1: %%

2: %% Beginning of file 'sample.tex'

3: %%

4: %% Modified 2005 December 5

5: %%

6: %% This is a sample manuscript marked up using the

7: %% AASTeX v5.x LaTeX 2e macros.

8: 

9: %% The first piece of markup in an AASTeX v5.x document

10: %% is the \documentclass command. LaTeX will ignore

11: %% any data that comes before this command.

12: 

13: %% The command below calls the preprint style

14: %% which will produce a one-column, single-spaced document.

15: %% Examples of commands for other substyles follow. Use

16: %% whichever is most appropriate for your purposes.

17: %%

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

19: 

20: %% manuscript produces a one-column, double-spaced document:

21: 

22: %\documentclass[manuscript]{aastex}

23: 

24: \documentclass{emulateapj}

25: 

26: %% preprint2 produces a double-column, single-spaced document:

27: 

28: %\documentclass[preprint2]{aastex}

29: %\documentclass[preprint,longabstract]{aastex}

30: %\documentclass[12pt,preprint,longabstract]{aastex}

31: \usepackage{apjfonts}

32: \usepackage{lscape}

33: \usepackage{amsmath}

34: 

35: \bibliographystyle{apj}

36: 

37: 

38: %% Sometimes a paper's abstract is too long to fit on the

39: %% title page in preprint2 mode. When that is the case,

40: %% use the longabstract style option.

41: 

42: %% \documentclass[preprint2,longabstract]{aastex}

43: 

44: %% If you want to create your own macros, you can do so

45: %% using \newcommand. Your macros should appear before

46: %% the \begin{document} command.

47: %%

48: %% If you are submitting to a journal that translates manuscripts

49: %% into SGML, you need to follow certain guidelines when preparing

50: %% your macros. See the AASTeX v5.x Author Guide

51: %% for information.

52: 

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

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

55: 

56: \newcommand{\ltappeq}{\raisebox{-0.6ex}{$\,\stackrel{\raisebox{-.2ex}{$\textstyle <$}}{\sim}\,$}}

57: \newcommand{\gtappeq}{\raisebox{-0.6ex}{$\,\stackrel{\raisebox{-.2ex}{$\textstyle >$}}{\sim}\,$}}

58: 

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

60: 

61: %\slugcomment{Accepted for publication in ApJ}

62: 

63: %% If you wish, you may supply running head information, although

64: %% this information may be modified by the editorial offices.

65: %% The left head contains a list of authors,

66: %% usually a maximum of three (otherwise use et al.).  The right

67: %% head is a modified title of up to roughly 44 characters.

68: %% Running heads will not print in the manuscript style.

69: 

70: \shorttitle{Stellar Exotica in 47 Tuc}

71: \shortauthors{Knigge et al.}

72: 

73: %% This is the end of the preamble.  Indicate the beginning of the

74: %% paper itself with \begin{document}.

75: 

76: \begin{document}

77: 

78: %% LaTeX will automatically break titles if they run longer than

79: %% one line. However,you may use \\ to force a line break if

80: %% you desire.

81: 

82: \title{Stellar Exotica in 47 Tucanae\footnotemark[1]}

83: \footnotetext[1]{Based on observations made with the NASA/ESA Hubble Space Telescope,

84: obtained at the Space Telescope Science Institute, which is operated

85: by the Association of Universities for Research in Astronomy, Inc., 

86: under NASA contract NAS 5-26555.}

87: 

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

89: %% author and affiliation information.

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

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

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

93: %% As in the title, use \\ to force line breaks.

94: 

95: %\author{Christian Knigge\altaffilmark{2}, Andrea Dieball\altaffilmark{2}}

96: %\altaffiltext{2}{School of Physics \& Astronomy, University of Southampton, Southampton SO16 7AR, UK}

97: %\author{Jes\'{u}s Ma\'{i}z Apell\'{a}niz\altaffilmark{3,4}}

98: %\altaffiltext{3}{Instituto de Astrof\'{\i}sica de Andaluc\'{\i}a-CSIC, 18008 Granada, Spain}

99: %\altaffiltext{4}{Ram\'on y Cajal Fellow}

100: %\author{Knox S. Long\altaffilmark{5}}

101: %\altaffiltext{5}{Space Telescope Science Institute, Baltimore, MD 21218, USA}

102: %\author{David R. Zurek\altaffilmark{6}}

103: %\author{Michael M. Shara\altaffilmark{6}}

104: %\altaffiltext{6}{American Museum of Natural History, New York, NY 10024, USA}

105: 

106: \author{Christian Knigge, Andrea Dieball}

107: \affil{School of Physics \& Astronomy, University of Southampton, Southampton SO16 7AR, UK}

108: \author{Jes\'{u}s Ma\'{i}z Apell\'{a}niz}

109: \affil{Ram\'on y Cajal Fellow}

110: \affil{Instituto de Astrof\'{\i}sica de Andaluc\'{\i}a-CSIC, 18008 Granada, Spain}

111: \author{Knox S. Long}

112: \affil{Space Telescope Science Institute, Baltimore, MD 21218, USA}

113: \author{David R. Zurek, Michael M. Shara}

114: \affil{American Museum of Natural History, New York, NY 10024, USA}

115: 

116: 

117: %% Mark off your abstract in the ``abstract'' environment. In the manuscript

118: %% style, abstract will output a Received/Accepted line after the

119: %% title and affiliation information. No date will appear since the author

120: %% does not have this information. The dates will be filled in by the

121: %% editorial office after submission.

122: 

123: \begin{abstract}

124: 

125: We present far-ultraviolet (FUV) spectroscopy obtained with the {\em

126: Hubble Space Telescope (HST)} for 48 blue objects in

127: the core of 47~Tuc. Based on their position in a FUV-optical

128: colour-magnitude diagram, these were expected to include cataclysmic

129: variables (CVs), blue stragglers (BSs), white dwarfs (WDs) and other

130: exotic objects.  For a subset of these sources, we also construct

131: broad-band, FUV through near-infrared spectral energy distributions. 

132: Based on our analysis of this extensive data set, we

133: report the following main results. (1) We detect emission lines in three

134: previously known or suspected CVs and thus spectroscopically confirm

135: the status of these systems. We also detect new dwarf nova eruptions

136: in two of these CVs. (2) Only one other source in our spectroscopic

137: sample exhibits marginal evidence for line emission. Thus CVs are not

138: the only class of objects found in the gap between the WD and main

139: sequences, nor are they common amongst objects near

140: the top of the WD cooling sequence. Nevertheless, predicted and

141: observed numbers of CV agree to within a factor of about

142: 2-3. (3) We have discovered a hot

143: ($T_{eff} \simeq 8700$~K), low-mass ($M \simeq 0.05~M_{\odot}$)\ 

144: secondary star in a previously known 0.8~d binary system. This

145: exotic object completely dominates the binary's FUV-NIR output and is

146: probably the remnant of a subgiant that has been stripped of its

147: envelope. Since this object must be in a short-lived evolutionary

148: state, it may represent the  ``smoking gun'' of a recent dynamical

149: encounter. (4) We have found a Helium WD, only the

150: second such object to be optically detected in 47~Tuc, and the first

151: outside a millisecond pulsar system. (5) We have discovered a bright

152: BS with a young WD companion, the only BS-WD binary known in any

153: GC. (6) We have found two additional candidate WD binary systems, one

154: containing a  MS companion, the other containing a subgiant. (7) We

155: estimate the WD binary fraction in the core of 47~Tuc to be  

156: $15\%~^{+17\%}_{-9\%}~{\rm (stat)}~^{+8\%}_{-7\%}~{\rm (sys)}$. (8) The 

157: mass of the optically brightest BS in our sample may exceed twice the

158: cluster turn-off mass, but the uncertainties are too large for this to be

159: conclusive. Thus there is still no definitive example of such a

160: ``supermassive'' BS in any GC.  Taken as a whole, our study

161: illustrates the wide range of stellar exotica that are lurking in the

162: cores of GCs, most of which are likely to have undergone significant

163: dynamical encounters.

164: 

165: \end{abstract}

166: 

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

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

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

170: %% what keyword punctuation is appropriate.

171: 

172: \keywords{

173: globular clusters: individual(NGC 104) --- 

174: blue stragglers --- 

175: novae, cataclysmic variables --- 

176: white dwarfs -- 

177: binaries:close ---  

178: ultraviolet: stars --- 

179: techniques: spectroscopic

180: }

181: 

182: 

183: 

184: \section{Introduction}

185: 

186: Globular clusters (GCs) harbour an impressive array of exotic stellar 

187: populations, such as blue stragglers (BSs), millisecond pulsars

188: (MSPs), X-ray binaries (XRBs), cataclysmic variables (CVs) and Helium

189: white dwarfs (He WDs). The sizes of

190: these populations can be significantly enhanced in GCs, relative to

191: the Galactic field. This is a direct result of the high stellar

192: densities encountered in GC cores (up to at least $10^{6} \, M_{\odot}\,{\rm

193:   pc}^{-3}$). In such extreme environments, dynamical interactions

194: between cluster members occur rather frequently, which opens up new

195: and efficient production channels for virtually all types of ``stellar

196: exotica.'' Thus GCs are excellent laboratories for studying these

197: otherwise extremely rare objects (Maccarone \& Knigge 2007). 

198: 

199: This argument can also be turned around. Since exotic stellar

200: populations in GCs are preferentially produced by dynamical 

201: interactions, they can be used as tracers of a GC's dynamical

202: state. For example, the number of faint X-ray sources in GCs has been

203: shown to correlate well with the 

204: predicted stellar encounter rate (Pooley et al. 2003; Heinke et

205: al. 2003a; Gendre, Barret \& Webb 2003a; Pooley \& Hut 2006), and

206: deviations from this relationship can be related to specific 

207: dynamically-relevant factors (e.g. the highly elliptical orbit of

208: NGC~6397; see Pooley et al. 2003). 

209: 

210: However, exotic stellar populations are not merely passive tracers of

211: GC evolution. Instead, most of these populations are composed of close

212: binaries that actively drive the evolution of their host clusters

213: towards evaporation. Both primordial and dynamically-formed close

214: binaries can be thought 

215: of as heat sources in this context: in successive dynamical 

216: encounters with passing single stars, close binaries tend to give up

217: gravitational binding energy to their interaction partners, and hence

218: to their parent cluster as a whole. The binaries themselves become

219: increasingly tight in the process (i.e. they ``harden''). Thus stellar

220: exotica in GCs are not merely {\em formed} dynamically, but are part

221: of the central feedback loop that links the dynamical evolution of a GC

222: to the formation and evolution of its close binary population

223: (e.g. Hut et al. 1992 [Section 3.1.3]; Shara \& Hurley 2006; Hurley,

224: Aarseth \& Shara 2007).

225: 

226: 

227: \begin{figure*}

228: \epsscale{1.15}

229: \center

230: \includegraphics[scale=0.55,angle=90,keepaspectratio=true]{f1.eps}

231: %\plotone{f1.eps}

232: \caption{{\em Left Panel:} The direct FUV image of a roughly 

233: 26\arcsec$\times$23\arcsec\ region in the core of 47~Tucanae. Sources 

234: bright enough to be included in the spectral extraction are circled

235: and labelled with a spectroscopic identification number. The

236: cross marks the approximate position of the cluster center

237: (Guhathakurta et al. 1992). Essentially all of this image is inside

238: the $R_{c} = 24$\arcsec\ core radius of the cluster (Howell,

239: Guhathakurta \& Gillilan 2000). {\em Right

240:   Panel:} The slitless FUV spectral image of 

241: the same region. Each trail in the image is the dispersed spectrum of

242: a bright FUV source. Note the obvious emission lines in the spectrum

243: of the brightest source (Star~1 = AKO~9; c.f. Knigge et al. 2003).}

244: \label{fig:images}

245: \end{figure*}

246: 

247: Even though exotic stellar populations are abundant in GCs compared to

248: the Galactic field, they still represent only a small fraction of the

249: total number of stars in any cluster. Thus even in GCs, finding and

250: characterizing these populations is a challenging task. One 

251: useful feature common to most members of these populations is that

252: their spectral energy distributions (SEDs) tend to be ``bluer'' than

253: those of ordinary stars (i.e. a 

254: larger fraction of their radiation is emitted at higher

255: frequencies). This can be exploited. For example, deep X-ray

256: surveys have turned out to be a powerful method for locating 

257: XRBs and CVs in GCs (e.g. Grindlay et al. 2001ab; Pooley et al. 2002;

258: Heinke et al. 2003b; Gendre, Barret \& Webb 2003b; Heinke et al. 2005;

259: Webb, Wheatley \& Barret 2006). Similarly, far-ultraviolet (FUV)

260: surveys of GCs are sensitive to an even wider range

261: of ``hot'' exotic populations, while still avoiding the crowding

262: problems associated with optical searches (e.g. Ferraro et al. 2001;

263: Knigge et al. 2002 [hereafter Paper I]; Dieball et al. 2005a; Dieball

264: et al. 2007). However, the detection of 

265: X-ray emission and/or FUV excess is usually not enough to yield a

266: unique source classification. Thus additional information -- ideally

267: based on spectroscopic observations -- is required for a full census

268: of the exotic stellar populations in GCs. 

269: 

270: Here, we present FUV spectroscopy obtained with the {\em Hubble Space

271:   Telescope (HST)} of 48 FUV-excess sources in the 

272: core of 47 Tuc. Based mainly on their position in a FUV/optical

273: colour-magnitude diagram (CMD), these sources are expected to include

274: BSs, CVs and young WDs (Paper I). We find examples of all of

275: these classes in our spectroscopy, along with unexpected discoveries

276: of even more exotic objects. In Section~2, we describe the 

277: observations we use in our study, along with our data reduction and  

278: analysis methods. We also highlight the positions of our spectroscopic

279: targets in the CMD and present proper motion-based membership

280: information for a subset of sources. In Section~3, we present an

281: overview of all the individual spectra and assess the overall

282: relationship between CMD position and spectroscopic

283: classification. The heart of this paper is 

284: Section~4, where we take a detailed look at the most interesting

285: sources. We also construct global FUV through near-infrared (NIR) 

286: SEDs for these objects and use joint fits to the spectroscopic

287: and SED data to determine their nature. In Section~5, we discuss some

288: of the wider implications of our work and estimate the WD binary

289: fraction in 47~Tuc. Finally, in Section~6, we summarize our main

290: results and conclusions. 

291: 

292: %8 BS

293: %8 gap

294: %9 WDs

295: %23 no-opt

296: 

297: 

298: \section{Observations and Data Analysis}

299: 

300: \subsection{Far-Ultraviolet Imaging and Slitless Spectroscopy}

301: \label{sec:fuvdata}

302: 

303: We have carried out a deep spectrosopic and photometric survey of 47

304: Tuc in the FUV waveband. In total, we obtained 30 orbits of

305: {\em HST/STIS} observations for our program (G0-8279), split into 6 epochs

306: of 5 orbits each. Imaging exposures were obtained at the beginning and

307: end of each epoch, with the rest of the time being spent on slitless

308: spectroscopy. Exposure times were typically 600~s for both imaging and

309: spectroscopy. In total, we obtained approximately 14,600~s of FUV

310: imaging and 82,200~s of slitl<ess spectroscopy. 

311: 

312: All of the observations used the FUV-MAMA detectors, together with the

313: F25QTZ filter. This filter blocks geocoronal Ly $\alpha$\, O{\sc i}

314: 1304 \AA\, and O{\sc i}] 1356 \AA\ emission, which would otherwise

315: produce an unacceptably high background in our slitless

316: spectral images. The resulting photometric bandpass is strongly

317: asymmetric, with peak transmission at about 1487~\AA, just above the

318: sharp cut-off around 1450~\AA; the pivot wavelength is 1595~\AA, and

319: the full-width at half-maximum is 229~\AA. The 1024$\times$1024 pixel

320: FUV-MAMA detectors cover a field of view of about

321: 25\arcsec$\times$25\arcsec\ (corresponding to about 1/3 of 47~Tuc's

322: core) with a spatial sampling of about 24.5~mas~pix$^{-1}$. 

323: 

324: The data reduction and analysis steps carried out for the direct

325: images -- which include the construction of a combined master image, as

326: well as source detection and photometry -- have already  been described

327: in Paper I. We will use the resulting FUV source catalogue throughout

328: this paper, except for three small changes and 

329: additions. First, in Paper I, we restricted our attention to the 

330: spatial area common to {\em all} direct images. Even though the

331: spatial shifts between images taken in different epochs are quite

332: small, this restriction removed one bright FUV source for which we now

333: have a high-quality spectrum. We have therefore added this source --

334: Star 999 in the notation adopted below -- to the photometric

335: catalogue. Second, we have applied a small (0.086~mag) correction to

336: the FUV magnitudes obtained in Paper I in order to account for the

337: sensititivity loss of the FUV-MAMA detectors at the time of the

338: observations. This correction was not available at the time of Paper I's

339: publication. Third, as described in Knigge et al. (2006), we have

340: applied distortion and boresight corrections to the FUV source

341: positions. The boresight correction is designed to put our equatorial 

342: coordinates into the absolute astrometric frame defined by the Chandra

343: source positions listed in Heinke et al. (2005). 

344: 

345: Our slitless FUV spectroscopy was carried out with the G140L grating,

346: which provides a dispersion of 0.584~\AA\ pix$^{-1}$ and a spectral

347: resolution of about 1.2~\AA. This set-up can cover a spectral

348: range of roughly 1450~\AA\ -- 1800~\AA. However, note that in slitless

349: observations, the 

350: actual wavelength range covered for each source depends on its spatial

351: position on the detector. Since the observatory software automatically

352: applied a spatial offset to all spectroscopic observations, the

353: effective field of view for the spectral images was only about 85\% of

354: the field available in the direct images. 

355: 

356: Figure~\ref{fig:images} shows the co-added direct and spectral images

357: side-by-side. This figure illustrates the high efficiency of the

358: slitless approach, but also the technical challenges it poses. By

359: moving to the FUV waveband, we can drastically 

360: reduce the crowding that plagues optical images, since the vast

361: majority of ``normal'' cluster 

362: members are simply not hot enough to produce appreciable amounts of

363: FUV flux (see Figure~1 in Paper I for a graphic illustration of this

364: effect). As a result, multi-object slitless spectroscopy of dense GC

365: cores becomes feasible in the FUV and provides an efficient technique

366: for determining the nature of exotic stellar

367: populations. However, extracting single-source spectra from slitless

368: spectral images like that shown in Figure~\ref{fig:images} is not a

369: trivial task. Part of the difficulty stems from the fact that the 

370: spatial and spectral dimensions are not independent. However, the more 

371: important problem is the blending of sources in the spectral image. 

372: 

373: We deal with these technical challenges by using the {\sc multispec}

374: software package, which has been specifically designed to facilitate

375: the extraction of single-source spectra from multi-object slitless

376: spectral images (Ma\'{i}z-Apell\'{a}niz 2005, 2007). Briefly, {\sc 

377: multispec} uses the 

378: position, brightness and colour of each source (as determined from

379: direct images of the field) to generate an initial estimate of its 

380: spectrum. Next, all of these estimates are combined to construct 

381: a complete synthetic spectral image that can be directly compared (in a

382: $\chi^2$ sense) to the actual spectral image. The initial estimates of

383: the individual spectra 

384: are then iteratively optimized until the best possible match between

385: synthetic and real spectral images has been achieved. 

386: 

387: A few points are worth noting regarding this process. First, a key 

388: factor in the construction of the synthetic spectral image from the

389: individual spectra is the wavelength-dependent cross-dispersion

390: profile (CDP). This is effectively the point spread function (PSF) 

391: perpendicular to the dispersion direction in the spectral image. We

392: derived the CDP for our instrumental set-up from

393: calibration observations in the {\em HST} archive. Second, blending is

394: self-consistently accounted for in the extraction, since an arbitrary

395: number of sources can contribute to any given pixel in the synthetic 

396: spectral image. Third, the extraction

397: process also allows for a smoothly varying global detector background,

398: which is iteratively estimated from the residuals between real and

399: synthetic spectral images. Fourth, we extract source

400: spectra separately for each epoch. This avoids any problems associated

401: with the slight spatial offsets between the spectral images for each

402: epoch. Our final spectrum for each source is then the

403: exposure-weighted average of the optimized single-epoch spectra. 

404: 

405: \begin{figure} 

406: %\epsscale{1.2}

407: \epsscale{1.0}

408: \plotone{f2.eps}

409: \caption{A comparison of spectroscopically and photometrically

410: estimated FUV magnitudes for several bright FUV sources (see text for

411: details). The photometric magnitudes include the 0.086~mag correction

412: for the time-dependent sensitivity loss of the FUV-MAMA detectors (see

413: Section~\ref{sec:fuvdata}). Colours mark 

414: different source types, based on the photometric classification scheme

415: shown in Figure~\protect\ref{fig:cmd} and described in

416: Section~\protect\ref{sec:cmd}. Gap objects are shown in blue, white

417: dwarfs in red, blue stragglers in green, and sources without optical

418: counterparts in cyan. The solid black line corresponds to $x = y$, the

419: dashed black line corresponds to the best-fitting line of slope unity

420: to all data points except the blue stragglers. The $y$-intercept of

421: this line is -0.089 mag.}

422: \label{fig:fluxcal}

423: \end{figure}

424: 

425: The extracted source spectra will be good approximations to the true

426: source spectra, as long as (i) the CDP is sufficiently accurate, 

427: (ii) the input source catalogue is sufficiently accurate and complete,

428: and (iii) blending is not too severe. All of these conditions imply

429: that there is a limit to the ability of the algorithm in extracting

430: spectra for faint objects that are heavily blended with one or more

431: bright objects. In this situation, even very slight errors in the

432: extraction of the bright source spectra can lead to large systematic

433: errors in the extracted faint source spectra. 

434: 

435: In order to flag areas of a spectrum that might be compromised by 

436: severe blending, we calculate a wavelength-dependent ``blend ratio''

437: for each extracted source. This is defined as the ratio of counts from 

438: the source itself to the counts produced by all other extracted

439: objects, as estimated at the peak of the source CDP. We stress that

440: this blend ratio is no panacea. In particular, since our estimate of

441: it is necessarily 

442: based on the extracted source spectra, it can itself be affected by  

443: blending. For example, if the flux of a faint source has been

444: overestimated due to blending, the blend ratio associated with that

445: source will be underestimated. Nevertheless, the blend ratio does

446: provide an extremely useful, if qualitative, measure of the

447: wavelength-dependent reliability of each source spectrum. As a rule of

448: thumb, blending is not a serious issue whenever the blend ratio is

449: $\ltappeq 0.1$, whereas spectral regions with blend ratios $\gtappeq

450: 1$\  may be severely compromised.

451: 

452: 

453: % THIS IS WHERE F3 GOES IN EMULATEAPJ

454: 

455: In order to test our spectral extraction method, we have folded the

456: extracted spectra of several bright FUV sources with particularly good

457: spectral coverage through the STIS/FUV-MAMA/F25QTZ photometric 

458: response curve. This was done with the {\sc synphot} package within

459: {\sc iraf/stsdas} and yielded spectroscopic estimates of the FUV

460: magnitudes for these sources. These could be directly compared to the

461: photometric magnitudes, as shown in Figure~\ref{fig:fluxcal}. The

462: agreement between spectroscopic and photometric estimates is generally

463: quite good, except for the 3 BSs in the sample. These generally have

464: spectroscopic magnitudes that are systematically fainter than the

465: photometric ones, by 0.2~mag~-~0.3~mag. This disagreement is actually

466: to be expected, since we can only extract spectra out to

467: 1800~\AA. Beyond this, the throughput drops dramatically and is not well

468: calibrated. However, the FUV spectra of BSs rise extremely steeply

469: towards longer 

470: wavelengths (see Section~\ref{sec:bs_specs}), and the FUV-MAMA/F25QZ

471: photometric bandpass does retain some slight sensitivity there. Thus

472: wavelengths beyond our spectrosopic cut-off  

473: contribute non-negligibly for the BSs. For the other sources,

474: photometric and spectroscopic estimates agree quite well. The rms

475: scatter is only about 0.1~mag, which is acceptable, especially considering that our 

476: sample includes known variables (such as the CVs AKO~9, V1 and V2; see

477: Section~\ref{sec:gap_specs} below). There is marginal evidence

478: for a slight offset of 0.089 mag (in the sense of the spectroscopic 

479: estimates being brighter), so we apply this as a uniform correction to

480: all of our extracted spectra.

481: 

482: In total, 77 FUV sources were included in the spectral extraction

483: process and are highlighted in Figure~\ref{fig:images}. Useful spectra

484: could be obtained for 48 of these sources, and these are listed in

485: Table~\ref{tab:specdat} along with their basic properties.

486: 

487: 

488: \subsection{Optical Imaging}

489: \label{sec:optdata}

490: 

491: The FUV-optical CMD presented in Paper~I served as a starting point in 

492: the classification of our spectroscopic targets (see

493: Section~\ref{sec:overview}). The optical photometry for this CMD was

494: based on a deep, co-added WFPC2/PC/F336W (roughly U-band) image of the

495: cluster core and has already been described in Paper~I. 

496: 

497: For the subset of FUV sources discussed in Section~\ref{sec:sed}, 

498: we additionally constructed broad-band, FUV through NIR SEDs. With the

499: exception of the FUV data point (which comes 

500: from our own STIS/F25QTZ direct image), all of this photometry was

501: obtained from ACS/HRC observations of 47 Tuc in the {\em HST} archive. These

502: observations spanned the full complement of ACS/HRC broad-band

503: filters, including F250W ($\lambda_p = $\ 2716~\AA), F330W ($\lambda_p

504: =\ $3363~\AA), F435W ($\lambda_p = $\ 4311~\AA), F475W ($\lambda_p = 

505: $\ 4776~\AA), F555W ($\lambda_p = $\ 5256~\AA), F606W ($\lambda_p =

506: $\ 5888~\AA), F625W ($\lambda_p = $\ 6296~\AA), F775W ($\lambda_p = 

507: $\ 7665~\AA), F814W ($\lambda_p = $\ 8115~\AA), and F850LP ($\lambda_p = 

508: $\ 9145~\AA); here, $\lambda_p$\ is the pivot wavelength of each

509: filter. Due to the limited field of view and depth of the exposures we

510: used, only a subset of these filters are usually available for each

511: target.

512: 

513: 

514: All of our ACS/HRC photometry was carried out with Andrew Dolphin's

515: PSF-fitting package {\sc dolphot}

516: \footnote{http://purcell.as.arizona.edu/dolphot}, 

517: which is a

518: modified version of the WFPC2-optimized {\sc HSTphot} 

519: package (Dolphin 2000). The advantage of {\sc dolphot} for our

520: purposes is that it provides an ACS-optimized mode, in which

521: detector- and filter-specific simulated {\sc Tiny Tim} PSFs are used as

522: baseline PSF

523: models\footnote{http://www.stsci.edu/software/tinytim/tinytim.html}.  

524: In ACS mode, {\sc dolphot} applies CTE corrections automatically and

525: provides fully calibrated photometric measurements. 

526: 

527: 

528: Throughout this paper, all magnitudes are given on the STMAG system,

529: where 

530: \begin{equation}

531: m_{STMAG} = -2.5 \log{F_{\lambda}} - 21.1. 

532: \label{eq:stmag}

533: \end{equation}

534: For photometric measurements, $F_{\lambda}$\ is the (constant) flux 

535: of a flat-spectrum source that would produce the observed count rate

536: in the photometric band-pass. In Section~4, we will also adopt the

537: STMAG system for spectroscopic data, by  using

538: Equation~\ref{eq:stmag}\ as a monochromatic definition of

539: $m_{STMAG}$. This convention allows us to put spectroscopic and 

540: photometric measurements onto a common, easily interpretable scale.

541: 

542: \subsection{The FUV-optical Colour-Magnitude Diagram}

543: \label{sec:cmd}

544: 

545: Figure~\ref{fig:cmd} shows the FUV-optical CMD for sources in our

546: imaging field of view. The data and models shown in

547: Figure~\ref{fig:cmd} are exactly the same as in Paper~I, except for

548: (i) the slight updates to the FUV photometry described in

549: Section~\ref{sec:fuvdata}, (ii) the fact that we now also show

550: objects without optical counterparts, for which we adopt an F336W

551: magnitude of 23.5 (well below the actual optical detection

552: limit), and (iii) we have added an approximate He WD model sequence

553: (calculated by shifting the standard WD sequence to account for the

554: radius difference between a 0.5~$M_{\odot}$\ WD and

555: 0.25~$M_{\odot}$\ one).

556: \begin{figure*}

557: %\epsscale{1.1}

558: \epsscale{1.0}

559: \plotone{f3.eps}

560: \caption{The FUV-optical CMD of objects in the core of 47~Tuc. 

561: Objects detected in both the FUV and WFPC2/F336W images are shown as

562: black points, objects for which useful spectra could be extracted are

563: additionally marked with circles and labelled with their spectroscopic

564: ID number (c.f. Table~\protect\ref{tab:specdat}). Cyan circles

565: correspond to FUV sources with spectra but without optical

566: counterparts; all 23 of these 

567: are shown at $m_{F336W} = 23.5$ and without their ID numbers (to

568: avoid confusion). Several synthetic model sequences are shown and

569: labelled, as are photometric selection boxes for various

570: populations. We also show the location of the zero-age main sequence

571: in the SMC, which is located behind 47~Tuc. Note that Star~999 (in the

572: BS selection box) is not 

573: marked by a black dot since it was not in the photometric catalogue of

574: Paper~I (see Section~\protect\ref{sec:fuvdata}). The long-dashed diagonal

575: line marks the approximate spectroscopic extraction limit ($m_{FUV} = 

576: 21.0$). The short-dashed diagonal line corresponds to $m_{FUV}

577: =19.5$\ and is discussed in Section~\ref{sec:binfraction}.}

578: \label{fig:cmd} 

579: \end{figure*}

580: 

581: 

582: 

583: Objects for which we have been able to extract useful spectra are

584: circled in Figure~\ref{fig:cmd} and labelled with their spectroscopic

585: ID number (c.f. Figure~\ref{fig:images} and Table~\ref{tab:specdat}).

586: Each spectroscopic target is also assigned a preliminary photometric

587: classification, based on its position in

588: Figure~\ref{fig:cmd}. Possible classifications include white dwarf

589: (WD), ``gap object'' (sources between the WD and main sequences),

590: blue stragglers (BS), FUV source with no optical counterpart (NoOpt),

591: main-sequence turn-off star (MS-TO) and red giant/horizontal branch

592: star (RGB).\footnote{We speculated in Paper~I that the mismatch

593: between the predicted and observed position of stars on the RGB may

594: be due to chromospheric FUV emission from these objects. However, the

595: recent discovery of a significant red leak in the STIS/FUV-MAMA

596: response (STIS Instrument Handbook for Cycle~17; Section 5.3.4)

597: provides a more mundane explanation. This red leak is not yet

598: incorporated into the sensitivity curve used in {\sc

599: stsdas/synphot}. Fortunately, this has negligible impact on our

600: results, since all of the objects discussed in Section~\ref{sec:sed}

601: are relatively blue and FUV bright.}

602: Selection boxes for all of these objects

603: (except the NoOpts) are shown in Figure~\ref{fig:cmd}, and the

604: spectroscopic targets inside each box are marked with a distinct

605: colour. The photometric classifications are also given in

606: Table~\ref{tab:specdat}. 

607: 

608: %THIS IS WHERE FIG 4 GOES IN EMULATEAPJ

609: 

610: In the context of stellar exotica, the gap region is of particular

611: interest. This is the area 

612: in which CVs and detached WD-MS binaries may be expected to be

613: found (Paper~I; Townsley \& Bildsten 2002). Some such objects may also

614: fall into the WD and NoOpt  

615: selection boxes, e.g. if the hot WD still dominates the binary SED in

616: both FUV and F336W. However, as shown statistically in Paper~I, both

617: of the latter boxes are likely to contain primarily WDs. 

618: 

619: In total, we have been able to extract useful spectra for 48 objects,

620: including 8 gap sources, 8 BSs, 9 WDs and 23 NoOpts. All MS-TO and RGB 

621: objects were too faint to be included in our spectral extraction. The

622: long-dashed diagonal line corresponds to $m_{FUV} = 21.0$ and marks the

623: approximate spectroscopic extraction limit. Note that not all sources

624: above this limit are circled in Figure~\ref{fig:cmd}. This is because

625: some bright sources fell outside our spectroscopic field of view and

626: others were too severely blended with even brighter objects to yield

627: useful spectra. The short-dashed diagonal line corresponds to $m_{FUV}

628: = 19.5$ and marks the approximate spectroscopic completeness limit, as

629: discussed further in Section~\ref{sec:binfraction}. Using the same

630: method as in Paper~I, we estimate that $\simeq 1.4$\ objects in the CMD

631: brighter than $m_{FUV} = 21.0$, and $\simeq 0.3$ objects brighter than

632: $m_{FUV} = 19.5$ may be the result of a chance coincidence (i.e. a

633: superposition of two unrelated stars). These estimates are based on

634: the number of FUV sources brighter than the adopted cut-offs, the

635: number of optical sources, and the number of actual matches between

636: these source lists. 

637: %Amongst FUV 

638: %objects brighter than $m_{FUV} = 19.0$, the expected number of

639: %spurious coincidences is $\simeq 0.1$.

640: 

641: \subsection{Proper Motions and Cluster Membership}

642: \label{sec:proper}

643: 

644: \begin{figure}

645: %\epsscale{1.2}

646: \epsscale{1.0}

647: \plotone{f4.eps}

648: \caption{Proper motion diagram of stars near the central regions of

649:   47~Tuc (constructed from data in McLaughlin et al. [2006]). The 6

650:   FUV sources with useful spectra for which proper motions are

651:   available are highlighted and labelled with their identification

652:   numbers. Colours correspend to the different source types, based on

653:   the photometric classification scheme shown in

654:   Figure~\protect\ref{fig:cmd} and described in

655:   Section~\protect\ref{sec:cmd}. Blue stragglers are shown in green,

656:   and the one gap object in blue. The solid black ellipses correspond

657:   to 1$\sigma$, 2$\sigma$, 3$\sigma$\ contours (where

658:   $\sigma$\ refers to the 1-dimensional standard deviations in

659:   $\mu_\alpha$\ and $\mu_\delta$). These contours contain 42.4\%,

660:   87.3\% and 98.2\% of all sources, respectively. All proper

661:   motions are relative to the mean motion of 47~Tuc, and the

662:   dashed blue circle marks the escape velocity from the cluster. Stars

663:   belonging to the Small Magellanic Cloud (which is located behind the

664:   cluster along the line of sight) would lie around $\mu_\alpha =

665:   -4.7$~mas~yr$^{-1}$\ and $\mu_\delta = 1.3$~mas~yr$^{-1}$ in these

666:   co-ordinates (beyond the right edge of the plot).}

667: \label{fig:proper}

668: \end{figure}

669: 

670: 

671: McLaughlin et al. (2006) have recently created a large photometric and

672: proper motion catalogue that contains over 14,000 stars near the

673: central regions of 47~Tuc. A subset of 18 of our spectroscopic targets

674: have counterparts in their master catalogue, and 6 of these have

675: well-determined proper motions. In Figure~\ref{fig:proper} we show the

676: location of these stars in the proper motion diagram, relative to

677: other cluster members in the McLaughlin catalogue. All of our

678: targets (which include 5 BSs and 1 gap object according to the

679: CMD-based classification) are fully consistent with being cluster

680: members.

681: 

682: 

683: % THIS IS WHERE F5 GOES IN EMULATEAPJ

684: 

685: \section{FUV Spectroscopy of 48 Objects in the Core of 47 Tuc}

686: \label{sec:overview}

687: 

688: In this section, we present an overview of all the useful spectra 

689: we could extract from our slitless FUV spectral images. We have

690: grouped the spectra by their CMD-based classifications for this

691: purpose. 

692: 

693: 

694: \subsection{The Spectra of Gap Objects}

695: \label{sec:gap_specs}

696: 

697: The FUV spectra of our 8 gap sources are shown in

698: Figure~\ref{fig:gap_specs}. The diversity of these spectra is striking

699: and immediately suggests that these objects do not form a homogenous

700: group. All gap sources will be analyzed individually in

701: Section~\ref{sec:sed}, so we only present a brief overview

702: here. 

703: 

704: 

705: \begin{figure*} 

706: \center

707: \includegraphics[angle=-90,width=1.0\textwidth,viewport=40 40 520 730,clip]{f5.eps}

708: \caption{\scriptsize The FUV spectra of gap objects. The top panel for each object

709:   shows the extracted spectrum. The positions and identifications of

710:   several key spectral lines are also indicated. The red horizontal

711:   line marks the flux level corresponding to the FUV photometry (for a

712:   flat spectrum source; see Equation~\ref{eq:stmag} in

713:   Section~\protect\ref{sec:optdata}). Note that the wavelength range

714:   covered for each object depends on its position on the detector. 

715:   The bottom panel for each

716:   objects shows the wavelength-dependent blend ratio, as defined in

717:   Section~\protect\ref{sec:fuvdata}. The horizontal dashed lines mark

718:   blend ratios of 0.1 and 1.0; this is the range where blending can start

719:   to become a problem.}

720: \label{fig:gap_specs}

721: \end{figure*}

722: 

723: % THIS IS WHERE F6 GOES IN EMULATEAPJ

724: 

725: % THIS IS WHERE F7 GOES IN EMULATEAPJ

726: 

727: 

728: Clear emission lines are seen in Stars~1, 3, and 20, all of which were

729: previously known or suspected CVs (also known as AKO~9, 

730: V1, and V2, respectively [see Table~\ref{tab:specdat}]). The presence

731: of emission lines confirms the CVs nature of these objects. Our

732: data on AKO~9 has already been analyzed in detail elsewhere (Knigge et

733: al. 2003), but the spectroscopic confirmation of V1 and V2 as CVs is

734: reported here for the first time in the refereed literature.

735: \footnote{The review presented in Knigge (2004) inludes preliminary,

736:   rough spectral extractions for these objects.}

737: Only one other gap source -- Star~17 -- shows tentative evidence of

738: line emission. Unfortunately, the position of this object on the

739: detector is such that only a short segment of its spectrum is

740: available, with the red cut-off occuring just beyond the C~{\sc iv}

741: line. The line itself thus falls close to the edge of the detector,

742: which is why we do not consider the apparent flux excess near

743: 1550~\AA\ to be a definite detection of line emission. We will discuss

744: Star~17 in more detail in Section~\ref{sec:star17}. 

745: 

746: Of the 4 remaining objects, 3 exhibit flat or blue continua, but the

747: fourth -- Star~10 -- presents a rather strange, red continuum with

748: absorption lines. All 4 of these objects turn out to be interesting 

749: (see Section~\ref{sec:sed}), but Star~10, in particular, is

750: probably the most exotic and unusual object our FUV survey has

751: uncovered. This object is analyzed more closely in

752: Section~\ref{sec:star10}.

753: 

754: \subsection{The Spectra of Blue Stragglers}

755: \label{sec:bs_specs}

756: 

757: Figure~\ref{fig:bs_specs} shows the FUV spectra of our 8 BSs in the

758: core of 47~Tuc. Only the two brightest of these -- Stars~2 and 999 --

759: will be analyzed in more detail in Section~\ref{sec:sed}.

760: 

761: Unlike the gap sources, all but one of the BSs display rather similar

762: FUV spectra. More specifically, 7 of the 8 BSs show extremely red FUV

763: continua, with fluxes increasing by typically 2 orders of magnitude

764: between roughly 1500~\AA\ and 1800~\AA. This near-exponential rise is

765: expected, since BSs are stars of spectral type A-F for which the FUV waveband

766: falls on the Wien tail of their SEDs. In principle, this should make

767: the FUV flux and spectral shape a very sensitive thermometer for BSs, 

768: a possibility we will explore in Section~\ref{sec:star999}. In those

769: sources with sufficient S/N, the spectra also exhibit absorption lines

770: due to C~{\sc i} and Al~{\sc ii} (with a possible contribution from

771: Fe~{\sc ii}).

772: 

773: The obvious outlier among the BSs is Star~2, whose FUV continuum rises

774: much more slowly towards longer wavelengths. Moreover, the flux level

775: at the blue end of the spectrum (around 1500~\AA) is much higher than

776: for any other BS in our sample. Thus even though the red part of

777: Star~2's FUV spectrum may be affected by blending, the discrepancy 

778: between this object and the other BSs appears to be real. Star~2 is

779: analyzed more carefully in Section~\ref{sec:star2}. 

780: 

781: \subsection{The Spectra of White Dwarfs}

782: \label{sec:wd_specs}

783: 

784: The FUV spectra of the 9 hot WDs in our sample are shown in

785: Figure~\ref{fig:wd_specs}. Only the two brightest of these -- Stars~5

786: and 7 -- will be analyzed in more detail in

787: Section~\ref{sec:sed}. 

788: 

789: The WDs present a relatively homogenous set of FUV spectra. which can

790: be characterized as blue, featureless continua. This is as expected

791: for young WDs near the top of the cooling sequence. There is

792: little evidence for absorption or emission lines in any of these

793: spectra. The only features that are perhaps worth remarking

794: on are a hint of an absorption dip near C~{\sc iv} in the spectrum of

795: Star~5, and a hint of emission near the same line in the spectrum of

796: Star~29. However, we regard neither of these features as convincing. We

797: thus conclude that the top of the WD cooling sequence in 47~Tuc

798: does not hide a large population of WD-dominated CVs (see also

799: Section~\ref{sec:no_specs}). Surprises are still possible though, as

800: illustrated by the analysis of Star~7 in Section~\ref{sec:star7}.

801: 

802: \begin{figure*} 

803: \center

804: %\includegraphics[angle=-90,width=0.88\textwidth,viewport=40 40 520 730,clip]{f6.eps}

805: \includegraphics[angle=-90,width=1.0\textwidth,viewport=40 40 520 730,clip]{f6.eps}

806: \caption{The FUV spectra of blue stragglers. See caption to

807:   Figure~\protect\ref{fig:gap_specs} for details.}   

808: \label{fig:bs_specs}

809: \end{figure*}

810: 

811: 

812: 

813: \begin{figure*} 

814: \center

815: %\includegraphics[angle=-90,width=0.88\textwidth,viewport=40 40 520  730,clip]{f7.eps}

816: \includegraphics[angle=-90,width=1.0\textwidth,viewport=40 40 520  730,clip]{f7.eps}

817: \caption{The FUV spectra of hot white dwarfs. See caption to

818:   Figure~\protect\ref{fig:gap_specs} for details.}   

819: \label{fig:wd_specs}

820: \end{figure*}

821: 

822: % THIS IS WHERE F8 GOES IN EMULATEAPJ

823: 

824: 

825: \subsection{The Spectra of FUV Sources without Optical Counterparts}

826: \label{sec:no_specs}

827: 

828: Figure~\ref{fig:no_specs} shows the spectra of our 23 FUV sources 

829: without optical counterparts. This group of objects is also expected

830: to be dominated by hot WDs. As shown in Paper~I, the number of these

831: objects is consistent with expectations for the WD population, and our

832: optical (WFPC2/PC/F336W) photometry is certainly not complete for

833: these optically faint objects. Since we do not have optical

834: counterparts, none of these sources are included in

835: Section~\ref{sec:sed}.

836: 

837: \begin{figure*} 

838: \center

839: \includegraphics[angle=-90,width=0.95\textwidth,viewport=40 40 520 730,clip]{f8a.eps}

840: \includegraphics[angle=-90,width=0.95\textwidth,viewport=40 40 520 730,clip]{f8b.eps}

841: \caption{The FUV spectra of objects without optical counterparts. See caption to

842:   Figure~\protect\ref{fig:gap_specs} for details.}   

843: \label{fig:no_specs}

844: \end{figure*}

845: 

846: 

847: \begin{figure*}[t] 

848: \center

849: \figurenum{8 (cont)}

850: \includegraphics[angle=-90,width=0.95\textwidth,viewport=40 40 360 730,clip]{f8c.eps}

851: \caption{The FUV spectra of objects without optical counterparts.}

852: \end{figure*}

853: 

854: 

855: 

856: Figure~\ref{fig:no_specs} shows that the spectra of essentially all of

857: these FUV sources are consistent with expectations for single DA

858: WDs. In particular, the resemblance of the spectra to those of the

859: photometrically-classified WDs (Figure~\ref{fig:wd_specs}) is

860: obvious. The spectrum of Star~33 contains an obvious hump, but this is

861: caused by blending with a fainter FUV source. That source was not

862: included in the spectral extraction, which explains why the blend

863: ratio for Star~33 does not show this feature. None of the objects in

864: Figure~\ref{fig:no_specs} show convincing evidence for emission or

865: absorption lines. This confirms the conclusion of Section~\ref{sec:wd_specs}

866: that the top of the cooling sequence does not harbour numerous CVs.

867: 

868: 

869: \section{A Closer Look at Individual Sources}

870: \label{sec:sed}

871: 

872: 

873: We will now present a more detailed analysis of the most interesting

874: and representative sources in our spectroscopic sample. This includes

875: all of the gap sources, as well as the two brightest objects from the

876: CMD-based WD and BS groups. All sources analysed in this section are

877: listed in bold at the top of Table~\ref{tab:specdat}. The last column

878: in the table presents our final classification for each source. 

879: 

880: Throughout this section, we will use model fits to the combined

881: spectral and SED data sets to shed light on the physical nature of our

882: sources. The cluster parameters adopted in these models are those of

883: Gratton et al. (2003), i.e. $d = 4840$~pc, $E(B-V) = 0.024$, $[Fe/H] =

884: -0.66$. All stars, except WDs, are described by interpolating on the

885: latest, $\alpha$-enhanced Kurucz {\sc atlas9} model atmosphere

886: grids.

887: \footnote{The Kurucz model atmosphere grids are available for download

888:   at http://kurucz.harvard.edu/grids.html.} 

889: These grids use the updated opacity distribution functions

890: described by Castelli \& Kurucz (2001) and account for enhanced

891: $\alpha$-element abundances at a level of $[\alpha/Fe] = 0.4$. For the

892: purpose of modelling WDs, we rely on the grid of synthetic WD spectra

893: described in G\"ansicke, Beuermann \& de Martino (1995), which has

894: been kindly provided by Boris G\"ansicke. We generally adopt $M_{WD} =

895: 0.5~M_{\odot}$\ and $R_{wd} = 0.017~R_{\odot}$\ (corresponding to

896: $log{\;g} = 7.6$) to describe WDs in 47~Tuc. This is appropriate for

897: recently formed WDs near the top of the cooling sequence (Renzini \&

898: Fusi Pecci 1988; Wood 1995). We also use the WD models

899: to qualitatively describe other hot spectral components, such as

900: accretion disks in CVs. We will only be interested in rough estimates

901: of the characteristic temperatures and sizes of such generic hot

902: components, and the use of WD models should be adequate for this

903: purpose. 

904: 

905: The spectral resolution of the Kurucz models is relatively low, with a

906: typical wavelength step of 10~\AA\ in the FUV region. This turns out to

907: be quite adequate for our qualitative modelling, and so the observed

908: spectra (and also the WD models) are interpolated onto the wavelength

909: grid of the Kurucz spectra during our fitting procedure. We have also

910: carried out synthetic photometry on all models in the Kurucz and WD

911: grids. This was done using {\sc synspec} in {\sc iraf/stsdas}, which 

912: includes throughput files for all {\em HST} detector/filter

913: combinations. Most of our modelling is done in a least squares sense,

914: with every data point being given equal weight. Since we only consider

915: the coarse Kurucz wavelength grid when modelling the FUV spectrum, this

916: usually provides a reasonable compromise between the emphasis placed

917: on the spectrum relative to the broad-band SED. In a few cases, the

918: relative weights assigned to the spectroscopy and photometry were

919: adjusted slightly to improve the overall quality of the fits.

920: 

921: We note from the outset that our goal here is to find physically

922: plausible descriptions of the data, rather than arbitrary sets of 

923: best-fit parameters. Thus we will sometimes constrain certain

924: parameters or parameter combinations to have values consistent with,

925: for example, the expected location of the WD or main sequences. Of

926: course, we will only do this when adequate fits can actually be found

927: with the constrained parameters. We feel this approach is appropriate,

928: since the match between data and models is limited by systematic

929: effects (e.g. blending in the spectra, crowding in the photometry, and

930: uncertainties affecting the theoretical models) rather than by purely

931: statistical errors on the data. 

932: 

933: \subsection{Star~1 = AKO~9: A CV with a Subgiant Donor}

934: \label{sec:star1}

935: 

936: \begin{figure*} 

937: \center

938: \includegraphics[angle=-90,width=1.0\textwidth]{f9.eps}

939: \caption{

940: FUV spectrum (top panel) and broad-band SED (bottom panel) of Star~1 =

941: AKO~9, along with a two-component model fit to these observations. The

942: black histogram in the top panel is the observed FUV spectrum (shown

943: at the full resolution), while the blue, red and black lines (in both

944: panels) show the

945: model spectra of the hot, cool and combined components,

946: respectively. Only the wavelength regions shown with a thick solid line

947: were used in the spectral fit. Underneath the spectrum, we also show

948: the corresponding blend ratio, as defined in 

949: Section~\protect\ref{sec:fuvdata}.  

950: In the bottom panel, the filled black circles are 

951: the ACS/HRC photometric data we try to fit, whereas the open red

952: squares are the 

953: magnitudes predicted by the full model. The corresponding residuals

954: are also shown. The emission lines in the FUV spectrum

955: confirm this object as a CV, while the broad-band SED

956: establishes the evolved nature of its mass-losing secondary. The open

957: circle around 3400~\AA\ in the bottom panel is our older 

958: WFPC2/PC/F336W data point, and the open triangles are the ACS/HRC data

959: points obtained in July 2002. The system was caught in a dwarf nova

960: outburst during the 2002 observations (see also

961: Figure~\protect\ref{fig:star1phot}).}  

962: \label{fig:star1}

963: \end{figure*}

964: 

965: \begin{figure*}[h!]

966: \center

967: \includegraphics[angle=0,width=0.95\textwidth]{f10.eps}

968: \caption{Approximately U-band observations of AKO~9 obtained with {\em HST}

969: in July 1999 (left panel) and July 2002 (right panel). The difference in

970: brightness is obvious and implies that AKO~9 experienced a dwarf nova

971: eruption in July 2002. In these images, north is up, east is to the left, 

972: and the field of view is approximately 1.8\arcsec.} 

973: \label{fig:star1phot}

974: \end{figure*}

975: 

976: Star~1 is the brightest FUV source in 47~Tuc and is identical to the

977: previously known blue object AKO~9. A detailed analysis of our FUV

978: spectroscopy and photometry for this source has already been presented

979: in Knigge et al. (2003). Figure~\ref{fig:star1} shows the FUV spectrum

980: and broad-band SED for AKO~9, along with our suggested two-component

981: fit to these data. 

982: 

983: As shown in Knigge et al. (2003), AKO~9 is almost certainly a

984: CV with an orbital period of $P_{orb} = 1.1091$~d

985: and an evolved, probably subgiant donor star. This description also

986: turns out to provide a good fit to the more extensive broad-band

987: SED we have assembled here. In constructing this model, we have

988: assumed that the secondary is a turn-off mass object and have used the

989: orbital period-mean density relation for Roche-lobe filling stars,  

990: \begin{equation}

991: \left< \rho \right> \simeq 107 P_{orb,hrs}^{-2} {\rm g~cm^{-3}},

992: \label{eq:rho}

993: \end{equation}

994: as an additional constraint on the donor parameters (e.g. Warner

995: 1995). The hot component  

996: that dominates the FUV light is too bright to be a pure WD and is

997: instead likely to arise in an accretion disk. In our fit, we have

998: described this component with a $\log{~g} = 7$~WD model atmosphere. The

999: temperature was fixed at $T_{hot} = 30,000$~K, which yields an effective

1000: radius of $R_{hot} \simeq 0.07~R_{\odot}$ (for a pure WD, we would

1001: expect $R_{hot} \simeq 0.02~R_{\odot}$). However, the temperature and

1002: effective radius are highly correlated and not well constrained. 

1003: 

1004: In the process of constructing the broad-band SED, we noticed that a

1005: subset of optical observations obtained in July 2002 were

1006: systematically brighter than all of our other optical data. We show

1007: these 4 data points as open triangles in Figure~\ref{fig:star1}. Note,

1008: in particular, the almost 2~mag brightness difference between the

1009: roughly U-band data used in Paper~I (open circle in

1010: Figure~\ref{fig:star1}) and that obtained in July 2002. As shown in

1011: Figure~\ref{fig:star1phot}, this difference is easily noticeable in

1012: the images themselves.

1013: 

1014: We conclude that we have discovered a dwarf nova eruption of

1015: AKO~9. Two previous outburst of this object have been observed with

1016: {\em HST}, one in July 1986, the other in October 1992 (Minniti et al. 1997;

1017: see also Knigge al. al. 2003). The present discovery of an eruption in

1018: July 2002 confirms that the mean interoutburst recurrence time is at

1019: most 6-7 years. However, given the relatively sparse time coverage of

1020: the photometry investigated to date, a shorter recurrence time remains

1021: possible.

1022: 

1023: Even though all of the existing observational evidence is consistent

1024: with AKO~9 being a dwarf nova-type CV, the source warrants additional

1025: investigation. In particular, a radial velocity study would be

1026: invaluable for establishing the system parameters. As an incentive for

1027: such a study, we note that the X-ray, FUV and optical properties

1028: established so far do not completely preclude the possibility that the

1029: primary could be a black hole, rather than a WD.

1030: 

1031: \subsection{Star~2: A Blue Straggler with a White Dwarf Companion}

1032: \label{sec:star2}

1033: 

1034: \begin{figure*} 

1035: \center

1036: \includegraphics[angle=-90,width=1.0\textwidth]{f11.eps}

1037: \caption{The FUV spectrum (top panel) and broad-band SED (bottom

1038: panel) of Star~2, along with our best-bet model fit to the

1039: data (see text for details). Notation is as in

1040: Figure~\protect\ref{fig:star1}. The region of the spectrum beyond 1550

1041: \AA\ was excluded from the fit because of blending with other objects, 

1042: but there is clear evidence for strong, unblended emission shortward

1043: of this. Star~2 appears to be a blue straggler with a hot WD companion.}

1044: \label{fig:star2}

1045: \end{figure*}

1046: 

1047: 

1048: Star~2 lies on the BS sequence in the FUV-optical CMD

1049: (Figure~\ref{fig:cmd}), but its FUV spectrum rises much more slowly

1050: towards longer wavelengths than that of any other BS in our data

1051: (Figure~\ref{fig:bs_specs}). Even though our spectrum for this object

1052: is rather short, and may be affected by blending longwards of about

1053: 1550~\AA, we have no reason to doubt the reality of the FUV excess it

1054: implies (relative to other BSs). 

1055: 

1056: Figure~\ref{fig:star2} shows the FUV spectrum and broad-band SED for

1057: this object, along with our best-bet model. All of the photometry is

1058: well described by a single component with parameters appropriate for a

1059: BS in 47 Tuc. However, no single component model is able to also match

1060: the FUV spectrosopy.

1061: \footnote{The surface gravity (and hence mass) of the BS is not well 

1062: constrained by the fit and was therefore fixed at a physically

1063: plausible value. Also, the predicted FUV spectrum of the BS may not be

1064: very accurate (see Section~\ref{sec:star999}). However, our conclusion

1065: that the observed FUV spectrum shows a strong excess at the shortest

1066: wavelengths is insensitive to these caveats.}

1067: 

1068: The simplest and most plausible way to account for the observed FUV

1069: excess is to invoke a WD companion to the BS. As shown in 

1070: Figure~\ref{fig:star2}, a WD with $T_{eff} \simeq 34,000$~K provides a

1071: reasonable match to the FUV spectrum without significantly affecting 

1072: the broad-band SED. Thus a BS-WD binary system provides a good

1073: match to the entire spectroscopic and potometric data set.

1074: 

1075: Is it possible that we are being fooled by a chance superposition?

1076: There are two questions to consider in this context. First, could

1077: the FUV flux we have identified with the short-wavelength spectrum of

1078: Star~2 actually be 

1079: associated with a different FUV source? Inspection of the direct image

1080: (Figure~\ref{fig:images}) shows that there are two FUV-bright 

1081: sources -- Stars~17 and 30 -- that lie close enough to Star~2 in 

1082: the spatial direction to warrant consideration. However, Star~17

1083: lies to the right of Star~2 in the direct 

1084: image and so cannot be responsible for any short-wavelength flux in

1085: the spectrum of Star~2. Star~30, on the other hand, has photometric 

1086: and spectroscopic properties consistent with a single WD and lies

1087: about 585 pixels (corresponding to about 340~\AA) to the left of

1088: Star~2. If the short-wavelength flux attributed to Star~2 were really

1089: due to Star~30, essentially of the observed counts would have to be

1090: associated with wavelengths well beyond 1800~\AA. However, the FUV

1091: bandpass has very little sensitivity at such long wavelengths, and the

1092: spectroscopic countrates of even the reddest FUV-bright sources in our

1093: sample -- the BSs -- peak well shortward of 1800~\AA. Thus Star~30 is

1094: almost certainly not the source of any flux assigned to Star~2.

1095: 

1096: The second question is whether the FUV source we see in the direct

1097: image could be a chance blend of the bright BS with a comparably

1098: bright WD. It is worth noting here that the BS is {\em expected} to be 

1099: detected in the FUV, so it is not plausible that the WD should 

1100: completely dominate the FUV output. We have inspected 

1101: the direct FUV image in the vicinity of Star~2 and find no evidence to

1102: support the idea that the FUV source is a blend. However, it should be

1103: acknowledged that the FUV imaging PSF 

1104: is rather complex and asymmetric, and that there are not many

1105: FUV-bright objects against which Star~2 could be meaningfully

1106: compared. By the same token, however, the 

1107: probability of two such bright objects lying close to each other by 

1108: chance is very small. Figure~\ref{fig:star2} shows that both the WD

1109: and the BS would be brighter than $m_{FUV} \simeq 19$ on their

1110: own. In total, there are only 20 objects (including Star~2) that are

1111: this bright in our FUV photometry. The probability that two such

1112: objects should lie close enough to each other to form an

1113: unrecognizable blend (say within 3~pixels) is only $\simeq 1$\%. Thus

1114: it is highly likely that the hot WD is physically associated with the

1115: BS.

1116: 

1117: Even though Star~2 is the first BS-WD binary that has been identified

1118: in any GC to date, the existence of such objects should probably be

1119: expected. One straightforward way to form such a system

1120: is directly from a MS-MS binary. As the more massive object in this

1121: progenitor system ascends the giant branch, it may overflow its Roche

1122: lobe and thus turn its companion into a BS via Case~B or Case~C mass

1123: transfer

1124: \footnote{Here, Case~B refers to mass transfer occuring prior to He

1125: ignition (during the ascent of the red giant branch), which may

1126: produce a He WD; Case~C refers to mass transfer after He ignition

1127: (during the ascent of the asymptotic giant branch), which would leave 

1128: behind a CO WD (e.g. Paczy{\'n}ski 1971).}

1129: 

1130: In either case, the outcome would be a BS-WD

1131: binary. This formation channel will be available in GC cores as long 

1132: as the original binary is hard (since otherwise it would not survive

1133: until the onset of mass transfer). The hard/soft boundary in GCs

1134: corresponds to binary separations on the order of a few AU

1135: (e.g. Davies 1997), so there is certainly room for BS-WD progenitor

1136: systems in the parameter space of hard binaries. 

1137: 

1138: An alternative way to form BS-WD binaries is directly via dynamical

1139: encounters. In particular, since single BSs are the most massive stars 

1140: found in GCs, any 3-body interaction involving such a BS would 

1141: tend to leave it with a companion (usually the more massive member of

1142: the original binary). Thus a system like Star~2 could have been

1143: formed, for example, in an exchange encounter involving a BS and a

1144: WD-MS binary. 

1145: 

1146: \subsection{Star~3 = V1: A CV with a Main-Sequence Donor}

1147: \label{sec:star3}

1148: 

1149: \begin{figure*} 

1150: \center

1151: \includegraphics[angle=-90,width=1.0\textwidth]{f12.eps}

1152: \caption{The FUV spectrum (top panel) and broad-band SED (bottom

1153: panel) of Star~3 = V1, along with our best-bet model fit to the

1154: data (see text for details). Notation is as in 

1155: Figure~\protect\ref{fig:star1}. Star~3 is a CV.}

1156: \label{fig:star3}

1157: \end{figure*}

1158: 

1159: Star~3 = V1 is a previously suspected CV

1160: (Paresce, De Marchi \& Ferraro 1992; Grindlay et al. 2001a; Ferraro et

1161: al. 2001; Knigge et al. 2002). Its orbital period is either 3.5 hrs or 7.0 hrs, 

1162: depending on whether the observed optical variability is due to

1163: ellipsoidal variations or not (Edmonds et al. 2003b). 

1164: 

1165: Figure~\ref{fig:star3} shows our FUV spectrum and broad-band SED for

1166: V1. As already noted in Section~\ref{sec:gap_specs}, the detection of 

1167: C~{\sc iv} in emission spectroscopically confirms V1 as a CV. The

1168: emission line is double-peaked, which suggests that it was formed in a

1169: rotating medium, such as an accretion disk (e.g. Smak 1981; Horne \&

1170: Marsh 1986) or a rotating disk wind 

1171: viewed at high inclination (e.g. Knigge, Woods \& Drew 1995). The FUV-NIR

1172: SED is consistent with a single, hot component dominating the flux at

1173: all wavelengths. 

1174: 

1175: Since the donor of V1 must fill its Roche lobe, we can use the

1176: period-mean density relation (Equation~\ref{eq:rho}) to constrain its

1177: nature. If the orbital period is 7.0~hrs, and the donor lies close to

1178: the cluster MS, the donor mass would have to be roughly 0.75~$M_{\odot}$.

1179: Such a star would be 0.5 mag - 1.0 mag brighter than

1180: V1 at $\lambda \gtappeq 5000$~\AA; this makes the long-period option

1181: unlikely. 

1182: \footnote{Strictly speaking, unevolved CV donors are slightly bloated

1183: relative to isolated MS stars (e.g. Knigge 2006). However, even if we

1184: assume that the secondary in V1 is a full 20\% larger than a MS star 

1185: (and adjust its parameters so as to still satisfy the

1186: period-density relation), it would be significantly ($\simeq 0.3$~mag)

1187: brighter than the observations at $\lambda \gtappeq 6000$~\AA. Any 

1188: contribution from the hot component that must dominate at shorter

1189: wavelengths would make this discrepancy even worse.} 

1190: If the period is 3.5~hrs instead (which would imply that the

1191: observed variability is not ellipsoidal in nature), the expected donor

1192: mass is about 0.5~$M_{\odot}$. As shown in Figure~\ref{fig:star3},

1193: such a donor would be a relatively minor contributor to the flux at

1194: all FUV-NIR wavelengths. This is self-consistent, since a faint 

1195: donor like this would not produce a significant ellipsoidal signal. We

1196: therefore favour the 3.5~hr period and suggest that the donor lies on

1197: or near the lower MS of the cluster. 

1198: 

1199: Both the FUV continuum and the FUV-NIR SED are reasonably described by

1200: a 12,000~K optically thick component with an effective radius of

1201: 0.18~$R_{\odot}$ (where we have again used a $\log{\;g} = 7.0$\ WD model

1202: to represent this component). Very similar models have been found to

1203: match the accretion disk-dominated SEDs of nova-like CVs in the

1204: Galactic field (e.g. Knigge et al. 1998). The identification of V1 as

1205: a nova-like CV containing a bright accretion disk is consistent with

1206: the fact that no dwarf nova outbursts have so far been discovered in

1207: this source. However, we do note that the observed SED in

1208: Figure~\ref{fig:star3} exhibits undulations (e.g. between 4000~\AA\ and

1209: 6000~\AA), which can also be seen as large (up to 0.5~mag) residuals

1210: between model and data. These undulations and residuals are probably

1211: due to variability. Even though nova-like CVs do not show the

1212: eruptions associated with dwarf novae, they do produce a wide range of

1213: variability, including flickering, orbital and even sizeable long-term

1214: variations (e.g. Honeycutt, Robertson \& Turner 1998; Honeycutt \&

1215: Kafka 2004; Knigge et al. 1998, 2000, 2004). 

1216: 

1217: 

1218: \subsection{Star~4: A Helium WD}

1219: 

1220: \label{sec:star4}

1221: 

1222: \begin{figure*} 

1223: \center

1224: \includegraphics[angle=-90,width=1.0\textwidth]{f13.eps}

1225: \caption{The FUV spectrum (top panel) and broad-band SED (bottom

1226: panel) of Star~4, along with our best-bet model fit to the

1227: data (see text for details). Notation is as in 

1228: Figure~\protect\ref{fig:star1}. Star~4 is a Helium white dwarf.}

1229: \label{fig:star4}

1230: \end{figure*}

1231: 

1232: Star~4 is a gap object photometrically, but one that lies close to the WD

1233: cooling sequence in Figure~\ref{fig:cmd}. The FUV spectrum and

1234: broad-band SED in Figure~\ref{fig:star4} are blue and featureless, but

1235: it is impossible to fit these data with an ordinary WD model. The

1236: observed spectral slope implies an effective temperature of around

1237: 20,000~K, but a normal WD with this temperature is much too faint to

1238: match the observed flux. Instead, the required stellar radius is about 

1239: 0.05~$R_{\odot}$, 3 times larger than the radius of a CO-core WD. 

1240: 

1241: Such a radius is exactly what is expected for low-mass He-core WDs in

1242: GCs (Serenelli et al. 2002). In fact, both the radius and 

1243: temperature we infer for Star~4 are extremely similar to those of 

1244: the He WD in NGC~6397 found by Edmonds et al. (1999). Motivated by

1245: this, we fix the surface gravity of our WD model fit at $\log{\;g} =

1246: 6.25$ (Edmonds et al. 1999) and then find an excellent,

1247: self-consistent description of the entire data set with $T_{eff} =

1248: 21,000$~K, $M = 0.2~M_{\odot}$ and $R_{\odot} = 0.055~R_{\odot}$. 

1249: 

1250: We conclude that Star~4 is a He WD in 47 Tuc. This is only the second

1251: such object to be optically identified in this cluster, the other

1252: being the low-mass companion to millisecond pulsar MSP-U (Edmonds et

1253: al. 2001). Indeed, only a handful of He WDs have so far been optically

1254: detected in all GCs combined, and the only ones not associated with

1255: MSPs or ultra-compact X-ray binaries (e.g. Dieball et al. 2005b) are

1256: the so-called  ``non-flickerers'' in NGC~6397 (Cool et al. 1998;

1257: Edmonds et al. 1999) and the double-degenerate binary V46 in M4

1258: (O'Toole et al. 2006). 

1259: 

1260: In order to form a He WD, its progenitor must lose sufficient mass to

1261: avoid He ignition at the top of the RGB. The obvious way to truncate

1262: the progenitor evolution is via mass loss during a mass-transfer

1263: episode in a binary system. In a GC, He WDs can therefore be formed

1264: either in primordial hard binaries or in binaries formed by dynamical

1265: encounters (most likely exchange interactions). The channels available

1266: for He WD formation in 47 Tuc have been discussed in detail by

1267: Hansen, Kalogera \& Rasio (2003). They find that both the primordial

1268: binary and exchange interaction channels can produce He WDs in 47

1269: Tuc. In either case, the companion of the He WD is likely to be a NS

1270: or CO-core WD. However, in a dense GC core, any He WD in a binary system

1271: will be vulnerable to ejection during subsequent exchange

1272: encounters. Thus isolated He WDs could be produced in GCs by dynamical

1273: processes.

1274: 

1275: In the case of Star~4, there is certainly no sign of any binary

1276: companion in the SED. However, only a radial velocity study will allow

1277: us to determine if this is because the He WD is currently single or

1278: because it has a dark, massive (NS or WD) companion. As noted by

1279: Hansen, Kalogera \& Rasio (2003), He WDs are unique tracers of the

1280: compact object populations in GCs and can provide insight into the

1281: dynamical processes that drive GC evolution. Thus follow-up studies of

1282: the known He WDs in GCs (including Star~4), as well as searches for

1283: additional candidates, are likely to be extremely valuable.

1284: 

1285: %6752 -- Ferraro et al. 2003; Cocozza et al. 2006; Bassa et al. 2006 (MSP; association with cluster uncertain)

1286: %M15 -- maybe Dieball....

1287: %6397 -- Cool et al. 1998, Taylor et al. 2001; Edmonds et al. (1999)

1288: %47 Tuc -- MSPs (e.g. Camilo et al. 2000) -- optically, MSP-U (Edmonds

1289: %et al. 2001) -- Hansen, Kalogera & Rasio: could have He WDs formed by

1290: %binary, exchange interactions involving NS and CO white dwarfs.

1291: %M4 -- possibly V46 (O'Toole et al. 2005)

1292: 

1293: 

1294: \subsection{Star~5: A Simple WD}

1295: \label{sec:star5}

1296: 

1297: \begin{figure*} 

1298: \center

1299: \includegraphics[angle=-90,width=1.0\textwidth]{f14.eps}

1300: \caption{The FUV spectrum (top panel) and broad-band SED (bottom

1301: panel) of Star~5, along with our best-bet model fit to the

1302: data (see text for details). Notation is as in 

1303: Figure~\protect\ref{fig:star1}. Star~5 is a white dwarf.}

1304: \label{fig:star5}

1305: \end{figure*}

1306: 

1307: Star~5 is a representative example of hot, young WDs in 47~Tuc 

1308: (c.f. Figure~\ref{fig:cmd}. The FUV spectrum and (very  

1309: limited) broad-band photometry are shown in Figure~\ref{fig:star5}. The

1310: object is too faint to be detected/measured in most of our optical

1311: images, but all of the data are consistent with the SED expected for a

1312: single, hot ($T_{eff} \simeq 39,000$~K) CO-core WD near the top of the

1313: cooling sequence.

1314: 

1315: \subsection{Star~7: A WD-MS Binary}

1316: \label{sec:star7}

1317: 

1318: \begin{figure*} 

1319: \center

1320: \includegraphics[angle=-90,width=1.0\textwidth]{f15.eps}

1321: \caption{The FUV spectrum (top panel) and broad-band SED (bottom

1322: panel) of Star~7, along with our best-bet model fit to the data (see

1323: text for details). Notation is as in

1324: Figure~\protect\ref{fig:star1}. Star~7 is a WD-MS binary system.}  

1325: \label{fig:star7}

1326: \end{figure*}

1327: 

1328: Star~7 was originally included in this section as another

1329: representative hot, young WD. However, despite its location close to

1330: the WD cooling sequence in Figure~\ref{fig:cmd}, the broad-band

1331: photometry in Figure~\ref{fig:star7} reveals a composite SED. As

1332: expected from the CMD location, the WD dominates at short wavelengths

1333: ($\lambda \ltappeq 4000$~\AA). Beyond this, there is a red excess that

1334: is well described by a 0.6~$M_{\odot}$\ MS star in the cluster. Even

1335: though the expected number of false matches among FUV sources brighter

1336: than $m_{FUV} = 19.5$\ is only $\simeq 0.3$\ (Section~\ref{sec:cmd}), it is,

1337: of course, impossible to rule out completely that the composite SED

1338: could arise from a chance coincidence between the WD and a low-mass MS

1339: star. However, we see little evidence for blending at

1340: intermediate wavelengths, where both components contribute

1341: significantly to the total flux. We therefore believe Star~7 is 

1342: likely to be a genuine WD-MS binary in 47~Tuc. 

1343: 

1344: \subsection{Star~10 = PC1-V36: A Close Binary with a Dark Primary and a Stripped Secondary}

1345: \label{sec:star10}

1346: 

1347: \begin{figure*} 

1348: \center

1349: \includegraphics[angle=-90,width=1.0\textwidth]{f16.eps}

1350: \caption{The FUV spectrum (top panel) and broad-band SED (bottom

1351: panel) of Star~10, along with our best-bet model fit to the

1352: data (see text for details). Notation is as in 

1353: Figure~\protect\ref{fig:star1}. Star~10 is a binary system whose SED

1354: is dominated by a peculiar low-mass secondary star. This secondary is

1355: probably the remnant of a subgiant that has been stripped of its

1356: envelope.}  

1357: \label{fig:star10}

1358: \end{figure*}

1359: 

1360: Star~10 = PC1-V36 is arguably the most interesting and exotic object in our

1361: entire sample. In Figure~\ref{fig:cmd}, it lies squarely in the gap

1362: between the WD and main sequences, but we already noted in

1363: Section~\ref{sec:gap_specs} that its FUV spectrum is rather

1364: peculiar. In Figure~\ref{fig:star10}, we show this FUV spectrum again,

1365: but now alongside with the broad-band SED we have constructed from the

1366: ACS/HRC images. 

1367: 

1368: In one sense, the spectrum and FUV-NIR SED are extremely simple: a

1369: single spectral component with $T_{eff} = 8700 \pm 40$~K, $\log{\;g} =

1370: 3.63 \pm 0.15$ and $R = 0.60\pm 0.01~R_{\odot}$ (corresponding to a

1371: mass of $M = 0.056 \pm 0.018~M_{\odot}$) provides an excellent

1372: description of our entire spectroscopic and photometric data

1373: set. 

1374: \footnote{We note in passing that an optical spectrum of Star~10 has

1375: been presented and analyzed by De Marco et al. (2005). Their preferred

1376: parameters for this object (N104-3 

1377: in their notation) are $T_{eff} = 7650 \pm 250$~K, $\log{\;g} \simeq 4.2

1378: \pm 0.3$, $R = 0.9 \pm 0.2~R_{\odot}$. Such a model is 

1379: inconsistent with the data presented here. For example, a star with

1380: these parameters would be more than 1.5~mag fainter in the FUV

1381: waveband than we observe. This discrepancy could be a sign of

1382: large-amplitude variability on long time scales. However, our own SED

1383: shows little sign of this, despite being constructed from 

1384: non-simultaneous observations (see

1385: Figure~\protect\ref{fig:star10}). Another possibility is that the

1386: spectrum used by De Marco et al. (2005) was affected by blending, which

1387: could bias the inferred stellar parameters. While it would clearly be

1388: important to distinguish between these alternatives, doing so is

1389: beyond the scope of the present paper.} However, this combination of

1390: parameters obviously does not describe any kind of normal star in

1391: 47~Tuc. This reflects the highly unusual SED of Star~10, which

1392: does not appear to be composite, yet is clearly much bluer than that

1393: of normal main sequence, red giant and horizontal branch stars in 47

1394: Tuc, and much redder than that of hot WDs. Note that we provide errors

1395: here, because the parameters of this object are so unusual and 

1396: there is no evidence for systematic residuals that might seriously

1397: bias the fit. Also, unlike for most other objects, no parameters were

1398: constrained {\em a priori} in the fit.  

1399: 

1400: Perhaps the most obvious question raised by the strange SED is

1401: whether Star~10 is actually a cluster member. Fortunately, this object 

1402: is the only non-BS in our sample for which a reliable proper motion

1403: measurement is available in McLaughlin et al. (2006). As shown in

1404: Figure~\ref{fig:proper}, these data are entirely consistent with

1405: membership of 47~Tuc.

1406: 

1407: Additional confirmation of the inferred parameters comes from the

1408: variability of the object. Albrow et al. (2001) found a periodic

1409: signal with $P = 0.4$~d in their optical time-series photometry for

1410: Star~10 (=~PC1-V36 in their notation). The light curve shape was

1411: consistent with ellipsoidal modulations, in 

1412: which case the orbital period of the binary system would be $P_{orb} =

1413: 0.8$~d. If this object {\em is} an ellipsoidal variable, the star that

1414: dominates the SED must be Roche lobe-filling and should obey the

1415: orbital period-mean density relationship given in 

1416: Equation~\ref{eq:rho}. Combining $P_{orb} = 0.8$~d with our

1417: well-determined radius of $R = 0.60~R_{\odot}$, this relationship

1418: predicts a mass of $M = 0.045~M_{\odot}$. This is completely

1419: consistent with the mass inferred from the spectral/SED fit. 

1420: We conclude that Star~10 is a 0.8~d binary system containing a 

1421: bright, low-mass secondary star that fills (or nearly fills) its Roche

1422: lobe. There is no sign of the primary in the SED, which rules out

1423: MS-TO stars and  other optically bright objects.  

1424: 

1425: With an effective temperature higher than any BS, a mass below the

1426: Hydrogen-burning limit, and a radius comparable to a MS star, the

1427: secondary in the Star~10 binary system must be in an extreme and

1428: short-lived evolutionary state. In our view, the most likely 

1429: interpretation is that the secondary star is the remnant of a

1430: subgiant whose Hydrogen envelope has been almost completely

1431: stripped off. A similar description has been suggested for the

1432: donor star in the X-ray binary AC~211 in M15 (van Zyl et al. 2004). 

1433: The requisite stripping could have occurred as a result

1434: of mass transfer in a primordial or dynamically-formed binary

1435: system. Alternatively, and more excitingly, it might have occurred

1436: during the very dynamical encounter that produced the binary

1437: system (such as a physical collision with a NS; e.g. Lombardi et

1438: al. 2006). In this case, Star~10 might be the ``smoking gun'' of a 

1439: recent dynamical encounter involving a compact object. 

1440: 

1441: 

1442: \subsection{Star~15: A WD-Subgiant Binary?}

1443: \label{sec:star15}

1444: 

1445: \begin{figure*} 

1446: \center

1447: \includegraphics[angle=-90,width=1.0\textwidth]{f17.eps}

1448: \caption{The FUV spectrum (top panel) and broad-band SED (bottom

1449: panel) of Star~15, along with our best-bet model fit to the data (see

1450: text for details). Notation is as in

1451: Figure~\protect\ref{fig:star1}. Star~15 is either a WD-subgiant binary

1452: system or a chance superposition of two such stars.}

1453: \label{fig:star15}

1454: \end{figure*}

1455: 

1456: Star~15 is a gap source and located very close to Star~10~=~PC1-V36 in

1457: the FUV/optical CMD (Figure~\ref{fig:cmd}). However, its FUV spectrum and

1458: broad-band SED in Figure~\ref{fig:star15} reveal that its properties

1459: are completely different from those of Star~10. Our suggested model

1460: for this system consists of a hot WD (which dominates the FUV) and a

1461: MS-TO/subgiant secondary star (which dominates everything else). This

1462: describes 

1463: the data fairly well, although the observed FUV spectrum is somewhat

1464: flatter than expected for a WD that matches the FUV flux level. We

1465: have not attempted to fit any more complicated models, since the FUV

1466: spectrum suffers from at least mild blending across much of the

1467: spectral range.  

1468: 

1469: As usual, we have to allow for the possibility that the apparently

1470: composite nature of Star~15 is due to a chance coincidence. Even

1471: though the expected number of false matches among FUV sources brighter

1472: than $m_{FUV} = 19.5$\ is only $\simeq 0.3$\ (Section~\ref{sec:cmd}), 

1473: this concern deserves particular attention in this case, because the offset

1474: between FUV and optical positions listed in Table~\ref{tab:specdat} is

1475: relatively large (1.29 FUV pixels). While this is well within the

1476: hard cut-off adopted in Paper~I (1.5~pixels) that was also used for

1477: estimating the expected number of chance coincidences, the probability

1478: of a {\em given} match being due to chance clearly increases with

1479: increasing offset. Based on careful visual inspection and  

1480: blinking of the FUV and optical images, we think there is a strong 

1481: possibility that the FUV and optical sources associated with Star~15 

1482: will turn out to be unrelated, altough it is impossible to be

1483: certain. In order to flag this, we have 

1484: therefore added a ``?''  to the spectroscopic/SED classification in

1485: Table~\ref{tab:specdat}.

1486: 

1487: 

1488: \subsection{Star~17: A Probable WD-Subgiant Binary and Possible CV}

1489: \label{sec:star17}

1490: 

1491: 

1492: \begin{figure*} 

1493: \center

1494: \includegraphics[angle=-90,width=1.0\textwidth]{f18.eps}

1495: \caption{The FUV spectrum (top panel) and broad-band SED (bottom

1496: panel) of Star~17, along with our best-bet model fit to the data (see

1497: text for details). The data point near 4800~\AA\ was taken from

1498: McLaughlin et al. (2006) and is therefore shown with a different

1499: symbol. Other notation is as in

1500: Figure~\protect\ref{fig:star1}. Star~17 is a probable WD-subgiant

1501: binary system  and possible CV (based on the tentative detection of

1502: C~{\sc iv} in emission).}

1503: \label{fig:star17}

1504: \end{figure*}

1505: 

1506: Star~17 is another gap source and located close to Stars~10 and 15 in

1507: the FUV/optical CMD (Figure~\ref{fig:cmd}). Its broad-band SED

1508: (Figure 18, bottom panel) is also

1509: very similar to that of Star~15 and suggests that Star~17 is another

1510: WD-subgiant binary system.  However, as already noted in

1511: Section~\ref{sec:gap_specs}, there is marginal evidence that C~{\sc

1512:   iv} 1550~\AA\ is in emission in this sytem (Figure 18, top panel). The

1513: detection is not 

1514: compelling, partly because the line happens to lie close to the edge

1515: of the detector and is therefore cut off, and partly because the

1516: spectrum of Star~17 suffers from moderate blending. 

1517: 

1518: In assessing the likelihood that Star~17 may be a previously unknown

1519: CV in 47~Tuc, it is worth noting that this FUV source was not found

1520: to be variable in Paper~I, and is also not a viable counterpart to any

1521: of the Chandra X-ray sources in Heinke et al. (2005). On balance, we

1522: are currently skeptical that this system will be confirmed as

1523: a CV in the future. 

1524: 

1525: The status of Star~17 as a WD-subgiant binary is somewhat more secure,

1526: although here again we have to consider the possibility that the match

1527: between FUV and optical sources is due to a chance coincidence. We

1528: note again that the expected number of false matches among FUV sources

1529: brighter than $m_{FUV} = 19.5$\ is only $\simeq 0.3$\ 

1530: (Section~\ref{sec:cmd}). However, as in the case of Star~15, the

1531: offset between FUV and optical positions is relatively large for

1532: Star~17 (1.20~FUV pixels), so the possibility of a false match

1533: deserves to be taken seriously nonetheless. We have therefore again

1534: inspected and blinked the FUV and optical images in the vicinity of

1535: Star~17. While we cannot rule out the possibility of a random match

1536: based on this, we do think Star~17 is more likely to be a genuine

1537: match than Star~15. Statistically, it would certainly be rather

1538: surprising if {\em both} Stars~15 and 17 turned out to be chance

1539: coincidences. Nevertheless, we have again conservatively marked the

1540: binary classification for Star~17 in Table~\ref{tab:specdat} with a

1541: ``?''.

1542: 

1543: 

1544: \subsection{Star~20~=~V2: A CV with a Main-Sequence Donor}

1545: \label{sec:star20}

1546: 

1547: 

1548: \begin{figure*} 

1549: \center

1550: \includegraphics[angle=-90,width=1.0\textwidth]{f19.eps}

1551: \caption{The FUV spectrum (top panel) and broad-band SED (bottom

1552: panel) of Star~20, along with our best-bet model fit to the data (see

1553: text for details). Notation is as in

1554: Figure~\protect\ref{fig:star1}. Star~20 is a CV.}

1555: \label{fig:star20}

1556: \end{figure*}

1557: 

1558: Star 20~=~V2 is a previously known CV that was originally discovered as an

1559: erupting dwarf nova by Paresce \& De Marchi (1994). A second outburst

1560: was found by Shara et al. (1996). V2 is also known to be an X-ray source

1561: (Grindlay et al. 2001a), as well as a UV-excess object located in the

1562: gap region of the CMD (Ferraro et al. 2001; Knigge et

1563: al. 2002). 

1564: 

1565: The FUV spectrum and broad-band SED we have obtained for V2 are shown

1566: in Figure~\ref{fig:star20}. As already noted in

1567: Section~\ref{sec:gap_specs}, the presence of C~{\sc iv}~1550~\AA\ and

1568: He~{\sc ii}~1640~\AA\ emission lines spectroscopically confirms the CV

1569: classification of this source. The SED can be decomposed into a hot

1570: component with parameters appropriate for a massive WD, and a cool

1571: component with parameters appropriate for a low-mass MS star in 47

1572: Tuc. 

1573: 

1574: Edmonds et al. (2003b) found evidence for variability on periods of 

1575: 6.8~hrs and 3.0~hrs in their optical photometry for V2. Allowing for

1576: the possibility that one of the signals is ellipsoidal in nature (in

1577: which case the orbital period is twice the observed period), this

1578: means that there are four candidate orbital periods for this

1579: system. For comparison, the donor parameters suggested by our SED fit

1580: (in which we constrained the donor to lie on the cluster MS), would

1581: imply an orbital period of 4.8~hrs (via Equation~\ref{eq:rho}). We

1582: will not pursue this issue further here, since neither the orbital

1583: period nor our estimate of the donor properties is secure. We do note

1584: that donors with significantly higher mass than that implied by our

1585: fit are excluded, since they would be brighter than the observed SED

1586: at optical and NIR wavelengths. Thus the broader classification of the

1587: secondary as a lower MS star {\em is} fairly secure. 

1588: 

1589: In the process of constructing our broad-band SED, we also found

1590: evidence of two additional eruptions of V2, which are shown in

1591: Figure~\ref{fig:v2_outburst}. First, the ACS/HRC/F330W data

1592: point in Figure~\ref{fig:star20} is 1.7~mag fainter than the older

1593: WFPC2/PC/F336W measurement that was used in the construction of the

1594: CMD (Figure~\ref{fig:cmd}. This suggests that V2 was in outburst in

1595: July 1999, the epoch that dominates in our co-added PC/F336W

1596: image. This is consistent with the fact that Ferraro et al. (2001)

1597: report an even brighter WFPC2/PC/F336W magnitude for V2 from data

1598: obtained exclusively in July 1999. The fainter ACS/HRC/F330W

1599: measurement would move V2 much closer to the WD sequence in

1600: Figure~\ref{fig:cmd}. Second, there is a 1.0 mag difference between

1601: the two ACS/HRC/F625 magnitude estimates we have for this source. The

1602: bright estimate comes from an image taken in March 2005, which is the

1603: only image obtained at this time in our optical data set. The

1604: discovery of two eruptions in our limited optical data set confirms

1605: that V2 is a dwarf nova with a relatively high duty cycle (c.f. Shara

1606: et al. 1996).  

1607: 

1608: 

1609: \begin{figure*} 

1610: \center

1611: \includegraphics[angle=0,width=0.95\textwidth]{f20a.eps}

1612: \includegraphics[angle=0,width=0.95\textwidth]{f20b.eps}

1613: \caption{

1614: {\em Top Row:} Approximately U-band observations of

1615: Star~20~=~V2 obtained with {\em HST} in July 1999 (left panel) and July 2002

1616: (right panel). The difference in brightness is obvious and implies

1617: that V2 experienced a dwarf nova eruption in or near July 1999.

1618: {\em Bottom Row:} Approximately R-band observations of

1619: Star~20~=~V2 obtained with {\em HST} in April 2002 (left panel) and March

1620: 2005 (right panel). The difference in brightness is obvious and implies

1621: that V2 experienced a dwarf nova eruption in or near March 2005. In all panels, 

1622: north is up, east is to the left, and the field of view is approximately 

1623: 1.8\arcsec.}   

1624: \label{fig:v2_outburst}

1625: \end{figure*}

1626: 

1627: 

1628: \subsection{Star~27: An SMC Interloper}

1629: \label{sec:star27}

1630: 

1631: \begin{figure*} 

1632: \center

1633: \includegraphics[angle=-90,width=1.0\textwidth]{f21.eps}

1634: \caption{The FUV spectrum (top panel) and broad-band SED (bottom

1635: panel) of Star~27, along with our best-bet model fit to the data (see

1636: text for details). Notation is as in

1637: Figure~\protect\ref{fig:star1}. Star~27 is a massive star that is

1638: located behind 47~Tuc in the SMC}

1639: \label{fig:star27}

1640: \end{figure*}

1641: 

1642: Star~27 is another gap source in Figure~\ref{fig:cmd}, and its FUV

1643: spectrum and broad-band SED are shown in

1644: Figure~\ref{fig:star27}. Despite several attempts, we were unable to

1645: find a physically plausible one- or two-component fit to this data

1646: that was consistent with the source being a cluster member. 

1647: 

1648: We therefore considered the possibility that Star~27 is a background

1649: star. As discussed in Paper~I, the outskirts of the SMC are located

1650: directly behind 47~Tuc, and roughly one SMC interloper may be expected to

1651: contaminate our FUV/optical CMD. Figure~\ref{fig:cmd} shows that

1652: Star~27 is a good candidate, since it lies close to the expected

1653: location of the SMC main sequence. Note that this object is not

1654: included in Figure~\ref{fig:proper}, since there is no proper motion

1655: information for it in McLaughlin et al. (2006). Thus a location in the

1656: SMC cannot be excluded for Star~27.

1657: 

1658: In order to test this idea, we changed the distance adopted in

1659: our fits to 60~kpc (the average metallicity in the SMC is similar to

1660: that of 47~Tuc). We then find that a single-component model with $T_{eff}

1661: \simeq 10,500$~K, $R = 1.6~R_{\odot}$ and $\log{\;g} = 4.3$ provides a

1662: good description of all of the datal. These parameters are entirely

1663: reasonable for a slightly evolved $2 M_{\odot}$\ star in the SMC. We

1664: therefore conclude that Star~27 is an SMC interloper and unrelated to

1665: 47~Tuc. 

1666: 

1667: \subsection{Star~999: A Bright and Probably Massive Blue Straggler}

1668: \label{sec:star999}

1669: 

1670: \begin{figure*} 

1671: \center

1672: \includegraphics[angle=-90,width=1.0\textwidth]{f22.eps}

1673: \caption{The FUV spectrum (top panel) and broad-band SED (bottom

1674: panel) of Star~20, along with two model fits to the data (see

1675: text for details). Notation is as in

1676: Figure~\protect\ref{fig:star1}. Star~999 is a blue straggler.} 

1677: \label{fig:star999}

1678: \end{figure*}

1679: 

1680: Star~999 is the optically brightest BS in our spectroscopic

1681: sample. Its FUV spectrum and broad-band SED are shown in

1682: Figure~\ref{fig:star999}, and imply $T_{eff} \simeq 7500$~K and $R

1683: \simeq 3~R_{\odot}$\ for this object. These are reasonable parameters

1684: for a BS in 47~Tuc. However, in order to obtain an acceptable single

1685: component fit to all of the data, a surprisingly low surface gravity

1686: is also required ($\log{\;g} \simeq 3.0$; red line in

1687: Figure~\ref{fig:star999}. Taken at face value, this combination of

1688: radius and surface gravity would point to an extremely low mass of $M

1689: \simeq 0.3~M_{\odot}$\ for Star~999. This does not seem reasonable for

1690: a BS.

1691: 

1692: There could be physical explanations for such an abnormally low

1693: surface gravity, such as fast rotation (Porter \& Townsend 2005;

1694: Knigge et al. 2006) or the presence of a disk around the star (De

1695: Marco et al. 2004). However, before resorting to such explanations, it

1696: is worth checking that the spectral models we are using can actually

1697: be trusted in the regime we are using them here. The obvious worry

1698: here is the reliability of the models in FUV region, since this

1699: waveband is on exponential Wien tail of the BS SED where even

1700: relatively slight inaccuracies in the model may result in large flux

1701: changes.  

1702: 

1703: The work by Castelli \& Cacciari (2001) suggests that this worry is

1704: well founded. They carried out an extensive FUV analysis of Pop~II

1705: A-stars, based on the same set of Kurucz models that we employ 

1706: here. Their main finding was that, for $T_{eff} \ltappeq 8700$~K,

1707: model fits to the FUV spectra produce biased surface gravity estimates 

1708: that are systematically lower than those implied by optical

1709: data. For hotter stars, there was no systematic discrepancy between

1710: the two sets of estimates.

1711: 

1712: As a test, we therefore repeated the one-component fit, but excluded

1713: the FUV spectrum and photometry. The result of this exercise 

1714: confirmed our suspicions: the best model fit to the NUV-NIR SED (the

1715: blue line in Figure~\ref{fig:star999}) has a similar temperature and

1716: radius to the global best fit, but a much higher surface gravity

1717: ($\log{\;g} = 3.8$). The BS mass implied by this fit is therefore also

1718: much higher ($M \simeq 1.9~M_{\odot}$). 

1719: 

1720: Two points regarding the fits carried out in previous sections are

1721: worth making at this stage. First, the effective  temperature of

1722: Star~10 is just about high enough for our model fit to 

1723: yield a meaningful surface gravity estimate. Second, the FUV excess seen

1724: in the BS Star~2 (and attributed to a WD companion) cannot be

1725: attributed to a similar model failure. This can be seen by comparing

1726: the FUV properties of Stars~2 and 999 directly. The broad-band FUV

1727: magnitude of both stars is dominated by a BS with $T_{eff} \simeq

1728: 7500$~K, with Star~2 being only 0.2~mag brighter than

1729: Star~999. However, at the shortest wavelengths available in the FUV

1730: spectra (1500\AA\ - 1600\AA), 

1731: Star~2 is a full 2~mag brighter than Star~999. Thus the massive FUV

1732: excess exhibited by Star~2 is not just relative to theoretical models,

1733: but also relative to other BSs in the cluster. 

1734: 

1735: Returning to Star~999 itself, it is interesting to note that the mass

1736: inferred from our fit to the NUV-NIR SED is just over twice the

1737: turn-off mass in 47~Tuc ($M_{TO} \simeq 0.9~M_{\odot}$). This is 

1738: interesting, since the formation of a BS with $M > 2 M_{TO}$\ must

1739: involve three progenitors. Such a 

1740: ``supermassive'' BS (SMBS) cannot be formed in a single collision

1741: between two MS stars, nor via mass transfer in an ordinary binary

1742: system. Unfortunately, the uncertainties on the mass of Star~999 are

1743: too large to establish it as a SMBS (the statistical error alone is 

1744: about 0.4~$M_{\odot}$). The sample of BSs analysed by De Marco et

1745: al. (2005) also contained several candidate SMBSs (including one in

1746: NGC~6793 for which Saffer et al. [2002] had derived a preliminary mass

1747: estimate of about $3~M_{TO}$). However, none of these candidates could

1748: be confirmed conclusively as SMBSs. At the moment, this leaves the

1749: BS-MS binary discovered by Knigge et al. (2006) as the 

1750: most convincing candidate for a BS system with three progenitors in a GC

1751: core. However, the first definitive SMBS remains to be discovered. 

1752: 

1753: \section{Discussion}

1754: \label{sec:discuss}

1755: 

1756: Since we have discovered and analysed a fair number of interesting and

1757: exotic objects in 47~Tuc, it is worth considering some of the wider 

1758: implications of our results. In the following sections, we will therefore

1759: take a look at three key issues: (i) the binary fraction in 47~Tuc;

1760: (ii) the nature and abundance of CVs in GCs; (iii) the significance of

1761: systems that are unique even by the standard of most other exotica

1762: (such as Star~10 in our sample).

1763: 

1764: 

1765: \subsection{The White Dwarf Binary Fraction in 47~Tuc}

1766: \label{sec:binfraction}

1767: 

1768: We can use our results to obtain a rough estimate of the binary

1769: fraction amongst hot, young WDs in 47~Tuc. As a starting point, we

1770: note that our FUV photometry 

1771: includes 25 objects brighter than $m_{FUV} = 19.5$\ and within our

1772: spectroscopic field of view. Figure~\ref{fig:cmd} shows that, above

1773: this limit, we have useable spectra for all gap objects, all but one

1774: BSs and all objects on the WD cooling sequence. Thus we use $m_{FUV} =

1775: 19.5$\ as an approximate completeness limit for the discovery of WD

1776: binary systems in our spectroscopic sample. 

1777: 

1778: There are 5 objects without optical counterparts above this limit, as

1779: well as 5 objects on the WD cooling sequence for which we have no

1780: evidence of a companion. As a check, we visually inspected the

1781: vicinity of each of these 10 FUV sources in the deepest red optical 

1782: image available. For most objects, this was a 600~s F814W exposure. In

1783: a few cases, the location of interest fell outside the F814W field of

1784: view or on an image defect, so we used a 

1785: 350~s F555W exposure instead.\footnote{One object -- Star~6 -- was

1786: outside the field of view of both images, so in this case we

1787: inspected a shallower 60~s F814W exposure.} No additional red

1788: counterparts were found, down to an estimated limit of $m_{STMAG}

1789: \simeq 21$\ or better. This excludes the existence of MS companions

1790: as bright as that in Star~7.

1791: 

1792: Thus there are probably 10 single WDs in our bright ($m_{FUV} <

1793: 19.5$) sub-sample. This same sub-sample also contains two very

1794: probable binary systems containing hot, young WDs 

1795: (Stars~2 and 7), and one additional possible one (Star~15). CVs have

1796: been excluded here, because their WDs are probably 

1797: accretion-heated, rather than young. We also exclude Star~17, partly

1798: because it is a candidate CV, and partly because it, too, could be a

1799: chance superposition. A rough estimate 

1800: of the binary fraction amongst hot WDs in the core of 47~Tuc is then

1801: 2/13 (assuming that Star~15 is {\em not} a binary), i.e. $f_{bin}=

1802: 15\%~^{+17\%}_{-9\%}~{\rm (stat)}~^{+8\%}_{-7\%}~{\rm (sys)}$. The

1803: statistical error on this estimate is solely due to small 

1804: number statistics, whereas the quoted systematic error shows how the

1805: maximum-likelihood estimate would change if the true number of WD

1806: binaries in our sample was 1 or 3, instead of 2.

1807: 

1808: Clearly, this estimate needs to be viewed with great caution. Aside

1809: from the issues of small number statistics and chance coincidences, we 

1810: cannot rule out the possibility of faint, lower MS companions to some

1811: of the apparently single WDs in our sample. On the other hand, the

1812: number of WDs (objects near the 

1813: cooling sequence in Figure~\ref{fig:cmd}, as well as objects without a 

1814: F336W counterpart) rises considerably more steeply towards fainter

1815: magnitudes than the numbers of gap sources and BSs. Since Stars~2 and

1816: 15 were found amongst the latter two classes, our binary candidate

1817: haul may have been unusually lucky.  

1818: 

1819: While the allowed range we have determined is not yet

1820: very constraining, it is certainly consistent with the much more accurate

1821: estimates of the binary fraction determined by 

1822: Albrow et al. (2001) in the core of 47~Tuc. They found $13\% \pm6\%

1823: $\ based on the number of eclipsing binaries in their data, and

1824: $14\%\pm4\% $\ based on the number of W~UMa  

1825: stars. Even though the estimate presented here is much less accurate,

1826: it is the first to be derived from the WD  population of a GC and

1827: demonstrates the 

1828: potential of FUV surveys in this area. Given the importance of core

1829: binary fractions to cluster dynamics -- and the present controversy

1830: surrounding the theoretical predictions (Ivanova et al. 2005; Hurley,

1831: Aarseth \& Shara 2007; Fregeau 2007) -- more accurate WD-based

1832: estimates could be extremely valuable. 

1833: 

1834: \subsection{Cataclysmic Variables in 47 Tuc and other GCs}

1835: \label{sec:CVs}

1836: 

1837: Theoretical models for CV formation in GCs predict that a cluster like

1838: 47~Tuc should harbour $\sim 200$, mostly dynamically-formed CVs (di

1839: Stefano \&  Rappaport 1994; Ivanova et al. 2006; but also see Townsley

1840: \& Bildsten 2005). Even though 

1841: recent surveys have at last begun to uncover a sizeable population of

1842: CV candidates in GCs, there is still a clear shortfall in the observed

1843: numbers relative to these predictions. For example, the combination

1844: of deep Chandra X-ray surveys and optical follow-up has so far led

1845: to the identification of $\sim 20$~probable CVs in 47~Tuc (Grindlay et

1846: al. 2001a; Edmonds et al. 2003ab; Heinke et al. 2005). However, it is

1847: still not clear if this discrepancy points to a problem with our

1848: understanding of CV formation and/or evolution, or if it is simply due 

1849: to observational incompleteness (see Heinke et al. 2005, Ivanova et

1850: al. 2006 and Maccarone \& Knigge 2007 for discussions of this point). 

1851: 

1852: The present results are relevant to this issue for several

1853: reasons. First, the discovery of emission lines in three previously

1854: suggested CV candidates in our spectroscopic sample 

1855: (Star~1~=~AKO~9; Star~3~=~V1; Star~20~=~V2) confirms that combined X-ray-,

1856: colour- and variability-based searches are, in fact, finding genuine

1857: CVs. This is rather important, since only seven other CVs have so far

1858: been spectroscopically confirmed in all Galactic GCs combined.

1859: \footnote{These are CVs 1-4 in NGC~6397 (Grindlay et al. 1995; Edmonds

1860:   et al. 1999), V101 in M5 (Margon, Downes \& Gunn 1981; Naylor et

1861:   al. 1989; Shara, Moffat \& Potter 1990); Nova 1938 in M4 (Shara,

1862:   Moffat \& Potter 1990); and Star~1 in NGC~6624 (Deutsch et

1863:   al. 1999). Note that the spectroscopically confirmed dwarf nova V4

1864:   in the field of M30 (Margon \& Downes 1983; Shara, Moffat \& Potter

1865:   1990) is probably a foreground object (Machin et al. 1991).}

1866: The only good CV candidate that was not confirmed spectroscopically is

1867: Star~10 = PC1-V36. This object was highlighted as a possible CV in

1868: Paper~I, based on 

1869: its FUV brightness, variability and possible match to an X-ray

1870: <source.\footnote{The status of Star~10=PC1-V36 as an X-ray

1871: emitter remains uncertain: see the ``note added'' in Paper~I, as well

1872: as Edmonds et al. (2003ab) and Heinke et al. (2005).}

1873: As discussed in Section~\ref{sec:star10}, we now find that Star~10 is

1874: an even more exotic system. 

1875: 

1876: Second, we have detected the secondary stars in two of our CVs. In

1877: Star~1~=~AKO~9, the donor is a subgiant, whereas in Star~20~=~V2, it

1878: is a lower-MS star. Moreover, our SED fit to Star~3~=~V1 constrains

1879: the secondary in this systems to be a lower-MS star also. In principle,

1880: the distribution of donor properties in the CV population is a 

1881: strong constraint on theoretical models. For example, di Stefano \&

1882: Rappaport (1994) predict quite a sizeable population of

1883: bright, long-period CVs with evolved donors (like AKO~9) in 47~Tuc,

1884: whereas such systems are relatively rare in the simulations of

1885: Ivanova et al. (2006). As the number of known GC CVs grows, the

1886: distributions of orbital period and donor spectral type will become

1887: a powerful way to test models of CV formation and evolution.

1888: 

1889: Third, we estimated in Paper~I that $\simeq 15$~CVs should be

1890: contained in our FUV imaging survey, based on the predictions by di

1891: Stefano \& Rappaport (1994). Thus theory and observation would agree

1892: perfectly if essentially all gap objects in Figure~\ref{fig:cmd} were

1893: CVs. Even though our  spectroscopic sample comprises only half (8/16)

1894: of the photometrically identified gap objects, it is clear that this

1895: scenario is too optimistic. More specifically, even if we include the

1896: marginal detection of line emission in Star~17, only half (4/8) of the

1897: gap objects in the spectroscopic sample are confirmed as CVs. Is it 

1898: possible that some or all of the other four gap objects in the

1899: spectroscopic sample are nevertheless CVs that we have caught in a

1900: state of weak or absent line emission? We believe this is extremely

1901: unlikely. The spectroscopic/SED classifications for three of these gap

1902: objects 

1903: convincingly show that they are {\em not} CVs: Star~4 is a He~WD,

1904: Star~10 is a different kind of exotic binary containing a stripped

1905: subgiant; Star~27 is an SMC interloper. This leaves Star~15 as the

1906: only other viable candidate. However, the possibility that it is

1907: really a CV seems remote, given the lack of X-ray emission and FUV

1908: variability (Paper~I), as well as the possibility that the match

1909: between FUV and optical sources could be a chance coincidence

1910: (Section~\ref{sec:star15}). 

1911: 

1912: Overall then, our results suggest that CVs comprise no more than 

1913: $\sim 50$\% of the population in the gap, at least among the brighter objects. 

1914: Nevertheless, taken at face value, this implies that predicted and observed 

1915: numbers differ by ``only'' a factor of about two or three. However, this 

1916: conclusion is not yet completely robust, both due to the small number

1917: statistics involved, and because different theoretical models predict

1918: different CV luminosity functions and hence different numbers of

1919: observable systems (di Stefano \& Rappaport 1994; Ivanova et al. 2006). 

1920: 

1921: 

1922: \subsection{The More Exotic, the Better...}

1923: \label{sec:exotica}

1924: 

1925: We finally want to briefly highlight what are arguably the three most

1926: exotic objects we have uncovered. These are Star~2 (a BS-WD binary),

1927: Star~4 (a He WD) and Star~10~=~PC1-V36 (a binary 

1928: containing a dark primary and stripped subgiant secondary). We have

1929: already outlined possible formation scenarios for these systems in

1930: Sections~\ref{sec:star2}, \ref{sec:star4}, and \ref{sec:star10}, so we

1931: will not repeat these here. Instead, we just note that our FUV survey

1932: has turned out to be an efficient way of finding and 

1933: classifying such remarkable systems. The gap 

1934: between the WD and main sequences in the FUV-optical CMD seems to be a

1935: particularly fertile hunting ground in this respect: both Stars~4

1936: and 10 are found amongst only 8 gap sources in our spectroscopic

1937: sample.

1938: 

1939: Aside from being intrinsically interesting, the most extreme objects

1940: in GCs are important precisely because they are probably quite hard to

1941: form. Star~10 may be a good example of this. The very existence of

1942: such a system may point to a single, specific formation

1943: mechanism. This contrasts with the 

1944: relatively more common types of stellar exotica, for which there is

1945: usually more than one formation channel (for an example in the CV

1946: context, see Figure~7 in Ivanova et al. 2006). Thus a single unique

1947: object can be the smoking gun that establishes the importance of a

1948: particular dynamical process in GCs. 

1949: 

1950: \section{Summary \& Conclusions}

1951: 

1952: We have presented and analysed FUV spectroscopy for 48 blue objects in

1953: the core of 47~Tucanae. For the 12 most interesting and representative

1954: cases, we have also assembled broad-band, FUV-NIR spectral energy

1955: distributions in order to determine their nature. Our main results are

1956: as follows:

1957: 

1958: \begin{enumerate}

1959: 

1960: \item We have spectroscopically confirmed three previously known or

1961:   suspected CVs in the cluster core (V1, V2 and AKO~9). For two of

1962:   these (V2 and AKO~9), we have also found photometric evidence of

1963:   dwarf nova eruptions in addition to those that were previously

1964:   known. 

1965: 

1966: \item Only one other source -- Star~17, a ``gap object'' located

1967:   between the WD and main sequences in the CMD -- exhibits marginal evidence for line

1968:   emission in its spectrum. Thus the gap region is not exclusively

1969:   populated by CVs, nor are these systems common amongst objects near

1970:   the top of the WD cooling sequence. Nevertheless, predicted and

1971:   observed numbers of CVs agree to within a factor of about 2-3. 

1972: 

1973: \item We have discovered a hot ($T_{eff} \simeq 8700$~K), large ($R

1974:   \simeq 0.6~R_{\odot}$), but very low-mass ($M \simeq 0.06~

1975:   M_{\odot}$) secondary star in a previously known 0.8~d binary

1976:   system. This exotic object, Star~10, fills or nearly fills its

1977:   Roche-lobe and completely dominates the binary's FUV-NIR output. We suggest that

1978:   this star is the remains of a subgiant that has been almost

1979:   completely stripped of its envelope. This stripping could have

1980:   occurred as a result of mass

1981:   transfer in the binary system, or during the dynamical interaction that

1982:   actually formed the system. Since the stripped secondary must

1983:   be in a short-lived evolutionary state, this object may represent

1984:   the  ``smoking gun'' of a recent dynamical encounter. 

1985: 

1986: \item We have also found a Helium WD in 47~Tuc (Star~4). This is only

1987:   the second optically detected such object in this cluster, and the first

1988:   outside an MSP system. The He WD could have been formed in a

1989:   primordial binary, or in a binary system formed in an exchange

1990:   encounter involving a massive NS or WD. No sign of a companion is seen

1991:   in our data.

1992: 

1993: \item We have discovered a bright BS with a young WD companion

1994:   (Star~2). This is the only BS-WD binary known in any GC. However,

1995:   the apparent rarity of such objects might be a selection

1996:   effect. Since BS are much brighter than WDs at optical wavelengths, such binaries have been

1997:   hard to find until now. In our FUV spectrum, the existence of a hot

1998:   WD companion to the BS can be inferred from a strong FUV excess at the

1999:   shortest wavelengths. 

2000: 

2001: \item In addition to the objects already discussed, we have found two

2002:   more candidate WD binary systems. In one of these (Star~15), the

2003:   apparent companion is a MS star, but this composite object may well

2004:   be the product of a chance coincidence. However, the other system

2005:   (Star~7) {\em is} likely to be a genuine binary, and the companion

2006:   in this case is a sub-giant.

2007: 

2008: \item We have used the number of WD binary systems we have found to

2009:   place a crude constraint on the WD binary fraction in the core of

2010:   47~Tuc. We find $f_{bin} = 15\%~^{+17\%}_{-9\%}~{\rm stat}~^{+8\%}_{-7\%}~{\rm

2011:     sys}$. Much stronger constraints can in principle be obtained from

2012:   larger WD samples constructed from FUV surveys.

2013: 

2014: \item An SED fit to the optically brightest BS in our spectroscopic

2015:   sample suggests that the mass of this star may exceed twice the

2016:   turn-off mass. However, the uncertainties on the mass estimate are

2017:   too large for this to be conclusive. There is still no definitive

2018:   example of such a ``supermassive'' BS in any GC. 

2019: 

2020: \end{enumerate}

2021: 

2022: Overall, we feel that the present study is an excellent illustration

2023: of the wide range of stellar exotica that are lurking in the cores of

2024: GCs. In particular, the gap region of the CMD seems to harbour quite a 

2025: variety of weird and wonderful objects. Most of these systems will

2026: have undergone significant dynamical encounters or perhaps have even

2027: been formed in one. All of this reenforces the motivation for studying

2028: these systems. 

2029: 

2030: We also hope to have shown that FUV surveys -- and particularly

2031: multi-object slitless FUV specroscopy -- provide an excellent and 

2032: efficient way of achieving this goal. The power of this approach is

2033: greatest, however, when it is used in combination with data spanning

2034: the widest possible range in wavelength (e.g. the FUV-NIR range). With

2035: such panchromatic data sets, we can determine the nature of most of

2036: the exotic stellar systems that are hiding in GC cores.

2037: 

2038: \acknowledgments

2039: We would like to thank the anonymous referee of this paper for a very

2040: insightful and helpful report. 

2041: Based on observations made with the NASA/ESA Hubble Space Telescope,

2042: obtained at the Space Telescope Science Institute, which is operated

2043: by the Association of Universities for Research in Astronomy, Inc., 

2044: under NASA contract NAS 5-26555. All FUV observations are associated 

2045: with program \#8279. Support for program \#8279 was provided by NASA

2046: through a grant from the Space Telescope Science Institute, which is

2047: operated by the Association of Universities for Research in Astronomy,

2048: Inc., under NASA contract NAS 5-26555. C.K. and A.D. acknowledge

2049: support from STFC through rolling grant

2050: PP/D001013/1. J.M.A. acknowledges support from  the Spanish Government 

2051: through grants AYA2004-08260-C03, 

2052: AYA2004-05395, and AYA2007-64052 and from FEDER funds".

2053: 

2054: %% To help institutions obtain information on the effectiveness of their

2055: %% telescopes, the AAS Journals has created a group of keywords for telescope

2056: %% facilities. A common set of keywords will make these types of searches

2057: %% significantly easier and more accurate. In addition, they will also be

2058: %% useful in linking papers together which utilize the same telescopes

2059: %% within the framework of the National Virtual Observatory.

2060: %% See the AASTeX Web site at http://www.journals.uchicago.edu/AAS/AASTeX

2061: %% for information on obtaining the facility keywords.

2062: 

2063: %% After the acknowledgments section, use the following syntax and the

2064: %% \facility{} macro to list the keywords of facilities used in the research

2065: %% for the paper.  Each keyword will be checked against the master list during

2066: %% copy editing.  Individual instruments or configurations can be provided 

2067: %% in parentheses, after the keyword, but they will not be verified.

2068: 

2069: {\it Facilities:} \facility{HST (STIS)}, \facility{HST (ACS)}

2070: 

2071: 

2072: \begin{thebibliography}{}

2073: 

2074: 

2075: \bibitem[Albrow et al.(2001)]{2001ApJ...559.1060A} Albrow, M.~D., 

2076: Gilliland, R.~L., Brown, T.~M., Edmonds, P.~D., Guhathakurta, P., \& 

2077: Sarajedini, A.\ 2001, \apj, 559, 1060 

2078: 

2079: \bibitem[Auriere et al.(1989)]{1989A&A...214..113A} Auri\`ere, M.,

2080:   Koch-Miramond, L., \& Ortolani, S.\ 1989, \aap, 214, 113  

2081: 

2082: \bibitem[Castelli \& Cacciari(2001)]{2001A&A...380..630C} Castelli, F., \& 

2083: Cacciari, C.\ 2001, \aap, 380, 630 

2084: 

2085: \bibitem[Castelli \& Kurucz(2001)]{2001A&A...372..260C} Castelli, F., \& 

2086: Kurucz, R.~L.\ 2001, \aap, 372, 260 

2087: 

2088: \bibitem[Cool et al.(1998)]{1998ApJ...508L..75C} Cool, A.~M., Grindlay, 

2089: J.~E., Cohn, H.~N., Lugger, P.~M., \& Bailyn, C.~D.\ 1998, \apjl, 508,

2090: L75 

2091: 

2092: \bibitem[Davies(1997)]{1997MNRAS.288..117D} Davies, M.~B.\ 1997, \mnras, 

2093: 288, 117

2094: 

2095: \bibitem[De Marco et al.(2004)]{2004ApJ...606L.151D} De Marco, O., Lanz, 

2096: T., Ouellette, J.~A., Zurek, D., \& Shara, M.~M.\ 2004, \apjl, 606,

2097: L151 

2098: 

2099: \bibitem[De Marco et al.(2005)]{2005ApJ...632..894D} De Marco, O., Shara, 

2100: M.~M., Zurek, D., Ouellette, J.~A., Lanz, T., Saffer, R.~A., \& Sepinsky, 

2101: J.~F.\ 2005, \apj, 632, 894 

2102: 

2103: \bibitem[Deutsch et al.(1999)]{1999AJ....118.2888D} Deutsch, E.~W., Margon, 

2104: B., Anderson, S.~F., \& Downes, R.~A.\ 1999, \aj, 118, 2888 

2105: 

2106: \bibitem[Dieball et al.(2007)]{2007ApJ...670..379D} Dieball, A., Knigge, 

2107: C., Zurek, D.~R., Shara, M.~M., Long, K.~S., Charles, P.~A., \& 

2108: Hannikainen, D.\ 2007, \apj, 670, 379 

2109: 

2110: \bibitem[Dieball et al.(2005)a]{2005ApJ...634L.105D} Dieball, A., Knigge, 

2111: C., Zurek, D.~R., Shara, M.~M., Long, K.~S., Charles, P.~A., Hannikainen, 

2112: D.~C., \& van Zyl, L.\ 2005, \apjl, 634, L105 

2113: 

2114: \bibitem[Dieball et al.(2005)b]{2005ApJ...625..156D} Dieball, A., Knigge, 

2115: C., Zurek, D.~R., Shara, M.~M., \& Long, K.~S.\ 2005, \apj, 625, 156 

2116: 

2117: \bibitem[Di Stefano \& Rappaport(1994)]{1994ApJ...423..274D} Di Stefano, 

2118: R., \& Rappaport, S.\ 1994, \apj, 423, 274 

2119: 

2120: \bibitem[Dolphin(2000)]{2000PASP..112.1383D} Dolphin, A.~E.\ 2000, \pasp, 

2121: 112, 1383 

2122: 

2123: \bibitem[Edmonds et al.(1999)]{1999ApJ...516..250E} Edmonds, P.~D., 

2124: Grindlay, J.~E., Cool, A., Cohn, H., Lugger, P., \& Bailyn, C.\ 1999, \apj, 

2125: 516, 250 

2126: 

2127: \bibitem[Edmonds et al.(2001)]{2001ApJ...557L..57E} Edmonds, P.~D., 

2128: Gilliland, R.~L., Heinke, C.~O., Grindlay, J.~E., \& Camilo, F.\ 2001, 

2129: \apjl, 557, L57 

2130: 

2131: \bibitem[Edmonds et al.(2003)a]{2003ApJ...596.1177E} Edmonds, P.~D., 

2132: Gilliland, R.~L., Heinke, C.~O., \& Grindlay, J.~E.\ 2003a, \apj, 596,

2133: 1177 

2134: 

2135: \bibitem[Edmonds et al.(2003)b]{2003ApJ...596.1197E} Edmonds, P.~D., 

2136: Gilliland, R.~L., Heinke, C.~O., \& Grindlay, J.~E.\ 2003b, \apj, 596,

2137: 1197 

2138: 

2139: \bibitem[Ferraro et al.(2001)]{2001ApJ...561..337F} Ferraro, F.~R., 

2140: D'Amico, N., Possenti, A., Mignani, R.~P., \& Paltrinieri, B.\ 2001, \apj, 

2141: 561, 337 

2142: 

2143: \bibitem[Fregeau(2007)]{2007arXiv0710.3167F} Fregeau, J.~M.\ 2007,

2144:   IAUS 246, ``Dynamical Evolution of Dense Stellar Systems'',

2145:   ed. E. Vesperini, in press (arXiv:0710.3167)

2146: 

2147: \bibitem[G{\"a}nsicke et al.(1995)]{1995A&A...303..127G} G{\"a}nsicke, 

2148: B.~T., Beuermann, K., \& de Martino, D.\ 1995, \aap, 303, 127 

2149: 

2150: \bibitem[Gendre et al.(2003)a]{2003A&A...403L..11G} Gendre, B., Barret, D., 

2151: \& Webb, N.\ 2003a, \aap, 403, L11 

2152: 

2153: \bibitem[Gendre et al.(2003)b]{2003A&A...400..521G} Gendre, B., Barret, D., 

2154: \& Webb, N.~A.\ 2003b, \aap, 400, 521 

2155: 

2156: \bibitem[Gratton et al.(2003)]{2003A&A...408..529G} Gratton, R.~G., 

2157: Bragaglia, A., Carretta, E., Clementini, G., Desidera, S., Grundahl, F., \& 

2158: Lucatello, S.\ 2003, \aap, 408, 529 

2159: 

2160: \bibitem[Grindlay et al.(1995)]{1995ApJ...455L..47G} Grindlay, J.~E., Cool, 

2161: A.~M., Callanan, P.~J., Bailyn, C.~D., Cohn, H.~N., \& Lugger, P.~M.\ 1995, 

2162: \apjl, 455, L47 

2163: 

2164: \bibitem[Grindlay et al.(2001)a]{2001Sci...292.2290G} Grindlay, J.~E., 

2165: Heinke, C., Edmonds, P.~D., \& Murray, S.~S.\ 2001a, Science, 292, 2290 

2166: 

2167: \bibitem[Grindlay et al.(2001)b]{2001ApJ...563L..53G} Grindlay, J.~E., 

2168: Heinke, C.~O., Edmonds, P.~D., Murray, S.~S., \& Cool, A.~M.\ 2001b, \apjl, 

2169: 563, L53 

2170: 

2171: \bibitem[Guhathakurta et al.(1992)]{1992AJ....104.1790G} Guhathakurta, P., 

2172: Yanny, B., Schneider, D.~P., \& Bahcall, J.~N.\ 1992, \aj, 104, 1790 

2173: 

2174: \bibitem[Hansen et al.(2003)]{2003ApJ...586.1364H} Hansen, B.~M.~S., 

2175: Kalogera, V., \& Rasio, F.~A.\ 2003, \apj, 586, 1364 

2176: 

2177: \bibitem[Heinke et al.(2003)a]{2003ApJ...598..501H} Heinke, C.~O., Grindlay, 

2178: J.~E., Lugger, P.~M., Cohn, H.~N., Edmonds, P.~D., Lloyd, D.~A., \& Cool, 

2179: A.~M.\ 2003a, \apj, 598, 501 

2180: 

2181: \bibitem[Heinke et al.(2003)b]{2003ApJ...598..516H} Heinke, C.~O., Grindlay, 

2182: J.~E., Edmonds, P.~D., Lloyd, D.~A., Murray, S.~S., Cohn, H.~N., \& Lugger, 

2183: P.~M.\ 2003b, \apj, 598, 516 

2184: 

2185: \bibitem[Heinke et al.(2005)]{2005ApJ...625..796H} Heinke, C.~O., Grindlay, 

2186: J.~E., Edmonds, P.~D., Cohn, H.~N., Lugger, P.~M., Camilo, F., Bogdanov, 

2187: S., \& Freire, P.~C.\ 2005, \apj, 625, 796 

2188: 

2189: \bibitem[Honeycutt et al.(1998)]{1998AJ....115.2527H} Honeycutt, R.~K., 

2190: Robertson, J.~W., \& Turner, G.~W.\ 1998, \aj, 115, 2527 

2191: 

2192: \bibitem[Honeycutt \& Kafka(2004)]{2004AJ....128.1279H} Honeycutt, R.~K., 

2193: \& Kafka, S.\ 2004, \aj, 128, 1279 

2194: 

2195: \bibitem[Horne \& Marsh(1986)]{1986MNRAS.218..761H} Horne, K., \& Marsh, 

2196: T.~R.\ 1986, \mnras, 218, 761 

2197: 

2198: \bibitem[Howell et al.(2000)]{2000PASP..112.1200H} Howell, J.~H., 

2199: Guhathakurta, P., \& Gilliland, R.~L.\ 2000, \pasp, 112, 1200 

2200: 

2201: \bibitem[Hurley et al.(2007)]{2007ApJ...665..707H} Hurley, J.~R., Aarseth, 

2202: S.~J., \& Shara, M.~M.\ 2007, \apj, 665, 707 

2203: 

2204: \bibitem[Hut et al.(1992)]{1992PASP..104..981H} Hut, P., et al.\ 1992,

2205:   \pasp, 104, 981  

2206: 

2207: \bibitem[Ivanova et al.(2005)]{2005MNRAS.358..572I} Ivanova, N., 

2208: Belczynski, K., Fregeau, J.~M., \& Rasio, F.~A.\ 2005, \mnras, 358,

2209: 572 

2210: 

2211: \bibitem[Ivanova et al.(2006)]{2006MNRAS.372.1043I} Ivanova, N., Heinke, 

2212: C.~O., Rasio, F.~A., Taam, R.~E., Belczynski, K., \& Fregeau, J.\ 2006, 

2213: \mnras, 372, 1043 

2214: 

2215: \bibitem[Knigge et al.(1995)]{1995MNRAS.273..225K} Knigge, C., Woods, 

2216: J.~A., \& Drew, J.~E.\ 1995, \mnras, 273, 225 

2217: 

2218: \bibitem[Knigge et al.(1998)]{1998ApJ...499..414K} Knigge, C., Long, K.~S., 

2219: Wade, R.~A., Baptista, R., Horne, K., Hubeny, I., \& Rutten, R.~G.~M.\ 

2220: 1998, \apj, 499, 414 

2221: 

2222: \bibitem[Knigge et al.(2000)]{2000ApJ...539L..49K} Knigge, C., Long, K.~S., 

2223: Hoard, D.~W., Szkody, P., \& Dhillon, V.~S.\ 2000, \apjl, 539, L49 

2224: 

2225: \bibitem[Knigge et al.(2002)]{2002ApJ...579..752K} Knigge, C., Zurek, 

2226: D.~R., Shara, M.~M., \& Long, K.~S.\ 2002, \apj, 579, 752 

2227: 

2228: \bibitem[Knigge et al.(2003)]{2003ApJ...599.1320K} Knigge, C., Zurek, 

2229: D.~R., Shara, M.~M., Long, K.~S., \& Gilliland, R.~L.\ 2003, \apj, 599, 

2230: 1320 

2231: 

2232: \bibitem[Knigge(2004)]{2004MPLA...19.2013K} Knigge, C.\ 2004, Modern 

2233: Physics Letters A, 19, 2013 

2234: 

2235: \bibitem[Knigge et al.(2004)]{2004ApJ...615L.129K} Knigge, C., 

2236: Araujo-Betancor, S., G{\"a}nsicke, B.~T., Long, K.~S., Szkody, P., Hoard, 

2237: D.~W., Hynes, R.~I., \& Dhillon, V.~S.\ 2004, \apjl, 615, L129 

2238: 

2239: \bibitem[Knigge et al.(2006)]{2006ApJ...641..281K} Knigge, C., Gilliland, 

2240: R.~L., Dieball, A., Zurek, D.~R., Shara, M.~M., \& Long, K.~S.\ 2006, \apj, 

2241: 641, 281 

2242: 

2243: \bibitem[Knigge(2006)]{2006MNRAS.373..484K} Knigge, C.\ 2006, \mnras, 373, 

2244: 484 

2245: 

2246: \bibitem[Lombardi et al.(2006)]{2006ApJ...640..441L} Lombardi, J.~C., Jr., 

2247: Proulx, Z.~F., Dooley, K.~L., Theriault, E.~M., Ivanova, N., \& Rasio, 

2248: F.~A.\ 2006, \apj, 640, 441 

2249: 

2250: \bibitem[Maccarone \& Knigge(2007)]{2007A&G....48e..12M} Maccarone, T., \& 

2251: Knigge, C.\ 2007, Astronomy and Geophysics, 48, 12 

2252: 

2253: \bibitem[Machin et al.(1991)]{1991MNRAS.250..602M} Machin, G., Callanan, 

2254: P.~J., Allington-Smith, J., Charles, P.~A., \& Hassall, B.~J.~M.\ 1991, 

2255: \mnras, 250, 602 

2256: 

2257: \bibitem[bla1]{bla1} Ma\'{i}z Apell\'{a}niz, J.\ 2005, {\em HST/STIS} Instrument

2258:   Science Report, ISR 2005-2

2259: 

2260: \bibitem[bla2]{bla2} Ma\'{i}z Apell\'{a}niz, J.\ 2007, {\sc multispec:} A code

2261:   for the extraction of slitless spectra in crowded fields, 

2262:   http://dae45.iaa.csic.es:8080/~jmaiz/software/multispec/handbook.pdf

2263: 

2264: \bibitem[Margon \& Downes(1983)]{1983ApJ...274L..31M} Margon, B., \& 

2265: Downes, R.~A.\ 1983, \apjl, 274, L31 

2266: 

2267: \bibitem[Margon et al.(1981)]{1981ApJ...247L..89M} Margon, B., Downes, 

2268: R.~A., \& Gunn, J.~E.\ 1981, \apjl, 247, L89 

2269: 

2270: \bibitem[McLaughlin et al.(2006)]{2006ApJS..166..249M} McLaughlin, D.~E., 

2271: Anderson, J., Meylan, G., Gebhardt, K., Pryor, C., Minniti, D., \& Phinney, 

2272: S.\ 2006, \apjs, 166, 249 

2273: 

2274: \bibitem[Minniti et al.(1997)]{1997ApJ...474L..27M} Minniti, D., Meylan, 

2275: G., Pryor, C., Phinney, E.~S., Sams, B., \& Tinney, C.~G.\ 1997, \apjl, 

2276: 474, L27 

2277: 

2278: \bibitem[Naylor et al.(1989)]{1989MNRAS.241P..25N} Naylor, T., et al.\ 

2279: 1989, \mnras, 241, 25P 

2280: 

2281: \bibitem[O'Toole et al.(2006)]{2006BaltA..15...61O} O'Toole, S.~J., 

2282: Napiwotzki, R., Heber, U., Drechsel, H., Frandsen, S., Grundahl, F., \& 

2283: Bruntt, H.\ 2006, Baltic Astronomy, 15, 61 

2284: 

2285: \bibitem[Paczy{\'n}ski(1971)]{1971ARA&A...9..183P} Paczy{\'n}ski,

2286:   B.\ 1971, \araa, 9, 183  

2287: 

2288: \bibitem[Paresce \& de Marchi(1994)]{1994ApJ...427L..33P} Paresce, F.,

2289:   \&  de Marchi, G.\ 1994, \apjl, 427, L33 

2290: 

2291: \bibitem[Paresce et al.(1992)]{1992Natur.360...46P} Paresce, F., de Marchi, 

2292: G., \& Ferraro, F.~R.\ 1992, \nat, 360, 46 

2293: 

2294: \bibitem[Pooley et al.(2002)]{fooo} Pooley, D., et al.\ 2002, \apj, 573, 184

2295: 

2296: \bibitem[Pooley et al.(2003)]{2003ApJ...591L.131P} Pooley, D., et al.\ 2003, \apjl, 591, L131 

2297: 

2298: \bibitem[Pooley \& Hut(2006)]{2006ApJ...646L.143P} Pooley, D., \& Hut, P.\ 

2299: 2006, \apjl, 646, L143

2300: 

2301: \bibitem[Porter \& Townsend(2005)]{2005ApJ...623L.129P} Porter, J.~M., \& 

2302: Townsend, R.~H.~D.\ 2005, \apjl, 623, L129 

2303: 

2304: \bibitem[Renzini \& Fusi Pecci(1988)]{1988ARA&A..26..199R} Renzini, A., \& 

2305: Fusi Pecci, F.\ 1988, \araa, 26, 199 

2306: 

2307: \bibitem[Serenelli et al.(2002)]{2002MNRAS.337.1091S} Serenelli, A.~M., 

2308: Althaus, L.~G., Rohrmann, R.~D., \& Benvenuto, O.~G.\ 2002, \mnras, 337, 

2309: 1091 

2310: 

2311: \bibitem[Saffer et al.(2002)]{2002ASPC..263..157S} Saffer, R.~A.,

2312:   Sepinski, J.~F., Demarchi, G., Livio, M., Paresce, F., Shara, M.~M., 

2313: \& Zurek, D.\ 2002, in ``Stellar Collisions, Mergers and their

2314: Consequences'', ASP Conf. Ser., Vol. 263, ed: M. M. Shara (ASP: San

2315: Francisco), p157 

2316: 

2317: \bibitem[Shara et al.(1990)]{1990AJ.....99.1858S} Shara, M.~M., Potter, M., 

2318: \& Moffat, A.~F.~J.\ 1990, \aj, 99, 1858 

2319: 

2320: \bibitem[Shara et al.(1996)]{1996ApJ...471..804S} Shara, M.~M., Bergeron, 

2321: L.~E., Gilliland, R.~L., Saha, A., \& Petro, L.\ 1996, \apj, 471, 804 

2322: 

2323: \bibitem[Shara and Hurley (2006)]{blablabla} Shara, M.~M., \& Hurley,

2324:   J.~R.\ 2006, \apj, 646, 464

2325: 

2326: \bibitem[Smak(1981)]{1981AcA....31..395S} Smak, J.\ 1981, Acta Astronomica, 

2327: 31, 395 

2328: 

2329: \bibitem[Town1]{Town1} Townsley, D.~M. \& Bildsten L.\ 2002, \apjl,

2330:   565, L35

2331: 

2332: \bibitem[Town2]{Town2} Townsley, D.~M. \& Bildsten L.\ 2005, \apj,  628, 395

2333: 

2334: \bibitem[van Zyl et al.(2004)]{2004MNRAS.350..649V} van Zyl, L., Charles, 

2335: P.~A., Arribas, S., Naylor, T., Mediavilla, E., \& Hellier, C.\ 2004, 

2336: \mnras, 350, 649

2337: 

2338: \bibitem[Webb et al.(2006)]{2006A&A...445..155W} Webb, N.~A., Wheatley, 

2339: P.~J., \& Barret, D.\ 2006, \aap, 445, 155 

2340: 

2341: \bibitem[bla1111]{bla1111} Warner, B. 1995, ``Cataclysmic Variable

2342:   Stars'', Cambridge Astrophysics Series, Cambridge University Press 

2343: 

2344: \bibitem[Wood(1995)]{1995whdw.conf...41W} Wood, M.~A.\ 1995, in: 9th

2345:   European Workshop on White Dwarfs, Lecture Notes in Physics, Vol.~443, 

2346: ed: D. Koester \& K. Werner (Berlin: Springer-Verlag), p.41 

2347: 

2348: \end{thebibliography}

2349: 

2350: %\LongTables

2351: %\begin{landscape}

2352: %\tablewidth{1.0\textwidth}

2353: %\begin{deluxetable*}{rllllllllll}

2354: \begin{deluxetable}{rllllllllll}

2355: \tablewidth{1.0\textwidth}

2356: \tabletypesize{\tiny}

2357: %\rotate

2358: \tablecaption{Basic data on all FUV sources with useful spectra. Objects listed in bold at the top of the table are analyzed in detail in Section~\protect\ref{sec:sed}.\label{tab:specdat}}

2359: \tablehead{

2360: \colhead{ID} & 

2361: \colhead{RA} & 

2362: \colhead{Dec} & 

2363: \colhead{$m_{FUV}$} & 

2364: \colhead{$m_{F336W}$} &

2365: \colhead{CMD} & 

2366: \colhead{Sep} & 

2367: \colhead{Var?} &

2368: \colhead{PM?} & 

2369: \colhead{Other Names} & 

2370: \colhead{Spec/SED Class}\\

2371: \colhead{(1)} &

2372: \colhead{(2)} &

2373: \colhead{(3)} &

2374: \colhead{(4)} &

2375: \colhead{(5)} &

2376: \colhead{(6)} &

2377: \colhead{(7)} &

2378: \colhead{(8)} &

2379: \colhead{(9)} &

2380: \colhead{(10)} &

2381: \colhead{(11)} 

2382: }

2383: \startdata

2384: \bf{1}    & \bf{00:24:04.915}  & \bf{-72:04:55.40}  & \bf{15.62}  & \bf{17.99} & \bf{Gap}  & \bf{0.95}    & \bf{Yes}	& \bf{\ldots} & \bf{AKO 9, W36, PC1-V11,M055581}    & \bf{CV (WD-SG)}                               \\

2385: \bf{2}    & \bf{00:24:06.541}  & \bf{-72:05:03.20}  & \bf{17.37}  & \bf{15.72} & \bf{BS}   & \bf{0.74}    & \bf{Yes}	& \bf{\ldots} & \bf{M048212}	                    & \bf{Binary (BS-WD)}                           \\       

2386: \bf{3}    & \bf{00:24:04.248}  & \bf{-72:04:57.99}  & \bf{17.53}  & \bf{19.11} & \bf{Gap}  & \bf{1.05}    & \bf{Yes}	& \bf{\ldots} & \bf{V1, X9, W42, PC1-V47, M053129}  & \bf{CV (WD-MS)}                               \\                  

2387: \bf{4}    & \bf{00:24:06.353}  & \bf{-72:04:49.17}  & \bf{17.70}  & \bf{19.96} & \bf{Gap}  & \bf{1.00}    & \bf{No} 	& \bf{\ldots} & \bf{M061563}	                    & \bf{He WD}                                    \\       

2388: \bf{5}    & \bf{00:24:05.557}  & \bf{-72:05:02.15}  & \bf{18.03}  & \bf{20.92} & \bf{WD}   & \bf{1.01}    & \bf{No} 	& \bf{\ldots} & \bf{M049123}	                    & \bf{WD}                                       \\  

2389: \bf{7}    & \bf{00:24:05.794}  & \bf{-72:04:57.13}  & \bf{18.26}  & \bf{20.85} & \bf{WD}   & \bf{1.07}    & \bf{No} 	& \bf{\ldots} & \bf{\ldots}	                    & \bf{Binary (WD-MS)}                           \\

2390: \bf{10}   & \bf{00:24:06.489}  & \bf{-72:04:52.27}  & \bf{18.35}  & \bf{18.07} & \bf{Gap}  & \bf{0.59}    & \bf{No} 	& \bf{Yes}    & \bf{W75?, PC1-V36, DM-3, M058539}   & \bf{Binary (??-Stripped SG)}                  \\

2391: \bf{15}   & \bf{00:24:06.563}  & \bf{-72:05:04.38}  & \bf{18.63}  & \bf{18.03} & \bf{Gap}  & \bf{1.29}    & \bf{No} 	& \bf{\ldots} & \bf{M047149}	                    & \bf{Binary? (WD-SG)\tablenotemark{c}}                           \\  

2392: \bf{17}   & \bf{00:24:06.390}  & \bf{-72:05:07.37}  & \bf{18.77}  & \bf{18.61} & \bf{Gap}  & \bf{1.20}    & \bf{No} 	& \bf{\ldots} & \bf{M044437}	                    & \bf{Binary? (WD-SG)\tablenotemark{c}; CV?}                      \\

2393: \bf{20}   & \bf{00:24:05.992}  & \bf{-72:04:56.11}  & \bf{18.99}  & \bf{19.94} & \bf{Gap}  & \bf{1.07}    & \bf{Yes}	& \bf{\ldots} & \bf{V2, X19, W30, PV1-V53, M054898} & \bf{CV (WD-MS)}                               \\ 

2394: \bf{27}   & \bf{00:24:06.850}  & \bf{-72:04:53.35}  & \bf{19.41}  & \bf{20.66} & \bf{Gap}  & \bf{0.75}    & \bf{No}     & \bf{\ldots} & \bf{M057483}	                    & \bf{SMC Interloper}                           \\       

2395: \bf{999}  & \bf{00:24:04.009}  & \bf{-72:04:51.24}  & \bf{17.61}  & \bf{15.72} & \bf{BS}   & \bf{\ldots}  & \bf{\ldots} & \bf{\ldots} & \bf{M059651\tablenotemark{a}}       & \bf{BS}                                       \\ 

2396:      6    &     00:24:08.459   &     -72:04:50.58   &     18.15   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots                                        \\

2397:      8    &     00:24:04.912   &     -72:04:51.67   &     18.27   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\

2398:      9    &     00:24:07.443   &     -72:05:07.43   &     18.33   &     21.10  &     WD    &     1.43     &     No 	& \ldots      & \ldots                              & \ldots	                                    \\

2399:      12   &     00:24:07.919   &     -72:04:52.99   &     18.41   &     15.87  &     BS    &     0.48     &     Yes	& Yes         & M057848	                            & \ldots	                                    \\       

2400:      13   &     00:24:04.890   &     -72:05:03.61   &     18.46   &     21.36  &     WD    &     1.24     &     No 	& \ldots      & \ldots                              & \ldots	                                    \\

2401:      14   &     00:24:05.554   &     -72:04:53.94   &     18.51   &     16.15  &     BS    &     0.44     &     No 	& Yes         & DM-9, M056964                       & \ldots	                                    \\        

2402:      16   &     00:24:06.511   &     -72:04:50.28   &     18.76   &     16.34  &     BS    &     0.43     &     Yes	& Yes         & M060466\tablenotemark{b}            & \ldots	                                    \\  

2403:      18   &     00:24:05.266   &     -72:04:59.02   &     18.99   &     21.74  &     WD    &     1.00     &     No 	& \ldots      & \ldots                              & \ldots	                                    \\

2404:      23   &     00:24:06.398   &     -72:04:53.92   &     19.10   &     16.26  &     BS    &     1.34     &     No 	& Yes         & DM-5,M056966\tablenotemark{b}       & \ldots	                                    \\       

2405:      24   &     00:24:06.122   &     -72:04:51.84   &     19.14   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2406:      25   &     00:24:08.023   &     -72:05:06.30   &     19.18   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2407:      26   &     00:24:07.456   &     -72:04:57.08   &     19.25   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2408:      29   &     00:24:07.933   &     -72:04:57.06   &     19.48   &     22.14  &     WD    &     1.18     &     No 	& \ldots      & M053994                             & \ldots	                                    \\  

2409:      30   &     00:24:07.024   &     -72:04:49.13   &     19.51   &     21.93  &     WD    &     0.63     &     No 	& \ldots      & \ldots	                            & \ldots	                                    \\       

2410:      31   &     00:24:04.246   &     -72:04:51.39   &     19.58   &     22.29  &     WD    &     0.75     &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2411:      32   &     00:24:03.646   &     -72:05:04.15   &     19.59   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2412:      33   &     00:24:05.986   &     -72:04:58.37   &     19.62   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2413:      36   &     00:24:04.171   &     -72:05:02.45   &     19.71   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2414:      37   &     00:24:05.234   &     -72:05:02.06   &     19.72   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2415:      38   &     00:24:07.375   &     -72:04:51.13   &     19.77   &     22.33  &     WD    &     0.82     &     No 	& \ldots      & \ldots		                    & \ldots	                                    \\       

2416:      42   &     00:24:05.982   &     -72:04:55.34   &     19.99   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2417:      46   &     00:24:07.228   &     -72:05:01.21   &     20.16   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2418:      47   &     00:24:04.623   &     -72:04:47.95   &     20.20   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2419:      49   &     00:24:06.085   &     -72:05:06.72   &     20.27   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2420:      52   &     00:24:05.964   &     -72:04:48.94   &     20.40   &     16.73  &     BS    &     0.52     &     No 	& Yes         & M061787\tablenotemark{a}            & \ldots	                                    \\  

2421:      54   &     00:24:05.736   &     -72:04:53.31   &     20.47   &     16.80  &     BS    &     0.27     &     No 	& Yes         & M057553                             & \ldots	                                    \\  

2422:      59   &     00:24:07.383   &     -72:05:05.76   &     20.55   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2423:      66   &     00:24:04.299   &     -72:05:05.01   &     20.68   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2424:      67   &     00:24:04.096   &     -72:04:51.27   &     20.69   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2425:      73   &     00:24:08.222   &     -72:04:52.19   &     20.86   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2426:      78   &     00:24:04.374   &     -72:04:58.26   &     20.99   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2427:      79   &     00:24:08.240   &     -72:04:53.58   &     21.00   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2428:      81   &     00:24:06.308   &     -72:05:05.27   &     21.05   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2429:      82   &     00:24:06.800   &     -72:05:00.12   &     21.06   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2430:      84   &     00:24:07.678   &     -72:04:55.88   &     21.09   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots	                                    \\      

2431:      85   &     00:24:06.341   &     -72:04:51.16   &     21.17   &     \ldots &     NoOpt &     \ldots   &     No 	& \ldots      & \ldots                              & \ldots                                        

2432: \enddata

2433: \tablenotetext{a}{The master image of McLaughlin et al. (2006) is saturated and/or confused in the vicinity of this object.}

2434: \tablenotetext{b}{This object has a near neighbour that was not detected in McLaughlin et al. (2006).}

2435: \tablenotetext{c}{The binary status of this system is not certain due to the possibility of a chance superposition.}

2436: %\tabletypesize{\scriptsize}

2437: %\tabletypesize{\tiny}

2438: %% Text for table notes should follow after the \enddata but before

2439: %% the \end{deluxetable}. Make sure there is at least one \tablenotemark

2440: %% in the table for each \tablenotetext.

2441: \tablecomments{Column definitions:\\

2442: (1) Spectrosopic identification number; c.f. Figure~\protect\ref{fig:images}\\

2443: (2) Right ascension (J2000); boresight corrected to match positions in Heinke et al. (2005).\\

2444: (3) Declination (J2000); boresight corrected to match positions in Heinke et al. (2005).\\

2445: (4) STIS/FUV-MAMA/F25QTZ magnitude (STMAG system).\\

2446: (5) WFPC2/PC/F336W magnitude (STMAG system).\\

2447: (6) CMD-based classification; see Section~\protect\ref{sec:cmd} and Figure~\protect\ref{fig:cmd}.\\ 

2448: (7) Offset between FUV and F336W position in STIS/FUV-MAMA pixels (1 pix $\simeq$\ 0.025\arcsec).\\

2449: (8) Variable in FUV photometry? (See Paper~I for details.)\\

2450: (9) Proper motion consistent with cluster membership? (Only if available in McLaughlin et al. [2006].)\\

2451: (10) Alternative names are taken from the following catalogues:\\

2452: \hspace*{0.5cm} AKO~n = Star n in Auri\`ere, Koch-Miramond \& Ortolani (1989); \\

2453: \hspace*{0.5cm} Vn = Variable star n as defined by Paresce, De Marchi \& Ferraro (1992) and Paresce \& De Marchi(1994)\\

2454: \hspace*{0.5cm} Wn = Star n in Grindlay et al. (2001a) and Heinke et al. (2005); \\

2455: \hspace*{0.5cm} PC1-Vn = Variable star n on the PC1 chip in Albrow et al. (2001);\\

2456: \hspace*{0.5cm} Mn = Star n in the master catalogue of McLaughlin et al. (2006);\\

2457: \hspace*{0.5cm} DM-n = Star 104-n in De Marco et al. (2005).\\

2458: (11) Final classification based on FUV spectrum and FUV-NIR SED (Section~\protect\ref{sec:sed}). Notation as in text, except G=subgiant. 

2459: }

2460: %\vspace*{0.5cm}

2461: %

2462: %\tablenotetext{a}{Spectrosopic identification number; c.f. Figure~\protect\ref{fig:images}.}

2463: %\tablenotetext{b}{Right ascension (J2000); boresight corrected to match positions in Heinke et al. (2005).}

2464: %\tablenotetext{c}{Declination (J2000); boresight corrected to match positions in Heinke et al. (2005).}

2465: %\tablenotetext{d}{STIS/FUV-MAMA/F25QTZ magnitude (STMAG system).}

2466: %\tablenotetext{e}{WFPC2/PC/F336W magnitude (STMAG system).}

2467: %\tablenotetext{f}{CMD-based classification; see Section~\protect\ref{sec:cmd} and Figure~\protect\ref{fig:cmd}.}

2468: %\tablenotetext{g}{Offset between FUV and F336W position in STIS/FUV-MAMA pixels (1 pix $\simeq$\ 0.025\arcsec).}

2469: %\tablenotetext{h}{Variable in FUV photometry? (See Paper~I for details.)}

2470: %\tablenotetext{i}{AKO~n = Star n in Auri\'ere, Koch-Miramond \& Ortolani (1989); \\

2471:                   %Vn = Star n in Paresce, De Marchi \& Ferraro (1992); \\

2472:                   %Wn = Star n in Grindlay et al. (2001) and Heinke et al. (2005); \\

2473:                   %PC1-Vn = Star n on the PC1 chip in Albrow et al. (2001);\\

2474:                   %Mn = Star n in the master catalogue of McLaughlin et al. (2006);\\

2475:                   %DM-n = Star n in the 47 Tuc data of De Marco et

2476:                   %al. (2005).}                                                                                             

2477: %\tablenotetext{j}{Final classification based on the analysis in Section~\protect\ref{sec:sed}.}

2478: %\end{deluxetable*}

2479: %\end{landscape}

2480: \end{deluxetable}

2481: 

2482: 

2483: 

2484: 

2485: \end{document}

2486: 

2487: 

2488: 

2489: %begin{figure} 

2490: %\epsscale{0.85}

2491: %\plotone{sample.eps}

2492: %\caption{Sample figure showing important results.}

2493: %\end{figure}

2494: 

2495: %\begin{figure}

2496: %\plottwo{fig2a.eps}{fig2b.eps}

2497: %\caption{{\sc Fig.~2} (\emph{a}) field surrounding target \#1; (\emph{b})

2498: %   detail of the inner 70~kpc.\label{targetfov}}

2499: %\end{figure}

2500: 

2501: %If you need to rotate or make other transformations to a figure, you may use the \plotfiddle command:

2502: 

2503: %\plotfiddle{PSFILE}{VSIZE}{ROTANG}{HSCALE}{VSCALE}{HTRANS}{VTRANS}

2504: %\plotfiddle{sample.eps}{2.6in}{-90.}{32.}{32.}{-250}{225}

2505: 

2506: 

2507: %begin{figure} 

2508: %\plotfiddle{PSFILE}{VSIZE}{ROTANG}{HSCALE}{VSCALE}{HTRANS}{VTRANS}

2509: %\plotfiddle{sample.eps}{2.6in}{-90.}{32.}{32.}{-250}{225}

2510: %\plotone{sample.eps}

2511: %\caption{Sample figure showing important results.}

2512: %\end{figure}

2513: 

2514: 

2515: