot-star winds
95: (Hillier \& Miller, 1999; Puls et al. 2005; Hamann et al. 2006 ) resulted in a
96: consensual downward revision of mass-loss rates by a factor of $\geq 3$,
97: confirmed by numerous independent observations (e.g., Moffat \& Robert 1994).
98:
99: Though wind-clumping was widely anticipated to operate also in low-Z environments
100: (Hamann \& Koesterke 1998; Crowther
101: et al. 2002; Bouret et al. 2003), any {\it {direct}} proof was lacking. Here we report the detection of
102: outmoving spectral features in the line profiles of 3 SMC W-R stars. We compare the general
103: characteristics of these clumps to those in a Galactic sample of W-R stars, finding them to be
104: strikingly similar.
105:
106: \section{Observations}
107:
108: We targeted three presumably single (Foellmi et al., 2003) early-type WN stars in the SMC:
109: SMC-WR 1 (WN3ha), WR 2 (WN5ha) and WR 4 (WN6h), thus forming a representative sample of the small SMC W-R population
110: (12 known Population I W-R stars, most of them early-type WNs: Massey et al. 2003).
111: %These stars have spectral types WN3ha, WN4ha and WN6h (Foellmi et al. 2003).
112: %[Sergey: pls check these types!]
113: We monitored the stars in continuous, $\sim$1h 40 min-long loops for two consecutive
114: nights in August, 2006 (between HJD 2,453,974.538 and ...75.877), alternating among the 3 targets:
115: WR 4 - WR 2 - WR 1... .
116: We used the UVES spectrograph at the ESO-VLT-UT2 8-m telescope (Kueyen),
117: sampling the region $\lambda 3927-6031$\AA\ .
118: During the routine reduction we experimented with different binning factors, finding a
119: reasonable compromise between
120: desired spectral resolution, $\lesssim 0.5$\AA\ /pix, and signal-to-noise ratio, $S/N> 100$, by binning
121: the available spectra to 0.25\AA\ (the region around HeII 4686\AA\ ) and 0.31\AA\ (HeII 5412\AA\ ) bins.
122: With 40, 30 and 20 min exposures, this resulted in 5-6
123: spectra/night/star, with comparable signal-to-noise ratios
124: $S/N \simeq 120$, $\simeq 140$
125: and $S\simeq 160$ ($\pm 20$) for WR1, WR2 and WR4, respectively,
126: estimated from the adjacent continua around measured spectral lines.
127:
128: \section{Properties of the Clumps}
129:
130: The high quality of the spectra allowed us to apply a rather straightforward procedure to
131: detect the line-profile variability. We rectified the profiles of all relatively prominent HeII
132: lines (the only lines strong enough to provide sufficient precision, namely,
133: HeII 4686, 4859 and 5412 \AA\ ) by linear interpolation between the line-free regions
134: located in the immediate vicinity of a given line. Then we created unweighted average profiles
135: of each line for each night of observation and subtracted these averages from the individual
136: profiles. We found that all the major emission lines in all 3 stars can be considered as variable at a
137: level substantially exceeding the predicted instrumental noise level.
138:
139: For a proper comparison of the new results with previous studies of
140: inhomogeneities in the winds of Galactic W-R stars, we closely follow the already developed
141: approaches specifically designed to reveal basic properties of clumps.
142:
143: (i) First of all, we attempt to detect similarities in the variability patterns of major emission lines
144: by cross-correlating the difference (individual - average) spectra
145: of different line transitions. We use
146: the average cross-correlation functions (see L\'epine et al. 2000, equations 7, 8): namely,
147: we produce an unweighted average of the
148: individual (transition-to-transition) cross-correlation functions for a given star and a given spectral window
149: (usually, $\pm v_\infty$, centered on the transition's $\lambda _0+\Delta \lambda$, where $\Delta \lambda$
150: corresponds to the SMC's systemic velocity). The maxima of the
151: average cross-correlation functions reach 0.2-0.4, which is comparable to $I_{max}=0.2-0.8$ measured in the
152: CIII and CIV profiles of the Galactic star
153: WR 135. Taking into account
154: that (a) the considered transitions arise from different, though overlapping, line-forming regions,
155: (b) there is some hydrogen present in the $\lambda 4860$\AA\ blend, and (c) there might be some
156: transition-dependent optical-depth effects (see below), we did not anticipate a very high
157: degree of correlation. Nevertheless, for general consistency, plus a slight gain
158: in S/N, we combine the appropriately shifted (to the rest-frame of
159: the HeII 4686 transition) difference spectra of the three lines, smooth them with a
160: gaussian filter (FWHM=7 pix), linearly time-interpolate the neighboring spectra in order to fill
161: the unavoidable gaps, and plot the results in Fig. 1. The continuous, ${\bold {\rm V}}$-shaped
162: structures present in the difference spectra of the SMC W-R stars are remarkably
163: similar to the variations
164: seen in {\it all} appropriately studied Galactic WR stars (L\'epine \& Moffat 1999),
165: with the latter firmly linked to the presence of numerous
166: small-scale inhomogeneities moving outwards in the accelerating parts of the stellar winds.
167:
168: (ii) We proceed with measurements of the clump properties closely following the approaches
169: developed in L\'epine \& Moffat (1999) and in L\'epine et al. (2000). Namely, we use the
170: degradation function to estimate the average acceleration of the outmoving emission features
171: (Fig. 2) and the net intrinsic variability levels, $\sigma_{intr}$,
172: in different transitions (Fig. 3). We remind the reader that the degradation function
173: (L\'epine et al. 2000, equation 2) measures the mean standard deviation between
174: spectra separated by a [variable] interval $\Delta t$. In addition, this function also accounts for a possible
175: motion of the clumps (predominantly, away from the line's center), by including the average acceleration
176: as a free parameter. The levels of net intrinsic variability
177: are estimated across emission-line profiles (L\'epine et al. 2000, equations 4,5) by using the
178: variability in line-free continuum regions adjacent to emission lines and treating it as a
179: source of instrumental noise,
180: then adjusting this noise level by allowing for the difference in count statistics across the profile of
181: a prominent emission line.
182:
183: (iii) In order to compare our measurements of the individual
184: fluxes and full-widths at half-maxima of the detected clumps with corresponding values
185: for Galactic WR stars, we follow the general approach and published measurements from Robert
186: (1992), thus using only the HeII 5412\AA\ line. Namely, we measure the fluxes, maximum intensities
187: and FWHMs from the least-square fits of Gaussian profiles to the recognizable features in the
188: difference-spectra (individual spectrum - night average).
189: Fig. 2 shows that, within the
190: statistical uncertainties, which are larger for the SMC stars mainly due to the relatively lower
191: S/N, the properties of the Galactic clumps are indistinguishable from those in the SMC sample.
192: Note that for the estimates of the clump properties we use the unsmoothed difference-spectra.
193:
194: The broad spectral coverage of the data allows us to address the question of optical depth of the
195: clumps by comparing the intrinsic variability patterns in different transitions. One should note
196: (Fig. 3) the most instructive contrast between HeII 4686\AA\ (transition 3-4) and HeII 5412\AA\
197: (transition 4-7). In order
198: to make quantitative inter-comparisons of the intrinsic variability levels, we normalise $\sigma_{intr}$
199: by the line-profile intensity of the corresponding transition.
200: Both SMC-WR 1 and WR 2 show a reasonably consistent behavior in both transitions (Fig. 3;
201: compare to Fig. 6 from L\'epine et al. 2000), if
202: one takes into account the pronounced difference in the extensions of the line-forming regions: compare
203: the velocity extensions of the average profiles of HeII 4686\AA\ to their HeII 4859 and 5412\AA\ counterparts.
204: Surprisingly, the variability patterns of WR 4 come as a sharp contrast to the predictably congruent
205: behavior of the patterns in WR 1 and WR 2.
206:
207: \section {Discussion and Conclusions}
208:
209: First of all, we concentrate on the intrinsic line variability as related
210: to optical-depth effects introduced by the clumps. Relying on the differences in the
211: total optical depths of the studied transitions and assuming that these are also applicable to the
212: population of discrete density enhancements, one expects to detect different
213: $\sigma_{intr}$ dependencies across the profiles. Most notably, the significant optical
214: depth of HeII 4686\AA\ may result in a blue-shifted peak of $\sigma_{intr}$ relative to
215: HeII 4859\AA\ and HeII 5412\AA\ . Quite predictably, SMC-WR 1 and
216: WR 2 show little, if any, line-opacity effects: note the pronounced (Fig. 3) absorption components
217: in the profiles of HeII 4859\AA\ and HeII 5412\AA\ , which are generally linked to low
218: optical depths of the winds. Concurrently, L\'epine \& Moffat (1999) find only mild clump-related
219: optical-depth effects in the [optically thin] CIII 5696\AA\ line formed in the winds of
220: Galactic WC stars, along with the possible presence of opacity-related effects in HeII 5412\AA\
221: seen in the Galactic WN stars.
222: Hence, for WR 1 and WR 2 one may conclude that either (a) the clumps are optically thick in all
223: transitions {\it and} the winds have rather low volume filling factors
224: (e.g., Owocki \& Cohen 2006), or (b) all clumps are optically thin. Unfortunately, our present limited
225: data set does not allow one to choose between these two alternatives.
226: The only noticeable difference between WR 1 and WR 2 is the secondary maximum
227: in $\sigma_{intr}$ of HeII 4859\AA\ for WR 2, which may be caused by the presence of relatively high quantities of hydrogen in the wind.
228:
229: WR 4 tells a different story. The intensity of the HeII 4686\AA\ transition, as well as the lack of
230: developed P Cygni absorptions, calls for a denser wind. If the global properties of the wind are
231: related to the properties of clumps, then one may detect different wavelength
232: dependencies of $\sigma_{intr}$. This clearly pertains to WR 4, where the blueward displacement
233: of the HeII 4686\AA\ peak may point to a high optical density of clumps. However, the
234: disparity of the HeII 4859 and 5412\AA\ distributions prompts an alternative explanation:
235: the spectral variability could be, in part, related to the presence of co-rotating interacting
236: regions (CIRs: Massa et al. 1995; Morel et al. 1999, and references therein), as seen in several
237: Galactic W-R stars. One may notice
238: the synchronized {\it redward} migration of weak emission features in the first half of Night 1 (Fig. 1, dashed
239: lines: $\lambda \sim 4673$ shifting to $\lambda \sim 4676$, $\sim 4689$ gradually moving to $\sim 4693$,
240: and $\sim 4698$ moving to $\sim 4702$), as well as the gradual disappearance of the
241: broad emission feature at $\lambda \sim 4690$ during Night 2, both phenomena reminiscent of a
242: CIR. In addition, along with general expectations, WR 4 is a periodic photometric variable
243: (Foellmi et al. 2003: P=6.55 d), and not an obvious binary. This period is then likely related to rotation.
244: However, this does not mean that WR4 lacks clumps.
245:
246: One should also notice the lack of any dependence of clump
247: FWHMs on spectral class or terminal
248: velocity of the W-R winds (Fig. 2, lower section). The recent comprehensive
249: analysis of turbulence in colliding supersonic flows (Folini \& Walder 2006)
250: shows that the velocity dispersions in the shock-bound cold dense layers
251: (the clumps?) depend only on the upwind Mach-numbers, $M_u$.
252: We consider a `generic' W-R wind with $v_\infty=2000$ km/s,
253: estimating the dispersion of turbulent velocities as $\sigma(v_{turb}) \sim (0.1-0.2) v_\infty$
254: (Prinja et al. 1990; Marchenko \& Moffat 1999).
255: We assume an `asymptotic' wind temperature $T=2\cdot10^4$K. This provides
256: an estimate for the sound speed in the clump-formation zone,
257: $\sim 5-100 R_\star$ (see the models in L\'epine et al. 2000).
258: Using $M_{rms} \sim \eta^{-1/2} M_u$ with
259: density contrast
260: $\eta \sim 30$ (Folini \& Walder 2006), one may estimate the typical width of a clump as:
261: $FWHM \sim (1.4-2.7)$\AA\ , or 90-170 km/s at $\lambda = 4686$\AA\ . This rough estimate
262: falls tantalizingly close
263: to the measurements from Fig. 3, which give $FWHM=100-300$ km/s. One may match these
264: figures even more closely under
265: the assumption that the hydrodynamically-modeled density contrasts fall below the pre-specified
266: value of $\eta=30$, by adopting $\eta\sim10$ as frequently used in models of W-R spectra
267: (Hamann \& Koesterke 1998).
268:
269: Hence, quite justifiably, we favor the notion in which the clumps have
270: the $\sim$same internal velocity dispersion defined by compressible
271: turbulence in a supersonic flow, but different emissivity volumes
272: (cf. L\'epine \& Moffat 1999).
273:
274: Concluding our study, we can state that:
275:
276: \hskip -0.8cm (a) despite the differences in ambient metallicity, both the winds of Galactic and SMC
277: W-R stars show a clear presence of small-scale structures;\\
278: (b) the general properties of these structures, namely average acceleration rate, average
279: individual flux and average velocity dispersion inside the clump, are similar for
280: both populations of W-R stars;\\
281: (c) point (b) comes as a big surprise, considering that the wind-driving force (and thus
282: the radiatively-driven instability) sensitively depends
283: on the presence of heavy elements (Vink \& de Koter 2005);\\
284: (d) a wind-clumping factor must be included in the models of massive-star winds
285: in low-Z environments, just as in high-Z environments. Our understanding of the first population of stars in the
286: early Universe, mostly all very massive, as well as processes of enrichment of the interstellar
287: medium, will depend on reliable estimates of mass-loss rates, i.e., the
288: assumed wind-clumping factors. Indeed, wind-clumping could be rather extreme in a low-Z
289: environment: one may recall f=0.01 used in the models of MPG 324 (O4 V((f)) star in SMC: Bouret et al. 2003),
290: and f=0.06 for Br43 (WC4 star in LMC: Hamann \& Koesterke 1998), to be contrasted with the more modest $f\sim 0.1-0.3$
291: values for Galactic WR stars. However, note that the recent data on Galactic OB stars
292: also provide $f \ll 0.1$ (Bouret et al. 2005; Fullerton et al. 2006). This may signify that: either the
293: filling factors depend on the evolutionary status of massive stars, or the invoked wind-clumping factors
294: are biased by the model approach and the choice of diagnostic lines. The latter, in turn,
295: may be related to a [strong] radial dependence of the wind-clumping factor. This obviously calls for a
296: thorough investigation which is far beyond the scope of this paper.
297:
298: The extreme clumping may dramatically decrease
299: estimates of the mass loss rates, thus helping to retain the expected high rotational velocities
300: for the progenitors of the
301: long/soft $\gamma$-ray bursts, as well as affect the appearance of their immediate environments,
302: thus the observable reaction (echo) on the incoming burst.
303:
304: \acknowledgments
305: This paper is based on
306: observations made with ESO Telescopes at the Paranal Observatory
307: under programme ID 077.D-0029.
308: C.F. thanks C. Esparza for the efficient and friendly technical support during observations.
309: AFJM is grateful for financial assistance to NSERC (Canada) and FQRNT (Qu\'ebec).
310: FM acknowledges support from the Alexander von Humboldt foundation.
311: \clearpage
312:
313: \begin{thebibliography}{}
314:
315: \bibitem []{} Bouret, J.-C., et al. 2003, ApJ, 595, 1182
316: \bibitem []{} Bouret, J.-C., Lanz, T., \& Hillier, D.J. 2005, A\&A, 438, 301
317: \bibitem []{} Crowther, P.A. 2000, A\&A, 356, 191
318: \bibitem []{} Crowther, P.A. 2006, Stellar Evolution at Low Metallicity: Mass-Loss, Explosions, Cosmology,
319: H. Lamers, N. Langer, \& T. Nugis, ASP Conf. Series, 353, 157
320: \bibitem []{} Crowther, P.A., Dessart, L., Hillier, D.J., Abbott, J.B., \& Fullerton, A.W. 2002, A\&A, 392, 653
321: \bibitem []{} Dessart, L., \& Owocki, S. P. 2005, A\&A, 437, 657
322: \bibitem []{} Eversberg, T., L\'epine, S., \& Moffat, A.F.J. 1998, ApJ, 494, 799
323: \bibitem []{} Foellmi, C., Moffat, A. F. J., \& Guerrero, M. A. 2003, MNRAS, 338, 360
324: \bibitem []{} Folini, D., \& Walder, R. 2006, A\&A, in press (astro-ph/0606753)
325: \bibitem []{} Fullerton, A.W., Massa, D.L., \& Prinja, R.K. 2006, ApJ, 637, 1025
326: \bibitem []{} Gr{\"a}fener, G., \& Hamann, W.-R. 2006, Stellar Evolution at Low Metallicity: Mass-Loss, Explosions, Cosmology,
327: H. Lamers, N. Langer, \& T. Nugis, ASP Conf. Series, 353, 171
328: \bibitem []{} Hamann, W.-R., Gr{\"a}efener, G., \& Liermann, A. 2006, A\&A, in press
329: \bibitem []{} Hamann, W.-R., \& Koesterke, L. 1998, A\&A, 335, 1003
330: \bibitem []{} Hillier, D.J., \& Miller, D.L. 1999, ApJ, 519, 354
331: \bibitem []{} Lamers, H.J.G.L.M, \& Cassinelli, J.P. 1999, Introduction to Stellar Winds,
332: Cambridge: Cambridge University Press
333: \bibitem []{} L\'epine, S., \& Moffat, A.F.J. 1999, ApJ, 514, 909
334: \bibitem []{} L\'epine, S., et al. 2000, AJ, 120, 3201
335: \bibitem []{} Lucy, L.B., \& Solomon, P.M. 1970, ApJ, 159, 879
336: \bibitem []{} Marchenko, S.V., \& Moffat, A.F.J. 1999, A\&A, 341, 211
337: \bibitem []{} Massa, D., et al. 1995, ApJL, 452, L53
338: \bibitem []{} Massey, P., Olsen, K. A. G., \& Parker, J. W. 2003, PASP, 115, 1265
339: \bibitem []{} Moffat, A.F.J., et al. 1988, ApJ, 334, 1038
340: \bibitem []{} Moffat, A.F.J., \& Robert, C. 1994, ApJ, 421, 310
341: \bibitem []{} Morel, T., et al. 1999, ApJ, 518, 428
342: \bibitem []{} Owocki, S.P., \& Cohen, D.H. 2006, ApJ, 648, 565
343: \bibitem []{} Prinja, R.K., Barlow, M.J., \& Howarth, I.D.1990, ApJ, 361, 607
344: \bibitem []{} Puls, J., et al. 2005, A\&A, 454, 625
345: \bibitem []{} Robert, C. 1992, Ph.D. Thesis, Universit\'e de Montr\'eal
346: \bibitem []{} Vink, J.S., \& de Koter, A. 2005, A\&A, 442, 587
347: \bibitem []{} Willis, A.J., et al. 2004, ApJS, 154, 651
348:
349:
350: \end{thebibliography}
351:
352: \clearpage
353:
354: \begin{figure}
355: \plotone{f1.eps}
356: \caption{Grayscale plots of time-interpolated and smoothed
357: difference (individual - night-average) spectra of WR1 (left panels),
358: WR2 (central panels) and WR4 (right panels) for Night 1 (lower panels)
359: and Night 2 (upper panels). The intensity ranges are -0.03 (black) to +0.03
360: (white) in the local continuum units ($\equiv 1$). The dashed lines in the lower-right panel
361: mark the features presumably related to CIRs (see text). \label{fig1}}
362: \end{figure}
363:
364: \clearpage
365:
366: \begin{figure}
367: \epsscale{0.35}
368: \plotone{f2.eps}
369: \caption{The upper panel shows the average
370: accelerations of outmoving clumps in the winds
371: for SMC stars (filled circles), as well as the measured ranges for Galactic WC
372: (open squares) and WN (open triangles) stars (see L\'epine \& Moffat 1999).
373: The data are arranged by the corresponding
374: terminal velocities of the WR winds.
375: The middle panel shows the average fluxes
376: ($F$) of the clumps detected in the HeII 5412\AA\ line and normalized by
377: the maximum intensity of this emission profile ($I_{max} -1$), in order to be compared
378: with the measurements of Robert (1992).
379: The lower panel shows the average
380: FWHMs of the clumps observed in the HeII 5412\AA\ profile.
381: Note that WR 134 was omitted from the lower panel due to the very
382: high $FWHM\sim 800 km/s$ (see explanations in L\'epine \& Moffat 1999).
383: The $v_\infty$ estimates are rather uncertain for WR 1 and WR 2 (P. Crowther,
384: priv. comm.); for WR 4 they come from Crowther (2000) and Willis et al.
385: (2004). \label{fig2}}
386: \end{figure}
387:
388: \clearpage
389:
390: \begin{figure}
391: \epsscale{1.0}
392: \plotone{f3.eps}
393: \caption{The intrinsic variability across the line profiles in WR 1, WR2 and WR 4 (lower panels, left-to-right):
394: full lines correspond to the HeII 4686\AA\ line, dotted lines show HeII+H 4860\AA\ , dashed lines trace
395: HeII 5412\AA\ . The upper panels show average rectified profiles of the transitions. \label{fig2}}
396: \end{figure}
397:
398: \end{document}
399: