1: 

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

3: 

4: \topmargin -0.1in

5: \begin{document}

6: 

7: \title{The Chemical Compositions of the Type II Cepheids -- The BL Her and W Vir Variables} 

8: 

9: 

10: \author{Thomas Maas}

11: \affil{The W.J. McDonald Observatory, The University of Texas; Austin, TX 78712-1083

12: \\ thomas$_{-}$maas@hotmail.com}

13: 

14: \author{

15: Sunetra Giridhar}

16: \affil{Indian Institute of Astrophysics;

17: Bangalore,  560034 India\\

18: giridhar@iiap.res.in}

19: 

20: \author{David L.\ Lambert }

21: \affil{The W.J. McDonald Observatory; University of

22: Texas; Austin, TX 78712-1083

23: \\ dll@astro.as.utexas.edu}

24: 

25: 

26: 

27: \begin{abstract}

28: 

29: Abundance analyses from high-resolution optical spectra are presented for

30: 19 Type II Cepheids in the Galactic field. The sample includes both

31: short-period (BL Her) and long-period (W Vir) stars. This is the

32: first extensive abundance analysis of these variables. 

33: The C, N, and O abundances with similar spreads for the BL Her and

34: W Vir show evidence for an atmosphere contaminated with $3\alpha$-process

35: and CN-cycling products.

36:  A notable

37: anomaly of the BL Her stars is an overabundance of Na by a factor

38: of about five relative to their presumed initial abundances. This

39: overabundance is not seen in the W Vir stars.

40:  The abundance

41: anomalies  running from mild to extreme  in W Vir stars 

42: but not seen in the BL Her stars

43: are attributed to dust-gas separation that provides an atmosphere

44: deficient in elements of high condensation temperature, notably

45: Al, Ca, Sc,  Ti, and $s$-process elements.

46:  Such anomalies have previously been seen among RV Tau

47: stars which represent a long-period extension of the variability enjoyed

48: by the Type II Cepheids.  

49: Comments are offered on how the contrasting abundance anomalies of BL Her

50: and W Vir stars may be explained in terms of the stars' evolution from the

51: blue horizontal branch.

52: 

53:  

54: {\it Subject headings: stars:abundances -- stars:AGB and post-AGB --

55: stars: variables:other (RV\,Tauri)}

56: 

57: 

58: 

59: \end{abstract}

60: 

61: 

62: 

63: \section{Introduction}

64: 

65: 

66: Wallerstein (2002)

67: remarks that  Type II Cepheids

68: ``include most intrinsic variables

69: with periods between 1 and about 50 days, except for the classical Cepheids and the

70: shortest semi-regular variables of type M.'' These bounds on the

71: periods place the Type II Cepheids in the instability strip

72:  between the RR Lyrae stars at the short

73: period limit and the

74: RV Tau variables at the long period limit.

75: Type II Cepheids fall into two classes:  BL Her stars with periods of 1 to

76: 5 days and the 

77:  W Vir stars with periods longer than

78: about 10 days with an indistinct boundary at

79: about 20-30 days separating these stars from the RV Tau stars.

80:  Kraft (1972) drew attention

81: to the period gap (6 to 9 days) between the BL Her and W Vir

82: stars in globular clusters.  

83: 

84: 

85: This paper is devoted to abundance determinations of field

86: Type II Cepheids.

87: Compositions of these stars

88:  have received scant attention from spectroscopists despite

89: hints  of unusual compositions.

90: Oddly, neither  W Vir nor BL Her, the two prototypes,  have

91: been subjected to a modern analysis.

92: Quantitative spectroscopy based on modern model atmospheres and

93: CCD spectra seems to be limited to

94:  TX Del (Andrievsky et al. 2002)\footnote{Andrievsky et al. identified

95: TX Del as a first-overtone Type I Cepheid - see below.}

96:  ST Pup, a star with RV Tau-like

97: abundance anomalies (Gonzalez \& Wallerstein 1996),

98:  and the  C-rich stars V553 Cen and

99: RT TrA (Wallerstein \& Gonzalez  1996; Wallerstein et al. 2000).

100: Model atmosphere analyses based on image-tube photographic spectra

101: were reported for BL Her (Caldwell \& Butler 1978),

102: $\kappa$ Pav (Luck \& Bond 1989) and

103: AU Peg (Harris, Olszewski, \& Wallerstein 1984). Curve of growth

104: analyses and photographic spectra provided abundance estimates

105: for TW Cap (Anderson \& Kraft 1971), W Vir (Barker et al. 1971), and

106: $\kappa$ Pav (Rodgers \& Bell 1963, 1968). This very mixed bag of

107: abundance estimates is an inadequate basis from

108: which to draw conclusions concerning the evolution of Type II Cepheids

109: and other potential origins for abundance anomalies.

110: 

111:  

112:    

113: The principal prior indication of an anomalous composition for Type II Cepheids

114: concerns 

115: the reported

116:  deficiency of the $s$-process elements (relative to the iron abundance) 

117: --

118: see Rodgers \& Bell (1963, 1968),  Barker et al. (1971) and

119: especially  Luck \& Bond (1989). Our interest was driven in part by this

120: indication, but also by the lack of a thorough study of field Type II Cepheids, and

121: finally by their close relationship to the RV Tau stars for which abundance

122: anomalies exist.

123: The   principal 

124: anomaly for some

125:  RV Tau stars is one in which the atmosphere is deficient in those

126: elements that first  condense into grains as gas is cooled, i.e.,

127: there has been a separation or winnowing of dust from gas and

128: accretion of gas by the star (Giridhar et al. 2005). An apparently

129: rarer anomaly is one in which elements with a low first ionization

130: potential are underabundant (Rao \& Reddy 2005). Other RV Tau

131: stars have a normal composition.

132: There is ample evidence that the W Vir and RV Tau stars have much in common.

133: For example, photometry of Type II Cepheids and RV Tau stars

134:  in the LMC  shows clearly

135: that the RV Tau stars and the Type II Cepheids define a common

136: Period-Luminosity-Color relation (Alcock et al. 1998).

137: Given this commonality, the question naturally arises --

138: What abundance anomalies are shown by the BL Her and W Vir stars

139: and are those anomalies similar to those shown by the RV Tau stars?

140: 

141: 

142: The observations and abundance analyses of 19 Type II Cepheids

143:  are described in Section 2.

144: Results for the individual stars and remarks on their classification

145: as Type II Cepheids are provided in Section 3. Discussion of the possible

146: anomalous abundances and their relation to the prior evolution of the

147:  stars and to other processes is provided in Section 4. Brief remarks on  

148: the spectroscopic calibration of photometric  measurements of

149: metallicity are given in Section 5. Concluding remarks

150: are offered in Section 6.

151: 

152: 

153: \section{Observations and Analyses}

154: 

155: 

156: The program stars were observed with

157:  the W.J.  McDonald Observatory's 2.7m Harlan J. Smith reflector and the

158: CCD-equipped `2dcoud\'{e}' spectrograph (Tull et al. 1995) in observing runs 

159: in 2004 and 2005. A spectral

160: resolving power

161: $R = \lambda/\Delta\lambda \simeq  60,000$ was used and a broad spectral range was covered

162: in a single exposure.

163: 

164: Spectra were rejected

165: if they showed line doubling, markedly asymmetric lines, or

166: strong emission in the Balmer lines. 

167:  It is presumed that the spectra not

168: showing these characteristics represent the atmosphere at a

169: time when standard theoretical models may be applicable.

170: This presumption should be tested by analysis of a series of spectra taken

171: over the pulsational cycle. This remains to be done but 

172: we have analysed three stars using spectra taken at

173: different phases and obtained consistent results.

174: The program stars and dates of observation are listed in Table 1.

175: 

176: 

177: The abundance analyses were performed as described in our

178: earlier papers on the RV Tau variables. Atmospheric parameters are

179: determined  from the Fe\,{\sc i} and Fe\,{\sc ii}

180: lines by demanding excitation and ionization equilibrium, and

181: that the iron abundance be independent of the equivalent width.

182: The adopted parameters listed in Table 1  were determined using

183: Kurucz model atmospheres and a normal helium abundance (He/H = 0.1).

184: The full abundance analysis used a Kurucz atmosphere

185: with the parameters in Table 1.

186:  Abundances are

187: referred to the recommended solar photospheric abundances

188: given by  Asplund, Grevesse, \& Sauval (2005).

189: Table 1 shows the anticipated average difference in atmospheric

190: parameters

191:  ($T_{\rm eff}, \log$ g)

192:  between the BL Her

193: and W Vir stars: 

194:  ($T_{\rm eff}, \log$ g) $=$ (5970,1.4) and (5380, 0.6) for the BL Her and

195: W Vir stars, respectively.

196: 

197: \section{The Program Stars}

198: 

199: Harris (1985) provided what is regarded as the reference catalog of

200: Type II Cepheids. His principal criterion for distinguishing a Type II

201: from a Classical (Type I)

202:   Cepheid was distance from the Galactic plane ($|$Z$|$). Our selection

203: of stars was based on

204: his  catalog but includes some recent discoveries. Various

205: photometric indices extracted from the light curves have been

206: proposed as distinctive marks of a Type II or a Type I Cepheid but

207: the distinction remains tricky, as many authors remark. In addition,

208: longer period 

209: Type II Cepheids overlap in pulsational properties (i.e., period and

210: luminosity) with the RV Tau variables. Therefore, not only must

211: population type be examined but variable type too. In the following

212: text, we comment on the our selection of Type II

213: Cepheids and remark on those with RV Tau proclivities.

214: 

215: In this paper, the distinction between a BL Her and a

216:  W Vir variable is maintained; the reasons for this will become

217: obvious when the compositions are presented. Here, a BL Her

218: variable is one with a pulsation period of four days and shorter,

219: and a W Vir variable is a star with a period of 11 days or longer.

220: Four stars in the sample are of intermediate period with

221: two judged by composition to be BL Her stars and two W Vir

222: stars. 

223: 

224: The stars are discussed  in order of increasing pulsation period.

225: 

226: 

227: {\bf BX Del:} BX Del with an estimated $|$Z$|$ of 0.3 kpc

228: was listed as a Type II Cepheid by Harris (1985). 

229: The

230: metallicity [Fe/H] $= -0.2$ places BX Del at about the upper limit for

231: thick disk stars (Reddy et al. 2006).

232: 

233: 

234: {\bf VY Pyx:} This variable was identified as a Type II Cepheid

235: by Sanwal \& Sarma (1991).

236: The composition shows similarities with that of BX Del.

237: 

238: 

239: {\bf BL Her:} This Type II Cepheid is the prototype of the subclass of

240: the shorter period  variables.

241: 

242:  

243: {\bf SW Tau:} The light curve of this bright BL Her star is that of a

244: Type II Cepheid. The metallicity, [Fe/H] of $+0.2$,  places SW Tau

245: as the most metal-rich star in our sample.

246:  SW Tau is C-rich with the

247: C\,{\sc i} lines  strong by inspection (Figure 1).

248: This figure shows a portion of  the

249: spectrum of SW Tau, IX Cas (a very C-poor star),

250:  CC Lyr 

251: (a star of normal C abundance but having an extreme Fe-deficiency),

252: and BL Her. The four stars have similar atmospheric parameters and, therefore,

253: to first order the variations in the strength of a given line from

254: star-to-star reflect abundance differences between the stars.

255: 

256:   

257: {\bf AU Peg:} AU Peg is a single-lined spectroscopic binary (Harris,

258: Olszewski, \& Wallerstein 1984) with an orbital period of

259: 53.3 days. Harris et al. argue that the secondary is a more

260: massive compact object.

261: A feature of AU Peg is

262: that it lies close to or just beyond the red edge of

263:  the instability strip (Harris et al 1984; Vink\'{o}. Szabados, \&

264: Szatm\'{a}ry 1993).

265:  Presently, AU Peg is close to filling its

266: Roche lobe and mass transfer between the two stars almost certainly

267: occured at an earlier time. Dust, as indicated by an IR-excess (McAlary \&

268: Welch 1986), surrounds the binary. Outflow from the system is suggested

269: by the appearance of  P Cygni-like feature accompanying AU Peg's

270: H$\alpha$ profile (Vink\'{o} et al. 1998).

271: 

272: {\bf DQ And:} Harris (1985) listed this star as a Type II Cepheid

273: on the basis of the estimated distance from the Galactic plane:

274: $|$Z$| = 0.6$ kpc. 

275:  Balog, Vink\'{o}, \& Kasz\'{a}s (1997),

276: put

277: $|$Z$|$ at 2.3 kpc. The radial velocity of $-231$ km s$^{-1}$

278: (Harris \& Wallerstein 1984) is certainly not that expected of a Pop. I

279: star.

280: However, the Baade-Wesselink radius  

281: puts DQ And on the period-radius (P-R) relation defined

282: by Classical Cepheids (Balog et al. 1997).

283: The star is here considered to be of Type II.

284: 

285: 

286: {\bf UY Eri:} With a [Fe/H] of $-1.8$ and

287: a radial velocity of 171 km s$^{-1}$ (Harris \& Wallerstein 1984),

288:  UY Eri is obviously a Type II Cepheid

289: and member of the Galactic halo.

290: Its composition is typical of a halo star with [Fe/H] of $-1.8$

291: (McWilliam 1997).

292: 

293:  

294: {\bf TX Del:} Harris (1985) includes the star in his table of

295: Type II Cepheids. Harris \& Welch (1989) found  that the star

296: is a single-lined spectroscopic binary with an orbital period of

297: 133 days. Schmidt et al. (2005) show that the star has varied in

298: mean brightness, a characteristic not associated with a Classical

299: Cepheid. Andrievsky et al. (2002) consider TX Del to be a

300: first-overtone Classical Cepheid.\footnote{Our derived abundances

301: are in quite good agreement with those by Andrievsky et al. (2002);

302: the mean difference in absolute abundance from 12 elements in common is

303: $-0.15$ dex.}

304:  Although this identification may

305: ease an explanation for the star's  solar metallicity,

306: it places this Pop. I object more than  1 kpc from the

307: Galactic plane. The Baade-Wesselink radius (Balog et al. 1997)

308: places TX Del, as it does DQ And,

309:  on the P-R relation for Classical Cepheids. 

310: An explanation as a runaway star  seems

311: unlikely given that TX Del is a binary. We consider TX Del

312: to be a Type II Cepheid and recall, as did Harris \& Welch, the

313: association of Pop. II characteristics and near-solar metallicity

314: held by metal-rich RR Lyrae stars.

315: 

316: 

317: 

318: {\bf IX Cas:}  A Type II star according to Harris (1985) who gave

319: $|$Z$|$ as 0.7 kpc. This is the primary of a spectroscopic binary (Harris \&

320: Welch 1989).

321: The Baade-Wesselink radius puts the star on the Type II P-R

322: line. 

323: The pulsation period of 9.2 days is long for a BL Her star but the

324: high Na abundance which  distinguishes BL Her from W Vir stars

325: places this star among the BL Her class.

326: The  outstanding mark of IX Cas's spectrum  is

327: the weakness of the C\,{\sc i} lines.

328:  Figure 1 shows that the C\,{\sc i} lines

329: typically used in our analyses are weak or absent from the spectrum of

330: IX Cas. Lines near 9050 \AA\ which are normally too strong for use in the

331: abundance analysis

332: had to be used in the case of IX Cas.

333: The low C abundance is

334: reminiscent of values found for the

335: weak G-band giants (Sneden et al. 1978).

336: 

337: 

338: 

339: {\bf AL Vir:} AL Vir is a Type II Cepheid according to distance from the

340: Galactic plane (Harris 1985), light curve (Schmidt et al. 2004a), He\,{\sc i}

341: 5876 \AA\ emission (Schmidt et al. 2004b), and the Baade-Wesselink

342: radius (Balog et al. 1997).

343: AL Vir with a pulsation period of 10.3 days marks the transition from

344: BL Her to W Vir stars in our listing of variable by increasing period.

345: The composition (particularly the lack of a Na

346: overabundance) suggests AL Vir is a W Vir star.

347: 

348: 

349: {\bf AP Her:} Harris (1985) considered AP Her a Type II Cepheid on the

350: basis of the estimated $|$Z$|$ of 0.4 kpc. 

351: At [Fe/H] $\simeq -0.8$, this may be considered a thick disk star.

352: The literature on

353: Type II Cepheids is peppered with remarks about the difficulties in

354: distinguishing between Type I and II Cepheids. This is well illustrated

355: by Schmidt et al. (2004a) who, in a discussion of stars with large period

356: changes, put  AP Her with stars that `are likely type II Cepheids'

357: but later place 

358: AP Her  among the variables

359:  `for which the predominance

360: of the evidence indicates  type I classification'.

361: The [Fe/H] 

362: confirms that AP Her is not a Classical Cepheid. 

363: 

364: 

365: 

366: {\bf CO Pup:} The literature on this star is extremely sparse. Following

367: Harris's (1985) listing of CO Pup as a Type II Cepheid, the star has

368: featured in just five papers (according to SIMBAD), all providing

369: or regurgitating photometry with not a single spectroscopic observation

370: reported. There appears to be confusion as to whether it is a Type I or

371: II star. Our estimate of the intrinsic [Fe/H] of $-0.6$ 

372: suggests that it is not a Classical

373: Cepheid.

374: 

375: {\bf SZ Mon:} This Cepheid is not listed by Harris (1985). Apparently, it

376: was assumed to be a Classical Cepheid. This assumption

377: was rejected

378:  by Stobie (1970) and Lloyd Evans (1970), largely on the

379: grounds that alternating minima are of different depths. The star is either a

380: W Vir or a RV Tau variable. If the latter description

381: is correct, the fundamental period is approximately 32.6 days.

382: SZ Mon has a marked infrared excess  

383:  (McAlary \& Welch 1986).

384: 

385: {\bf W Vir:} The prototypical Type II Cepheid with an extensive record of

386: published photometric and spectroscopic observations.

387: As befits the prototype, its composition is broadly representative of the other

388: W Vir stars. Our results are in fair accord with the first results

389: given long ago by Barker et al. (1971).

390: 

391: {\bf MZ Cyg:} Type II according to Harris (1985) with $|$Z$|$ of 0.9 kpc.

392: The light curve (Schmidt et al. 2004a) is unlike that of a Classical

393: 21 day Cepheid.  Emission in H$\alpha$ and He\,{\sc i} 5876 \AA\ occuring

394: coincident with the bump on the rising light curve 

395: is taken to 

396: be a discriminant between Type II and Type I Cepheids. Schmidt

397: et al. (2004b) report such emission for MZ Cyg and thus

398: confirm its designation as a Type II Cepheid. 

399: The [Fe/H]  reminds one of the metal-rich RR Lyraes with

400: Pop. II kinematics.

401: 

402: 

403: {\bf CC Lyr:} Harris (1985) estimated a $|$Z$|$ of 2.7 kpc

404:  and, hence, identified it as a Type II Cepheid.

405: Photometry shows alternating

406: minima of different depths, a characteristic of RV Tau variables

407: (Schmidt et al. 2004a). CC Lyr has a strong infrared excess, also

408: invariably a characteristic of RV Tau variables. 

409: This 

410: star's spectrum betrays an obvious signature

411:  of dust-gas separation, a mark of warm RV Tau variables.

412: CC Lyr's spectrum

413: contains  few lines, a reflection of the fact that the atmosphere

414: is highly depleted in many elements including Fe.

415: Figure 1 shows C\,{\sc i} lines but not  the Fe\,{\sc i}

416: and Ni\,{\sc i} lines seen in spectra of the other three stars.

417: The intrinsic metallicity is about [A/H] $\simeq -0.8$ as indicated by the S and Zn

418: abundances which we assume are undepleted. The measured iron abundance is

419: [Fe/H] $\simeq -4$ which corresponds to a dust-gas depletion of more

420: than three orders of

421: magnitude!

422: 

423: 

424: {\bf RX Lib:} Harris (1985) put RX Lib in his table; the estimated

425: $|$Z$|$ was 3.2 kpc. Earlier, Harris \& Wallerstein (1984) placed RX Lib in

426: a category `Stars thought not to be Cepheids' remarking that the

427: type for this star is `uncertain'. Apparently, observations -- spectroscopic

428: and photometric -- have not been published since the mid-1980s on 

429: this star. 

430: Although the type may be `uncertain', RX Lib's composition is 

431: similar to that of W Vir.

432: 

433: 

434: 

435: {\bf TW Cap:} The metallicity, [Fe/H] $= -1.8$, and essentially

436: a normal composition for a halo star suffice to identify this as

437: a Type II Cepheid. The star's position in the P-R diagram

438: supports the Type II designation (B\"{o}hm-Vitense et al. 1974).

439:  Harris (1985) classified the star thus on the

440: basis of its estimated 2 kpc distance from the Galactic plane.

441: 

442: 

443: {\bf V1711 Sgr:} Harris (1985) listed this star as a Population II

444: Cepheid. In the limited literature on V1711 Sgr, there is

445: a lone dissenting voice about the classification. Berdnikov \& Szabados (1998)

446: find a difference in amplitude between their light curve and that obtained

447: about two decades earlier by Dean et al. (1977) and suggest that the

448: star may be a SRd variable. 

449: Presence of an IR excess (Lloyd Evans 1985; McAlary \& Welch 1986;

450: Smith 1998) and a composition bearing

451: the signature of dust-gas separation (see below) together

452: with the long period suggest that the star may be related to the

453: RV Tau variables.

454: We follow Harris and include the star among our sample of Type II

455:  Cepheids. The period of 28.56 days reported Berdnikov \& Szabados (1998)  

456: is adopted and not the 30.5 days listed by Harris (1985).

457: 

458: 

459: 

460: 

461: 

462: 

463: 

464: 

465: 

466: \section{Implications of the  Chemical Compositions}

467: 

468: \subsection{The Evolutionary Context}

469: 

470: 

471: Type II Cepheids in the instability strip above the horizontal

472: branch have evolved from stars on the 

473: blue  end of the horizontal branch (BHB) where the stars are He-core burners. 

474: Placement of a star on  the horizontal branch

475: following evolution to the tip of the first red giant branch

476: requires substantial mass loss by processes

477: as yet unidentified.

478: Seminal calculations about Type II Cepheids

479:  were reported by Gingold (1974, 1976, 1977, 1985).

480: Gingold (1985, Figure 1) shows two evolutionary tracks starting from

481: the BHB that may bracket the

482: origins of the Type II Cepheids. Figure 2 is an adaptation of Gingold's

483: figure.

484: The form of the track appears to depend principally on the

485: mass of the envelope (relative to the core mass) and weakly on the

486: initial metallicity.

487: 

488: In the simpler of the two tracks,   the blue horizontal branch

489: star evolves to the red, crosses  the instability strip as a BL Her variable,

490: and evolves up the AGB to endure thermal pulses

491:  before leaving the AGB

492: tip for rapid evolution to the blue across the instability

493: strip as a RV Tau star and  a brief experience as

494: as a post-AGB star. We refer to this type of evolution as following a track-direct.

495: 

496: In the alternate path, a star from the  more extreme blue part of

497: the BHB

498: evolves to the red, crosses the instability strip to

499: approach the AGB but  experiences a structural readjustment

500: between the H and He shell burning shells that directs the evolutionary

501: track back to the blue. The star executes a `bluenose' (Gingold's

502: parlance) involving two more transits across the instability strip

503: and returns to the AGB from which after increasing in luminosity

504:  it can make

505: its final departure across the instability strip to the blue as a

506: post-AGB star. This track makes four crossings of the instability

507: strip: the first three with a period representative of BL Her variables and 

508: the final one as a W Vir or RV Tau variable. 

509: We refer to this type of evolution as following a track-bluenose.

510: 

511: In both  evolutionary tracks, the

512: luminosity difference between  the earlier one or three  and the final

513: crossings of the instability strip is the key to  

514: the period gap in the distribution function of

515: Type II Cepheids between the BL Her and W Vir

516: variables. This distribution necessarily  depends

517: on the  time taken to transit the instability strip on each

518: crossing

519: (Gingold 1985) and on the rate of production of stars evolving along a

520: track-direct and a track-bluenose.

521: 

522: 

523: Metal-rich RR Lyraes discovered by Preston (1959) long posed

524: a puzzle because

525: metal-rich low-mass stars after the He-core flash

526: settle as He-core burning giants to the red of the RR Lyrae gap, the

527: so-called clump giants.

528: A clump He-core burning

529:  giant evolves   up the AGB where

530: episodic thermal pulses begin at high luminosity.

531:  Such pulses

532: are not predicted

533: to lead to excursions into the instability strip to the

534: blue of the AGB (see Iben 1991 - his Fig. 5 and Gingold 1985 - his Fig. 1).

535:  Taam,

536: Kraft, \& Suntzeff (1976) proposed that severe mass loss would

537: place  giants with a low mass envelope away from the clump and on the 

538: horizontal branch in or to the blue of the RR Lyrae gap.

539: Subsequent

540: evolution of these horizontal branch stars 

541: produces a metal-rich BL Her or W Vir, as above.

542:   Gingold (1977)

543: also showed that even stars placed on the red horizontal

544: branch would, if the envelope mass were reduced sufficiently, evolve 

545: first to the blue. 

546: Taam et al.  showed that kinematically the metal-rich RR Lyraes were

547: members of the old (now, thick) disk. 

548: 

549: How thick-disk giants lose the required amount of mass (about $0.5M_{\odot}$

550: -- Taam et al. 1976) and reduce their envelope to a few per cent of a solar

551: mass is  unknown. Observers may speculate that the He-core flash that

552: precedes He-core burning may lead in rare but sufficient cases to

553: internal violence and mass loss. This oft-invoked speculation

554: has yet to find a resonance in theoretical calculations -- see recent

555: reports by Deupree (1996) and Dearborn, Lattanzio, \& Eggleton (2006).

556: In the absence of observationally confirmed 

557: theoretical ideas on how severe mass loss

558: is achieved, it is impossible to predict the changes of surface

559: composition expected of BL Her, W Vir, and RR Lyr stars relative

560: to their initial (main sequence) composition.

561: Horizontal branch stars including RR Lyr variables must be  affected

562: by the first dredge-up occuring at the base of the

563: red-giant branch and by other subsequent events such as the

564: mass loss required to account for a star's position on the

565: horizontal branch. Principal effects will be on C, N, and O,

566: possibly He, and certainly Li, Be, and B.

567: 

568: 

569: Three BL Her stars in our sample (IX Cas, TX Del, and AU Peg)

570: and the W Vir star ST Pup (Gonzalez \& Wallerstein 1996) are known spectroscopic

571: binaries for which previous mass transfer may have directed their

572: primary star to the instability strip. Although other binaries

573:  may lurk undetected

574: in the sample, well observed stars like W Vir must, if accompanied

575: by a companion, have a very low velocity amplitude.

576: 

577: \subsection{Observed composition and period}

578: 

579: The BL Her stars and the

580: W Vir stars occupy different positions

581:  along the evolutionary tracks from the blue horizontal

582: branch to the post-AGB departure from the AGB;  the BL Her stars are

583: closest to the blue horizontal branch and the W Vir stars are presumed

584: to be leaving the

585: AGB. In light of this difference in staging along the evolutionary

586: tracks,

587: we distinguish the two classes in discussing their

588: compositions. 

589: 

590: The BL Her variables are stars with pulsation periods of four

591: days or shorter plus TX Del and IX Cas, stars of intermediate

592: period but with the  signature of BL Her stars, i.e, an overabundance of

593: Na. The BL Her stars have on average a higher [Fe/H] than the W Vir

594: stars.

595: The W Vir variables are stars

596: with pulsation periods of greater than about

597:  10 days plus the intermediate

598: period stars AL Vir and AP Her.

599: Among the W Vir stars  is CC Lyr which, as noted

600: above, has a composition indicative of severe dust-gas separation,

601: a common feature of RV Tau variables. Other W Vir stars offer

602: indications of dust-gas separation and some

603: have light curves suggestive of RV Tau-like pulsations. 

604: 

605: The compositions of the stars are given in Table 2 for the

606: BL Her stars, Table 3 for the intermediate period quartet, and

607: Table 4 for the W Vir stars. In Figures 2 to 5 displaying the results

608: as [X/Fe] versus [Fe/H], the BL Her stars (including TX Del and IX Cas)

609: are represented by unfilled circles, the W Vir stars (including

610: AL Vir and AP Her) by filled circles. In addition to our sample, we plot

611: the previously published results for

612: the C-rich BL Her stars  -- V553 Cen (Wallerstein \& Gonzalez 1996)

613: and RT TrA (Wallerstein, Matt, \& Gonzalez 2000) -- as unfilled

614: squares, the W Vir stars ST Pup (Gonzalez \& Wallerstein 1996) 

615: and $\kappa$ Pav (Luck \& Bond 1989) as filled triangles.

616: 

617: Studies of the kinematics of Type II Cepheids show that the examples

618: in the solar vicinity are primarily  thick disk stars with

619: an admixture of halo stars. 

620: Initial compositions of thick disk and halo stars are now known

621: as a function of metallicity.

622: The thin and thick disk do not have identical compositions

623: as a function of metallicity (Bensby et al. 2005; Reddy et al.

624: 2006). The differences  at a given [Fe/H] are not large. 

625: Of relevance to our discussion is the fact that at a given [Fe/H]

626: the scatter in initial abundance ratios, i.e., [X/Fe], is very small

627: for both the thin and the thick disk, at least for the stars now in the

628: solar neighborhood (Edvardsson et al. 1993; 

629: Reddy et al. 2003, 2006). At a common  low [Fe/H], the composition of

630: thick disk and halo stars appear to merge but the scatter in [X/Fe]

631: at a given [Fe/H] may be more significant for the halo population. Since

632: our sample includes just two halo stars, this scatter

633: in composition, if real, is unimportant here.

634: 

635: The compositions of the variables are judged relative to their

636: presumed initial compositions as inferred (usually) from their

637: [Fe/H], but, an alternative is considered in the cases where

638: dust-gas separation is suspected.

639: 

640: 

641: 

642: \subsection{The BL Herculis variables}

643: 

644: Our sample is augmented by published analyses of the 

645: BL Her C-rich variables V553 Cen and RT TrA.\footnote{Other 

646:  C-rich variables

647: listed by Lloyd Evans (1983) have periods of 20 days or longer. None

648: have been subject to quantitative analysis.}

649: Consideration of the C, N, and O abundances led Wallerstein

650: and colleagues to propose that the atmospheres of the

651: two C-rich variables had been enriched

652: in $^{12}$C from the $3\alpha$--process followed

653: by operation of the H-burning CN-cycle.

654: The high C abundances pointed to the $3\alpha$-process. Operation of the

655: CN-cycle was indicated by the high N abundances.

656: Measurement of the $^{12}$C/$^{13}$C ratio for both stars

657: gave results

658:  equal to the equilibrium value for the CN-cycle, an indication that

659: CN-cycling occurred following the $^{12}$C enrichment from the

660: $3\alpha$-process.

661: A Na enrichment was attributed to proton capture on $^{22}$Ne 

662: occuring at the time of  H-burning.

663: This combined operation of the H-burning CN cycle and the He-burning $3\alpha$-process

664: and attendant  reactions may account for the BL Her stars

665: analyzed here.  (The halo variable UY Eri is not discussed

666: further.)

667: 

668: Prior to a star's transfer to the horizontal branch, the first dredge-up

669: increased the surface N abundance at the expense of the C

670: abundance. The maximum possible N abundance post dredge-up is obviously

671: equal to the sum of the initial C and N abundances. Oxygen is not

672: predicted to be reduced by the dredge-up. The C and N

673: abundances show clearly that the first dredge-up has not been the

674: primary influence on the stars' surface abundances. The C

675: abundance for all but one star (IX Cas) equals or exceeds the initial

676: abundance. The N abundance except for IX Cas and DQ And exceeds by

677: about 0.6 dex that

678: predicted by complete conversion of initial C and N to N.

679: The N abundance is approximately that

680: expected by complete conversion of initial C, N, and O to N but the

681: O abundance is not depleted but has an abundance similar to or

682: even slightly greater than the presumed initial abundance.

683: One star -- SW Tau -- is C-rich with a C/O ratio of about 3.  A

684: second -- BL Her -- is borderline C-rich with C/O of 0.9. With

685: the remarkable exception of IX Cas (C/O $= 0.005$), SW Tau, and BL Her,

686:  the other BL Her

687: stars have a C/O ratio in the range 0.1 to 0.4.

688: The C, N, and O abundances  signal the presence of

689: $3\alpha$-process  and CN-cycle products.

690: 

691: A sodium overabundance 

692: is an obvious feature of the eight BL Her variables 

693:  (Figure 3).

694: The [Na/Fe] ratio over the interval $-0.5 <$ [Fe/H] $< +0.2$

695: is independent of [Fe/H] with a spread of about 0.5 dex and

696:  a mean [Na/Fe] $= +0.73$, a

697: substantial increase over the ratio of $+0.12$ for the

698: thick disk (Reddy et al. 2006). Wallerstein and colleagues

699: obtained a  Na overabundance for their pair of

700: C-rich BL Her variables: [Na/Fe] =  $+0.74$ (RT TrA) and $+0.43$ (V553 Cen).

701: The mean [Na/Fe] for the W Vir stars excluding the halo star TW Cap, and

702: the stars (ST Pup, CC Lyr, and V1711 Sgr) affected by dust-gas separation is

703: [Na/Fe] $= +0.23$, a value consistent with the ratio of unevolved thick

704: disk stars.

705: 

706: The Al abundances of the BL Her stars

707: give a mean 

708: [Al/Fe] $= +0.13$, a value confirmed by the pair of C-rich

709: variables. The mean [Al/Fe] is 

710: slightly less than that of thick disk stars for which Reddy et al. (2006)

711: gave [Al/Fe] $= +0.30$.  The mean [Na/Al] of the BL Her stars is

712: $+0.60$ but the presumed initial ratio for the thick disk stars

713: is $-0.18$, a factor six change in the Na/Al ratio occuring in the

714: course of evolution to a BL Her variable. The W Vir stars show a

715: wide range in Al abundance  which we attribute to 

716: dust-gas separation but the upper limit in the range is consistent

717: with the ratio for thick disk stars.

718: 

719: The $\alpha$-elements, here Mg, Si, S, Ca, and Ti, conform to

720: expectation. At thick disk metallicities,

721:  the ratios [$\alpha$/Fe] are positive for the dwarfs:

722: [$\alpha$/Fe] =  $+0.32$ (Mg), $+0.22$ (Si), $+0.18$ (Ca), and

723: $0.21$ (Ti)

724:  at [Fe/H] $\simeq -0.5$ (Reddy et al. 2006).

725:  (Sulphur was not examined by Reddy et al. but a value

726: for [S/Fe] close to those of Mg and Si is expected.) Thick disk stars

727: with [Fe/H] $> -0.3$ have thin disk abundances with a decline of [$\alpha$/Fe]

728: to zero at about [Fe/H] = 0. 

729: The observed [$\alpha$/Fe] ratios versus [Fe/H] are shown in Figure 4. They

730:  match expectation well for the composite

731: index from Mg, Si, and S. The observed [Ca/Fe] for [Fe/H] $< 0$

732: stars is as expected but for the low [Ca/Fe] for AU Peg.

733: AU Peg also shows a low [Ti/Fe]. 

734: 

735: 

736: 

737: Iron-group abundances (here, Cr, Mn, and Ni) shown in Figure 5

738: as a function of [Fe/H] follow well the trends expected of

739: disk stars (Reddy et al. 2006):  [Cr/Fe] $\simeq$ [Ni/Fe] $\simeq 0$

740: and [Mn/Fe] $\leq 0$.

741: Two of the stars -- BX Del and TX Del -- appear to be Sc-poor

742: relative to expectation. Both are also Ti-poor.

743:  The Zn abundances follow expectation,

744: i.e., [Zn/Fe] $\simeq 0$, with the possible exception of

745: DQ And. 

746: 

747: 

748: 

749: 

750: 

751: Abundances of heavy elements in Type II Cepheids  have occasioned

752: considerable comment beginning with Rodgers \& Bell's (1963)

753: analysis of $\kappa$ Pav where they reported an underabundance

754: by about a factor of seven of $s$-process elements (relative to

755: the Fe abundance). A new analysis of $\kappa $ Pav was made by

756: Luck \& Bond (1989) who found [$s$/Fe] $= -0.4$ at [Fe/H] $= 0.0$.

757: The $s$-process contribution is here compiled as a simple mean of the

758: derived Y, Zr, La, and Ce abundances. Neodymium with roughly equal

759: $s$- and $r$-process contributions at the solar composition (Burris et al. 2000)

760: was not included but its inclusion would not alter any

761: conclusions. The expected [$s$/Fe] for the thick disk is

762: close to zero. Figure 6 shows that this expectation is

763: found for five of our seven BL Her variables, and 

764: C-rich V553 Cen, and RT TrA. Three BL Her stars -- BX Del, VY Pyx,

765: and TX Del -- have [$s$/Fe] $\simeq -0.5$, a value far from

766: the range shown by thick (and thin) disk unevolved stars. 

767: There is a tantalising hint that the [$s$/Fe] values offer a

768: bimodal distribution. The low-$s$ stars are discussed

769: further in the next section on the W Vir stars for which

770: underabundance of $s$-process elements is common.

771: 

772: 

773: 

774: Europium is taken as the sole measure of the $r$-process

775: contribution to the stellar composition; Eu in the solar

776: composition is about 97\% $r$-process (Burris et al. 2000).\footnote{Luck

777:  \& Bond

778: adopted the mean of their measured elements from La to Eu

779:  as an indicator  of the

780: $r$-process. This

781: choice includes La and Ce which in the solar mixtures

782: are

783: about 80\% $s$-process and  only 20\% $r$-process.}

784: Europium abundances for the BL Her  variables

785: are shown in Figure 7. A tight relation is seen with a

786: tendency for [Eu/Fe] to increase  with decreasing [Fe/H].

787: This relation mimics that for the thick disk but is displaced

788: at a given [Fe/H] to lower [Eu/Fe] by about 0.2 dex. 

789: One presumes that this is an artifact of the analysis and not a

790: consequence of stellar evolution.

791: 

792: Among the W Vir stars,  dust-gas separation is

793: unmistakably severe in the case of CC Lyr and suspected in

794: several other cases. The signatures of severe dust-gas separation

795: are [S/Fe] and [Zn/Fe] ratios in excess of their normal

796: ratios for thick disk stars

797:  of about $+0.3$ and 0.0, respectively, and negative 

798: ratios for [Al/Fe], [Sc/Fe], and [Ti/Fe]. 

799: In mild cases,  the effects of separation are limited to the

800: elements of highest condensation temperature (Al, Sc, and Ti) and, then 

801: [Al/Fe], [Sc/Fe], and [Ti/Fe] show negative values but [S/Fe] and [Zn/Fe]

802: have  normal values. 

803: In our sample of BL Her variables, none show the signature of

804: severe dust-gas separation. The trio with a low [$s$/Fe], also a

805: signature of depletion of the highest condensation

806: temperature elements, provide weak hints of depletion of Al, Sc, and Ti.

807: 

808: 

809: In summary, the composition of the BL Her variables, as suggested by Wallerstein

810: and colleagues from their analyses of the C-rich pair of variables, has

811: been set by mixing with the products of $3\alpha$- processing, 

812: CN-cycling, and $p$-capture on $^{22}$Ne. Other abundance ratios relative to

813: Fe have their expected values with the  exception of low $s$-process

814: abundances for a minority of the group. 

815: 

816: 

817: 

818: \subsection{The W Virginis Variables}

819: 

820: The W Vir stars as a class show evidence in their C, N, and O

821: abundances for the presence of $3\alpha$-process and CN-cycle

822: products in their atmosphere but not the Na overabundance

823: seen for the BL Her variables. A signature of weak dust-gas

824: separation seems present in some stars with

825: a clear signature of severe separation present for

826: CC Lyr and ST Pup (Gonzalez \& Wallerstein 1996).

827: The sample of W Vir stars has on average a lower (intrinsic) [Fe/H]

828: than the  sample of BL Her stars.

829: 

830: The first dredge-up is clearly not the defining part of the evolutionary

831: history. For four stars, although the N abundance may be equal to

832: the sum of the initial C and N abundances, the C abundance either

833: remains equal to the initial abundance or exceeds it by as much as

834: 0.8 dex. In two cases, the N abundance exceeds the sum of the

835: initial C and N abundances. The C/O ratios range from

836: about 2 for MZ Cyg and SZ Mon to 0.06 for W Vir.

837:  Although lacking a

838: counterpart to IX Cas with its low C abundance, the spread in

839: C, N, and O abundances is similar to that of the BL Her variables.

840: 

841: 

842: 

843: 

844: 

845: 

846: The striking difference in the [Na/Fe] ratios of the BL Her and

847: W Vir stars  (Figure 3) is a clue to their different origins.

848: Setting ST Pup and V1711 Sgr

849: aside, the [Na/Fe] of the

850: W Vir stars appears independent of [Fe/H] and pulsation period, and consistent

851: with that found for the thick disk dwarfs but significantly

852: less than that of the BL Her variables.

853: The exceptions, both with the high [Na/Fe] representative of the BL Her

854: variables, are ST Pup and V1711 Sgr. ST Pup (Gonzalez \& Wallerstein 1996) is

855: seriously affected by dust-gas separation such that neither the present Fe and

856: nor most likely also the Na abundance are unaffected. It is impossible at

857: present to correct the observed [Na/Fe] for the effects of dust-gas

858: separation but a downward revision is certain.

859: V1711 Sgr may be a Na-rich W Vir star but distortion of the [Na/Fe]

860: ratio by dust-gas separation effects cannot be excluded.

861: Although V1711 Sgr is not as dramatically affected by dust-gas

862: separation as ST Pup, the [Na/Fe] may have been increased

863: sufficiently (say, by 0.3 dex) to place it among the BL Her stars.

864:  

865: Aluminum is in the main less abundant in W Vir than in BL

866: Her stars (Figure 3). All W Vir stars show a [Al/Fe] less than

867: that of the thick disk dwarfs. ST Pup's positive [Al/Fe]

868: is anomalous for a star strongly affected by dust-gas separation

869: (see below). 

870: 

871: Abundances of the $\alpha$-elements (Figure 4) show a

872: greater spread for [Ca/Fe] and [Ti/Fe] versus [Fe/H] than is the case for the

873: BL Her stars. The spread is greater for [Ti/Fe]

874: than for [Ca/Fe] and almost absent for [Mg/Fe], [Si/Fe], and

875: [S/Fe]. The appearance of negative [Ca/Fe] and [Ti/Fe] is consistent

876: with the operation of dust-gas separation. 

877: 

878: The $r$-process element Eu defines a similar  [Eu/Fe] versus [Fe/H]

879: relation for the BL Her and W Vir stars  which incorporates the two C-rich

880: BL Her stars (Figure 6).

881: 

882: The $s$-process abundances show apparently bimodal values of

883: [$s$/Fe] of about 0.0 or $-0.7$, a tendency similar to that

884: provided by the BL Her stars. Given that [$s$/Fe] $\simeq 0.0$

885: is representative of the thick disk, the anomalous stars

886: are those with negative [$s$/Fe]  values. 

887: As noted above, the $s$-poor stars have negative [Ti/Fe] and not the

888: positive values expected of thick disk stars and negative [Sc/Fe] with

889: values of about $-1$ in four cases. (The W Vir star $\kappa$ Pav, as

890: analysed by Luck \& Bond (1989),  is a typical $s$-process deficient

891: member of the class.)

892: 

893: Longer-period W Vir stars and shorter-period  RV Tau stars  overlap in

894: period and, therefore, it of interest to comment on similarities and

895: differences in composition. 

896: The C, N, and O abundances of the two groups have very similar

897: ranges and mean values (Giridhar et al. 2005).

898: Of especial interest in the [Na/Fe] ratio of

899: RV Tau stars. Since many RV Tau stars show a Fe-depletion from dust-gas

900: separation, we compare not [Na/Fe] but [Na/Zn] because Zn is generally not

901: depleted and [Zn/Fe] $\simeq 0$ in unevolved stars. 

902:  We compile [Na/Zn] for RV Tau stars

903:  from results given by Giridhar et al. (2005)

904: and earlier papers in that series. The mean value from 21 stars is

905: [Na/Zn] $= +0.17\pm0.03$ to be compared with [Na/Zn] $= +0.20$ for our

906: W Vir stars and $+0.85$ for our BL Her stars. Excluded from our sample of

907: RV Tau stars are two stars with the high [Na/Zn] characteristic of the BL Her

908: stars, CE Vir where the abundance anomalies are not due to dust-gas

909: separation (Rao \& Reddy 2005), and EP Lyr where the dust-gas separation is

910: so severe that Na is somewhat depleted. The conclusions are that the

911: distribution of [Na/Zn] ratios  (i.e., the [Na/Fe] in the absence of dust-gas separation) of the W Vir

912:  and the  RV Tau stars 

913:  are identical with the majority showing no sodium enrichment but a

914: minority of perhaps 10 per cent showing the sodium enrichment of the

915: BL Her stars. 

916: 

917: 

918: 

919: \subsection{Accounting for the anomalies}

920: 

921: Introduction to  the stellar surface of products of internal

922: nucleosynthesis is a well known phenomenon accounting in principle  

923: for a surface composition differing from the 

924: presumed initial composition. The classic case must be that

925: of those S stars with Tc present at the surface. The

926: qualifying phrase `in principle' recognises that the details

927: of the internal nucleosynthesis and dredge up to the

928: surface may not yet be understood. Often, the particular processes of

929: nucleosynthesis involved are identifiable and it is the modes of transport of

930: the nucleosynthetic products to the atmosphere that are in

931: doubt.

932: 

933: Such would appear to be the case for the BL Her stars. The

934:  atmospheres are contaminated,

935:  as Wallerstein and colleagues first showed from their

936: analyses of two C-rich BL Her stars, with

937: $3\alpha$-processed material followed by  exposure to

938: the CN-cycle.

939: This recipe which accounts in principle for the C, N, O, and

940: Na abundances of the BL Her stars

941: has yet to be

942: incorporated into a stellar evolution model that ties

943: the nucleosynthesis and mixing to  events in the life of

944: a low-mass star. 

945: The sole remaining abundance anomaly is the appearance

946: of a low [$s$/Fe] in three stars. There are hints of anomalously

947: low Ca, Ti, and Sc in a minority. Such hints lack an obvious

948: nucleosynthetic explanation, even in principle.

949: 

950: In searching for the origin of the abundance anomalies of the

951: Population II Cepheids, one may begin with the Na  abundance

952: difference between the W Vir and  BL Her stars. 

953: Recall that the W Vir stars are considered to have evolved  from

954: BL Her stars. Sodium destruction is highly unlikely to have

955: occured along the track between the BL Her and W Vir phases. 

956: Thus, it seems clear that the W Vir stars observed here

957: cannot be direct descendants of the BL Her stars observed here;

958: V1711 Sgr may be an exception.

959: Since, sodium production is highly unlikely to be a feature of

960: the short evolutionary phase from the BHB to the instability strip, the

961: overabundance of sodium in BL Her stars

962:  must be attributed to the as yet

963: unknown processes involving mixing and mass loss

964:  that placed the BL Her progenitors on the BHB.

965: Continuing this thought we conjecture that the processes

966: which placed the W Vir progenitors on the BHB did not

967: provide for a sodium overabundance. 

968: The unknown processes in both cases involved $3\alpha$-process and

969: the CNO-cycles to similar degrees. Given that the distributions of

970:  C, N, and O abundances are

971: similar for BL Her and W Vir stars, the difference involving the unknown

972: processes would appear to have

973: been the temperature at which the CNO-cycles operated.  Sodium 

974: production by proton capture on $^{22}$Ne occurs at `high' but not

975: at `low' temperatures. We speculate that mixing and mass loss involving

976: CNO-cycling at high temperatures placed the star near the red end on the

977: BHB with the star evolving along a track-direct to become a Na-rich

978: BL Her star. When CNO-cycling at low temperatures was involved, the

979: star was deposited  at the blue end of the BHB. 

980: This disposition of progenitors along the BHB and evolution of BHB stars

981: to BL Her and W Vir stars is discussed further in Section 6.

982: 

983: 

984: The   anomalies common   among  W Vir stars but 

985: uncommon among BL Her stars 

986: are very unlikely to have a nucleosynthetic

987: origin. Anomalies include the

988: low [Al/Fe], 

989: [Ca/Fe], [Sc/Fe],  [Ti/Fe], and

990: [$s$/Fe] ratios. 

991: A catalog of explanations for the anomalies includes

992:  the standard invocation of errors in the

993: atmospheric parameters, non-LTE effects,  the inapplicability of standard

994: atmospheres invoking hydrostatic equilibrium to a low density atmosphere

995: subject to a pulsation, 

996:  non-LTE overionization

997: of atoms and ions by Lyman continuum photons from a shock wave in the pulsating

998: atmosphere,

999:  substantial helium enrichment

1000: leading to a systematic overestimate of (say) [Fe/H] but to small(er) effects

1001: on ratios such as [$s$/Fe],

1002: and  reduction of the abundance of refractory elements by dust-gas separation in

1003: the upper atmosphere or in a circumbinary disk.

1004: 

1005: Luck \& Bond (1989) showed that errors  in the atmospheric

1006: parameters cannot erase the low [$s$/Fe] values, and by extension of 

1007: their arguments, other  anomalous underabundances (relative to Fe)

1008: cannot be attributed to injudicious choices of atmospheric

1009: parameters. Non-LTE effects computed for standard atmospheres deserve

1010: examination but are unlikely to account for example for differences

1011: in abundance anomalies between and within the samples of

1012:  BL Her and W Vir variables where stars of similar atmospheric

1013: parameters can differ in the degree of their anomalies.

1014: 

1015: The possibility of Lyman continuum emission leading to departures from

1016: LTE ionization equilibrium was mentioned by Barker et al. (1971)

1017: following a suggestion by Wallerstein, and discussed in more detail

1018: by Luck \& Bond (1989). Since the observable heavy elements from Y to Eu

1019: are present in these atmospheres

1020:  predominantly as singly-charged ions, all with ionization

1021: potentials less than ionization potential of H, the idea has an

1022: appeal.

1023:  Emission in Balmer H$\alpha$ and the He\,{\sc i}

1024: 5876 \AA\ lines is a feature of the rising branch for the longer-period

1025: Type II Cepheids for which $s$-process and other underabundances are

1026: common.

1027:  The fact that Eu does not show an abundance scatter like the $s$-process

1028: elements is a strike against the proposal. It also

1029:  does not account for the underabundance of

1030: Al;  Al$^+$ has an ionization potential of 18.83 eV and Lyman emission at

1031: 18.8 eV must be weak and, certainly, Mg present in the atmosphere as Mg$^+$

1032: with an ionization potential of 15.04 eV is never underabundant.

1033:  The idea might be given a  stringent test

1034: by observing a star at intervals throughout its pulsation and

1035: seeing if the derived composition correlates with the presence of a shock as

1036: revealed by line splitting. The weight of the evidence is that emission from

1037: the shock wave is not a major source of the abundance anomalies. 

1038:  

1039: Addition of extensive amounts of $3\alpha$-processed and CN-cycled

1040: material might foster the idea that the reported abundance anomalies are

1041: a consequence of analysing the spectrum of a He-rich atmosphere on the

1042: assumption that it has a normal He abundance. To first order, this

1043: incorrect assumption results in [X/Fe] ratios close to the

1044: actual values but a [Fe/H] that is overestimated.

1045:  Luck \& Bond (1989)

1046: speculated that a  low [$s$/Fe] ratio,

1047:  a feature of normal stars of [Fe/H] $< -2$,

1048: was being associated with a higher [Fe/H] because of an overlooked

1049: helium enrichment.

1050: The reader is referred to

1051: their paper for arguments leading to the conjecture's rejection. Here, we

1052: note their  point that the kinematics of the  W Vir stars

1053: show that they belong to the

1054: thick disk and not to the very metal-poor halo.  Yet, a thorough attempt to

1055: measure the He/H ratio of the metal-rich variables would be welcomed.

1056: 

1057: 

1058: Dust-gas separation is  evident in RV Tau variables (Giridhar

1059: et al. 2005; Maas, Van Winckel, \& Waelkens 2002)

1060:  except for the  metal-poor and cool variables. In oxygen-rich gas, the heavy

1061: elements are among the first to be removed from the gas as

1062: cooling occurs.

1063: The demonstration by Alcock et al. (1998) that the RV Tau and Type

1064: II Cepheids in the LMC define a single period-luminosity-color

1065: relation suggests that the longer period Type II Cepheids in our sample

1066: may be susceptible to  dust-gas separation. Earlier, we noted that several of

1067: our stars show RV Tau-like photometric behavior and might be

1068: promoted from W Vir to RV Tau status.

1069:  The

1070: signature of dust-gas

1071:  separation, previously reported for ST Pup (Gonzalez \&

1072: Wallerstein 1996),  is strikingly evident here for CC Lyr where

1073: the spectrum contains lines of elements having a low condensation

1074: temperature ($T_{\rm C}$),  i.e.,

1075: lines of C\,{\sc i}, N\,{\sc i}, O\,{\sc i}, Na\,{\sc i}, S\,{\sc i}, and

1076: Zn\,{\sc i}, but lines of higher condensation

1077: temperature are poorly represented or not detected at all.

1078:  Severe underabundances

1079: are measured, e.g., [Mg/H] $\simeq$

1080: [Fe/H] $\simeq -4$ but [S/H] $\simeq$ [Zn/H] $\simeq -0.5$. Figure 7

1081: constructed for CC Lyr

1082: shows the characteristic pattern for a strongly affected RV Tau variable;

1083: the abundances [X/H] decline smoothly with increasing condensation

1084: temperature.

1085: The Mg and Fe deficiencies of CC Lyr are more extreme than any

1086: reported previously for a RV Tau variable. Lines of the elements normally the

1087: most depleted -- Al, Ca, Sc, and Ti, for example -- could not be detected

1088: in our spectra.

1089:  CC Lyr has a considerable

1090: infrared excess from circumstellar dust. Its Fe deficiency is typical of

1091: values reached in the warm post-AGB stars known to be spectroscopic binaries

1092: (Van Winckel, Waelkens, \& Waters 1995).

1093: One supposes that CC Lyr may also be a spectroscopic binary.

1094: ST Pup, a W Vir star with a 

1095: strong infrared excess and signatures of dust-gas separation,

1096:  was shown by Gonzalez \& Wallerstein

1097:  (1996) to be a spectroscopic

1098: binary with an orbital period of 410 days.\footnote{The

1099: dust-gas  signatures for ST Pup are rather ragged but this is possibly

1100: attributable to larger than normal uncertainties in the derived

1101: abundances resulting from the very few lines  available for most of

1102: the elements:

1103: [Fe/H] $= -1.47$ with [S/Fe] $= +1.29$ and [Zn/Fe] $= +1.41$ are

1104: customary markers of severe dust-gas separation but [Al/Fe] $= +0.19$ is not.} 

1105: 

1106: 

1107: 

1108: The signature of dust-gas separation seems present in other W Vir stars,

1109: all of which show negative [Sc/Fe] values. Since

1110: these stars are likely thick-disk members of different initial [Fe/H], we

1111: correct the derived abundances [X/H] for the initial [X/H] of the

1112: thick disk (Reddy et al. 2006) before presenting the $T_{\rm C}$ versus [X/H]

1113: plots; all corrections are small.

1114:  Figure 8 shows the [X/H] versus $T_{\rm C}$ plots for

1115: CO Pup and W Vir.

1116: In such cases, as in the examples of mildly affected RV Tau stars,

1117: the scatter in [X/H] at a given condensation temperature may

1118: be significant. This scatter is attributed to the understandable

1119: failure of the condensation temperature to represent fully the

1120: complex processes of dust formation and accretion of gas but not dust by

1121: the star. 

1122:  The decrease in [X/H] at the highest $T_{\rm C}$ is

1123: clear for CO Pup. For W Vir, this signature of dust-gas separation is

1124: critically dependent on the Sc abundance.

1125: Here and in Figures 9 and 10, the $s$-process elements are

1126: denoted by unfilled circles. Lack of a clear separation between the

1127: unfilled and filled circles referring to elements of similar $T_{\rm C}$

1128: show that the $s$-process products did not mix to the surface in the

1129: time that these W Vir stars were on the AGB.

1130: A clearer example of dust-gas separation

1131:  is provided by V1711 Sgr with Al, Ti, and Sc, all 

1132: underabundant by more than one dex relative to Fe (Figure 9). 

1133: The signature of dust-gas separation also seems evident for SZ Mon, the

1134: W Vir star with a RV Tau-like light curve and a strong infrared excess

1135: (Figure 9). The evidence seems fair   for MZ Cyg but weaker for

1136: RX Lib (Figure 10) where it relies on the low

1137: Sc abundance. In all cases, the $s$-process elements are underabundant,

1138: as expected when Al, Sc, and Ti are underabundant.

1139:  Although several  key elements have not been measured for TW Cap, a halo star,

1140: the indication is that dust-gas separation has not affected this star;

1141: Ca, Ti, Sc, and $s$-process elements have the abundances expected of a 

1142: star with [Fe/H] $-1.8$, a result anticipated from our previous reports on the

1143: absence of dust-gas separation in intrinsically 

1144: metal-poor RV Tau variables (Giridhar et al. 2005). 

1145: 

1146: Binarity has been advanced as a principal key to the dust-gas separation

1147: exhibited by RV Tau stars;  Van Winckel (2003)

1148: suggested that `binarity may very well be a common phenomenon among

1149: RV Tau stars'. In the present sample, the known spectroscopic

1150: binaries -- TX Del, IX Cas, and AU Peg --

1151: are  not strongly affected by dust-gas separation but all are BL Her stars

1152: for which dust-gas separation is not observed.

1153: The affected W Vir star ST Pup is a binary (Gonzalez \& Wallerstein 1996).

1154: Investigating the binary nature of the W Vir stars, not only those

1155: affected by dust-gas separation, will call for intensive

1156: radial velocity monitoring in order to resolve the orbital from the

1157: pulsational velocity variations. It is noteworthy  that the

1158: Fe-deficiency of CC Lyr is similar to that of the A-type

1159: post-AGB stars, all known to be spectroscopic binaries 

1160: (Van Winckel et al. 1995).

1161: 

1162: 

1163: 

1164: 

1165: 

1166: 

1167: In short, the suggestion is clear: the principal reason for the various

1168: abundance anomalies -- C, N, and O apart -- among the W Vir stars is that

1169: their atmospheres have been partially cleansed of refractory elements.

1170: The mass of the

1171: convective envelope must be small to sustain a large deficiency of

1172: refractory elements. Perhaps, not coincidentally a

1173:  small envelope is also a condition required

1174: to place the immediate progenitors of these variables

1175: on the horizontal branch to the blue of the

1176: red clump. There may also be a correlation between the magnitude of the

1177: infrared excess and the visibility of the dust-gas separation. Stars

1178: with a weak dust-gas separation are returning to a normal

1179: composition following dissolution of circumstellar dust.  

1180: The BL Her stars have yet to evolve to the upper AGB and to commence

1181: dust production and, hence, absence of evidence of dust-gas separation

1182: is  not surprising.

1183: 

1184: In discussing the abundance anomalies, we have overlooked 

1185: the  possibility that a contributing factor

1186: might be found in conditions existing in

1187: the progenitors resident on the blue horizontal branch. 

1188: Stars with the lowest ratios of envelope to core mass reside on 

1189: the horizontal branch at effective temperatures ($T_{\rm eff} > 11500$ K)

1190: such that severe atmospheric abundance anomalies are created by gravitational

1191: settling and radiative levitation (Behr 2003), e.g., Fe overabundances

1192: by a factor of 2 dex have been recorded. These stars are

1193: likely to evolve along a track-bluenose. However, one supposes that such

1194: anomalies are erased by the time that the stars have evolved off the 

1195: horizontal branch and into the instability strip. This

1196: supposition deserves observational and theoretical attention.   

1197: 

1198: 

1199: 

1200: 

1201: 

1202: \section{Metallicities from photometry}

1203: 

1204: Accumulation and analysis of high-resolution spectra is a time

1205: consuming process. A variety of statistical properties of the Type II Cepheids

1206: may be more readily examined using a photometric estimate of metallicity

1207: provided that  the estimate is reliable.

1208: Here, we assess the reliability for three photometric systems

1209: which have been applied to Type II Cepheids.

1210: 

1211: Harris (1981)

1212:  used Washington photometry to obtain  the

1213: metallicity [A/H] for a large sample of Type II Cepheids --

1214:  see  Harris \& Wallerstein (1984) for two revisions. There are

1215: 13 stars not including CC Lyr in common with our sample. The comparison

1216: of our [Fe/H] and [A/H] is shown in Figure 11. (The [A/H] for CC Lyr with its

1217: highly non-standard composition is expected to depart appreciably from

1218: the spectroscopic [Fe/H]: indeed, [A/H] of $-2.2$ versus [Fe/H] of about $-4$.)

1219: It is clear from the figure that the photometric and spectroscopic

1220: estimates are well correlated but offset such that [A/H] is overestimated

1221: by about 0.5 dex. The offset is not dependent on the pulsation

1222: period of the star.

1223: The conclusion is that the Washington photometry provides, after a 0.5

1224: dex downward revision,  metallicity estimates of satisfactory quality for

1225: statistical purposes -- see, for example, the distribution functions

1226: for metallicity of Type II Cepheid (Harris 1981).

1227: 

1228: Diethelm (1990) used Walraven {\it VBLUW} photometry to estimate the

1229: metallicity of the shorter period Type II Cepheids. Five stars are

1230: in common with our sample including a  RR Lyrae IK Hya (not discussed

1231: here). Again, 

1232: there is an offset between photometric and spectroscopic estimates:

1233: [A/H] is overestimated by about 0.5 dex for the metal-rich stars

1234: and possibly by 0.8 dex at low [Fe/H] (Figure 11). 

1235: 

1236: Str\"{o}mgren photometry was applied by Meakes, Wallerstein, \&

1237: Opalko (1991). Excluding CC Lyr ([A/H] of $-2.3$), the  common stars

1238: show that  [A/H] and [Fe/H] are  roughly correlated (Figure 11). 

1239:  

1240: In summary, the metallicities may be reliably provided by photometry

1241: except for stars seriously affected by dust-gas separation.

1242: 

1243: 

1244: 

1245: \section{Concluding remarks}

1246: 

1247: Our sample of field Type II Cepheids is the first for which extensive

1248: data on chemical compositions are provided. The study

1249: complements prior modern analyses of $\kappa$ Pav, ST Pup, V553 Cen, and

1250: RT TrA.  Our data on just two

1251: halo (metal-poor) Type II

1252:  Cepheids omit key elements -- e.g., C, N, and Na -- so that

1253: no more than a perfunctory interpretation of these stars is offered.

1254: Common to the BL Her and W Vir stars is evidence for addition to

1255: their atmospheres of $3\alpha$-processed and CN-cycled material.

1256: The distinguishing mark of the BL Her stars, the shorter period Type II

1257: Cepheids, is their overabundance of Na, an overabundance not

1258: detected among the W Vir stars. The distinguishing mark of the

1259: W Vir stars, the longer period Type II Cepheids, is a composition

1260: modified by dust-gas separation to varying degrees. 

1261: 

1262: 

1263: Our following discussion assumes that the post-BHB evolution

1264: follows the lines given in Section 4.1 (Figure 2): evolution follows either a

1265: track-direct or a track-bluenose and the order of development is

1266: BHB $\rightarrow$ BL Her $\rightarrow$ W Vir $\rightarrow$ post-AGB.

1267: Then, there  are two  evolutionary puzzles presented by these 

1268: distinguishing marks:

1269: 

1270: 1: BL Her stars are Na-rich and evolve to become W Vir stars but none of the analysed

1271: W Vir stars with the possible exception of V1711 Sg are Na-rich.

1272: 

1273: 2. W Vir stars have normal sodium abundances and have evolved from BL Her stars

1274: but all of the analysed BL Her stars are Na-rich.

1275: 

1276: 

1277: 

1278: In the case of the first puzzle, the simplest explanation  is that a long-period

1279:  (W Vir or RV Tau) variable resulting from a Na-rich BL Her star

1280: transits the instability strip much faster than the W Vir star of normal

1281: Na abundance. Our discussion of Na abundances of RV Tau stars showed that

1282: BL Her-like Na overabundances are seen in approximately 10 per cent of

1283: the sample.  If V1711 Sgr is Na-rich, a similar fraction of our

1284: sample of ten W Vir  stars is Na-rich.  

1285: 

1286: Possible resolution of the second puzzle starts with the

1287: conjecture  that changes to the sodium abundance are

1288: determined prior to residence on the BHB by events

1289: associated with the mass loss (and mixing) required to divert the

1290: star from the red clump to the BHB. 

1291: Sodium production by proton capture on $^{22}$Ne occurs in

1292: association with H-burning by the CNO-cycles only at `high'

1293: temperatures. Thus, the BHB may be

1294: populated by Na-rich and Na-normal stars.

1295: Stars deposited on the BHB without experiencing sodium enrichment

1296: evolve to Na-normal  BL Her stars before  becoming Na-normal

1297: W Vir stars.

1298:  The fact that our sample of nine BL Her stars

1299: does not include a Na-normal star

1300: may be a consequence of  a lower production rate of Na-normal

1301: to Na-rich BHB stars, and a more rapid crossing of the instability strip 

1302: by the Na-normal 

1303: than the Na-rich BL Her stars.  

1304: Given that a star on a track-bluenose makes three crossings of the instability

1305: strip at a period appropriate for a BL Her star, we identify the Na-rich

1306: BL Her stars with these tracks. These tracks begin at the blue end of the

1307: BHB. 

1308: The stars on a track-direct start at the red end of the BHB, cross

1309: the instability strip once as a BL Her star and then evolve up the AGB to

1310: enjoy thermal pulses before entering the instability strip as a W Vir or

1311: RV Tau star. Such BHB stars are presumed to lack a sodium enrichment.

1312: Examination of a larger sample of BL Her stars is, then, expected to

1313: provide an example or two of a Na-rich variable.

1314: 

1315: 

1316: 

1317: 

1318: 

1319: In seeking to  resolve puzzles in stellar evolution,

1320:  one endeavors to connect the stars with their progenitors, 

1321: descendants, and close relatives among field and cluster stars.

1322:  The vast majority of our sample

1323: are probably thick disk citizens. Close relatives are the

1324: RR Lyrae stars at one extreme and RV Tau and post-AGB stars at the other

1325: extreme. We have already noted similarities between the compositions of the

1326: W Vir and RV Tau stars.

1327:  Progenitors of the BL Her and W Vir stars are  on the blue

1328: horizontal branch. Identified BHB stars have halo kinematics

1329: and metallicities and are surely progenitors of very metal-poor variables

1330: such as UY Eri and TW Cap. Thick disk BHB stars have proven

1331: elusive: `There are, however, {\it disk} RR Lyrae stars in the nearby

1332: field, but there are (as far as we know) no corresponding field BHB stars that

1333: have disk kinematics' (Kinman et al. 2000). It would be extremely

1334: valuable to locate some thick disk BHB stars and to

1335: determine their compositions, especially their Na abundances. 

1336: Our attribution of BL Her stars to a track-bluenose and W Vir stars to a

1337: track-direct implies the redder BHB stars should have a normal

1338: Na abundance and the bluer BHB stars should be sodium rich. 

1339: 

1340: 

1341:  Fresh insights into the phenomenon of the Type II Cepheids may

1342: be gleaned from abundance analyses of these variables in

1343: globular clusters. Published results apply to W Vir or RV Tau stars

1344: (i.e., periods greater than 10 days) in four

1345: globular clusters and five post-HB stars including three variables

1346: in $\omega$ Cen (cf. Wallerstein's [2002] review).

1347: All variables bar one in $\omega$ Cen have

1348: a period representative of W Vir stars and the exception with a

1349: period of 4.5 days is likely a BL Her star.

1350: Results for the globular clusters M2, M5, M10, M12, and M28 -- one

1351: from each cluster with the exception of two from M5 -- are provided

1352: variously by Gonzalez \& Lambert (1997), Carney et al. (1998), and

1353: Klochkova et al. (2003). Gonzalez \& Wallerstein (1994) give

1354: abundances for the $\omega$ Cen stars. The clusters have metallicities

1355: from [Fe/H] of $-1.2$ to $-1.7$ with the stars from $\omega$ Cen

1356: spanning the range from $-1.7$ to $-2.1$. Thus, the sample is

1357: collectively more metal-poor than the large majority of the variables

1358: discussed in this paper.

1359: 

1360: Perhaps, the most significant conclusion to be drawn from these

1361: metal-poor stars is that there is no evidence that their atmospheres

1362: have been affected by dust-gas separation, a result expected from

1363: observations of metal-poor field RV Tau stars (Giridhar et al. 2005).

1364: Interpretation of light element abundances (C to Al) is complicated

1365: by the fact that many clusters show star-to-star correlated

1366: abundance variations that are entirely missing from samples of field stars.

1367: Since such variations are present among main sequence members of a cluster,

1368: they cannot be explained wholly as internal processing and mixing.

1369: In this respect, Carney et al. (1998) show that the variable V42 in M5

1370: has the  Na enrichment and O deficiency seen in the most extreme Na-rich and

1371: O-poor red giants, i.e., the Na and O abundances of V42 may be unrelated to

1372: its transition from the red giant to the horizontal branch.

1373: Notwithstanding this novel complication presented by the clusters and

1374: their variables, the preliminary forays into abundance analyses of

1375: cluster Type II Cepheids and their horizontal branch progenitors

1376: deserve to be extended: for example, the key elements of C, N, and O

1377: remain almost untouched by the published analyses (see Gonzalez \& Wallerstein

1378: for C, N, and O among $\omega$ Cen stars, and Klochkova et al. (2003) for

1379: C, N, and O for V1 in M12).\footnote{A  remarkable result concerns the

1380: Al abundance in the $\omega$ Cen stars. In four of the five stars, Al was

1381: underabundant by $-0.3$ to $-1.1$ as measured by [Al/Fe] but without the

1382: attendant Ca and Sc deficiencies expected from dust-gas separation. Yet more

1383: remarkable is the observation that in the fifth star, Gonzalez \& Wallerstein

1384: (1994) found a strong Al overabundance ([Al/Fe] = $+1.2$) in an O-deficient

1385: star with a normal Na abundance.} Extension of the work on cluster

1386: variables to the BL Her stars would also be of interest. This would be

1387: of special interest if these BL Her stars are shown to be Na-rich. Material

1388: rich in Na from proton capture on $^{22}$Ne is possibly also He-rich.

1389: The pulsational properties of Type II Cepheids are dependent on the

1390: He abundance as well as mass and metallicity (see Bono et al. [1997] for

1391: references to these dependencies). Since BL Her stars in globular

1392: clusters have an accurately determined absolute luminosity in

1393: contrast to field stars where the luminosity is never well

1394: determined, it may be possible to determine their helium

1395: abundance.

1396:  

1397: 

1398: 

1399: Immediate   descendants of the Type II Cepheids are red giants and

1400:  must be picked out

1401: from the (presumably) more numerous giant stars evolved from clump

1402: giants. Anomalous compositions are a likely indicator.

1403: Wallerstein \& Gonzalez (1996) and Wallerstein et al. (2000) noted the

1404: similarities between the compositions of the C-rich Type II Cepheids

1405: and the R-type carbon stars (Dominy 1984) where products of the

1406: $3\alpha$-process are greatly in evidence. At the other extreme, IX Cas and

1407: the Weak G-band giants may be related examples of stars where the stellar

1408: atmosphere is dominated by CNO-cycled products and $3\alpha$--processed

1409: products are absent or just a minor contaminant. Although intriguing, these

1410: suggested links between Type II Cepheids and peculiar red giants do not

1411: directly suggest an interpretation  in terms of stellar evolution. 

1412: The observers' {\it Deus ex machina} for linking a Type II

1413: Cepheid to a peculiar red giant  may be the He-core flash but,

1414: as noted earlier, theoreticians discount this possibility. 

1415: Perhaps,  the appeal should be  not to an episode in the life of a single star

1416: but to one  in the evolution of a binary

1417: star.  Such an  appeal, however, may be a forlorn one in the case of the R-type

1418: carbon stars and Weak G-band giants because neither class

1419: seem to have an unusual degree of binarity (Tomkin, Sneden, \&

1420: Cottrell 1984; McClure 1985). 

1421: 

1422: 

1423: 

1424: 

1425: 

1426: We thank Ron Wilhelm for helpful correspondence.

1427: This research has made use of NASA's Astrophysics Data System and the

1428: Centre de Don\'{e}es de Strasbourg's  SIMBAD

1429: astronomical database.

1430:  We also thank the anonymous referee for valuable suggestions.

1431: This research has been supported in part by the Robert A. Welch Foundation of

1432: Houston, Texas. 

1433: 

1434: \begin{references}

1435: 

1436: \reference{Alc98}

1437: Alcock, C., Allsman, R.A., Alves, D.R., et al. 1998, AJ, 115, 1921

1438: \reference{And71}

1439: Anderson, K.S., \& Kraft, R.P. 1971, ApJ, 167, 119

1440: \reference{And02}

1441: Andrievsky, S.M., Kovtyukh, V.V., Luck, R.E., et al. 2002, A\&A, 381, 32

1442: \reference{Asp05}

1443: Asplund, M., Grevesse, N., \& Sauval, A.J. 2005, ASP Conf. Ser., 336, 25

1444: \reference{Bal97}

1445: Balog, Z., Vink\'{o}, J., \& Kasz\'{a}s, G. 1997, AJ, 113, 1833

1446: \reference{Bar71}

1447: Barker, T., Baumgert, L.D., Butler, D., et al. 1971, ApJ, 165, 67

1448: \reference{Beh03}

1449: Behr, B.B. 2003, ApJS, 149, 67

1450: \reference{Ben05}

1451: Bensby, T., Feltzing, S., \& Lundstr\"{o}m, I. 2005, A\&A, 433, 185

1452: \reference{Ber98}

1453: Berdnikov, L.N., \& Szabados, L. 1998, AcA, 48, 763

1454: \reference{Boh74}

1455: B\"{o}hm-Vitense, E., Szkody, P., Wallerstein, G. 1974, ApJ, 194, 125

1456: \reference{Bon97}

1457: Bono, G., Caputo, F., \& Santolamazza, P. 1997, A\&A, 317, 171

1458: \reference{Bur00}

1459: Burris, D.L., Pilachowski, C.A., Armandroff, T.E., Sneden, C., Cowan, J.J., \& Roe, H. 2000, ApJ, 544, 302

1460: \reference{Cal78}

1461: Caldwell, C.N., \& Butler, D. 1978, AJ, 83, 1190

1462: \reference{Car98}

1463: Carney, B., Fry, A.M., \& Gonzalez, G. 1998, AJ, 116, 2984

1464: \reference{Dea77}

1465: Dean, J.F., Cousins, A.W.F., Bywater, R.A., \& Warren, P.R. 1977, Mem. RAS, 83, 69

1466: \reference{Dea06}

1467: Dearborn, D.S.P., Lattanzio, J.C., \& Eggleton, P.P. 2006, ApJ, 639, 405

1468: \reference{Deu96}

1469: Deupree, R.G. 1996, ApJ, 471, 377

1470: \reference{Die90}

1471: Diethelm, R. 1990, A\&A, 239, 186

1472: \reference{Dom84}

1473: Dominy, J.F. 1984, ApJS, 55, 27

1474: \reference{Edv93}

1475: Edvardsson, B., Andersen, J., Gustafsson, B., Lambert, D.L., Nissen, P.E.,

1476: \& Tomkin, J. 1993, A\&A, 275, 101

1477: \reference{Gin74}

1478: Gingold, R.A. 1974, ApJ, 193, 177

1479: \reference{Gin76}

1480: Gingold, R.A. 1976, ApJ,  204, 116

1481: \reference{Gin77}

1482: Gingold, R.A. 1977, MNRAS, 178, 533

1483: \reference{Gin85}

1484: Gingold, R.A. 1985, Mem. Soc. Astron. Italiana, 56, 169

1485: \reference{Gir05}

1486: Giridhar, S., Lambert, D.L., Reddy, B.E., Gonzalez, G., \& Yong, D. 2005,

1487: ApJ, 627, 432

1488: \reference{Gon94}

1489: Gonzalez, G., \& Wallerstein, G. 1994, AJ, 108, 1325

1490: \reference{Gon96}

1491: Gonzalez, G. \& Wallerstein, G. 1996, MNRAS, 280, 515

1492: \reference{Gon97}

1493: Gonzalez, G., \& Lambert, D.L. 1997, AJ, 114, 341

1494: \reference{Har81}

1495: Harris, H.C. 1981, AJ, 86, 719

1496: \reference{Har95}

1497: Harris, H.C. 1985, AJ, 90, 756

1498: \reference{Har84a}

1499: Harris, H.C., Olzewski, E.W., \& Wallerstein, G. 1984, AJ, 89, 119

1500: \reference{Har84b}

1501: Harris, H.C., \& Wallerstein, G. 1984, AJ, 89, 379

1502: \reference{Har89}

1503: Harris, H.C., \& Welch, D.L. 1989, AJ, 98, 981

1504: \reference{Ibe91}

1505: Iben, I. Jr. 1991, ApJS, 76, 55

1506: \reference{Kin00}

1507: Kinman, T., Castelli, F., Cacciari, C. et al. 2000, A\&A, 364, 102

1508: \reference{Klo03}

1509: Klochkova, V.G., Panchuk, V.E., Tavolganskaya, N.S., \&

1510:    Kovtyukh, V.V. 2003, Astr. Letters, 29, 748

1511: \reference{Kra72}

1512: Kraft, R.P. 1972, Dudley Obs. Rept, No. 4, 69

1513: \reference{Llo70}

1514: Lloyd Evans, T. 1970, Obs, 90, 254

1515: \reference{Llo83}

1516: Lloyd Evans, T. 1983, Obs, 103, 276

1517: \reference{Lyo85}

1518: Lloyd Evans, T. 1985, MNRAS, 217, 493

1519: \reference{Luc89}

1520: Luck, R.E., \& Bond, H.E. 1989, ApJ, 342, 476

1521: \reference{Maa02}

1522: Maas, T., Van Winckel, H., \& Waelkens, C. 2002, A\&A, 386, 504 

1523: \reference{McA86}

1524: McAlary, C.W., \& Welch, D.L. 1986, AJ, 91, 1209

1525: \reference{McC85}

1526: McClure, R.D. 1985, JRASC, 79, 277

1527: \reference{McW97}

1528: McWilliam, A. 1997, ARAA, 35, 503

1529: \reference{Mea91}

1530: Meakes, M., Wallerstein, G, \& Opalko, J.F. 1991, AJ, 101, 1795

1531: %\reference{Pol00}

1532: %Pollard, K.R., Alcock, C., Allsman, R.A. et al. 2000, ASP Conf. Ser. 203, 89

1533: \reference{Pre59}

1534: Preston, G.W. 1959, ApJ, 130, 507

1535: \reference{Pri02}

1536: %Pritzl, B.J., Smith, H.A., Catelan, M., \& Sweigart, A.V. 2002, AJ, 124, 949

1537: %\reference{Pri03}

1538: %Pritzl, B.J., Smith, H.A., Stetson, P.B., Catelan, M., et al. 2003, AJ, 126, 1381

1539: %\reference{Rao05}

1540: Rao, N.K., \& Reddy, B.E. 2005, MNRAS, 357, 235

1541: \reference{Red06}

1542: Reddy, B.E., Lambert, D.L., \& Allende Prieto, C. 2006, MNRAS, 367, 1329

1543: \reference{Red03}

1544: Reddy, B.E., Tomkin, J., Lambert, D.L., \& Allende Prieto, C. 2003, MNRAS, 340, 304

1545: \reference{Rod63}

1546: Rodgers, A.W., \& Bell, R.A. 1963, MNRAS, 125, 487

1547: \reference{Rod68}

1548: Rodgers, A.W., \& Bell, R.A. 1968, MNRAS, 139, 175

1549: \reference{San91}

1550: Sanwal, N.N., \& Sarma, M.B.K. 1991, JAA, 12, 119

1551: \reference{Sch04a}

1552: Schmidt, E.G., Johnston, D., Langan, S., \& Lee, K.M. 2004a, AJ, 128, 1748

1553: \reference{Sch05}

1554: Schmidt, E.G., Johnston, D., Langan, S., \& Lee, K.M. 2005, AJ, 130, 832

1555: \reference{Sch04b}

1556: Schmidt, E.G., Johnston, D., Lee, K.M., \& Langan, S. 2004b, AJ, 1288, 2988

1557: \reference{Smi98}

1558: Smith, G.H. 1998, PASP, 110, 1119

1559: \reference{Sne78}

1560: Sneden, C., Lambert, D.L., Tomkin, J., \& Peterson, R.C. 1978, ApJ, 222, 585

1561: \reference{Sto70}

1562: Stobie, R.S. 1970, MNRAS, 148, 1

1563: \reference{Taa76}

1564: Taam, R.E., Kraft, R.P., \& Suntzeff, N. 1976, ApJ, 204, 842

1565: \reference{Tom84}

1566: Tomkin, J., Sneden, C, \& Cottrell, P.L. 1984, PASP, 96, 609

1567: \reference{Tul95}

1568: Tull, R.G., MacQueen, P.J., Sneden, C., \& Lambert, D.L. 1995, PASP, 107, 251

1569: \reference{Van03}

1570: Van Winckel, H. 2003, ARAA, 41, 391

1571: \reference{Van95}

1572: Van Winckel, H., Waelkens, C., \& Waters, L.B.F.M. 1995, A\&A, 293, L25

1573: \reference{Vin98}

1574: Vink\'{o}, J., Remage Evans, N., Kiss, L., \& Szabados, L. 1998, MNRAS, 296, 824

1575: \reference{Vin93}

1576: Vink\'{o}, J., Szabados, L., Szatm\'{a}ry, K. 1993, A\&A, 279, 410

1577: \reference{Wal02}

1578: Wallerstein, G.  2002, PASP, 114, 689

1579: \reference{Wal96}

1580: Wallerstein, G., \& Gonzalez, G. 1996, MNRAS, 282, 1236

1581: \reference{Wal00}

1582: Wallerstein, G., Matt, S., Gonzalez, G. 2000, MNRAS, 311, 414

1583: 

1584: \end{references}

1585: 

1586: 

1587: %1

1588: \begin{figure}

1589: \plotone{f1.ps}

1590: \figcaption{The interval 7105--7135 \AA\ for CC Lyr, SW Tau, IX Cas, and BL Her.

1591: Note the 

1592: absence of the C\,{\sc i} lines for IX Cas, and the absence of the Fe\,{\sc i}

1593: and Ni\,{\sc i} lines for CC Lyr.}

1594: \end{figure}

1595: 

1596: %2

1597: 

1598: \begin{figure}

1599: %\plotone{f2.eps}

1600: \figcaption{A HR diagram showing the location of the Population II instability

1601: strip and approximate lines of constant period. Superimposed  are

1602: two evolutionary tracks for two stars: the track for the  0.6$M_\odot$ star is

1603: a track-direct and that for the 0.536$M_\odot$ is a track-bluenose (see text).

1604: The location of the RR Lyr variables and the approximate positions of

1605: RV Tau variables are marked. (After Gingold 1985)}

1606: \end{figure}

1607:  

1608: 

1609: 

1610: %3

1611: \begin{figure}

1612: \plotone{f3.eps}

1613: \figcaption{The [Na/Fe] (top panel) and [Al/Fe] (bottom panel) ratios

1614: versus [Fe/H]. The key to the symbols is given in the top panel.

1615: Symbols: unfilled circles -- Our BL Her stars,

1616:  filled

1617: circles -- Our W Vir stars, unfilled squares -- The C-rich BL Her

1618: stars RT TrA and V553 Cen, filled triangles-- The W Vir stars $\kappa$ Pav at

1619: [Fe/H] $\simeq 0.0$ and ST Pup at [Fe/H] $\simeq -1.5$.} 

1620: \end{figure}

1621: 

1622: 

1623: %4

1624: \begin{figure}

1625: \plotone{f4.eps}

1626: \figcaption{The run of the abundances of $\alpha$-elements with [Fe/H].

1627: The top panel shows the ratio [$\alpha$/Fe] versus [Fe/H] where

1628: [$\alpha$/Fe] is the mean of the ratios [Mg/Fe], [Si/Fe], and [S/Fe].

1629: The middle panel shows [Ca/Fe] versus [Fe/H]. The bottom panel

1630: shows [Ti/Fe] versus [Fe/H]. Symbols are as in Figure 3. 

1631: }

1632: \end{figure}

1633: 

1634: %5

1635: \begin{figure}

1636: \plotone{f5.eps}

1637: \figcaption{The run of the ratios [Cr/Fe] (top panel), [Mn/Fe] (middle

1638: panel), and [Ni/Fe] (bottom panel) versus [Fe/H]. Symbols are as in

1639: Figure 3.}

1640: \end{figure}

1641: 

1642: 

1643: %6

1644: \begin{figure}

1645: \plotone{f6.eps}

1646: \figcaption{The run of the ratio [$s$/Fe] (top panel) and [Eu/Fe]

1647: (bottom panel) versus [Fe/H]. See text for the elements

1648: contributing to the mean [$s$/Fe].  Symbols are as in Figure 3.}

1649: \end{figure}

1650: 

1651: %7

1652: \begin{figure}

1653: \plotone{f7.eps}

1654: \figcaption{Abundances [X/H] versus condensation temperature $T_C$

1655: for CC Lyr. Elements are identified by their chemical symbol.}

1656: \end{figure}

1657: 

1658: 

1659: %8

1660: \begin{figure}

1661: \plotone{f8.eps}

1662: \figcaption{Adjusted abundances (see text) [X/H] for CO Pup (top

1663: panel) and W Vir (bottom panel) versus

1664: condensation temperature $T_C$. Elements are identified by their

1665: chemical symbol. Unfilled circles refer to elements made principally

1666: by the $s$-process whose abundance might have been enhanced by thermal

1667: pulses on the AGB.}

1668: \end{figure}

1669: 

1670: 

1671: %9

1672: \begin{figure}

1673: \plotone{f9.eps}

1674: \figcaption{Adjusted abundances (see text) [X/H] for

1675: V1711 Sgr (top panel) and SZ Mon (bottom panel) versus

1676: condensation temperature $T_C$. Elements are identified by their

1677: chemical symbol.}

1678: \end{figure}

1679: 

1680: 

1681: %10

1682: \begin{figure}

1683: \plotone{f10.ps}

1684: \figcaption{Adjusted abundances [X/H] for RX Lib (top panel)

1685:  and MZ Cyg (bottom panel) versus

1686: condensation temperature $T_C$. Elements are identified by their

1687: chemical symbol.}

1688: \end{figure}

1689: 

1690: %11

1691: \begin{figure}

1692: \plotone{f11.eps}

1693: \figcaption{Comparisons of our spectroscopic and published

1694: photometric metallicities. The lefthand panel shows the comparison

1695: with the results from Washington photometry (Harris 1985),

1696: the middle panel with  Walraven

1697: photometry (Diethelm 1990), and the righthand panel with Str\"{o}mgren

1698: photometry (Meakes et al. 1991). The dashed line in each panel

1699: indicates equality between spectroscopic and photometric metallicity. The key to the symbols is given in Figure 3.}

1700: \end{figure}

1701: 

1702: 

1703: \input{tab1.tex}

1704: 

1705: 

1706: \input{tab2.tex}

1707: 

1708: 

1709: \input{tab3.tex}

1710: 

1711: \input{tab4.tex}

1712: 

1713: 

1714: \end{document}

1715: 

1716: 

1717: 

1718: 

1719: 

1720: 

1721: