1: \input{psfig.sty}
2: \documentclass[12pt,preprint]{aastex}
3: \usepackage{graphicx}
4:
5: \shorttitle{Optical Depth in stellar coronae}
6: \shortauthors{Testa et al.}
7: \usepackage{amsmath}
8:
9:
10: %\received{2004 January 26}
11: \begin{document}
12: \title{Detection of X-ray Resonance Scattering in Active Stellar Coronae}
13: \author{Paola Testa\altaffilmark{1,2,3}, Jeremy
14: J. Drake\altaffilmark{2}, Giovanni Peres\altaffilmark{3} and Edward
15: E.\ DeLuca\altaffilmark{2}}
16: \altaffiltext{1}{SAO Predoctoral Fellow}
17: \altaffiltext{2}{Smithsonian Astrophysical Observatory, MS 3, 60 Garden
18: Street, Cambridge, MA 02138, USA; jdrake@head.cfa.harvard.edu}
19: \altaffiltext{3}{DSFA, Sezione di Astronomia, Universit\`a di Palermo
20: Piazza del Parlamento 1, 90134 Palermo, Italy}
21:
22: \begin{abstract}
23: An analysis of Lyman series lines arising from hydrogen-like oxygen
24: and neon ions in the coronae of the active RS~CVn-type binaries II~Peg
25: and IM~Peg, observed using the {\it Chandra} High Resolution
26: Transmission Grating Spectrograph, shows significant decrements in the
27: Ly$\alpha$/Ly$\beta$ ratios as compared with theoretical predictions
28: and with the same ratios observed in similar active binaries. We
29: interpret these decrements in terms of resonance scattering of line
30: photons out of the line-of-sight; these observations present the first
31: strong evidence for this effect in active stellar coronae. The net
32: line photon loss implies a non-uniform and asymmetric surface
33: distribution of emitting structures on these stars. Escape
34: probability arguments, together with the observed line ratios and
35: estimates of the emitting plasma density, imply typical line-of-sight
36: sizes of the coronal structures that dominate the X-ray emission
37: of $10^{10}$~cm at temperatures of $3\times 10^6$~K and of $10^8$~cm at
38: $10^7$~K. These sizes are an order of magnitude larger than predicted
39: by simple quasi-static coronal loops models, but are still very small
40: compared to the several $10^{11}$~cm radii of the underlying stars.
41: \end{abstract}
42:
43: \keywords{Radiative transfer --- X-rays: stars --- stars: coronae
44: --- stars: late type}
45:
46: \section{Introduction}
47: \label{s:intro}
48:
49: The Sun is the only star for which we can presently image coronal
50: X-ray emission, and in order to understand how coronae on other stars
51: might be structured we must resort to indirect means. One potentially
52: powerful diagnostic of the characteristic size of X-ray emitting
53: regions is line quenching through resonance scattering. The escape
54: probability of a photon emitted by a resonance line in a low density
55: homogeneous plasma is dependent on the line-of-sight path length
56: through the plasma region. Provided the plasma density and the abundance
57: fraction of the ion in question are known, in the regime where the
58: plasma is only marginally thick the comparison of strong and weak lines
59: subject to different scattering losses can yield an estimate of the
60: typical emitting region path length.
61:
62: Resonance scattering in the solar corona has been studied
63: predominantly in the light of the strong ($gf=2.66$) $2p^6\, ^1S_0
64: \rightarrow 2p^5 3d ^1P_1$ resonance line of Fe~XVII at 15.02~\AA\ as
65: compared to nearby weaker Fe~XVII lines, though with controversial
66: results concerning whether optical depth effects were seen or not
67: (Phillips et al.\ 1996; Phillips et al.\ 1997; Schmelz
68: et al.\ 1997; Saba et al.\ 1999).
69: Studies of the same
70: transition seen in different stars observed at high resolution
71: ($\lambda /\Delta\lambda \sim 1000$) by {\it Chandra} have also
72: recently been presented by Phillips et al.\ (2001) and Ness et
73: al.\ (2003). Both stellar studies failed to find evidence for
74: significant optical depth.
75:
76: There are two problems with using the prominent Fe~XVII soft X-ray
77: complex for optical depth studies. Recently, it has been shown by
78: Doron \& Behar\ (2002) and Gu\ (2003) that the indirect processes of
79: radiative recombination, dielectronic recombination, and resonance
80: excitation involving the neighbouring charge states are important for
81: understanding the relative strengths of Fe~XVII--XX lines. Secondly,
82: the coronae of active stars have been found to be Fe-poor by factors
83: of up to 10 compared with a solar or local cosmic composition (e.g.\
84: reviews by Drake et al.\ 2003, Audard et al.\ 2003), reducing the
85: sensitivity of Fe lines as optical depth indicators. Indeed, the
86: spectral lines likely to exhibit the largest optical depths in the
87: coronae of active stars are the Lyman $\alpha$ lines of hydrogen-like
88: O and Ne---elements that are often seen to be enhanced relative to Fe
89: (e.g.\ Drake et al.\ 2001, Audard et al.\ 2003).
90:
91:
92: In this {\em Letter}, we present an analysis of
93: the Ly$\alpha$ to Ly$\beta$ line strength ratios in the active
94: binaries II~Peg and IM~Peg, and show that the Ly$\alpha$ lines
95: of Ne and O are significantly quenched. We use inferred optical
96: depths to make the first direct estimates of the dimensions of
97: coronal structures in active stars.
98:
99:
100: \section{Observations and Analysis}
101: \label{s:analysis}
102:
103: Observations and relevant characteristics of the program stars are
104: summarised in Table~\ref{tab1}. All spectra were obtained by the {\em
105: Chandra} High Energy Transmission Grating (HETG) in conjunction with
106: the Advanced CCD Imaging System spectroscopic detector (ACIS-S), and
107: were downloaded from the {\em Chandra} Data
108: Archive\footnote{http://cxc.harvard.edu/cda}. Spectra from multiple
109: observations (IM~Peg and AR~Lac) were combined, and positive and
110: negative orders were summed, keeping HEG and MEG spectra separate.
111: The same observations of HR~1099, II~Peg and AR~lac are described in
112: more detail by Drake et al.\ (2001), Huenemoerder et al.\ (2001) and
113: Huenemoerder et al.\ (2003), respectively.
114:
115: Resonance scattering of Ly$\alpha$ and Ly$\beta$ photons can be
116: diagnosed by comparison of the measured Ly$\alpha$/Ly$\beta$ ratio
117: with respect to the theoretical ratio. Spectral line fluxes were
118: measured using the FITLINES utility in the
119: PINTofALE\footnote{http://hea-www.harvard.edu/PINTofALE/}
120: IDL\footnote{Interactive Data Language, Research Systems Inc.}
121: software package (Kashyap \& Drake\ 2000). In the case of O~VIII, the
122: Ly$\beta$ transition is blended with an Fe~XVIII line
123: ($2s^22p^5~^2P_{3/2} - 2s^22p^4(^3P)3s~^2P_{3/2}$,
124: $\lambda=16.004$~\AA). We estimated its intensity by scaling the
125: observed intensity of the neighbouring Fe~XVIII 16.071~\AA\
126: ($2s^2 2p^5~^2P_{3/2} - 2s^2 2p^4(^3P)3s~^4P_{5/2}$)
127: transition, that shares the same upper level, by the ratio of their
128: theoretical line strengths (0.76) as predicted by the APED database
129: (Smith et al.\ 2001).
130: Uncertainties involved in this deblending procedure are negligible for
131: II~Peg (whose Fe~XVIII is weak) and similar to or less than Poisson
132: errors for the other stars for, e.g., an error of $\sim 20\%$ in the
133: theoretical Fe XVIII ratio. He-like resonance lines of Ne~IX and
134: O~VII were also measured in order to derive a temperature estimate
135: from the Ly$\alpha$/He-like~$r$ ratio. Measured line fluxes and
136: statistical errors are listed in Table~\ref{tab2}.
137:
138: Observed and theoretical O and Ne Ly$\alpha$/Ly$\beta$ ratios are
139: compared in Figure~\ref{fig1}. The observed ratios are shown at the
140: temperatures at which observed H-like to
141: He-like $2p$-$1s$ line intensities matched their theoretical (APED)
142: values (see Table~\ref{tab3}).
143: While it is an approximation to assume that the lines are
144: formed at a single temperature, this is not critical because the
145: temperature dependences of these Ly$\alpha$/Ly$\beta$ ratios are not
146: steep for $T > 2$~MK. Both O and Ne H-like
147: lines are formed at temperatures significantly lower than those at
148: which the theoretical Ly$\alpha$/Ly$\beta$ ratios reach their
149: asymptotic limits.
150:
151: The observed Ly$\alpha$/Ly$\beta$ ratios are lower than the
152: theoretical values for both the Ne~X and the O~VIII lines in the case
153: of IM~Peg, and for O~VIII in II~Peg.
154: Huenemoerder et al.\ (2001) noticed similar
155: discrepancies in II~Peg but excluded optical depth effects on the grounds that
156: this deviation was less than $\sim 2\sigma$ from the theoretical
157: value; it appears, however, that this assessment was based on the high
158: temperature asymptotic ratio of $\sim 6.25$ (in photon
159: units). Our measured ratio for II~Peg is instead more than 3$\sigma$
160: lower than the expected ratio at the temperature of formation of the
161: O~VIII lines. In the case of IM~Peg, the observed O~VIII ratio
162: lies $\sim 1.8 \sigma$ below the theoretical value, while the Ne~X
163: ratio in both HEG and MEG, are $>3\sigma$ lower.
164: For comparison with IM~Peg and II~Peg, we also present similar results
165: for AR~Lac and HR~1099: these ratios instead do not show significant
166: departure from the expected optically thin values.
167:
168: Though uncertainties in the theoretical ratio are not included,
169: these are not expected to exceed 10\%\ based on good agreement with
170: recent laboratory experiments (G.~Brown, private communication;
171: Beiersdorfer\ 2003).
172: If we assume a 10\%\ error in theoretical
173: ratios, the Ne Ly$\alpha$/Ly$\beta$ of IM~Peg still departs by
174: $3\sigma$ for MEG ($\sim 1.7 \sigma$ for HEG), while the O ratios
175: lie $2.5\sigma$ for II~Peg and $1.3\sigma$ for IM~Peg below the
176: theoretical curve.
177: In Figure~\ref{fig1},
178: we also plot the O~VIII Ly$\alpha$/Ly$\beta$ ratios obtained from
179: the line fluxes measured by Raassen et al.\ (2002) from
180: {\em Chandra}/LETGS and {\em XMM-Newton}/RGS1 spectra of Procyon.
181: In the Procyon spectrum Fe~XVIII is not detected and these data provide
182: further validation of the APED O~VIII Ly$\alpha$/Ly$\beta$ ratio.
183:
184: These quantitative results are reinforced by a visual comparison of
185: spectra in Figure~\ref{fig2}, in which the Ly$\alpha$ lines in II~Peg
186: and IM~Peg are visibly weaker relative to Ly$\beta$ that those of
187: AR~Lac and HR~1099. There do not appear to be plausible explanations
188: for the discrepant ratios other than by the quenching of Ly$\alpha$
189: relative to Ly$\beta$ through resonance scattering within the emitting
190: plasma. Intervening photoelectric absorption could cause similar
191: effects but H column densities of order a few $10^{21}$~cm$^2$ would
192: be required---two or three orders of magnitude higher than inferred
193: for these stars (Mewe et al.\ 1997; Mitrou et al.\ 1997).
194:
195:
196: \subsection{Path Length Estimate}
197:
198: Our measured line ratios allow us to derive an effective optical depth
199: $\tau$, and a typical photon path length within the emitting plasma.
200: We use the escape probability, $p(\tau)$, derived by Kastner \&
201: Kastner\ (1990), which for $\tau \lesssim 50$
202: can be approximated by (e.g.\ Kaastra \& Mewe\ 1995)
203: \begin{equation}
204: p(\tau) \sim \frac{1}{1+0.43~\tau}. \label{eq:ptau}
205: \end{equation}
206: This is the escape probability for line photons emitted at optical
207: depths between 0 and $\tau$, averaged over a Gaussian line profile
208: due to thermal Doppler broadening, and it assumes that each scattered
209: photon is completely lost from the line of sight.
210: This probability is significantly larger than the $e^{-\tau}$
211: transmittance of the simple absorption case, since it assumes
212: emission over the whole line of sight through the plasma.
213: The line center optical depth, $\tau$, can be written (e.g.,
214: Acton\ 1978):
215: \begin{equation}
216: \tau = 1.16 \cdot 10^{-14} \cdot \frac{n_{\rm i}}{n_{\rm el}}
217: A_{\rm Z} \frac{n_{\rm H}}{n_{\rm e}} \lambda f
218: \sqrt{\frac{M}{T}} n_{\rm e} \ell
219: \label{eq:tau}
220: \end{equation}
221: for ion fraction $n_{\rm i}/n_{\rm el}$ (from
222: Mazzotta et al.\ 1998), element abundance $A_{\rm Z}$,
223: oscillator strength $f$, temperature $T$, electron density
224: $n_{\rm e}$, atomic weight $M$, where
225: $n_{\rm H}/n_{\rm e} \sim 0.85$, and $\ell$ is
226: the total path length along the line of sight through the
227: emitting plasma. In the above, neglect of non-thermal
228: broadening should not be important: non-thermal velocities
229: of, e.g., 50~km~s$^{-1}$---similar to the thermal velocities
230: for O and Ne at their characteristic temperatures of
231: formation---would result in $\ell$ being underestimated by
232: a factor of 1.4.
233:
234: Electron densities were adopted from the survey of Testa et al.\ (2004;
235: hereafter Paper~I): for $\tau({\rm O~VIII})$ we assumed the
236: $n_{\rm e}$ derived from the O~VII He-like triplet; the Ne~IX
237: triplet is strongly affected by blends with Fe lines and so for
238: $\tau({\rm Ne~X})$ we assumed the $n_{\rm e}$ derived from
239: the Mg~XI He-like triplet that forms at a similar temperature. In the
240: case of IM~Peg, the low signal in the O~VII intercombination and
241: forbidden lines precludes a firm density estimate; we therefore
242: assumed a value of $2\times 10^{10}$~cm$^{-3}$, which is typical of
243: values obtained for all the measurable spectra of RS~CVns in Paper~I.
244: The assumed $n_{\rm e}$ are listed in Table~\ref{tab3}, while the
245: adopted O and Ne abundances are listed in Table~\ref{tab4}.
246:
247: In order to derive an estimate of the path length, $\ell$, we treated
248: the fine structure components of Ly$\alpha$ (1:~$^2P_{3/2} \rightarrow
249: ^2S_{1/2}$ at $\lambda=18.9671$~\AA\ and 2:~$^2P_{1/2} \rightarrow
250: ^2S_{1/2}$ at $\lambda=18.9726$~\AA) separately, since their splitting
251: ($\Delta \lambda/ \lambda \sim 0.0003$) is larger than the thermal
252: width ($\sim \sqrt{k_BT/M}/c \sim 0.00013$). The observed and
253: theoretical intensities, $I_{\rm obs}$ and $I_{\rm th}$,
254: are then related as follows:
255:
256: \begin{equation}
257: I_{Ly\alpha~{\rm obs}} = I_{1~{\rm obs}}+ I_{2~{\rm obs}} =
258: \frac{I_{1~{\rm th}}}{1+0.43 \tau_1}
259: + \frac{I_{2~{\rm th}}}{1+0.43 \tau_2} .
260: \end{equation}
261: Since for hydrogenic ions $f_2=f_1/2=0.2776/2$ (e.g.\ Morton\
262: 2003), $I_{2~{\rm th}}=I_{1~{\rm th}}/2$,
263: and $\tau_i \sim C(\ell) \cdot f_i$
264: (see Eq.~\ref{eq:tau}), we obtain:
265:
266: \begin{equation}
267: \frac{I_{Ly\alpha~{\rm obs}}}{I_{Ly\alpha~{\rm th}}} = \frac{1}{3}
268: \left[\frac{2}{1+0.43 C(\ell) f_1}
269: + \frac{1}{1+0.43 C(\ell) f_1/2}\right] .
270: \end{equation}
271: For Ly$\beta$ the splitting of the components is smaller than the
272: thermal width and the escape probability for Ly$\beta$ is given
273: directly by Eq.\ref{eq:ptau}.
274: The combined equation for $C(\ell)$ was then solved to obtain the
275: path length.
276:
277: The path length estimates, $\ell_{\tau}$, derived from the
278: measurements are listed in Table~\ref{tab4}.
279: For comparison, we also list the stellar radii and loop lengths
280: expected for a standard hydrostatic loop model (e.g.\ Rosner et
281: al.\ 1978, RTV hereafter), $L_{\rm RTV}$, corresponding to the
282: observed temperatures and densities.
283:
284:
285: \section{Discussion and Conclusions}
286: \label{s:discuss}
287:
288: The discrepant O~VIII and Ne~X Ly$\alpha$/Ly$\beta$ ratios found here
289: for II~Peg and IM~Peg represent the first clear evidence of resonant
290: scattering in coronal X-ray emission lines.
291: The photon path lengths inferred from the observed ratios
292: (Table~\ref{tab4}) are:
293: (1) about two orders of magnitude different from each other,
294: reflecting the differences in the plasma densities found in Paper~I
295: for the characteristic temperatures of formation of the O~VIII and
296: Ne~X lines;
297: (2) very small with respect to the stellar radius;
298: (3) an order of magnitude larger than the loop lengths derived
299: from RTV model scaling laws. In the ratio
300: $\ell_{\tau}/L_{\rm RTV}$ the $n_e$ terms cancel, so that this
301: conclusion is completely independent of plasma density measurements.
302:
303: We note that the path lengths obtained from this type of analysis
304: should be treated as lower limits because addition of Ly$\alpha$
305: photons by any optically thin emission or any scattering
306: into the line-of-sight are not taken into account. In the case of a
307: uniform spherically-symmetric arrangement of emitting structures over
308: the stellar surface, scattering into and out of the line-of-sight
309: would be expected to compensate one another. The detection of
310: resonance scattering implicitly suggests a non-uniform coronal
311: distribution.
312:
313: That we find different optical depths in emitting regions on what are
314: ostensibly similar active RS~CVn-type binaries is puzzling. One
315: possibility is that this is transient behaviour that might occur on
316: all similarly active stars for a preferential arrangement of one or
317: more coronal structures. We also note that, whereas both AR~Lac and
318: HR~1099 comprise two late-type stars of more similar type, II~Peg and
319: IM~Peg have unseen companions of unknown spectral type, though how
320: this arrangement might benefit larger coronal photon path lengths
321: is not obvious.
322:
323:
324: \begin{acknowledgements}
325: We thank the referee, Jeffrey L. Linsky, for helpful comments
326: that enabled us to improve the manuscript.
327: PT was partially supported by {\it Chandra} grants GO1-20006X
328: and GO1-2012X under the SAO Predoctoral Fellowship program.
329: JJD was supported by NASA contract NAS8-39073 to the
330: {\em Chandra X-ray Center}; EED was supported by NASA grant
331: NAG5-10872. GP and PT were partially supported by MIUR and
332: by ASI.
333: \end{acknowledgements}
334:
335: \clearpage
336:
337: \input{tab1}
338:
339: \clearpage
340:
341: \input{tab2}
342:
343: \clearpage
344:
345: \input{tab3}
346:
347: \clearpage
348:
349: \input{tab4}
350:
351: \clearpage
352:
353: \begin{figure}[!ht]
354: \rotatebox{90}{
355: \epsscale{0.45}
356: \plotone{f1b.eps}
357: \plotone{f1a.eps}}
358: \caption{O~VIII ({\em top}) and Ne~X ({\em bottom}) Ly$\alpha/$Ly$\beta$
359: ratios for the analyzed sources vs.\ the temperature derived from
360: the Ly$\alpha/$He-like~$r$ diagnostics.
361: The solid curve is the theoretical ratio from the APED database.
362: For the Ne~X we show the results from both the HEG (filled symbols)
363: and the MEG (empty symbols) measurements; the HEG ratios are
364: shifted by +5\% on the T axis in order to distinguish
365: the error bars corresponding to HEG and MEG measurements for the
366: same source.
367: \label{fig1}}
368: \end{figure}
369:
370: \clearpage
371:
372: \begin{figure}[!ht]
373: \rotatebox{90}{
374: \epsscale{0.3}
375: \plotone{f2b.epsi}
376: \plotone{f2a.epsi}
377: }
378: \caption{Comparison of the spectra for sources showing optical
379: depth effects and sources appearing effectively optically thin.
380: {\em Top --} O~VIII Ly$\alpha,\beta$ spectral region for
381: II~Peg (filled profile) and HR~1099 (shifted in $\lambda$
382: by +0.06\AA~for better readability). Spectra are normalised
383: such that they have the same {\em unblended} Ly$\beta$ line
384: strength, i.e.\ after correction for the Fe~XVIII line flux
385: (see text). The different strengths of the Fe line complex
386: near 17~\AA\ reflects primarily a different Fe/O abundance
387: ratio in the two stars (Drake et al.\ 2001, Huenemoerder et
388: al.\ 2001).
389: {\em Bottom --} Ne~X Ly$\alpha,\beta$ spectral region for
390: IM~Peg (filled spectrum) and AR~Lac (shifted in $\lambda$
391: by +0.03\AA~for better readability). Spectra are again
392: normalized to the observed intensities of the Ly$\beta$ line.
393: \label{fig2}}
394: \end{figure}
395:
396: \clearpage
397:
398:
399:
400: \begin{thebibliography}{}
401: \bibitem[]{acton78} Acton, L.W.\ 1978, \apj, 225, 1069
402: \bibitem[]{audard03} Audard, M., G{\" u}del, M., Sres, A.,
403: Raassen, A.J.J. \& Mewe, R.\ 2003, \aap, 398, 1137
404: \bibitem[]{beiersd} Beiersdorfer, P.\ 2003, \araa, 41, 343
405: \bibitem[]{berd1998} Berdyugina, S.V.,
406: Jankov, S., Ilyin, I., Tuominen, I., Fekel, F.C.\
407: 1998, \aap, 334, 863
408: \bibitem[]{berd1999} Berdyugina, S.V.,
409: Ilyin, I., Tuominen, I.\ 1999, \aap, 347, 932
410: \bibitem[]{doron2002} Doron, R. \& Behar, E. \ 2002, \apj, 574, 518
411: \bibitem[]{drake03} Drake, J.J.\ 2003, Advances in
412: Space Research, 32, 945
413: \bibitem[]{dra2001} Drake, J.J., Brickhouse, N.S., Kashyap, V.,
414: Laming, J.M., Huenemoerder, D.P., Smith, R., Wargelin,
415: B.J. \ 2001, \apjl, 548, L81
416: \bibitem[]{gu2003} Gu, M.F. \ 2003, \apj, 593, 1249
417: \bibitem[]{Huen2001} Huenemoerder, D.P.,
418: Canizares, C.R., Schulz, N.S.\ 2001, \apj, 559, 1135
419: \bibitem[]{Huen2003} Huenemoerder, D.P., Canizares, C.R.,
420: Drake, J.J., Sanz-Forcada, J.\ 2003, \apj, 595, 1131
421: \bibitem[]{kaastra95} Kaastra, J.S. \& Mewe, R.\ 1995, \aap, 302, L13
422: \bibitem[]{kastner90} Kastner, S.O. \& Kastner, R.E.\ 1990, JQSRT, 44, 275
423: \bibitem[]{kash2000} Kashyap, V. \& Drake, J.J. \ 2000, Bull.
424: Astron. Soc. India, 28, 475
425: \bibitem[]{mazzotta98} Mazzotta, P., Mazzitelli, G., Colafrancesco, S.
426: \& Vittorio, N.\ 1998, \aaps, 133, 403
427: \bibitem[]{mewe97} Mewe, R., Kaastra, J.S., van den Oord, G.H.J.,
428: Vink, J., Tawara, Y.\ 1997, \aap, 320, 147
429: \bibitem[]{mitrou97} Mitrou, C.K., Mathioudakis, M., Doyle, J.G.
430: \& Antonopoulou, E.\ 1997, \aap, 317, 776
431: \bibitem[]{morton03} Morton, D.C.\ 2003, \apjs, 149, 205
432: \bibitem[]{ness2003} Ness, J.-U., Schmitt, J.H.M.M., Audard, M.,
433: G{\" u}del, M., Mewe, R.\ 2003, \aap, 407, 347
434: \bibitem[]{phill2001} Phillips, K.J.H., Mathioudakis, M.,
435: Huenemoerder, D.P., Williams, D.R., Phillips, M.E.,
436: Keenan, F.P.\ 2001, \mnras, 325, 1500
437: \bibitem[]{phill1997} Phillips, K.J.H.,Greer, C.J., Bathia, A.K.,
438: Coffey, I.H., Barnsley, R. \& Keenan, F.P.\ 1997, \aap, 324, 381
439: \bibitem[]{phill1996} Phillips, K.J.H., Greer, C.J., Bathia, A.K.
440: \& Keenan, F.P.\ 1996, \apjl, 469, L57
441: \bibitem[]{raassen02} Raassen, A.J.J. et al.\ 2002, \aap, 389, 228
442: \bibitem[]{RTV} Rosner, R., Tucker, W.H. \& Vaiana, G.S.\ 1978,
443: \apj, 220, 643
444: \bibitem[]{saba} Saba, J.L.R., Schmelz, J.T., Bathia, A.K. \&
445: Strong, K.T.\ 1999, \apj, 510, 1064
446: \bibitem[]{schmelz} Schmelz, J.T., Saba, J.L.R., Chauvin, J.C. \&
447: Strong, K.T.\ 1997, \apj, 477, 509
448: \bibitem[]{smith2001} Smith, R.K., Brickhouse, N.S., Liedahl, D.A.
449: \& Raymond, J.C. \ 2001, \apjl, 556, L91
450: \bibitem[]{strass93} Strassmeier, K.G., Hall, D.S., Fekel, F.C. \&
451: Scheck, M.\ 1993, \aaps, 100, 173
452: \bibitem[]{testa04} Testa, P., Drake, J.J. \& Peres, G.\ 2004,
453: \apj, in press
454: \end{thebibliography}
455:
456:
457: \end{document}
458: