1: 

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

3: \usepackage{amsmath}

4: 

5: % \documentclass[manuscript]{aastex}

6: 

7: \newcommand{\vdag}{(v)^\dagger}

8: \newcommand{\myemail}{skywalker@galaxy.far.far.away}

9: 

10: %% You can insert a short comment on the title page using the command below.

11: 

12: %\slugcomment{Not to appear in Nonlearned J., 45.}

13: 

14: %\shorttitle{UX UMa}

15: %\shortauthors{Linnell et al.}

16: 

17: \begin{document}

18: 

19: \title{Modeling 

20: UX~Ursae~Majoris\footnotemark[1]~: An abundance of challenges.}

21: \footnotetext[1]

22: {Based on observations made with the NASA/ESA Hubble Space Telescope, obtained at the

23: Space Telescope Science Institute, which is operated by the Association of Universities 

24: for Research in Astronomy, Inc. under NASA contract NAS5-26555, and the NASA-CNES-CSA

25: {\it Far Ultraviolet Explorer}, which is operated for NASA by the Johns Hopkins University

26: under NASA contract NAS5-32985.}

27: 

28: 

29: %% Use \author, \affil, and the \and command to format

30: %% author and affiliation information.

31: %% Note that \email has replaced the old \authoremail command

32: %% from AASTeX v4.0. You can use \email to mark an email address

33: %% anywhere in the paper, not just in the front matter.

34: %% As in the title, you can use \\ to force line breaks.

35: 

36: 

37: \author{Albert P. Linnell$^2$, Patrick Godon$^3$, Ivan Hubeny$^4$, Edward M. Sion$^5$,

38: Paula Szkody$^6$}

39: 

40: \affil{$^2$Department of Astronomy, University of Washington, Box 351580, Seattle,

41: WA 98195-1580\\

42: $^3$Department of Astronomy and Astrophysics, Villanova University,

43: Villanova, PA 19085\\

44: visiting at the Space Telescope Institute, Baltimore, MD.\\

45: $^4$Steward Observatory and Department of Astronomy,

46: University of Arizona, Tucson, AZ 85721\\

47: $^5$Department of Astronomy and Astrophysics, Villanova University,

48: Villanova, PA 19085\\

49: $^6$Department of Astronomy, University of Washington, Box 351580, Seattle,

50: WA 98195-1580\\

51: }

52: 

53: \email{$^2$linnell@astro.washington.edu\\

54: $^3$godon@stsci.edu\\

55: $^4$hubeny@as.arizona.edu\\

56: $^5$edward.sion@villanova.edu\\

57: $^6$szkody@astro.washington.edu\\

58: }

59: 

60: \begin{abstract}

61:  We present a system model for optical and far UV spectra of the nova-like variable 

62:  UX~UMa	involving a white dwarf, secondary star, gas stream, hot spot and accretion 

63:  disk using our code BINSYN and based on an initially adopted system distance.

64:  Calculated SED intensity data successfully fit successive

65:  tomographically-extracted annuli longward of the Balmer limit but require

66:  a postulated `iron curtain' shortward of the Balmer limit that is applied to the

67:  annulus section closest to the secondary star, while postulated

68:  recombination emission fills in the model SED shortward of the Balmer limit and

69:  is applied to the annulus section more remote from the secondary star.

70:  The same model fits $UBV$ 1954 light curves by Walker and Herbig. Fits to

71:  $HST$ $FOS$ spectra are approximate but require assumed time-variable

72:  changes in the SED. Comparable effects, possibly involving variable absorption, afflict $FUSE$

73:  spectra. Fits to $IUE$ spectra by the model show time-dependent residuals

74:  that indicate changes in the accretion disk temperature profile, possibly indicative of a

75:  slightly variable $\dot{M}$ from the secondary star. Using model-based component

76:  light contributions and the improvement on the Bailey relation by Knigge we determine the system distance

77:  and mass transfer rate. 

78: 

79: 

80: \end{abstract}

81: 

82: %% Keywords should appear after the \end{abstract} command. The uncommented

83: %% example has been keyed in ApJ style. See the instructions to authors

84: %% for the journal to which you are submitting your paper to determine

85: %% what keyword punctuation is appropriate.

86: 

87: \keywords{accretion, accretion disks --- novae, cataclysmic variables --- stars:

88: individual(UX~UMa) --- ultraviolet: 

89: stars --- white dwarfs}

90: %% From the front matter, we move on to the body of the paper.

91: %% In the first two sections, notice the use of the natbib \citep

92: %% and \citet commands to identify citations.  The citations are

93: %% tied to the reference list via symbolic KEYs. The KEY corresponds

94: %% to the KEY in the \bibitem in the reference list below. We have

95: %% chosen the first three characters of the first author's name plus

96: %% the last two numeral of the year of publication as our KEY for

97: %% each reference.

98: 

99: 

100: \section{Introduction}

101: \label{s-intro}

102: 

103: Cataclysmic variables (CVs) are

104: semi-detached binary stars in 

105: which a late-type typically main sequence star loses mass onto a white dwarf (WD) 

106: via Roche lobe overflow and  

107: accretion proceeds through a viscous disk. In the nova-like (NL) subclass, of which UX UMa 

108: is the prototype, the mass 

109: transfer rate from the secondary star 

110: is large enough that dwarf nova outbursts do not occur. 

111: (See \citet{warner95} for a history of UX UMa studies and a thorough review of CV types and properties.)

112: Important early studies of UX UMa by \citet{wh1954,jph1954} and \citet{kw1963} established the

113: system photometric properties, while \citet{nr1974} discovered coherent oscillations.

114: $IUE$ spectra identified UX UMa as a UV emission line variable \citep{holm1982,king1983}.

115: X-ray observations \citep{becker1981,wood1995} detected a soft source, further refined \citep{pratt2004} to

116: a noneclipsed soft component and a deeply eclipsed hard component. An accretion disk wind was

117: discovered by \citet{mason1995} and modeled by \citet{knigge1997}.

118: 

119: \citet{knigge1998a} had mixed success in modeling the spectral energy distribution (SED) of UX UMa

120: with a combination of stellar atmosphere synthetic spectra. A similar approach by \citet{froning2003}

121: (hereafter FR2003)

122: in a fit to the $FUSE$ spectrum produced an estimated $\dot{M}=1.6{\times}10^{-9}~M_{\odot}~{\rm yr}^{-1}$.

123: However, a mass transfer rate of

124: $\dot{M}=1.6{\times}10^{-9}~M_{\odot}~{\rm yr}^{-1}$ is small enough that the system would be

125: unstable against outburst, contrary to observation.

126: Neither of these two studies included a SED contribution from the WD;

127: on the other hand, tomographic reconstruction of the UX UMa disk \citep{ru1992} is consistent with

128: a standard model accretion disk with $\dot{M}=5{\times}10^{-9}~M_{\odot}~{\rm yr}^{-1}$.

129: A comparable study by \citet{bap1995} hereafter BA1995, with a different estimated distance to the system, finds

130: $\dot{M}=1{\times}10^{-8}~M_{\odot}~{\rm yr}^{-1}$.

131: \citet{frank1981} modeled IR and optical light curves and determined system parameters.	

132: \citet{smak1994a} modeled optical region light curves and derived bright spot parameters. 

133: \citet{mason1997} argue

134: that a more complicated structure than a bright spot on an 

135: otherwise unmodified accretion disk rim is necessary. 

136: BA1995 derived a WD $T_{\rm eff}$ of 52,000K to 70,000K, but FR2003 points out that a

137: $T_{\rm eff}$ that large is inconsistent with FUV spectra.

138: Neither of the tomographic reconstruction studies has been the basis of an effort to model

139: the spectra.

140: These continuing inconsistencies are the motivation to attempt to derive a single model that represents

141: a substantial part of the disparite observational data.    

142: 

143: \section{Observational Data: Spectroscopic and Photometric Observations}

144: 

145: Table~1 lists the spectroscopic observations used in this study. 

146: We obtained the $IUE$ spectra from the $IUE$ archive and processed them with the standard

147: IUEDAC package.

148: The $HST/FOS$ observations are from \citet{knigge1998a}, further analyzed in

149: \citet{bap1998}, hereafter BA1998, and described in that publication: 

150: the G160L data sets each include 691 spectra and

151: the PRISM data sets each include 870 spectra; these time-resolved spectra have a resolution of 5.3 s.

152: We obtained the time-integrated spectrum of each of the G160L data sets from the MAST archive for

153: our analysis and designate this data set DS1.

154: We used

155: spectroscopic data from the \citet{ru1994} paper, 

156: consisting of optical spectra obtained in June 1992; we designate this data set DS2.

157: For that data set we used the Dexter facility of the

158: Astrophysics Data Service (ADS) to digitize the spectrally-resolved disk spectra for regions A-F in 

159: their Figures~5a-f. 

160: 

161: In agreement with \citet{ru1992}, we adopt an interstellar reddening of $E(B-V)=0.0$.

162: \citet{bruch1994} list $E(B-V)=0.02$. This small value is uncertain because of the variability of the system

163: light curve discussed below and we prefer to leave the reddening at 0.0 for our analysis.

164: Unless stated otherwise, changing to $E(B-V)=0.02$ has a negligible effect on our results.

165: The $FUSE$ spectra, from the MAST archive (and used by FR2003), were processed with CalFUSE v3.1. 

166: The dataset includes 

167: multiple exposures; we

168: have selected two for this study.

169: The $HST/FOS$ spectral lines were identified and studied by \citet{knigge1998a} and BA1998 while

170: the $FUSE$ spectral lines were identified and studied by FR2003; we do not repeat those identifications.

171: 

172: \section{Initial System Parameters}

173: 

174: \citet{ru1992} list several determinations of the distance to UX UMa and adopt a compromise value

175: of 250pc. Additional determinations are

176: \citet{smak1994a} ($328{\pm}12$pc), \citet{frank1981} ($340{\pm}110$pc), and BA1995 ($345{\pm}34$pc). 

177: FR2003 challenge 

178: the BA1995

179: system parameters, and our analysis agrees with some of the FR2003 points. 

180: (We have concerns about the BA1995 determination; see \S6.)

181: We choose to adopt the distance of 250pc for our working model, initially as an illustrative test and

182: subject to subsequent consistency verification.	The distance uncertainty remains a major obstacle and

183: there is a degeneracy between the

184: system distance and the adopted $\dot{M}$. 

185: 

186: \citet{ru1994} determined an approximate spectral type of M0 for the secondary star in UX UMa and a

187: corresponding main sequence mass of ${\sim}0.5M_{\odot}$. This determination, together with the

188: orbital period, is consistent with the \citet{smak1994a} system parameters of 

189: $M_{\rm wd}=0.70{\pm}0.2M_{\odot}$, $M_{\rm s}=0.49M_{\odot}$, and $q=0.70$. The period-secondary mass relation of

190: \citet[eq. 2.100]{warner95} gives a secondary mass of $M_{\rm s}=0.48M_{\odot}$, while the \citet{knigge2006}

191: period-secondary mass relation finds $M_{\rm s}=0.44M_{\odot}$.

192: A radial velocity study by \citet{sh1984} finds that 

193: $M_{\rm s}{\la}0.5M_{\odot}$ and the mass ratio of the system is greater than ${\sim}0.8$.

194: \citet{frank1981} find $0.2<M_{\rm s}/M_{\odot}<0.5$ and $0.1<M_{\rm wd}/M_{\odot}<0.5$.

195: They also find $\dot{M}=8{\times}10^{-9}~M_{\odot}~{\rm yr}^{-1}$ (see below). It is worth mentioning

196: that, at an orbital period of $4.7^h$, UX UMa lies in a region of parameter space where nuclear

197: evolution of secondaries becomes significant, adding a measure of uncertainty to standard relationships.

198:  

199:  The BA1995 $M_{\rm wd}=0.47M_{\odot}$ depends on the radial velocity value

200:  $K_{\rm wd}=160~{\rm km}~{\rm s}^{-1}$ \citep{sh1984} determined from the emission wings of H$\alpha$.

201:  \citet{sch1983} found a range of $K_{\rm wd}$ values for different spectral features.	FR2003 used 

202:  a cross-correlation procedure

203:  to determine a $K_{\rm wd}$ of $70~{\rm km}~{\rm s}^{-1}$, while absorption reversals in emission lines determined

204:  $K_{\rm wd}$ values from $140~{\rm km}~{\rm s}^{-1}$ for S~IV and S~III to $200~{\rm km}~{\rm s}^{-1}$ for C~III.

205:  Thus there is some uncertainty in the $M_{\rm wd}$ determination; a more secure value of $M_{\rm wd}$ depends on

206:  an improved radial velocity curve attributable to the WD. 

207:  Note that the value $q=1$ is at the limit of stability

208:  against dynamical mass transfer which occurs for $q>1$. 

209: We provisionally adopt the BA1995 model, $M_{\rm wd}=M_{\rm s}=0.47M_{\odot}$. This model	closely

210: agrees with the model adopted by \citet[Table~4]{ru1992}, $M_{\rm wd}=M_{\rm s}=0.45M_{\odot}$.

211: 

212: From \citet{p2000}, the zero temperature radius of a $0.47M_{\odot}$ WD is 

213: $R_{\rm wd,0}=0.01375R_{\odot}=9.57{\times}10^8{\rm cm}$ and, 

214: from their Fig.4a, we find that the radius of a 20,000K,

215: $0.47M_{\odot}$ WD, used later in our discussion,

216: would be $1.15{\times}10^9{\rm cm}$. 

217: BA1995 argue that, if 1/2 of the WD is visible,

218: it has a $T_{\rm eff}=70,000$K. This temperature is off the Figure~4a plot of \citet{p2000}, but we estimate 

219: a radius of $3.5{\times}10^9{\rm cm}$. A 70,000K WD with this radius (and visible to the observer)

220: would completely dominate 

221: the system flux and would lead to a seriously discrepant synthetic spectrum (see below).	

222: (Also see the discussion by FR2003.)

223: 

224: Table~2 lists the system parameters	used in this paper. Except for $i$ the parameters are adopted as discussed

225: herein and so have no errors attached.

226: The Roche potential for the WD produces a radius appropriate to a $0.47M_{\odot}$, 20,000K WD.

227: The secondary star Roche potential causes it to fill its Roche lobe.

228: \citet{ru1992} find $i=71{\arcdeg}$ for their $q=1.0$ model. They investigate the sensitivity of this

229: result to variation of $q$ and $i$, and find their result is not very sensitive provided the eclipse

230: width remains unchanged. In our light curve simulations (\S4.1) we found a slight

231: improvement in the fit to eclipse depth with $i=70.2{\arcdeg}$, but a detectably poorer fit to

232: eclipse width with $i=70.0{\arcdeg}$. We adopt $i=70.2{\arcdeg}$.

233: 

234: Using a maximum-entropy eclipse mapping algorithm (the MEM method), \citet{ru1992} determine a 

235: temperature profile for

236: the UX UMa accretion disk (their Figure~4b). The procedure uses four optical-wavelength eclipse light

237: curves that constitute the observational data. A two-dimensional array of surface elements (pixels)

238: covers the model accretion disk which is projected on the sky at inclination $i$. A default intensity

239: is assigned to each pixel at each of the observed wavelengths and an assumed distance together with

240: calibrated absolute magnitudes for the observational data produces intensities in physical units. 

241: The algorithm then iterates to vary

242: the individual intensities until the entire calculated array of pixel intensities has maximum entropy, thereby

243: fitting the observed light curves with calculated light curves. A black body fit to individual pixel

244: intensities at the four observed wavelengths determines $T_{\rm eff}$, pixel by pixel. The temperature

245: profile then is a plot of the pixel $T_{\rm eff}$ values as function of the pixel distance from the

246: center of the WD.

247: 

248: We used the ADS Dexter facility to digitize the

249: points in their plot. 

250: We also used the Dexter facility to digitize the BA1998 `back' points (discussed in \S4.) from their

251: Figure~7 for November 1994.

252: Figure~1 compares the temperature profile in our model with the points

253: from Rutten et al. (crosses) and Baptista et al. (triangles). Note that the plotted points do not

254: represent directly observed quantities; rather, they are derived quantities from the maximum

255: entropy maps.

256: Based on the Figure~1 plot, our model adopts a standard model \citep{fkr92}

257: $\dot{M}=5.0{\times}10^{-9}M_{\odot}{\rm yr}^{-1}$ temperature profile as an initial approximation.	

258: By eye estimate, the temperature profile for our model is a fairly good fit to the plotted data points.

259: The data points depart from	a standard model near the L1 point where the bright spot and tidal heating

260: exert an effect (discussed below). 

261: 

262: There are no data points at extremely small radii; we find 

263: ambiguous evidence for obscuration of the WD and/or for a hot boundary layer, raising doubts about the

264: justification of modeling the theoretical temperature downturn at very small radii. The theoretical

265: downturn assumes a slowly-rotating central star \citep{fkr92}; with faster rotation the downturn

266: becomes smaller and we have no information on the rotation rate of the WD.

267: Our Figure~1 profile shows only the start of a temperature drop near the WD and so fails to fit the 

268: theoretical $\dot{M}$ profiles at small radii.	

269: Our model calculates only one annulus with

270: radius less than the radius of the temperature maximum at $R=1.36R_{\rm wd}$. 

271: 

272: Our program uses the \citet[Equation~2.61]{warner95} relation for the tidal cutoff radius and for $q=1.0$ this leads

273: to $R_{\rm cutoff}=0.6R_{\rm L1}$. \citet{ru1992} and BA1998 both tabulate data

274: points to $R=0.7R_{\rm L1}$. This difference explains the failure of our model to extend to the tabular

275: terminus of the Rutten et al. data in Figure~1. We subsequently (\S4.1) model light curves with the aid of a bright spot

276: on the accretion disk rim. \citet[Table~7]{ru1992} determine $R_{\rm spot}/R_{\rm L1}=0.57{\pm}0.05$, in

277: agreement with the rim location in our model. If our model allowed the accretion disk to extend to 

278: $R_{\rm cutoff}=0.7R_{\rm L1}$

279: the $T_{\rm eff}$ of the outermost part of the accretion disk would be below the temperature limit for stability 

280: \citep{osa96}. We do not believe the tidal cutoff sets a sharp boundary beyond which no emitting gas can be found

281: but for formal internal consistency	we adopt $R_{\rm cutoff}=0.6R_{\rm L1}$.

282: 

283: \subsection{Calculation of synthetic spectra}

284: 

285:  Our model uses the BINSYN suite (\citet{linnell1996,linnell2008} and references therein).

286:  This model includes contributions from the accretion disk, the WD, the secondary star, and the

287:  accretion disk rim.

288:  \citet{smak2002} shows that the structure of the outer parts of accretion disks is an important consideration.

289:  Smak shows that allowance for tidal heating produces a rim $T_{\rm eff}$ comparable to that of the

290:  outermost annulus. On a finer scale, which Smak does not discuss in detail, the part of the rim close to the

291:  orbital plane has a higher $T_{\rm eff}$ than at a higher $z$ value. We set the rim $T_{\rm eff}$ equal to

292:  that of the outermost annulus and make the rim isothermal. 

293:  The program requires a set of individually-calculated accretion disk annulus spectra which are based on

294:  annulus models calculated with program TLUSTY \citep{h1988,hubeny1990,hl1995,hh1998}.

295:  The annulus models do not include irradiation by the central star although this is a TLUSTY option; 

296:  we will find there is some

297:  uncertainty concerning the WD $T_{\rm eff}$.

298:  We emphasize that the TLUSTY option for accretion disk annuli (without irradiation) produces a standard 

299:  model radial temperature

300:  profile and includes explicit treatment of the mass transfer rate and the WD mass and radius. 

301:  It would be impractical to recalculate TLUSTY annulus model arrays for different WD radii depending on the radii

302:  appropriate to different assumed $T_{\rm eff}$ values.	We calculate a single array appropriate to

303:  a zero temperature WD and handle the effect of a non-zero temperature WD in the BINSYN model.

304:  Table~3 lists the annuli used in our model and gives some of the individual annulus properties.

305:  The first column lists the radius of the annulus; the second column lists the $T_{\rm eff}$; the third

306:  column lists the column mass to the central plane in ${\rm gm/}{\rm cm}^2$; the fourth column

307:  lists the temperature at the cental plane; the fifth column lists log~$g$ (cgs) at a Rosseland optical depth

308:  of approximately 0.6; the sixth column lists the vertical distance above the central plane of the

309:  optical depth 0.6 point, in cm; the final column lists the Rosseland optical depth of the central plane.

310:  The annulus model calculations through $r/r_{\rm wd,0}=12.0$ converged. The remaining models are grey model

311:  solutions and are subject to some scatter in the calculated $z_{\rm H}$.	

312:  The TLUSTY models have standard model \citep{fkr92} $T_{\rm eff}$ values.

313: 

314:  Program SYNSPEC \citep{hlj1994} uses the TLUSTY output to calculate individual annulus synthetic spectra.

315:  The spectra were calculated with a spectral resolution of 1\AA. An important point is that SYNSPEC

316:  produces both of two 

317:  different output formats. In one, the output is in Eddington flux units 

318:  (${\rm erg}~{\rm cm}^{-2}~{\rm s}^{-1}~{\rm \AA}^{-1}$). In the other the output is in intensities

319:  (${\rm erg}~{\rm cm}^{-2}~{\rm s}^{-1}~{\rm hz}^{-1}~{\rm sr}^{-1}$) for a specified array of directions

320:  relative to the outward normal. In our calculations we specified ten directions equally spaced in

321:  $cos({\gamma})$, where ${\gamma}$ is the angle between the chosen direction and the outward normal to the

322:  local photosphere. We will specifically require the intensity-based data for some of the subsequent discussion.

323: 

324:  BINSYN sets up a separate array of annuli to represent the accretion disk; in this study we used 45

325:  division circles to produce 44 annuli, with inner radius at the WD boundary and outer radius at the

326:  tidal cutoff boundary. This choice provides adequate radial resolution for accurate calculation of

327:  both synthetic spectra and light curves. One half of the WD is visible in this model.

328:  BINSYN uses the \citet[eq.~5.45]{fkr92} equation set to calculate 

329:  a standard model accretion disk, including the rim height.

330:  Table~4 lists the array properties. In calculating flux from annulus segments we use the $T_{\rm eff}$

331:  of the annulus inner edge. 

332:  The tabulation of accretion disk annuli from TLUSTY, Table~3,

333:  begins at the accretion disk temperature maximum since following the theoretical temperature drop to

334:  smaller radii would lead to double-value ambiguity (either side of the maximum) in the BINSYN interpolation.

335:  The algorithm for assigning radii in BINSYN places at least one division circle interior to the

336:  temperature maximum. The two initial entries with the same $T_{\rm eff}$ represent the inner and outer radii

337:  of the first annulus; thereafter the table lists the outer radius.

338: 

339:  The $T_{\rm eff}$ values in Table~4 are appropriate for a 20,000K WD, with its radius determined

340:  as in \S3. The change in WD radius from a zero temperature WD produces a clearly detectable change in the

341:  tabular values of the BINSYN annuli $T_{\rm eff}$ values. Compare the $T_{\rm eff}$ values at

342:  $r/r_{\rm wd}=1.36$ in Table~3 and Table~4; the difference is due to the different values of $r_{\rm wd}$.

343:  BINSYN determines flux values for segments on a given

344:  annulus by using the Table~4 $T_{\rm eff}$ value to interpolate, temperature-wise, within Table~3,

345:  making use of the flux values corresponding to the Table~3 entries. Note that Table~3 is based on

346:  $r_{\rm wd,0}=9.57{\times}10^8~{\rm cm}$ while Table~4	is based on 

347:  $r_{\rm wd}({\rm T=20,000K})=1.15{\times}10^9~{\rm cm}$.

348:  

349:  By option, the Table~4 profile may be replaced by a different non-standard model profile. This option

350:  has been used in a number of instances (e.g., \citet{linnell2008}).

351:  

352:  

353:  \section{Comparison of model with DS1 and DS2}

354: 

355:  Our comparisons will use eclipse maps associated with the DS1 and DS2 data and these maps will require default

356:  intensities for the map pixels.

357:  Calculation of default intensities requires a source distance to determine the subtended solid angle.

358:  DS1 adopted a distance of 345pc; DS2, 250pc. Our working model adopts 250pc (\S3.), so

359:  to produce a common standard of comparison we divide the first data set intensities by $(345/250)^2=1.9$.

360:  Separate tests show that our results remain essentially unchanged if we adopt a distance of 345pc,

361:  normalize observed intensities to that distance, and adopt ${\dot{M}}=1.0{\times}10^{-8}M_{\odot}~{\rm yr}^{-1}$.

362: 

363: 

364: Production of a MEM eclipse map, which determines intensities pixel by pixel, permits segregation of pixels 

365: into isolated regions on the accretion disk.

366: Thus, DS1 divides the accretion disk into concentric rings, each ring further subdivided into a 'back', a 'front',

367: and a 'stream'.	Figure~2a illustrates the DS1 divisions. From \S2, the observational data include

368: both the PRISM and G160L as separate spectral sources. Consequently, it was necessary to introduce identifiers for

369: the source of the data used as well as the position of the ring subdivision.

370:   Thus, in DS1,

371:  identifier `prism.center' refers to the radial range 0.000$r_{\rm L1}$-0.075$r_{\rm L1}$, while the identifier

372:  `prism.back.20' refers to the azimuthal range $90{\arcdeg}$ to $270{\arcdeg}$ on the annulus

373:  extending from $0.175r_{\rm L1}$ to $0.225r_{\rm L1}$. Table~5 lists the subdivisions of the accretion disk

374:  and may be compared with Figure~2a. DS2 follows a similar but coarser scheme of geometric resolution; its 

375:  subdivisions also are listed

376:  in Table~5.

377:  

378:  The PRISM spectra were divided into 127 passbands and a separate MEM map was constructed for each passband.

379:  For each ring subdivision, concatenation of the 127 passband intensities constituted an extremely low resolution

380:  spectrum. It is that 'spectrum', for a given ring subdivision, that we fit with our model synthetic spectrum.

381:  Since the data being fit consist of intensities, it is necessary to use the corresponding intensity output from 

382:  SYNSPEC as discussed

383:  in \S3.1. 

384:  

385:  The G160L spectra were divided into 59 passbands and

386:  processed as with the PRISM data. DS2 used low-resolution optical wavelength spectra and followed a similar

387:  basic procedure.  

388:  Note that DS2 are corrected for extinction in the Earth's atmosphere

389:  while DS1 requires no such correction.

390:  

391:  Figure~2b illustrates the BINSYN model and the 44 annuli. Software keys isolate annuli subdivisions

392:  ('front', 'back','stream') for comparison with DS1. 

393:  Given the Table~5 radial ranges, in each comparison we select an annulus from Table~4 that is centered 

394:  on the DS1 or DS2 annulus in question.

395:  

396:  The \S3 system parameters and the $\dot{M}$ from Figure~1 set the model scale and the

397:  radiation properties of the individual photospheric segments in physical units. Production of a system 

398:  synthetic spectrum

399:  involves integration over the entire (projected) system model. Software keys restrict the contributing

400:  segments to corresponding DS1 or DS2 regions. The final step is division of the integrated model intensity

401:  by the contributing model area. There are no adjustable parameters to apply to the model in a comparison

402:  with the DS1 or DS2 data. We stress that the DS1 and DS2 data are not directly observed quantities but

403:  result from MEM maps which require an assumed distance for expression in physical units.

404: 

405:  Both data sets provide intensities for a central region that includes the WD; we have found that the

406:  most consistent overall choice for the WD $T_{\rm eff}$ is $\approx20,000$K. Figure~3 compares our model

407:  with DS2 region `A', Table~5.

408:  The error bars are an average from the 

409:  \citet{ru1994} Table~1. 

410: The reasonably close agreement between our model and

411:  the region 'A' data is gratifying. 

412:  Based on the error bars, the discrepancy shortward of

413:  5000\AA~probably is real (i.e., not a data artifact such as, e.g., incomplete correction for atmospheric

414:  extinction at a 2$\sigma$ level) but

415:  we remain

416:  cautious in interpreting the discrepancy because of the results for Figure~4a and Figure~4b, discussed

417:  below. The discrepancy could be explained by an opacity source not included in our model.

418: 

419:  Figure~4a compares a 40,000K WD model (upper synthetic spectrum), a 20,000K WD model (middle synthetic spectrum), 

420:  and a zero temperature model (lower synthetic spectrum) with the G160L 

421:  and prism data. The synthetic spectrum for the accretion disk contribution corresponds to BINSYN annuli

422:  1-6. Minor corrections, not discussed here, 

423:  would raise the 40,000K synthetic

424:  spectrum close to the peaks of the observed spectra.

425: 

426:  BA1998

427:  note that there was a $70\%$ change in $\dot{M}$ between August 1994 (G160L) and November 1994 (prism)

428:  in the sense that the disk intensities are larger for the prism data. 

429:  We empirically found that the

430:  G160L spectra accord well with the prism spectra if the G160L intensities are divided by 

431:  $1.2{\times}10^{-3}$ (and used in the plots) rather than 

432:  the $1.9{\times}10^{-3}$ factor used with the prism spectra.

433: 

434:  The 40,000K WD contributes substantially to the upper synthetic spectrum 

435:  while the inner accretion disk annuli are the major contributors to the 20,000K WD model; the data

436:  do not discriminate the latter model from a zero temperature WD model. 

437:  (We produced the system synthetic 

438:  spectrum for a 

439:  zero temperature WD

440:  by suppressing the WD contribution to the integrated intensity but left the total

441:  projected area unchanged.)

442:  Note that the

443:  peak intensity for the 20,000K WD model is at an ordinate value of 3.5, substantially discrepant from 

444:  the peak observed intensity.

445:  The Figure~4a data are in conflict with the Figure~3 data. The WD $T_{\rm eff}$ is one of the

446:  principal issues to be addressed and the Figure~4a data (plus the Figure~4b data: see below) constitute

447:  the only direct evidence for a 40,000K WD while the remaining data, discussed below, provide inconclusive

448:  (\S6.) support for a 20,000K WD.

449:  

450:  Figure~4b shows the continuation of 

451:  Figure~4a to longer wavelengths. Note the different spectral gradient at wavelengths longward of about 

452:  5500\AA~but possibly extending to shorter wavelengths. 

453:  BA1998 describe a dropoff in the prism spectrograph response in that region and beyond while 

454:  the DS2

455:  data set (Figure~3) agrees with our model; 

456:  the prism data appear to have a calibration issue (BA1998) that is insignificant at the short wavelength end

457:  but increases to longer wavelengths.

458: 

459:  The Figure~4a error bars are small and the DS1 fractional errors gradually

460:  increase as the measured intensity drops in successive figures; to avoid excessive clutter we 

461:  provide DS1 error bars only for

462:  Figure~4a and Figure~4j.

463:  Neither synthetic spectrum (the 20,000K case or the 40,000K case in Figure~4a) models the observed 

464:  deep absorption features of Si II ${\lambda}1300$;

465:  Si IV ${\lambda}1394,1403$; C IV ${\lambda}1548,1552$; He II ${\lambda}1640$; N V ${\lambda}1718$;

466:  Si III ${\lambda}1892$ and Mg II ${\lambda}2800$. It is noteworthy that the C IV ${\lambda}1548,1552$

467:  doublet is in emission; \citet{knigge1997} show that there is a wind

468:  associated with the emission feature and that there is an underlying slow-moving `chromosphere' that can produce

469:  narrow absorption reversals.

470:  The question arises whether a change in the physical conditions in the system, in the two-year interval

471:  between the Figure~3 data (obtained in 1992) and the Figure~4a data (obtained in 1994), could explain 

472:  the discrepancy between the two figures.

473:  There is evidence for a high temperature source located near where

474:  a transition layer would be expected. The evidence includes a hard X-ray source \citep{pratt2004},

475:  an eclipsed flickering source \citep{bruch2000}, a 29-second oscillation source that is eclipsed 

476:  \citep{knigge1998b}, and a source of excess radiation shortward of 965\AA~(see below). 

477:  Further discussion of the WD $T_{\rm eff}$	and the hot source is in \S6.

478: 

479:  Figure~4c shows the DS1 data for Table~5 radial designation 0.10.

480:  In this and subsequent plots the heavy grey line is the 'front' region, the light grey line is the

481:  'back' region and the light continuous line is the 'stream'. The heavy continuous line is the model

482:  synthetic spectrum. Note that the synthetic spectrum fits the DS1 data very well near 4000\AA. The

483:  'back' shows a higher intensity than the 'front' and the synthetic spectrum lies between them. At

484:  this annulus the 'stream' is in fair agreement with the 'front'.

485: 

486:  Baptista et al. discuss the lower flux levels in the 

487:  `front' annulus regions as compared with the `back'. They propose absorption due to a large

488:  number of blended lines of FeII similar to the `iron curtain', hereafter IC,

489:  invoked by \citet{hor1994}, hereafter HO1994. The disk $T_{\rm eff}$ at this radius is about 21,000K. 

490:  The IC calculated by HO1994 for

491:  OY Car (their Figure~8) produced absorption shortward of the Balmer discontinuity and 

492:  increased the depth of the Balmer jump. Thus, a similar mechanism is credible to explain

493:  the `front' discrepancy from the model synthetic spectrum.

494:  There is

495:  excess flux in the `back' Balmer continuum, and there is no Balmer jump.

496:  \citet{knigge1998a} propose H recombination emission from an accretion disk chromosphere (ADC) as a

497:  mechanism to fill in the Balmer discontinuity. Their ADC is at the base of a biconical wind and

498:  has a vertical height of order $10R_{\rm wd}$ to provide an emission measure adequate to fill

499:  in the Balmer jump. 

500:  Note the slight difference between the amount of absorption for the `stream'

501:  and the 'front`.  The ADC, the IC, and the biconical wind probably are part of a single structure.

502:  The IC absorbing material most likely is located above the outer part of the accretion disk and is

503:  seen projected on the inner annuli where most of the radiative flux is produced. 

504: 

505:  The problem with the prism data calibration is apparent. If it were possible to adjust the synthetic spectrum

506:  fit empirically, the wavelength at which dropoff starts could be moved to a longer wavelength from

507:  around 5000\AA. There are no adjustable parameters; it would otherwise make sense to truncate

508:  the prism spectrum at some appropriate wavelength but the fit changes from figure to figure so there is no

509:  good way to choose a truncation wavelength. These plots, including their agreement with the

510:  DS2 data (see below), serve as documentation of the prism

511:  calibration problem.

512: 

513:  Figure~4d compares the model with prism data for radial designation 0.15. 

514:  The dotted line is the data for DS2 region `B'. 

515:  Note the

516:  agreement of the DS2 profile with the model spectrum. As in Figure~4c, the model spectrum,

517:  shortward of the Balmer limit, is intermediate between the `front' and `back' and 

518:  is in excellent agreement with the `stream'. Ignoring the unmodeled emission lines, the `front' Balmer

519:  jump agrees with the synthetic spectrum. The two mechanisms of Figure~4c (IC and recombination emission)

520:  also may explain the

521:  residuals here.

522: 

523:  Figure~4e compares the model with prism data for radial designation 0.20.

524:  As in Figure~4d, shortward of the Balmer limit, the model spectrum is intermediate

525:  between the 'front' and 'back' but the 'stream' now lies closer to the 'back'.

526:  

527:  Figure~4f shows data for radial designation 0.25. 

528:  The dotted line is the DS2

529:  annulus `C' spectrum.

530:  In contrast to Figure~4d, the DS2 data indicates a higher intensity than the

531:  corresponding DS1 data, while the model spectrum lies half way between the two data plots.

532:  A reduction in the model $T_{\rm eff}$ from the 11,307K of Table~4 annulus \#~19 to 10,898K

533:  produces a close fit to the DS1 data longward of 4000\AA~and a fairly close fit shortward, 

534:  consistent with a local

535:  departure from the standard model; a corresponding but smaller increase of the model $T_{\rm eff}$

536:  produces a close fit to the DS2 data. 

537: 

538:  Figure~4g begins regions in the outer half of the accretion disk. The `stream' here

539:  agrees well with the `back', but departs increasingly in the successive regions. In Figure~4g there still

540:  is good agreement between the `stream' and 'back'.  The Table~4 annulus \# 23 radial position fits that of

541:  the observational data. The synthetic spectrum, with $T_{\rm eff}=9855$K, fits the `back' and `stream'

542:  well shortward of 4000\AA~but has too large intensity on the longward side. As in Figure~4f, a reduction

543:  of the model annulus $T_{\rm eff}$ of a few hundred Kelvins produces a close fit to the DS1 data

544:  longward of 4000\AA~but interpolates half way between the `front' and `back' on the shortward side.

545:  

546:  Figure~4h presents DS1 data for radial designation 0.40. The `stream'

547:  intensity now exceeds that from the `back' longward of 4000\AA~but agrees shortward of there. 

548:  The synthetic spectrum is in good agreement with the prism data shortward of 4000\AA~and continues

549:  agreement with the `stream' data to 5000\AA. The `front', `back', and `stream' now begin to show

550:  differences over much of the spectral region covered. 

551: 

552:  The dotted curve in Figure~4h is DS2 annulus `D'. 

553:  It would compare with prism `.35' data, half way between Figure~4g

554:  and Figure~4h, if those data were available.

555:  The discrepancy, in part, is due to observational data originating

556:  over a finite annular width; the hotter inner edge provides an enhanced contribution.

557:  Annulus `D' closely agrees with the DS1 data of Figure~4g shortward of 5000\AA~but disagrees with the

558:  slope of the Figure~4g synthetic spectrum, more closely fitting the Figure~4h synthetic spectrum

559:  SED while displaced to a larger intensity. Annulus `D' could be plotted either in Figure~4g or

560:  Figure~4h; for mimimum congestion we have chosen Figure~4h.

561: 

562:  Figure~4i plots DS1 data for radial region 0.50. 

563:  The corresponding

564:  Table~4 annulus is \# 37 with standard model $T_{\rm eff}=6985$K (inner edge), represented by the 

565:  lower synthetic spectrum.

566:  DS2 annulus `E' (the lower dotted line), 

567:  fits the lower 

568:  synthetic spectrum well. 

569:  The annulus \#~37 segment $0\arcdeg$ to $90\arcdeg$ is

570:  set to $T_{\rm eff}=7500$K and is represented by the upper synthetic spectrum. 

571:  The upper dotted line is DS2 annulus `F', corresponding to the same radial region as

572:  annulus `E' but azimuth $0\arcdeg$ to $90\arcdeg$ (the 'stream'). Since annulus `F' covers a much larger radial

573:  region than the upper synthetic spectrum we do not regard the difference as serious.

574:  The (lower) standard model synthetic spectrum agrees well with the `back'; its representative

575:  point, at log~${\rm r}/{\rm r_{L1}}=-0.3$, Figure~1, falls on the dotted line below the observed data points.

576:  The higher temperature annular segment falls on the Figure~1 heavy continuous curve passing 

577:  through the region of

578:  observed points.

579: 

580:  Figure~4j shows DS1 data for radial region 0.60. The corresponding Table~4 annulus

581:  is \# 44, $T_{\rm eff}=6158$K (inner edge). The lower heavy line is that model. To represent the `stream' we have 

582:  reset the annulus \# 44 $T_{\rm eff}$, between azimuth $0\arcdeg$ and $90\arcdeg$, to 8600K. The

583:  upper synthetic spectrum represents that annular segment. 

584:  The DS1 `stream' and `back' agree shortward

585:  of 4000\AA~but the very large `stream' Balmer jump is larger than the synthetic spectrum represents.

586:  The differences between the `front' and `back' are close to their errors and we do not regard

587:  their differences or their departure from the synthetic spectrum as significant.

588:  Note the change in ordinate scales from Figure~4a to Figure~4j; the intensity values change by roughly

589:  a factor 50.

590: 

591:  We have interpolated $T_{\rm eff}$ values for the azimuth $0\arcdeg$ to $90\arcdeg$ segments at

592:  annuli between \# 30 and \# 44 where the `stream' differs significantly from the `back'. This completes

593:  our accretion disk model and prepares a comparison with other data sets.

594: 

595: 

596: \subsection{Light curve simulations}

597: 

598:  We simulate $UBV$ light curves via synthetic photometry. We calculated synthetic system spectra at 81

599:  orbital phases, distributed to cover the variable parts of the light curve with adequate phase resolution.

600:  The synthetic spectra were based on the intensity version of SYNSPEC, which provides automatic

601:  wavelength-dependent correction for limb darkening, but converted to flux-based spectra for the final output. 

602: Each synthetic spectrum was weighted by the pass bands for the V,B, and U standard filters and the products

603: integrated. The integrated flux in V,B, and U, normalized to the flux maxima, as function of phase constitute the

604: theoretical light curves. 

605: 

606:  The software represents the bright spot as a rim section of elevated $T_{\rm eff}$ \citep{wood1986,wood1989};

607:  this is a modification of the (otherwise) isothermal rim (\S3.1).

608:  The bright spot covers the full rim height; the structure of our present program does not allow

609:  a variable rim thickness with azimuth.

610:  The rim semi-thickness, $H$, Table~2, follows from the standard model \citep[eq.5.39]{fkr92}.

611:  It was necessary to assign the region of elevated temperature to an extended azimuth

612:  region on the rim, in agreement with \citet{smak1994a}, to fit the \citet{wh1954} observations 

613:  (see Table~6 for the spot parameters).

614:  Hydrodynamical 2D models of the stream

615:  impact \citet{rozy1985,rozy1988} identify two shock waves: (1) a shock on a plane perpendicular to the

616:  orbital plane, roughly bisecting the

617:  angle between the stream and the rim and terminating at the upstream edge of the stream, and (2) a 

618:  shock slightly more inclined to the stream and extending far into the disk. Although the simulation is

619:  2D, \citet{rozy1985} states that a bow shock will develop, prospectively leading to vertical expansion

620:  upstream. \citet{livio1986} and \citet{ar1998} perform a 3D simulation and find that material from the 

621:  stream flows over

622:  the disk if cooling is efficient, applicable to low $\dot{M}$ cases, and is more like an explosion in

623:  high $\dot{M}$ cases, leading to a bulge extending along the disk rim. 

624:  The inclined shock wave plane suggests that a more complex model 

625:  would provide a better physical representation of the bright

626:  spot than adopted in this study. HO1994 developed a model of this type in their study of

627:  OY Car, in contrast to the `painted on' rim model of \citet{wood1989} for the same star. 

628: 

629:  Figure~5 shows the fit to the $V$ light curve, Figure~6 shows the fit to the $B$ light curve, and

630:  Figure~7 shows the fit to the $U$ light curve with a depth residual. A $U$ eclipse depth residual,

631:  differing from the

632:  good $V$ and $B$ fits, is suggestively familiar from the binary star literature; 

633:  it results from the poor representation of

634:  stellar SEDs by a black body over the Balmer discontinuity. That explanation is not possible here

635:  since the synthetic spectra simulate the Balmer discontinuity. The far edge of the accretion disk remains

636:  uneclipsed and it is this region that shows extra unmodeled light shortward of the Balmer discontinuity (Figure~4).

637:  We suggest that it is this unmodeled, uneclipsed light that reduces the observed depth of U eclipse

638:  below the model.

639:  These light curves adopt a 20,000K WD as

640:  discussed in the previous section.	The eclipse depths are more strongly correlated with orbital inclination

641:  than WD $T_{\rm eff}$

642:  and so are not useful in distinguishing between a 20,000K WD and a 40,000K WD. 

643:  We call attention to the downward trend in the Figure~6 residuals between orbital phases 0.2 and 0.8.

644:  \S6 discusses this feature in terms of a vertically extended rim.

645:  It is of interest that the model provides a good light curve fit over the range of dates from April~17, 1953

646:  ($B$)

647:  to June~13, 1953 ($U$).

648: 

649:  We have tested the sensitivity of our model to our choice of $R_{\rm cutoff}=0.6R_{\rm L1}$ by setting

650:  $R_{\rm cutoff}=0.7R_{\rm L1}$ and recalculating the model light curves. The eclipse widths remain the

651:  same to visual detectability limits while the eclipse depths become systematically smaller by small

652:  amounts; the $U$ calculated depth now fits the observations precisely while the $V$ and $B$ calculated depths are

653:  slightly too shallow. 

654:  

655:  \section{SED fits to observed spectra}

656: 

657:  We apply the model developed in previous sections,  including the rim 

658:  bright spot from the immediately

659:  previous section, hereafter the \S4. model, to represent observed spectra. The SYNSPEC 

660:  synthetic spectra

661:  for the annuli of Table~3 are produced in both an intensity format and a flux-based format; the latter format

662:  requires adoption of a wavelength-independent limb darkening coefficient.

663:  The intensity simulations of \S4 required the intensity-based synthetic spectra. 

664:  We used the intensity-based synthetic

665:  spectra, which automatically correct for wavelength-dependent limb darkening, and converted the

666:  output to flux units for comparison with the observed spectra which are tabulated in physical flux units.

667: 

668:  \subsection{SED fits to $FUSE$ and $FOS$ spectra}

669: 

670:  The $FOS$ spectra have already been used by BA1998

671:  to produce the MEM data modeled in Figure~4.

672:  Our objective in this section is to combine the $FOS$ and $FUSE$ spectra and

673:  model those observed spectra. The $FOS$ and $FUSE$ spectra have exposure times listed in Table~1. 

674:  $FUSE$ orbit03 exposure starts at orbital phase 

675:  $0.4333$ and ends at $0.5117$. $FUSE$ orbit04 exposure starts at orbital

676:  phase $0.7863$ and ends at orbital phase $0.8750$.	Both of the $FUSE$ exposures

677:  start and end outside eclipse. 

678:  The first G160L exposure, data set Y2AH0201T, starts at

679:  orbital phase $-0.090$ and ends at orbital phase $0.130$. The second G160L exposure, data set

680:  Y2AH0401T, starts at orbital phase $-0.063$ and ends at orbital phase $0.156$.

681:  As BA1998 indicate, in each case 

682:  the $FOS$ data set consists of 691 spectra produced in `rapid readout' mode.

683:  In each $FOS$ case, our spectrum is the sum of the exposures in the data set. 

684: 

685:  To simulate the sum of the exposures for the $FOS$ spectra, we calculated 33 synthetic system spectra

686:  equally spaced in phase between phase -0.063 and 0.156, corresponding to the Y2AH0401T data set.

687:  These spectra all used a 20,000K model for the WD, specifically for 

688:  $\dot{M}=5.0{\times}10^{-9}M_{\odot}~{\rm yr}^{-1}$. BA1998 note that there was a $70\%$ reduction

689:  in $\dot{M}$ from the DS1 observations to the $FOS$ observations. Experiment shows that a change

690:  like this does not make a large change in the calculated spectral gradient; the major change is in the flux

691:  level. Rather than calculate a complete new set of TLUSTY annuli and a corresponding new BINSYN

692:  model, we preserve the adopted $\dot{M}$ and subsume the effect of the $\dot{M}$ change in the

693:  normalization factor to superpose the model on the observed spectra (see below).

694:  We summed the spectra and divided by 33 to produce an average spectrum through eclipse. Comparison

695:  of the average spectrum with one that is outside eclipse shows that the average spectrum closely

696:  fits the outside-eclipse profile but has a flux level that is $88\%$ of the outside-eclipse

697:  spectrum. We use this average spectrum for comparison with both of the $FOS$ spectra. 

698:  

699:  Figure~8 compares two synthetic spectra with the YAH0201T spectrum. The lower synthetic spectrum is the

700:  average spectrum described above and the upper synthetic spectrum is the 40,000K model. It is apparent

701:  that the synthetic spectrum departs significantly from the observed spectrum. 

702:  In the following

703:  section we will find that the same is true of the $IUE$ spectra and that the SED is appreciably variable,

704:  temporally, with $T_{\rm eff}$ profiles that differ from the standard model.  

705:  A fair overall fit can be achieved 

706:  shortward of 2100\AA~with a normalizing factor of

707:  $7.5{\times}10^{41}$, corresponding to a distance of 281pc. 

708:  This distance is too large because of the failure to allow for the reduced $\dot{M}$ drescribed above.

709:  A correction moves the calculated distance toward the adopted 250pc.

710:  The 40,000K WD model shown (calculated for an outside-eclipse phase), with the same normalizing factor, 

711:  lies well above the observed spectrum and

712:  would require a larger normalizing factor to fit the observed spectrum. Reducing to $88\%$ of the calculated

713:  flux to allow for effects of eclipse still leaves a large discrepancy.

714:   

715:  A completely self-consistent loop for the \S4. model would determine a normalizing factor that

716:  reproduces	the assumed initial model distance of 250pc. Note that this analysis, while adopting values

717:  of $M_{\rm wd}$ and $\dot{M}$, leaves the distance as a parameter to be determined. By contrast, in Figure~4, the

718:  data to be fit involve an assumed distance but there is no adjustable parameter in the comparison

719:  with the model. 

720: 

721:  Figure~9 presents the FUV spectrum for the Figure~8 configuration. The $FUSE$ spectrum is outside eclipse so

722:  the synthetic spectrum for the 20,000K model also represents that condition, as does the 40,000K synthetic

723:  spectrum. 

724:  The \S4. model 

725:  (lower synthetic spectrum) fits

726:  the $FUSE$ spectrum approximately but does not reproduce the very large number of fairly deep

727:  absorption features. The absorption features in the 1120\AA~to 1150\AA~region of

728:  the synthetic spectrum are replicated with much larger amplitude in the $FUSE$ spectrum.

729:  FR2003 discuss the $FUSE$ spectra in detail.

730:  Reddening has a strong effect; a change from E(B-V)=0.00 to E(B-V)=0.01

731:  raises the $FUSE$ spectrum by $17\%$ without a detectable change in the slope.

732:  The reddening-corrected spectrum fits the synthetic spectrum better but, as noted below, the $FUSE$ spectrum

733:  shows phase-wise variation that makes the value of the fit questionable. We do not consider the improved

734:  fit necessarily as support for a value of E(B-V)=0.01. 

735:  Note that the emission excess shortward of 965\AA, described by

736:  FR2003, is clearly present. Also note the blueward displacement of the Ly$\beta$ and Ly$\gamma$

737:  absorption features; the model approximately reproduces their depths. The correponding Doppler

738:  shift is ${\sim}3000~{\rm km}~{\rm s}^{-1}$.

739: 

740:  Figure~10 combines the G160L spectrum from the Y2AH0401T set with the $FUSE$ orbit04 spectrum

741:  (the phase of the $FUSE$ spectrum is unrelated to the phase of the G160L spectrum;

742:  they have been plotted together for economy--otherwise separate plots would be required).

743:  Because of the same total phase range for the two $FOS$ spectra and the same number of contributing

744:  individual spectra they

745:  would be expected to be essentially identical.

746:  Yet the G160L spectrum

747:  shows an overall flux reduction,

748:  is substantially fainter in the 1200\AA~to 1500\AA~region, but now shows an excellent fit longward

749:  of 2000\AA. The accretion disk SED has changed between the times of the two observation sets.

750: 

751:  Figure~11, which shows the FUV part of Figure~10, poses a problem.	As with Figure~9, the 20,000K

752:  model represents an outside-eclipse phase. The $FUSE$ spectrum is outside eclipse

753:  so it would be expected

754:  to show little difference from Figure~9, yet the difference is striking. Not only is the flux level

755:  lower, there appear to be absorption bands that are not prominent in Figure~9.

756:  Note the broader, deeper and more complex Ly$\beta$ and Ly$\gamma$ features, but the emission excess

757:  shortward of 965\AA~is unaffected.	The $FUSE$ spectrum shows variation which may be both/either

758:  temporal or phase-dependent.

759: 

760: 

761: 

762:  \subsection{SED fits to $IUE$ spectra}

763: 

764:  The $IUE$ archive lists 31 LWP and LWR	spectra, and 37 SWP spectra, mostly

765:  observed in 1980 and mostly taken in pairs including a LWR exposure and a SWP exposure.

766:  We used the BA1995

767:  ephemeris  $T_{\rm min}= {\rm HJD}~2443904.87872+0.196671278E$ to calculate orbital

768:  phases. 

769:  The UX UMa orbital period is short enough, and the $IUE$ exposure time long enough, that the second 

770:  exposure of a pair occurred at an

771:  orbital phase differing appreciably from the first. For example, in the pair SWP10128+LWR08799

772:  the first exposure started at orbital phase $0.2887$ while the second started at 

773:  $0.4069$. In cases

774:  where both exposures are outside eclipse 

775:  the phase displacement between exposures makes an 

776:  undetectable change in the synthetic spectra; consequently we combine the observed spectra of a

777:  pair for analysis. Separate tests with outside-eclipse synthetic spectra show nearly undetectable

778:  phase-wise variation in the SED even though the model includes the rim bright spot.

779:  The observed spectra do show cycle to cycle changes 

780:  and there are year to year

781:  changes that are clearly apparent \citep{holm1982} (see below).

782:  We find that our model with a 20,000K WD provides an appreciably better fit, in all cases tested,

783:  than a 40,000K WD. A 40,000K WD produces a too-steep spectral gradient (see below).

784:  

785:  Figure~12 presents a fit to SWP10371+LWR09051. The orbital phase at the start of the SWP exposure

786:  was $0.2867$ and the phase at the start of the LWR exposure was $0.1615$. Both 

787:  exposures were

788:  outside eclipse. The synthetic spectrum has been divided by $5.2{\times}10^{41}$ for this

789:  comparison, showing a reasonably close fit to the $IUE$ spectra except beyond 2600\AA;

790:  the corresponding distance is 234pc. This fit appears to provide support for our adopted distance of 250pc,

791:  but, as we see in the following discussion, the support is ambiguous.

792:  The upturn at 2600\AA~is not due to the secondary star since BINSYN includes an explicit model of the 

793:  secondary star in the simulation; this effect mimics the Figure~8 anomaly.

794:  Plots of the $IUE$ spectra and the G160L plus prism spectra show a substantially different flux level in

795:  the two cases, indicating a likely change in $\dot{M}$ between the times of observation.

796:  (The $IUE$ 1350\AA~continuum flux in Figure~12 is about 3.0 ordinate units	while the corresponding

797:  level in Figure~8 is about 2.2 ordinate units.)

798:  

799:  Figure~13 presents a fit to SWP10128+LWR08798. The orbital phase at the start of the SWP exposure

800:  was $0.2887$ and the phase at the start of the LWR exposure was $0.4069$.

801:  The synthetic spectrum has been divided by $5.2{\times}10^{41}$, as in Figure~12. The discrepancy

802:  beyond 2600\AA~in Figure~12 now afflicts Figure~13 beyond 1700\AA. A possible postulate to explain

803:  Figure~13 is a lower accretion disk temperature gradient. The required change is drastic: A 20,000K

804:  WD and a 12,000K isothermal accretion disk provide a good fit, with a normalizing factor of 

805:  $3.5{\times}10^{41}$, placing the system at a distance of 192pc. But the excess radiation above the 

806:  synthetic spectrum with the Figure~12 normalizing

807:  factor, nearly consistent with the adopted distance of UX UMa, indicates an actual {\it increase} in total 

808:  disk luminosity, contradicting the implication of the

809:  reduced normalizing divisor. 

810:  Thus, the isothermal accretion disk model fits

811:  the $IUE$ spectrum but does not provide a believable system model.

812: 

813:  Finally, Figure~14 presents a fit to SWP10677+LWR09388. This spectrum permits an excellent fit by a

814:  standard model, but with a mass transfer rate of $\dot{M}=3.0{\times}10^{-9}M_{\odot}~{\rm yr}^{-1}$;

815:  the corresponding distance is 165pc. The actual mass transfer rate in UX UMa may, at times, equal

816:  the value quoted, but if the accretion disk reaches equilibrium, the flux values will differ appreciably from the

817:  Figure~14 values.

818: 

819:  Our conclusions from simulation of these and other $IUE$ spectra are:

820:  (1) The Figure~8 anomaly occurs frequently but may start at shorter or longer wavelengths than 2000\AA.

821:  The source of the anomaly is unknown.

822:  (2) In some cases, the SED in a variable $\dot{M}$ system can be accurately fit by a nonstandard model

823:  (e.g., isothermal) that does not lead to a believable system model; the fit cannot be used even to

824:  constrain the disk luminosity.

825:  (3) As a caution to disk modelers, a good standard model fit in a system that shows spectrum variability

826:  cannot be taken at face value,

827:  even if the $\dot{M}$ indicates that the accretion disk is stable against outburst, unless there is independent

828:  evidence concerning the system distance.  

829: 

830: 

831: \section{Discussion}

832: 

833:  An important part of the system analysis for UX UMa depends on knowledge of its distance. 

834:  BA1995 calculate a value of 345pc with an estimated error

835:  of 34pc. The 345pc determination depends on fits in a color-magnitude diagram, with one wavelength at 1523\AA.

836:  The fits use a theoretical color-magnitude relation derived on assumed radiation characteristics

837:  of the source. If the source can be represented by a black body, the derived distance is 401pc, and if

838:  standard Kurucz model atmospheres, 312pc. We feel the residuals in the SED fits found in the present study are

839:  large enough that an assumed radiation characteristic for the source should be treated with caution, and that

840:  the estimated error in the 345pc determination could be larger.

841:  Other distance determinations are in \S3. The \citet{knigge2006} method is an important improvement on the 

842:  \citet{bailey1981} relation;

843:  as cited in \S3., this method leads to a distance of 215pc if the secondary provides all of the system K-flux and

844:  376pc if the secondary provides 1/3 of the system flux. Based on the parameters of our \S4. model, and adopting a

845:  secondary $T_{\rm eff}=3575$K \citep[Table~3]{knigge2006} the secondary provides 0.48 of the system flux at

846:  2.2$\mu$, leading to a calculated distance of 312pc. 

847:  Consider the sensitivity of the calculated distance to variation of system parameters.

848:  The flux ratio secondary/(secondary+disk) is sensitive to 

849:  the accretion disk flux which in

850:  turn depends on $\dot{M}$. (The contribution of the adopted 20,000K WD is only a few tenths of a percent at 2.2$\mu$.)  

851:  From Figure~1 we estimate that the scatter of the observational data constrains $\dot{M}$ within a factor of 

852:  about 2 for the adopted system distance

853:  of 250pc. 

854:  Increasing $\dot{M}$ to

855:  $1.0{\times}10^{-8}M_{\odot}~{\rm yr}^{-1}$ produced a new calculated distance of 332pc. 

856:  We increased $M_{\rm wd}$ by $10\%$, while maintaining $M_{\rm sec}$ fixed, to test the calculated distance

857:  sensitivity to $M_{\rm wd}$ variation. The new distance was 323pc.

858:  Variation in the

859:  observed K magnitude of a few times 0.01 produces variation of only a few parsecs in the calculated distance.

860:  We propose a distance of $312\pm30$pc as the best currently 

861:  available distance determination to UX UMa and note the accordance

862:  with the value found by BA1995 using a fit to Kurucz model atmospheres.

863: 

864: This study finds that a standard model 

865: ${\dot{M}}=5.0{\times}10^{-9}M_{\odot}~{\rm yr}^{-1}$ accretion disk surrounding a 20,000K (but see below),

866: $0.47M_{\odot}$ WD at a distance of 250pc provides a model that reasonably fits spectral intensity 

867: data (Figure~4) and spectral flux

868: data (Figure~8 through Figure~11). 

869: All of the DS1 and DS2 data can be equally well represented if the distance is 345pc and

870: ${\dot{M}}=1.0{\times}10^{-8}M_{\odot}~{\rm yr}^{-1}$. 

871: In particular, the observational data of Figure~1 depend only on the adopted distance. 

872: Interpolating between the 250pc and 345pc calibrations, for our preferred distance of 312pc,

873: the corresponding $\dot{M}=8{\times}10^{-9}M_{\odot}~{\rm yr}^{-1}$. We take this value to be our final

874: result for the mass transfer rate. In principle it would be possible to iterate our solution, starting over

875: with the new distance and $\dot{M}$ determination but, based on our study of the

876: ${\dot{M}}=1.0{\times}10^{-8}M_{\odot}~{\rm yr}^{-1}$ case at 345pc, we believe there would be no

877: improvement in any of the plots presented in this paper. 

878: 

879: \citet{puebla2007} (hereafter P2007) use a separate method to study accretion rates and, for UX~UMa 

880: (their Table~2) find

881: ${\dot{M}}=1.7{\times}10^{-8}M_{\odot}~{\rm yr}^{-1}$ from black body fits and

882: ${\dot{M}}=1.4{\times}10^{-8}M_{\odot}~{\rm yr}^{-1}$ for their model accretion disk fits, adopting a

883: distance to UX UMa of 340pc and $M_{\rm wd}/M_{\odot}$ in the range 0.4-0.8. This result is in approximate

884: agreement with our results. 

885: In their Figure~6 P2007 fit $IUE$ spectra of

886: UX~UMa with two models. Their second model, for $M_{\rm wd}/M_{\odot}=0.4$, finds

887: ${\dot{M}}=5.5{\times}10^{-9}M_{\odot}~{\rm yr}^{-1}$, in close agreement with our model.

888: P2007 parameterize the WD contribution with ${\zeta}=f_{\rm wd}/f_{\rm disk}$ and define

889: a ``disk-dominated'' system as one in which ${\zeta}<0.1$, where the flux values are integrated contributions

890: from 1500\AA~to 3250\AA. 

891: We would prefer

892: to define ``disk-dominated'' in terms of an integration extending to 950\AA~since a large part of the

893: flux from a hot WD can occur shortward of 1500\AA.

894: Because of the ambiguity of the WD $T_{\rm eff}$ in UX~UMa

895: (see below), we

896: do not estimate a value for $\zeta$ except to note that, from Figure~4a, the accretion disk supplies well

897: over $90\%$ of the system flux if the WD $T_{\rm eff}=20,000$K (compare with the zero temperature WD where

898: the accretion disk supplies $100\%$ of the system flux, excluding the secondary). 

899: 

900: Several studies (\S3) support the adopted $M_{\rm sec}=0.47M_{\odot}$. The adoption of $q=1.0$ places the

901: system at the boundary of instability against dynamical scale mass transfer and a $M_{\rm wd}$ larger than 

902: $0.47M_{\odot}$ seems likely. \citet{bap1995} state that adoption of the \citet{smak1994a} 

903: $M_{\rm wd}=0.70M_{\odot}$ increases the system distance from 345pc to 386pc; the effect of a smaller $q$

904: is to increase the size of the accretion disk and so make it more luminous. 

905: A WD mass greater than $0.47M_{\odot}$ would be smaller and would produce a deeper potential well so, for the same

906: $\dot{M}$, the accretion disk would be hotter and the spectral gradient would be steeper. Preservation of a fit to the

907: calibrated temperatures of Figure~1 would require either a reduction in $\dot{M}$ or a nonstandard model

908: temperature profile. 

909: 

910: The evidence concerning the WD $T_{\rm eff}$ is ambiguous. BINSYN requires specification of a WD $T_{\rm eff}$

911: and our adopted \S4. model includes a 20,000K WD

912: but, as Figure~4a shows, this model differs almost negligibly from a zero temperature WD; the hotter

913: inner annuli, Table~4, contributions dominate the system synthetic spectrum. 

914: A 40,000K or hotter WD, directly visible to

915: the observer, would be strongly

916: inconsistent with the Figure~3 intensity data while the SED data, Figure~8 through Figure~11, all are 

917: inconsistent with a

918: hot WD. The only data supporting a hot WD are Figure~4a and Figure~4b. 

919: Although the DS1 data plotted in Figure~4a and Figure~4b superficially support a 40,000K WD, we

920: find it more attractive to attribute the excess flux source to something like a boundary layer in common

921: with the other unmodeled hot sources described in connection with Figure~4a.

922: We argue that, in Figure~4a, 

923: the central region is seen through an absorbing layer, likely the ``transition region''

924: between the accretion disk and the fast wind \citep{knigge1997}, which produces the deep absorption features.

925: 

926: Studies of other cataclysmic variable systems generally support a WD $T_{\rm eff}$ hotter than 34,000K 

927: \citep{sion1999,knigge2000,sion2008} at the orbital period of UX UMa.

928: It is informative that the WD temperature determination for DW UMa obtained during a low state of

929: accretion when the WD was clearly visible \citep{knigge2000} were much higher than during the normal high

930: accretion state, indicating that the accretion disk can occult the WD. Since the orbital period of UX UMa 

931: is outside the range for systems that undergo low states, there is no opportunity to determine how much disk

932: occultation occurs.

933: \citet[Table~3]{sion1999} lists CV systems

934: for which absorbing curtains obscuring the WDs have been calculated. Note that our application would be more

935: complex since, in addition to obscuring the central region, we require obscuration of the `front' of 

936: particular annular regions, separate from the WD. 

937: 

938: As we have demonstrated, an accretion disk with constant $\dot{M}$ fits both the DS2 and DS1

939: observations, and this datum (a fixed, unchanging accretion disk) might

940: imply that a changed WD explains the difference between Figure~3 and Figure~4b. 

941: But if $\dot{M}$ has remained 

942: nearly constant we see no 

943: mechanism to heat the WD (and we do not believe a changed $\dot{M}$ could heat the WD from 20,000K to

944: 40,000K), while there are documented changes in the accretion disk	(Figure~8 to Figure~13)

945: and one such change might be formation of a hot emitting region between the times of the two data sets

946: (June 4, 6, and 7, 1992 for DS2 and Nov. 11, 1994 for DS1).

947: 

948: Our relatively simple bright spot model, \S4.1, represents the $UBV$ light curves of Figure~5 through

949: Figure~7, including the luminosity maximum just before eclipse. 

950: These simulations are the basis for our determination of the system orbital inclination. 

951: UX UMa occasionally shows

952: anomalous light curves that can not be represented by even a sophisticated model like that

953: of HO1994. An example is in Figure~15. Less extreme cases show a depressed light level from phase

954: 0.2 to 0.8, with a downward tilt to phase 0.8, followed by a rapid rise to light maximum at phase 0.9. These

955: features become more pronounced from $V$ to $B$ to $U$. An example is (our) Figure~6 or Figure~3 of \citet{jph1954}.

956:  \citet{mason1997} attribute this effect to a disk `bulge' upstream from the bright spot and tie it to a 

957:  similar effect in UV spectra. In Figure~15 we suggest that, except for a temporally enlarged `bulge' 

958:  or associated disturbance, the system would 

959:  show a brightness

960:  peak at orbital phase 0.9 with a brightness reduction to the light level seen at phase 0.15.

961:  The same `dips' are seen in LMXRBs, discussed by \citet{livio1993} and with the same proposed explanation.

962:  This subject is discussed further by \citet{knigge1998a} and FR2003. 

963:  This explanation for anomalous light curves differs from

964:  the \citet{smak1994b} proposal for circumdisk absorbing

965:  material. 

966:  

967: An important result of this study is that, starting with an adopted distance of 250pc to UX UMa, 

968: intensity-based observed spectra, flux-based observed spectra,

969: and photometric data all can be approximately represented by a standard model with 

970: ${\dot{M}}=5.0{\times}10^{-9}M_{\odot}~{\rm yr}^{-1}$ and with an elevated temperature in the accretion

971: disk quadrant that includes the stream impact region. Application of the \citet{knigge2006} method then

972: leads to an improved distance estimate of 312pc followed by a revised estimate of the (average) mass transfer rate

973: (at the times of the DS1 and DS2 data sets). In spite of this success, SED fits show departures from the

974: standard model

975: in agreement with BINSYN studies of

976: IX Vel \citep{linn2007} and QU Car \citep{linnell2008}

977: which also exhibit departures from a standard model.	

978:  We postpone attempting to model the IC and the recombination spectrum that fills in the Balmer 

979:  continuum for the

980:  `back' spectra, Figure~4c to Figure~4j, to a subsequent publication.  

981:   

982: We thank the referee for a prompt report; responding to it substantially improved this paper.

983: We

984: are grateful to Dr. Baptista for supplying data from BA1998 for this investigation. 

985: PG wishes to thank Mario Livio for his kind hospitality at the Space Telescope Science Institute

986: where part of thi work was carried out.

987: Support for this work was provided by NASA through grant number 

988: HST-AR-10657.01-A  to Villanova University (P. Godon) from the Space

989: Telescope Science Institute, which is operated by the Association of

990: Universities for Research in Astronomy, Incorporated, under NASA

991: contact NAS5-26555. 

992: PS is supported by HST grant GO-09724.06A.

993: 

994: This research was partly based on observations made with the NASA/ESA Hubble Space Telescope, 

995: obtained at the

996: Space Telescope Science Institute, which is operated by the Association of Universities 

997: for Research in Astronomy, Inc. under NASA contract NAS5-26555, and the NASA-CNES-CSA

998: {\it Far Ultraviolet Explorer}, which is operated for NASA by the Johns Hopkins University

999: under NASA contract NAS5-32985. 

1000:  

1001: 

1002: 

1003: %%% BIBLIOGRAPHY

1004: \begin{thebibliography}{}

1005: 

1006: \bibitem[Armitage \& Livio(1998)]{ar1998}

1007: Armitage,~P.~J., \& Livio,~M. 1998, \apj, 493, 898

1008: 

1009: \bibitem[Bailey(1981)]{bailey1981}

1010: Bailey,~J. 1981, \mnras, 197, 31

1011: 

1012: \bibitem[Baptista et al.(1995)]{bap1995}

1013: Baptista,~R.,~Horne,~K.,Hilditch,~R.~W.,Mason,~K.~O.,\& Drew,~J.~E. 1995, \apj, 448, 395  (BA1995)

1014: 

1015: \bibitem[Baptista et al.(1998)]{bap1998}

1016: Baptista,~R.,Horne,~K.,Wade,~R.~A.,Hubeny,~I.,Long,~K~S.,\& Rutten,~R.~G.~M. 1998, \mnras, 298, 1079 (BA1998)

1017: 

1018: \bibitem[Becker(1981)]{becker1981}

1019: Becker,~R.~H. 1981, \apj, 251, 626

1020: 

1021: \bibitem[Bruch(2000)]{bruch2000}

1022: Bruch,~A. 2000, \aap, 359, 998

1023: 

1024: \bibitem[Bruch \& Engel(1994)]{bruch1994}

1025: Bruch,~A., \& Engel,~A. 1994, \aaps, 104, 79

1026: 

1027: \bibitem[Frank et al.(1981)]{frank1981}

1028: Frank,~J., King,~A.~R., Sherrington,~M.~R., Jameson,~R.~F., \& Axon,~D.~J. 1981, \mnras, 195, 505

1029: 

1030: \bibitem[Frank, King, \& Raine(1992)]{fkr92} 

1031: Frank,~J., King,~A., 

1032: \& Raine,~D.\ 1992, Accretion Power in Astrophysics (Cambridge:\ Univ.Press)

1033: 

1034: \bibitem[Froning et al.(2003)]{froning2003}

1035: Froning,~C.~S., Long,~K.~S., \& Knigge,~C. 2003, \apj, 584, 433 (FR2003)

1036: 

1037: \bibitem[Holm, Panek, \& Schiffer(1982)]{holm1982}

1038: Holm,~A.~V., Panek,~R.~J., \& Schiffer,~F.~H.,III 1982, \apj, 252, L35

1039: 

1040: \bibitem[Horne et al.(1994)]{hor1994}

1041: Horne,~K., Marsh,~T.~R., Cheng,~F.-H., Hubeny,~I., \& Lanz,~T., 1994, \apj, 426, 294 (HO1994)

1042: 

1043: \bibitem[Hubeny(1988)]{h1988}

1044: Hubeny,~I. 1988, Comp. Phys. Comm., 52, 103

1045: 

1046: 

1047: \bibitem[Hubeny(1990)]{hubeny1990}

1048: Hubeny,~I. 1990, \apj, 351, 632

1049: 

1050: \bibitem[Hubeny, Lanz, \& Jeffery(1994)]{hlj1994}

1051: Hubeny,~I., Lanz,~T., \& Jeffery,~C.~S. 1994, in Newsletter on 

1052: Analysis of Astronomical Spectra No.~20, ed. C.~S. Jeffery (CCP7;St. Andrews: St. Andrews Univ.), 30

1053: 

1054: \bibitem[Hubeny \& Lanz(1995)]{hl1995}

1055: Hubeny,~I., \& Lanz,~T. 1995, \apj, 439, 875

1056: 

1057: \bibitem[Hubeny \& Hubeny(1998)]{hh1998}

1058: Hubeny,~I., \& Hubeny,~V. 1998, \apj, 505, 558

1059: 

1060: \bibitem[Johnson, Perkins, \& Hiltner(1954)]{jph1954}

1061: Johnson,~H.~L.,Perkins,~B., \& Hiltner,~W.~A. 1954, \apjs, 1, 91

1062: 

1063: \bibitem[King et al.(1983)]{king1983}

1064: King,~A.~R., Frank,~J., Jameson,~R.~F., \& Sherrington,~M.~R. 1983, \mnras, 203, 677

1065: 

1066: \bibitem[Knigge(2006)]{knigge2006}

1067: Knigge,~C., 2006, \mnras, 373, 484

1068: 

1069: \bibitem[Knigge \& Drew(1997)]{knigge1997}

1070: Knigge,~C., \& Drew,~J.~E. 1997, \apj, 486, 445

1071: 

1072: \bibitem[Knigge et al.(1998a)]{knigge1998a}

1073: Knigge,~C., Long,~K.~S., Wade,~R.~A., Baptista,~R., Horne,~K., Hubeny,~I., \& Rutten,~R.~G.~M.

1074: 1998, \apj, 499, 414

1075: 

1076: \bibitem[Knigge et al.(1998b)]{knigge1998b}

1077: Knigge,~C., Drake,~N., Long,~K.~S., Wade,~R.~A., Horne,~K., \& Baptista,~R. 1998, \apj, 499, 429

1078: 

1079: \bibitem[Knigge et al.(2000)]{knigge2000}

1080: Knigge,~C., Long,~K.~S., Hoard,~D.~W., Szkody,~P., \& Dhillon,~V.~S. 2000, \apj, 539, L49

1081: 

1082: \bibitem[Krzemi\'{n}ski \& Walker(1963)]{kw1963}

1083: Krzemi\'{n}ski,~W., \& Walker,~M.~F. 1963, \apj, 138, 146

1084: 

1085: \bibitem[Linnell \& Hubeny(1996)]{linnell1996}

1086: Linnell,~A.~P., \& Hubeny,~I. 1996, \apj, 471, 958

1087: 

1088: \bibitem[Linnell et al.(2007)]{linn2007}

1089: Linnell,~A.~P., Godon,~P., Hubeny.~I., Sion,~E.~M., \& Szkody,~P. 2007, \apj, 662, 1204

1090: 

1091: \bibitem[Linnell et al.(2008)]{linnell2008}

1092: Linnell,~A.~P., Godon,~P., Hubeny,~I., Sion,~E.~M., Szkody,~P., \& Barrett,~P. 2008, 

1093: \apj, 676, 1266 

1094: 

1095: \bibitem[Livio(1993)]{livio1993}

1096: Livio,~M. 1993, in Accretion Disks in Compact Stellar Systems 

1097: ed. J. Craig Wheeler (Singapore:World Scientific), p.243

1098: 

1099: \bibitem[Livio, Soker, \& Dgani(1986)]{livio1986}

1100: Livio,~M., Soker,~N., \& Dgani,~R. 1986, \apj, 305, 267

1101: 

1102: \bibitem[Mason et al.(1995)]{mason1995}

1103: Mason,~K.~O., Drew,~J.~E., C\'{o}rdova,~F.~A., Horne,~K., Hilditch,~R., Knigge,~C., Lanz,~T., 

1104: \& Meylan,~T. 1995, \mnras, 274, 271

1105: 

1106: \bibitem[Mason, Drew, \& Knigge(1997)]{mason1997}

1107: Mason,~K.~O., Drew,~J.~E., \& Knigge,~C. 1997, \mnras, 290, L23

1108: 

1109: \bibitem[Nather \& Robinson(1974)]{nr1974}

1110: Nather,~R.~E., \& Robinson,~E.~L. 1974, \apj, 190, 637

1111: 

1112: \bibitem[Osaki(1996)]{osa96}

1113: Osaki,~Y. 1996, \pasp, 108, 39

1114: 

1115: \bibitem[Panei, Althaos, \& Benvenuto(2000)]{p2000}

1116: Panei,~J.~A., Althaos,~L.~G., \& Benvenuto,~O.~G. 2000, \aap, 353, 970 

1117: 

1118: \bibitem[Pratt et al.(2004)]{pratt2004}

1119: Pratt,~G.~W., Mukai,~K., Hassall,~B.~J.~M., Naylor,~T., \& Wood,~J.~H. 2004, \mnras, 348, L49

1120: 

1121: \bibitem[Puebla et al.(2007)]{puebla2007}

1122: Puebla,~R., Diaz,~M., \& Hubeny,~I. 2007, \aj, 134, 1923 (P2007)

1123: 

1124: \bibitem[R\'{o}\.{z}yczka(1985)]{rozy1985}

1125: R\'{o}\.{z}yczka,~M. \& Schwarzenberg-Czerny,~A. 1987, Acta Astr., 37, 141

1126: 

1127: \bibitem[R\'{o}\.{z}yczka(1988)]{rozy1988}

1128: R\'{o}\.{z}yczka,~M. 1988, Acta Astr., 38, 175

1129: 

1130: \bibitem[Rutten et al.(1992)]{ru1992}

1131: Rutten,~R.~G.~M., van Paradijs,~J., \& Tinbergen,~J. 1992, \aap, 260, 213

1132: 

1133: \bibitem[Rutten et al.(1994)]{ru1994}

1134: Rutten,~R.~G.~M., Dhillon,~V.~S., Horne,~K., \& Kuulkers,~E. 1994, \aap, 283, 441

1135: 

1136: \bibitem[Schlegel, Honeycutt, \& Kaitchuck(1983)(]{sch1983}

1137: Schlegel,~E.~M., Honeycutt,~R.~K., \& Kaitchuck,~R.~H. 1983, \apjs, 53, 397

1138: 

1139: \bibitem[Shafter(1984)]{sh1984}

1140: Shafter,~A.~W. 1984, \aj, 89, 1555

1141: 

1142: \bibitem[Shakura \& Sunyaev(1973)]{ss1973}

1143: Shakura,~N.~I., \& Sunyaev,~R.~A. 1973, \aap, 24, 337

1144: 

1145: \bibitem[Sion(1999)]{sion1999}

1146: Sion,~E.~M. 1999, \pasp, 111, 532

1147: 

1148: \bibitem[Sion et al.(2008)]{sion2008}

1149: Sion,~E.~M., G\"{a}nsicke,~B.~T., Long,~K.~S., Szkody,~P., Knigge,~C., Hubeny,~I.,

1150: DeMartino,~D., \& Godon,~P. 2008, \apj, 681, 543 

1151: 

1152: \bibitem[Smak(1994a)]{smak1994a}

1153: Smak,~J. 1994a, Acta Astron., 44, 59

1154: 

1155: \bibitem[Smak(1994b)]{smak1994b}

1156: Smak,~J. 1994b, Acta Astron., 44, 257

1157: 

1158: \bibitem[Smak(2002)]{smak2002}

1159: Smak,~J.2002, Acta Astron., 52, 263

1160: 

1161: \bibitem[Walker \& Herbig(1954)]{wh1954}

1162: Walker,~M.~F., \& Herbig,~G.~H. 1954, \apj, 120, 278

1163: 

1164: \bibitem[Warner(1995)]{warner95}

1165: Warner,~B.\ 1995, Cataclysmic Variable Stars (Cambridge:\ Cambridge 

1166: University Press)

1167: 

1168: \bibitem[Wood et al.(1986)]{wood1986}

1169: Wood,~J.~H., Horne,~K., Berriman,~G., Wade,~R., O'Donoghue,~D., \& Warner,~B.

1170: 1986, 219, 629

1171: 

1172: \bibitem[Wood et al.(1989)]{wood1989}

1173: Wood,~J.~H., Horne,~K., Berriman,~G., \& Wade,~R.~A. 1989, \apj, 341, 974

1174: 

1175: \bibitem[Wood, Naylor, \& Marsh(1995)]{wood1995}

1176: Wood,~J.~H., Naylor,~T., \& Marsh,~T.~R. 1995, \mnras, 274, 31

1177: 

1178: 

1179: \end{thebibliography}

1180: \clearpage

1181: 

1182: 

1183: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1184: %%% TABLES

1185: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1186: 

1187: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1188: \input{tab1}

1189: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1190: \clearpage

1191: 

1192: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1193: \input{tab2}

1194: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1195: \clearpage

1196: 

1197: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1198: \input{tab3}

1199: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1200: 

1201: \clearpage

1202: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1203: \input{tab4}

1204: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1205: 

1206: \clearpage

1207: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1208: \input{tab5}

1209: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1210: 

1211: \clearpage

1212: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1213: \input{tab6}

1214: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1215: 

1216: \clearpage

1217: 

1218: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1219: %%% FIGURES

1220: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1221: \begin{figure}[tb]

1222: \figurenum{1}

1223: \epsscale{0.97}

1224: \plotone{F1.eps}

1225: \epsscale{1.00}

1226: \vspace{0.5cm}

1227: \figcaption{The continuous line is the $T_{\rm eff}$ profile

1228: for the system model, with restricted annular azimuthal sections in

1229: the outer part of the accretion disk. The dotted line is the same

1230: profile but for the remainder of the azimuthal annular sections in the

1231: outer part of the accretion disk.

1232: Crosses mark plotted points from 

1233: \citet[Figure~4a]{ru1992}. Triangles mark `back' side temperatures from

1234: \citet{bap1998} for November 1994 transformed to our adopted distance of 250pc.

1235: Continuous curves, from top to bottom,

1236: are standard models for $\dot{M}/M_{\odot}({\rm yr}^{-1})=5.0{\times}10^{-8}, 5.0{\times}10^{-9},

1237: 5.0{\times}10^{-10}, {\rm and}~5.0{\times}10^{-11}$. See the text for details.

1238: \label{F1}}

1239: \end{figure}

1240: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1241: 

1242: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1243: \begin{figure}[tb]

1244: \figurenum{}

1245: \includegraphics[scale=0.5,angle=90]{F2a.eps}

1246: \vspace{0.5cm}

1247: %\label{F2a}}

1248: \end{figure}

1249: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1250: 

1251: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1252: \begin{figure}[tb]

1253: \figurenum{2}

1254: \includegraphics[scale=0.5,angle=90]{F2b.eps}

1255: \figcaption{(a): Segregation of DS1 pixels into concentric rings.

1256: (b): Corresponding BINSYN model with 44 annuli.} 

1257: %\label{F2b}}

1258: \end{figure}

1259: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1260: 

1261: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1262: \begin{figure}[tb]

1263: \figurenum{3}

1264: \epsscale{0.97}

1265: \plotone{F3.eps}

1266: \epsscale{1.00}

1267: \vspace{0.5cm}

1268: \figcaption{Fit to \citet{ru1994} central region (grey line, region A, Table~5).

1269: The continuous line is a 20,000K WD plus $\dot{M}=5.0{\times}10^{-9}M_{\odot}{\rm yr}^{-1}$

1270: standard model accretion disk. The accretion disk region corresponding to the

1271: observational data extends to 

1272: $0.10{\rm r_{L1}}$, as for the model data.	Except for the adopted 

1273: distance used to produce the data points there are no free parameters in this fit.

1274: See the text for details. 

1275: \label{F3}}

1276: \end{figure}

1277: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1278: 

1279: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1280: \begin{figure}[tb]

1281: \figurenum{4a-4d}

1282: \epsscale{0.97}

1283: \plotone{F4a_d.eps}

1284: \epsscale{1.00}

1285: \vspace{0.5cm}

1286: %\figcaption{

1287: %\label{F4a_d}}

1288: \end{figure}

1289: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1290: 

1291: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1292: \begin{figure}[tb]

1293: \figurenum{4e-4h}

1294: \epsscale{0.97}

1295: \plotone{F4e_h.eps}

1296: \epsscale{1.00}

1297: \vspace{0.5cm}

1298: %\figcaption{

1299: %\label{F4e_h}}

1300: \end{figure}

1301: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1302: 

1303: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1304: \begin{figure}[tb]

1305: \figurenum{4a-4j}

1306: \epsscale{0.97}

1307: \plotone{F4i_j.eps}

1308: \epsscale{1.00}

1309: \vspace{0.5cm}

1310: \figcaption{

1311: Fits of synthetic spectra to DS1 and DS2 data. In Figure~4a

1312: and Figure~4b

1313: the broad grey plot is the PRISM and the light grey plot is G160L;

1314: they are the 'center' in Table~5.

1315: In Figure~4c and thereafter the heavy grey line is the 'front',

1316: the light grey line is the 'back', and the light continous line is

1317: the 'stream'. The heavy continuous line is the synthetic spectrum.

1318: In all cases the synthetic spectrum 'front' is the same as the 'back'.

1319: Figure~4c is radial designation (Table~5) 0.10; Figure~4d is radial

1320: designation 0.15. The dotted plot is DS2 radial designation B.

1321: Figure~4e is DS1 radial designation (Table~5) 0.20; Figure~4f

1322: is radial designation 0.25; the dotted line is DS2 radial designation C.

1323: Figure~4g is DS1 radial designation 0.30;

1324: Figure~4h is DS1 radial designation 0.40; the dotted line is DS2 radial

1325: designation D. Note that the 'stream' now is higher than either the 'front' or 'back'.  

1326: Figure~4i is DS1 radial designation (Table~5) 0.50. The lower synthetic

1327: spectrum simulates the 'back' and the upper synthetic spectrum simulates

1328: the 'stream'. The lower dotted line is DS2 radial designation E and the 

1329: upper dotted line is DS2 radial designation F. 

1330: Figure~4j

1331: is DS1 radial designation 0.60. The lower synthetic spectrum simulates

1332: the 'back' and the upper synthetic spectrum simulates the 'stream'.

1333: Note that the DS1 'stream' greatly exceeds either the 'front' or 'back'.

1334: See the text

1335: for details.

1336: \label{F4i_j}}

1337: \end{figure}

1338: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1339: 

1340: \clearpage

1341: 

1342: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1343: \begin{figure}[tb]

1344: \figurenum{5}

1345: \epsscale{0.97}

1346: \plotone{F5.eps}

1347: \epsscale{1.00}

1348: \vspace{0.5cm}

1349: \figcaption{Synthetic light curve fit to V data \citep{wh1954} for

1350: May 21, 1953.

1351: See the text for details.

1352: \label{F5}}

1353: \end{figure}

1354: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1355: 

1356: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1357: \begin{figure}[tb]

1358: \figurenum{6}

1359: \epsscale{0.97}

1360: \plotone{F6.eps}

1361: \epsscale{1.00}

1362: \vspace{0.5cm}

1363: \figcaption{Synthetic light curve fit to B data \citep{wh1954} for

1364: April 17, 1953.	Note the systematic trend in the residuals between orbital

1365: phases 0.2 and 0.8.

1366: See the text for details.

1367: \label{F6}}

1368: \end{figure}

1369: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1370: 

1371: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1372: \begin{figure}[tb]

1373: \epsscale{0.97}

1374: \figurenum{7}

1375: \plotone{F7.eps}

1376: \epsscale{1.00}

1377: \vspace{0.5cm}

1378: \figcaption{Synthetic light curve fit to U data \citep{wh1954} for

1379: June 13, 1953.

1380: See the text for details.

1381: \label{F7}}

1382: \end{figure}

1383: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1384: \clearpage

1385: 

1386: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1387: \begin{figure}[tb]

1388: \figurenum{8}

1389: \epsscale{0.97}

1390: \plotone{F8.eps}

1391: \epsscale{1.00}

1392: \vspace{0.5cm}

1393: \figcaption{Grey line, ${\lambda}>1200$\AA: $FOS$ data set Y2AH0201T.

1394: Grey line, ${\lambda}<1200$\AA: $FUSE$ orbit04 spectrum. Upper synthetic spectrum:

1395: Model with 40,000K WD. Lower synthetic spectrum: Model with 20,000K WD.

1396: Both synthetic spectra have been divided by $7.5{\times}10^{41}$,

1397: corresponding to a distance of 281pc. See the text for details.

1398: \label{F8}}

1399: \end{figure}

1400: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1401: 

1402: 

1403: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1404: \begin{figure}[tb]

1405: \figurenum{9}

1406: \epsscale{0.97}

1407: \plotone{F9.eps}

1408: \epsscale{1.00}

1409: \vspace{0.5cm}

1410: \figcaption{FUV plot of same data as Figure~8.

1411: Lower synthetic spectrum: Model with 20,000K WD. 

1412: Upper synthetic spectrum: Model with 40,000K WD.

1413: See the text for details.

1414: \label{F9}}

1415: \end{figure}

1416: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1417: 

1418: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1419: \begin{figure}[tb]

1420: \figurenum{10}

1421: \epsscale{0.97}

1422: \plotone{F10.eps}

1423: \epsscale{1.00}

1424: \vspace{0.5cm}

1425: \figcaption{Grey line, ${\lambda}>1200$\AA: $FOS$ data set Y2AH0401T.

1426: Grey line, ${\lambda}<1200$\AA: $FUSE$ orbit03 spectrum. 

1427: The 

1428: synthetic spectrum, using

1429: a 20,000K WD, has been divided by $7.5{\times}10^{41}$. 

1430: See the text for details.

1431: \label{F10}}

1432: \end{figure}

1433: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1434: 

1435: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1436: \begin{figure}[tb]

1437: \figurenum{11}

1438: \epsscale{0.97}

1439: \plotone{F11.eps}

1440: \epsscale{1.00}

1441: \vspace{0.5cm}

1442: \figcaption{FUV detail of Figure~10. Note the contrast to Figure~9.

1443: The synthetic spectrum, using

1444: a 20,000K WD, has been divided by $7.5{\times}10^{41}$ and plotted as the

1445: heavy line. See the text for details.

1446: \label{F11}}

1447: \end{figure}

1448: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1449: 

1450: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1451: \begin{figure}[tb]

1452: \figurenum{12}

1453: \epsscale{0.97}

1454: \plotone{F12.eps}

1455: \epsscale{1.00}

1456: \vspace{0.5cm}

1457: \figcaption{Grey line; SWP10371+LWR09051, observed 16/10/80.

1458:  Heavy line: 20,000K WD and standard model accretion disk with

1459: $\dot{M}=5.0{\times}10^{-9}M_{\odot}{\rm yr}^{-1}$ mass transfer rate

1460: and at a distance of 234pc.	See the text for details.

1461: \label{F12}}

1462: \end{figure}

1463: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1464: 

1465: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1466: \begin{figure}[tb]

1467: \figurenum{13}

1468: \epsscale{0.97}

1469: \plotone{F13.eps}

1470: \epsscale{1.00}

1471: \vspace{0.5cm}

1472: \figcaption{Grey line: SWP10128+LWR08799, observed 14/09/80. 

1473: Heavy line: 20,000K WD and standard model accretion disk with

1474: $\dot{M}=5.0{\times}10^{-9}M_{\odot}{\rm yr}^{-1}$ mass transfer rate

1475: and at a distance of 234pc.	See the text for details.

1476: \label{F13}}

1477: \end{figure}

1478: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1479: 

1480: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1481: \begin{figure}[tb]

1482: \figurenum{14}

1483: \epsscale{0.97}

1484: \plotone{F14.eps}

1485: \epsscale{1.00}

1486: \vspace{0.5cm}

1487: \figcaption{Grey line: SWP10677+LWR09388, observed 25/11/80. 

1488: Synthetic spectrum: 20,000K WD and standard model accretion disk with

1489: $\dot{M}=3.0{\times}10^{-9}M_{\odot}{\rm yr}^{-1}$ mass transfer rate

1490: and at a distance of 165pc. 

1491: See the text for details.

1492: \label{F14}}

1493: \end{figure}

1494: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1495: 

1496: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1497: \begin{figure}[tb]

1498: \figurenum{15}

1499: \epsscale{0.97}

1500: \plotone{F15.eps}

1501: \epsscale{1.00}

1502: \vspace{0.5cm}

1503: \figcaption{Fit of synthetic light curve of Figure~7 to an

1504: observed U light curve for May 29, 1952 \citep{jph1954}. 

1505: An extreme discrepancy from the system model, with light maximum

1506: occurring $after$ eclipse. See the text for a proposed explanation.

1507: \label{F15}}

1508: \end{figure}

1509: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1510: 

1511: \end{document}

1512: 

1513: