1: \documentclass[10pt,aas2pp4]{aastex}
2:
3: \usepackage{psfig}
4:
5: \def\etal{{et\,al.}}
6: \def\msun{M$_{\odot}$}
7: \def\mdot{$\dot M$}
8: \def\degs{\ifmmode ^{\circ}\else$^{\circ}$\fi}
9: \def\amin{\ifmmode ^{\prime}\else$^{\prime}$\fi}
10: \def\asec{\ifmmode ^{\prime\prime}\else$^{\prime\prime}$\fi}
11: \def\fss{\hbox{$.\!\!^{\rm s}$}} % Fractions of seconds
12: \def\farcs{\hbox{$.\!\!^{\prime\prime}$}} % Fractions of arcseconds
13: \def\farcm{\hbox{$.\mkern-4mu^\prime$}} % Fractions of arcminutes
14: \def\fm{\hbox{$.\!\!^{\rm m}$}} % Fractions of magnitudes
15: \def\fd{\hbox{$.\!\!^{\rm d}$}} % Fractions of days
16: % \gax is a math symbol with a '>' over '\sim' (a sqiggle), and
17: % \lax is a math symbol with a '<' over '\sim' (a sqiggle),
18: \newbox\grsign \setbox\grsign=\hbox{$>$}
19: \newdimen\grdimen \grdimen=\ht\grsign
20: \newbox\laxbox \newbox\gaxbox
21: \setbox\gaxbox=\hbox{\raise.5ex\hbox{$>$}\llap
22: {\lower.5ex\hbox{$\sim$}}}\ht1=\grdimen\dp1=0pt
23: \setbox\laxbox=\hbox{\raise.5ex\hbox{$<$}\llap
24: {\lower.5ex\hbox{$\sim$}}}\ht2=\grdimen\dp2=0pt
25: \def\gax{\mathrel{\copy\gaxbox}}
26: \def\lax{\mathrel{\copy\laxbox}}
27: \def\h{$^{\rm h}$}\def\m{$^{\rm m}$}
28: \def\s{$^{\rm s}$}
29: \def\grs{GRS 1915+105}
30: \def\gro{GRO J1655-40}
31: \def\1e17{1E 1740.7-2942}
32: \def\gr17{GRS 1758-258}
33: \def\4u{4U~1630-47}
34: \def\cyg{Cyg X-1}
35: \def\gx{GX 339-4}
36: \renewcommand{\textfraction}{0.05}
37:
38: %\received{2002 June 17}
39: \begin{document}
40:
41: % \thesaurus{06 % A&A Section 6: Form. struct. and evolut. of stars
42: % (02.01.2; % Accretion, accretion disks
43: % 08.02.1: % binaries: close
44: % 08.02.3; % binaries: general
45: % 08.09.2: % Stars: individual:
46: %% 09.18.1: % (ISM:) reflection nebulae
47: % 13.25.5)} % X-rays: stars
48:
49: \title{The 590\,days long term periodicity of the microquasar GRS 1915+105}
50:
51: \author{A. Rau\altaffilmark{1,2}, J. Greiner\altaffilmark{1,2}, M. L. McCollough\altaffilmark{3}}
52:
53: \altaffiltext{1}{Astrophysical Institute Potsdam, An der Sternwarte 16, 14482 Potsdam, Germany}
54: \altaffiltext{2}{Max-Planck-Institute for extraterrestrial Physics, Giessenbachstrasse,
55: 85748 Garching, Germany; arau@mpe.mpg.de, jcg@mpe.mpg.de}
56: \altaffiltext{3}{National Space Science and Technology Center, 320 Sparkman
57: Drive, Huntsville, AL 35805; mccolml@head-cfa.harvard.edu}
58:
59: %\date{Received ?? 2003; accepted ?? 2003}
60:
61: \begin{abstract}
62: We report on the discovery of a 590\,days long term periodicity in the
63: hard X-ray component of the microquasar \grs\ found from a
64: comprehensive study of more than four years of {\it RXTE}
65: observations. The periodicity is also observed in the hard X-ray flux
66: observed by BATSE, and in the radio flux as seen with the {\it Green
67: Bank Interferometer} and the {\it Ryle Telescope}. We discuss various
68: possible explanations, including the precession of a radiation induced
69: warped accretion disk.
70: \end{abstract}
71:
72: %
73: \keywords{Physical data and processes: accretion, accretion disk --- X-rays: binaries: --- X-rays: individual (GRS 1915+105) }
74:
75:
76: \section{Introduction}
77:
78: The galactic X-ray binary system \grs\ is the most prominent
79: microquasar showing dramatic variability in its light curve (Greiner
80: \etal\ 1996), quasi-periodic oscillations, phase lags and coherence
81: behavior (Morgan \etal\ 1997, Muno \etal\ 2001). Microquasars are
82: thought to be down-scaled analogs to quasars harboring a stellar mass
83: black hole which accretes matter from a companion star. They show
84: superluminal ejections (Mirabel \& Rodriguez 1994) making the sources
85: potential laboratories for studying accretion and relativistic jets
86: near black holes (Mirabel \etal\ 1992). \grs\ is the most energetic
87: object known in our galaxy with a luminosity of $\sim5\cdot 10^{39}$
88: erg/s in the high state and $\sim 10^{38}$ erg/s in the low state. The
89: binary system, located at a distance of $\sim$12~kpc, contains a
90: 1\,\msun\ late type giant (Greiner \etal\ 2001a) in a 33.5\,days orbit
91: around a 14\,\msun\ black hole (Greiner \etal\ 2001b), making it the
92: most massive stellar black hole known.
93:
94: Originally discovered by GRANAT (Castro-Tirado \etal\ 1992) \grs\ has
95: been extensively monitored by the {\it RXTE} satellite since 1996, and
96: a number of investigations of these data have been published
97: (e.g. Greiner \etal 1996, Morgan \etal\ 1997, Belloni \etal\ 1997 \&
98: 2000, Muno \etal\ 1999 \& 2001, Rau \& Greiner 2003; hereafter RG03).
99:
100: In a recent paper (RG03) we presented a comprehensive study of the
101: X-ray spectral behavior of \grs\ in the $\chi$-state (Belloni \etal\
102: 2000) for over nearly four years of observation with {\it
103: RXTE}. Within this study, a remarkable long term periodicity was found
104: which was mentioned briefly by Kuulkers \etal\ (1997). They found a
105: 19\,month long term variability comparing the X-ray outbursts observed
106: with BASTE and {\it RXTE} data from August 1992, March 1994, October
107: 1995 and May 1997. This is consistent with our presented result.
108:
109: A number of X-ray binaries, including LMXB as well as HMXB, exhibit
110: long term variability in their X-ray flux with periods remarkably
111: longer then their orbital period. Tananbaum \etal\ (1972) found a
112: 35\,days period in Her X-1 which they explained by the precession of a
113: warped accretion disk in the system. The same process is used to
114: describe the long term variations in SMC X-1 (Wojdowski \etal\ 1998),
115: LMC X-3 (Cowley \etal\ 1991) and LMC X-4 (Heemskerk \& van Paradijs
116: 1989).
117:
118: In this letter we discuss the detection of a new long term periodicity
119: in \grs.
120:
121: \section{Data reduction and analysis}
122:
123: The analysis presented here is based on the data reduction described
124: in full detail in RG03. We investigated public RXTE data from \grs\
125: from November 1996 to September 2000 obtained from the HEASARC. We
126: analyzed data of the PCU0 detector of the Proportional Counter Array
127: (PCA) from 3--25\,keV and data of the High Energy X-ray Timing
128: Experiment (HEXTE) cluster 0 from 20--190\,keV for 139 $\chi$-state
129: observations from 89 different days. $\chi$-states correspond to the
130: low/hard state of \grs, exhibiting relatively low flux from the
131: accretion disk and showing a hard power law tail. These states are
132: characterized by the lack of obvious variations in light curve and
133: spectrum and are connected by continuous radio emission of varying
134: strength.
135:
136: For the spectral fitting in XSPEC 11.0 (Arnaud 1996) we used a model
137: consisting of (i) photo-electric absorption (WABS), (ii) a spectrum
138: from an accretion disk consisting of multiple blackbody components
139: (DISKBB) and (iii) a power law spectrum reflected from an ionized
140: relativistic accretion disk (REFSCH; Fabian \etal\ 1989, Magdziarz \&
141: Zdziarski 1995).
142:
143: The analysis of the X-ray model parameters for all $\chi$-state
144: observations of \grs\ reveals a remarkably periodic behavior of the
145: power law slope, $\Gamma$, which dominates the hard X-ray spectrum. It
146: shows a nearly sinusoidal variation (Fig.~\ref{fitres}, see also RG03)
147: on a time scale much longer then the orbital period, $P_{orb}$. The
148: recently found correlation of the power law slope and the radio flux,
149: $F_R$, in the $\chi$-states (RG03) suggests that one should also
150: search for a long term periodicity in the radio data of the {\it Green
151: Bank Interferometer} ({\it GBI}; 2.25 and 8.3\,GHz) and the {\it Ryle
152: Telescope} ({\it RT}; 15\,GHz).
153:
154: \begin{figure}
155: \begin{center}
156: \psfig{figure=f1.eps,width=8.0cm,%
157: bbllx=1.9cm,bblly=3.5cm,bburx=19.0cm,bbury=26.1cm,clip=}\par
158: \caption[]{\begin{it}1.5--12\,keV {\it RXTE} All Sky Monitor (ASM)
159: count rate (top panel), 15\,GHz {\it RT} (2nd from top) and
160: 2.25\,GHz {\it GBI} flux (3rd) from November 1996 to
161: September 2000. ({\it GBI} was shutdown some time before
162: JD $\sim$2450400 and following JD $\sim$2451660.) The lower
163: panels show the power law slope, $\Gamma$, (4rd) and the
164: accretion disk temperature, $T_{bb}$, (5th) for all observed
165: $\chi$-states of \grs. Error bars are 1$\sigma$ for each
166: parameter of interest.\end{it}
167: \label{fitres}}
168: % plot_fitres_4letter
169: \vspace{-1.0cm}
170: \end{center}
171: \end{figure}
172:
173: For the determination of a cycle duration an analysis of variance for
174: several model parameters and X-ray and radio fluxes was performed
175: using the Fisher-Snedecor distribution function (Schwarzenberg-Czerny
176: 1989). The results for $\Gamma$, the 20--200\,keV X-ray {\it
177: CGRO}/BATSE flux and the 8.3\,GHz {\it RT} flux are given in the
178: periodogram in Fig.~\ref{tsa_alpha}. It shows a maximum at a period,
179: $P_{long}$, of 590$\pm$40(FWHM)\,days. $\Gamma$ and $F_R$ show
180: several identical local maxima around 200, 300 and 400 days which are
181: caused by the three major radio outbursts (see right panel of
182: Fig.~\ref{phase}) during a 590\,days period. This local maxima result
183: from the time separation of the outbursts at phase $\sim$0.6 and
184: $\sim$0.85 (200\,days), $\sim$0.2 and $\sim$0.6 (300\,days) and
185: $\sim$0.2 and $\sim$0.85 (400\,days). These local maxima in the
186: analysis of the variance of Gamma arise from a combination of two
187: factors. First the chi-states are usually connected with radio
188: emission and second the chi-states have non-stochastic distribution
189: within the RXTE observations. The later factor results in an uneven
190: coverage of data points within the time period.
191:
192: %\clearpage
193:
194: \begin{figure}[h]
195: \begin{center}
196: \vbox{\psfig{figure=f2.eps,width=8cm,%
197: bbllx=7.8cm,bblly=2.5cm,bburx=18.3cm,bbury=20.0cm,angle=-90,clip=}}\par
198: \caption[Varianzanalyse der Potenzgesetzkomponente]{\begin{it}Analysis
199: of variance of the 8.3\,GHz GBI radio flux (top curve), the
200: 20-200\,keV BATSE X-ray flux (middle) and the power law slope (bottom)
201: for JD 2450300-2451800. The y-axis is in units of the test statistic
202: of the Fisher-Snedecor distribution function. The test statistic has
203: larger values for the radio flux due to the higher number of {\it GBI}
204: observations compared to BATSE and {\it RXTE}. A maximum at
205: 590$\pm$40\,days is clearly seen in each of the three
206: graphs. \end{it}}
207: \label{tsa_alpha}
208: \vspace{-0.7cm}
209: \end{center}
210: % arbeit/tsa/plot_tsa
211: \end{figure}
212:
213: %\clearpage
214:
215: Fig.~\ref{phase} shows the power law slope and the {\it GBI}
216: radio flux at 2.25\,GHz displayed in segments of 590\,days each. These
217: light curves demonstrate, that 590\,days is the preferred period rather
218: than 200,300 or 400 days. Even though the coverage with {\it GBI} data
219: is shorter than with {\it RXTE} observations the periodicity is
220: clearly visible in the radio outbursts. Both the broad outburst with
221: exponential cutoff at phase $\sim$0.2 and the shorter and steeper
222: outbursts repeat periodically.
223:
224: %\clearpage
225:
226: \begin{figure}[h] \begin{center}
227: \vbox{\psfig{figure=f3.eps,width=8cm,%
228: bbllx=1.1cm,bblly=10.5cm,bburx=19.5cm,bbury=24.5cm,clip=}}\par
229: \caption[Phasengefaltete Potenzgesetzsteigung und 2.25\,GHz
230: Fl\"usse]{\begin{it}Power law slope, $\Gamma$, of all analyzed
231: $\chi$-state observations (left) and of the 2.25\,GHz radio flux from
232: {\it GBI} (right) from JD 2450350--2450940, JD 2450940--2451530 and
233: JD 2451530--2452120 (top to bottom panels). Each panel covers a
234: 590\,days period and JD=2450225 was chosen as the zero point for
235: plotting reasons only. The solid lines visualize the best fitting
236: sinusoidal function (left) and illustrate the repeating outbursts in
237: the radio flux (right). \end{it}}
238: %Die Periodizit\"at in $\Gamma$ und Radiofluss ist deutlich erkennbar. (0.346879*SIN((6.317*(:jd\_nor+0.5)-1.894)*180/3.14) + 3.09
239: \vspace{-1.0cm}
240: \label{phase}
241: \end{center}
242: \end{figure}
243:
244: %\clearpage
245:
246: The soft X-ray component did not show any periodicity in the
247: $\chi$-states. This is true for the ASM count rate and hardness ratio
248: as well as for the fit parameters of the accretion disk,
249: i.e. effective temperature and the inner disk radius.
250:
251: %\clearpage
252:
253: \begin{figure}[h]
254: \begin{center}
255: \vbox{\psfig{figure=f4.eps,width=8cm,%
256: bbllx=1.1cm,bblly=14.5cm,bburx=19.5cm,bbury=27.0cm,clip=}}\par
257: \caption[]{\begin{it} BATSE 20-200\,keV flux from JD 2448580-2452120
258: with a time binning of 57\,days. The dotted lines devide the light
259: curve into 590\,day segments. \end{it}}
260: % /home/arau/work/grs1915/work4ApJLetter/plotLc
261: \vspace{-1.0cm}
262: \label{batse}
263: \end{center}
264: \end{figure}
265:
266: %\clearpage
267:
268: The 590\, days period in the hard X-rays is also seen in the
269: 20--200\,keV BATSE earth occultation fluxes
270: (Tab.\ref{tsaresultate}). However, the long term periodicity seems to
271: be variable on larger time scale. Fig.~\ref{batse} shows the
272: 20-200\,keV light curve with a 52\,day (orbital precession period of
273: {\it CGRO}) binning. This binning ensures the removal of any
274: periodicity arising from the precession of the satellite
275: orbit. Although the 590\,day period is also visible when analyzing the
276: data from JD 2448300 to JD 2450300 (before {\it RXTE} observations
277: where performed, Tab.~\ref{tsaresultate}) the structures in the light
278: curve show remarkable differences and a somewhat shorter period of
279: 545$\pm$25~days is found when analyzing all available BATSE data from
280: JD 2448300 to JD 2451800 together. A first hint for this variation is
281: seen when fitting the power law slope behavior with a simple
282: sinusoidal function with a 590\,days period (solid line in left panel
283: of Fig.~\ref{phase}). The minimum in he power law slope of the long
284: duration $\chi$-state from JD 2450400--2450600 lies outside the
285: minimum of the sinusoidal function (Fig.~\ref{phase} top). Also, an
286: additional maximum is seen around phase=0.6, particularly in the
287: bottom panel. The inpersistent behavior is not surprising, and
288: probably reflects the known transient behavior of \grs.
289:
290: %\clearpage
291:
292: \begin{table}[h]
293: \caption[Ergebnisse der Varianzanalyse]{\begin{it} Results of the
294: analysis of variance for several model parameters and energy
295: ranges for the analysis over the time span of JD 2450300--2451800 with
296: exceptions explicitly stated otherwise. (n.d. = periodicity not detected)\end{it}}
297: \begin{center}
298: \begin{tabular}{lc}
299: \hline \hline
300: data & days \\
301: \hline \hline
302: $T_{bb}$ & n.d. \\
303: $\Gamma$ & 590$\pm$40 \\
304: $R$ & 590$\pm$40 \\
305: 2.25\,GHz GBI flux & 600$\pm$40 \\
306: 8.3\,GHz GBI flux & 590$\pm$40 \\
307: 15\,GHz RT flux & 600$\pm$40 \\
308: GBI spectral index & 590$\pm$40 \\
309: 1.5--12\,keV ASM cts/s & n.d. \\
310: 20--200\,keV BATSE flux & 600$\pm$40 \\
311: 20--200\,keV BATSE flux ({\tiny JD 2448300--2450300}) & 590$\pm$40 \\
312: 20--200\,keV BATSE flux ({\tiny JD 2448300--2451800}) & 545$\pm$25 \\
313: \hline \hline
314: \end{tabular}
315: \label{tsaresultate}
316: \end{center}
317: \end{table}
318:
319: %\clearpage
320:
321: \section{Interpretation}
322:
323: The observation of the 590\,days periodicity in \grs\ suggests a
324: precessing warped accretion disk, analogous to what is seen in other
325: X-ray binaries. However, in all previous sources it was the soft X-ray
326: flux which showed the long-term modulation. In contrast, in the case
327: of \grs, it is the spectral slope of the power law and the radio
328: flux. The radio flux is connected with the hard X-ray component and
329: probably originates near the inner part of the disk. It is unlikely
330: that the periodic behavior of $F_R$ and $\Gamma$ can be explained by a
331: geometric effect of obscuration of the inner disk by the outer
332: accretion disk.
333:
334: It has long been known (Katz 1973), that a precession of a warped disk
335: can be caused by tidal forces of the donor star. Assuming the
336: accretion disk consists of small concentric rings which are tilted
337: away from the orbital plane. Because of their fast rotation the rings
338: behave gyroscopically. Without interaction of the various individual
339: ring elements a retrograde precession around an axis perpendicular to
340: the orbital plane will be excited by the tidal torque. The precession
341: rate depends on the radius of the ring element which will lead to a tilted
342: disk. In a fluid disk, an internal torque between the ring elements
343: acts against the tilting. The resulting torque can enforce a
344: precession of the disk. But the internal torque arises only if the
345: disk is inclined with respect to the orbital plain and is connected
346: with a dissipation of energy and a backward aligned movement of the
347: disk into the orbital plane.
348:
349: For tidally forced precession, a correlation of $P_{long}/P_{orb}$ and
350: $q$ is theoretically expected but not found (Wijers \& Pringle 1999).
351: Neither the existence of a long term period nor the ratio
352: $P_{long}/P_{orb}$ are correlated with the mass ratios of the
353: components of the binary systems, ruling out tidal forces as the
354: generally dominant mechanism for the long term periodicities in these
355: systems. Thus, another mechanism is required which provides continuous
356: warping of the accretion disk. Radiation driven warping is one
357: possible mechanism (Petterson 1977, Pringle 1996, Maloney \etal\ 1996,
358: Ogilvie \& Dubus 2001). The outer, tilted part of the accretion disk
359: is irradiated by the central source. If the radiation is absorbed and
360: re-emitted parallel to the local disk gradient a torque due to gas
361: pressure affects the disk. Therefore, a small warp grows exponentially
362: if the luminosity is sufficient to depress the dissipative processes
363: forcing the disk back into the orbital plane (Petterson 1977).
364:
365: In principle, for every system a combination of viscosity and
366: accretion efficiency can be found which makes the system stable,
367: unstable or highly unstable against radiation driven warping. A deeper
368: knowledge of the internal physics of accretion disk is therefore
369: required to determine the cause of the warping. Recently, Ogilvie \&
370: Dubus (2001) found that radiation driven warping in LMXB becomes
371: relevant when the orbital period is above $\sim$1\,day, which makes it
372: possibly important for \grs.
373:
374: Whether or not a warped accretion disk is the relevant cause of the
375: long term periodicity in \grs\ remains uncertain, because the
376: 590\,days period is visible in the non-thermal part of the X-ray
377: spectrum and in the radio flux, but not in the disk
378: parameters. However, it is worth mentioning two issues: (i) a possibly
379: existing long term variability in the soft X-ray flux and the disk
380: parameters can easily be masked by the high short term variability in
381: timing and spectral behavior of \grs. (ii) The observed periodicity
382: can originate due to variability of the mass accretion rate or the
383: viscosity of the inner part of the disk. One can assume that the Roche
384: lobe overflow of the donor star deposits it's mass at different disk
385: radii, depending on the mass transfer rate, thus differing mass
386: density. The radial flow in the inner part of the accretion disk is
387: therefore not constant, leading to a periodic variation of the amount
388: of soft seed photons being comptonized in the corona. Therefore, the
389: electron temperature in the corona varies on the same time scale due
390: to Compton cooling, as the power law slope. The amount of matter to be
391: ejected in a jet changes with similar periodicity which explains the
392: overall periodicity in the radio flux.
393:
394: In the same way, periodic instabilities in the secondary affect the
395: amount of accreted matter and may therefore cause or at least
396: influence the observed periodicity. Wind-driven limit cycles (Shields
397: \etal\ 1986) as discussed for LMC X-3 (Wilms \etal\ 2001) can also
398: exist in \grs. In order to drive a wind from the outer parts of the
399: accretion disk the sound speed for the inverse Compton temperature,
400: $kT_{IC}$, has to exceed the escape velocity in the outer disk. With
401: an orbital separation of $a\sim7.5\cdot 10^{12}$\,cm the outer disk
402: radii is likely of the order of $\sim5\cdot 10^{12}$\,cm. Thus, \grs\
403: fullfils the criterium given in Equ.(3) of Wilms \etal\ (2001) for
404: $kT_{IC}>1$\,keV and may therefore be driving a substantial
405: Compton-heated wind.
406:
407: A warped accretion disk would temporarily shadow the irradiated donor
408: star. A search for the 590\,days periodicity in the IR flux of the
409: donor would also help in the understanding of the origin and mechanism
410: of the long term behavior of \grs.
411:
412: \section{Conclusion}
413:
414: We present the discovery of a 590\,days long term periodicity in the
415: power law slope of the hard X-ray spectrum, the hard X-ray flux and
416: the radio flux (at several frequencies) of the microquasar \grs. No
417: such periodicity is seen in the soft X-ray component. We discussed the
418: observed behavior in the context of a warped accretion disk and of
419: wind-driven limit cycles. We have found, that a periodic mass density
420: variation in the disk, produced by the warp and/or by periodic
421: instabilities in the donor, can describe the observations, by
422: producing a varying amount of soft seed photons for the Comptonization
423: process and for the relativistic mass ejecta, very well. A wind-driven
424: limit cycles is also a possible origin of the 590\,days periodicity.
425:
426: \begin{acknowledgements}
427: The authors thank E. H. Morgan (MIT) and G. G. Pooley (MRAO) for
428: providing the ASM and {\it RT} data and W. A. Heindl ({\it RXTE}) and
429: K. A. Arnaud (XSPEC) for their comments. We thank R. Schwarz (AIP) for
430: the kind help with the analysis of variance. {\it The Ryle Telescope}
431: is supported by PPARC. The {\it Green Bank Interferometer} is a
432: facility of the National Science Foundation operated by the NRAO in
433: support of NASA High Energy Astrophysics programs.
434: \end{acknowledgements}
435:
436: \begin{thebibliography}{}
437:
438: \bibitem[Arnaud 1996]{437} Arnaud K. A., 1996, in Astronomical Data
439: Analysis Software and Systems, eds. Jacoby G. H., Barnes J., ASP
440: Conf. Series 101, 17
441: \bibitem[Belloni \etal\ 1997]{440} Belloni T., Mendez M., King A. R., van der Klis M., van Paradijs J., 1997, ApJ 488, L109
442: \bibitem[]{bkm00} Belloni T., Klein-Wolt M., Mendez M., van der Klis
443: M., van Paradijs J., 2000, A\&A 355, 271
444: \bibitem[]{cbl92} Castro-Tirado A.J., Brandt S., Lund N., 1992, IAU
445: Circ., 5590
446: \bibitem[Cowley \etal\ 1991]{446} Cowley A. P., Schmidtke P. C., Ebisawa K., Makino F., Remillard R. A., Crampton D., Hutchings J. R., Kitamoto S., Treves A., 1991, ApJ 381, 526
447: \bibitem[Fabian \etal\ 1989]{447} Fabian A. C., Rees M. J., Stella L., White N. E., 1989, MN 238, 729
448: \bibitem[Greiner \etal\ 1996]{gmr96} Greiner J., Morgan E. H., Remillard R. A., 1996, ApJ 473, L107
449: \bibitem[Greiner \etal\ 2001a]{gcm01a} Greiner J., Cuby J. G., McCaughrean M. J., Castro-Tirado A. J., Mennickent R. E., 2001a, A\&A 373, 37
450: \bibitem[Greiner \etal\ 2001b]{452} Greiner J., Cuby J. G., McCaughrean M. J., 2001b, Nature 414, 522
451: \bibitem[Heemskerk \& van Paradijs 1989]{453} Heemskerk M. H. M., van Paradijs J., 1989, A\&A 223, 154
452: \bibitem[Katz 1973]{k73} Katz J. I., 1973, Nature Phys. Sci., 246, 87
453: \bibitem[Kuulkers \etal\ 1997]{kpk97} Kuulkers E., Parmar A. N., Kitamoto S., Cominsky L. R., Sood R. K., 1997, MN 291, 81
454: \bibitem[Magdziarz \& Zdziarski 1995]{456} Magdziarz P., Zdziarski A. A., 1995, MN 273, 837
455: \bibitem[Maloney \etal\ 1996]{457} Maloney P. R., Begelman M. C., Pringle
456: J. E., 1996, ApJ 472, 582
457: \bibitem[Mirabel \etal\ 1992]{459} Mirabel I. F., Rodriguez L. F., Cordier B., Paul J., Lebrun F., 1992, Nature 358, 215
458: \bibitem[Mirabel \& Rodriguez 1994]{460} Mirabel I. F., Rodriguez L. F., 1994, Nature 371, 46
459: \bibitem[Morgan \etal\ 1997]{463} Morgan E. H., Remillard R. A., Greiner J., 1997, ApJ 482, 993
460: \bibitem[Morrison \& McGammon 1983]{464} Morrison R., McCammon D., 1983, ApJ 270, 119
461: \bibitem[Muno \etal\ 1999]{mmr99} Muno M. P., Morgan E. H., Remillard R. A., 1999, ApJ 527, 321
462: \bibitem[Muno \etal\ 2001]{mrm01} Muno M. P., Remillard R. A., Morgan E. H., Waltman E. B., Dhawan V., Hjellming R. M., Pooley G., 2001, ApJ 556, 515
463: \bibitem[Ogilvie \& Dubus 2001]{od01} Ogilvie G. I., Dubus G., 2001, MN 320, 485
464: \bibitem[Petterson 1977]{pe77} Petterson J.A., 1977, ApJ 216, 827
465: \bibitem[Pringle 1996]{469} Pringle J. E., 1996, MN 281, 357
466: \bibitem[Rau \& Greiner 2003]{RG03} Rau A., Greiner J., 2003, A\&A
467: 397, 711
468: \bibitem[Schwarzenberg-Czerny 1989]{472} Schwarzenberg-Czerny A., 1989, MN 241, 153
469: \bibitem[Shields \etal\ 1986] {} Shields G. A., McKee C. F., Begelman
470: M. C., 1986, 90
471: \bibitem[Tananbaum \etal\ 1972]{475} Tananbaum H, Gursky H., Kellogg E., Levinson R., Schreier E., Giaconni R., 1972, ApJ 174, L143
472: \bibitem[Wijers \& Pringle 1999]{476} Wijers R. A. M. J., Pringle J. E., 1999, MN 308, 207
473: \bibitem[Wojdowski \etal\ 1998]{477} Wojdowski P. S., Clark G. W., Levin A. M., Woo J. W., Zhang S. N., 1998, ApJ 502, 253
474: \bibitem[Wilms \etal\ 2001]{478} Wilms J., Nowak M. A., Pottschmidt K.,
475: Heindl W. A., Dove J. B., Begelman M. C., 2001, MN 320, 327
476: \end{thebibliography}
477:
478:
479: \end{document}
480:
481:
482:
483:
484:
485:
486:
487:
488:
489: