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

2: %\usepackage{graphicx}

3: 

4: \def\simlt{\mathrel{\hbox{\rlap{\hbox{\lower4pt\hbox{$\sim$}}}\hbox{$<$}}}}

5: \def\simgt{\mathrel{\hbox{\rlap{\hbox{\lower4pt\hbox{$\sim$}}}\hbox{$>$}}}}

6: 

7: \def\nodata{---}

8: 

9: \def\lesssim{\mathbin{\lower 3pt\hbox 

10:       {$\rlap{\raise 5pt\hbox{$\char'074$}}\mathchar"7218$}}} 

11: \def\gtrsim{\mathbin{\lower 3pt\hbox

12:       {$\rlap{\raise 5pt\hbox{$\char'076$}}\mathchar"7218$}}}  

13: 

14: 

15: \newcommand{\ns}{neutron star}

16: \newcommand{\rxte}{\textit{RXTE}}

17: \newcommand{\exos}{\textit{EXOSAT}}

18: \newcommand{\chandra}{\textit{Chandra}}

19: \newcommand{\xmm}{\textit{XMM}-Newton}

20: 

21: \newcommand{\erg}{erg cm$^{-2}$ s$^{-1}$}

22: \newcommand{\Te}{\mbox{$T_{\rm eff}$}}

23: \newcommand{\Tc}{\mbox{$T_{\rm c}$}}

24: \newcommand{\fedd}{\mbox{$F_{\rm Edd}$}}

25: \newcommand{\ledd}{\mbox{$L_{\rm Edd}$}}

26: \newcommand{\msol}{\mbox{$M_{\odot}$}}

27: 

28: \begin{document}

29: 

30: \title{The Magnetar Nature and the Outburst Mechanism of a Transient

31: Anomalous X-ray Pulsar}

32: 

33: \author{Tolga G\"uver$^{1}$, Feryal \"Ozel$^{2}$, Ersin

34: G\"o\u{g}\"u\c{s}$^{3}$ and Chryssa Kouveliotou$^{4}$}

35: 

36: \affil{$^{1}$Istanbul University, Science Faculty, Astronomy \& Space

37: Sciences Department, Beyaz\i t, Istanbul, 34119}

38: \affil{$^{2}$University of Arizona, Department of Physics, 1118 E. 4th

39: St., Tucson, AZ 85721}

40: \affil{$^{3}$Sabanc\i~University, Faculty of Engineering Natural Sciences, 

41: 34956 Istanbul, Turkey}

42: \affil{$^{4}$NASA/MSFC, VP 62, 320 Sparkman Drive Huntsville, AL 35805, USA }

43: 

44: \begin{abstract}

45: Anomalous X-ray Pulsars (AXPs) belong to a class of neutron stars

46: believed to harbor the strongest magnetic fields in the universe, as

47: indicated by their energetic bursts and their rapid

48: spindowns. However, an unambiguous measurement of their surface field

49: strengths has not been made to date. It is also not known whether AXP

50: outbursts result from changes in the neutron star magnetic field or

51: crust properties.  

52: Here we report a spectroscopic measurement of the

53: surface magnetic field strength of an AXP, XTE J1810$-$197, and

54: solidify its magnetar nature. The field strength obtained from

55: detailed spectral analysis and modeling, B =

56: (2.72$\pm$0.03)$\times10^{14}$~G, is remarkably close to the value

57: inferred from the rate of spindown of this source and remains nearly

58: constant during numerous observations spanning over an order of

59: magnitude in source flux. The surface temperature, on the other hand,

60: declines steadily and dramatically following the 2003 outburst of this

61: source. Our findings demonstrate that heating occurs in the upper

62: neutron star crust during an outburst and sheds light on the transient

63: behaviour of AXPs.

64: \end{abstract}

65: 

66: \keywords{pulsars: general --- stars: individual (XTE J1810--197) --- 

67: stars: neutron --- X-rays: stars}

68: \section{Introduction}

69: 

70: The X-ray pulsar XTE J1810$-$197 was discovered (Ibrahim et al.\ 2004)

71: in 2003 when it suddenly brightened to more than 100 times its

72: quiescent value (Halpern \& Gotthelf 2005) during an outburst. The

73: source showed a steady decline of its X-ray flux thereafter,

74: accompanied by significant spectral changes (Gotthelf \& Halpern

75: 2006). The 5.54~s period of the source, as well as the large period

76: derivative $\dot P \approx 10^{-11}$~s~s$^{-1}$ were established

77: (Ibrahim et al.\ 2004), confirming the source as an Anomalous X-ray

78: Pulsar (AXP). The detection of characteristic X-ray bursts (Woods et

79: al.\ 2005), similar to those seen in other AXPs (Gavriil, Kaspi, \&

80: Woods 2002, 2006), further strengthened this

81: classification.

82: 

83: XTE J1810$-$197 resides on one extreme of the diverse spectrum of

84: variability properties observed from AXPs. These span at least four

85: types of different activity in pulsed and persistent emission, ranging

86: from the short-lived energetic bursts to outbursts that are

87: characterized by sudden increases and subsequent long-timescale decays

88: in the persistent flux (Kaspi 2006). Moreover, XTE J1810$-$197 is also

89: unique in its unusually low quiescent flux levels that have been

90: determined from archival XTE and ROSAT data (Gotthelf \& Halpern

91: 2006), almost two orders of magnitude fainter than the other known

92: AXPs, earning it the title of the transient AXP.

93: 

94: As with the X-ray spectra of the other known AXPs, the spectra of XTE

95: J1810$-197$ have so far been analyzed by fitting empirical functions

96: such as two blackbodies or a blackbody plus a power-law to the data

97: (Gotthelf \& Halpern 2005). Such analyses are typically used for

98: providing a rough estimate of the surface temperature of AXPs, even

99: though neutron star surfaces do not emit like blackbodies. The photon

100: energy-dependent radiation processes in their atmospheres strongly

101: distort the emission originating deep in the neutron stars away from a

102: blackbody spectrum (see, e.g., Shibanov et al.\ 1992). The strong

103: magnetic fields that the sources are thought to possess, based on the

104: rapid spindowns (Kouveliotou et al.\ 1998), also leave distinctive

105: imprints on the spectra both by altering the radiation processes in

106: their atmospheres (\"Ozel 2003) and by giving rise to moderate

107: scattering optical depths in the magnetospheres (Thompson, Lyutikov,

108: \& Kulkarni 2002).

109: 

110: In this Letter, we analyze the spectra of XTE J1810$-$197 for the

111: first time with a physical model that incorporates emission from the

112: magnetar surface and its reprocessing in the magnetosphere, as

113: described in G\"uver, \"Ozel, \& Lyutikov (2006). Unlike empirical

114: fits, we account for the known radiative processes that take place on

115: and around a magnetar for a range of magnetic field strengths and

116: surface temperatures and base our analysis on the resulting set of

117: models. The fits, therefore, can yield the physical parameters of the

118: source. Conversely, successfully reproducing in detail the spectral

119: characteristics of every epoch with a single physical model, while

120: keeping consistent values for these parameters, would indicate that

121: this model captures all the relevant physical effects that take place

122: on a magnetar and is thus a validation of the theoretical model. We

123: choose XTE J1810$-$197 a prime candidate for this detailed study,

124: because it has gone through extreme variations in its X-ray flux and

125: spectrum over its short lifetime and, thus, it is not a priori obvious

126: that such a wide range of spectra can be fit with a model that depends

127: only on four physical parameters.

128: 

129: We describe our physical model in Section 2 and present the data

130: analysis and results for XTE J1810$-$197 in Section 3. In Section 4,

131: we discuss the implications of the spectroscopically determined

132: magnetic field strength and conclude with a discussion of the

133: mechanism responsible for the outburst of the transient magnetar based

134: on the results of our analysis.

135: 

136: \section{The Surface Thermal Emission and Magnetospheric Scattering Model}

137: 

138: We base our spectral analysis on the physical model of magnetars that

139: we have developed (G\"uver et al.\ 2006), which for the first time

140: takes into account the relevant mechanisms that take place both on the

141: atmosphere and in the magnetosphere of a magnetar. The {\it Surface Thermal

142: Emission and Magnetospheric Scattering} model (STEMS) depends only on four

143: physical parameters that describe the surface magnetic field strength

144: and temperature of the neutron star, as well as the density and the

145: energetics of charges in its magnetosphere.

146: 

147: In our detailed calculations, we address the polarization-mode

148: dependent transport of radiation, treating absorption, emission, and

149: scattering processes that take place in the fully ionized plasmas of

150: hot ($\sim 0.1-0.6$~keV) magnetar atmospheres (\"Ozel 2003). The model

151: also incorporates the interaction of photons with the protons in the

152: plasma that gives rise to absorption features at the proton cyclotron

153: energy. Furthermore, we fully calculate the effects of vacuum

154: polarization resonance, which leads to an enhanced conversion between

155: photons of different polarization modes as they propagate outward

156: through the atmosphere. We have calculated spectra spanning the range

157: of surface magnetic field strengths $B = 5 \times 10^{13} - 5 \times

158: 10^{15}$~G and surface temperatures $T = 0.1-0.6$~keV, in line with

159: the physical processes incorporated into the calculations.

160: 

161: In the stellar magnetospheres, we include a treatment of resonant

162: scattering (G\"uver, \"Ozel, \& Lyutikov 2006). The enhanced current

163: density in the magnetosphere of a magnetar significantly increases the

164: optical depth to electron scattering experienced by the outgoing

165: atmospheric photons (Thompson et al.\ 2002). The resulting

166: upscattering modifies both the high-energy continuum and the

167: equivalent widths of the proton cyclotron absorption features (Lyutikov

168: \& Gavriil, 2006; G\"uver et al., 2006).

169: 

170: Finally, because the surface photons originate in the strong

171: gravitational field of the neutron star, we follow the general

172: relativistic propagation of the photons to an observer on Earth. This

173: last step depends on the mass-to-radius ratio of the neutron star (for

174: which we assume a fixed fiducial value of $z= (1-2M/R)^{-1/2}-1 =

175: 0.3$) and is necessary to make the physical models directly comparable

176: to the observations of AXPs. Here, $M$ and $R$ are the mass and radius

177: of the neutron star, respectively, given in gravitational units.

178: 

179: In order to compare observed X-ray spectra with the STEMS model, we

180: numerically calculated model X-ray spectra (in the 0.05 - 8.12 keV

181: range) and created a table model which can be used within XSPEC

182: (Arnaud 1996).

183: 

184: \section{Data Analysis and Results} 

185: 

186: XTE J1810$-$197 was observed for a total of 170 ks in seven pointings

187: between August 9, 2003 and March 12, 2006 with EPIC-PN onboard

188: XMM-Newton. During these observations, the unabsorbed $0.5-7$~keV flux

189: of the source varied from

190: 57.96$\times~10^{-12}$~erg~s$^{-1}$~cm$^{-2}$ at its peak to

191: 4.17$\times~10^{-12}$~erg~s$^{-1}$cm$^{-2}$ during the last

192: observation. Fast read out of EPIC-PN cameras and decreasing flux of

193: the source with time ensures that the spectra were not affected by

194: photon pile-up in any of the observations.

195: 

196: We calibrated all observations using the Science Analysis Software

197: (SAS, v. 7.0.0) and the latest (January 2007) available calibration

198: files. We eliminated the segments of data which were highly affected

199: by high-energy particle background. We grouped all the spectra before

200: background subtraction so that each spectral bin has at least 25

201: counts. We then fit the spectra, using XSPEC v11.3.2, with the STEMS

202: model we discussed above.  We allowed for interstellar extinction from

203: a cold medium with cosmic abundances.

204: 

205: Figure~1 shows the spectra observed in the seven different epochs, the

206: best fit models, and the residuals. The model describes in detail the

207: salient features of the spectra in the entire energy range. This is

208: especially remarkable because the significant evolution of the spectra

209: during the decay of the outburst can be fit with a single physical

210: model. The $\chi^2_{\nu}/$d.o.f. for the spectral fits, in order of

211: decreasing source flux are 1.07/732, 0.95/548, 1.11/820, 1.12/772,

212: 1.21/653, 1.02/423, 1.02/302.  Even more compelling than the low

213: $\chi^2_{\nu}$ values is the flatness of residuals that demonstrate the

214: ability of the model to reproduce the observations without the need

215: for any additional ad hoc components such as a blackbody or a

216: power-law function.

217: 

218: In the analyses of all seven observations, we obtain a nearly

219: constant value of $N_{\rm H} = 0.63 \pm 0.08 \times10^{22}$~cm$^{-2}$

220: for the equivalent hydrogen column density responsible for the

221: interstellar extinction, even though we allow this parameter to vary

222: between observations. This value is also in agreement with an

223: independent study of this source (Gotthelf \& Halpern 2006). The

224: scattering optical depth $\tau$ and the velocity distribution $\beta$

225: of the electrons in the magnetosphere also remain fairly constant,

226: around values $\tau \approx 4.5 \pm 1.1$ and $\beta \approx 0.22 \pm

227: 0.05$.  These results show that there is no significant variation in

228: the magnetosphere of the source during the window of the XMM-{\it

229: Newton} observations. If the onset of the outburst introduced any

230: magnetospheric changes, these must have stabilized within the months

231: between the peak of the outburst and the first XMM-{\it Newton}

232: observation.

233: 

234: \section{Discussion}

235: 

236: Our analysis of the 0.5-7~keV spectra of XTE J1810$-$197 with the

237: STEMS model allows for a tight and unique constraint of the magnetic

238: field strength of the source. The best-fit values for the surface

239: magnetic field and the surface temperature of the neutron star,

240: obtained from the detailed fits, as well as the 1- and 2-sigma

241: confidence limits are shown in Figure~2. The magnetic field strength

242: ranges from $B=(2.25 \pm 0.05) \times 10^{14}$~G in the earliest

243: observation to $B=(3.3 \pm 0.1) \times 10^{14}$~G in the last one,

244: while the temperature declines from $T=0.49 \pm 0.02$~keV to $T=0.22

245: \pm 0.03$~keV in the same interval. Note that for the last two

246: observations which have very low flux levels and poorer statistics,

247: the confidence contours were drawn by freezing the other model

248: parameters. For comparison, we also obtained the value of the magnetic

249: field strength from fitting all the data sets simultaneously, which

250: yields $(2.72 \pm 0.03) \times 10^{14}$~G.

251: 

252: The confidence contours show that the magnetic field can be tightly

253: constrained in each observation. This is because of the presence of

254: significant broad features in the magnetar spectra imparted by

255: weakened proton cyclotron lines and the vacuum polarization resonance

256: that have a strong dependence on the magnetic field strength. As we

257: previously discussed in G\"uver et al.\ (2006), these features allow

258: for a precise determination of the magnetic field strength from

259: continuum spectra. To demonstrate this effect, we plot in Figure~3 the

260: deviation of the model from the data obtained on March 18, 2005 (i.e.,

261: the longest observation) when the magnetic field strength is

262: artificially set to 8-sigma higher than the best-fit value, while all

263: the other parameters remain at their best-fit values (obtained before

264: modifying the field strength). The deviations are due to the broad

265: features that can be seen easily in the residuals. It is the detection

266: of these unique modifications in the spectra of XTE J1810$-$197 that

267: allows for the measurement of its surface magnetic field strength.

268: 

269: The contours shown in Figure 2 indicate that the surface magnetic

270: field strength and the surface temperature are not correlated. We find

271: that this is also true for all the other model parameters. As an

272: example, when we perform the above experiment where we set the

273: magnetic field strength to an articially higher value but, instead,

274: allow the other model parameters to change in the fit, we find a local

275: minimum with statistic value 805/653, which is higher than the global

276: minimum.

277: 

278: The measured magnetic field strength remains nearly constant during

279: the decline of the outburst. The slight evolution of the field

280: strength seen in Figure~2 likely arises from using phase-averaged

281: spectra in conjunction with the geometric simplicity of our physical

282: model that does not take into account variations of the field strength

283: on the neutron star surface. It may also be partially affected by the

284: spatial evolution of the hotspot that gave rise to the outburst on the

285: neutron star surface. Therefore, confirming or rejecting this trend

286: requires modeling that includes more details about the magnetic field

287: topology. The analysis of the pulse-phase resolved spectra (or

288: equivalently, the energy-dependent pulse profiles) of this source will

289: be reported elsewhere.

290: 

291: Our measurement is also in good agreement with the value of the

292: magnetic field inferred from the average spindown rate of the pulsar

293: using XMM-Newton and RXTE observations of the pulsar (Gotthelf \&

294: Halpern, 2006, Ibrahim et al.\ 2004), as well as from the

295: short-timescale measurements of the period derivative from radio

296: observations (Camilo et al.\ 2007).  Such an accord is unanticipated

297: given the numerous assumptions involved in inferring the magnetic

298: field strength with a vacuum dipole spindown formula (Spitkovsky

299: 2006). Thus, our independent, spectroscopic

300: measurement of the magnetic field strength is a validation of the use

301: of the dipole spindown formula.

302: 

303: The surface temperature is also well constrained, as can be seen in

304: the tight confidence limits in Figure~2. The time arrow shows the

305: monotonic decline of the temperature during the sequence of the seven

306: XMM-{\it Newton} observations. The decay of the source flux during the

307: same time interval can be explained entirely by the cooling of the

308: neutron star crust, as described by the single temperature parameter,

309: without significant changes to the emitting area on the neutron star

310: surface. Assuming a distance of 3.3 kpc (Camilo et al. 2006), the

311: radius of this hot region remains approximately $3.7$~km, likely

312: corresponding to the area that is heated during the outburst. As

313: discussed earlier, the scattering optical depth $\tau$ and the

314: velocity distribution $\beta$ of electrons in the magnetosphere also

315: remain fairly constant, despite the changes in the hardness of the

316: spectra as the source cools. Indeed, the spectral changes are best

317: described by a change in the temperature alone, without accompanying

318: changes to any other parameter describing the neutron star surface or

319: its magnetosphere.

320: 

321: This physical model allows us to track the changes in the AXP during

322: its decline from its outburst and probe the mechanism that produces

323: the transient behavior. Suggested ideas for the observed flares and

324: outbursts for the AXPs and SGRs rely either on the injection of heat

325: deep in the crust or a sudden change in the topology of the field

326: lines in the magnetosphere. Our analysis, which disentangles the

327: contributions of the processes in the magnetosphere from those on the

328: stellar surface, shows that it is the release of heat in the crust,

329: and not changes in the magnetosphere, that is responsible for this long 

330: timescale AXP outburst.

331: 

332: We can identify the depth in the crust where the heat is released to

333: produce the outburst of XTE J1810$-$197 by considering the energetics

334: of the outburst and the measured evolution of the

335: temperature. Assuming that the heat is deposited over a surface area

336: $S$ at a depth $h$, where the particle density and temperature are

337: given by $N_{\rm d}$ and $T_{\rm d}$, respectively, we can calculate

338: the total energy of the outburst $E$ as $E \approx L \Delta t \approx

339: 3 N_{\rm d} k_{\rm B} \Delta T_{\rm d} S h$. Here, $\Delta T_{\rm d}$

340: is the resulting increase in the temperature in the deep layer, which

341: is related to the change in the effective temperature by $\Delta

342: T_{\rm d} / T_{\rm d} \approx \Delta T_{\rm eff} / T_{\rm eff}$ by the

343: Eddington-Barbier relation. Assuming a distance of 3.3 kpc (Camilo et

344: al.\ 2006) and a timescale of $\Delta t \simeq 1$~yr, we estimate a

345: total energy of $10^{42}$~erg for the outburst using the typical

346: luminosity $L \simeq 3 \times 10^{34}$. Finally, we calculate the

347: particle density at a given depth using the detailed surface model of

348: the neutron star used in fitting the spectral data. Requiring that,

349: during the outburst, $\Delta T_{\rm eff}/T_{\rm eff}$ is larger than

350: $0.27$~keV$/0.22$~keV, as inferred from the temperature evolution, we

351: find that the energy release occurred at a depth of $\simeq 2.5$~m,

352: which corresponds to a column depth of $2 \times

353: 10^{11}$g~cm$^{-2}$. This shows that the currents carrying the

354: magnetic field must be decaying in the upper crust. For the transient

355: AXP, the lack of subsequent energy release at such depths allows the

356: crust to cool completely, and fade back to the very low quiescent flux level.

357: 

358: 

359: \acknowledgements

360: 

361: It is a pleasure to thank Jules Halpern for suggesting XTE J1810-197

362: as an ideal candidate for spectral studies. We also thank Pat Slane,

363: Harvey Tannanbaum, Jeff McClintock, and Anatoly Spitkovsky for useful

364: discussions, and Dimitrios Psaltis, Ali Alpar and Fernando Camilo for

365: detailed comments on the manuscript. We also would like to thank the

366: referee Fotis Gavriil for useful suggestions. F.\"O. acknowledges

367: support from a Visiting Faculty fellowship from the Scientific and Technological

368: Research Council of TURKEY (TUBITAK). 

369: E.G. acknowledges partial support from the Turkish Academy

370: of Sciences through grant E.G/TUBA-GEB\.IP/2004-11.

371: 

372: \begin{references}

373: 

374: \reference{a96} Arnaud, K.~A., 1996, Astronomical

375: Data Analysis Software and Systems V\ eds. Jacoby G. and Barnes J.,

376: p17, ASP Conf. Series volume 101

377: 

378: \reference{c06} Camilo F. et al., 2006, Nature, 442, 892

379: 

380: \reference{c07} Camilo F. et al., 2007, ApJ, in press, ArXiv

381: Astrophysics e-prints arXiv:astro-ph/0610685

382: 

383: \reference{gkw02} Gavriil, F.~P., Kaspi, V.~M.,

384: Woods, P.~M., 2002, Nature 419, 142

385: 

386: \reference{gkw06} Gavriil, F.~P., Kaspi, V.~M.,

387: Woods, P.~M., 2006, \apj, 641, 418

388: 

389: \reference{gh05} Gotthelf, E.~V., Halpern, J.~P., 2005., 

390: \apj, 632, 1075

391: 

392: \reference{gh06} Gotthelf, E.~V., 

393: Halpern, J.~P., 2006, ArXiv Astrophysics 

394: e-prints arXiv:astro-ph/0608473

395: 

396: \reference{gol06} G\"uver, T., \"Ozel, F., Lyutikov, M.\ 2006, 

397: ArXiv Astrophysics e-prints arXiv:astro-ph/0611405

398: 

399: \reference{hg05} Halpern, J.~P., 

400: Gotthelf, E.~V., 2005, \apj, 618, 874

401: 

402: \reference{ibrahim} Ibrahim, A.~I., et al., 2004, \apj, 609, L21

403: 

404: \reference{kaspi} Kaspi, V.\ M.\ 2006, ArXiv Astrophysics e-prints

405: arXiv:astro-ph/0610304

406: 

407: \reference{k98} Kouveliotou, C., et al., 1998, Nature, 393, 235

408: 

409: \reference{l2006} Lyutikov, M., \& Gavriil,

410:   F.P. 2006, \mnras, 368, 690L

411: 

412: \reference{ozel03} {\"O}zel, F.\ 2003, \apj, 583, 

413: 402

414: 

415: \reference{shibanov92} Shibanov, I.~A., Zavlin, V.~E., Pavlov, G.~G.,

416: \& Ventura, J.\ 1992, \aap, 266, 313

417: 

418: \reference{a06} Spitkovsky, A.\ 2006, \apj, 648, 51.

419: 

420: \reference{tlk02} Thompson, C., 

421: Lyutikov, M., Kulkarni, S.~R., 2002, \apj, 574, 332

422: 

423: \reference{woods05} Woods, P.~M., Kouveliotou, C., Gavriil, F.~P.,

424: Kaspi, V.~M., Roberts, M.~S.~E., Ibrahim, A., Markwardt, C.~B., Swank,

425: J.~H., Finger, M.~H., 2005, \apj, 629, 985

426: 

427: \end{references}

428: 

429: \clearpage

430: 

431: \begin{figure}

432: \includegraphics[angle=-90,scale=0.5]{f1.ps}

433: \caption{Comparison of theoretical magnetar spectra to XMM-Newton

434: observations of XTE J1810$-$197 obtained over three years while the

435: source was declining from its 2003 outburst. Different colors

436: correspond to the seven different epochs of observations. Solid lines

437: show the best-fit theoretical models while the lower panel shows the

438: residuals of the fits in units of standard deviation demonstrating the

439: ability of the model to account in detail for the observed spectra.}

440: \end{figure}

441: 

442: \clearpage

443: 

444: \begin{figure}

445: \includegraphics[angle=0,scale=0.6]{f2.ps}

446: \caption{The spectroscopically measured surface magnetic field

447: strength and temperature during the decline of the 2003 outburst of

448: XTE J1810$-$197. The contours show the one- and two-sigma confidence

449: limits on the parameters obtained from the individual

450: observations. The hatch-filled area shows the magnetic field inferred

451: from the observed rate of spindown, assuming magnetic dipole braking,

452: from the X-ray observations of Gotthelf \& Halpern (2006) and Ibrahim

453: et al.\ (2004), as well as the radio measurements of Camilo et al.\

454: (2007). The spectroscopically determined field strength is remarkably

455: close to the range of values inferred from the dipole spindown

456: formula. The monotonic and rapid decline of the measured effective

457: temperature is the only significant change in the source properties

458: during the outburst.}

459: \end{figure}

460: 

461: \clearpage

462: 

463: \begin{figure}

464: \includegraphics[scale=1.0]{f3.ps}

465: \caption{The deviation of a test model from the data obtained on March

466:   18, 2005, when the magnetic field strength is artificially set

467:   to $2.92 \times 10^{14}$~G, that is 8-sigma higher than the best-fit

468:   value, while all the other parameters remain at their best-fit

469:   values.}

470: \end{figure}

471: 

472: \end{document}

473: 

474: