1: \documentclass[preprint,12pt]{aastex}

2: \usepackage{epsfig}

3: \usepackage{emulateapj5}

4: \usepackage{apjfonts}

5: 

6: \newcommand{\chandra}{{\it Chandra}}

7: \newcommand{\asca}{{\it ASCA}}

8: \newcommand{\rosat}{{\it ROSAT}}

9: \newcommand{\sax}{{\it BeppoSAX}}

10: \newcommand{\xmm}{{\it XMM-Newton}}

11: \newcommand{\einstein}{{\it Einstein}}

12: \newcommand{\lum}{\thinspace\hbox{$\hbox{erg}\thinspace\hbox{s}^{-1}$}}

13: \newcommand{\flux}{\thinspace\hbox{$\hbox{erg}\thinspace\hbox{cm}^{-2}\thinspace\hbox{s}^{-1}$}}

14: \newcommand{\sss}{XMMU\,J005510.7-373855}

15: 

16: \begin{document}

17: 

18: \def\spose#1{\hbox to 0pt{#1\hss}}

19: \def\laeq{\mathrel{\spose{\lower 3pt\hbox{$\mathchar"218$}}

20:      \raise 2.0pt\hbox{$\mathchar"13C$}}}

21: \def\gaeq{\mathrel{\spose{\lower 3pt\hbox{$\mathchar"218$}}

22:      \raise 2.0pt\hbox{$\mathchar"13E$}}}

23: 

24: 

25: \title{A Luminous Recurrent Supersoft X-ray Source in NGC 300}

26: 

27: \author{A.K.H.~Kong\altaffilmark{1} and R.~Di\,Stefano\altaffilmark{1,2}}

28: \altaffiltext{1}{Harvard-Smithsonian Center for Astrophysics, 60

29: Garden Street, Cambridge, MA 02138; akong@cfa.harvard.edu}

30: \altaffiltext{2}{Department of Physics and Astronomy, Tufts

31: University, Medford, MA 02155}

32: 

33: \begin{abstract}

34: We report the results of \xmm\ observations for an especially luminous

35: supersoft X-ray source (SSS) with bolometric luminosity of $10^{39}$

36: \lum\ in

37: the spiral galaxy NGC 300. The source was detected as a SSS in 1992 and

38: disappeared in subsequent X-ray observations. 

39: The source was active again during recent \xmm\ observations. 

40: It appeared to be very soft ($kT\sim 60$ eV) and very luminous ($\sim

41: 10^{38}-10^{39}$ \lum). The two \xmm\

42: observations also reveal that the source went from a ``high'' state to

43: a ``low'' state in 6 days. We also found a 5.4-hr periodicity during

44: the ``low'' state. We consider white dwarf, black hole,

45: and neutron star models to explain the nature of the source.

46: 

47: \end{abstract}

48: 

49: \keywords{galaxies: individual (NGC 300)  --- X-rays:

50: binaries --- X-rays: galaxy}

51: 

52: \section{Introduction}

53: 

54: Supersoft X-ray sources (SSSs) form a distinct class of objects, first

55: established through 

56: \rosat\ observations. The hallmarks of SSSs are very soft X-ray emission

57: ($kT$ typically $<100$ eV) and bolometric luminosities of $10^{36}-10^{38}$

58: \lum. The advent of \chandra\ and \xmm\ provides good opportunities to

59: detect and study 

60: SSSs in nearby galaxies. Luminous ($10^{38-40}$\lum) SSSs have been

61: found in several nearby galaxies including M31 (Kong et al. 2002;

62: Di\,Stefano et al. 2002a), 

63: M81 (Swartz et al. 2002), M101 (Pence et al. 2001; Mukai et al. 2003;

64: Di\,Stefano \& Kong 2003 [DK03]),

65: NGC4697 (Sarazin, Irwin \& Bregman 2001; DK03),

66: M51 (DK03), and M83 (Soria \& Wu 2003; DK03). 

67: 

68: NGC 300 is an SA(s)d galaxy, seen near face-on (inclination angle

69: 46$^{\circ}$; Tully 1988) at a distance of $2.0\pm0.1$ Mpc (Freedman

70: et al. 2001). The galaxy has been observed by

71: \rosat\ (Read et al. 1997; Read \& Pietsch 2001) and \xmm\ (Soria \&

72: Kong 2003). 

73: 

74: In this Letter, we report the reappearance of a luminous SSS in NGC

75: 300 as observed with \xmm.

76: 

77: \section{Observations and Data Reduction}

78: 

79: \subsection{\rosat}

80: 

81: NGC 300 was observed by \rosat\ five times from 1991 to 1997. A

82: detailed analysis of the \rosat\ data was done by Read \& Pietsch

83: (2001). Briefly, the datasets consist of 2 Position Sensitive

84: Proportional Counter (PSPC) and 3 High Resolution Imager (HRI)

85: pointings. The exposures range from $\sim 9$ ks to $\sim 37$ ks. The

86: luminous supersoft source was only detected in a 37 ks PSPC observation

87: taken on 1992 May and June (see Read \& Pietsch 2001). Spectral

88: analysis was also done by Read et al. (1997). We have

89: reanalyzed the PSPC spectrum taken on 1992 May/June, and used other data to set upper

90: limits on the luminosity for the long-term lightcurve.

91: 

92: We extracted the source spectrum from a $30''$ circular region, while

93: background was from an annulus region ($45''$ and $60''$ radii)

94: centered on the source. 

95: The spectrum was grouped into at least 20 counts per spectral bin

96: to allow $\chi^2$ statistics to be used.

97: 

98: \subsection{\xmm}

99: 

100: The \xmm\ instrument modes were full-frame, medium filter for the three European Photon

101: Imaging Cameras (EPIC).

102: The first observation was taken on 2000 December 26 for 37 ks and the

103: second observation was on 2001 January 1 for about 47 ks. 

104: After rejecting intervals with a high

105: background level, we considered a good time interval of $\sim 28$ ks

106: and $\sim 40$ ks for the first and second observation, respectively. 

107: Data were reduced and analyzed

108: with the \xmm\ SAS package v5.3.3.

109: %\footnote{http://xmm.vilspa.esa.es/external/xmm\_sw\_cal/sas\_frame.shtml}.

110: 

111: We used here the MOS images to determine the position of the source

112: because the spatial resolution of MOS detector is slightly better than

113: that of pn. The source is located at 

114: R.A.=00h55m10s.7, Dec.=$-37^{\circ}38'55''$ (J2000), $\sim 4'$ (2.4 kpc) from the

115: galactic center; the derived positions from the two \xmm\ observations agree

116: with each other. This position is about $5''$ off from previous \rosat\ PSPC

117: observation, and is consistent with the positional error ($7.3''$)

118: quoted by Read \& Pietsch (2001). 

119: 

120: Source spectra and lightcurves of \sss\ were extracted with the SAS task

121: {\sc xmmselect}. Source-free regions were used for background to avoid

122: the chip boundary and a nearby faint source.

123: In order to allow $\chi^2$ statistics to be used, all

124: the spectra were binned such that there are at least 20 counts per

125: spectral bin. Response matrices were created by {\sc rmfgen} and {\sc arfgen}. 

126: 

127: \section{Analyses and Results}

128: 

129: %\subsection{Spectral Analysis}

130: 

131: Spectral analysis was performed by making use of XSPEC v11.2.

132: %\footnote{http://heasarc.gsfc.nasa.gov/docs/xanadu/xspec/index.html}.

133: Table 1 shows the best-fitting spectral parameters for the three

134: \rosat\ and \xmm\ observations.

135: 

136: For the \rosat\ observation, the spectrum (see Figure 1) can be fit with a blackbody

137: model with $N_H=1.1\times10^{21}$ cm$^{-2}$ and $kT=48.7$ eV; the

138: 0.2--2 keV luminosity is $10^{39}$ \lum. We note that Read et

139: al. (1997) fit the spectrum with a thermal bremsstrahlung model

140: with temperature of a 0.1 keV and we confirmed that it is also an

141: acceptable model.  Using the thermal bremsstrahlung model, the 

142: luminosity becomes $3.8\times10^{38}$ \lum\ [the luminosity quoted by

143: Read et al. (1997) is $1.6\times10^{37}$ \lum, which is an absorbed

144: luminosity corrected for

145: absorption in our own Galaxy], a factor of 2.5 lower than 

146: the blackbody model. 

147: While we cannot distinguish between the two models

148: statistically, we prefer the blackbody model as subsequent \xmm\

149: observations confirm the supersoft nature of the source (see

150: below). Also, the \xmm\ observations would have detected the high

151: energy photons associated with the thermal bremsstrahlung model, had

152: the model been correct. Finally, the \xmm\ observations provide more

153: photons than \rosat.

154: 

155: We fit the pn, MOS1 and MOS2 data simultaneously with

156: several single-component models with interstellar absorption

157: (including absorbed power-law, thermal bremsstrahlung, blackbody and

158: Raymond-Smith); only the blackbody model provides an acceptable

159: fit. The blackbody temperature of both observations ranges between 57

160: eV and 67 eV, while the $N_H$ varies from $1.5\times10^{21}$ cm$^{-2}$

161: in the first observation to $5.3\times10^{20}$ cm$^{-2}$ in the second

162: observation. The 0.2--2 keV luminosity also drops from

163: $9\times10^{38}$ \lum\ to $1.2\times10^{38}$ \lum, indicating that the

164: source is a variable on timescale of days. The spectra of the

165: two observations are shown in Figure 1.

166: 

167: %\subsection{Temporal Analysis}

168: 

169: %\subsubsection{Long-term Variability}

170: The combined \rosat\ and \xmm\

171: long-term lightcurve of the SSS is shown in Figure 2, which is constructed

172: from a series of \rosat\ and \xmm\ pointings. The source was below the

173: detection limit of

174: other \rosat\ observations. In these cases, we estimated the $3\sigma$

175: limits, assuming a blackbody model

176: with mean $N_H$ ($10^{21}$ cm$^{-2}$) and $kT$ (58 eV). The value of 

177: $N_H$ plays an important role in estimating the luminosity. For instance,

178: if we lower the $N_H$ to $5.3\times10^{20}$ cm$^{-2}$, as found in the

179: second \xmm\

180: observation, the upper limits decrease by a

181: factor of $\sim 3$. In other words, the source varies by as much as

182: factor of 30 between the ``low'' state and the ``high'' state spanning

183: 8.5 years.

184: 

185: %\subsubsection{Short-term Variability}

186: 

187: To search for short-term variability, we computed the Lomb-Scargle

188: periodogram (LSP; Lomb 1976; Scargle 1982), a modification of the

189: discrete Fourier transform which is generalized to the case of uneven

190: spacing. In each observation, we extracted the combined background

191: subtracted lightcurve

192: from pn and MOSs to increase the signal-to-noise. Individual

193: lightcurves from each of

194: the three detectors were also used to verify the result. By

195: applying the LSP to the combined lightcurve, we found that there is a

196: sharp peak at 5.4 hr in

197: the ``low'' state observation (see Figure 3). Independent checks from the pn and

198: MOS data also confirmed the periodicity. We determined the 99.9\% confidence

199: level by generating Gausiaan noise datasets with the same time

200: intervals and variance as the true data, and then performed the LSP

201: analysis on the resulting datasets (see Kong, Charles \& Kuulkers

202: 1998). The peak power in each periodogram

203: (which must be purely due to noise) was then recorded. This was

204: repeated 10000 times to obtain good statistics. The peak at 5.4 hr is well

205: above the 99.9\% confidence level. The folded lightcurve of the ``low'' state data in 5.4 hr

206: is also shown in Figure 3.

207: 

208: Similar analysis was also carried out in the ``high'' state data, but

209: there is no significant peak in the LSP. The folded lightcurve shows

210: no obvious periodic variability at 5.4 hr (see Figure 3). It is not

211: clear if it is due to geometric effect but the exposure time of the ``high'' state 

212: covers less than 8 hr, corresponding to 1.4 cycles of the 5.4-hr

213: period in contrast to the 2.1 cycles in the ``low'' state. We

214: therefore performed 

215: Kolmogorov-Smirnov test to test

216: for variability and found significant variability at the 99.9\%

217: levels. We also search for variability of the hardness ratio but no

218: significant change was found in both observations.

219: 

220: %\subsection{Optical}

221: 

222: We also examined an Digitized Sky Survey image of NGC 300 to search for

223: possible optical counterpart. Within a $4''$ error circle of the \xmm\

224: position, there is no obvious optical counterpart with limiting $B$ and $R$

225: magnitudes of 21 and 18.1, respectively. However, the source is

226: surrounded by H{\sc ii} regions (Deharveng et al. 1988; Blair \& Long 1997), OB associations

227: (Pietrzynski et al. 2001) and supernova remnants (Blair \& Long

228: 1997). The nearest H{\sc ii} region (source 136; Deharveng et al. 1988)

229: is located about $20''$ ($\sim 200$ pc) south-east of the X-ray

230: source. Within $50''$ ($\sim 500$ pc) of the SSS, we found an OB

231: association (AS99;

232: Pietrzynski et al. 2001), two supernova remnant candidates (N300-S21,

233: N300-N23; Blair \& Long 1997) and a H{\sc ii} region (N300-H20; Blair \&

234: Long 1997). 

235: We also looked for UV counterpart

236: from the Optical Monitor (OM) image (see Soria \& Kong 2003 for details of the OM

237: observations); no 

238: counterpart is found at the position of the SSS in both observations.

239: 

240: \section{Discussion}

241: 

242: It is clear from the spectral fits of all \rosat\ and \xmm\

243: observations that the \sss\ is very soft ($kT \sim 48-67$ eV) and is

244:  highly variable. Such a soft spectrum is consistent with SSSs seen in

245: our own Galaxy, the Magellanic Clouds, M31 (Greiner 2000 \footnote{see

246: http://www.aip.de/$\sim$jcg/sss/ssscat.html for updated catalog}) and

247: several nearby galaxies (see \S\,1). Since the Galactic and Magellanic

248: Cloud SSSs are conjectured

249: to be white dwarfs (WDs), it is reasonable to consider    

250: a WD model for the source in NGC 300. 

251: Since, however, the luminosity in the high

252: state may be too large to be consistent with the standard SSS WD model,

253: we also consider black hole (BH) and neutron star (NS) models. 

254: 

255: %\subsection{WD Models}

256: 

257: %In this model, the emission comes from a WD that burns accreting

258: The X-ray emission of the SSS could come from a WD that burns accreting

259: hydrogen in a quasi-steady manner. The maximum luminosity is the

260: Eddington luminosity for a $1.4M_{\odot}$ object. WD models therefore

261: seem to be ruled out by the high bolometric luminosities during 2 of

262: the observations. It is nevertheless worth noting that the WD model

263: temperatures are comparable to those we observed, and that

264: X-ray variability by a factor

265: of a few over times of days and months has been observed in SSSs.

266: (see e.g., Greiner and Di Stefano 2002 and references therein.)  

267: If the 5.4-hour variation detected during the X-ray low state corresponds 

268: to an orbital period, then this system may be similar

269: to 1E0035.4-7230, a Magellanic Cloud SSS that appears to have a

270:  $4$ hour period (Kahabka 1996). In such a short-period binary, it is likely that 

271: a radiation-driven wind plays an important role in the

272: binary evolution. (see, e.g., van Teeseling \& King 1998)

273: A WD model would perhaps be the most natural explanation for the

274: SSS in NGC 300, were it not for the high estimated bolometric

275: luminosities. 

276: Note that it may be difficult to use beaming to circumvent the

277: Eddington limit in these systems, since the inner region of the accretion

278: disk can also become a luminous emitter of SSS radiation 

279: (Popham \& Di\thinspace Stefano 1996).  

280: 

281: %\subsection{Black Hole Models}

282: 

283: SSS emission is expected from accreting BHs.

284: Modeling the accretion disk as a thin disk which is optically thick,

285: we can derive a relationship between the minimum mass of the accretor,

286: and the observed temperature and luminosity (Frank, King, \& Raine 2002). Using $kT=60$ eV and

287: $L_X=10^{38}$\lum\ ($10^{39}$\lum), we find that the accretor mass is

288: greater than $\approx 890 M_{\odot}$ ($2800 M_{\odot}$). Although the

289: fact that the observed luminosity is $10^2-10^3$ times smaller than

290: the Eddington luminosity may call the validity of the disk model into

291: question, this calculation suggests that the accretor is an

292: intermediate-mass BH. If the accretor is a BH, we may use $L_{obs}=0.1

293: \dot m c^2 (\eta/0.1)$, where $\dot m$ is the accretion rate and

294: $\eta$, the efficiency factor, is likely to be close to 0.1. This

295: yields $\dot m\approx 1.8\times10^{-8} M_{\odot}$/yr

296: ($1.8\times10^{-7} M_{\odot}$/yr) for $L_{obs}=10^{38}$\lum\

297: ($L_{obs}=10^{39}$\lum). This would be consistent with an

298: irradiation-driven wind from a low-mass donor. Measurement of the

299: orbital period could provide supporting evidence. Indeed, if the donor

300: fills its Roche lobe, then $P_{orb}\approx(8.9~\mathrm{hrs})(M_d/M_{\odot})$,

301: where $M_d$ is the mass of the donor star. If $P_{orb}=5.4$ hrs, then

302: $M_d\approx0.61 M_{\odot}$.

303: 

304: %\subsection{Neutron Star Models}

305: 

306: It is of course possible that the observed emission emanates from an

307: accreting NS. In this case the SSS emission would presumably emanate

308: from a photosphere which is much larger than the neutron star itself.

309: The photospheric radius would be different in each observation:

310: $3.7 \times 10^9$ cm during the {\it ROSAT} observation,

311: $2.6 \times 10^9$ cm during the \xmm\  high-state observation, 

312: and $6.8 \times 10^8$ cm during the \xmm\ low-state observation,

313: corresponding to the different temperatures and luminosities.

314: 

315: At this point, neutron star models seem conceptually unattractive

316: because there is no obvious explanation for why the photosphere should

317: achieve these relatively large sizes, and no way to

318: relate them to the system's fundamental physical parameters.

319: In addition, just as in the WD models, the luminosity

320: appears to be super-Eddington during two of the observations, 

321: unless the neutron star has a mass as large as $3-4\, M_\odot.$

322: In this case, if the efficiency of turning accretion energy into X-rays is the 

323: same during all $3$ observations,  

324: then the accretion rate must have changed by an order of magnitude,

325: approaching $2\times 10^{-8} M_\odot$ yr$^{-1}$ during the high state.

326: (see Di\thinspace Stefano et al. 2002b for a similar model of the

327: X-ray source in the globular cluster Bo 375.)  

328: If, on the other hand, the mass of the neutron star is close to $1.4\, M_\odot,$

329: then other ways around the Eddington limit must be found; e.g.,

330: the energy could be beamed into a smaller solid

331: angle during the $2$ high-state observations.

332: In this case, the accretion rate could be significantly lower.  

333: 

334: \section{Summary}

335: 

336: We have studied a recurrent luminous SSS in NGC300 with \xmm. The

337: source was seen by \rosat\ in 1992 and fell below the detection limit

338: in subsequent \rosat\ observations. It reappeared in recent \xmm\

339: observations with bolometric luminosities between $10^{38}$ and

340: $10^{39}$ \lum. During the ``low'' state, the source showed a 5.4-hr

341: periodicity. If we consider the periodicity is due to the orbital

342: period, then the X-ray emission can be explained by WD, BH, and NS

343: models. However, WD and NS models appear to be unlikely due to the high X-ray

344: luminosity during the ``high'' state. Further repeated X-ray

345: observations of the SSS in different states may permit discrimination among these

346: models.

347: 

348: \begin{acknowledgements}

349: This work was supported by NASA under an LTSA grant,

350: NAG5-10705. A.K.H.K. acknowledges support from the Croucher Foundation.

351: %This work is based on observations obtained with \xmm, an ESA mission

352: %with

353: %instruments and contributions directly funded by ESA member states and

354: %the US (NASA). 

355: \end{acknowledgements}

356: 

357: \begin{references}

358: 

359: \reference{} Blair, W.P., \& Long, K.S. 1997, ApJS, 108, 261

360: 

361: \reference{} Deharveng, L., Caplan, J., Lequeux, J., et al. 1988, A\&AS, 73, 407

362: 

363: \reference{} Dickey, J.M., \& Lockman, F.J. 1990, ARA\&A, 28, 215

364: 

365: \reference{} Di\,Stefano, R., \& Kong, A.K.H. 2003, ApJ, in press

366: (astro-ph/0301162), DK03

367: 

368: \reference{} Di\,Stefano, R., Greiner, J., Kong, A., Garcia, M.R.,

369: Primini, F.A., Barmby, P., Murray, S.S., Curry, S. 2002a, American

370: Physical Society, APRN17072

371: 

372: \reference{} Di\,Stefano, R., Kong, A.K.H., Garcia, M.R., Barmby, P.,

373: Greiner, J., Murray, S.S., \& Primini, F.A. 2002b, ApJ, 570, 618

374: 

375: \reference{} Frank, J., King, A., \& Raine, D.J. 2002, Accretion Power

376: in Astrophysics: Third Edition, CUP

377: 

378: \reference{} Freedman, W.L., et al. 2001, ApJ, 553, 47

379: 

380: \reference{} Greiner, J. 2000, NewA, 5, 137

381: 

382: \reference{} Greiner, J., \& Di\,Stefano, R. 2002, ApJ, 578, L59

383: 

384: \reference{} Kahabka, P. 1996, A\&A, 306, 795

385: 

386: \reference{} Kong, A.K.H., Charles, P.A., \& Kuulkers, E. 1998, NewA, 3, 301

387: 

388: \reference{} Kong, A.K.H., Garcia, M.R., Primini, F.A., Murray, S.S.,

389: Di\,Stefano, R., \& McClintock, J. 2002, ApJ, 577, 738

390: 

391: 

392: \reference{} Lomb, N.R. 1976, Ap\&SS, 39, 447

393: 

394: 

395: \reference{} Mukai, K., Pence, W.D., Snowden, S.L., Kuntz, K.D. 2003,

396: ApJ, 582, 184

397: 

398: \reference{} Pence, W.D., Snowden, S.L., Mukai, K., \& Kuntz,

399: K.D. 2001, ApJ, 561, 189

400: 

401: \reference{} Pietrzynski, G., Gieren, W., Fouque, P., \& Pont, F 2001,

402: A\&A, 371, 497

403: 

404: \reference{} Popham, R., \& Di\,Stefano, R. 1996, in Supersoft X-Ray

405: Sources, Proceedings of the International Workshop Held in Garching,

406: Germany, 28 February - March 1996. Lecture Notes in Physics, Vol. 472,

407: edited by Jochen Greiner. Springer-Verlag, Berlin Heidelberg New York, p.65

408: 

409: 

410: 

411: \reference{} Read, A.M., Ponman, T.J., \& Strickland, D.K. 1997,

412: MNRAS, 286, 626

413: 

414: \reference{} Read, A.M., \& Pietsch, W. 2001, A\&A, 373, 473

415: 

416: \reference{} Sarazin, C.L., Irwin, J.A., \& Bregman, J.N. 2001, ApJ,

417: 556, 533

418: 

419: \reference{} Scargle, D. 1982, ApJ, 263, 835

420: 

421: 

422: \reference{} Soria, R., \& Kong, A.K.H. 2003, A\&A, submitted

423: 

424: \reference{} Soria, R., \& Wu, K. 2003, A\&A, submitted

425: 

426: \reference{} Strickland, D.K., Colbert, E.J.M., Heckman, T.M., Weaver,

427: K.A., Dahlem, M., \& Stevens, I.R. 2001, ApJ, 560, 707

428: 

429: \reference{} Swartz, D.A., Ghosh, K.K., Suleimanov, V., Tennant, A.F.,

430:  Wu, K. 2002, ApJ, 574, 382

431: 

432: \reference{} Tully, R. 1988, Nearby Galaxies Catalog (Cambridge:

433: Cambridge Univ. Press)

434: 

435: \reference{} van Teeseling, A., \& King, A.R. 1998, A\&A, 338, 957

436: 

437: \end{references}

438: 

439: \newpage

440: 

441: \begin{deluxetable}{cccccc}

442: \tabletypesize{\small}

443: \tablecaption{Best-fitting Spectral Parameters}

444: \tablewidth{0pt}

445: \tablehead{

446: \colhead{Date} & \colhead{$N_H$} & \colhead{$kT$} &

447: \colhead{$L_X$}\tablenotemark{a} & \colhead{$L_{bol}$}\tablenotemark{b}

448: & \colhead{$\chi^2_{\nu}$/d.o.f}\\

449:  &  ($\times10^{21}$ cm$^{-2})$ &(eV) & & &\\

450: }

451: \startdata

452: 1992 May 26  & $1.08^{+0.31}_{-0.31}$ & $48.7^{+9.1}_{-9.1}$

453: & 10 & 20& 0.74/11\\

454: 2000 Dec 26 & $1.50^{+0.32}_{-0.20}$ &

455: $56.8^{+2.2}_{-3.3}$ & 9 & 15 & 1.07/39\\

456: 2001 Jan 1 & $0.53^{+0.20}_{-0.07}$ &

457: $66.5^{+2.4}_{-3.6}$ & 1.1 & 1.9 & 1.15/48\\ 

458: \enddata

459: \tablecomments{All quoted uncertainties are 1$\sigma$.}

460: \tablenotetext{a}{0.2--2 keV luminosity ($\times 10^{38}$\lum),

461: assuming a distance of 2 Mpc}

462: \tablenotetext{b}{Bolometric luminosity ($\times 10^{38}$\lum)}

463: 

464: \end{deluxetable}

465: 

466: 

467: \newpage

468: 

469: \begin{figure}

470: \begin{center}

471: {\rotatebox{-90}{\psfig{file=f1a.ps,width=2.5in}}}

472: {\rotatebox{-90}{\psfig{file=f1b.ps,width=2.5in}}}

473: {\rotatebox{-90}{\psfig{file=f1c.ps,width=2.5in}}}

474: \end{center}

475: \caption{Energy spectra of \sss\ (see Table 1 for spectral parameters).

476: From left to right: \rosat\ PSPC spectrum on 1992 May/June (high state), \xmm\ spectrum on

477: 2000 December 26 (high state) and \xmm\ spectrum on 2001 January 1

478: (low state). For the \xmm\ spectra, pn, MOS1, and MOS2 data are marked

479: as triangles, solid squares and circles.}

480: \end{figure}

481: 

482: \begin{figure}

483: \begin{center}

484: \psfig{file=f2.ps,width=4in}

485: \end{center}

486: \caption{Long-term lightcurve of the supersoft ULX from 1991 to 2001

487: (circles: \rosat\ PSPC; squares: \rosat\ HRI; triangles: \xmm). The 0.2--2 keV

488: luminosity is from spectral fits in Table 1, while the $3\sigma$ upper

489: limits are assuming a blackbody model with $N_H=10^{21}$ cm$^{-2}$ and

490: $kT=58 eV$.}

491: \end{figure}

492: 

493: \begin{figure}

494: \begin{center}

495: \psfig{file=f3.ps,width=7in}

496: \end{center}

497: \caption{a) Lomb-Scargle periodogram of \sss\ as obtained by \xmm\ on

498: 2001 January. The horizontal dotted line is the 99.9\% confidence

499: level; b) Folded lightcurve of the ``low'' state (2001 January) data

500: on a period of 5.4 hr; c) Folded lightcurve of the ``high'' state (2000 December) data

501: on a period of 5.4 hr. $T_0$ of both lightcurves are set at the time of the first

502: data point.}

503: \end{figure}

504: 

505: 

506: 

507: \end{document}

508: 

509: