1: %\documentclass[preprint]{aastex}
2: %\usepackage{emulateapj5}
3: %\input epsf
4:
5: %\documentclass[preprint2]{aastex}
6: \documentclass[12pt,preprint]{aastex}
7: %\documentclass[manuscript]{aastex}
8:
9: \newcommand{\Einstein}{{\it Einstein}}
10: \newcommand{\EUVE}{{\it EUVE}}
11: \newcommand{\Boo}{44{\it $\iota$\/}~Boo}
12: %\newcommand{\Capella}{{\it Capella}}
13: \newcommand{\Capella}{Capella}
14: \newcommand{\IUE}{{\it IUE}}
15: \newcommand{\EXOSAT}{{\it EXOSAT}}
16: \newcommand{\ROSAT}{{\it ROSAT}}
17: \newcommand{\ASCA}{{\it ASCA}}
18: \newcommand{\Chandra}{{\it Chandra}}
19: \newcommand{\XMM}{{\it XMM}}
20:
21: \newcommand{\lam}{$\lambda$}
22: \newcommand{\cmsq}{cm$^{-2}$}
23: \newcommand{\cmcube}{cm$^{-3}$}
24:
25: \begin{document}
26:
27: %DRAFT: \today
28:
29: \title{X-ray Doppler Imaging of \Boo\ with \Chandra}
30:
31: \author{N. S. Brickhouse, A. K. Dupree, and P. R. Young}
32: \affil{Harvard-Smithsonian Center for Astrophysics, 60 Garden
33: Street, Cambridge, MA 02138}
34: \email{nbrickhouse@cfa.harvard.edu, adupree@cfa.harvard.edu, pyoung@cfa.harvard.edu}
35:
36: \doublespace
37:
38: \begin{abstract}
39:
40: \Chandra\ High-Energy Transmission Grating observations of the bright
41: eclipsing contact binary \Boo\ show X-ray line profiles which are
42: Doppler-shifted by orbital motions. The X-ray emission spectrum
43: contains a multitude of lines superimposed on a weak continuum, with
44: strong lines of \ion{O}{8}, \ion{Ne}{10}, \ion{Fe}{17}, and
45: \ion{Mg}{12}. The profiles of these lines from the total observed
46: spectrum show Doppler-broadened widths of $\sim 550$ km s$^{-1}$. Line
47: centroids vary with orbital phase, indicating velocity changes of $>
48: 180$ km s$^{-1}$. The first-order light curve shows significant variability,
49: but no clear evidence for either primary or secondary eclipses. Flares
50: are observed for all spectral ranges; additionally, the light curve
51: constructed near the peak of the emission measure distribution ($T_e =
52: 5$ to $8 \times 10^6$ K) shows quiescent variability as well as
53: flares. The phase-dependences of line profiles and light curves
54: together imply that at least half of the emission is localized at high
55: latitude. A simple model with two regions on the primary star at
56: relatively high latitude reproduces the observed line profile shifts
57: and quiescent light curve. These first clear X-ray Doppler shifts of
58: stellar coronal material illustrate the power of \Chandra.
59:
60: \end{abstract}
61:
62: \keywords{atomic processes --- binaries: close --- stars: activity --- stars:
63: individual (\Boo) --- stars: late-type --- X-rays: stars}
64:
65: \section{Introduction}
66:
67: We report the first ``X-ray Doppler imaging'' of a late-type binary
68: system, made possible by the high spectral resolution and stable
69: wavelength scale of the \Chandra\ High-Energy Transmission Grating
70: (HETG). Phase-dependent line profile measurements of the contact
71: binary \Boo, together with temperature-selected light curves, provide
72: powerful new diagnostic capability.
73:
74: Optical Doppler imaging of rapidly rotating RS CVn systems has
75: demonstrated coronal activity patterns distinctly different from those
76: of the Sun, with relatively stable polar active regions co-existing
77: with more transient equatorial active regions (e.g. Vogt et al. 1999;
78: Donati \& Collier Cameron 1997). While optical Doppler imaging has
79: been applied to RS CVn systems for more than a decade, only recently
80: have Doppler images of a contact binary been reported, for the highly
81: variable VW~Cep, showing the presence of large polar spots, as well as
82: low latitude features (Hendry \& Mochnacki 2000).
83:
84: In X-rays, the spectral resolution has until now been inadequate for
85: line profile analysis of late-type stars. Eclipse mapping from light
86: curve analysis has been used to infer the sizes and locations of
87: flares (e.g. on Algol, Schmitt \& Favata 1999; and on VW Cep, Choi \&
88: Dotani 1998); however, it is generally difficult to disentangle
89: variability due to flaring from the modulation produced by either
90: rotation or eclipses. The \EUVE\ Deep Survey light curve for \Boo\
91: over 19.6 epochs shows no eclipses, but reveals quiescent modulation
92: at the 50\% level, associated with the rotation of the primary star;
93: furthermore, EUV density diagnostics from the \EUVE\ spectrometers
94: suggest high density, implying localized coronal regions with small
95: scale size $l \sim .004 R_{\star}$, which must be located at high
96: latitude (Brickhouse \& Dupree 1998; hereafter, BD98). Determining the
97: sizes and locations of coronal structures on contact binaries is an
98: important step toward understanding the rotation/activity correlation,
99: given that contact binaries are anomalously underluminous in X-rays,
100: presenting normalized X-ray fluxes ($L_x/L_{bol}$) 4 -- 5 times weaker than those of the fastest
101: rotating single stars (St{\c{e}}pien, Schmitt, \& Voges 2001).
102:
103: \section{Observations}
104:
105: \Boo\ (HD~133640) is a triple star system containing a contact binary
106: of the W~UMa type, located a distance of 12.76 pc (ESA 1998). X-ray
107: emission, first observed by Cruddace \& Dupree (1984), is expected to
108: originate with the contact binary components B and C of the
109: system. These two stars, of type G0 Vn (Gray, Napier, \& Winkler 2001),
110: orbit with a period of 6.4 hours and an
111: inclination of 72.8$^\circ$ (Hill, Fisher, \& Holmgren 1989).
112:
113: Time-tagged HETG spectra of \Boo, using the Advanced CCD Imaging
114: Spectrometer-S (ACIS-S) detector, were obtained by \Chandra\ on 2000
115: April 25 (ObsID 14). The system was observed for 59142 s, continuously
116: covering 2.56 epochs of the orbit. The event lists were
117: obtained from the Chandra X-ray Observatory Center (CXC) public
118: archive as Level 1 files, and were then processed, using CXC software CIAO V2.1 and
119: calibration database CALDB V5.0. High-Energy Grating (HEG) and
120: Medium-Energy Grating (MEG) spectra were extracted as described by
121: Canizares et al. (2001), and source-specific effective area curves
122: were generated. Positive and negative first-order spectra were
123: co-added.
124:
125: \section{Data Analysis}
126:
127: Here we link phase-dependent changes in the line profiles to the
128: phase-binned behavior of the light curve, by applying the
129: techniques of both eclipse mapping and X-ray Doppler imaging.
130: Emission lines from the available Fe ionization stages provide broad
131: coverage of electron temperature $T_e$ and allow construction of a
132: continuous emission measure distribution (EMD) model. The model EMD
133: allows us to determine the $T_e$ range
134: contributing to the emission lines of interest for profile analysis,
135: as well as to predict which other spectral features arise from the same
136: $T_e$ range. Photons from all of these features are then selected from the
137: event list to construct a new light curve characteristic of the
138: line-emitting $T_e$.
139:
140: \subsection{Total Spectrum and Emission Measure Distribution}
141:
142: The total first-order spectrum of \Boo\ shows strong lines of H-like
143: and He-like O, Ne, Mg, and Si, along with Fe L-shell lines from
144: several ionization states. A weak continuum is also apparent. Table~1
145: gives the measured line fluxes and linewidths for the four strongest
146: emission lines (\ion{O}{8}, \ion{Ne}{10},
147: and \ion{Mg}{12} Ly$_{\alpha}$ and \ion{Fe}{17} $2p^53d~^1P_1 - 2p^6~^1S_0$),
148: as determined from Gaussian fits to the
149: continuum-subtracted line profiles using the {\it Sherpa} package in
150: CIAO. Line profiles are wider than the instrumental FWHM (240 and 460
151: km s$^{-1}$ at 15 \AA, for HEG and MEG, respectively), with widths of $\sim$
152: 550 km s$^{-1}$ corresponding to the spread of velocities in the system.
153:
154: Strong, isolated Fe lines are used to construct the EMD: \ion{Fe}{17} $\lambda
155: \lambda$16.780; 17.051; 17.096, \ion{Fe}{18}I $\lambda \lambda$14.208;
156: 15.625, \ion{Fe}{19} $\lambda$14.664, \ion{Fe}{22} $\lambda$11.770,
157: and \ion{Fe}{23} $\lambda$11.736. The
158: method is described by BD98 and we use the spectral models from the Astrophysical
159: Plasma Emission Code (APEC; Smith et al. 2001). The EMD is characterized by a narrow
160: peak at $8 \times 10^6$ K, similar to that found by BD98, but with the
161: peak emission measure about 20\% lower and the emission measure for
162: $T_e > 10^7$ K about 50\% higher.
163:
164: Although the four strongest lines potentially represent a broad range
165: of $T_e$, as characterized by their temperature of maximum emissivity
166: $T_{max}$ (see Table~1), in the \Boo\ spectrum they are all formed
167: within a much narrower range. The EMD bump over the $T_e$ range, $T_L
168: = 5$ to $8 \times 10^6$ K, produces more than half of the flux for
169: each of these lines (65, 83, 92, and 68\%, respectively, for
170: \ion{O}{8}, \ion{Ne}{10}, \ion{Fe}{17}, and \ion{Mg}{12}). Models
171: show that the spectral region between 14 and 20 \AA\ is primarily
172: composed of line emission produced at $T_L$, and thus photons from
173: this spectral region are used to construct the $T_L$ light curve.
174:
175: \subsection{Light Curves}
176:
177: Figure~1 shows the optical light curve of Gherega et al. (1994), which
178: has subsequently been confirmed by eclipse timing measurements
179: (Pribulla, Chochol, \& Parimucha 1999; Albayrak \& Gurol 2001).
180: Primary eclipse (phase 0.00) corresponds to the eclipse of the
181: secondary by the primary. Also shown are the \Chandra\ total
182: first-order light curve and the $T_L$ light curve, limited to photons
183: falling between 14 and 20 \AA. The total light curve shows
184: significant variability, but no clear evidence for either primary or
185: secondary eclipses. A light curve constructed for high energy photons
186: ($\lambda < 8$ \AA; not shown) tracks the features of the total light
187: curve, suggesting that the largest variations in total count rate
188: are produced at the highest temperatures in the corona. The $T_L$
189: light curve shows a different pattern of variability.
190:
191: Four flares are identified in the total light curve, with the largest
192: flare producing a peak count rate 3 times higher than the total count
193: rate minimum. The $T_L$ light curve shows three of the flares, with
194: the largest flare producing only twice the count rate at peak compared
195: to the $T_L$ count rate minimum.
196:
197: Excluding the times of obvious flares, the $T_L$ light curve is
198: notable in several respects which may reflect its ``quiescent''
199: behavior: (1) The $T_L$ light curve is variable at the $\sim$20\%
200: level, consistent with the X-ray variability found with \ROSAT\
201: (McGale, Pye, \& Hodgkin 1996). (2) Narrow dips occur immediately
202: following each of the three primary eclipses, with remarkably similar
203: shapes on each occurrence (Fig.~1). These dips indicate that emission
204: is being absorbed, eclipsed, or rotated out of view. (3) A somewhat
205: broader dip is observed near one of the two secondary eclipses that
206: occur during the \Chandra\ pointing. This dip may also indicate
207: occultation or a region rotating out of view; however, such an
208: interpretation is not well supported because the other secondary eclipse
209: coincides with a flare, observed in the total light curve, but not in
210: the $T_L$ light curve. Perhaps a dip effectively masks the $T_L$
211: flare. (4) Excluding flares and dips, the $T_L$ light curve is
212: consistent with the \EUVE\ finding (BD98) of sinusoidal variation with
213: only one maximum, rather than two maxima as found in the optical light
214: curve. Periodicity at or near the optical period, as found in the 19.6
215: epoch long \EUVE\ observation, is not confirmed.
216:
217: \subsection{Phase-binned Line Profiles}
218:
219: Figure~2 shows the \ion{Ne}{10} and \ion{Fe}{17} line profiles summed
220: over four equal phase intervals centered at Phase 0.00, 0.25, 0.50,
221: and 0.75. Also shown are instrumental line profiles, centered at the
222: laboratory wavelengths. While the MEG profiles have greater
223: signal-to-noise, the factor of two greater resolving power of HEG
224: allows much better isolation of different velocity components. The
225: \ion{O}{8} and \ion{Mg}{12} profiles (not shown) are similar, but with
226: lower spectral resolution (HEG does not cover the \ion{O}{8} line) or
227: signal-to-noise ratio.
228:
229: The rest wavelengths of these lines are well known. The H-like
230: Ly$_{\alpha}$ wavelengths are taken as the weighted
231: average of the doublet, assuming they are optically thin and
232: collisionally excited; the \ion{Fe}{17} wavelength is 15.014 $\pm$
233: .001 \AA\ (Brown et al. 1998). The absolute wavelength scale of the
234: HETG is currently only accurate to about 100 km s$^{-1}$; however, the scale
235: appears to be very stable, such that {\it relative} wavelength
236: measurements are secure to better than 100 parts per million, or about
237: 30 km s$^{-1}$ (CXC Proposers' Observatory Guide 2001). The reality of
238: individual features in the line profiles is more difficult to assess,
239: but we expect that significant deviations from a Gaussian profile,
240: occurring over several bins, reflect real features. The profiles
241: contain multiple components representing a range of
242: velocities. Roughly one third of the emission is always present within
243: the instrumental line profile, and cannot be resolved. Excluding data
244: during the flare times (Fig.~1) does not affect these components.
245:
246: Median wavelengths are determined from MEG line profiles as the
247: interpolated center of the count distribution, with Poisson errors
248: assumed. Continuum and weaker background are not subtracted, as they
249: appear to contribute at most 20 counts to any of the total integrated
250: MEG lines. Figure~3 shows the shift of the median wavelengths for
251: each line, for the same four orbital phase bins as in
252: Figure~2. Centroids determined from separate MEG and HEG fits to
253: Gaussian line profiles are consistent with these medians, but give
254: larger errors where the profiles appear distinctly non-Gaussian. The
255: four ions studied show similar patterns of velocity shifts as a
256: function of orbital phase. Relative velocity changes with phase exceed
257: the instrumental uncertainties, and suggest net velocity changes of
258: 180 km s$^{-1}$ over the orbit. \ion{Ne}{10} has sufficient counts to verify
259: that the patterns of line shifts repeat from one epoch to the next.
260:
261: \section{Modeling the Coronal Activity}
262:
263: We present a simple 2-component model of the variability due to
264: quiescent coronal structures. The lack of eclipse signatures (two
265: maxima and two minima per orbit) severely limits the amount of
266: emitting material that can be evenly distributed near the surfaces of
267: both stars, since the equatorial regions would then be eclipsed.
268: Diffuse extended emission is also strictly limited, since the
269: integrated broad line profiles show distinct velocity components which
270: change with phase. The $T_L$ light curve shows about 20\% quiescent
271: modulation, implying that most of the material is always visible,
272: i.e. located at high latitude. Flares are not correlated with phase,
273: consistent with the active regions being visible most of the time.
274:
275: Our model attempts to match the observed features: (a) phase-dependent
276: line shifts; (b) broad sinusoidal $T_L$ light curve modulation found
277: from \EUVE\ (BD98) and confirmed with the present observation; and (c)
278: narrow dips in the $T_L$ light curve. Two regions contribute roughly
279: equally to the total emission in order to account for similar drops in
280: the count rate during the narrow dips and broad minima. Since
281: line centroids show velocity shifts consistent with the motion of the
282: primary star, we place both active regions on the primary star. They
283: are assumed to lie on the star's surface. The distinctive dip in the
284: light curve just after phase 0.0 can only be explained as a small
285: emitting region R1 on the inner face of the primary, which briefly
286: rotates out of view. R1 must extend below a latitude of 72.8$^\circ$
287: in order to be occulted, but not extend below $\approx$ 70$^\circ$ so
288: that the dip remains narrow. (Of course, if the material is extended
289: above the stellar surface, the latitude constraint is weakened.) For
290: the same reason R1 cannot extend more than $\approx 10^\circ$ in
291: longitude.
292:
293: To reproduce the broader modulation requires a larger region R2 on the
294: outer face of the primary in order to yield a light curve maximum at
295: phase $\approx 0.2$. R2 must also be at high latitude in order to
296: produce the weak $\sim 20$\% modulation. Figure~4 shows that the
297: light curve produced from a particular choice of the regions R1 and R2
298: mimics the key features of the Chandra $T_L$ light curve. Simulations
299: of the system with different active region locations and sizes,
300: including radially extended regions, are underway to study the effects
301: on light curves and line profiles.
302:
303: \section{Conclusions}
304:
305: The \Chandra\ observations of \Boo\ provide the first clear evidence
306: from line profile velocity measurements for localized activity at high
307: latitude on the primary star. Our interpretation is supported by the
308: phase-dependent light curves for emission produced at the temperature
309: of the EMD bump. Location of activity on the primary star has been
310: considered in the past to explain peculiarities and changes in the
311: photometric behavior of W UMa binaries, although a strong theoretical
312: basis for such a preference is lacking (Rucinski 1985). Coronal
313: enhancements that we find associated with the primary provide added
314: support for the dominance of the primary in magnetic activity as other
315: studies have indicated (BD98; Choi \& Dotani 1998; Barden 1985). These results demonstrate the power of \Chandra\ for
316: understanding coronal structure, using eclipse mapping and X-ray
317: Doppler imaging.
318:
319: \acknowledgements
320:
321: This work was supported in part by NASA NAG5-3559 and the Chandra
322: X-Ray Observatory Center NASA NAS8-39073 from NASA to the Smithsonian
323: Astrophysical Observatory and by Johns Hopkins Contract \#2480-60016.
324:
325: \begin{references}
326:
327: \reference{}
328: Albayrak, B., \& Gurol, B. 2001, IBVS, No. 5069
329:
330: \reference{}
331: Barden, S. C. 1985, \apj, 295, 162
332:
333: \reference{}
334: Brickhouse, N. S., \& Dupree, A. K. 1998 (BD98), \apj, 502, 918
335:
336: \reference{}
337: Brown, G. V., Beiersdorfer, P., Liedahl, D. A., Widmann, K.,
338: \& Kahn, S. M. 1998, \apj, 502, 1015
339:
340: \reference{}
341: Canizares, C. R., et al. 2001, \apj, 539, L41
342:
343: \reference{}CXC Proposers Observatory Guide V.3 2001, (Cambridge: Chandra X-ray
344: Observatory Center)
345:
346: \reference{}
347: Choi, C. S., \& Dotani, T. 1998, \apj, 429, 761
348:
349: \reference{}
350: Cruddace, R. G. \& Dupree, A. K. 1984, \apj, 177, 263
351:
352: \reference{}
353: Donati, J.-F. et al. 1992, \aap, 265, 682
354:
355: \reference{}
356: ESA, 1998, The Hipparcos Catalogue, ESA SP-1200 (Noordwijk: ESA)
357:
358: \reference{}
359: Favata, F., \& Schmitt, J. H. M. M. 1999, \aap, 350, 900
360:
361: \reference{}Gherega, O., Farkas, L., \& Horvath, A. 1994, IBVS, No. 4045
362:
363: \reference{}Gray, R. O., Napier, M. G., \& Winkler, L.~I. 2001, \apj,
364: 121, 2148
365:
366: \reference{}
367: Hendry, P. D., \& Mochnacki, S. W. 2000, \apj, 531, 467
368:
369: \reference{}Hill, G., Fisher, W. A., \& Holmgren, D. 1989, \aap,
370: 211, 81
371:
372: \reference{}
373: McGale, P. A., Pye, J. P., \& Hodgkin, S. T. 1996, MNRAS, 280, 627
374:
375: \reference{}
376: Pribulla, T., Chochol, D., \& Parimucha, S. 1999, IBVS, No. 4751
377:
378: \reference{}
379: Rucinski, S. M. 1985, in Interacting Binaries, ed. P. P. Eggleton \&
380: J. E. Pringle, (Boston: D. Riedel), 13
381:
382: \reference{}
383: Schmitt, J.~H. M. M., \& Favata, F. 1999, Nature, 410, 44
384:
385: \reference{}
386: Smith, R. K., Brickhouse, N. S., Raymond, J. C., \& Liedahl,
387: D. A. 2001, \apj, 556, L91
388:
389: \reference{}
390: St{\c{e}}pien, K., Schmitt, J. H. M. M., \& Voges, W. 2001,
391: \aap, 370, 157
392:
393: \reference{}
394: Vogt, S. S., Hatzes, A. P., Misch, A. A., \& Kurster, M. 1999, \apj
395: Supp, 121, 547
396:
397: \end{references}
398:
399: \clearpage
400:
401: \begin{deluxetable}{lrrrcccc}
402: \small
403: \tablecolumns{10}
404: \tablewidth{0pc}
405: \tablecaption{Strong Line Measurements}
406: \tablehead{
407: \colhead{Ion} &
408: \colhead{$\lambda$} &
409: \colhead{MEG} &
410: \colhead{HEG} &
411: \colhead{Flux} &
412: \colhead{MEG FWHM} &
413: \colhead{HEG FWHM} &
414: \colhead{$T_{max}$} \\
415: \colhead{} &
416: \colhead{(\AA)} &
417: \colhead{(Cnts)} &
418: \colhead{(Cnts)} &
419: \colhead{(ph cm$^{-2}$ s$^{-1})$} &
420: \colhead{(m\AA)} &
421: \colhead{(m\AA)} &
422: \colhead{(K)} }
423: \startdata
424: \ion{O}{8} & 18.969 & 721 & \nodata & $1.21 \pm 0.05 \times 10^{-3}$ & 34.2 $\pm$ 1.4 & \nodata & $3 \times 10^6$ \\
425: \ion{Ne}{10}\tablenotemark{a} & 12.134 & 1544 & 434 & $5.18 \pm 0.14 \times 10^{-4}$ & 29.4 $\pm$ 0.1 & 23.4 $\pm$ 1.4 & $5 \times 10^6$ \\
426: \ion{fe}{17} & 15.014 & 401 & 121 & $3.87 \pm 0.23 \times 10^{-4}$ & 33.2 $\pm$ 2.3 & 25.4 $\pm$ 3.6 & $5 \times 10^6$ \\
427: \ion{Mg}{12} & 8.421 & 347 & 163 & $5.94 \pm 0.38 \times 10^{-5}$ & 23.7 $\pm$ 1.7 & 17.2 $\pm$ 2.0 & $8 \times 10^6$
428:
429: \tablenotetext{a}{The contribution of Fe XVII $\lambda$12.124 is
430: estimated to be only $\sim 7$\%, based on ratios with \ion{Fe}{17} $\lambda$17.051.}
431: \enddata
432: \normalsize
433: \end{deluxetable}
434:
435:
436: \clearpage
437: \begin{figure}
438: \plottwo{f1a.ps}{f1b.ps}
439: \caption{{\it Left}: Count rate vs MJD (s) for 1000 s time
440: intervals. The uppermost curve is the total first-order light
441: curve. The middle curve represents the low temperature $T_L$ ($14$
442: \AA\ $< \lambda < 20 $ \AA) light curve (count rate $\times 3$). The
443: lowest curve is the scaled optical light curve, with the primary (P)
444: and secondary (S) minima marked. Four flares (F) are identified, with
445: vertical bars denoting their extent. {\it Right}: Count rate vs
446: orbital phase for the same $T_L$ light curve shown above. Data
447: ($\diamond$) are phase-folded over two epochs. Plus symbols ($+$) are
448: overplotted for the flare times marked with vertical bars on the
449: figure above ({\it left}). A sine curve is overlaid on the light curve
450: for illustration.
451: }
452: \end{figure}
453:
454: \begin{figure}
455: \plottwo{f2a.ps}{f2b.ps}
456: \caption{\ion{Ne}{10} ({\it left}) and \ion{Fe}{17} ({\it right}) MEG
457: and HEG line profiles (solid) for each phase. Binsizes are 0.005 and
458: 0.0025 \AA, respectively, and spectra are box-car smoothed over 3
459: bins. The zero points of the spectra are shifted up for each phase,
460: but the scales are the same. The exposure time is not the same for
461: each phase bin, so integrated counts will differ. For each profile,
462: the instrumental line profile (dash-dotted) is shown for comparison,
463: with peak set to the data peak and centroid set at the laboratory
464: wavelength.
465: }
466: \end{figure}
467:
468: \begin{figure}
469: \plotone{f3.ps}
470: \caption{Velocities as a function of orbital phase, as measured from
471: MEG profiles. Velocities are derived from the median centroid shifts
472: for the phase bins indicated by vertical dashed lines. The four
473: symbols represent \ion{O}{8} ($\diamond$), \ion{Ne}{10} ($\bullet$),
474: \ion{Fe}{17} ($\Delta$), and \ion{Mg}{12} ($\square$). Error bars
475: represent statistical errors only. The different symbols are shifted
476: slightly from their phase centers for plotting. Radial velocity
477: curves for the primary (solid) and secondary (dash-dotted) are from
478: Hill et al. (1989).
479: }
480: \end{figure}
481:
482:
483: \begin{figure}
484: \plotone{f4.eps}
485: \caption{Simulation of two active regions R1 and R2, described in
486: text, with model light curve.
487: }
488: \end{figure}
489:
490:
491:
492: \end{document}
493: